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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1a7ab7b4-5d3e-42da-955c-aef2a917badf | molecular-dipole-moment-learning-via | 2205.15510 | null | https://arxiv.org/abs/2205.15510v1 | https://arxiv.org/pdf/2205.15510v1.pdf | Molecular Dipole Moment Learning via Rotationally Equivariant Gaussian Process Regression with Derivatives in Molecular-orbital-based Machine Learning | This study extends the accurate and transferable molecular-orbital-based machine learning (MOB-ML) approach to modeling the contribution of electron correlation to dipole moments at the cost of Hartree-Fock computations. A molecular-orbital-based (MOB) pairwise decomposition of the correlation part of the dipole moment is applied, and these pair dipole moments could be further regressed as a universal function of molecular orbitals (MOs). The dipole MOB features consist of the energy MOB features and their responses to electric fields. An interpretable and rotationally equivariant Gaussian process regression (GPR) with derivatives algorithm is introduced to learn the dipole moment more efficiently. The proposed problem setup, feature design, and ML algorithm are shown to provide highly-accurate models for both dipole moment and energies on water and fourteen small molecules. To demonstrate the ability of MOB-ML to function as generalized density-matrix functionals for molecular dipole moments and energies of organic molecules, we further apply the proposed MOB-ML approach to train and test the molecules from the QM9 dataset. The application of local scalable GPR with Gaussian mixture model unsupervised clustering (GMM/GPR) scales up MOB-ML to a large-data regime while retaining the prediction accuracy. In addition, compared with literature results, MOB-ML provides the best test MAEs of 4.21 mDebye and 0.045 kcal/mol for dipole moment and energy models, respectively, when training on 110000 QM9 molecules. The excellent transferability of the resulting QM9 models is also illustrated by the accurate predictions for four different series of peptides. | ['Thomas F. Miller III', 'Lixue Cheng', 'Jiace Sun'] | 2022-05-31 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [ 2.60566920e-01 -1.36674002e-01 -1.40320465e-01 -4.01802599e-01
-1.01589060e+00 -2.01654032e-01 4.79157060e-01 4.48550284e-01
-3.60008627e-01 1.15936625e+00 -2.54062712e-01 -3.47944051e-01
-3.16059321e-01 -6.82314336e-01 -7.08618045e-01 -1.42378008e+00
-4.57657427e-01 6.17508888e-01 -1.57691523e-01 -2.01915547e-01
6.31064832e-01 9.50133026e-01 -1.49433029e+00 2.13338479e-01
1.44945276e+00 7.68697083e-01 2.17368364e-01 4.24060255e-01
4.54424292e-01 6.10118151e-01 -1.34404063e-01 -1.74274653e-01
-9.75769311e-02 -1.82493120e-01 -6.81903183e-01 -1.19037822e-01
2.19826177e-01 1.80544361e-01 -6.57474622e-02 7.33318329e-01
6.28213406e-01 7.56346345e-01 1.37758839e+00 -7.73659587e-01
-6.41110718e-01 -5.42782769e-02 -3.92680436e-01 -2.54934371e-01
4.33059782e-01 7.43883196e-03 9.63864684e-01 -1.27844179e+00
7.87466109e-01 1.24193072e+00 4.79573280e-01 2.36470401e-01
-1.67962623e+00 -4.59768772e-01 -4.80774343e-01 3.76557082e-01
-1.54686654e+00 -3.27828109e-01 7.49639571e-01 -4.90666330e-01
1.61208737e+00 3.99540752e-01 3.72686148e-01 8.26859355e-01
9.34803128e-01 6.96022958e-02 9.42259014e-01 -7.08346546e-01
5.00652492e-01 2.73957491e-01 1.40660241e-01 6.26999080e-01
3.40347826e-01 -4.28834669e-02 -5.67092955e-01 -7.39115477e-01
3.24946374e-01 4.46811765e-02 3.09834424e-02 -6.47617817e-01
-8.47571909e-01 1.09868419e+00 2.12854058e-01 -1.12466708e-01
-5.94293594e-01 -1.81866869e-01 9.93565992e-02 -5.20572960e-02
2.76314437e-01 6.72439635e-01 -5.62296152e-01 1.19293079e-01
-7.03448892e-01 3.16125363e-01 1.05207324e+00 5.64723492e-01
9.72067416e-01 -1.74931437e-02 1.71457812e-01 5.42139053e-01
6.21622145e-01 6.01671040e-01 1.70410603e-01 -7.33057439e-01
2.47331962e-01 2.51920372e-01 2.53833979e-01 -1.01615202e+00
-4.86184508e-01 -5.56982197e-02 -6.74881399e-01 -1.19210154e-01
-1.98399648e-03 -1.05522230e-01 -5.68371713e-01 1.40137136e+00
5.61088324e-01 -6.56455934e-01 3.19573879e-01 3.80354702e-01
7.14396954e-01 1.00566387e+00 3.79580528e-01 -6.46659851e-01
1.09183455e+00 -7.28011727e-01 -4.16788638e-01 3.30706865e-01
1.00206828e+00 -7.01250017e-01 5.15818954e-01 4.53513682e-01
-9.39181268e-01 -4.96356994e-01 -6.92039490e-01 2.54214089e-02
-9.77507904e-02 3.22415322e-01 1.15391564e+00 5.42053699e-01
-6.83465838e-01 1.11966729e+00 -1.11398137e+00 -1.89142197e-01
6.79038763e-02 9.92124856e-01 -7.20875382e-01 8.12396854e-02
-9.28601801e-01 9.20453906e-01 5.69842935e-01 1.64802484e-02
-9.13596392e-01 -6.29785657e-01 -7.57378697e-01 -2.13998836e-02
-3.18073392e-01 -3.72393280e-01 6.00417078e-01 -7.98961043e-01
-1.82510853e+00 4.12843347e-01 -5.69869220e-01 -3.10639918e-01
-1.88647270e-01 1.75644785e-01 -6.03503942e-01 4.94773060e-01
-1.63247287e-02 7.79437304e-01 5.46155632e-01 -1.06976295e+00
1.10034488e-01 -2.60681599e-01 -4.56441432e-01 1.64889649e-01
-1.61811739e-01 1.65248346e-02 2.76433229e-01 -7.48330131e-02
2.80732125e-01 -9.98923242e-01 -4.98468131e-01 -7.05225229e-01
-6.37315631e-01 -1.83680654e-01 1.98929876e-01 -6.11573637e-01
9.28185642e-01 -1.93013740e+00 4.25730735e-01 6.62716508e-01
-1.30296305e-01 -1.46721840e-01 -1.30762279e-01 1.17929375e+00
-5.65062582e-01 -7.51752555e-02 -3.82634252e-01 3.46584246e-02
-2.09314257e-01 1.59517810e-01 -6.54737353e-02 6.19148970e-01
3.52855355e-01 4.96762246e-01 -6.16999865e-01 -3.17784131e-01
2.74724126e-01 9.38810647e-01 -1.07308328e+00 1.51197180e-01
1.49885073e-01 6.36734128e-01 -5.75459123e-01 7.21127450e-01
9.42209125e-01 -1.75202355e-01 5.97043455e-01 -4.44736063e-01
-4.11886930e-01 3.72670889e-01 -9.00821328e-01 1.40354323e+00
8.51544663e-02 1.36877373e-01 -3.96031916e-01 -1.07663536e+00
1.08352983e+00 2.77436405e-01 8.89589906e-01 -5.38770735e-01
-2.43469432e-01 3.56024772e-01 5.18040240e-01 -4.81189758e-01
3.90855432e-01 -4.56953526e-01 2.14662611e-01 -2.05877200e-01
6.39445007e-01 -2.37268358e-02 2.75523216e-01 2.23491564e-02
5.03185928e-01 3.81154656e-01 4.23713535e-01 -8.70942891e-01
9.17499781e-01 8.96554068e-02 1.70714274e-01 4.49810386e-01
2.95700163e-01 1.91046223e-01 3.99627268e-01 -4.02552813e-01
-8.51906836e-01 -8.87034774e-01 -9.32132721e-01 1.05069411e+00
-6.84328824e-02 -6.21710598e-01 -7.41356432e-01 -9.67803895e-02
1.06155202e-01 8.01909566e-01 -3.40016574e-01 -3.85020792e-01
-1.27966821e-01 -1.65551436e+00 6.08439408e-02 3.25139947e-02
-6.60937652e-02 -1.07342994e+00 1.31385885e-02 4.24931169e-01
1.71917811e-01 -5.77199459e-01 1.65321544e-01 8.45349073e-01
-9.83997285e-01 -9.79482412e-01 -2.08763897e-01 -4.26673710e-01
7.61334777e-01 -6.19219057e-02 4.93665308e-01 -4.34915185e-01
-3.52011532e-01 3.03112447e-01 -1.36903217e-02 -2.39619315e-01
-4.89069909e-01 -5.89291677e-02 4.64864343e-01 1.46995082e-01
4.99835908e-01 -8.00213099e-01 -6.08195543e-01 1.02865852e-01
-6.44983172e-01 -2.58368760e-01 6.78225219e-01 8.88689935e-01
1.00965178e+00 -1.80301696e-01 4.27293539e-01 -6.14845157e-01
3.58828068e-01 -6.88629270e-01 -4.53041703e-01 4.92385179e-02
-6.85611546e-01 4.11887258e-01 6.28847241e-01 -3.86795670e-01
-1.24460196e+00 1.78834125e-01 -2.81695724e-01 2.57791758e-01
-2.89897233e-01 6.22415781e-01 -4.08617556e-01 -6.90548956e-01
7.69267917e-01 2.93654710e-01 -2.61414975e-01 -5.30837417e-01
1.01482552e-02 5.68608820e-01 5.14819361e-02 -1.08427000e+00
4.02204931e-01 3.21487606e-01 7.59147108e-01 -1.27982926e+00
-3.98946851e-01 -5.42922854e-01 -8.10104489e-01 4.03665930e-01
8.52262318e-01 -9.22869325e-01 -1.21279585e+00 5.51621139e-04
-9.74263966e-01 6.10199757e-03 1.10156536e-01 9.11859453e-01
-5.84483862e-01 8.44183028e-01 -5.46049476e-01 -1.11322343e+00
-4.80357051e-01 -1.48392415e+00 9.54870641e-01 7.34901521e-03
1.81456786e-02 -1.14104760e+00 2.33175129e-01 4.32569116e-01
3.97749571e-03 5.24329424e-01 1.22876549e+00 -9.13687170e-01
-3.99262816e-01 -2.82582164e-01 2.56853968e-01 3.15520197e-01
-1.07298508e-01 3.04960608e-01 -9.05209124e-01 -4.84197468e-01
-1.23604298e-01 -2.19280332e-01 7.38319933e-01 5.76585174e-01
1.01368260e+00 -2.63465792e-01 -4.22409207e-01 6.00188673e-01
1.57739592e+00 5.35917640e-01 7.30680585e-01 2.79401511e-01
8.87236953e-01 3.27015370e-01 5.33973634e-01 6.31166577e-01
1.63902193e-01 6.70812488e-01 3.43662322e-01 5.49901314e-02
7.82957613e-01 -1.36494949e-01 7.19557345e-01 9.87035334e-01
-8.89333308e-01 -1.33638799e-01 -7.55118549e-01 -1.47135451e-01
-1.52520978e+00 -1.12212479e+00 -7.32902467e-01 2.49673867e+00
8.78226221e-01 -2.39820510e-01 -1.49207622e-01 -9.64282230e-02
2.86974728e-01 -2.70135194e-01 -3.22028488e-01 -7.29196966e-01
-9.35236886e-02 5.74428856e-01 5.09480894e-01 7.18396962e-01
-1.08977330e+00 7.11590827e-01 6.11253881e+00 9.94156063e-01
-1.02969015e+00 -1.59808189e-01 4.60380644e-01 2.41532668e-01
-3.17003697e-01 3.98674965e-01 -9.91107702e-01 3.16994488e-01
1.36476922e+00 1.66931197e-01 4.32591170e-01 8.99215043e-01
3.40731353e-01 -3.02197427e-01 -1.03889966e+00 8.88114512e-01
-1.64259464e-01 -1.08186877e+00 3.89708728e-01 4.47564662e-01
9.51088965e-01 -1.43242590e-02 -3.71207707e-02 2.41045415e-01
-5.24705589e-01 -1.21225297e+00 3.16366315e-01 7.93299735e-01
6.52576327e-01 -1.06695366e+00 5.37324250e-01 2.73293346e-01
-8.44333887e-01 1.24135189e-01 -9.35078919e-01 -3.60765234e-02
-1.38429075e-01 4.66428787e-01 -7.79302537e-01 8.89584303e-01
2.92988807e-01 5.03400207e-01 -2.44433105e-01 7.27257907e-01
1.08921640e-01 4.90246385e-01 -3.11169028e-01 3.19158356e-03
2.05094531e-01 -9.33507025e-01 3.85615855e-01 1.11201739e+00
4.31318641e-01 1.11954212e-01 4.00990136e-02 7.35636592e-01
1.79152533e-01 7.34433711e-01 -2.85591424e-01 -3.35200340e-01
1.97946250e-01 1.38838172e+00 -3.94134909e-01 -5.64917810e-02
-2.34056532e-01 7.92621374e-01 1.78225681e-01 5.75396836e-01
-6.78397000e-01 -3.71789038e-01 5.72361171e-01 -8.72614048e-03
3.00374925e-01 -2.41672248e-01 6.35252297e-01 -1.20456553e+00
-2.75443107e-01 -6.75146639e-01 -1.59317143e-02 -5.35826266e-01
-1.12232387e+00 4.34897840e-01 3.13732997e-02 -6.81609750e-01
-7.52181709e-02 -1.04844344e+00 -5.08027673e-01 1.02977741e+00
-1.55241024e+00 -8.80616963e-01 2.32631728e-01 4.25506592e-01
-1.25566632e-01 -1.47336468e-01 1.41019022e+00 1.10743336e-01
-6.50108278e-01 2.65469253e-01 9.65292990e-01 -6.11543179e-01
6.90904915e-01 -1.30556250e+00 -2.55426139e-01 2.68139690e-01
-9.09598917e-02 1.04203033e+00 7.41920829e-01 -5.48458576e-01
-1.81341720e+00 -9.29169297e-01 8.31888616e-01 -6.03598058e-01
4.80069995e-01 -2.22353816e-01 -9.53160226e-01 3.67197067e-01
-1.28930405e-01 5.47762327e-02 1.29294765e+00 -1.66006479e-02
8.79816636e-02 1.25670895e-01 -1.17559004e+00 2.37441495e-01
4.72309530e-01 -5.95148027e-01 -3.53533477e-01 9.01017785e-01
2.02306956e-01 -1.61595136e-01 -1.52313793e+00 2.92221844e-01
4.17565554e-01 -1.08953536e+00 1.27382231e+00 -9.25767243e-01
8.22402388e-02 -1.10659353e-01 -5.12314200e-01 -8.16945374e-01
-4.12139416e-01 -7.98322499e-01 -1.27749890e-01 7.54126787e-01
4.74358767e-01 -7.27417231e-01 3.61736000e-01 7.75938272e-01
-2.67838120e-01 -9.73182559e-01 -1.00150728e+00 -4.11946744e-01
1.79554000e-01 -2.52988935e-01 1.76053613e-01 8.25059414e-01
2.96475053e-01 2.10253879e-01 -3.85206610e-01 3.35440010e-01
6.64153457e-01 1.02438211e-01 4.23379391e-01 -1.53967345e+00
-3.42573285e-01 2.09157258e-01 -3.55331987e-01 -6.31735146e-01
2.87720025e-01 -1.17169797e+00 -3.21570069e-01 -1.09613538e+00
4.38683033e-01 -2.47312695e-01 -4.23977584e-01 1.94696650e-01
4.00372818e-02 -2.73684263e-01 -1.82845488e-01 3.71717811e-01
-5.84093571e-01 8.50336969e-01 1.01228654e+00 1.46474823e-01
-5.06285548e-01 -3.40481959e-02 -3.44496340e-01 6.07212484e-01
6.08604491e-01 -7.30260730e-01 -1.52332619e-01 5.34028292e-01
1.68080643e-01 2.01302052e-01 2.85828590e-01 -1.02177691e+00
-9.96674299e-02 -2.13674918e-01 6.75856709e-01 -4.05308574e-01
5.81554174e-01 -5.53478956e-01 4.32200313e-01 4.78327960e-01
-9.70051214e-02 -3.34874958e-01 1.36442795e-01 4.93957609e-01
-2.78772324e-01 -3.00735891e-01 5.49620569e-01 4.51123081e-02
-3.03902000e-01 4.08618331e-01 -5.86470127e-01 -5.29959619e-01
7.91032970e-01 -2.03157336e-01 1.75646871e-01 1.02598578e-01
-8.92551541e-01 -3.84666622e-01 3.90109330e-01 -4.19643968e-01
4.36631829e-01 -1.15166807e+00 -3.16277474e-01 3.47303659e-01
5.27760796e-02 -2.88999230e-01 5.97540140e-01 1.33233571e+00
-6.61812127e-01 9.37099457e-01 -1.68256417e-01 -5.38089395e-01
-1.01576078e+00 5.65714240e-01 5.42315423e-01 -1.51162699e-01
5.79772852e-02 5.15060246e-01 2.97427207e-01 -6.12969220e-01
-3.53154510e-01 -1.43308058e-01 -1.63640752e-01 -1.70752630e-02
1.91021234e-01 5.95278740e-01 1.66348487e-01 -1.02694607e+00
-5.92164397e-01 6.15465820e-01 3.80226299e-02 3.40179116e-01
1.58946872e+00 1.82151243e-01 -3.35727096e-01 7.15401489e-03
1.40272856e+00 2.98524410e-01 -1.13298869e+00 3.82494658e-01
-7.79232308e-02 1.14887811e-01 4.59024590e-03 -4.25581694e-01
-2.55272359e-01 1.10602713e+00 5.77629030e-01 -2.88409412e-01
6.03054762e-01 -2.62027502e-01 1.75629646e-01 1.03716362e+00
4.29590583e-01 -8.80386055e-01 -2.95880288e-01 2.84084827e-01
6.98663890e-01 -1.19016647e+00 1.58760145e-01 -4.99609739e-01
-6.90170944e-01 1.55230558e+00 1.62437677e-01 -1.32607576e-02
3.85427743e-01 -5.48998773e-01 -5.85590601e-01 -3.18188697e-01
-4.80749875e-01 2.05333978e-01 6.86213911e-01 6.80372894e-01
8.23667109e-01 1.35862306e-01 -3.82785350e-01 6.09103322e-01
1.58518210e-01 -3.29972595e-01 3.53502333e-01 7.25384533e-01
-4.27937865e-01 -1.34911382e+00 -3.20592195e-01 1.43154174e-01
-4.49469268e-01 -4.35590833e-01 -5.85688055e-02 7.66870737e-01
1.25753850e-01 9.00308013e-01 -2.84334153e-01 8.57603103e-02
4.55343239e-02 3.31864208e-01 6.18022978e-01 -4.65850115e-01
-2.15744182e-01 2.12156266e-01 8.19349363e-02 -4.25087899e-01
-6.78780138e-01 -5.76234758e-01 -1.56106734e+00 -2.60737896e-01
-7.40585148e-01 9.24416602e-01 6.98508859e-01 9.37847197e-01
5.38886964e-01 4.73466143e-02 7.83426702e-01 -1.23649049e+00
-6.65620029e-01 -8.95816445e-01 -9.36547637e-01 3.73164982e-01
-2.91000996e-02 -7.74835825e-01 -3.96869689e-01 -1.08170815e-01] | [5.163029193878174, 5.355057239532471] |
e7b2eb01-2f9f-4bee-b273-aae8e0c7191b | an-image-quality-assessment-dataset-for | 2304.05772 | null | https://arxiv.org/abs/2304.05772v1 | https://arxiv.org/pdf/2304.05772v1.pdf | An Image Quality Assessment Dataset for Portraits | Year after year, the demand for ever-better smartphone photos continues to grow, in particular in the domain of portrait photography. Manufacturers thus use perceptual quality criteria throughout the development of smartphone cameras. This costly procedure can be partially replaced by automated learning-based methods for image quality assessment (IQA). Due to its subjective nature, it is necessary to estimate and guarantee the consistency of the IQA process, a characteristic lacking in the mean opinion scores (MOS) widely used for crowdsourcing IQA. In addition, existing blind IQA (BIQA) datasets pay little attention to the difficulty of cross-content assessment, which may degrade the quality of annotations. This paper introduces PIQ23, a portrait-specific IQA dataset of 5116 images of 50 predefined scenarios acquired by 100 smartphones, covering a high variety of brands, models, and use cases. The dataset includes individuals of various genders and ethnicities who have given explicit and informed consent for their photographs to be used in public research. It is annotated by pairwise comparisons (PWC) collected from over 30 image quality experts for three image attributes: face detail preservation, face target exposure, and overall image quality. An in-depth statistical analysis of these annotations allows us to evaluate their consistency over PIQ23. Finally, we show through an extensive comparison with existing baselines that semantic information (image context) can be used to improve IQA predictions. The dataset along with the proposed statistical analysis and BIQA algorithms are available: https://github.com/DXOMARK-Research/PIQ2023 | ['Jean Ponce', 'Sira Ferradans', 'Theo Cayla', 'Davide Garcia-Civiero', 'Ana-Stefania Calarasanu', 'Nicolas Chahine'] | 2023-04-12 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chahine_An_Image_Quality_Assessment_Dataset_for_Portraits_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chahine_An_Image_Quality_Assessment_Dataset_for_Portraits_CVPR_2023_paper.pdf | cvpr-2023-1 | ['image-quality-assessment'] | ['computer-vision'] | [ 3.33598882e-01 -1.56071231e-01 -2.35237986e-01 -5.18203318e-01
-1.07452571e+00 -7.68978417e-01 5.50948024e-01 1.35584781e-02
-1.38805613e-01 5.32852292e-01 1.86715290e-01 -6.50510192e-02
-1.00756191e-01 -5.13829112e-01 -5.72328269e-01 -4.47812557e-01
3.72990191e-01 9.03477818e-02 -1.19036593e-01 2.30712327e-03
2.46882170e-01 4.50205863e-01 -1.68498969e+00 3.50241274e-01
1.13195860e+00 1.34942698e+00 -4.35626134e-02 6.07237935e-01
1.85598552e-01 3.72796446e-01 -7.20286608e-01 -1.44522035e+00
3.93925756e-01 -2.61086524e-01 -6.01327300e-01 4.86537933e-01
1.10871422e+00 -5.37637532e-01 -7.99200833e-02 1.00518370e+00
7.18340337e-01 -2.97492027e-01 5.67184031e-01 -1.68643332e+00
-7.69774258e-01 -1.47064582e-01 -4.49617445e-01 -2.56366492e-03
6.84861779e-01 5.00792325e-01 1.10345912e+00 -9.13090050e-01
4.68325645e-01 1.26172888e+00 8.20425451e-01 3.15983236e-01
-1.12776899e+00 -9.63877082e-01 -1.89904004e-01 5.08010089e-01
-1.55565190e+00 -8.11900854e-01 6.34557068e-01 -4.60785747e-01
4.91655111e-01 3.87572169e-01 8.38693202e-01 1.20306587e+00
-1.31104961e-02 6.03244781e-01 1.59727061e+00 -2.29429454e-01
3.04554731e-01 4.03471321e-01 -2.97462940e-01 5.58005333e-01
1.90534964e-01 -1.29430229e-02 -6.98961020e-01 -1.64827913e-01
5.19318581e-01 -2.19145611e-01 -1.72683358e-01 -2.56953984e-01
-8.17798555e-01 4.57970768e-01 3.75602841e-01 -1.88418888e-02
-9.30669382e-02 -2.09176913e-01 1.44261748e-01 2.59593755e-01
5.04808009e-01 2.62413502e-01 -2.58288383e-01 -1.64810807e-01
-1.10433578e+00 8.98385122e-02 5.22872210e-01 9.24581826e-01
7.19882607e-01 -1.76384047e-01 -2.48765066e-01 1.13576007e+00
1.87873766e-01 9.29782510e-01 1.46012262e-01 -1.43659151e+00
3.86404485e-01 7.06102133e-01 3.74844462e-01 -1.27694809e+00
-7.69687211e-03 -4.93069105e-02 -6.30148590e-01 3.34706336e-01
5.64407229e-01 2.66444176e-01 -6.27076864e-01 1.43929088e+00
1.90354735e-01 -3.76652949e-03 -2.65276372e-01 8.50178301e-01
6.84295058e-01 3.43011290e-01 2.12042436e-01 -1.43506229e-01
1.51109779e+00 -5.39908469e-01 -6.86048448e-01 -1.14960909e-01
5.70087247e-02 -1.00258064e+00 1.50467205e+00 6.42077982e-01
-1.17552400e+00 -6.99128985e-01 -9.56381798e-01 -1.37826994e-01
-3.59375805e-01 4.21497017e-01 4.14459795e-01 1.36185098e+00
-1.21058154e+00 3.88995171e-01 -1.84473321e-01 -4.82536376e-01
9.47852433e-01 3.30440849e-01 -6.41200840e-01 -4.42275405e-01
-9.69593287e-01 6.86619461e-01 -2.25150824e-01 5.69372512e-02
-9.76059258e-01 -8.05345237e-01 -7.56003439e-01 -2.61530280e-01
2.28764668e-01 -5.12038231e-01 1.10598207e+00 -1.56426930e+00
-1.48250723e+00 1.21793664e+00 -3.57910722e-01 -1.46168232e-01
6.37819648e-01 -4.85541373e-02 -7.08293259e-01 3.13245267e-01
2.43027315e-01 8.32089126e-01 1.06992817e+00 -1.28568327e+00
-5.52897811e-01 -5.35376251e-01 2.56438136e-01 1.90055162e-01
-4.56108153e-01 1.84497431e-01 -7.52644897e-01 -6.17153883e-01
-4.28015113e-01 -9.44062889e-01 2.89124131e-01 3.91657293e-01
-3.51127476e-01 7.60170212e-03 3.83820862e-01 -9.17565048e-01
9.64835167e-01 -2.18200135e+00 -2.70607144e-01 1.46386743e-01
1.97476447e-02 2.86042720e-01 -2.86417633e-01 1.05337530e-01
9.23982784e-02 3.11290711e-01 -3.01269948e-01 -6.62284076e-01
9.13668498e-02 -4.07268405e-02 6.63082600e-02 5.43160141e-01
3.36112946e-01 9.03437912e-01 -6.92485273e-01 -7.20042348e-01
3.63567770e-01 4.89131927e-01 -2.82953978e-01 2.04959363e-01
5.52632734e-02 3.01222831e-01 7.72759318e-02 1.15024042e+00
9.37084615e-01 -2.07974479e-01 1.52143583e-01 -5.97685397e-01
2.64076382e-01 -1.77375644e-01 -1.03835511e+00 1.39146829e+00
-6.05135500e-01 7.47750461e-01 2.42665410e-02 -4.12022769e-01
5.92832625e-01 1.82942495e-01 3.90413821e-01 -1.09237218e+00
8.54842141e-02 1.69861570e-01 -2.94301689e-01 -3.99128497e-01
5.01482904e-01 1.63968384e-01 6.85909986e-02 4.97563109e-02
8.04919153e-02 -1.93112195e-01 2.12162748e-01 3.82472984e-02
5.97596347e-01 -2.50773847e-01 2.05038294e-01 -8.74211043e-02
6.06602073e-01 -1.34809494e-01 4.89934891e-01 3.60482693e-01
-5.51031709e-01 9.73984659e-01 3.15457821e-01 -1.23004362e-01
-1.16678584e+00 -1.23279142e+00 -2.41800204e-01 7.72247493e-01
1.83316872e-01 -2.73280501e-01 -1.05487299e+00 -5.28048158e-01
-6.65613636e-02 3.54318887e-01 -6.12897515e-01 5.44997938e-02
-1.90854128e-02 -5.19115210e-01 5.07589519e-01 2.38399908e-01
9.80783284e-01 -8.00265074e-01 -2.81595141e-02 -3.55572253e-01
-3.94478709e-01 -1.43054473e+00 -6.46244466e-01 -8.82497549e-01
-4.62911636e-01 -1.41974998e+00 -9.81184006e-01 -2.63194174e-01
5.92964768e-01 3.19803059e-01 1.25413895e+00 -6.25354145e-03
-2.29240343e-01 7.77578115e-01 -3.20248485e-01 -3.20736676e-01
-3.76928002e-01 -1.46326542e-01 9.02240276e-02 4.32843506e-01
3.12796891e-01 -3.51370066e-01 -1.01835799e+00 8.28289986e-01
-7.87984192e-01 -1.85985222e-01 6.34594679e-01 4.39292967e-01
5.46156228e-01 1.03046969e-01 2.94419885e-01 -4.83128339e-01
5.67728221e-01 -1.99632302e-01 -6.69327438e-01 4.53447729e-01
-7.41924345e-01 -7.21434712e-01 2.42874220e-01 -3.43928963e-01
-1.05998361e+00 -1.67380288e-01 -7.89318010e-02 -2.78343350e-01
-2.53153890e-01 -6.94400743e-02 -7.59064078e-01 -3.28432083e-01
5.36878109e-01 -5.38099632e-02 3.23077627e-02 -1.17048495e-01
3.60175401e-01 9.05090988e-01 5.78612626e-01 -1.75306618e-01
6.98151290e-01 5.86881280e-01 -1.58093199e-01 -9.23388720e-01
-5.43954849e-01 -3.44591528e-01 -3.93024862e-01 -7.04124033e-01
7.37138212e-01 -1.17286336e+00 -9.07127798e-01 7.79310107e-01
-7.19187975e-01 -3.31540555e-01 -8.53041634e-02 1.68160245e-01
-3.87440383e-01 5.84385097e-01 -2.42692903e-01 -9.58106756e-01
-1.40194133e-01 -1.32387221e+00 1.31550324e+00 3.54759932e-01
-2.45037764e-01 -7.24778414e-01 -3.39008540e-01 1.17449033e+00
1.72907069e-01 4.98834252e-02 4.74564731e-01 -1.64917707e-02
-5.02552629e-01 -2.36421868e-01 -4.77299303e-01 6.68664217e-01
1.81444332e-01 1.60226658e-01 -1.29453003e+00 -3.14606488e-01
-3.74391109e-01 -3.08633715e-01 3.48641247e-01 5.21320283e-01
1.13933837e+00 -3.94189626e-01 5.83912432e-02 3.90424848e-01
1.33605111e+00 -6.67357743e-02 1.02323818e+00 3.61410379e-02
6.17026627e-01 7.96619117e-01 7.83594549e-01 4.35735762e-01
5.75442374e-01 9.55119252e-01 4.77012992e-01 -9.52184424e-02
-3.62299651e-01 -3.17829132e-01 6.38347268e-01 3.11451852e-01
-7.12362304e-02 -3.50476235e-01 -8.35821748e-01 6.49794340e-01
-1.15543962e+00 -8.13277423e-01 -2.48683430e-02 2.28219295e+00
6.88120604e-01 -2.38511071e-01 5.01025200e-01 3.68547410e-01
7.62196481e-01 3.32122557e-02 -4.59927052e-01 -2.57247686e-01
-4.44877416e-01 -1.19188033e-01 6.62705958e-01 3.81844372e-01
-8.97662520e-01 6.55216694e-01 6.26244068e+00 1.14455831e+00
-8.23852539e-01 2.59694099e-01 1.23197639e+00 -4.01509441e-02
-2.95280874e-01 -1.86754093e-01 -5.43298066e-01 7.20275402e-01
9.89012361e-01 1.76659524e-01 6.32816315e-01 6.53787494e-01
4.36561137e-01 -4.63716418e-01 -8.07661533e-01 1.51665330e+00
3.49547595e-01 -1.10945940e+00 -2.41762847e-02 1.97076038e-01
9.07020688e-01 -3.25302184e-01 6.37514234e-01 -2.35318288e-01
-5.99990524e-02 -1.13761771e+00 8.08574617e-01 5.04176795e-01
1.23124611e+00 -4.75023001e-01 6.90054417e-01 -2.93052793e-01
-9.46017325e-01 -3.55090618e-01 -2.78384060e-01 1.94567904e-01
1.76223228e-03 6.14813805e-01 -5.10766804e-01 4.83357757e-01
1.04730630e+00 6.53239906e-01 -1.16191113e+00 1.02762890e+00
-6.65235445e-02 5.78187227e-01 -1.01590892e-02 2.66256690e-01
-2.80754238e-01 -2.46637955e-01 2.39625514e-01 7.74946272e-01
5.00138223e-01 -1.66154280e-02 -3.30729365e-01 7.09551632e-01
-3.94535184e-01 3.14489335e-01 -3.71611685e-01 -6.69451803e-02
5.61936617e-01 1.28259182e+00 -5.53086579e-01 -1.38269946e-01
-6.00486159e-01 1.21982980e+00 -3.18981707e-01 3.49099487e-01
-8.46876204e-01 9.51803029e-02 8.42402399e-01 4.29806948e-01
7.66955987e-02 9.50678438e-02 -3.96038145e-01 -9.06627893e-01
2.62428552e-01 -1.25560629e+00 2.21291453e-01 -1.23032701e+00
-1.58950067e+00 4.44761217e-01 5.02193021e-03 -1.33866751e+00
1.29706547e-01 -7.01484025e-01 -3.23863238e-01 8.39845240e-01
-1.49425483e+00 -1.45151174e+00 -7.65268326e-01 7.39266515e-01
5.08247793e-01 -1.84161037e-01 4.48886603e-01 4.96484399e-01
-5.02151847e-01 9.50962663e-01 -8.38118047e-02 9.76209268e-02
9.29539084e-01 -1.00367439e+00 2.13630021e-01 6.81976438e-01
1.62252933e-01 2.61366189e-01 4.97012019e-01 -4.95627582e-01
-1.10194516e+00 -1.03968430e+00 6.45498574e-01 -7.60084093e-01
2.80768275e-01 -3.74211758e-01 -5.43985069e-01 1.42020807e-01
2.81763911e-01 5.11190780e-02 8.00749898e-01 -1.18354052e-01
-4.54648346e-01 -6.56228483e-01 -1.41140699e+00 5.44242024e-01
9.86281455e-01 -8.35381269e-01 8.64672288e-02 1.95293203e-01
2.16312051e-01 -3.02617718e-02 -1.09044158e+00 2.98558444e-01
8.63679349e-01 -1.31667936e+00 1.04300165e+00 1.38597429e-01
2.60531276e-01 -3.77015561e-01 -2.92446584e-01 -9.70599890e-01
6.16157502e-02 -5.42086482e-01 2.81185180e-01 1.57975447e+00
3.45698982e-01 -3.54919046e-01 6.62182450e-01 8.68253827e-01
2.74397999e-01 -4.29610044e-01 -9.51073349e-01 -7.36708224e-01
-1.12077206e-01 -6.23672366e-01 8.13491881e-01 6.83424950e-01
-4.66094166e-01 -4.22570072e-02 -5.14231384e-01 2.98873901e-01
6.65092170e-01 -2.23866507e-01 8.78577769e-01 -1.03255177e+00
7.75969476e-02 -4.15604323e-01 -6.65730417e-01 -4.73165214e-01
9.43296496e-03 -4.09441650e-01 -3.75014931e-01 -1.33714890e+00
3.70404720e-01 -2.43023753e-01 3.27654071e-02 2.16061935e-01
-1.03361428e-01 9.38665390e-01 1.97369188e-01 3.81325036e-01
-6.77873552e-01 3.79247695e-01 1.12505829e+00 -3.34627658e-01
1.58473358e-01 -7.45242685e-02 -5.82311392e-01 5.65135062e-01
8.13096523e-01 -7.20350668e-02 -5.10016739e-01 -3.00575316e-01
4.33367938e-01 -1.23384058e-01 8.16924036e-01 -1.23203993e+00
-2.60411408e-02 -8.02155510e-02 5.91640770e-01 -1.14532448e-01
5.06592751e-01 -9.91299808e-01 4.00690854e-01 7.25320652e-02
-1.33859962e-01 1.81665733e-01 1.47272095e-01 5.14158010e-01
-2.68499196e-01 7.25785792e-02 8.73623610e-01 8.00401792e-02
-7.74046361e-01 4.61386085e-01 -9.38065425e-02 6.99561313e-02
8.37033153e-01 -6.71061695e-01 -2.65145719e-01 -7.04549789e-01
-4.17980134e-01 -3.62882130e-02 1.01428103e+00 3.79975557e-01
6.46126091e-01 -1.39592588e+00 -5.86818099e-01 -1.75973065e-02
4.77409244e-01 -6.74243808e-01 6.78788304e-01 8.93300653e-01
-3.71588707e-01 7.74986818e-02 -4.33885127e-01 -5.64414024e-01
-1.48828888e+00 4.79531467e-01 3.51501256e-01 1.37569726e-01
1.12369150e-01 5.85673213e-01 -8.41932297e-02 -1.28355369e-01
8.45705792e-02 -7.10314214e-02 -1.76468134e-01 2.13241532e-01
6.18671596e-01 6.45406067e-01 1.73520669e-01 -1.12784076e+00
-3.38545173e-01 6.78314984e-01 3.24145019e-01 -1.88647717e-01
7.81564534e-01 -5.72019815e-01 3.43383402e-02 1.56952113e-01
1.17350864e+00 2.01823547e-01 -1.26459622e+00 1.63245589e-01
-2.86330611e-01 -9.56651032e-01 -1.58740491e-01 -1.17464697e+00
-1.14856601e+00 8.10581565e-01 1.21611083e+00 3.89730185e-02
1.46428895e+00 -9.86584798e-02 5.94215155e-01 -1.62874132e-01
3.48075330e-01 -1.10826671e+00 2.32744962e-01 -3.24657321e-01
9.12578821e-01 -1.68577743e+00 5.16832694e-02 -5.74065626e-01
-9.22272384e-01 5.81431329e-01 4.03491229e-01 3.24193299e-01
3.83429378e-01 -2.66398996e-01 1.70611709e-01 1.57083981e-02
-1.46920189e-01 -2.39808008e-01 5.82939029e-01 1.02617073e+00
1.50813252e-01 7.33619630e-02 -3.51888034e-03 4.80790585e-01
-2.47114852e-01 3.57743390e-02 3.17344040e-01 2.51950175e-01
9.91437584e-02 -1.12787187e+00 -4.77343917e-01 4.03691411e-01
-5.43705583e-01 3.17017734e-02 -5.80975831e-01 7.26081967e-01
4.54416811e-01 1.48133278e+00 -1.06882080e-01 -3.36506367e-01
3.33451658e-01 -2.05360755e-01 4.99593198e-01 -1.08804472e-01
-4.00231332e-01 -3.77101421e-01 1.79900751e-01 -8.67063284e-01
-8.04571211e-01 -7.58101761e-01 -2.95652956e-01 -6.03795230e-01
-9.65635404e-02 -2.12495714e-01 6.65263295e-01 5.96657217e-01
4.14372116e-01 -6.28171414e-02 6.42394602e-01 -6.06643260e-01
8.76146257e-02 -8.05092990e-01 -5.25843740e-01 7.77711153e-01
1.03645369e-01 -6.67493343e-01 -3.51831377e-01 2.80146062e-01] | [11.8433198928833, -1.8169310092926025] |
5ded4fec-fab9-4d37-9416-32466fbb81f1 | homography-estimation-with-convolutional | 2010.01041 | null | https://arxiv.org/abs/2010.01041v2 | https://arxiv.org/pdf/2010.01041v2.pdf | Homography Estimation with Convolutional Neural Networks Under Conditions of Variance | Planar homography estimation is foundational to many computer vision problems, such as Simultaneous Localization and Mapping (SLAM) and Augmented Reality (AR). However, conditions of high variance confound even the state-of-the-art algorithms. In this report, we analyze the performance of two recently published methods using Convolutional Neural Networks (CNNs) that are meant to replace the more traditional feature-matching based approaches to the estimation of homography. Our evaluation of the CNN based methods focuses particularly on measuring the performance under conditions of significant noise, illumination shift, and occlusion. We also measure the benefits of training CNNs to varying degrees of noise. Additionally, we compare the effect of using color images instead of grayscale images for inputs to CNNs. Finally, we compare the results against baseline feature-matching based homography estimation methods using SIFT, SURF, and ORB. We find that CNNs can be trained to be more robust against noise, but at a small cost to accuracy in the noiseless case. Additionally, CNNs perform significantly better in conditions of extreme variance than their feature-matching based counterparts. With regard to color inputs, we conclude that with no change in the CNN architecture to take advantage of the additional information in the color planes, the difference in performance using color inputs or grayscale inputs is negligible. About the CNNs trained with noise-corrupted inputs, we show that training a CNN to a specific magnitude of noise leads to a "Goldilocks Zone" with regard to the noise levels where that CNN performs best. | ['Avinash Kak', 'David Niblick'] | 2020-10-02 | null | null | null | null | ['homography-estimation'] | ['computer-vision'] | [ 1.27068996e-01 -3.02631110e-01 1.91591024e-01 -2.96297610e-01
-6.31314576e-01 -5.67214072e-01 5.73441803e-01 -2.25638971e-01
-5.44617712e-01 4.68423158e-01 -4.97042574e-03 -1.89942077e-01
-4.28651050e-02 -8.44237566e-01 -1.03913438e+00 -4.36677933e-01
-9.86295100e-03 1.67983934e-01 1.21951833e-01 -5.99933803e-01
2.87526727e-01 9.79887545e-01 -1.84090889e+00 -1.16493583e-01
3.56960207e-01 1.08912551e+00 -2.45213434e-01 6.31295264e-01
2.01917619e-01 2.63004482e-01 -5.48453033e-01 -2.07615271e-01
7.44534254e-01 -1.72030464e-01 -4.06243473e-01 -1.62448421e-01
1.27154183e+00 -4.13390607e-01 -6.88236535e-01 1.03249073e+00
4.56336677e-01 1.21523209e-01 1.60600588e-01 -1.10809469e+00
-6.87494636e-01 -6.35458380e-02 -4.08106267e-01 -9.56302360e-02
6.33527398e-01 2.00388625e-01 7.12526321e-01 -9.34197783e-01
9.28510010e-01 1.18109107e+00 1.08978295e+00 1.07117280e-01
-1.18497753e+00 -5.21100760e-01 -2.51571447e-01 -1.25608668e-01
-1.37307632e+00 -4.69059497e-01 7.52835691e-01 -2.34055653e-01
1.08506238e+00 6.94341511e-02 6.70123279e-01 8.67096424e-01
4.45616782e-01 2.84839422e-01 9.99457598e-01 -4.28560525e-01
5.93160316e-02 -1.63472459e-01 -2.29810014e-01 8.51361275e-01
5.34028113e-01 4.71491456e-01 -4.80796963e-01 -8.79544988e-02
1.11664021e+00 -6.97752088e-02 -3.62176687e-01 -7.59765804e-01
-1.46561003e+00 6.82866037e-01 8.11259329e-01 2.68263191e-01
-1.54598191e-01 6.21918738e-01 1.35228261e-01 4.24771607e-01
1.57802999e-01 7.44499743e-01 -3.84447813e-01 -4.14218288e-03
-9.74739730e-01 1.87989935e-01 6.91772938e-01 1.11122847e+00
1.21819448e+00 1.68795094e-01 3.29580843e-01 4.45821196e-01
4.16561216e-02 5.09263277e-01 1.75972804e-01 -1.16527295e+00
4.55767930e-01 3.07762921e-01 2.60422081e-01 -1.43883121e+00
-6.16785944e-01 -6.19957209e-01 -5.80620944e-01 5.65047562e-01
6.26884401e-01 -2.26388454e-01 -9.72563148e-01 1.65562117e+00
3.21496762e-02 3.75658534e-02 -8.31664912e-03 1.04505181e+00
4.74823743e-01 1.34270385e-01 -5.95463514e-01 3.49607199e-01
1.09893382e+00 -6.29944503e-01 -6.43670142e-01 -4.03346241e-01
5.78810096e-01 -1.05807972e+00 9.69684541e-01 3.65490377e-01
-7.66604900e-01 -5.04778862e-01 -1.60336542e+00 -3.70179683e-01
-8.08442652e-01 2.14704201e-01 7.76726305e-01 7.25699127e-01
-1.39635873e+00 9.33204234e-01 -7.17874229e-01 -7.08300591e-01
8.13488290e-02 5.98051012e-01 -7.85461366e-01 -2.37431414e-02
-1.03584576e+00 1.10759711e+00 3.61711271e-02 2.69298732e-01
-3.38470548e-01 -5.20176768e-01 -1.10143733e+00 6.29638061e-02
3.91846485e-02 -5.90602040e-01 9.87314701e-01 -9.14826810e-01
-1.36811078e+00 7.12549627e-01 -5.67314029e-02 -3.72504294e-01
7.78042138e-01 -3.30574721e-01 -1.76829144e-01 5.78183345e-02
-3.39301527e-02 8.07354867e-01 5.82641423e-01 -1.25279486e+00
-4.34776008e-01 -3.94264281e-01 1.79668441e-02 3.85198966e-02
2.52750758e-02 -2.51143038e-01 -5.01336157e-01 -2.74316072e-01
8.01982462e-01 -1.16873837e+00 -1.54530674e-01 3.37952048e-01
-1.95319742e-01 5.71234047e-01 8.52564633e-01 -3.49590838e-01
3.57363999e-01 -2.25365019e+00 -2.35198677e-01 3.84476215e-01
5.24063595e-02 -1.14833437e-01 -1.00639828e-01 4.76969808e-01
-9.09235403e-02 -7.22295344e-02 1.84104457e-01 -3.72462243e-01
-2.30001230e-02 1.49419338e-01 -1.22567676e-01 9.90842938e-01
5.35729751e-02 8.64666820e-01 -7.99070716e-01 -1.25019088e-01
4.93757188e-01 7.95742869e-01 -3.79029989e-01 -2.17169657e-01
2.49095500e-01 3.13516766e-01 3.42691615e-02 7.49757648e-01
7.58087754e-01 -3.08504067e-02 1.41934484e-01 -4.38508004e-01
-3.76988262e-01 7.23876059e-02 -1.39139402e+00 1.82794785e+00
-6.80249989e-01 1.19471800e+00 1.76278815e-01 -4.70465988e-01
1.11517131e+00 2.10351814e-02 3.56756240e-01 -9.36767459e-01
3.68205667e-01 5.50679862e-01 3.21439803e-02 6.82530925e-02
9.90482867e-01 1.59247950e-01 1.13866329e-01 5.20171747e-02
1.51011601e-01 -3.14486265e-01 -2.05837741e-01 -9.97712556e-03
1.06251442e+00 1.42138183e-01 1.01426020e-01 -2.96127796e-01
1.19989090e-01 7.53098652e-02 2.62463391e-01 9.56592560e-01
-1.86353713e-01 9.24980581e-01 3.63298267e-01 -7.24781871e-01
-1.22269154e+00 -7.67469406e-01 -2.23417416e-01 5.51870525e-01
4.20623690e-01 -1.05271429e-01 -3.50827366e-01 -7.84655735e-02
2.82988966e-01 1.09155886e-01 -6.97264910e-01 -6.02331944e-02
-6.08927488e-01 -3.82184207e-01 6.54506266e-01 5.42470813e-01
7.60601103e-01 -7.04711676e-01 -8.98427844e-01 2.77210120e-02
1.46794796e-01 -1.35703242e+00 -1.10847078e-01 3.47828090e-01
-9.18269813e-01 -1.03385174e+00 -7.44285643e-01 -5.45410275e-01
6.25634968e-01 4.67320830e-01 8.61917734e-01 1.16225854e-01
-1.91803053e-01 5.17704427e-01 -2.22296640e-01 -2.18793660e-01
1.47937402e-01 1.24887206e-01 6.51275516e-02 -2.34943047e-01
1.69041201e-01 -7.16179550e-01 -6.81480050e-01 4.00012314e-01
-6.93057299e-01 -1.83542222e-01 5.43327332e-01 7.83167958e-01
3.93060833e-01 -3.71237695e-01 -1.91997036e-01 -6.89960420e-01
3.85911793e-01 -4.21889387e-02 -1.00364041e+00 -1.41710006e-02
-5.74402153e-01 1.25593722e-01 5.44731498e-01 -3.88756931e-01
-6.17672265e-01 5.36293864e-01 -2.82826647e-03 -6.49888575e-01
-2.56207913e-01 3.13118100e-01 1.07200645e-01 -1.03681445e+00
8.86229873e-01 -1.07145645e-01 7.87796900e-02 -3.53627115e-01
3.75555575e-01 2.37571463e-01 8.00473511e-01 -3.16888869e-01
1.00561976e+00 9.29441214e-01 4.13236052e-01 -6.94155395e-01
-3.09439987e-01 -4.04477239e-01 -7.65358984e-01 -9.57136676e-02
6.44565880e-01 -8.44164968e-01 -6.70966506e-01 4.79176641e-01
-1.24895978e+00 -2.15853691e-01 -1.28476331e-02 7.11288393e-01
-6.98586166e-01 2.83083767e-01 -5.11321127e-01 -5.47370493e-01
1.36720359e-01 -1.37180531e+00 1.15794635e+00 2.54137456e-01
7.43057132e-02 -8.45761776e-01 -2.07026917e-02 -7.77455121e-02
6.41159534e-01 5.59726059e-01 5.46938896e-01 -5.18447220e-01
-7.63964236e-01 -6.29346907e-01 -3.66482407e-01 6.07485697e-02
3.57439741e-02 -1.33188898e-02 -1.25101006e+00 -3.32985461e-01
-3.45801771e-01 -1.22736655e-01 7.90954888e-01 2.94042736e-01
6.05353713e-01 5.82782924e-02 -1.12079598e-01 1.18213379e+00
1.79147637e+00 1.64044827e-01 9.93048072e-01 8.42840195e-01
7.87085831e-01 5.17855823e-01 4.40085083e-01 8.83818120e-02
1.17620558e-01 8.21898401e-01 7.33050346e-01 -5.31520903e-01
-1.46306738e-01 -1.97949439e-01 -7.68018886e-02 2.09243506e-01
-1.52747827e-02 7.95379877e-02 -9.05494332e-01 4.81326282e-01
-1.63245738e+00 -6.11548603e-01 -1.52561724e-01 2.48892498e+00
1.48344696e-01 5.92335165e-02 -2.82620728e-01 5.87469228e-02
5.10662436e-01 2.26395980e-01 -2.36981869e-01 -3.61107409e-01
-3.99853170e-01 1.50212884e-01 1.13366222e+00 5.80571055e-01
-1.15868640e+00 9.43533599e-01 6.95925140e+00 1.11331031e-01
-1.60915375e+00 -2.11688921e-01 2.88529433e-02 6.41666129e-02
1.04613779e-02 2.56244779e-01 -5.70552111e-01 9.84028354e-03
5.94237566e-01 4.13783163e-01 5.84946930e-01 9.60067630e-01
-1.27437770e-01 -2.56064594e-01 -1.04570198e+00 1.20966566e+00
2.08428264e-01 -1.35996056e+00 -2.39241719e-01 2.55692422e-01
8.62865448e-01 3.78887653e-01 2.80168027e-01 -6.06648959e-02
1.76633283e-01 -9.40557003e-01 7.11781383e-01 4.22980964e-01
7.39630997e-01 -5.23743689e-01 1.09132135e+00 -1.20348111e-01
-1.06708026e+00 5.20635433e-02 -4.94143426e-01 -9.57277864e-02
-1.60712097e-02 3.40657502e-01 -6.48045719e-01 6.14725828e-01
6.75183773e-01 4.49296385e-01 -6.87062442e-01 1.20012987e+00
2.19913963e-02 -1.14432953e-01 -5.49281776e-01 1.04799852e-01
3.75045389e-01 -6.07432015e-02 2.78799444e-01 1.10100508e+00
5.15641272e-01 -2.86282390e-01 -2.11661607e-02 7.68380463e-01
-2.79073929e-03 -4.82735187e-02 -1.08509290e+00 3.20856333e-01
4.26032960e-01 1.13390493e+00 -8.30407739e-01 -1.90250203e-01
-5.18474221e-01 9.12408948e-01 1.23844735e-01 5.67855000e-01
-4.49993461e-01 -8.58843982e-01 6.88064575e-01 3.15859839e-02
2.95844942e-01 -7.43486345e-01 -4.99599427e-01 -1.11539066e+00
6.41277060e-02 -6.86702788e-01 -7.89001361e-02 -9.79253948e-01
-7.98034370e-01 5.79918861e-01 -1.83248833e-01 -1.26441038e+00
-1.96928903e-01 -9.42122817e-01 -4.80770379e-01 9.93415177e-01
-1.62954950e+00 -1.00919628e+00 -7.99040258e-01 5.33703327e-01
3.18414979e-02 8.07796940e-02 6.37565374e-01 3.52264643e-01
-1.18273515e-02 5.25448143e-01 2.62978137e-01 3.84064913e-01
8.42127442e-01 -1.06651783e+00 6.23419166e-01 9.31761026e-01
2.98384607e-01 1.00087821e+00 9.91009712e-01 -5.57411730e-01
-1.79840446e+00 -7.57037759e-01 7.40402341e-01 -5.07926166e-01
5.20754099e-01 -5.03359377e-01 -5.79698980e-01 9.04574096e-01
-5.27301095e-02 4.01776880e-01 -7.18420546e-04 1.78580493e-01
-6.08678102e-01 -4.55097482e-02 -1.10915112e+00 4.59564507e-01
9.60596681e-01 -8.54586720e-01 -1.94564715e-01 2.30363831e-01
5.46165109e-01 -8.83582532e-01 -5.75843155e-01 4.06433791e-01
1.06207657e+00 -1.38917136e+00 1.04656947e+00 -2.82839924e-01
1.19820260e-01 -3.74590755e-01 -4.71635550e-01 -1.17165077e+00
-2.20713884e-01 -4.07314569e-01 4.75787938e-01 6.08002603e-01
2.77746767e-01 -9.50068235e-01 8.57890546e-01 3.13948005e-01
-1.60424337e-01 -3.72721434e-01 -9.96662974e-01 -1.06009758e+00
-4.34054360e-02 -3.26876730e-01 6.24610484e-01 9.53476667e-01
-4.02005702e-01 -1.78183675e-01 -3.57831150e-01 4.45874035e-01
4.57946390e-01 1.30821383e-02 1.14043653e+00 -1.20534277e+00
-4.10102047e-02 -3.02920520e-01 -9.91773248e-01 -8.23203623e-01
-8.85110795e-02 -4.26471442e-01 2.02416167e-01 -1.33374786e+00
-4.63041574e-01 -2.27525234e-01 -5.95117249e-02 4.07748252e-01
4.16308492e-01 6.57980144e-01 3.67857516e-01 2.19148010e-01
-2.12258443e-01 2.01868758e-01 1.01497149e+00 -3.30433585e-02
-2.04397395e-01 -1.78331852e-01 -2.08429307e-01 5.74642777e-01
7.27007389e-01 -2.85515994e-01 -1.01604477e-01 -6.72477603e-01
4.49666411e-01 -1.16571650e-01 7.40375221e-01 -1.51013458e+00
4.95687187e-01 2.26389095e-01 6.64143443e-01 -5.49009740e-01
6.12435699e-01 -8.86507273e-01 4.31717694e-01 5.14413297e-01
-5.07318787e-02 4.67423677e-01 3.35916072e-01 2.90072203e-01
-1.90578654e-01 6.43093511e-02 6.58678532e-01 -3.03300679e-01
-7.40007043e-01 1.26748189e-01 -1.48423761e-01 -3.48337531e-01
4.32989955e-01 -6.05635405e-01 -5.05580485e-01 -7.58909583e-01
-5.30248404e-01 -2.99904495e-01 9.56234276e-01 3.63279879e-01
4.95477945e-01 -1.27589178e+00 -1.97414935e-01 4.00439084e-01
9.81542021e-02 -4.06019837e-01 -1.90169923e-02 9.21600640e-01
-1.09351707e+00 5.94138682e-01 -5.89041352e-01 -5.86374760e-01
-9.96899188e-01 4.97427523e-01 5.86133718e-01 2.58293062e-01
-4.89305645e-01 4.45424736e-01 -1.21323422e-01 -5.20172238e-01
4.12020981e-01 -4.72340643e-01 2.11549625e-01 -2.02148601e-01
1.42694041e-01 3.31486940e-01 4.78897005e-01 -7.31393814e-01
-4.25927252e-01 9.19900119e-01 2.59834081e-01 -3.55042279e-01
1.06722367e+00 -8.23511218e-05 1.06050506e-01 1.88173622e-01
1.38470662e+00 1.30999431e-01 -1.12424529e+00 1.36188150e-03
-1.34287432e-01 -7.02567339e-01 1.38008520e-01 -5.50717115e-01
-1.19957554e+00 1.07998574e+00 8.46219480e-01 -3.23771648e-02
8.03273797e-01 -3.87995899e-01 4.44470078e-01 6.11543834e-01
6.86144412e-01 -8.79935384e-01 -1.40657350e-01 7.43879795e-01
6.76629126e-01 -1.21284032e+00 2.13296011e-01 -3.35174471e-01
-1.30047411e-01 1.34183562e+00 5.19291759e-01 -5.26755929e-01
5.52692831e-01 2.79549628e-01 4.62082505e-01 -3.48170638e-01
-1.14418522e-01 -3.90812814e-01 2.11500496e-01 7.84525514e-01
3.06648374e-01 -2.50240535e-01 -2.69670952e-02 -2.66798556e-01
-4.65455532e-01 -2.24947378e-01 4.90693867e-01 1.09220111e+00
-3.48164767e-01 -8.51599514e-01 -6.36855900e-01 7.14272261e-02
-1.73565984e-01 -1.18464097e-01 -6.17009759e-01 1.41128111e+00
1.72871619e-01 5.76823890e-01 2.13745579e-01 -6.46907270e-01
5.02149940e-01 -1.11903198e-01 5.13868034e-01 -1.02927379e-01
-6.86179399e-01 -2.10250989e-02 -1.39413523e-02 -1.00732911e+00
-2.93257952e-01 -4.33865458e-01 -9.39565063e-01 -4.55628186e-01
-2.99958974e-01 -2.70207077e-01 1.13940275e+00 7.58020997e-01
3.47089171e-01 2.62657017e-01 3.18946689e-01 -1.29096782e+00
-4.39374775e-01 -6.51194215e-01 -5.19716799e-01 2.50196785e-01
6.48113549e-01 -8.15646470e-01 -4.83737379e-01 -4.32239115e-01] | [7.772150993347168, -2.0093822479248047] |
03e046d4-dd28-4367-8169-82900456af24 | assessing-the-effectiveness-of-gpt-3-in | 2306.08190 | null | https://arxiv.org/abs/2306.08190v1 | https://arxiv.org/pdf/2306.08190v1.pdf | Assessing the Effectiveness of GPT-3 in Detecting False Political Statements: A Case Study on the LIAR Dataset | The detection of political fake statements is crucial for maintaining information integrity and preventing the spread of misinformation in society. Historically, state-of-the-art machine learning models employed various methods for detecting deceptive statements. These methods include the use of metadata (W. Wang et al., 2018), n-grams analysis (Singh et al., 2021), and linguistic (Wu et al., 2022) and stylometric (Islam et al., 2020) features. Recent advancements in large language models, such as GPT-3 (Brown et al., 2020) have achieved state-of-the-art performance on a wide range of tasks. In this study, we conducted experiments with GPT-3 on the LIAR dataset (W. Wang et al., 2018) and achieved higher accuracy than state-of-the-art models without using any additional meta or linguistic features. Additionally, we experimented with zero-shot learning using a carefully designed prompt and achieved near state-of-the-art performance. An advantage of this approach is that the model provided evidence for its decision, which adds transparency to the model's decision-making and offers a chance for users to verify the validity of the evidence provided. | ['Mars Gokturk Buchholz'] | 2023-06-14 | null | null | null | null | ['misinformation'] | ['miscellaneous'] | [-3.22092354e-01 2.86239505e-01 -6.62009776e-01 -1.87526330e-01
-1.02819729e+00 -5.87099493e-01 1.20184815e+00 6.90740407e-01
-5.22839725e-01 7.51489878e-01 6.12891197e-01 -6.80265248e-01
2.29015082e-01 -8.25150013e-01 -5.38529277e-01 -2.12071855e-02
1.80569172e-01 1.29916683e-01 3.24818671e-01 -4.72913414e-01
9.16703701e-01 1.34632681e-02 -1.13717651e+00 5.52133918e-01
9.58197534e-01 8.01316023e-01 -7.58148193e-01 3.65656883e-01
-2.12106004e-01 1.27984333e+00 -6.58657014e-01 -1.27726281e+00
6.43901005e-02 -1.08009808e-01 -7.25815952e-01 -4.49387878e-01
6.01848781e-01 -5.22432327e-01 -4.82059538e-01 1.22157836e+00
3.84993106e-01 -3.76262605e-01 6.83954000e-01 -1.05821276e+00
-1.25230527e+00 7.14265227e-01 -6.35674715e-01 3.45349759e-01
5.05100369e-01 2.47521654e-01 1.01180661e+00 -9.12264287e-01
6.98020101e-01 1.21530950e+00 7.18880296e-01 3.64577502e-01
-9.00185704e-01 -1.09255588e+00 -2.72263408e-01 5.84873021e-01
-1.16269982e+00 -6.61638379e-01 6.57286465e-01 -7.16153622e-01
7.13911951e-01 1.48230776e-01 6.07380450e-01 1.50397873e+00
3.66190761e-01 7.17739761e-01 1.76216018e+00 -6.51154935e-01
3.35139424e-01 2.13044211e-01 4.19632941e-01 7.46450603e-01
5.74920654e-01 2.53177416e-02 -9.27814603e-01 -6.83352470e-01
1.77291796e-01 -3.51426631e-01 1.71069186e-02 2.21512780e-01
-7.08659589e-01 1.36278105e+00 2.91762829e-01 4.20339733e-01
-2.65938580e-01 -1.00027151e-01 5.41958272e-01 1.76133394e-01
1.04459834e+00 4.73712534e-01 -1.03052460e-01 -5.42120993e-01
-1.22121191e+00 3.11097652e-01 9.39913034e-01 3.72873664e-01
3.26364607e-01 -3.36333930e-01 -3.10266882e-01 6.33244455e-01
4.60653454e-01 3.26807767e-01 7.26859212e-01 -9.28310096e-01
6.10655248e-01 4.25673097e-01 4.19113785e-01 -1.59450543e+00
-1.54722825e-01 -2.80630410e-01 -5.39162040e-01 -6.20314591e-02
5.44699848e-01 7.39969090e-02 -8.23312938e-01 1.34509397e+00
6.39297888e-02 1.12142228e-01 -7.08419923e-03 8.25534761e-01
8.28662217e-01 6.08655214e-01 -3.35077010e-02 -6.61502257e-02
1.32381952e+00 -6.16612732e-01 -8.72838080e-01 -3.27697873e-01
9.00381684e-01 -8.31867158e-01 9.22102094e-01 4.34647888e-01
-6.75410628e-01 4.34685946e-02 -1.06619680e+00 -1.70341820e-01
-4.02373970e-01 -4.20017779e-01 6.85002267e-01 9.75591838e-01
-6.95723951e-01 5.51973879e-01 -4.80902463e-01 -4.70667541e-01
8.72893572e-01 -4.76423085e-01 -2.70517528e-01 -1.46027789e-01
-1.72632110e+00 1.33569217e+00 -7.21805245e-02 -3.50746036e-01
-6.31473601e-01 -4.85059261e-01 -7.58888006e-01 -1.10771842e-01
3.14299136e-01 -1.23339415e-01 1.19572008e+00 -7.03328609e-01
-1.38072252e+00 1.32718480e+00 -1.16288595e-01 -8.41387212e-01
7.99727738e-01 -4.20267195e-01 -6.47320569e-01 6.70540035e-02
4.67370480e-01 6.15021996e-02 8.53927016e-01 -8.30836713e-01
-3.39964598e-01 -4.22901839e-01 -1.15536764e-01 -3.32519591e-01
-5.70140064e-01 5.63574493e-01 2.45636493e-01 -5.77679098e-01
-1.57388240e-01 -7.68836141e-01 3.28206331e-01 -5.33774793e-02
-3.83574605e-01 -2.99252987e-01 6.75233662e-01 -1.19218636e+00
1.49398124e+00 -1.93813908e+00 -5.99522471e-01 -3.64277847e-02
4.34397697e-01 6.64470136e-01 2.71536142e-01 7.06942141e-01
3.36041480e-01 7.94409394e-01 -4.70955670e-02 -3.10845256e-01
1.83385424e-02 -2.99134642e-01 -3.88828099e-01 9.58414078e-01
2.09238892e-03 1.04358459e+00 -8.02406967e-01 -4.55148309e-01
2.80749500e-02 1.23082325e-01 -3.23097110e-01 -2.84969032e-01
3.78442369e-02 8.78217220e-02 -1.90795168e-01 6.37337089e-01
3.90670985e-01 -2.51504600e-01 4.64554876e-02 1.12449653e-01
-3.59349579e-01 7.71432936e-01 -5.16597331e-01 1.00667071e+00
-1.90420210e-01 1.10506594e+00 -1.33744463e-01 -7.26609588e-01
8.62717748e-01 2.55460232e-01 -2.39157528e-02 -7.19159126e-01
2.72450596e-01 3.62814963e-01 -1.57755792e-01 -6.12863541e-01
4.92843747e-01 -1.84110463e-01 -2.99607784e-01 7.59928107e-01
-3.58380049e-01 -6.84470683e-02 -1.13987245e-01 4.42762315e-01
9.90965843e-01 -2.75875032e-01 6.23313785e-01 -2.98247904e-01
3.14691991e-01 7.07601681e-02 3.70599568e-01 9.89660621e-01
-5.94278097e-01 2.38323212e-01 6.26013994e-01 -5.38645566e-01
-1.15922701e+00 -6.14933729e-01 -2.10974947e-01 9.36144352e-01
-1.96741462e-01 -5.84988177e-01 -6.02947772e-01 -7.22077489e-01
3.14763099e-01 1.43930280e+00 -8.86600912e-01 -1.12445638e-01
-3.17902952e-01 -5.78577459e-01 9.48329329e-01 -1.57458052e-01
7.08105505e-01 -5.33815920e-01 -8.13093603e-01 9.98009890e-02
-4.77919161e-01 -1.02132213e+00 -1.54897839e-01 -5.52965105e-01
-4.36856806e-01 -1.02809095e+00 -4.22471285e-01 -1.04928695e-01
3.42089623e-01 9.92559120e-02 6.16177499e-01 7.67620876e-02
6.50768876e-02 -5.61195090e-02 -4.32778567e-01 -5.82649112e-01
-7.40742862e-01 -3.36564898e-01 3.48020270e-02 -3.55785433e-03
4.96084005e-01 -1.72986820e-01 -1.79901540e-01 -2.64093149e-02
-6.63181603e-01 -7.72995874e-02 3.02306592e-01 8.25853765e-01
-1.89851761e-01 -2.19919756e-01 6.41504943e-01 -9.69231606e-01
9.84878778e-01 -5.11631072e-01 -4.79471475e-01 7.72710592e-02
-9.09997106e-01 -1.99277043e-01 3.67575645e-01 -2.40749836e-01
-9.36879933e-01 -1.14338648e+00 9.23278369e-03 -2.39030216e-02
2.04322815e-01 8.02548587e-01 4.05948222e-01 -2.37813458e-01
8.48384023e-01 -2.33391696e-03 -5.67084067e-02 -4.10425603e-01
4.59436774e-01 1.26231790e+00 1.60545751e-01 -1.76416487e-01
6.89617217e-01 4.19282645e-01 -3.81353587e-01 -6.38230979e-01
-1.36940026e+00 -3.64907116e-01 -4.39690828e-01 -2.63814539e-01
5.10480881e-01 -9.07642245e-01 -3.88795465e-01 7.13238895e-01
-1.16762435e+00 -1.58282027e-01 3.57618898e-01 4.40257698e-01
-2.23545730e-01 5.97037077e-01 -9.47988570e-01 -1.15115356e+00
-2.80031532e-01 -5.69390118e-01 5.26585639e-01 -1.08618326e-01
-6.20361567e-01 -8.74512553e-01 6.11141883e-02 1.03615487e+00
5.37654459e-01 3.91005576e-01 7.74345756e-01 -1.05316997e+00
-2.02132985e-02 -5.51014543e-01 -4.24531192e-01 1.30748004e-01
-6.14314573e-03 1.08494759e-01 -1.03247750e+00 -2.95440614e-01
2.21459195e-01 -6.95827067e-01 9.71382976e-01 1.89061780e-02
8.95608187e-01 -1.09598064e+00 -1.60160854e-01 -3.31154838e-02
1.04590058e+00 -2.59988189e-01 5.44847906e-01 8.44113529e-01
3.34441245e-01 6.13867402e-01 6.33180022e-01 3.76137674e-01
5.78628421e-01 5.49638987e-01 1.47373348e-01 4.50249135e-01
1.24454789e-01 -5.46711743e-01 3.76418650e-01 5.66348970e-01
2.05898836e-01 -1.36484146e-01 -1.23623049e+00 5.13293207e-01
-2.01286674e+00 -1.30663574e+00 -1.27615213e-01 2.22423840e+00
8.82772565e-01 5.65776110e-01 -5.06415479e-02 -2.50691827e-03
8.31518888e-01 4.53983903e-01 -2.05357313e-01 -7.74633765e-01
-2.12391540e-01 -2.79192775e-01 6.39306664e-01 5.60921431e-01
-1.02467787e+00 1.02174509e+00 6.22031832e+00 8.79505932e-01
-1.18096840e+00 4.44973975e-01 8.50579262e-01 -9.55525637e-02
-4.39803243e-01 4.31538001e-02 -5.34189403e-01 9.19896305e-01
1.12458205e+00 -5.40114999e-01 2.47629777e-01 6.26984835e-01
2.83515036e-01 -3.43854398e-01 -4.25516814e-01 8.20604742e-01
5.98636985e-01 -1.71462119e+00 -2.39738122e-01 2.78481752e-01
8.26518238e-01 3.28994125e-01 -9.06453840e-03 2.99992323e-01
4.98960525e-01 -1.15033042e+00 1.13616562e+00 4.02631253e-01
6.50733352e-01 -4.69439238e-01 1.03749752e+00 6.84283555e-01
3.99571806e-02 -2.88735658e-01 -1.57487318e-01 -6.32175565e-01
3.94130707e-01 1.00146866e+00 -6.91510260e-01 3.62401396e-01
6.83834851e-01 6.39499724e-01 -7.27512360e-01 6.29326761e-01
-6.84727132e-01 1.16568089e+00 -4.81050797e-02 -5.20460069e-01
4.06608045e-01 9.91636887e-02 5.63762844e-01 1.24004173e+00
1.60443604e-01 6.76208064e-02 -1.32596180e-01 6.00993931e-01
-4.45106149e-01 2.16211140e-01 -6.56026840e-01 -5.00390530e-01
5.92643499e-01 9.07668352e-01 -3.12466025e-01 -5.54619253e-01
-4.98113066e-01 6.21538639e-01 6.54823124e-01 2.72032395e-02
-5.50288200e-01 -4.17842194e-02 2.82803535e-01 3.87191772e-01
8.62917900e-02 -2.04928562e-01 -4.95227277e-01 -1.24483740e+00
4.83373292e-02 -1.08439815e+00 3.91558111e-01 -6.57534301e-01
-1.61161304e+00 -3.42101753e-02 -9.24564824e-02 -7.99680650e-01
-2.99998801e-02 -4.95965213e-01 -4.35623646e-01 7.57242739e-01
-1.36816311e+00 -1.23096085e+00 1.50657922e-01 -5.48756942e-02
1.45411298e-01 -2.86749035e-01 8.13516498e-01 -2.14296263e-02
-3.70920420e-01 3.75736088e-01 2.75273293e-01 3.89655739e-01
1.15096104e+00 -7.50145793e-01 5.63687384e-01 8.35409164e-01
4.98078414e-04 8.02039206e-01 8.32799137e-01 -1.11180651e+00
-1.04942584e+00 -7.12880194e-01 1.45285809e+00 -7.48138189e-01
1.44472134e+00 -3.59393835e-01 -9.02404904e-01 6.85334206e-01
3.00902039e-01 -2.02583253e-01 1.01600647e+00 2.79701710e-01
-8.62949073e-01 3.22247505e-01 -1.49734449e+00 4.55890566e-01
5.56454599e-01 -8.39051843e-01 -9.12529290e-01 4.16842461e-01
4.69701022e-01 -2.58529961e-01 -6.32388234e-01 -8.11945647e-03
8.25197101e-01 -1.08795214e+00 3.81015360e-01 -9.85198021e-01
8.43278170e-01 1.47922397e-01 -1.08775996e-01 -1.26313972e+00
-4.26292121e-01 -5.30854106e-01 -3.85621458e-01 1.27611160e+00
5.12019515e-01 -9.37589467e-01 3.42576593e-01 7.56738901e-01
2.99604416e-01 -6.50895000e-01 -1.18535376e+00 -7.31282055e-01
4.44973648e-01 -4.19951737e-01 2.04972893e-01 1.42713773e+00
4.74970132e-01 2.09484830e-01 -6.98458791e-01 -1.18869066e-01
7.54456580e-01 -5.44018596e-02 6.24629974e-01 -1.16972399e+00
6.98704347e-02 -4.45746690e-01 -4.96183723e-01 -3.85684282e-01
3.42576534e-01 -7.38429487e-01 -4.42443728e-01 -1.57208204e+00
5.97012758e-01 -8.19980800e-02 -9.35855433e-02 3.80009383e-01
-3.41995537e-01 3.20089847e-01 3.29925179e-01 6.36046886e-01
-4.40344989e-01 3.23031962e-01 7.89394438e-01 -2.26839915e-01
3.28763008e-01 -2.94847637e-01 -1.06074607e+00 7.98361897e-01
1.08539903e+00 -6.32341862e-01 2.80295551e-01 -3.04336429e-01
4.42518622e-01 -4.37701881e-01 6.56980634e-01 -4.42995161e-01
3.56530696e-01 -2.09386393e-01 2.58557022e-01 -1.99134633e-01
2.18412638e-01 -1.39587432e-01 -3.41343969e-01 7.60900915e-01
-5.09200931e-01 -9.83401909e-02 1.23420835e-01 5.86097240e-01
-2.35810559e-02 -3.58843237e-01 7.74203181e-01 -2.33434200e-01
-3.73643011e-01 -2.21579924e-01 -5.03740609e-01 2.22616524e-01
8.53191435e-01 1.35047376e-01 -1.09756732e+00 -7.98097134e-01
-7.11670443e-02 -4.08761390e-03 7.12375343e-01 5.40235758e-01
3.66961718e-01 -1.14794075e+00 -1.15475845e+00 -1.60430312e-01
3.26432228e-01 -8.53553712e-01 -1.82814643e-01 1.02814138e+00
-4.13384527e-01 6.59029901e-01 -5.07864654e-02 -3.28471549e-02
-1.05608475e+00 5.29253066e-01 -2.09008262e-01 -2.23618567e-01
-4.20424551e-01 5.10396242e-01 -5.99495709e-01 -2.48987556e-01
-2.02007055e-01 3.34597021e-01 -1.31670654e-01 3.63241166e-01
7.63236880e-01 5.98042607e-01 -1.14496298e-01 -8.57389271e-01
-5.19571900e-01 8.08307230e-02 -2.62771815e-01 -3.69440883e-01
1.29143298e+00 -1.13842025e-01 -2.92713821e-01 8.04260492e-01
9.81555641e-01 5.21198332e-01 -4.61416632e-01 -3.75506580e-01
2.15683728e-01 -9.27519441e-01 3.30312282e-01 -1.15258992e+00
-3.75171423e-01 6.75279260e-01 1.26266360e-01 5.38202047e-01
1.54802874e-01 -6.72334731e-02 7.00795650e-01 1.55072108e-01
6.47699356e-01 -1.33620679e+00 7.77426735e-02 4.53502595e-01
7.48483300e-01 -1.60703874e+00 2.56202936e-01 -2.78426617e-01
-6.20899916e-01 1.04796946e+00 1.50858313e-01 2.43116468e-01
5.00873387e-01 -1.26362175e-01 2.72611469e-01 -2.27956459e-01
-6.14303231e-01 3.48327518e-01 2.27631956e-01 1.76018000e-01
6.49317205e-01 3.64028335e-01 -9.65193510e-01 5.02233505e-01
-3.33391279e-01 1.81458190e-01 8.58612001e-01 8.48719001e-01
-7.27967680e-01 -6.73161089e-01 -5.07110178e-01 8.65536630e-01
-8.12941074e-01 -3.77069980e-01 -7.36346424e-01 5.78166842e-01
-1.89419538e-01 1.42292833e+00 -3.76895040e-01 -2.30174690e-01
-4.62727621e-02 2.42279634e-01 1.44574732e-01 -5.43853462e-01
-5.64487755e-01 -4.29171473e-01 8.53826463e-01 -5.87840199e-01
-1.99733749e-01 -7.63986647e-01 -6.84968829e-01 -7.60636210e-01
-3.85439247e-01 2.10931361e-01 7.13304341e-01 1.14440119e+00
5.03042817e-01 -1.49615929e-01 3.96844715e-01 -9.71882492e-02
-9.35249090e-01 -1.25732112e+00 -3.05757850e-01 4.86987263e-01
3.02960277e-01 -5.69395840e-01 -7.06754863e-01 -2.09433079e-01] | [8.267525672912598, 10.19655990600586] |
698adc4a-9880-4454-a1d2-f53fbb5bc03e | driving-digital-engineering-integration-and | 2206.10454 | null | https://arxiv.org/abs/2206.10454v1 | https://arxiv.org/pdf/2206.10454v1.pdf | Driving Digital Engineering Integration and Interoperability Through Semantic Integration of Models with Ontologies | Engineered solutions are becoming more complex and multi-disciplinary in nature. This evolution requires new techniques to enhance design and analysis tasks that incorporate data integration and interoperability across various engineering tool suites spanning multiple domains at different abstraction levels. Semantic Web Technologies (SWT) offer data integration and interoperability benefits as well as other opportunities to enhance reasoning across knowledge represented in multiple disparate models. This paper introduces the Digital Engineering Framework for Integration and Interoperability (DEFII) for incorporating SWT into engineering design and analysis tasks. The framework includes three notional interfaces for interacting with ontology-aligned data. It also introduces a novel Model Interface Specification Diagram (MISD) that provides a tool-agnostic model representation enabled by SWT that exposes data stored for use by external users through standards-based interfaces. Use of the framework results in a tool-agnostic authoritative source of truth spanning the entire project, system, or mission. | ['Zhongyuan Yu', 'Dinesh Verma', 'Benjamin Kruse', 'Steven Hespelt', 'John Dzielski', 'Mark Blackburn', 'Thomas Hagedorn', 'Daniel Dunbar'] | 2022-06-08 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [-2.11997434e-01 9.51945111e-02 2.51120955e-01 -3.41861069e-01
-1.04751222e-01 -8.00744712e-01 5.44073880e-01 2.73809433e-01
2.48988360e-01 1.16719224e-01 2.49277562e-01 -3.75473887e-01
-1.11494005e+00 -1.10136223e+00 -2.53721289e-02 3.21135223e-01
3.37281615e-01 5.07483006e-01 5.10249674e-01 -7.01665461e-01
3.31786215e-01 5.41140497e-01 -1.97664583e+00 5.78167915e-01
6.55895650e-01 9.81950462e-01 3.16934675e-01 3.00407499e-01
-7.62133181e-01 6.15506828e-01 -5.04237235e-01 -2.20075063e-02
2.79667765e-01 7.87633955e-02 -1.07376409e+00 -9.44699049e-02
1.78387910e-02 1.80115998e-01 5.02959490e-01 6.57428741e-01
2.95104563e-01 -4.47070152e-01 1.10053547e-01 -1.68315446e+00
-6.98174000e-01 -3.58783640e-02 1.76983833e-01 -4.40120727e-01
6.91408098e-01 5.32836802e-02 5.39215803e-01 -9.46240187e-01
9.32442009e-01 1.03398800e+00 6.69322789e-01 1.66793495e-01
-1.02433920e+00 -4.85231102e-01 -3.60110760e-01 2.12105587e-01
-1.25580895e+00 -2.85557240e-01 6.59103394e-01 -9.32748020e-01
1.38290489e+00 6.88101888e-01 9.02251124e-01 3.72678906e-01
1.42800272e-01 -3.77938092e-01 7.98143268e-01 -7.61771560e-01
3.38675052e-01 5.62607884e-01 5.50665319e-01 3.49814147e-01
9.55017924e-01 -2.14226097e-01 -7.06369460e-01 -2.23728344e-01
6.74687743e-01 6.57670721e-02 1.08011886e-01 -7.38811672e-01
-8.97255898e-01 4.73748008e-03 -3.47285181e-01 7.60995984e-01
-2.93112904e-01 -1.43505512e-02 3.42492938e-01 7.15030253e-01
-1.54688716e-01 8.60562503e-01 -7.22890973e-01 -4.28425968e-01
-1.91565290e-01 2.75107473e-01 1.08371270e+00 1.43576276e+00
9.35700536e-01 2.70957239e-02 4.57737535e-01 7.24421561e-01
8.99303734e-01 -7.05110505e-02 1.97931051e-01 -1.30057991e+00
-3.55378091e-02 1.39207363e+00 6.35566294e-01 -9.82111037e-01
-3.52116764e-01 -5.37581742e-01 3.39493901e-01 1.02408695e+00
-9.14722383e-02 2.36919686e-01 -4.97834235e-01 1.10771418e+00
6.20484114e-01 -8.41299653e-01 2.71250457e-01 3.75918299e-01
7.51242459e-01 3.84532288e-02 3.81511092e-01 5.81549287e-01
1.41266489e+00 -3.21277082e-01 -1.03580058e+00 -1.39298841e-01
6.42141223e-01 -8.45901012e-01 1.05518734e+00 4.22984481e-01
-1.06134796e+00 -5.01908839e-01 -1.55383337e+00 2.11275881e-03
-1.22319067e+00 -6.23180985e-01 3.61612529e-01 6.76968336e-01
-8.95937502e-01 3.70581537e-01 -4.74156559e-01 -9.83076513e-01
1.41425475e-01 4.45500389e-02 -3.35274458e-01 2.06854984e-01
-1.09899533e+00 1.62138546e+00 6.60244584e-01 -3.82293373e-01
-3.84760112e-01 -1.30740857e+00 -7.38946497e-01 -1.05380170e-01
3.22178036e-01 -8.71533573e-01 1.14286661e+00 -6.29062593e-01
-1.07188857e+00 5.34314394e-01 4.97805953e-01 2.79175229e-02
3.52632105e-01 -2.05662578e-01 -1.23708987e+00 -3.71643960e-01
3.39177161e-01 -2.24514291e-01 -1.63839236e-01 -1.47174525e+00
-6.94036305e-01 -4.17162478e-01 2.85636812e-01 -1.50950328e-01
-4.90008831e-01 4.67759341e-01 -8.87385756e-02 -1.93571165e-01
1.15278974e-01 -3.81818116e-02 1.93273440e-01 4.21000451e-01
3.47363502e-01 3.69519264e-01 1.18153107e+00 -5.72316766e-01
1.68036079e+00 -1.72828448e+00 -8.37228894e-02 3.73758465e-01
3.05431366e-01 2.51981288e-01 2.04352394e-01 1.48125732e+00
-3.53314169e-02 4.00114208e-01 -9.65493917e-02 5.18757701e-01
8.65517139e-01 2.66877532e-01 4.45780396e-01 -3.72029334e-01
-2.29029819e-01 3.78542155e-01 -6.63225234e-01 -1.14677511e-01
6.27280593e-01 3.10212046e-01 -1.69596717e-01 -1.35019710e-02
-3.24576288e-01 2.39848033e-01 -4.43513334e-01 8.01263690e-01
5.34311831e-01 -1.79565549e-01 6.52464092e-01 -2.80564010e-01
-6.69208229e-01 2.78060082e-02 -1.79338074e+00 2.15227842e+00
-1.08578181e+00 -2.92785242e-02 4.29710299e-01 -3.75763685e-01
1.30104256e+00 8.34226429e-01 3.27465564e-01 -5.46798587e-01
-1.13706775e-01 6.72904134e-01 -1.91827282e-01 -1.13963914e+00
1.36635348e-01 -1.18023649e-01 -1.40272602e-01 5.49894929e-01
1.46396205e-01 -2.07278833e-01 2.67002136e-01 -1.97295472e-01
1.10913730e+00 9.04532075e-01 4.24828261e-01 -5.95444024e-01
4.59484100e-01 1.51053369e-01 4.52111721e-01 1.49298340e-01
4.72718447e-01 -1.51580130e-03 -2.59556044e-02 -7.50183582e-01
-1.26103592e+00 -1.09957790e+00 -3.05105001e-01 7.19865263e-01
2.56805211e-01 -1.18295801e+00 -4.55395967e-01 -1.01050235e-01
1.09627908e-02 9.65926647e-01 -3.30846548e-01 1.19927796e-02
3.47124189e-02 -6.93953736e-03 9.47627798e-02 6.50410295e-01
3.80745173e-01 -5.48416972e-01 -1.17104471e+00 5.81931770e-01
3.76687974e-01 -9.35547352e-01 5.02567589e-01 -2.48146638e-01
-6.11222386e-01 -1.01669371e+00 2.16337383e-01 -3.77583206e-01
1.87531993e-01 -1.90918043e-01 1.31784272e+00 1.60215676e-01
-5.32529473e-01 8.15671563e-01 -4.82400417e-01 -8.42660606e-01
-7.15988994e-01 -4.34128940e-01 -3.70221078e-01 -3.91431719e-01
4.35323685e-01 -1.02537656e+00 -2.43635222e-01 5.09757221e-01
-1.19863021e+00 2.58611053e-01 -9.31290165e-02 -1.26458546e-02
1.09858207e-01 1.57208070e-01 1.11872649e+00 -6.83220387e-01
7.83541560e-01 -5.52773237e-01 -5.50071657e-01 6.43432796e-01
-9.55930293e-01 -6.95957765e-02 2.76666552e-01 2.30288610e-01
-1.36944318e+00 -5.35396099e-01 1.40640423e-01 2.71033376e-01
-5.32557011e-01 9.52437520e-01 -9.53746080e-01 -9.54219326e-02
4.70399290e-01 -4.96568710e-01 2.42630944e-01 -9.99630690e-01
2.76829928e-01 9.55847323e-01 2.88113058e-01 -8.74289334e-01
7.41685688e-01 2.36781508e-01 -2.26899937e-01 -5.02534747e-01
-1.26174420e-01 -1.09815322e-01 -5.77739477e-01 -7.76298761e-01
7.66728640e-01 -3.13421994e-01 -4.83311445e-01 7.43441358e-02
-1.17338383e+00 2.13461325e-01 -7.15894341e-01 3.43520075e-01
-3.20103675e-01 4.59539006e-03 2.74406075e-01 -8.38732123e-01
-1.38808772e-01 -8.91600609e-01 5.76942444e-01 1.37632992e-02
-1.06757426e+00 -1.09730697e+00 2.71628231e-01 4.75966513e-01
7.95312107e-01 8.62853110e-01 1.25703740e+00 -5.47445655e-01
-5.94881773e-01 -6.12118006e-01 -5.42004071e-02 3.71506959e-01
5.55729270e-01 1.95663840e-01 -6.16729259e-01 2.41855696e-01
-2.56507903e-01 3.70720387e-01 -9.29538488e-01 -6.09113753e-01
4.96246010e-01 6.89865649e-02 -2.87783802e-01 3.02624702e-02
1.95528591e+00 6.65887058e-01 7.31349707e-01 1.06645119e+00
4.28147346e-01 1.15719700e+00 5.92041433e-01 4.29617465e-01
3.22112471e-01 8.73162150e-01 2.85382628e-01 -1.83063075e-02
-3.82627517e-01 2.19271079e-01 -3.34086478e-01 6.08712912e-01
-2.12699816e-01 2.86925018e-01 -1.56965697e+00 5.39223373e-01
-2.02849650e+00 -7.51739562e-01 -5.24417937e-01 2.14807296e+00
4.89852607e-01 -4.87719178e-02 -1.29991308e-01 2.37138465e-01
4.76755679e-01 -5.25125802e-01 -2.35659316e-01 -6.31081223e-01
4.53635752e-01 4.64076191e-01 1.67186763e-02 3.93457323e-01
-4.04271819e-02 1.90268859e-01 5.98858261e+00 1.58605620e-01
-2.86433578e-01 5.61574578e-01 -7.95499086e-01 2.42097735e-01
-6.08839333e-01 5.14277339e-01 -2.41641194e-01 1.91823840e-01
1.17086935e+00 -9.95836318e-01 2.43375376e-01 7.63772547e-01
4.52121198e-01 2.45901152e-01 -1.12325323e+00 4.55411762e-01
-3.67844611e-01 -2.08100367e+00 -8.57648030e-02 7.42845163e-02
5.79327524e-01 -3.56034786e-01 -6.61926270e-01 -3.95001352e-01
4.86910075e-01 -4.73114222e-01 1.16848409e+00 1.17180061e+00
4.85656291e-01 -5.60362279e-01 4.86956179e-01 -1.75305948e-01
-1.37636840e+00 -2.14240372e-01 3.74823958e-01 -3.22944492e-01
3.65493864e-01 2.48064980e-01 -2.83929259e-01 1.45786154e+00
1.11306405e+00 3.00038099e-01 -3.15052241e-01 1.03992009e+00
3.93876463e-01 -3.61099839e-01 -2.58478940e-01 4.06583220e-01
-4.23421383e-01 -3.44814748e-01 3.86413664e-01 1.03159904e+00
5.89554489e-01 -5.77313066e-01 -1.13690339e-01 1.07365072e+00
7.12015331e-01 4.41308953e-02 -7.72615135e-01 -8.27568024e-02
8.75152290e-01 1.08311200e+00 -2.57276148e-01 -1.61756948e-01
-8.41739595e-01 3.24028671e-01 -4.58991885e-01 3.90122592e-01
-4.37396228e-01 -1.06055570e+00 1.02950883e+00 8.10922801e-01
-1.79108027e-02 -4.01748747e-01 -5.47133267e-01 -5.32224059e-01
2.29830801e-01 -8.52262616e-01 3.54159147e-01 -1.22433650e+00
-1.31644595e+00 3.95691246e-01 4.50595707e-01 -1.40590072e+00
-1.89721063e-01 -6.00603878e-01 -3.09873998e-01 1.07718551e+00
-9.06418800e-01 -1.52222133e+00 -6.69460535e-01 3.94940330e-03
-7.37581551e-02 -5.44546023e-02 1.21502912e+00 6.55617118e-01
-1.23841673e-01 -3.97234619e-01 -1.11977905e-01 -2.60720283e-01
3.46635342e-01 -1.18127024e+00 1.26829877e-01 5.92520893e-01
-6.16435766e-01 8.77400160e-01 7.41330087e-01 -7.39777029e-01
-1.48494399e+00 -8.62208426e-01 6.81755900e-01 -7.84852922e-01
1.02272344e+00 -2.41597444e-01 -9.32745934e-01 7.52168596e-01
5.21048725e-01 -3.53669673e-01 1.28222847e+00 -2.89994359e-01
-3.84543151e-01 -3.15945655e-01 -1.52649271e+00 5.06625175e-01
1.16158950e+00 -6.58976674e-01 -9.30347860e-01 8.29106662e-03
5.81528306e-01 1.59143329e-01 -1.99525452e+00 4.14108425e-01
9.37341154e-01 -1.02382851e+00 9.25263941e-01 -7.24116564e-01
-1.41324028e-01 -1.03155160e+00 -5.03871083e-01 -6.66169286e-01
-2.54201233e-01 -7.06554592e-01 8.30851346e-02 1.57457948e+00
4.99799579e-01 -1.22355998e+00 -4.31008637e-02 1.35054255e+00
-6.66446924e-01 -1.25900537e-01 -5.79557598e-01 -1.16482961e+00
-2.83647090e-01 -8.12635243e-01 1.50220478e+00 1.11467803e+00
5.14752567e-01 1.20881619e-02 3.95016760e-01 2.60905057e-01
5.43731868e-01 -2.41577625e-02 7.49528170e-01 -1.95217443e+00
-1.11688323e-01 -2.22585887e-01 -8.46567869e-01 2.79486507e-01
-6.37945592e-01 -9.45710540e-01 -6.35757029e-01 -2.53562450e+00
-5.73682725e-01 -2.48094410e-01 5.25775217e-02 4.75616306e-01
9.79746044e-01 -1.93091244e-01 1.43433109e-01 2.47442529e-01
-6.47035018e-02 -5.40008806e-02 8.56992841e-01 3.48740548e-01
8.08808580e-02 -7.58351922e-01 -9.07456934e-01 8.59977663e-01
6.66273713e-01 -6.17272973e-01 -5.69388151e-01 -5.68039656e-01
6.77213609e-01 -2.66244084e-01 4.58416581e-01 -1.49414706e+00
9.43544433e-02 -5.17420948e-01 6.04382670e-03 -3.32594365e-02
1.81741595e-01 -1.69752264e+00 1.85446262e+00 3.71143430e-01
6.03036024e-02 -4.52637784e-02 2.26351187e-01 1.31758764e-01
-2.33605392e-02 -6.88705623e-01 4.68491733e-01 -2.47219026e-01
-8.16378355e-01 -2.09351465e-01 -4.42899704e-01 -4.17101920e-01
1.36444056e+00 -8.13629985e-01 -6.42538369e-01 3.84379029e-01
-1.07359874e+00 5.38153462e-02 1.06097734e+00 6.26392066e-01
-3.71813253e-02 -1.29203713e+00 -1.08061589e-01 2.61614799e-01
7.07526028e-01 -4.30378944e-01 1.10841386e-01 1.44471467e-01
-7.03343987e-01 1.70499310e-01 -6.70573115e-01 -1.29667342e-01
-7.55063474e-01 2.78854251e-01 7.13242412e-01 3.05553794e-01
-5.04007399e-01 -1.73387341e-02 -3.97811115e-01 -8.31174314e-01
-2.51026154e-01 -9.66081843e-02 1.21882327e-01 -1.23336725e-02
6.13797843e-01 8.76264691e-01 4.94672298e-01 -2.36841932e-01
-5.60889244e-01 8.27755153e-01 5.28394461e-01 -3.71580005e-01
1.53473258e+00 -2.98539549e-01 -3.63170862e-01 8.81534398e-01
6.16465092e-01 -2.48541653e-01 -7.53177881e-01 5.56702614e-02
8.21171045e-01 -5.87383330e-01 1.00948205e-02 -1.51061785e+00
-5.07046103e-01 3.72272640e-01 7.37020433e-01 9.14400816e-01
7.51004457e-01 2.31951162e-01 2.46677235e-01 8.33978876e-02
7.20960557e-01 -1.51055634e+00 -3.75677556e-01 -2.42184415e-01
1.45680296e+00 -2.03708634e-01 8.32344741e-02 -5.51501751e-01
-2.67486989e-01 1.47861123e+00 6.83271587e-01 5.35074234e-01
1.00870216e+00 9.08079743e-01 3.43258888e-01 -8.70106220e-01
-7.76453853e-01 -2.17868820e-01 -1.29249573e-01 1.31811023e+00
5.87278068e-01 -3.89286846e-01 -7.27654457e-01 9.13320661e-01
3.38669121e-01 7.05537975e-01 4.49911416e-01 1.97989571e+00
-5.53178251e-01 -1.85080636e+00 -5.82926869e-01 2.18227848e-01
-2.73365825e-02 4.03998345e-01 -3.12809139e-01 9.88239765e-01
5.57667732e-01 1.04916692e+00 -6.79864585e-02 -2.96852916e-01
1.05981064e+00 4.16226000e-01 4.90844280e-01 -7.27514386e-01
-9.25274134e-01 -3.62580895e-01 7.72463679e-01 -5.30892193e-01
-4.47079271e-01 -4.12216663e-01 -1.38416505e+00 -1.86058450e-02
-1.23681566e-02 3.05187374e-01 1.20000720e+00 7.77159691e-01
1.14392078e+00 1.05217803e+00 -6.96578547e-02 -3.87671441e-01
-5.97689152e-02 -6.28292620e-01 -6.33481801e-01 2.33972952e-01
-3.68831396e-01 -1.01423669e+00 4.52388078e-02 4.98437345e-01] | [9.062145233154297, 7.666627883911133] |
e29d2da3-3aae-41d7-98cb-07db1949c2c3 | a-comprehensive-study-on-the-robustness-of | 2306.12111 | null | https://arxiv.org/abs/2306.12111v1 | https://arxiv.org/pdf/2306.12111v1.pdf | A Comprehensive Study on the Robustness of Image Classification and Object Detection in Remote Sensing: Surveying and Benchmarking | Deep neural networks (DNNs) have found widespread applications in interpreting remote sensing (RS) imagery. However, it has been demonstrated in previous works that DNNs are vulnerable to different types of noises, particularly adversarial noises. Surprisingly, there has been a lack of comprehensive studies on the robustness of RS tasks, prompting us to undertake a thorough survey and benchmark on the robustness of image classification and object detection in RS. To our best knowledge, this study represents the first comprehensive examination of both natural robustness and adversarial robustness in RS tasks. Specifically, we have curated and made publicly available datasets that contain natural and adversarial noises. These datasets serve as valuable resources for evaluating the robustness of DNNs-based models. To provide a comprehensive assessment of model robustness, we conducted meticulous experiments with numerous different classifiers and detectors, encompassing a wide range of mainstream methods. Through rigorous evaluation, we have uncovered insightful and intriguing findings, which shed light on the relationship between adversarial noise crafting and model training, yielding a deeper understanding of the susceptibility and limitations of various models, and providing guidance for the development of more resilient and robust models | ['Lap-Pui Chau', 'Mingyang Ma', 'Yuru Su', 'Xiaofei Wang', 'Jiawei Lian', 'Shaohui Mei'] | 2023-06-21 | null | null | null | null | ['adversarial-robustness', 'benchmarking', 'benchmarking'] | ['adversarial', 'miscellaneous', 'robots'] | [ 4.66078222e-01 -4.89244074e-01 3.34127396e-01 -2.36078829e-01
-6.23695135e-01 -9.47927296e-01 6.38126612e-01 -1.55494377e-01
-4.06333327e-01 3.70777667e-01 1.83654413e-01 -6.66282058e-01
-2.03941077e-01 -8.30301046e-01 -6.68120146e-01 -8.34281027e-01
-3.13570678e-01 -4.26850289e-01 -5.86234815e-02 -5.74136734e-01
9.83517393e-02 1.07097483e+00 -1.60021424e+00 1.94439813e-02
7.37958729e-01 9.26332831e-01 -3.65853757e-02 6.93459630e-01
5.90645075e-01 7.96144903e-01 -7.25611091e-01 -6.59842372e-01
4.85220551e-01 -5.43295182e-02 -6.58122241e-01 -1.57711759e-01
4.48285460e-01 -4.69218075e-01 -9.00794446e-01 1.21595430e+00
1.00519407e+00 2.34163821e-01 4.68152732e-01 -8.74797225e-01
-1.23341095e+00 6.24334693e-01 -2.23172873e-01 7.47488678e-01
-7.15368018e-02 7.54531980e-01 9.31938946e-01 -5.72926342e-01
4.23344582e-01 1.29251075e+00 1.06916308e+00 6.48352027e-01
-1.09034669e+00 -7.78476954e-01 1.50891885e-01 -9.81279532e-04
-1.28934336e+00 -8.30142438e-01 4.38941747e-01 -3.96230161e-01
7.03844786e-01 5.06043673e-01 3.12415302e-01 1.49940622e+00
3.23574021e-02 5.88302195e-01 1.21826661e+00 -8.30801576e-02
7.80543163e-02 -1.17018029e-01 -1.26476809e-01 1.36850476e-01
1.81632042e-01 6.38918400e-01 -3.99886258e-02 -2.37793654e-01
6.92293882e-01 -2.42912155e-02 -5.25746882e-01 4.19640422e-01
-8.49621058e-01 8.19420755e-01 8.91750813e-01 1.84967369e-01
-2.10659698e-01 1.03920527e-01 5.38328111e-01 1.70610115e-01
5.14119029e-01 5.31253159e-01 -4.11393672e-01 3.99548739e-01
-5.32623887e-01 2.13648796e-01 4.53725606e-01 4.44710970e-01
3.52103859e-01 5.15225291e-01 -1.26252279e-01 9.86477137e-01
-5.17060906e-02 8.59670877e-01 1.34854257e-01 -6.95954442e-01
2.53110528e-01 1.40563071e-01 -1.56160919e-02 -1.55718613e+00
-4.03884023e-01 -6.54897332e-01 -1.19765508e+00 2.21972585e-01
1.85650304e-01 -1.79018244e-01 -8.97235990e-01 1.85095108e+00
-6.20072782e-02 1.91831648e-01 1.00557409e-01 8.41941476e-01
9.22275007e-01 3.58466834e-01 3.53704661e-01 3.21434110e-01
9.51650679e-01 -4.58840728e-01 -3.57204497e-01 -3.82228166e-01
1.83860466e-01 -7.40565062e-01 1.08726287e+00 1.68545499e-01
-6.74628377e-01 -5.92048824e-01 -9.52512980e-01 2.04918668e-01
-5.15752196e-01 -1.45578638e-01 7.60119200e-01 7.55978942e-01
-8.12948406e-01 8.62036407e-01 -8.90334666e-01 -5.89486361e-01
7.82419205e-01 -8.05516541e-02 -5.51667392e-01 -3.11134398e-01
-1.37058091e+00 1.10822666e+00 1.09252766e-01 7.08301544e-01
-1.21402347e+00 -7.12053895e-01 -7.31535554e-01 -2.84769446e-01
6.71616122e-02 -4.01465654e-01 1.05674279e+00 -8.00923944e-01
-9.30684209e-01 8.20522070e-01 2.66429871e-01 -5.42101920e-01
4.91210103e-01 -1.63558275e-01 -8.46927226e-01 4.57698889e-02
-8.42060894e-02 2.65111506e-01 7.52947330e-01 -1.32897902e+00
-5.61854951e-02 -4.88480985e-01 6.80411905e-02 -5.40988073e-02
-3.65157485e-01 4.14899111e-01 8.66272673e-02 -1.16147459e+00
1.18487976e-01 -8.40747833e-01 -3.65189314e-01 1.36004269e-01
-5.91521382e-01 2.49697328e-01 5.86936831e-01 -7.00935125e-01
1.08008182e+00 -2.49181700e+00 -2.95459032e-01 4.77680862e-02
2.45064244e-01 7.21172988e-01 -4.83403444e-01 5.43076754e-01
-3.09646606e-01 7.56924450e-01 -5.11222124e-01 5.35185337e-02
-2.11553983e-02 1.94334954e-01 -8.43238235e-01 8.12323630e-01
5.49264729e-01 1.01854289e+00 -7.74440289e-01 1.74747854e-01
2.62768924e-01 6.51000738e-01 -1.87674075e-01 1.68743283e-01
2.33301744e-01 4.80612755e-01 -4.07430559e-01 9.35773790e-01
8.10689330e-01 8.47533047e-02 -3.55164826e-01 -2.17018619e-01
4.81456108e-02 1.10415012e-01 -8.57057154e-01 7.98802853e-01
-2.27527499e-01 9.18283641e-01 2.30386242e-01 -8.86720181e-01
8.01421881e-01 1.29288644e-01 9.48531330e-02 -7.40149319e-01
5.41587137e-02 1.23164609e-01 -9.00142640e-02 -5.73212028e-01
3.26135576e-01 -2.92144090e-01 4.85264920e-02 2.74986565e-01
-2.61357397e-01 -2.42798582e-01 -1.81349859e-01 -4.04814184e-02
1.19845426e+00 -2.70331711e-01 6.52652830e-02 -5.09693772e-02
1.61760956e-01 -1.16769180e-01 4.14117754e-01 1.36231363e+00
-4.88881022e-01 7.13555276e-01 7.67014846e-02 -6.70475066e-01
-8.08804274e-01 -1.17454982e+00 -5.53855479e-01 1.13840115e+00
-2.10178763e-01 8.47470760e-02 -4.43396568e-01 -4.18861508e-01
1.47899345e-01 3.43893051e-01 -8.22330475e-01 -4.48378980e-01
-3.33632350e-01 -1.51216090e+00 1.41916239e+00 6.13151073e-01
6.47093832e-01 -1.01807666e+00 -1.91133946e-01 -8.67018998e-02
-1.69106856e-01 -1.23922205e+00 8.51105154e-02 1.55237332e-01
-8.29338729e-01 -1.25561345e+00 -3.08365375e-01 -2.41519734e-01
3.85684520e-01 6.29822850e-01 1.19045436e+00 3.29550803e-01
-4.85351831e-01 2.59288043e-01 -5.01069367e-01 -5.93534946e-01
-5.90794683e-01 -1.27360955e-01 2.60091364e-01 -4.03279543e-01
9.09636766e-02 -5.71836829e-01 -6.28328681e-01 5.08746207e-01
-1.54743648e+00 -7.96166778e-01 6.74589336e-01 6.25495911e-01
1.67027161e-01 3.39626729e-01 6.93220258e-01 -6.13218844e-01
7.41175294e-01 -7.88846135e-01 -4.29979384e-01 8.42116326e-02
-3.77939314e-01 -2.76465446e-01 7.34426141e-01 -2.33622715e-01
-8.68519425e-01 -2.58602113e-01 -5.63295722e-01 -5.80310449e-02
-4.37153548e-01 7.04265475e-01 -4.92976792e-02 -5.00638247e-01
1.32128119e+00 1.69816837e-01 -1.88923702e-01 -5.31909406e-01
4.15386647e-01 7.47931361e-01 8.11993718e-01 -3.77401590e-01
1.12902355e+00 6.60022199e-01 -1.69476375e-01 -1.13428605e+00
-7.99064636e-01 -1.77321136e-01 -2.74002194e-01 -7.47255459e-02
4.10884500e-01 -1.00828075e+00 -3.23749185e-01 9.79646206e-01
-6.80383027e-01 -4.97147113e-01 -2.29067393e-02 1.05464928e-01
-1.79783031e-02 6.51592910e-01 -6.75398588e-01 -8.24918568e-01
-3.03021848e-01 -1.10268319e+00 7.45540261e-01 2.67676003e-02
1.03931744e-02 -9.35018063e-01 6.87792823e-02 2.62031615e-01
7.77316391e-01 5.46669126e-01 6.25690281e-01 -6.02720737e-01
-4.36369985e-01 -4.59429055e-01 -3.31673265e-01 6.44505203e-01
1.14569381e-01 5.46694040e-01 -1.50431275e+00 -2.79421240e-01
-7.08594220e-03 -5.61551452e-01 1.30273521e+00 3.54929835e-01
1.19880033e+00 -3.68795902e-01 -6.91029802e-02 1.03281641e+00
1.47427630e+00 -2.19387308e-01 9.91762102e-01 6.73843920e-01
5.51866353e-01 4.41340268e-01 1.19445384e-01 3.05099159e-01
2.31513921e-02 2.99852341e-01 9.10401583e-01 -4.25980866e-01
-1.37091056e-02 2.93798074e-02 1.89763576e-01 -1.13721481e-02
-2.42320314e-01 -5.44906735e-01 -1.01572943e+00 4.23438847e-01
-1.23515117e+00 -1.19568372e+00 2.81700911e-03 1.97252238e+00
6.72962964e-01 -1.51288599e-01 -1.43001020e-01 2.18582422e-01
7.28145242e-01 6.14799380e-01 -7.65171230e-01 8.47729146e-02
-7.95459569e-01 1.51969209e-01 7.03424633e-01 1.55492589e-01
-1.57632816e+00 1.07614219e+00 7.86080313e+00 5.84267437e-01
-1.23916972e+00 -3.49764258e-01 7.36935854e-01 1.18496783e-01
-2.29904994e-01 -4.00334060e-01 -3.69554579e-01 1.96606770e-01
7.34147668e-01 1.63156733e-01 5.58538914e-01 8.12360406e-01
4.96328354e-01 7.52044618e-02 -7.27940917e-01 5.45045912e-01
-9.48409587e-02 -1.23259020e+00 2.52479110e-02 -1.56250089e-01
8.78153443e-01 6.85177386e-01 3.88672858e-01 2.00921088e-01
6.88052654e-01 -1.33391285e+00 4.81904149e-01 4.31912780e-01
6.24202609e-01 -5.70380747e-01 8.94459307e-01 8.67443979e-02
-8.46433282e-01 -2.66485393e-01 -5.85728765e-01 -2.26779655e-01
-1.62512824e-01 6.30853474e-01 -2.79200941e-01 4.52432364e-01
1.08779716e+00 8.57937932e-01 -7.36021042e-01 7.95085013e-01
-1.93933979e-01 8.56981754e-01 -1.90993741e-01 5.08269012e-01
9.85791460e-02 1.47720337e-01 5.58784902e-01 1.28142631e+00
2.37297639e-02 2.13327631e-01 -1.46444216e-01 8.59849989e-01
-2.22045928e-01 -2.97470480e-01 -1.02515316e+00 -3.35464090e-01
6.07544839e-01 1.13372219e+00 -5.27406812e-01 2.12069780e-01
-2.57785112e-01 6.61949873e-01 2.60264277e-01 6.54700458e-01
-6.85932696e-01 -2.50143260e-01 1.19731033e+00 -1.93003029e-01
-1.98395979e-02 -4.34746623e-01 -4.98683870e-01 -1.10626614e+00
-1.88827645e-02 -1.11632907e+00 4.48227704e-01 -7.88479507e-01
-1.75317764e+00 7.07852423e-01 -2.90312558e-01 -9.95809376e-01
3.67489755e-01 -7.03677356e-01 -8.16287220e-01 8.66384685e-01
-1.79635572e+00 -9.97154236e-01 -4.72668946e-01 4.75788951e-01
9.30502936e-02 -1.88612685e-01 8.76749456e-01 1.38233945e-01
-1.03569746e+00 6.06978178e-01 3.78959864e-01 4.06496137e-01
5.24240613e-01 -6.37233555e-01 9.31165695e-01 1.35021424e+00
-1.89921543e-01 9.74987805e-01 7.81722188e-01 -5.43632746e-01
-1.27439225e+00 -1.46468759e+00 2.03341186e-01 -6.78603351e-01
9.52704728e-01 -6.93753455e-03 -1.13912964e+00 8.52596700e-01
-2.39847943e-01 1.83171019e-01 8.35305810e-01 -2.87265986e-01
-5.70984483e-01 -1.22011356e-01 -1.15068126e+00 7.17151523e-01
1.08123481e+00 -7.85586417e-01 -3.71943027e-01 2.52642006e-01
5.92934728e-01 -1.24753855e-01 -8.44960153e-01 6.99338078e-01
4.11744744e-01 -1.16384888e+00 1.44933224e+00 -7.48078048e-01
5.12234747e-01 -2.66487718e-01 -4.47753459e-01 -1.30070233e+00
-5.66162765e-01 -4.36419845e-01 3.47191066e-01 1.12715065e+00
2.44489402e-01 -7.66256273e-01 1.31403252e-01 6.15669787e-01
-3.36888433e-01 -3.65403295e-01 -6.98462784e-01 -7.50335097e-01
3.03379327e-01 -8.19867671e-01 6.11401260e-01 1.06339633e+00
-6.49852633e-01 -3.73471826e-01 -3.32399577e-01 8.01000595e-01
5.43477297e-01 -4.56189424e-01 6.30548596e-01 -1.10745311e+00
2.41338268e-01 -6.53050780e-01 -4.43725467e-01 -5.08380234e-01
2.62880683e-01 -7.25769818e-01 1.07756317e-01 -1.23229635e+00
3.65771502e-02 -4.69004333e-01 -4.30679858e-01 5.78604400e-01
-6.36493921e-01 8.22170377e-01 -3.17845754e-02 3.99094880e-01
-8.05056095e-02 4.70071554e-01 9.93864238e-01 -1.66068971e-01
5.91882877e-02 1.57648355e-01 -1.26436198e+00 7.45084882e-01
1.23440516e+00 -4.85731542e-01 -1.41831294e-01 -8.32966745e-01
3.50453168e-01 -5.84362924e-01 8.93343747e-01 -8.50288272e-01
-1.37434572e-01 -4.60631073e-01 4.22106236e-01 -1.93592072e-01
-5.71303479e-02 -5.65089285e-01 6.58199715e-05 3.28032404e-01
-2.35654771e-01 -6.63951486e-02 6.60157323e-01 7.03506410e-01
1.95641350e-03 1.08686201e-01 1.03268087e+00 -4.15378273e-01
-8.82090628e-01 4.51347023e-01 -5.27879596e-01 8.27118754e-02
6.11276269e-01 -2.13259444e-01 -6.37529373e-01 -2.66274720e-01
-5.58786213e-01 4.69332142e-03 5.58837354e-01 4.25475389e-01
5.48695445e-01 -1.01310825e+00 -9.76821542e-01 2.73940861e-01
2.40674078e-01 -9.94888023e-02 5.58863401e-01 4.35212731e-01
-4.71398711e-01 6.94061294e-02 -2.90496260e-01 -3.58470917e-01
-7.47870743e-01 5.42965591e-01 6.46559656e-01 2.44053662e-01
-4.63168323e-01 9.20967638e-01 1.08746082e-01 -6.58961058e-01
1.04136467e-01 -1.05111964e-01 -8.79344866e-02 -1.69582307e-01
6.77353501e-01 7.62108743e-01 2.13048920e-01 -7.48786747e-01
-3.97824317e-01 3.42295051e-01 2.26755351e-01 2.13488549e-01
1.58004785e+00 -4.75330465e-02 -2.69526452e-01 3.89948301e-02
8.38312685e-01 -7.50628859e-02 -1.32898092e+00 -1.98919758e-01
-2.89445817e-01 -5.17896593e-01 5.87163642e-02 -8.93641651e-01
-1.15342093e+00 7.55827963e-01 5.46307921e-01 4.17093724e-01
1.37734139e+00 -1.48762181e-01 4.60936964e-01 5.12189090e-01
2.44216956e-02 -7.39856839e-01 -3.41334268e-02 6.78481460e-01
1.21624637e+00 -1.37563241e+00 2.60757264e-02 -7.56208301e-02
-5.11340976e-01 8.81293833e-01 3.81727427e-01 4.53279261e-03
5.17935634e-01 1.79559752e-01 4.53148872e-01 -9.10702720e-02
-2.35888675e-01 -3.05802315e-01 1.50483012e-01 9.91411805e-01
1.57979831e-01 4.22240533e-02 3.17189872e-01 4.61238950e-01
-2.63670385e-01 -1.60160527e-01 3.61176044e-01 7.55979717e-01
-3.94264877e-01 -6.60351038e-01 -5.73564053e-01 3.52995038e-01
-7.63975739e-01 -4.84608978e-01 -5.29274583e-01 7.30477810e-01
-1.68522224e-01 1.37482584e+00 -3.97900492e-01 -5.31479180e-01
5.16779363e-01 -2.21404836e-01 -8.32171887e-02 -4.28899705e-01
-7.30893433e-01 -4.99846101e-01 -2.74109319e-02 -5.08769214e-01
-3.19502473e-01 -5.25623262e-01 -4.67946231e-01 -6.81128442e-01
-3.96767646e-01 -3.08442861e-01 5.43588996e-01 8.97960424e-01
3.97745401e-01 3.58013868e-01 8.37901831e-01 -9.68850374e-01
-8.55672061e-01 -1.01864743e+00 -6.50274754e-01 5.59090197e-01
6.55407369e-01 -3.41108114e-01 -7.99215674e-01 -6.20141514e-02] | [5.607115268707275, 7.875911235809326] |
80680571-611c-4ce5-950d-809980270979 | faithful-knowledge-distillation | 2306.04431 | null | https://arxiv.org/abs/2306.04431v2 | https://arxiv.org/pdf/2306.04431v2.pdf | Faithful Knowledge Distillation | Knowledge distillation (KD) has received much attention due to its success in compressing networks to allow for their deployment in resource-constrained systems. While the problem of adversarial robustness has been studied before in the KD setting, previous works overlook what we term the relative calibration of the student network with respect to its teacher in terms of soft confidences. In particular, we focus on two crucial questions with regard to a teacher-student pair: (i) do the teacher and student disagree at points close to correctly classified dataset examples, and (ii) is the distilled student as confident as the teacher around dataset examples? These are critical questions when considering the deployment of a smaller student network trained from a robust teacher within a safety-critical setting. To address these questions, we introduce a faithful imitation framework to discuss the relative calibration of confidences, as well as provide empirical and certified methods to evaluate the relative calibration of a student w.r.t. its teacher. Further, to verifiably align the relative calibration incentives of the student to those of its teacher, we introduce faithful distillation. Our experiments on the MNIST and Fashion-MNIST datasets demonstrate the need for such an analysis and the advantages of the increased verifiability of faithful distillation over alternative adversarial distillation methods. | ['Krishnamurthy Dj Dvijotham', 'Francisco Eiras', 'Philip H. S. Torr', 'M. Pawan Kumar', 'Rudy Brunel', 'Tom A. Lamb'] | 2023-06-07 | null | null | null | null | ['adversarial-robustness'] | ['adversarial'] | [ 1.47116020e-01 7.13158071e-01 -3.73411924e-01 -2.10202277e-01
-6.34399235e-01 -1.02295315e+00 4.73760903e-01 2.95119673e-01
-5.39403975e-01 7.82790959e-01 1.58126913e-02 -4.92628813e-01
-4.23514038e-01 -7.95333266e-01 -1.12644124e+00 -6.87851131e-01
-1.29123852e-01 5.74992537e-01 1.84437633e-01 -1.57962903e-01
-1.20468736e-01 4.57934618e-01 -1.00823617e+00 -3.02267760e-01
9.56695676e-01 9.14501905e-01 -4.17318970e-01 6.67367458e-01
5.75207710e-01 1.06199765e+00 -7.94285953e-01 -1.05722666e+00
6.72188997e-01 -2.17145875e-01 -9.38299477e-01 -2.72783488e-01
8.05966258e-01 -7.06577539e-01 -8.32934499e-01 1.39864993e+00
4.32159662e-01 2.85786837e-01 4.64843333e-01 -1.72493863e+00
-7.12392211e-01 1.15834486e+00 -2.59245425e-01 4.65737849e-01
-1.18893355e-01 2.07649380e-01 9.82191324e-01 -2.30354503e-01
4.09883589e-01 1.03171599e+00 5.34223497e-01 5.40406883e-01
-1.24359608e+00 -9.22897637e-01 1.25478327e-01 9.89893004e-02
-1.19469190e+00 -4.67041403e-01 5.83055556e-01 -4.07035351e-01
3.11049044e-01 3.52511704e-02 3.23294699e-01 1.23231733e+00
-8.34784359e-02 7.55892813e-01 1.04883087e+00 -1.28072232e-01
2.71857917e-01 6.75202727e-01 1.64898753e-01 6.14016533e-01
5.95623732e-01 6.11995935e-01 -4.29025590e-01 -6.43884763e-02
6.33509099e-01 -2.37056538e-01 -4.52747256e-01 -4.85623389e-01
-9.21414375e-01 9.87985849e-01 6.52584255e-01 7.09472001e-02
-7.56117404e-02 4.56451327e-01 3.95328403e-01 6.57965779e-01
4.24876213e-01 7.14387774e-01 -4.10029471e-01 -2.24157706e-01
-1.00867724e+00 1.41429782e-01 1.05561352e+00 1.12842929e+00
3.87215555e-01 3.67914826e-01 -3.01618566e-04 1.85556993e-01
-4.90942597e-03 4.26902294e-01 2.23182768e-01 -1.01307034e+00
5.91046929e-01 5.16011491e-02 -1.80198699e-01 -1.00840485e+00
1.78967550e-01 -5.82305908e-01 -7.69014001e-01 2.61312038e-01
5.85056782e-01 -4.46494728e-01 -6.63886786e-01 2.13603354e+00
4.21280712e-01 7.36746192e-01 2.35970408e-01 6.52842581e-01
4.67011958e-01 2.75853872e-01 -1.76892936e-01 -2.89123040e-03
9.37698483e-01 -7.36265957e-01 -3.38703752e-01 -6.80781677e-02
4.03304905e-01 -3.92744899e-01 6.81101441e-01 3.82681072e-01
-1.23181283e+00 -4.68360007e-01 -1.22622597e+00 4.47912253e-02
-2.54899412e-01 -5.11922300e-01 3.59645396e-01 8.11688423e-01
-9.09701169e-01 9.26242590e-01 -8.84895861e-01 8.02168697e-02
5.49770832e-01 4.59423602e-01 -4.59435552e-01 1.31695122e-01
-1.41455030e+00 1.12108803e+00 4.88035411e-01 -1.56036541e-01
-1.47656024e+00 -1.07582486e+00 -7.54558563e-01 1.97164223e-01
5.51889181e-01 -5.31909108e-01 1.41977286e+00 -1.03647220e+00
-1.43032920e+00 6.21917725e-01 6.20370746e-01 -9.38039958e-01
1.00988472e+00 -1.60419717e-01 -1.02273606e-01 4.62454528e-01
-2.06661552e-01 5.92435956e-01 9.68555093e-01 -1.18627656e+00
-6.72673941e-01 -1.72186956e-01 5.75106144e-01 3.31496559e-02
-4.67793256e-01 -3.41995567e-01 -5.41969389e-03 -8.15070152e-01
-2.42631987e-01 -1.07029939e+00 -3.43032889e-02 2.12575555e-01
-4.90894020e-01 -1.77562669e-01 9.03198481e-01 -4.77396578e-01
8.87948275e-01 -2.25463104e+00 7.11393058e-02 4.83269572e-01
4.69797939e-01 5.45431614e-01 -6.41818568e-02 1.02270417e-01
-3.53164017e-01 2.21726269e-01 -5.50278462e-02 -1.51068598e-01
9.83934775e-02 5.23587763e-01 -8.12482655e-01 8.46467137e-01
2.97201902e-01 7.49329388e-01 -1.19366384e+00 -3.62887233e-01
-1.39395982e-01 4.71437544e-01 -7.16277957e-01 4.66419488e-01
1.05564050e-01 3.53329629e-01 -3.01824212e-01 3.86583358e-01
5.26266754e-01 -1.91568993e-02 4.49957997e-02 -7.93075040e-02
5.33924758e-01 2.76745588e-01 -1.31837952e+00 1.02012944e+00
-6.80474341e-02 6.42330468e-01 3.64451319e-01 -1.29140365e+00
7.52018929e-01 4.56218839e-01 2.45119259e-01 -2.07120419e-01
1.54022500e-01 1.63178399e-01 2.30234489e-01 -2.38765612e-01
6.25603914e-01 -4.63283777e-01 -4.24420647e-02 6.62447214e-01
3.83380413e-01 -5.08366644e-01 -1.18523329e-01 6.10076427e-01
1.03814292e+00 -2.72556633e-01 -5.28540201e-02 -3.21197659e-01
1.87933341e-01 -3.35830182e-01 5.68555295e-01 8.92127156e-01
-6.17531240e-01 2.49860287e-01 7.12862968e-01 -5.19603826e-02
-1.09719765e+00 -1.18710744e+00 -1.30534574e-01 9.08026040e-01
1.75857440e-01 -4.60691117e-02 -8.06379557e-01 -1.09875369e+00
4.72303063e-01 6.92385256e-01 -8.13485205e-01 -6.12184525e-01
-5.04911065e-01 -1.01700284e-01 1.12085319e+00 7.18634725e-01
5.00445247e-01 -4.78306532e-01 -4.68774527e-01 -9.60382447e-02
1.18948907e-01 -1.11322594e+00 -5.48233151e-01 4.68430012e-01
-6.26759231e-01 -1.36701643e+00 -3.67092699e-01 -5.51065564e-01
8.62892210e-01 1.44128993e-01 1.05014658e+00 2.64744251e-03
2.56873935e-01 7.14556754e-01 -1.06932722e-01 -5.33980966e-01
-8.47126544e-01 9.59821194e-02 5.83280802e-01 -3.93530905e-01
-1.13629494e-02 -7.26144075e-01 -3.87546360e-01 3.46796721e-01
-1.15596068e+00 -3.35450649e-01 2.87828445e-01 7.57823646e-01
3.25344265e-01 3.98634732e-01 5.55861831e-01 -9.33813512e-01
7.39587843e-01 -6.15835965e-01 -8.31270874e-01 3.46139044e-01
-8.99685085e-01 1.65159002e-01 8.35346401e-01 -8.32542896e-01
-6.17412567e-01 -3.79112273e-01 9.87029374e-02 -9.21386063e-01
2.25691065e-01 2.87102222e-01 6.32597879e-02 -2.78227121e-01
6.66533411e-01 -1.14789851e-01 1.34758860e-01 -6.89464584e-02
6.25276566e-01 3.50946665e-01 1.09313142e+00 -1.27665901e+00
1.51544189e+00 3.86564314e-01 5.73702939e-02 -3.30255777e-01
-9.53338623e-01 1.42184660e-01 -6.35585606e-01 -1.36908349e-02
4.07399029e-01 -8.57418358e-01 -9.68612731e-01 1.67076677e-01
-7.85552979e-01 -4.22772884e-01 -1.04859197e+00 5.12591958e-01
-5.59198081e-01 3.80458772e-01 -5.95253050e-01 -6.01922214e-01
-2.46199936e-01 -1.44434249e+00 2.39488900e-01 3.42692345e-01
-9.47408006e-02 -1.19826198e+00 -2.76836157e-02 4.81565952e-01
4.26915616e-01 3.46994430e-01 7.99353123e-01 -1.23725641e+00
-4.21583563e-01 -3.48849565e-01 4.76095006e-02 8.91468108e-01
-2.69046247e-01 1.47966504e-01 -1.02445757e+00 -6.41790211e-01
1.03608012e-01 -8.04480791e-01 6.77698731e-01 2.71608438e-02
1.10843897e+00 -8.08116198e-01 8.25937539e-02 7.42744625e-01
1.01204085e+00 -2.12965161e-01 9.28187892e-02 2.20644787e-01
5.32983601e-01 5.05021572e-01 4.13876623e-01 1.59262806e-01
2.62973130e-01 2.31140867e-01 7.62422562e-01 1.48522884e-01
2.00897064e-02 -8.06135654e-01 4.10461426e-01 7.52885699e-01
2.45939806e-01 -7.34640881e-02 -5.96163750e-01 5.09431839e-01
-1.54224074e+00 -9.09512460e-01 4.34065312e-01 2.25227523e+00
1.26841378e+00 4.23091590e-01 -1.00511350e-02 3.00457865e-01
6.05672717e-01 2.26611316e-01 -7.66169012e-01 -4.96882677e-01
2.05385759e-02 2.42646933e-01 6.76754177e-01 5.38570046e-01
-9.99024332e-01 6.44502103e-01 6.39074850e+00 8.79580498e-01
-1.09198880e+00 1.81154519e-01 8.68356824e-01 -1.51500747e-01
-3.17913592e-01 1.94952741e-01 -7.81501889e-01 5.10965466e-01
1.03876507e+00 -5.13473868e-01 6.08867347e-01 1.10943449e+00
-2.40296483e-01 7.08736330e-02 -1.60559928e+00 3.93099576e-01
-8.67292583e-02 -1.26372313e+00 -2.02232897e-01 4.74615619e-02
1.02532053e+00 -1.91669576e-02 5.32171965e-01 4.33594704e-01
1.06214345e+00 -1.15349936e+00 9.25334513e-01 1.52818367e-01
7.19263792e-01 -9.01541352e-01 6.30322576e-01 4.53921646e-01
-6.82130516e-01 -4.81502935e-02 -3.34212184e-01 2.13603824e-01
-3.35559189e-01 5.37281036e-01 -1.04359436e+00 4.06042159e-01
4.84750509e-01 3.51177514e-01 -3.82650942e-01 6.18078947e-01
-7.35937238e-01 9.39857960e-01 -4.37618047e-01 4.20080036e-01
3.26531947e-01 1.40399665e-01 5.70101738e-01 9.95754778e-01
-1.16896555e-02 8.80711898e-03 1.01704568e-01 9.81851578e-01
-6.70188725e-01 -6.12720370e-01 -5.39238930e-01 -2.08387330e-01
8.27826500e-01 1.01380599e+00 -2.15619907e-01 -4.65955436e-01
1.37420252e-01 4.45754707e-01 4.29404616e-01 1.85062423e-01
-8.85255158e-01 -2.90620416e-01 7.29714751e-01 -2.05582865e-02
1.61416829e-01 -9.17715281e-02 -4.76952717e-02 -1.05715322e+00
-3.80769446e-02 -1.08727324e+00 6.33678019e-01 -4.51893419e-01
-1.29761302e+00 2.36900270e-01 1.26035720e-01 -1.06387186e+00
-2.22565502e-01 -3.06593806e-01 -8.22282851e-01 6.39184117e-01
-1.72963691e+00 -8.35352242e-01 -1.43463546e-02 9.02882755e-01
5.73250018e-02 -4.54762243e-02 5.85372031e-01 1.86456814e-01
-7.26938009e-01 1.31623268e+00 1.54444516e-01 5.05328119e-01
8.06547940e-01 -1.36390615e+00 -3.47070284e-02 1.18734300e+00
1.48146451e-01 7.24774003e-01 9.39635217e-01 -4.53604788e-01
-1.42273569e+00 -1.19668949e+00 3.52357566e-01 -6.50722742e-01
1.00626934e+00 5.35613708e-02 -8.36606264e-01 1.00314415e+00
-1.29295914e-02 4.78507012e-01 4.61595088e-01 -1.37537077e-01
-7.72008300e-01 -2.58176774e-01 -1.57755077e+00 2.56941676e-01
5.60918093e-01 -7.65349329e-01 -6.58641934e-01 2.83167511e-01
1.07870746e+00 -6.60159588e-01 -1.23368549e+00 3.55662793e-01
3.94488126e-01 -7.88932741e-01 9.87119794e-01 -8.08111608e-01
5.92896283e-01 -1.38679206e-01 -3.50598693e-02 -1.55450797e+00
4.05265242e-02 -8.02518964e-01 -5.25551558e-01 1.35920048e+00
6.47045970e-02 -6.45023286e-01 8.20622444e-01 8.06287169e-01
9.35886502e-02 -7.05510914e-01 -1.21971822e+00 -1.05358136e+00
6.49366140e-01 -3.61173391e-01 5.99628627e-01 1.35199046e+00
6.94178045e-02 -7.39646927e-02 -4.34317350e-01 4.41891253e-01
6.39887810e-01 -1.20964587e-01 8.19247901e-01 -1.00606954e+00
-5.49490511e-01 -3.41042459e-01 -5.92488647e-01 -8.80893528e-01
3.03515017e-01 -1.06751859e+00 -1.65164154e-02 -7.16633022e-01
5.24037927e-02 -7.84285069e-01 -3.28949571e-01 5.23401082e-01
-3.03385615e-01 1.07082449e-01 2.83170342e-01 9.17411968e-02
-4.67728913e-01 3.88340980e-01 1.15014613e+00 -3.16822499e-01
2.08989874e-01 2.33484000e-01 -1.09100020e+00 6.48137033e-01
6.92881346e-01 -7.34936237e-01 -6.20904326e-01 -3.78641307e-01
2.80247509e-01 1.89687293e-02 5.52460730e-01 -7.27852821e-01
6.33543551e-01 -2.79251546e-01 6.15894012e-02 -5.10240979e-02
1.23781607e-01 -1.01536572e+00 -1.38170436e-01 7.92334139e-01
-6.63088441e-01 1.60170272e-02 -3.81107326e-04 6.38433576e-01
-1.37858421e-01 -4.59546417e-01 1.12951756e+00 7.47335777e-02
-1.45565420e-01 4.80909735e-01 8.84550363e-02 6.00408018e-01
1.18081105e+00 -2.09312011e-02 -6.80552840e-01 -4.59024310e-01
-4.56641793e-01 4.23452437e-01 3.98118138e-01 3.00533295e-01
4.18167740e-01 -1.10424054e+00 -7.00864434e-01 2.21985027e-01
-1.69527292e-01 4.82103497e-01 -1.38720036e-01 7.17094660e-01
-1.92140073e-01 1.39391601e-01 -4.86419350e-03 -3.32136750e-01
-1.09380877e+00 7.26662636e-01 5.44448793e-01 -2.47827351e-01
-4.91236001e-01 1.15725744e+00 1.72199849e-02 -4.29511279e-01
7.79923081e-01 -4.82124746e-01 3.75873506e-01 -1.68827444e-01
3.19214761e-01 4.95393783e-01 -7.20457286e-02 -3.48864079e-01
-7.41642118e-02 -2.00997535e-02 -1.97055370e-01 4.06480432e-02
1.29950893e+00 2.76437968e-01 1.51293978e-01 -1.49365123e-02
1.07599366e+00 7.61084929e-02 -1.54566586e+00 -5.23426175e-01
-1.83367401e-01 -5.26121140e-01 -6.47768229e-02 -6.97480023e-01
-1.48031938e+00 9.41336393e-01 2.48742059e-01 3.41135353e-01
7.58538544e-01 -4.38375138e-02 6.14477873e-01 5.26636899e-01
1.72611117e-01 -1.08936131e+00 1.56850979e-01 3.02748442e-01
5.43016374e-01 -1.14756310e+00 1.79353416e-01 1.00876829e-02
-5.65963507e-01 9.41780388e-01 6.30235970e-01 -3.08647841e-01
6.45132661e-01 4.40385282e-01 -2.23795161e-01 5.62897287e-02
-7.69799769e-01 3.91630024e-01 9.38955769e-02 5.92780709e-01
-1.06177248e-01 1.99549839e-01 3.30350757e-01 4.62946713e-01
-5.96949100e-01 -1.36165693e-01 8.16267312e-01 8.73877645e-01
-3.55089664e-01 -7.79879749e-01 -3.49142015e-01 3.56260717e-01
-7.03269362e-01 1.02261640e-01 -2.03920439e-01 7.60840893e-01
2.19829828e-01 1.00550902e+00 -1.11006975e-01 -5.88933766e-01
2.97335714e-01 -2.23777860e-01 4.74312127e-01 -3.79280388e-01
-9.21776891e-01 -5.68296731e-01 -7.36248642e-02 -3.76216054e-01
-1.49996623e-01 -3.75615031e-01 -9.43515837e-01 -8.54690969e-01
-5.64031780e-01 5.50067663e-01 5.13455331e-01 9.18890476e-01
-5.13376966e-02 3.91698629e-01 9.06701684e-01 -4.45245296e-01
-1.56917799e+00 -6.01022959e-01 -5.43223321e-01 2.94320881e-01
5.66610456e-01 -3.85413915e-01 -8.47973704e-01 -2.19978184e-01] | [5.623227596282959, 7.834003448486328] |
f111f4f0-ab15-4fbc-a78b-9886e32ef755 | feature-transformation-for-cross-domain-few | 2203.02270 | null | https://arxiv.org/abs/2203.02270v1 | https://arxiv.org/pdf/2203.02270v1.pdf | Feature Transformation for Cross-domain Few-shot Remote Sensing Scene Classification | Effectively classifying remote sensing scenes is still a challenge due to the increasing spatial resolution of remote imaging and large variances between remote sensing images. Existing research has greatly improved the performance of remote sensing scene classification (RSSC). However, these methods are not applicable to cross-domain few-shot problems where target domain is with very limited training samples available and has a different data distribution from source domain. To improve the model's applicability, we propose the feature-wise transformation module (FTM) in this paper. FTM transfers the feature distribution learned on source domain to that of target domain by a very simple affine operation with negligible additional parameters. Moreover, FTM can be effectively learned on target domain in the case of few training data available and is agnostic to specific network structures. Experiments on RSSC and land-cover mapping tasks verified its capability to handle cross-domain few-shot problems. By comparison with directly finetuning, FTM achieves better performance and possesses better transferability and fine-grained discriminability. \textit{Code will be publicly available.} | ['Wei Luo', 'Zhihao Chen', 'Qiaoling Chen'] | 2022-03-04 | null | null | null | null | ['cross-domain-few-shot'] | ['computer-vision'] | [ 6.38780355e-01 -3.85072201e-01 -1.08168185e-01 -6.72217667e-01
-8.37104559e-01 -6.18366957e-01 6.23983741e-01 -1.94828138e-01
-3.91482353e-01 7.42519498e-01 -1.74207807e-01 -2.88036227e-01
-6.72214210e-01 -1.09094787e+00 -3.78907442e-01 -8.50694299e-01
-3.61873917e-02 2.39664197e-01 2.22500727e-01 -4.16113764e-01
-1.08501449e-01 6.08035803e-01 -1.43974817e+00 4.57928963e-02
1.20834053e+00 9.28955197e-01 5.85227311e-01 4.02261108e-01
-5.93881458e-02 3.72947216e-01 -2.74879038e-01 1.57929137e-01
6.94729567e-01 -3.59894276e-01 -7.32123494e-01 -7.68000400e-03
6.97892904e-01 -2.15193897e-01 -3.31018627e-01 1.26064408e+00
7.08927691e-01 4.36014593e-01 8.31895888e-01 -1.03405666e+00
-6.36984408e-01 3.74244690e-01 -7.23387957e-01 4.34106529e-01
-3.94901812e-01 -1.00605220e-01 8.66991818e-01 -8.05986822e-01
4.04509306e-01 1.00912523e+00 7.52657533e-01 4.08614784e-01
-1.30382597e+00 -6.56171203e-01 3.58429313e-01 8.95956531e-02
-1.67738271e+00 -2.80376315e-01 5.90676188e-01 -4.77136523e-01
6.58908546e-01 2.74233043e-01 2.32032448e-01 6.61660850e-01
-2.16181517e-01 4.93912697e-01 1.34102130e+00 -9.23685208e-02
4.38735694e-01 1.14859343e-01 1.81865767e-01 1.57795951e-01
3.99976261e-02 1.09636344e-01 -5.34097701e-02 -6.71408474e-02
8.55029166e-01 3.46989483e-01 -3.90401691e-01 -3.04742157e-01
-9.60009933e-01 1.02650428e+00 7.06738830e-01 4.36504483e-01
-3.52957278e-01 -9.34865251e-02 1.46988899e-01 6.36513829e-01
7.79852629e-01 3.49557579e-01 -4.25953090e-01 2.88524121e-01
-1.12513328e+00 3.23846899e-02 5.66367209e-01 9.00997877e-01
1.00646830e+00 3.00520450e-01 -1.63086373e-02 1.19489479e+00
-1.41930655e-01 1.03682709e+00 4.56527054e-01 -5.90300679e-01
4.70945209e-01 4.25211281e-01 -1.50275871e-01 -9.83239055e-01
-4.53328967e-01 -7.51408160e-01 -1.13748920e+00 1.08205199e-01
1.19930074e-01 -1.79550171e-01 -9.89159107e-01 1.55850196e+00
2.67508328e-01 3.28239799e-01 1.91717550e-01 9.53635216e-01
7.34853327e-01 7.42073119e-01 2.90209614e-02 1.66882738e-01
1.02393842e+00 -6.54153824e-01 -1.34951785e-01 -6.55844092e-01
4.26637143e-01 -4.21588957e-01 1.27662551e+00 -1.80321276e-01
-1.74481049e-01 -4.12098467e-01 -8.40605855e-01 3.38997573e-01
-6.13598049e-01 2.29423732e-01 6.49886250e-01 6.37937546e-01
-7.96768248e-01 4.94052917e-01 -5.94685972e-01 -6.78672075e-01
7.65774250e-01 2.99687162e-02 -4.12747353e-01 -4.79677469e-01
-1.16201949e+00 7.85457075e-01 5.04931927e-01 2.39576120e-02
-7.28173554e-01 -9.39882636e-01 -7.90624499e-01 1.70572504e-01
3.96319538e-01 -1.75961405e-01 8.38834584e-01 -1.08954120e+00
-1.25673878e+00 6.02711022e-01 3.27650994e-01 -2.01983243e-01
3.86356324e-01 -3.62409055e-02 -5.74977100e-01 2.19873905e-01
3.15564692e-01 5.11078179e-01 9.64444935e-01 -1.05570054e+00
-6.89204812e-01 -4.43604350e-01 1.14722721e-01 2.94907451e-01
-4.46099997e-01 -1.64667159e-01 4.91603091e-02 -7.47376859e-01
2.78522104e-01 -9.29797530e-01 -2.67054737e-01 2.86887676e-01
1.13647021e-01 2.08289191e-01 9.76573288e-01 -3.58628988e-01
7.93936193e-01 -2.43034267e+00 -2.67446101e-01 3.16670090e-01
-1.65920332e-01 6.37087703e-01 -6.39855802e-01 4.03162122e-01
-1.02832861e-01 4.43923660e-02 -5.48938513e-01 4.29286838e-01
-1.53266266e-01 3.51359665e-01 -4.64516044e-01 6.12067103e-01
3.63188446e-01 6.99629784e-01 -8.58198345e-01 -2.09951431e-01
1.82566553e-01 3.47265780e-01 -2.64221042e-01 -1.30673245e-01
-2.31759530e-02 4.46128160e-01 -8.08560848e-01 6.71442091e-01
1.08482933e+00 -5.14116049e-01 1.37948185e-01 -9.22847912e-02
3.52791585e-02 -1.72163010e-01 -1.21589386e+00 1.49174118e+00
-6.59763694e-01 5.05284667e-01 2.74428707e-02 -1.30300570e+00
1.20325613e+00 -8.91729221e-02 3.22033107e-01 -7.69708753e-01
-7.29247779e-02 1.90509364e-01 -1.14533909e-01 -2.64904588e-01
3.69995326e-01 -3.67442757e-01 -1.00514404e-01 3.74792546e-01
-5.91059774e-02 -1.64390489e-01 -1.19201764e-01 -5.93189448e-02
8.87095034e-01 1.03913650e-01 6.08456731e-01 -4.80559617e-01
3.79809350e-01 2.67067552e-01 5.17646432e-01 1.04693413e+00
-2.16376022e-01 5.83513200e-01 -1.51442155e-01 -2.79118568e-01
-8.10927987e-01 -1.02028000e+00 -5.46379387e-01 1.42515409e+00
3.58444452e-01 2.02523500e-01 -3.20504904e-01 -6.30969465e-01
2.17148542e-01 5.81888199e-01 -5.40690720e-01 -1.27015367e-01
-1.47669360e-01 -1.07698131e+00 5.83522201e-01 5.00460088e-01
9.30698335e-01 -5.69407880e-01 -5.65903127e-01 1.30277857e-01
-2.50387162e-01 -1.05636537e+00 -2.59805143e-01 2.13635176e-01
-9.39664483e-01 -8.96839261e-01 -8.15537214e-01 -5.98703802e-01
5.45234025e-01 1.03046870e+00 7.68936157e-01 -3.85827869e-01
-3.35649610e-01 1.55700684e-01 -5.73419034e-01 -2.09240839e-01
-7.79596493e-02 2.75619626e-01 -2.60762334e-01 1.65362373e-01
4.24117833e-01 -7.43799746e-01 -4.98793155e-01 5.26128113e-01
-1.23735178e+00 -1.63120911e-01 6.96696281e-01 9.59782720e-01
4.42721993e-01 4.97185707e-01 7.06319690e-01 -1.01553822e+00
1.41519919e-01 -7.16036320e-01 -5.81169665e-01 3.15340698e-01
-5.71938753e-01 -1.76236868e-01 5.84374309e-01 -4.26554590e-01
-1.29312778e+00 1.65841758e-01 2.43547216e-01 -2.96592653e-01
-3.92915398e-01 7.80406892e-01 -2.07674354e-02 -3.23802084e-01
9.80673552e-01 3.02124441e-01 8.28886628e-02 -6.98427558e-01
2.85515904e-01 8.50551248e-01 3.57473105e-01 -4.33169782e-01
1.11067545e+00 6.82874024e-01 -1.40829785e-02 -1.01844728e+00
-1.04361367e+00 -7.20991492e-01 -6.38855815e-01 -8.25668126e-02
5.36296368e-01 -1.39935720e+00 5.87310940e-02 6.16670549e-01
-5.01289427e-01 -5.25057137e-01 -3.17985386e-01 6.34232819e-01
-2.37448648e-01 1.03188328e-01 -1.20247394e-01 -5.89696884e-01
-2.10717276e-01 -5.72425485e-01 1.10725987e+00 8.75645354e-02
3.78882170e-01 -1.03466964e+00 -9.46516544e-02 -6.51344657e-02
7.91136324e-01 2.11962119e-01 8.73828590e-01 -6.65799141e-01
-3.75738800e-01 -2.14415491e-01 -6.08283758e-01 6.72597706e-01
4.94076639e-01 -2.50158489e-01 -1.08501327e+00 -5.44052899e-01
-2.78914571e-02 -3.45841914e-01 1.10531020e+00 3.53323698e-01
1.04603362e+00 -1.47027954e-01 -2.20931977e-01 7.96894193e-01
1.96218443e+00 -1.22327439e-01 5.95575154e-01 3.78909111e-01
6.08210444e-01 6.92924917e-01 9.61460471e-01 5.36766291e-01
2.71259874e-01 5.59960127e-01 4.22814965e-01 -3.00668210e-01
-8.80967379e-02 5.56849688e-02 1.62817925e-01 2.26438403e-01
-1.42573431e-01 -8.77439976e-02 -1.14455938e+00 5.00565588e-01
-1.73642027e+00 -1.24562073e+00 6.15146272e-02 1.94362867e+00
6.15737855e-01 -3.17147970e-01 -2.82794684e-01 -3.90243940e-02
8.76226306e-01 4.25630510e-01 -8.10588717e-01 3.81330729e-01
-3.56417984e-01 1.70797855e-01 9.83401239e-01 2.81854540e-01
-1.42058170e+00 1.00810313e+00 5.86788654e+00 1.20441151e+00
-1.53811955e+00 2.53653646e-01 2.92072088e-01 1.80657115e-02
-1.65395170e-01 -3.26215252e-02 -7.43689775e-01 2.94927120e-01
5.82442284e-01 -1.49903461e-01 3.21072489e-01 9.72474396e-01
1.19218610e-01 -6.82744980e-02 -7.77333200e-01 8.69706035e-01
-7.37160519e-02 -1.20847225e+00 1.43257063e-02 -5.67699932e-02
8.22754860e-01 5.92475235e-01 2.95734107e-02 4.36911374e-01
3.76989186e-01 -7.74977565e-01 4.72509623e-01 3.99785280e-01
1.25094497e+00 -3.68491769e-01 6.27227068e-01 5.39078295e-01
-1.26663613e+00 -3.39122742e-01 -8.41625512e-01 -9.10089016e-02
-2.95508951e-01 6.09133542e-01 -7.47680783e-01 8.00693750e-01
7.05349684e-01 1.05733693e+00 -6.38438106e-01 8.82436335e-01
1.20984800e-01 5.64693749e-01 -4.21614587e-01 2.12870151e-01
2.94939697e-01 -3.45682800e-01 4.22946274e-01 1.08048105e+00
5.38657844e-01 2.56520599e-01 5.23801386e-01 6.10359192e-01
2.42018655e-01 8.44942331e-02 -7.81067252e-01 4.91441600e-02
6.35674417e-01 1.27901411e+00 -6.01576209e-01 -2.44274721e-01
-3.76788825e-01 8.33474815e-01 -9.91868507e-03 5.02873182e-01
-7.06200838e-01 -4.82261926e-01 6.92657471e-01 1.97308436e-02
4.75764155e-01 -3.00041828e-02 -6.41737431e-02 -1.35937202e+00
-1.25664830e-01 -8.51974189e-01 4.14737284e-01 -5.99101484e-01
-1.73533475e+00 6.51655555e-01 2.10988805e-01 -1.78143966e+00
1.26260281e-01 -5.41377425e-01 -4.76844788e-01 1.00543153e+00
-2.14240170e+00 -1.32677579e+00 -6.84024632e-01 9.05184627e-01
3.59619677e-01 -3.18108737e-01 9.64565098e-01 4.54715192e-01
-2.45784968e-01 3.75462562e-01 7.02071905e-01 3.34103182e-02
6.67627573e-01 -8.94815266e-01 1.70322329e-01 9.88925695e-01
-2.16819897e-01 2.98751682e-01 4.08783019e-01 -3.84676486e-01
-1.09488046e+00 -1.82401550e+00 5.08551121e-01 1.30364701e-01
8.42172027e-01 -1.78684309e-01 -1.06400752e+00 3.69007498e-01
-5.15063584e-01 4.34640378e-01 8.49750698e-01 -1.08775325e-01
-6.82998180e-01 -6.37578845e-01 -1.28260875e+00 5.39642125e-02
1.04923511e+00 -8.33790779e-01 -3.70101869e-01 4.25518423e-01
2.56059408e-01 -1.57548320e-02 -8.97329926e-01 4.37698215e-01
2.68163860e-01 -6.13385379e-01 8.98741722e-01 -3.75287265e-01
2.36348137e-01 -3.77791375e-01 -7.83451438e-01 -1.36612833e+00
-5.99757075e-01 6.08931892e-02 6.90348744e-01 1.03678405e+00
2.07241967e-01 -9.95453179e-01 3.86480778e-01 2.05135733e-01
-8.68763179e-02 -6.55686855e-02 -8.48630667e-01 -1.34839535e+00
1.56776339e-01 -1.53614312e-01 7.75909126e-01 1.24378228e+00
-5.46935558e-01 2.19894215e-01 -3.92116994e-01 6.41160429e-01
5.71954966e-01 6.93901360e-01 7.42389083e-01 -1.62988365e+00
-4.12122667e-01 -2.22582757e-01 -3.20327014e-01 -9.23098385e-01
7.62570351e-02 -1.16868341e+00 2.00312644e-01 -1.41016555e+00
3.00876766e-01 -8.32539856e-01 -3.63279819e-01 8.62983644e-01
-6.55557141e-02 4.92731810e-01 2.62680054e-01 5.34589708e-01
-1.81590095e-01 6.35026991e-01 1.16660559e+00 -3.26529324e-01
-1.41991705e-01 1.85684994e-01 -7.31837630e-01 4.67176378e-01
9.45525646e-01 -5.56790709e-01 -5.38089216e-01 -5.61863959e-01
-1.41707480e-01 -1.36466190e-01 6.02577746e-01 -1.09205842e+00
-3.36331539e-02 -5.20018876e-01 1.55046165e-01 -4.12227094e-01
1.93083733e-01 -8.72984290e-01 2.70842344e-01 2.53216416e-01
-1.76244408e-01 -6.72652900e-01 8.09974149e-02 6.52917027e-01
-3.28855664e-01 -2.44568959e-01 1.09510756e+00 -8.65067095e-02
-1.21688676e+00 6.54991925e-01 -1.82484210e-01 -6.60518482e-02
9.37094033e-01 -3.15807939e-01 -4.66114819e-01 -8.99953768e-02
-5.77670455e-01 5.66312633e-02 3.84824276e-01 3.59457463e-01
3.52737278e-01 -1.24655199e+00 -9.67023313e-01 3.49526525e-01
4.58384633e-01 2.83420202e-03 4.77117419e-01 6.56054020e-01
-2.56847352e-01 1.69496030e-01 -3.84121388e-01 -7.08056390e-01
-1.12868702e+00 1.47067145e-01 4.93778855e-01 8.59004073e-03
-6.38298929e-01 6.06698692e-01 4.26190287e-01 -8.04636836e-01
-2.90093035e-01 -3.81578915e-02 -1.28562599e-01 2.08040774e-01
5.93766809e-01 3.65038455e-01 6.00566640e-02 -6.76746547e-01
-4.68619883e-01 7.10665166e-01 5.43812029e-02 3.32579345e-01
1.80286479e+00 -2.58527398e-01 -3.81467491e-02 4.02790219e-01
1.18842494e+00 -3.55251491e-01 -1.58418941e+00 -8.63814533e-01
-2.11991161e-01 -7.42050886e-01 3.46804082e-01 -8.61032963e-01
-1.18316007e+00 8.75779331e-01 9.21535790e-01 1.40410677e-01
1.43022811e+00 -3.57120819e-02 2.32628599e-01 7.13020325e-01
4.89448696e-01 -1.16740751e+00 -1.99551597e-01 7.39661396e-01
6.90763891e-01 -1.44652486e+00 -4.45423201e-02 -4.75186974e-01
-5.73454440e-01 1.01032543e+00 3.76941264e-01 -1.11865453e-01
9.22011554e-01 1.10536935e-02 1.46482617e-01 -1.43469080e-01
-3.18272501e-01 -5.15134335e-01 2.10097551e-01 8.77916098e-01
1.51592877e-03 2.69497633e-01 -1.86617449e-02 2.39986420e-01
1.41039521e-01 1.86842140e-02 2.83276767e-01 9.34610844e-01
-6.47392213e-01 -9.03105795e-01 -2.33304352e-01 6.72332346e-01
-1.74693182e-01 -3.40397656e-01 6.93851635e-02 7.25024462e-01
2.94400509e-02 8.83066177e-01 1.60856843e-01 -1.89278618e-01
2.94270366e-01 -2.38578677e-01 1.47814512e-01 -8.30203414e-01
-2.73901403e-01 3.88474055e-02 -1.25258684e-01 -4.76489753e-01
-7.86056161e-01 -6.69588089e-01 -9.46610749e-01 -3.47072363e-01
-3.84483129e-01 -4.36931737e-02 6.24104798e-01 8.50812435e-01
4.71470088e-01 2.51200974e-01 1.01833832e+00 -7.08596408e-01
-1.12343383e+00 -9.20780301e-01 -1.19385660e+00 2.51409978e-01
3.72493565e-01 -5.96791387e-01 -2.86954343e-01 4.48963931e-03] | [9.670011520385742, -1.3688151836395264] |
ab619110-5729-4ca6-8e94-4ecd3726c668 | the-ability-of-image-language-explainable | 2209.09310 | null | https://arxiv.org/abs/2209.09310v1 | https://arxiv.org/pdf/2209.09310v1.pdf | The Ability of Image-Language Explainable Models to Resemble Domain Expertise | Recent advances in vision and language (V+L) models have a promising impact in the healthcare field. However, such models struggle to explain how and why a particular decision was made. In addition, model transparency and involvement of domain expertise are critical success factors for machine learning models to make an entrance into the field. In this work, we study the use of the local surrogate explainability technique to overcome the problem of black-box deep learning models. We explore the feasibility of resembling domain expertise using the local surrogates in combination with an underlying V+L to generate multi-modal visual and language explanations. We demonstrate that such explanations can serve as helpful feedback in guiding model training for data scientists and machine learning engineers in the field. | ['Ujjwal Ratan', 'Anna Zapaishchykova', 'Petrus Werner'] | 2022-09-19 | null | null | null | null | ['explainable-models'] | ['computer-vision'] | [ 1.71931200e-02 6.73322618e-01 -2.16098636e-01 -5.12719810e-01
-4.46295142e-01 -3.88500869e-01 5.30772984e-01 3.14263672e-01
-1.37631312e-01 5.87005794e-01 3.21668088e-01 -8.28890026e-01
-6.57187626e-02 -3.58377188e-01 -7.07475185e-01 -2.82240629e-01
3.84720415e-01 5.92786312e-01 -4.05656308e-01 -1.96180400e-02
2.17432380e-01 4.51762736e-01 -1.05117142e+00 6.83401167e-01
1.11302912e+00 5.06491899e-01 2.25740463e-01 4.40643251e-01
-4.07457441e-01 1.03484404e+00 -4.03149724e-01 -3.36318970e-01
4.85934950e-02 -3.87541562e-01 -6.71822190e-01 7.36464113e-02
2.68991619e-01 -2.74774641e-01 1.48000181e-01 8.46593261e-01
2.83914268e-01 -2.50199556e-01 7.85960197e-01 -1.38334417e+00
-1.10011947e+00 4.60121155e-01 -4.45382565e-01 -2.24153604e-02
3.09106469e-01 4.92437422e-01 7.56125569e-01 -9.65969205e-01
7.35264897e-01 1.16313064e+00 7.32248306e-01 1.03845584e+00
-1.35790133e+00 -4.20502841e-01 3.14759701e-01 2.28532583e-01
-1.05510318e+00 -9.94349495e-02 7.28393376e-01 -8.49215031e-01
8.68442237e-01 1.40927777e-01 9.54856575e-01 1.12705064e+00
2.26454496e-01 8.09327364e-01 1.30855322e+00 -6.40973866e-01
1.86500251e-01 8.56154144e-01 2.19301179e-01 1.09486413e+00
4.61449176e-01 3.74253690e-01 -5.98901510e-01 -5.91546893e-02
1.07344902e+00 9.29569751e-02 -3.87680918e-01 -5.17597616e-01
-1.00968862e+00 1.12536156e+00 8.56973827e-01 2.79239744e-01
-5.74969530e-01 2.84907132e-01 -1.03472114e-01 1.02806740e-01
3.07746351e-01 8.80140781e-01 -3.09841990e-01 2.13649571e-01
-8.48883092e-01 2.43192136e-01 4.03232664e-01 4.39175844e-01
5.71217060e-01 1.27190769e-01 -3.85076463e-01 3.30879778e-01
7.18011081e-01 3.85650128e-01 1.89094380e-01 -1.05695748e+00
8.55603218e-02 9.58300412e-01 2.25780308e-01 -8.18511546e-01
-4.83143896e-01 -4.81132507e-01 -5.50288200e-01 7.34529853e-01
4.15470183e-01 2.98806597e-02 -1.19197345e+00 1.54547310e+00
2.71298856e-01 -2.59440076e-02 2.70750821e-01 1.08789337e+00
1.12935162e+00 4.36146498e-01 4.61276740e-01 2.36739963e-01
1.22112644e+00 -1.07460070e+00 -7.28321612e-01 -5.05028069e-01
8.63973558e-01 -5.31264663e-01 1.37272072e+00 2.50831962e-01
-1.06535459e+00 -6.14486337e-01 -7.06609368e-01 -1.63651064e-01
-1.82692200e-01 2.36826763e-01 7.25749314e-01 3.09134573e-01
-1.20071733e+00 4.05377835e-01 -6.90342069e-01 -4.71792161e-01
7.21111000e-01 2.60242224e-01 -4.00960475e-01 -1.92263499e-01
-8.85962844e-01 1.25386298e+00 7.61426687e-02 2.74986029e-03
-9.04372156e-01 -9.13228750e-01 -8.50546777e-01 -1.04417400e-02
-1.25531945e-02 -1.36544824e+00 1.29789138e+00 -1.39264119e+00
-9.27880168e-01 1.00461400e+00 -2.37254009e-01 -6.07243836e-01
7.82435775e-01 -1.33336648e-01 -2.91684251e-02 1.01349533e-01
5.80981113e-02 1.03276527e+00 7.22249746e-01 -1.72658455e+00
-4.16737378e-01 -2.93401420e-01 2.56536573e-01 1.10209256e-01
-5.30851595e-02 -3.11731666e-01 -1.95195258e-01 -3.83725613e-01
-1.72597080e-01 -6.29067242e-01 -6.64126039e-01 4.69447225e-01
-3.71540368e-01 -5.20193800e-02 6.79429412e-01 -8.12488317e-01
9.79302049e-01 -1.87989831e+00 -2.87134480e-02 3.40289772e-01
7.97363281e-01 4.79185760e-01 -9.28899199e-02 4.01481003e-01
8.54939688e-03 3.36747110e-01 1.01885870e-01 -1.79613546e-01
-1.48311034e-01 3.31285417e-01 -2.06558540e-01 -9.81411245e-03
3.36713523e-01 1.30716622e+00 -8.24269176e-01 -6.11098647e-01
3.06052536e-01 8.52123857e-01 -5.96771002e-01 2.40808249e-01
-3.41362953e-01 9.74552870e-01 -5.76248884e-01 2.45501488e-01
3.16892952e-01 -7.74923265e-01 -4.15065745e-03 1.85361486e-02
-2.35685892e-02 1.16542600e-01 -5.81827998e-01 1.33747494e+00
-6.74871624e-01 7.90376604e-01 1.35540903e-01 -6.99469447e-01
7.28626549e-01 4.18304861e-01 1.94205754e-02 -6.36466682e-01
-5.02662659e-02 2.74822891e-01 1.29425615e-01 -8.51226211e-01
-2.30558142e-02 -6.60508513e-01 6.11524403e-01 5.30612767e-01
-4.12642241e-01 -6.36111423e-02 -2.37676367e-01 4.24802363e-01
6.67386055e-01 1.35675848e-01 4.12863016e-01 -1.38030360e-02
3.03223014e-01 5.53502321e-01 2.32000768e-01 8.42312574e-01
-1.84713513e-01 6.82453156e-01 3.50096583e-01 -8.37550521e-01
-9.55060780e-01 -7.22357988e-01 2.31652498e-01 6.40659153e-01
-1.26499563e-01 -2.22698525e-01 -9.72662628e-01 -9.39246714e-01
1.57049417e-01 1.00709844e+00 -9.47449148e-01 -2.61739671e-01
-2.48078167e-01 -1.13877118e-01 1.71672404e-01 8.80435765e-01
9.45407674e-02 -1.20297134e+00 -1.14813781e+00 6.65977225e-02
-8.45443159e-02 -9.25129235e-01 -1.40174225e-01 -1.75225409e-03
-9.37209249e-01 -1.08542407e+00 -9.85177338e-01 -5.85373878e-01
1.24714315e+00 2.94937700e-01 1.19487011e+00 6.39520526e-01
-3.01885843e-01 7.12801754e-01 -2.46042386e-01 -8.81759226e-01
-9.24932241e-01 -3.59750330e-01 -2.65222728e-01 -2.56875157e-01
3.10161442e-01 -4.37275320e-03 -7.24009454e-01 4.27276008e-02
-8.15245330e-01 8.55356038e-01 5.33878088e-01 1.03164840e+00
3.73896211e-01 -7.07264900e-01 4.40142661e-01 -1.12318707e+00
9.86494839e-01 -1.98452517e-01 -2.80306727e-01 4.80686337e-01
-9.61029470e-01 2.95995712e-01 5.11231661e-01 -3.77062201e-01
-9.29371834e-01 8.60767886e-02 3.20301056e-02 -5.66224515e-01
-1.36796489e-01 8.83417189e-01 3.55965316e-01 -1.96141958e-01
9.36594665e-01 -1.38426647e-01 2.14744762e-01 -4.58667248e-01
5.24704278e-01 4.43104446e-01 1.88097805e-01 -2.53596127e-01
7.25443661e-01 5.28088212e-01 -1.93417668e-01 -3.85728896e-01
-8.31552982e-01 -1.16103828e-01 -3.64363760e-01 -2.47636512e-01
1.01133764e+00 -9.33661580e-01 -3.81839246e-01 -2.00867891e-01
-1.46656537e+00 -4.13195252e-01 -4.11444336e-01 3.96031469e-01
-3.64726186e-01 2.84780357e-02 -6.65592104e-02 -8.63616526e-01
-2.05354363e-01 -1.16709387e+00 9.09591854e-01 4.75775421e-01
-7.67018378e-01 -1.33751905e+00 -1.32810399e-01 7.66827941e-01
5.27930975e-01 2.66605526e-01 1.30894017e+00 -6.95728540e-01
-6.46092057e-01 -7.67308474e-02 -4.01890278e-01 8.50206614e-02
-9.31461621e-03 -5.65015562e-02 -1.06432164e+00 -6.59891218e-02
-1.13826342e-01 -2.54416585e-01 7.45151401e-01 6.57132924e-01
1.02178490e+00 -4.22865301e-01 -5.30785441e-01 4.74453688e-01
1.22370827e+00 1.03936456e-01 1.59570828e-01 3.75095725e-01
8.83450687e-01 9.92952943e-01 5.03314376e-01 2.27951363e-01
6.07759833e-01 5.16146302e-01 4.69127804e-01 -9.57963347e-01
-3.30586910e-01 -4.17062193e-01 3.28796618e-02 3.14286202e-01
-6.31084666e-02 2.18022734e-01 -1.54005444e+00 6.07533514e-01
-2.12504530e+00 -6.25783384e-01 -3.38567317e-01 1.80630326e+00
6.03150010e-01 -4.14926596e-02 -1.60353422e-01 -1.69036627e-01
3.90107244e-01 -3.25157225e-01 -5.06998360e-01 -6.59932852e-01
4.30142768e-02 -7.53570348e-02 1.64516196e-01 6.96974814e-01
-7.31699646e-01 9.09082651e-01 6.63702202e+00 2.47080877e-01
-1.18323910e+00 8.25729594e-02 8.02329063e-01 1.27011552e-01
-9.58114445e-01 8.53418857e-02 -3.45081240e-01 -5.10205626e-02
5.93225002e-01 -4.62287217e-02 2.63095081e-01 8.60535085e-01
7.00494587e-01 -6.53784499e-02 -1.34061754e+00 9.36274171e-01
-6.58166632e-02 -1.84046650e+00 4.28821623e-01 1.63831178e-03
8.48342717e-01 -3.45886528e-01 3.68456453e-01 -1.10779837e-01
3.79755944e-01 -1.57023430e+00 7.13650703e-01 6.92369103e-01
6.01563632e-01 -4.00950611e-01 5.94308913e-01 3.86277527e-01
-6.25297427e-01 -2.17689663e-01 1.80874281e-02 -2.36934870e-01
1.61485583e-01 2.50752270e-01 -1.44020712e+00 3.30500722e-01
5.75940192e-01 4.73481566e-01 -4.85443264e-01 8.87115359e-01
-5.28253078e-01 4.64520007e-01 1.61060184e-01 6.07151650e-02
3.65597785e-01 9.87037718e-02 2.89449602e-01 1.06885767e+00
1.80141315e-01 -2.96768360e-02 -1.64068229e-02 1.41396356e+00
2.67353535e-01 4.21606638e-02 -9.35536385e-01 -4.17454809e-01
-1.56382360e-02 8.95576894e-01 -5.82645535e-01 -3.15245390e-01
-4.41561818e-01 7.18290925e-01 3.86480808e-01 6.53589427e-01
-6.48566365e-01 2.21762300e-01 5.95176816e-01 4.68896598e-01
2.27003489e-02 6.26235968e-03 -9.52374458e-01 -8.13067138e-01
-1.38361290e-01 -1.19319475e+00 1.68686658e-01 -1.20403326e+00
-1.17591238e+00 5.75480044e-01 -2.73314863e-01 -9.88369703e-01
-2.64601380e-01 -8.03718984e-01 -6.05056107e-01 1.11193204e+00
-1.78420138e+00 -1.60106289e+00 -3.69076788e-01 4.43526864e-01
4.01107669e-01 -2.06093147e-01 8.76691997e-01 -2.44053483e-01
-2.48660088e-01 2.31327578e-01 -1.03884272e-01 -1.35929808e-01
5.03898621e-01 -1.21399820e+00 4.24417287e-01 4.85246092e-01
4.37246889e-01 1.01743376e+00 1.02112269e+00 -7.44404912e-01
-9.36252773e-01 -9.38133240e-01 1.03597951e+00 -7.03497887e-01
2.35563532e-01 -5.63429184e-02 -9.82155800e-01 6.88055277e-01
3.02177072e-01 -1.46299541e-01 7.24300981e-01 6.74124516e-04
-3.07812721e-01 3.04482609e-01 -1.15436780e+00 7.87853122e-01
6.80316925e-01 -6.02568150e-01 -6.18527591e-01 2.97080368e-01
5.34451485e-01 -2.69766927e-01 -2.58543491e-01 3.26955289e-01
4.42031562e-01 -1.02525723e+00 9.77994144e-01 -1.00825679e+00
6.53618515e-01 -2.32109204e-01 2.59035021e-01 -1.36463058e+00
-2.70710051e-01 -4.57573444e-01 6.06846390e-03 7.48510122e-01
7.47270525e-01 -5.47669411e-01 7.48179615e-01 1.13665962e+00
1.11511769e-02 -9.08269942e-01 -6.66463077e-01 -2.52661616e-01
1.21822953e-01 -2.66163260e-01 3.64996314e-01 9.99983013e-01
-7.31231421e-02 1.79662645e-01 -3.58322978e-01 1.32002130e-01
3.01376641e-01 1.18886121e-01 7.67810106e-01 -1.41304207e+00
-2.65408963e-01 -4.71549928e-01 -9.00230631e-02 -7.69259453e-01
-6.40453994e-02 -9.04879510e-01 -2.63969362e-01 -2.43236852e+00
4.18797731e-01 -3.33038539e-01 -2.61204481e-01 7.64414370e-01
-3.32853258e-01 -1.99488439e-02 4.85858649e-01 2.63987213e-01
-2.25256816e-01 1.16165742e-01 1.53145003e+00 -1.38962686e-01
-1.79235801e-01 -1.64026052e-01 -1.31299961e+00 8.62676978e-01
7.88516164e-01 -5.46088040e-01 -5.95802069e-01 -7.18132198e-01
3.25208306e-01 -4.18413281e-02 8.82714629e-01 -5.34898996e-01
1.20042339e-01 -3.28262240e-01 4.88889128e-01 1.17012486e-02
2.58340299e-01 -1.04699826e+00 2.14864343e-01 8.19375813e-01
-5.50907433e-01 1.36484310e-01 4.30283278e-01 4.72851306e-01
-1.90435365e-01 -7.83359557e-02 6.13447487e-01 -5.13245821e-01
-6.46968305e-01 -1.40715346e-01 -3.39122981e-01 -1.61042452e-01
1.07975960e+00 -5.66331863e-01 -3.50525051e-01 -6.90278053e-01
-8.58440995e-01 4.56472605e-01 6.82988942e-01 3.84526998e-01
9.64312911e-01 -1.12874198e+00 -6.89687550e-01 2.91655630e-01
2.55580068e-01 -2.94977099e-01 2.50811875e-01 8.90515506e-01
-7.00363040e-01 6.66016579e-01 -3.22247416e-01 -5.41999698e-01
-1.38532603e+00 3.83542448e-01 5.11911929e-01 -2.87978023e-01
-6.81244433e-01 9.78079200e-01 7.55087495e-01 -4.29424942e-01
3.32569629e-01 -3.18484813e-01 -3.67503256e-01 -3.35617870e-01
5.07421732e-01 -2.93865912e-02 -4.40141886e-01 -3.53561640e-01
-4.21437621e-01 5.20803094e-01 -9.13164243e-02 -3.50427210e-01
1.32222712e+00 -3.49833593e-02 3.07460427e-01 4.42168683e-01
5.68560660e-01 -2.92656720e-01 -1.43520391e+00 -1.76362897e-04
7.95424059e-02 -4.42215025e-01 -1.65522434e-02 -1.48107040e+00
-8.98468852e-01 1.51278698e+00 6.61550701e-01 9.61600021e-02
8.04580510e-01 2.90953934e-01 3.81604850e-01 1.99083593e-02
9.24493968e-02 -8.42038155e-01 3.05866033e-01 2.66779959e-02
1.15410554e+00 -1.55255187e+00 -1.19790025e-01 -2.07888454e-01
-1.25803685e+00 1.01146197e+00 7.00992644e-01 2.71316588e-01
3.60175222e-01 -2.46868096e-02 1.04204667e+00 -5.92290699e-01
-6.83735728e-01 -1.41717032e-01 6.91247225e-01 9.15247858e-01
6.53267026e-01 6.75319359e-02 -1.49453476e-01 6.31697297e-01
1.73102617e-01 2.86157340e-01 4.27585185e-01 7.26460636e-01
-3.52771908e-01 -1.13174832e+00 -4.49808896e-01 2.50764132e-01
-2.75164306e-01 -1.91470280e-01 -7.68624306e-01 7.51739681e-01
1.04027309e-01 9.19741631e-01 -2.76454300e-01 -1.75559014e-01
2.24809647e-01 1.72694176e-01 2.83662677e-01 -7.17493296e-01
-8.51182282e-01 -3.02457325e-02 5.60408980e-02 -4.43763822e-01
-2.93911278e-01 -2.45904103e-01 -1.55093670e+00 4.10491861e-02
-1.54078491e-02 1.26819044e-01 7.36119866e-01 1.04211366e+00
5.77361226e-01 5.73251128e-01 -2.84210034e-02 -5.18835843e-01
-4.67674553e-01 -6.00036681e-01 -1.26425266e-01 5.32817662e-01
6.65625811e-01 -4.70021486e-01 -1.11234210e-01 1.63114399e-01] | [8.935934066772461, 5.5338239669799805] |
0131001b-b052-4b6e-897b-c18c6370fec3 | deductive-verification-of-chain-of-thought | 2306.03872 | null | https://arxiv.org/abs/2306.03872v2 | https://arxiv.org/pdf/2306.03872v2.pdf | Deductive Verification of Chain-of-Thought Reasoning | Large Language Models (LLMs) significantly benefit from Chain-of-Thought (CoT) prompting in performing various reasoning tasks. While CoT allows models to produce more comprehensive reasoning processes, its emphasis on intermediate reasoning steps can inadvertently introduce hallucinations and accumulated errors, thereby limiting models' ability to solve complex reasoning tasks. Inspired by how humans engage in careful and meticulous deductive logical reasoning processes to solve tasks, we seek to enable language models to perform explicit and rigorous deductive reasoning, and also ensure the trustworthiness of their reasoning process through self-verification. However, directly verifying the validity of an entire deductive reasoning process is challenging, even with advanced models like ChatGPT. In light of this, we propose to decompose a reasoning verification process into a series of step-by-step subprocesses, each only receiving their necessary context and premises. To facilitate this procedure, we propose Natural Program, a natural language-based deductive reasoning format. Our approach enables models to generate precise reasoning steps where subsequent steps are more rigorously grounded on prior steps. It also empowers language models to carry out reasoning self-verification in a step-by-step manner. By integrating this verification process into each deductive reasoning stage, we significantly enhance the rigor and trustfulness of generated reasoning steps. Along this process, we also improve the answer correctness on complex reasoning tasks. Code will be released at https://github.com/lz1oceani/verify_cot. | ['Hao Su', 'Roland Memisevic', 'Mingu Lee', 'Zhiao Huang', 'Xuanlin Li', 'Yunhao Fang', 'Zhan Ling'] | 2023-06-06 | null | null | null | null | ['logical-reasoning'] | ['reasoning'] | [-2.50855219e-02 8.01081240e-01 8.68778825e-02 -4.03306365e-01
-3.09330344e-01 -6.24508977e-01 8.23592186e-01 1.88824102e-01
2.06111534e-03 6.43155575e-01 2.58663356e-01 -9.33143795e-01
-1.36318430e-01 -1.02035320e+00 -3.79497498e-01 1.20991752e-01
4.57392246e-01 4.39828336e-01 6.85099736e-02 -3.26200396e-01
4.79058981e-01 5.13710856e-01 -1.10126507e+00 5.46442270e-01
1.13675284e+00 3.05673063e-01 -8.74447823e-02 4.81169909e-01
-3.07914972e-01 2.22303796e+00 -2.02551663e-01 -8.34377348e-01
-3.23994756e-02 -5.66311419e-01 -1.49794877e+00 -8.52990970e-02
-1.84786513e-01 -5.61390579e-01 7.33865052e-02 1.20749831e+00
-1.66937336e-01 -5.70002086e-02 3.79159123e-01 -1.41670060e+00
-8.77505243e-01 1.00652599e+00 3.73127051e-02 -2.73076504e-01
1.02957022e+00 6.17844284e-01 8.30792427e-01 -6.75448954e-01
5.16631722e-01 1.40214205e+00 7.51727164e-01 6.89290464e-01
-1.14424312e+00 -6.57393157e-01 8.82847384e-02 3.91421169e-01
-1.21845698e+00 -5.37045658e-01 7.33032048e-01 -5.08452535e-01
1.01912177e+00 4.24358428e-01 8.19118381e-01 1.04987431e+00
3.96452844e-01 6.46779478e-01 1.39986026e+00 -5.48182428e-01
5.18303454e-01 5.26071668e-01 3.25331330e-01 8.20816875e-01
3.49705905e-01 -7.45314509e-02 -3.94541502e-01 -2.09421396e-01
7.26964235e-01 5.28461523e-02 -2.31469631e-01 -1.35881111e-01
-1.11846220e+00 6.59927607e-01 3.22878778e-01 4.15885508e-01
-6.67566955e-01 3.75332125e-02 2.09272996e-01 4.83254790e-01
-1.35009959e-01 7.93633282e-01 -1.51683792e-01 -1.69536442e-01
-9.01416302e-01 5.84698617e-01 1.22021532e+00 7.34251380e-01
5.99974155e-01 -1.99758142e-01 -2.93685645e-01 2.11912960e-01
6.62756324e-01 3.56947899e-01 2.43408382e-01 -1.37216246e+00
5.92018627e-02 1.10676146e+00 4.19293165e-01 -1.01293993e+00
-1.82094857e-01 -2.07013920e-01 -6.22648239e-01 4.54434544e-01
3.48183900e-01 4.67138626e-02 -2.72545129e-01 1.56735301e+00
4.67502445e-01 -2.90163040e-01 2.95521915e-01 8.50642383e-01
3.75428885e-01 3.11108947e-01 3.32796991e-01 -2.89060384e-01
1.57260013e+00 -9.69151258e-01 -8.59089255e-01 -1.02294661e-01
8.76144052e-01 -4.67871785e-01 1.37070990e+00 5.85107803e-01
-1.50764179e+00 -2.63107210e-01 -7.62824416e-01 -3.17075998e-01
-1.19824745e-02 -3.28698039e-01 1.02376521e+00 3.80997032e-01
-1.15437770e+00 4.56146955e-01 -6.72527611e-01 -6.45565689e-02
3.77704620e-01 -6.41221926e-02 -2.16017649e-01 -1.71548381e-01
-1.46240115e+00 1.27098775e+00 3.14769626e-01 3.58083546e-01
-9.20568645e-01 -6.61076009e-01 -8.70803058e-01 1.16994701e-01
7.25419164e-01 -1.03231621e+00 1.72688544e+00 -8.06757867e-01
-1.52125347e+00 7.75885522e-01 -3.18030298e-01 -5.85907221e-01
1.01273906e+00 -1.62073597e-01 -2.43579552e-01 3.59849542e-01
2.80120701e-01 4.02155578e-01 3.84213746e-01 -1.35078585e+00
-3.76280956e-02 -1.06572382e-01 6.15418553e-01 -5.76438345e-02
1.94339916e-01 2.39734694e-01 1.55743152e-01 -3.47691998e-02
9.97074470e-02 -6.20943785e-01 -3.15164447e-01 2.35112756e-01
-3.37432325e-01 -3.00024152e-01 1.46904960e-01 -6.18437290e-01
1.24187076e+00 -1.85508370e+00 -2.24383801e-01 6.62629232e-02
7.83766448e-01 3.38189662e-01 2.37626180e-01 8.28825057e-01
-7.26542100e-02 3.35261106e-01 1.29473582e-01 -1.08528502e-01
5.10619342e-01 9.25236032e-04 -5.80993831e-01 -6.90836012e-02
3.68612707e-01 1.29652119e+00 -9.92810547e-01 -6.29746795e-01
1.47434011e-01 2.28114113e-01 -7.05686986e-01 4.28120852e-01
-4.45814878e-01 2.80326128e-01 -5.54661691e-01 4.73043263e-01
6.13572836e-01 -6.39877141e-01 3.21076870e-01 2.52426445e-01
-1.07842181e-02 7.41015792e-01 -1.01638019e+00 1.25210702e+00
-6.19519949e-01 1.56439066e-01 3.74631723e-03 -4.68793124e-01
8.34769249e-01 6.29877269e-01 -2.73316264e-01 -6.04605496e-01
5.72188944e-02 1.29202440e-01 1.88738868e-01 -8.29552650e-01
2.72548676e-01 -7.84064412e-01 1.67252719e-01 9.88506377e-01
-3.59966248e-01 -5.84779859e-01 2.33027473e-01 7.74259806e-01
1.00664258e+00 1.91996247e-01 6.85335279e-01 -3.55350487e-02
8.21268678e-01 4.17381048e-01 4.27911669e-01 7.52768397e-01
-2.97351539e-01 -2.55102348e-02 7.36468554e-01 -4.88749415e-01
-8.32036734e-01 -9.25030828e-01 2.32417554e-01 3.46842676e-01
-1.05875231e-01 -6.40746295e-01 -8.40097785e-01 -5.24493992e-01
-3.19799662e-01 1.54065776e+00 -4.63076085e-01 -3.88561398e-01
-8.85504708e-02 4.35693227e-02 8.18331778e-01 3.78787696e-01
8.16309214e-01 -1.41578209e+00 -1.01909959e+00 1.65195793e-01
-3.18653196e-01 -9.43523467e-01 2.43437022e-01 -1.79947183e-01
-7.22735763e-01 -1.18859017e+00 9.29046720e-02 -2.29107559e-01
7.37961471e-01 1.55894801e-01 9.90552783e-01 7.66753435e-01
2.59702682e-01 3.43902946e-01 -4.18855697e-01 -3.68721992e-01
-1.05501163e+00 -5.12022614e-01 -7.38122314e-02 -3.99661362e-01
5.20404935e-01 -6.62030518e-01 -2.08379388e-01 9.45245475e-02
-9.20227349e-01 8.25070918e-01 3.20335686e-01 3.80276680e-01
-1.52356625e-01 3.43451917e-01 2.74636954e-01 -8.98770869e-01
1.18768525e+00 -3.62912744e-01 -3.79371405e-01 5.35317719e-01
-9.05724108e-01 1.59226984e-01 9.89675879e-01 -2.50815332e-01
-1.37532568e+00 -6.43807352e-01 3.91449556e-02 -1.65167361e-01
-2.99031347e-01 6.31576538e-01 1.80024162e-01 1.51498094e-01
8.69617820e-01 2.54215091e-01 1.55649647e-01 1.32384166e-01
2.85836756e-01 4.80977744e-01 3.52791548e-01 -1.01574767e+00
1.12092304e+00 2.22586155e-01 -3.59653801e-01 -2.01362967e-02
-9.38925028e-01 7.97899812e-02 -4.34047371e-01 -3.84534955e-01
5.43281078e-01 -8.35462570e-01 -1.36773467e+00 1.06365688e-01
-1.31039262e+00 -7.86726832e-01 -2.93185115e-01 3.49295855e-01
-5.86795926e-01 5.04065871e-01 -8.24982703e-01 -1.39793897e+00
-4.24771726e-01 -8.88040900e-01 4.90733773e-01 2.00157270e-01
-1.01673770e+00 -1.05170763e+00 -3.96529958e-02 7.41344273e-01
4.23701704e-01 -1.84518814e-01 1.06656551e+00 -7.28942513e-01
-5.94080865e-01 -1.80492446e-01 -1.79495394e-01 2.44120613e-01
-1.69632912e-01 5.88574521e-02 -8.41584146e-01 3.42435151e-01
6.40979588e-01 -7.00553894e-01 -3.24738994e-02 -4.11795646e-01
6.57244265e-01 -6.60666168e-01 1.48336902e-01 9.39180106e-02
1.02139354e+00 1.21442378e-01 8.20775568e-01 4.19071019e-01
2.49122247e-01 7.38339126e-01 6.41207218e-01 2.64116406e-01
8.30952406e-01 6.84622601e-02 -8.86933282e-02 6.56059161e-02
6.02534302e-02 -6.82408988e-01 4.37758863e-01 5.26433408e-01
-6.90141991e-02 3.98456365e-01 -1.33102286e+00 3.07935715e-01
-1.86644208e+00 -1.36122787e+00 -4.39988226e-01 1.70426250e+00
1.26397300e+00 2.05519512e-01 -3.26647013e-01 4.07972604e-01
2.97003776e-01 -2.09124550e-01 -2.20161587e-01 -7.90359974e-01
4.22240555e-01 -7.93309286e-02 -5.16858220e-01 8.47688794e-01
-3.91913578e-03 1.12794363e+00 5.92784262e+00 3.46598476e-01
-8.55948746e-01 -1.04568571e-01 3.16797554e-01 2.25406379e-01
-9.78300095e-01 5.01588821e-01 -2.88485616e-01 -2.40241475e-02
8.78460705e-01 -3.59409600e-01 8.38108540e-01 6.39948964e-01
6.51230812e-01 -4.75128889e-01 -1.31123745e+00 5.18031657e-01
-1.73124686e-01 -1.31707931e+00 2.68704534e-01 -3.58165324e-01
2.51978099e-01 -6.67537689e-01 -5.16351163e-01 6.05733097e-01
6.25083148e-01 -1.06209624e+00 1.07227910e+00 6.18146658e-01
7.24269450e-02 -3.56068879e-01 6.13345861e-01 9.49511945e-01
-6.48624182e-01 -1.04494229e-01 -4.89914455e-02 -9.64930177e-01
1.97414920e-01 7.55237281e-01 -1.05531621e+00 5.54291248e-01
2.59633005e-01 2.04039618e-01 -5.72346807e-01 5.31864941e-01
-9.81411099e-01 3.69143367e-01 1.94679722e-01 -1.95670679e-01
4.40726429e-02 -3.37569207e-01 1.83066189e-01 8.99475873e-01
-2.03658149e-01 5.21407545e-01 -8.41745734e-02 1.68427253e+00
3.08121234e-01 -1.80456758e-01 -5.71061671e-01 -2.28599697e-01
5.24140120e-01 1.14690220e+00 -6.37170136e-01 -7.15412855e-01
-9.87864286e-02 7.90987551e-01 5.69250226e-01 4.62663502e-01
-9.04824078e-01 -1.47527847e-02 2.47443974e-01 1.79316193e-01
-3.56188655e-01 -2.19857037e-01 -6.82867467e-01 -1.40281820e+00
1.88071385e-01 -1.37290645e+00 -4.69572656e-02 -1.58591640e+00
-1.06276476e+00 4.86285388e-01 1.12265140e-01 -6.47209704e-01
-3.02405149e-01 -3.52719128e-01 -7.21966445e-01 1.02559018e+00
-1.46166456e+00 -1.22975278e+00 -3.01263392e-01 6.63929820e-01
2.61863500e-01 4.86891538e-01 9.88078356e-01 -3.05578411e-01
-3.02669644e-01 3.59668061e-02 -1.13614857e+00 1.58752143e-01
4.10923034e-01 -9.68253314e-01 2.67293930e-01 9.51453447e-01
-2.05254838e-01 1.53379846e+00 8.43207061e-01 -8.57185662e-01
-1.47465158e+00 -6.41311467e-01 1.46755075e+00 -5.68314672e-01
1.07313049e+00 -2.71138340e-01 -1.16534173e+00 1.01963127e+00
1.59198567e-01 -4.78880346e-01 6.48465574e-01 9.83693823e-02
-5.78027427e-01 2.91740179e-01 -1.28774440e+00 1.19030678e+00
1.00857008e+00 -1.02123725e+00 -1.41348457e+00 4.23420876e-01
6.86081529e-01 -2.44113356e-01 -6.72046423e-01 8.04386009e-03
3.63461494e-01 -1.19501555e+00 6.12452626e-01 -6.52562022e-01
8.85300338e-01 -6.46339715e-01 2.08125651e-01 -7.58758366e-01
-4.18190479e-01 -8.69161427e-01 -3.19228806e-02 1.25411522e+00
4.20089066e-01 -9.17224109e-01 1.38107195e-01 1.47219527e+00
1.34360403e-01 -5.39274931e-01 -2.25258142e-01 -3.77957791e-01
2.11691838e-02 -9.50758815e-01 6.54459357e-01 1.17844796e+00
9.74946976e-01 3.59924674e-01 -4.07986157e-02 2.50967145e-01
4.01600868e-01 2.08076954e-01 8.44273329e-01 -1.12101054e+00
-4.48603094e-01 -2.58135051e-01 1.89510837e-01 -5.30902624e-01
2.88134307e-01 -9.24135506e-01 -2.46556774e-02 -1.88195562e+00
3.92251849e-01 -2.03172848e-01 3.15066487e-01 8.75121951e-01
-2.19830349e-01 -2.38698885e-01 4.11474824e-01 2.65298307e-01
-7.94754982e-01 2.62347400e-01 1.43542600e+00 2.11968794e-01
2.48497515e-03 -2.08286479e-01 -1.13428891e+00 8.42091024e-01
7.88008809e-01 -4.64691460e-01 -7.56367624e-01 -1.33019447e-01
8.14895332e-01 3.39152455e-01 8.92460585e-01 -9.45062757e-01
6.07047021e-01 -6.28611684e-01 4.08513807e-02 -2.14015413e-02
-4.23521623e-02 -8.17977011e-01 5.66582799e-01 8.00578177e-01
-6.84558749e-01 1.02609508e-02 8.65697637e-02 -1.26980260e-01
-1.69050217e-01 -3.72304648e-01 4.44324821e-01 -5.10401189e-01
-3.87069941e-01 -3.87954801e-01 -7.77296126e-01 -2.14750126e-01
1.04937148e+00 -2.24943787e-01 -3.37345302e-01 -4.88113403e-01
-7.21445024e-01 4.22154456e-01 7.62714148e-01 -6.65132329e-02
5.89952290e-01 -9.45202053e-01 -4.29930836e-01 2.84936428e-01
2.90463702e-03 1.66011140e-01 2.28111386e-01 1.27243853e+00
-6.83415771e-01 3.94124478e-01 -2.53635257e-01 -1.60007834e-01
-8.87134969e-01 8.42810869e-01 4.41611826e-01 -3.91273737e-01
-6.62638783e-01 5.64347267e-01 3.25160362e-02 -6.11534238e-01
-5.63453548e-02 -6.47818148e-01 -1.16288893e-01 -3.10650021e-01
9.41846430e-01 4.22042422e-02 -1.68554753e-01 3.54053453e-02
-3.74167532e-01 1.38698339e-01 -6.59303144e-02 -3.20175707e-01
9.36514258e-01 -2.49618277e-01 -5.93279600e-01 5.70333242e-01
2.33829886e-01 1.58968240e-01 -1.04473150e+00 -2.62747724e-02
9.98880155e-03 -3.61425072e-01 -2.66136795e-01 -1.33583367e+00
-2.14523464e-01 7.34349012e-01 -5.92035234e-01 4.52007115e-01
8.67484272e-01 -2.39477590e-01 3.77401471e-01 5.83832085e-01
6.27044439e-01 -6.29613578e-01 -5.57358153e-02 4.76198554e-01
1.00390625e+00 -9.63786423e-01 2.29356512e-02 -4.78548706e-01
-9.08334196e-01 1.26306570e+00 6.13243639e-01 2.01412588e-01
-2.84985770e-02 5.01121819e-01 3.21480811e-01 -3.91929716e-01
-1.30197656e+00 3.40302110e-01 -3.19459796e-01 4.70066220e-01
5.82870245e-01 3.03991325e-02 -3.53227854e-01 7.74885058e-01
-4.22571003e-01 8.57928574e-01 7.26018667e-01 1.03339577e+00
-2.75299430e-01 -9.39008594e-01 -6.88314497e-01 -1.58426136e-01
-8.37754011e-02 -3.05222869e-01 -6.89136028e-01 6.67615056e-01
-2.01541856e-01 1.30908525e+00 -3.82836670e-01 -1.06005825e-01
1.74777701e-01 3.36251378e-01 3.71838629e-01 -5.06360590e-01
-1.01163745e+00 -5.47889233e-01 4.31068987e-01 -6.69512272e-01
-2.56851703e-01 -2.97838718e-01 -1.74834788e+00 -1.04510546e+00
6.07575588e-02 5.71043491e-01 2.09121585e-01 1.37466037e+00
1.54716492e-01 4.56811935e-01 1.66700616e-01 -3.74745816e-01
-9.46489155e-01 -7.15815783e-01 -1.35877520e-01 4.81166840e-01
1.12377129e-01 -2.53135055e-01 -3.64101917e-01 2.50399038e-02] | [9.515277862548828, 7.367082118988037] |
04f90199-131d-453f-8606-f152f7860d3d | learning-selective-self-mutual-attention-for | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Liu_Learning_Selective_Self-Mutual_Attention_for_RGB-D_Saliency_Detection_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Liu_Learning_Selective_Self-Mutual_Attention_for_RGB-D_Saliency_Detection_CVPR_2020_paper.pdf | Learning Selective Self-Mutual Attention for RGB-D Saliency Detection | Saliency detection on RGB-D images is receiving more and more research interests recently. Previous models adopt the early fusion or the result fusion scheme to fuse the input RGB and depth data or their saliency maps, which incur the problem of distribution gap or information loss. Some other models use the feature fusion scheme but are limited by the linear feature fusion methods. In this paper, we propose to fuse attention learned in both modalities. Inspired by the Non-local model, we integrate the self-attention and each other's attention to propagate long-range contextual dependencies, thus incorporating multi-modal information to learn attention and propagate contexts more accurately. Considering the reliability of the other modality's attention, we further propose a selection attention to weight the newly added attention term. We embed the proposed attention module in a two-stream CNN for RGB-D saliency detection. Furthermore, we also propose a residual fusion module to fuse the depth decoder features into the RGB stream. Experimental results on seven benchmark datasets demonstrate the effectiveness of the proposed model components and our final saliency model. Our code and saliency maps are available at https://github.com/nnizhang/S2MA.
| [' Junwei Han', ' Ni Zhang', 'Nian Liu'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['rgb-d-salient-object-detection'] | ['computer-vision'] | [ 1.85900137e-01 -6.58846349e-02 -7.91539773e-02 -5.28934717e-01
-6.57146037e-01 3.99223641e-02 3.95160913e-01 2.21106857e-01
-3.78446847e-01 4.51029211e-01 4.77906376e-01 1.58796191e-01
1.61443919e-01 -5.95160484e-01 -7.29020834e-01 -7.31853962e-01
5.56413591e-01 -4.25992817e-01 9.77233529e-01 -2.06718996e-01
4.55482930e-01 -6.72750026e-02 -1.70727408e+00 2.60346860e-01
1.14552867e+00 1.41548789e+00 7.96747148e-01 3.53449911e-01
-1.55118793e-01 9.63819623e-01 -2.00392649e-01 4.09455001e-02
6.65601566e-02 -3.58844161e-01 -5.50479174e-01 -8.99221748e-02
-5.38273109e-03 -5.86044312e-01 -4.00890350e-01 1.17389226e+00
7.18658805e-01 6.51403293e-02 -6.94767684e-02 -1.36103714e+00
-8.71787727e-01 4.75523263e-01 -9.02119279e-01 4.77560163e-01
4.69766974e-01 2.64832854e-01 5.93301952e-01 -1.01254725e+00
2.52788246e-01 1.22142196e+00 3.11707109e-01 3.29779476e-01
-4.34789717e-01 -6.03385627e-01 5.63957512e-01 6.77642286e-01
-1.27341723e+00 -4.22977269e-01 1.08500767e+00 3.71336145e-03
6.09975755e-01 1.55341431e-01 6.64734006e-01 6.36132419e-01
9.63687152e-02 1.42618668e+00 1.01996672e+00 -2.52357215e-01
-2.32050754e-02 2.14885578e-01 3.76795083e-02 6.21234000e-01
-4.38199379e-02 -8.24072212e-02 -6.21828973e-01 2.14921951e-01
7.63033092e-01 5.28068304e-01 -4.39629823e-01 -1.43927857e-01
-1.28219581e+00 5.12389362e-01 1.19320965e+00 3.29280168e-01
-4.52783108e-01 1.76076204e-01 1.67249531e-01 -1.45912036e-01
6.55638099e-01 -1.72605112e-01 -2.41671503e-01 1.75032705e-01
-8.29415381e-01 -2.54456662e-02 3.38050574e-02 1.08739555e+00
1.07614183e+00 -2.09492460e-01 -3.61431032e-01 5.92803299e-01
6.87831521e-01 4.97266233e-01 7.02089131e-01 -5.20029008e-01
4.63064700e-01 8.83098185e-01 1.10492535e-01 -1.11585784e+00
-4.55579698e-01 -4.22850132e-01 -5.78988135e-01 2.25332770e-02
-1.70745999e-01 1.79991741e-02 -1.09947824e+00 1.62638354e+00
5.10611415e-01 5.74305773e-01 -9.72849354e-02 1.53705215e+00
1.14497054e+00 5.89912176e-01 8.53769779e-02 8.98475945e-02
1.06706977e+00 -1.29138494e+00 -7.49803066e-01 -3.67356688e-01
2.27667153e-01 -8.03092360e-01 9.49678719e-01 -1.59484476e-01
-1.25835431e+00 -6.38343871e-01 -1.15326643e+00 -6.41287565e-01
-3.32636029e-01 1.06384918e-01 5.06129503e-01 5.59013709e-02
-1.08533037e+00 2.84305811e-01 -1.02380598e+00 -1.70612723e-01
6.39921546e-01 4.09682214e-01 6.14467338e-02 -1.36781991e-01
-1.38826799e+00 8.04019332e-01 3.07866961e-01 4.90910947e-01
-6.85515642e-01 -4.58097279e-01 -8.68650854e-01 -6.93762153e-02
3.87910247e-01 -6.43026590e-01 1.18249643e+00 -1.01856291e+00
-1.25836134e+00 3.42992187e-01 -4.39692587e-01 4.25269920e-03
1.76102653e-01 -2.73880005e-01 -2.27199554e-01 7.90334940e-02
1.60215899e-01 6.99316204e-01 7.07117140e-01 -1.31423080e+00
-1.09365308e+00 -4.87659216e-01 2.02497721e-01 6.12435162e-01
-2.09582731e-01 -9.93564054e-02 -7.12546170e-01 -4.42999214e-01
5.04826248e-01 -6.28021896e-01 -2.05114573e-01 -1.10458367e-01
-5.02844691e-01 -1.52562752e-01 9.08977628e-01 -6.07218564e-01
1.26048374e+00 -2.19492674e+00 4.01695758e-01 -3.26706469e-02
2.89252132e-01 1.22591592e-01 4.30263951e-02 -7.69961998e-02
7.35952556e-02 -5.75539097e-02 -4.04133290e-01 -6.36761069e-01
-1.45611733e-01 -1.03043638e-01 -1.30680636e-01 4.00533587e-01
4.94049817e-01 1.07466602e+00 -1.07465911e+00 -5.71593225e-01
3.92649561e-01 6.83779418e-01 -3.86161417e-01 2.58326113e-01
-2.48084660e-03 5.17144024e-01 -8.04694057e-01 8.57681930e-01
9.39251840e-01 -3.00742775e-01 -5.07669985e-01 -4.52608198e-01
-3.32859874e-01 3.70103508e-01 -9.92399335e-01 2.24155927e+00
-3.41986775e-01 3.07910085e-01 -1.39163882e-01 -6.25186741e-01
8.22964966e-01 -3.52347195e-02 2.17864737e-01 -8.91086221e-01
3.33281040e-01 4.11350012e-01 -1.54438689e-01 -4.50149983e-01
5.86994052e-01 -1.42758498e-02 7.08945170e-02 3.59873921e-01
9.48298201e-02 2.37472877e-02 -2.91995704e-01 2.94824690e-01
7.76369095e-01 3.78532767e-01 -2.79523525e-02 3.91199924e-02
7.53483415e-01 -2.45384082e-01 6.91498041e-01 2.95993030e-01
-3.82613033e-01 9.73708332e-01 1.61662444e-01 -7.33580738e-02
-6.09630942e-01 -8.03455651e-01 1.88709259e-01 1.09450984e+00
1.07745337e+00 -1.91696197e-01 -4.90071028e-01 -6.94520354e-01
-1.12791993e-01 3.53367627e-01 -7.34537125e-01 -5.25878966e-01
-3.90854329e-01 -5.67847133e-01 -5.19837774e-02 6.06224239e-01
8.88439417e-01 -1.06982040e+00 -7.89339244e-01 1.44812390e-01
-3.29649925e-01 -8.17187905e-01 -5.71279228e-01 -9.24174301e-03
-7.76358902e-01 -7.55456984e-01 -1.00932860e+00 -8.21580410e-01
6.34437501e-01 6.93835437e-01 4.89852965e-01 3.26794833e-01
5.96142225e-02 3.45957398e-01 -6.68548703e-01 -4.80280489e-01
2.96291590e-01 2.19940603e-01 -3.19743216e-01 1.90149575e-01
3.52892041e-01 -4.69223887e-01 -1.05801630e+00 2.26353556e-01
-1.07292593e+00 4.30910587e-01 7.65543640e-01 5.91181159e-01
6.38550401e-01 -3.21955949e-01 7.22968102e-01 -2.68726259e-01
2.91148841e-01 -8.85920703e-01 -1.91770315e-01 1.83473408e-01
-3.06397140e-01 7.11170083e-04 8.39142501e-02 -3.29721123e-02
-1.29088378e+00 1.08352855e-01 -1.54928505e-01 -7.02820957e-01
-9.85222161e-02 4.10812467e-01 -4.82573688e-01 9.38978884e-03
-5.89133762e-02 4.49546874e-01 -2.32383758e-01 -5.70844114e-01
2.60211647e-01 6.89344525e-01 4.16511208e-01 -6.37692288e-02
5.10577261e-01 5.11232197e-01 -4.42782879e-01 -2.32114360e-01
-8.88736367e-01 -4.11245406e-01 -3.65404695e-01 -3.44226867e-01
8.94505084e-01 -1.13500655e+00 -5.48667133e-01 7.14087367e-01
-1.23109901e+00 -3.00598964e-02 -1.37885630e-01 4.94420350e-01
-3.22025180e-01 2.43126586e-01 -3.22067857e-01 -6.86938524e-01
-2.74330258e-01 -1.48450577e+00 1.42679703e+00 8.46182048e-01
4.70400184e-01 -6.80154204e-01 -2.14740843e-01 -7.07225222e-03
5.44563293e-01 1.18974283e-01 2.51965731e-01 -2.18018562e-01
-9.22261238e-01 5.20882905e-02 -7.88260818e-01 1.05417714e-01
3.34636748e-01 -3.88376117e-01 -1.07520103e+00 8.47754721e-03
2.70487934e-01 -1.03727229e-01 1.35641968e+00 3.77321333e-01
1.12948251e+00 -6.28892556e-02 -3.76756728e-01 6.23948812e-01
1.47332764e+00 1.11072257e-01 6.36209309e-01 3.44162166e-01
1.14564836e+00 4.93321121e-01 8.83440793e-01 4.80753213e-01
8.48199248e-01 5.28933942e-01 7.89870679e-01 -2.42144808e-01
-1.86840981e-01 -2.60096133e-01 3.07654232e-01 9.75614965e-01
-1.04218878e-01 -6.83222264e-02 -8.08598638e-01 7.59514689e-01
-2.08025360e+00 -6.90871298e-01 -7.20266476e-02 1.94271946e+00
8.68242264e-01 1.81382611e-01 -6.63408563e-02 5.79565056e-02
9.72886801e-01 2.26348206e-01 -7.40586400e-01 -1.12105522e-03
-2.57022321e-01 -8.74294043e-02 4.70178485e-01 4.72721100e-01
-9.93554235e-01 7.66073048e-01 4.32728291e+00 9.14702415e-01
-1.40167642e+00 4.35735315e-01 6.38046384e-01 -4.59062338e-01
-7.41435230e-01 1.38902187e-01 -6.64708912e-01 8.30646694e-01
3.75374675e-01 -1.66220605e-01 1.54472142e-01 4.74508852e-01
1.61271870e-01 -6.15672886e-01 -6.85703158e-01 9.40609276e-01
2.38697857e-01 -1.03417075e+00 -1.03539817e-01 -2.54305333e-01
7.21427739e-01 1.83033898e-01 2.69600242e-01 -3.83694209e-02
-9.62533876e-02 -6.10103846e-01 1.04052210e+00 9.49597299e-01
4.31246996e-01 -7.93226361e-01 9.08125937e-01 9.69997719e-02
-1.38427293e+00 -1.80584356e-01 -3.69694710e-01 -7.76303411e-02
3.43482912e-01 7.47286499e-01 -3.01421940e-01 8.66108000e-01
8.57706130e-01 1.26330924e+00 -8.80582035e-01 1.36070943e+00
-3.39176476e-01 1.18073039e-01 -3.49999279e-01 -1.01836719e-01
3.82905453e-01 1.83791444e-01 5.64064503e-01 8.54890943e-01
4.23744470e-01 2.99323261e-01 -3.39219868e-02 8.34073126e-01
1.69208273e-01 3.76278604e-03 -1.38139561e-01 4.39915955e-01
3.90376300e-01 1.12994969e+00 -7.01785564e-01 -3.54952842e-01
-6.14524603e-01 1.16867793e+00 1.71777666e-01 2.84003675e-01
-1.04801130e+00 -5.85072756e-01 3.98423165e-01 -1.06524557e-01
4.53922510e-01 4.72909659e-02 -4.83405769e-01 -1.19937706e+00
2.10465536e-01 -1.73213437e-01 1.01687320e-01 -1.15337801e+00
-8.96098971e-01 5.74594796e-01 -2.11702183e-01 -1.46215558e+00
4.18382674e-01 -1.00456767e-01 -6.04308724e-01 1.04546630e+00
-2.06209540e+00 -1.38412416e+00 -6.51844501e-01 7.30501533e-01
5.66758335e-01 2.62508422e-01 7.87558183e-02 3.86748672e-01
-6.24380052e-01 3.10878009e-01 -3.04662883e-01 -3.07704628e-01
6.72543645e-01 -1.08432508e+00 1.23449035e-01 8.65255237e-01
-4.78348225e-01 4.45383728e-01 4.61338878e-01 -6.61850333e-01
-1.20466113e+00 -1.03336763e+00 7.05922306e-01 -1.44976988e-01
3.29421669e-01 -1.25897378e-01 -1.01212752e+00 5.33606708e-01
3.20877910e-01 1.56791225e-01 2.88035035e-01 -3.94132525e-01
-8.37330532e-04 -2.25943252e-01 -1.02857447e+00 3.35095972e-01
1.07242393e+00 -4.40117836e-01 -6.67360604e-01 -9.25143138e-02
1.23096013e+00 -5.79956353e-01 -5.72965622e-01 5.64581633e-01
4.31204885e-01 -1.16739881e+00 7.70534098e-01 1.00757517e-01
7.14830160e-01 -7.55961418e-01 -1.07956462e-01 -1.05801976e+00
-2.10741594e-01 -8.75948742e-02 -3.83140408e-02 1.40438473e+00
1.91203341e-01 -4.43177700e-01 4.32765722e-01 5.74387729e-01
-5.07076681e-01 -9.63763714e-01 -8.34619462e-01 -9.28717852e-02
-4.40156639e-01 -3.40712339e-01 9.27062809e-01 6.39346838e-01
9.07171071e-02 2.48804435e-01 -3.81580710e-01 2.20256805e-01
4.05140072e-01 2.86890179e-01 3.68809700e-01 -7.86637366e-01
5.30910268e-02 -4.73659873e-01 -4.62234885e-01 -1.30623651e+00
-1.44679666e-01 -8.43671978e-01 3.78886729e-01 -1.75534749e+00
2.75501221e-01 -5.72767138e-01 -8.65382195e-01 5.89586675e-01
-7.20005393e-01 2.65576929e-01 2.49027327e-01 1.76361695e-01
-1.07009339e+00 1.10657382e+00 1.33546662e+00 -7.81643540e-02
-3.15193713e-01 -2.58238733e-01 -9.45406556e-01 6.45163417e-01
8.34066212e-01 -3.30535829e-01 -5.04270494e-01 -8.69101822e-01
1.36205489e-02 -1.54743567e-01 7.11246967e-01 -1.15051806e+00
6.69525266e-01 -7.54504874e-02 4.96643960e-01 -8.61549556e-01
3.36006761e-01 -8.30417216e-01 -3.57539386e-01 1.40876740e-01
-1.51734307e-01 -1.52753264e-01 3.02241027e-01 6.47558510e-01
-3.87667418e-01 1.29757091e-01 5.92796922e-01 -9.41663459e-02
-1.04710269e+00 5.00735819e-01 8.55181068e-02 -1.67915866e-01
1.00177658e+00 -1.09260075e-01 -3.35738242e-01 -2.23065570e-01
-6.35524035e-01 5.34209847e-01 5.36308646e-01 6.84136093e-01
1.05733299e+00 -1.57883942e+00 -5.48997641e-01 2.99224973e-01
1.68887258e-01 3.55297238e-01 7.13973641e-01 1.17732430e+00
-1.39444545e-01 2.51756221e-01 -3.40806216e-01 -7.65576005e-01
-7.69891262e-01 4.90892440e-01 1.83034271e-01 1.98696151e-01
-2.37525970e-01 1.17130685e+00 4.10346657e-01 1.89724043e-02
-1.35315480e-02 -5.66649437e-01 -3.28693032e-01 -4.91680168e-02
5.98710477e-01 2.08275691e-01 -1.42611250e-01 -8.35630894e-01
-7.79355347e-01 6.47267938e-01 -1.09877307e-02 -3.14955227e-02
1.31395733e+00 -7.31653154e-01 -1.62491411e-01 5.66358030e-01
1.18963087e+00 -2.54497409e-01 -1.46337891e+00 -4.70263571e-01
-2.18055561e-01 -6.85789824e-01 3.36599499e-01 -5.80191016e-01
-1.37469196e+00 9.72859859e-01 8.24756384e-01 1.91159938e-02
1.61875796e+00 2.04094633e-01 1.01827443e+00 -4.02193457e-01
1.80591002e-01 -7.80042708e-01 3.48086804e-02 2.59246677e-01
9.56911623e-01 -1.53660285e+00 7.25627095e-02 -3.64398390e-01
-6.33507788e-01 5.48046708e-01 8.53487790e-01 -2.60877460e-01
8.13184440e-01 5.55251464e-02 -1.18369095e-01 6.36674762e-02
-6.59139097e-01 -6.17319465e-01 2.32000545e-01 3.56722206e-01
1.81509614e-01 -2.52895862e-01 -4.03856426e-01 1.03234327e+00
2.67705441e-01 3.07643801e-01 3.93389344e-01 1.16489816e+00
-6.65181398e-01 -7.82708585e-01 -1.52611583e-01 3.04135621e-01
-1.44387737e-01 -2.43939593e-01 -1.44538417e-01 2.36458555e-01
4.59117472e-01 1.08634627e+00 2.41973475e-02 -6.30622149e-01
3.22874546e-01 -1.69158056e-01 2.64407545e-01 -4.66258764e-01
-4.37975168e-01 2.32299387e-01 -4.92169559e-01 -6.65403843e-01
-9.08907652e-01 -6.66236103e-01 -1.44539440e+00 -3.34311500e-02
-6.17681205e-01 -5.81390150e-02 6.63345397e-01 9.90223110e-01
6.16711318e-01 8.19518447e-01 7.58994699e-01 -1.16925156e+00
1.22936770e-01 -9.91206229e-01 -3.67808104e-01 8.26246440e-02
7.39406168e-01 -9.18058932e-01 -3.76942992e-01 -1.11096345e-01] | [9.712965965270996, -0.7540121674537659] |
b42a16fb-590b-4439-b295-2cd89ff5d813 | unravela-decipherment-toolkit | null | null | https://aclanthology.org/P15-2090 | https://aclanthology.org/P15-2090.pdf | UNRAVEL---A Decipherment Toolkit | null | ['Malte Nuhn', 'Hermann Ney', 'Julian Schamper'] | 2015-07-01 | unravel-a-decipherment-toolkit | https://aclanthology.org/P15-2090 | https://aclanthology.org/P15-2090.pdf | ijcnlp-2015-7 | ['decipherment'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.277866840362549, 3.6416878700256348] |
2d2f6b2a-ee85-420b-a02e-53fbafabcb5d | can-open-domain-qa-reader-utilize-external | 2211.12707 | null | https://arxiv.org/abs/2211.12707v1 | https://arxiv.org/pdf/2211.12707v1.pdf | Can Open-Domain QA Reader Utilize External Knowledge Efficiently like Humans? | Recent state-of-the-art open-domain QA models are typically based on a two stage retriever-reader approach in which the retriever first finds the relevant knowledge/passages and the reader then leverages that to predict the answer. Prior work has shown that the performance of the reader usually tends to improve with the increase in the number of these passages. Thus, state-of-the-art models use a large number of passages (e.g. 100) for inference. While the reader in this approach achieves high prediction performance, its inference is computationally very expensive. We humans, on the other hand, use a more efficient strategy while answering: firstly, if we can confidently answer the question using our already acquired knowledge then we do not even use the external knowledge, and in the case when we do require external knowledge, we don't read the entire knowledge at once, instead, we only read that much knowledge that is sufficient to find the answer. Motivated by this procedure, we ask a research question "Can the open-domain QA reader utilize external knowledge efficiently like humans without sacrificing the prediction performance?" Driven by this question, we explore an approach that utilizes both 'closed-book' (leveraging knowledge already present in the model parameters) and 'open-book' inference (leveraging external knowledge). Furthermore, instead of using a large fixed number of passages for open-book inference, we dynamically read the external knowledge in multiple 'knowledge iterations'. Through comprehensive experiments on NQ and TriviaQA datasets, we demonstrate that this dynamic reading approach improves both the 'inference efficiency' and the 'prediction accuracy' of the reader. Comparing with the FiD reader, this approach matches its accuracy by utilizing just 18.32% of its reader inference cost and also outperforms it by achieving up to 55.10% accuracy on NQ Open. | ['Chitta Baral', 'Man Luo', 'Neeraj Varshney'] | 2022-11-23 | null | null | null | null | ['triviaqa', 'open-domain-question-answering'] | ['miscellaneous', 'natural-language-processing'] | [-1.79114968e-01 4.15608704e-01 -6.03938885e-02 -1.29846230e-01
-1.50237596e+00 -1.04562461e+00 4.56281900e-01 2.39395186e-01
-6.39553428e-01 9.75522459e-01 6.33025467e-02 -5.77791989e-01
-2.44733840e-01 -1.07197547e+00 -8.59671414e-01 -1.78059623e-01
6.84634566e-01 1.13190186e+00 7.33663619e-01 -6.80207670e-01
2.97166646e-01 -8.52376074e-02 -1.47213006e+00 2.34201521e-01
1.39351976e+00 8.38776827e-01 3.14834058e-01 6.13821328e-01
-3.36567998e-01 9.64608371e-01 -7.50229657e-01 -7.63676465e-01
6.19911589e-02 -3.05222362e-01 -1.61156738e+00 -6.04378998e-01
2.55019426e-01 -3.80837262e-01 -1.70969665e-01 8.34476054e-01
4.90666240e-01 3.51346940e-01 4.85738099e-01 -5.77668071e-01
-8.60106349e-01 4.84954268e-01 -1.67665616e-01 3.78696918e-01
9.75565791e-01 1.54754922e-01 1.31374526e+00 -5.83866298e-01
6.82393789e-01 8.50614190e-01 5.43625057e-01 3.66509229e-01
-1.05665791e+00 -3.17974329e-01 -1.37602702e-01 4.85458076e-01
-1.32033610e+00 -3.79249513e-01 3.04285377e-01 -1.44791111e-01
1.18655562e+00 2.88137913e-01 3.78854841e-01 7.52845109e-01
3.52431238e-02 5.88634729e-01 1.14014900e+00 -6.98114157e-01
1.24954537e-01 2.40664631e-01 5.39731443e-01 6.64494574e-01
-1.28732878e-03 -3.07293911e-03 -5.79063416e-01 -2.16596723e-01
1.94462717e-01 -4.78981227e-01 -4.56126839e-01 1.77240536e-01
-8.58716249e-01 9.17273581e-01 3.34038645e-01 1.47959396e-01
-4.43904638e-01 -2.00199962e-01 1.34828091e-01 6.22501671e-01
2.29372606e-01 8.55052173e-01 -7.34177828e-01 -6.23420238e-01
-1.10461640e+00 4.14372444e-01 1.34334075e+00 7.14276075e-01
1.02388203e+00 -8.38199854e-01 -2.61412859e-01 9.84676480e-01
1.37624741e-01 6.43352628e-01 4.03256088e-01 -1.18492830e+00
5.05361676e-01 6.64917588e-01 4.64803427e-01 -7.92277277e-01
-1.47769570e-01 -4.33900774e-01 -1.45698071e-01 -3.76546115e-01
7.56126642e-01 -2.59758979e-01 -9.09226000e-01 1.75309968e+00
3.36821735e-01 -2.49984741e-01 2.25587428e-01 7.77792096e-01
7.31511533e-01 7.83334196e-01 5.81056997e-02 -9.36167911e-02
1.69968176e+00 -9.68843400e-01 -6.69679821e-01 -2.44629309e-01
8.10470700e-01 -8.15710664e-01 1.16707802e+00 3.40597510e-01
-1.28905737e+00 -4.59130764e-01 -8.43078434e-01 -5.19938409e-01
-3.56719524e-01 7.25876726e-03 2.33984560e-01 3.89280051e-01
-9.52025294e-01 4.11632419e-01 -4.62604225e-01 -2.32160449e-01
1.59424618e-02 2.58835673e-01 -6.92963526e-02 -4.84582067e-01
-1.75536740e+00 1.40165710e+00 3.72749269e-01 -1.32327050e-01
-7.86401987e-01 -8.53593409e-01 -5.49444318e-01 3.38891506e-01
7.29827881e-01 -1.19262004e+00 1.61627245e+00 -6.98568165e-01
-1.54416132e+00 5.35670042e-01 -5.86395741e-01 -3.94442946e-01
4.39472377e-01 -6.32514119e-01 -2.69611090e-01 4.54909980e-01
1.96734011e-01 5.10676086e-01 4.30796742e-01 -1.23151493e+00
-6.64675176e-01 -2.54564166e-01 8.76488149e-01 4.17612165e-01
-1.52079100e-02 -7.01013878e-02 -5.58255970e-01 2.52592918e-02
1.66729823e-01 -8.90991509e-01 1.63818687e-01 -3.93658757e-01
-1.60677433e-01 -6.69462919e-01 3.93759847e-01 -8.75852287e-01
1.49446344e+00 -1.71553373e+00 6.18351856e-03 3.70593488e-01
3.08990926e-01 4.38305795e-01 2.51983795e-02 6.12388253e-01
5.88801861e-01 1.84391011e-02 -7.82343224e-02 3.52726877e-03
8.24494734e-02 4.07561570e-01 -5.07682264e-01 -1.54273435e-01
-7.46168718e-02 1.18080461e+00 -1.04813623e+00 -6.31384313e-01
-1.98848486e-01 1.25043795e-01 -6.84507966e-01 3.92914176e-01
-6.88616574e-01 2.59850740e-01 -5.39951086e-01 3.08879405e-01
4.52050805e-01 -8.15926492e-01 1.20440170e-01 -1.94419529e-02
2.83588469e-01 8.76081467e-01 -9.15789783e-01 1.55511761e+00
-6.66263103e-01 3.38307381e-01 -1.90462112e-01 -7.58757353e-01
6.43927634e-01 3.18434387e-01 -1.48120463e-01 -9.33433950e-01
-1.77005157e-01 4.24162596e-01 4.92118718e-03 -6.35052562e-01
5.49371541e-01 -2.34553531e-01 5.26079945e-02 6.21430278e-01
-5.70840500e-02 -5.72652221e-02 1.52541906e-01 5.87109089e-01
1.35702407e+00 2.10438043e-01 1.04936190e-01 -6.90241084e-02
6.75963283e-01 5.24662018e-01 3.21128100e-01 1.04377103e+00
1.02140896e-01 1.80033237e-01 3.55466336e-01 -4.29801457e-03
-7.19378471e-01 -1.07794774e+00 3.85387801e-02 1.33907568e+00
2.73901969e-01 -5.22210896e-01 -9.53883171e-01 -6.95466220e-01
-1.61545232e-01 1.03173482e+00 -3.50692093e-01 -1.51340157e-01
-6.10002697e-01 -1.40528262e-01 8.55078340e-01 4.55410510e-01
7.73276746e-01 -9.16671455e-01 -6.99662566e-01 3.23018402e-01
-8.53486061e-01 -9.85445321e-01 -1.73930079e-02 -2.37226542e-02
-5.54779708e-01 -9.41057503e-01 -4.75780129e-01 -4.50710833e-01
3.63775998e-01 1.75287221e-02 1.53758335e+00 4.60395068e-01
3.45705837e-01 5.96270978e-01 -6.67062461e-01 -2.95682251e-01
-4.36611265e-01 4.71506149e-01 -3.42591703e-01 -5.92119038e-01
6.77133441e-01 -3.27596366e-01 -6.21762037e-01 2.31280506e-01
-7.23214865e-01 -6.62534311e-02 6.23777568e-01 9.12482202e-01
4.36050206e-01 -1.15684830e-01 9.39091504e-01 -1.03464234e+00
8.19157183e-01 -6.22572005e-01 -3.95890981e-01 6.98121428e-01
-6.81938648e-01 2.36112624e-01 7.38483846e-01 -3.18761408e-01
-1.32159054e+00 -5.16228974e-01 -4.43568617e-01 -1.10354900e-01
-2.10454362e-03 9.78734016e-01 9.60193723e-02 8.61756504e-02
8.50703478e-01 2.90224046e-01 -7.95203969e-02 -4.94690746e-01
5.03194332e-01 8.35189283e-01 4.41700011e-01 -9.86243069e-01
7.75825858e-01 1.34535238e-01 -4.03391153e-01 -4.87900376e-01
-1.59139144e+00 -6.11722529e-01 -4.24713552e-01 1.27151802e-01
7.79434979e-01 -8.13943446e-01 -9.23720002e-01 6.54814616e-02
-1.22525060e+00 -2.85683841e-01 -2.98652589e-01 2.25003913e-01
-4.89979029e-01 5.27832091e-01 -6.69507205e-01 -6.22651339e-01
-4.78922218e-01 -9.85197783e-01 8.50750268e-01 3.03139597e-01
-4.60017145e-01 -1.10701370e+00 7.34036416e-02 1.08353806e+00
5.67645669e-01 -4.90487754e-01 1.26246035e+00 -1.07243395e+00
-7.01967895e-01 -1.71058197e-02 -2.00759560e-01 2.90248960e-01
-1.45421788e-01 -7.43377209e-01 -8.88606608e-01 -1.10840045e-01
-3.52264494e-02 -6.17810547e-01 4.65827644e-01 -1.51700139e-01
7.51662254e-01 -3.12154949e-01 -1.84700593e-01 -4.59284224e-02
1.33858263e+00 -1.10769115e-01 7.78835654e-01 4.60423440e-01
4.08238977e-01 5.94664693e-01 8.60143661e-01 3.97350900e-02
9.26968455e-01 6.08927429e-01 -7.70366713e-02 2.93982804e-01
-2.97000650e-02 -4.21594560e-01 2.24545240e-01 9.29647267e-01
-2.05299810e-01 -1.96237862e-01 -1.07894504e+00 7.09111452e-01
-1.71012127e+00 -8.41796696e-01 1.42373249e-01 2.17811871e+00
1.37155712e+00 1.48763046e-01 -2.45821014e-01 -4.10202593e-02
2.25863427e-01 -2.44835004e-01 -5.05574703e-01 -5.96884668e-01
8.86286125e-02 6.69560850e-01 1.78221852e-01 7.92466283e-01
-3.40454638e-01 1.07825065e+00 5.98975658e+00 1.06568122e+00
-6.60482645e-01 1.80696487e-01 2.22771883e-01 8.74841958e-02
-6.20580852e-01 5.45720160e-01 -1.02153265e+00 4.75828528e-01
1.34303129e+00 -9.63318795e-02 5.79216897e-01 4.58814293e-01
-1.51550859e-01 -5.97373724e-01 -1.08954680e+00 4.66258079e-01
1.41107691e-02 -1.30267513e+00 2.32117593e-01 -2.68948320e-02
4.72394377e-01 -1.20460931e-02 -2.83499569e-01 7.25401998e-01
4.60506558e-01 -1.09397912e+00 5.09797931e-01 8.47678781e-01
4.92557079e-01 -6.56102836e-01 1.00068462e+00 1.00964224e+00
-7.60806322e-01 -2.43024528e-01 -3.27140450e-01 -1.61019444e-01
3.15481961e-01 3.79030585e-01 -6.94110572e-01 8.87936890e-01
6.86699331e-01 -9.96126905e-02 -5.53066075e-01 7.42900014e-01
-5.09965241e-01 7.93470740e-01 -4.83659565e-01 -7.14056045e-02
2.65855014e-01 -1.26936324e-02 3.04369301e-01 9.21501398e-01
7.45807588e-02 7.21656024e-01 1.11468993e-01 8.73342812e-01
-1.67010069e-01 -7.16680512e-02 -8.79755467e-02 1.93138793e-01
8.47319126e-01 7.53323674e-01 7.72952512e-02 -6.96712315e-01
-4.51422662e-01 8.82010162e-01 7.40851939e-01 2.47962803e-01
-7.14735389e-01 -3.88828099e-01 4.80921306e-02 1.07408509e-01
4.54340726e-01 9.79359029e-04 -1.89580079e-02 -8.97218585e-01
3.02588165e-01 -1.19195652e+00 6.08674049e-01 -1.02694011e+00
-1.34735072e+00 3.44268709e-01 4.47689034e-02 -7.29678392e-01
-6.23174667e-01 -2.65162110e-01 -1.84104368e-01 1.21056688e+00
-2.04540920e+00 -1.12786043e+00 -5.58418371e-02 6.28533661e-01
3.41294199e-01 3.14966887e-01 9.73219991e-01 1.78539708e-01
-1.43177770e-02 7.05770314e-01 9.08989161e-02 7.84521252e-02
7.71350980e-01 -1.23019195e+00 -8.67234990e-02 5.59915066e-01
1.23517700e-01 1.26846826e+00 7.41388917e-01 -6.44168377e-01
-1.53815269e+00 -5.14070213e-01 1.31067920e+00 -9.44597661e-01
6.24818444e-01 2.03975543e-01 -1.27927732e+00 6.96602166e-01
1.95023268e-01 -2.35554501e-01 7.29392231e-01 4.08672273e-01
-4.12591457e-01 1.26229394e-02 -1.16322625e+00 4.32626903e-01
7.09518015e-01 -8.10732007e-01 -1.45952451e+00 3.06046337e-01
8.83508086e-01 -6.08641028e-01 -1.20599544e+00 3.83619756e-01
5.00473678e-01 -7.30578125e-01 8.81528258e-01 -5.77208400e-01
4.11340594e-01 -4.29116964e-01 -1.60122022e-01 -1.15826869e+00
-5.69882095e-02 -3.75527024e-01 -5.67098260e-01 1.04820597e+00
7.85868287e-01 -8.39822888e-01 5.23023844e-01 9.23827946e-01
-4.58781309e-02 -1.05269492e+00 -9.64010775e-01 -6.45502150e-01
4.14961755e-01 -2.02147305e-01 5.58421671e-01 6.65913284e-01
1.66549727e-01 6.84243858e-01 1.51875585e-01 3.26196402e-01
1.62166402e-01 3.88701826e-01 7.45011628e-01 -1.20301235e+00
-5.60678720e-01 1.80981621e-01 1.68932155e-01 -1.79037058e+00
1.06792904e-01 -7.03202069e-01 9.96057317e-02 -1.73794758e+00
3.29924971e-01 -6.68440342e-01 -1.08737551e-01 4.56532121e-01
-4.68934655e-01 -4.14459668e-02 1.96569264e-01 4.16464150e-01
-9.78401959e-01 1.58039495e-01 1.31952369e+00 2.01107726e-01
-8.94654989e-02 -2.10778564e-01 -9.44826782e-01 5.00384152e-01
4.79804248e-01 -2.54886895e-01 -5.23509324e-01 -6.05273068e-01
6.46414697e-01 3.94244045e-01 4.31312978e-01 -7.63830841e-01
6.48719490e-01 6.11781776e-02 2.13325992e-01 -4.68780726e-01
5.53759813e-01 -6.79801404e-01 -1.40255839e-01 2.02584267e-01
-2.48266563e-01 4.29190993e-02 1.64167359e-01 5.53640187e-01
-2.37230465e-01 -5.82564712e-01 5.42396128e-01 -3.94993842e-01
-5.59987545e-01 -2.01658294e-01 -2.58985847e-01 4.93952096e-01
5.89891970e-01 -1.51591077e-01 -7.28780806e-01 -5.72518051e-01
-6.55565441e-01 4.06105965e-01 3.66871685e-01 6.19316846e-02
4.01824802e-01 -6.64495349e-01 -5.34677744e-01 -2.62153566e-01
5.97588569e-02 1.76614597e-01 1.89731911e-01 9.13981855e-01
-3.19794714e-01 6.87185705e-01 2.67973810e-01 -4.79562819e-01
-9.33280051e-01 3.71446609e-01 2.49997020e-01 -7.04708993e-01
-4.13397193e-01 7.26303279e-01 -2.50496894e-01 -5.24843991e-01
-9.63610336e-02 8.92926529e-02 -3.20410252e-01 5.01095615e-02
4.50223923e-01 3.95159811e-01 2.79704213e-01 -3.95431399e-01
-4.51124273e-02 5.31955063e-01 -4.09097105e-01 -2.76139736e-01
9.36678529e-01 -4.37245250e-01 -1.74876079e-01 2.74885118e-01
9.42758441e-01 3.24530303e-01 -6.49184644e-01 -5.42680383e-01
3.78189683e-02 -3.44893157e-01 -6.35854676e-02 -1.46872151e+00
-4.19896126e-01 7.59283662e-01 2.54787773e-01 1.71387643e-01
1.02620828e+00 1.89384818e-01 1.24612617e+00 9.32634413e-01
6.19666576e-01 -1.13971007e+00 3.14986780e-02 9.28604305e-01
4.52188253e-01 -1.21668959e+00 -7.97532266e-04 -3.07558954e-01
-6.02212608e-01 7.81317770e-01 5.99650323e-01 2.07100838e-01
2.64455885e-01 -1.61554113e-01 2.13888779e-01 -4.93821055e-01
-9.13004518e-01 -2.68179446e-01 2.89239883e-01 3.10836911e-01
2.86026984e-01 -9.27893445e-02 -3.16857129e-01 7.12026834e-01
-6.70898557e-01 9.95723903e-02 3.47329825e-01 9.32179749e-01
-7.81518281e-01 -1.10273516e+00 -2.44180530e-01 6.79463029e-01
-6.07201040e-01 -3.87869239e-01 -1.71219021e-01 6.02464199e-01
9.96187981e-03 1.38551331e+00 -2.21987337e-01 -1.03067756e-01
3.27327400e-01 3.88837993e-01 4.20140505e-01 -7.28243411e-01
-7.06316173e-01 -5.12293100e-01 5.33200026e-01 -4.47207719e-01
-3.56454790e-01 -3.45215559e-01 -1.42165971e+00 -5.59655488e-01
-7.56585717e-01 7.06295907e-01 3.07557225e-01 1.46918643e+00
5.20482302e-01 1.83725804e-01 1.58776060e-01 1.10865876e-01
-1.03178203e+00 -9.62153375e-01 -3.12042721e-02 3.07361603e-01
2.66583383e-01 -6.78617835e-01 -3.78993422e-01 -9.71830338e-02] | [11.17094612121582, 7.974584102630615] |
41819506-fa79-4249-8d95-53b018d802f6 | voxelnet-end-to-end-learning-for-point-cloud | 1711.06396 | null | http://arxiv.org/abs/1711.06396v1 | http://arxiv.org/pdf/1711.06396v1.pdf | VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection | Accurate detection of objects in 3D point clouds is a central problem in many
applications, such as autonomous navigation, housekeeping robots, and
augmented/virtual reality. To interface a highly sparse LiDAR point cloud with
a region proposal network (RPN), most existing efforts have focused on
hand-crafted feature representations, for example, a bird's eye view
projection. In this work, we remove the need of manual feature engineering for
3D point clouds and propose VoxelNet, a generic 3D detection network that
unifies feature extraction and bounding box prediction into a single stage,
end-to-end trainable deep network. Specifically, VoxelNet divides a point cloud
into equally spaced 3D voxels and transforms a group of points within each
voxel into a unified feature representation through the newly introduced voxel
feature encoding (VFE) layer. In this way, the point cloud is encoded as a
descriptive volumetric representation, which is then connected to a RPN to
generate detections. Experiments on the KITTI car detection benchmark show that
VoxelNet outperforms the state-of-the-art LiDAR based 3D detection methods by a
large margin. Furthermore, our network learns an effective discriminative
representation of objects with various geometries, leading to encouraging
results in 3D detection of pedestrians and cyclists, based on only LiDAR. | ['Yin Zhou', 'Oncel Tuzel'] | 2017-11-17 | voxelnet-end-to-end-learning-for-point-cloud-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Zhou_VoxelNet_End-to-End_Learning_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Zhou_VoxelNet_End-to-End_Learning_CVPR_2018_paper.pdf | cvpr-2018-6 | ['birds-eye-view-object-detection'] | ['computer-vision'] | [-2.44943440e-01 -1.53461561e-01 -2.83921659e-02 -5.21977425e-01
-4.96534944e-01 -4.02058154e-01 5.95613182e-01 1.10605940e-01
-4.90187168e-01 1.48158416e-01 -2.71271944e-01 -4.64511395e-01
2.40228668e-01 -1.05590141e+00 -1.01324880e+00 -3.64664972e-01
-1.06997930e-01 7.64179528e-01 7.27331996e-01 -1.80766404e-01
1.63304403e-01 1.09615767e+00 -1.61809933e+00 -1.54049769e-01
6.32374883e-01 1.26784432e+00 2.70506531e-01 2.15598658e-01
-2.66558439e-01 -6.22845776e-02 -3.43747199e-01 -5.20570017e-02
5.59548676e-01 4.81983751e-01 -1.71133988e-02 9.25576761e-02
6.07998490e-01 -6.27506256e-01 -3.40295166e-01 8.42612565e-01
3.66816103e-01 2.05812216e-01 8.76678348e-01 -1.40619385e+00
-4.63683784e-01 2.45021492e-01 -8.90397906e-01 1.33498624e-01
2.08623484e-01 4.32854980e-01 7.74961531e-01 -1.41453540e+00
4.39607054e-01 1.66793191e+00 7.09255219e-01 2.74559051e-01
-1.16362417e+00 -9.40019727e-01 3.90563816e-01 4.21310961e-02
-1.80156100e+00 1.97737906e-02 8.62740874e-01 -6.03606045e-01
1.21374953e+00 3.52491438e-03 9.70424533e-01 7.40476668e-01
1.66910514e-01 8.30149174e-01 6.10754490e-01 6.66902140e-02
1.02797225e-01 1.11510225e-01 2.22323872e-02 8.14129412e-01
4.84828383e-01 3.75038087e-01 -1.69362292e-01 -9.53365043e-02
9.70142543e-01 3.78147483e-01 9.82675850e-02 -1.13070440e+00
-9.76537883e-01 1.11828899e+00 1.16580105e+00 -2.96532422e-01
-3.58961701e-01 2.54808038e-01 4.22980487e-01 -2.30408058e-01
6.28304899e-01 -1.54753253e-01 -2.01721266e-01 5.20458341e-01
-8.40224385e-01 6.55700088e-01 2.69499213e-01 1.08622062e+00
8.47783744e-01 -7.74052143e-02 -7.26435333e-02 5.51262259e-01
6.84562683e-01 6.99476361e-01 -3.80451903e-02 -6.65397525e-01
6.48874164e-01 1.03280807e+00 9.11388621e-02 -9.07672524e-01
-5.08864224e-01 -5.42545736e-01 -7.34257877e-01 6.15831792e-01
-7.79631063e-02 3.16074580e-01 -9.78679597e-01 1.37133384e+00
7.86370218e-01 1.44451067e-01 -3.99967045e-01 1.35489404e+00
1.01381564e+00 6.33680940e-01 -3.84342968e-02 4.43925321e-01
1.41237307e+00 -5.78532100e-01 2.08473116e-01 -3.13036948e-01
3.71756375e-01 -2.96821117e-01 7.50145257e-01 7.65986741e-02
-8.88768733e-01 -7.96157837e-01 -1.04604876e+00 -3.16861302e-01
-5.66455960e-01 2.33945623e-01 5.34175336e-01 5.69325089e-01
-7.57193029e-01 2.15434417e-01 -8.64231884e-01 -2.39421517e-01
9.45258617e-01 4.77406174e-01 -3.74932170e-01 -1.62266955e-01
-6.08345449e-01 9.94435668e-01 4.47564363e-01 1.15647785e-01
-9.55269396e-01 -7.90113091e-01 -1.14913487e+00 1.27594426e-01
3.76811951e-01 -1.01582217e+00 8.99710119e-01 8.95835012e-02
-1.04359889e+00 1.06798160e+00 -1.03579931e-01 -6.44022703e-01
7.09821343e-01 -3.15377623e-01 -5.92253357e-02 -1.96844637e-01
4.44142520e-01 1.10599673e+00 7.89217949e-01 -1.24816906e+00
-8.70291293e-01 -7.12697268e-01 -1.59192502e-01 1.10414676e-01
1.78414896e-01 -1.34093761e-01 -4.88384664e-01 -1.21151678e-01
5.40745080e-01 -8.40064466e-01 -3.56290370e-01 5.00578821e-01
-6.14942431e-01 -7.06689000e-01 1.11471117e+00 -5.69503345e-02
4.05633092e-01 -2.17256403e+00 -1.11078225e-01 3.31568718e-01
5.50602317e-01 1.17164567e-01 -4.21612300e-02 -5.39985523e-02
7.06225112e-02 -4.90471646e-02 -9.40518379e-02 -4.48240757e-01
2.47296974e-01 -4.45027687e-02 -4.41315055e-01 7.11224735e-01
6.60563946e-01 1.02895582e+00 -8.52984786e-01 -5.78058183e-01
8.29084158e-01 9.04783189e-01 -6.37304008e-01 -2.85866624e-03
-2.29597107e-01 2.05932677e-01 -6.86671257e-01 8.24367642e-01
1.11395407e+00 1.05245970e-01 -6.01835847e-01 3.63313290e-03
-4.87351388e-01 3.63737941e-01 -1.19261754e+00 1.61556280e+00
-2.92226940e-01 5.39394617e-01 -8.85813609e-02 -6.75784826e-01
1.41065645e+00 -1.49597630e-01 4.68278587e-01 -4.17141736e-01
2.25006372e-01 2.82673210e-01 -2.97993392e-01 1.32266730e-01
5.72403789e-01 1.19463183e-01 -2.80395508e-01 1.60448804e-01
-9.64929238e-02 -4.69466269e-01 -9.64727029e-02 6.43907264e-02
9.44157958e-01 1.11965306e-01 1.93511039e-01 1.36995047e-01
5.59075296e-01 2.26283118e-01 4.42399114e-01 6.72075510e-01
-3.46520096e-01 6.69773996e-01 1.25582829e-01 -6.73787415e-01
-1.11607111e+00 -1.45847356e+00 -4.27518398e-01 7.28852153e-01
2.98666120e-01 -2.90125757e-01 -3.21685046e-01 -6.65851891e-01
7.24842370e-01 5.63575029e-01 -3.69477689e-01 -1.49266139e-01
-7.59274244e-01 -2.22162440e-01 1.87889874e-01 7.73844481e-01
4.86768514e-01 -7.65724301e-01 -1.10215306e+00 2.05070227e-01
4.30408180e-01 -1.19992721e+00 -3.18808168e-01 3.25454086e-01
-8.99730027e-01 -9.23276067e-01 -4.55568105e-01 -6.22038543e-01
5.73860466e-01 8.71081531e-01 9.44104970e-01 -1.08869612e-01
-5.12015104e-01 1.58540383e-01 -1.10575721e-01 -7.79088557e-01
1.58683944e-03 5.65355606e-02 2.07562640e-01 -4.06671733e-01
8.49774241e-01 -4.95450705e-01 -8.29244494e-01 3.18417519e-01
-3.06884408e-01 -1.33180588e-01 6.89020813e-01 5.71308494e-01
9.89603996e-01 -8.15491900e-02 1.00532487e-01 -4.08589900e-01
2.69745439e-01 -4.58970845e-01 -1.03675830e+00 -3.31955940e-01
-1.40935153e-01 -2.62823850e-01 2.83779114e-01 -2.43024945e-01
-4.32853222e-01 6.78487241e-01 -1.30721495e-01 -9.48802114e-01
-3.59028816e-01 2.59908903e-02 -1.68206826e-01 -2.64332503e-01
6.67391479e-01 3.00625145e-01 -8.10091868e-02 -4.81568605e-01
6.73219979e-01 5.69781601e-01 6.24041140e-01 -2.73445278e-01
1.28004587e+00 5.73128879e-01 1.77748322e-01 -6.79427505e-01
-5.29739916e-01 -6.33969605e-01 -1.11000359e+00 -2.91256875e-01
8.16466153e-01 -1.28741336e+00 -9.09457386e-01 -3.67508046e-02
-1.41084242e+00 4.83639985e-02 -3.99425685e-01 4.58241075e-01
-5.15850961e-01 8.71781111e-02 -1.54995933e-01 -9.17942464e-01
-3.39852810e-01 -1.14906979e+00 1.74821961e+00 1.97811455e-01
2.65634134e-02 -2.46002376e-01 -1.68437853e-01 2.04760224e-01
-3.90311726e-03 4.31244969e-01 7.30490386e-01 -3.84577602e-01
-1.16249728e+00 -5.65397680e-01 -6.80467427e-01 1.04874708e-01
-3.57047588e-01 -7.62231201e-02 -9.67304528e-01 -2.79560238e-01
-2.68841624e-01 -1.27699137e-01 1.16395199e+00 3.97899419e-01
1.24603188e+00 3.28620553e-01 -8.37026715e-01 7.56700039e-01
1.22438788e+00 -1.70727015e-01 1.37165800e-01 1.67757437e-01
6.91536427e-01 2.68440396e-01 7.09477484e-01 4.63971227e-01
5.09172916e-01 6.92006946e-01 9.85727549e-01 -1.56396292e-02
-3.34809348e-02 -5.13336182e-01 5.03395759e-02 4.91342582e-02
4.74678054e-02 1.42157048e-01 -7.91735470e-01 4.57345665e-01
-1.68523467e+00 -6.91014826e-01 -1.55842081e-01 2.14478636e+00
2.28585497e-01 6.19372904e-01 3.54707420e-01 2.71224231e-02
7.33054340e-01 3.83805856e-02 -8.36001754e-01 -3.80973108e-02
2.14619741e-01 1.23906292e-01 5.53933322e-01 6.78419322e-02
-1.36962914e+00 1.13675976e+00 5.02253008e+00 5.08250475e-01
-1.15081346e+00 1.03159539e-01 1.38864994e-01 -2.66380638e-01
7.18815532e-03 -2.28371724e-01 -1.28376794e+00 2.91479081e-01
4.01053548e-01 3.62854749e-02 -8.94164443e-02 1.39369524e+00
3.03737253e-01 1.64629668e-01 -1.12016952e+00 1.06759048e+00
-2.44922370e-01 -1.29643869e+00 1.37245700e-01 3.66324127e-01
2.28531778e-01 5.04441321e-01 7.88235441e-02 5.71489334e-01
2.12826371e-01 -1.02577436e+00 1.10826993e+00 7.57758915e-02
6.73261464e-01 -9.29498732e-01 4.68728602e-01 6.30595982e-01
-1.59748399e+00 -1.37724444e-01 -1.06085694e+00 7.75444806e-02
2.72736579e-01 6.90502226e-01 -1.27954566e+00 1.83097616e-01
8.18258226e-01 5.59529185e-01 -4.81381327e-01 1.52932703e+00
-1.43442571e-01 1.36432841e-01 -7.44096458e-01 -9.34836045e-02
4.60684150e-01 -2.22604975e-01 7.61805236e-01 1.15490091e+00
4.55383182e-01 -3.10337245e-02 5.47199845e-01 1.51151288e+00
-1.05489410e-01 1.54346898e-02 -1.02151847e+00 5.66728950e-01
7.69180059e-01 1.37567782e+00 -8.33995104e-01 -1.94397509e-01
-3.36244911e-01 5.28549552e-01 4.64425504e-01 -9.70831513e-02
-8.34054708e-01 -2.80657321e-01 9.53758061e-01 4.20394093e-01
7.56181657e-01 -6.42543435e-01 -2.55397558e-01 -7.53126740e-01
1.79300234e-01 -2.75803357e-02 7.40965549e-03 -7.77453244e-01
-1.14366877e+00 4.56921458e-01 7.18562976e-02 -1.41288888e+00
-1.82901219e-01 -8.23019683e-01 -6.81178927e-01 1.10927320e+00
-1.74162531e+00 -1.30890965e+00 -6.48323596e-01 4.59710300e-01
5.00633597e-01 -6.73307329e-02 3.98238182e-01 1.75126731e-01
-4.43735451e-01 2.77087748e-01 -3.56700391e-01 2.57581860e-01
2.92333364e-01 -1.15277076e+00 1.06032348e+00 5.42034090e-01
2.38300368e-01 4.41016942e-01 2.88670033e-01 -8.09690893e-01
-1.44819939e+00 -1.74021709e+00 5.90293705e-01 -6.06048584e-01
2.83207327e-01 -7.80751526e-01 -8.36494505e-01 5.98608971e-01
-5.47774255e-01 5.75959086e-01 2.55530804e-01 -1.45546809e-01
-5.71420670e-01 -1.44577235e-01 -1.19600248e+00 3.66195261e-01
1.39738929e+00 -3.98012966e-01 -4.40336972e-01 4.60045576e-01
8.97387445e-01 -7.19650209e-01 -5.77274740e-01 4.12020773e-01
4.08172965e-01 -8.39355230e-01 1.51486897e+00 -5.10272682e-01
4.52396981e-02 -6.06029809e-01 -1.13151595e-01 -1.08716238e+00
-5.26275575e-01 8.90226513e-02 -1.37619227e-01 7.27667928e-01
3.25571224e-02 -4.46837276e-01 1.08780491e+00 2.30293810e-01
-5.64679623e-01 -7.36567438e-01 -1.27878559e+00 -8.28967512e-01
1.56042511e-02 -7.86191106e-01 9.03715789e-01 3.77598494e-01
-6.40908301e-01 4.07170594e-01 1.89945593e-01 4.84191835e-01
8.74076247e-01 3.05181593e-01 1.11125827e+00 -1.80677462e+00
2.98262477e-01 -8.05459261e-01 -9.11616683e-01 -1.44947672e+00
1.75367087e-01 -1.11210871e+00 2.02387959e-01 -1.48798394e+00
5.83354849e-03 -7.50667393e-01 -1.13697154e-02 2.88374692e-01
3.22889611e-02 3.64793837e-01 3.08274776e-01 1.44887492e-01
-4.73780781e-01 8.10985863e-01 1.10398257e+00 -3.16808045e-01
-3.21387768e-01 3.64886284e-01 -4.68475968e-01 6.85441136e-01
4.78501886e-01 -4.46133763e-01 -1.11916192e-01 -4.30047542e-01
-1.66557997e-01 -3.18748176e-01 1.01117110e+00 -1.15266371e+00
2.21366212e-01 4.98000197e-02 7.55115867e-01 -1.49122429e+00
8.07664096e-01 -9.06471670e-01 -2.78331369e-01 4.67327356e-01
1.26460463e-01 -2.02212185e-01 1.80567309e-01 6.46008968e-01
1.25225142e-01 1.19701684e-01 7.43900299e-01 -2.54668713e-01
-8.27406287e-01 6.97269976e-01 7.74173811e-02 -4.08034861e-01
1.16529465e+00 -6.22088552e-01 -4.22376879e-02 1.59077585e-01
-3.64987522e-01 4.94925559e-01 4.39878911e-01 5.67469776e-01
1.12308240e+00 -1.42875004e+00 -7.75162518e-01 4.99908328e-01
2.34891430e-01 8.52132857e-01 3.72484028e-02 4.83435988e-01
-3.62038553e-01 7.05797315e-01 -1.16654277e-01 -1.36005628e+00
-1.12959397e+00 5.55597246e-01 3.51582170e-01 2.43428469e-01
-1.12799537e+00 8.96908581e-01 3.44242036e-01 -6.73308611e-01
3.01386118e-01 -7.85891831e-01 -1.13921136e-01 -1.63003609e-01
3.25664192e-01 1.29164785e-01 1.13437004e-01 -7.69296944e-01
-5.44724464e-01 8.14344883e-01 -9.81702730e-02 3.58204216e-01
1.39506030e+00 1.68851301e-01 3.18571150e-01 1.86535239e-01
1.07427526e+00 -5.16055584e-01 -1.46465683e+00 -3.85296464e-01
-2.26064637e-01 -6.60866022e-01 2.33847335e-01 -2.13332444e-01
-1.01469970e+00 1.04569793e+00 8.41657460e-01 -3.15474644e-02
4.42065895e-01 2.91523814e-01 7.98551023e-01 6.24971449e-01
6.06009364e-01 -5.47241092e-01 -8.65047798e-02 5.88133633e-01
8.68746996e-01 -1.42409825e+00 1.75899819e-01 -6.15258038e-01
-1.83338091e-01 1.02348614e+00 7.36696661e-01 -4.81346548e-01
7.47613668e-01 5.25676534e-02 -3.40752721e-01 -3.46065938e-01
-4.46956635e-01 -4.41688448e-01 3.56882393e-01 8.90897691e-01
8.52533244e-03 2.84011722e-01 2.70979553e-01 6.50763869e-01
-4.23517764e-01 -1.72246665e-01 -2.94665489e-02 8.26476157e-01
-9.55685854e-01 -5.85211992e-01 -5.94426572e-01 5.21158576e-01
4.23410028e-01 2.20030010e-01 -3.89352173e-01 1.06270969e+00
3.97942096e-01 3.71873617e-01 5.06245077e-01 -4.39144462e-01
6.86638117e-01 -7.80858994e-02 3.25473964e-01 -9.36621666e-01
-4.23925906e-01 -8.17175061e-02 -4.11830902e-01 -6.31735444e-01
1.42223448e-01 -7.98077106e-01 -1.38509250e+00 -2.27319524e-01
-3.63829255e-01 -2.88127065e-01 1.01749730e+00 7.20760345e-01
4.05596495e-01 3.19123864e-01 5.61307967e-01 -1.56121969e+00
-6.20981812e-01 -7.46583581e-01 -4.55135882e-01 1.44301072e-01
2.74353206e-01 -1.14029753e+00 5.85053042e-02 -5.82961738e-01] | [7.744967937469482, -2.8010919094085693] |
d28977e4-95d4-4310-8c95-df076e2a4276 | dsc-iit-ism-at-semeval-2020-task-6-boosting | 2009.08180 | null | https://arxiv.org/abs/2009.08180v1 | https://arxiv.org/pdf/2009.08180v1.pdf | DSC IIT-ISM at SemEval-2020 Task 6: Boosting BERT with Dependencies for Definition Extraction | We explore the performance of Bidirectional Encoder Representations from Transformers (BERT) at definition extraction. We further propose a joint model of BERT and Text Level Graph Convolutional Network so as to incorporate dependencies into the model. Our proposed model produces better results than BERT and achieves comparable results to BERT with fine tuned language model in DeftEval (Task 6 of SemEval 2020), a shared task of classifying whether a sentence contains a definition or not (Subtask 1). | ['Priyanshu Kumar', 'Aman Sinha', 'Aadarsh Singh'] | 2020-09-17 | null | https://aclanthology.org/2020.semeval-1.93 | https://aclanthology.org/2020.semeval-1.93.pdf | semeval-2020 | ['definition-extraction'] | ['natural-language-processing'] | [ 2.39656731e-01 6.60706580e-01 -3.11470479e-01 -4.55228478e-01
-6.68749392e-01 -8.55274081e-01 8.69689286e-01 2.56539404e-01
-3.27466995e-01 9.11423206e-01 7.50366747e-01 -1.20364511e+00
3.57864380e-01 -1.32134509e+00 -8.58025253e-01 2.60578454e-01
-1.28281638e-01 4.56216186e-01 3.32442105e-01 -5.74425220e-01
-2.77805477e-01 9.44354236e-02 -4.42767829e-01 7.10189164e-01
4.37443316e-01 8.00502896e-01 1.39666900e-01 9.67161894e-01
-3.34151328e-01 1.59544086e+00 -9.39360142e-01 -1.06114280e+00
-1.73041970e-01 -3.88224095e-01 -1.62767720e+00 -2.26992190e-01
6.66704655e-01 -2.24081859e-01 -5.45213878e-01 9.78835762e-01
2.47183815e-01 -2.65831035e-02 6.57848775e-01 -9.19112980e-01
-9.37943280e-01 1.63941431e+00 -2.55461067e-01 7.15337038e-01
4.20511901e-01 8.81573856e-02 1.79921150e+00 -7.09209085e-01
9.13510621e-01 1.40720916e+00 8.04089129e-01 4.15088534e-01
-1.26115727e+00 -6.29450083e-01 2.60556459e-01 4.21011835e-01
-1.11055863e+00 -5.12307405e-01 6.95159912e-01 -2.21135631e-01
2.10375738e+00 1.20174110e-01 5.12930274e-01 1.26620543e+00
6.84298694e-01 7.55328953e-01 5.29146314e-01 -1.27821550e-01
-2.60575235e-01 -3.15234244e-01 6.10835075e-01 1.06864500e+00
6.24176934e-02 -2.56789774e-01 -2.97505111e-01 5.48285581e-02
3.77419591e-01 -6.38706028e-01 -3.56332570e-01 4.45085824e-01
-1.02955711e+00 8.31989944e-01 6.76381171e-01 4.08717752e-01
-3.15090150e-01 7.71577835e-01 9.76536334e-01 7.60267735e-01
4.62951273e-01 6.42102122e-01 -7.04782248e-01 -1.41509786e-01
-8.30664515e-01 1.29376516e-01 1.03435409e+00 1.46144629e+00
4.61495638e-01 2.12640435e-01 -6.57434225e-01 4.99208421e-01
3.56179804e-01 2.20443383e-01 1.57824919e-01 -2.72827268e-01
1.05270469e+00 6.50986612e-01 -4.69724476e-01 -5.50438583e-01
-4.00887012e-01 -9.67186868e-01 -8.07914495e-01 -5.06899476e-01
-1.47152737e-01 -4.32572126e-01 -9.15703535e-01 1.59386754e+00
-3.20528001e-01 6.76718205e-02 1.07804753e-01 3.95609915e-01
1.68042147e+00 5.84295452e-01 2.15310395e-01 3.01156994e-02
1.18497252e+00 -9.52119470e-01 -9.77008641e-01 -3.01756024e-01
9.39999878e-01 -4.68135029e-01 6.66112661e-01 1.07551597e-01
-1.15158689e+00 -5.77771068e-01 -1.24329007e+00 -4.54983830e-01
-4.27044749e-01 -1.33357840e-02 7.77917802e-01 2.78938293e-01
-1.36145210e+00 4.28880036e-01 -6.43403649e-01 -5.72563112e-02
4.79198098e-01 3.08850050e-01 -2.58543104e-01 3.08142364e-01
-1.90446246e+00 1.16812968e+00 6.34509146e-01 1.32495776e-01
-1.45222712e+00 -6.60053372e-01 -1.38689578e+00 1.45144030e-01
2.28139356e-01 -8.20289135e-01 1.40524876e+00 -7.11758971e-01
-1.25369132e+00 8.62812221e-01 -1.48682639e-01 -1.20356786e+00
-2.59268284e-03 -1.57721236e-01 -5.50162613e-01 -1.35875061e-01
1.93158641e-01 2.26536915e-01 3.33133042e-01 -8.18879306e-01
-4.04517651e-01 6.92963451e-02 1.05416882e+00 -2.43190438e-01
-6.29128888e-04 2.09239513e-01 -2.97327965e-01 -4.50589955e-01
-6.63032591e-01 -2.81043798e-01 -1.78149417e-01 -6.78050339e-01
-8.82545233e-01 -9.49228704e-01 5.48708856e-01 -1.00494289e+00
1.62176454e+00 -1.60886729e+00 2.36506030e-01 -3.99940133e-01
7.02067792e-01 1.81701019e-01 -1.32575989e-01 8.73025358e-01
-3.34513366e-01 5.17197311e-01 -7.82174096e-02 -6.23327732e-01
2.50678778e-01 4.11326051e-01 -4.04232383e-01 1.65292472e-01
3.33150536e-01 1.53539491e+00 -1.12240517e+00 -4.88191426e-01
1.90176859e-01 3.23355466e-01 -3.90732974e-01 4.03353840e-01
-5.74313104e-01 -2.50508666e-01 -2.50496179e-01 3.21646959e-01
6.02166533e-01 -4.64726985e-01 4.76573467e-01 -4.32622850e-01
3.11781138e-01 1.41034639e+00 -4.27449286e-01 1.68553162e+00
-9.73890901e-01 9.01461720e-01 3.01031698e-03 -8.48328829e-01
8.03304911e-01 5.47571421e-01 -2.01517448e-01 -6.90601408e-01
1.14082895e-01 -2.51431555e-01 8.85664895e-02 -3.07752520e-01
5.97633421e-01 -3.19773316e-01 -2.74351031e-01 9.67654660e-02
7.08509743e-01 -3.55216652e-01 4.84882623e-01 8.04511309e-01
1.71234095e+00 6.37707859e-02 4.65000600e-01 -3.00831944e-01
6.78858817e-01 -7.19300136e-02 4.97447997e-01 5.87579966e-01
1.28144979e-01 2.53583014e-01 1.12474847e+00 -4.10075068e-01
-6.24252379e-01 -1.21621370e+00 1.12732388e-02 7.48693466e-01
-2.36478522e-01 -1.26471639e+00 -2.63019562e-01 -1.32538140e+00
-1.16982482e-01 1.30059040e+00 -7.58439302e-01 -1.37943089e-01
-8.78933132e-01 -2.96180278e-01 1.14206874e+00 7.49791086e-01
8.88176858e-01 -1.08701813e+00 -1.43328294e-01 3.73236805e-01
-2.10141376e-01 -1.43866897e+00 -4.07685220e-01 4.98600364e-01
-3.91317904e-01 -1.23135853e+00 1.28856059e-02 -1.00157464e+00
1.39680803e-01 -5.36995351e-01 2.15303922e+00 4.41349810e-04
3.29998046e-01 7.43309334e-02 -5.32152593e-01 -1.16545670e-01
-6.49666488e-01 5.61119497e-01 -6.86450601e-01 -6.01525247e-01
2.37276092e-01 -4.21248794e-01 -1.16575703e-01 -4.29977685e-01
-7.26738751e-01 2.40541279e-01 1.87931970e-01 8.44208896e-01
1.23433895e-01 -1.14221368e-02 7.08993554e-01 -1.50908113e+00
1.06961572e+00 -4.82101917e-01 -4.08658147e-01 3.96646589e-01
-6.44567013e-01 1.67193636e-01 1.11745095e+00 2.65674204e-01
-9.51931119e-01 -4.16568965e-01 -6.80638552e-01 -2.44780686e-02
2.77647704e-01 9.51971054e-01 -2.58973926e-01 5.61701834e-01
4.33950216e-01 1.00811489e-01 -5.75639665e-01 -4.64412093e-01
6.34319425e-01 7.19237804e-01 6.17147982e-01 -5.75427175e-01
7.22479880e-01 -2.38777965e-01 -2.51939207e-01 -2.93462098e-01
-1.11102402e+00 -2.06042677e-01 -7.75087237e-01 6.07131422e-03
8.02836478e-01 -9.79940534e-01 -4.18944150e-01 -2.56563798e-02
-1.74396908e+00 -5.81397831e-01 -1.53420746e-01 -1.16022810e-01
-1.77353501e-01 9.66644213e-02 -9.67987716e-01 -4.14036244e-01
-9.18593645e-01 -8.35185349e-01 1.08062959e+00 -2.91072220e-01
-1.95076376e-01 -1.60927331e+00 8.32179636e-02 2.97486931e-01
3.95046145e-01 4.95718658e-01 1.25404513e+00 -4.78850931e-01
-2.81623453e-01 2.99240313e-02 -3.20913643e-01 6.62753165e-01
2.93900698e-01 -6.03995211e-02 -7.15589166e-01 -3.52539629e-01
-5.08259773e-01 -5.23878813e-01 1.19973969e+00 3.12916301e-02
1.12939322e+00 -5.30955613e-01 -2.01836556e-01 5.55985212e-01
1.53984571e+00 -2.09616333e-01 6.82978272e-01 1.91863198e-02
6.61404967e-01 -8.30230564e-02 -1.00592002e-01 1.55591547e-01
7.64386117e-01 4.60702449e-01 6.62261665e-01 -1.72784746e-01
-9.26514983e-01 -5.14984548e-01 5.39874256e-01 1.14502037e+00
1.84250236e-01 -8.26774538e-01 -1.01926661e+00 9.82252121e-01
-1.42599261e+00 -8.35859537e-01 -4.63491738e-01 1.50155449e+00
1.20310605e+00 6.88997746e-01 -2.49629408e-01 -2.53892951e-02
3.75130951e-01 7.86021292e-01 5.49191721e-02 -1.05525720e+00
-3.19264084e-01 9.76572633e-01 7.18230426e-01 8.84033024e-01
-1.21868563e+00 1.32387745e+00 6.76533127e+00 7.17624068e-01
-8.32707226e-01 3.70078444e-01 3.00518900e-01 4.41158861e-01
-8.89647484e-01 2.40868956e-01 -8.25463235e-01 2.42408186e-01
1.36705518e+00 -3.62299472e-01 4.64636087e-02 2.63514310e-01
-8.91774222e-02 4.50935960e-01 -1.45867765e+00 2.17304349e-01
1.76885545e-01 -1.85182619e+00 4.11169827e-01 -3.43532026e-01
3.79671752e-01 4.40205693e-01 -3.98540050e-01 9.33744431e-01
8.95146489e-01 -1.48125994e+00 6.44619048e-01 2.57184982e-01
1.12595952e+00 -6.72632158e-01 1.06778264e+00 2.91623510e-02
-1.50326502e+00 4.58503395e-01 -3.16752583e-01 -5.30767262e-01
2.18023285e-01 6.69963479e-01 -1.46821630e+00 9.20567989e-01
-4.22814079e-02 1.45440912e+00 -7.65142918e-01 4.38511014e-01
-1.04770076e+00 1.27301133e+00 1.84456930e-01 -3.23369771e-01
4.72328752e-01 3.58310699e-01 5.25185049e-01 1.81750751e+00
-7.39743173e-01 -2.53222615e-01 3.41542512e-02 1.10017800e+00
-8.23755682e-01 -3.62860233e-01 -6.51825428e-01 -2.61619449e-01
2.86179245e-01 1.09383118e+00 -1.44019753e-01 -6.47160470e-01
-6.32799327e-01 9.12120461e-01 8.60536277e-01 4.25420970e-01
-1.04190493e+00 -5.53718030e-01 3.36342782e-01 -1.63334921e-01
4.01345938e-01 -5.14858663e-01 -1.39713570e-01 -1.36253309e+00
-1.42810583e-01 -7.24549651e-01 6.00387514e-01 -6.37912810e-01
-1.15835762e+00 9.06128705e-01 -1.60944045e-01 -6.86576307e-01
-3.19413751e-01 -6.33387566e-01 -7.01876581e-01 8.42181742e-01
-1.66825640e+00 -1.63469684e+00 1.55482620e-01 5.54556668e-01
6.75403535e-01 8.49604085e-02 1.13914406e+00 5.29714346e-01
-3.94791573e-01 7.67940640e-01 -6.04669690e-01 9.10746694e-01
2.93165863e-01 -1.81840301e+00 1.06764662e+00 1.07358992e+00
4.43869591e-01 6.53334856e-01 6.08627796e-01 -6.65564299e-01
-1.34970403e+00 -1.53892732e+00 1.52950406e+00 -4.75791246e-01
1.03979945e+00 -7.21054077e-01 -3.92731667e-01 1.28216195e+00
1.02851117e+00 -1.81205019e-01 3.13420385e-01 5.52866340e-01
-6.53521836e-01 4.23985422e-02 -6.32891178e-01 2.59699076e-01
1.25518167e+00 -8.69599998e-01 -1.21458757e+00 5.74432790e-01
1.36404455e+00 -6.73256636e-01 -1.07228255e+00 4.57259566e-01
9.13341343e-02 -4.57035899e-01 6.58098459e-01 -1.04013073e+00
9.02746081e-01 -2.80643534e-02 -6.81125149e-02 -1.38529909e+00
-4.59714741e-01 -6.76044285e-01 -4.06596184e-01 1.39787602e+00
1.09224319e+00 -3.70604098e-01 3.41530651e-01 -4.93312394e-03
-3.54026675e-01 -8.33749890e-01 -1.11540461e+00 -7.21012235e-01
4.86707628e-01 -6.59286797e-01 6.08559072e-01 6.64822698e-01
-6.03596754e-02 1.27151537e+00 -1.50736436e-01 -1.27682220e-02
1.89560905e-01 3.55806142e-01 4.03344154e-01 -4.51975524e-01
-4.31129485e-01 -2.73030907e-01 -5.51553011e-01 -1.39821362e+00
7.48635113e-01 -1.63669622e+00 -2.60787487e-01 -2.42114782e+00
1.59287646e-01 1.71449706e-01 -5.05071640e-01 6.73960626e-01
-3.22258100e-03 -4.80380811e-04 -2.72349231e-02 -5.09380400e-01
-8.06691289e-01 7.50385940e-01 1.23979390e+00 -8.51710260e-01
4.69616383e-01 -1.06711522e-01 -6.20561481e-01 8.84380564e-02
3.42519939e-01 -2.23671317e-01 -3.36112261e-01 -1.16929626e+00
4.51205254e-01 1.56557828e-01 1.89474210e-01 -5.04083753e-01
-5.78364497e-03 2.46209368e-01 3.30051705e-02 -5.77524543e-01
-2.95123761e-03 -3.52200210e-01 -3.45204353e-01 3.86966467e-01
-7.00964987e-01 4.64997798e-01 1.37026995e-01 2.65248269e-01
-5.51646590e-01 1.26848109e-02 2.62110144e-01 -1.74102604e-01
-6.55801237e-01 3.53986830e-01 -5.23012400e-01 5.09510934e-01
2.40638003e-01 5.95235705e-01 -6.22260511e-01 -5.64256668e-01
-7.55176246e-01 5.72782815e-01 -1.63526267e-01 5.57665169e-01
7.12099552e-01 -1.09779644e+00 -1.10238945e+00 -1.86690286e-01
-7.46463090e-02 8.14050809e-02 -5.85863590e-01 6.13845646e-01
-6.42118514e-01 5.67238331e-01 3.56326401e-01 -8.54360014e-02
-1.08064079e+00 3.82150590e-01 8.02385330e-01 -8.90314519e-01
-7.75882304e-01 1.33983111e+00 5.22542447e-02 -4.97278959e-01
5.72077706e-02 -8.94347429e-01 -3.82222146e-01 -2.37434328e-01
3.05289060e-01 -1.31156296e-02 3.31295669e-01 -3.72178018e-01
-8.24509203e-01 -1.07366815e-01 -2.26554424e-01 2.50302851e-01
1.47480273e+00 1.83036700e-01 -5.38876295e-01 3.16468835e-01
1.49360573e+00 9.53885145e-04 -5.15978992e-01 -3.71616155e-01
1.69938520e-01 1.98098555e-01 3.69353563e-01 -1.01117003e+00
-1.05711043e+00 9.72115755e-01 -1.94406927e-01 1.50587782e-01
1.03403616e+00 1.95512950e-01 1.17629826e+00 3.88406932e-01
5.12596011e-01 -7.10036933e-01 -1.64603323e-01 1.13544631e+00
1.34224498e+00 -8.74319136e-01 -3.82246971e-02 -4.73196000e-01
-3.73019636e-01 1.27984774e+00 4.93256986e-01 -5.25271714e-01
6.21320307e-01 8.65246654e-01 -2.67740399e-01 -5.70140243e-01
-1.42797613e+00 -3.14200133e-01 3.01912993e-01 4.16275620e-01
1.02320755e+00 5.17705023e-01 -5.02839148e-01 7.30843484e-01
-4.80585247e-01 -1.07407503e-01 5.76920807e-01 5.22946656e-01
1.72180310e-01 -1.10546815e+00 7.11832225e-01 6.03585064e-01
-7.46945977e-01 -8.95203292e-01 -7.95755267e-01 7.23757923e-01
1.78144231e-01 1.25287688e+00 -2.09406868e-01 -6.94035113e-01
2.92113602e-01 -3.97321675e-03 6.52730823e-01 -1.14116991e+00
-1.21275365e+00 -4.06529158e-01 1.30626321e+00 -4.87846375e-01
-3.00303429e-01 -5.63746318e-02 -1.29588485e+00 -2.28518978e-01
-3.39356840e-01 2.53692180e-01 3.30672711e-02 9.59406972e-01
-1.98914260e-02 1.29896581e+00 4.16567832e-01 3.05769980e-01
-4.49014395e-01 -1.34809375e+00 -5.00364542e-01 2.69100547e-01
5.10963857e-01 -2.05330655e-01 -2.41711184e-01 6.15764521e-02] | [9.968160629272461, 9.0113525390625] |
b6471f4b-485f-4446-8725-e93b92ff70cf | how-to-be-fair-and-diverse | 1610.07183 | null | http://arxiv.org/abs/1610.07183v1 | http://arxiv.org/pdf/1610.07183v1.pdf | How to be Fair and Diverse? | Due to the recent cases of algorithmic bias in data-driven decision-making,
machine learning methods are being put under the microscope in order to
understand the root cause of these biases and how to correct them. Here, we
consider a basic algorithmic task that is central in machine learning:
subsampling from a large data set. Subsamples are used both as an end-goal in
data summarization (where fairness could either be a legal, political or moral
requirement) and to train algorithms (where biases in the samples are often a
source of bias in the resulting model). Consequently, there is a growing effort
to modify either the subsampling methods or the algorithms themselves in order
to ensure fairness. However, in doing so, a question that seems to be
overlooked is whether it is possible to produce fair subsamples that are also
adequately representative of the feature space of the data set - an important
and classic requirement in machine learning. Can diversity and fairness be
simultaneously ensured? We start by noting that, in some applications,
guaranteeing one does not necessarily guarantee the other, and a new approach
is required. Subsequently, we present an algorithmic framework which allows us
to produce both fair and diverse samples. Our experimental results on an image
summarization task show marked improvements in fairness without compromising
feature diversity by much, giving us the best of both the worlds. | ['Tarun Kathuria', 'Amit Deshpande', 'L. Elisa Celis', 'Nisheeth K. Vishnoi'] | 2016-10-23 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [ 5.79910874e-01 2.85896808e-01 -3.15279603e-01 -6.45311594e-01
-5.13700604e-01 -3.22982907e-01 7.30295777e-01 3.53942305e-01
-5.84626377e-01 1.07782936e+00 4.45077032e-01 -1.95380241e-01
-3.56285632e-01 -7.23399818e-01 -3.07495147e-01 -9.82778668e-01
4.41876143e-01 4.26869601e-01 -6.59118816e-02 -2.16770962e-01
6.87781334e-01 4.68670696e-01 -1.89942539e+00 9.91559774e-03
1.19375205e+00 8.32794845e-01 -3.42102557e-01 5.46838343e-01
-2.10444987e-01 8.69001985e-01 -7.09913433e-01 -6.77393377e-01
3.22956324e-01 -8.98621619e-01 -9.49865699e-01 1.67073831e-01
5.06662607e-01 -8.78550634e-02 4.05408055e-01 1.29903328e+00
6.42337143e-01 1.44822896e-01 9.64352071e-01 -1.21024346e+00
-3.01658928e-01 7.04173267e-01 -6.93883598e-01 -6.27659336e-02
8.83405730e-02 7.00121745e-02 1.16110528e+00 -2.78796613e-01
4.94014472e-01 1.23931587e+00 2.65076905e-01 6.70317113e-01
-1.33310556e+00 -5.41010678e-01 7.24482015e-02 -7.56769180e-02
-9.73564744e-01 -7.52682388e-01 6.88705385e-01 -4.52878207e-01
2.96805233e-01 7.91100919e-01 7.32061327e-01 8.81561816e-01
1.69175282e-01 6.95662916e-01 1.30125046e+00 -5.33508599e-01
5.76207697e-01 4.22595263e-01 2.37303972e-01 1.60681278e-01
8.12970102e-01 -1.39033264e-02 -3.93102705e-01 -4.85570163e-01
2.42796883e-01 -1.79268822e-01 -3.31501603e-01 -4.67698574e-01
-9.10044849e-01 1.08669174e+00 3.54353964e-01 2.75971532e-01
-4.46481824e-01 2.43571401e-02 5.10670364e-01 4.02402699e-01
5.54790974e-01 7.86742032e-01 -5.30692637e-02 3.63691747e-02
-1.15653968e+00 6.91176057e-01 7.87403584e-01 3.73943299e-01
4.50372279e-01 -1.60793722e-01 -2.56449003e-02 5.96386850e-01
-1.84548691e-01 1.65500060e-01 4.36211675e-01 -1.05164254e+00
-3.27247679e-02 7.24116087e-01 1.61016732e-01 -1.01163089e+00
-5.85434698e-02 -3.88531238e-01 -1.09353483e+00 6.53259456e-01
6.28524065e-01 3.11737619e-02 -4.40655321e-01 1.91575563e+00
6.01990461e-01 -5.65704584e-01 -8.24177563e-02 7.97684431e-01
2.73507714e-01 2.10983500e-01 1.17777117e-01 -7.46513486e-01
1.21304905e+00 -2.95676827e-01 -6.51439130e-01 8.51868186e-03
4.33703274e-01 -7.01940536e-01 9.38069761e-01 5.58008552e-01
-1.07427192e+00 -1.71789333e-01 -1.04763126e+00 -2.86417603e-02
-1.28272802e-01 -3.40355903e-01 4.16076332e-01 8.94636691e-01
-5.97280324e-01 7.76871860e-01 -1.98282957e-01 -4.46116894e-01
7.90510654e-01 2.25006729e-01 -3.61890197e-01 6.84580654e-02
-8.54884923e-01 1.11728895e+00 3.47312182e-01 -1.16541766e-01
-2.27776766e-01 -5.28300703e-01 -3.74758124e-01 1.43719137e-01
6.61945403e-01 -9.32601392e-01 9.17578101e-01 -1.39074636e+00
-1.04285324e+00 1.12938154e+00 -6.67008013e-02 -4.61664528e-01
1.18434525e+00 2.24102475e-02 2.72387825e-02 -3.18054646e-01
3.86968330e-02 4.64482754e-01 9.15396392e-01 -1.39223719e+00
-9.58636224e-01 -5.45285642e-01 -2.06123013e-02 1.88525096e-01
-2.11069837e-01 8.57501999e-02 5.87314904e-01 -5.16751707e-01
-1.22567371e-01 -8.20285738e-01 -4.37437773e-01 9.06961113e-02
-5.56477964e-01 -3.82013321e-01 3.75273764e-01 -2.09051222e-01
1.14552438e+00 -2.08875012e+00 1.18565634e-01 1.81566715e-01
4.91825819e-01 3.52523863e-01 3.59247893e-01 2.41979942e-01
-1.23044774e-01 3.13869566e-01 -4.19339031e-01 -1.72587335e-01
1.01574712e-01 1.22660190e-01 -4.53277588e-01 6.60610616e-01
1.61443010e-01 5.54603815e-01 -8.00029516e-01 -5.45147955e-01
1.33783102e-01 3.70401628e-02 -6.81833267e-01 6.19945936e-02
-2.50859223e-02 3.45702261e-01 -4.64212358e-01 2.45242268e-01
5.97380996e-01 1.27354816e-01 5.39330244e-02 1.11764446e-01
-1.01500295e-01 1.59220919e-01 -1.29151428e+00 9.18986380e-01
4.46344428e-02 4.78288740e-01 -3.85341309e-02 -1.23821700e+00
7.51648366e-01 6.19009994e-02 3.99646938e-01 -5.20317018e-01
4.82879758e-01 5.25147974e-01 2.38454565e-01 -4.14921701e-01
8.27859104e-01 -7.42008448e-01 -1.22961365e-01 7.94563472e-01
-5.51673412e-01 -3.64355892e-01 2.12832108e-01 2.95063239e-02
7.13511348e-01 -3.70797813e-01 9.91041899e-01 -6.31176114e-01
5.48026979e-01 2.50912368e-01 7.86162198e-01 7.23154843e-01
-3.91335636e-01 6.54600382e-01 9.24638033e-01 -6.22606695e-01
-1.12586272e+00 -7.51380265e-01 -2.26451859e-01 9.01842296e-01
-1.54482290e-01 -6.92965975e-03 -7.64588535e-01 -8.66434991e-01
1.40744686e-01 9.90963876e-01 -8.64049852e-01 -2.76800573e-01
-3.31078082e-01 -9.95252669e-01 2.26547211e-01 1.06367767e-02
2.30578661e-01 -8.98255765e-01 -1.32198513e+00 9.07949880e-02
-1.25870854e-01 -5.82185566e-01 -2.11736575e-01 2.83292025e-01
-8.43517661e-01 -1.32336497e+00 -6.29818261e-01 -1.66488066e-01
8.39834929e-01 2.02020645e-01 1.05908322e+00 2.14373812e-01
-2.38563061e-01 -8.90642032e-02 -3.09121728e-01 -9.74714637e-01
-7.76827991e-01 1.12464614e-02 4.44386415e-02 -3.03064939e-02
4.36150789e-01 -5.74597299e-01 -5.23323417e-01 1.33579403e-01
-1.13810492e+00 4.52254117e-02 3.96259576e-01 7.90130854e-01
1.96412250e-01 1.67926356e-01 8.09120238e-01 -1.19970560e+00
7.62364268e-01 -2.61948138e-01 -4.52069581e-01 2.52897322e-01
-7.11924911e-01 5.71930967e-02 6.54406071e-01 -2.29383469e-01
-7.73692250e-01 -2.23607659e-01 -3.60464752e-02 1.27359226e-01
-2.82668263e-01 2.42353335e-01 -4.00932014e-01 1.95166796e-01
9.88423944e-01 -1.07837580e-02 4.25512820e-01 -1.80300012e-01
3.59279484e-01 8.10006022e-01 2.24257067e-01 -4.62263882e-01
5.33877790e-01 6.84258103e-01 1.84332207e-01 -8.88256073e-01
-9.35533166e-01 -1.17585592e-01 -5.09694397e-01 -1.79338381e-01
4.92743731e-01 -3.58106017e-01 -4.96501148e-01 2.16659859e-01
-8.01481903e-01 6.77951425e-02 -8.32519352e-01 1.56601980e-01
-7.01360762e-01 1.41352147e-01 3.05072635e-01 -1.00623441e+00
-3.92235905e-01 -1.15557051e+00 5.06040990e-01 2.67656177e-01
-6.39028132e-01 -6.10912383e-01 1.45146658e-03 4.50488687e-01
5.15771151e-01 5.67449987e-01 1.19738662e+00 -9.18397546e-01
-9.11744162e-02 -4.01785910e-01 -1.61962546e-02 5.01314402e-01
4.77048069e-01 2.51810700e-01 -1.02756631e+00 -2.08921701e-01
4.03507054e-01 -2.70768642e-01 8.46421897e-01 6.11328006e-01
1.00645888e+00 -4.76044714e-01 3.84193063e-02 1.49900317e-01
1.37220573e+00 -1.52453199e-01 5.72303832e-01 2.27692097e-01
3.00113440e-01 1.29065514e+00 5.80684066e-01 6.28185630e-01
2.16858625e-01 4.96360481e-01 4.14399385e-01 7.19671920e-02
1.00334622e-01 8.99345949e-02 -5.74107878e-02 3.05330455e-01
-1.48619741e-01 -9.81807411e-02 -5.86728156e-01 5.24141371e-01
-1.96982872e+00 -1.31649721e+00 -3.52291077e-01 2.73787642e+00
8.77460778e-01 9.25138816e-02 4.99692559e-01 5.87831974e-01
7.58052945e-01 1.53973728e-01 -5.22677064e-01 -7.43502676e-01
-9.93294194e-02 -2.81145066e-01 1.74632251e-01 4.37069744e-01
-8.94962251e-01 4.46542680e-01 6.18634701e+00 7.98853040e-01
-1.19554746e+00 -3.46911162e-01 9.92978573e-01 -3.55668753e-01
-5.09534061e-01 -4.94290665e-02 -3.13259661e-01 4.50529337e-01
5.48896194e-01 -6.74103260e-01 1.46206126e-01 7.52591968e-01
3.85606498e-01 -6.01450980e-01 -1.10336030e+00 6.98508084e-01
2.83675715e-02 -1.04194760e+00 2.52332836e-01 3.34738672e-01
6.66494966e-01 -6.00388527e-01 -7.80599490e-02 -1.13917150e-01
4.65089768e-01 -1.24516690e+00 8.11696887e-01 2.88815171e-01
4.62264270e-01 -1.04125047e+00 7.21016228e-01 6.11772716e-01
-4.08950925e-01 7.32463151e-02 -5.29780626e-01 -3.30258995e-01
3.61512136e-03 1.16922665e+00 -4.56198484e-01 5.74562788e-01
3.17360759e-01 2.13362917e-01 -3.73578638e-01 1.10359848e+00
-5.11972094e-03 2.34359369e-01 -1.90032959e-01 -2.12643728e-01
4.21325117e-02 -1.20150000e-01 6.83468103e-01 8.42042506e-01
8.97274092e-02 -7.29375146e-03 -2.30662048e-01 7.75783896e-01
-1.80551931e-02 2.95972645e-01 -6.40291452e-01 9.10476521e-02
2.66114891e-01 1.20371449e+00 -7.40049481e-01 -2.85864294e-01
2.00220663e-02 5.77894449e-01 2.73820847e-01 -9.28202718e-02
-5.57004511e-01 -2.29299903e-01 9.14863110e-01 2.38683254e-01
-1.45172268e-01 2.75342017e-01 -7.04052091e-01 -1.05241048e+00
-6.23867214e-02 -1.27364695e+00 4.54265565e-01 -1.74334124e-01
-1.38171637e+00 3.31412047e-01 2.65972577e-02 -1.11495519e+00
-3.23232561e-01 -2.47455955e-01 -5.35568178e-01 8.50676477e-01
-1.53459346e+00 -6.82855546e-01 -2.17506602e-01 3.75263482e-01
2.23943010e-01 7.66737200e-03 5.68994999e-01 1.27005912e-02
-3.23908865e-01 3.47983211e-01 2.07142849e-02 -3.06338787e-01
8.39343429e-01 -1.24792826e+00 -6.46848977e-02 8.34279299e-01
-3.59490067e-02 6.51568353e-01 1.27631903e+00 -4.63929683e-01
-1.01954687e+00 -8.06776106e-01 1.23095977e+00 -3.81802022e-01
2.56747663e-01 -8.39282125e-02 -9.16084588e-01 1.43496230e-01
7.94081166e-02 -1.55809298e-01 9.30418551e-01 9.58695859e-02
-2.21533999e-01 -1.26561731e-01 -1.46915281e+00 6.82778358e-01
7.38916993e-01 -2.31477502e-03 -7.09907234e-01 -8.12704265e-02
8.05627182e-02 -8.31920356e-02 -3.10077220e-01 3.26716423e-01
7.68758655e-01 -1.57774150e+00 5.76256394e-01 -6.97233379e-01
6.80690050e-01 -1.48061618e-01 -1.72254354e-01 -1.47067726e+00
-1.31159738e-01 -4.73960131e-01 5.00027478e-01 1.33424461e+00
1.62676364e-01 -7.13406146e-01 8.68860066e-01 9.60726917e-01
3.99822712e-01 -6.53907716e-01 -9.46880400e-01 -6.21983290e-01
5.26032269e-01 -3.82109582e-01 7.38818586e-01 1.05615640e+00
-7.29081854e-02 1.69010386e-01 -4.69875574e-01 -4.23075110e-01
8.64356935e-01 2.58016825e-01 1.07453918e+00 -1.51339889e+00
2.13640571e-01 -1.00655508e+00 -4.29872453e-01 -3.75036120e-01
-3.26025523e-02 -6.40512347e-01 5.13961762e-02 -1.42693508e+00
5.07099867e-01 -3.89928639e-01 -5.18552363e-02 1.20177120e-01
-5.31809270e-01 5.61112761e-02 2.81232536e-01 2.71325499e-01
-1.86457857e-01 5.55610836e-01 1.15125918e+00 6.39462546e-02
-7.36176074e-02 2.76895672e-01 -1.46873093e+00 7.44811475e-01
7.51202464e-01 -6.98070168e-01 -3.29663128e-01 2.11982597e-02
3.89048070e-01 -3.50571990e-01 3.84103954e-01 -7.88788736e-01
1.46469310e-01 -6.09639823e-01 1.65771797e-01 -1.73536673e-01
-1.24395281e-01 -9.82282102e-01 1.71121836e-01 7.22948492e-01
-7.03854263e-01 -3.42161268e-01 -2.65422821e-01 3.66962254e-01
-9.11576897e-02 -4.33851272e-01 1.22438192e+00 -2.15880767e-01
-1.64532438e-01 1.77675616e-02 -1.72795370e-01 2.59382635e-01
1.11421633e+00 -2.92320400e-01 -4.02025253e-01 -2.92313874e-01
-1.75592080e-01 2.78491467e-01 7.66259611e-01 1.81681380e-01
2.59412795e-01 -1.14619696e+00 -1.10870802e+00 4.35631983e-02
1.57712042e-01 -2.55164783e-02 1.42070547e-01 9.52069223e-01
-3.77411842e-01 1.02358937e-01 -2.65922397e-01 -3.20810407e-01
-1.42959344e+00 4.33252960e-01 2.92448759e-01 -4.59019653e-02
-4.79631096e-01 6.71806276e-01 1.32138446e-01 -1.68662623e-01
2.29136527e-01 -7.05892742e-02 -1.92868277e-01 6.48440778e-01
8.19127858e-01 4.37611520e-01 9.28065181e-03 -6.03176773e-01
-3.35759223e-01 2.50626385e-01 -9.34754685e-02 1.24547824e-01
1.41911721e+00 -2.00527325e-01 -2.93829978e-01 5.66387892e-01
8.41945231e-01 1.03289753e-01 -9.21524644e-01 -1.31983489e-01
5.27487844e-02 -8.56885612e-01 3.24507244e-02 -6.09733522e-01
-7.77512312e-01 7.88340747e-01 1.18687920e-01 6.75425828e-01
1.23394775e+00 -2.69584447e-01 3.65294904e-01 1.69745702e-02
4.13477808e-01 -1.16963840e+00 -3.90513420e-01 1.54240793e-02
9.77493584e-01 -1.52505445e+00 4.78515327e-01 -6.97275177e-02
-8.41310740e-01 1.06111681e+00 1.36741742e-01 -2.91839153e-01
2.74033695e-01 1.39189260e-02 3.66529226e-02 -1.20960943e-01
-8.04598391e-01 -1.20973155e-01 2.64900297e-01 3.58515918e-01
5.28541803e-01 8.29948932e-02 -9.16860223e-01 3.41526747e-01
-3.85542512e-01 1.16979234e-01 7.04070508e-01 8.76044154e-01
-8.16569984e-01 -1.08671951e+00 -5.33079863e-01 8.68544996e-01
-6.63291097e-01 3.32085609e-01 -7.52972007e-01 6.80494726e-01
1.31446034e-01 1.01745284e+00 -2.31898665e-01 -1.16437875e-01
3.33596170e-01 -4.38029543e-02 4.00076061e-01 -4.11908299e-01
-7.30807006e-01 -1.16183877e-01 1.00068159e-01 -3.67644280e-01
-6.16779804e-01 -8.39581251e-01 -8.06401312e-01 -7.94145226e-01
-4.27313477e-01 3.22116226e-01 5.72743952e-01 1.04356468e+00
-4.74591367e-02 3.54183197e-01 8.11909258e-01 -6.54337466e-01
-9.38244879e-01 -5.63555300e-01 -7.29312718e-01 4.82733399e-01
4.81261313e-01 -3.73717308e-01 -7.75851667e-01 -2.26024494e-01] | [8.668256759643555, 5.202047824859619] |
c460c5da-3c24-4c7e-a0db-9b8c2f2393ea | variational-inference-posterior-threshold | 2301.04771 | null | https://arxiv.org/abs/2301.04771v1 | https://arxiv.org/pdf/2301.04771v1.pdf | Variational Inference: Posterior Threshold Improves Network Clustering Accuracy in Sparse Regimes | Variational inference has been widely used in machine learning literature to fit various Bayesian models. In network analysis, this method has been successfully applied to solve the community detection problems. Although these results are promising, their theoretical support is only for relatively dense networks, an assumption that may not hold for real networks. In addition, it has been shown recently that the variational loss surface has many saddle points, which may severely affect its performance, especially when applied to sparse networks. This paper proposes a simple way to improve the variational inference method by hard thresholding the posterior of the community assignment after each iteration. Using a random initialization that correlates with the true community assignment, we show that the proposed method converges and can accurately recover the true community labels, even when the average node degree of the network is bounded. Extensive numerical study further confirms the advantage of the proposed method over the classical variational inference and another state-of-the-art algorithm. | ['Can M. Le', 'Xuezhen Li'] | 2023-01-12 | null | null | null | null | ['community-detection'] | ['graphs'] | [ 1.73178986e-01 2.51304537e-01 -1.45113423e-01 -2.75055505e-02
-4.31102276e-01 -3.48977089e-01 3.37622136e-01 2.84522176e-01
-4.22577143e-01 8.85301590e-01 -4.91342604e-01 -8.70868564e-02
-3.43534201e-01 -9.10869777e-01 -5.61327875e-01 -1.14461541e+00
-1.35945708e-01 8.69143784e-01 2.75558978e-01 2.35913396e-01
2.47139022e-01 3.00347090e-01 -1.14022410e+00 -4.01818693e-01
9.87333715e-01 4.17745113e-01 1.56023055e-01 2.51388550e-01
-3.76483016e-02 2.55885690e-01 -4.06072408e-01 -5.63587248e-01
-1.87293459e-02 -2.48579443e-01 -5.21267295e-01 1.67434797e-01
1.08722128e-01 -8.09331238e-03 -2.96281278e-01 1.58330667e+00
1.26162991e-01 5.50476350e-02 8.87441933e-01 -1.33642006e+00
-1.44275231e-02 8.05081069e-01 -1.12042356e+00 9.92097631e-02
9.21273511e-03 -4.32637036e-01 1.17432785e+00 -6.59150422e-01
6.34075284e-01 1.24529338e+00 7.75443256e-01 1.37048766e-01
-1.51984632e+00 -8.06275249e-01 2.09736638e-02 1.34551302e-01
-1.95276642e+00 -1.17739484e-01 1.00650716e+00 -7.37265587e-01
1.60444662e-01 -1.68009147e-01 5.69446921e-01 7.69035459e-01
-3.59639712e-02 4.97075468e-01 1.08745539e+00 -3.83089006e-01
2.96739519e-01 1.83339849e-01 2.03115717e-01 6.75668240e-01
7.49428272e-01 -3.28550220e-01 -5.14611639e-02 -6.05249345e-01
7.89147973e-01 5.62086366e-02 -3.43910664e-01 -6.92602694e-01
-9.17238295e-01 1.22394192e+00 3.95433486e-01 3.81475925e-01
-4.82836664e-01 2.16915339e-01 9.65350792e-02 -1.62451088e-01
5.51171601e-01 -2.07932338e-01 2.16797575e-01 1.60390675e-01
-1.39744902e+00 2.64676362e-01 9.19787526e-01 6.77024901e-01
8.47287059e-01 -1.88407570e-01 1.64058417e-01 7.03704834e-01
6.62908018e-01 5.29793322e-01 -4.82161909e-01 -1.10355854e+00
2.45206222e-01 3.49474818e-01 2.36558825e-01 -1.33804750e+00
-1.39373973e-01 -7.22403944e-01 -1.38989425e+00 7.90121092e-04
8.50919306e-01 -1.68107212e-01 -3.11410964e-01 1.80487442e+00
4.07950491e-01 6.32304728e-01 -3.23319376e-01 8.24898720e-01
2.86084771e-01 5.17501652e-01 -1.64620593e-01 -6.44985497e-01
8.87859583e-01 -3.12534004e-01 -7.64046431e-01 1.86021462e-01
3.59253660e-02 -5.55700302e-01 4.56688821e-01 5.61929822e-01
-9.59550202e-01 -5.56713529e-02 -9.08355534e-01 6.74293876e-01
1.31399602e-01 -6.56490400e-02 6.32942259e-01 9.31572139e-01
-9.23661947e-01 5.45243204e-01 -1.00994110e+00 -5.60063541e-01
4.54860926e-01 2.39553258e-01 -7.33312219e-02 -3.05892050e-01
-1.03793299e+00 4.68784064e-01 3.64882857e-01 5.30516684e-01
-8.96490872e-01 -3.14046979e-01 -5.04902303e-01 2.10596219e-01
5.68406343e-01 -5.28823197e-01 7.24520206e-01 -5.87767839e-01
-1.32918572e+00 5.30742407e-01 -4.89454865e-01 -4.07159925e-01
7.46324360e-01 3.27897698e-01 4.22547199e-02 3.37344885e-01
7.23590255e-02 1.29994124e-01 8.35037589e-01 -1.28555441e+00
-1.80355176e-01 -3.47929925e-01 1.03388736e-02 -1.42341703e-01
-2.43379503e-01 -2.34568909e-01 -4.80585128e-01 -2.96520382e-01
5.22812366e-01 -9.44872856e-01 -4.27569538e-01 1.16095571e-02
-6.40510201e-01 -3.40203553e-01 6.40030086e-01 -3.26454163e-01
1.03451395e+00 -1.99734080e+00 3.35638523e-01 7.96253860e-01
4.47265446e-01 -1.83653176e-01 3.35271686e-01 4.88738805e-01
2.38636553e-01 2.55429566e-01 -6.23627305e-01 -3.37188631e-01
-1.16687350e-01 2.66692013e-01 5.90265729e-02 1.08614242e+00
-5.55307865e-02 1.97095454e-01 -9.04517770e-01 -7.29492843e-01
1.40051708e-01 6.64931238e-01 -6.97197080e-01 -2.30565339e-01
-3.21219712e-02 4.90458131e-01 -4.72791791e-01 2.16712296e-01
9.93663907e-01 -6.84285820e-01 8.36999416e-01 -6.20849207e-02
-2.93122195e-02 -3.34211200e-01 -1.67493200e+00 1.08171773e+00
9.53536555e-02 6.45406008e-01 6.44414723e-01 -1.53795004e+00
7.54949808e-01 4.38838035e-01 6.58878446e-01 3.00599515e-01
1.28897563e-01 1.14757143e-01 1.23662621e-01 -2.26757154e-01
1.49386242e-01 -4.05026913e-01 2.26945922e-01 3.10866445e-01
-1.98813900e-01 1.59347519e-01 4.79875356e-01 4.14934218e-01
7.71464229e-01 -3.38142604e-01 2.36579731e-01 -4.73041683e-01
6.41162992e-01 -2.28717744e-01 6.77550972e-01 9.54262912e-01
-6.57513589e-02 4.04790103e-01 9.20257092e-01 4.02884990e-01
-1.01160836e+00 -1.22510660e+00 -4.74444985e-01 4.56144989e-01
3.95580053e-01 -9.61852819e-02 -1.07563186e+00 -3.03536564e-01
1.17067462e-02 3.01537454e-01 -4.77384418e-01 2.56889105e-01
-2.79889405e-01 -1.13303125e+00 4.16796714e-01 3.89305314e-05
3.98672044e-01 -4.64933395e-01 -1.62302889e-02 3.27720463e-01
-5.28854609e-01 -1.19383729e+00 -9.27655473e-02 -2.60726303e-01
-1.09455407e+00 -1.38800490e+00 -8.19050074e-01 -5.61427176e-01
8.74249995e-01 2.98396170e-01 8.64612043e-01 3.57388556e-01
5.95235005e-02 2.62848526e-01 -1.14442743e-01 1.41183972e-01
-4.87214446e-01 1.57301202e-01 1.07771896e-01 3.39152902e-01
1.52089506e-01 -7.02180624e-01 -3.50027680e-01 3.26113671e-01
-7.40683377e-01 -1.43669888e-01 4.05649662e-01 8.11565816e-01
4.00640875e-01 6.29202366e-01 5.77036083e-01 -1.00212908e+00
5.83090305e-01 -7.14306414e-01 -9.43204999e-01 1.21919684e-01
-6.07804239e-01 -5.36800064e-02 3.24334443e-01 -1.45220459e-01
-1.01364136e+00 -8.20141062e-02 6.72207847e-02 -3.88127744e-01
-1.49498892e-03 9.54534590e-01 -5.51972724e-02 -1.18313968e-01
9.07267630e-02 6.44452916e-03 1.98564485e-01 -5.34947932e-01
1.76584378e-01 5.56154072e-01 3.45137477e-01 -6.03181005e-01
1.01381767e+00 7.63816774e-01 4.83681172e-01 -1.28229702e+00
-6.94430292e-01 -5.51609695e-01 -7.37100780e-01 -3.54346186e-01
6.61161482e-01 -7.64270246e-01 -1.11160243e+00 3.72681737e-01
-1.12957704e+00 6.53106198e-02 3.77181828e-01 5.31729043e-01
-2.79833317e-01 8.49702656e-01 -6.53349519e-01 -1.27871144e+00
1.02504566e-01 -1.00886941e+00 4.66475874e-01 1.68209597e-01
6.88823462e-02 -1.34024739e+00 9.60762054e-02 2.67958075e-01
7.58241862e-02 4.23351645e-01 9.07335579e-01 -3.21585000e-01
-6.57432020e-01 -1.40612528e-01 -4.09275115e-01 1.67537257e-01
-1.44233614e-01 4.86573905e-01 -5.92391014e-01 -5.25502384e-01
-1.79120451e-01 1.71700284e-01 8.68536890e-01 8.89340043e-01
7.97351241e-01 2.61501465e-02 -6.53429985e-01 2.30868325e-01
1.64236927e+00 -2.68411875e-01 5.01503348e-01 -2.11453006e-01
5.27261376e-01 4.78928596e-01 4.47461516e-01 6.26374125e-01
2.81523794e-01 6.70882523e-01 5.68011880e-01 9.46511775e-02
2.79981852e-01 -5.51972352e-02 1.16093621e-01 8.90048325e-01
-2.14193583e-01 -2.14358494e-01 -8.56604755e-01 5.64336419e-01
-2.09706855e+00 -1.17706478e+00 -5.98211706e-01 2.37769389e+00
7.65673578e-01 2.91346282e-01 1.82738885e-01 2.58250594e-01
1.24471021e+00 1.20932916e-02 -3.60314369e-01 1.35033846e-01
-2.02383384e-01 2.08225250e-02 4.42715824e-01 8.13390136e-01
-1.00624287e+00 6.14808142e-01 6.84732389e+00 9.13298428e-01
-6.94265425e-01 3.47468942e-01 2.39987597e-01 3.38209093e-01
-2.24469692e-01 2.50839859e-01 -6.34106040e-01 4.55814540e-01
5.75368464e-01 -2.36768112e-01 3.01855415e-01 5.53193450e-01
5.67327738e-01 -3.22321445e-01 -7.42094219e-01 7.48563290e-01
-1.66776240e-01 -1.21390522e+00 -3.86110514e-01 4.73456472e-01
8.71348619e-01 -1.78310931e-01 -1.38966858e-01 -8.43042806e-02
2.24691197e-01 -8.57310355e-01 3.00114274e-01 5.99622667e-01
5.33857167e-01 -8.48757029e-01 9.48445499e-01 5.54860651e-01
-1.20811772e+00 1.57341257e-01 -5.06892920e-01 -2.12505892e-01
3.35954815e-01 1.24051976e+00 -9.38926220e-01 4.49967831e-01
4.87693816e-01 8.38940978e-01 -2.95548230e-01 1.40462303e+00
-2.56411344e-01 9.37805235e-01 -7.63137817e-01 -8.98745358e-02
1.09895177e-01 -6.91886961e-01 9.26402986e-01 9.67809260e-01
3.05117786e-01 -1.84428796e-01 2.38200828e-01 1.22229564e+00
-5.86849302e-02 -1.72809325e-02 -4.60435241e-01 -1.32692590e-01
6.58433974e-01 1.18669188e+00 -1.11587703e+00 -2.97334075e-01
-2.83983499e-01 5.96694887e-01 2.50445336e-01 6.07344329e-01
-7.10949123e-01 -1.55833393e-01 4.20291305e-01 3.44815850e-01
4.61052775e-01 -3.40786129e-01 -1.37011021e-01 -1.14218342e+00
-1.51038706e-01 -4.04439449e-01 2.68660814e-01 -3.46988320e-01
-1.41225672e+00 7.42590129e-02 2.83343434e-01 -8.31591129e-01
-3.22694480e-01 -3.78887624e-01 -5.79298675e-01 7.46753931e-01
-1.11647904e+00 -7.84003675e-01 -1.90648958e-01 4.30667847e-01
-3.14981155e-02 6.39165491e-02 5.27244031e-01 5.23443997e-01
-9.62681770e-01 3.84735554e-01 7.10104287e-01 1.68150470e-01
2.82851696e-01 -1.21651840e+00 -3.42009008e-01 1.06234276e+00
3.24453004e-02 7.50767529e-01 1.27451432e+00 -6.90581441e-01
-1.09888804e+00 -6.95580959e-01 6.19994521e-01 2.91201826e-02
8.21694195e-01 -3.96189272e-01 -9.99345779e-01 4.18887824e-01
1.52594700e-01 -1.05984226e-01 4.88085002e-01 2.96999872e-01
-3.48174497e-02 1.86836906e-02 -1.27823985e+00 4.83755589e-01
7.23585486e-01 -2.54147023e-01 -1.26248524e-01 4.20782387e-01
1.02938533e-01 9.02340785e-02 -9.69859600e-01 2.51695514e-01
5.02443373e-01 -1.01329005e+00 9.76151109e-01 -1.31958589e-01
9.83115509e-02 -5.02747655e-01 -8.78731310e-02 -1.06326628e+00
-2.84201711e-01 -6.38307154e-01 -2.32627332e-01 1.33636701e+00
1.20724931e-01 -7.35481083e-01 1.07961249e+00 6.96837902e-02
5.22773266e-01 -3.82247478e-01 -1.20823264e+00 -6.54721379e-01
1.46138996e-01 -2.02780813e-01 4.82915007e-02 9.93629098e-01
-7.18463734e-02 2.50304192e-01 -3.20840716e-01 3.91154408e-01
1.52832043e+00 4.18265611e-02 5.73921323e-01 -1.92246222e+00
-2.78461546e-01 -6.13742292e-01 -3.33247215e-01 -9.91271794e-01
4.96915787e-01 -8.43701839e-01 5.31649552e-02 -1.43931901e+00
7.39132583e-01 -6.49951220e-01 -4.74463403e-02 -5.45196421e-02
-1.81830376e-01 3.20767134e-01 -4.40051407e-02 3.30698669e-01
-4.21505064e-01 4.26921129e-01 1.14678526e+00 -1.09058827e-01
1.24752119e-01 3.72366160e-01 -4.70310777e-01 8.61879408e-01
6.63107932e-01 -7.43432701e-01 -2.74298131e-01 2.14852765e-01
4.17613477e-01 1.21294901e-01 5.66441655e-01 -7.30606675e-01
3.44060689e-01 -7.50482902e-02 -1.31110057e-01 -8.18524420e-01
3.70501220e-01 -8.92962217e-01 4.23614264e-01 5.01594603e-01
-7.41170123e-02 -3.99573237e-01 -3.02102357e-01 1.09187603e+00
-1.70782775e-01 -7.12503791e-01 1.02071083e+00 1.00893915e-01
-6.08632676e-02 2.38918051e-01 -5.22914767e-01 5.43696433e-02
7.98527062e-01 -1.05527170e-01 -7.61637278e-03 -6.54347420e-01
-8.94067228e-01 2.09115490e-01 5.24475455e-01 -3.51639420e-01
4.64982986e-01 -1.13050377e+00 -9.45016742e-01 -1.40567988e-01
-2.41611093e-01 -1.49866253e-01 2.41772383e-01 1.35506570e+00
-4.76817697e-01 1.66112885e-01 2.99877971e-01 -9.78530169e-01
-1.15906954e+00 4.46948320e-01 2.28353471e-01 -2.52133727e-01
-5.23443460e-01 4.47284341e-01 -5.94583265e-02 -1.07171200e-01
3.27721477e-01 4.50687855e-02 -2.46162996e-01 7.48885870e-02
1.73403516e-01 6.50709331e-01 -4.07652646e-01 -7.10011065e-01
-4.47047919e-01 6.80180132e-01 1.27755702e-01 -2.65048623e-01
1.24574852e+00 -3.49733084e-01 -6.13468587e-01 5.55665791e-01
1.00075686e+00 4.03098613e-02 -1.05327976e+00 -3.94265383e-01
-2.94261891e-03 -3.57708901e-01 1.09583013e-01 5.50057739e-02
-1.22216439e+00 1.04433608e+00 3.53461355e-01 6.91934049e-01
5.71651697e-01 -1.93385761e-02 1.80157930e-01 2.59697974e-01
4.64010030e-01 -9.42903936e-01 -2.54792303e-01 2.94574589e-01
4.30904835e-01 -1.19484520e+00 2.60404021e-01 -9.01200354e-01
-2.29207836e-02 9.82100844e-01 8.23728219e-02 -3.44532460e-01
1.04027712e+00 6.77656978e-02 -5.37160039e-01 -1.43207610e-01
-3.42187345e-01 -3.02738231e-03 -6.97428659e-02 5.45599639e-01
4.39728677e-01 2.30155528e-01 -4.69501406e-01 2.94260085e-01
1.75543979e-01 -1.39520600e-01 8.28064382e-01 3.61562937e-01
-4.67966080e-01 -9.96452272e-01 -5.62188864e-01 3.61769348e-01
-5.70691168e-01 -2.77608056e-02 -1.69681907e-01 8.39311898e-01
-1.76065683e-01 1.16404772e+00 3.21694970e-04 2.08866969e-01
-1.68151736e-01 -1.32616192e-01 6.45649791e-01 -4.54862118e-01
-3.16055082e-02 5.03027320e-01 -2.17831463e-01 -1.31763503e-01
-8.65322530e-01 -1.04835713e+00 -1.03949809e+00 -6.65013015e-01
-6.56307340e-01 5.27186811e-01 3.54550838e-01 1.02043116e+00
-1.13254860e-01 3.98130536e-01 4.96507168e-01 -7.65933335e-01
-5.02948463e-01 -1.03235435e+00 -9.39741254e-01 8.69723335e-02
2.50183523e-01 -1.11330271e+00 -7.36483872e-01 -1.34614468e-01] | [6.947624206542969, 5.183107852935791] |
02c8445f-7630-4300-8dca-397d63a10a66 | optimal-quadratic-binding-for-relational | 2204.07186 | null | https://arxiv.org/abs/2204.07186v1 | https://arxiv.org/pdf/2204.07186v1.pdf | Optimal quadratic binding for relational reasoning in vector symbolic neural architectures | Binding operation is fundamental to many cognitive processes, such as cognitive map formation, relational reasoning, and language comprehension. In these processes, two different modalities, such as location and objects, events and their contextual cues, and words and their roles, need to be bound together, but little is known about the underlying neural mechanisms. Previous works introduced a binding model based on quadratic functions of bound pairs, followed by vector summation of multiple pairs. Based on this framework, we address following questions: Which classes of quadratic matrices are optimal for decoding relational structures? And what is the resultant accuracy? We introduce a new class of binding matrices based on a matrix representation of octonion algebra, an eight-dimensional extension of complex numbers. We show that these matrices enable a more accurate unbinding than previously known methods when a small number of pairs are present. Moreover, numerical optimization of a binding operator converges to this octonion binding. We also show that when there are a large number of bound pairs, however, a random quadratic binding performs as well as the octonion and previously-proposed binding methods. This study thus provides new insight into potential neural mechanisms of binding operations in the brain. | ['Haim Sompolinsky', 'Naoki Hiratani'] | 2022-04-14 | null | null | null | null | ['relational-reasoning'] | ['natural-language-processing'] | [ 1.81934386e-01 -1.26863644e-01 2.38373484e-02 -2.60176033e-01
-2.83643454e-01 -5.92881143e-01 6.27479494e-01 4.83007967e-01
-7.71019161e-01 7.42992520e-01 1.71108872e-01 -2.91208386e-01
-3.67813438e-01 -1.02036858e+00 -7.97146082e-01 -6.48869753e-01
-3.72047722e-01 7.16361463e-01 5.99624336e-01 -6.90985382e-01
3.85063201e-01 5.44866204e-01 -1.54704797e+00 3.38861078e-01
7.87236571e-01 8.76052797e-01 4.37877089e-01 4.82843846e-01
5.63179925e-02 5.75844407e-01 -3.48953903e-01 -4.72598106e-01
1.61569968e-01 -4.07310069e-01 -7.93115675e-01 -7.60075092e-01
1.78810939e-01 -1.85784057e-01 -3.02189440e-01 1.23499107e+00
3.90681386e-01 4.22437966e-01 8.99805367e-01 -6.18181288e-01
-8.01215470e-01 7.41220593e-01 -1.11218065e-01 6.03086531e-01
4.84707117e-01 -4.02408183e-01 1.41285491e+00 -1.15768337e+00
4.71093476e-01 1.15079582e+00 2.88776994e-01 3.18273306e-01
-1.41208589e+00 -6.46929622e-01 -3.57137173e-02 6.87140584e-01
-1.57608581e+00 -4.03298855e-01 6.51964903e-01 -6.03942513e-01
9.66286063e-01 5.54697752e-01 9.07146394e-01 6.81751311e-01
5.84757388e-01 2.41594836e-01 1.16774189e+00 -6.86117589e-01
1.75156698e-01 -2.15634063e-01 5.25719106e-01 6.73304617e-01
5.32582402e-01 2.25146208e-02 -1.04048908e+00 -7.10531324e-02
8.19031775e-01 -2.42912009e-01 -2.54112691e-01 -2.57577181e-01
-1.35032094e+00 6.62443221e-01 5.11586487e-01 5.10411382e-01
-3.64452243e-01 1.24784432e-01 -9.25864279e-03 7.93803558e-02
2.45479923e-02 4.85845774e-01 -2.68966705e-02 8.83751735e-02
-4.84019905e-01 3.21580023e-01 8.58333051e-01 6.01025283e-01
8.91733527e-01 -3.71424228e-01 4.09483025e-03 8.03803563e-01
3.86596113e-01 6.44313812e-01 -1.13931410e-01 -5.93494534e-01
5.38122237e-01 3.79082412e-01 1.66642636e-01 -1.41974235e+00
-7.47043967e-01 -3.10709417e-01 -9.00222898e-01 -2.11450309e-01
6.90617621e-01 3.61127764e-01 -3.70156378e-01 2.07472301e+00
1.01702549e-01 -2.39711598e-01 -1.14651576e-01 1.12673450e+00
7.29714036e-01 5.65092623e-01 -1.05489321e-01 -3.29935759e-01
1.56540501e+00 -1.50468275e-01 -8.69734943e-01 -3.01365137e-01
4.47750300e-01 -5.88606060e-01 6.81829095e-01 3.41274053e-01
-1.63397014e+00 -2.62856275e-01 -1.19939613e+00 -3.62542838e-01
-4.67867047e-01 -1.95299849e-01 9.80690837e-01 6.15381002e-01
-1.04573846e+00 4.38645214e-01 -6.27701938e-01 -1.27950937e-01
-4.70287129e-02 7.80982256e-01 -4.62129980e-01 1.15694657e-01
-1.68612611e+00 1.42971706e+00 2.76479065e-01 6.13492310e-01
-4.87583846e-01 -2.45550290e-01 -5.35715818e-01 1.35184661e-01
2.38960177e-01 -5.22693634e-01 9.16533709e-01 -3.92355591e-01
-1.17015803e+00 8.51879895e-01 -1.38881341e-01 -1.38202235e-01
-2.08844379e-01 4.11415212e-02 -4.92110699e-02 1.69194490e-01
-2.59294957e-02 3.60803992e-01 3.27078491e-01 -9.44581985e-01
1.13779992e-01 -5.22232413e-01 5.27358234e-01 4.23581839e-01
-5.85704744e-02 3.71413119e-02 -1.69969052e-01 -2.67139196e-01
9.31424677e-01 -8.83349836e-01 -7.29055256e-02 -2.03261524e-01
-2.77611632e-02 -3.46077635e-04 -4.47934657e-01 -6.46142781e-01
1.22397768e+00 -2.19375229e+00 9.12296176e-01 5.84989965e-01
5.87492108e-01 -2.07655713e-01 -2.94727124e-02 7.18812048e-01
-2.09555611e-01 -2.33171601e-02 -8.27131942e-02 3.89116824e-01
2.96123683e-01 3.19243044e-01 -2.00698107e-01 6.46945953e-01
-1.73562244e-01 9.37896848e-01 -7.67790437e-01 -2.74181515e-01
-1.28438458e-01 1.57482415e-01 -7.78078616e-01 -1.14416018e-01
-5.24051152e-02 1.62861422e-01 -3.83049816e-01 4.59506959e-01
7.27598310e-01 -2.04594702e-01 5.12402892e-01 -5.06403983e-01
-1.39415458e-01 5.15128374e-01 -1.14208627e+00 1.29765558e+00
-1.05927147e-01 5.06976843e-01 2.18108252e-01 -1.02496612e+00
7.00539351e-01 1.45568267e-01 2.56222486e-01 -9.53908563e-01
2.46501744e-01 2.37174273e-01 7.07645357e-01 -1.09555811e-01
2.89376557e-01 -4.86506462e-01 -2.72585750e-01 4.15835619e-01
-3.23881023e-02 -4.19399589e-02 3.24096054e-01 5.22470891e-01
1.27902436e+00 -4.13813412e-01 4.97268736e-01 -5.30071557e-01
3.86343747e-01 -5.44362664e-01 3.77713025e-01 7.92358935e-01
-3.78760323e-02 2.39854321e-01 9.09016371e-01 -2.66518027e-01
-7.42680609e-01 -1.25331175e+00 -3.59498352e-01 1.25838339e+00
4.55023855e-01 -6.85107708e-01 -8.57236326e-01 2.40072295e-01
-2.42730960e-01 4.43057925e-01 -8.16502750e-01 -3.38642865e-01
-3.96687061e-01 -1.12573397e+00 4.33428615e-01 4.43241894e-01
1.92166194e-01 -9.07298207e-01 -5.23736060e-01 8.39626119e-02
-3.87326330e-01 -9.73163128e-01 -2.40027487e-01 5.71510613e-01
-6.03641272e-01 -9.97717500e-01 -1.48918897e-01 -5.75931847e-01
5.91567457e-01 1.80032566e-01 9.20184016e-01 2.24444777e-01
-1.47527784e-01 2.82415539e-01 -1.17763661e-01 -1.88039884e-01
-5.01273125e-02 -7.44474605e-02 7.78250173e-02 1.09637920e-02
3.14165354e-01 -7.52290130e-01 -4.69694138e-01 3.25753242e-01
-8.40757310e-01 2.28784621e-01 7.39002109e-01 9.97097790e-01
3.40438843e-01 -3.76533866e-01 2.33827293e-01 -7.18318522e-01
7.47486293e-01 -4.45716053e-01 -6.66575551e-01 2.29630709e-01
-1.33744538e-01 3.46482396e-01 1.97287872e-01 -3.84472847e-01
-8.43021929e-01 -9.40675437e-02 -1.05893061e-01 3.76700282e-01
5.15757143e-01 8.38659763e-01 -2.91157871e-01 -2.65214473e-01
9.02604520e-01 1.99987426e-01 -1.79182351e-01 -1.38806462e-01
4.27536726e-01 2.95440942e-01 1.61725238e-01 -1.09503043e+00
3.36297661e-01 2.86147892e-01 4.54321474e-01 -8.81701350e-01
-7.35946596e-01 -3.86355251e-01 -1.06025505e+00 -2.93451011e-01
1.02638781e+00 -4.21771318e-01 -1.15322626e+00 1.95653826e-01
-1.61270845e+00 -8.47195014e-02 2.44705871e-01 6.75443769e-01
-7.42181599e-01 2.77312845e-01 -7.63568580e-01 -8.36561263e-01
1.57679781e-01 -1.17421031e+00 8.06453526e-01 -2.25318924e-01
-2.54325867e-01 -8.05124283e-01 8.81814510e-02 2.64498293e-01
9.92702469e-02 -2.71465898e-01 1.32238388e+00 -5.25351524e-01
-9.89275455e-01 -1.44545659e-02 -3.01603049e-01 -1.77014798e-01
-4.30456340e-01 -4.96686399e-01 -5.30386567e-01 1.42140046e-01
2.03760967e-01 -2.66967230e-02 9.22051430e-01 1.01600327e-01
6.93621695e-01 -2.03740209e-01 -3.55572224e-01 3.41728747e-01
1.11529934e+00 4.60896254e-01 7.43700862e-01 7.34053366e-03
5.41047037e-01 7.72363484e-01 1.67785704e-01 2.34072044e-01
4.18128043e-01 8.01103294e-01 5.89083493e-01 3.99451077e-01
3.28452647e-01 1.56853810e-01 1.17607623e-01 1.26609623e+00
-5.94846010e-01 -2.00465843e-02 -1.12907493e+00 2.13539481e-01
-1.90429378e+00 -1.29902613e+00 -3.68102640e-01 2.34999132e+00
1.04512322e+00 2.53383696e-01 -2.29500830e-01 1.40804425e-01
4.99121934e-01 2.67452206e-02 -2.99350798e-01 -1.30638152e-01
-3.91715199e-01 5.82192719e-01 4.85664099e-01 8.26604187e-01
-5.60841739e-01 8.89943302e-01 7.46796322e+00 4.87642050e-01
-7.63725638e-01 4.21079278e-01 -8.87276530e-02 -6.29819483e-02
-4.38425630e-01 7.83417299e-02 -7.05287814e-01 2.35964552e-01
8.71556580e-01 1.41721576e-01 9.62164879e-01 9.74330455e-02
-1.89676434e-01 -4.07115042e-01 -1.23911130e+00 1.15691936e+00
2.17331290e-01 -1.32680166e+00 -3.83359939e-02 1.95417076e-01
1.33585259e-01 -2.19824001e-01 -1.16056673e-01 -5.61948530e-02
6.42744871e-03 -1.11893821e+00 9.20990169e-01 1.00266814e+00
3.23048949e-01 -4.03272241e-01 3.85991871e-01 4.45545644e-01
-9.99426484e-01 4.88424189e-02 -6.02763236e-01 -5.68294346e-01
4.86245304e-02 6.13725424e-01 -3.15108031e-01 2.82107830e-01
2.43181750e-01 3.46091598e-01 -3.64010513e-01 7.66523004e-01
-1.96700439e-01 3.71100128e-01 -4.71196890e-01 -5.00370622e-01
-1.31664306e-01 -5.48401833e-01 3.62737715e-01 8.64921451e-01
5.39595038e-02 8.66422653e-01 -2.52574444e-01 1.04569590e+00
1.36385977e-01 2.39029780e-01 -4.45539743e-01 2.52391752e-02
5.14987946e-01 9.80301857e-01 -1.00699031e+00 -6.22329377e-02
-3.79676014e-01 6.97233737e-01 7.66947985e-01 4.21903014e-01
-6.98383093e-01 -1.05664462e-01 4.46452945e-01 1.96091279e-01
1.15631327e-01 -7.68791378e-01 -2.73978859e-01 -1.11776710e+00
2.28279494e-02 -5.21754801e-01 5.09022139e-02 -8.76759410e-01
-1.03795481e+00 3.57237786e-01 2.89209425e-01 -4.74782497e-01
6.47070482e-02 -9.71281409e-01 5.75405508e-02 9.40802574e-01
-6.39666319e-01 -7.47946024e-01 1.00592665e-01 5.75941145e-01
-3.63451153e-01 1.55076221e-01 9.59350884e-01 4.09871370e-01
-4.30172682e-01 1.73997700e-01 1.44538237e-02 3.38014439e-02
4.05313432e-01 -1.19550955e+00 -9.94877070e-02 6.12170398e-01
3.66703093e-01 1.18366408e+00 6.46588206e-01 -6.29772007e-01
-1.72356308e+00 -1.94178879e-01 9.59189773e-01 -6.20149195e-01
6.71275258e-01 -1.01623380e+00 -9.63619292e-01 7.38304496e-01
-2.45801862e-02 -1.16531208e-01 9.16305900e-01 4.91988301e-01
-5.04308224e-01 -1.09167978e-01 -6.81434095e-01 6.72146201e-01
1.34756374e+00 -8.47231984e-01 -7.92433977e-01 5.92865586e-01
3.45622182e-01 -5.95329821e-01 -8.38463128e-01 1.07004464e-01
8.00513029e-01 -1.20309842e+00 1.10070205e+00 -4.52628136e-01
7.71297961e-02 -2.84493446e-01 -7.56874800e-01 -1.13571990e+00
-7.15287030e-01 -3.13328564e-01 1.11351527e-01 3.67087513e-01
3.60199749e-01 -8.35164070e-01 3.10610801e-01 6.74360812e-01
-8.30131769e-02 -7.14674830e-01 -1.19350564e+00 -4.90622818e-01
8.06892365e-02 -4.88798320e-01 2.82396734e-01 7.43414879e-01
4.93560821e-01 9.22697306e-01 -1.30310208e-01 1.77079499e-01
5.06610870e-01 -1.64229915e-01 2.50600874e-01 -1.55503547e+00
-2.92623878e-01 -5.15806019e-01 -6.65606141e-01 -1.17179012e+00
3.07269961e-01 -1.30343878e+00 -1.15155324e-01 -1.54563546e+00
4.24670756e-01 -5.33328652e-01 -3.15038621e-01 4.66717720e-01
1.78038359e-01 1.61195338e-01 1.94149464e-01 2.25613028e-01
-5.20951033e-01 4.03122216e-01 1.24551618e+00 7.95847476e-02
1.17469355e-01 -3.32600892e-01 -7.44506776e-01 6.15552545e-01
3.88006747e-01 -3.91177088e-01 -2.66186655e-01 -1.60791516e-01
1.08724308e+00 3.90920013e-01 4.16506976e-01 -6.99680626e-01
5.98689318e-01 -3.50311726e-01 3.32637221e-01 -6.31898701e-01
7.79365003e-01 -5.82157671e-01 3.33789378e-01 3.20283741e-01
-1.94796458e-01 1.04885668e-01 1.11591004e-01 3.50144297e-01
-9.76714399e-03 -4.86113340e-01 5.62822878e-01 -6.03921860e-02
-4.40222770e-01 -1.42374858e-01 -5.99171817e-01 -4.18303199e-02
8.27804208e-01 -1.09089635e-01 -2.35373780e-01 -2.52636433e-01
-9.84497726e-01 -3.81274134e-01 -3.57852504e-02 -9.10316855e-02
5.74415267e-01 -1.40248311e+00 -3.26577544e-01 2.66512901e-01
-1.92899913e-01 -2.52794832e-01 2.13480726e-01 1.07708085e+00
-4.67834204e-01 7.26593792e-01 -4.11181867e-01 -4.49419469e-01
-8.70016932e-01 4.73089755e-01 4.14966166e-01 8.02766979e-02
-4.68647219e-02 8.50393951e-01 6.88018739e-01 -2.28437468e-01
-1.11493796e-01 -5.36271393e-01 -2.34511793e-01 4.38905597e-01
4.82098311e-01 2.96644866e-01 2.58410513e-01 -8.98931265e-01
-6.44008696e-01 6.68500423e-01 8.86076093e-02 -3.91600609e-01
1.01511467e+00 8.93059596e-02 -9.04157400e-01 7.48286009e-01
6.17037833e-01 2.81030893e-01 -5.85764170e-01 -8.43180418e-02
-2.69819260e-01 -2.27481842e-01 -1.65277600e-01 -6.68430984e-01
-5.04877687e-01 9.40742850e-01 8.91089439e-02 3.47672075e-01
9.20422077e-01 2.32217684e-01 7.03268945e-02 6.77667022e-01
6.60383761e-01 -6.75887823e-01 1.29922256e-01 7.74136364e-01
1.10481250e+00 -6.86066508e-01 -1.02071702e-01 -7.72844136e-01
-1.09431930e-01 9.70212519e-01 5.18639326e-01 -9.27600861e-02
9.07535076e-01 1.65108889e-01 -5.71155429e-01 -2.89759099e-01
-7.60494590e-01 -2.84559995e-01 5.45749903e-01 4.15904731e-01
5.56955159e-01 2.66756058e-01 -7.24213660e-01 8.73726010e-01
-4.16091800e-01 -5.23165762e-01 5.03989637e-01 7.37365723e-01
-6.12166226e-01 -1.10914636e+00 -6.41952634e-01 5.10944843e-01
-1.92559332e-01 -5.19816637e-01 -4.32785153e-01 5.92999041e-01
7.59601668e-02 6.31360650e-01 4.60816808e-02 -2.21138135e-01
3.80707055e-01 -3.14953625e-02 1.25408292e+00 -8.59610498e-01
-2.65130281e-01 -2.00882316e-01 3.23138665e-03 -4.51755881e-01
-3.98128957e-01 -8.02182317e-01 -1.46940434e+00 -3.46780092e-01
-4.93205398e-01 2.69845992e-01 6.66510701e-01 1.16518354e+00
1.17173858e-01 4.20180053e-01 -2.45019034e-01 -8.29223156e-01
-4.44226891e-01 -8.04851055e-01 -7.57911325e-01 8.05784911e-02
1.05555616e-01 -1.27891517e+00 -2.25469917e-01 -3.01140875e-01] | [7.422172546386719, 5.029465675354004] |
1f48f27c-f954-44ee-934f-6c0520a90974 | rlas-biabc-a-reinforcement-learning-based | 2301.02807 | null | https://arxiv.org/abs/2301.02807v1 | https://arxiv.org/pdf/2301.02807v1.pdf | RLAS-BIABC: A Reinforcement Learning-Based Answer Selection Using the BERT Model Boosted by an Improved ABC Algorithm | Answer selection (AS) is a critical subtask of the open-domain question answering (QA) problem. The present paper proposes a method called RLAS-BIABC for AS, which is established on attention mechanism-based long short-term memory (LSTM) and the bidirectional encoder representations from transformers (BERT) word embedding, enriched by an improved artificial bee colony (ABC) algorithm for pretraining and a reinforcement learning-based algorithm for training backpropagation (BP) algorithm. BERT can be comprised in downstream work and fine-tuned as a united task-specific architecture, and the pretrained BERT model can grab different linguistic effects. Existing algorithms typically train the AS model with positive-negative pairs for a two-class classifier. A positive pair contains a question and a genuine answer, while a negative one includes a question and a fake answer. The output should be one for positive and zero for negative pairs. Typically, negative pairs are more than positive, leading to an imbalanced classification that drastically reduces system performance. To deal with it, we define classification as a sequential decision-making process in which the agent takes a sample at each step and classifies it. For each classification operation, the agent receives a reward, in which the prize of the majority class is less than the reward of the minority class. Ultimately, the agent finds the optimal value for the policy weights. We initialize the policy weights with the improved ABC algorithm. The initial value technique can prevent problems such as getting stuck in the local optimum. Although ABC serves well in most tasks, there is still a weakness in the ABC algorithm that disregards the fitness of related pairs of individuals in discovering a neighboring food source position. | ['Saeed Shiry Ghidary', 'Azam Bastanfard', 'Javad Mohammadzadeh', 'Hamid Gharagozlou'] | 2023-01-07 | null | null | null | null | ['imbalanced-classification', 'open-domain-question-answering', 'answer-selection'] | ['miscellaneous', 'natural-language-processing', 'natural-language-processing'] | [ 3.17970484e-01 2.69481480e-01 -2.56376982e-01 -3.94705892e-01
-4.83623445e-01 -1.37017861e-01 -1.08842999e-02 2.84404546e-01
-5.66039979e-01 8.38155627e-01 -1.89289004e-01 -4.81519818e-01
-2.55714595e-01 -1.18396544e+00 -6.26105607e-01 -9.49794412e-01
2.24012583e-01 7.17364013e-01 5.47870696e-01 -4.96134877e-01
3.30857128e-01 1.04655392e-01 -1.64339209e+00 6.25547111e-01
1.08414245e+00 1.35131192e+00 2.76948124e-01 6.07542455e-01
-6.17147923e-01 1.22163880e+00 -9.92821038e-01 -6.97987676e-01
7.72357136e-02 -8.79188061e-01 -1.13208902e+00 -4.42048103e-01
-2.03891948e-01 -5.60034364e-02 1.36898816e-01 1.02744710e+00
4.64732379e-01 7.77853355e-02 4.12785321e-01 -1.43032217e+00
-9.91999507e-01 5.23402452e-01 -3.00988406e-01 2.91702718e-01
2.05563813e-01 2.37335473e-01 1.25896704e+00 -6.63053453e-01
1.97157905e-01 1.20452738e+00 3.99968475e-01 6.89759910e-01
-8.41473162e-01 -6.01653516e-01 3.55098218e-01 6.56706274e-01
-9.12613690e-01 -5.91593571e-02 7.28931546e-01 -1.09242298e-01
9.36342299e-01 3.72077554e-01 8.23774040e-01 7.52507508e-01
2.67234683e-01 9.20163691e-01 6.31856263e-01 -6.47067308e-01
5.32835484e-01 2.75421828e-01 5.53466499e-01 6.58218205e-01
1.94793254e-01 2.19997957e-01 -2.92885840e-01 -1.94262251e-01
1.96217448e-01 -8.75140578e-02 -2.37620801e-01 -1.46037951e-01
-8.24168622e-01 1.21773422e+00 1.06413710e+00 4.55316722e-01
-6.01777673e-01 -2.66198465e-03 5.54955788e-02 8.45977962e-01
1.09544732e-01 7.70577431e-01 -5.66976309e-01 1.31275445e-01
-3.94823313e-01 2.29217306e-01 8.33730340e-01 2.89767206e-01
9.42137063e-01 -1.56879008e-01 -4.53933358e-01 9.84744251e-01
2.06610024e-01 1.62222818e-01 9.69005823e-01 -7.09571302e-01
4.77770388e-01 1.14612198e+00 4.94555160e-02 -1.10601127e+00
-1.83590531e-01 -4.24091280e-01 -6.67014182e-01 1.78667441e-01
5.28626382e-01 -1.99333638e-01 -7.07176447e-01 1.64920151e+00
3.73295695e-01 -1.39626116e-01 3.04720551e-01 9.31806445e-01
9.23471212e-01 7.81310618e-01 -3.55188176e-02 -1.34963885e-01
1.28245485e+00 -1.50520730e+00 -7.65768290e-01 -5.81065834e-01
5.69653273e-01 -3.54136765e-01 1.14912176e+00 1.26207739e-01
-8.43048513e-01 -4.99393582e-01 -1.11426187e+00 1.04296118e-01
-7.05906451e-01 -2.02456079e-02 3.20906401e-01 4.46708798e-01
-7.16934085e-01 5.83506465e-01 -1.80612937e-01 -4.90899272e-02
3.03509772e-01 5.05520999e-01 1.27114160e-02 -6.01235479e-02
-1.70094669e+00 1.06736851e+00 4.29370821e-01 2.98293948e-01
-2.67627835e-01 -1.23684689e-01 -7.61328638e-01 3.08429241e-01
3.48601043e-01 -6.50377572e-01 1.13709211e+00 -1.64860880e+00
-1.70542455e+00 6.93303108e-01 -8.59080851e-02 -5.53860784e-01
1.41647920e-01 2.69658595e-01 -4.63800788e-01 -5.48899025e-02
-6.68971147e-03 8.39620948e-01 1.19330847e+00 -9.54635441e-01
-7.91409254e-01 -4.83292282e-01 3.12206239e-01 3.49892467e-01
-4.63238001e-01 -5.87890996e-03 9.69247520e-03 -4.12747204e-01
1.23099387e-01 -4.84323055e-01 -7.67901093e-02 -7.30074048e-02
-5.66666014e-02 -5.20395398e-01 5.94452918e-01 -5.90140581e-01
1.43265641e+00 -1.98192501e+00 1.15382634e-01 9.78553966e-02
-5.75568117e-02 3.52302581e-01 -5.01074255e-01 1.02437800e-02
-2.01557785e-01 -5.87905571e-03 -3.50204438e-01 2.01574340e-01
-7.16008395e-02 3.87190670e-01 -3.10080945e-01 -4.54461798e-02
6.62197709e-01 9.38201547e-01 -9.64456320e-01 -2.21978620e-01
-2.55634516e-01 -7.91235268e-02 -6.38843000e-01 5.48918188e-01
-4.10812259e-01 -1.73844118e-02 -4.55544710e-01 5.36334932e-01
5.73334515e-01 -3.46694976e-01 -6.74975738e-02 1.49351239e-01
8.43174681e-02 3.03975344e-01 -1.13132429e+00 9.28445756e-01
-2.47531652e-01 1.39194533e-01 8.09050873e-02 -1.23539507e+00
1.20450640e+00 4.18926738e-02 -5.48357591e-02 -9.49622631e-01
2.17756301e-01 4.17892635e-01 5.09365618e-01 -8.11338902e-01
2.86918074e-01 7.71680055e-03 1.80639014e-01 5.05196333e-01
3.61440964e-02 -6.73998445e-02 8.18649232e-02 -2.13963509e-01
1.15962923e+00 -7.47707160e-03 2.02888265e-01 5.13257720e-02
9.58003640e-01 3.21272202e-02 6.87532067e-01 6.53418005e-01
-4.10682797e-01 3.77872050e-01 6.65821791e-01 -6.42728388e-01
-5.41117072e-01 -5.94791293e-01 4.05226313e-02 1.45188022e+00
2.89805263e-01 4.00411747e-02 -7.63824105e-01 -9.33479190e-01
7.82332271e-02 8.27653229e-01 -7.27751374e-01 -7.60108352e-01
-6.93974018e-01 -8.52384865e-01 2.65242547e-01 2.52558470e-01
6.58329487e-01 -1.78363085e+00 -8.82366359e-01 2.13520631e-01
-2.35909626e-01 -2.79470354e-01 -1.31212249e-01 6.15170717e-01
-6.28065944e-01 -1.20017517e+00 -6.90262973e-01 -1.16113567e+00
7.24978447e-01 -1.17745720e-01 1.16018951e+00 4.84525710e-01
9.65823829e-02 -2.67694801e-01 -4.54119891e-01 -3.23918492e-01
-3.20384383e-01 -2.16899496e-02 -3.34717900e-01 2.66391188e-01
7.30289280e-01 2.73355749e-02 -5.16534269e-01 4.38889802e-01
-6.11269295e-01 -2.12493658e-01 4.61657435e-01 1.34605634e+00
3.88991594e-01 -3.07491086e-02 1.15834582e+00 -5.08166194e-01
1.05068684e+00 -7.23589003e-01 -3.33923876e-01 5.04149675e-01
-4.87406552e-01 1.45753488e-01 8.14954102e-01 -5.53309917e-01
-7.79164076e-01 -3.18512708e-01 -3.54442894e-01 -1.04689986e-01
6.77050799e-02 4.69415814e-01 -1.27408862e-01 -1.16651371e-01
7.00917542e-01 9.83531028e-02 1.03418753e-01 -1.32714316e-01
1.14561915e-01 9.24031913e-01 1.12183712e-01 -2.62023896e-01
2.84354419e-01 -5.79557084e-02 -6.97180212e-01 -2.54774332e-01
-8.00287962e-01 -4.76294160e-02 -2.34315753e-01 -2.24207968e-01
7.92088389e-01 -2.18677714e-01 -6.30216658e-01 3.71955395e-01
-1.20553625e+00 -2.87859857e-01 -5.75612009e-01 1.40129760e-01
-2.07892433e-01 -1.80049285e-01 -5.51069677e-01 -8.91992152e-01
-3.81401956e-01 -1.26075554e+00 4.18979496e-01 5.95820010e-01
-2.09806561e-01 -6.94899559e-01 -5.90959527e-02 4.90001023e-01
6.89103842e-01 -3.57450128e-01 1.50823033e+00 -1.03079188e+00
-1.79949909e-01 -2.72249997e-01 -4.22695018e-02 5.34014940e-01
-8.50522295e-02 -2.02823967e-01 -8.12324226e-01 -1.36612177e-01
3.35067421e-01 -6.77562594e-01 9.25395608e-01 8.01089853e-02
1.18221974e+00 -3.34854424e-01 -5.50498031e-02 8.60531926e-02
1.00340915e+00 6.73969805e-01 5.69604516e-01 7.43504882e-01
2.80845165e-01 8.45426083e-01 5.73201239e-01 -5.31295128e-02
4.00634766e-01 2.37237632e-01 7.11700082e-01 8.88161212e-02
3.09212804e-02 -7.44274780e-02 4.56989855e-01 5.86647272e-01
4.45142299e-01 -4.20029402e-01 -7.19125032e-01 4.99523312e-01
-1.81618178e+00 -8.37837815e-01 -5.21951122e-03 2.37618709e+00
9.28465009e-01 1.93309918e-01 -1.41072452e-01 5.04543841e-01
9.50560033e-01 7.41894543e-02 -8.50141048e-01 -7.02690721e-01
-1.60416633e-01 3.18141878e-01 -1.01318985e-01 4.83553529e-01
-8.28212619e-01 9.56883490e-01 5.43352795e+00 8.28320086e-01
-1.13910270e+00 3.77153642e-02 8.79372060e-01 9.99877080e-02
-3.52536917e-01 -8.07342231e-02 -6.11653030e-01 5.75690091e-01
8.17817032e-01 2.60979354e-01 5.27801096e-01 7.45366454e-01
-2.82269895e-01 -3.54038000e-01 -1.00210929e+00 8.43189955e-01
1.43974110e-01 -1.06519544e+00 1.27998725e-01 -4.35661435e-01
4.77732599e-01 -2.04517379e-01 5.95768951e-02 7.52421618e-01
1.81029230e-01 -9.10368979e-01 6.32067502e-01 3.25583041e-01
1.40145183e-01 -7.39341438e-01 1.04007077e+00 7.34062672e-01
-6.68467164e-01 -7.54515707e-01 -5.63786507e-01 -2.60646045e-01
-2.08720878e-01 2.96951175e-01 -2.29657307e-01 3.81291807e-01
8.42878580e-01 2.42435768e-01 -5.86342692e-01 1.03171575e+00
-5.18930852e-01 4.83011603e-01 -2.08342865e-01 -5.01251578e-01
3.61987710e-01 -3.67823899e-01 3.92477781e-01 5.68550348e-01
7.98342824e-02 8.33430514e-02 -6.61957562e-02 8.06669652e-01
-9.19051021e-02 1.72558337e-01 -2.48434544e-01 6.32713884e-02
4.35572028e-01 9.32702959e-01 -4.97620255e-01 -3.36591721e-01
-1.54498622e-01 1.02133477e+00 7.06264138e-01 4.30912226e-01
-7.00219691e-01 -7.17513442e-01 1.27498969e-01 -1.89140141e-01
3.97421598e-01 6.66527569e-01 -4.98162687e-01 -9.09653425e-01
8.14047158e-02 -1.00831485e+00 8.65323246e-01 -7.99793780e-01
-1.46769130e+00 9.18606937e-01 -6.09737456e-01 -8.97115767e-01
-1.48319006e-01 -4.84011114e-01 -8.25652003e-01 8.87837112e-01
-1.80075371e+00 -6.17110550e-01 -1.89264312e-01 4.90652114e-01
4.86474633e-01 -3.68846029e-01 9.23556209e-01 3.18626404e-01
-8.06109667e-01 6.90656006e-01 1.42812906e-02 5.63971885e-02
5.35934031e-01 -1.08645856e+00 -2.08050817e-01 4.96884078e-01
-1.23448402e-01 3.37519199e-01 3.15951824e-01 -3.92672807e-01
-8.54604065e-01 -8.55349243e-01 1.37399733e+00 -6.01337627e-02
4.82259214e-01 7.41182175e-03 -1.16200805e+00 3.49772841e-01
1.41064867e-01 1.24408212e-02 6.68662608e-01 -1.85667917e-01
-4.82160039e-02 -2.98313171e-01 -1.40930462e+00 4.27224219e-01
4.60208595e-01 -1.66596159e-01 -9.06395912e-01 3.36590677e-01
7.58181751e-01 -1.97008461e-01 -4.08382922e-01 3.20286214e-01
2.72324711e-01 -9.57682490e-01 6.87272251e-01 -8.93182576e-01
6.06362939e-01 -3.53828013e-01 5.12233377e-02 -1.43987608e+00
-4.61033612e-01 -2.92193983e-02 -1.99201018e-01 9.37420785e-01
5.84724903e-01 -8.52197230e-01 6.69622242e-01 4.74863499e-01
7.88758397e-02 -1.27476346e+00 -1.04167986e+00 -4.22871470e-01
2.46268868e-01 5.10989912e-02 8.27202499e-01 9.02494550e-01
-6.26727790e-02 6.80036128e-01 -3.81576009e-02 -9.20391083e-02
6.67496473e-02 3.60907704e-01 1.87277123e-01 -1.21482706e+00
-2.30370820e-01 -6.77789867e-01 -2.97120726e-03 -1.11353016e+00
1.83846638e-01 -8.40380907e-01 2.00284272e-02 -1.37981260e+00
-9.01695266e-02 -6.57752216e-01 -5.11205912e-01 5.30283093e-01
-4.74586010e-01 -4.39642258e-02 1.55524448e-01 1.01851635e-01
-5.80597699e-01 7.96761572e-01 1.21998715e+00 -3.94740582e-01
-2.79173613e-01 2.60962486e-01 -6.70132637e-01 7.18891442e-01
8.68820250e-01 -7.08194554e-01 -2.21634552e-01 -5.67770362e-01
4.97819155e-01 9.06173065e-02 2.56091118e-01 -7.54268169e-01
3.17960143e-01 -1.34030534e-02 3.97056133e-01 -3.87069166e-01
2.54320174e-01 -8.76095593e-01 -4.99859720e-01 9.76646185e-01
-6.50872052e-01 1.45324290e-01 -2.32784003e-01 4.61604297e-01
-4.16717082e-01 -8.69693398e-01 8.46724689e-01 -1.84048995e-01
-5.56481659e-01 7.85861909e-02 -3.27801913e-01 5.28384261e-02
1.02556014e+00 -3.52323353e-01 -4.94986564e-01 -3.44395787e-01
-5.24969637e-01 8.21524084e-01 -2.53075305e-02 5.48106074e-01
7.30138302e-01 -1.35799670e+00 -6.23149872e-01 4.63529706e-01
9.23546106e-02 -1.30484495e-02 2.47233491e-02 4.50325668e-01
-4.47417945e-01 1.78095087e-01 -1.45262554e-01 -2.98224151e-01
-9.33506906e-01 5.60448647e-01 8.19181144e-01 -4.99700814e-01
-5.34617566e-02 1.28478205e+00 -6.75881952e-02 -9.51627612e-01
3.94993603e-01 -2.12502018e-01 -9.75170851e-01 2.04164356e-01
6.40718579e-01 2.72668302e-01 2.12887168e-01 -3.97885680e-01
-3.12835604e-01 1.91472188e-01 -8.59587342e-02 1.39581919e-01
1.19829774e+00 1.89755350e-01 -4.92509991e-01 4.24405992e-01
9.85697865e-01 -4.39228803e-01 -7.19355524e-01 -1.30273998e-01
-3.69697660e-02 -2.45154738e-01 7.01478273e-02 -1.17102313e+00
-1.00110614e+00 8.84297669e-01 6.59893572e-01 4.97855037e-01
1.08038628e+00 -1.93726882e-01 9.05336261e-01 5.58306754e-01
1.55686736e-01 -1.22020006e+00 3.57612401e-01 8.68231177e-01
1.03106093e+00 -1.18491101e+00 -5.80529988e-01 7.88210854e-02
-7.19090521e-01 1.08777106e+00 1.10286891e+00 -1.97348684e-01
4.18180943e-01 -2.06726387e-01 3.12434323e-02 -9.64607745e-02
-9.21309650e-01 -6.25396743e-02 -4.89562005e-02 3.79455030e-01
2.74861064e-02 -1.93771139e-01 -4.78728443e-01 8.84144723e-01
-2.34981522e-01 -1.63692668e-01 -6.71459083e-03 9.63482499e-01
-6.79316282e-01 -1.01690555e+00 -5.36648333e-01 6.28306866e-01
-2.23862320e-01 -3.17679495e-02 -4.23336506e-01 2.60998905e-01
5.53269565e-01 1.17471325e+00 4.85294640e-01 -5.17032444e-01
2.91045338e-01 2.20371693e-01 1.12207867e-01 -5.32652676e-01
-1.20772946e+00 -5.11937380e-01 -9.94781256e-02 -3.18756402e-01
-1.33156896e-01 -1.55823275e-01 -1.35707927e+00 -1.10074498e-01
-6.59928441e-01 5.33123970e-01 2.16947854e-01 1.04740143e+00
3.21603477e-01 6.46844804e-01 7.26702154e-01 -3.19634289e-01
-8.25078070e-01 -8.80393863e-01 -2.05204561e-01 3.33680272e-01
4.44055378e-01 -6.28179252e-01 -5.50818443e-01 -4.80128139e-01] | [11.065961837768555, 8.071281433105469] |
cafb469b-ca59-4f74-96da-1eb426b997ce | identification-of-twitter-bots-based-on-an | 2112.04913 | null | https://arxiv.org/abs/2112.04913v2 | https://arxiv.org/pdf/2112.04913v2.pdf | Identification of Twitter Bots Based on an Explainable Machine Learning Framework: The US 2020 Elections Case Study | Twitter is one of the most popular social networks attracting millions of users, while a considerable proportion of online discourse is captured. It provides a simple usage framework with short messages and an efficient application programming interface (API) enabling the research community to study and analyze several aspects of this social network. However, the Twitter usage simplicity can lead to malicious handling by various bots. The malicious handling phenomenon expands in online discourse, especially during the electoral periods, where except the legitimate bots used for dissemination and communication purposes, the goal is to manipulate the public opinion and the electorate towards a certain direction, specific ideology, or political party. This paper focuses on the design of a novel system for identifying Twitter bots based on labeled Twitter data. To this end, a supervised machine learning (ML) framework is adopted using an Extreme Gradient Boosting (XGBoost) algorithm, where the hyper-parameters are tuned via cross-validation. Our study also deploys Shapley Additive Explanations (SHAP) for explaining the ML model predictions by calculating feature importance, using the game theoretic-based Shapley values. Experimental evaluation on distinct Twitter datasets demonstrate the superiority of our approach, in terms of bot detection accuracy, when compared against a recent state-of-the-art Twitter bot detection method. | ['Sotiris Ioannidis', 'Despoina Antonakaki', 'Christos Tzagkarakis', 'Alexander Shevtsov'] | 2021-12-08 | null | null | null | null | ['twitter-bot-detection'] | ['miscellaneous'] | [-2.84080863e-01 2.88636744e-01 -6.54289901e-01 1.67005673e-01
1.32494822e-01 -4.95606661e-01 1.10187948e+00 2.46426046e-01
-2.56402969e-01 6.36505127e-01 2.67949291e-02 -6.64878786e-01
1.00134753e-01 -1.12507200e+00 4.91018519e-02 -6.48642838e-01
9.31053907e-02 4.79239374e-01 2.38466620e-01 -7.57005453e-01
8.83515418e-01 2.41825759e-01 -1.70308006e+00 6.95469137e-03
1.04096663e+00 5.87480128e-01 7.41601810e-02 3.53701860e-01
-3.12092632e-01 1.07707095e+00 -6.58706069e-01 -9.08170700e-01
1.90053254e-01 -4.03922498e-01 -5.18462121e-01 -1.52945518e-01
-3.07070524e-01 -1.06853247e-01 -4.08573896e-02 1.19702506e+00
3.27445447e-01 -2.72487760e-01 6.51431143e-01 -1.60925221e+00
-3.75372589e-01 8.61752033e-01 -8.27189028e-01 2.03718588e-01
2.31256023e-01 4.87651378e-01 1.14945662e+00 -3.83255422e-01
8.65110815e-01 1.28366518e+00 6.22235775e-01 3.44538242e-01
-9.59058702e-01 -9.93449390e-01 -2.75810231e-02 3.41590762e-01
-9.23179090e-01 8.33747759e-02 9.82956171e-01 -7.80343890e-01
3.81547898e-01 4.22590792e-01 8.10321212e-01 1.16727376e+00
3.14776838e-01 3.73822331e-01 1.30922210e+00 -2.48654738e-01
3.84255707e-01 7.08846271e-01 2.72873312e-01 6.26634955e-01
5.35302877e-01 -7.01383576e-02 -4.02069926e-01 -5.94435394e-01
3.11686262e-03 -8.24623108e-02 1.80494741e-01 -9.82850790e-03
-5.55159807e-01 1.56726158e+00 5.95420122e-01 5.42163730e-01
-5.97241640e-01 -1.64182976e-01 6.25159204e-01 2.11901814e-01
9.55898106e-01 5.03516018e-01 -3.98997702e-02 -2.24321350e-01
-6.07129395e-01 3.04012626e-01 8.56754184e-01 2.66735464e-01
7.47625768e-01 -6.91276044e-02 -2.00376570e-01 3.51560265e-01
5.24183393e-01 3.21550280e-01 6.54845119e-01 -4.78805423e-01
3.82504016e-01 1.10632575e+00 5.32279313e-02 -1.69031298e+00
-4.20477092e-01 -5.88415682e-01 -6.19935632e-01 2.76287019e-01
4.08885092e-01 -1.47990733e-01 -3.45540717e-02 1.44305110e+00
7.54774094e-01 1.59346275e-02 -3.33530754e-01 7.25787580e-01
7.23496020e-01 4.80548799e-01 3.79675686e-01 -4.08956140e-01
1.55100787e+00 -5.33522308e-01 -7.13386238e-01 -1.80392131e-01
6.54010534e-01 -5.56965292e-01 1.04111850e+00 1.39023900e-01
-4.84505951e-01 -1.27976939e-01 -6.76009536e-01 4.12097543e-01
-6.90562665e-01 -3.81792188e-01 6.26934946e-01 1.16748691e+00
-6.66841865e-01 5.46727717e-01 -2.81898618e-01 -5.23807764e-01
4.90973949e-01 2.27222711e-01 1.14879586e-01 5.37616730e-01
-1.38065732e+00 8.87857616e-01 2.25705117e-01 -4.38113183e-01
-3.01630467e-01 -4.85776633e-01 -4.10864711e-01 -1.04881763e-01
3.01086247e-01 -4.91037399e-01 8.59986782e-01 -9.71301734e-01
-1.51968122e+00 1.00919878e+00 3.92591208e-01 -6.80896699e-01
8.12807798e-01 4.71339792e-01 -9.26181227e-02 -1.07972726e-01
5.86835146e-01 2.09270209e-01 1.07836246e+00 -1.13084018e+00
-6.46509111e-01 -5.32317162e-01 3.47175628e-01 -1.59708168e-02
-6.74335420e-01 3.72395605e-01 2.01934516e-01 -4.85674590e-01
-9.96651500e-02 -1.07439339e+00 -1.50321797e-01 -7.01314747e-01
-6.38362348e-01 -5.19286275e-01 1.17869186e+00 -7.26009727e-01
1.51929390e+00 -1.93024135e+00 -1.22005582e-01 3.92430305e-01
6.45286381e-01 4.11959976e-01 6.34405911e-01 6.92257404e-01
2.84754395e-01 5.17383754e-01 1.59476876e-01 -2.00317711e-01
1.84523351e-02 -2.76756257e-01 -3.41193438e-01 5.66272140e-01
-2.81021178e-01 6.10925794e-01 -1.05728483e+00 -5.96103728e-01
3.50127339e-01 2.54750289e-02 -6.21131480e-01 -5.20607792e-02
-2.08086878e-01 6.06035829e-01 -6.92615390e-01 4.63507354e-01
6.00153863e-01 -3.71109009e-01 5.67963243e-01 2.04518810e-01
-4.71878648e-01 2.15905190e-01 -6.44966006e-01 4.38450366e-01
-2.87070781e-01 6.64004207e-01 -1.14057353e-02 -6.41676784e-01
9.71676052e-01 -6.50324523e-02 4.30436134e-01 -3.12688768e-01
9.46614265e-01 2.24287346e-01 8.30737278e-02 -5.49357116e-01
5.32541335e-01 1.96179003e-01 -4.58550081e-03 8.09894979e-01
-4.88864869e-01 1.70980185e-01 2.24443510e-01 3.95388067e-01
9.25566196e-01 -4.10915405e-01 8.32243025e-01 -2.45371312e-01
5.99628448e-01 3.53179306e-01 1.83809265e-01 7.92620420e-01
-4.53183770e-01 -2.78970718e-01 9.59302247e-01 -4.37066138e-01
-7.86447465e-01 -2.79122591e-01 4.70114425e-02 1.41369092e+00
4.04438198e-01 -5.89103281e-01 -8.66560996e-01 -8.08547974e-01
1.44966781e-01 9.67661083e-01 -6.43731892e-01 1.20955713e-01
-4.23612565e-01 -9.70230460e-01 4.81225014e-01 -5.01490772e-01
7.60955989e-01 -1.15415919e+00 -6.57294273e-01 1.67840809e-01
-3.40653718e-01 -8.35638344e-01 2.40720034e-01 -4.00688171e-01
-5.26668251e-01 -1.51026070e+00 -1.44683495e-02 -2.56621361e-01
3.71437401e-01 5.85280299e-01 6.58312201e-01 4.32612777e-01
7.75093734e-02 -5.73791116e-02 -5.17997444e-01 -5.89070618e-01
-9.11949575e-01 2.50265241e-01 -3.79729364e-03 2.65349984e-01
5.39969981e-01 -6.02949142e-01 -4.25585598e-01 3.72296542e-01
-5.94692945e-01 1.80424735e-01 2.37404168e-01 5.86872578e-01
-3.91509444e-01 8.04858953e-02 5.65258861e-01 -1.25525129e+00
1.11776876e+00 -1.10356343e+00 -6.15148604e-01 -3.00940633e-01
-7.61803687e-01 -3.97187620e-01 7.09823728e-01 -4.05113637e-01
-7.73544490e-01 -5.38420081e-01 1.84606090e-01 7.79699460e-02
2.04413459e-01 4.63446438e-01 2.84656256e-01 -1.85251877e-01
1.03497195e+00 -8.33695903e-02 3.18112522e-01 -1.42434284e-01
2.41686657e-01 1.18468034e+00 -3.09751987e-01 5.91651872e-02
9.30316567e-01 6.51647389e-01 -9.56534520e-02 -7.36002326e-01
-5.55441678e-01 -4.73779559e-01 -1.40234903e-01 -8.66470218e-01
6.96520925e-01 -4.54671711e-01 -1.38824391e+00 5.00694811e-01
-1.15026760e+00 1.51434645e-01 2.53451198e-01 1.75974026e-01
-2.57918090e-01 2.95127988e-01 -5.47993362e-01 -1.13206518e+00
-4.02997196e-01 -1.07551539e+00 5.75215578e-01 7.86984935e-02
-4.61756974e-01 -9.48516011e-01 3.54342051e-02 8.69390190e-01
5.16474009e-01 4.38539028e-01 8.50093067e-01 -9.43467081e-01
-2.96560705e-01 -2.07191244e-01 -3.22930515e-01 -1.01565607e-01
-2.00957693e-02 2.00697526e-01 -8.15818608e-01 -4.53711301e-02
9.94357765e-02 -6.52778074e-02 4.31797773e-01 3.18664402e-01
7.71044970e-01 -7.33029246e-01 -5.16120255e-01 -8.95843729e-02
1.18256378e+00 -1.22620218e-01 3.14215243e-01 9.67383802e-01
4.40689236e-01 8.88637602e-01 6.83989286e-01 7.99561799e-01
4.70782101e-01 6.58125460e-01 1.04385161e+00 2.79738426e-01
1.97327569e-01 -2.98879355e-01 3.28236073e-01 4.75175232e-01
-2.50269085e-01 -8.17594156e-02 -1.06122673e+00 6.81615695e-02
-2.03228426e+00 -1.20214868e+00 -5.17595291e-01 1.87818074e+00
3.83825362e-01 3.40964377e-01 5.56792617e-01 3.16096216e-01
1.33908093e+00 3.86596084e-01 -2.27000445e-01 -4.39258665e-01
1.72295630e-01 -2.54112959e-01 7.72294343e-01 5.73334575e-01
-1.03480387e+00 1.05420721e+00 5.16553783e+00 1.11104274e+00
-1.31210053e+00 5.18019795e-01 6.47095919e-01 3.46228510e-01
8.69015697e-03 -1.12557448e-02 -6.83873594e-01 7.45710731e-01
7.97511339e-01 -4.43984717e-01 5.31425834e-01 1.05134034e+00
6.52027249e-01 5.14916927e-02 -7.89100304e-02 6.71435595e-01
-1.96819901e-02 -1.54293561e+00 -1.73576653e-01 4.61886525e-01
6.01850390e-01 4.42521684e-02 1.87604144e-01 3.30660492e-01
3.24071497e-01 -4.86700982e-01 7.31686115e-01 4.80482206e-02
3.33801299e-01 -6.24630094e-01 7.11965203e-01 8.45794678e-01
-6.26342297e-01 -6.51496172e-01 -8.38805363e-02 -6.09244764e-01
-1.12346262e-01 5.71604431e-01 -1.36981535e+00 2.88976043e-01
4.54642087e-01 5.86484611e-01 -5.90289116e-01 4.98991251e-01
-4.48900163e-01 8.69571209e-01 2.57876627e-02 -8.40215623e-01
4.24158871e-01 -3.75768900e-01 9.66485083e-01 8.83923769e-01
-5.96513320e-03 -1.46642372e-01 7.31204972e-02 7.96756506e-01
-7.30780959e-02 6.23789608e-01 -9.06589389e-01 6.67805374e-02
6.96097732e-01 1.48600435e+00 -1.01239657e+00 -1.70200869e-01
6.76210364e-03 3.17748964e-01 2.63703644e-01 -1.40883222e-01
-9.68278646e-01 -4.82178964e-02 5.01972258e-01 5.77796817e-01
-1.14252940e-01 1.94292173e-01 -5.48971355e-01 -8.50981116e-01
-4.30076808e-01 -1.02884221e+00 2.69833237e-01 -3.08528364e-01
-1.35363996e+00 2.89722294e-01 -5.26149496e-02 -1.28894806e+00
-3.94667149e-01 -4.05418456e-01 -7.88177431e-01 4.09299195e-01
-1.11097109e+00 -1.22739494e+00 -2.93299347e-01 2.81234920e-01
3.52616906e-01 -5.38495541e-01 3.66556853e-01 -7.35024735e-02
-6.44374192e-01 1.43102035e-01 5.82106784e-02 1.96900475e-03
1.47324324e-01 -9.50499058e-01 1.10961676e-01 4.75987673e-01
-3.56282413e-01 4.46548998e-01 1.20914757e+00 -8.46514702e-01
-1.10155213e+00 -9.98639703e-01 1.05188417e+00 -2.92045236e-01
1.16236925e+00 -3.79976928e-01 -1.75427780e-01 3.65445346e-01
1.82310984e-01 -5.89010715e-01 7.12346971e-01 1.96391255e-01
-2.49873221e-01 9.67773274e-02 -1.51415133e+00 9.57669616e-01
7.45544434e-01 -2.41841406e-01 -2.68185705e-01 7.52369583e-01
2.81288594e-01 2.43907254e-02 -1.65134281e-01 1.58633795e-02
5.26295125e-01 -1.45628679e+00 5.16019940e-01 -5.90472519e-01
5.62707126e-01 -1.72360718e-01 1.77101076e-01 -1.16280127e+00
-3.24318588e-01 -8.68514299e-01 -8.46855268e-02 1.11355376e+00
2.38747820e-01 -1.02359509e+00 8.75122309e-01 3.40647310e-01
4.86557633e-01 -5.51562846e-01 -9.64323759e-01 -3.91583055e-01
-2.98713684e-01 -3.98934901e-01 5.84303916e-01 1.28569865e+00
3.39430988e-01 4.55115736e-01 -5.61138511e-01 2.43347744e-03
6.62228227e-01 1.14866942e-02 1.25804114e+00 -1.42043459e+00
-7.87786767e-02 -8.39131117e-01 -8.30020487e-01 -2.83940732e-01
2.63053536e-01 -8.56941879e-01 -6.98457301e-01 -1.05189407e+00
1.86286077e-01 -3.76945645e-01 2.83450872e-01 1.45047709e-01
-1.60205122e-02 4.68843222e-01 3.55475932e-01 5.98637640e-01
-3.80743414e-01 2.94992834e-01 1.01782012e+00 -1.93637580e-01
-3.61983359e-01 3.07639599e-01 -1.06621182e+00 9.96401906e-01
1.05826032e+00 -7.93488264e-01 -1.21522203e-01 4.10337180e-01
5.97584844e-01 -2.35515401e-01 3.66309941e-01 -7.04264045e-01
1.62834421e-01 -2.69062459e-01 -3.37839812e-01 -3.92589599e-01
1.04729176e-01 -6.56255603e-01 8.43533203e-02 1.14177191e+00
-1.05955236e-01 -1.53607041e-01 -3.01422060e-01 6.15511775e-01
5.59136309e-02 -2.71677047e-01 8.86110842e-01 -4.80216816e-02
-2.29661986e-01 -8.10056850e-02 -7.01779962e-01 -1.55445799e-01
1.55582583e+00 -2.95136690e-01 -7.24637032e-01 -7.14662433e-01
-3.24032158e-01 -3.59055363e-02 3.40026557e-01 3.56677860e-01
5.84223606e-02 -1.03045666e+00 -9.09676433e-01 -1.13542594e-01
5.10096140e-02 -1.08489871e+00 8.37860815e-03 1.03739142e+00
-7.27521777e-01 1.95535734e-01 -2.13416815e-01 -4.42911118e-01
-1.39411902e+00 4.09863651e-01 4.00774963e-02 -5.37404001e-01
-4.72042769e-01 3.23381901e-01 -2.19940096e-01 -5.17379761e-01
-1.15151674e-01 4.43315506e-01 -8.79800379e-01 4.63243484e-01
5.12341857e-01 8.58636498e-01 -1.60045415e-01 -1.04074335e+00
-1.65892959e-01 -1.15279868e-01 9.94083583e-02 1.31966963e-01
1.07917023e+00 -8.16134810e-02 -5.24809837e-01 2.50750005e-01
8.19266200e-01 4.29624736e-01 -4.48989660e-01 1.13591865e-01
1.27807885e-01 -7.73165524e-01 1.50466152e-03 -4.17285830e-01
-7.44245768e-01 2.73838580e-01 2.52962917e-01 1.23065424e+00
5.56992888e-01 -7.04733953e-02 4.04431939e-01 1.22663230e-01
4.55987424e-01 -9.40403283e-01 -6.21404722e-02 4.74113375e-01
5.76224267e-01 -1.23526621e+00 -3.27571854e-02 -6.48728788e-01
-6.19915485e-01 9.57216144e-01 4.44281399e-01 -2.04085827e-01
8.50548446e-01 -3.68276596e-01 5.89959733e-02 -3.99771541e-01
-4.50000465e-01 -1.49926513e-01 -2.76114911e-01 3.96380842e-01
1.94537207e-01 3.18316370e-01 -1.11219227e+00 5.47745466e-01
-5.74191451e-01 -4.32882786e-01 6.75706625e-01 6.34084046e-01
-7.30342567e-01 -9.11607206e-01 -5.98356247e-01 6.20643795e-01
-6.16370738e-01 -2.77904351e-03 -8.66541564e-01 9.77221608e-01
2.31061235e-01 1.40545392e+00 -1.90963417e-01 -7.13891327e-01
-1.53851017e-01 -3.17096293e-01 -2.97902197e-01 -3.90481353e-01
-1.20412970e+00 -4.99467731e-01 2.69119143e-01 -2.41511941e-01
-5.25497913e-01 -5.23253143e-01 -8.05005848e-01 -1.18572140e+00
-7.14227617e-01 3.87440026e-01 1.06422830e+00 1.03126657e+00
2.40746379e-01 6.02184758e-02 1.29035091e+00 -8.11003327e-01
-5.55356979e-01 -1.15054393e+00 -5.08455515e-01 4.76361096e-01
-2.27644429e-01 -8.85449886e-01 -7.64229834e-01 -6.04112148e-01] | [8.086346626281738, 10.130343437194824] |
e79f7384-8536-4839-b84a-6460a55ec3d7 | a-stock-prediction-model-based-on-dcnn | 2009.03239 | null | https://arxiv.org/abs/2009.03239v1 | https://arxiv.org/pdf/2009.03239v1.pdf | A Stock Prediction Model Based on DCNN | The prediction of a stock price has always been a challenging issue, as its volatility can be affected by many factors such as national policies, company financial reports, industry performance, and investor sentiment etc.. In this paper, we present a prediction model based on deep CNN and the candle charts, the continuous time stock information is processed. According to different information richness, prediction time interval and classification method, the original data is divided into multiple categories as the training set of CNN. In addition, the convolutional neural network is used to predict the stock market and analyze the difference in accuracy under different classification methods. The results show that the method has the best performance when the forecast time interval is 20 days. Moreover, the Moving Average Convergence Divergence and three kinds of moving average are added as input. This method can accurately predict the stock trend of the US NDAQ exchange for 92.2%. Meanwhile, this article distinguishes three conventional classification methods to provide guidance for future research. | ['Ningning Liu', 'Qiao Zhou'] | 2020-09-07 | null | null | null | null | ['stock-prediction'] | ['time-series'] | [-9.51278925e-01 -8.66572738e-01 -2.13196918e-01 -3.71539772e-01
6.75931275e-02 -4.76362169e-01 5.70566833e-01 -2.16709867e-01
-3.13471615e-01 7.35606670e-01 1.15882210e-01 -5.55348575e-01
-3.11582424e-02 -1.23634744e+00 -3.15589368e-01 -6.36629522e-01
-9.93178859e-02 -3.91818024e-02 1.39330938e-01 -5.95870793e-01
6.49099231e-01 5.46381891e-01 -1.38315260e+00 8.85597840e-02
6.70306683e-01 1.94947755e+00 -1.41451359e-01 -1.22466320e-02
-6.64163828e-01 1.09158516e+00 -7.68508494e-01 -4.46676910e-01
6.92397356e-01 -1.47034705e-01 6.67973049e-03 -2.32106745e-01
-2.00950757e-01 -5.95616698e-01 -3.34914684e-01 1.29674935e+00
2.15619087e-01 -2.23930217e-02 4.31390643e-01 -1.17872655e+00
-9.09123957e-01 8.76312673e-01 -6.30443275e-01 7.15781212e-01
-6.96444571e-01 -3.10851121e-03 9.10836399e-01 -7.86205947e-01
2.12036192e-01 8.79379570e-01 6.74220383e-01 -3.43988556e-03
-6.50977910e-01 -1.36604941e+00 3.59174371e-01 3.66848528e-01
-8.27112257e-01 1.44042611e-01 8.35500419e-01 -8.13368917e-01
6.63078129e-01 3.53195734e-04 1.10353518e+00 4.32464987e-01
8.69518638e-01 7.29639292e-01 1.04390132e+00 3.64191890e-01
8.09137896e-02 7.22563267e-02 3.57700437e-01 -6.99966401e-02
4.97061402e-01 2.42916942e-01 2.60779485e-02 2.78346062e-01
7.14800179e-01 4.56536174e-01 -1.29922450e-01 3.77370447e-01
-8.95043612e-01 9.54552531e-01 6.43835664e-01 4.45476264e-01
-4.37550038e-01 -1.88837036e-01 6.90999746e-01 7.18915045e-01
8.34248960e-01 2.73247361e-01 -9.58703041e-01 -1.99111458e-02
-1.00700593e+00 5.81778407e-01 6.68598890e-01 4.23776001e-01
3.59516114e-01 8.40373635e-01 -6.62566572e-02 4.75138277e-01
1.69021785e-01 4.31734085e-01 1.09545588e+00 -4.96942967e-01
5.39085329e-01 7.05235720e-01 1.89264938e-01 -1.36901951e+00
-6.71174347e-01 -9.39736545e-01 -1.13018048e+00 4.81709361e-01
1.97495922e-01 -4.32113230e-01 -7.14688659e-01 1.10867023e+00
-2.20176518e-01 7.39667425e-03 2.22049817e-01 7.98012614e-01
6.74039066e-01 1.10630035e+00 -3.34614575e-01 -4.68646646e-01
1.05489814e+00 -8.52017343e-01 -1.12483931e+00 1.21993415e-01
2.13005170e-01 -5.87784171e-01 3.27992409e-01 5.21325290e-01
-8.66781831e-01 -5.92333138e-01 -1.09935355e+00 3.55429620e-01
-8.30212533e-01 1.59879923e-01 4.20631051e-01 2.59234190e-01
-6.89181030e-01 9.24297690e-01 -5.57406843e-01 6.35047019e-01
3.25841159e-01 8.84145573e-02 1.87871099e-01 5.65343976e-01
-1.64097524e+00 9.29409087e-01 7.57863224e-01 4.48741078e-01
4.58934642e-02 -6.99748397e-01 -4.73936886e-01 3.61907631e-01
-2.14804590e-01 -9.47687328e-02 1.32601380e+00 -1.05618155e+00
-1.43587053e+00 1.37694523e-01 5.18711507e-01 -8.22273612e-01
7.44390726e-01 7.17089698e-02 -8.78380477e-01 -4.15044606e-01
-7.67038316e-02 -2.95913108e-02 4.93470103e-01 -4.40355390e-01
-1.04257298e+00 -4.33338463e-01 -4.44180638e-01 -1.87692821e-01
-2.78164297e-01 1.62614640e-02 -4.78912517e-02 -1.22856307e+00
2.21263215e-01 -6.14185929e-01 -9.54542533e-02 -6.66836023e-01
-7.16086477e-02 -2.90586948e-01 8.58021736e-01 -9.16951895e-01
1.49692440e+00 -2.12817502e+00 -4.26587731e-01 4.53916669e-01
3.27328444e-02 2.25388154e-01 3.27033371e-01 2.35516414e-01
-4.46202636e-01 3.22165579e-01 -2.41933041e-03 3.14630568e-01
1.16561517e-01 -2.00076729e-01 -8.69442225e-01 2.83313721e-01
2.83908993e-01 1.00196731e+00 -3.14384609e-01 9.65017155e-02
6.53247610e-02 2.56162137e-01 1.75785020e-01 -7.51510561e-02
-2.04539299e-01 1.98097646e-01 -4.89480853e-01 5.18731952e-01
1.06928134e+00 -3.67337108e-01 -3.20813119e-01 -6.35856017e-02
-7.69473493e-01 2.83687413e-01 -1.23039925e+00 6.08430862e-01
-1.06152399e-02 8.28562558e-01 -3.77373725e-01 -8.32285106e-01
1.55506873e+00 1.61673889e-01 5.25046766e-01 -1.11841726e+00
4.18740720e-01 6.84771597e-01 3.22989404e-01 -1.70598745e-01
5.84859788e-01 -2.20063150e-01 2.30221495e-01 3.41858327e-01
-5.81885159e-01 2.59096473e-01 2.05216467e-01 -4.17864680e-01
3.80497456e-01 -3.10911000e-01 1.00303672e-01 -3.82561773e-01
3.36192638e-01 -7.94494078e-02 9.75594163e-01 8.77842084e-02
-1.96371853e-01 3.03987831e-01 7.52802432e-01 -1.10133278e+00
-8.72577012e-01 -4.65016991e-01 -5.90402186e-01 4.71179515e-01
-3.25713456e-02 2.15384379e-01 -1.95816621e-01 -2.04483464e-01
4.66070622e-01 6.85847521e-01 -7.42867112e-01 -7.23304376e-02
-3.09727609e-01 -1.00789440e+00 2.09879026e-01 6.68576181e-01
1.00769246e+00 -1.37286973e+00 -4.90253806e-01 4.39636469e-01
3.74614626e-01 -5.25146544e-01 -2.19493598e-01 1.03423700e-01
-9.82688546e-01 -9.22340453e-01 -9.65482712e-01 -5.76594055e-01
-2.84273736e-02 -4.40048613e-02 1.06860924e+00 4.55985479e-02
4.53186899e-01 -5.56998670e-01 -1.04912803e-01 -9.83758688e-01
7.23085850e-02 1.67986467e-01 1.23002883e-02 1.57271907e-01
5.12775004e-01 -2.67345369e-01 -7.20443487e-01 7.06459954e-02
-8.07169437e-01 -1.48889244e-01 4.46062893e-01 6.98331416e-01
3.97891641e-01 6.89014792e-01 8.65534067e-01 -4.78482693e-01
8.70934486e-01 -6.84260070e-01 -1.35164058e+00 -8.57371092e-02
-1.24411237e+00 -2.44403109e-02 6.05138719e-01 -2.84085989e-01
-9.34450805e-01 -4.53605235e-01 4.09974866e-02 -5.98995328e-01
4.39899296e-01 1.01782095e+00 2.81218827e-01 2.99824923e-01
-1.07575893e-01 4.76236582e-01 1.63745612e-01 -5.68501770e-01
-3.52789372e-01 4.65555042e-01 2.47891456e-01 2.80236423e-01
7.11347461e-01 6.92462325e-02 -1.81601405e-01 -1.81845605e-01
-7.36364305e-01 -4.12685610e-03 -4.16305393e-01 -2.80736119e-01
6.43347263e-01 -9.46274042e-01 -8.94031048e-01 1.10956228e+00
-1.10576212e+00 -1.49326911e-02 3.22299525e-02 6.81499064e-01
8.97297040e-02 -2.11463854e-01 -1.06690347e+00 -9.84111011e-01
-6.74217522e-01 -1.17309260e+00 2.72941768e-01 4.69619989e-01
3.04114580e-01 -1.11395979e+00 -9.99548752e-03 -3.15052688e-01
7.20143974e-01 3.78197074e-01 6.73189461e-01 -9.78890777e-01
-5.65617800e-01 -4.83577877e-01 -3.91715258e-01 6.74566209e-01
1.54513270e-01 4.01359469e-01 -6.09581530e-01 -8.00020471e-02
1.92042738e-01 1.30472481e-01 1.01692426e+00 6.85670853e-01
1.14501357e+00 -5.51116288e-01 2.92174518e-01 8.30588758e-01
1.48595226e+00 1.06724155e+00 7.48098195e-01 1.12552691e+00
3.97469431e-01 5.57029009e-01 5.27359068e-01 5.72416306e-01
1.66133240e-01 -6.15313984e-02 4.18092608e-01 1.77988842e-01
8.06259573e-01 6.59429953e-02 3.12297672e-01 1.21247017e+00
-1.28209308e-01 9.93218869e-02 -9.97029960e-01 9.65093821e-02
-1.71047390e+00 -1.36595786e+00 -3.25409353e-01 1.81158638e+00
5.16943574e-01 4.96683180e-01 1.04739860e-01 2.47622609e-01
7.02545166e-01 3.51426810e-01 -9.90820050e-01 -1.51818410e-01
-3.65105510e-01 -1.85537830e-01 8.58073950e-01 -5.75547554e-02
-1.25121045e+00 4.69463438e-01 6.30326653e+00 5.86227953e-01
-1.85405362e+00 -4.52252775e-01 1.15539849e+00 -9.73108970e-03
-2.62141019e-01 -5.29086769e-01 -9.79626000e-01 1.18370175e+00
8.15976501e-01 -7.05185115e-01 1.57777682e-01 9.97211695e-01
2.54661888e-01 3.20880264e-01 -3.05484295e-01 9.39887881e-01
-2.88167655e-01 -1.75686550e+00 3.07417307e-02 2.50928432e-01
7.52010465e-01 2.17996702e-01 5.52323103e-01 5.68280160e-01
1.69772595e-01 -8.59928846e-01 9.11696076e-01 9.46863711e-01
2.43470967e-01 -1.17593384e+00 1.34525645e+00 3.36947978e-01
-1.41020167e+00 -4.97597575e-01 -4.02607054e-01 -5.15531361e-01
-3.52761149e-02 7.86810696e-01 -4.49919924e-02 5.39143503e-01
9.84842300e-01 1.05625081e+00 -3.33972335e-01 9.86930132e-01
3.09280068e-01 4.44446355e-01 -8.85964781e-02 -3.45397234e-01
4.92145807e-01 -8.40829790e-01 -2.28798296e-02 7.41438508e-01
7.72032440e-01 1.46783501e-01 -4.19451743e-02 7.42064476e-01
8.93887598e-03 3.99752736e-01 -3.90352219e-01 -1.44523859e-01
2.74174660e-01 8.80211711e-01 -5.79560876e-01 -5.07574677e-01
-5.94766438e-01 2.21447572e-01 -4.57344837e-02 3.01370561e-01
-5.75410962e-01 -6.70759439e-01 7.04819441e-01 -1.37186246e-02
4.66149360e-01 -4.46383655e-02 -7.63945341e-01 -1.33636892e+00
2.32542306e-01 -5.15577495e-01 3.91780376e-01 -7.36420333e-01
-1.56584799e+00 6.34407222e-01 -3.91233325e-01 -1.72641027e+00
-2.11144596e-01 -1.04933500e+00 -1.14808929e+00 1.12893713e+00
-1.97624815e+00 -2.85717547e-01 -6.50917813e-02 2.50877500e-01
4.22458798e-01 -1.00673234e+00 2.18245149e-01 3.32962930e-01
-8.87750804e-01 1.18762910e-01 8.47095907e-01 7.49583244e-01
1.94745123e-01 -1.17992914e+00 6.20470822e-01 7.27155626e-01
-3.42740417e-01 3.64991874e-01 3.86409819e-01 -7.46146798e-01
-6.97653413e-01 -1.15660751e+00 6.95131421e-01 1.58764765e-01
1.30486608e+00 3.04411203e-01 -1.25847268e+00 7.23727524e-01
2.80528575e-01 -5.58239184e-02 3.95823479e-01 -2.42216200e-01
-4.06313151e-01 -7.06476450e-01 -8.66800010e-01 3.20883214e-01
1.39629636e-02 5.47124855e-02 -5.37263691e-01 -9.43657905e-02
7.78570175e-01 -3.84814233e-01 -9.93814886e-01 4.78729159e-01
6.72272980e-01 -1.22553742e+00 4.64914829e-01 -5.95169961e-01
6.19969130e-01 -1.10421300e-01 1.16945453e-01 -1.50159192e+00
-6.09219909e-01 4.31254283e-02 1.30818076e-02 1.05862463e+00
5.30304790e-01 -1.17278159e+00 6.06866002e-01 6.92686915e-01
-1.30520966e-02 -9.77894485e-01 -7.87822247e-01 -9.04579222e-01
4.72111404e-01 -3.73743355e-01 1.42853391e+00 1.18969762e+00
-2.50245184e-01 9.14916992e-02 -3.18660855e-01 -1.30416319e-01
1.69292554e-01 5.75355649e-01 4.86712426e-01 -1.59251487e+00
2.37338647e-01 -1.18385696e+00 -2.89421588e-01 -7.07854152e-01
1.61036715e-01 -5.50831258e-01 -4.51173991e-01 -1.32560706e+00
-3.51103902e-01 -2.99666941e-01 -9.46189702e-01 1.72758833e-01
-3.10670137e-02 -2.54923046e-01 3.35474074e-01 6.29901290e-01
2.01556668e-01 6.72463894e-01 1.32516134e+00 -3.66234481e-01
-1.78565666e-01 5.34720063e-01 -4.11762267e-01 6.22025788e-01
1.33925235e+00 1.05445273e-01 -1.37971103e-01 -3.08398843e-01
5.65660000e-01 1.72463283e-01 -1.40690491e-01 -9.06941891e-01
1.00692943e-01 -3.87545586e-01 8.63390565e-01 -1.30663264e+00
-1.15339182e-01 -7.61853516e-01 2.94131130e-01 7.83296943e-01
-2.49968052e-01 8.56708169e-01 2.82068700e-01 3.16376328e-01
-8.41553390e-01 -7.54499510e-02 7.01352537e-01 -1.87252879e-01
-7.56236315e-01 8.31896901e-01 -2.90757388e-01 -1.55919358e-01
9.90533113e-01 -4.01145332e-02 -3.79086763e-01 -4.72375840e-01
-3.69670451e-01 5.30347764e-01 -3.85417603e-02 4.56280440e-01
4.82960552e-01 -1.79390895e+00 -7.98408389e-01 2.45324790e-01
-2.23514676e-01 -1.42755508e-01 1.68418258e-01 5.96412659e-01
-8.28649104e-01 5.36219954e-01 -4.16188329e-01 -4.64201868e-02
-5.77926934e-01 6.72713518e-01 7.21090019e-01 -2.83344477e-01
-5.25379777e-01 4.58435684e-01 -4.52601649e-02 -7.89858028e-02
1.31777093e-01 -6.98731124e-01 -8.43976617e-01 7.69133449e-01
9.02417064e-01 3.21321160e-01 2.49278858e-01 -5.25269032e-01
8.08643028e-02 5.98826170e-01 -1.71805069e-01 1.22140832e-01
1.67417836e+00 2.50274181e-01 -3.26090753e-01 1.04304039e+00
1.32841384e+00 -3.50350469e-01 -1.15156794e+00 -1.28937513e-01
3.95869225e-01 -3.40631574e-01 1.91292077e-01 -5.09510696e-01
-1.89418757e+00 8.71722877e-01 5.86211801e-01 9.05081868e-01
1.06013405e+00 -6.88779414e-01 1.01203597e+00 3.47063690e-01
2.85535473e-02 -1.27701652e+00 -3.27622265e-01 9.55052912e-01
9.14165080e-01 -1.37440932e+00 -1.25036702e-01 2.58310199e-01
-6.87289536e-01 1.52110267e+00 5.92214942e-01 -4.50729102e-01
1.31590641e+00 2.12146342e-01 5.38949788e-01 -4.34477031e-02
-8.41268539e-01 1.43004164e-01 3.20444286e-01 -1.30689830e-01
4.84035164e-01 2.53236573e-02 -4.41719174e-01 9.89562273e-01
-7.55962193e-01 -1.07511863e-01 4.33043927e-01 4.68621969e-01
-4.41994309e-01 -4.62546468e-01 -3.74694198e-01 8.90154779e-01
-8.29965770e-01 -1.52520895e-01 -7.36918207e-03 8.75298440e-01
-1.54232830e-01 4.28516090e-01 9.60265815e-01 -6.46137834e-01
4.23016131e-01 1.82565346e-01 -5.31533599e-01 -6.30326793e-02
-6.65570855e-01 2.79229991e-02 -4.66887832e-01 -1.14998966e-01
-7.72569329e-02 -6.71179295e-01 -1.24236917e+00 -6.33109748e-01
-1.20522991e-01 2.08242387e-01 5.86956918e-01 8.70062590e-01
1.30653903e-01 5.70983648e-01 1.24249935e+00 -6.99420035e-01
-7.77891994e-01 -1.04621601e+00 -1.15873957e+00 2.68385738e-01
6.13978684e-01 -6.52324200e-01 -5.16350091e-01 -2.22676948e-01] | [4.451918125152588, 4.227987289428711] |
09cab97a-d075-466c-801f-f97bf3df31a1 | box-supervised-instance-segmentation-with | 2207.09055 | null | https://arxiv.org/abs/2207.09055v1 | https://arxiv.org/pdf/2207.09055v1.pdf | Box-supervised Instance Segmentation with Level Set Evolution | In contrast to the fully supervised methods using pixel-wise mask labels, box-supervised instance segmentation takes advantage of the simple box annotations, which has recently attracted a lot of research attentions. In this paper, we propose a novel single-shot box-supervised instance segmentation approach, which integrates the classical level set model with deep neural network delicately. Specifically, our proposed method iteratively learns a series of level sets through a continuous Chan-Vese energy-based function in an end-to-end fashion. A simple mask supervised SOLOv2 model is adapted to predict the instance-aware mask map as the level set for each instance. Both the input image and its deep features are employed as the input data to evolve the level set curves, where a box projection function is employed to obtain the initial boundary. By minimizing the fully differentiable energy function, the level set for each instance is iteratively optimized within its corresponding bounding box annotation. The experimental results on four challenging benchmarks demonstrate the leading performance of our proposed approach to robust instance segmentation in various scenarios. The code is available at: https://github.com/LiWentomng/boxlevelset. | ['Lei Zhang', 'Xiansheng Hua', 'Miaomiao Cui', 'Jianke Zhu', 'Wenyu Liu', 'Wentong Li'] | 2022-07-19 | null | null | null | null | ['box-supervised-instance-segmentation'] | ['computer-vision'] | [ 3.70471865e-01 3.35514218e-01 -2.97726333e-01 -7.06254900e-01
-9.11357760e-01 -2.80815274e-01 3.64225239e-01 7.95553327e-02
-4.80828702e-01 5.83829105e-01 -4.69032437e-01 1.11711606e-01
-1.61849856e-02 -8.03654194e-01 -8.17048609e-01 -7.90690303e-01
1.92797974e-01 5.89381278e-01 4.66124237e-01 5.44695482e-02
3.25985998e-01 2.78748691e-01 -1.22113705e+00 1.23177900e-03
1.30648792e+00 1.21134591e+00 1.96720064e-01 3.52210969e-01
-4.61511940e-01 3.70984495e-01 -4.31424171e-01 -1.99301004e-01
4.86605883e-01 -5.34882843e-01 -7.02315569e-01 5.02998292e-01
2.97428727e-01 -5.19768558e-02 -4.97412235e-02 1.19335103e+00
2.48788297e-01 4.26267654e-01 6.02116942e-01 -9.70506489e-01
-2.99998730e-01 4.52674925e-01 -8.69893014e-01 3.71923819e-02
-2.92139560e-01 3.87823015e-01 8.58091593e-01 -9.72191572e-01
5.39675951e-01 8.16257656e-01 4.25303191e-01 4.36625481e-01
-1.16859865e+00 -4.50365603e-01 3.54775786e-01 -1.25791103e-01
-1.46332121e+00 9.70053300e-03 1.15238166e+00 -4.56347078e-01
3.01819026e-01 -7.56611116e-03 8.50247025e-01 2.77425319e-01
-3.00522763e-02 8.92716527e-01 1.15401971e+00 -2.94611871e-01
3.00831079e-01 2.45888576e-01 3.47384751e-01 1.02701879e+00
-7.82825276e-02 -7.94068798e-02 -5.36234677e-02 1.85245782e-01
9.24505830e-01 2.12646365e-01 -2.98348784e-01 -6.30596638e-01
-9.54758704e-01 6.41483724e-01 8.64630461e-01 2.25979298e-01
-1.26022696e-01 1.33508801e-01 2.98535764e-01 -2.15731815e-01
8.05348814e-01 1.97820023e-01 -4.24049914e-01 1.83767840e-01
-1.23765385e+00 1.38434157e-01 4.99378949e-01 7.31759429e-01
1.21738839e+00 -1.24163777e-01 -5.03345907e-01 7.47740448e-01
2.40902305e-01 -3.25636640e-02 2.21372724e-01 -9.73967075e-01
3.21351320e-01 1.12744725e+00 1.40721798e-01 -6.40711248e-01
-2.29656622e-01 -5.93531728e-01 -7.54009128e-01 3.75013143e-01
5.69476783e-01 -1.07866913e-01 -1.22117555e+00 1.18267894e+00
8.67896616e-01 4.59308684e-01 -2.22414568e-01 1.04697478e+00
7.53798068e-01 8.15906227e-01 -7.89667889e-02 -2.14861751e-01
1.16841733e+00 -1.33229983e+00 -6.30524695e-01 -1.93115488e-01
5.18787742e-01 -2.75292665e-01 1.33871579e+00 2.27618545e-01
-1.23701680e+00 -6.14703774e-01 -1.06518924e+00 -1.09538853e-01
-3.15537781e-01 1.77013710e-01 3.50825250e-01 4.54314440e-01
-7.22246110e-01 6.65181160e-01 -1.00986660e+00 1.86681703e-01
1.09707093e+00 3.27325821e-01 6.44684732e-02 1.05134778e-01
-8.49764347e-01 3.16286474e-01 7.05637276e-01 3.78769398e-01
-7.26992488e-01 -7.57463157e-01 -9.18299735e-01 4.32540067e-02
7.39686728e-01 -3.46600235e-01 1.04677165e+00 -1.08961582e+00
-1.60328531e+00 1.07308757e+00 -8.10448825e-02 -4.77641046e-01
8.46601844e-01 -1.43630370e-01 2.93375790e-01 2.13781461e-01
1.08397938e-01 7.58703172e-01 9.10170019e-01 -1.41848505e+00
-4.96048927e-01 -3.52193236e-01 -1.08739480e-01 1.80561841e-01
-2.08216056e-01 -1.03783071e-01 -7.58180380e-01 -5.43323040e-01
8.71444717e-02 -4.73994285e-01 -5.05775571e-01 2.29192272e-01
-5.89177370e-01 -2.46732563e-01 9.17554915e-01 -4.35722619e-01
1.36698771e+00 -1.99413848e+00 2.51418680e-01 2.82190859e-01
1.75518349e-01 1.66465670e-01 4.30328250e-01 1.11422846e-02
2.25195885e-01 2.05949262e-01 -1.46775460e+00 -4.85077471e-01
-1.51337832e-01 4.60623764e-02 6.76888824e-02 6.11554325e-01
4.47441906e-01 9.56753731e-01 -8.39568198e-01 -7.89670169e-01
4.51488912e-01 5.31956971e-01 -4.25473362e-01 3.27835351e-01
-6.86674178e-01 7.79946804e-01 -4.41547930e-01 6.07610106e-01
7.87732720e-01 -2.93383688e-01 -3.05051059e-01 5.14155924e-02
-3.28410864e-01 -3.70206386e-01 -1.14238513e+00 1.88168347e+00
-1.93683475e-01 3.35971683e-01 1.89087301e-01 -1.11801779e+00
1.27729273e+00 -3.14986967e-02 5.37277818e-01 -3.29124212e-01
3.70135933e-01 3.07190895e-01 -1.80031642e-01 -2.97432482e-01
9.24414471e-02 -2.51626492e-01 -2.03384116e-01 1.08171389e-01
-8.53126347e-02 -4.46939588e-01 3.28624398e-01 7.37465620e-02
4.63158876e-01 5.75950027e-01 8.07697773e-02 -3.88059020e-01
8.83936346e-01 2.27960154e-01 7.28957236e-01 4.31961775e-01
-2.47793421e-01 9.21021223e-01 5.06251931e-01 -5.37316620e-01
-8.39993656e-01 -7.12843776e-01 -4.99115646e-01 6.42523825e-01
6.01675391e-01 -1.12209618e-01 -1.33393323e+00 -9.30035412e-01
-2.81364918e-01 5.50955713e-01 -7.09063113e-01 4.49214056e-02
-7.43271649e-01 -5.73301315e-01 1.22638769e-01 4.91547614e-01
9.16955590e-01 -1.23862684e+00 -7.48386145e-01 2.29364485e-01
1.65040240e-01 -7.83599794e-01 -8.80371928e-01 2.04236671e-01
-9.10264492e-01 -9.12287831e-01 -1.03133142e+00 -9.86511111e-01
1.05154550e+00 -1.96113557e-01 7.71188080e-01 3.36169362e-01
-5.86653054e-01 -1.04435915e-02 -5.17926291e-02 -1.25241593e-01
4.80159943e-04 8.93662423e-02 -4.95824069e-01 4.64320421e-01
-7.19871670e-02 -2.44384795e-01 -1.07831335e+00 3.71475041e-01
-9.65409398e-01 1.75333977e-01 4.14165050e-01 7.54186988e-01
1.35790217e+00 2.49205567e-02 4.52530861e-01 -1.09403384e+00
2.19181046e-01 -2.86726803e-01 -9.51991200e-01 1.77753512e-02
-4.03551251e-01 -1.21367872e-01 5.79991579e-01 -2.32716814e-01
-1.13333571e+00 3.44402134e-01 -1.03530511e-01 -4.66265321e-01
-3.22814792e-01 2.08025485e-01 -3.07678103e-01 -2.27578208e-02
3.33922237e-01 3.03796142e-01 1.43393241e-02 -4.34813112e-01
4.92773771e-01 3.69951248e-01 5.69784701e-01 -6.06867254e-01
8.35917056e-01 7.66009092e-01 -1.84929103e-01 -5.59541345e-01
-1.17340958e+00 -5.47343314e-01 -1.04119515e+00 -4.27159041e-01
1.15972948e+00 -4.26245779e-01 -4.17762011e-01 5.66167593e-01
-9.76028621e-01 -8.48585069e-01 -6.12976193e-01 5.05951941e-02
-6.39388442e-01 2.49174833e-01 -4.18433994e-01 -7.84488022e-01
-4.65834528e-01 -1.22370076e+00 1.23770189e+00 6.42026365e-01
2.14958951e-01 -9.49481070e-01 3.36670801e-02 4.56361830e-01
-8.89474601e-02 6.93679750e-01 6.49872720e-01 -3.82399738e-01
-7.79668212e-01 -3.11800271e-01 -2.90682524e-01 5.16408384e-01
4.48965328e-03 7.64267519e-02 -7.13393331e-01 -1.08226880e-01
7.54792094e-02 -2.68528014e-01 9.34226871e-01 6.32559359e-01
1.72813630e+00 -2.46471941e-01 -3.51499766e-01 1.11284137e+00
1.65429735e+00 2.75066435e-01 4.75029737e-01 2.43628770e-01
9.39798713e-01 5.58217645e-01 8.46910179e-01 2.73225456e-01
1.22858666e-01 5.48148930e-01 4.14797038e-01 -3.93803686e-01
9.54618901e-02 -2.84110337e-01 -4.96084541e-02 4.10916746e-01
9.01205465e-02 -3.64725292e-02 -9.39992785e-01 5.41099429e-01
-1.81579423e+00 -6.03371501e-01 -2.55716652e-01 2.18168092e+00
8.67551029e-01 4.63333577e-01 2.76355326e-01 -1.72551312e-02
8.98827612e-01 3.19747865e-01 -9.26204801e-01 -2.20908463e-01
2.59524792e-01 2.24442333e-01 3.01741898e-01 6.63990796e-01
-1.20893359e+00 1.20156264e+00 4.25113058e+00 1.02423203e+00
-1.02791250e+00 1.56371802e-01 1.21619546e+00 -8.84457771e-03
-8.30041915e-02 3.70410942e-02 -7.66856134e-01 7.51073897e-01
2.20414400e-01 -8.15800279e-02 7.48726875e-02 7.77218819e-01
3.53602737e-01 -9.95742008e-02 -7.79243767e-01 6.37439132e-01
-1.20187476e-01 -1.38449156e+00 -3.06070119e-01 -1.67736948e-01
1.00087869e+00 -2.99973220e-01 2.40275543e-02 2.63101995e-01
-1.34106934e-01 -9.84642684e-01 6.84458971e-01 5.39276600e-01
9.64756310e-01 -8.84540439e-01 3.99952114e-01 4.82912064e-01
-1.38725424e+00 2.73091700e-02 -2.29203269e-01 1.20898582e-01
3.50561470e-01 6.17229581e-01 -4.95217681e-01 3.52746725e-01
5.55151939e-01 5.86116433e-01 -3.11590135e-01 1.22549736e+00
-2.87125379e-01 6.86930776e-01 -3.79816234e-01 8.18919539e-02
5.09971440e-01 -7.07108796e-01 5.23663819e-01 1.08556104e+00
-1.31327406e-01 2.41677731e-01 5.68457901e-01 1.36572385e+00
-2.66599387e-01 3.60393852e-01 -1.87889621e-01 2.95414627e-01
1.62213519e-01 1.53554285e+00 -1.45757771e+00 -3.43001157e-01
-9.52577069e-02 1.04115927e+00 4.47533458e-01 2.55116820e-01
-8.77466261e-01 -5.10041177e-01 9.58259851e-02 3.17171484e-01
3.59442830e-01 -2.73968410e-02 -6.31014824e-01 -8.11034918e-01
1.27356157e-01 -2.84531564e-01 3.39743942e-01 -4.19619501e-01
-1.00199842e+00 5.10677814e-01 1.28304884e-01 -1.17183769e+00
3.24139982e-01 -3.54254216e-01 -1.26049018e+00 7.26316154e-01
-1.54747355e+00 -9.25861478e-01 -5.28052032e-01 3.26894641e-01
9.16589856e-01 2.54262149e-01 1.48720071e-01 1.98772922e-02
-1.05061209e+00 3.37267697e-01 5.56455851e-02 2.19520852e-01
2.12242961e-01 -1.49615180e+00 2.49249071e-01 7.77617097e-01
-2.21239459e-02 3.42849851e-01 3.86803359e-01 -7.00934827e-01
-8.09289396e-01 -1.30899239e+00 6.31423146e-02 -2.28804171e-01
4.98753160e-01 -5.35472155e-01 -1.30460083e+00 4.73419189e-01
1.46783203e-01 4.67621893e-01 2.07477525e-01 -6.41665936e-01
2.08541110e-01 -2.21629664e-01 -1.26899207e+00 4.85606402e-01
9.97580469e-01 1.01481214e-01 -3.42035294e-01 4.68838066e-01
9.78123546e-01 -7.00090766e-01 -7.69481301e-01 5.63381612e-01
8.59375298e-02 -8.29227984e-01 7.42194951e-01 -2.96762079e-01
5.07139802e-01 -5.80319285e-01 5.18921077e-01 -1.01500273e+00
3.95100266e-02 -8.01655591e-01 -1.79879040e-01 1.19049680e+00
3.40975076e-01 -5.52845776e-01 1.09021664e+00 5.76574922e-01
-3.68198335e-01 -1.71192694e+00 -9.21416461e-01 -5.50229847e-01
1.83794856e-01 -2.29066402e-01 5.81151068e-01 6.17132366e-01
-4.69178408e-01 -2.79663913e-02 -2.66641472e-02 7.27965608e-02
9.39767778e-01 3.52741271e-01 6.57593131e-01 -1.10841644e+00
-2.55684227e-01 -5.81732750e-01 -1.08745046e-01 -1.12211776e+00
9.38246548e-02 -1.10145998e+00 3.75762135e-01 -1.72168899e+00
6.27977252e-02 -6.30605161e-01 -2.49690354e-01 2.83666492e-01
-5.34355998e-01 2.67917991e-01 3.32191102e-02 1.28643513e-01
-5.80691040e-01 7.16003776e-01 1.55005014e+00 -2.78017908e-01
-6.68342948e-01 1.69966802e-01 -5.11898458e-01 9.11953926e-01
9.55436468e-01 -4.77247208e-01 -3.92943859e-01 -1.61550239e-01
-2.65269965e-01 -3.99116352e-02 4.11859244e-01 -9.77139592e-01
2.13507354e-01 -1.46892145e-01 3.30366611e-01 -6.63481653e-01
2.89229691e-01 -7.55233407e-01 -3.03544641e-01 3.89088959e-01
-3.94585192e-01 -4.92772490e-01 9.93274599e-02 6.75011992e-01
-1.19357988e-01 -4.69294935e-01 1.16984582e+00 -3.12350512e-01
-5.80370009e-01 8.81037951e-01 2.60909915e-01 3.52444947e-01
1.41308463e+00 -5.46313643e-01 1.08319864e-01 1.87405556e-01
-6.70402586e-01 6.47329330e-01 5.61234295e-01 1.43848434e-02
7.78908134e-01 -1.05280948e+00 -6.40287519e-01 1.45171776e-01
2.08752081e-02 1.02109623e+00 4.51480627e-01 7.77071655e-01
-6.08557343e-01 -1.76143244e-01 1.93246379e-01 -9.44221735e-01
-8.54424119e-01 3.88438731e-01 7.92978704e-01 -2.12396652e-01
-1.05427611e+00 1.13314939e+00 3.76989841e-01 -4.26336169e-01
3.14045429e-01 -3.41309309e-01 -7.26954937e-02 -1.23056404e-01
3.02687764e-01 2.57914841e-01 -3.00498664e-01 -5.98495781e-01
-2.18540713e-01 7.99495637e-01 -1.73703656e-01 -1.27598256e-01
1.25731468e+00 1.23278052e-01 -6.97003156e-02 6.03273034e-01
1.34481490e+00 -2.35235959e-01 -1.88409352e+00 -1.30187035e-01
-3.42209972e-02 -4.59823310e-01 1.79333985e-01 -5.72757363e-01
-1.30704081e+00 8.81241620e-01 5.81096292e-01 2.74202842e-02
9.87518311e-01 1.59074873e-01 1.08883965e+00 -2.08408237e-01
1.00551024e-01 -1.13980782e+00 1.46839797e-01 6.35770783e-02
7.01383710e-01 -1.29620230e+00 -1.12475418e-01 -5.45438766e-01
-6.53293133e-01 8.44475150e-01 9.52499151e-01 -4.95491862e-01
5.93675435e-01 2.66569376e-01 3.50074135e-02 -3.77286494e-01
-1.90412179e-01 -2.22571969e-01 3.46799403e-01 2.95315325e-01
1.62462413e-01 -3.80641810e-04 -4.42361116e-01 5.87220788e-01
1.68815359e-01 -8.80525485e-02 1.17826842e-01 7.93107986e-01
-7.09533155e-01 -8.53917360e-01 -4.11755264e-01 5.58944643e-01
-3.21722090e-01 1.40137583e-01 -2.56171525e-01 7.13797688e-01
4.02336150e-01 4.49068695e-01 2.56066889e-01 1.86847314e-01
2.60254830e-01 9.36811939e-02 2.53594875e-01 -9.88421261e-01
-6.57699943e-01 8.52006525e-02 -6.03077948e-01 -4.02829081e-01
-2.72668213e-01 -5.44857025e-01 -1.90321529e+00 4.98643845e-01
-3.48378450e-01 1.24529712e-01 3.97586048e-01 8.78464580e-01
3.30674537e-02 5.60633302e-01 7.18898058e-01 -1.05603409e+00
-7.87440613e-02 -7.10259318e-01 -2.70115644e-01 3.49845856e-01
2.11227566e-01 -5.95759809e-01 -3.62913251e-01 5.53832352e-02] | [9.601234436035156, 0.27757441997528076] |
6a767f1e-30d1-484b-857a-d422a1f6c486 | duet-2d-structured-and-approximately | 2306.16058 | null | https://arxiv.org/abs/2306.16058v2 | https://arxiv.org/pdf/2306.16058v2.pdf | DUET: 2D Structured and Approximately Equivariant Representations | Multiview Self-Supervised Learning (MSSL) is based on learning invariances with respect to a set of input transformations. However, invariance partially or totally removes transformation-related information from the representations, which might harm performance for specific downstream tasks that require such information. We propose 2D strUctured and EquivarianT representations (coined DUET), which are 2d representations organized in a matrix structure, and equivariant with respect to transformations acting on the input data. DUET representations maintain information about an input transformation, while remaining semantically expressive. Compared to SimCLR (Chen et al., 2020) (unstructured and invariant) and ESSL (Dangovski et al., 2022) (unstructured and equivariant), the structured and equivariant nature of DUET representations enables controlled generation with lower reconstruction error, while controllability is not possible with SimCLR or ESSL. DUET also achieves higher accuracy for several discriminative tasks, and improves transfer learning. | ['Luca Zappella', 'Dan Busbridge', 'Jason Ramapuram', 'Chen Huang', 'Arno Blaas', 'T. Anderson Keller', 'Federico Danieli', 'Xavier Suau'] | 2023-06-28 | null | null | null | null | ['self-supervised-learning', 'transfer-learning'] | ['computer-vision', 'miscellaneous'] | [ 2.65855581e-01 3.44886541e-01 -4.01481390e-01 -4.69942510e-01
-4.64562088e-01 -1.02917314e+00 9.52506959e-01 1.14319600e-01
1.06169209e-01 5.27421415e-01 7.28662133e-01 -7.81130453e-04
-2.73959249e-01 -6.55733883e-01 -8.86367977e-01 -5.05444705e-01
6.57356950e-03 3.19775522e-01 -1.03645004e-01 -4.05619830e-01
-2.53869444e-01 6.41981900e-01 -1.60557926e+00 3.61549228e-01
5.70522368e-01 6.47414684e-01 1.93660066e-01 3.68828446e-01
1.63950771e-01 1.00300324e+00 9.69054997e-02 3.69525738e-02
4.19597238e-01 -4.17341650e-01 -6.66082740e-01 3.32784727e-02
6.14804804e-01 3.59651297e-02 -6.09975398e-01 8.70493233e-01
1.57639384e-01 3.32290262e-01 1.30332899e+00 -1.34884381e+00
-9.30367053e-01 4.31899935e-01 -3.01512271e-01 -2.53222466e-01
4.04191524e-01 -4.54667658e-02 1.43895066e+00 -1.12747526e+00
9.95039284e-01 1.41975129e+00 5.62713444e-01 6.65679991e-01
-1.69995975e+00 -5.69398999e-01 3.65608394e-01 5.20731695e-03
-1.26397538e+00 -6.16021276e-01 7.79472947e-01 -7.25526452e-01
8.57350349e-01 5.76518476e-01 5.92205703e-01 1.21807384e+00
4.79610056e-01 9.59402502e-01 9.89094913e-01 -2.63293415e-01
1.65341809e-01 -1.54934991e-02 8.50220099e-02 7.46054590e-01
2.67523438e-01 2.75072288e-02 -4.85578507e-01 1.46045819e-01
8.94808829e-01 2.08996654e-01 -1.43686950e-01 -1.14261436e+00
-1.45322073e+00 9.78194654e-01 6.32769346e-01 4.22107786e-01
-2.56851196e-01 -4.12771069e-02 4.33657199e-01 7.37169683e-01
2.71338671e-01 7.05845594e-01 -3.49717677e-01 3.20110023e-01
-5.74126720e-01 -4.66732122e-02 2.44836658e-01 1.27939117e+00
1.03443789e+00 5.95588982e-01 -3.65491807e-01 7.18489230e-01
1.17780477e-01 4.78764355e-01 8.90264332e-01 -9.50362027e-01
4.54075217e-01 8.50057840e-01 -3.09683651e-01 -9.81572807e-01
-6.17845833e-01 -4.71474081e-01 -1.19657576e+00 1.78907990e-01
7.28057250e-02 2.09001094e-01 -9.00909901e-01 2.20245910e+00
-5.91806173e-02 -4.31775957e-01 4.06031996e-01 5.66675127e-01
9.55287278e-01 6.77331507e-01 -5.00500798e-02 -1.67239487e-01
9.42463040e-01 -7.26564050e-01 -6.84849739e-01 -2.67831057e-01
7.33904779e-01 -4.32643086e-01 1.13834941e+00 -2.51829308e-02
-9.53607440e-01 -6.90818429e-01 -1.08182693e+00 -3.49584937e-01
-4.49917376e-01 -4.81128842e-02 7.24420249e-01 2.45605230e-01
-1.10645151e+00 5.09890854e-01 -8.54496658e-01 -3.69489878e-01
4.20801997e-01 3.89739722e-01 -1.01157808e+00 -2.40118336e-02
-1.16694057e+00 9.04632568e-01 2.76071161e-01 -4.19380337e-01
-1.03128469e+00 -8.55443299e-01 -1.43314588e+00 -2.30673570e-02
1.36165209e-02 -6.26897156e-01 8.88822377e-01 -1.00799596e+00
-1.41233921e+00 8.05875063e-01 -1.13152497e-01 -3.46701026e-01
6.08386397e-01 -8.72237831e-02 -2.50827879e-01 -5.15087089e-03
3.41962099e-01 9.78464067e-01 1.09360981e+00 -1.11023462e+00
8.63527060e-02 -4.67725039e-01 1.12402625e-01 4.08165455e-01
-5.97818792e-01 -5.10894716e-01 -6.94345683e-02 -8.29462469e-01
5.30955195e-01 -9.37136948e-01 -3.00184220e-01 1.19184365e-03
-4.10948306e-01 -1.27562642e-01 1.07123399e+00 -5.51852405e-01
7.83651114e-01 -2.17283392e+00 6.59055293e-01 6.85984120e-02
3.39491129e-01 7.04691261e-02 -4.53666687e-01 7.01313734e-01
-6.37264848e-01 2.46625580e-02 -1.81356564e-01 -2.54689921e-02
9.28319022e-02 3.53014201e-01 -4.80210185e-01 5.75379789e-01
1.16674632e-01 1.19942236e+00 -8.11998487e-01 -3.35754782e-01
4.37888086e-01 2.85419315e-01 -7.13650048e-01 1.84393629e-01
-1.46166652e-01 7.01096177e-01 -5.43734133e-01 4.90824312e-01
2.10075825e-01 -1.63752466e-01 2.09898323e-01 -5.76449156e-01
-6.85869530e-02 4.40604948e-02 -1.13970590e+00 1.65247750e+00
-7.25827277e-01 6.02680326e-01 -1.67584389e-01 -1.18885338e+00
1.06674576e+00 4.31634456e-01 6.52515471e-01 -7.04385340e-01
-1.06914170e-01 -6.67598248e-02 -2.13263735e-01 -1.20809115e-01
4.17653710e-01 -2.09484443e-01 -3.06454480e-01 4.03332502e-01
4.95408654e-01 -2.09753558e-01 8.38827938e-02 3.92245591e-01
9.65365469e-01 4.02540416e-01 7.20871568e-01 -5.18562257e-01
4.98756438e-01 -2.26048246e-01 7.08376467e-01 6.64656997e-01
6.76090196e-02 6.81032956e-01 3.72930676e-01 -5.35897255e-01
-1.10081851e+00 -1.40155065e+00 -1.16855644e-01 1.01418543e+00
4.55668867e-02 -6.23858333e-01 -2.82632530e-01 -5.91177762e-01
5.07520474e-02 7.58408844e-01 -7.81361222e-01 -5.52556574e-01
-3.63498420e-01 -2.02195838e-01 3.76609653e-01 6.79991901e-01
5.35473049e-01 -6.14729702e-01 -3.72258693e-01 1.90142337e-02
-9.63073969e-02 -9.11446393e-01 -6.71492279e-01 3.08416873e-01
-9.68639076e-01 -8.47425699e-01 -4.87027913e-01 -6.78154707e-01
9.40769255e-01 3.48379880e-01 8.35651457e-01 -7.14375138e-01
-2.49037057e-01 7.48895884e-01 -2.31058642e-01 -1.75278842e-01
-5.24530530e-01 1.69524718e-02 4.64030832e-01 2.39174128e-01
-4.27948385e-01 -7.67165363e-01 -1.70433089e-01 2.77596533e-01
-9.47635233e-01 4.29560781e-01 4.55837131e-01 1.07371724e+00
8.10932815e-01 -2.92056173e-01 5.50805807e-01 -1.00670195e+00
2.41020262e-01 -2.16374800e-01 -3.03314567e-01 2.75593191e-01
-5.77273369e-01 4.67585415e-01 9.74021435e-01 -4.70076472e-01
-1.02904344e+00 3.93853009e-01 2.58783191e-01 -9.78064179e-01
1.46394610e-01 1.84663221e-01 -4.00809854e-01 3.16385701e-02
9.37892437e-01 3.78870249e-01 2.29269117e-01 -5.02476513e-01
5.91573715e-01 1.29359409e-01 2.67545968e-01 -5.28908730e-01
1.16260374e+00 4.40993190e-01 2.88393587e-01 -7.08745718e-01
-9.10777986e-01 -1.94297448e-01 -1.12024283e+00 -1.47623137e-01
8.18444788e-01 -1.07636166e+00 -3.31648350e-01 4.63783666e-02
-6.73697948e-01 -1.19840190e-01 -8.65070999e-01 5.23315907e-01
-9.74812508e-01 1.93802819e-01 -3.33802700e-01 -3.63639981e-01
-2.95995474e-01 -1.02276444e+00 9.28533375e-01 -3.40601832e-01
-6.59418225e-01 -1.21876240e+00 -8.89864266e-02 6.65297136e-02
4.05673474e-01 5.65480769e-01 1.25894380e+00 -7.45114207e-01
-2.73231745e-01 -3.32961500e-01 1.77981392e-01 5.12737334e-01
5.54094613e-01 -3.13566595e-01 -8.90082300e-01 -6.74794078e-01
-2.24683017e-01 -6.12425506e-01 8.71068954e-01 1.45288706e-01
1.11578202e+00 -6.34705603e-01 -1.72652915e-01 7.27153122e-01
1.16060448e+00 5.41672809e-03 3.23351145e-01 1.55418381e-01
9.41679001e-01 7.45642483e-01 3.14127028e-01 1.02138288e-01
2.45078772e-01 5.71006656e-01 3.93364757e-01 -2.13328853e-01
-3.02160323e-01 -7.69155622e-01 7.06432462e-01 1.02608240e+00
-8.24948698e-02 2.28120327e-01 -6.44612491e-01 3.42599273e-01
-1.87561774e+00 -1.02704668e+00 -3.16907205e-02 1.98481917e+00
6.51741207e-01 -2.07698911e-01 -2.07846954e-01 1.88827757e-02
6.83773339e-01 5.42969763e-01 -9.04394269e-01 -4.28228796e-01
-4.53395486e-01 7.11971223e-02 3.61218542e-01 4.84102041e-01
-1.06477833e+00 1.07704663e+00 6.57088470e+00 7.28433788e-01
-9.55599546e-01 -4.69316132e-02 3.14963996e-01 2.55037159e-01
-6.96370304e-01 -8.62000789e-03 -4.41441238e-01 -2.22270593e-01
5.24671316e-01 -4.52014923e-01 3.24226916e-01 8.55339229e-01
-1.09308571e-01 6.60714209e-01 -1.61446583e+00 8.22123826e-01
1.69700384e-01 -1.51398659e+00 7.74932921e-01 -1.41242012e-01
9.38234091e-01 -1.80566698e-01 3.37653071e-01 5.38171470e-01
5.40632427e-01 -9.66944516e-01 8.00348938e-01 5.57406425e-01
1.02387035e+00 -5.25953650e-01 1.95974752e-01 3.48942369e-01
-1.23741102e+00 4.50852811e-02 -4.97749418e-01 1.48788951e-02
-3.17997754e-01 2.90022731e-01 -6.63069665e-01 7.52269328e-01
4.17787582e-01 1.48721457e+00 -6.93616152e-01 9.14521962e-02
-3.61567944e-01 4.34168339e-01 5.95000349e-02 2.32499823e-01
1.17916763e-01 -4.09252793e-01 7.89038479e-01 9.43647504e-01
1.97582781e-01 -2.25137040e-01 2.25494906e-01 8.68003607e-01
-1.89347342e-01 1.52360797e-01 -1.26674223e+00 -8.75372142e-02
1.52508765e-01 1.27657664e+00 -3.49358380e-01 -2.88960725e-01
-2.41992489e-01 8.87147546e-01 3.59811306e-01 4.41800386e-01
-2.81750798e-01 -3.49552818e-02 7.91314781e-01 -2.16087885e-02
1.72493652e-01 -2.73252517e-01 -1.54679328e-01 -1.54652822e+00
-1.17054284e-01 -9.23244059e-01 4.74931508e-01 -6.88179553e-01
-1.25659120e+00 4.25388694e-01 1.74335167e-01 -1.83040106e+00
-4.51896131e-01 -6.34965241e-01 -3.53214234e-01 4.60096180e-01
-1.03756046e+00 -1.61604106e+00 -6.01159297e-02 9.74090576e-01
6.52162850e-01 -6.05544209e-01 1.16813278e+00 -2.84327298e-01
-3.54580462e-01 6.50930703e-01 2.23089918e-01 6.91875070e-02
6.02009773e-01 -1.18107033e+00 2.64560521e-01 5.76804399e-01
3.76016200e-01 7.49369621e-01 5.78907669e-01 -4.72115397e-01
-1.94781554e+00 -1.63048804e+00 5.60606837e-01 -5.86718619e-01
6.21288359e-01 -6.09873474e-01 -7.23966956e-01 1.28708243e+00
1.02729648e-02 1.90523013e-01 7.16535330e-01 1.05440408e-01
-9.99536991e-01 -2.70841271e-01 -1.07392502e+00 8.42067122e-01
1.46237087e+00 -7.28581309e-01 -6.85973942e-01 4.41817313e-01
8.75563025e-01 -3.39892238e-01 -1.08813143e+00 5.64283907e-01
6.38695240e-01 -7.81123459e-01 1.05986118e+00 -1.01151407e+00
3.00483644e-01 -1.69503987e-01 -6.73570871e-01 -1.59925771e+00
-8.18499625e-01 -4.47688788e-01 1.17525786e-01 1.17432117e+00
3.32639128e-01 -8.16957772e-01 2.66717762e-01 3.03409159e-01
-3.79027635e-01 -3.47645223e-01 -8.34116220e-01 -1.24885261e+00
3.43953997e-01 -2.41239294e-01 5.17084897e-01 1.32607818e+00
-6.71809018e-02 6.17609918e-01 -4.42206681e-01 -1.34865819e-02
5.48496366e-01 4.16112810e-01 7.02061057e-01 -1.31880605e+00
3.38823237e-02 -1.53567553e-01 -7.90081978e-01 -5.83861053e-01
6.55872762e-01 -1.71167624e+00 -4.90495056e-01 -1.55557585e+00
2.04308912e-01 -1.26590967e-01 -2.04868793e-01 9.51844513e-01
3.45134914e-01 1.35664910e-01 2.65367538e-01 5.51074326e-01
-5.10706186e-01 9.35314834e-01 1.41388249e+00 -3.56126040e-01
-1.13514632e-01 -2.15902239e-01 -7.20699430e-01 7.46204972e-01
7.88556159e-01 -4.08626050e-02 -7.34180331e-01 -2.92272806e-01
-1.37808755e-01 -1.06123239e-01 2.11312264e-01 -7.98792899e-01
-1.64616611e-02 -2.88246930e-01 6.82857156e-01 -2.52262294e-01
4.47646469e-01 -7.31414199e-01 3.57328355e-01 4.88542974e-01
-7.11902499e-01 2.00233847e-01 9.99467373e-02 5.58171034e-01
-3.48436058e-01 6.52135313e-02 9.08307910e-01 -1.66973650e-01
-6.55095577e-01 4.54271525e-01 -4.70222890e-01 1.03804097e-01
1.04987752e+00 -2.53889799e-01 -1.51246935e-01 -6.30195260e-01
-8.87342691e-01 9.62746143e-02 4.09764826e-01 7.72614598e-01
7.06603944e-01 -1.80823052e+00 -6.32762194e-01 6.13100588e-01
4.51049387e-01 -6.83609545e-02 1.73570916e-01 4.30050999e-01
-7.72696957e-02 5.23005188e-01 -6.08298540e-01 -5.64138532e-01
-1.12415564e+00 6.66142762e-01 1.83921397e-01 -1.79016769e-01
-9.51728463e-01 3.73461604e-01 7.83140421e-01 -8.15941393e-01
-7.04733655e-02 -1.05615869e-01 -4.25974965e-01 2.95864284e-01
1.88340396e-01 3.39661390e-01 -1.99553408e-02 -9.28036094e-01
-1.93627238e-01 6.58986866e-01 -1.28632694e-01 -7.41366446e-02
1.42488039e+00 2.44049802e-01 -7.87664950e-02 8.07475209e-01
1.48228681e+00 -2.15165541e-01 -1.17187643e+00 -4.41181988e-01
-1.31578460e-01 -1.98358610e-01 1.27951242e-02 -1.95100114e-01
-9.15922999e-01 7.77559578e-01 4.17307496e-01 -3.84883992e-02
8.88650596e-01 -1.35886911e-02 1.56139746e-01 6.77066863e-01
3.77705812e-01 -8.69196475e-01 3.43338162e-01 5.73162019e-01
1.49052179e+00 -9.67152894e-01 -3.23205739e-02 -3.67091864e-01
-8.09006691e-01 1.06426799e+00 6.91087723e-01 1.16106439e-02
5.08413672e-01 -1.87059250e-02 -2.42082879e-01 3.55370231e-02
-9.00942028e-01 -8.38600919e-02 5.40496588e-01 9.09926116e-01
3.26757729e-01 2.03837082e-01 8.22068602e-02 2.25313857e-01
-4.20691431e-01 -6.79100275e-01 3.34942102e-01 8.55540574e-01
-2.75834233e-01 -1.10955656e+00 -1.75237194e-01 6.19655132e-01
2.57240534e-01 1.40881494e-01 -5.46037376e-01 9.81091022e-01
-2.30701789e-01 6.78814590e-01 9.35975462e-02 -5.12656212e-01
6.84493184e-01 2.30139807e-01 4.94112819e-01 -8.44153762e-01
-3.05949837e-01 -6.58749938e-02 2.13718992e-02 -6.22793734e-01
-3.21514547e-01 -1.02507055e+00 -1.25538528e+00 1.93088979e-01
1.43122718e-01 -1.12616289e-02 1.26736626e-01 7.56595790e-01
4.16453570e-01 4.78923053e-01 9.85981345e-01 -7.09119260e-01
-6.03700578e-01 -8.65812957e-01 -6.19162142e-01 7.86098659e-01
4.82802033e-01 -8.33787084e-01 -3.26337188e-01 2.28651762e-01] | [9.050681114196777, 2.7041268348693848] |
8825e9b8-0e75-4e1d-85ec-ef8452f3eae5 | magneto-an-efficient-deep-learning-method-for-1 | 2011.04349 | null | https://arxiv.org/abs/2011.04349v1 | https://arxiv.org/pdf/2011.04349v1.pdf | MAGNeto: An Efficient Deep Learning Method for the Extractive Tags Summarization Problem | In this work, we study a new image annotation task named Extractive Tags Summarization (ETS). The goal is to extract important tags from the context lying in an image and its corresponding tags. We adjust some state-of-the-art deep learning models to utilize both visual and textual information. Our proposed solution consists of different widely used blocks like convolutional and self-attention layers, together with a novel idea of combining auxiliary loss functions and the gating mechanism to glue and elevate these fundamental components and form a unified architecture. Besides, we introduce a loss function that aims to reduce the imbalance of the training data and a simple but effective data augmentation technique dedicated to alleviates the effect of outliers on the final results. Last but not least, we explore an unsupervised pre-training strategy to further boost the performance of the model by making use of the abundant amount of available unlabeled data. Our model shows the good results as 90% $F_\text{1}$ score on the public NUS-WIDE benchmark, and 50% $F_\text{1}$ score on a noisy large-scale real-world private dataset. Source code for reproducing the experiments is publicly available at: https://github.com/pixta-dev/labteam | ['Ngoc C. Lê', 'Trung Thanh Tran', 'Giang Nam Ngo', 'Lam Thanh Do', 'Tung Dinh Nguyen', 'Anh Tuan Vu', 'Hieu Trong Phung'] | 2020-11-09 | magneto-an-efficient-deep-learning-method-for | https://arxiv.org/abs/2011.04349 | https://arxiv.org/pdf/2011.04349 | null | ['extractive-tags-summarization'] | ['natural-language-processing'] | [ 1.63234979e-01 5.83481900e-02 -3.10135335e-02 -5.94974339e-01
-1.03387308e+00 -2.37556607e-01 4.64118391e-01 2.24473789e-01
-7.53492773e-01 6.01136684e-01 3.39112252e-01 9.05173272e-02
2.56009281e-01 -5.08010209e-01 -8.61177027e-01 -7.67441094e-01
2.44546443e-01 -4.14080210e-02 3.22937936e-01 -4.10813652e-02
9.05081779e-02 7.38833472e-02 -1.43511796e+00 2.18369678e-01
1.08551800e+00 1.24264967e+00 3.89323562e-01 2.20081672e-01
-2.04721943e-01 6.20720029e-01 -4.47910458e-01 -4.29652929e-01
1.08366080e-01 -3.24150056e-01 -5.43519914e-01 2.98294485e-01
3.63459438e-01 -9.68580246e-02 -3.99150103e-01 1.18404317e+00
7.17818558e-01 1.64255723e-01 4.31932181e-01 -8.39043975e-01
-6.18933797e-01 5.06471932e-01 -8.24022174e-01 1.92858130e-01
-1.72463864e-01 1.93617478e-01 1.01132309e+00 -9.26711380e-01
3.53132278e-01 7.61301935e-01 4.30916578e-01 5.02894461e-01
-8.93403590e-01 -6.50856733e-01 4.11450326e-01 1.59411728e-01
-1.30140197e+00 -3.97867650e-01 7.45278358e-01 -2.69843131e-01
4.27906394e-01 1.84113398e-01 3.66331220e-01 1.21433985e+00
-3.57583612e-02 1.07630706e+00 9.39421594e-01 -3.92869979e-01
-6.69483170e-02 5.85661791e-02 2.61351198e-01 9.56903577e-01
1.40281171e-01 -4.63467777e-01 -3.68897766e-01 1.08396120e-01
4.02859658e-01 2.01644063e-01 -3.89737725e-01 -2.14227110e-01
-1.17340231e+00 6.27259731e-01 7.12445021e-01 3.78363609e-01
-2.66482234e-01 1.78860158e-01 4.65099454e-01 -2.58815438e-01
7.68209279e-01 2.78053939e-01 -5.13542295e-01 3.05392668e-02
-8.82183731e-01 -1.19021907e-01 2.01678813e-01 8.18514168e-01
8.21409881e-01 4.62589040e-02 -5.81298947e-01 1.04353583e+00
2.04786286e-01 2.96586841e-01 6.00563705e-01 -7.49188721e-01
5.59145153e-01 7.18067527e-01 -3.46152261e-02 -7.65796602e-01
-2.93204248e-01 -9.71847594e-01 -1.03490889e+00 -1.90262988e-01
-9.33144093e-02 -1.93469435e-01 -1.31891143e+00 1.88506174e+00
1.68991759e-01 2.73764074e-01 -1.92779168e-01 8.45412791e-01
1.03514063e+00 8.17531526e-01 2.33601764e-01 -1.17865928e-01
1.41914427e+00 -1.25268435e+00 -6.06219530e-01 -3.74480098e-01
4.55634981e-01 -6.95455611e-01 1.30518866e+00 1.22080214e-01
-1.00016522e+00 -7.09694684e-01 -1.12257314e+00 -3.45854461e-01
-3.03961277e-01 6.58852518e-01 4.84371543e-01 8.29088017e-02
-8.47323716e-01 4.02245104e-01 -8.59114945e-01 -2.45316610e-01
6.48968220e-01 2.93224394e-01 -3.93802583e-01 -9.64026153e-02
-9.53644395e-01 4.48403776e-01 4.62451935e-01 3.30337912e-01
-7.10653484e-01 -4.19449329e-01 -9.10562694e-01 3.02315384e-01
7.39070237e-01 -7.23017454e-01 1.01411402e+00 -9.81351435e-01
-1.24815369e+00 9.65051055e-01 -1.47550091e-01 -2.55403012e-01
4.15978223e-01 -4.05361354e-01 -1.94553986e-01 -2.83709690e-02
2.73394465e-01 7.13783264e-01 6.02472544e-01 -1.29151917e+00
-6.50306344e-01 -4.69920486e-01 -2.16166377e-01 1.55455589e-01
-7.39466369e-01 -2.47189432e-01 -9.67133820e-01 -8.43927264e-01
4.55339886e-02 -7.89279938e-01 -3.00376624e-01 -3.37174565e-01
-5.59271336e-01 -1.97472200e-01 6.09665930e-01 -6.39800012e-01
1.24394774e+00 -2.29540324e+00 7.39222541e-02 -1.55219451e-01
3.24205756e-01 6.10696375e-01 -2.61762679e-01 2.81791002e-01
3.50290090e-02 1.49892852e-01 -3.82823944e-01 -6.59240305e-01
9.62572359e-03 -1.96034983e-02 -1.72541127e-01 2.61653990e-01
1.91317588e-01 8.64752650e-01 -6.64294243e-01 -3.97653610e-01
8.88896734e-02 3.56362402e-01 -3.96421313e-01 2.18833283e-01
-3.26958686e-01 4.77725685e-01 -6.30593359e-01 3.73070270e-01
7.27639794e-01 -3.79295588e-01 -1.54218629e-01 -2.59583771e-01
-1.78829476e-01 1.12337545e-01 -9.81809437e-01 1.96157587e+00
-2.34297559e-01 3.37745875e-01 1.28315151e-01 -1.16032088e+00
8.63932967e-01 5.92693947e-02 5.12756467e-01 -7.22086608e-01
4.53193218e-01 8.54591578e-02 -3.31530511e-01 -5.29834270e-01
5.19123852e-01 2.48573124e-01 -2.72497416e-01 1.53714448e-01
2.96232253e-01 3.24324697e-01 3.03071529e-01 3.84510458e-01
1.00498962e+00 1.35441333e-01 1.99652597e-01 -1.63806230e-01
5.92385828e-01 -1.18015669e-01 8.46562922e-01 4.97471780e-01
-5.61971334e-04 7.15098619e-01 5.32644451e-01 -3.20716798e-01
-8.78341377e-01 -7.07579136e-01 7.74099156e-02 1.15781748e+00
1.89539358e-01 -5.31435192e-01 -7.32736111e-01 -8.99700403e-01
-2.59896696e-01 5.43981731e-01 -7.99193382e-01 -2.52281964e-01
-3.60170037e-01 -1.02380836e+00 3.18583816e-01 5.91858268e-01
7.86624134e-01 -1.00591910e+00 -2.02332258e-01 -1.94036718e-02
-4.69210744e-01 -1.16304219e+00 -6.02119386e-01 2.22036257e-01
-5.84549427e-01 -7.68992126e-01 -7.65201509e-01 -7.70074248e-01
7.17942476e-01 2.30644807e-01 8.63931596e-01 8.23212788e-02
-1.01687767e-01 -7.81926885e-02 -4.86000389e-01 -5.28836131e-01
3.60605679e-02 4.28172201e-01 -4.54576701e-01 2.92813927e-01
2.13938415e-01 -4.78426635e-01 -7.66669631e-01 1.45684496e-01
-9.74484861e-01 3.53566319e-01 8.32681775e-01 8.02295089e-01
8.25917423e-01 -2.63940513e-01 5.23518503e-01 -8.73004138e-01
3.30613673e-01 -3.82663846e-01 -4.74229425e-01 9.51378196e-02
-4.40214723e-01 1.12955347e-01 6.67462468e-01 -1.66651279e-01
-9.32780802e-01 1.23484530e-01 -3.71330827e-01 -3.98663878e-01
-1.58916727e-01 3.83344501e-01 -3.95981818e-01 3.58089954e-01
3.41892928e-01 4.24125612e-01 -3.00427318e-01 -7.77989209e-01
4.05627549e-01 7.36841559e-01 7.09902942e-01 -3.58380347e-01
5.90487599e-01 5.26769102e-01 -2.46072471e-01 -3.88777524e-01
-1.33156538e+00 -6.37208581e-01 -5.03184676e-01 4.22531851e-02
8.88615072e-01 -1.05449224e+00 -5.83675206e-01 5.24451196e-01
-9.58570957e-01 -1.66836545e-01 -2.97492594e-01 4.23700213e-01
-2.80440569e-01 5.22225797e-01 -4.01343793e-01 -4.17420536e-01
-7.12241590e-01 -1.14991212e+00 1.18035889e+00 4.49127406e-01
3.41259837e-01 -4.70080733e-01 -7.86055326e-02 5.81292152e-01
2.58018732e-01 3.50361854e-01 7.41464317e-01 -7.98088193e-01
-5.61111867e-01 -2.13182002e-01 -4.62819993e-01 7.52021372e-01
6.10867999e-02 -8.73556882e-02 -1.00317764e+00 -3.07810068e-01
-1.41851515e-01 -2.59567142e-01 1.63523424e+00 3.10584992e-01
1.51997185e+00 -2.09238067e-01 -3.38122785e-01 6.06129050e-01
1.44656384e+00 4.82141711e-02 6.30890965e-01 2.72297919e-01
9.36983168e-01 4.20814216e-01 5.73380947e-01 4.39912200e-01
5.02578735e-01 6.34620428e-01 5.57597876e-01 -3.39291781e-01
-1.68645307e-01 -1.73017591e-01 1.96098149e-01 8.64740968e-01
-2.58768313e-02 -6.10862136e-01 -7.83653915e-01 6.20669544e-01
-2.13373804e+00 -7.66607881e-01 -1.13507167e-01 2.13555002e+00
7.13442743e-01 1.70076966e-01 -1.44222334e-01 -2.15584680e-01
8.84752154e-01 4.32130724e-01 -4.18328106e-01 5.25221415e-02
-1.10780187e-01 1.25553519e-01 4.41741377e-01 2.47500047e-01
-1.44601476e+00 1.06216609e+00 4.53394842e+00 1.07273853e+00
-1.21091187e+00 2.38464460e-01 9.36750770e-01 -2.51221955e-01
-8.74850526e-02 -2.67818570e-01 -6.92901611e-01 7.43606508e-01
6.63675249e-01 1.29553556e-01 -1.20913401e-01 7.23713279e-01
3.35462421e-01 -1.41388342e-01 -6.51692927e-01 7.97256827e-01
2.29926437e-01 -1.27044487e+00 9.04946029e-02 5.56793958e-02
7.32107043e-01 2.88747221e-01 1.07067019e-01 3.07803929e-01
1.64941773e-01 -7.18602777e-01 8.07608664e-01 5.71990728e-01
7.29410827e-01 -7.54301548e-01 1.08442807e+00 2.40289196e-01
-1.13835669e+00 -1.10190570e-01 -3.42475623e-01 2.22004622e-01
-1.66823287e-02 7.64391184e-01 -4.34575826e-01 8.65708470e-01
7.18374133e-01 7.71957695e-01 -7.55458176e-01 1.04342055e+00
-3.61355692e-01 5.91492176e-01 -1.49044499e-01 2.13430241e-01
5.13285160e-01 -2.29214877e-02 3.58933777e-01 1.23373890e+00
3.72669548e-01 1.19179755e-01 3.35754663e-01 5.84676564e-01
-5.28750479e-01 2.76714146e-01 -2.39096165e-01 1.34026915e-01
1.44763887e-01 1.54731679e+00 -6.34375334e-01 -4.61147308e-01
-4.29388195e-01 1.03678203e+00 4.41361248e-01 2.85659224e-01
-9.34065223e-01 -3.65472078e-01 3.45458865e-01 1.40104517e-01
5.77053368e-01 -1.02940790e-01 -2.08125934e-01 -1.30757558e+00
2.55249888e-01 -5.61491847e-01 4.80470538e-01 -7.66782582e-01
-9.75898564e-01 7.35769808e-01 -2.14367062e-01 -1.23343694e+00
2.98204403e-02 -2.23612666e-01 -6.04183733e-01 6.67622745e-01
-1.60928261e+00 -1.27118063e+00 -6.80941522e-01 4.83795404e-01
6.22286797e-01 -1.60441250e-01 5.01544118e-01 4.41153705e-01
-9.14524376e-01 5.53366840e-01 3.59318823e-01 2.91176468e-01
9.43887770e-01 -1.09317350e+00 3.63099247e-01 1.08917356e+00
1.46785676e-01 3.47414255e-01 4.53387260e-01 -4.19380963e-01
-8.09975445e-01 -1.32787383e+00 7.09466100e-01 -9.47863832e-02
3.69020671e-01 -4.96358246e-01 -9.49521840e-01 6.84466541e-01
4.48922396e-01 2.55994827e-01 4.62754935e-01 -7.84335881e-02
-2.21182689e-01 -3.41765374e-01 -8.79050672e-01 4.10153776e-01
1.00147521e+00 -7.71945193e-02 -4.13400233e-01 2.51423538e-01
7.87765205e-01 -3.01641434e-01 -5.04577041e-01 6.28165722e-01
1.32684708e-01 -8.54972720e-01 6.80880725e-01 -3.51296276e-01
5.81052423e-01 -3.97017390e-01 -1.60181522e-02 -1.24531901e+00
-4.67041016e-01 -4.43276227e-01 1.88527301e-01 1.56659770e+00
2.98329204e-01 -3.48387629e-01 6.86535954e-01 1.18270248e-01
-6.38502538e-01 -1.09818125e+00 -6.93573833e-01 -3.77711147e-01
-2.45497540e-01 -1.30039632e-01 2.83131272e-01 8.09559643e-01
-3.58390808e-01 5.86814821e-01 -4.81759638e-01 7.99945518e-02
5.01875162e-01 1.47710918e-02 6.85241520e-01 -1.07960761e+00
2.89188437e-02 -2.30808377e-01 -2.88759083e-01 -9.95862484e-01
1.54562607e-01 -7.87369370e-01 6.87003881e-02 -1.77408421e+00
5.59372962e-01 -3.34971875e-01 -7.25312352e-01 8.22157621e-01
-4.63143736e-01 4.13526118e-01 2.94293284e-01 1.94666281e-01
-1.03305805e+00 9.41634059e-01 1.10808766e+00 -2.54337430e-01
-2.04322319e-02 -1.09022874e-02 -1.00644565e+00 7.28843987e-01
7.82406867e-01 -4.82800901e-01 -3.31155360e-01 -7.14564741e-01
-1.26527622e-01 -3.04090798e-01 3.30902368e-01 -1.00835621e+00
1.62025616e-01 1.04483098e-01 3.71208400e-01 -5.97166598e-01
3.21796983e-01 -6.44321799e-01 -2.11619735e-01 2.24208266e-01
-3.65328431e-01 -1.37475412e-02 2.85757780e-01 5.57103992e-01
-2.97771931e-01 -1.37928665e-01 9.36965942e-01 -1.89319372e-01
-8.29278469e-01 5.04771471e-01 5.42859137e-02 4.25306782e-02
9.85588670e-01 1.43251210e-01 -5.18567741e-01 -2.54981846e-01
-5.88373065e-01 4.32483733e-01 4.41989630e-01 3.72468740e-01
3.25933397e-01 -1.21784639e+00 -8.48874152e-01 7.32042566e-02
3.48937988e-01 1.12874269e-01 5.57892919e-01 8.59308481e-01
-2.63116181e-01 3.06235790e-01 -1.58245355e-01 -4.71620351e-01
-1.20690894e+00 5.13602018e-01 1.88310176e-01 -3.75870258e-01
-5.34446657e-01 9.29630220e-01 4.52862948e-01 -5.62864169e-02
2.90189713e-01 -3.00364792e-01 -1.95089012e-01 1.87903836e-01
5.02183139e-01 5.07351235e-02 2.18562633e-01 -6.64123952e-01
-3.68943930e-01 4.51649427e-01 -3.47268075e-01 1.14962250e-01
1.55572879e+00 -2.65132338e-01 -1.47277955e-02 2.56814420e-01
1.28402960e+00 4.05213870e-02 -1.32655025e+00 -5.13504803e-01
-2.93438911e-01 -2.28716731e-01 1.44953534e-01 -8.08775902e-01
-1.32918096e+00 9.03399169e-01 7.07750380e-01 -3.33569869e-02
1.35247970e+00 1.63615987e-01 9.09870028e-01 1.47053152e-01
-1.30214572e-01 -1.07130897e+00 2.36940861e-01 3.83284897e-01
7.35122263e-01 -1.32308650e+00 3.70659716e-02 -3.01871806e-01
-7.15559244e-01 6.07352078e-01 6.07311606e-01 -6.47090822e-02
2.93973505e-01 1.23323143e-01 6.69729114e-02 -1.14982560e-01
-6.51535213e-01 -5.03615916e-01 4.34510678e-01 1.29890203e-01
3.59956890e-01 -1.57925040e-01 -5.81991374e-01 9.07492876e-01
1.96975201e-01 -7.55578578e-02 3.14654797e-01 6.38898492e-01
-5.57064235e-01 -9.66466188e-01 5.80209889e-04 4.23659235e-01
-8.82572591e-01 -1.32427305e-01 -1.90972656e-01 6.85669601e-01
3.80880833e-01 6.99421167e-01 -2.43725721e-02 -4.13397610e-01
4.09065425e-01 1.60438418e-02 8.62421393e-02 -6.86877072e-01
-5.15765786e-01 3.01286191e-01 -9.84069929e-02 -5.37824869e-01
-5.46230376e-01 -4.23151344e-01 -1.21390855e+00 -1.08438313e-01
-2.18912706e-01 1.12893574e-01 5.98679602e-01 8.88679922e-01
5.72408974e-01 7.71450758e-01 6.51021540e-01 -8.05649102e-01
-2.36194402e-01 -1.18355775e+00 -3.99523020e-01 5.34955382e-01
1.67733610e-01 -4.72592026e-01 -9.66526940e-02 8.13092366e-02] | [9.76876449584961, 0.35662877559661865] |
938d876a-127c-4daa-8ba9-0f0ae2a9096d | analysis-of-hydrological-and-suspended | 1911.12466 | null | https://arxiv.org/abs/1911.12466v2 | https://arxiv.org/pdf/1911.12466v2.pdf | Analysis of Hydrological and Suspended Sediment Events from Mad River Watershed using Multivariate Time Series Clustering | Hydrological storm events are a primary driver for transporting water quality constituents such as turbidity, suspended sediments and nutrients. Analyzing the concentration (C) of these water quality constituents in response to increased streamflow discharge (Q), particularly when monitored at high temporal resolution during a hydrological event, helps to characterize the dynamics and flux of such constituents. A conventional approach to storm event analysis is to reduce the C-Q time series to two-dimensional (2-D) hysteresis loops and analyze these 2-D patterns. While effective and informative to some extent, this hysteresis loop approach has limitations because projecting the C-Q time series onto a 2-D plane obscures detail (e.g., temporal variation) associated with the C-Q relationships. In this paper, we address this issue using a multivariate time series clustering approach. Clustering is applied to sequences of river discharge and suspended sediment data (acquired through turbidity-based monitoring) from six watersheds located in the Lake Champlain Basin in the northeastern United States. While clusters of the hydrological storm events using the multivariate time series approach were found to be correlated to 2-D hysteresis loop classifications and watershed locations, the clusters differed from the 2-D hysteresis classifications. Additionally, using available meteorological data associated with storm events, we examine the characteristics of computational clusters of storm events in the study watersheds and identify the features driving the clustering approach. | ['Byung Suk Lee', 'Ali Javed', 'Scott D. Hamshaw', 'Donna M. Rizzo'] | 2019-11-28 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [-7.05160871e-02 -5.74795425e-01 1.77600220e-01 -3.11698675e-01
-3.24407727e-01 -8.79218221e-01 5.55027187e-01 6.96328163e-01
-2.04212397e-01 7.55266786e-01 5.13062656e-01 -6.31419957e-01
-7.19970405e-01 -1.10805643e+00 -2.12524757e-01 -9.83613670e-01
-8.07778776e-01 2.46929064e-01 -3.76208201e-02 -4.46085006e-01
2.81648546e-01 1.00341868e+00 -1.54715228e+00 3.96355279e-02
9.21881855e-01 2.08859116e-01 1.54065639e-01 6.63227618e-01
-5.28236151e-01 -7.64867365e-02 -5.86172998e-01 6.92006946e-01
2.79837132e-01 -6.09731615e-01 -2.68811285e-01 1.36122763e-01
1.17829321e-02 1.16773974e-02 5.26134633e-02 7.65548289e-01
2.30122253e-01 1.98783800e-01 7.51063049e-01 -9.02352989e-01
1.06666647e-01 4.84875858e-01 -6.95968866e-01 5.74991465e-01
-5.02196029e-02 1.77836642e-01 8.21556211e-01 -5.82237542e-01
4.82696742e-01 1.61045265e+00 6.85363889e-01 -3.37397039e-01
-1.71743035e+00 -6.75984085e-01 1.55483261e-01 -3.40456158e-01
-1.34074879e+00 -2.25719124e-01 3.07681203e-01 -9.29229438e-01
8.68006825e-01 2.24580690e-01 1.12947142e+00 2.57267147e-01
6.82994545e-01 1.43270031e-01 1.14596319e+00 -5.79358824e-02
4.29962337e-01 -4.10941899e-01 2.51336813e-01 -3.38711590e-01
6.03221834e-01 5.23753285e-01 -1.19598173e-01 -5.86535394e-01
5.03616869e-01 4.39526290e-01 -3.21859658e-01 2.52142131e-01
-6.77766562e-01 1.08704543e+00 4.17387664e-01 3.54059905e-01
-5.25619090e-01 -2.23139688e-01 1.74865514e-01 4.38079417e-01
5.84399581e-01 4.00629759e-01 -6.11349523e-01 -1.81579739e-01
-1.16047263e+00 2.05494702e-01 8.30502689e-01 4.00076210e-01
9.20813560e-01 8.54969025e-02 4.04619902e-01 2.88636476e-01
6.99757993e-01 1.15724015e+00 1.10985979e-01 -4.22736317e-01
2.42246866e-01 5.65973818e-01 2.82010853e-01 -9.40146208e-01
-9.12767887e-01 -1.29091829e-01 -7.11720943e-01 1.99058264e-01
5.61650813e-01 -4.55532551e-01 -7.53394008e-01 1.27990270e+00
9.49326456e-02 -4.77242395e-02 3.12910885e-01 7.37223506e-01
4.55656886e-01 9.84792173e-01 3.58294487e-01 -6.11874223e-01
1.28871632e+00 3.33998203e-01 -7.49003172e-01 2.50365525e-01
4.62050259e-01 -5.23304462e-01 7.68958688e-01 -2.97723532e-01
-5.49409568e-01 -2.68748015e-01 -4.73317862e-01 1.04937613e+00
-4.69965428e-01 -4.77962822e-01 2.64722466e-01 5.00041664e-01
-6.67428315e-01 7.03605533e-01 -1.30988812e+00 -8.75006855e-01
-6.00018445e-03 8.04539919e-02 9.90238786e-03 5.71163893e-01
-1.26644623e+00 6.24405622e-01 2.01024845e-01 4.59449977e-01
-4.58944172e-01 -7.66318262e-01 -6.52304769e-01 1.36813000e-01
-5.35170138e-01 1.93492249e-01 6.83752120e-01 -3.29864562e-01
-9.31463540e-01 3.00861806e-01 -5.00550151e-01 -9.16144177e-02
2.12101251e-01 9.90633890e-02 -7.65331924e-01 9.78172868e-02
4.60102886e-01 2.20348127e-02 4.30716395e-01 -1.36085486e+00
-1.00294089e+00 -7.03045607e-01 -5.76809645e-01 3.37002007e-03
-1.51143432e-01 2.30906177e-02 5.13225913e-01 -5.25073826e-01
6.32394731e-01 -9.72244501e-01 -6.40190244e-01 -4.42647874e-01
1.25598907e-01 3.37974839e-02 1.11096001e+00 -2.15146214e-01
1.27583873e+00 -2.42811346e+00 -4.26486760e-01 6.98340297e-01
1.39248565e-01 -1.18443795e-01 1.60138652e-01 9.60058272e-01
2.47881301e-02 2.20990062e-01 -7.40208030e-01 3.06214064e-01
-2.28207737e-01 7.65044808e-01 -5.82540274e-01 1.01630473e+00
3.60906750e-01 2.72200197e-01 -1.21118236e+00 -6.99657351e-02
5.05405366e-01 2.06008583e-01 -8.34736153e-02 -3.09261065e-02
2.37380013e-01 6.18586302e-01 -2.05212221e-01 5.51497042e-01
1.19314623e+00 7.74185956e-02 2.37236664e-01 2.58445263e-01
-1.05411780e+00 7.41890818e-02 -1.15503228e+00 5.74422717e-01
8.68400484e-02 1.00112617e+00 3.31633776e-01 -7.56006718e-01
1.28420055e+00 3.86571318e-01 7.62676835e-01 -6.66992486e-01
-5.31123459e-01 2.14472383e-01 4.03937936e-01 -8.32262039e-01
5.37524283e-01 -5.77397645e-01 5.95183037e-02 6.41440451e-01
-7.10305095e-01 -3.45340632e-02 3.18102598e-01 -5.31769320e-02
7.14271009e-01 -3.85821372e-01 8.89186338e-02 -1.05430269e+00
-2.57137008e-02 4.49925125e-01 6.90957010e-01 5.21108687e-01
-2.42728278e-01 3.50313872e-01 6.46742344e-01 -4.45175171e-01
-9.36190724e-01 -1.34521210e+00 -1.05692494e+00 7.26700664e-01
2.28097171e-01 -1.43092901e-01 2.45888725e-01 2.79523224e-01
6.29819512e-01 6.66221917e-01 -8.12208891e-01 2.35999823e-01
-5.92114508e-01 -1.63029420e+00 4.71260071e-01 4.23443556e-01
2.36644477e-01 -8.34735036e-01 -9.92381752e-01 9.02446449e-01
9.23284367e-02 -4.92903262e-01 2.05280811e-01 7.57119596e-01
-1.47940218e+00 -1.04623830e+00 -1.30085304e-01 -2.97497094e-01
6.19273782e-01 3.29722673e-01 1.11108768e+00 -3.94077241e-01
-2.45204926e-01 4.03776690e-02 -4.92992669e-01 -4.85428512e-01
-3.51353973e-01 -3.40792328e-01 -8.06874260e-02 -2.07838804e-01
7.69139767e-01 -7.77620852e-01 -6.27627730e-01 4.47277158e-01
-1.07807100e+00 -7.79420257e-01 3.26260030e-01 5.47365665e-01
1.90979972e-01 3.02742422e-01 7.58223712e-01 -6.06164396e-01
6.86764657e-01 -1.04223287e+00 -7.62482107e-01 -1.12701669e-01
-4.40415740e-01 -2.17906550e-01 3.97951871e-01 -1.48338720e-01
-9.35846865e-01 9.71973911e-02 4.36297446e-01 1.59286022e-01
-5.83138883e-01 1.03572595e+00 3.39645535e-01 3.45417529e-01
8.34870458e-01 3.05097222e-01 1.93392739e-01 -5.08281112e-01
-2.17204019e-02 8.94254148e-01 2.94677228e-01 -3.39426368e-01
7.84296691e-01 9.46536541e-01 3.23503464e-02 -1.59459090e+00
-1.54708520e-01 -9.26959157e-01 -8.75036299e-01 -4.00364608e-01
7.33040273e-01 -1.09056652e+00 -6.29734874e-01 1.37421787e-01
-7.43358195e-01 -1.92718133e-01 -1.52552098e-01 9.89318013e-01
8.85383785e-02 -2.40477361e-02 -3.67198080e-01 -1.03402388e+00
-2.80056987e-02 -9.22381103e-01 8.39124143e-01 1.20864019e-01
-4.95481253e-01 -1.22181857e+00 5.37515581e-01 -7.47536600e-01
5.70248604e-01 9.00561571e-01 9.73916829e-01 -2.96666145e-01
-2.31831684e-03 1.54717267e-01 -1.20119795e-01 -4.72586930e-01
7.64279664e-01 6.82957053e-01 -5.60445309e-01 -4.78193313e-01
4.62667719e-02 5.90486169e-01 6.77997887e-01 8.88131142e-01
1.00636289e-01 -2.71581672e-02 -4.07789379e-01 5.97107708e-01
1.47092092e+00 7.07878411e-01 3.22167724e-01 6.87528253e-01
1.82956174e-01 8.52934718e-01 2.69494832e-01 9.26923096e-01
2.13428080e-01 1.64927632e-01 2.57765710e-01 -2.71252245e-01
3.78398716e-01 3.43675315e-01 5.15164316e-01 4.93113935e-01
3.07860207e-02 2.33621031e-01 -1.31204677e+00 8.95002782e-01
-1.71179771e+00 -1.34186482e+00 -1.06556702e+00 2.00125122e+00
5.05266190e-01 -2.15068102e-01 2.78347731e-01 7.57996142e-02
9.08591926e-01 1.54101908e-01 -4.75449353e-01 -4.67358500e-01
-5.22130489e-01 1.02838010e-01 7.69203603e-01 6.13631964e-01
-1.04396427e+00 6.01670146e-01 6.83821917e+00 -3.97809297e-01
-1.39437628e+00 -4.78199929e-01 -9.62801278e-02 7.29268566e-02
-4.07611728e-01 1.92750528e-01 -8.36830735e-01 4.40131426e-01
1.14663875e+00 -4.29547817e-01 5.94210997e-03 2.54739702e-01
1.41664827e+00 -3.61624449e-01 -6.64490402e-01 2.34558746e-01
-7.81701446e-01 -1.02330744e+00 -1.54362589e-01 3.41206938e-01
7.43610620e-01 2.62171775e-01 -2.81655759e-01 6.36707060e-03
5.26526332e-01 -8.84502709e-01 4.22676414e-01 3.79314423e-01
3.86511356e-01 -6.35352969e-01 5.70693851e-01 -6.17903061e-02
-1.73776948e+00 -1.86205208e-01 -3.27077597e-01 -6.25048161e-01
4.76812541e-01 9.96330142e-01 -7.36607492e-01 3.10261250e-01
1.15414298e+00 1.22554469e+00 -1.96933001e-01 1.14969361e+00
6.50807470e-02 1.11002684e+00 -7.38170743e-01 1.48808926e-01
7.31462836e-01 -7.73491144e-01 7.85781026e-01 1.30298090e+00
4.18460697e-01 6.11795485e-01 6.55205399e-02 7.26035237e-01
7.13094354e-01 6.91908300e-02 -7.62339830e-01 -4.42770571e-02
8.17201018e-01 7.97625363e-01 -1.05893612e+00 -3.15980881e-01
-4.03224856e-01 -4.81865406e-02 -6.95820689e-01 5.44218481e-01
-3.06643009e-01 -6.87992454e-01 1.30567741e+00 1.16198637e-01
2.66600698e-01 -4.69087303e-01 -5.84876180e-01 -7.75940239e-01
-1.61335468e-01 -3.56590152e-01 4.42162305e-01 -2.62835026e-01
-1.42336547e+00 2.99182504e-01 2.55222112e-01 -1.60973966e+00
-1.27189159e-01 -4.25105803e-02 -9.73177850e-01 1.20991623e+00
-1.66953480e+00 -3.17521751e-01 -4.97195810e-01 6.01639509e-01
9.03990939e-02 2.87548423e-01 9.55574453e-01 -4.84090336e-02
-3.80896062e-01 -3.44461530e-01 1.27875233e+00 4.96072881e-02
5.18587589e-01 -1.38056457e+00 3.74538094e-01 6.50320888e-01
-6.64108634e-01 6.19086266e-01 9.51145232e-01 -1.11805594e+00
-1.22699142e+00 -1.45290184e+00 9.26346064e-01 2.31864657e-02
1.09412420e+00 -7.00223446e-02 -1.12993717e+00 4.68981773e-01
-1.52577814e-02 -2.93511271e-01 1.11954713e+00 8.03811550e-02
2.17213765e-01 -9.75125283e-02 -1.10768831e+00 5.00085533e-01
4.33653295e-01 -2.27838218e-01 -9.26282763e-01 3.94452006e-01
-1.90114919e-02 2.62348771e-01 -1.43074846e+00 3.15432101e-01
5.89652479e-01 -6.85264349e-01 3.80112022e-01 -6.79855466e-01
1.86393410e-01 -6.90600872e-01 -3.18715215e-01 -1.53379381e+00
-4.65468228e-01 -4.49662089e-01 8.17697704e-01 8.13180864e-01
3.23369056e-01 -8.48759830e-01 6.25292957e-01 3.47107083e-01
-1.05777591e-01 9.58382711e-02 -8.94710839e-01 -8.23263884e-01
4.47547764e-01 -1.31635129e-01 7.15088308e-01 1.03777993e+00
2.98951149e-01 -1.02813452e-01 3.37007701e-01 7.77835786e-01
8.89790893e-01 8.04439723e-01 5.68356574e-01 -1.86605036e+00
4.48148370e-01 -5.91693401e-01 -3.16962034e-01 -5.16832888e-01
-1.64151311e-01 -6.40759349e-01 1.31995678e-01 -1.47704995e+00
-2.30870962e-01 -4.59579587e-01 1.59404412e-01 3.72757614e-01
-1.43764243e-01 -3.80276859e-01 3.54374424e-02 7.97653139e-01
5.45595467e-01 6.90496922e-01 8.03767085e-01 3.61749567e-02
-1.04909587e+00 -8.70336313e-03 -1.44542217e-01 4.04100090e-01
7.63950646e-01 -6.56664968e-01 4.51044515e-02 -2.34274045e-01
-1.80049110e-02 4.44059968e-01 -2.15507653e-02 -7.12605298e-01
1.42429978e-01 -6.27965689e-01 4.35327023e-01 -9.98378456e-01
-2.61053950e-01 -1.00568426e+00 5.46499431e-01 9.50763881e-01
-7.89250359e-02 4.24411714e-01 5.99919617e-01 6.39370382e-01
-5.55889428e-01 2.69310296e-01 7.23542631e-01 -1.46370500e-01
-5.16725123e-01 3.92562635e-02 -1.31102431e+00 -3.31286430e-01
1.06078184e+00 -4.70337987e-01 -5.01101613e-01 -2.96965212e-01
-8.83285105e-01 7.16909289e-01 4.81838286e-01 2.33550161e-01
3.59809458e-01 -8.56150150e-01 -1.12734616e+00 3.94962698e-01
1.26215413e-01 -8.45235735e-02 5.55235222e-02 8.56010556e-01
-7.77275264e-01 2.58347481e-01 -4.00544614e-01 -1.13769984e+00
-7.84017742e-01 -1.78478315e-01 4.18248922e-01 1.74806356e-01
-9.76927221e-01 7.75946081e-02 -4.34346385e-02 -3.62377584e-01
-3.45871657e-01 -3.80988926e-01 -4.94408339e-01 8.49865437e-01
4.17889029e-01 5.49482107e-01 2.84704193e-02 -6.75298989e-01
-4.64341432e-01 6.70141160e-01 4.38219488e-01 -2.11992338e-01
1.88401878e+00 -3.44092786e-01 -5.08281827e-01 9.07270849e-01
1.16148376e+00 -3.26258123e-01 -1.35925770e+00 7.71894306e-02
3.99687439e-01 -5.37721336e-01 -5.69669306e-02 -1.07155599e-01
-9.99445975e-01 6.45381749e-01 6.94892585e-01 6.77381277e-01
9.78748381e-01 -2.15813145e-01 3.35369408e-01 3.77363563e-01
-3.30936700e-01 -7.17780828e-01 -7.31449902e-01 7.74405122e-01
4.66532648e-01 -9.82852757e-01 -1.21564180e-01 1.11804321e-01
-1.98289335e-01 1.48797631e+00 -3.08762528e-02 -8.48654136e-02
1.28515732e+00 5.26085854e-01 5.64807057e-01 -5.08597314e-01
-5.36851704e-01 -3.35090518e-01 -6.66030288e-01 3.24196875e-01
5.41206837e-01 5.31324685e-01 -6.43522739e-01 -3.60144705e-01
-4.83715743e-01 -5.55586100e-01 8.44764352e-01 1.25987577e+00
-7.08141327e-01 -6.73139572e-01 -1.05750108e+00 8.59806120e-01
-1.29089564e-01 -6.40370473e-02 -4.84588593e-02 7.40642488e-01
-3.39120477e-01 1.31510484e+00 7.71719575e-01 -7.92723224e-02
5.04497230e-01 2.08702888e-02 -4.73417044e-01 -5.22810519e-01
-7.85890996e-01 5.15254915e-01 -3.37156922e-01 -6.92855343e-02
-7.47639477e-01 -1.46009123e+00 -1.50316334e+00 -5.59002399e-01
-2.84552574e-01 7.95105398e-01 7.42157459e-01 7.99855530e-01
4.29117605e-02 3.08220118e-01 1.14453828e+00 -9.63026464e-01
-1.94099888e-01 -1.11059320e+00 -1.35004675e+00 4.79824126e-01
6.12889290e-01 -3.65413338e-01 -9.08564687e-01 1.81339100e-01] | [6.494519233703613, 3.120260238647461] |
161ea0ee-7641-4731-a2c1-5f8ae63aa280 | icdar-2021-competition-on-scientific | 2106.14616 | null | https://arxiv.org/abs/2106.14616v1 | https://arxiv.org/pdf/2106.14616v1.pdf | ICDAR 2021 Competition on Scientific Literature Parsing | Scientific literature contain important information related to cutting-edge innovations in diverse domains. Advances in natural language processing have been driving the fast development in automated information extraction from scientific literature. However, scientific literature is often available in unstructured PDF format. While PDF is great for preserving basic visual elements, such as characters, lines, shapes, etc., on a canvas for presentation to humans, automatic processing of the PDF format by machines presents many challenges. With over 2.5 trillion PDF documents in existence, these issues are prevalent in many other important application domains as well. Our ICDAR 2021 Scientific Literature Parsing Competition (ICDAR2021-SLP) aims to drive the advances specifically in document understanding. ICDAR2021-SLP leverages the PubLayNet and PubTabNet datasets, which provide hundreds of thousands of training and evaluation examples. In Task A, Document Layout Recognition, submissions with the highest performance combine object detection and specialised solutions for the different categories. In Task B, Table Recognition, top submissions rely on methods to identify table components and post-processing methods to generate the table structure and content. Results from both tasks show an impressive performance and opens the possibility for high performance practical applications. | ['Douglas Burdick', 'Xu Zhong', 'Antonio Jimeno Yepes'] | 2021-06-08 | null | null | null | null | ['table-recognition'] | ['computer-vision'] | [ 2.04771250e-01 -6.61700889e-02 -7.82948285e-02 -1.54172868e-01
-1.06672812e+00 -1.17669213e+00 6.36170685e-01 8.37503850e-01
-2.30246902e-01 8.01408410e-01 -3.58912423e-02 -5.00458181e-01
-1.75396770e-01 -7.67547607e-01 -1.17300439e+00 -2.96088994e-01
-1.11280652e-02 5.95518768e-01 -2.39950512e-02 2.98178077e-01
7.82341242e-01 8.34413707e-01 -1.25739610e+00 8.43268633e-01
7.26636469e-01 9.57913876e-01 3.68288219e-01 9.09540594e-01
-1.08329034e+00 5.22860646e-01 -1.09816253e+00 -7.30701327e-01
8.87341332e-03 -1.90503709e-02 -6.97921872e-01 -1.45637259e-01
9.78149414e-01 8.75953287e-02 -1.03987962e-01 9.93590891e-01
4.59899336e-01 -3.73181343e-01 7.52419770e-01 -1.03231537e+00
-8.32573056e-01 9.55170035e-01 -8.25529814e-01 3.29006463e-01
6.23791039e-01 -3.39671262e-02 1.02241242e+00 -8.51805806e-01
1.28153253e+00 1.51551139e+00 5.81158161e-01 2.11373419e-01
-1.05627024e+00 -5.56249440e-01 -3.47120911e-02 3.15774173e-01
-9.06853557e-01 -3.68396997e-01 4.56914157e-01 -5.70711792e-01
1.01472485e+00 2.66709179e-01 2.65318543e-01 1.21000791e+00
3.45832556e-01 1.00015199e+00 9.45871413e-01 -4.35484499e-01
2.94085801e-01 1.80065468e-01 4.60663468e-01 5.51491618e-01
8.94800901e-01 -8.04201722e-01 -7.79917300e-01 8.62576663e-02
4.85942751e-01 -3.16998601e-01 9.47631374e-02 -8.65761712e-02
-1.22130728e+00 5.60907900e-01 -3.99301127e-02 2.27313831e-01
-1.07847184e-01 -2.97803134e-01 8.79205942e-01 -6.03435736e-04
5.53499535e-02 8.70093942e-01 -4.22556490e-01 -3.05762380e-01
-8.84991825e-01 5.59189558e-01 1.03075492e+00 1.42824554e+00
4.67789173e-02 -2.74145484e-01 -4.72070366e-01 8.96041691e-01
-4.77024801e-02 4.53307062e-01 1.24719039e-01 -6.50599837e-01
1.22318292e+00 7.07350969e-01 -1.61702111e-01 -1.23520708e+00
-4.13241059e-01 -4.00850475e-01 -9.45844233e-01 4.37475219e-02
5.99116921e-01 1.73981547e-01 -1.06365454e+00 9.04008746e-01
8.45502540e-02 -6.54794157e-01 -2.39849593e-02 2.89450377e-01
1.53580928e+00 8.92817199e-01 2.08241180e-01 1.18461713e-01
1.82312310e+00 -6.87541127e-01 -1.01688421e+00 -1.09778807e-01
4.36372489e-01 -1.20730948e+00 8.06707680e-01 1.11158872e+00
-1.20223212e+00 -5.51865101e-01 -1.25227296e+00 -6.86315656e-01
-1.13136971e+00 2.18181431e-01 5.67843378e-01 5.71400285e-01
-5.86822689e-01 6.57259762e-01 -3.23559433e-01 -2.16369614e-01
1.10782325e+00 1.74706459e-01 -5.10336757e-01 -4.84915614e-01
-6.13210261e-01 6.15531564e-01 4.02957737e-01 5.05326241e-02
-1.22135036e-01 -1.23529351e+00 -9.40894365e-01 1.89416111e-01
4.98225003e-01 -4.72288549e-01 9.98218119e-01 4.83343974e-02
-1.00814712e+00 9.73360896e-01 1.14149600e-01 -6.12056017e-01
6.69558346e-01 -3.60328674e-01 -4.04659331e-01 2.86486477e-01
2.71451890e-01 7.18900621e-01 4.65402603e-01 -9.68237936e-01
-7.46920943e-01 -4.86342728e-01 -5.02511740e-01 -1.83450520e-01
-1.77389070e-01 1.94654673e-01 -8.92898798e-01 -9.27848339e-01
-4.37429994e-02 -3.82313073e-01 1.89003453e-01 1.91472828e-01
-9.10692096e-01 -4.13451552e-01 1.04091823e+00 -9.30182159e-01
1.02341354e+00 -1.82960153e+00 -1.89358100e-01 1.15877137e-01
2.88136929e-01 1.32840320e-01 1.42068730e-03 3.43072593e-01
1.60448207e-03 6.29348934e-01 -2.09729344e-01 -4.21052933e-01
2.38431782e-01 -1.68103874e-01 -5.84564328e-01 1.66581467e-01
2.89724708e-01 9.28134799e-01 -5.93870401e-01 -7.39285290e-01
1.28785864e-01 5.14135480e-01 -4.32867147e-02 -3.54758985e-02
-4.93598998e-01 -9.39123482e-02 -5.69696128e-01 8.96501243e-01
9.62529778e-01 -4.61496353e-01 1.58062637e-01 -1.24603271e-01
-2.67066658e-01 2.51198024e-01 -1.20219350e+00 1.66564107e+00
-2.47259945e-01 1.06356561e+00 2.34759226e-01 -8.96897495e-01
8.73484731e-01 -6.38838932e-02 2.20735028e-01 -8.25298548e-01
-6.72129691e-02 1.52646318e-01 -1.12445816e-01 -3.68502229e-01
6.11825049e-01 6.92268312e-01 -1.78306535e-01 8.97383019e-02
1.76891591e-02 -3.66926134e-01 1.13130999e+00 5.21885216e-01
1.11267602e+00 2.40384236e-01 1.87497213e-01 -2.80963868e-01
4.39194888e-01 3.15354049e-01 5.41369505e-02 1.00560439e+00
2.80892968e-01 7.79280245e-01 8.92532170e-01 -5.09486258e-01
-1.14014328e+00 -8.46992731e-01 -4.21218455e-01 6.97775006e-01
-3.31823766e-01 -7.75794744e-01 -7.47985721e-01 -5.95680714e-01
2.77245820e-01 6.41250491e-01 -5.65800071e-01 5.68822861e-01
-7.44738460e-01 -6.30299211e-01 5.26134610e-01 5.87195933e-01
3.43523771e-01 -1.37054884e+00 -5.16412795e-01 1.69539332e-01
3.23416650e-01 -1.48127997e+00 -1.97676599e-01 4.79552418e-01
-8.28356981e-01 -1.13462949e+00 -8.45156789e-01 -7.77539909e-01
6.86166823e-01 -4.16486673e-02 1.42251730e+00 -2.51159817e-01
-9.48713601e-01 1.07687920e-01 -1.01852335e-01 -9.53204751e-01
-3.72657597e-01 2.63541758e-01 -4.80812430e-01 -7.53102481e-01
1.88515022e-01 -1.40043229e-01 -7.13082328e-02 -4.05992836e-01
-9.22146499e-01 1.66713968e-01 8.52154791e-01 4.87982869e-01
8.08286846e-01 5.25678955e-02 2.90884823e-01 -1.42683399e+00
7.50242233e-01 -2.35485375e-01 -9.64880347e-01 4.31897283e-01
-4.78421450e-01 2.75150359e-01 1.14329422e+00 -6.18670210e-02
-9.36153293e-01 -3.59615423e-02 -1.45856589e-01 8.86276271e-03
-5.33121347e-01 5.32206357e-01 -5.27270257e-01 3.60704929e-01
3.65615785e-01 1.16904661e-01 -4.75678891e-01 -9.96994793e-01
4.79637861e-01 5.32127798e-01 8.24592113e-01 -7.42569149e-01
6.84700429e-01 1.13344878e-01 2.87139863e-01 -1.06817770e+00
-8.39666128e-01 -2.96727300e-01 -6.36852503e-01 2.35477135e-01
8.41192901e-01 -4.81907398e-01 -8.25330079e-01 2.19716236e-01
-1.55025971e+00 1.64868698e-01 -8.96179304e-02 -1.70903787e-01
-1.44416168e-01 2.89550245e-01 -6.30755484e-01 -5.38798869e-01
-6.11931324e-01 -1.03666437e+00 1.18146765e+00 4.86425757e-01
-2.23627523e-01 -6.72931612e-01 -2.81095743e-01 3.06491792e-01
-4.24516015e-02 3.21144998e-01 1.46450365e+00 -8.33416700e-01
-8.43414783e-01 -1.61433280e-01 -5.54858327e-01 -8.86404663e-02
-7.98126012e-02 4.26983237e-01 -9.35204148e-01 1.42943427e-01
-5.56514740e-01 -2.26114959e-01 1.07906330e+00 4.30786490e-01
1.82573414e+00 -3.62529486e-01 -5.42688370e-01 6.26013458e-01
1.16935837e+00 5.52419722e-01 7.75282443e-01 7.10198700e-01
9.08465624e-01 8.50942552e-01 2.56087989e-01 2.74235278e-01
6.31799623e-02 4.35988724e-01 8.84899199e-02 3.90935093e-02
-4.28869963e-01 -1.70279190e-01 -1.09487087e-01 5.75436532e-01
4.54783291e-01 -7.09354222e-01 -1.15647089e+00 4.08723623e-01
-1.31175113e+00 -9.00654852e-01 -5.17743230e-01 1.91981268e+00
9.91991580e-01 5.37262976e-01 -9.72041190e-02 2.47338086e-01
6.30099475e-01 -2.74200272e-02 -5.27298450e-01 -6.79843545e-01
-5.51891804e-01 4.29149508e-01 5.82861006e-01 -7.97386020e-02
-1.34520471e+00 7.06990838e-01 5.99552011e+00 1.19535315e+00
-8.89161348e-01 -5.44583023e-01 1.12776041e+00 1.36345923e-01
-4.82861213e-02 -5.88032782e-01 -1.30366278e+00 4.95695621e-01
9.24161673e-01 -1.53309882e-01 -1.55076031e-02 9.35175598e-01
-4.13129851e-02 -2.09479958e-01 -1.22033417e+00 1.31386125e+00
5.34110377e-03 -2.14690375e+00 5.18698871e-01 4.74379305e-03
3.93403441e-01 -4.72187191e-01 2.82134265e-01 1.11970402e-01
1.58810467e-01 -1.48112011e+00 7.22988129e-01 3.69987190e-01
8.45149636e-01 -9.25918937e-01 6.37509108e-01 -1.36611015e-01
-8.68342757e-01 2.15726644e-01 -3.89523298e-01 4.28891301e-01
-1.48652360e-01 7.58125663e-01 -8.77825141e-01 5.03943622e-01
1.00614572e+00 8.69245231e-01 -1.05963659e+00 1.30157816e+00
-8.87638554e-02 3.84816170e-01 -2.69976586e-01 -3.12357873e-01
2.62341518e-02 -1.62571415e-01 4.12649989e-01 1.93522167e+00
2.40665898e-01 -3.00314367e-01 -3.18993658e-01 9.66350198e-01
-6.59557879e-01 2.96879023e-01 -5.66799641e-01 -7.53579080e-01
3.43018562e-01 1.45214021e+00 -1.34719920e+00 -4.49259162e-01
-2.56296545e-01 5.90313673e-01 7.55529329e-02 -2.80035306e-02
-4.05666471e-01 -8.27778697e-01 2.84924448e-01 -9.87845436e-02
6.38746917e-01 -1.77382216e-01 -1.08186460e+00 -6.46305680e-01
3.28054428e-01 -1.06463480e+00 5.02006352e-01 -7.33559132e-01
-1.29652643e+00 4.77349967e-01 -2.43447855e-01 -8.56049180e-01
-6.37564203e-03 -1.28713369e+00 -4.05194461e-01 7.82590508e-01
-1.19317806e+00 -9.92529035e-01 -1.38997346e-01 -9.56150219e-02
9.68408763e-01 -2.97034144e-01 7.69218862e-01 2.83323437e-01
-8.16644132e-01 4.90782976e-01 4.81643021e-01 3.04496706e-01
9.35085237e-01 -1.74549603e+00 8.42690289e-01 6.41201079e-01
4.35668141e-01 7.75425076e-01 7.55636990e-01 -7.65996993e-01
-1.95881808e+00 -9.27470982e-01 7.59029984e-01 -6.31745398e-01
5.43650866e-01 -9.95559454e-01 -1.07903457e+00 1.19522400e-01
3.35445851e-01 -2.25041613e-01 4.63644654e-01 -1.25935093e-01
-6.13210320e-01 7.46076927e-02 -1.04853368e+00 5.30340016e-01
6.87485456e-01 -1.54805228e-01 -4.97738391e-01 6.88846350e-01
5.56782544e-01 -7.97531128e-01 -7.37015784e-01 1.60288643e-02
4.95092332e-01 -5.43097198e-01 1.15636480e+00 -6.94756866e-01
1.26127875e+00 -1.00247137e-01 3.17066014e-01 -7.24203408e-01
1.02369279e-01 -6.87257588e-01 -3.27008843e-01 1.33244002e+00
6.06700540e-01 7.37869442e-02 9.57011938e-01 4.63154525e-01
-1.46046847e-01 -5.86533248e-01 -5.98011374e-01 -7.29576468e-01
2.55677074e-01 -3.40191871e-01 3.77503365e-01 6.27232194e-01
-2.62379199e-01 6.43409491e-01 1.51415184e-01 -2.41891608e-01
7.01948643e-01 2.59121686e-01 7.09926188e-01 -1.41091621e+00
7.62826130e-02 -9.83186781e-01 -1.54347435e-01 -8.08672607e-01
-1.77910373e-01 -8.73611093e-01 -2.50166893e-01 -2.16330838e+00
2.99280137e-01 -6.34557307e-02 2.30516084e-02 4.13789749e-01
9.95862763e-03 1.45702243e-01 3.96470696e-01 1.03510823e-02
-7.37428963e-01 -1.00910537e-01 1.14199066e+00 -5.18484771e-01
-3.44139598e-02 -3.71043652e-01 -8.06546092e-01 5.60366452e-01
5.28121591e-01 -4.91722345e-01 5.72227798e-02 -2.85854965e-01
3.92298162e-01 -2.15403005e-01 5.21728992e-02 -9.78937209e-01
3.80472869e-01 -2.31622327e-02 1.18180013e+00 -1.39655995e+00
-2.78625302e-02 -4.75386471e-01 -3.10517460e-01 1.40513688e-01
-4.14338768e-01 3.43206108e-01 8.68834317e-01 3.63302529e-01
-1.64044742e-02 -3.17558408e-01 4.45512265e-01 -2.19225243e-01
-5.10164499e-01 6.71699084e-03 -3.33703786e-01 2.42291659e-01
5.27489722e-01 -3.27955753e-01 -9.95235920e-01 1.42846882e-01
-3.97608757e-01 2.06045374e-01 2.06904933e-01 6.14101171e-01
5.07053971e-01 -7.75608838e-01 -6.48887277e-01 -1.48371262e-02
6.61126971e-02 2.89465308e-01 2.19658390e-01 2.35261902e-01
-9.97424424e-01 9.88654852e-01 -4.17146176e-01 -3.59174311e-01
-1.24134839e+00 6.24471486e-01 -4.44460839e-01 -5.44924140e-01
-9.01601076e-01 8.35231841e-01 1.64250538e-01 -5.86242527e-02
5.38600445e-01 -4.58109826e-01 -4.28311020e-01 2.28673264e-01
1.00551665e+00 4.54338014e-01 4.78719294e-01 1.56494882e-03
-3.67680013e-01 3.47896278e-01 -5.28322875e-01 1.67559981e-01
1.82842994e+00 4.13735360e-01 -1.40904844e-01 2.25299835e-01
1.11135936e+00 4.80142027e-01 -8.34202826e-01 2.83165723e-01
6.17794633e-01 -2.13524982e-01 -1.46862373e-01 -1.36094487e+00
-7.72422016e-01 1.09236467e+00 2.79030591e-01 3.11438709e-01
6.91772342e-01 -1.07084215e-01 5.38685918e-01 7.41834879e-01
-1.38586342e-01 -1.15217984e+00 -1.76212460e-01 5.76404452e-01
1.13749492e+00 -1.08916628e+00 5.41990221e-01 -6.57171190e-01
-2.45104849e-01 1.71776581e+00 4.52652246e-01 2.58478522e-01
1.50907606e-01 8.51939440e-01 -1.28247127e-01 -2.67528653e-01
-5.02762258e-01 4.53635603e-01 7.11707175e-01 6.61251903e-01
6.75974786e-01 -2.29540706e-01 -1.40031680e-01 8.45144928e-01
-4.77238446e-01 -2.93422371e-01 6.53564453e-01 1.10980165e+00
-1.75981790e-01 -1.20770991e+00 -7.32951403e-01 9.13874745e-01
-7.30986357e-01 -2.68688083e-01 -7.23785162e-01 9.68469322e-01
-1.36834204e-01 5.54341495e-01 1.92300484e-01 4.78946805e-01
3.89480472e-01 -1.34258661e-02 6.01675808e-01 -6.61755681e-01
-6.12289310e-01 1.06587090e-01 1.23667173e-01 -3.22053045e-01
2.15053841e-01 -6.85127318e-01 -1.32264364e+00 -2.84248769e-01
4.48378116e-01 3.57432999e-02 1.13062060e+00 5.84879458e-01
5.54968059e-01 8.77475679e-01 -3.77791494e-01 -4.35204059e-01
-2.50269413e-01 -7.95241475e-01 -4.02825266e-01 1.83771417e-01
8.14429000e-02 -3.32052559e-01 1.05364069e-01 5.29378116e-01] | [11.670382499694824, 2.840211868286133] |
924fcd22-1091-413e-9176-b5605b260fdc | road-genome-a-topology-reasoning-benchmark | 2304.10440 | null | https://arxiv.org/abs/2304.10440v2 | https://arxiv.org/pdf/2304.10440v2.pdf | OpenLane-V2: A Topology Reasoning Benchmark for Scene Understanding in Autonomous Driving | Accurately depicting the complex traffic scene is a vital component for autonomous vehicles to execute accurate judgments. However, existing benchmarks tend to oversimplify the scene by solely focusing on lane perception tasks. Observing that human drivers rely on both lanes and traffic signals to operate their vehicles safely, we present OpenLane-V2, the first dataset on topology reasoning for traffic scene structure. The objective of the presented dataset is to advance research in understanding the structure of road scenes by examining the relationship between perceived entities, such as traffic elements and lanes. Leveraging existing datasets, OpenLane-V2 consists of 2,000 annotated road scenes that describe traffic elements and their correlation to the lanes. It comprises three primary sub-tasks, including the 3D lane detection inherited from OpenLane, accompanied by corresponding metrics to evaluate the model's performance. We evaluate various state-of-the-art methods, and present their quantitative and qualitative results on OpenLane-V2 to indicate future avenues for investigating topology reasoning in traffic scenes. | ['Wei zhang', 'Junchi Yan', 'Ping Luo', 'Bangjun Wang', 'Peijin Jia', 'Chonghao Sima', 'Li Chen', 'Tianyu Li', 'Hongyang Li', 'Hang Xu', 'Feng Wen', 'Shengyin Jiang', 'Yuting Wang', 'Yang Li', 'Zhenbo Liu', 'Huijie Wang'] | 2023-04-20 | null | null | null | null | ['3d-lane-detection', 'lane-detection'] | ['computer-vision', 'computer-vision'] | [-1.94989443e-01 3.16818029e-01 -1.65828168e-01 -9.59097207e-01
-2.92131275e-01 -6.89119041e-01 8.75237703e-01 7.75797591e-02
-2.04873960e-02 3.52411687e-01 1.86170414e-01 -8.93557727e-01
-2.13650465e-02 -9.03597355e-01 -7.06680298e-01 -1.11450672e-01
-2.15822935e-01 6.92880392e-01 8.35008562e-01 -5.99328756e-01
4.26795691e-01 8.61682236e-01 -1.90756273e+00 3.19819063e-01
7.61110902e-01 1.17002094e+00 -1.17439710e-01 6.11687243e-01
-4.35797691e-01 8.72256219e-01 -3.57200027e-01 -7.04630733e-01
1.93216592e-01 3.62426251e-01 -4.78780448e-01 1.87639326e-01
1.10854745e+00 -2.73273915e-01 -7.24731863e-01 9.39511955e-01
-1.37001097e-01 1.45830050e-01 7.51590073e-01 -1.90661597e+00
-1.26545474e-01 3.67478460e-01 -1.52558908e-01 3.91479462e-01
2.79721677e-01 5.78705370e-01 1.01287293e+00 -8.96967053e-01
7.43795216e-01 1.78709698e+00 6.36589170e-01 -5.19455934e-05
-1.04044163e+00 -4.99495327e-01 4.90472257e-01 7.99208462e-01
-1.37031627e+00 -8.34337771e-01 8.94811392e-01 -6.19441807e-01
9.67896223e-01 3.13427210e-01 4.61169749e-01 9.80038285e-01
2.62672752e-01 7.42323875e-01 1.18149233e+00 2.59220690e-01
7.12694526e-02 2.10214093e-01 4.46893036e-01 9.14994001e-01
5.01914263e-01 3.96375209e-01 -2.89711267e-01 4.18495357e-01
3.35016340e-01 -6.86301470e-01 4.23459411e-01 -8.38416576e-01
-1.19181693e+00 4.96752828e-01 7.03462958e-01 -3.96462619e-01
-1.33519858e-01 1.89259216e-01 5.69579542e-01 1.45454854e-01
7.96593577e-02 1.52304962e-01 1.66675940e-01 -1.62377104e-01
-3.38382274e-01 5.37906170e-01 7.49714732e-01 1.42435956e+00
9.64864373e-01 -3.03150024e-02 -2.54415691e-01 5.58946550e-01
3.25162619e-01 7.70745754e-01 -8.44708443e-01 -1.33838832e+00
9.97706592e-01 6.28909349e-01 4.74604778e-02 -1.35755396e+00
-6.29208863e-01 -2.14380935e-01 -4.15112495e-01 4.19572592e-01
4.59821612e-01 2.92043626e-01 -8.37602556e-01 1.39213347e+00
1.16399445e-01 3.39113355e-01 -2.45592762e-02 8.69247496e-01
1.11563849e+00 3.88078332e-01 3.65060121e-01 4.87021744e-01
1.41048586e+00 -1.16317177e+00 -6.06765091e-01 -5.63578069e-01
6.49875820e-01 -5.77434421e-01 9.47930515e-01 3.53724100e-02
-8.31381261e-01 -8.31686020e-01 -9.86336172e-01 -4.00193661e-01
-9.38911498e-01 2.16665678e-02 6.70658350e-01 6.12006426e-01
-9.74235773e-01 -1.96065858e-01 -2.32453346e-01 -1.53379858e-01
8.27701449e-01 -2.90711850e-01 -5.11372268e-01 -2.27240339e-01
-1.32398999e+00 1.42502904e+00 2.53751010e-01 3.25031489e-01
-1.14762950e+00 -7.75632501e-01 -1.22685266e+00 -5.90270236e-02
7.09116042e-01 -5.68842053e-01 1.17119253e+00 6.57157153e-02
-8.90338719e-01 9.46556926e-01 -2.95561761e-01 -6.55519247e-01
9.62004244e-01 1.45619273e-01 -9.18973863e-01 7.40338564e-02
4.16684538e-01 9.86241758e-01 5.10198355e-01 -1.85598505e+00
-1.19179511e+00 -9.08895582e-02 3.32688808e-01 3.76583748e-02
6.90484762e-01 -4.08749610e-01 -7.04685032e-01 -3.35915089e-02
5.75001016e-02 -8.49732459e-01 -3.18028986e-01 2.61791676e-01
-9.04044926e-01 -3.95434111e-01 1.07567728e+00 -7.49684423e-02
1.05436385e+00 -2.11617422e+00 -5.54795206e-01 4.12781656e-01
6.40078068e-01 1.69163823e-01 -2.89710999e-01 2.86283433e-01
2.16741666e-01 2.64053363e-02 -1.05034195e-01 -3.63656461e-01
5.04055142e-01 4.41482872e-01 -7.48032570e-01 3.15647095e-01
2.82899916e-01 1.22002113e+00 -9.80156302e-01 -7.48309076e-01
7.72538900e-01 1.89520717e-01 -1.69691250e-01 -8.35404247e-02
-2.01651424e-01 1.91435963e-01 -4.29771394e-01 7.20030844e-01
1.00698555e+00 3.35113972e-01 -2.36186221e-01 -4.89811897e-01
-4.25875396e-01 6.14757776e-01 -9.95216906e-01 1.08592439e+00
-3.62713814e-01 1.46495461e+00 6.07525557e-03 -6.63485348e-01
9.09287751e-01 -3.73499662e-01 7.50929341e-02 -1.13038683e+00
6.75794929e-02 -6.54274672e-02 -7.14187995e-02 -8.33046138e-01
7.44721591e-01 4.50426847e-01 -5.18035531e-01 -8.08525160e-02
-6.05176330e-01 -2.78042197e-01 5.92221022e-01 3.34229648e-01
9.28880930e-01 -1.24368712e-01 -3.58237512e-02 -1.72804043e-01
5.45001626e-01 4.75931883e-01 3.01486880e-01 9.33113039e-01
-7.86507547e-01 1.74588561e-02 9.00641978e-01 -6.72719479e-01
-1.05950856e+00 -1.70818198e+00 -3.96687686e-01 7.53010750e-01
7.61104882e-01 -3.04902673e-01 -4.53245670e-01 -7.94479311e-01
5.54367721e-01 1.15435612e+00 -7.51810133e-01 -1.06927818e-02
-5.90030849e-01 1.56337664e-01 8.00252199e-01 7.24077761e-01
7.86034524e-01 -7.91684091e-01 -6.80301964e-01 2.24202219e-02
-2.47033581e-01 -1.91314995e+00 -4.59340513e-02 -4.67563897e-01
-2.55932868e-01 -1.34161603e+00 2.51120955e-01 -4.35393035e-01
6.39950514e-01 8.10193181e-01 1.51569784e+00 -4.39123884e-02
-1.44419625e-01 2.43923634e-01 6.11821413e-02 -6.23107076e-01
-5.06293595e-01 -6.56381994e-02 -8.36158395e-02 8.48293230e-02
4.89787549e-01 -1.49517864e-01 -5.34774542e-01 1.01533616e+00
-2.55071938e-01 3.34986836e-01 7.67153919e-01 4.84112911e-02
3.99909079e-01 2.33255729e-01 2.61253774e-01 -1.05794275e+00
3.06746423e-01 -5.60342669e-01 -7.39728034e-01 1.09732516e-01
-2.64018774e-01 -1.15400590e-01 2.14484259e-01 2.10148901e-01
-1.11643040e+00 -1.52357161e-01 -9.92099717e-02 -2.82002956e-01
-6.22768521e-01 9.48097110e-02 -4.80364919e-01 7.58507941e-03
3.95296663e-01 -1.18335247e-01 -1.75363362e-01 -9.58906189e-02
7.98021555e-01 3.63923728e-01 9.51010466e-01 -4.98722821e-01
9.84685361e-01 8.40566576e-01 4.24654931e-01 -9.28112924e-01
-8.51421833e-01 -5.01533985e-01 -8.59556794e-01 -8.35632980e-01
8.61231208e-01 -7.51225889e-01 -1.20779335e+00 1.88918129e-01
-1.01078463e+00 -3.57394040e-01 -2.93985493e-02 2.27315620e-01
-7.79340088e-01 2.09473610e-01 -2.31980121e-05 -6.53363049e-01
7.54027307e-01 -1.35130894e+00 1.26029623e+00 -2.52828717e-01
6.27545416e-02 -9.12485301e-01 -3.41045946e-01 7.34576464e-01
4.18068200e-01 1.94288507e-01 1.15537918e+00 -2.36212313e-01
-1.09998906e+00 -4.64370735e-02 -8.32745373e-01 -1.23582847e-01
-9.04704034e-02 2.10910484e-01 -1.14672613e+00 3.67480278e-01
-7.95231462e-01 1.73883140e-01 1.13072109e+00 2.38397438e-02
1.28459871e+00 9.61719081e-02 -7.23456442e-01 3.74401122e-01
8.82864773e-01 8.57282802e-03 9.33506906e-01 2.86368430e-01
8.83360922e-01 1.27316797e+00 1.08272994e+00 -1.30346462e-01
1.15921521e+00 7.70330429e-01 8.84452760e-01 -1.97759539e-01
-5.26729465e-01 -5.87509453e-01 5.82456309e-03 2.64043629e-01
2.23132357e-01 -4.01418686e-01 -1.23035288e+00 5.69153786e-01
-1.74400115e+00 -1.03669691e+00 -8.30794394e-01 1.78792083e+00
-4.42376100e-02 8.47575188e-01 2.70489067e-01 7.87673593e-02
6.45339370e-01 4.53115284e-01 -6.12993836e-01 -5.07341862e-01
-1.95285305e-01 -7.20284939e-01 8.17586064e-01 6.91949248e-01
-1.27161825e+00 1.39154387e+00 7.09060192e+00 7.19905257e-01
-7.06796706e-01 -3.42366397e-01 5.71176112e-01 3.43531728e-01
-4.77177501e-01 1.66659787e-01 -1.11827171e+00 2.01489478e-01
7.87971020e-01 1.48243353e-01 1.28795639e-01 8.04863453e-01
4.98621285e-01 -2.60223508e-01 -1.22320020e+00 7.44710743e-01
-1.19760700e-01 -1.54157948e+00 2.94692487e-01 6.71527609e-02
4.08720464e-01 3.89819667e-02 2.69884944e-01 5.89788079e-01
2.83599257e-01 -1.11891913e+00 1.03814244e+00 8.09876204e-01
7.19086885e-01 -4.99889493e-01 5.03995538e-01 1.53395176e-01
-1.58004093e+00 -2.04702035e-01 -2.53135473e-01 4.77055274e-02
6.16946101e-01 3.65210533e-01 -1.06458724e+00 2.74202824e-01
5.28090417e-01 8.50442946e-01 -9.44049835e-01 1.01676548e+00
-2.32432783e-01 3.15718323e-01 -1.56573996e-01 2.02683911e-01
5.77994108e-01 -2.36431465e-01 8.90655637e-01 1.34918916e+00
-1.45167664e-01 -2.06090361e-01 1.91457883e-01 1.13587999e+00
1.02970913e-01 -4.20615941e-01 -1.20384634e+00 2.35433415e-01
7.84563124e-01 1.09088933e+00 -5.60557723e-01 -4.32243526e-01
-6.11465156e-01 -7.66589195e-02 3.51550490e-01 4.80990171e-01
-1.30640078e+00 -2.50826508e-01 1.31334805e+00 4.77225423e-01
2.52188712e-01 -6.47753119e-01 -5.41822672e-01 -4.29822773e-01
1.05028450e-01 -3.18511695e-01 7.68358679e-03 -1.09913516e+00
-1.04699254e+00 6.06445849e-01 4.56845492e-01 -1.47607970e+00
2.86600858e-01 -8.15665305e-01 -5.39033234e-01 3.98022264e-01
-2.14552426e+00 -1.22735548e+00 -8.12127888e-01 2.62912154e-01
6.59548461e-01 -1.35680079e-01 6.95770383e-02 2.65867800e-01
-5.15734255e-01 3.49393100e-01 -5.04298866e-01 1.94467343e-02
4.48147833e-01 -1.13285124e+00 1.09679031e+00 6.48068190e-01
-4.25518826e-02 1.65037945e-01 7.63835788e-01 -5.77481031e-01
-1.33798611e+00 -1.48406160e+00 8.83704066e-01 -1.08760512e+00
8.41339707e-01 -6.28915250e-01 -7.02627301e-01 8.04530263e-01
-1.80474550e-01 1.84932724e-01 2.78966706e-02 -8.87350589e-02
-6.36663258e-01 -4.44397807e-01 -8.46088886e-01 9.38711584e-01
1.70990241e+00 -5.83375514e-01 -6.35584116e-01 -4.59672185e-03
5.26065469e-01 -4.90183741e-01 -5.67975938e-01 5.95010936e-01
4.93570149e-01 -1.19906175e+00 1.30074966e+00 -6.77153587e-01
1.81669593e-01 -7.57880449e-01 -2.66522139e-01 -1.09210348e+00
-3.06901604e-01 -2.66001791e-01 8.15021470e-02 8.54364216e-01
5.61964035e-01 -8.12921107e-01 5.94644010e-01 4.82853472e-01
-7.91389525e-01 -5.40237367e-01 -9.68838155e-01 -7.52393484e-01
-2.78485358e-01 -1.12267518e+00 8.25478494e-01 3.79329145e-01
-5.14261961e-01 2.69020051e-01 1.26661211e-01 5.04148245e-01
8.33171308e-01 1.03319988e-01 1.24495137e+00 -1.36119032e+00
7.62324393e-01 -9.19661760e-01 -9.09553707e-01 -1.49400854e+00
5.47267079e-01 -7.96192348e-01 2.54892588e-01 -1.62840533e+00
-3.34370643e-01 -9.12291825e-01 5.42730540e-02 1.63370937e-01
2.02121168e-01 2.23837703e-01 1.80146351e-01 -1.11081906e-01
-9.87518728e-01 3.93926859e-01 1.57389116e+00 -4.92548466e-01
2.24240988e-01 7.57827312e-02 -5.60508668e-01 7.75538325e-01
5.99474728e-01 9.66395512e-02 -7.37304747e-01 -4.95688647e-01
2.47835174e-01 -8.96883234e-02 8.82634699e-01 -1.00061452e+00
3.68598282e-01 -4.05502141e-01 -1.01285540e-01 -1.38513768e+00
6.03380501e-01 -9.36280012e-01 -1.09412357e-01 -6.83722645e-03
-5.25897861e-01 1.30296335e-01 3.77577156e-01 7.89519072e-01
-2.10952505e-01 4.48785394e-01 4.24567342e-01 2.11099342e-01
-1.74374270e+00 3.04875582e-01 -6.25186741e-01 8.27133358e-02
1.47783029e+00 -5.94596803e-01 -1.04692996e+00 -3.41595113e-01
-7.08955467e-01 9.77199078e-01 2.98176140e-01 9.78090465e-01
8.38762283e-01 -1.44370854e+00 -6.69611692e-01 4.20981944e-01
7.28535295e-01 -5.47159985e-02 3.72249693e-01 8.39441717e-01
-7.40997314e-01 9.68722284e-01 -4.11597311e-01 -9.59748328e-01
-1.08988059e+00 6.66520834e-01 3.95553440e-01 3.52343053e-01
-7.70910501e-01 4.03232485e-01 4.97040004e-01 -4.19643074e-01
2.03607455e-01 -6.93037033e-01 -4.77479994e-01 -1.65388852e-01
2.68923670e-01 7.41163254e-01 4.14106287e-02 -1.16825855e+00
-4.78074580e-01 6.52527153e-01 1.76032767e-01 1.86746687e-01
5.89749992e-01 -6.71507657e-01 1.38179168e-01 4.57834363e-01
1.06085742e+00 -7.07738101e-02 -1.54206026e+00 -1.54511496e-01
1.12314492e-01 -7.68953204e-01 -4.35162000e-02 -6.86980903e-01
-7.78279901e-01 1.07322550e+00 3.05379927e-01 2.68427223e-01
4.03492957e-01 1.17582284e-01 8.08202326e-01 6.16466880e-01
6.11991107e-01 -1.04795063e+00 -2.04441309e-01 7.46718705e-01
9.35100675e-01 -1.52229667e+00 -2.92789578e-01 -1.28918386e+00
-8.67816508e-01 9.66789067e-01 8.50666702e-01 1.36932656e-01
7.47015297e-01 1.07448414e-01 3.09983104e-01 -5.86734056e-01
-8.67693782e-01 -6.76251054e-01 6.23774111e-01 8.35208178e-01
-2.91870952e-01 5.29140294e-01 5.08880317e-01 -9.64254141e-02
-5.85767508e-01 -5.67297757e-01 3.82245570e-01 5.20624697e-01
-6.62672043e-01 -6.04953945e-01 -1.32754788e-01 4.96538311e-01
5.97730517e-01 3.98060888e-01 -5.22408426e-01 1.06080556e+00
2.46341109e-01 1.27480161e+00 5.59923708e-01 -4.07011837e-01
8.73335540e-01 -3.19535911e-01 1.64922491e-01 -2.16629699e-01
2.31591925e-01 -7.49435186e-01 9.12924051e-01 -1.01208425e+00
2.61108726e-02 -5.29864371e-01 -1.25153708e+00 -6.80849373e-01
3.79771113e-01 -1.96061358e-01 7.86524832e-01 9.43154395e-01
4.78946179e-01 6.66932464e-01 7.22778976e-01 -8.15767169e-01
5.96080767e-03 -3.13592583e-01 -1.16432101e-01 5.22587955e-01
5.84426999e-01 -1.29404020e+00 -1.58718094e-01 -3.44734460e-01] | [8.014110565185547, -1.6064603328704834] |
e5399e18-5a9e-4872-bf3c-5db7dc5f8d83 | discriminatory-and-orthogonal-feature | 2210.11519 | null | https://arxiv.org/abs/2210.11519v1 | https://arxiv.org/pdf/2210.11519v1.pdf | Discriminatory and orthogonal feature learning for noise robust keyword spotting | Keyword Spotting (KWS) is an essential component in a smart device for alerting the system when a user prompts it with a command. As these devices are typically constrained by computational and energy resources, the KWS model should be designed with a small footprint. In our previous work, we developed lightweight dynamic filters which extract a robust feature map within a noisy environment. The learning variables of the dynamic filter are jointly optimized with KWS weights by using Cross-Entropy (CE) loss. CE loss alone, however, is not sufficient for high performance when the SNR is low. In order to train the network for more robust performance in noisy environments, we introduce the LOw Variant Orthogonal (LOVO) loss. The LOVO loss is composed of a triplet loss applied on the output of the dynamic filter, a spectral norm-based orthogonal loss, and an inner class distance loss applied in the KWS model. These losses are particularly useful in encouraging the network to extract discriminatory features in unseen noise environments. | ['Hanseok Ko', 'David K. Han', 'Kyungdeuk Ko', 'Donghyeon Kim'] | 2022-10-20 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 4.19370115e-01 -1.87064022e-01 6.81810081e-02 -3.53970110e-01
-7.80601859e-01 -3.32654655e-01 2.81037241e-01 9.03444588e-02
-5.31392813e-01 4.46534723e-01 5.51236495e-02 -3.46175700e-01
-2.70012796e-01 -3.72002006e-01 -4.39173549e-01 -8.54541004e-01
-3.14609185e-02 -5.41850448e-01 2.58232117e-01 7.35432357e-02
-3.83203849e-02 5.97231388e-01 -1.33219802e+00 -1.02496132e-01
8.68198514e-01 1.63793504e+00 4.81606454e-01 7.88055122e-01
3.11779410e-01 4.52603310e-01 -6.57606006e-01 -1.19053535e-01
3.89794856e-01 -4.06641006e-01 -2.37539709e-01 -4.76242781e-01
6.84142783e-02 -1.19328924e-01 -5.52659392e-01 1.07154763e+00
1.15656948e+00 4.69490349e-01 4.69799101e-01 -1.17717659e+00
1.30154610e-01 4.70077217e-01 -1.61144465e-01 4.03452247e-01
3.69474977e-01 1.30828723e-01 1.07817876e+00 -9.39047158e-01
-8.17976706e-03 8.49437892e-01 8.27410460e-01 3.22729617e-01
-1.18653631e+00 -7.63230324e-01 1.59913972e-01 3.97502929e-01
-1.39139020e+00 -6.40088320e-01 8.77341807e-01 -1.06882349e-01
8.97503674e-01 5.66852450e-01 3.40447843e-01 9.57874298e-01
1.22937568e-01 8.87188256e-01 8.43557060e-01 -3.95088404e-01
3.21514040e-01 4.49116439e-01 1.63903832e-02 4.10054058e-01
-5.69551066e-02 1.08159170e-01 -9.10497367e-01 -8.03919360e-02
2.98750311e-01 -1.50348857e-01 -7.10094273e-01 -2.64236510e-01
-7.27290154e-01 6.90953076e-01 5.27724802e-01 2.54643619e-01
-2.62243062e-01 9.34810564e-02 3.90745729e-01 3.81765872e-01
3.07468235e-01 4.86163557e-01 -3.89623195e-01 -3.33572894e-01
-1.05421937e+00 -1.38495117e-01 6.97225749e-01 5.04860044e-01
3.10638279e-01 1.58068180e-01 -4.61385876e-01 8.73746991e-01
2.50006735e-01 5.26871026e-01 3.44675124e-01 -4.94742483e-01
6.01973295e-01 1.88859537e-01 -2.42826696e-02 -1.01733565e+00
-6.60006940e-01 -7.84005940e-01 -8.48106205e-01 -1.42389119e-01
7.45810047e-02 -2.45541900e-01 -6.41991735e-01 1.79472101e+00
2.91010141e-01 2.82425612e-01 4.17345436e-03 8.93009961e-01
5.88883698e-01 4.71721679e-01 -9.34138596e-02 -2.33757392e-01
1.02118897e+00 -5.34663677e-01 -9.94917393e-01 -1.19138554e-01
4.09917802e-01 -7.93330669e-01 1.26682734e+00 3.59807968e-01
-9.75192308e-01 -3.98379266e-01 -1.45090210e+00 2.12784708e-01
-4.42071974e-01 4.12149489e-01 9.00122449e-02 8.74856174e-01
-6.69732392e-01 7.36615241e-01 -6.51428223e-01 -2.05171943e-01
3.26457441e-01 5.38725019e-01 -1.03235710e-03 1.01152264e-01
-1.45578170e+00 7.35318959e-01 2.96497732e-01 3.13678861e-01
-5.49132705e-01 -6.23142719e-01 -8.12572181e-01 3.39892566e-01
5.81681848e-01 -2.78329700e-01 9.48731065e-01 -3.53982508e-01
-1.74541080e+00 1.82947293e-01 1.55872822e-01 -6.69324636e-01
5.84470689e-01 -3.71674538e-01 -7.69445360e-01 2.36325637e-01
-2.24064633e-01 1.12548627e-01 1.19548237e+00 -1.00654936e+00
-6.24803364e-01 -3.06518644e-01 3.76932248e-02 2.89527267e-01
-8.46190214e-01 -4.33366857e-02 -3.83416086e-01 -1.01648867e+00
4.25477363e-02 -8.52386475e-01 -6.39309734e-02 -1.80140153e-01
-5.66717684e-01 -1.33593529e-01 1.15553343e+00 -9.29019511e-01
1.47367287e+00 -2.51345396e+00 -2.36018598e-01 5.73467314e-01
-4.53322530e-02 6.57080352e-01 8.11157599e-02 1.53493479e-01
8.75973925e-02 -1.16761200e-01 -4.51793224e-02 -2.48803526e-01
5.59594035e-02 -2.61454314e-01 -1.71535254e-01 5.37453771e-01
2.12400451e-01 3.84180427e-01 -7.13519752e-01 5.82948588e-02
1.63849607e-01 6.49834812e-01 -3.59146386e-01 4.52967435e-01
1.58664495e-01 1.83581114e-01 -3.62547040e-01 3.13592613e-01
4.93593752e-01 2.27664853e-03 -2.41100013e-01 -5.70377052e-01
1.46404117e-01 2.81649143e-01 -1.54936051e+00 1.46104169e+00
-8.77943099e-01 5.34038126e-01 2.69447893e-01 -1.03084683e+00
7.62145936e-01 3.34384352e-01 2.70160854e-01 -8.30185175e-01
2.17306688e-01 2.01295584e-01 -2.15046808e-01 -4.24945563e-01
3.75936151e-01 3.44356030e-01 -1.26955211e-01 1.76633656e-01
-1.56368706e-02 7.63264969e-02 -6.11048490e-02 1.78797573e-01
1.11916494e+00 -2.29803339e-01 3.38522010e-02 -3.21561515e-01
6.10246539e-01 -8.03131342e-01 6.54656708e-01 8.96654069e-01
8.54246914e-02 5.46862125e-01 2.10997462e-01 -3.34557518e-02
-6.21058941e-01 -9.51688588e-01 -2.20106781e-01 8.68381202e-01
2.70528406e-01 -4.48928028e-01 -5.19333065e-01 -8.04756284e-01
-1.20790862e-01 8.58542681e-01 -3.38208750e-02 -7.35867977e-01
-3.12958628e-01 -7.57182419e-01 6.14939094e-01 5.77743590e-01
6.36201560e-01 -4.95813549e-01 -7.72959411e-01 1.38466030e-01
-3.58822905e-02 -1.42563164e+00 -9.44140017e-01 8.25859129e-01
-2.62467772e-01 -8.71215403e-01 -5.73956370e-01 -3.89975876e-01
6.31088614e-01 2.14751229e-01 4.80376035e-01 -1.70748636e-01
-3.88611495e-01 3.82751346e-01 -5.03940642e-01 -5.44309914e-01
2.45681629e-02 6.56134114e-02 3.67561519e-01 3.35578144e-01
1.73291177e-01 -4.21348393e-01 -7.13085949e-01 4.02634919e-01
-8.74354362e-01 -3.49679410e-01 4.12652373e-01 1.03855813e+00
2.83856928e-01 5.09022057e-01 5.28338194e-01 -2.92050183e-01
8.04012001e-01 -2.25639284e-01 -6.83991194e-01 8.06969404e-02
-6.53844178e-01 1.10114984e-01 9.63258326e-01 -5.89023411e-01
-7.70597339e-01 8.03666115e-02 -3.06157827e-01 -1.80136412e-01
3.58569026e-01 3.50306153e-01 -4.37967688e-01 -3.41718018e-01
5.52252591e-01 2.15938702e-01 -2.94634104e-01 -5.31000257e-01
4.18075696e-02 1.07894194e+00 3.42194349e-01 -1.08392596e-01
1.00118041e+00 1.35541186e-01 6.30973876e-02 -1.15780175e+00
-7.66091347e-01 -7.35488474e-01 -2.46030241e-01 -7.41743222e-02
4.73058254e-01 -8.28464985e-01 -8.67349744e-01 3.11798304e-01
-7.04555631e-01 -1.00099072e-02 -3.80931377e-01 6.73982620e-01
-2.70111084e-01 2.63482511e-01 3.98897231e-02 -1.02466333e+00
-6.31123900e-01 -1.24223650e+00 9.11513150e-01 4.56099719e-01
-1.67482167e-01 -6.83576465e-01 -5.77052236e-01 9.67735872e-02
6.95052862e-01 6.66343793e-02 7.21306920e-01 -8.88410449e-01
-5.32748885e-02 -6.29693449e-01 -1.38846800e-01 8.36319029e-01
5.17709590e-02 -5.09439468e-01 -1.21556497e+00 -4.79534924e-01
2.67186671e-01 -2.74377316e-01 9.01497245e-01 1.17140330e-01
1.62008202e+00 -4.45752412e-01 -1.92477375e-01 8.06664467e-01
1.28253615e+00 3.63080025e-01 3.29422623e-01 8.32228288e-02
5.07903337e-01 1.70534566e-01 4.16685611e-01 4.98834372e-01
-9.64499861e-02 8.72432709e-01 1.60737455e-01 -5.18517233e-02
-6.73557073e-02 -7.34688640e-02 5.29228806e-01 7.35187232e-01
4.51579183e-01 -4.46511388e-01 -5.37777364e-01 3.18044305e-01
-1.62554872e+00 -6.76844597e-01 3.73362541e-01 2.52066493e+00
7.51679957e-01 4.90175903e-01 -9.04987976e-02 6.39994979e-01
4.60057467e-01 2.84136504e-01 -8.55433345e-01 -4.28044766e-01
-2.28282332e-01 1.18910104e-01 6.30092740e-01 3.92710119e-01
-1.28009689e+00 2.99470961e-01 5.20914793e+00 1.19712341e+00
-1.38939369e+00 -2.00449452e-02 4.66598570e-01 -2.61230975e-01
-4.09800708e-02 -1.46662220e-01 -1.00598073e+00 7.98870683e-01
1.10979366e+00 5.81633262e-02 3.29517454e-01 8.16953838e-01
4.97817963e-01 -2.32167318e-01 -9.05359805e-01 1.12830007e+00
2.55218185e-02 -9.29381967e-01 -5.47077000e-01 -1.29057169e-01
3.35368693e-01 -7.47311041e-02 3.33310038e-01 1.11054443e-03
-1.39831007e-01 -8.92722547e-01 5.67573309e-01 3.85353982e-01
7.93329835e-01 -8.44495356e-01 7.46564031e-01 3.02605093e-01
-1.14845288e+00 -3.43564361e-01 -7.88358748e-02 4.62508410e-01
1.18059114e-01 1.04812634e+00 -8.60427856e-01 4.78722394e-01
8.19224417e-01 1.47155583e-01 -2.13537380e-01 1.19035292e+00
-1.44436643e-01 6.68427646e-01 -6.66642666e-01 -2.03631222e-01
1.38205424e-01 -2.90526636e-03 9.04809713e-01 1.34150445e+00
3.04160565e-01 -1.68935418e-01 2.89210528e-01 4.36516732e-01
-2.19529912e-01 5.30069992e-02 -3.36097091e-01 4.36257683e-02
4.27602887e-01 1.38988781e+00 -5.45557380e-01 6.41904399e-02
-2.17336550e-01 1.16105556e+00 -1.10709876e-01 3.16477895e-01
-6.95579886e-01 -9.89971757e-01 6.73954487e-01 1.48154408e-01
4.76228356e-01 -1.60768270e-01 -2.79101223e-01 -8.16302061e-01
3.74900997e-01 -7.64361739e-01 4.60051745e-02 -3.33129734e-01
-1.08716881e+00 5.85716248e-01 -2.89659590e-01 -1.32564950e+00
-3.03701162e-02 -4.11815286e-01 -5.89644253e-01 1.11101723e+00
-1.59986722e+00 -6.75083816e-01 -3.23217094e-01 5.11409938e-01
4.09441024e-01 -2.81841546e-01 6.81815624e-01 5.66842318e-01
-6.91766381e-01 1.19163692e+00 3.33514035e-01 -3.81524675e-02
5.04691720e-01 -1.13747919e+00 8.38521048e-02 1.02562535e+00
5.34010381e-02 6.12020075e-01 9.58169162e-01 -3.39368194e-01
-1.25364482e+00 -1.05978632e+00 6.40519917e-01 3.01341832e-01
4.64327067e-01 -7.09212542e-01 -6.93467915e-01 -1.39610052e-01
-5.27663119e-02 3.98687780e-01 8.09019566e-01 -2.68810451e-01
-3.69284302e-02 -4.29392993e-01 -1.18931210e+00 4.57848370e-01
6.81269526e-01 -6.87860608e-01 6.40190840e-02 4.23508048e-01
7.41117895e-01 -3.55872959e-01 -8.74525428e-01 6.23347938e-01
6.29128754e-01 -5.82308412e-01 7.56478965e-01 -2.87098557e-01
-3.50286335e-01 -3.84782374e-01 -2.71215618e-01 -1.39572191e+00
-1.79450903e-02 -1.13137281e+00 -3.41668576e-01 1.27732515e+00
5.20180762e-01 -5.41539431e-01 6.68776512e-01 4.81194735e-01
3.35049368e-02 -9.62934256e-01 -1.02732110e+00 -1.05579221e+00
-7.17615843e-01 -5.60779274e-01 3.79667938e-01 3.84636045e-01
-1.56577408e-01 2.77848721e-01 -5.10450900e-01 2.93239921e-01
5.04725933e-01 -4.72000182e-01 3.57847065e-01 -1.07364905e+00
-3.61520708e-01 -2.87360370e-01 -3.64564627e-01 -1.18383789e+00
-1.40865535e-01 -6.21943653e-01 3.08752000e-01 -1.09775996e+00
-2.50655621e-01 -4.76719916e-01 -5.55451989e-01 1.28666848e-01
-3.23034346e-01 1.18215837e-01 2.07672358e-01 -1.92221850e-01
-4.17501152e-01 8.19348276e-01 5.09647131e-01 -2.08718762e-01
-3.82106215e-01 5.57439148e-01 -6.39730155e-01 8.89965773e-01
9.45787668e-01 -5.13019562e-01 -3.98675025e-01 -3.59820314e-02
1.75457388e-01 -3.21592480e-01 6.51166439e-02 -1.24375451e+00
6.23723626e-01 2.70555675e-01 4.47344005e-01 -4.29409802e-01
6.70327067e-01 -1.12724102e+00 -3.28453183e-01 4.10912424e-01
-4.02135640e-01 -2.65518457e-01 2.68598460e-02 6.63573861e-01
-1.83411196e-01 -1.90362766e-01 9.31720734e-01 3.50654334e-01
-5.36904991e-01 1.17772169e-01 -3.35098593e-03 -9.43678394e-02
8.25185776e-01 -3.26941550e-01 1.77446917e-01 -5.13864338e-01
-5.77348650e-01 3.71867597e-01 -1.93032131e-01 5.81245184e-01
7.90542543e-01 -1.24029708e+00 -3.05755436e-01 4.43678260e-01
-6.40664622e-02 -1.24598145e-01 1.42928541e-01 9.35187638e-01
-1.76836744e-01 4.77186024e-01 3.48092735e-01 -3.34149331e-01
-1.41050434e+00 9.81743410e-02 2.57470936e-01 -2.50860482e-01
-5.81694245e-01 1.09082913e+00 -3.13917130e-01 -1.73048809e-01
9.14624333e-01 -1.34082392e-01 -2.17228338e-01 8.19783062e-02
6.78369284e-01 5.07402539e-01 5.62208295e-01 -5.73575795e-01
-4.88441586e-01 3.14373702e-01 -3.87017354e-02 4.29303572e-02
1.27537489e+00 -2.50994861e-01 3.26503843e-01 3.61371368e-01
1.52773869e+00 1.38172626e-01 -1.30616152e+00 -6.37663007e-01
4.95573645e-03 -4.60922033e-01 5.62786520e-01 -7.01655865e-01
-9.67866778e-01 6.60847604e-01 1.02938247e+00 2.97566056e-01
1.61937809e+00 -3.65643978e-01 1.10891402e+00 2.92789489e-01
-1.75075326e-02 -1.62141669e+00 1.71432331e-01 4.95060235e-01
7.45363414e-01 -9.91847992e-01 -9.42585841e-02 -2.88462341e-01
-4.15305614e-01 9.43214953e-01 2.73943275e-01 2.09876761e-01
9.98793423e-01 6.09579861e-01 -2.00553268e-01 1.96102977e-01
-3.21938932e-01 -2.44709522e-01 5.13729870e-01 5.25251865e-01
1.47983013e-02 -1.54307246e-01 -2.14710131e-01 8.52527022e-01
-2.35057727e-01 -3.53187710e-01 1.61222562e-01 6.80512011e-01
-3.58431786e-01 -8.05984259e-01 -2.66050190e-01 7.41179228e-01
-8.05533946e-01 -9.63889882e-02 -5.92780747e-02 -6.08518627e-03
8.22649524e-02 1.27009630e+00 -2.73570806e-01 -9.22752738e-01
7.06022382e-01 -6.27947301e-02 3.19448323e-03 -2.55908847e-01
-5.33838272e-01 8.21792856e-02 -3.57578471e-02 -6.47200286e-01
-1.29110917e-01 -4.86315161e-01 -9.57447171e-01 1.24530680e-01
-7.55689681e-01 1.44351199e-01 1.03027523e+00 1.04908943e+00
2.72723585e-01 7.70410001e-01 1.22059083e+00 -5.98632574e-01
-9.48498785e-01 -7.44817674e-01 -7.23750651e-01 2.01301485e-01
4.38729137e-01 -5.07820487e-01 -5.50126076e-01 -2.91476935e-01] | [14.577825546264648, 6.050532817840576] |
08456197-4d76-4e1a-8e07-2138ea01a454 | bivariate-beta-lstm | 1905.10521 | null | https://arxiv.org/abs/1905.10521v3 | https://arxiv.org/pdf/1905.10521v3.pdf | Bivariate Beta-LSTM | Long Short-Term Memory (LSTM) infers the long term dependency through a cell state maintained by the input and the forget gate structures, which models a gate output as a value in [0,1] through a sigmoid function. However, due to the graduality of the sigmoid function, the sigmoid gate is not flexible in representing multi-modality or skewness. Besides, the previous models lack modeling on the correlation between the gates, which would be a new method to adopt inductive bias for a relationship between previous and current input. This paper proposes a new gate structure with the bivariate Beta distribution. The proposed gate structure enables probabilistic modeling on the gates within the LSTM cell so that the modelers can customize the cell state flow with priors and distributions. Moreover, we theoretically show the higher upper bound of the gradient compared to the sigmoid function, and we empirically observed that the bivariate Beta distribution gate structure provides higher gradient values in training. We demonstrate the effectiveness of bivariate Beta gate structure on the sentence classification, image classification, polyphonic music modeling, and image caption generation. | ['Il-Chul Moon', 'JoonHo Jang', 'Seung jae Shin', 'Kyungwoo Song'] | 2019-05-25 | null | null | null | null | ['music-modeling'] | ['music'] | [ 1.05319194e-01 1.20406106e-01 -1.61988571e-01 -3.62974137e-01
-2.95250535e-01 -4.21207368e-01 5.97061932e-01 -2.72111874e-02
-4.29711670e-01 8.56063485e-01 2.93441772e-01 -2.94296890e-01
1.08598046e-01 -1.10451853e+00 -1.00244284e+00 -1.00158143e+00
6.23659603e-02 1.02974288e-03 2.11342767e-01 -5.32699190e-02
2.49509633e-01 -5.86226396e-02 -1.21105015e+00 6.34820402e-01
5.08721292e-01 1.04625130e+00 5.91722310e-01 7.62055933e-01
-2.49744117e-01 7.10407019e-01 -6.46470010e-01 -3.39571744e-01
-2.09509835e-01 -5.81156850e-01 -4.06867802e-01 -3.88898611e-01
-1.00176059e-01 -1.19017720e-01 -4.21152949e-01 1.02935183e+00
5.86850464e-01 4.08577509e-02 7.61022866e-01 -1.03807557e+00
-1.00638282e+00 9.62600470e-01 -1.68443203e-01 7.86131695e-02
1.77702427e-01 1.69881076e-01 8.77547562e-01 -6.06141627e-01
2.48519421e-01 1.07288623e+00 6.29500508e-01 7.17220962e-01
-9.91470814e-01 -4.70156312e-01 4.74232644e-01 4.04684953e-02
-1.39733553e+00 -5.26566710e-03 6.33031547e-01 -3.20332795e-01
9.15740252e-01 4.00807112e-02 7.93413699e-01 1.42655349e+00
5.97893775e-01 6.83070779e-01 1.15335059e+00 -5.71494281e-01
1.31910309e-01 3.05604517e-01 5.12712123e-03 6.31231070e-01
-1.82526320e-01 1.67533353e-01 -8.00311625e-01 -5.62858256e-03
1.15663671e+00 -8.13317224e-02 -2.26967931e-01 1.87750131e-01
-1.25286007e+00 5.30562162e-01 5.73829830e-01 4.20388281e-01
-2.51739681e-01 6.04093432e-01 1.80315211e-01 2.53505617e-01
1.55761778e-01 2.32173935e-01 -4.17441785e-01 -2.05052108e-01
-8.07556570e-01 -1.64932549e-01 7.23004341e-01 8.55216682e-01
4.60043073e-01 1.27156377e-01 -6.23464644e-01 6.84736490e-01
5.44998229e-01 4.79319423e-01 6.38108790e-01 -4.09718752e-01
4.16192949e-01 1.32295117e-01 -5.78365996e-02 -8.49911809e-01
-2.00048313e-01 -8.53329241e-01 -8.30538094e-01 -1.02304772e-01
5.29097140e-01 -3.00545432e-02 -1.03604305e+00 2.23525357e+00
-3.47880751e-01 2.11536616e-01 1.70467243e-01 6.99615777e-01
7.57125020e-01 8.21955502e-01 2.90099174e-01 -2.04062238e-01
1.27168977e+00 -8.10211539e-01 -9.62840855e-01 -3.43161434e-01
4.12654579e-01 -6.43798828e-01 1.54039919e+00 3.24357778e-01
-8.99800956e-01 -6.78388298e-01 -1.07252359e+00 -9.68596041e-02
-5.00312746e-01 1.85550407e-01 7.95451880e-01 7.08889961e-01
-1.11238372e+00 5.99627674e-01 -7.69727468e-01 -2.39199221e-01
2.35117525e-02 3.85412753e-01 1.35663316e-01 2.66590178e-01
-1.56896281e+00 9.20238197e-01 5.27946711e-01 4.13864881e-01
-6.74434066e-01 -3.10222477e-01 -6.25397027e-01 3.40134770e-01
-3.41432989e-01 -1.01838160e+00 8.79731834e-01 -9.54730511e-01
-1.71757329e+00 5.86272776e-01 -1.74395800e-01 -6.76459372e-01
2.73581088e-01 2.03339875e-01 -2.35370159e-01 -2.07302988e-01
-5.39796889e-01 1.10678291e+00 9.85588431e-01 -1.18390930e+00
-2.59851336e-01 -8.44400674e-02 2.03656450e-01 2.36973315e-01
-6.19254708e-01 -4.15442884e-01 -2.52913982e-01 -8.45296443e-01
2.63962805e-01 -8.66173983e-01 7.15620369e-02 -2.28154540e-01
-2.80509263e-01 3.97149101e-02 1.94836959e-01 -5.37009716e-01
1.64036715e+00 -2.39790750e+00 -2.10533485e-01 2.37122208e-01
-3.88001174e-01 -3.26343209e-01 -4.06686962e-02 3.07893574e-01
-6.99229632e-03 1.54115856e-01 -2.22548902e-01 -1.70243487e-01
7.10620880e-02 4.57855701e-01 -8.23498368e-01 -3.00560091e-02
9.38257426e-02 8.62767279e-01 -8.92465115e-01 -2.79337198e-01
-1.55438602e-01 6.66272461e-01 -5.41202962e-01 -1.10956980e-02
-2.69882560e-01 3.42457891e-01 -1.09463006e-01 3.48836511e-01
5.58714211e-01 -2.54488498e-01 2.92758774e-02 -1.78403214e-01
3.16153318e-02 3.94862413e-01 -9.13846672e-01 1.79026806e+00
-5.68326592e-01 4.80339080e-01 -2.64522612e-01 -5.10149896e-01
1.14700067e+00 4.80495781e-01 -8.22299719e-02 -4.81369078e-01
2.12964892e-01 1.38281316e-01 1.58470318e-01 -2.76309282e-01
4.83849734e-01 -4.64409620e-01 -1.24786440e-02 3.35020661e-01
2.22966582e-01 -1.28220156e-01 3.79643999e-02 2.80549712e-02
4.95707273e-01 1.52704611e-01 -1.53451875e-01 -2.17937484e-01
2.92086154e-01 -7.23895073e-01 3.86310428e-01 1.00722766e+00
1.38993248e-01 8.97418201e-01 6.45035565e-01 -2.35395774e-01
-7.12354481e-01 -1.25001419e+00 -1.96513280e-01 1.10676074e+00
2.39485260e-02 -2.82369286e-01 -7.59879947e-01 -2.48980552e-01
-2.94721156e-01 9.48767066e-01 -7.37021506e-01 -5.22764206e-01
-2.98679650e-01 -1.04815626e+00 7.69901752e-01 6.69316828e-01
6.32218480e-01 -1.23724091e+00 -3.26841801e-01 2.63364345e-01
-4.26831186e-01 -9.11312461e-01 -6.27635658e-01 4.84558582e-01
-1.04781115e+00 -4.04281765e-01 -7.00531125e-01 -9.75594044e-01
8.14990342e-01 -4.73603547e-01 8.86086941e-01 -2.14280784e-01
3.66342425e-01 1.36751756e-01 -1.29448041e-01 -4.16979849e-01
-4.12218809e-01 2.28575513e-01 -1.65424481e-01 2.74810679e-02
1.21701889e-01 -7.30862558e-01 -5.44501364e-01 1.57636479e-02
-9.36530948e-01 3.91153276e-01 5.07790506e-01 1.13306463e+00
5.50498724e-01 -9.83531699e-02 7.84133673e-01 -6.04806960e-01
7.19272196e-01 -3.22305590e-01 -1.51730761e-01 3.30680311e-01
-5.65315664e-01 4.55174416e-01 4.51400697e-01 -8.03520143e-01
-1.17465651e+00 -1.09776400e-01 -4.49946672e-01 -1.46942377e-01
2.33678058e-01 6.58042848e-01 -7.26416931e-02 2.61939317e-01
5.18262327e-01 5.03442347e-01 -2.16684073e-01 -1.29062876e-01
3.57527494e-01 3.85465086e-01 4.86654311e-01 -7.09359765e-01
2.10916251e-01 3.71060878e-01 -1.08215384e-01 -4.09042567e-01
-7.90476501e-01 2.91550875e-01 -3.80927503e-01 -1.45903707e-01
9.04698074e-01 -8.77314925e-01 -6.61632955e-01 6.50128782e-01
-1.28975356e+00 -3.63602549e-01 -3.51311415e-01 7.23257422e-01
-6.45078003e-01 -9.56147239e-02 -1.01945257e+00 -8.95676792e-01
-2.81071782e-01 -8.48451018e-01 6.97489202e-01 3.58971894e-01
-2.88975358e-01 -1.34037638e+00 -3.28686446e-01 -1.70861691e-01
5.81230521e-01 -3.64153609e-02 1.22984695e+00 -1.14644609e-01
-4.72213805e-01 -5.05282171e-02 9.09156650e-02 4.84265029e-01
-8.44330043e-02 -4.35647294e-02 -1.37950826e+00 -2.09228799e-01
2.93650299e-01 -4.25670668e-02 1.27759528e+00 6.14269197e-01
1.25869334e+00 -2.39791974e-01 -1.76913403e-02 6.97785795e-01
1.28792477e+00 3.18239808e-01 1.07140875e+00 1.67387709e-01
4.39421535e-01 1.68441176e-01 1.68823585e-01 4.13763076e-01
3.33465159e-01 1.63748965e-01 2.86107242e-01 -4.72523971e-03
-4.97679412e-02 -7.07589149e-01 6.31002724e-01 1.06481087e+00
7.73233250e-02 -4.17460054e-01 -5.92690408e-01 4.49357390e-01
-1.64660966e+00 -9.40385878e-01 1.31969258e-01 2.33971238e+00
1.16708815e+00 6.48309410e-01 -5.26294708e-01 6.56715557e-02
7.07795501e-01 1.51680574e-01 -2.78518289e-01 -6.29356742e-01
-5.32069325e-01 1.91297024e-01 5.04853249e-01 5.98741949e-01
-7.12735534e-01 8.99022520e-01 6.99231911e+00 1.04025567e+00
-1.45780277e+00 -2.82971580e-02 8.19062233e-01 -7.14422669e-03
-5.65087974e-01 4.30116802e-02 -9.40953493e-01 6.76125467e-01
8.57614100e-01 5.61020747e-02 4.18558806e-01 3.22659343e-01
1.41566336e-01 -3.12913388e-01 -1.05218601e+00 9.60683525e-01
6.46740384e-03 -9.50175941e-01 5.96995533e-01 -1.16047718e-01
7.44969189e-01 -2.36206159e-01 5.17390072e-01 4.11278218e-01
-1.13707311e-01 -1.11696482e+00 9.89085853e-01 9.66186106e-01
7.08424032e-01 -5.82872212e-01 6.72504187e-01 5.13172567e-01
-9.82314289e-01 -3.19325894e-01 -4.34609860e-01 -2.91260809e-01
1.24249801e-01 6.64828718e-01 -8.26386034e-01 1.76205739e-01
3.05329531e-01 2.58499354e-01 -5.04175544e-01 8.00581932e-01
-4.88034278e-01 7.45915830e-01 -1.93768680e-01 -1.46162927e-01
2.17462257e-02 -1.30196899e-01 2.87928432e-01 1.37034881e+00
6.72138393e-01 -2.44548410e-01 -4.84251939e-02 1.07063019e+00
-4.47669160e-03 -2.49665067e-01 -4.97009873e-01 -1.10535964e-01
4.21541840e-01 8.73051167e-01 -7.40992725e-01 -2.23977640e-01
-9.84567106e-02 8.73308718e-01 -9.22840908e-02 7.44962096e-01
-1.12437248e+00 -3.13454211e-01 3.93061573e-03 9.96116772e-02
2.79493570e-01 -2.73208082e-01 -9.65719402e-01 -9.67452765e-01
1.15818325e-02 -3.27733427e-01 3.79999667e-01 -1.02695465e+00
-1.28070736e+00 7.21355379e-01 -8.98755863e-02 -1.10808349e+00
-2.30088294e-01 -4.86165702e-01 -5.76988459e-01 1.12286806e+00
-1.38542867e+00 -1.25969315e+00 2.19492111e-02 5.51281691e-01
1.02860115e-01 1.04692817e-01 1.03047001e+00 1.97432175e-01
-2.10780293e-01 6.39492810e-01 -1.55797288e-01 -2.64857784e-02
6.94729805e-01 -1.15282559e+00 1.07299231e-01 5.50995767e-01
7.58507028e-02 1.01475430e+00 6.82761669e-01 -5.02810955e-01
-8.44319403e-01 -8.95102084e-01 1.11678612e+00 -1.85905010e-01
4.35576349e-01 -5.24374127e-01 -9.23320472e-01 6.13220334e-01
3.07581246e-01 -8.84606093e-02 5.12950182e-01 -3.34929042e-02
-2.90123701e-01 -3.10177177e-01 -8.78592312e-01 6.52685583e-01
7.26649046e-01 -7.88901210e-01 -5.05126655e-01 -1.41952820e-02
6.08076334e-01 -4.27902967e-01 -6.32844150e-01 3.49344045e-01
7.34781921e-01 -7.82890379e-01 6.75783813e-01 -9.26565826e-02
2.97105610e-01 -3.50235254e-01 -1.82589561e-01 -1.41437995e+00
-2.71439403e-01 -3.00354630e-01 -1.91963777e-01 1.22812819e+00
8.47076833e-01 -7.38556802e-01 4.69243437e-01 4.51770306e-01
-7.61786327e-02 -9.40244377e-01 -9.70211923e-01 -6.56262398e-01
1.60840556e-01 -4.25321549e-01 4.23198640e-01 5.81767678e-01
-1.10929728e-01 1.72532573e-01 -2.89173365e-01 6.45989105e-02
1.31879643e-01 -8.46317783e-02 1.16043657e-01 -6.48180902e-01
-7.13330925e-01 -4.53841329e-01 -1.30659923e-01 -1.42477417e+00
-5.90322316e-02 -9.35364962e-01 1.12486906e-01 -1.59796953e+00
2.80098349e-01 -3.71992618e-01 -8.16845536e-01 4.69485015e-01
-1.30147576e-01 1.11067280e-01 8.13621730e-02 -1.35411605e-01
-7.47015849e-02 7.38615274e-01 1.44378579e+00 -1.40610069e-01
-2.49507889e-01 2.28797570e-02 -6.13305449e-01 6.75676763e-01
7.67836332e-01 -5.46254933e-01 -5.78447461e-01 -7.16518581e-01
5.90971112e-01 -2.18385998e-02 4.32085842e-01 -1.05636954e+00
3.88821006e-01 2.14846376e-02 5.63090801e-01 -4.88340467e-01
4.79062200e-01 -4.61297750e-01 3.66535001e-02 4.39107537e-01
-5.40497363e-01 6.10831566e-02 1.50588393e-01 5.35408139e-01
-3.65315199e-01 -2.34458715e-01 6.66426003e-01 -2.49704853e-01
-3.75165761e-01 4.61965539e-02 -4.42533463e-01 -2.03523234e-01
5.12451410e-01 -3.14746320e-01 -9.13371667e-02 -5.21928608e-01
-1.04085970e+00 5.21073490e-02 2.47840267e-02 4.65572476e-01
6.50221586e-01 -1.37473774e+00 -3.46184611e-01 4.67168123e-01
-2.30078995e-01 -2.65759766e-01 1.46957919e-01 7.92653263e-01
-1.81168854e-01 2.40177974e-01 -2.43096605e-01 -6.48069501e-01
-7.80131936e-01 1.98330313e-01 3.92318577e-01 -3.23688775e-01
-2.20703743e-02 9.55374539e-01 2.66807288e-01 -2.73668081e-01
3.52928698e-01 -8.16068351e-01 -2.48835668e-01 1.26414716e-01
2.99734861e-01 -1.14796013e-01 -4.01273444e-02 -1.16903663e-01
-6.56917617e-02 3.12317759e-01 2.10551322e-01 -6.03546858e-01
9.97510314e-01 -1.83200955e-01 -2.16572553e-01 1.15545654e+00
8.28914702e-01 -2.45368361e-01 -1.32549691e+00 7.89034739e-02
-2.01379374e-01 -8.39384869e-02 -6.19082246e-03 -1.00784552e+00
-7.02323556e-01 1.15222466e+00 5.68464637e-01 4.10761386e-02
1.16671610e+00 -2.60049790e-01 7.32895792e-01 2.96420306e-01
3.81792068e-01 -1.12138438e+00 1.00095317e-01 9.88009632e-01
8.94719839e-01 -6.41246378e-01 -5.23818612e-01 -1.04894847e-01
-5.37795305e-01 1.31298196e+00 5.38199604e-01 8.86475220e-02
7.16372430e-01 5.01570523e-01 7.40353912e-02 2.34124899e-01
-9.67201948e-01 2.72539780e-02 1.25595674e-01 3.19197565e-01
7.33199954e-01 5.63533045e-02 -3.98712546e-01 9.17892575e-01
-4.53530729e-01 8.13892931e-02 5.03339708e-01 7.79534757e-01
-2.99508244e-01 -1.06964779e+00 -2.57969558e-01 2.66032577e-01
-3.81590307e-01 -5.05106628e-01 9.27934274e-02 3.47144902e-01
3.93964440e-01 7.61263907e-01 3.35449010e-01 -2.45367110e-01
7.76414648e-02 4.71090198e-01 7.45440781e-01 -5.71372688e-01
-6.67440474e-01 1.26008525e-01 -1.85198009e-01 1.49839699e-01
-1.13873728e-01 -1.84466079e-01 -1.50502634e+00 6.52705655e-02
-5.20769000e-01 9.01617408e-02 7.61410534e-01 8.70411158e-01
2.63915151e-01 7.46248305e-01 3.83000821e-01 -3.85832071e-01
-6.26708567e-01 -1.19424868e+00 -6.44236386e-01 2.68364966e-01
8.79464522e-02 -4.07469034e-01 -3.27041775e-01 2.77578235e-01] | [10.760590553283691, 6.490073204040527] |
ae516642-6e5c-41da-8e31-c61aa9bc40da | bridgeformer-bridging-video-text-retrieval | 2201.04850 | null | https://arxiv.org/abs/2201.04850v2 | https://arxiv.org/pdf/2201.04850v2.pdf | Bridging Video-text Retrieval with Multiple Choice Questions | Pre-training a model to learn transferable video-text representation for retrieval has attracted a lot of attention in recent years. Previous dominant works mainly adopt two separate encoders for efficient retrieval, but ignore local associations between videos and texts. Another line of research uses a joint encoder to interact video with texts, but results in low efficiency since each text-video pair needs to be fed into the model. In this work, we enable fine-grained video-text interactions while maintaining high efficiency for retrieval via a novel pretext task, dubbed as Multiple Choice Questions (MCQ), where a parametric module BridgeFormer is trained to answer the "questions" constructed by the text features via resorting to the video features. Specifically, we exploit the rich semantics of text (i.e., nouns and verbs) to build questions, with which the video encoder can be trained to capture more regional content and temporal dynamics. In the form of questions and answers, the semantic associations between local video-text features can be properly established. BridgeFormer is able to be removed for downstream retrieval, rendering an efficient and flexible model with only two encoders. Our method outperforms state-of-the-art methods on the popular text-to-video retrieval task in five datasets with different experimental setups (i.e., zero-shot and fine-tune), including HowTo100M (one million videos). We further conduct zero-shot action recognition, which can be cast as video-to-text retrieval, and our approach also significantly surpasses its counterparts. As an additional benefit, our method achieves competitive results with much shorter pre-training videos on single-modality downstream tasks, e.g., action recognition with linear evaluation. | ['Ping Luo', 'XiaoHu Qie', 'Ying Shan', 'Dian Li', 'Xihui Liu', 'Yixiao Ge', 'Yuying Ge'] | 2022-01-13 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Ge_Bridging_Video-Text_Retrieval_With_Multiple_Choice_Questions_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Ge_Bridging_Video-Text_Retrieval_With_Multiple_Choice_Questions_CVPR_2022_paper.pdf | cvpr-2022-1 | ['zero-shot-action-recognition', 'video-text-retrieval', 'text-to-video-search'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 1.19668171e-01 -5.30525923e-01 -5.22337437e-01 -1.95885420e-01
-1.23232996e+00 -6.27561092e-01 8.57807159e-01 -3.90160680e-01
-6.14994824e-01 3.64854574e-01 5.82720160e-01 1.00336611e-01
-1.05168089e-01 -4.34978634e-01 -9.05152082e-01 -6.85293317e-01
2.06324220e-01 2.94917017e-01 3.24875861e-01 -1.59993902e-01
2.33949840e-01 1.22457668e-02 -1.70280004e+00 6.27498090e-01
4.45903808e-01 1.16717267e+00 4.85872030e-01 6.81114852e-01
-8.32348987e-02 9.29896474e-01 -3.87963414e-01 -4.19843286e-01
1.48733348e-01 -4.49380994e-01 -7.47900486e-01 9.73224416e-02
6.12548292e-01 -9.38781023e-01 -1.00591838e+00 5.56779683e-01
6.17475510e-01 4.47699904e-01 7.66528606e-01 -1.10792482e+00
-8.65636945e-01 3.83568376e-01 -4.09185112e-01 1.98622927e-01
6.87435031e-01 2.74551660e-01 1.37092650e+00 -1.21585643e+00
8.70397687e-01 1.43696356e+00 1.90707490e-01 4.91673648e-01
-8.31780374e-01 -6.00564837e-01 3.14850688e-01 6.49597943e-01
-1.48448467e+00 -7.69215405e-01 4.69995797e-01 -3.12589556e-01
1.12393892e+00 3.05440843e-01 5.44313312e-01 1.48422778e+00
1.86019167e-01 1.27753532e+00 3.91997993e-01 -1.57751143e-01
-1.75760046e-01 -2.25642785e-01 -1.99768752e-01 6.44172132e-01
-3.23166311e-01 -1.85180172e-01 -9.67467904e-01 1.31480977e-01
7.82113612e-01 4.16974306e-01 -6.27078831e-01 -9.53156427e-02
-1.50444794e+00 8.43700707e-01 1.89823821e-01 3.79584372e-01
-2.85935521e-01 3.68253380e-01 7.16156840e-01 6.70696735e-01
3.66088867e-01 1.12821653e-01 -4.17429686e-01 -4.07768667e-01
-1.02613330e+00 6.22491874e-02 6.88772500e-01 1.12057579e+00
7.26686656e-01 -1.56541914e-01 -6.94425583e-01 8.79846692e-01
4.76106882e-01 6.54362381e-01 8.08884621e-01 -8.69394600e-01
7.50207424e-01 2.65370548e-01 -9.19565558e-02 -9.98359978e-01
1.45807043e-01 7.63359740e-02 -6.11215174e-01 -6.93658173e-01
6.20137118e-02 1.09749354e-01 -1.03380549e+00 1.50593519e+00
1.02904767e-01 4.16892827e-01 2.51267329e-02 1.09933734e+00
1.06642032e+00 9.53318954e-01 -5.71891777e-02 -2.21508697e-01
1.37152898e+00 -1.39009368e+00 -8.09940338e-01 9.89801902e-03
7.98690975e-01 -7.87351966e-01 1.07451975e+00 1.23301342e-01
-1.10178745e+00 -5.50978780e-01 -5.45926094e-01 -4.58470911e-01
-3.09088409e-01 1.63751155e-01 3.61267984e-01 -4.42072973e-02
-1.14405560e+00 4.15024281e-01 -7.40397990e-01 -4.76748586e-01
2.59194851e-01 2.47896403e-01 -6.11470520e-01 -4.23620105e-01
-1.42621779e+00 6.04281127e-01 2.92068660e-01 -4.28554369e-03
-1.23456132e+00 -4.67668951e-01 -9.68299329e-01 2.10145891e-01
7.06199110e-01 -8.84203553e-01 1.13223803e+00 -1.05364454e+00
-1.61666489e+00 7.23437488e-01 -2.37945989e-01 -1.56214207e-01
3.63108397e-01 -4.28096265e-01 -2.22449973e-01 9.64403152e-01
2.13962078e-01 8.64908338e-01 1.38621604e+00 -7.02782333e-01
-5.99886596e-01 -1.31142735e-01 5.22494972e-01 4.32104021e-01
-7.97334373e-01 3.29984814e-01 -1.36972129e+00 -1.01324952e+00
-2.52779782e-01 -9.77550805e-01 2.75323868e-01 2.13097095e-01
-2.11146772e-02 -3.97078961e-01 9.51459825e-01 -7.10384667e-01
1.36584282e+00 -2.29884601e+00 6.62625790e-01 -2.63970375e-01
1.54075369e-01 1.09494053e-01 -6.20242894e-01 7.49748468e-01
7.47327432e-02 1.06878273e-01 1.67879805e-01 -4.60612744e-01
1.54444844e-01 2.60935694e-01 -3.39053333e-01 4.59839463e-01
8.89687762e-02 1.28416955e+00 -8.37396979e-01 -8.34857404e-01
2.28685841e-01 5.19659519e-01 -6.99939251e-01 4.55050707e-01
-2.35962480e-01 2.06567630e-01 -7.88748682e-01 7.12108731e-01
1.52971372e-01 -5.29124200e-01 5.90298846e-02 -3.56709659e-01
2.14922890e-01 5.66743389e-02 -7.57339239e-01 2.08784199e+00
-4.61246729e-01 7.41943181e-01 4.59532700e-02 -1.24219453e+00
3.55702341e-01 5.54886162e-01 7.00446367e-01 -9.01842713e-01
8.01283866e-03 1.59259275e-01 -5.59938610e-01 -9.51281548e-01
4.60732251e-01 1.54909730e-01 -1.70272067e-01 3.44099760e-01
4.11956787e-01 2.08911955e-01 2.83899605e-01 4.87958670e-01
1.15545344e+00 5.15316188e-01 2.57060137e-02 2.16290042e-01
5.67575574e-01 -2.91555256e-01 1.81686059e-01 7.12549627e-01
1.05155081e-01 6.45446777e-01 1.83709547e-01 1.65466089e-02
-5.82992256e-01 -7.35693812e-01 1.30490422e-01 1.60270357e+00
3.53277922e-01 -7.26603389e-01 -4.18434530e-01 -6.59975648e-01
-5.69853261e-02 7.23678470e-02 -4.83697593e-01 -2.69112319e-01
-6.41012728e-01 -1.73015490e-01 5.32873631e-01 5.18698633e-01
5.14408529e-01 -1.17797923e+00 -3.65800560e-01 9.86834168e-02
-6.85795486e-01 -1.32892728e+00 -1.03590536e+00 -1.57470018e-01
-7.38887310e-01 -1.07691658e+00 -1.15025926e+00 -8.77592862e-01
2.66405940e-01 7.26528525e-01 9.99195933e-01 2.49193832e-01
-1.24151260e-01 1.12315547e+00 -9.87470388e-01 3.17396522e-01
1.93435848e-01 -4.59000394e-02 -1.53861910e-01 3.38214248e-01
2.29526758e-01 -2.84407824e-01 -9.10595417e-01 5.63110530e-01
-1.32837796e+00 -1.21592708e-01 6.93409443e-01 1.10555410e+00
5.84070146e-01 -3.16430956e-01 4.03661102e-01 -4.24964249e-01
3.18736464e-01 -7.68891394e-01 -3.94920520e-02 5.38730145e-01
-1.63997989e-02 7.70526305e-02 6.14024460e-01 -6.60249352e-01
-1.02114916e+00 -1.72859833e-01 6.49768934e-02 -1.08713877e+00
9.22681838e-02 7.83959091e-01 -1.14409409e-01 -4.56151320e-03
1.16359510e-01 4.02367026e-01 -1.51905775e-01 -5.04000366e-01
4.63413894e-01 6.86749458e-01 2.41199777e-01 -5.27211905e-01
7.13444352e-01 4.40615982e-01 -3.37952822e-01 -8.93704236e-01
-6.56495512e-01 -9.34775293e-01 -4.15206254e-01 -3.32476288e-01
9.77658570e-01 -1.23986137e+00 -5.22453725e-01 4.70886588e-01
-1.06323123e+00 -3.61600280e-01 4.01447713e-03 6.04434371e-01
-6.42752290e-01 7.23586082e-01 -9.51109290e-01 -3.38122159e-01
-3.07306290e-01 -1.16910326e+00 1.68349218e+00 -1.49710387e-01
6.78117946e-02 -9.22834814e-01 -2.45416969e-01 5.93778729e-01
2.66192198e-01 -2.94146180e-01 6.23514533e-01 -5.88225901e-01
-8.66787732e-01 -1.15197584e-01 -3.11613947e-01 1.96257874e-01
-3.20918933e-02 -3.75338234e-02 -7.49881566e-01 -5.71614921e-01
-6.23876825e-02 -8.06773961e-01 1.27231562e+00 2.30610564e-01
1.31423187e+00 -3.81720692e-01 -4.43452477e-01 5.48553407e-01
1.20351827e+00 1.59967281e-02 8.71504128e-01 2.06587330e-01
7.62821019e-01 5.09979010e-01 9.17482078e-01 4.70815808e-01
5.02924860e-01 9.41467643e-01 2.70322561e-01 2.52393633e-01
-2.21797645e-01 -2.90256649e-01 8.29307616e-01 1.20805860e+00
-9.55259874e-02 -6.06160581e-01 -4.59584326e-01 4.49244529e-01
-2.20484352e+00 -1.17157733e+00 3.11484009e-01 1.84552014e+00
7.24094212e-01 -3.20143282e-01 -1.66557714e-01 -2.31025890e-01
4.41245049e-01 4.88005728e-01 -4.47204202e-01 2.97428727e-01
4.37042378e-02 1.13213792e-01 2.48436425e-02 2.16369584e-01
-1.13387609e+00 1.06104565e+00 5.26378727e+00 1.26625824e+00
-1.10652459e+00 2.00170830e-01 1.49468675e-01 -4.59309161e-01
-3.18768173e-01 -7.14582577e-02 -7.33037889e-01 5.83927691e-01
8.83787334e-01 3.15817371e-02 4.57089782e-01 5.15180111e-01
2.19163537e-01 -2.05345408e-04 -1.26439822e+00 1.30414212e+00
5.80320954e-01 -1.25787592e+00 4.37607944e-01 -1.88959226e-01
5.09626925e-01 6.70697093e-02 -2.99940426e-02 7.57854939e-01
-2.38912508e-01 -7.85321236e-01 7.54256964e-01 7.74269044e-01
1.23785150e+00 -2.31554911e-01 5.59878349e-01 2.62079209e-01
-1.62248790e+00 -2.19404012e-01 -2.45323360e-01 3.19606721e-01
2.46117309e-01 2.67665014e-02 -4.01699133e-02 7.93944359e-01
8.57165337e-01 1.27646661e+00 -4.57512081e-01 8.14567804e-01
4.23302911e-02 3.47365797e-01 -2.68760055e-01 4.58532535e-02
4.67233658e-01 -6.53211698e-02 3.98246288e-01 1.23695612e+00
4.77137506e-01 2.47481003e-01 4.26285982e-01 2.13430718e-01
-3.43402594e-01 2.17935339e-01 -6.04379773e-01 -3.14526886e-01
2.89241999e-01 1.06144536e+00 -4.05032516e-01 -5.26934981e-01
-7.94870496e-01 1.36528456e+00 2.49919608e-01 7.46189117e-01
-1.03638554e+00 -2.68641442e-01 4.01060373e-01 -1.91375408e-02
5.99378884e-01 -6.01056293e-02 8.20003867e-01 -1.79525673e+00
2.83146411e-01 -1.11876297e+00 5.50332129e-01 -8.84019256e-01
-1.21866059e+00 3.71890277e-01 2.32922956e-01 -1.44830012e+00
-3.69622141e-01 -5.37587225e-01 -2.01384664e-01 4.48388308e-01
-1.66696739e+00 -1.33948457e+00 -3.39224905e-01 1.09820664e+00
1.10624087e+00 -8.36174116e-02 6.07051790e-01 7.32887208e-01
-4.68533605e-01 6.85642540e-01 1.52070254e-01 1.70425922e-01
1.20599854e+00 -6.89917445e-01 -1.34642601e-01 5.53134918e-01
3.65367740e-01 5.21349192e-01 2.26274449e-02 -5.24861932e-01
-1.98001337e+00 -1.16680455e+00 7.52251208e-01 -3.70057613e-01
8.80637288e-01 -2.21827880e-01 -7.22349763e-01 7.87306428e-01
2.70261407e-01 9.59186852e-02 5.59610963e-01 -2.86201417e-01
-3.99232149e-01 -5.36211506e-02 -4.52759773e-01 7.48328447e-01
1.21211004e+00 -1.01253605e+00 -5.94665468e-01 4.70316231e-01
9.75917339e-01 -2.86780387e-01 -8.71750593e-01 2.82393992e-01
8.08165729e-01 -6.80487394e-01 1.09491420e+00 -5.70970953e-01
6.77286029e-01 -3.68849598e-02 -4.14405793e-01 -8.66300583e-01
-2.12386057e-01 -7.52841234e-01 -4.86008883e-01 1.06899154e+00
9.60538313e-02 -4.11811799e-01 3.49523157e-01 3.08393985e-01
-1.82432115e-01 -7.77879298e-01 -9.34526145e-01 -6.41449153e-01
-2.50651956e-01 -2.84653991e-01 1.28560066e-01 8.24193895e-01
-6.21770211e-02 5.24607420e-01 -7.33797252e-01 -2.81729311e-01
1.02322511e-01 2.41106674e-01 7.31145561e-01 -8.28040361e-01
-5.40448666e-01 -3.81395489e-01 -2.90134072e-01 -1.92627215e+00
4.62540716e-01 -9.38216388e-01 7.43481740e-02 -1.39008415e+00
4.72792119e-01 5.31976894e-02 -2.80505717e-01 5.05336165e-01
-2.59013861e-01 4.05855656e-01 4.23416167e-01 5.92636228e-01
-1.21518004e+00 9.49270666e-01 1.50656009e+00 -3.58224332e-01
1.23658262e-01 -4.71274048e-01 -2.35248670e-01 4.46782261e-01
3.28275621e-01 -3.81014198e-01 -5.11120439e-01 -7.26012230e-01
1.82344303e-01 6.02223039e-01 4.51544553e-01 -6.19409919e-01
3.21661621e-01 -4.60570157e-02 1.61885068e-01 -5.10056317e-01
7.04566240e-01 -8.15805197e-01 -8.83002803e-02 -8.53610225e-03
-4.51480240e-01 1.76254176e-02 -1.04658753e-02 9.53134477e-01
-6.45007133e-01 -3.64717335e-01 1.77140608e-01 -1.94030657e-01
-9.93257165e-01 6.86861694e-01 -4.13397282e-01 1.15405366e-01
8.93993378e-01 -2.00399846e-01 -2.61685818e-01 -7.40773201e-01
-6.09538198e-01 5.93815863e-01 2.51609027e-01 7.20753729e-01
9.20540273e-01 -1.44529331e+00 -6.03698909e-01 -8.42858255e-02
3.42915475e-01 -2.96754628e-01 5.35821676e-01 1.02112210e+00
-1.60423413e-01 7.10135758e-01 1.21675812e-01 -7.29638934e-01
-1.33121240e+00 6.26707673e-01 8.26592930e-03 -3.09965104e-01
-7.56699920e-01 7.99675703e-01 5.73070109e-01 5.58723398e-02
4.51478988e-01 -1.10495605e-01 -3.58218551e-01 4.60041553e-01
5.35322011e-01 7.85534531e-02 -2.92949170e-01 -7.83278883e-01
-2.37122893e-01 8.67735326e-01 -2.13063166e-01 -1.18899094e-02
1.16993821e+00 -3.53252798e-01 2.67271921e-02 3.17274541e-01
1.74962032e+00 -3.63501251e-01 -1.16197419e+00 -3.95745486e-01
-2.65485287e-01 -6.02849305e-01 7.51825199e-02 -3.70942444e-01
-1.29113185e+00 9.41491067e-01 2.27685362e-01 -1.30538881e-01
1.20759070e+00 1.00748621e-01 1.11820924e+00 7.82982111e-01
3.68410110e-01 -1.15891266e+00 7.16359735e-01 4.98724282e-01
9.90088999e-01 -1.41354716e+00 -1.49870485e-01 -1.20013155e-01
-6.24592423e-01 1.19102585e+00 4.76307720e-01 2.64633838e-02
5.36248744e-01 -2.76640028e-01 -2.22509980e-01 -1.68704599e-01
-1.04502451e+00 -3.10970187e-01 5.86529672e-01 1.39300674e-01
2.24262580e-01 -3.17890704e-01 -1.37803823e-01 3.87637556e-01
5.21788597e-01 9.37215537e-02 3.36934365e-02 9.19248402e-01
-2.35078946e-01 -9.45222855e-01 -9.61471722e-02 6.17726922e-01
-4.70085353e-01 -3.33109826e-01 -1.86274856e-01 7.98373878e-01
-2.72302032e-01 9.41155672e-01 7.31204776e-03 -3.54254156e-01
1.46432295e-01 4.66644391e-02 4.19264793e-01 -6.13864541e-01
-4.85396653e-01 4.77036119e-01 9.42102745e-02 -1.07499015e+00
-8.27688038e-01 -4.71964419e-01 -9.20229793e-01 -1.03029519e-01
-5.25357068e-01 6.22603521e-02 1.42354086e-01 1.10751271e+00
5.29179156e-01 4.14156973e-01 6.15535855e-01 -1.04966629e+00
-5.52089989e-01 -8.74376535e-01 -3.67962331e-01 5.29624045e-01
3.85878921e-01 -7.84953475e-01 -4.42197591e-01 3.03967357e-01] | [10.304606437683105, 0.9250437617301941] |
640e7cf9-afca-45f4-8084-7540d31ac890 | compact-model-training-by-low-rank-projection | 2204.05566 | null | https://arxiv.org/abs/2204.05566v2 | https://arxiv.org/pdf/2204.05566v2.pdf | Compact Model Training by Low-Rank Projection with Energy Transfer | Low-rankness plays an important role in traditional machine learning, but is not so popular in deep learning. Most previous low-rank network compression methods compress the networks by approximating pre-trained models and re-training. However, the optimal solution in the Euclidean space may be quite different from the one in the low-rank manifold. A well-pre-trained model is not a good initialization for the model with low-rank constraints. Thus, the performance of a low-rank compressed network degrades significantly. Compared to other network compression methods such as pruning, low-rank methods attracts less attention in recent years. In this paper, we devise a new training method, low-rank projection with energy transfer (LRPET), that trains low-rank compressed networks from scratch and achieves competitive performance. First, we propose to alternately perform stochastic gradient descent training and projection onto the low-rank manifold. Compared to re-training on the compact model, this enables full utilization of model capacity since solution space is relaxed back to Euclidean space after projection. Second, the matrix energy (the sum of squares of singular values) reduction caused by projection is compensated by energy transfer. We uniformly transfer the energy of the pruned singular values to the remaining ones. We theoretically show that energy transfer eases the trend of gradient vanishing caused by projection. Third, we propose batch normalization (BN) rectification to cut off its effect on the optimal low-rank approximation of the weight matrix, which further improves the performance. Comprehensive experiments on CIFAR-10 and ImageNet have justified that our method is superior to other low-rank compression methods and also outperforms recent state-of-the-art pruning methods. Our code is available at https://github.com/BZQLin/LRPET. | ['Xiangmin Xu', 'Fang Liu', 'Xiaofen Xing', 'Zhenquan Lin', 'Kailing Guo'] | 2022-04-12 | null | null | null | null | ['low-rank-compression'] | ['computer-code'] | [ 1.8316190e-01 1.8989526e-01 -4.0042838e-01 -2.8684407e-01
-3.0512387e-01 -5.3609982e-02 3.0964038e-01 -3.0939299e-01
-7.3633265e-01 6.4889121e-01 3.2377955e-01 -2.2637258e-01
-4.3253082e-01 -6.7882907e-01 -1.0573913e+00 -7.0954722e-01
2.8787235e-02 5.2599704e-01 1.4878546e-01 2.5196031e-02
-6.3461192e-02 5.3637952e-01 -1.3615104e+00 5.2038491e-01
8.7868452e-01 8.4135741e-01 5.8753002e-01 2.1722621e-01
6.0778771e-02 6.9829839e-01 -3.2195514e-01 -4.4089356e-01
5.0375015e-01 -4.3365771e-01 -7.7487224e-01 -4.4659817e-01
7.0155185e-01 -5.4598719e-01 -1.0570688e+00 1.4682240e+00
2.2181639e-01 3.1808615e-01 3.8229671e-01 -8.2501864e-01
-4.2655924e-01 9.6926463e-01 -4.6539891e-01 1.3819775e-01
-2.0520788e-01 -3.6245406e-01 1.0939493e+00 -1.0942253e+00
6.8102366e-01 1.1479267e+00 8.2732838e-01 7.8696579e-01
-1.2419459e+00 -8.0815750e-01 1.9344661e-01 4.4188365e-01
-1.4987211e+00 -5.0501728e-01 1.0102202e+00 -8.1877522e-03
8.8112175e-01 3.6984801e-01 8.5393727e-01 8.1749403e-01
-2.2553252e-01 7.8831822e-01 7.1664184e-01 -3.7858194e-01
8.6361028e-02 -7.5320951e-03 1.3322775e-01 8.2103783e-01
4.3947712e-01 2.3346193e-02 -5.2518755e-01 -1.0916714e-01
8.3873284e-01 4.3748093e-01 -5.4213375e-01 -4.6096072e-01
-9.4098794e-01 6.5813142e-01 6.9489956e-01 4.6580204e-01
-3.0866140e-01 4.0748036e-01 2.0815341e-01 2.6824671e-01
3.6061540e-01 4.2862096e-01 -4.3896744e-01 -1.5164550e-01
-1.4320525e+00 -3.9993212e-02 6.9169128e-01 7.5574875e-01
6.0422963e-01 3.7024755e-02 7.2840978e-03 1.1891516e+00
1.7400584e-01 -3.8555001e-03 6.5467370e-01 -1.1244466e+00
7.0044637e-01 5.2250504e-01 -5.3735399e-01 -1.0541078e+00
-1.7548117e-01 -8.7185091e-01 -1.3613048e+00 -7.7367127e-02
2.4679515e-01 1.2819484e-01 -7.9615963e-01 1.9463346e+00
-4.1007411e-02 5.9493709e-01 9.7508594e-02 9.8339242e-01
6.8851137e-01 7.3703027e-01 -4.0485203e-01 -2.7492988e-01
8.9174253e-01 -1.1810461e+00 -5.4739481e-01 -2.5354642e-01
6.2659770e-01 -3.1564060e-01 1.1493696e+00 6.1399001e-01
-1.2035767e+00 -3.4211391e-01 -1.2516557e+00 -1.3215430e-01
-1.0001927e-03 2.9396895e-01 6.0801685e-01 2.6797161e-01
-1.0091699e+00 1.4228321e+00 -9.6304202e-01 1.1568424e-01
7.6291060e-01 5.1575524e-01 -4.8152450e-01 -5.2104938e-01
-1.1931351e+00 7.2834325e-01 5.3679711e-01 5.1868513e-02
-6.2028986e-01 -8.7707537e-01 -4.9486673e-01 3.4009808e-01
3.2254487e-01 -4.7213110e-01 9.1474491e-01 -7.6111966e-01
-1.4273458e+00 4.4737482e-01 -1.8010244e-01 -6.2677103e-01
4.8468331e-01 -5.1396936e-01 -6.4320281e-02 3.4144181e-01
-3.4683973e-01 6.9746256e-01 8.3410978e-01 -1.0172929e+00
-1.4235176e-01 -1.5272713e-01 -1.1192471e-02 2.8543913e-01
-1.0817068e+00 -2.7224806e-01 -7.8039050e-01 -7.5331360e-01
6.7917663e-01 -8.6059058e-01 -3.2804260e-01 9.8336130e-02
-2.7592248e-01 -2.0292399e-03 7.1174532e-01 -6.2042791e-01
1.5658017e+00 -2.1997635e+00 3.7203637e-01 3.4111452e-01
3.5418159e-01 4.9784437e-01 -2.8172803e-01 9.0441287e-02
-3.8122588e-01 2.6710492e-01 -2.8523394e-01 -6.0928816e-01
-2.3378958e-01 5.4598570e-01 -4.2103004e-01 4.7273752e-01
-6.6160575e-02 6.4505994e-01 -7.9154205e-01 -4.5787403e-01
-9.2492186e-02 7.7106166e-01 -9.8640436e-01 -2.7824217e-01
1.9010478e-01 -2.8835256e-02 -1.6016267e-01 2.7789927e-01
7.5910091e-01 -3.1599340e-01 1.3005060e-01 -6.4210379e-01
2.0182467e-01 5.3689307e-01 -1.1325108e+00 1.7751175e+00
-2.6879847e-01 4.7511646e-01 5.8479838e-02 -1.2643815e+00
5.1492608e-01 1.3048814e-01 6.0455132e-01 -5.6213647e-01
-2.9018408e-02 3.2932585e-01 9.9363208e-02 7.5182281e-02
3.2281676e-01 6.3266344e-02 5.4914331e-01 2.8114793e-01
2.0577718e-01 7.4786887e-02 4.3476918e-01 4.0071520e-01
1.1228286e+00 -6.4736664e-02 -1.6343695e-01 -1.4868400e-01
3.6144668e-01 -4.3142626e-01 8.0184895e-01 5.6073046e-01
1.5875690e-01 6.7937464e-01 5.1989877e-01 -3.0413458e-01
-1.0809067e+00 -7.2178894e-01 -4.2286113e-01 9.9910152e-01
-3.5911694e-02 -7.9549044e-01 -8.3661824e-01 -6.0151738e-01
-1.7667310e-01 5.1993543e-01 -2.2923332e-01 -5.6805146e-01
-9.0293306e-01 -8.6251253e-01 5.3383803e-01 4.5444435e-01
7.0225555e-01 -7.3118126e-01 -1.4613256e-01 3.4715176e-02
-1.6449848e-01 -1.0478511e+00 -4.0016019e-01 5.2296436e-01
-1.5880692e+00 -8.3492672e-01 -6.4378744e-01 -8.6092931e-01
1.0717943e+00 3.3963263e-01 9.6437722e-01 3.8963202e-01
2.0707227e-02 -1.8409367e-01 -1.8385530e-01 9.0652473e-02
-2.1094900e-01 1.9867499e-01 3.3695307e-01 -3.4879494e-01
2.0219518e-01 -9.4250464e-01 -5.3902614e-01 2.5064111e-01
-9.0888888e-01 2.7231517e-01 8.4832746e-01 1.0231097e+00
1.0355074e+00 4.9672842e-01 2.5365549e-01 -8.9623737e-01
4.7091320e-01 -5.8540691e-02 -4.5959875e-01 8.3386265e-03
-6.6633612e-01 3.3754003e-01 8.9330119e-01 -7.3802757e-01
-6.0998750e-01 2.2380295e-01 -6.7615345e-02 -1.0323038e+00
3.1766835e-01 6.4104456e-01 -1.2978221e-01 -1.0675506e-01
6.5697885e-01 7.5827532e-02 -8.0012284e-02 -8.6932212e-01
3.5869136e-01 9.3805358e-02 5.8255398e-01 -6.0180950e-01
1.0399995e+00 4.4763535e-01 1.5139008e-01 -7.4068367e-01
-9.5195055e-01 -3.4587404e-01 -5.9367645e-01 7.6527588e-02
3.0117598e-01 -9.2294854e-01 -1.4522491e-01 1.2437315e-01
-1.0456039e+00 -3.8226473e-01 -6.9679916e-01 8.4253412e-01
-1.8323284e-01 3.9495900e-01 -9.7635317e-01 -3.6076936e-01
-3.9900976e-01 -8.5973650e-01 4.3565196e-01 -9.3571998e-02
1.2442602e-01 -6.5619230e-01 -2.2631848e-02 1.8610290e-01
2.6991576e-01 -2.8238228e-01 9.3052661e-01 -5.3822923e-01
-4.9146605e-01 -2.5748911e-01 -1.7254709e-01 8.0080652e-01
-4.2749700e-01 -2.7037135e-01 -7.8710443e-01 -4.4813457e-01
3.1027123e-01 -2.8517726e-01 1.5097198e+00 3.2456607e-01
1.5843171e+00 -7.2818083e-01 -3.8273206e-01 1.0578027e+00
1.3293957e+00 -2.9171297e-01 6.6833824e-01 2.0671962e-01
9.6262282e-01 3.3562031e-01 2.6959532e-01 1.5052816e-01
-1.6729988e-01 5.3342837e-01 5.7309270e-01 -5.3221986e-02
-2.9232815e-01 -3.6161682e-01 4.7309721e-01 1.4099880e+00
-2.6804954e-01 2.1230201e-01 -6.7045951e-01 2.5174001e-01
-1.8957951e+00 -1.0535784e+00 7.9373337e-02 2.2752016e+00
1.0991466e+00 3.7597021e-01 -3.3047536e-01 3.6921382e-01
6.4053595e-01 2.1564674e-01 -6.1983907e-01 -7.8410290e-02
-2.2974131e-01 1.9699554e-01 6.8285358e-01 5.4685032e-01
-9.9876457e-01 8.7963349e-01 5.4525919e+00 1.1120820e+00
-1.0350839e+00 3.4035197e-01 6.9935727e-01 -7.4639171e-01
-2.6044545e-01 -6.7575574e-02 -8.5190648e-01 2.0583166e-01
8.9089054e-01 1.0689231e-01 9.1897893e-01 1.0877522e+00
5.7116531e-02 4.0325844e-01 -1.2612889e+00 1.1347495e+00
7.3373787e-02 -1.4786407e+00 2.0849074e-01 1.9031848e-01
1.0080158e+00 4.6997994e-01 1.4343877e-01 4.3105945e-01
3.0340686e-02 -1.0519952e+00 6.2290114e-01 3.8936326e-01
9.2166263e-01 -9.5255578e-01 6.0321069e-01 5.0442821e-01
-1.0609046e+00 -1.9312869e-01 -1.1104425e+00 1.5078081e-01
6.2412275e-03 1.0904974e+00 -3.5541952e-01 1.3972472e-01
7.8994900e-01 9.4387388e-01 -4.8136756e-01 9.6765733e-01
-2.6793355e-01 7.7679843e-01 -7.3221755e-01 4.2659786e-01
1.3545659e-01 -5.9910548e-01 6.1403668e-01 1.0611955e+00
3.5729450e-01 1.6124213e-01 -3.5781696e-02 8.8450408e-01
-6.9306201e-01 5.7754643e-02 -4.7505593e-01 -1.1017619e-01
3.0931476e-01 1.1632142e+00 -4.8250738e-01 -3.5129437e-01
-4.5696929e-01 9.2782384e-01 5.7644761e-01 4.5994496e-01
-7.7466875e-01 -2.5019598e-01 4.4413376e-01 3.5927847e-01
3.3658364e-01 -1.1452609e-01 -1.8022960e-01 -1.3841816e+00
4.0642306e-01 -7.9936689e-01 1.7796083e-01 -3.5594696e-01
-9.0797096e-01 6.1635566e-01 2.2088639e-02 -1.1637298e+00
2.8543487e-02 -5.1080179e-01 -3.9381179e-01 5.2121133e-01
-1.4839560e+00 -6.0792911e-01 -6.1926190e-02 5.3939927e-01
3.8093865e-01 -2.1951310e-01 5.5134380e-01 8.6880386e-01
-7.3998636e-01 8.8838434e-01 4.9753940e-01 9.3613483e-02
3.7129214e-01 -8.7049973e-01 -2.2067422e-02 9.2381781e-01
3.6771899e-01 8.8089550e-01 3.5079181e-01 -5.5769122e-01
-1.2447246e+00 -1.2555597e+00 7.7344376e-01 1.6808131e-01
5.6029427e-01 -3.1838325e-01 -1.2575788e+00 4.5340466e-01
-3.6356649e-01 2.4944513e-01 4.2320347e-01 -4.3751050e-02
-4.5378852e-01 -4.0375286e-01 -1.0077426e+00 7.4076933e-01
1.3620478e+00 -6.7497164e-01 -4.9761569e-01 5.8211493e-01
8.5159945e-01 -2.5033951e-01 -7.1593493e-01 6.3788182e-01
3.8757387e-01 -8.0543089e-01 1.0958517e+00 -5.2842164e-01
5.5258107e-01 -8.9244448e-02 -3.1812984e-01 -1.0514748e+00
-4.4147858e-01 -5.8997005e-01 -8.3989787e-01 8.7997067e-01
3.5172871e-01 -4.3271393e-01 1.3261298e+00 5.1238322e-01
-4.2787638e-01 -1.3045063e+00 -1.1202843e+00 -9.4734061e-01
1.5179394e-01 -4.1012451e-01 3.2967424e-01 9.5196050e-01
-1.9298658e-03 1.9622263e-01 -3.7301737e-01 -4.3550555e-02
5.5144376e-01 -1.5570895e-01 3.0427659e-01 -1.3206019e+00
-5.8536088e-01 -6.3220531e-01 -1.4024359e-01 -1.5883628e+00
2.6267415e-01 -1.3490211e+00 -1.8625090e-01 -1.4667912e+00
3.3774891e-01 -5.2055830e-01 -6.3685179e-01 6.0209560e-01
2.2035502e-01 3.9908126e-01 2.4402148e-01 6.1942476e-01
-5.7048124e-01 7.5743037e-01 1.2785593e+00 -4.5149252e-02
-3.1042457e-01 -1.0147416e-01 -5.2679563e-01 9.4756866e-01
9.5333654e-01 -8.0594718e-01 -4.4368902e-01 -5.5306327e-01
4.5065182e-01 -2.4837671e-01 1.8546966e-01 -1.3340422e+00
4.6641552e-01 1.7517547e-01 3.5594872e-01 -5.7757741e-01
4.1975316e-01 -9.2510796e-01 3.5130866e-02 6.8051738e-01
-3.7680203e-01 -1.0488432e-01 2.3584630e-02 5.2466732e-01
-2.8448632e-01 -7.0779711e-01 9.6936423e-01 -1.1899891e-01
-1.6580674e-01 5.6393993e-01 3.3851802e-02 -2.3833890e-02
4.3245742e-01 -3.1886995e-01 -2.2392043e-01 -2.0766531e-01
-6.9364190e-01 -1.2426434e-01 4.2991692e-01 1.3802779e-01
9.2702979e-01 -1.4948831e+00 -5.2146053e-01 2.6110905e-01
-5.3634214e-01 4.1852441e-01 8.2607768e-02 8.1250286e-01
-5.4512358e-01 3.4070465e-01 -1.5283373e-02 -4.0599835e-01
-1.2228408e+00 4.5176634e-01 3.2698518e-01 -6.0318166e-01
-8.6280197e-01 9.5117712e-01 1.3533466e-01 -2.9997435e-01
6.2062132e-01 -4.2542374e-01 -9.9449798e-02 -1.4113767e-01
5.1387566e-01 5.0261801e-01 1.2184743e-01 -5.6861991e-01
-1.3441150e-01 5.6879520e-01 -3.4304672e-01 -2.3931542e-02
1.7427430e+00 3.5905853e-01 -3.1215748e-01 1.3902080e-01
1.5944115e+00 -1.5764868e-01 -1.3964283e+00 -4.0571722e-01
-2.9447934e-01 -2.7496713e-01 4.3990073e-01 -2.4134712e-01
-1.6262422e+00 8.7751466e-01 4.4380593e-01 -3.0416930e-01
1.1339599e+00 -1.6972809e-01 8.3361334e-01 9.5848680e-01
1.9063875e-01 -1.2940708e+00 3.6623651e-01 7.7901244e-01
9.8644412e-01 -7.8408539e-01 5.0803989e-01 -3.5320765e-01
-1.8424027e-01 9.5481557e-01 6.2316358e-01 -4.1408256e-01
7.9265773e-01 9.2573151e-02 -5.6788462e-01 -8.9074969e-02
-7.1467704e-01 3.3700383e-01 4.0069631e-01 1.7863987e-01
2.1093583e-01 -2.2710544e-01 -1.9208774e-01 6.4251626e-01
-4.5464933e-01 -2.0604256e-01 3.0354923e-01 6.8068331e-01
-5.3289330e-01 -1.0820693e+00 -1.4051434e-01 8.7758875e-01
-5.5108565e-01 -4.5826182e-01 -2.2376652e-01 5.1172525e-01
1.5455861e-01 4.4695330e-01 6.4708013e-03 -5.1297504e-01
1.9821264e-01 -5.8880714e-03 4.6516782e-01 -6.0685468e-01
-3.4316945e-01 1.7188153e-01 -1.5660182e-01 -8.0772477e-01
-1.2082622e-01 -4.3611574e-01 -1.3606479e+00 -3.5362056e-01
-5.1671493e-01 2.4517299e-01 6.9548744e-01 4.7439876e-01
2.9407975e-01 5.2731466e-01 2.1731415e-01 -1.0054665e+00
-7.0432496e-01 -8.7258500e-01 -3.9687458e-01 4.4438574e-01
1.3438372e-01 -5.4026800e-01 -6.5281022e-01 -1.8133515e-01] | [8.496973037719727, 3.255361318588257] |
5d52943a-b54b-4883-9cfd-2d4be550e9b3 | sc2-supervised-compression-for-split | 2203.08875 | null | https://arxiv.org/abs/2203.08875v2 | https://arxiv.org/pdf/2203.08875v2.pdf | SC2 Benchmark: Supervised Compression for Split Computing | With the increasing demand for deep learning models on mobile devices, splitting neural network computation between the device and a more powerful edge server has become an attractive solution. However, existing split computing approaches often underperform compared to a naive baseline of remote computation on compressed data. Recent studies propose learning compressed representations that contain more relevant information for supervised downstream tasks, showing improved tradeoffs between compressed data size and supervised performance. However, existing evaluation metrics only provide an incomplete picture of split computing. This study introduces supervised compression for split computing (SC2) and proposes new evaluation criteria: minimizing computation on the mobile device, minimizing transmitted data size, and maximizing model accuracy. We conduct a comprehensive benchmark study using 10 baseline methods, three computer vision tasks, and over 180 trained models, and discuss various aspects of SC2. We also release sc2bench, a Python package for future research on SC2. Our proposed metrics and package will help researchers better understand the tradeoffs of supervised compression in split computing. | ['Stephan Mandt', 'Marco Levorato', 'Ruihan Yang', 'Yoshitomo Matsubara'] | 2022-03-16 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [ 4.85361814e-01 -1.45270318e-01 -7.37375200e-01 -7.78741837e-01
-8.27234149e-01 -1.80806339e-01 3.17613602e-01 3.75640988e-02
-6.81155264e-01 4.46579516e-01 1.21246874e-01 -3.73136938e-01
-5.97187579e-02 -7.08181441e-01 -8.65526497e-01 -5.64862311e-01
7.07143545e-02 4.18999463e-01 7.11752698e-02 6.35451555e-01
1.01599470e-01 3.59324723e-01 -1.58540595e+00 7.02447593e-01
5.19852042e-01 1.30115914e+00 4.52225894e-01 6.10243082e-01
3.57405066e-01 9.58837390e-01 -4.51521993e-01 -5.10631144e-01
2.42152780e-01 -1.89301506e-01 -7.90490985e-01 -3.52484435e-01
4.63040560e-01 -5.40926278e-01 -4.68000919e-01 9.91242588e-01
6.97860539e-01 1.05524652e-01 4.19220626e-01 -1.34270906e+00
-3.08600605e-01 1.05797184e+00 -1.62658095e-01 6.33969605e-01
-3.68576139e-01 -2.32321307e-01 8.70839357e-01 -6.97920799e-01
6.44313097e-01 6.10879600e-01 9.48352516e-01 4.47585553e-01
-9.11075771e-01 -7.43354619e-01 -1.19124174e-01 4.90958244e-01
-1.35809731e+00 -1.01481187e+00 4.60195184e-01 1.22681357e-01
1.51435292e+00 2.79105037e-01 6.56140804e-01 1.03270447e+00
7.34856650e-02 1.18559861e+00 7.52767980e-01 -2.62009799e-01
4.43083793e-01 9.95484442e-02 4.02356625e-01 6.13521397e-01
4.65673864e-01 -5.41576669e-02 -1.03392994e+00 -7.57726207e-02
3.42724949e-01 5.42604387e-01 -1.57255441e-01 4.70630005e-02
-7.14696705e-01 5.28202355e-01 3.11315119e-01 1.85219228e-01
-4.14915234e-01 5.80363512e-01 7.72035241e-01 3.95876557e-01
7.25912750e-01 1.20406948e-01 -7.33117223e-01 -7.48350382e-01
-1.50313926e+00 5.30707240e-02 7.97796547e-01 1.32304418e+00
4.88162637e-01 -4.82832864e-02 -8.96736681e-02 7.55902708e-01
-6.10167459e-02 1.54900730e-01 8.76397550e-01 -1.38135719e+00
6.68018043e-01 2.12663636e-01 -6.39703751e-01 -7.51778603e-01
-1.63469195e-01 -8.64760101e-01 -7.89546013e-01 -4.56642896e-01
-1.96117207e-01 -2.88287610e-01 -3.98606271e-01 1.47351038e+00
-1.89169317e-01 5.16571760e-01 7.00756684e-02 6.30734324e-01
7.98293352e-01 5.93886793e-01 -8.72806385e-02 -1.54957011e-01
7.53838301e-01 -1.67773664e+00 -5.38908124e-01 -2.54602194e-01
1.32868993e+00 -4.03108448e-01 9.49284196e-01 4.18871552e-01
-1.20510209e+00 -2.47056261e-01 -1.03565037e+00 -4.43704396e-01
-2.60109752e-01 3.17334652e-01 9.10480201e-01 7.73003995e-01
-1.39386082e+00 1.05785179e+00 -1.41211033e+00 -1.77527919e-01
9.08102572e-01 3.92464519e-01 6.03648275e-03 -2.52332062e-01
-5.41300714e-01 4.37346876e-01 2.77824908e-01 -4.57447320e-01
-9.89930749e-01 -7.50642717e-01 -4.27863181e-01 5.16547084e-01
1.80708408e-01 -5.38502455e-01 1.49347568e+00 -7.73718774e-01
-1.16745877e+00 9.23307240e-01 -2.19095752e-01 -8.65475416e-01
1.76641807e-01 -4.69728976e-01 -1.98309392e-01 2.04207495e-01
-2.17214674e-01 6.81465864e-01 5.04011810e-01 -6.56473041e-01
-6.01295412e-01 -7.54455566e-01 -3.75284702e-01 3.66823137e-01
-1.00858855e+00 -2.49548312e-02 -9.88739610e-01 -6.20248735e-01
5.53934313e-02 -1.00727057e+00 -1.45715341e-01 -2.17180010e-02
-3.32816839e-01 -2.60237157e-01 9.47696388e-01 -5.49291611e-01
1.35441148e+00 -2.30393577e+00 -1.90258831e-01 -1.16876066e-01
3.80450845e-01 2.76599407e-01 -1.88932285e-01 -3.00392266e-02
1.90828532e-01 1.60694361e-01 6.41017556e-02 -8.48250389e-01
-2.63723016e-01 1.26630440e-01 -1.46238178e-01 1.32984668e-01
-4.05301571e-01 1.02689683e+00 -5.88296175e-01 -5.05286872e-01
-4.77974713e-01 5.38279653e-01 -7.94024408e-01 -3.30320112e-02
-6.69036061e-02 -3.92882943e-01 -2.56170690e-01 8.00911009e-01
5.46929359e-01 -6.67508185e-01 2.65153885e-01 -4.74270582e-02
4.63352233e-01 6.63319767e-01 -7.13820279e-01 1.91553962e+00
-4.71737027e-01 9.27161157e-01 2.19537213e-01 -1.15116954e+00
5.81970572e-01 1.95675015e-01 5.57808816e-01 -6.78658605e-01
1.21585771e-01 1.22886486e-01 -5.11089981e-01 -3.55609268e-01
6.96359932e-01 4.91011381e-01 3.84829432e-01 7.30449975e-01
-3.03039961e-02 1.65787011e-01 -2.39018619e-01 4.18563604e-01
1.46974015e+00 -1.07841983e-01 -1.77116767e-01 9.75506753e-03
-4.94219840e-01 -5.41909151e-02 5.93871772e-01 6.71789229e-01
-5.03475785e-01 4.23180312e-01 3.00543994e-01 -3.79126281e-01
-8.31918418e-01 -7.27205217e-01 -8.83737057e-02 1.50684428e+00
5.18137775e-03 -1.02206993e+00 -1.08398139e+00 -6.53884113e-01
-1.36782154e-01 5.04873931e-01 -1.22372814e-01 -3.83161068e-01
-2.95205772e-01 -8.15301239e-01 9.08626080e-01 1.08794022e+00
7.98859417e-01 -6.90059006e-01 -9.70192492e-01 -3.87049675e-01
-3.05444062e-01 -1.33052969e+00 -5.47126591e-01 5.57009637e-01
-1.72176790e+00 -6.71195149e-01 -3.87454301e-01 -9.02911544e-01
5.49416184e-01 8.05900812e-01 1.23636186e+00 3.57215792e-01
2.94211954e-02 1.13194093e-01 -3.01996917e-01 -4.71809417e-01
1.02460608e-01 5.86689889e-01 1.19248740e-01 -6.17465556e-01
6.85091555e-01 -6.21898949e-01 -5.02809703e-01 2.07615811e-02
-8.70134890e-01 2.67807186e-01 6.47642553e-01 6.44007802e-01
7.95230925e-01 2.13647470e-01 3.52481186e-01 -1.15454042e+00
7.64076293e-01 -6.58389330e-01 -1.64032102e-01 2.02162996e-01
-1.27685344e+00 1.16501683e-02 6.00082099e-01 -2.40992695e-01
-8.76367748e-01 1.12460628e-01 -6.67384192e-02 -7.29396164e-01
4.60815094e-02 7.01797843e-01 1.49724241e-02 2.22866312e-01
6.87852800e-01 2.36703888e-01 -7.11941300e-03 -5.86697519e-01
3.17437691e-03 9.78641152e-01 3.56647670e-01 -4.09404874e-01
-5.74017242e-02 3.11664283e-01 -1.89211801e-01 -6.77682877e-01
-6.10817194e-01 -4.31467533e-01 -3.16360235e-01 1.48298249e-01
2.90144295e-01 -1.20625663e+00 -2.66678512e-01 3.67083281e-01
-9.70541954e-01 -6.71410918e-01 -2.75999784e-01 5.37235618e-01
-4.46098983e-01 5.94057590e-02 -1.04178441e+00 -4.81665164e-01
-7.24087477e-01 -1.21582866e+00 1.21809816e+00 1.92948319e-02
1.22381076e-01 -7.53741503e-01 -2.34167576e-01 6.47345781e-01
6.41986132e-01 -3.95038962e-01 8.33994091e-01 -8.34516406e-01
-5.40359974e-01 -3.46720994e-01 -4.57067013e-01 4.29860979e-01
-3.75131935e-01 -3.81669253e-01 -1.27194285e+00 -4.15450960e-01
2.13996088e-03 -6.76194191e-01 1.14900053e+00 4.84749466e-01
1.98444867e+00 -3.44556451e-01 -7.32877970e-01 1.07745397e+00
1.34651649e+00 1.37938216e-01 4.47916061e-01 -6.77206218e-02
4.50660855e-01 2.29497924e-01 1.37286857e-01 4.33407873e-01
2.75080711e-01 5.84021807e-01 3.54460835e-01 2.10891306e-01
-1.22958750e-01 -3.32193077e-01 5.12553811e-01 1.27360737e+00
-2.02936962e-01 -1.72540322e-01 -7.56011486e-01 4.00389075e-01
-1.62671363e+00 -7.52658725e-01 1.60790190e-01 2.10229945e+00
8.75609577e-01 1.87895522e-01 -5.47737479e-02 2.23399922e-01
4.76815879e-01 4.11055982e-02 -9.15713251e-01 -1.68430090e-01
1.63467363e-01 3.87753844e-01 6.40659630e-01 1.41300589e-01
-8.96831155e-01 9.52742457e-01 7.29437685e+00 1.07314527e+00
-1.20516527e+00 5.78252912e-01 1.17930079e+00 -8.55832160e-01
-2.58031547e-01 -1.16162069e-01 -9.98294413e-01 5.86056054e-01
1.52234221e+00 -1.17305130e-01 6.08672142e-01 1.51050246e+00
-6.93093762e-02 1.38458923e-01 -1.34132028e+00 1.47682118e+00
1.78010032e-01 -1.59991455e+00 -1.93750381e-01 2.07142934e-01
8.39879751e-01 8.70905995e-01 1.18528537e-01 3.60047132e-01
9.63351503e-02 -9.72730994e-01 3.46271157e-01 3.63235623e-01
1.08261466e+00 -5.36086738e-01 7.35354722e-01 5.76596320e-01
-9.42447841e-01 -1.38295725e-01 -4.44778085e-01 -2.30755195e-01
-1.24603562e-01 7.11460769e-01 -6.24523640e-01 -4.08278182e-02
1.08888018e+00 1.02961159e+00 -8.02935362e-01 7.54717886e-01
3.20639104e-01 9.44232643e-01 -4.94731784e-01 -2.25188285e-01
-1.76175833e-01 1.13136247e-01 -7.22497329e-03 1.21235454e+00
3.76837462e-01 -1.62580878e-01 -9.10797864e-02 5.52289963e-01
-7.15164542e-01 -3.03599909e-02 -7.90856242e-01 -2.89311379e-01
8.59145343e-01 9.40261602e-01 -8.57400239e-01 -5.20359755e-01
-3.41676354e-01 1.17774212e+00 5.55469990e-01 1.30922228e-01
-7.51423359e-01 -3.70316476e-01 5.26314557e-01 4.00913395e-02
2.52977703e-02 -3.07527930e-01 -7.39098132e-01 -1.23063827e+00
2.03715444e-01 -7.98114836e-01 3.24759215e-01 -7.15522408e-01
-7.27497160e-01 5.37003398e-01 -4.73928601e-02 -7.44543195e-01
-4.76920418e-02 -5.56441605e-01 -3.63631427e-01 2.15102151e-01
-1.26174068e+00 -6.98111951e-01 -4.34352458e-01 4.43247586e-01
8.10880244e-01 -4.46693271e-01 9.01320159e-01 6.05936348e-01
-7.64274597e-01 1.11162496e+00 5.22340000e-01 -3.92602049e-02
4.39560294e-01 -6.29352093e-01 4.02070552e-01 7.19665229e-01
4.38064963e-01 5.23413360e-01 4.84049059e-02 -5.29519379e-01
-1.44086707e+00 -1.20926678e+00 9.63911593e-01 -2.73505867e-01
1.07469119e-01 -2.56332636e-01 -5.73024631e-01 9.31894600e-01
7.49533176e-02 1.19164623e-01 9.75708246e-01 2.28244320e-01
-3.09767187e-01 -4.73317504e-01 -9.73973453e-01 3.42113912e-01
1.55975199e+00 -7.24002242e-01 2.17478916e-01 5.75266957e-01
1.12937593e+00 -1.66484505e-01 -7.41691291e-01 3.16638291e-01
5.39958179e-01 -9.12823319e-01 6.98525310e-01 -7.01251388e-01
7.18640506e-01 6.11968398e-01 -5.09544134e-01 -9.96706545e-01
-3.46049406e-02 -4.44710314e-01 -7.03714073e-01 6.71086311e-01
5.10244608e-01 -2.13954270e-01 1.60440183e+00 8.44723642e-01
-4.04214680e-01 -1.30107605e+00 -6.79590404e-01 -7.95105338e-01
-2.03300938e-01 -9.47946131e-01 5.59115469e-01 9.94317114e-01
4.74243648e-02 3.37141782e-01 -1.65470079e-01 -2.94406742e-01
3.12423497e-01 -1.34452015e-01 4.37755615e-01 -1.10079634e+00
-6.18501067e-01 -3.98475498e-01 -3.96935016e-01 -1.45974839e+00
1.22811384e-01 -1.33634508e+00 -2.76662141e-01 -1.17371881e+00
5.87793350e-01 -6.51701510e-01 -2.98954278e-01 7.18450546e-01
3.08838397e-01 3.86925697e-01 2.63341367e-01 6.61008835e-01
-9.35507894e-01 5.36346138e-01 7.74787366e-01 -7.41853267e-02
-1.55741364e-01 1.46105140e-01 -9.09119189e-01 6.41075969e-01
9.88222063e-01 -5.85079670e-01 -7.52946496e-01 -1.20563734e+00
2.58091241e-01 -2.31035367e-01 1.23610543e-02 -1.41259146e+00
8.19707692e-01 1.67069837e-01 1.79459825e-01 -4.94199663e-01
4.16132838e-01 -8.98941219e-01 -1.94121465e-01 5.23345947e-01
-7.03633010e-01 4.14926082e-01 -4.60142605e-02 5.54553866e-01
-8.14176425e-02 -3.71029645e-01 5.89354992e-01 -2.79527046e-02
-5.56803584e-01 4.57021475e-01 -2.14722291e-01 -1.91936642e-02
9.55691278e-01 -1.60233423e-01 -3.56320381e-01 -4.43661362e-01
-7.50693440e-01 -8.29814095e-03 2.93844581e-01 2.24317417e-01
8.49463701e-01 -1.20023572e+00 -2.99666792e-01 3.37862492e-01
-2.51312226e-01 1.80859402e-01 3.80582027e-02 7.36903727e-01
-7.29064226e-01 6.47395253e-01 1.00489512e-01 -5.01087546e-01
-1.50888550e+00 2.60798931e-01 1.46547168e-01 -3.66521776e-01
-4.88255352e-01 1.04519689e+00 -5.71097672e-01 -4.29233909e-01
7.89122224e-01 -3.99801940e-01 2.37022251e-01 -2.75918931e-01
5.74113548e-01 8.29798639e-01 6.71467304e-01 -1.69447407e-01
-6.15261048e-02 -8.33517686e-02 -2.29166940e-01 7.19750300e-02
1.43769884e+00 6.47035465e-02 -4.56224829e-02 2.19252110e-01
1.63786590e+00 -4.88372773e-01 -1.02431643e+00 -5.19371629e-01
-1.21366695e-01 -2.51696199e-01 7.07758844e-01 -4.87607896e-01
-1.57433522e+00 9.10672426e-01 9.34906304e-01 -6.03604838e-02
1.46902955e+00 -4.87393960e-02 1.05413663e+00 8.55264127e-01
4.19388771e-01 -1.49916673e+00 1.90701529e-01 6.06794775e-01
2.95862496e-01 -1.05729687e+00 1.17903747e-01 -4.47575718e-01
-5.13484657e-01 6.93043709e-01 7.94429600e-01 2.63234019e-01
1.13127327e+00 7.08998680e-01 -2.38558918e-01 -1.87781900e-01
-1.18649387e+00 4.07617122e-01 -2.39840269e-01 5.02665401e-01
6.12347126e-01 1.34895831e-01 -6.82843924e-02 9.70795929e-01
-4.51231301e-01 4.87195492e-01 7.64607117e-02 1.11611378e+00
-1.54020458e-01 -8.67938876e-01 1.68603569e-01 1.08955908e+00
-5.55154264e-01 -3.76928955e-01 -1.83475181e-01 7.25690499e-02
5.25242575e-02 8.19243073e-01 4.18360770e-01 -9.01648521e-01
-2.15660095e-01 2.30180785e-01 1.20171130e-01 -6.81536913e-01
-4.07239079e-01 -1.53107345e-01 5.05164899e-02 -8.75483215e-01
-3.58348995e-01 -5.27012348e-01 -1.20840681e+00 -6.44506097e-01
-4.69137013e-01 -6.60069063e-02 1.07843924e+00 8.34718347e-01
1.14908051e+00 3.99389088e-01 2.61409640e-01 -8.19262087e-01
-7.94118345e-01 -9.42724288e-01 -6.21785343e-01 1.75443918e-01
-1.64445326e-01 -2.29609776e-02 -1.18613116e-01 1.83427036e-01] | [8.557530403137207, 3.0465469360351562] |
87e60a34-70ad-4270-915f-1a4ce9855550 | ignore-previous-prompt-attack-techniques-for | 2211.09527 | null | https://arxiv.org/abs/2211.09527v1 | https://arxiv.org/pdf/2211.09527v1.pdf | Ignore Previous Prompt: Attack Techniques For Language Models | Transformer-based large language models (LLMs) provide a powerful foundation for natural language tasks in large-scale customer-facing applications. However, studies that explore their vulnerabilities emerging from malicious user interaction are scarce. By proposing PromptInject, a prosaic alignment framework for mask-based iterative adversarial prompt composition, we examine how GPT-3, the most widely deployed language model in production, can be easily misaligned by simple handcrafted inputs. In particular, we investigate two types of attacks -- goal hijacking and prompt leaking -- and demonstrate that even low-aptitude, but sufficiently ill-intentioned agents, can easily exploit GPT-3's stochastic nature, creating long-tail risks. The code for PromptInject is available at https://github.com/agencyenterprise/PromptInject. | ['Ian Ribeiro', 'Fábio Perez'] | 2022-11-17 | null | null | null | null | ['real-world-adversarial-attack', 'adversarial-text'] | ['adversarial', 'adversarial'] | [ 2.48673875e-02 2.98593193e-01 -2.74506390e-01 -4.12546359e-02
-1.04635942e+00 -1.25914168e+00 8.17436337e-01 -1.40480444e-01
-1.93793830e-02 5.04268408e-01 1.54795080e-01 -9.63133037e-01
3.13994735e-01 -5.29661357e-01 -6.99877858e-01 -4.90968674e-01
-3.09573743e-03 5.24440050e-01 -2.54972100e-01 -4.92508858e-01
-3.20014954e-02 3.64640772e-01 -2.04153001e-01 2.08170757e-01
7.32719183e-01 4.45758253e-01 -3.35804224e-01 8.85073423e-01
9.03097615e-02 1.35861039e+00 -7.25249887e-01 -1.04286420e+00
4.70607430e-01 -3.02189052e-01 -7.23653436e-01 -4.42148954e-01
1.80804074e-01 -3.44829410e-01 -4.08594906e-01 1.21917892e+00
4.92857188e-01 -2.13460743e-01 1.51576057e-01 -1.76603413e+00
-7.54159749e-01 1.23325300e+00 -3.57868910e-01 1.23241797e-01
6.47931874e-01 1.11701775e+00 1.08029640e+00 -3.49280119e-01
2.65973210e-01 1.37865591e+00 6.73028946e-01 8.81997466e-01
-1.43855429e+00 -9.52283382e-01 1.74613193e-01 -4.40508187e-01
-1.14490616e+00 -3.37795824e-01 8.24565411e-01 -4.56957549e-01
1.07133710e+00 5.86680233e-01 6.74374998e-02 1.82638323e+00
4.70468789e-01 7.09505439e-01 1.20552921e+00 -1.14013895e-01
-1.68762188e-02 1.68540373e-01 2.08510220e-01 5.79095840e-01
1.23142391e-01 5.63702047e-01 -3.43543589e-01 -9.86752331e-01
6.06130183e-01 -2.83215821e-01 5.83331361e-02 1.53394639e-01
-1.18386364e+00 9.97476578e-01 -7.73199573e-02 1.65434480e-01
-5.90359330e-01 4.41658527e-01 4.47021574e-01 5.24468958e-01
2.21632928e-01 8.33525598e-01 -5.96376896e-01 -5.34946442e-01
-3.21490586e-01 6.65540397e-01 1.09155524e+00 1.04554260e+00
1.46694615e-01 4.24559772e-01 -3.45305026e-01 3.05586994e-01
1.30276546e-01 7.35014677e-01 1.88423812e-01 -8.33903909e-01
6.96993589e-01 1.24745056e-01 1.07855320e-01 -9.16553199e-01
-3.36851507e-01 -2.38666072e-01 -5.06197691e-01 2.74861716e-02
6.08949304e-01 -7.08914876e-01 -3.28974217e-01 2.25013280e+00
1.55975550e-01 1.08946092e-01 1.08362786e-01 5.03156245e-01
2.48247057e-01 4.55506474e-01 3.15553933e-01 -1.46610498e-01
1.48226643e+00 -8.99745047e-01 -4.47350621e-01 -5.93803942e-01
4.90866244e-01 -7.33614802e-01 1.38110387e+00 3.20677489e-01
-1.23673534e+00 6.22139014e-02 -5.33574224e-01 3.34005773e-01
-1.32939517e-01 -6.91536963e-01 8.57537389e-01 1.02427018e+00
-8.64931583e-01 1.58042133e-01 -7.24346220e-01 3.36059108e-02
4.43364501e-01 3.41929257e-01 -2.26536676e-01 2.58741826e-01
-1.38181269e+00 8.63787532e-01 1.41531620e-02 -2.87915319e-01
-1.42760766e+00 -7.28043199e-01 -8.09386671e-01 -1.36086062e-01
7.75323272e-01 -7.73041129e-01 2.03227758e+00 -7.39154756e-01
-1.66770720e+00 5.71336627e-01 2.06744611e-01 -7.15748012e-01
6.70906782e-01 -2.39629820e-01 -3.54823530e-01 -2.09347263e-01
1.47975413e-02 1.49500951e-01 1.07654405e+00 -1.03029156e+00
-1.22168183e-01 -2.28394456e-02 3.95762444e-01 -1.28007919e-01
-1.26428664e-01 8.50670516e-01 1.94988862e-01 -1.01298070e+00
-8.46037209e-01 -1.17278564e+00 -7.65451372e-01 -6.34449959e-01
-9.65126514e-01 4.83565358e-03 3.70070398e-01 -7.45974064e-01
1.29866612e+00 -1.95465910e+00 -3.92338932e-02 1.93381861e-01
3.80646080e-01 4.37570155e-01 -6.47230148e-01 7.42317140e-01
-1.96592957e-01 5.60759962e-01 -4.33985703e-02 -1.58116877e-01
5.14979005e-01 -1.05092771e-01 -8.40194464e-01 1.39872774e-01
1.85156673e-01 1.46458542e+00 -1.01953828e+00 5.30991480e-02
-5.98219456e-03 5.42435236e-02 -7.29700863e-01 5.31668305e-01
-6.94339395e-01 5.48749208e-01 -5.84910870e-01 6.41750336e-01
3.36695284e-01 -5.88890389e-02 2.20224664e-01 4.50527012e-01
2.96856552e-01 4.02082145e-01 -4.56661701e-01 1.28244126e+00
-5.01581192e-01 2.05503330e-01 2.44798288e-01 -2.57586420e-01
4.39628482e-01 4.35781598e-01 2.92587399e-01 -4.82402951e-01
5.06013691e-01 8.54826272e-02 1.84997171e-01 -2.48746991e-01
3.50175411e-01 -2.59168565e-01 -7.66490459e-01 9.95809376e-01
-1.46845177e-01 -4.19692487e-01 -5.11770621e-02 4.87724334e-01
1.39497566e+00 -3.77558410e-01 4.47012246e-01 -5.44758588e-02
2.73685813e-01 -2.34528169e-01 5.18049538e-01 1.15355992e+00
-2.58379072e-01 2.36403435e-01 9.14697051e-01 -4.14223403e-01
-9.44913924e-01 -9.99020100e-01 5.64183414e-01 9.89858270e-01
-3.90314996e-01 -6.41483903e-01 -1.07279086e+00 -1.02925730e+00
1.63685873e-01 1.29295874e+00 -9.65192243e-02 -4.62616593e-01
-6.93302393e-01 -5.07341266e-01 1.35590434e+00 3.18029851e-01
-3.09504922e-02 -1.24706948e+00 -2.96303570e-01 3.00187588e-01
-2.76995897e-01 -1.32854879e+00 -1.22289217e+00 2.14830767e-02
-2.59737730e-01 -1.05483425e+00 -2.91435927e-01 -2.77866066e-01
4.64501143e-01 -1.01733640e-01 1.25837719e+00 -8.43988433e-02
2.76192874e-02 5.32940447e-01 -1.07922241e-01 -4.64906931e-01
-1.11974967e+00 9.50505137e-02 3.97884846e-01 -1.28855422e-01
4.01368171e-01 -7.40538418e-01 -6.17430583e-02 3.44373465e-01
-8.21265876e-01 -2.02801451e-01 1.98566794e-01 6.53587818e-01
-1.50271565e-01 -8.18906054e-02 4.72832948e-01 -1.02314651e+00
1.41612697e+00 -4.66242999e-01 -9.21737731e-01 1.19496956e-01
-4.58213687e-01 1.13724116e-02 9.09631133e-01 -9.65725183e-01
-6.95418060e-01 -4.31800872e-01 -3.62478852e-01 -4.68162328e-01
-2.73869872e-01 1.39383554e-01 -2.66132474e-01 -3.49990875e-01
7.10389435e-01 2.71816730e-01 -2.23389510e-02 -2.53762156e-01
6.20707870e-01 3.95558864e-01 5.50216854e-01 -1.16682768e+00
1.39009893e+00 -6.69382587e-02 -4.18361306e-01 -2.16469601e-01
-5.55836916e-01 1.14530481e-01 1.52600616e-01 1.73135251e-01
7.12757409e-01 -7.36603141e-01 -1.31843853e+00 8.29216719e-01
-1.42283094e+00 -9.12867785e-01 -3.01552922e-01 -4.88280840e-02
-6.15881145e-01 3.70225608e-01 -1.12075973e+00 -9.09749448e-01
-7.40490258e-01 -1.24611115e+00 6.89745367e-01 1.87401459e-01
-7.74407387e-01 -9.80599999e-01 2.37562314e-01 5.27609110e-01
6.83472335e-01 1.29692510e-01 9.94602203e-01 -1.41671443e+00
-6.44578755e-01 -3.84188354e-01 3.78231615e-01 4.72725928e-01
4.23960648e-02 -2.06719130e-01 -7.33325124e-01 -2.55395740e-01
3.74600708e-01 -4.46424037e-01 -6.62261993e-02 2.44142883e-03
7.68053114e-01 -1.29357326e+00 6.18420877e-02 4.80344296e-01
8.64555299e-01 2.97349304e-01 4.22940731e-01 2.65051723e-01
7.82159567e-01 3.25159490e-01 3.53274882e-01 5.69575965e-01
3.29867095e-01 7.39681244e-01 3.54129374e-01 1.85106546e-01
4.30293322e-01 -7.52990901e-01 8.71823132e-01 5.53577483e-01
2.94082046e-01 -3.96388978e-01 -9.56519246e-01 1.74434930e-01
-1.60276747e+00 -1.15243042e+00 2.21575692e-01 2.01270628e+00
1.10262299e+00 4.58976179e-01 3.79156023e-01 -3.87501627e-01
3.73445213e-01 4.35456723e-01 -4.48293209e-01 -1.00019169e+00
-2.50644293e-02 2.22638860e-01 6.58846140e-01 8.40571284e-01
-9.28795695e-01 1.26157892e+00 6.32517147e+00 1.03540409e+00
-8.94614100e-01 4.70962107e-01 6.88262701e-01 -1.72964826e-01
-7.19749570e-01 1.80356756e-01 -6.64566576e-01 7.00599074e-01
1.15003836e+00 -7.23253131e-01 1.01012874e+00 8.60891700e-01
2.59298295e-01 5.02104580e-01 -1.22337115e+00 7.04011858e-01
-3.54641527e-02 -1.06414437e+00 -1.46422699e-01 2.20729932e-01
5.63063025e-01 3.36989835e-02 4.70847070e-01 5.39579272e-01
1.26611912e+00 -1.15916157e+00 9.21795368e-01 -1.05005335e-02
5.49597740e-01 -7.41304040e-01 3.86549145e-01 6.60893917e-01
-7.05690026e-01 -3.66379470e-01 2.76691765e-01 -4.37295027e-02
7.44164824e-01 2.69232690e-01 -7.92208731e-01 2.14670867e-01
5.91602847e-02 -1.09211318e-01 -2.78656006e-01 2.62892544e-01
-6.19089484e-01 7.69812047e-01 -3.18389565e-01 1.48192897e-01
2.58109123e-01 -2.41604936e-03 1.06378508e+00 9.77976620e-01
7.07598925e-02 9.08455029e-02 3.94785881e-01 1.08158886e+00
-1.90038741e-01 -3.58061582e-01 -9.32907343e-01 -7.35187352e-01
5.70067048e-01 1.33021200e+00 -1.60620749e-01 -6.32468089e-02
-2.38568783e-01 9.56393123e-01 1.52039483e-01 4.21242833e-01
-1.28838027e+00 -4.77641646e-04 1.18528879e+00 -2.01439615e-02
-2.28788108e-01 -4.36476469e-01 -3.86701882e-01 -1.20985997e+00
-2.22397484e-02 -2.12231135e+00 3.15378428e-01 -6.71194196e-01
-1.44270504e+00 8.62418413e-01 -8.79981965e-02 -7.04405546e-01
-6.57263875e-01 -3.61948848e-01 -9.95144308e-01 8.85125697e-01
-8.81702542e-01 -1.50065506e+00 3.86625916e-01 7.09500492e-01
4.25851732e-01 -5.20093679e-01 8.56300235e-01 1.20177075e-01
-6.69108868e-01 1.22694945e+00 -4.73767340e-01 1.64896220e-01
3.48428637e-01 -9.49819028e-01 1.29028094e+00 1.09933734e+00
3.17607045e-01 1.02798927e+00 9.93025005e-01 -7.55345702e-01
-1.55173910e+00 -9.11718130e-01 9.51710701e-01 -1.11555421e+00
1.25848889e+00 -6.87683702e-01 -4.95282739e-01 1.20103765e+00
4.18797165e-01 -3.13017875e-01 5.68694055e-01 -1.49537951e-01
-8.58602941e-01 3.09868455e-01 -1.24198461e+00 1.23774934e+00
9.34861481e-01 -1.01982343e+00 -3.39673311e-01 5.95384717e-01
1.32917500e+00 -4.95893300e-01 -6.54451489e-01 3.35085876e-02
3.29066038e-01 -6.45783007e-01 1.08959103e+00 -1.11179304e+00
1.85059324e-01 1.60636410e-01 -1.77117422e-01 -1.25046754e+00
-2.23912820e-01 -1.87251067e+00 -2.45033354e-01 1.42649186e+00
4.90657449e-01 -1.23194313e+00 5.91866910e-01 1.04476488e+00
2.41790071e-01 -4.35349882e-01 -7.36950994e-01 -1.06020510e+00
4.94186074e-01 -7.22210705e-01 8.48127663e-01 9.42909539e-01
2.99534261e-01 2.49921083e-01 -7.32505858e-01 2.89921194e-01
5.59028506e-01 -3.49961430e-01 1.02603710e+00 -4.87480193e-01
-7.23679125e-01 -5.12518167e-01 2.05403626e-01 -6.87732875e-01
5.40384471e-01 -8.69332612e-01 -1.40382603e-01 -5.71171522e-01
1.04708433e-01 -4.32634056e-01 -1.36601880e-01 9.43652928e-01
-1.65792629e-01 -1.37494266e-01 6.85478866e-01 -9.28113088e-02
-3.19685996e-01 3.24106663e-01 8.68659675e-01 -2.69449025e-01
-7.40844756e-02 3.18185002e-01 -1.15427077e+00 6.85537934e-01
1.17010033e+00 -8.64756346e-01 -3.51182938e-01 -2.17057303e-01
3.83383542e-01 2.66703576e-01 5.51542342e-01 -3.72050196e-01
1.05530038e-01 -6.83042645e-01 -6.61687315e-01 2.07375184e-01
3.20321321e-02 -6.61893368e-01 2.71247983e-01 4.93999809e-01
-6.92659914e-01 6.36014819e-01 3.79158407e-01 1.97234347e-01
1.97273940e-01 -1.18919827e-01 3.95409852e-01 -3.10496569e-01
2.77748052e-02 5.59496343e-01 -7.30526209e-01 4.77309763e-01
9.84926283e-01 4.53925073e-01 -5.14247775e-01 -7.59192586e-01
-2.54299581e-01 1.04462951e-01 5.27792573e-01 6.11113846e-01
3.13177705e-01 -1.21621096e+00 -6.96247280e-01 2.02644855e-01
-1.77476615e-01 -4.58034843e-01 1.75643131e-01 8.56030166e-01
-2.59231299e-01 4.12408739e-01 4.21112441e-02 9.73759517e-02
-1.30485034e+00 1.07208157e+00 3.03636402e-01 -8.62785220e-01
-3.12741965e-01 1.01863205e+00 4.14943159e-01 -7.95018017e-01
1.05629517e-02 -1.47494167e-01 4.57631171e-01 -6.34105742e-01
4.78445888e-01 9.47134569e-03 -4.31752086e-01 -4.48825002e-01
-4.07564998e-01 6.91194385e-02 -3.12415272e-01 -4.37464952e-01
8.11000109e-01 2.23060697e-01 -2.50334740e-01 -1.62030488e-01
7.72115946e-01 4.33083892e-01 -9.46658015e-01 -2.37714931e-01
1.01441279e-01 -3.54433209e-01 -5.16313374e-01 -7.82751322e-01
-7.77943552e-01 6.26320124e-01 -7.72450715e-02 4.37462330e-01
7.70080924e-01 -1.26277775e-01 1.16250265e+00 4.05705273e-01
8.20640504e-01 -5.69209278e-01 2.35684991e-01 6.36120617e-01
1.01430738e+00 -8.31171572e-01 -6.10701144e-01 -2.23142758e-01
-1.07867503e+00 4.53489602e-01 6.87779129e-01 2.27327701e-02
2.54859686e-01 1.00520229e+00 3.07690799e-01 -3.75925452e-02
-9.96743321e-01 3.87250304e-01 -2.88256735e-01 8.45087051e-01
-6.99867532e-02 3.41957062e-01 1.03234082e-01 1.18514037e+00
-4.00793314e-01 -2.44641542e-01 7.28450656e-01 7.86272883e-01
2.86290824e-01 -1.61065066e+00 -4.99910235e-01 4.36982103e-02
-9.96075273e-01 -6.62616670e-01 -5.99437714e-01 3.86801958e-01
-4.12543952e-01 1.09293425e+00 -5.14123678e-01 -8.00881982e-01
1.81645274e-01 4.79464866e-02 2.76640505e-01 -6.07861280e-01
-1.17865336e+00 -2.49968797e-01 5.54382265e-01 -8.32106888e-01
5.06180227e-01 -4.97696698e-01 -7.90641844e-01 -8.81379724e-01
2.53720209e-02 1.08672477e-01 1.82072416e-01 6.95578754e-01
3.51376772e-01 2.32353300e-01 8.20732951e-01 -5.70105970e-01
-1.24356306e+00 -6.72532260e-01 -1.93723798e-01 3.40032220e-01
2.64732867e-01 -1.22883037e-01 -4.32250649e-01 -2.63913155e-01] | [6.062373638153076, 8.04647445678711] |
9639f085-fe12-4d29-a5f7-9ef4e12fe2d2 | mutual-information-divergence-a-unified | 2205.13445 | null | https://arxiv.org/abs/2205.13445v1 | https://arxiv.org/pdf/2205.13445v1.pdf | Mutual Information Divergence: A Unified Metric for Multimodal Generative Models | Text-to-image generation and image captioning are recently emerged as a new experimental paradigm to assess machine intelligence. They predict continuous quantity accompanied by their sampling techniques in the generation, making evaluation complicated and intractable to get marginal distributions. Based on a recent trend that multimodal generative evaluations exploit a vison-and-language pre-trained model, we propose the negative Gaussian cross-mutual information using the CLIP features as a unified metric, coined by Mutual Information Divergence (MID). To validate, we extensively compare it with competing metrics using carefully-generated or human-annotated judgments in text-to-image generation and image captioning tasks. The proposed MID significantly outperforms the competitive methods by having consistency across benchmarks, sample parsimony, and robustness toward the exploited CLIP model. We look forward to seeing the underrepresented implications of the Gaussian cross-mutual information in multimodal representation learning and the future works based on this novel proposition. | ['Sang-Woo Lee', 'Kang Min Yoo', 'Jiyoung Lee', 'Yunji Kim', 'Jin-Hwa Kim'] | 2022-05-25 | null | null | null | null | ['human-judgment-correlation', 'human-judgment-classification'] | ['reasoning', 'reasoning'] | [ 6.60687208e-01 3.76441538e-01 -1.37830526e-01 -5.52464247e-01
-1.19720912e+00 -6.46260798e-01 1.24019384e+00 -1.25591457e-01
-4.71516103e-01 7.94182777e-01 3.84504914e-01 -5.71843572e-02
-3.65150981e-02 -4.82616186e-01 -7.63770401e-01 -8.23372602e-01
2.57890731e-01 5.94714105e-01 -3.53712887e-01 8.05563703e-02
3.89425814e-01 3.99085134e-02 -1.72379851e+00 3.80479723e-01
8.56183946e-01 1.05231297e+00 1.43814772e-01 7.51481831e-01
-1.34371489e-01 6.17303193e-01 -5.55874646e-01 -9.73135173e-01
9.32655707e-02 -6.72996938e-01 -6.74686491e-01 -6.64923154e-03
4.12884504e-01 1.22585408e-01 9.53236371e-02 1.05043876e+00
6.84827983e-01 -1.28915071e-01 1.31257260e+00 -1.44231772e+00
-9.65791404e-01 6.65283501e-01 -5.77625573e-01 -4.32898581e-01
6.76361263e-01 2.31322214e-01 1.05575037e+00 -8.71758461e-01
8.54950130e-01 1.39420056e+00 4.73933607e-01 5.66805542e-01
-1.39232671e+00 -3.68340760e-01 -1.48231402e-01 8.12044274e-03
-1.32151604e+00 -2.76979357e-01 6.50393605e-01 -5.60592055e-01
4.66972828e-01 3.75385374e-01 -7.11727962e-02 1.54279208e+00
1.78643819e-02 8.67289424e-01 1.32945871e+00 -5.82849145e-01
1.19997248e-01 6.12158775e-01 -2.70707607e-01 5.12778997e-01
5.37571348e-02 1.48109645e-01 -5.36373854e-01 -1.65720627e-01
4.38410014e-01 -5.49217939e-01 -3.42985690e-01 -4.22041237e-01
-1.45451903e+00 9.36806023e-01 1.17966928e-01 1.91065162e-01
-1.31914213e-01 5.87376542e-02 3.29810917e-01 1.29091784e-01
3.20560932e-01 4.80809242e-01 -1.19892344e-01 -3.06880295e-01
-1.06179547e+00 2.78340191e-01 7.78996706e-01 8.81001711e-01
6.33274138e-01 -1.38229225e-02 -6.56574130e-01 7.64998913e-01
4.01163310e-01 6.33568287e-01 6.22273684e-01 -8.14878345e-01
4.72397327e-01 4.98333871e-01 1.26628920e-01 -1.01603079e+00
-9.80843902e-02 -3.71970326e-01 -1.01440346e+00 8.87266099e-02
4.55796063e-01 -1.41150847e-01 -6.92472279e-01 2.08544636e+00
4.02476266e-02 -3.02251220e-01 1.73115611e-01 9.54589427e-01
7.06649363e-01 5.37863195e-01 9.63761732e-02 -1.97416425e-01
1.16432035e+00 -8.13186228e-01 -6.55441761e-01 1.61976188e-01
2.71340072e-01 -9.78124440e-01 1.15555954e+00 3.70370746e-01
-1.11212146e+00 -7.52139151e-01 -9.83051062e-01 3.07080690e-02
-3.55810136e-01 1.73571751e-01 6.55098557e-01 7.80797184e-01
-1.17771113e+00 5.91235101e-01 -4.16956544e-01 -2.76616007e-01
1.69523075e-01 1.14194462e-02 -5.54929256e-01 1.32358551e-01
-9.85484660e-01 9.18203473e-01 4.37217683e-01 -1.56482548e-01
-7.87721813e-01 -4.26499069e-01 -8.71301532e-01 -1.17838740e-01
-1.62150189e-01 -1.04732454e+00 1.09261286e+00 -1.08832192e+00
-1.66139126e+00 1.05871582e+00 -8.59263167e-02 -4.10212666e-01
9.29788232e-01 -1.08273841e-01 -1.42283157e-01 1.29606828e-01
-4.98486869e-02 1.18406796e+00 1.00625074e+00 -1.56864893e+00
-9.98360887e-02 -2.43858323e-01 -1.27075627e-01 1.77685112e-01
-2.29231685e-01 -2.75399953e-01 -2.34738842e-01 -6.47698939e-01
2.59916726e-02 -8.47126424e-01 -1.86580291e-03 -1.62152693e-01
-5.69534183e-01 -2.21113995e-01 2.15794295e-01 -4.63656485e-01
1.04477501e+00 -1.83601737e+00 2.82013088e-01 6.29624277e-02
-1.12504289e-01 -1.03853876e-02 -4.96208638e-01 5.79955995e-01
-1.69393808e-01 2.58133411e-01 -3.15818876e-01 -6.49812639e-01
5.32782376e-01 -4.68748324e-02 -3.35327268e-01 2.67760754e-01
4.17607576e-01 1.02259827e+00 -9.01441991e-01 -6.56461179e-01
1.29558519e-01 7.24814534e-01 -3.56411219e-01 2.98085600e-01
-2.53158778e-01 5.29679835e-01 -1.83161885e-01 4.17248905e-01
7.82747149e-01 -2.89821655e-01 -1.21216625e-02 -4.88787800e-01
1.19797952e-01 -2.23958850e-01 -8.09727430e-01 1.88198531e+00
-2.79982656e-01 7.41388857e-01 -4.00332063e-01 -8.15811992e-01
1.01645899e+00 3.51647079e-01 8.15885793e-03 -6.46320939e-01
2.87563622e-01 2.28378177e-01 -1.78217947e-01 -5.29683053e-01
5.46920836e-01 -2.43785456e-01 -1.11485936e-01 2.86802888e-01
4.00073230e-01 -3.00984591e-01 4.57855791e-01 1.88745216e-01
6.79257810e-01 5.40919602e-01 2.15550691e-01 -1.69400141e-01
5.03736675e-01 -3.13286692e-01 -2.16476824e-02 9.15275633e-01
1.21117815e-01 1.13100481e+00 7.02238202e-01 1.52742714e-02
-1.20334077e+00 -1.32814884e+00 -1.77215561e-01 8.34826231e-01
-5.92015348e-02 -2.22896054e-01 -9.23260152e-01 -5.73182702e-01
-2.73218811e-01 9.70482588e-01 -7.95245826e-01 -5.19757457e-02
1.15844496e-01 -7.14816630e-01 6.83337033e-01 2.54570693e-01
2.26214692e-01 -8.76727462e-01 -3.63490850e-01 -1.46419838e-01
-3.51760656e-01 -1.14934421e+00 -4.03010100e-01 -1.42884463e-01
-5.42210639e-01 -8.00244570e-01 -1.23652053e+00 -4.47975278e-01
4.99590695e-01 -2.08244607e-01 1.41393554e+00 -5.87862670e-01
-2.17955858e-01 6.93937421e-01 -4.35353369e-01 -2.91307867e-01
-5.95963836e-01 -7.19393045e-02 -1.96678936e-01 2.31206551e-01
1.96639866e-01 -6.40199244e-01 -6.54370248e-01 3.65375817e-01
-1.10667908e+00 4.11401391e-01 9.66543138e-01 8.63183200e-01
4.67455596e-01 -6.45789921e-01 5.33158362e-01 -5.26497483e-01
9.10417616e-01 -4.85603541e-01 -4.14436013e-01 5.58719695e-01
-6.70935869e-01 5.55572331e-01 2.37321690e-01 -4.86981928e-01
-1.15262341e+00 -5.56995394e-03 2.22050890e-01 -3.79617840e-01
-3.02216023e-01 4.75248218e-01 -2.06026956e-01 1.87088743e-01
5.81530869e-01 3.78877193e-01 1.45819888e-01 -1.78347886e-01
9.00046825e-01 4.97338712e-01 6.66611910e-01 -6.15513444e-01
7.39318907e-01 3.33117515e-01 1.70298412e-01 -5.76489210e-01
-8.37970257e-01 -1.46640062e-01 -4.55268711e-01 -2.25355715e-01
1.15697610e+00 -7.45371222e-01 -8.09251904e-01 3.20190310e-01
-1.59899962e+00 1.57209694e-01 -6.74158782e-02 5.34475267e-01
-8.51479888e-01 5.24205565e-01 -3.01549703e-01 -1.19860446e+00
-3.67564917e-01 -1.12230277e+00 1.43450391e+00 2.85646707e-01
-2.59511620e-01 -1.03907657e+00 3.66910785e-01 5.02348363e-01
5.46984553e-01 5.82977712e-01 7.58671999e-01 -5.49795985e-01
-4.45695907e-01 -1.85067400e-01 -5.12747169e-01 5.73606789e-01
-1.82265565e-01 -6.31224597e-04 -1.24961877e+00 1.24394204e-02
-1.52464792e-01 -4.62485671e-01 1.04529762e+00 2.27740049e-01
9.00339484e-01 -2.88459182e-01 -2.01074984e-02 3.44502300e-01
1.36927629e+00 -1.42282978e-01 8.80165040e-01 -4.08439226e-02
3.67863268e-01 8.82392883e-01 4.16587234e-01 5.81357598e-01
3.31424057e-01 4.65044349e-01 3.70335370e-01 -2.76753046e-02
-8.48332122e-02 -4.20971483e-01 4.42921698e-01 7.58394659e-01
-1.63953096e-01 -6.60556614e-01 -6.00818813e-01 2.92939991e-01
-1.74451637e+00 -1.08891761e+00 -2.96619944e-02 2.37392402e+00
9.13682222e-01 -6.27060700e-03 -1.03918865e-01 -2.05874145e-01
7.90910840e-01 3.08572687e-02 -1.21054098e-01 -3.97460848e-01
-5.24636269e-01 1.37155905e-01 2.42542759e-01 4.30435985e-01
-1.00397909e+00 5.45433521e-01 6.46194315e+00 1.04459977e+00
-1.00380647e+00 4.01791260e-02 9.30191875e-01 1.76706880e-01
-5.98118544e-01 -3.50891799e-02 -5.64977288e-01 4.22559083e-01
9.29155529e-01 -1.16204873e-01 2.23247468e-01 6.94276035e-01
-5.85480034e-02 -2.68996954e-01 -1.37798440e+00 1.32455838e+00
6.01190805e-01 -9.88701940e-01 4.34108824e-01 -1.99095104e-02
9.62593555e-01 -1.90785185e-01 5.05038142e-01 1.78410098e-01
3.86760123e-02 -1.25630641e+00 7.50462651e-01 1.08014929e+00
7.07489848e-01 -5.27570188e-01 8.41998994e-01 7.96285793e-02
-7.93727458e-01 3.04531962e-01 -1.81044340e-01 1.82959124e-01
3.04304332e-01 5.56981325e-01 -9.82493043e-01 7.74410844e-01
8.51850305e-03 2.46818751e-01 -7.51271248e-01 9.31472719e-01
1.12094778e-04 3.45189482e-01 -6.20994866e-02 -1.77829981e-01
2.68824995e-01 -3.66156608e-01 6.07693195e-01 1.32337058e+00
6.17965221e-01 -6.65794373e-01 -1.61117435e-01 1.33193338e+00
6.07524849e-02 2.00153023e-01 -6.31846666e-01 -5.77226222e-01
2.77068913e-02 1.38412476e+00 -5.39535761e-01 -2.49649480e-01
-2.16121629e-01 1.26414442e+00 -3.47267166e-02 4.21914488e-01
-1.07307172e+00 -2.83458501e-01 2.11154535e-01 -5.68832196e-02
1.15057334e-01 -4.92090769e-02 -2.13758633e-01 -1.06537592e+00
1.06134161e-01 -7.33070850e-01 1.07624218e-01 -1.00795519e+00
-1.67988551e+00 9.58163321e-01 2.58258611e-01 -1.37299943e+00
-7.41285920e-01 -7.08128095e-01 -4.14455891e-01 7.34749317e-01
-1.33931625e+00 -1.33262467e+00 -3.96866739e-01 2.73931533e-01
2.37784892e-01 -2.75003254e-01 9.86856043e-01 1.81895554e-01
-2.34605059e-01 8.23350132e-01 6.40580580e-02 -7.18146637e-02
9.37447667e-01 -1.28774929e+00 5.33673577e-02 4.38946307e-01
3.27661097e-01 5.27847230e-01 9.92353022e-01 -2.78326064e-01
-1.15083456e+00 -6.92409575e-01 7.74382830e-01 -6.87958896e-01
5.40686369e-01 -2.60685712e-01 -5.46239734e-01 1.98120892e-01
7.60020494e-01 -4.29423302e-01 7.38202929e-01 -2.47307107e-01
-6.19512022e-01 9.43715498e-02 -9.19164300e-01 5.25158167e-01
7.96073914e-01 -4.41140711e-01 -4.72874254e-01 4.07907456e-01
6.01745546e-01 3.58555987e-02 -6.96513057e-01 4.38518375e-01
6.42264485e-01 -1.20051694e+00 7.17754841e-01 -3.87259215e-01
6.86804116e-01 -1.70513883e-01 -4.81529564e-01 -1.10693741e+00
1.25384152e-01 -9.03327286e-01 3.16980988e-01 1.56873310e+00
7.94878721e-01 -4.27945644e-01 6.01540983e-01 5.96148074e-01
5.89952357e-02 -5.93440294e-01 -8.13754797e-01 -7.52468824e-01
8.82676691e-02 -5.01654565e-01 4.23969388e-01 6.27440274e-01
-2.24026497e-02 5.27217150e-01 -5.83572686e-01 -3.96194875e-01
7.57701278e-01 -1.35054057e-02 8.34035277e-01 -1.15719008e+00
-4.47753489e-01 -6.67711437e-01 -5.37134469e-01 -9.47799504e-01
1.94790855e-01 -1.01692796e+00 -7.20964326e-03 -1.26372266e+00
5.67522943e-01 6.59390986e-02 -2.24405497e-01 3.62578966e-02
-7.01353922e-02 4.56355393e-01 2.84088731e-01 5.69742098e-02
-7.88430393e-01 8.28176439e-01 1.36643267e+00 -2.32615858e-01
1.92585990e-01 -1.83238178e-01 -5.79187930e-01 6.20163620e-01
5.44386744e-01 -2.03284889e-01 -5.26370585e-01 -2.33760908e-01
4.15636033e-01 4.94574793e-02 5.31561077e-01 -1.01144803e+00
1.04377270e-02 1.28312940e-02 3.51448983e-01 -5.17028928e-01
5.99545062e-01 -4.92942899e-01 2.33374894e-01 6.61230609e-02
-6.26969516e-01 8.62465501e-02 -1.95906207e-01 4.73399550e-01
-4.53796268e-01 -4.39050078e-01 6.10706091e-01 -1.72883838e-01
-3.53761345e-01 2.99498476e-02 3.41874473e-02 1.50942370e-01
8.66491139e-01 -2.50872344e-01 -2.62513816e-01 -8.16594779e-01
-5.71284950e-01 -2.77834505e-01 3.44486982e-01 5.45031786e-01
6.05493605e-01 -1.46332788e+00 -1.05901837e+00 -2.14716066e-02
3.02244246e-01 -6.16821408e-01 2.60343909e-01 9.98022437e-01
-2.57529020e-01 5.84278703e-01 -1.66492298e-01 -7.46600211e-01
-1.02749538e+00 3.76231700e-01 5.50779030e-02 -3.55565548e-01
1.81728199e-01 7.51561940e-01 4.07383949e-01 -3.75638276e-01
2.49647871e-01 -1.00922167e-01 -5.21662086e-02 1.74377277e-01
5.16340971e-01 1.78876534e-01 -1.46045670e-01 -6.87983096e-01
-8.04084688e-02 4.55040067e-01 1.48237661e-01 -6.12681508e-01
9.15289342e-01 -1.52669787e-01 -1.44621134e-01 6.16369843e-01
1.36810076e+00 -1.89173713e-01 -1.09715950e+00 9.55603197e-02
4.66116220e-02 -2.50413775e-01 -3.80145997e-01 -9.34915245e-01
-6.64756715e-01 9.73283470e-01 8.98157418e-01 1.97981119e-01
9.05272782e-01 2.57366244e-02 3.46690238e-01 2.17741415e-01
2.60318190e-01 -9.15561080e-01 3.63796085e-01 2.32573241e-01
1.19454765e+00 -1.48049212e+00 -2.16308042e-01 -1.58100516e-01
-1.08259320e+00 1.04523146e+00 4.26578820e-01 1.79824501e-01
4.25949335e-01 -7.50796730e-03 5.11346292e-03 2.55665928e-01
-7.24459648e-01 -1.12766668e-01 7.77884066e-01 8.82237613e-01
8.13424289e-01 1.38776004e-01 -4.40211892e-01 6.37348235e-01
-3.74177158e-01 -1.05892628e-01 1.97606638e-01 2.85853297e-01
-9.31045264e-02 -1.33129132e+00 -2.19128489e-01 1.30838558e-01
-3.66582513e-01 -1.81117937e-01 -5.05387247e-01 6.20116413e-01
9.98214632e-02 9.11684096e-01 -6.77611679e-02 -3.53325576e-01
-1.81163952e-01 2.07730561e-01 7.10581362e-01 -1.69242486e-01
-2.41726190e-01 -4.99241166e-02 -7.37425685e-02 -4.24937963e-01
-6.27643406e-01 -5.44918478e-01 -6.71931624e-01 1.69125199e-01
-3.60181123e-01 2.44242370e-01 9.95250463e-01 7.44769454e-01
5.71795166e-01 1.40382871e-01 4.77007598e-01 -1.07660055e+00
-7.50825167e-01 -1.35338533e+00 -3.04961711e-01 7.76316226e-01
-1.22679263e-01 -5.54538429e-01 -5.28005838e-01 2.41773173e-01] | [11.07171630859375, 0.8958476185798645] |
bada5dc0-6ce8-4c3f-aaea-8c137db3497d | disentangled-phonetic-representation-for | 2305.14783 | null | https://arxiv.org/abs/2305.14783v1 | https://arxiv.org/pdf/2305.14783v1.pdf | Disentangled Phonetic Representation for Chinese Spelling Correction | Chinese Spelling Correction (CSC) aims to detect and correct erroneous characters in Chinese texts. Although efforts have been made to introduce phonetic information (Hanyu Pinyin) in this task, they typically merge phonetic representations with character representations, which tends to weaken the representation effect of normal texts. In this work, we propose to disentangle the two types of features to allow for direct interaction between textual and phonetic information. To learn useful phonetic representations, we introduce a pinyin-to-character objective to ask the model to predict the correct characters based solely on phonetic information, where a separation mask is imposed to disable attention from phonetic input to text. To avoid overfitting the phonetics, we further design a self-distillation module to ensure that semantic information plays a major role in the prediction. Extensive experiments on three CSC benchmarks demonstrate the superiority of our method in using phonetic information. | ['Qifan Wang', 'Xiaojun Quan', 'Zihong Liang'] | 2023-05-24 | null | null | null | null | ['spelling-correction'] | ['natural-language-processing'] | [ 5.48909664e-01 -1.30824402e-01 -1.26140207e-01 -5.07848859e-01
-6.61879659e-01 -3.69583249e-01 5.45203686e-01 1.43757671e-01
-5.96953571e-01 5.03354073e-01 4.55835730e-01 -5.27436018e-01
5.09900570e-01 -6.04122519e-01 -5.61026037e-01 -6.33547664e-01
6.75878584e-01 7.50468969e-02 1.56884998e-01 7.57212043e-02
4.93565470e-01 1.60862461e-01 -1.47622502e+00 6.16530895e-01
1.51781762e+00 6.17184341e-01 5.96218944e-01 5.81208348e-01
-4.66644198e-01 8.00317168e-01 -7.36675620e-01 -4.14224833e-01
1.39879182e-01 -6.70284629e-01 -6.41118348e-01 -1.33318990e-01
2.98613578e-01 -1.42167926e-01 -1.86396405e-01 1.39449012e+00
3.13570112e-01 1.01962186e-01 7.93267608e-01 -6.99177146e-01
-8.77870500e-01 1.01913285e+00 -3.85421515e-01 2.05013692e-01
2.04977673e-02 4.53009494e-02 1.25777590e+00 -9.70833600e-01
2.35837474e-01 1.27358854e+00 4.78680432e-01 6.18572772e-01
-1.07094777e+00 -5.96883059e-01 4.62116122e-01 4.73105133e-01
-1.29583967e+00 -4.38479185e-01 8.02764714e-01 -2.81351209e-01
8.78455102e-01 6.05201244e-01 3.13755125e-01 1.07574821e+00
7.55105466e-02 1.16738081e+00 8.85424495e-01 -6.65062428e-01
1.28092244e-01 2.04049692e-01 1.18169360e-01 1.14311799e-01
1.15571730e-01 -1.58674911e-01 -3.39488447e-01 1.51674211e-01
5.57625234e-01 -2.06954665e-02 -3.30816299e-01 1.39219165e-01
-9.45200324e-01 7.73976028e-01 3.14006507e-01 3.59850019e-01
-1.95641704e-02 7.99916089e-02 4.67598617e-01 2.00621292e-01
3.45034033e-01 6.64525628e-01 -5.02711892e-01 -4.39668357e-01
-7.21061051e-01 -6.08938225e-02 2.42725387e-01 9.17404592e-01
6.05881929e-01 3.21259499e-01 -4.53714162e-01 1.18331599e+00
2.43031103e-02 3.94641429e-01 9.88138556e-01 -6.41169071e-01
6.77859426e-01 5.04405200e-01 -3.42124142e-02 -8.82577479e-01
6.50544912e-02 -3.27009946e-01 -6.97021961e-01 -2.61489362e-01
3.95788401e-01 2.38438360e-02 -9.90767598e-01 1.52253795e+00
-1.82655394e-01 1.75921112e-01 2.56585045e-05 1.04337895e+00
4.88361597e-01 6.94212496e-01 1.70793578e-01 -7.19757155e-02
1.08903360e+00 -1.14755309e+00 -8.52761328e-01 -5.28318286e-01
8.44703019e-01 -9.54261541e-01 1.58396304e+00 2.61938304e-01
-8.16259801e-01 -8.02546322e-01 -9.72474813e-01 -3.53574336e-01
-2.43403375e-01 5.36242187e-01 2.04134732e-01 6.62686110e-01
-4.44958240e-01 6.84272587e-01 -7.90082753e-01 2.95940693e-03
1.84021205e-01 8.10522959e-02 3.82008404e-02 -5.89281097e-02
-1.31046236e+00 9.11546111e-01 5.89868307e-01 2.45998099e-01
-4.16994870e-01 -4.04027104e-01 -7.61522949e-01 3.82891625e-01
3.66963267e-01 -2.22830102e-01 1.28877246e+00 -1.33451116e+00
-1.62431157e+00 4.15619642e-01 -4.90517795e-01 -2.66695261e-01
5.51343560e-01 -4.30033863e-01 -4.23884064e-01 -2.97373589e-02
8.78539681e-02 5.70680976e-01 8.44587088e-01 -9.80381489e-01
-8.27376723e-01 -1.01148255e-01 -3.69275004e-01 3.57885242e-01
-5.46793878e-01 1.93451092e-01 -7.64309824e-01 -1.26726925e+00
2.32432723e-01 -1.01458371e+00 -1.85356721e-01 -4.50340182e-01
-5.94032347e-01 -2.05361530e-01 5.50290048e-01 -8.18103969e-01
1.39416707e+00 -2.21570611e+00 -2.69505894e-03 1.61588266e-01
-1.17182173e-01 4.94540840e-01 -2.06918970e-01 1.04228891e-01
6.17602505e-02 2.91661739e-01 -3.34412694e-01 -6.17931008e-01
-2.07764208e-01 4.47473019e-01 -3.49490911e-01 1.90465704e-01
4.80772257e-01 7.76536167e-01 -8.11268032e-01 -2.93534815e-01
6.17483854e-02 3.65815550e-01 -8.18761349e-01 9.97838899e-02
-1.88289806e-01 3.81755531e-01 -3.60151500e-01 2.52545208e-01
7.49395788e-01 1.01200745e-01 2.96288908e-01 2.17256382e-01
-1.51503175e-01 1.12773252e+00 -9.71104503e-01 1.23357236e+00
-4.36915278e-01 3.91363651e-01 -2.57448286e-01 -9.32105243e-01
1.03009343e+00 1.77688718e-01 -4.54672938e-03 -9.20797765e-01
1.59366980e-01 3.07317287e-01 3.11699033e-01 -3.20731729e-01
7.31919885e-01 -8.36481601e-02 -9.19581875e-02 1.60190076e-01
-4.73394632e-01 -3.21970023e-02 -4.84557860e-02 -3.94324698e-02
5.99687099e-01 5.06122820e-02 9.43340659e-02 -2.14207768e-01
6.76415503e-01 -6.53976351e-02 1.06912613e+00 8.97839069e-01
-1.47518516e-01 1.00883555e+00 5.58003724e-01 -4.86186482e-02
-1.02574646e+00 -7.28121400e-01 1.38579181e-03 1.15305865e+00
1.25565380e-01 -6.57967150e-01 -9.48625982e-01 -8.69993448e-01
-7.83813279e-03 1.03500819e+00 -5.71756124e-01 -4.55886185e-01
-8.72574329e-01 -6.36436641e-01 6.53486907e-01 8.90128672e-01
1.52217180e-01 -9.62524354e-01 -2.51067489e-01 2.61264592e-01
-2.73626149e-01 -9.97058213e-01 -9.11665857e-01 3.67942780e-01
-7.25871682e-01 -7.27497935e-01 -6.78203404e-01 -8.88653278e-01
6.52399838e-01 3.27490330e-01 5.31697273e-01 1.79124311e-01
1.25010461e-01 -5.02840519e-01 -5.96595883e-01 -2.94045269e-01
-6.36788130e-01 3.40989202e-01 8.73415338e-05 -5.64665883e-04
5.79692543e-01 -1.66194290e-01 -2.52615362e-01 3.26703370e-01
-8.37024033e-01 3.63399744e-01 5.62342286e-01 8.39852393e-01
4.98218715e-01 -1.97320767e-02 4.06770200e-01 -1.14224601e+00
5.39250493e-01 -1.59955353e-01 -4.08145577e-01 7.34281465e-02
-7.24867940e-01 3.43796521e-01 1.09365010e+00 -5.50168335e-01
-1.03801119e+00 8.96270722e-02 -5.08640528e-01 -4.34275657e-01
-1.96331397e-01 6.83584154e-01 -4.78928655e-01 4.60547447e-01
2.33772695e-01 5.47758758e-01 -1.66168079e-01 -9.06352818e-01
3.18159491e-01 8.37980449e-01 5.91896832e-01 -5.11372745e-01
5.81176043e-01 5.79638705e-02 -7.46963978e-01 -6.42549157e-01
-8.42984915e-01 -3.59993160e-01 -6.75259829e-01 2.64767200e-01
8.22109878e-01 -8.04609716e-01 -4.28381532e-01 4.15564835e-01
-1.17365766e+00 -4.85932827e-02 -1.08353831e-01 5.58975220e-01
-2.04150245e-01 6.14001453e-01 -8.83645236e-01 -7.08492279e-01
-1.46685541e-01 -1.26291323e+00 7.29852200e-01 2.57923961e-01
-3.59637260e-01 -6.18359625e-01 -1.59093529e-01 2.66168982e-01
4.31022763e-01 -5.24348021e-01 1.18489456e+00 -8.48137498e-01
-3.20316732e-01 -3.92715167e-03 -2.78540611e-01 7.69686818e-01
3.83978963e-01 1.81995064e-01 -1.29184580e+00 -3.89461741e-02
-1.14837673e-03 1.64898127e-01 1.22720242e+00 1.32538572e-01
1.56796801e+00 -6.23838663e-01 -1.42447636e-01 5.94448090e-01
1.04472303e+00 3.70077312e-01 8.21860135e-01 3.14571351e-01
1.07949042e+00 5.04469991e-01 5.49385428e-01 3.07653606e-01
3.43734413e-01 6.86708391e-01 2.39475742e-01 7.09129944e-02
-2.06136376e-01 -5.93457401e-01 4.39489067e-01 1.25320005e+00
1.73324123e-01 -1.38019443e-01 -7.90338516e-01 4.41704541e-01
-1.71441388e+00 -6.54669404e-01 -3.04675967e-01 2.22804856e+00
1.27065933e+00 4.25328016e-01 -2.65086472e-01 3.51334125e-01
8.21739972e-01 1.58707723e-01 -3.35033387e-01 -6.31959856e-01
-2.82947809e-01 4.84565198e-02 4.79266256e-01 5.67430735e-01
-1.04821336e+00 1.29239511e+00 5.66963911e+00 1.15783346e+00
-1.40596902e+00 -1.82598263e-01 5.24409473e-01 9.13791209e-02
-6.89515114e-01 1.50472388e-01 -9.02718723e-01 8.53279591e-01
7.92826712e-01 1.04135469e-01 2.80757725e-01 8.62230062e-01
4.35405225e-01 9.86503139e-02 -9.67887700e-01 6.80451870e-01
-2.47697886e-02 -1.09174013e+00 2.85413742e-01 -2.17290998e-01
7.56314635e-01 -1.62357166e-01 7.19093531e-02 3.98978561e-01
1.79058388e-01 -1.04128647e+00 1.08729112e+00 4.19264674e-01
4.51307297e-01 -8.67795050e-01 6.84332013e-01 5.07505655e-01
-9.41400766e-01 -4.41035703e-02 -6.48679137e-01 -2.41872370e-01
-9.38603804e-02 6.02940679e-01 -9.26342368e-01 3.26273173e-01
3.90976071e-01 8.13465774e-01 -7.95238316e-01 1.05561376e+00
-6.27272666e-01 9.53202128e-01 5.34271868e-03 -1.43633515e-01
2.37472177e-01 -2.09200606e-01 3.09764415e-01 1.35348785e+00
3.02270085e-01 -1.17297910e-01 1.28647223e-01 9.09769654e-01
-2.61593815e-02 3.35867673e-01 -1.42104238e-01 -1.62887856e-01
5.50449431e-01 7.28313863e-01 -6.36254191e-01 -4.87751573e-01
-3.72800589e-01 1.34381747e+00 2.69238681e-01 2.72592396e-01
-7.97771394e-01 -5.81651628e-01 8.68608236e-01 -5.39826825e-02
5.91040969e-01 -1.85385644e-01 -8.58443141e-01 -1.29342067e+00
6.75410479e-02 -1.03715098e+00 4.55805473e-02 -5.28736472e-01
-1.11035240e+00 5.28393507e-01 -5.72813869e-01 -1.19457781e+00
-2.78739721e-01 -5.43168664e-01 -7.68542588e-01 1.26460469e+00
-1.71476519e+00 -7.87735820e-01 6.91164285e-02 2.24960908e-01
8.59563589e-01 1.33611292e-01 5.78455150e-01 3.62288475e-01
-8.89515579e-01 9.90951598e-01 3.74401420e-01 3.53803188e-01
9.06911075e-01 -1.19758153e+00 6.15766883e-01 1.16493106e+00
2.39103101e-02 8.16383362e-01 6.74608767e-01 -8.29698980e-01
-9.80759144e-01 -1.18699777e+00 1.29255128e+00 -3.69605094e-01
2.68483609e-01 -5.50473154e-01 -1.35157847e+00 5.41775882e-01
-6.15126677e-02 -2.22516388e-01 5.78522861e-01 4.94503342e-02
-5.10493338e-01 8.77427533e-02 -5.19670665e-01 8.25442374e-01
6.81095243e-01 -7.00727046e-01 -7.55585194e-01 -1.56952724e-01
8.27854216e-01 -1.82621598e-01 -2.31350631e-01 3.20580155e-01
3.58102024e-01 -7.67095983e-01 5.87021768e-01 -5.77645183e-01
6.20469272e-01 -4.14866418e-01 -4.44069430e-02 -1.49194467e+00
-3.46139669e-01 -3.90593112e-01 2.03377143e-01 1.40714955e+00
5.23325562e-01 -4.14692581e-01 7.45650589e-01 3.01652133e-01
-4.63295400e-01 -3.93073380e-01 -7.78368235e-01 -7.99705327e-01
3.89682353e-01 -6.08388722e-01 7.01348007e-01 1.08396137e+00
1.06777214e-01 2.47015059e-01 -5.62551916e-01 1.72493458e-01
-1.21151216e-01 -7.54703209e-02 4.22314346e-01 -1.13306057e+00
-3.24389070e-01 -7.09720254e-01 1.07730016e-01 -1.42985606e+00
2.47841269e-01 -1.03267241e+00 4.36671883e-01 -1.20531344e+00
1.26051500e-01 -6.67799413e-01 -3.48709345e-01 5.20020247e-01
-8.75491083e-01 7.29673728e-02 3.29347223e-01 3.71380687e-01
-2.34621257e-01 8.50423455e-01 1.14400792e+00 -1.06195003e-01
-1.75716996e-01 1.53090522e-01 -9.04627979e-01 7.24348009e-01
8.68559897e-01 -6.25815451e-01 -2.36553833e-01 -8.27018142e-01
-9.25135389e-02 -2.67425239e-01 -1.59189656e-01 -7.53747165e-01
1.12877846e-01 -3.59748602e-01 3.66520971e-01 -4.12864178e-01
2.27185741e-01 -5.26080966e-01 -3.48092109e-01 4.06468183e-01
-6.66390419e-01 1.12366751e-01 7.00981617e-02 2.94097751e-01
-3.97632092e-01 -5.13402402e-01 9.11263764e-01 9.15959477e-02
-5.77785611e-01 -2.05918401e-01 -7.52290547e-01 -3.25180553e-02
4.58946794e-01 -3.19110632e-01 -2.28662744e-01 -2.28665143e-01
-3.56925219e-01 1.37969419e-01 5.44257939e-01 6.99068546e-01
5.29660642e-01 -1.28873897e+00 -7.03119457e-01 5.93164802e-01
9.59450230e-02 -2.25757480e-01 1.20711513e-01 6.51269436e-01
-5.35662532e-01 5.90802729e-01 6.20289370e-02 -2.86673516e-01
-1.18362010e+00 5.61426163e-01 1.88316986e-01 -1.80070940e-02
-6.05543196e-01 8.20689142e-01 3.15077782e-01 -4.95216370e-01
3.84918779e-01 -7.25208282e-01 -3.67200851e-01 -4.00958918e-02
6.05748653e-01 1.80724144e-01 2.70166755e-01 -5.84330082e-01
-2.54505783e-01 2.71971285e-01 -4.82260674e-01 6.11131713e-02
1.00978780e+00 -3.48036468e-01 1.95561603e-01 4.28772926e-01
1.13447917e+00 4.65540588e-01 -1.22990191e+00 -2.79031217e-01
3.52263004e-01 -6.34680271e-01 -1.30925002e-02 -7.11168110e-01
-9.45797145e-01 1.22382045e+00 2.31253371e-01 -5.92663512e-02
8.01591635e-01 -3.23598742e-01 9.38675582e-01 2.01758400e-01
-2.34210134e-01 -1.38574803e+00 -5.09109721e-02 9.85975087e-01
5.67507625e-01 -1.12384558e+00 -4.95329320e-01 -5.83690286e-01
-1.03391981e+00 1.07249033e+00 8.05567861e-01 -6.54106885e-02
3.18382084e-01 2.40630776e-01 2.57790774e-01 6.08707011e-01
-6.12086356e-01 -1.68136328e-01 3.48135173e-01 4.41801906e-01
6.78717494e-01 1.42630368e-01 -4.35893834e-01 9.17158067e-01
-4.38690931e-01 -3.84473979e-01 7.50591457e-01 7.81547725e-01
-6.25024974e-01 -1.34506977e+00 -3.19193661e-01 3.47682029e-01
-6.04837239e-01 -5.92213929e-01 -4.19027090e-01 3.25159729e-01
1.15461163e-01 7.88759172e-01 2.64771134e-01 -3.46129805e-01
2.42655292e-01 1.75275430e-01 -4.36685905e-02 -8.03939998e-01
-6.75614476e-01 5.45386493e-01 -9.55403671e-02 -2.40444273e-01
2.51281619e-01 -7.47668564e-01 -1.39774501e+00 -1.12184502e-01
-4.81185675e-01 3.03530484e-01 5.52056193e-01 1.00230181e+00
2.88717955e-01 6.81063533e-01 6.58000708e-01 -4.21227574e-01
-7.62906790e-01 -9.90395367e-01 -5.52301645e-01 2.70184010e-01
2.64192998e-01 -3.23785335e-01 -3.53586525e-01 -7.12592229e-02] | [10.872296333312988, 10.719463348388672] |
ffae0a86-1058-4da7-be4c-c0b7ebbaee2d | pragmatics-in-grounded-language-learning | 2211.08371 | null | https://arxiv.org/abs/2211.08371v2 | https://arxiv.org/pdf/2211.08371v2.pdf | Pragmatics in Language Grounding: Phenomena, Tasks, and Modeling Approaches | People rely heavily on context to enrich meaning beyond what is literally said, enabling concise but effective communication. To interact successfully and naturally with people, user-facing artificial intelligence systems will require similar skills in pragmatics: relying on various types of context -- from shared linguistic goals and conventions, to the visual and embodied world -- to use language effectively. We survey existing grounded settings and pragmatic modeling approaches and analyze how the task goals, environmental contexts, and communicative affordances in each work enrich linguistic meaning. We present recommendations for future grounded task design to naturally elicit pragmatic phenomena, and suggest directions that focus on a broader range of communicative contexts and affordances. | ['Aida Nematzadeh', 'Roma Patel', 'Jennifer Hu', 'Nicholas Tomlin', 'Daniel Fried'] | 2022-11-15 | null | null | null | null | ['grounded-language-learning'] | ['natural-language-processing'] | [ 8.86217132e-02 4.25298721e-01 -1.50809288e-01 -5.79949260e-01
-2.82731742e-01 -6.98180318e-01 7.90457726e-01 1.64742202e-01
-3.53270739e-01 7.01960385e-01 1.30728436e+00 -2.65249908e-01
-2.74386823e-01 -4.24846768e-01 1.89481229e-01 -4.52438369e-02
6.85350969e-02 2.28327468e-01 -2.85076797e-01 -8.09688926e-01
2.02983290e-01 2.21295699e-01 -1.42573357e+00 5.32493591e-01
5.32027304e-01 1.49875149e-01 5.00987053e-01 5.27527869e-01
-5.76501369e-01 1.29705667e+00 -6.39124274e-01 -8.46135430e-03
-1.78478152e-01 -4.19832051e-01 -1.19592440e+00 6.40128553e-02
-2.31137052e-01 -4.41536337e-01 6.55773655e-02 6.98669553e-01
4.65928167e-01 6.29975379e-01 2.30817944e-01 -8.76809478e-01
-1.29374766e+00 9.66731489e-01 3.00312549e-01 -2.15856284e-01
1.21743703e+00 5.11364460e-01 1.01365817e+00 -3.77259642e-01
9.99929547e-01 1.59251356e+00 7.04838395e-01 1.02835464e+00
-1.13604581e+00 -5.27383983e-02 5.90827763e-01 -1.73293263e-01
-1.03206980e+00 -8.74617696e-01 8.89203846e-01 -5.40724576e-01
1.46669519e+00 5.23704648e-01 1.14723778e+00 1.21684074e+00
-3.45409304e-01 5.24043262e-01 1.14775610e+00 -8.70850384e-01
2.78003752e-01 2.73325652e-01 4.54715580e-01 3.51333916e-01
-2.96190986e-03 1.52604550e-01 -8.94784927e-01 -2.17810661e-01
7.02220857e-01 -1.47056669e-01 -3.18873823e-01 -1.37951076e-01
-1.64766610e+00 4.36104804e-01 3.36793721e-01 9.41627502e-01
-4.98931438e-01 5.26579857e-01 2.27697536e-01 3.08738828e-01
4.03380543e-02 1.21304297e+00 -3.92242759e-01 -7.39134729e-01
2.53459185e-01 3.29946578e-01 1.09709227e+00 1.28483427e+00
3.38538766e-01 -3.05630445e-01 -2.24062979e-01 1.03630388e+00
6.57916605e-01 7.01637387e-01 9.61133987e-02 -1.48213589e+00
8.59443322e-02 6.92526996e-01 8.31548572e-01 -8.19079638e-01
-8.06435168e-01 3.36843789e-01 2.22789273e-01 2.72601902e-01
6.12878859e-01 -4.14371789e-01 -3.40822548e-01 1.98639584e+00
6.27198517e-01 -5.98453403e-01 3.50384444e-01 1.00877810e+00
1.13893580e+00 3.02302599e-01 8.53357732e-01 -1.51471749e-01
1.71919417e+00 -5.80241501e-01 -1.31003869e+00 -5.32729268e-01
1.09793019e+00 -8.38150263e-01 2.10970044e+00 -1.54607192e-01
-1.11548913e+00 6.91390038e-03 -9.00078177e-01 -6.30396068e-01
-4.04823542e-01 -2.13799611e-01 1.03879058e+00 5.01288295e-01
-9.39122498e-01 3.26614380e-01 -4.28981572e-01 -9.12295699e-01
7.37754479e-02 -3.16954926e-02 -2.32332155e-01 2.81635970e-01
-1.08450222e+00 1.20229030e+00 2.75597066e-01 -1.52424257e-02
-2.12474808e-01 -1.78528205e-01 -1.26892757e+00 -3.03220779e-01
4.04770255e-01 -9.34234023e-01 1.90488613e+00 -9.71815884e-01
-1.81744957e+00 9.32168603e-01 1.36719439e-02 1.04883082e-01
1.04191877e-01 -4.18704659e-01 -3.50977331e-01 5.10489941e-02
2.98319370e-01 8.20929646e-01 -1.29573599e-01 -1.34972882e+00
-5.86686671e-01 -2.03908250e-01 8.41115117e-01 1.03094411e+00
-2.12880552e-01 5.08019328e-01 2.15325858e-02 -4.35242951e-01
8.27756897e-03 -6.96857691e-01 -1.79564670e-01 2.49532089e-01
-1.65398702e-01 -2.83389688e-01 6.48324847e-01 -2.16764644e-01
1.20333397e+00 -2.21557426e+00 2.16447726e-01 -2.81167746e-01
4.78097111e-01 -3.07173908e-01 1.09600592e-02 9.60439146e-01
5.36718130e-01 2.60157436e-01 2.75360256e-01 -3.86449516e-01
6.44764483e-01 4.15107250e-01 1.01509154e-01 2.09347785e-01
-3.02448183e-01 1.01295507e+00 -1.39780927e+00 -6.15632355e-01
6.92092061e-01 4.79314774e-01 -5.44285595e-01 8.23361650e-02
-4.22780305e-01 5.58827996e-01 -8.48442256e-01 4.30920660e-01
-2.24943742e-01 -2.51166612e-01 4.91908818e-01 2.79292852e-01
-3.51633012e-01 9.45772707e-01 -9.27988410e-01 2.10383368e+00
-1.01477110e+00 6.90668344e-01 3.30160618e-01 7.31899738e-02
4.79184926e-01 4.90496427e-01 2.61217244e-02 -6.05382919e-01
1.76399663e-01 9.17888880e-02 8.04620534e-02 -1.28958404e+00
3.61797720e-01 -5.63215315e-01 -4.87775296e-01 6.28241837e-01
-5.60012579e-01 -8.24060798e-01 -1.43906891e-01 5.68663143e-02
6.93753839e-01 5.51645398e-01 6.71704471e-01 -7.23396480e-01
4.35733981e-02 2.07561344e-01 3.26934248e-01 5.99536896e-01
-5.74002385e-01 -1.08210489e-01 1.10611312e-01 -7.81078577e-01
-7.92114019e-01 -7.70351410e-01 1.00832246e-02 1.59589994e+00
5.22848964e-01 -4.51055080e-01 -8.27259660e-01 -1.11291170e-01
-4.81534213e-01 1.36768496e+00 -3.79083961e-01 2.26939961e-01
-5.75138986e-01 1.04500867e-01 3.98861468e-01 3.93221468e-01
3.53151739e-01 -1.71439052e+00 -1.50186706e+00 2.72325933e-01
-4.67163444e-01 -1.19503355e+00 -3.59576076e-01 -1.02043770e-01
-4.31339800e-01 -8.00043821e-01 -1.22471601e-01 -8.55574071e-01
1.81169257e-01 2.14509055e-01 1.20583236e+00 3.51701021e-01
9.32157934e-02 1.00372231e+00 -5.48905313e-01 -5.49956501e-01
-4.36968416e-01 -5.65844536e-01 -9.86462757e-02 -6.13371670e-01
7.34453022e-01 -5.58068156e-01 -3.49143595e-01 9.82694700e-02
-2.73480654e-01 8.39488387e-01 1.05989054e-02 5.66890419e-01
4.82458137e-02 -6.55424297e-01 3.84736687e-01 -6.09999895e-01
1.14975846e+00 -2.25831255e-01 3.40017140e-01 2.96421528e-01
1.03202492e-01 -2.00492755e-01 1.70765325e-01 -6.12051606e-01
-1.47214925e+00 -3.50039274e-01 7.49311820e-02 4.83483553e-01
-5.99249244e-01 3.40318263e-01 -3.61020446e-01 2.21287832e-01
1.07668042e+00 -5.00326395e-01 -3.78877148e-02 -3.09953690e-01
9.90365028e-01 1.03098166e+00 3.83747250e-01 -1.31730890e+00
8.70025158e-02 5.25735021e-01 -3.65852356e-01 -1.21267080e+00
-7.06650078e-01 -2.71235138e-01 -3.37062061e-01 -3.79259467e-01
1.07295859e+00 -6.18039668e-01 -1.00462079e+00 -3.94913070e-02
-1.23430598e+00 -9.74328041e-01 -9.34018493e-01 6.80958748e-01
-9.07845318e-01 1.89598039e-01 -5.42987168e-01 -1.27995884e+00
-6.01138510e-02 -1.14773166e+00 1.20033157e+00 2.35379502e-01
-1.30731308e+00 -1.23809564e+00 -3.21428001e-01 3.15715164e-01
5.13870239e-01 4.90216523e-01 7.49563813e-01 -4.94282305e-01
-1.28583580e-01 2.50190943e-01 9.37097066e-05 -4.29087192e-01
7.72032976e-01 -2.87949353e-01 -8.14621508e-01 3.52904916e-01
3.01636100e-01 -5.34478307e-01 -5.31148136e-01 2.48440161e-01
2.31777340e-01 -5.83091676e-01 -4.93750721e-01 1.24760494e-01
8.98008049e-01 5.14534116e-01 2.73714691e-01 4.52538908e-01
4.08142716e-01 1.17215240e+00 6.76968932e-01 4.28572625e-01
7.68596172e-01 7.86522210e-01 -2.75826633e-01 1.93942428e-01
-3.60524714e-01 -3.95577163e-01 -2.37704930e-03 3.15448105e-01
-1.83649417e-02 8.72573815e-03 -1.01690125e+00 5.82548201e-01
-1.96055150e+00 -9.12624419e-01 3.24679427e-02 1.59900963e+00
1.07896709e+00 -1.28704354e-01 8.87142569e-02 -2.04468638e-01
4.70110774e-01 1.38789818e-01 -1.42485708e-01 -7.54007339e-01
6.87246248e-02 -4.34610963e-01 -4.70335186e-01 1.02121365e+00
-7.04610825e-01 1.29846025e+00 7.32220602e+00 5.44983931e-02
-6.38392270e-01 2.86302328e-01 2.58106202e-01 -3.38814735e-01
-7.74229467e-01 -1.09744612e-02 -2.64223129e-01 -4.78881970e-02
4.46190625e-01 -1.56210779e-04 9.34469402e-01 6.98354661e-01
6.85514629e-01 -1.52215227e-01 -1.40994394e+00 1.04029000e+00
-3.62399012e-01 -1.17717659e+00 -3.06088537e-01 -1.97506681e-01
2.59980232e-01 -5.10983288e-01 -2.42890298e-01 9.80091169e-02
6.16062105e-01 -1.03682494e+00 1.15081096e+00 4.92895722e-01
7.61711001e-01 -3.53693925e-02 1.20499231e-01 3.49384427e-01
-1.06807733e+00 -2.58330047e-01 1.98930100e-01 -1.12927556e+00
7.95649469e-01 -2.58986056e-01 -2.93234169e-01 -3.15093547e-01
7.99682260e-01 3.88981074e-01 3.84232014e-01 3.41864854e-01
-3.63993585e-01 -1.92812711e-01 -6.18124962e-01 -8.52430403e-01
2.57443011e-01 -1.41694367e-01 9.49086607e-01 1.20053744e+00
-3.24100032e-02 7.78271556e-01 2.96850920e-01 7.93653786e-01
4.64506984e-01 3.25332195e-01 -1.14199829e+00 -3.46622080e-01
1.01255167e+00 7.71453619e-01 -5.79505146e-01 -2.41968021e-01
-6.38207436e-01 5.84644854e-01 1.81973368e-01 5.24410486e-01
-2.90877908e-01 -1.30357258e-02 1.14436555e+00 2.80708913e-02
-6.81236446e-01 -4.63922948e-01 -7.27961481e-01 -8.98555040e-01
-8.05495754e-02 -1.02582562e+00 8.52423310e-02 -1.14267969e+00
-1.13582599e+00 3.70415032e-01 4.08756971e-01 -7.11868644e-01
-2.33461142e-01 -5.06989300e-01 -3.67491186e-01 7.62616098e-01
-8.57618272e-01 -1.71411455e+00 -4.01057303e-01 2.07607388e-01
6.18577540e-01 2.34919876e-01 1.22583330e+00 -1.94991991e-01
8.64132345e-02 -3.83576043e-02 -6.53279901e-01 -3.11237425e-01
3.52974027e-01 -1.17212486e+00 4.26771939e-01 3.24954271e-01
-9.85468179e-02 1.35798454e+00 1.09292531e+00 -6.60269618e-01
-1.53262925e+00 -6.43398613e-02 1.15141511e+00 -7.73156404e-01
6.88878477e-01 -2.75799751e-01 -4.16579276e-01 1.07553387e+00
7.05713630e-01 -4.77492362e-01 1.04272401e+00 6.23664856e-01
-1.80283368e-01 5.18802106e-01 -1.54042518e+00 1.53106391e+00
1.89807713e+00 -7.41090417e-01 -1.37962770e+00 3.90343279e-01
1.35145879e+00 -6.08184218e-01 -6.48712397e-01 4.10200618e-02
7.72688329e-01 -8.21042299e-01 8.80590796e-01 -7.73855925e-01
-9.04827416e-02 -4.41422999e-01 -5.35160124e-01 -9.81384635e-01
-3.15084010e-01 -1.36452091e+00 2.97998339e-01 1.13988483e+00
2.54686177e-01 -7.50829399e-01 -2.58277338e-02 1.46961653e+00
-4.48084772e-01 -3.08828771e-01 -6.72949433e-01 -1.39374018e-01
-1.08755581e-01 -8.61658871e-01 1.03961837e+00 9.43921983e-01
1.25170624e+00 6.64224625e-01 -9.24979746e-02 -2.04187989e-01
1.60470232e-02 -1.33374259e-01 7.21378386e-01 -1.15577734e+00
-2.59833455e-01 -6.56648099e-01 1.52922437e-01 -1.25338578e+00
-8.01333264e-02 -2.56787479e-01 2.06973672e-01 -1.78701508e+00
-1.91158086e-01 -7.71828771e-01 4.28823113e-01 5.23338139e-01
-1.10771889e-02 -1.40682146e-01 4.66955423e-01 1.70659453e-01
-6.88711941e-01 3.49830985e-01 1.61933267e+00 1.87695235e-01
-8.90672386e-01 -6.35978103e-01 -1.42614198e+00 1.13425648e+00
5.84097266e-01 3.45243141e-02 -7.55140722e-01 -6.59861982e-01
6.26503766e-01 -4.03393321e-02 1.31396472e-01 -6.42113805e-01
7.45568424e-03 -1.00309074e+00 -1.39923066e-01 2.01776311e-01
7.02666521e-01 -1.00435436e+00 3.13727289e-01 1.87672585e-01
-6.48542285e-01 -1.92109376e-01 3.23039949e-01 2.15491846e-01
4.58462477e-01 -2.75103450e-01 3.00443232e-01 -3.88084710e-01
-9.03978705e-01 -6.60796285e-01 -7.17810512e-01 9.30925086e-02
1.16287339e+00 -4.70433950e-01 -3.29761833e-01 -6.52643144e-01
-1.27038133e+00 4.06362772e-01 1.01737475e+00 5.01419783e-01
4.09103900e-01 -1.23939133e+00 -2.03429297e-01 -2.12543309e-01
3.44640046e-01 3.49287875e-02 4.38206606e-02 1.35707900e-01
-6.52479529e-01 3.17324966e-01 -4.64038132e-03 -2.68819690e-01
-8.91173840e-01 5.17834783e-01 6.12280965e-01 5.30910194e-01
-1.14449465e+00 7.64463842e-01 4.17127103e-01 -5.97849667e-01
3.00643623e-01 -6.35788560e-01 -4.77479130e-01 -2.14841366e-01
8.02157760e-01 1.07971564e-01 -8.28248203e-01 -7.68763125e-01
-4.34690624e-01 5.55435300e-01 6.68315887e-01 -6.70001209e-01
9.37003791e-01 -7.87930727e-01 -1.28301546e-01 8.67670476e-01
4.71371979e-01 1.31310001e-01 -1.10378158e+00 -1.76868930e-01
7.54557401e-02 -4.52141941e-01 -2.74602324e-01 -1.04648101e+00
1.46782935e-01 4.97723907e-01 1.64898664e-01 5.49069107e-01
7.81315863e-01 3.23371023e-01 3.83758992e-01 7.67144978e-01
7.80687928e-01 -1.30691552e+00 1.81912094e-01 2.39715770e-01
1.37170780e+00 -8.10318589e-01 -2.14282647e-01 -6.13013268e-01
-1.06525230e+00 9.21905220e-01 5.44751942e-01 4.30928111e-01
6.02052629e-01 5.11120439e-01 4.99190718e-01 -5.05274177e-01
-6.42227709e-01 -4.68958586e-01 -1.60133511e-01 1.22306001e+00
7.91275263e-01 4.49154347e-01 -5.96618772e-01 5.88660121e-01
-4.51680809e-01 2.29556814e-01 2.86954612e-01 1.29017854e+00
-3.93036395e-01 -1.01154184e+00 -3.82809967e-01 -1.57097913e-02
-1.20792232e-01 4.44958061e-02 -8.08160424e-01 8.94095242e-01
6.97975978e-02 1.72133863e+00 -6.52263537e-02 -2.13299021e-02
5.33623040e-01 8.91562477e-02 4.94182259e-01 -1.02136481e+00
-9.26495075e-01 9.85488296e-02 9.77347910e-01 -6.42225146e-01
-7.24936664e-01 -7.87419617e-01 -1.76135719e+00 -3.79170626e-01
-1.38730913e-01 -2.27341857e-02 4.17150676e-01 1.14413893e+00
3.14656854e-01 2.57140428e-01 -9.19336602e-02 -9.40645695e-01
-1.58300087e-01 -1.12858713e+00 -1.69346660e-01 7.27425516e-01
3.14181387e-01 -6.68273449e-01 -1.53177604e-01 -1.32893547e-01] | [9.315995216369629, 6.8219757080078125] |
c25f00bf-cebd-4131-ace8-cf904df4706e | eformer-edge-enhancement-based-transformer | 2109.08044 | null | https://arxiv.org/abs/2109.08044v2 | https://arxiv.org/pdf/2109.08044v2.pdf | Eformer: Edge Enhancement based Transformer for Medical Image Denoising | In this work, we present Eformer - Edge enhancement based transformer, a novel architecture that builds an encoder-decoder network using transformer blocks for medical image denoising. Non-overlapping window-based self-attention is used in the transformer block that reduces computational requirements. This work further incorporates learnable Sobel-Feldman operators to enhance edges in the image and propose an effective way to concatenate them in the intermediate layers of our architecture. The experimental analysis is conducted by comparing deterministic learning and residual learning for the task of medical image denoising. To defend the effectiveness of our approach, our model is evaluated on the AAPM-Mayo Clinic Low-Dose CT Grand Challenge Dataset and achieves state-of-the-art performance, $i.e.$, 43.487 PSNR, 0.0067 RMSE, and 0.9861 SSIM. We believe that our work will encourage more research in transformer-based architectures for medical image denoising using residual learning. | ['Santosh Yadav', 'Abhishek Iyer', 'Tanish Mittal', 'Harsh Sulakhe', 'Achleshwar Luthra'] | 2021-09-16 | null | null | null | null | ['medical-image-denoising'] | ['computer-vision'] | [ 2.73189783e-01 1.64524227e-01 1.48524314e-01 -4.58659738e-01
-1.27851677e+00 1.55517340e-01 6.49254471e-02 5.70604652e-02
-7.36721814e-01 4.49133962e-01 4.18425500e-01 -3.91543329e-01
2.40292177e-02 -5.84309101e-01 -7.54177809e-01 -9.77040529e-01
-2.97924995e-01 -4.10666913e-01 1.97749466e-01 -2.89703190e-01
1.01470672e-01 2.00782135e-01 -7.82427788e-01 9.54518080e-01
8.92869890e-01 1.27122343e+00 2.18624696e-01 8.77699614e-01
4.81248260e-01 1.35375512e+00 -3.35059345e-01 -5.25303721e-01
1.86151519e-01 -4.50373232e-01 -6.86004817e-01 -1.69273764e-01
2.19813943e-01 -6.92164183e-01 -7.44226038e-01 9.56766844e-01
1.07541132e+00 -6.22199476e-02 5.68815768e-01 -4.04899687e-01
-7.26525187e-01 5.43304920e-01 -7.27820575e-01 8.79174352e-01
-1.42779186e-01 1.96970195e-01 6.85553730e-01 -8.70914519e-01
4.34495568e-01 9.66372371e-01 1.25734961e+00 5.84000945e-01
-1.23059368e+00 -4.42527115e-01 2.79272459e-02 5.23094952e-01
-1.20784903e+00 -4.18805957e-01 6.72799230e-01 -2.60885023e-02
1.13989091e+00 2.90522635e-01 3.29994678e-01 8.81430030e-01
1.33032787e+00 8.28395128e-01 1.19469607e+00 -3.45732719e-01
7.57436901e-02 -1.11639082e-01 3.36469151e-02 8.73409033e-01
-8.96464661e-02 4.01487827e-01 -4.02689189e-01 2.92211652e-01
7.06797540e-01 -4.20569442e-02 -2.83448070e-01 1.47011384e-01
-8.67708921e-01 8.12505782e-01 9.89847898e-01 3.92673343e-01
-6.99026287e-01 6.25782251e-01 6.83123708e-01 4.73538607e-01
8.84358764e-01 -1.56126465e-04 -1.15210786e-01 2.20980123e-01
-1.03690422e+00 -1.90310463e-01 1.60199046e-01 5.07574797e-01
1.64535314e-01 2.33118564e-01 -4.33704764e-01 7.90850401e-01
3.13884437e-01 2.32891351e-01 6.04548395e-01 -1.00467777e+00
2.07913697e-01 1.45278633e-01 -2.91785687e-01 -7.76822925e-01
-4.52779174e-01 -8.12140226e-01 -1.21950126e+00 5.81595778e-01
-1.76130787e-01 -2.04315081e-01 -1.42385721e+00 1.46679842e+00
-7.26351291e-02 5.45857787e-01 1.30328059e-01 7.70982862e-01
1.23861349e+00 5.04465044e-01 2.71205783e-01 -1.95446461e-01
1.31143916e+00 -1.18126726e+00 -9.82400537e-01 -1.91498235e-01
5.75971544e-01 -6.90999389e-01 7.94934332e-01 4.23709571e-01
-1.57610464e+00 -4.95377749e-01 -1.14275157e+00 -3.25305641e-01
6.31326064e-03 8.44413415e-02 4.11686331e-01 6.78429127e-01
-1.38504040e+00 9.60052490e-01 -1.35914648e+00 9.00128633e-02
6.82558894e-01 5.56433380e-01 -1.90345556e-01 -2.72216499e-01
-1.03519726e+00 1.01472092e+00 -1.59310669e-01 4.52058256e-01
-1.20929885e+00 -8.61581683e-01 -8.63597929e-01 9.60309654e-02
-1.00011356e-01 -9.87994552e-01 1.37540674e+00 -7.56511271e-01
-1.14316452e+00 8.55372131e-01 -1.84342563e-01 -8.68228078e-01
5.21034300e-01 -1.60579905e-01 -4.53018278e-01 3.52171302e-01
-2.01480146e-02 6.49831414e-01 7.39476442e-01 -1.10931385e+00
-5.61796308e-01 -3.91570538e-01 -1.52318776e-01 1.86642945e-01
-1.21849239e-01 -5.32021895e-02 -3.73137802e-01 -1.00302184e+00
2.77855814e-01 -6.24467909e-01 -7.07503319e-01 2.94465683e-02
-1.83474645e-01 2.40360364e-01 5.51884770e-01 -1.23041761e+00
1.39622688e+00 -2.21764231e+00 -2.57601976e-01 2.06275377e-03
4.68812972e-01 7.84018636e-02 -2.41724730e-01 1.58733670e-02
-4.49324399e-01 8.07453170e-02 -4.74047750e-01 -5.12384355e-01
-4.12328303e-01 -7.09652677e-02 9.02660117e-02 5.90613365e-01
5.20405471e-02 9.96203959e-01 -5.55396795e-01 -3.89337838e-01
3.14893246e-01 1.00405931e+00 -6.97596073e-01 6.16704486e-02
5.25050640e-01 3.18446815e-01 -2.22437337e-01 5.24425268e-01
7.72252738e-01 -7.40715712e-02 -2.40483493e-01 -6.74367964e-01
-1.73702817e-02 1.16904296e-01 -7.77688801e-01 1.95888472e+00
-6.33992493e-01 6.36064112e-01 3.07275862e-01 -1.01601195e+00
4.32046771e-01 4.92927045e-01 7.79500604e-01 -9.94568706e-01
4.46020901e-01 1.67103827e-01 1.48159623e-01 -6.04285359e-01
9.91848335e-02 -6.04297042e-01 2.69166350e-01 1.06258802e-01
6.09207787e-02 9.61101130e-02 7.49650132e-03 9.28713083e-02
1.46945715e+00 -1.09575950e-01 4.27801572e-02 -4.25669372e-01
5.19335866e-01 -2.54765809e-01 3.86539817e-01 7.55891681e-01
-4.38691288e-01 7.99795210e-01 2.18375310e-01 -5.50225794e-01
-8.25936556e-01 -1.19096947e+00 -3.36568922e-01 9.33301508e-01
4.60021291e-03 -1.61294535e-01 -8.56313109e-01 -5.61954975e-01
-3.47777933e-01 6.75949574e-01 -9.94559765e-01 -4.87785041e-01
-7.03761339e-01 -1.12366295e+00 4.03106719e-01 8.93908620e-01
6.45038307e-01 -1.05668700e+00 -6.46804690e-01 4.89107937e-01
-3.80212933e-01 -7.73323238e-01 -8.07070255e-01 5.45265615e-01
-1.19965112e+00 -7.27170348e-01 -8.91744196e-01 -1.15583992e+00
6.91771209e-01 1.51894569e-01 1.20153999e+00 1.00843623e-01
-5.40846348e-01 1.78780064e-01 -1.15387604e-01 -2.73498058e-01
-2.96513140e-01 -2.18456149e-01 -4.89163220e-01 -3.60049278e-01
9.85651538e-02 -4.71122652e-01 -1.23784542e+00 4.58719470e-02
-1.06616020e+00 -1.30139545e-01 8.17094147e-01 1.29737318e+00
7.25766838e-01 2.22494632e-01 4.43936557e-01 -7.69621849e-01
7.98772216e-01 -1.08864591e-01 -7.61798173e-02 1.80260167e-01
-7.27405608e-01 1.99072435e-01 3.25303644e-01 -3.48921953e-04
-1.22182751e+00 -2.97067482e-02 -8.05374920e-01 -1.56884179e-01
2.61564851e-01 5.64245284e-01 4.47707623e-01 -3.01981509e-01
7.30849266e-01 1.50525242e-01 3.95537242e-02 -2.82412320e-01
1.39174506e-01 5.16144514e-01 7.15403914e-01 3.09523474e-02
3.04997206e-01 6.25797212e-01 -1.58977196e-01 -4.93556201e-01
-7.22770691e-01 -4.02291596e-01 -1.57727510e-01 -1.05869330e-01
1.10491121e+00 -1.23815084e+00 -6.86924517e-01 5.11172771e-01
-8.90731037e-01 -3.53667051e-01 -2.93929964e-01 4.44050938e-01
-5.35928845e-01 3.52316797e-01 -1.34864140e+00 -4.61532116e-01
-1.14308226e+00 -1.58891773e+00 9.47658777e-01 -8.12268183e-02
8.91325399e-02 -1.09112191e+00 2.81779375e-02 3.50898027e-01
7.12460101e-01 3.87988418e-01 6.82970464e-01 -1.01312563e-01
-4.26963300e-01 -1.55202672e-01 -2.79568821e-01 6.95764124e-01
8.44301016e-04 -6.93771958e-01 -9.79881287e-01 -5.77497959e-01
4.23661292e-01 -2.16042712e-01 1.45132148e+00 1.05883014e+00
1.48089135e+00 -3.37870908e-03 -2.53016442e-01 9.24236059e-01
1.68645263e+00 3.02755922e-01 1.23872054e+00 2.24640131e-01
4.00487125e-01 1.05022877e-01 1.43365070e-01 3.28336209e-01
2.52410263e-01 1.87129378e-01 4.19748217e-01 -7.08470225e-01
-4.93002057e-01 2.08254814e-01 2.78738707e-01 6.78710699e-01
-1.66361481e-01 -1.86506063e-01 -7.43088961e-01 5.69755375e-01
-1.53433907e+00 -9.93953645e-01 -7.76287466e-02 1.71779621e+00
8.33444417e-01 2.34670654e-01 -4.13604587e-01 1.20222829e-01
3.85380834e-01 7.72711635e-02 -5.10261536e-01 -4.31526035e-01
1.72494814e-01 8.38423491e-01 6.53425276e-01 7.21718013e-01
-1.29065919e+00 5.73123574e-01 6.47190571e+00 8.32251251e-01
-1.13395894e+00 4.18813229e-01 1.31190836e+00 -1.57015864e-02
-5.80115505e-02 -7.06642091e-01 -2.42212981e-01 2.70157486e-01
1.01830006e+00 2.00315356e-01 2.02530339e-01 4.81399924e-01
4.53254789e-01 -1.80781230e-01 -8.07444215e-01 9.96417701e-01
1.23201020e-01 -1.47837210e+00 -4.09849197e-01 -2.17119411e-01
5.48857868e-01 2.58989275e-01 3.37117791e-01 3.19800705e-01
2.07678914e-01 -1.26679790e+00 2.78098196e-01 4.09665257e-01
8.86840343e-01 -9.53006983e-01 1.15586376e+00 -5.82725368e-02
-1.01150727e+00 -3.97538841e-02 -2.31413603e-01 3.10045183e-01
2.37277493e-01 6.68410659e-01 -6.11214221e-01 4.11019534e-01
1.06329668e+00 6.11557364e-01 -3.95002872e-01 1.15088606e+00
-6.69512525e-02 5.51512599e-01 3.60812619e-02 5.21796644e-01
3.32820415e-01 1.08704120e-01 1.25084311e-01 1.48777425e+00
3.28813970e-01 4.31204379e-01 -1.91986144e-01 3.80675912e-01
-2.28056505e-01 -6.67016357e-02 -1.66822389e-01 5.71414530e-01
-3.22188467e-01 1.15211546e+00 -7.92971730e-01 -3.02706122e-01
-4.18789446e-01 1.25504231e+00 -5.74452616e-02 3.36093098e-01
-1.16736054e+00 -4.00914133e-01 4.14345413e-01 2.28380889e-01
4.59756732e-01 1.79212615e-01 -7.92436540e-01 -9.44947839e-01
-2.04844773e-01 -9.05387104e-01 4.91464883e-01 -6.70401990e-01
-1.23503828e+00 1.10109901e+00 -4.61842179e-01 -9.68991160e-01
1.85342148e-01 -5.79893231e-01 -6.07777119e-01 8.40408385e-01
-1.75172889e+00 -9.96224284e-01 -3.22720081e-01 6.06903315e-01
5.51604271e-01 1.94783226e-01 6.72180772e-01 5.36749065e-01
-4.44986701e-01 7.46082008e-01 2.87796527e-01 1.66446805e-01
7.40903020e-01 -1.20494747e+00 3.92157882e-01 9.20052409e-01
-2.74944305e-01 3.38701904e-01 6.96434975e-01 -5.74425280e-01
-1.08264506e+00 -1.05327678e+00 7.59086132e-01 3.31900455e-02
3.61191064e-01 -1.44430278e-02 -7.74303377e-01 6.50814950e-01
8.55594099e-01 2.54664391e-01 5.54251790e-01 -3.08656216e-01
-3.26902643e-02 -4.15331393e-01 -1.56234038e+00 5.69581568e-01
7.32007682e-01 -1.95214510e-01 -4.79185998e-01 2.86724389e-01
5.59736848e-01 -7.31821418e-01 -1.06175041e+00 6.52316093e-01
3.04718763e-01 -1.09711921e+00 1.29572046e+00 -1.53474003e-01
7.73200154e-01 -6.47019595e-02 -1.95286109e-03 -1.40096378e+00
-6.17051125e-01 -3.24479252e-01 2.58454114e-01 5.13917863e-01
4.79374945e-01 -2.90963531e-01 9.41265404e-01 3.59743893e-01
-7.08749413e-01 -1.11103702e+00 -1.12730742e+00 -3.01511496e-01
2.85117328e-01 -5.18012047e-01 -8.44316855e-02 5.57896256e-01
-1.41112790e-01 2.10605547e-01 -4.57706451e-01 9.76206884e-02
6.97569609e-01 -4.98051494e-01 -2.02189893e-01 -5.24200976e-01
-2.84273803e-01 -2.83303082e-01 -2.19907582e-01 -9.80853081e-01
-2.44763061e-01 -7.80083001e-01 3.25089246e-01 -1.75546825e+00
3.65306377e-01 -5.53478226e-02 -8.69803309e-01 5.06454289e-01
-2.67874688e-01 7.58748293e-01 -5.92628419e-02 -3.36731464e-01
-5.31108856e-01 5.43860912e-01 1.29901803e+00 -4.13949877e-01
-8.22533108e-03 -1.19816467e-01 -8.09745252e-01 6.11673355e-01
7.93140829e-01 -5.93725920e-01 -3.18485916e-01 -8.09411526e-01
-4.37487900e-01 1.46728024e-01 3.82552326e-01 -1.15105796e+00
3.29560459e-01 4.63530719e-01 6.90619826e-01 -5.67665100e-01
3.06475788e-01 -9.12654936e-01 -9.03212801e-02 1.08407450e+00
-4.31638211e-01 2.58945614e-01 3.96622717e-01 3.70179921e-01
-2.79437244e-01 -8.78873840e-02 1.21843064e+00 -2.97601670e-01
-6.46813512e-01 2.15234295e-01 -4.57856476e-01 -7.53316581e-02
8.74620259e-01 -6.17179042e-03 -3.96191813e-02 -2.97295928e-01
-1.06053174e+00 3.24005750e-03 -1.75211072e-01 -1.56643376e-01
1.16596687e+00 -1.13344610e+00 -9.85820532e-01 9.72260460e-02
-2.41868824e-01 -3.89568597e-01 7.38509297e-01 1.26157320e+00
-7.98145354e-01 1.54153004e-01 -8.76979828e-02 -6.22358978e-01
-1.47268283e+00 4.10749644e-01 6.46119893e-01 -7.83411860e-01
-9.76906776e-01 1.17296684e+00 1.69964537e-01 6.57959953e-02
3.69226098e-01 -6.10079169e-01 8.50785300e-02 -3.24195713e-01
6.47570610e-01 3.10528874e-01 5.47068238e-01 -2.28799671e-01
-4.19424027e-01 2.77664393e-01 -4.48186398e-01 -2.22801343e-02
1.67432559e+00 -1.05394848e-01 -1.37113929e-01 -1.99374065e-01
1.43747747e+00 -3.27935010e-01 -1.31437171e+00 -1.52393803e-01
-1.83597103e-01 -9.25158113e-02 9.37804937e-01 -1.13224399e+00
-1.47817957e+00 8.67699623e-01 1.46830130e+00 -2.87138581e-01
1.89267766e+00 -2.25792125e-01 1.12560630e+00 9.22312811e-02
-1.55222744e-01 -9.97079968e-01 2.70646513e-01 2.70069987e-01
1.04478967e+00 -1.37894726e+00 1.07448854e-01 -3.16840142e-01
-6.12191439e-01 9.82459009e-01 3.53829414e-01 -4.27530706e-01
8.58986199e-01 7.63672948e-01 2.88913697e-01 -2.26186946e-01
-7.42978096e-01 -7.33307824e-02 3.10337573e-01 4.24807459e-01
1.00272429e+00 -2.64466316e-01 -6.15042686e-01 4.38975424e-01
1.49637252e-01 2.24491239e-01 4.59482163e-01 1.03709626e+00
-4.53466564e-01 -9.07544374e-01 -2.39467368e-01 5.26500702e-01
-1.02400112e+00 -6.45405948e-01 3.88546735e-01 5.42710304e-01
1.15742847e-01 9.89356041e-01 -1.99539423e-01 -3.09184909e-01
5.26334882e-01 -3.42626452e-01 4.89630610e-01 -2.96755850e-01
-1.08034050e+00 4.93174404e-01 -3.28557828e-04 -6.48491621e-01
-3.69462341e-01 -3.53199542e-01 -1.23959517e+00 -3.20700496e-01
-8.96827579e-02 -5.03855348e-02 5.24799764e-01 5.28413713e-01
2.10109860e-01 1.30413580e+00 5.15085816e-01 -4.36220139e-01
-5.46862602e-01 -8.77934694e-01 -4.08945203e-01 3.20531994e-01
6.23093963e-01 -1.37019753e-01 -2.68150032e-01 1.83251053e-01] | [13.491364479064941, -2.5111496448516846] |
3d3701b0-7a76-420c-a4b0-a1b4d698a448 | metagad-learning-to-meta-transfer-for-few | 2305.10668 | null | https://arxiv.org/abs/2305.10668v1 | https://arxiv.org/pdf/2305.10668v1.pdf | MetaGAD: Learning to Meta Transfer for Few-shot Graph Anomaly Detection | Graph anomaly detection has long been an important problem in various domains pertaining to information security such as financial fraud, social spam, network intrusion, etc. The majority of existing methods are performed in an unsupervised manner, as labeled anomalies in a large scale are often too expensive to acquire. However, the identified anomalies may turn out to be data noises or uninteresting data instances due to the lack of prior knowledge on the anomalies. In realistic scenarios, it is often feasible to obtain limited labeled anomalies, which have great potential to advance graph anomaly detection. However, the work exploring limited labeled anomalies and a large amount of unlabeled nodes in graphs to detect anomalies is rather limited. Therefore, in this paper, we study a novel problem of few-shot graph anomaly detection. We propose a new framework MetaGAD to learn to meta-transfer the knowledge between unlabeled and labeled nodes for graph anomaly detection. Experimental results on six real-world datasets with synthetic anomalies and "organic" anomalies (available in the dataset) demonstrate the effectiveness of the proposed approach in detecting anomalies with limited labeled anomalies. | ['Kai Shu', 'Canyu Chen', 'Kaize Ding', 'Xiongxiao Xu'] | 2023-05-18 | null | null | null | null | ['graph-anomaly-detection'] | ['graphs'] | [ 3.92720103e-01 1.06752120e-01 9.71296951e-02 -2.11109176e-01
-2.59579718e-01 -3.86450976e-01 4.40948248e-01 7.37531364e-01
9.92036536e-02 6.24955297e-01 -4.73594636e-01 -3.59095782e-01
-9.09992903e-02 -1.08911538e+00 -6.39834344e-01 -6.09476805e-01
-3.75096411e-01 3.39133620e-01 3.72823745e-01 -2.06931934e-01
4.68530774e-01 4.95292276e-01 -1.45390022e+00 -9.97336954e-02
1.10462761e+00 7.46323407e-01 -4.31785315e-01 3.05123717e-01
-5.74792087e-01 9.39465523e-01 -5.85072696e-01 -3.44154596e-01
3.51705521e-01 -6.04083776e-01 -6.66167200e-01 6.24630988e-01
3.14518422e-01 -6.79977909e-02 -2.39259675e-01 1.59120309e+00
1.48732498e-01 1.64371118e-01 5.07440686e-01 -1.71741283e+00
-4.25210774e-01 3.72157425e-01 -1.14207959e+00 6.90177739e-01
2.69158244e-01 -3.29263508e-02 8.54308546e-01 -5.92557311e-01
4.64284271e-01 1.08730710e+00 5.34260511e-01 4.85106826e-01
-9.10714865e-01 -5.60370922e-01 5.24258792e-01 2.79096097e-01
-1.17015612e+00 -2.46747166e-01 1.16682851e+00 -3.06260407e-01
4.90948379e-01 1.69382557e-01 4.96422619e-01 9.60719883e-01
-2.11419109e-02 5.68748593e-01 8.59307766e-01 -5.25245488e-01
3.88104558e-01 3.20788808e-02 3.68195891e-01 8.24121833e-01
6.70246363e-01 -2.59558439e-01 -3.15971553e-01 -4.13314611e-01
3.00952345e-01 3.59120250e-01 2.13767518e-03 -4.45657015e-01
-8.42718422e-01 7.95252800e-01 4.97467667e-01 3.87255520e-01
-3.13319266e-01 -2.87106097e-01 6.59003615e-01 7.37063527e-01
8.91696632e-01 3.32323492e-01 -6.02996051e-02 1.53084427e-01
-3.84346336e-01 -8.58861879e-02 8.08880687e-01 8.41149688e-01
9.33063865e-01 4.90985423e-01 3.07405829e-01 6.88560843e-01
2.42963195e-01 2.27510571e-01 4.33253407e-01 -1.97382778e-01
5.13994336e-01 1.25309646e+00 -1.15605317e-01 -1.50257194e+00
-2.09836587e-01 -3.55839521e-01 -1.03144062e+00 -3.32397521e-02
4.87189800e-01 -1.07299112e-01 -1.12688136e+00 1.40521634e+00
5.55317521e-01 7.52882779e-01 -3.05251461e-02 5.68486631e-01
6.46137595e-01 5.23285508e-01 -5.37632732e-03 -4.16341335e-01
8.71264875e-01 -8.35341334e-01 -7.37550795e-01 -3.60087872e-01
1.00437486e+00 -5.39166987e-01 9.93599415e-01 2.04829291e-01
-3.49064916e-01 -5.55510297e-02 -8.67864072e-01 6.37508988e-01
-6.07025087e-01 -5.37013173e-01 5.99798083e-01 8.34211349e-01
-6.37601316e-01 5.13566613e-01 -7.89077997e-01 -8.22439492e-01
4.62235481e-01 8.31759423e-02 -3.18594307e-01 -3.56987596e-01
-9.64119494e-01 4.03418750e-01 7.11728096e-01 1.59297884e-01
-7.95357227e-01 -1.85787752e-01 -9.40572977e-01 -5.69341555e-02
1.02630413e+00 -5.69316521e-02 8.38357389e-01 -1.19096410e+00
-6.75047696e-01 7.41415977e-01 4.34938259e-02 -5.74150920e-01
4.37197924e-01 1.40531659e-01 -9.76539731e-01 -2.43890081e-02
1.63214296e-01 -1.91671282e-01 9.07457888e-01 -1.08012760e+00
-6.22574627e-01 -7.03630030e-01 -1.29655823e-02 7.15671405e-02
-7.61064529e-01 -1.30719841e-01 -3.71665172e-02 -6.89680874e-01
4.17791754e-01 -8.11811090e-01 -3.76526505e-01 -4.51127023e-01
-7.15027153e-01 -1.57919258e-01 1.42097402e+00 -3.94168824e-01
1.19001079e+00 -2.25494552e+00 -3.67252290e-01 7.11593091e-01
3.52052689e-01 2.14721352e-01 -1.84935275e-02 5.58068991e-01
-2.28389621e-01 2.10846037e-01 -6.40908360e-01 4.95606624e-02
-3.64212185e-01 3.54206771e-01 -2.45825380e-01 5.22358000e-01
1.61255628e-01 4.51477617e-01 -1.11890841e+00 -5.32019734e-01
1.15320392e-01 -1.89164549e-01 -2.65727133e-01 2.03391179e-01
-2.99310714e-01 8.18428278e-01 -8.54123592e-01 1.08843553e+00
6.17263556e-01 -4.81443971e-01 5.55662587e-02 2.85885364e-01
3.64825547e-01 -3.65517437e-01 -1.34201515e+00 1.32118559e+00
1.27030790e-01 3.18799466e-01 -2.51316577e-01 -1.50313306e+00
9.63777184e-01 2.73771733e-01 6.08239233e-01 -5.09937346e-01
-3.52076702e-02 3.24642658e-01 1.98894024e-01 -6.17702127e-01
2.53504127e-01 5.52089550e-02 6.84303045e-02 5.46653450e-01
-1.69402391e-01 3.78966898e-01 5.64435959e-01 4.45538670e-01
1.52690649e+00 -4.18988138e-01 3.28353375e-01 -1.20451497e-02
7.35066712e-01 1.73001438e-01 6.18159056e-01 6.65039718e-01
-1.94351673e-01 3.55949134e-01 7.92024553e-01 -5.89071572e-01
-8.44153345e-01 -7.76040912e-01 2.54498720e-01 7.70620883e-01
1.27725720e-01 -4.33027774e-01 -6.79220200e-01 -1.25878906e+00
-1.66249990e-01 5.90672493e-01 -4.31038916e-01 -3.41545075e-01
-3.92327994e-01 -1.09727430e+00 4.22543138e-01 1.26021788e-01
5.49024820e-01 -1.32545745e+00 -2.85240971e-02 2.56840110e-01
-1.45478575e-02 -1.15000188e+00 -1.58411935e-01 -1.85433149e-01
-1.01522994e+00 -1.56886590e+00 -1.54019386e-01 -7.00395286e-01
1.32252872e+00 4.12779063e-01 9.12394047e-01 5.58017552e-01
-4.98572528e-01 4.15135235e-01 -6.16664767e-01 -5.69772005e-01
-5.91806889e-01 -1.79352872e-02 3.67587730e-02 5.58671534e-01
6.01449311e-01 -6.10267043e-01 -3.49478811e-01 1.44213691e-01
-1.19083118e+00 -6.13080859e-01 4.13923949e-01 6.83642566e-01
3.78294140e-01 5.09447157e-01 8.63298118e-01 -1.69070983e+00
7.49611437e-01 -9.20943499e-01 -6.84390306e-01 2.42777884e-01
-8.54532242e-01 -1.80415645e-01 8.25214326e-01 -2.48853356e-01
-9.08667624e-01 -1.99217945e-01 1.68002307e-01 -4.98764873e-01
-4.45908934e-01 7.76069582e-01 -1.41223609e-01 -2.92677730e-01
6.10286653e-01 1.69204190e-01 -9.14725289e-03 -4.05337274e-01
-3.16261090e-02 3.29664320e-01 2.32206792e-01 -4.23861951e-01
1.00359964e+00 3.85227829e-01 3.67518187e-01 -1.08107805e+00
-8.49140882e-01 -7.33461559e-01 -4.64406580e-01 -1.76625267e-01
4.07355845e-01 -5.56495726e-01 -9.41984206e-02 6.18621290e-01
-5.82432032e-01 2.51167446e-01 -3.88483673e-01 1.98296651e-01
-1.84142999e-02 9.55488205e-01 -3.59198093e-01 -7.80053496e-01
-3.67120653e-01 -7.91656077e-01 5.84120393e-01 1.79590046e-01
1.12961099e-01 -1.22099960e+00 6.77647889e-02 3.81574668e-02
1.48488656e-01 7.81627893e-01 1.15822375e+00 -1.38994575e+00
-5.17678738e-01 -6.70493543e-01 -1.43649638e-01 1.83313429e-01
6.53065860e-01 1.52298436e-01 -7.83208191e-01 -5.23060024e-01
-5.02127707e-02 -2.52606869e-01 5.17903984e-01 -6.72013313e-02
1.30508471e+00 -2.48775020e-01 -4.00128484e-01 8.45705047e-02
1.27375865e+00 2.57105291e-01 4.18803632e-01 1.83065102e-01
1.05255449e+00 6.08425558e-01 7.65402257e-01 5.88807881e-01
-5.33852130e-02 6.37128996e-03 6.05864406e-01 5.68099469e-02
4.06123400e-01 -2.57777333e-01 1.60771459e-01 1.02393579e+00
-3.82471383e-02 -5.16980112e-01 -1.25503588e+00 6.15022779e-01
-1.93056536e+00 -8.64897549e-01 -3.56184900e-01 2.29353619e+00
1.18177757e-01 4.16363031e-01 1.19137868e-01 5.57042658e-01
1.22371078e+00 2.37473667e-01 -7.39529252e-01 -1.37281328e-01
9.99655426e-02 -1.35614067e-01 1.77761674e-01 -1.21263359e-02
-1.20634639e+00 7.31646299e-01 4.88611507e+00 6.53705657e-01
-9.65665758e-01 -5.60139343e-02 6.05704725e-01 4.59929556e-01
-2.21926361e-01 2.13210300e-01 -3.45657080e-01 4.96109784e-01
8.38774979e-01 -2.94845641e-01 1.46552056e-01 9.00013804e-01
-1.14075989e-01 -3.24794166e-02 -9.79079962e-01 8.11111391e-01
1.03314675e-01 -7.47674882e-01 2.04319611e-01 6.62773252e-02
7.92917013e-01 -1.77878708e-01 -1.45026788e-01 3.46175700e-01
1.57638490e-01 -8.12428713e-01 -1.97266936e-01 2.26963297e-01
3.58479798e-01 -9.48437512e-01 8.44159782e-01 6.11404419e-01
-1.19835389e+00 -1.34380400e-01 -3.69499773e-01 1.06487185e-01
-1.90569192e-01 7.45953321e-01 -1.01665413e+00 7.32817590e-01
6.39890075e-01 9.96101737e-01 -8.62710297e-01 1.15692627e+00
-2.24841945e-02 8.17693591e-01 -2.57108599e-01 1.34625897e-01
3.38025451e-01 -4.80077505e-01 7.66224086e-01 5.44215918e-01
5.06833434e-01 1.52823210e-01 5.86997390e-01 3.76634717e-01
-2.56957740e-01 5.52689016e-01 -1.30590260e+00 -3.80710870e-01
2.27653578e-01 1.20952260e+00 -1.24634802e+00 -1.76813439e-01
-7.62645245e-01 7.33156741e-01 2.36298800e-01 3.05073261e-01
-5.48153162e-01 -3.55536431e-01 4.65073995e-03 2.55082577e-01
-1.23195663e-01 1.46690145e-01 2.11680353e-01 -1.36562157e+00
7.08201453e-02 -9.49751556e-01 1.05670917e+00 -1.88908726e-01
-1.61293638e+00 4.35886025e-01 -1.28083766e-01 -1.62451768e+00
-2.79401660e-01 -3.26492071e-01 -1.05431819e+00 3.35629582e-01
-1.21964371e+00 -9.86188412e-01 -5.91395080e-01 8.32067132e-01
4.89107013e-01 -5.01005948e-01 7.26170421e-01 4.00607735e-01
-8.11046958e-01 4.05885875e-01 1.72463477e-01 3.52753282e-01
5.99918723e-01 -1.25657713e+00 4.55186933e-01 1.30699670e+00
3.18349391e-01 2.64597356e-01 6.25469089e-01 -1.01652038e+00
-1.18196619e+00 -1.26143575e+00 3.37770849e-01 5.75190596e-02
8.48452389e-01 -1.07767642e-01 -1.40937138e+00 9.21444356e-01
1.31558776e-02 6.28950119e-01 5.64088047e-01 9.14098993e-02
-7.19755962e-02 -1.11902528e-01 -1.37962401e+00 5.28009117e-01
1.04309344e+00 -2.48497635e-01 -3.35740536e-01 6.96360350e-01
4.13402826e-01 -1.60056859e-01 -4.81435716e-01 5.72412491e-01
-2.31561229e-01 -8.57708037e-01 5.09349465e-01 -9.37423468e-01
-2.77258344e-02 -2.89605737e-01 1.64354578e-01 -1.39212322e+00
1.51528284e-01 -3.01997423e-01 -3.65319878e-01 1.40901554e+00
2.79683858e-01 -9.69767272e-01 1.05221629e+00 3.95168811e-01
-2.36356650e-02 -2.56143212e-01 -8.21488321e-01 -6.78662837e-01
-5.22597969e-01 -2.34415710e-01 6.21254086e-01 1.55183184e+00
-1.64128587e-01 1.80899709e-01 -3.37661028e-01 5.67279160e-01
9.15825963e-01 1.32831708e-01 7.75689244e-01 -1.67088068e+00
1.11199662e-01 -4.49423268e-02 -8.34176064e-01 -7.96536058e-02
2.64791995e-01 -9.00026679e-01 -2.69645423e-01 -1.18335021e+00
-7.18212780e-03 -4.38304096e-01 -6.33293450e-01 4.30758357e-01
-2.98486710e-01 1.03261128e-01 -3.95812124e-01 1.87347516e-01
-7.29180753e-01 4.63792711e-01 9.41323161e-01 -2.79004186e-01
-6.96951151e-02 3.59124333e-01 -3.15974057e-01 1.13168800e+00
8.75640750e-01 -6.11136436e-01 -8.03695261e-01 3.19689997e-02
1.64304748e-01 -3.05198855e-03 2.02139750e-01 -9.53283131e-01
2.16814846e-01 -1.81260288e-01 3.64911519e-02 -4.71525490e-01
-1.59184784e-01 -1.01321399e+00 -4.88223620e-02 4.44304377e-01
2.56148353e-02 2.51858652e-01 -3.89134735e-02 1.18039739e+00
-5.29130757e-01 -3.64058614e-01 6.12841904e-01 -3.71731520e-01
-1.15403628e+00 6.96153641e-01 -2.87876651e-02 4.37991291e-01
1.41504407e+00 -1.68337241e-01 -2.77622789e-01 -4.46877688e-01
-7.31522501e-01 3.18717033e-01 4.36767161e-01 4.56106931e-01
7.72075176e-01 -1.17541683e+00 -5.69081426e-01 3.09706807e-01
5.85403025e-01 2.23552212e-01 2.79169351e-01 7.06981421e-01
-5.87557435e-01 -1.62302658e-01 -2.12642059e-01 -5.88800967e-01
-1.23689687e+00 8.79900098e-01 8.98809284e-02 -2.95555532e-01
-7.09223986e-01 4.25258666e-01 -1.33261934e-01 -4.08158869e-01
2.02536479e-01 2.18356386e-01 -4.17027205e-01 7.37816393e-02
2.78105855e-01 4.83596861e-01 1.85981452e-01 -5.20753980e-01
-1.68540329e-01 3.15732062e-01 -4.56688941e-01 5.46956420e-01
1.10918975e+00 -5.77832498e-02 -3.65341693e-01 6.19737029e-01
7.48151064e-01 3.99795175e-02 -5.70397735e-01 -6.15404189e-01
5.42498708e-01 -6.92140400e-01 -4.39335793e-01 -3.80268879e-02
-1.30239785e+00 8.17634702e-01 5.47491729e-01 8.12704086e-01
1.15234458e+00 -2.65917450e-01 6.30848050e-01 7.52951443e-01
4.83263999e-01 -9.60058272e-01 2.30281889e-01 1.64726436e-01
3.80850732e-01 -1.66156304e+00 -4.19308841e-02 -7.32535362e-01
-5.25620401e-01 1.03752756e+00 1.09560061e+00 -2.06783041e-01
7.71937311e-01 -2.24466637e-01 7.68302428e-03 -4.91789430e-01
-3.33352655e-01 -1.05384953e-01 6.81204349e-02 4.27202851e-01
1.77565470e-01 -1.74437985e-01 -1.88939422e-01 -8.85264501e-02
3.87479275e-01 -4.96984392e-01 8.56670916e-01 1.30750847e+00
-4.20425206e-01 -1.06618273e+00 -4.68309671e-01 9.92115200e-01
-7.11814225e-01 1.54897451e-01 -4.72381145e-01 7.99139977e-01
-1.49025634e-01 9.91971195e-01 2.72378419e-02 -1.06535554e-01
3.62110078e-01 2.01538846e-01 2.46708784e-02 -7.92289019e-01
-1.81657344e-01 -1.76713929e-01 2.19883099e-02 -2.85281748e-01
-5.07508934e-01 -4.04157281e-01 -1.16012371e+00 -3.24454337e-01
-3.85340333e-01 3.73396337e-01 2.16487095e-01 9.84877050e-01
1.60915390e-01 5.01519322e-01 7.79253960e-01 -1.42300844e-01
-4.43866998e-01 -1.00610256e+00 -9.97708917e-01 8.32716286e-01
2.01731667e-01 -6.82019055e-01 -5.94159365e-01 -2.15606913e-01] | [6.624269962310791, 5.76399040222168] |
0bf47dd9-99db-4803-bc89-dd6817cfce07 | mobile-user-interface-element-detection-via-1 | 2305.09699 | null | https://arxiv.org/abs/2305.09699v1 | https://arxiv.org/pdf/2305.09699v1.pdf | Mobile User Interface Element Detection Via Adaptively Prompt Tuning | Recent object detection approaches rely on pretrained vision-language models for image-text alignment. However, they fail to detect the Mobile User Interface (MUI) element since it contains additional OCR information, which describes its content and function but is often ignored. In this paper, we develop a new MUI element detection dataset named MUI-zh and propose an Adaptively Prompt Tuning (APT) module to take advantage of discriminating OCR information. APT is a lightweight and effective module to jointly optimize category prompts across different modalities. For every element, APT uniformly encodes its visual features and OCR descriptions to dynamically adjust the representation of frozen category prompts. We evaluate the effectiveness of our plug-and-play APT upon several existing CLIP-based detectors for both standard and open-vocabulary MUI element detection. Extensive experiments show that our method achieves considerable improvements on two datasets. The datasets is available at \url{github.com/antmachineintelligence/MUI-zh}. | ['Weiqiang Wang', 'Changhua Meng', 'Jun Lan', 'Haoxing Chen', 'Zhuoer Xu', 'Zhangxuan Gu'] | 2023-05-16 | mobile-user-interface-element-detection-via | http://openaccess.thecvf.com//content/CVPR2023/html/Gu_Mobile_User_Interface_Element_Detection_via_Adaptively_Prompt_Tuning_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Gu_Mobile_User_Interface_Element_Detection_via_Adaptively_Prompt_Tuning_CVPR_2023_paper.pdf | cvpr-2023-1 | ['optical-character-recognition'] | ['computer-vision'] | [ 2.76003748e-01 -5.04946828e-01 -4.61965591e-01 -1.64061084e-01
-8.92775118e-01 -6.72291338e-01 6.35491073e-01 4.14241292e-02
-3.95851433e-01 9.90359858e-02 1.50540844e-01 -3.61788720e-01
2.96198487e-01 -2.32842058e-01 -6.91594899e-01 -2.85643756e-01
4.62621510e-01 3.42749953e-01 4.55046326e-01 -7.95025155e-02
4.90000695e-01 1.59337506e-01 -1.67858899e+00 5.21020532e-01
7.78240502e-01 1.04121757e+00 6.61280155e-01 1.02366388e+00
-1.20816156e-01 6.35353029e-01 -5.54702401e-01 -2.35651776e-01
1.36086449e-01 -2.32386276e-01 -5.53635418e-01 1.61953017e-01
8.12613189e-01 -4.05226588e-01 -4.86397505e-01 1.06239462e+00
5.71877122e-01 -4.14494053e-02 7.72051632e-01 -1.37733948e+00
-8.48149478e-01 4.12036002e-01 -8.70064437e-01 4.27257866e-01
5.73481679e-01 2.06312299e-01 1.16086793e+00 -1.19564152e+00
6.14445686e-01 9.99349117e-01 5.42353213e-01 5.77673316e-01
-8.54059696e-01 -7.35747159e-01 8.95720199e-02 3.17050099e-01
-1.64990425e+00 -5.85485518e-01 6.70126438e-01 -4.64764386e-01
1.00421977e+00 5.31740367e-01 4.65125889e-01 9.96797323e-01
-1.09521197e-02 1.45545924e+00 6.98643982e-01 -7.44714797e-01
-1.33698627e-01 4.00559962e-01 4.62118268e-01 8.71962428e-01
9.75047573e-02 -2.55221307e-01 -5.99242747e-01 6.70323968e-02
5.66482604e-01 7.45782033e-02 -2.82440424e-01 -1.88052937e-01
-1.22367370e+00 3.17006201e-01 2.05272824e-01 2.37555489e-01
2.05405742e-01 7.16973171e-02 1.47402957e-01 2.04236120e-01
3.35095882e-01 2.05444396e-01 -2.99337417e-01 -2.35055789e-01
-8.97382319e-01 1.96912885e-01 4.86494780e-01 1.37454128e+00
8.15137148e-01 -3.21535587e-01 -6.04053020e-01 1.20103097e+00
2.46905059e-01 6.47283018e-01 5.24136662e-01 -6.95342362e-01
6.06594026e-01 8.26548278e-01 2.19969660e-01 -8.48198473e-01
-2.97725350e-01 -1.38857171e-01 -4.06932443e-01 -1.71375796e-01
2.14005083e-01 6.49948940e-02 -1.08292997e+00 1.39726198e+00
2.71493644e-01 5.83208129e-02 -4.95413542e-01 9.20951843e-01
1.02894044e+00 5.91924191e-01 8.20212439e-02 2.87681758e-01
1.50448608e+00 -1.10569882e+00 -5.25517941e-01 -5.50382137e-01
5.68210900e-01 -1.16571188e+00 1.39848387e+00 1.09568141e-01
-9.85488951e-01 -6.83171570e-01 -1.08041430e+00 -2.50948787e-01
-3.44152212e-01 7.60362685e-01 3.52544159e-01 5.57126820e-01
-1.07718968e+00 4.42681015e-02 -4.35037345e-01 -5.28274715e-01
4.07343477e-01 2.48948485e-01 -1.14313632e-01 -1.20104939e-01
-8.09346199e-01 4.88634408e-01 1.98828861e-01 -2.76666403e-01
-7.11502969e-01 -5.14012277e-01 -8.33756030e-01 9.35468003e-02
6.03887439e-01 -4.59924608e-01 1.45426106e+00 -1.12032783e+00
-1.24414337e+00 1.11433244e+00 -2.79946089e-01 -2.49941554e-02
3.58363569e-01 -1.60974011e-01 -4.12465572e-01 1.83246046e-01
1.66805327e-01 8.75251770e-01 1.35511291e+00 -1.22576010e+00
-1.05312335e+00 -2.40794688e-01 4.86115664e-02 1.73729062e-01
-7.30230451e-01 4.02089983e-01 -1.37144434e+00 -7.49937892e-01
-9.19791982e-02 -9.78829503e-01 2.88801372e-01 3.20274048e-02
-6.22552395e-01 -4.07056481e-01 7.74909139e-01 -6.08231068e-01
1.65791321e+00 -2.22327256e+00 -1.74553037e-01 6.71024546e-02
3.34504336e-01 4.19501126e-01 -3.61460865e-01 1.19733028e-01
2.16453865e-01 4.57159132e-02 1.54999614e-01 -5.33311009e-01
4.82482724e-02 -1.74567357e-01 -5.30333025e-03 3.43713433e-01
-7.61166494e-03 1.06875300e+00 -6.14436626e-01 -7.54210949e-01
1.69301927e-01 2.48418882e-01 -5.88450849e-01 2.41975144e-01
-3.12789470e-01 -9.74611416e-02 -3.16639692e-01 1.09759438e+00
5.76993287e-01 -4.63836372e-01 -8.18265323e-03 -1.78510293e-01
-1.36180773e-01 7.43086683e-03 -1.09460914e+00 1.46011388e+00
-2.29647860e-01 9.53340292e-01 1.18523002e-01 -6.39033914e-01
6.39904618e-01 1.99183404e-01 3.11380357e-01 -8.16387653e-01
3.61995280e-01 5.39142936e-02 -2.43200794e-01 -4.03562129e-01
9.45123136e-01 8.27085316e-01 -1.43366203e-01 3.59851450e-01
-1.26505256e-01 2.58776903e-01 3.34589720e-01 3.35608423e-01
1.00105786e+00 4.19702716e-02 2.38601893e-01 -8.32531303e-02
6.22962832e-01 -1.01474533e-02 2.89179236e-01 1.09988785e+00
-5.18841863e-01 8.73772442e-01 5.95521554e-02 -4.39289138e-02
-1.18858278e+00 -8.24167430e-01 -5.39911687e-02 1.62418854e+00
4.23960447e-01 -6.08349383e-01 -8.22313786e-01 -5.13318777e-01
-5.42574339e-02 3.80585432e-01 -5.51785469e-01 8.60886276e-02
-3.53152514e-01 -4.41179246e-01 3.51073533e-01 7.08223879e-01
4.02311236e-01 -9.38774884e-01 -3.81064177e-01 -9.62747335e-02
-5.27411580e-01 -1.02147305e+00 -1.14855325e+00 -7.16038719e-02
-3.40726882e-01 -1.00142062e+00 -7.16577351e-01 -9.95274246e-01
5.85400879e-01 9.69134808e-01 1.06057751e+00 3.25300187e-01
-6.79788411e-01 8.92846286e-01 -3.46666515e-01 -4.57530230e-01
-4.35207248e-01 2.05558732e-01 -1.76064566e-01 9.33578685e-02
6.81133926e-01 -2.24609356e-02 -8.85751009e-01 6.67559683e-01
-7.36877441e-01 2.25684121e-01 3.74962538e-01 6.21188939e-01
6.38607562e-01 -3.49946737e-01 -1.12372609e-02 -6.07698917e-01
5.62407374e-01 -1.71706498e-01 -5.79785764e-01 4.80146736e-01
-4.75047469e-01 -2.01484993e-01 9.19160917e-02 -7.30019689e-01
-1.06174767e+00 1.28564611e-01 5.55503778e-02 -5.89405835e-01
-1.34214461e-01 1.70622289e-01 -2.58040637e-01 -7.62769654e-02
4.43923831e-01 2.88799614e-01 -3.54200363e-01 -5.57905853e-01
3.38584453e-01 1.19440329e+00 7.33189821e-01 -3.55527610e-01
6.69858575e-01 3.66893619e-01 -6.89631522e-01 -9.13551509e-01
-7.10984826e-01 -1.10321903e+00 -3.84669185e-01 -3.33105326e-01
7.70929098e-01 -1.15698564e+00 -5.89638293e-01 6.14649296e-01
-1.07529259e+00 -4.20028389e-01 1.72233656e-01 1.05854988e-01
-3.75845879e-01 5.40331602e-01 -5.70785761e-01 -6.92493081e-01
-4.60880190e-01 -1.09334636e+00 1.42421937e+00 3.68263990e-01
-1.82884663e-01 -5.23665428e-01 -8.81802812e-02 5.46489716e-01
2.88677692e-01 -4.47984129e-01 6.08257234e-01 -4.46450144e-01
-6.33688509e-01 -6.26306474e-01 -5.03627777e-01 3.73381339e-02
5.19957487e-03 2.23686203e-01 -1.05228305e+00 -3.50821018e-01
-5.36647499e-01 -2.18069762e-01 9.61492062e-01 2.91760206e-01
1.43999434e+00 -2.01788768e-01 -5.33602178e-01 6.22453153e-01
1.29307520e+00 1.96625769e-01 6.96847498e-01 5.52186012e-01
9.73861992e-01 1.24706127e-01 7.50542879e-01 5.96506596e-01
4.41946298e-01 1.03006577e+00 2.99193263e-01 -3.19900289e-02
-4.02969390e-01 -2.18178809e-01 3.90089303e-01 5.45585155e-01
-7.03242347e-02 -4.26847845e-01 -8.10591400e-01 5.52338898e-01
-1.86950803e+00 -9.99204874e-01 -5.17104194e-02 2.14169574e+00
8.64076376e-01 7.96012580e-02 3.37055743e-01 -5.30505925e-02
8.53081584e-01 1.01061419e-01 -3.41148078e-01 -1.09161593e-01
-3.13368924e-02 -2.98379570e-01 6.31227970e-01 4.67951000e-01
-1.44654489e+00 1.09302807e+00 5.87012434e+00 1.20688796e+00
-1.03176224e+00 3.27232748e-01 4.57193673e-01 -1.24919817e-01
-2.42109090e-01 -2.22722262e-01 -1.34329319e+00 5.98476231e-01
6.49454832e-01 2.37579681e-02 3.71118814e-01 1.08597088e+00
4.44918200e-02 -1.90468356e-01 -9.36990023e-01 1.25121212e+00
4.35004443e-01 -1.27712798e+00 -7.39228949e-02 -1.27893567e-01
6.01704359e-01 1.18605867e-01 1.84835136e-01 2.58197874e-01
-8.69713649e-02 -4.93428171e-01 1.02766454e+00 2.30780661e-01
9.99174893e-01 -3.94874692e-01 2.10290834e-01 1.09541558e-01
-1.42322147e+00 -2.13980615e-01 -4.71851915e-01 2.56824106e-01
-3.30364764e-01 -7.19081908e-02 -9.80789125e-01 -3.92151549e-02
8.54407787e-01 5.93358576e-01 -1.14895380e+00 1.25626683e+00
-4.76159193e-02 6.37757838e-01 -3.16139072e-01 -2.07946479e-01
-3.06002237e-03 4.82511856e-02 7.16898024e-01 1.58360350e+00
1.71739995e-01 -2.28976160e-01 4.21743572e-01 6.03649676e-01
-2.17060655e-01 2.51406312e-01 -3.46803665e-01 -5.65604679e-02
6.16231084e-01 1.31483984e+00 -7.93795466e-01 -4.92407680e-01
-6.65429235e-01 1.41083586e+00 2.46305540e-01 2.98824996e-01
-9.65418518e-01 -4.76034880e-01 6.97721422e-01 1.81347936e-01
5.20400345e-01 -1.50237918e-01 -1.12130672e-01 -1.32360554e+00
6.43513277e-02 -9.24269259e-01 5.30109227e-01 -9.78889287e-01
-8.37240934e-01 3.50354403e-01 -5.75067848e-02 -1.43045163e+00
-4.36669812e-02 -8.27650130e-01 -5.42418003e-01 4.25842911e-01
-1.25301409e+00 -1.49559069e+00 -5.60995996e-01 7.77340889e-01
1.05304444e+00 -2.05468968e-01 4.17472243e-01 5.96321344e-01
-8.86008978e-01 1.21125531e+00 3.15986067e-01 3.44506949e-01
9.51003075e-01 -1.14556766e+00 4.32901859e-01 1.03868592e+00
4.04341817e-01 4.69651312e-01 5.56427658e-01 -6.78556979e-01
-1.44832063e+00 -1.05647755e+00 8.91528904e-01 -8.38634312e-01
5.92205107e-01 -5.44386029e-01 -8.64200413e-01 7.69093156e-01
2.05686152e-01 -2.04333037e-01 4.19766247e-01 -7.63542801e-02
-5.35814941e-01 -6.17004409e-02 -5.87185800e-01 8.54982495e-01
1.06380296e+00 -6.13498628e-01 -3.68319184e-01 4.73548204e-01
5.57400525e-01 -4.74029571e-01 -4.11549598e-01 1.10008202e-01
6.09024048e-01 -6.99305058e-01 1.17179143e+00 -2.46985450e-01
2.18797907e-01 -3.36026907e-01 -1.27172545e-01 -6.65256798e-01
-5.17920077e-01 -5.24745345e-01 -3.07905465e-01 1.39934468e+00
4.23223227e-01 -1.83403209e-01 6.19545460e-01 5.86557329e-01
-4.74517532e-02 -3.79959643e-01 -6.01769686e-01 -6.47659898e-01
-5.16783118e-01 -6.65287256e-01 3.65692526e-01 5.77798724e-01
-1.31988488e-02 3.71715456e-01 -6.05871737e-01 6.22377396e-02
5.24526775e-01 1.15087986e-01 9.12675321e-01 -8.90560091e-01
-3.65067065e-01 -6.86022878e-01 -3.70165646e-01 -1.23593259e+00
-1.17195413e-01 -7.68574119e-01 1.90845534e-01 -1.27732563e+00
6.24357283e-01 -2.42097855e-01 -3.93963277e-01 6.28130019e-01
-5.86027265e-01 6.23224020e-01 3.33942562e-01 5.77179909e-01
-1.16351914e+00 1.45861194e-01 9.60001767e-01 -6.37578487e-01
-3.43949825e-01 7.05837086e-02 -6.92124188e-01 7.46203780e-01
7.79810429e-01 -1.77825406e-01 -2.00832665e-01 -4.75845039e-01
2.38122474e-02 -4.20260370e-01 4.22027200e-01 -1.06718075e+00
2.93989509e-01 -1.33275956e-01 4.87054706e-01 -8.76960158e-01
2.15802312e-01 -5.57288468e-01 -3.33488286e-01 1.16890326e-01
-4.03753102e-01 3.83281149e-02 2.68604726e-01 6.51872814e-01
1.15811929e-01 -3.78318459e-01 6.60264134e-01 1.95193999e-02
-1.31566226e+00 3.44058752e-01 -5.38503885e-01 -1.94596667e-02
7.29136646e-01 -3.61528248e-01 -3.01834285e-01 -3.42862040e-01
-2.11135045e-01 2.98219085e-01 6.72269762e-01 8.12987685e-01
7.72549868e-01 -1.12551844e+00 -4.96709526e-01 1.71686023e-01
6.55473590e-01 -5.57502568e-01 1.41450062e-01 6.55008674e-01
-3.70232970e-01 4.69738036e-01 -8.84209052e-02 -7.38719285e-01
-1.81412971e+00 7.76285946e-01 2.13482738e-01 4.40572426e-02
-5.19845486e-01 9.47046340e-01 3.23837042e-01 -8.97384807e-02
6.64787591e-01 -1.35904700e-01 -2.89075226e-01 2.05165483e-02
8.41524005e-01 2.95103580e-01 -6.04481027e-02 -6.42910480e-01
-3.98748577e-01 5.66364050e-01 -3.62882346e-01 4.23584692e-02
7.37366617e-01 -6.72473967e-01 1.50702432e-01 1.94280565e-01
1.19572115e+00 -4.83041108e-02 -1.13288021e+00 -5.35364628e-01
-2.74574012e-02 -6.28557861e-01 -5.47391437e-02 -7.07998097e-01
-7.38979280e-01 8.00839961e-01 9.62129653e-01 -2.33407035e-01
1.06232703e+00 1.15029052e-01 7.12709606e-01 3.88423085e-01
2.42522478e-01 -1.43986440e+00 4.28255945e-01 4.60132718e-01
7.83874750e-01 -1.49354970e+00 -9.23656821e-02 -4.03119057e-01
-7.47202694e-01 8.69300365e-01 9.26000834e-01 2.56954432e-01
4.80969369e-01 1.70292944e-01 2.24250749e-01 -4.42998530e-03
-4.71999735e-01 -4.95721996e-01 6.37967706e-01 5.29892147e-01
3.94791782e-01 7.61033744e-02 -9.79723483e-02 4.32849914e-01
2.23705754e-01 -1.37507319e-01 3.09020132e-01 9.03156042e-01
-7.08037138e-01 -9.80354011e-01 -5.42285442e-01 5.11089683e-01
-3.07439119e-01 -1.74731642e-01 -3.66993815e-01 6.76310241e-01
6.84897322e-03 1.00698984e+00 1.46106377e-01 -5.80791891e-01
1.14615254e-01 4.11536880e-02 3.77698898e-01 -5.08913279e-01
-4.01111573e-01 2.93760329e-01 -1.36208117e-01 -6.61728442e-01
-1.76155820e-01 -5.75525880e-01 -9.64070618e-01 -1.41879171e-01
-3.81251305e-01 -2.96138436e-01 5.67490816e-01 6.88297629e-01
4.63665396e-01 3.42759818e-01 3.66790771e-01 -9.63508129e-01
-3.96893889e-01 -8.82345557e-01 -3.61161470e-01 5.53696334e-01
3.04289520e-01 -6.93169773e-01 -7.81766921e-02 2.46096104e-01] | [10.866461753845215, 1.9094696044921875] |
efc52455-20e1-4c79-8e79-3820b03f47ab | a-joint-model-for-graph-based-chinese | null | null | https://aclanthology.org/2020.ccl-1.76 | https://aclanthology.org/2020.ccl-1.76.pdf | A Joint Model for Graph-based Chinese Dependency Parsing | In Chinese dependency parsing, the joint model of word segmentation, POS tagging and dependency parsing has become the mainstream framework because it can eliminate error propagation and share knowledge, where the transition-based model with feature templates maintains the best performance. Recently, the graph-based joint model (Yan et al., 2019) on word segmentation and dependency parsing has achieved better performance, demonstrating the advantages of the graph-based models. However, this work can not provide POS information for downstream tasks, and the POS tagging task was proved to be helpful to the dependency parsing according to the research of the transition-based model. Therefore, we propose a graph-based joint model for Chinese word segmentation, POS tagging and dependency parsing. We designed a charater-level POS tagging task, and then train it jointly with the model of Yan et al. (2019). We adopt two methods of joint POS tagging task, one is by sharing parameters, the other is by using tag attention mechanism, which enables the three tasks to better share intermediate information and improve each other’s performance. The experimental results on the Penn Chinese treebank (CTB5) show that our proposed joint model improved by 0.38% on dependency parsing than the model of Yan et al. (2019). Compared with the best transition-based joint model, our model improved by 0.18%, 0.35% and 5.99% respectively in terms of word segmentation, POS tagging and dependency parsing. | ['Yufeng Chen', 'Jinan Xu', 'Yujie Zhang', 'Mingtong Liu', 'Xingchen Li'] | null | null | null | null | ccl-2020-10 | ['chinese-word-segmentation'] | ['natural-language-processing'] | [-3.90578300e-01 1.72738299e-01 -1.95990279e-01 -2.11932406e-01
-7.89136946e-01 -4.59887028e-01 5.12377024e-02 1.21495388e-01
-6.33859456e-01 7.59467542e-01 3.17712665e-01 -6.68618441e-01
3.62280339e-01 -7.85453141e-01 -4.10169750e-01 -6.69041395e-01
1.99110359e-01 2.54378617e-01 8.60247076e-01 -5.84496818e-02
1.93478227e-01 1.50319397e-01 -6.26364648e-01 4.44900580e-02
1.13972104e+00 3.53382677e-01 7.35855162e-01 5.76589882e-01
-7.18610823e-01 3.50621074e-01 -6.09885275e-01 -5.50406814e-01
-2.08565697e-01 -4.90067542e-01 -9.17729497e-01 -3.35620433e-01
-6.17355943e-01 -8.88567641e-02 -1.59570560e-01 1.28909552e+00
4.37166601e-01 3.02146282e-02 2.34897569e-01 -8.01669955e-01
-9.57826734e-01 1.21974087e+00 -6.54594898e-01 2.13049546e-01
-8.38943943e-02 -2.23116145e-01 1.20142436e+00 -3.78184468e-01
3.15800369e-01 1.33015430e+00 6.61594093e-01 7.19227314e-01
-5.64576626e-01 -7.33878970e-01 6.76254332e-01 7.83839002e-02
-1.11204934e+00 2.16138542e-01 5.11721671e-01 -4.15504090e-02
1.25337541e+00 -2.15759993e-01 6.07605398e-01 6.98294401e-01
3.45863789e-01 9.49742019e-01 1.09663391e+00 -6.05838299e-01
-1.24568395e-01 -1.99879676e-01 6.63958073e-01 7.41319418e-01
3.52736712e-01 -3.53881329e-01 -4.90832143e-02 6.60428554e-02
8.41587722e-01 -1.78723782e-01 -7.75530115e-02 4.11862999e-01
-8.59660447e-01 7.94915438e-01 3.24752718e-01 9.03769553e-01
-2.09605440e-01 -4.53704409e-02 1.37942925e-01 -2.72671968e-01
5.58389664e-01 9.98430848e-02 -9.21776712e-01 -2.54723549e-01
-4.30110633e-01 -9.99850556e-02 8.70708942e-01 1.15802813e+00
6.33618295e-01 -9.23528522e-02 -1.58371180e-01 8.64035547e-01
8.14501643e-01 5.40255725e-01 7.40076900e-01 -4.24120963e-01
6.91679478e-01 5.54297328e-01 -6.04374446e-02 -4.95574027e-01
-5.35734653e-01 -8.76005664e-02 -5.54740012e-01 -3.64352435e-01
3.56076568e-01 -7.81693399e-01 -1.24382162e+00 1.86233985e+00
1.23300664e-01 7.71709979e-02 2.60778785e-01 6.95935249e-01
8.99115205e-01 9.78741169e-01 8.38800251e-01 -4.35346216e-01
1.56793964e+00 -1.23524857e+00 -1.20955372e+00 -2.63275206e-01
1.05593216e+00 -1.16291296e+00 7.07970560e-01 1.21371940e-01
-9.32630539e-01 -6.08416200e-01 -7.80159414e-01 3.15405354e-02
-2.34308824e-01 2.82118898e-02 7.79927850e-01 9.66794729e-01
-9.58654881e-01 5.04562914e-01 -1.22261989e+00 -3.32314670e-01
2.11071491e-01 4.31517750e-01 -3.26527297e-01 -4.74464893e-02
-1.41371882e+00 8.24422002e-01 7.39522636e-01 2.71667451e-01
-2.55821198e-01 -5.04448786e-02 -6.82249963e-01 9.15975273e-02
2.33866379e-01 -2.50394911e-01 1.28828597e+00 -7.51996994e-01
-1.69368613e+00 3.22491020e-01 -2.13686571e-01 -5.03425784e-02
-8.59871209e-02 -5.06683946e-01 -4.66874599e-01 -7.44008347e-02
2.88842320e-01 5.02628446e-01 9.69261825e-02 -8.53043914e-01
-1.00931728e+00 -2.20709383e-01 2.16991641e-02 2.14290231e-01
-1.25068530e-01 4.64691162e-01 -9.61250126e-01 -4.77442890e-01
3.68576050e-01 -9.70164776e-01 -6.30935848e-01 -1.03838122e+00
-3.64334226e-01 -6.38754785e-01 6.24094427e-01 -1.11320460e+00
1.59826267e+00 -2.02277875e+00 -1.20330401e-01 -2.47274980e-01
-2.78206974e-01 7.50052512e-01 -1.54557824e-01 4.73819226e-01
-2.10229866e-02 7.94508576e-01 -4.42329407e-01 -5.43350041e-01
-2.20157117e-01 6.39153361e-01 3.10273468e-01 -1.54826259e-02
3.36190581e-01 9.53365088e-01 -7.87689149e-01 -8.79912317e-01
-5.48263118e-02 2.40765437e-01 -4.48726445e-01 4.12212163e-01
-2.11768784e-02 4.48585063e-01 -8.59256387e-01 7.22156584e-01
9.81518328e-01 2.19184518e-01 6.73602819e-01 -4.67386702e-03
-5.35588801e-01 3.31799507e-01 -1.01159632e+00 1.65923774e+00
-3.72300357e-01 -8.05764496e-02 6.16286462e-03 -8.17235827e-01
1.13425064e+00 5.69208860e-01 2.01617524e-01 -5.03180563e-01
1.89646259e-01 2.37903744e-01 3.47065806e-01 -5.09145021e-01
4.28050578e-01 -2.52451539e-01 -5.14189541e-01 3.01813841e-01
2.26635844e-01 1.74211144e-01 1.46750480e-01 1.57523543e-01
1.12406123e+00 5.02890646e-01 1.95001692e-01 -3.78806651e-01
4.49968576e-01 1.95775643e-01 1.17516017e+00 5.58727622e-01
-4.40683335e-01 5.85681677e-01 6.79358959e-01 6.11374676e-02
-5.87287366e-01 -8.20732355e-01 -6.28621057e-02 9.24195230e-01
2.24354297e-01 -3.36672664e-01 -9.68907893e-01 -1.16100800e+00
-5.45839012e-01 5.21426260e-01 -2.60795861e-01 1.13087207e-01
-1.15868616e+00 -1.23019087e+00 6.43587470e-01 7.94167995e-01
8.46321523e-01 -1.31124735e+00 7.98106492e-02 5.89207172e-01
-3.93496007e-01 -1.29360592e+00 -3.14907223e-01 4.32198226e-01
-9.28101778e-01 -9.41028953e-01 -6.37557805e-01 -1.29991746e+00
5.56709111e-01 1.07514396e-01 7.01528311e-01 2.44081676e-01
4.09361959e-01 -1.43824339e-01 -1.22247636e+00 -3.38058025e-01
-2.81042337e-01 3.71562511e-01 -3.88618261e-01 -2.93878466e-01
3.44107538e-01 -3.92162263e-01 -3.15193832e-01 6.59606010e-02
-7.53611684e-01 -2.16755979e-02 9.31958616e-01 7.88558125e-01
5.00775039e-01 -2.14281380e-01 6.16902947e-01 -1.27893889e+00
4.61260021e-01 -3.96731049e-01 -4.69497979e-01 2.23678082e-01
-6.32287264e-01 -4.11157310e-02 6.08356118e-01 -1.16555758e-01
-1.71880376e+00 1.09843038e-01 -7.66887903e-01 2.18019351e-01
-2.54441589e-01 5.87122619e-01 -7.08027244e-01 2.15022892e-01
-2.53940135e-01 -4.17088047e-02 -5.59466064e-01 -1.14067256e+00
2.32562661e-01 7.28107452e-01 2.81542063e-01 -4.49568063e-01
4.72488254e-01 -2.10924283e-01 -1.71627745e-01 -3.41342866e-01
-8.92855525e-01 -4.87485141e-01 -1.14567792e+00 4.77630556e-01
1.59778309e+00 -8.45646262e-01 -3.94788623e-01 7.19037652e-01
-1.64166176e+00 -3.56280327e-01 3.02909315e-01 8.99224222e-01
-8.65931287e-02 9.56262112e-01 -1.07873058e+00 -8.22557449e-01
-2.42757112e-01 -1.04325342e+00 7.11581826e-01 7.05160439e-01
3.15272152e-01 -1.17679191e+00 1.55143023e-01 7.43389204e-02
-5.93857951e-02 -5.27731739e-02 1.07990968e+00 -9.87268567e-01
-3.26708287e-01 -1.24971394e-03 -3.85942698e-01 4.72801000e-01
1.91734701e-01 5.98979704e-02 -6.81715250e-01 1.06922530e-01
-9.09075961e-02 2.60849446e-01 7.78878272e-01 3.28100324e-01
6.73314393e-01 6.15254641e-02 -4.54732150e-01 4.03522074e-01
1.51696837e+00 7.62893498e-01 6.51568413e-01 2.21432954e-01
9.44292188e-01 5.38193941e-01 1.07393241e+00 -2.67326962e-02
6.23487353e-01 2.68220186e-01 3.12919796e-01 -1.44109339e-01
-1.17994145e-01 -2.72503495e-01 4.07066435e-01 1.56088567e+00
-2.86270589e-01 -4.23105985e-01 -1.01312661e+00 6.44556582e-01
-2.17568898e+00 -3.68721068e-01 -8.22081506e-01 1.65105057e+00
8.35952342e-01 3.67204636e-01 -1.19708084e-01 -1.66335464e-01
1.08464253e+00 7.83787668e-02 1.67004600e-01 -8.33035648e-01
6.16098642e-02 6.10961080e-01 6.79231644e-01 5.66784978e-01
-1.18194735e+00 1.61548829e+00 5.70127916e+00 7.29368150e-01
-6.77172065e-01 4.27479714e-01 3.10286313e-01 8.23176146e-01
-2.07179233e-01 5.27710378e-01 -1.29224455e+00 5.61920345e-01
1.03553164e+00 1.81758568e-01 -1.44647256e-01 6.26763105e-01
5.42925894e-02 -2.22275645e-01 -4.19438720e-01 4.19840366e-01
-2.74846464e-01 -7.29106367e-01 -1.15508802e-01 -1.50976226e-01
5.30342937e-01 -5.53871542e-02 -6.05780363e-01 6.79064929e-01
8.50806892e-01 -7.56365657e-01 3.31678718e-01 2.26845190e-01
1.59189910e-01 -7.84108102e-01 1.39052486e+00 4.90288377e-01
-1.50800216e+00 2.42492169e-01 -6.05998933e-01 -3.01386118e-01
6.70381904e-01 4.62912381e-01 -4.16854650e-01 1.02011120e+00
7.15757668e-01 5.00860691e-01 -4.10717577e-01 8.37063372e-01
-8.47494960e-01 1.17370868e+00 -1.11172743e-01 -3.82187843e-01
5.63565850e-01 -3.52087080e-01 7.50062764e-02 1.59747398e+00
2.34004512e-01 4.62364614e-01 3.86399657e-01 3.23928237e-01
2.48366401e-01 4.59814101e-01 1.49074313e-03 -3.08309365e-02
5.78672051e-01 1.38990879e+00 -9.94705200e-01 -3.07226151e-01
-7.53357470e-01 9.04353917e-01 4.22015518e-01 1.51733726e-01
-8.99216294e-01 -8.04283977e-01 1.40809923e-01 -5.19765854e-01
6.56968474e-01 -5.61817110e-01 -4.16609585e-01 -8.81832063e-01
-1.94459200e-01 -2.05396235e-01 4.00809109e-01 -5.02211154e-01
-9.91181731e-01 7.19917834e-01 6.68582395e-02 -7.08842158e-01
6.54538050e-02 -7.13738620e-01 -8.93386424e-01 1.41097498e+00
-1.66228640e+00 -1.40933442e+00 1.28291756e-01 2.08503067e-01
4.52254027e-01 2.61419892e-01 9.57581401e-01 2.90250093e-01
-1.04857612e+00 5.55695951e-01 -1.40590563e-01 5.93535364e-01
5.56223929e-01 -1.38158798e+00 6.92260861e-01 1.20998788e+00
-8.67949277e-02 5.60604632e-01 1.83071792e-01 -9.10826027e-01
-1.08634388e+00 -1.07725358e+00 1.44340396e+00 -2.41635591e-01
6.08790874e-01 -1.30196512e-01 -1.11638820e+00 8.91019702e-01
3.51013750e-01 -1.03091031e-01 6.84846878e-01 2.09242091e-01
-3.19574669e-04 1.29368082e-01 -9.06641245e-01 4.23043817e-01
1.09056711e+00 9.92233977e-02 -9.32727456e-01 7.36117959e-02
1.28216660e+00 -1.86623439e-01 -9.74924088e-01 4.16577041e-01
3.28986138e-01 -8.93504143e-01 3.62713337e-01 -4.04769391e-01
1.70375779e-01 -2.31896460e-01 -6.48764614e-03 -1.26642478e+00
-7.53907144e-01 -3.84869128e-01 5.43736696e-01 1.88426232e+00
6.99030340e-01 -8.31141591e-01 4.98348296e-01 4.50609654e-01
-7.07709372e-01 -6.17567062e-01 -7.99493611e-01 -6.99852526e-01
4.69832659e-01 -3.19560736e-01 7.46542156e-01 7.36749053e-01
8.28073770e-02 5.17071307e-01 -8.45212266e-02 3.34382445e-01
4.86155562e-02 2.07191203e-02 3.24621826e-01 -1.02191126e+00
-2.86296278e-01 -2.50649631e-01 1.84621111e-01 -1.29411626e+00
2.27423012e-01 -6.60554290e-01 3.90608579e-01 -1.99125183e+00
1.19528875e-01 -7.83986330e-01 -6.05470657e-01 7.74904072e-01
-8.07195783e-01 -2.42844578e-02 3.41132790e-01 2.32756451e-01
-5.03153741e-01 2.34610692e-01 1.49990034e+00 4.62462068e-01
-3.24892819e-01 1.00193312e-02 -8.24322104e-01 7.77693033e-01
1.06640077e+00 -7.89553642e-01 9.53918844e-02 -7.43111372e-01
-2.11864829e-01 3.00161958e-01 -4.73879188e-01 -6.91429079e-01
1.59002140e-01 -2.38589153e-01 7.05068484e-02 -5.49518764e-01
1.23201301e-02 -5.05415857e-01 -3.35885674e-01 6.47847176e-01
2.78044283e-01 2.86756456e-01 1.82112917e-01 4.80984747e-01
-2.82756865e-01 -6.79875195e-01 5.51375747e-01 -4.15452629e-01
-9.04169500e-01 3.13092142e-01 -4.59694833e-01 6.15242345e-04
8.97761047e-01 1.10221103e-01 -2.80533612e-01 2.90918350e-01
-8.46087694e-01 2.83322036e-01 -1.06599197e-01 2.82479107e-01
1.24728367e-01 -1.00966227e+00 -6.05568647e-01 -5.82469888e-02
-4.22712177e-01 2.21672088e-01 2.63917059e-01 7.11995423e-01
-6.19098902e-01 4.86672610e-01 -1.52913690e-01 -1.92415670e-01
-1.08379471e+00 3.03033918e-01 -1.39680147e-01 -8.27001631e-01
-2.61745989e-01 9.67770159e-01 1.68677881e-01 -4.33438927e-01
-2.69329488e-01 -4.95998085e-01 -7.59428203e-01 -1.46216810e-01
6.09761029e-02 1.42695993e-01 -1.34447753e-01 -6.32432461e-01
-6.13591492e-01 7.39052236e-01 -8.71244818e-02 -2.08781496e-01
1.11101282e+00 -1.61901191e-01 -3.14375579e-01 2.31682435e-01
8.64721358e-01 3.14095497e-01 -9.54236627e-01 4.55098459e-03
2.32548878e-01 -8.71668980e-02 -9.57143083e-02 -9.45404470e-01
-8.67895722e-01 1.08939052e+00 1.69746727e-01 3.55432004e-01
1.03400898e+00 3.00393682e-02 1.22353482e+00 4.46110517e-02
3.76487285e-01 -1.00244999e+00 -3.45084488e-01 1.00475359e+00
1.75269812e-01 -1.01051354e+00 -4.14263755e-01 -8.97628188e-01
-6.92224205e-01 1.09937894e+00 8.96819890e-01 -3.12214613e-01
9.72052693e-01 3.16377610e-01 2.83672422e-01 3.03206116e-01
-5.02841949e-01 -6.50663555e-01 -8.67184326e-02 8.06848228e-01
7.19531417e-01 4.14644480e-01 -1.30572176e+00 1.41919231e+00
2.17980091e-02 -3.36027950e-01 3.49873215e-01 1.06084406e+00
-7.36914158e-01 -1.87502849e+00 -1.88139617e-01 3.38913016e-02
-9.46603298e-01 -4.13140863e-01 -1.49521485e-01 9.59634304e-01
3.48968923e-01 1.25413013e+00 2.43322197e-02 -4.27337199e-01
4.02933031e-01 2.40132704e-01 2.28947878e-01 -1.12059915e+00
-7.43960142e-01 3.92831028e-01 3.11430424e-01 -2.03059152e-01
-6.03589475e-01 -5.69831967e-01 -1.77056944e+00 1.36709929e-01
-6.78856254e-01 7.19757140e-01 6.74693525e-01 1.15337789e+00
2.13344265e-02 6.74052417e-01 6.51062846e-01 -3.09364706e-01
-2.06014916e-01 -1.27003098e+00 -6.90800905e-01 -3.52732167e-02
-4.44367021e-01 -3.35343778e-01 -1.39043421e-01 -1.25508472e-01] | [10.009744644165039, 10.058658599853516] |
a4c1022d-4cb1-413c-8799-434473d1482e | multi-head-cascaded-swin-transformers-with | 2207.08412 | null | https://arxiv.org/abs/2207.08412v2 | https://arxiv.org/pdf/2207.08412v2.pdf | Multi-branch Cascaded Swin Transformers with Attention to k-space Sampling Pattern for Accelerated MRI Reconstruction | Global correlations are widely seen in human anatomical structures due to similarity across tissues and bones. These correlations are reflected in magnetic resonance imaging (MRI) scans as a result of close-range proton density and T1/T2 parameters. Furthermore, to achieve accelerated MRI, k-space data are undersampled which causes global aliasing artifacts. Convolutional neural network (CNN) models are widely utilized for accelerated MRI reconstruction, but those models are limited in capturing global correlations due to the intrinsic locality of the convolution operation. The self-attention-based transformer models are capable of capturing global correlations among image features, however, the current contributions of transformer models for MRI reconstruction are minute. The existing contributions mostly provide CNN-transformer hybrid solutions and rarely leverage the physics of MRI. In this paper, we propose a physics-based stand-alone (convolution free) transformer model titled, the Multi-head Cascaded Swin Transformers (McSTRA) for accelerated MRI reconstruction. McSTRA combines several interconnected MRI physics-related concepts with the transformer networks: it exploits global MR features via the shifted window self-attention mechanism; it extracts MR features belonging to different spectral components separately using a multi-head setup; it iterates between intermediate de-aliasing and k-space correction via a cascaded network with data consistency in k-space and intermediate loss computations; furthermore, we propose a novel positional embedding generation mechanism to guide self-attention utilizing the point spread function corresponding to the undersampling mask. Our model significantly outperforms state-of-the-art MRI reconstruction methods both visually and quantitatively while depicting improved resolution and removal of aliasing artifacts. | ['Zhaolin Chen', 'Gary Egan', 'Mehrtash Harandi', 'Kamlesh Pawar', 'Mevan Ekanayake'] | 2022-07-18 | null | null | null | null | ['de-aliasing'] | ['computer-vision'] | [ 1.28071234e-01 -2.99988016e-02 -3.69214304e-02 -3.10796738e-01
-9.62515593e-01 8.71409103e-03 2.75518298e-01 -1.61235947e-02
-3.12840581e-01 5.12964189e-01 6.81386709e-01 7.57874101e-02
-5.81496418e-01 -6.16664767e-01 -7.11615026e-01 -9.76995707e-01
-4.74960804e-01 3.10905188e-01 4.07934666e-01 -3.87807578e-01
-6.48968220e-02 5.22261441e-01 -9.20052707e-01 3.65579665e-01
8.26715767e-01 1.00125921e+00 4.95633006e-01 3.84613067e-01
1.16485365e-01 7.82159567e-01 -7.06573501e-02 -2.02181563e-01
1.08475246e-01 -4.33712572e-01 -8.83962512e-01 -4.25366849e-01
1.93082407e-01 -3.54443073e-01 -8.44431341e-01 1.03457153e+00
9.53742802e-01 2.11958393e-01 5.47452152e-01 -8.12946439e-01
-1.01212072e+00 7.70619810e-01 -7.47476339e-01 9.13211286e-01
-1.85413644e-01 1.56562969e-01 6.39954567e-01 -1.00809515e+00
3.88783723e-01 6.68015540e-01 1.19377625e+00 4.07826960e-01
-1.22368133e+00 -5.69936216e-01 -2.46415928e-01 4.83975261e-01
-1.18865502e+00 -5.71280047e-02 1.00982285e+00 -3.28333646e-01
1.12291157e+00 2.50191242e-01 8.04680586e-01 1.00829947e+00
8.88068974e-01 4.54674572e-01 1.23638499e+00 -3.61579917e-02
-1.25456154e-01 -2.91846991e-01 1.07126881e-03 7.08602905e-01
-3.15321326e-01 1.54854402e-01 -5.87561429e-01 -1.79258838e-01
1.16083789e+00 2.28698075e-01 -7.34859169e-01 -2.80104160e-01
-1.47790265e+00 8.00121844e-01 1.10265863e+00 7.03741252e-01
-7.06918657e-01 1.40195161e-01 6.05044842e-01 -4.28750329e-02
5.26574850e-01 3.61470342e-01 -2.84341305e-01 3.44751924e-01
-9.77518797e-01 8.16678479e-02 1.69934127e-02 5.29976428e-01
3.33788246e-01 1.53171301e-01 -4.61480916e-01 9.84475374e-01
4.96010184e-02 2.34939948e-01 1.13855171e+00 -5.78315079e-01
8.58242661e-02 9.81595144e-02 -5.91222048e-01 -9.85518217e-01
-9.43815827e-01 -1.00378537e+00 -1.20374703e+00 -1.21641599e-01
1.36320040e-01 2.81479210e-01 -8.87008965e-01 1.74308538e+00
4.89625335e-01 4.81739521e-01 -3.61014396e-01 1.28853762e+00
1.09128296e+00 3.06541115e-01 4.86278050e-02 -1.69600844e-01
1.59161413e+00 -1.13462210e+00 -8.71250033e-01 7.01246262e-02
3.51349592e-01 -6.03574395e-01 1.13292766e+00 9.67619289e-03
-1.32999277e+00 -4.95530337e-01 -1.00888741e+00 -3.20602804e-01
-1.39020815e-01 -1.90701187e-01 7.08825946e-01 4.45261359e-01
-1.19947243e+00 9.27823067e-01 -9.78099167e-01 1.16202630e-01
4.88195270e-01 4.77128893e-01 -3.80838126e-01 -6.88402504e-02
-1.37493837e+00 1.11121774e+00 -2.26638932e-02 2.60905206e-01
-8.72800946e-01 -1.57280457e+00 -6.95231557e-01 -6.45670071e-02
5.93331605e-02 -8.71758044e-01 9.28233981e-01 -6.42285943e-01
-1.45172930e+00 5.33667386e-01 2.09856316e-01 -4.99625683e-01
5.00355482e-01 -1.06575236e-01 -4.63248044e-01 4.86832172e-01
1.19925044e-01 4.60842937e-01 8.58086944e-01 -7.64884412e-01
3.14107537e-01 -5.56827664e-01 -2.86996335e-01 2.23734781e-01
-1.60216331e-01 1.38122186e-01 -1.81250900e-01 -1.05492067e+00
3.29071403e-01 -7.99010694e-01 -4.49247211e-01 -3.47075537e-02
-2.72956848e-01 4.11233753e-01 6.59428835e-01 -1.07823706e+00
1.04366791e+00 -1.95973074e+00 1.20330475e-01 1.41391262e-01
6.62144244e-01 -1.05574258e-01 -1.23176113e-01 -1.01545893e-01
-7.24918246e-01 -2.32132792e-01 -4.16822582e-01 -4.95477393e-03
-4.92891133e-01 -2.70226356e-02 -1.95424166e-02 8.26792359e-01
-3.07308342e-02 1.08671498e+00 -9.70282495e-01 -4.50707585e-01
3.87590885e-01 1.01405859e+00 -7.24653900e-01 2.74607353e-02
5.52121878e-01 7.71769106e-01 -3.73115629e-01 4.66540545e-01
9.05749440e-01 -2.44439006e-01 -5.83383031e-02 -1.07088029e+00
-8.52076709e-02 1.15005903e-01 -4.50146645e-01 2.16078997e+00
-5.58041513e-01 1.38476253e-01 1.54810399e-01 -1.23760712e+00
4.52197105e-01 3.86436403e-01 1.16449189e+00 -9.88282740e-01
2.09980100e-01 2.90754020e-01 1.56133831e-01 -4.82806921e-01
1.75859585e-01 -5.55172443e-01 3.70111048e-01 2.78285801e-01
3.24135691e-01 -8.87037888e-02 -5.01185775e-01 1.08358942e-01
1.21397007e+00 3.59761133e-03 3.84354256e-02 -5.51125288e-01
3.51721406e-01 -3.08387458e-01 3.57025415e-01 6.40755773e-01
-3.47156882e-01 1.04024911e+00 1.33884838e-02 -5.73126435e-01
-1.14771569e+00 -1.19034982e+00 -4.73714232e-01 7.42525697e-01
-7.78825954e-02 -2.68255889e-01 -7.22333610e-01 -5.36328018e-01
-3.01995814e-01 2.23256320e-01 -9.38443422e-01 -3.39965254e-01
-7.98811853e-01 -1.27707744e+00 3.91628265e-01 6.78457379e-01
4.55651164e-01 -8.37340593e-01 -6.17094517e-01 5.10581672e-01
-3.89913380e-01 -1.05734444e+00 -7.75352776e-01 3.74378681e-01
-1.17021811e+00 -8.62934589e-01 -1.15056455e+00 -3.25791717e-01
4.83935833e-01 3.06093723e-01 9.51770604e-01 -4.40085270e-02
-6.94616020e-01 2.18510374e-01 -1.41948700e-01 -1.46291248e-04
4.22882847e-02 -5.17290495e-02 -1.28538264e-02 -1.88971967e-01
-1.76805273e-01 -1.03206038e+00 -1.09513652e+00 1.79064706e-01
-8.73089135e-01 1.54727951e-01 7.25263953e-01 1.21488822e+00
7.37951517e-01 -4.99827638e-02 3.86941433e-01 -5.50439417e-01
4.98800546e-01 -7.35142469e-01 3.22607271e-02 1.74236044e-01
-3.35596770e-01 1.59482613e-01 5.47868490e-01 -4.67613220e-01
-9.15971160e-01 -2.86454022e-01 -4.25251365e-01 -6.40770674e-01
2.89916068e-01 3.43247086e-01 2.71187752e-01 -5.85838914e-01
5.73091209e-01 4.85187441e-01 2.25691170e-01 -3.44639838e-01
1.06104895e-01 7.14163408e-02 6.56781673e-01 -4.66657788e-01
4.21884477e-01 6.46721721e-01 1.64832577e-01 -6.75569296e-01
-8.49032342e-01 -2.67100573e-01 -5.39577782e-01 -3.72338533e-01
9.72435951e-01 -7.42756426e-01 -5.80208659e-01 4.84204978e-01
-9.06240880e-01 -1.07831255e-01 -4.12695616e-01 8.86611760e-01
-5.26009738e-01 5.13182521e-01 -1.11257184e+00 -1.64803132e-01
-9.11132276e-01 -1.72603750e+00 9.51177657e-01 -2.59905215e-02
6.62755743e-02 -9.23043668e-01 -1.95347294e-02 3.85392457e-01
1.05324519e+00 2.51696348e-01 1.13431251e+00 -3.90226722e-01
-3.59572262e-01 2.76940823e-01 -3.64781022e-01 2.30648130e-01
6.12548962e-02 -6.66835248e-01 -9.13342476e-01 -2.43585914e-01
3.85930538e-01 -1.44127030e-02 7.17819393e-01 1.14583051e+00
1.68165267e+00 -3.36162522e-02 -4.89831492e-02 1.06676924e+00
1.33454037e+00 -9.25090685e-02 7.42683589e-01 3.09247702e-01
8.28141391e-01 3.70855153e-01 -1.13243058e-01 2.02613547e-01
3.87522250e-01 8.08766365e-01 3.63640159e-01 -5.26346147e-01
-7.00815797e-01 1.57659799e-02 4.96923849e-02 1.32588089e+00
-2.67162144e-01 5.48047245e-01 -8.38367581e-01 5.94018400e-01
-1.52285099e+00 -7.75068223e-01 -2.80227721e-01 2.16912222e+00
8.37661743e-01 -2.36737669e-01 -1.29646808e-02 -7.70092085e-02
5.68598986e-01 1.15480378e-01 -6.86967969e-01 -2.01887917e-03
-2.91945669e-03 6.83747649e-01 8.05558741e-01 3.99612248e-01
-9.20551598e-01 3.17539483e-01 6.47480869e+00 1.00082040e+00
-1.33571661e+00 9.89859581e-01 6.78229094e-01 -1.98441744e-01
-4.27783549e-01 -3.72075051e-01 -1.64014325e-01 4.21847224e-01
8.65019739e-01 4.94024344e-02 4.15196121e-01 5.67081749e-01
1.26066342e-01 2.08891794e-01 -5.99670410e-01 1.08288968e+00
-1.59862787e-02 -1.42536604e+00 -4.30610105e-02 -3.60236503e-02
4.24328953e-01 4.89210725e-01 1.89941779e-01 8.24676901e-02
-1.40832201e-01 -1.16524947e+00 6.05027020e-01 5.32479167e-01
9.68821824e-01 -7.44908571e-01 7.15604365e-01 -1.83674455e-01
-1.20771563e+00 -1.98944397e-02 -2.72400081e-01 5.08720696e-01
2.22494125e-01 9.60449934e-01 -4.67434496e-01 7.97309756e-01
1.09554958e+00 5.51309586e-01 -3.54483455e-01 1.04633963e+00
2.00278372e-01 3.93043518e-01 -2.25577518e-01 6.01111174e-01
3.46228689e-01 -1.74877614e-01 5.13980329e-01 1.18049645e+00
3.91038567e-01 2.54884213e-01 -1.68199223e-02 9.89118278e-01
1.84136063e-01 1.81598037e-01 -2.53975511e-01 5.97278059e-01
-9.03791785e-02 1.41790402e+00 -7.52867281e-01 -2.13246971e-01
-4.24959987e-01 1.01518655e+00 1.18849918e-01 3.01711470e-01
-1.07142866e+00 -1.53313473e-01 4.56223547e-01 4.30277139e-01
2.98136901e-02 -4.29970808e-02 -2.65648842e-01 -1.21358192e+00
-1.18220478e-01 -6.42343223e-01 3.01565349e-01 -8.75713646e-01
-1.45892239e+00 8.96973968e-01 5.67662418e-02 -8.31148863e-01
3.03002000e-01 -2.89897084e-01 -5.36636710e-01 9.25412416e-01
-1.77942455e+00 -1.20337117e+00 -4.05153185e-01 9.32939827e-01
3.94251704e-01 1.50196925e-01 7.44257152e-01 7.66586185e-01
-2.46805280e-01 5.24266005e-01 -6.78181499e-02 -8.17360654e-02
6.03857160e-01 -1.09482646e+00 4.15173620e-02 5.35003185e-01
-3.73498648e-01 7.90384591e-01 5.70778728e-01 -5.95926821e-01
-1.36574864e+00 -9.07462716e-01 3.53966027e-01 1.09341949e-01
8.59430611e-01 -1.18986607e-01 -1.19368815e+00 4.87350941e-01
1.38934910e-01 7.56801367e-01 7.14021027e-01 -2.69413143e-01
-2.91933060e-01 -2.19537646e-01 -1.41536331e+00 2.41291553e-01
1.02015054e+00 -6.71203613e-01 -5.29828250e-01 6.09277427e-01
7.58567274e-01 -7.77036309e-01 -1.35058093e+00 4.81857121e-01
5.57933092e-01 -1.04568160e+00 1.48615015e+00 -2.86045313e-01
6.11953974e-01 -1.05512179e-01 1.49884662e-02 -1.46063197e+00
-7.97121346e-01 -3.04493845e-01 4.67489362e-02 6.18638039e-01
4.68181856e-02 -7.24972606e-01 5.39435089e-01 3.71350318e-01
-6.13606215e-01 -9.50489223e-01 -1.44133210e+00 -4.38178837e-01
2.22835883e-01 -4.28727627e-01 5.82049727e-01 1.14091492e+00
-2.14926109e-01 -8.41435269e-02 -3.37030143e-01 2.34381601e-01
9.70159948e-01 -2.11919159e-01 -1.67037666e-01 -6.86522007e-01
-4.40979362e-01 -4.68249351e-01 -3.68032247e-01 -6.33845091e-01
-5.80336601e-02 -1.14044583e+00 -2.96910912e-01 -1.13460433e+00
5.69945395e-01 -6.15428448e-01 -7.81581700e-01 2.98358589e-01
1.98584013e-02 5.51454425e-01 1.90426204e-02 1.50093123e-01
-3.41070741e-02 7.24159896e-01 1.60287869e+00 -8.42358768e-02
1.21264316e-01 -3.45128447e-01 -5.60290933e-01 5.48332512e-01
5.24027109e-01 -4.31570798e-01 -4.20935839e-01 -4.48244512e-01
-3.14417817e-02 3.25911880e-01 6.64709032e-01 -1.22801769e+00
2.31303886e-01 2.71616727e-01 6.14699960e-01 -4.50433582e-01
2.24723205e-01 -8.63841832e-01 3.24108720e-01 5.09190321e-01
-2.12065354e-01 3.57527405e-01 2.04234615e-01 2.14016244e-01
-1.72235519e-01 -1.32983169e-02 1.13329375e+00 -3.90395552e-01
-1.95709929e-01 7.35399961e-01 -2.23076269e-01 -1.36860445e-01
7.22010911e-01 -2.70647705e-02 1.16797339e-03 -4.46928628e-02
-8.39565396e-01 -2.36962661e-01 -6.37888312e-02 2.39607140e-01
8.41787398e-01 -1.40914035e+00 -5.57022750e-01 2.15941966e-01
-2.56480694e-01 -2.34464154e-01 1.13525200e+00 1.72269607e+00
-4.24613684e-01 2.35799506e-01 -4.64640886e-01 -7.08189249e-01
-8.54757190e-01 4.75272983e-01 7.96435595e-01 -6.21084869e-01
-1.29700470e+00 9.44694698e-01 4.83925015e-01 -6.11637771e-01
-2.92836487e-01 -5.44798851e-01 -4.63846736e-02 -2.88893849e-01
7.65243053e-01 8.45153704e-02 5.20630717e-01 -6.26964748e-01
-5.20151854e-01 8.15443277e-01 -2.02657238e-01 7.97681212e-02
1.79119051e+00 -5.02446219e-02 -1.92003533e-01 3.15573663e-02
1.37201822e+00 -1.10958472e-01 -1.08878231e+00 -4.03841168e-01
-5.45492589e-01 -3.11060905e-01 8.17118227e-01 -8.82933736e-01
-1.72355914e+00 9.15956080e-01 1.06355774e+00 -2.96704262e-01
1.25459087e+00 -9.16012228e-02 1.25755644e+00 -4.09126848e-01
2.84181535e-01 -8.23082626e-01 1.62128240e-01 1.83571935e-01
1.14115143e+00 -9.50640023e-01 9.31465179e-02 -2.99123555e-01
-6.51571274e-01 1.07181120e+00 2.89909124e-01 -1.38998866e-01
7.80597866e-01 5.92738986e-01 -1.99638739e-01 -6.17587984e-01
-1.88415080e-01 2.10340172e-01 3.91513646e-01 7.43592441e-01
5.54544330e-01 7.74577111e-02 -2.03566894e-01 7.05067933e-01
-1.64231420e-01 7.50003457e-02 2.66671896e-01 5.45380831e-01
1.79043859e-02 -7.10562050e-01 -4.26637739e-01 4.98755664e-01
-7.48536646e-01 -4.92798775e-01 5.25536418e-01 5.04455268e-01
1.22537732e-01 2.09732339e-01 -2.17393428e-01 -1.26584455e-01
2.22386524e-01 -2.53356099e-01 7.74011552e-01 -1.71364129e-01
-1.06177485e+00 2.13366583e-01 -3.97538632e-01 -8.47173929e-01
-3.86972278e-01 -4.74466324e-01 -1.38432002e+00 -2.85678267e-01
-2.23639816e-01 -1.18456222e-02 7.72578835e-01 8.10820580e-01
3.91017824e-01 1.13672316e+00 4.38858300e-01 -1.14074898e+00
-4.23752666e-01 -8.78435075e-01 -7.02373326e-01 3.79568696e-01
3.61555487e-01 -9.07038450e-01 -1.03336915e-01 -4.23097044e-01] | [13.614700317382812, -2.428156614303589] |
081dd35a-91c5-48f0-ac05-527d3d70ff1e | evaluating-diversity-of-multiword-expressions | null | null | https://aclanthology.org/2022.coling-1.290 | https://aclanthology.org/2022.coling-1.290.pdf | Evaluating Diversity of Multiword Expressions in Annotated Text | Diversity can be decomposed into three distinct concepts, namely: variety, balance and disparity. This paper borrows from the extensive formalization and measures of diversity developed in ecology in order to evaluate the variety and balance of multiword expression annotation produced by automatic annotation systems. The measures of richness, normalized richness, and two variations of Hill’s evenness are considered in this paper. We observe how these measures behave against increasingly smaller samples of gold annotations of multiword expressions and use their comportment to validate or invalidate their pertinence for multiword expressions in annotated texts. We apply the validated measures to annotations in 14 languages produced by systems during the PARSEME shared task on automatic identification of multiword expressions and on the gold versions of the corpora. We also explore the limits of such evaluation by studying the impact of lemmatization errors in the Turkish corpus used in the shared task. | ['Jean-Yves Antoine', 'Agata Savary', 'Yagmur Ozturk', 'Adam Lion-Bouton'] | null | null | null | null | coling-2022-10 | ['lemmatization'] | ['natural-language-processing'] | [ 1.10446543e-01 -4.81197946e-02 -3.27798054e-02 -3.42270494e-01
-6.30275130e-01 -1.02994978e+00 7.65927792e-01 4.02228236e-01
-9.71688330e-01 1.01409912e+00 6.10635936e-01 -1.74932793e-01
-1.71724379e-01 -5.67898095e-01 -2.43638664e-01 -6.05723262e-01
-3.02463789e-02 2.80362457e-01 -7.36771477e-03 -5.13203800e-01
4.30462599e-01 1.43749535e-01 -1.64420247e+00 1.82154000e-01
1.00102973e+00 5.17545223e-01 -4.12451550e-02 4.76255715e-01
-5.11651099e-01 6.38637245e-01 -1.09962559e+00 -8.24663997e-01
1.33140773e-01 -4.98558015e-01 -9.36974823e-01 -3.74257937e-02
4.44516599e-01 1.76696241e-01 5.15436530e-01 1.12505758e+00
4.06704605e-01 -3.23205054e-01 8.65252674e-01 -7.29483902e-01
-4.56665069e-01 1.13776171e+00 -3.57912749e-01 4.09595907e-01
3.83704662e-01 5.62532656e-02 1.26890194e+00 -6.22911811e-01
9.81404424e-01 1.29420304e+00 8.03269565e-01 6.54614791e-02
-1.08405638e+00 -6.14019811e-01 -6.43819422e-02 -3.08721781e-01
-1.52216125e+00 -6.29800618e-01 9.95139182e-02 -9.53289151e-01
1.00755596e+00 2.57220238e-01 6.00302994e-01 9.65012550e-01
-1.18750341e-01 1.69578105e-01 1.08388317e+00 -7.56622076e-01
9.00036693e-02 2.98451960e-01 2.20121950e-01 4.39835548e-01
7.18675792e-01 -3.94827455e-01 -3.36108506e-01 -3.32001209e-01
4.60961163e-01 -8.77298236e-01 -3.42104942e-01 2.09284186e-01
-1.21513760e+00 9.25384164e-01 -3.16392541e-01 9.80380714e-01
-3.23753178e-01 -1.51419044e-01 9.32951570e-01 3.74131799e-01
8.48796546e-01 8.00971091e-01 -6.36870980e-01 -2.91867256e-01
-6.59673035e-01 4.24081348e-02 9.85395730e-01 6.80441141e-01
7.01681674e-01 3.12831104e-02 2.27791131e-01 1.14243639e+00
1.66451186e-03 4.45428133e-01 7.77719438e-01 -7.65328169e-01
3.05961639e-01 5.58460653e-01 7.62836188e-02 -8.48558009e-01
-4.87058133e-01 -4.58808720e-01 -3.51030588e-01 -3.02181453e-01
5.62158167e-01 -4.00973052e-01 -3.66544873e-01 1.99592078e+00
3.86053175e-01 -3.67199570e-01 2.15124592e-01 4.46231812e-01
6.70103014e-01 3.64684641e-01 4.70603973e-01 -6.09510839e-01
1.24385488e+00 -3.84432614e-01 -6.67715371e-01 3.74363691e-01
1.08195627e+00 -7.58933365e-01 1.10558963e+00 4.05617326e-01
-8.39606166e-01 -3.09973001e-01 -9.06127393e-01 1.36687666e-01
-5.75048864e-01 -1.66809298e-02 4.98270661e-01 1.04755247e+00
-9.36362028e-01 5.09585083e-01 -2.93660432e-01 -5.56660950e-01
-1.44088864e-01 -8.29151720e-02 -3.48963022e-01 6.98227823e-01
-1.20077407e+00 1.11916161e+00 3.13173115e-01 -1.50403991e-01
-5.18570468e-02 -4.70763117e-01 -7.03702271e-01 -3.13228488e-01
2.01056749e-01 -1.40123263e-01 1.10481179e+00 -1.50327313e+00
-1.11548138e+00 1.64252245e+00 3.67966473e-01 -3.50851081e-02
3.86390120e-01 -8.91807303e-02 -4.75056589e-01 -4.65835892e-02
4.81344700e-01 8.90134275e-02 -6.18018359e-02 -9.28335428e-01
-6.93797946e-01 -4.43218172e-01 -8.89878571e-02 -1.85324162e-01
-3.96867752e-01 5.00215530e-01 2.87609667e-01 -7.63371348e-01
-2.42507115e-01 -7.12277293e-01 6.00815527e-02 -2.95551419e-01
-9.55656096e-02 -3.52640688e-01 -1.54615507e-01 -7.45798171e-01
1.42713308e+00 -2.06528115e+00 2.12405965e-01 2.38604859e-01
-1.55157268e-01 1.40356392e-01 -4.54113901e-01 6.32359445e-01
6.53855577e-02 7.11290598e-01 -4.02344167e-01 6.03957614e-03
1.83327049e-01 5.13141632e-01 -2.11386904e-01 6.58822834e-01
2.80720562e-01 4.34007764e-01 -1.05730069e+00 -5.58527410e-01
-2.65082359e-01 2.58969396e-01 -4.26229924e-01 -4.75700587e-01
-2.60900795e-01 1.46944806e-01 -1.47632390e-01 5.34688294e-01
2.45275199e-01 3.57587874e-01 5.44241071e-01 -4.31021452e-02
-6.78312898e-01 5.02915025e-01 -8.56355309e-01 1.48042345e+00
-5.39466619e-01 7.03395367e-01 -1.13276415e-01 -6.66985035e-01
1.02757359e+00 1.70174792e-01 2.53683209e-01 -4.90719885e-01
5.17836392e-01 6.41914666e-01 4.96559024e-01 -6.99915588e-01
9.43351090e-01 -2.64517367e-01 -5.59767962e-01 3.76437873e-01
2.77249902e-01 -6.66192994e-02 6.63337588e-01 -2.06882536e-01
9.04698849e-01 2.65302569e-01 7.90137589e-01 -9.57054675e-01
6.11632884e-01 2.64095485e-01 4.99097288e-01 3.01580340e-01
-2.46475488e-02 7.65141025e-02 7.87874699e-01 -2.89028674e-01
-1.23092878e+00 -8.19675624e-01 -7.31899083e-01 1.30190146e+00
-3.45252365e-01 -6.18920684e-01 -5.74523509e-01 -4.80410904e-01
-1.66481853e-01 8.05060506e-01 -7.81254470e-01 2.47517467e-01
-2.95932978e-01 -9.27637458e-01 1.17799330e+00 1.05176121e-01
1.88783675e-01 -9.20174479e-01 -9.95968938e-01 8.11520815e-02
-2.41528362e-01 -9.78149414e-01 -6.90855691e-03 8.85867625e-02
-2.14318588e-01 -1.19021249e+00 -5.21242976e-01 -4.97821450e-01
5.46608195e-02 -4.15172428e-01 1.27295375e+00 2.26515427e-01
-2.36634374e-01 2.07793802e-01 -7.81615615e-01 -4.95912194e-01
-8.21302414e-01 2.89006799e-01 1.49469823e-01 -4.45125729e-01
4.96763319e-01 -5.65018058e-01 1.61941528e-01 2.50107765e-01
-1.04844522e+00 -6.76713049e-01 2.45050266e-01 4.79883432e-01
3.19498956e-01 -3.29035789e-01 5.96354663e-01 -7.72116959e-01
6.00433111e-01 -5.12998879e-01 -4.85903800e-01 3.62329334e-01
-1.98829889e-01 2.34786034e-01 2.86267877e-01 -5.96365929e-01
-8.37560177e-01 -3.15589398e-01 -4.64343041e-01 2.04856277e-01
9.62416083e-02 7.48085141e-01 -1.05646953e-01 9.79718473e-03
8.68658960e-01 -2.55528957e-01 -3.93723696e-01 -3.68908137e-01
3.18201154e-01 9.57024872e-01 4.87503827e-01 -8.02507818e-01
2.41526738e-01 -5.05517162e-02 -1.83238357e-01 -1.42149401e+00
-6.86990738e-01 -3.78064334e-01 -7.79904664e-01 -4.08584736e-02
8.22928488e-01 -9.10188377e-01 -6.23198330e-01 2.78441191e-01
-1.10136855e+00 -3.56213689e-01 -3.05035174e-01 3.09265852e-01
-3.77396196e-01 4.62041646e-01 -2.87502319e-01 -1.04301155e+00
-2.01385319e-01 -7.36599743e-01 1.11962759e+00 -2.72028953e-01
-7.56063163e-01 -1.05430746e+00 8.31769347e-01 -3.35317671e-01
1.01003334e-01 6.17763340e-01 1.09129655e+00 -8.54615867e-01
4.23889935e-01 1.96808919e-01 1.76890761e-01 1.98248267e-01
1.76058374e-02 6.13868356e-01 -7.52652645e-01 2.33179554e-01
-1.18164428e-01 -2.37350911e-01 8.00862610e-01 -1.63374215e-01
2.49865621e-01 -3.99899036e-01 1.51204780e-01 3.08775276e-01
1.80250454e+00 -1.11828133e-01 4.84852970e-01 4.98057187e-01
3.22122693e-01 1.25545287e+00 4.63037252e-01 5.91923714e-01
2.10464567e-01 5.61470151e-01 5.44609455e-03 5.63356102e-01
2.55603403e-01 9.12264138e-02 4.79668736e-01 1.25633681e+00
-1.48811266e-01 -3.71227533e-01 -1.24008441e+00 7.77425051e-01
-1.41104901e+00 -9.79689777e-01 -6.04076684e-01 2.03457427e+00
9.18615460e-01 -1.75550908e-01 4.67449844e-01 5.46649396e-02
6.72802746e-01 1.71516109e-02 2.05932528e-01 -9.34197009e-01
-5.23420572e-01 4.66431469e-01 5.51768899e-01 5.63789189e-01
-6.72123253e-01 9.25654769e-01 7.15195417e+00 7.67413497e-01
-9.62415457e-01 8.87192115e-02 3.56528252e-01 1.09775640e-01
-4.15374786e-01 -2.38500461e-01 -5.97789109e-01 3.74411285e-01
1.11373532e+00 -2.40353733e-01 -1.78651810e-02 5.24869859e-01
-4.21233699e-02 -4.31686372e-01 -8.40081155e-01 4.90631580e-01
1.63483426e-01 -7.40414739e-01 -5.97857796e-02 -4.42943070e-03
6.19477153e-01 2.07787767e-01 -2.91428983e-01 2.07169533e-01
4.21179116e-01 -6.25769258e-01 1.04889524e+00 3.93411249e-01
9.10920739e-01 -6.09994709e-01 7.16879368e-01 1.22419991e-01
-8.27465177e-01 1.84430361e-01 -4.62352782e-01 -3.36687863e-01
4.91920412e-02 4.20694858e-01 -4.40538913e-01 4.26260561e-01
4.10244256e-01 2.78028667e-01 -6.78464890e-01 6.97989702e-01
-3.48015100e-01 4.92486238e-01 -4.13588583e-01 -3.33010644e-01
3.25352907e-01 -2.87509769e-01 5.75242043e-01 1.58065939e+00
3.81374985e-01 -1.85652614e-01 -1.63569063e-01 6.20538473e-01
1.29591227e-01 7.64753640e-01 -4.67348307e-01 -5.02536416e-01
5.47248960e-01 1.16982615e+00 -7.08317697e-01 -3.24184060e-01
-2.95815676e-01 6.88648224e-01 3.61305445e-01 -1.36747494e-01
-5.96845269e-01 -3.38615566e-01 8.13148618e-01 -1.51726194e-02
2.23034188e-01 -2.42912769e-01 -2.21376151e-01 -9.54943299e-01
-1.22394100e-01 -7.75667548e-01 4.22605276e-01 -4.13083822e-01
-1.58203351e+00 8.66426289e-01 2.52516091e-01 -9.15108025e-01
-4.11546946e-01 -7.51804948e-01 -3.35943222e-01 9.62366223e-01
-9.97942746e-01 -8.21187913e-01 -5.56521565e-02 -1.08791757e-02
2.54221469e-01 1.21693183e-02 9.76622820e-01 2.74584562e-01
-5.31804681e-01 4.46324408e-01 1.11902982e-01 1.48636103e-01
6.56921744e-01 -1.07409430e+00 3.27542365e-01 4.57805216e-01
1.13498792e-01 4.27460968e-01 8.94681573e-01 -6.74419999e-01
-4.58204448e-01 -7.18915045e-01 1.15448880e+00 -4.75518405e-01
1.18422771e+00 -1.72410458e-01 -7.41039574e-01 4.74308848e-01
2.85706758e-01 -7.15746522e-01 1.11249959e+00 2.90476441e-01
-4.44123775e-01 4.91788864e-01 -1.05943549e+00 5.20705640e-01
1.32776046e+00 -4.12851959e-01 -6.76478863e-01 2.68548042e-01
6.01352334e-01 1.63246438e-01 -1.18648434e+00 3.68602246e-01
1.00326002e+00 -1.01359165e+00 4.63453293e-01 -6.13288224e-01
5.59158504e-01 2.56992667e-03 -5.60799301e-01 -1.25571668e+00
-9.33896154e-02 -3.81664991e-01 1.00856864e+00 1.73208535e+00
5.03210664e-01 -6.45143449e-01 2.06353776e-02 9.67416093e-02
-1.19927734e-01 -3.31724465e-01 -9.22616541e-01 -8.30170274e-01
5.61161637e-01 -1.56854257e-01 5.69887936e-01 1.37620592e+00
2.69103914e-01 5.39647579e-01 1.41133815e-01 -2.81564444e-01
3.40545774e-01 -3.75192583e-01 7.03457773e-01 -1.40023077e+00
-2.84716576e-01 -6.96724057e-01 -6.76220536e-01 -1.26064107e-01
4.79613274e-01 -9.55435634e-01 8.54562670e-02 -9.99192059e-01
-5.51744923e-02 -3.38762671e-01 2.43407175e-01 2.90446192e-01
1.21356780e-02 4.54292685e-01 8.79836231e-02 2.71321058e-01
-3.13414156e-01 3.26905102e-01 5.95739901e-01 2.84994751e-01
-2.17078909e-01 -6.80876136e-01 -5.86450160e-01 9.67655599e-01
7.49632418e-01 -5.35109580e-01 8.03972408e-02 -3.87932032e-01
7.97840357e-01 -3.55893999e-01 7.19249323e-02 -7.67144322e-01
-4.72693712e-01 -1.67580381e-01 -2.95529868e-02 2.26555439e-03
-1.61106661e-01 -5.31713963e-01 3.03032875e-01 4.29965436e-01
-4.30482805e-01 4.74040061e-01 1.73008770e-01 -5.18266670e-02
-6.13242909e-02 -7.13098109e-01 6.38159931e-01 -3.18572015e-01
-5.32340884e-01 -3.26428652e-01 -7.59337366e-01 3.39465052e-01
7.48368919e-01 -2.30654955e-01 -4.29338366e-01 8.72328226e-03
-5.49340904e-01 -5.31591848e-02 7.75700390e-01 1.17872268e-01
-2.27412120e-01 -9.96456683e-01 -1.00404489e+00 -3.52792591e-01
2.98152030e-01 -5.99987805e-01 -2.61337072e-01 9.97441709e-01
-8.34894717e-01 9.13300458e-03 -4.28291291e-01 -3.48437071e-01
-1.18141747e+00 4.36649591e-01 3.68784040e-01 -1.07490905e-01
9.16398689e-02 5.59845209e-01 1.07619010e-01 -4.14538890e-01
-1.31218210e-01 -2.12004006e-01 -1.90555081e-01 5.94381869e-01
3.26867282e-01 3.88727129e-01 -9.32241976e-02 -1.29225802e+00
-4.37069982e-01 6.37259781e-01 4.30825442e-01 -4.48658615e-01
1.23581839e+00 -1.95881903e-01 -4.47092772e-01 1.03081048e+00
1.17052686e+00 6.09986603e-01 -3.07921767e-01 -3.31191197e-02
4.55233157e-01 -2.48703882e-01 -3.88497055e-01 -7.21278191e-01
-2.97484010e-01 4.61121291e-01 1.99573621e-01 4.74676788e-01
7.26138711e-01 1.31459627e-03 1.14305228e-01 2.81793773e-01
2.31529489e-01 -1.05799782e+00 -3.72388691e-01 6.26109660e-01
8.83718431e-01 -5.96683979e-01 -2.83410028e-02 -4.46506441e-01
-5.99663615e-01 1.14384151e+00 2.78165817e-01 -2.90646367e-02
3.03918123e-01 3.56114477e-01 2.98685461e-01 -2.89574951e-01
-6.05519772e-01 -7.02665448e-01 -1.14168515e-02 4.58566308e-01
9.10444200e-01 3.27827744e-02 -1.29929197e+00 4.42006141e-01
-6.56402349e-01 -4.29897428e-01 6.39600158e-01 5.44613183e-01
-6.70288324e-01 -1.07710326e+00 -3.12479377e-01 1.89835116e-01
-7.90642321e-01 -3.12795222e-01 -1.14098477e+00 1.19735050e+00
5.42718768e-01 7.31114209e-01 3.70939136e-01 -4.55953479e-01
2.39985108e-01 1.74942523e-01 7.70545065e-01 -6.40821457e-01
-1.07514560e+00 -1.03968121e-01 8.71279120e-01 -1.86238308e-02
-9.93861854e-01 -8.99354398e-01 -8.35742176e-01 -1.59319028e-01
-5.39453387e-01 4.10327196e-01 7.65975714e-01 9.32664752e-01
-3.01984161e-01 2.42680654e-01 2.91790068e-01 -6.10916078e-01
-4.91844833e-01 -1.31744862e+00 -6.67239368e-01 2.96204001e-01
-9.73973051e-02 -5.56894839e-01 -5.52646160e-01 -9.24000964e-02] | [10.441834449768066, 10.087299346923828] |
2bad04b6-afbd-4afb-b1f7-a6daa4de239a | a-lightweight-and-detector-free-3d-single | 2203.04232 | null | https://arxiv.org/abs/2203.04232v2 | https://arxiv.org/pdf/2203.04232v2.pdf | A Lightweight and Detector-free 3D Single Object Tracker on Point Clouds | Recent works on 3D single object tracking treat the task as a target-specific 3D detection task, where an off-the-shelf 3D detector is commonly employed for the tracking. However, it is non-trivial to perform accurate target-specific detection since the point cloud of objects in raw LiDAR scans is usually sparse and incomplete. In this paper, we address this issue by explicitly leveraging temporal motion cues and propose DMT, a Detector-free Motion-prediction-based 3D Tracking network that completely removes the usage of complicated 3D detectors and is lighter, faster, and more accurate than previous trackers. Specifically, the motion prediction module is first introduced to estimate a potential target center of the current frame in a point-cloud-free manner. Then, an explicit voting module is proposed to directly regress the 3D box from the estimated target center. Extensive experiments on KITTI and NuScenes datasets demonstrate that our DMT can still achieve better performance (~10% improvement over the NuScenes dataset) and a faster tracking speed (i.e., 72 FPS) than state-of-the-art approaches without applying any complicated 3D detectors. Our code is released at \url{https://github.com/jimmy-dq/DMT} | ['Uwe Stilla', 'Antoni B. Chan', 'Wei Li', 'Qiangqiang Wu', 'Yan Xia'] | 2022-03-08 | null | null | null | null | ['3d-single-object-tracking'] | ['computer-vision'] | [-3.33104968e-01 -5.73587894e-01 -3.00337642e-01 -1.17601082e-01
-5.88277221e-01 -7.06871331e-01 5.03241241e-01 -1.16522811e-01
-5.50728559e-01 2.23634288e-01 -5.29276252e-01 -4.02455032e-01
2.98586726e-01 -6.09758854e-01 -6.35341525e-01 -5.58368146e-01
7.62713179e-02 6.04133189e-01 1.08478856e+00 6.49685264e-02
3.87288770e-03 9.50117826e-01 -1.46381009e+00 -6.15665138e-01
5.52523196e-01 1.10221004e+00 3.67534369e-01 5.90848505e-01
-2.32762899e-02 1.24947228e-01 -3.67134601e-01 -2.23075133e-02
6.01617813e-01 8.05543885e-02 1.06719665e-01 2.85014845e-02
8.13891351e-01 -5.49198568e-01 -4.58509535e-01 9.81631756e-01
4.72930968e-01 -6.79500625e-02 4.30999994e-01 -1.35489702e+00
-5.04344516e-02 -1.66093677e-01 -9.95780170e-01 3.81545693e-01
1.01662576e-01 4.11818594e-01 5.06690323e-01 -9.74384725e-01
4.63119388e-01 1.27986872e+00 7.39048183e-01 5.21467328e-01
-1.14347363e+00 -1.18999922e+00 2.57547170e-01 -1.08843602e-01
-1.59103560e+00 -2.66461134e-01 6.31652236e-01 -6.06252670e-01
5.83924234e-01 7.11864457e-02 8.50780487e-01 8.32583547e-01
1.65149778e-01 6.66377008e-01 7.02567041e-01 1.55618992e-02
-2.55298265e-03 -1.51298612e-01 6.83241189e-02 6.81235611e-01
7.78091908e-01 5.03709435e-01 -3.59027237e-01 -5.18648103e-02
9.68338192e-01 3.54740292e-01 -5.61858565e-02 -1.00624073e+00
-1.37839258e+00 6.41463816e-01 6.58966482e-01 -1.02202177e-01
-1.52937934e-01 3.94432247e-01 2.04235345e-01 -1.57101437e-01
5.29053390e-01 -2.39013702e-01 -2.91198224e-01 3.37660164e-02
-9.39123273e-01 4.82870102e-01 3.90333861e-01 1.24807179e+00
6.68077648e-01 1.36827886e-01 -1.16249792e-01 1.73974156e-01
6.19619906e-01 1.17792213e+00 -1.81465149e-01 -9.35864449e-01
3.50821555e-01 7.22337842e-01 5.86353362e-01 -7.47334421e-01
-4.15853262e-01 -4.38774139e-01 -4.85709012e-01 6.66023612e-01
4.44098592e-01 -1.56013250e-01 -8.78815949e-01 1.46021700e+00
1.12309265e+00 4.35742289e-01 -4.07316148e-01 1.21886408e+00
7.62809992e-01 4.49813098e-01 4.22120951e-02 -4.02908307e-03
1.29851794e+00 -7.97191024e-01 -3.46795827e-01 -3.23230565e-01
5.78304529e-01 -8.09583783e-01 4.79727417e-01 -9.30470303e-02
-6.51551127e-01 -6.58565104e-01 -9.67494607e-01 1.11934103e-01
-1.14244685e-01 3.57861161e-01 5.78987539e-01 5.11907935e-01
-6.62667334e-01 2.86699444e-01 -1.23683047e+00 -4.48802948e-01
5.13542175e-01 3.54816139e-01 -2.21071288e-01 -8.59822519e-03
-6.41155541e-01 1.00830305e+00 4.83187854e-01 2.38331631e-02
-1.00409067e+00 -8.25293720e-01 -7.73783684e-01 -2.73728341e-01
7.29360700e-01 -8.21329892e-01 1.33119977e+00 -1.24393895e-01
-1.26761889e+00 7.48911262e-01 -3.01924229e-01 -3.35240155e-01
6.72822416e-01 -6.14558995e-01 -1.50847420e-01 -7.47736543e-02
1.64995372e-01 6.80974245e-01 8.34109306e-01 -1.08499432e+00
-9.17410254e-01 -5.52585065e-01 -1.39378592e-01 1.43902525e-01
1.30373105e-01 -5.12505509e-02 -8.83849621e-01 -3.95197868e-01
3.22502643e-01 -1.26605594e+00 -2.48355418e-01 7.89353549e-01
-2.75041640e-01 -3.00985247e-01 1.21073663e+00 -2.45261535e-01
8.82809043e-01 -2.18650031e+00 -3.01576734e-01 -2.22044840e-01
2.86917537e-01 5.81138313e-01 1.09596424e-01 1.05749585e-01
3.47493768e-01 -2.79480636e-01 7.25559071e-02 -2.83098429e-01
-1.55583415e-02 -1.23699076e-01 -1.60198256e-01 7.31550515e-01
1.12368785e-01 7.38367856e-01 -9.52537477e-01 -5.39221883e-01
6.79106653e-01 6.92868471e-01 -2.62064189e-01 1.73409295e-03
-3.32589656e-01 5.74162900e-01 -8.37977469e-01 1.03907502e+00
1.06966937e+00 -1.46200269e-01 -1.94380671e-01 -1.59843013e-01
-7.39132047e-01 1.90457702e-01 -1.30220127e+00 1.66398740e+00
5.09264804e-02 5.13297737e-01 1.47109866e-01 -3.71981502e-01
9.91323590e-01 -4.08964865e-02 6.34123385e-01 -2.29816645e-01
3.00419539e-01 1.99302182e-01 -9.00933519e-02 8.82738233e-02
4.63352233e-01 1.10877775e-01 -7.87790120e-02 1.83238775e-01
-2.62125522e-01 -2.13537127e-01 7.24170730e-02 8.94752294e-02
1.11648619e+00 6.29494846e-01 2.34393746e-01 -2.64027640e-02
3.76974314e-01 5.42814553e-01 9.56119180e-01 7.23760486e-01
-5.52865565e-01 3.62676412e-01 -2.34672144e-01 -3.45900476e-01
-8.38835955e-01 -1.21358836e+00 -1.38953611e-01 5.90437770e-01
6.72062337e-01 -3.23626697e-01 -3.19282055e-01 -6.23469770e-01
6.78681672e-01 2.80651093e-01 -1.78837135e-01 4.87750508e-02
-6.59846008e-01 -3.11284244e-01 2.82133222e-01 6.13418281e-01
4.52717930e-01 -3.69160324e-01 -9.69067276e-01 1.26029968e-01
1.71545103e-01 -1.23946977e+00 -5.64999044e-01 -2.05082539e-02
-1.16963923e+00 -1.07940233e+00 -5.45595825e-01 -3.55994642e-01
5.39077342e-01 1.10508835e+00 7.09852338e-01 9.59926918e-02
-2.70485252e-01 1.85567066e-01 -2.30744243e-01 -7.52199769e-01
1.25736907e-01 -1.78757146e-01 3.87893230e-01 -4.37501907e-01
6.03559613e-01 -3.14782947e-01 -7.62045562e-01 5.63314974e-01
-2.89112091e-01 2.40276251e-02 6.36871815e-01 3.21911514e-01
9.23939407e-01 -1.58846319e-01 -9.22622085e-02 -3.66897702e-01
-2.70527512e-01 -1.48977280e-01 -1.42131877e+00 -1.65906921e-01
-3.04363787e-01 -2.49331087e-01 1.20498210e-01 -7.99480975e-01
-7.40513504e-01 8.18531275e-01 2.35117763e-01 -1.06200755e+00
-7.87525699e-02 -1.44095883e-01 -6.70875162e-02 -5.13636053e-01
4.19755489e-01 8.59282389e-02 -7.06392825e-02 -7.19143689e-01
2.03264132e-01 2.91936278e-01 5.82627714e-01 -3.03692907e-01
1.69944465e+00 8.15129936e-01 2.24726215e-01 -6.05140924e-01
-9.48511422e-01 -8.63288283e-01 -9.21747625e-01 -5.11656880e-01
8.40735197e-01 -1.27971518e+00 -9.23656106e-01 4.45181519e-01
-1.12538302e+00 -1.18674211e-01 -7.39562437e-02 8.45982194e-01
-1.47734940e-01 3.17292541e-01 -2.84047633e-01 -9.37508166e-01
-4.09129471e-01 -9.52588141e-01 1.31939673e+00 4.58614200e-01
2.12796465e-01 -4.75610614e-01 1.56411365e-01 1.32882297e-01
1.56207800e-01 3.84070545e-01 1.06299244e-01 -2.31230542e-01
-1.20070064e+00 -5.96080065e-01 -3.66961181e-01 -1.95021987e-01
2.89911963e-02 1.74769208e-01 -7.69967675e-01 -4.57490474e-01
-2.39178449e-01 1.78241819e-01 7.13794172e-01 3.60887408e-01
7.48532712e-01 4.27988440e-01 -8.63789141e-01 6.62837505e-01
1.40297413e+00 9.00792256e-02 9.15333033e-02 3.92269552e-01
8.25991690e-01 1.03686482e-01 1.27514672e+00 4.72166300e-01
4.95117545e-01 1.03573966e+00 7.40451097e-01 1.30516663e-01
-2.39924118e-01 -4.47437704e-01 4.22897249e-01 3.72882515e-01
-9.62283760e-02 6.53146133e-02 -8.92660141e-01 4.23311800e-01
-1.95979643e+00 -8.10445786e-01 -5.58895111e-01 2.42378855e+00
3.61917078e-01 4.37468797e-01 2.46292695e-01 -3.90936971e-01
1.00067401e+00 1.14768654e-01 -9.04437602e-01 5.80590785e-01
2.95834064e-01 -1.00479648e-01 9.30377066e-01 2.92502850e-01
-1.36722720e+00 1.19331741e+00 5.48272228e+00 6.10447407e-01
-1.02391982e+00 2.02212021e-01 -3.78038704e-01 -4.03643787e-01
3.81558180e-01 3.55562657e-01 -1.57860446e+00 4.70781982e-01
6.33015871e-01 -1.40268862e-01 -1.19978756e-01 1.11112964e+00
3.79890144e-01 -2.40039289e-01 -8.39656293e-01 9.96415019e-01
-3.33669752e-01 -1.13441002e+00 -3.40379119e-01 2.34403998e-01
2.26478830e-01 4.05974239e-01 -2.66659856e-01 3.09491843e-01
3.64938766e-01 -2.49624342e-01 9.60444987e-01 3.76762599e-01
5.66299260e-01 -5.50903678e-01 4.21860129e-01 7.52973795e-01
-1.84102952e+00 1.98733732e-01 -6.89190865e-01 -7.73667172e-02
4.13461834e-01 6.96422994e-01 -8.01795244e-01 5.23113966e-01
9.82966840e-01 6.98358178e-01 -5.65898716e-01 1.54567802e+00
-2.93833941e-01 3.12565655e-01 -7.98583210e-01 -5.15477583e-02
7.14798570e-02 1.52104795e-02 1.05768621e+00 9.74180281e-01
6.02858305e-01 1.56869605e-01 6.02305770e-01 7.03712165e-01
9.87599343e-02 -3.32825392e-01 -6.48124695e-01 3.04951131e-01
7.41275072e-01 1.43951929e+00 -6.99669182e-01 -2.14099541e-01
-4.05065686e-01 5.82314789e-01 1.54720753e-01 -8.03906098e-02
-1.19426322e+00 -1.43605530e-01 9.63396668e-01 3.31444591e-01
8.67803633e-01 -7.33585656e-01 1.80588797e-01 -1.24105453e+00
7.99481496e-02 -2.20045075e-01 1.75782055e-01 -8.24913502e-01
-1.04585266e+00 2.69874305e-01 6.88701198e-02 -1.83962750e+00
-2.49144435e-02 -6.45422935e-01 -5.91129005e-01 7.93237090e-01
-1.54123354e+00 -1.14777374e+00 -5.77320099e-01 5.28981924e-01
2.80449390e-01 2.13207722e-01 2.06286594e-01 3.80331188e-01
-6.34766459e-01 2.38668978e-01 -8.07172582e-02 1.71646208e-01
7.58071303e-01 -8.77947032e-01 5.78243315e-01 1.07945025e+00
-6.55611157e-02 4.60801274e-01 5.83984494e-01 -1.07433248e+00
-1.86518896e+00 -1.29102731e+00 6.11937046e-01 -7.87233174e-01
7.86598504e-01 -5.44412851e-01 -7.42755830e-01 6.58168018e-01
-3.36772919e-01 4.07226801e-01 2.33440354e-01 -2.63127089e-01
-2.04912052e-01 -2.09968969e-01 -9.53065813e-01 4.62143362e-01
1.37549531e+00 -4.53597419e-02 -4.75261837e-01 2.56240219e-01
7.88817644e-01 -8.45658243e-01 -7.27039099e-01 5.79670548e-01
5.76352298e-01 -6.64161742e-01 1.21855342e+00 -1.34719945e-02
-5.37624478e-01 -1.17263234e+00 2.94806715e-02 -7.16217875e-01
-4.21394557e-01 -4.21884894e-01 -5.91554224e-01 1.00159192e+00
-7.75693953e-02 -5.24817228e-01 1.01218879e+00 2.49430731e-01
-1.91901043e-01 -3.50922853e-01 -1.17046475e+00 -1.22408283e+00
-4.03345615e-01 -4.91051137e-01 3.62184256e-01 5.07865727e-01
-6.98964596e-01 2.77802557e-01 -3.89226377e-01 6.00756764e-01
1.15915418e+00 5.21438181e-01 1.31519794e+00 -1.55844748e+00
1.06091730e-01 -2.04157725e-01 -5.97782135e-01 -1.58580232e+00
-1.40067741e-01 -7.21484601e-01 2.83404768e-01 -1.30675006e+00
1.93445999e-02 -5.32732666e-01 5.76377707e-03 4.03225482e-01
-8.07164460e-02 2.75913984e-01 5.21077394e-01 5.34133017e-01
-7.71463275e-01 5.60915887e-01 1.20664620e+00 -4.55467366e-02
-9.65033621e-02 4.14237618e-01 -2.72007912e-01 7.71039724e-01
5.74482918e-01 -8.36395562e-01 6.09413907e-03 -4.85635251e-01
-4.31529969e-01 1.29791384e-03 7.79046774e-01 -1.28887796e+00
5.18804491e-01 -1.79225475e-01 6.50183201e-01 -1.58317828e+00
6.03486061e-01 -9.88362551e-01 2.78236210e-01 7.96604574e-01
5.30540168e-01 7.94202928e-03 4.15377229e-01 6.79123521e-01
1.59821883e-01 -1.16975948e-01 7.62596071e-01 -8.63301083e-02
-8.77521455e-01 8.47470105e-01 -1.54311270e-01 -2.83444643e-01
1.28664696e+00 -3.33047837e-01 -2.42217615e-01 6.24845810e-02
-3.31237882e-01 4.57031161e-01 7.84027994e-01 4.60541457e-01
5.59374630e-01 -1.52732861e+00 -4.85125154e-01 -6.72325194e-02
1.85759962e-01 3.78127664e-01 1.04617074e-01 1.00283802e+00
-3.47258598e-01 5.49974322e-01 5.14218546e-02 -1.23434317e+00
-1.37562275e+00 5.63033581e-01 3.15579802e-01 3.08821481e-02
-9.69005644e-01 5.67972898e-01 2.07250834e-01 -4.06515479e-01
2.72231877e-01 -3.06318194e-01 1.66963160e-01 -2.53558278e-01
5.97563446e-01 3.59633654e-01 -2.94693857e-01 -7.35466361e-01
-7.32632220e-01 1.06564057e+00 2.06472185e-02 1.54054239e-01
1.17750072e+00 1.52043672e-03 3.90397549e-01 1.88496307e-01
5.46911299e-01 -3.69378440e-02 -1.78849101e+00 -3.09942573e-01
1.55459130e-02 -8.01143467e-01 1.50214091e-01 -2.58325070e-01
-1.10922754e+00 7.67790914e-01 9.14679050e-01 -1.32376879e-01
7.52528906e-01 1.04696527e-01 8.56325090e-01 3.68310571e-01
6.38244867e-01 -6.51648164e-01 -9.50486809e-02 6.88421249e-01
4.67225879e-01 -1.31400073e+00 3.65597308e-01 -5.79281449e-01
-1.81481153e-01 9.66922283e-01 9.74912465e-01 -2.38041952e-01
5.79949677e-01 3.30924451e-01 -2.82288454e-02 -1.49400339e-01
-4.40134197e-01 -5.70633769e-01 3.63427550e-01 6.42581344e-01
2.08490584e-02 -1.10872306e-01 2.14390513e-02 1.33017987e-01
1.70022532e-01 -3.69458161e-02 -1.88344915e-03 1.27353585e+00
-7.61082292e-01 -9.88794088e-01 -7.03199983e-01 1.52408525e-01
-3.01377177e-02 2.31352985e-01 -2.96552271e-01 1.21612692e+00
1.21436931e-01 5.98150253e-01 2.48976480e-02 -4.91612971e-01
5.58213353e-01 -2.56181300e-01 5.29536486e-01 -6.29093587e-01
-2.97205180e-01 3.54175985e-01 -2.39345744e-01 -8.06861937e-01
-5.90392351e-01 -9.05357957e-01 -1.34880328e+00 -4.59878027e-01
-6.66722953e-01 -2.60028094e-01 7.69087195e-01 6.60562873e-01
5.19641817e-01 1.72612488e-01 3.43828738e-01 -1.41543102e+00
-4.22721803e-01 -7.19476461e-01 -2.33010665e-01 -4.83075902e-02
3.86087626e-01 -1.26404631e+00 -3.13427389e-01 -3.81802469e-01] | [6.663346767425537, -2.2389767169952393] |
0593b9ce-2fe0-4b52-bff3-fece0e91c39f | p-tree-programming | 1707.03744 | null | http://arxiv.org/abs/1707.03744v1 | http://arxiv.org/pdf/1707.03744v1.pdf | P-Tree Programming | We propose a novel method for automatic program synthesis. P-Tree Programming
represents the program search space through a single probabilistic prototype
tree. From this prototype tree we form program instances which we evaluate on a
given problem. The error values from the evaluations are propagated through the
prototype tree. We use them to update the probability distributions that
determine the symbol choices of further instances. The iterative method is
applied to several symbolic regression benchmarks from the literature. It
outperforms standard Genetic Programming to a large extend. Furthermore, it
relies on a concise set of parameters which are held constant for all problems.
The algorithm can be employed for most of the typical computational
intelligence tasks such as classification, automatic program induction, and
symbolic regression. | ['Christian Oesch'] | 2017-07-12 | null | null | null | null | ['program-induction'] | ['computer-code'] | [ 3.85815233e-01 1.48872152e-01 -8.15500379e-01 -4.05478984e-01
-7.23413110e-01 -4.83544111e-01 2.45095491e-01 4.01995808e-01
-1.02536418e-01 8.52302849e-01 -4.79047179e-01 -5.23599803e-01
-3.00725013e-01 -1.03940582e+00 -8.44262183e-01 -5.20411730e-01
-1.38128236e-01 7.88773119e-01 5.88090301e-01 -7.31931254e-02
6.94012463e-01 2.94340372e-01 -1.77488136e+00 1.65046856e-01
1.06076884e+00 6.90714538e-01 1.32414117e-01 5.98856688e-01
-2.47767940e-01 5.95976591e-01 -6.18593574e-01 -1.77507505e-01
-2.38381132e-01 -2.48301670e-01 -8.50601494e-01 -4.00184184e-01
-4.19865906e-01 1.62343942e-02 1.69890821e-01 1.30968893e+00
-5.28686158e-02 1.44496664e-01 5.37028432e-01 -1.41047072e+00
-1.54397011e-01 1.19402480e+00 -5.26124835e-01 -1.25616938e-01
6.04925692e-01 -1.30231604e-01 9.86978948e-01 -5.63684881e-01
5.49490869e-01 1.32107937e+00 5.82021058e-01 2.42563292e-01
-1.71359217e+00 -5.68833470e-01 3.31192613e-02 1.97133750e-01
-1.52055085e+00 1.63377021e-02 7.45938778e-01 -5.99699736e-01
1.07751942e+00 2.82012790e-01 4.54658955e-01 4.01077747e-01
2.11546168e-01 6.82801068e-01 9.24461544e-01 -7.73530960e-01
5.72340786e-01 2.02154830e-01 6.07587695e-01 9.60678995e-01
2.29501650e-01 3.03962946e-01 -2.31385365e-01 -8.92712533e-01
2.55279601e-01 -4.33550090e-01 -8.01541656e-02 -5.77388167e-01
-9.10111904e-01 1.02585924e+00 -4.16131206e-02 2.24986836e-01
-5.10695614e-02 5.72459638e-01 3.99377078e-01 1.47727892e-01
-9.61537510e-02 6.14265501e-01 -4.87047493e-01 -2.39566267e-01
-7.75576711e-01 5.70545793e-01 1.06904340e+00 1.04384696e+00
1.00072753e+00 -5.73915914e-02 -2.48417556e-01 7.48781443e-01
3.53098303e-01 3.80049080e-01 3.07743281e-01 -9.13889110e-01
3.97989899e-01 7.80070126e-01 -2.02502264e-03 -1.00519395e+00
-1.72644243e-01 4.70423518e-04 -3.50233167e-02 3.80798668e-01
2.15471536e-01 -7.78726488e-02 -7.05190778e-01 1.55952549e+00
2.17063636e-01 -9.25098211e-02 -1.84716702e-01 1.90316528e-01
3.61923426e-01 9.66810942e-01 1.03473693e-01 -2.44775489e-01
9.21509326e-01 -6.80435717e-01 -4.84080523e-01 2.91752368e-02
9.36778188e-01 -4.75775540e-01 6.96774662e-01 7.34163880e-01
-9.27198887e-01 -3.03647935e-01 -1.20555675e+00 5.18126667e-01
-1.76797181e-01 2.25692853e-01 7.97733366e-01 1.11323452e+00
-9.92196620e-01 8.74750316e-01 -8.18817139e-01 6.37407675e-02
2.32058972e-01 8.14294755e-01 1.66992381e-01 2.54441742e-02
-7.36687839e-01 7.39417017e-01 9.23415482e-01 -6.93112090e-02
-7.24535942e-01 -4.71358389e-01 -9.39323723e-01 -3.00565846e-02
4.92195398e-01 -3.14741641e-01 1.34027934e+00 -8.28234434e-01
-1.67535985e+00 6.62398636e-01 -5.42912126e-01 -3.60999763e-01
9.72200930e-02 2.42168397e-01 -2.27172151e-01 -3.47893655e-01
-8.08131844e-02 3.05823565e-01 7.09978759e-01 -1.16076314e+00
-5.68191826e-01 -1.52113333e-01 -1.39503911e-01 -2.49786153e-01
1.73123941e-01 4.47273523e-01 -4.13580745e-01 -5.03417492e-01
5.79126365e-02 -1.20511687e+00 -6.10953689e-01 -4.67227966e-01
-5.94262123e-01 -4.86233979e-01 5.08737147e-01 -3.87615502e-01
1.60393810e+00 -2.13265586e+00 6.12069368e-01 9.29930031e-01
2.70736683e-02 -9.14720222e-02 2.04158247e-01 2.07780093e-01
-8.80526099e-03 2.98467666e-01 -6.68366253e-01 4.11703825e-01
2.55672425e-01 2.41569147e-01 -3.61118764e-01 1.76377296e-01
2.79388547e-01 7.11497962e-01 -8.19264531e-01 -6.42634749e-01
-9.27960798e-02 -2.52021700e-01 -9.80753362e-01 2.37355739e-01
-8.80778432e-01 -1.41812608e-01 -8.96635771e-01 5.98101676e-01
3.54366362e-01 -1.57847870e-02 4.06912446e-01 3.60748291e-01
-1.17321856e-01 1.90240473e-01 -1.24768090e+00 1.31246245e+00
-1.93604976e-01 3.35853130e-01 -4.55120981e-01 -1.06093323e+00
1.29544008e+00 -1.19663224e-01 1.65224344e-01 -1.14460392e-02
5.33890389e-02 3.50158840e-01 1.39959350e-01 -4.33148354e-01
5.20042777e-01 2.11035386e-01 -4.83998179e-01 5.61946094e-01
-1.88659117e-01 -5.22502363e-01 5.20259738e-01 -1.70664713e-01
1.16361344e+00 3.99018317e-01 6.47665679e-01 -3.03128600e-01
7.40575850e-01 4.86891180e-01 6.58625126e-01 9.01700377e-01
2.37638697e-01 8.01583380e-02 1.11194837e+00 -3.26938123e-01
-7.79885590e-01 -7.44647861e-01 -3.20947438e-01 1.20021462e+00
-5.24663925e-02 -6.46435678e-01 -7.46607900e-01 -6.78424597e-01
2.08710015e-01 1.10275209e+00 -5.53724051e-01 -2.10008249e-01
-7.29734242e-01 -7.09540725e-01 4.31549907e-01 3.86338830e-01
4.06053439e-02 -1.15035927e+00 -5.24375021e-01 4.05920535e-01
2.97974408e-01 -5.91348529e-01 -9.76339132e-02 5.72369933e-01
-9.29733515e-01 -1.07591152e+00 -7.17364177e-02 -9.56639051e-01
7.96055377e-01 -4.96878386e-01 1.08417821e+00 4.10475701e-01
-1.73395261e-01 -8.04196857e-03 -2.17835441e-01 -1.37233332e-01
-1.02745175e+00 6.64256513e-02 -3.19564492e-01 -4.54546124e-01
3.30165893e-01 -3.56404752e-01 3.97708744e-01 3.54759425e-01
-5.24665356e-01 -2.44236425e-01 2.98277080e-01 1.16617119e+00
8.73890221e-01 6.97963595e-01 3.47849965e-01 -1.12055099e+00
7.40535438e-01 -6.40038908e-01 -1.37708139e+00 5.88142753e-01
-6.35710895e-01 7.33963013e-01 5.40926874e-01 -6.08848572e-01
-9.80214655e-01 3.97655606e-01 1.41614929e-01 2.71695724e-04
-2.21458618e-02 8.94340873e-01 -1.65630713e-01 -2.97685295e-01
6.86326563e-01 5.11693098e-02 -3.27703327e-01 -8.30164701e-02
1.42868593e-01 3.07504952e-01 4.40198541e-01 -1.28937042e+00
7.07478344e-01 -4.12287921e-01 1.65856078e-01 -4.53790754e-01
-1.32965013e-01 1.70542613e-01 -3.46675903e-01 7.75195956e-02
5.13849795e-01 -1.51193410e-01 -7.68869519e-01 1.46602854e-01
-1.24087465e+00 -4.36925143e-01 9.88230482e-03 7.02302903e-02
-6.33737981e-01 2.20108524e-01 1.17230695e-02 -1.04001200e+00
8.24427754e-02 -1.69248664e+00 7.83032060e-01 1.84709132e-01
-7.17680454e-01 -7.47671127e-01 2.69385546e-01 -2.20315963e-01
-9.47736874e-02 2.22869560e-01 1.82315350e+00 -7.56468296e-01
-6.62655115e-01 -3.42427045e-01 1.98271379e-01 1.28967136e-01
-2.00245500e-01 6.74500704e-01 -4.23382342e-01 1.49723694e-01
-2.07363188e-01 -1.31991968e-01 4.83645976e-01 3.60721320e-01
1.60132945e+00 -1.87619969e-01 -8.48341167e-01 4.80347991e-01
1.25573468e+00 7.90047586e-01 5.19413650e-01 3.06303084e-01
3.93720329e-01 5.23457289e-01 7.39141464e-01 3.33906591e-01
9.58871469e-02 8.38994861e-01 1.76267162e-01 6.23168409e-01
5.49750507e-01 -2.93045640e-01 4.37015742e-01 3.99805307e-01
1.73695579e-01 1.61548480e-01 -1.41678715e+00 4.82152194e-01
-1.89833617e+00 -7.35559165e-01 -1.00174122e-01 2.35810328e+00
1.25228715e+00 3.09773803e-01 2.51305431e-01 4.38738972e-01
8.49295020e-01 -2.98060387e-01 -4.72598523e-01 -8.57447863e-01
3.15562814e-01 6.92009807e-01 4.51464742e-01 5.28528988e-01
-8.57289910e-01 8.30356598e-01 7.40641212e+00 7.30779231e-01
-7.55223572e-01 -4.09603685e-01 4.76203918e-01 3.42109263e-01
-5.43576181e-01 3.89458746e-01 -9.94069755e-01 3.32703561e-01
9.65323508e-01 -7.51815021e-01 7.35868812e-01 1.32791412e+00
-3.54061693e-01 -4.95519817e-01 -1.54109788e+00 6.13221824e-01
-2.65310675e-01 -1.23496795e+00 -2.54888952e-01 -2.57030010e-01
8.43580425e-01 -3.74117911e-01 6.38252720e-02 4.89359647e-01
7.08051562e-01 -1.08722031e+00 7.09005773e-01 2.71871299e-01
4.53556240e-01 -1.22570121e+00 5.34312904e-01 3.37188989e-01
-8.84284139e-01 -5.14226258e-01 -8.07645172e-02 8.60756189e-02
-2.42364541e-01 5.73411644e-01 -9.91769969e-01 2.60500342e-01
4.54159558e-01 2.20382318e-01 -5.27995467e-01 1.22284305e+00
-5.28478146e-01 5.98480046e-01 -4.18445349e-01 -5.67579269e-01
-9.43816975e-02 -2.08812416e-01 4.66409773e-01 1.00891960e+00
5.24913073e-01 -3.95071320e-02 2.23959252e-01 1.43331170e+00
2.22337350e-01 6.06042966e-02 -6.43783331e-01 -1.08447947e-01
7.52322912e-01 6.12148225e-01 -7.46943235e-01 -2.85231739e-01
-8.16424116e-02 2.18918025e-01 2.44423345e-01 2.65830308e-01
-8.17280948e-01 -7.09492207e-01 2.14164108e-01 -3.13002616e-01
4.44382161e-01 -8.06086808e-02 -5.69328368e-01 -5.97524405e-01
7.28907958e-02 -1.20081127e+00 4.10818279e-01 -4.45194542e-01
-5.25055170e-01 5.64264715e-01 5.59429109e-01 -5.58549583e-01
-8.08358133e-01 -6.83232963e-01 -7.20688164e-01 9.00393724e-01
-1.14305174e+00 -3.08752716e-01 7.78313279e-02 2.04318151e-01
4.42633145e-02 -2.98246443e-01 1.09378457e+00 -3.03574592e-01
-9.63423371e-01 5.91118038e-01 -7.91662410e-02 -2.26530358e-01
6.53854087e-02 -1.11851859e+00 3.72917175e-01 7.22032130e-01
-3.08568954e-01 7.69908071e-01 8.57152343e-01 -8.89887273e-01
-1.67954803e+00 -8.11902106e-01 6.99977517e-01 -2.55367815e-01
8.61249030e-01 -6.32710531e-02 -9.20672715e-01 5.57581365e-01
-3.45164359e-01 -1.68176234e-01 4.22722012e-01 2.69554466e-01
-3.59462291e-01 7.93718249e-02 -1.24461889e+00 5.31092227e-01
5.43242097e-01 -2.06959665e-01 -5.67114770e-01 1.75614759e-01
6.15084231e-01 -5.60066521e-01 -8.85077417e-01 3.72548521e-01
4.07171011e-01 -5.74309409e-01 6.69863522e-01 -5.13638437e-01
5.57421327e-01 -4.48347330e-01 -1.46565661e-01 -1.21360445e+00
-1.00995250e-01 -9.07538354e-01 -2.54338592e-01 1.16926885e+00
7.04127312e-01 -7.64597297e-01 7.66327798e-01 8.32887471e-01
7.38317147e-02 -7.53938138e-01 -9.40986931e-01 -6.97641015e-01
2.07634754e-02 -5.76764345e-01 1.04396403e+00 5.26174784e-01
3.85094881e-01 -1.29063308e-01 6.92760870e-02 1.95019230e-01
8.32263350e-01 4.78395492e-01 6.37997568e-01 -1.27128398e+00
-7.67842531e-01 -6.93073094e-01 -3.04719210e-01 -5.91918945e-01
6.72876835e-01 -1.07030272e+00 3.00220132e-01 -7.81024337e-01
1.76566377e-01 -9.53218520e-01 1.60768889e-02 6.28466249e-01
-1.90654486e-01 -5.80106795e-01 -2.38789871e-01 -1.03666127e-01
-2.19815090e-01 3.09578449e-01 5.94154656e-01 -8.62882659e-02
-6.01676464e-01 3.24954569e-01 -4.94355232e-01 8.22563291e-01
8.40676367e-01 -1.00533760e+00 -2.35542074e-01 -1.23862676e-01
5.31397164e-01 3.99888337e-01 7.31054917e-02 -8.06102872e-01
1.95713222e-01 -5.50044954e-01 -1.25277832e-01 -5.77247262e-01
-2.20455661e-01 -7.75526464e-01 5.47355354e-01 6.06818855e-01
-6.54614747e-01 7.66946003e-02 2.71525443e-01 3.77373338e-01
-9.72089320e-02 -9.99929667e-01 7.40151823e-01 2.03181326e-01
-6.21470332e-01 -1.65208161e-01 -4.56032723e-01 -1.37217313e-01
1.20121849e+00 -1.82639897e-01 -8.37485865e-03 9.47909877e-02
-5.08222103e-01 3.84947032e-01 6.32978916e-01 1.10679351e-01
4.55319554e-01 -1.23828638e+00 -3.18208992e-01 2.37280279e-01
1.43373877e-01 -3.83929424e-02 -4.23813611e-01 5.19408107e-01
-5.48893154e-01 3.64077419e-01 -2.39405520e-02 -7.07661092e-01
-1.24028265e+00 6.95635796e-01 1.66509837e-01 -3.92121673e-01
-2.74463862e-01 4.78402317e-01 -1.97081268e-01 -4.17079955e-01
3.18203270e-01 -5.59995294e-01 -1.36287168e-01 -4.10133362e-01
5.03494918e-01 3.82333159e-01 -1.37611404e-01 -1.94580346e-01
-3.79394531e-01 5.36070466e-01 1.32362127e-01 -9.48079005e-02
1.46014607e+00 6.27877951e-01 -6.59022212e-01 5.33627868e-01
8.92200530e-01 -9.78513211e-02 -6.87540293e-01 -2.61875898e-01
6.64333582e-01 -3.87896448e-01 -1.28009554e-03 -6.89629316e-01
-6.56804919e-01 3.26341003e-01 1.02537349e-01 1.68182805e-01
1.04037941e+00 -1.19152963e-01 -4.77036415e-03 7.97311544e-01
6.61813319e-01 -9.11473930e-01 -4.05550629e-01 6.96134090e-01
6.88491404e-01 -7.47810423e-01 2.22175494e-01 -5.51443100e-01
-3.32868874e-01 1.48021817e+00 3.77611279e-01 -2.09803879e-01
4.60078686e-01 7.85664022e-01 -7.91812181e-01 8.82812142e-02
-7.81185508e-01 1.61631346e-01 1.84400216e-01 7.45929420e-01
2.67018616e-01 1.32229358e-01 -3.75248164e-01 8.06632817e-01
-2.02332005e-01 2.02446058e-01 5.14925063e-01 1.04957962e+00
-5.34167111e-01 -1.56700075e+00 -7.91889548e-01 4.50045139e-01
-1.83677301e-01 2.31371280e-02 -1.34325102e-01 6.74881041e-01
-1.18960418e-01 6.43141747e-01 -3.48222136e-01 -4.67467606e-01
2.17163309e-01 2.68410057e-01 6.79359138e-01 -8.14341009e-01
-5.73411584e-01 -2.77449429e-01 4.50914860e-01 -4.30398703e-01
-8.05428848e-02 -7.89525986e-01 -1.48256862e+00 -2.40211248e-01
-4.56911892e-01 2.86736935e-01 7.52886415e-01 7.15885699e-01
2.41474826e-02 4.24168140e-01 6.49067879e-01 -6.68362319e-01
-7.28307605e-01 -2.63249099e-01 -2.47539654e-01 -2.80059546e-01
-1.25393700e-02 -8.56161714e-01 -2.33902141e-01 -6.33393228e-02] | [8.205163955688477, 7.181873798370361] |
a319a2db-a678-497a-bbee-bf2acab3f1f4 | unsupervised-learning-of-3d-scene-flow-from | 2206.03673 | null | https://arxiv.org/abs/2206.03673v1 | https://arxiv.org/pdf/2206.03673v1.pdf | Unsupervised Learning of 3D Scene Flow from Monocular Camera | Scene flow represents the motion of points in the 3D space, which is the counterpart of the optical flow that represents the motion of pixels in the 2D image. However, it is difficult to obtain the ground truth of scene flow in the real scenes, and recent studies are based on synthetic data for training. Therefore, how to train a scene flow network with unsupervised methods based on real-world data shows crucial significance. A novel unsupervised learning method for scene flow is proposed in this paper, which utilizes the images of two consecutive frames taken by monocular camera without the ground truth of scene flow for training. Our method realizes the goal that training scene flow network with real-world data, which bridges the gap between training data and test data and broadens the scope of available data for training. Unsupervised learning of scene flow in this paper mainly consists of two parts: (i) depth estimation and camera pose estimation, and (ii) scene flow estimation based on four different loss functions. Depth estimation and camera pose estimation obtain the depth maps and camera pose between two consecutive frames, which provide further information for the next scene flow estimation. After that, we used depth consistency loss, dynamic-static consistency loss, Chamfer loss, and Laplacian regularization loss to carry out unsupervised training of the scene flow network. To our knowledge, this is the first paper that realizes the unsupervised learning of 3D scene flow from monocular camera. The experiment results on KITTI show that our method for unsupervised learning of scene flow meets great performance compared to traditional methods Iterative Closest Point (ICP) and Fast Global Registration (FGR). The source code is available at: https://github.com/IRMVLab/3DUnMonoFlow. | ['Hesheng Wang', 'Ruiqi Ding', 'Xiaoyu Tian', 'Guangming Wang'] | 2022-06-08 | null | null | null | null | ['scene-flow-estimation'] | ['computer-vision'] | [ 1.10221151e-02 -4.67450410e-01 -1.66970402e-01 -3.18998396e-01
-1.88771114e-01 -4.61488366e-01 2.81054497e-01 -4.24075484e-01
-6.29885435e-01 6.71166718e-01 8.39602947e-02 -3.77422012e-02
7.11368024e-02 -8.99113655e-01 -6.05575621e-01 -8.30970228e-01
2.15140253e-01 1.52228624e-01 4.72149938e-01 2.04995587e-01
4.29734230e-01 6.23841465e-01 -1.32091665e+00 -2.32561767e-01
7.84929276e-01 8.00690591e-01 4.52947974e-01 6.78335428e-01
-2.34470591e-01 1.21944201e+00 -3.47789764e-01 -9.79468506e-03
4.88462538e-01 -4.77842629e-01 -7.05396354e-01 2.97431469e-01
6.41085863e-01 -7.41199911e-01 -8.36888313e-01 1.06408262e+00
3.85829657e-01 4.01398838e-01 5.05815804e-01 -1.40635729e+00
-2.42267236e-01 -1.14657376e-02 -6.36967719e-01 3.48750979e-01
5.26552558e-01 3.26368898e-01 5.39650500e-01 -8.38184953e-01
7.38380253e-01 9.67124760e-01 3.82094562e-01 4.48021799e-01
-7.34198213e-01 -6.44988000e-01 3.09176883e-03 1.97555348e-01
-1.42388380e+00 -2.96332359e-01 1.16603136e+00 -6.86789095e-01
5.52049160e-01 -3.21343653e-02 8.15347493e-01 7.13054895e-01
1.68452132e-02 8.05461705e-01 8.24252725e-01 -2.45064273e-01
7.46832639e-02 3.76952551e-02 -6.83505237e-02 7.37204909e-01
1.90901712e-01 4.03225899e-01 -2.96805650e-01 4.37785417e-01
1.29185534e+00 1.80272281e-01 -7.60651648e-01 -5.75980365e-01
-1.20730984e+00 6.20158374e-01 5.62215269e-01 1.62331998e-01
-1.07983584e-02 7.29543045e-02 1.78565785e-01 9.56212208e-02
3.45738769e-01 1.07172735e-01 -4.12887394e-01 -1.99757427e-01
-7.75248349e-01 -2.07029164e-01 7.11371541e-01 1.05555642e+00
1.27346087e+00 2.04688981e-01 3.16314369e-01 5.24853647e-01
5.13233125e-01 6.67684615e-01 5.80895424e-01 -1.25752902e+00
6.28679693e-01 7.29152858e-01 -1.36132678e-02 -1.51212740e+00
-2.64482319e-01 2.55846661e-02 -8.90028775e-01 2.19766185e-01
6.10439062e-01 -3.26829284e-01 -5.92993319e-01 1.42493260e+00
4.65175897e-01 7.19509065e-01 2.53357389e-03 1.29532075e+00
1.07113123e+00 8.24033737e-01 -4.33786035e-01 -3.44948024e-01
6.56533003e-01 -9.51127648e-01 -7.07953870e-01 -1.80141151e-01
6.35883927e-01 -7.83004045e-01 8.80114317e-01 2.89188355e-01
-9.41670179e-01 -8.49641442e-01 -9.60986495e-01 -2.52011895e-01
-1.81700960e-01 1.20719835e-01 7.24013567e-01 5.66369653e-01
-7.82182455e-01 1.72069490e-01 -9.70221460e-01 -4.12660569e-01
3.12673777e-01 2.18404040e-01 -6.05350614e-01 -3.57351333e-01
-9.97887075e-01 6.50695384e-01 5.45014262e-01 4.02507693e-01
-9.05402124e-01 -7.08447993e-01 -1.17802465e+00 -4.11784768e-01
3.41696471e-01 -7.00476587e-01 7.18033075e-01 -8.53874564e-01
-1.59393454e+00 8.29077363e-01 1.68524142e-02 -7.05800727e-02
6.76365674e-01 -2.86438435e-01 6.70257360e-02 2.50569910e-01
1.80233255e-01 6.24134839e-01 5.03479540e-01 -1.21064436e+00
-5.87432623e-01 -1.90696225e-01 1.09536864e-01 4.15945143e-01
-1.89361945e-01 -3.18490952e-01 -7.67434359e-01 -1.74387351e-01
3.02960843e-01 -7.07933187e-01 -1.54903561e-01 2.36587375e-01
-4.20912087e-01 1.51510134e-01 1.03086329e+00 -4.99342561e-01
1.06035197e+00 -2.06949544e+00 5.89999557e-03 -2.24717645e-04
1.14315026e-01 7.64649063e-02 7.26569891e-02 2.00891361e-01
-4.17964496e-02 -2.55506128e-01 -5.65127611e-01 -1.81903839e-01
-6.22700155e-01 2.88451701e-01 -1.26965284e-01 8.27396929e-01
-1.46488562e-01 7.40243196e-01 -1.15688705e+00 -7.68936932e-01
8.73211622e-01 3.94390970e-01 -5.75130761e-01 4.49122936e-01
1.44736543e-01 1.05579734e+00 -4.46229100e-01 2.93228775e-01
9.97673094e-01 -9.04033259e-02 -2.76767433e-01 -4.60443884e-01
-2.86659807e-01 -8.20731893e-02 -1.51760089e+00 2.07123971e+00
-3.12005013e-01 7.66881049e-01 -2.35652149e-01 -9.48804498e-01
1.03114617e+00 -1.86561532e-02 8.47511530e-01 -4.36245054e-01
3.44087899e-01 4.48924024e-03 -2.42544085e-01 -8.87036741e-01
3.18834484e-01 1.47688612e-01 1.50741577e-01 3.00899655e-01
2.47639529e-02 -6.66299403e-01 3.79117966e-01 1.01428583e-01
7.74672568e-01 4.40023899e-01 3.61729041e-02 -2.95857675e-02
9.96012449e-01 -1.00449808e-02 8.67495894e-01 2.63608605e-01
-2.64008969e-01 8.78315866e-01 2.21851438e-01 -5.87407291e-01
-9.79670525e-01 -1.00170219e+00 -1.30499288e-01 -1.05023589e-02
8.62518728e-01 -2.34577477e-01 -5.98854601e-01 -5.57909012e-01
-2.63339639e-01 2.66925931e-01 -2.68830508e-01 -7.90301710e-03
-8.10857773e-01 -6.20991170e-01 2.15859890e-01 3.43196452e-01
1.06406343e+00 -1.03393888e+00 -6.87988997e-01 -9.35664400e-02
-4.52645808e-01 -1.61093831e+00 -6.88631475e-01 -5.11472762e-01
-1.02868915e+00 -1.52640676e+00 -5.90111196e-01 -8.12837183e-01
8.63606691e-01 4.76614296e-01 7.86795557e-01 -1.33065823e-02
-3.18584561e-01 4.97405350e-01 -7.14770630e-02 9.92224440e-02
-6.01704083e-02 -1.47365868e-01 3.16105634e-02 2.91557908e-01
2.64833480e-01 -6.48591995e-01 -9.14444029e-01 6.21304274e-01
-9.35641527e-01 1.64701715e-01 2.34213457e-01 5.29996932e-01
5.00608563e-01 1.97197720e-01 -1.00169964e-01 -6.41282916e-01
5.33073172e-02 -1.71036035e-01 -9.56597507e-01 -1.96485564e-01
-2.06776664e-01 -3.82204086e-01 5.23095965e-01 -9.96765196e-02
-1.35568309e+00 3.77432734e-01 -2.94455662e-02 -7.63035715e-01
-3.15523267e-01 2.46524394e-01 -2.89143920e-01 -1.69389531e-01
5.76202452e-01 4.21067595e-01 1.09159850e-01 -1.76594585e-01
3.92075241e-01 4.62498367e-01 6.64517045e-01 -3.37308913e-01
1.01575851e+00 9.01584089e-01 1.42684802e-01 -9.15049553e-01
-8.08753788e-01 -7.43962646e-01 -1.05952168e+00 -4.79487240e-01
1.21911573e+00 -1.00395668e+00 -8.86324942e-01 8.39982986e-01
-1.31568038e+00 -4.93018329e-01 -2.69010544e-01 1.00933981e+00
-7.27158070e-01 8.39247108e-01 -6.59061909e-01 -5.46194315e-01
-7.90322945e-02 -1.15472460e+00 8.29645157e-01 3.73102635e-01
2.87375569e-01 -1.31435895e+00 2.88388997e-01 4.69218165e-01
-1.11767031e-01 3.94827843e-01 2.10558742e-01 3.74088973e-01
-1.16661060e+00 -4.08031642e-02 -3.20557117e-01 7.38950491e-01
4.10113603e-01 1.84453785e-01 -9.04567659e-01 -2.63353318e-01
4.03629184e-01 -8.52547586e-02 6.12845659e-01 7.04839766e-01
1.11253595e+00 6.17576204e-02 -1.46768734e-01 1.24314654e+00
1.72668219e+00 3.93031806e-01 9.72693086e-01 3.64770621e-01
1.28612971e+00 6.56588852e-01 6.87720358e-01 1.99386775e-01
3.89021784e-01 6.19907558e-01 4.83706743e-01 -1.11792564e-01
-1.42181054e-01 -4.90521610e-01 3.26500595e-01 1.05925000e+00
-1.76343396e-01 -7.74205998e-02 -8.86713028e-01 3.93186450e-01
-1.84017003e+00 -9.71280456e-01 -4.39592481e-01 2.37880039e+00
4.32853580e-01 -3.48940045e-02 -2.16873229e-01 1.12092420e-01
7.41430342e-01 2.14407966e-01 -5.27636945e-01 2.31898531e-01
-1.14418432e-01 -2.57937312e-01 6.22714937e-01 8.74481201e-01
-1.22197104e+00 1.02256918e+00 4.97501183e+00 4.61522460e-01
-1.22515833e+00 4.81026210e-02 5.55247247e-01 1.55701309e-01
2.72201318e-02 1.93948463e-01 -7.89729178e-01 6.18777871e-01
2.39186898e-01 -3.49725373e-02 2.01248899e-01 6.40851378e-01
4.85846698e-01 -3.24900895e-01 -9.68837738e-01 1.48043048e+00
2.31990948e-01 -1.13182247e+00 -9.20048282e-02 -5.60515299e-02
1.08106196e+00 7.05466494e-02 -3.54525208e-01 -2.00817972e-01
3.42455618e-02 -6.14468932e-01 2.74541676e-01 6.21266305e-01
7.22666204e-01 -4.48900819e-01 7.01644778e-01 4.93068010e-01
-1.23606229e+00 1.51811922e-02 -5.46681821e-01 -9.67513844e-02
5.56106985e-01 7.08067775e-01 -4.19450074e-01 7.08494723e-01
6.41158521e-01 1.52909327e+00 -3.81436169e-01 1.46151149e+00
-5.29301584e-01 2.55523503e-01 -3.11848074e-01 3.90589207e-01
1.72672734e-01 -7.00522840e-01 6.05281115e-01 8.20924222e-01
2.84633368e-01 5.01421466e-02 3.16578597e-01 9.00831401e-01
5.05347401e-02 1.29586041e-01 -8.81853998e-01 4.70959187e-01
2.39952862e-01 1.26923192e+00 -7.83476114e-01 -1.62794694e-01
-4.56439942e-01 8.99983048e-01 1.82255283e-02 4.30598170e-01
-6.72731042e-01 -3.01370740e-01 3.40594977e-01 5.91315888e-02
-1.36932477e-01 -4.89825517e-01 -1.52195141e-01 -1.62551641e+00
9.52436924e-02 -2.46208444e-01 3.48345965e-01 -1.06230116e+00
-1.18677676e+00 3.82094979e-01 1.36762246e-01 -1.77189255e+00
-9.46041420e-02 -5.76160192e-01 -8.24515343e-01 7.21938074e-01
-1.68956959e+00 -6.68392360e-01 -1.03115952e+00 8.27013731e-01
5.75055361e-01 -5.08833639e-02 1.16794579e-01 4.94952559e-01
-6.61447167e-01 1.88105196e-01 -6.85287043e-02 5.70932627e-01
6.45178020e-01 -9.33995068e-01 1.47852913e-01 1.09542763e+00
9.15359035e-02 1.81410953e-01 2.27643833e-01 -4.98902410e-01
-1.28243756e+00 -1.03781736e+00 5.12048781e-01 -7.08994865e-01
4.80856299e-01 -1.84640631e-01 -7.44863153e-01 7.20647216e-01
-2.38499288e-02 3.65792960e-01 1.48905456e-01 -5.78500986e-01
2.49375775e-01 -5.16619205e-01 -9.48088229e-01 3.07902932e-01
1.11708879e+00 -4.48483646e-01 -2.39504397e-01 3.50222796e-01
6.13947451e-01 -7.11706102e-01 -8.16710234e-01 4.25883979e-01
2.69496769e-01 -1.12036550e+00 1.10350466e+00 1.84420422e-02
6.84342742e-01 -7.17057824e-01 2.09666677e-02 -1.03322196e+00
2.36099765e-01 -5.40859044e-01 4.69968580e-02 1.13220394e+00
-1.18099645e-01 -6.93763971e-01 1.18115807e+00 4.64454740e-01
-1.52845263e-01 -5.29577672e-01 -6.89450681e-01 -6.01100206e-01
-1.02549881e-01 -5.48562288e-01 3.41277331e-01 1.14816940e+00
-3.73386651e-01 2.62500107e-01 -3.19177896e-01 2.19080448e-01
9.23636615e-01 2.63987556e-02 1.31833971e+00 -9.31051493e-01
-1.12965055e-01 -2.57549554e-01 -8.09673131e-01 -1.67688644e+00
1.87154070e-01 -7.19460070e-01 -3.38213239e-03 -1.55520201e+00
5.13094105e-02 -4.94399250e-01 1.66658744e-01 -4.62173447e-02
-4.14326638e-02 1.98291764e-01 1.85863629e-01 4.68181610e-01
-2.88622350e-01 7.54786670e-01 1.81646717e+00 -5.21092080e-02
-4.51923728e-01 -9.04559493e-02 -1.17475241e-01 9.23367143e-01
5.93567073e-01 -2.41755739e-01 -8.78675997e-01 -6.37771368e-01
-2.77311802e-02 3.41061264e-01 3.88672382e-01 -1.15185821e+00
6.55146301e-01 -2.79137611e-01 3.84568870e-01 -6.94524407e-01
3.25772852e-01 -1.02329218e+00 9.22875106e-02 3.49401176e-01
2.82028969e-02 -7.22076884e-03 2.58765742e-03 4.69890684e-01
-4.16176587e-01 -2.52607465e-01 7.26563394e-01 -3.31039816e-01
-1.13207340e+00 8.62626314e-01 3.02702952e-02 2.48405457e-01
1.03562307e+00 -6.03339970e-01 -1.84668720e-01 -4.60946351e-01
-4.36467916e-01 1.89193338e-01 4.80204225e-01 3.60612780e-01
9.93152261e-01 -1.26609266e+00 -6.56602919e-01 5.26828527e-01
-4.18770984e-02 6.83904767e-01 3.27725679e-01 8.43411684e-01
-1.25379944e+00 2.11248308e-01 -2.65374660e-01 -1.03464913e+00
-9.19398129e-01 4.43447590e-01 6.49405837e-01 1.52692571e-01
-6.78566098e-01 6.40246451e-01 6.52143300e-01 -6.77188635e-01
-2.04839986e-02 -1.18852548e-01 -2.00122133e-01 -2.29859442e-01
2.80742645e-01 4.66421992e-01 -2.94654459e-01 -9.07509863e-01
-3.04013938e-01 1.29780722e+00 2.03678280e-01 3.55503634e-02
9.49081481e-01 -4.03071314e-01 -2.85163999e-01 3.69589537e-01
1.68684149e+00 -1.67552173e-01 -1.63671768e+00 -4.30348255e-02
-5.03878355e-01 -1.04230237e+00 1.66460931e-01 -1.20422453e-01
-1.62530065e+00 1.07117558e+00 5.38263857e-01 -2.41081148e-01
1.21191382e+00 -3.12275290e-01 8.21321666e-01 1.94416910e-01
5.51018119e-01 -7.20332146e-01 1.58675253e-01 4.42253888e-01
4.06219184e-01 -1.34327221e+00 6.29882291e-02 -7.00320959e-01
-2.98994452e-01 1.20638239e+00 1.08678055e+00 -3.98283452e-01
8.38444114e-01 3.63972783e-02 2.81621337e-01 -2.82318355e-03
-1.45305827e-01 -1.34569153e-01 1.09214805e-01 5.49386621e-01
1.10217981e-01 -4.57304358e-01 -9.46066827e-02 -3.00704509e-01
-1.61444411e-01 1.22136183e-01 7.24580169e-01 6.92312181e-01
-2.02739760e-01 -8.88743579e-01 -2.45076045e-01 -5.21456450e-02
1.08076250e-02 1.45084679e-01 1.94724537e-02 9.18856859e-01
1.75434709e-01 8.34577918e-01 3.15301448e-01 -2.76572943e-01
4.32753533e-01 -5.35038173e-01 6.05546653e-01 -4.41941082e-01
9.20149311e-02 -1.50616154e-01 -3.48351777e-01 -6.60503089e-01
-8.72105360e-01 -5.50538123e-01 -1.17788339e+00 -2.66450644e-01
-1.70185372e-01 6.74988553e-02 4.86457258e-01 7.99789190e-01
-7.13261738e-02 1.68940797e-01 1.00973916e+00 -1.01834774e+00
2.10158452e-01 -6.80682242e-01 -4.72958833e-01 7.08841264e-01
3.36677551e-01 -6.20128214e-01 -7.39591241e-01 4.31776732e-01] | [8.592412948608398, -2.031944513320923] |
3291b837-d0b1-4ab0-8693-b712b0095fb7 | slsg-industrial-image-anomaly-detection-by | 2305.00398 | null | https://arxiv.org/abs/2305.00398v1 | https://arxiv.org/pdf/2305.00398v1.pdf | SLSG: Industrial Image Anomaly Detection by Learning Better Feature Embeddings and One-Class Classification | Industrial image anomaly detection under the setting of one-class classification has significant practical value. However, most existing models struggle to extract separable feature representations when performing feature embedding and struggle to build compact descriptions of normal features when performing one-class classification. One direct consequence of this is that most models perform poorly in detecting logical anomalies which violate contextual relationships. Focusing on more effective and comprehensive anomaly detection, we propose a network based on self-supervised learning and self-attentive graph convolution (SLSG) for anomaly detection. SLSG uses a generative pre-training network to assist the encoder in learning the embedding of normal patterns and the reasoning of position relationships. Subsequently, SLSG introduces the pseudo-prior knowledge of anomaly through simulated abnormal samples. By comparing the simulated anomalies, SLSG can better summarize the normal features and narrow down the hypersphere used for one-class classification. In addition, with the construction of a more general graph structure, SLSG comprehensively models the dense and sparse relationships among elements in the image, which further strengthens the detection of logical anomalies. Extensive experiments on benchmark datasets show that SLSG achieves superior anomaly detection performance, demonstrating the effectiveness of our method. | ['Zhaoyang Wu', 'Zhiwei Yang', 'Jing Liu', 'Minghui Yang'] | 2023-04-30 | null | null | null | null | ['one-class-classification'] | ['miscellaneous'] | [ 8.68952498e-02 2.31159270e-01 1.33765206e-01 -4.31198150e-01
1.08312324e-01 4.92200814e-02 5.04612803e-01 2.95215189e-01
2.54803807e-01 8.32131132e-02 -2.16471665e-02 -3.39572847e-01
-2.15710372e-01 -1.09197819e+00 -5.93685508e-01 -6.65315270e-01
-5.44029772e-01 7.25686252e-02 2.34138951e-01 -2.41335273e-01
3.11437577e-01 6.16267920e-01 -1.54320490e+00 2.00021833e-01
1.04712546e+00 1.26533961e+00 -2.49327362e-01 4.77325052e-01
-3.25084746e-01 8.53223026e-01 -7.38485754e-01 -2.94827759e-01
2.65412658e-01 -5.84357560e-01 -4.68762547e-01 4.70280051e-01
4.62060690e-01 -4.73451734e-01 -7.87576258e-01 1.31419671e+00
6.64354339e-02 8.07684883e-02 6.39237344e-01 -1.77358866e+00
-9.60097969e-01 3.70840341e-01 -5.88388324e-01 5.79003036e-01
3.25266004e-01 2.90314227e-01 1.21814883e+00 -7.34447479e-01
1.00891672e-01 1.13257694e+00 5.61567664e-01 2.35687122e-01
-8.94583762e-01 -4.60585177e-01 5.86651504e-01 6.20293319e-01
-1.26254666e+00 4.32521978e-04 1.14645362e+00 -2.42215604e-01
1.01114798e+00 3.61357301e-01 7.77451277e-01 1.01988912e+00
3.74428988e-01 7.35222697e-01 2.79287219e-01 -2.93627203e-01
1.68913245e-01 -2.23693684e-01 1.64833710e-01 1.10585606e+00
7.19159603e-01 -1.02001637e-01 -4.29586172e-01 -1.76723287e-01
5.73453903e-01 5.44350147e-01 -2.94479251e-01 -5.04148483e-01
-8.67503345e-01 9.28726673e-01 5.65012991e-01 3.04408103e-01
-3.47541124e-01 -7.26197511e-02 5.45603454e-01 5.00645101e-01
4.95710582e-01 4.79909271e-01 3.02732922e-02 2.88216323e-01
-4.32864547e-01 -3.08946759e-01 5.35643578e-01 8.75425696e-01
7.87098825e-01 5.81847072e-01 -5.78566156e-02 6.72690153e-01
4.34611559e-01 3.94513935e-01 4.87186462e-01 -4.16197062e-01
2.17592806e-01 1.27206421e+00 -6.17741466e-01 -1.63163829e+00
-2.69228697e-01 -8.08086157e-01 -1.14667428e+00 6.39713481e-02
2.45784000e-02 1.60073519e-01 -9.19242322e-01 1.46624577e+00
7.43098333e-02 5.91202855e-01 1.59649149e-01 5.78223646e-01
6.06455088e-01 6.61568165e-01 -3.61395180e-01 3.04498170e-02
9.84819114e-01 -8.37850571e-01 -6.62240744e-01 -3.90528738e-01
8.40258658e-01 -1.63621157e-01 9.88433897e-01 2.46240094e-01
-4.88323748e-01 -3.42199355e-01 -1.23392558e+00 3.35786939e-01
-5.40146410e-01 -1.13309428e-01 9.25374389e-01 4.55489784e-01
-9.39639568e-01 4.85837191e-01 -8.06113541e-01 -3.27820182e-01
8.14921558e-01 1.31170258e-01 -3.70945603e-01 -2.66219139e-01
-1.06630647e+00 4.41104561e-01 3.88621628e-01 1.99040309e-01
-5.93046963e-01 -4.44887400e-01 -1.48719811e+00 2.02426776e-01
4.06150132e-01 -3.44144464e-01 5.45514703e-01 -7.50045836e-01
-8.70312274e-01 7.12251544e-01 -2.41327798e-03 -6.84821367e-01
1.50032014e-01 -4.83887903e-02 -9.75184977e-01 4.63208020e-01
1.41103193e-01 1.38399377e-01 1.11580098e+00 -1.07815933e+00
-6.18888497e-01 -5.11008680e-01 3.26152891e-02 1.10673897e-01
-7.17563212e-01 -4.40043092e-01 -2.68540204e-01 -6.88429654e-01
6.35977626e-01 -5.03585160e-01 -2.26427138e-01 -1.55854104e-02
-7.65087187e-01 -2.82844305e-01 1.30767405e+00 -3.59510899e-01
1.41587877e+00 -2.55277300e+00 -2.57304221e-01 6.89877689e-01
5.77470660e-01 1.73664525e-01 -1.06232889e-01 4.68168497e-01
-4.77852166e-01 -5.17493933e-02 -4.10962075e-01 -2.07053885e-01
1.36791198e-02 6.26215518e-01 -7.63683379e-01 6.64915562e-01
7.43880570e-01 9.13347840e-01 -9.43414211e-01 -3.77907544e-01
3.85752708e-01 5.25861084e-02 -6.04557574e-01 2.50207871e-01
1.20674394e-01 1.63827986e-01 -5.48707843e-01 8.42578351e-01
6.53484464e-01 -4.86516178e-01 -1.74630687e-01 -9.07462239e-02
4.07680452e-01 1.25839803e-02 -1.04089391e+00 1.38458669e+00
-3.55073102e-02 5.42338789e-01 -3.96245271e-01 -1.56004548e+00
1.07105052e+00 -2.23778099e-01 5.30281961e-01 -7.73800552e-01
-1.81604147e-01 5.73822968e-02 1.14425868e-01 -5.84282458e-01
1.78641796e-01 3.76749814e-01 1.10577874e-01 4.60123867e-01
1.35985136e-01 1.95263937e-01 1.66958734e-01 6.18900836e-01
1.41466773e+00 -2.59744406e-01 2.36914426e-01 -4.94641438e-02
7.29488254e-01 -2.67553985e-01 5.19109070e-01 7.30323851e-01
-1.64548621e-01 5.07489741e-01 8.64125431e-01 -8.26137364e-01
-5.25570929e-01 -1.23566616e+00 -9.07847956e-02 7.67755806e-01
4.36341941e-01 -6.11292481e-01 -4.21300769e-01 -1.18187785e+00
1.09574072e-01 9.02299106e-01 -7.21583366e-01 -8.39660525e-01
-3.56577784e-01 -9.86157894e-01 4.00604606e-01 5.04237473e-01
5.92947423e-01 -1.08584094e+00 -3.31941694e-01 -1.66165084e-01
9.04665068e-02 -1.02859318e+00 -7.21986815e-02 1.78283170e-01
-7.54007220e-01 -1.25351083e+00 1.90008432e-01 -8.24609518e-01
1.30116606e+00 2.59326488e-01 8.69657636e-01 6.59487188e-01
-6.48089588e-01 5.94086945e-01 -4.23863590e-01 -3.78693134e-01
-2.87507862e-01 -4.27797914e-01 6.93703517e-02 3.38332683e-01
5.80564857e-01 -7.79940784e-01 -5.62631488e-01 2.28906557e-01
-1.14792120e+00 -3.27669412e-01 6.16421163e-01 8.37814033e-01
3.84212643e-01 5.17311156e-01 5.53439558e-01 -8.89693499e-01
4.44571972e-01 -6.67649031e-01 -3.10357481e-01 4.67495024e-02
-8.09588075e-01 7.87331760e-02 9.17079389e-01 -1.69869348e-01
-7.18215704e-01 -3.00416201e-01 -2.07367986e-01 -5.41447222e-01
-2.64336079e-01 2.37995461e-01 -1.98280826e-01 -1.32333025e-01
5.33961892e-01 6.55962467e-01 1.89326763e-01 -4.52102795e-02
4.72383499e-02 2.99520105e-01 5.39773285e-01 -2.28553772e-01
1.13292456e+00 5.98820567e-01 1.70223147e-01 -1.02446544e+00
-9.45358157e-01 -4.19028699e-01 -4.36572820e-01 -1.18089644e-02
4.90239590e-01 -6.30769193e-01 -3.70288730e-01 5.29748261e-01
-6.17896199e-01 8.56475681e-02 -6.15043938e-01 2.21670717e-01
-3.27540815e-01 9.68907237e-01 -5.40149629e-01 -7.17402577e-01
-1.05652303e-01 -8.19505751e-01 1.06304622e+00 -2.39429381e-02
9.09131095e-02 -1.26559901e+00 -3.34867328e-01 -2.02457562e-01
2.62534648e-01 4.26389068e-01 1.21744084e+00 -1.10658777e+00
-5.86709559e-01 -6.68554246e-01 -2.13046134e-01 5.80022514e-01
3.32305491e-01 -3.74073796e-02 -8.77865434e-01 -3.45852762e-01
1.30658507e-01 -9.22932178e-02 1.00018013e+00 9.27843153e-02
1.78565073e+00 -3.91549528e-01 -3.94596398e-01 8.85834217e-01
1.16184878e+00 1.38214096e-01 8.60566616e-01 3.03927511e-01
9.38787043e-01 4.09225136e-01 4.82362032e-01 6.11083150e-01
2.54827380e-01 5.67411073e-02 9.84207332e-01 -2.91368246e-01
1.09020926e-01 -2.31737226e-01 2.97260582e-01 7.55007863e-01
1.80518687e-01 -2.40119010e-01 -7.94039190e-01 3.53273392e-01
-1.77168453e+00 -9.31954622e-01 -8.50984678e-02 1.95972788e+00
1.87923029e-01 3.86613369e-01 -1.91268831e-01 4.39120412e-01
6.70440972e-01 3.83813709e-01 -7.41554201e-01 -2.94492781e-01
-3.05260122e-01 -5.43204769e-02 6.09961785e-02 1.26938447e-01
-1.27226007e+00 5.84774494e-01 5.93799067e+00 6.29099429e-01
-9.61542547e-01 -4.94713306e-01 6.36743724e-01 3.30934674e-01
-2.12686434e-01 -2.35787153e-01 -4.83788341e-01 5.52409887e-01
5.36474943e-01 -1.91615000e-01 1.74313486e-01 8.56835365e-01
-3.89365464e-01 8.20817724e-02 -1.17335296e+00 9.62437391e-01
5.15564442e-01 -1.12517369e+00 5.57336628e-01 1.35865554e-01
4.82201517e-01 -2.16794536e-01 1.54374763e-01 4.55605268e-01
-3.83589715e-02 -9.70319331e-01 1.93454012e-01 4.99418437e-01
5.23007214e-01 -7.97230840e-01 9.75556850e-01 2.68603832e-01
-1.06528497e+00 -4.06171709e-01 -6.44420147e-01 -9.20015499e-02
-2.26799682e-01 6.33837998e-01 -7.62102067e-01 9.18536067e-01
6.99712873e-01 1.21557081e+00 -1.03319788e+00 8.26702416e-01
-3.69991392e-01 6.39569461e-01 -2.45024264e-01 2.95849144e-01
3.37379605e-01 -3.54469866e-01 7.82603979e-01 1.04159522e+00
2.86819756e-01 -1.27447605e-01 2.93431759e-01 8.38827550e-01
1.64301068e-01 -1.31513597e-03 -1.06896830e+00 -2.50207305e-01
1.03490032e-01 1.19017303e+00 -8.01416934e-01 -2.18584105e-01
-6.69415712e-01 1.03102374e+00 3.18853825e-01 4.64276463e-01
-7.78125823e-01 -6.57310128e-01 7.02812254e-01 1.03199169e-01
3.28069210e-01 -3.33874822e-02 -1.36756003e-01 -1.27551579e+00
1.74304128e-01 -7.38386691e-01 8.72462511e-01 -4.41085637e-01
-1.64472497e+00 6.85617805e-01 -1.76121503e-01 -1.47992384e+00
-3.64054650e-01 -8.69873047e-01 -1.06539297e+00 4.36371177e-01
-1.44010854e+00 -1.01747990e+00 -6.05367243e-01 7.93820739e-01
4.02647495e-01 -4.78120655e-01 9.10967827e-01 1.66574046e-02
-9.26053524e-01 6.99891806e-01 -2.44189903e-01 5.25241852e-01
4.20729071e-01 -1.41051888e+00 5.20471454e-01 1.37291253e+00
3.35639119e-01 5.73947251e-01 3.72700959e-01 -7.74392962e-01
-1.27448571e+00 -1.38902080e+00 2.01628521e-01 -2.19496652e-01
9.04814243e-01 -3.79853964e-01 -1.33522367e+00 8.60858023e-01
-3.81571762e-02 5.51253319e-01 6.90890312e-01 6.39311746e-02
-4.92353827e-01 -1.49911553e-01 -1.02385879e+00 5.54390073e-01
1.30533326e+00 -4.73886132e-01 -6.80443227e-01 5.38107693e-01
7.09911346e-01 -3.51669043e-01 -4.54840094e-01 5.26042283e-01
-1.42347455e-01 -1.18374002e+00 9.91906881e-01 -8.12929988e-01
5.91444671e-01 -2.90212423e-01 -1.54521316e-01 -1.40901256e+00
-3.58048677e-01 -2.54340738e-01 -6.02451026e-01 8.94599736e-01
1.65899113e-01 -1.24121249e+00 5.63492000e-01 4.76361513e-02
-5.48600018e-01 -1.14844704e+00 -5.93385518e-01 -8.83000910e-01
-4.47488487e-01 -6.17938221e-01 6.78897321e-01 1.10263395e+00
-8.23398400e-03 3.45029309e-02 -5.94977811e-02 7.54458904e-01
7.12034225e-01 2.45429188e-01 5.46232462e-01 -1.37594438e+00
-1.03433505e-01 -3.62614185e-01 -1.34307086e+00 -7.48720646e-01
2.15119094e-01 -1.05046391e+00 -2.55527139e-01 -1.47941756e+00
-2.60758307e-02 -1.30174816e-01 -5.95062137e-01 5.39001107e-01
-2.95759112e-01 2.56568879e-01 -4.88387138e-01 9.40664671e-03
-8.91010165e-01 7.32343137e-01 1.14245450e+00 -8.93558711e-02
1.34120092e-01 -6.55969083e-02 -7.36035645e-01 9.29387450e-01
7.55096555e-01 -9.97903869e-02 -5.68146527e-01 -7.31431097e-02
1.79150462e-01 -5.57158113e-01 5.40246606e-01 -1.06203747e+00
1.96932018e-01 1.75420001e-01 6.48152411e-01 -6.93028569e-01
4.46407385e-02 -8.28036427e-01 -5.13553679e-01 5.28080463e-01
-1.11371256e-01 8.24650154e-02 2.03049228e-01 1.10814714e+00
-3.50449681e-01 -2.17293978e-01 4.58425909e-01 1.83899239e-01
-1.06607068e+00 6.00754082e-01 -1.24124266e-01 1.61945313e-01
1.22218037e+00 -4.60159063e-01 -2.77564764e-01 -4.82728690e-01
-6.79342031e-01 4.74977374e-01 3.84516805e-01 4.43187594e-01
1.08374250e+00 -1.40371811e+00 -3.80145550e-01 1.04205370e+00
6.10169232e-01 2.23674789e-01 2.99671620e-01 7.23646522e-01
-5.89492202e-01 4.76514213e-02 -2.11188659e-01 -9.07349885e-01
-8.57715905e-01 6.08577371e-01 3.72390449e-01 -9.61068049e-02
-1.34618771e+00 7.50118196e-01 5.96027851e-01 -1.53608024e-01
2.60422438e-01 -2.53346652e-01 -2.27066323e-01 -2.89016694e-01
6.03773952e-01 2.22146437e-01 3.57204229e-02 -4.01353240e-01
-3.90243679e-01 2.51149893e-01 -2.20954180e-01 6.73356056e-01
1.27940381e+00 -6.82000443e-02 -3.94916952e-01 4.18983370e-01
1.00533211e+00 1.25745339e-02 -1.08291209e+00 -3.37864697e-01
4.30906471e-03 -6.78407609e-01 -8.79641995e-02 -3.17950428e-01
-1.32424712e+00 8.38193119e-01 3.33933890e-01 6.05407059e-01
1.37763941e+00 -4.02804185e-03 6.27714455e-01 7.92647421e-01
6.85900301e-02 -7.70566761e-01 5.69929481e-01 5.07439315e-01
8.78805816e-01 -1.27680242e+00 -3.23312432e-02 -7.68881857e-01
-5.66580474e-01 1.27589309e+00 9.71244574e-01 -4.40711558e-01
6.10410810e-01 1.61300212e-01 -1.30490392e-01 -7.00147271e-01
-3.88941169e-01 -4.79155369e-02 4.99421448e-01 6.28076315e-01
-1.65652096e-01 -1.71360254e-01 3.75328302e-01 3.94623965e-01
-1.95330814e-01 -8.83616090e-01 5.07679999e-01 7.86728978e-01
-4.57373142e-01 -4.90489274e-01 -1.02805808e-01 9.19158816e-01
-2.17270255e-01 -9.33057442e-02 -3.37285429e-01 7.08805323e-01
3.04510780e-02 7.40765095e-01 5.66068470e-01 -4.66988534e-01
2.60068119e-01 1.60501391e-01 1.72806174e-01 -7.33335793e-01
1.42993271e-01 -3.98931831e-01 -2.45047972e-01 -9.29778934e-01
1.21628933e-01 -3.75532538e-01 -1.42161906e+00 -1.20452009e-02
-3.54810447e-01 1.64929166e-01 2.36606464e-01 9.65551972e-01
4.77746964e-01 9.27282512e-01 9.33149874e-01 -2.43941873e-01
-6.19102776e-01 -6.80570781e-01 -9.44517374e-01 7.93259680e-01
4.72258776e-01 -8.06852758e-01 -8.58915567e-01 -3.60092878e-01] | [6.635772228240967, 5.772824287414551] |
12928a98-32a8-4603-b591-edcdd98bae70 | nndetection-a-self-configuring-method-for | 2106.00817 | null | https://arxiv.org/abs/2106.00817v2 | https://arxiv.org/pdf/2106.00817v2.pdf | nnDetection: A Self-configuring Method for Medical Object Detection | Simultaneous localisation and categorization of objects in medical images, also referred to as medical object detection, is of high clinical relevance because diagnostic decisions often depend on rating of objects rather than e.g. pixels. For this task, the cumbersome and iterative process of method configuration constitutes a major research bottleneck. Recently, nnU-Net has tackled this challenge for the task of image segmentation with great success. Following nnU-Net's agenda, in this work we systematize and automate the configuration process for medical object detection. The resulting self-configuring method, nnDetection, adapts itself without any manual intervention to arbitrary medical detection problems while achieving results en par with or superior to the state-of-the-art. We demonstrate the effectiveness of nnDetection on two public benchmarks, ADAM and LUNA16, and propose 11 further medical object detection tasks on public data sets for comprehensive method evaluation. Code is at https://github.com/MIC-DKFZ/nnDetection . | ['Klaus H. Maier-Hein', 'Fabian Isensee', 'Paul F. Jaeger', 'Michael Baumgartner'] | 2021-06-01 | null | null | null | null | ['medical-object-detection'] | ['computer-vision'] | [ 2.75315970e-01 6.14196844e-02 -1.68360144e-01 -4.47699100e-01
-8.98752928e-01 -5.56549013e-01 3.95677924e-01 5.28281629e-01
-7.71537483e-01 2.91596472e-01 -3.28704983e-01 -5.54375112e-01
6.12095371e-02 -4.64230537e-01 -2.54271656e-01 -6.31257653e-01
-4.48537432e-02 7.86056221e-01 5.00307918e-01 1.03159592e-01
1.91227496e-01 4.65328068e-01 -1.27058411e+00 3.69226933e-01
5.44250131e-01 9.94578719e-01 2.61901379e-01 8.49150002e-01
6.33726493e-02 3.57304782e-01 -4.94383693e-01 -3.86745453e-01
2.21614033e-01 -3.15237403e-01 -9.94179070e-01 2.95772970e-01
4.81758863e-01 -1.53820768e-01 1.03976972e-01 1.13856363e+00
7.16777205e-01 -3.52000222e-02 7.01972246e-01 -6.96894050e-01
-3.27492923e-01 5.54473341e-01 -6.43573463e-01 5.98236918e-01
-8.38585049e-02 2.32072592e-01 9.10550475e-01 -7.88715065e-01
5.92041910e-01 8.88224959e-01 6.29213631e-01 6.91806197e-01
-1.17172110e+00 -2.80645877e-01 -4.09683026e-02 2.30367154e-01
-1.53865814e+00 -5.09135187e-01 2.97026604e-01 -4.84769821e-01
5.68287313e-01 6.78044915e-01 4.49247360e-01 7.37113059e-01
-8.31106119e-03 1.06208205e+00 1.06489587e+00 -3.62550795e-01
3.40788245e-01 1.57200009e-01 3.17758352e-01 7.50976801e-01
3.43428612e-01 -3.18436325e-01 1.68308660e-01 -1.62763909e-01
6.38198912e-01 1.05540074e-01 -2.78455675e-01 -2.60729730e-01
-1.39654708e+00 6.64388955e-01 6.24975264e-01 4.87980694e-01
-2.86406219e-01 1.92631949e-02 4.32090461e-01 1.76908262e-02
4.26205665e-01 6.71536982e-01 -4.19756740e-01 -1.30795330e-01
-9.98924673e-01 5.01484498e-02 5.86432219e-01 6.71042740e-01
2.23232806e-01 -5.34941375e-01 -5.13630569e-01 9.59782362e-01
-4.36838195e-02 1.75295249e-01 6.10592782e-01 -7.34293938e-01
-5.60884066e-02 6.88979208e-01 2.35908432e-03 -7.38445222e-01
-9.03112113e-01 -8.64431262e-01 -1.01324642e+00 9.96073931e-02
5.44265866e-01 -8.41354951e-02 -1.33135223e+00 1.19405866e+00
6.92859054e-01 2.24794701e-01 -2.40777254e-01 1.12414348e+00
1.13012362e+00 2.43476704e-01 7.08585158e-02 -2.83105625e-03
1.64059329e+00 -1.08451831e+00 -3.82282048e-01 -1.28585964e-01
7.51926839e-01 -7.20309496e-01 9.17855203e-01 4.86531049e-01
-1.00272202e+00 -3.34914565e-01 -8.42145801e-01 1.15936629e-01
-2.95888931e-01 3.11840355e-01 7.27058411e-01 5.70518374e-01
-9.84931052e-01 5.75529456e-01 -1.11260593e+00 -4.46592838e-01
8.75917673e-01 5.23740709e-01 -1.33013740e-01 -1.00076655e-02
-6.87237799e-01 6.94227636e-01 4.85721469e-01 2.27802098e-01
-6.55862689e-01 -6.03509188e-01 -5.40543675e-01 -2.11358726e-01
7.90331125e-01 -8.01872551e-01 1.49088430e+00 -7.09849894e-01
-1.19776857e+00 1.14713180e+00 2.37400025e-01 -5.70147991e-01
9.74906981e-01 -3.19498628e-02 -2.13492721e-01 2.55055636e-01
1.12824015e-01 9.18681502e-01 6.21394396e-01 -9.32090759e-01
-8.61402750e-01 -2.65018195e-01 5.49209490e-02 6.19803034e-02
-2.34686479e-01 1.35753602e-01 -1.03632402e+00 -6.63348258e-01
1.77351981e-01 -9.94594872e-01 -7.43214965e-01 1.96256507e-02
-6.62745893e-01 -3.26998025e-01 5.28525233e-01 -3.02704930e-01
1.10150158e+00 -2.07186961e+00 -1.83682412e-01 2.05166206e-01
6.56153023e-01 4.11262035e-01 2.80371439e-02 -2.28522122e-01
1.11864813e-01 9.68857259e-02 -4.15358841e-01 -4.22351390e-01
-1.60510287e-01 5.47021478e-02 3.69269371e-01 6.97836339e-01
1.22582749e-01 1.01919198e+00 -1.00134361e+00 -8.70694160e-01
2.98137397e-01 3.28953922e-01 -6.21379852e-01 5.71912527e-03
-1.60364330e-01 4.83267426e-01 -5.34357786e-01 1.00089157e+00
4.41801220e-01 -7.83420861e-01 9.16230157e-02 -2.26642862e-01
1.19440198e-01 3.00698001e-02 -1.19529521e+00 1.57213032e+00
-1.87857255e-01 3.93865168e-01 1.78142756e-01 -9.27461207e-01
4.47039098e-01 2.13174269e-01 7.41737247e-01 -4.94789541e-01
4.36062157e-01 2.03507110e-01 3.91753465e-01 -5.71016848e-01
3.15740854e-01 1.80592760e-01 3.79474089e-02 2.29801044e-01
-1.25156611e-01 6.56836182e-02 4.59457099e-01 1.19866088e-01
1.36583030e+00 -1.78179532e-01 5.68892360e-01 -3.69359612e-01
3.56044710e-01 2.74912000e-01 4.37602252e-01 1.06147003e+00
-5.28315902e-01 9.00900841e-01 3.93315613e-01 -4.60681558e-01
-7.41975069e-01 -8.04723203e-01 -5.76091170e-01 9.50094342e-01
2.41992727e-01 -2.89178759e-01 -9.61102486e-01 -1.07206452e+00
1.23880226e-02 3.15464437e-01 -9.81947243e-01 1.54003292e-01
-4.83725250e-01 -1.16415346e+00 3.46492440e-01 4.50903147e-01
1.68779522e-01 -1.18201947e+00 -1.07254291e+00 3.38048518e-01
-6.63999245e-02 -1.09795535e+00 -4.81629312e-01 1.51270404e-01
-7.03286767e-01 -1.20602787e+00 -9.40556765e-01 -6.31208718e-01
9.87467527e-01 2.34633580e-01 1.32522500e+00 3.78271163e-01
-1.07265604e+00 1.44256845e-01 -3.96470279e-01 -4.78931934e-01
-4.39679474e-01 3.82771283e-01 -3.66756290e-01 -3.70340534e-02
1.18307576e-01 -1.60120785e-01 -1.16453052e+00 5.60683310e-01
-9.56346512e-01 2.34192282e-01 9.22444701e-01 7.69699097e-01
9.46238160e-01 -1.87389076e-01 4.39823598e-01 -1.36677158e+00
3.26226175e-01 -3.33235532e-01 -5.68293452e-01 1.99127302e-01
-5.92824399e-01 -2.52548099e-01 1.76316887e-01 -4.37733203e-01
-5.31447172e-01 2.89233208e-01 -3.40289295e-01 -3.43273968e-01
-3.62865299e-01 2.84115046e-01 2.17192173e-01 -8.96373242e-02
7.81661272e-01 -1.47271499e-01 -1.08952083e-01 -5.63796461e-01
2.37292871e-01 7.06863642e-01 5.86224198e-01 -1.37024343e-01
3.53406310e-01 6.49642587e-01 2.34480277e-02 -5.79355121e-01
-8.87072802e-01 -9.78619933e-01 -4.92379248e-01 -1.54539555e-01
8.76454413e-01 -5.95941782e-01 -5.95235586e-01 3.29895973e-01
-9.72241819e-01 -4.10358787e-01 -2.87526727e-01 2.08237097e-01
-3.21431994e-01 1.86872661e-01 -7.09091306e-01 -4.64821637e-01
-5.52450180e-01 -1.36247766e+00 1.21042252e+00 1.63805738e-01
-2.34224871e-01 -8.30508530e-01 -1.45630524e-01 5.08084536e-01
4.04576749e-01 4.80987251e-01 4.69072342e-01 -8.02358866e-01
-4.65388894e-01 -4.58843678e-01 -3.59443218e-01 1.74405187e-01
3.32592279e-01 -1.61767110e-01 -8.64546418e-01 -3.35305274e-01
-2.22055092e-01 -1.86667778e-02 1.08432281e+00 5.33348143e-01
1.53084266e+00 -8.02649409e-02 -6.42152846e-01 7.26843178e-01
1.28308249e+00 -3.07219196e-02 1.79715931e-01 3.25833529e-01
7.08219111e-01 3.52920830e-01 7.30525076e-01 3.64883602e-01
3.02348107e-01 8.13872039e-01 5.24580657e-01 -6.34585023e-01
-1.97391585e-01 5.66370249e-01 -3.38859200e-01 3.23460907e-01
-1.63839366e-02 -1.97939858e-01 -1.32827199e+00 7.46025980e-01
-1.87979102e+00 -4.92016673e-01 -2.08313316e-01 1.90967619e+00
9.40716386e-01 2.83734828e-01 2.72080302e-01 -1.68777686e-02
7.82284856e-01 -2.11791813e-01 -5.86472571e-01 4.65057455e-02
2.82267094e-01 7.35731274e-02 6.93511605e-01 2.31504023e-01
-1.52098167e+00 7.82839894e-01 5.50579929e+00 1.10405147e+00
-1.23905361e+00 4.36901361e-01 1.23532355e+00 -1.62088171e-01
3.65453571e-01 -4.24520284e-01 -7.98687696e-01 4.86853898e-01
6.31172895e-01 3.76446038e-01 3.72036137e-02 7.33286500e-01
1.33272499e-01 -3.06157082e-01 -1.10926414e+00 9.59150732e-01
-1.04709111e-01 -1.41776419e+00 -2.21987560e-01 -3.75627056e-02
7.06246972e-01 2.50658363e-01 1.08813196e-01 3.64827961e-02
1.42894968e-01 -9.37831879e-01 4.93727982e-01 1.70746878e-01
8.65436256e-01 -4.91038948e-01 9.88176942e-01 1.53786525e-01
-8.71525407e-01 8.70607048e-02 -1.57751188e-01 4.81209338e-01
6.78685233e-02 7.56197274e-01 -1.23858500e+00 2.97921181e-01
8.87608469e-01 4.89887238e-01 -9.38325167e-01 1.56169772e+00
6.49520978e-02 7.72749186e-01 -3.37000996e-01 4.75372709e-02
3.55591625e-01 1.69746131e-01 5.48316896e-01 1.55717289e+00
4.67070527e-02 2.37332344e-01 3.14026028e-01 6.49201989e-01
-9.82925519e-02 3.38372290e-01 8.53838399e-03 1.86361715e-01
6.73894063e-02 1.74754822e+00 -1.50279975e+00 -5.06388247e-01
-3.28481663e-03 8.38133276e-01 5.72125539e-02 -1.08587928e-01
-8.76783371e-01 -1.79108009e-01 3.37074935e-01 2.24412739e-01
3.75956833e-01 5.15271798e-02 -4.38087016e-01 -7.58837044e-01
-5.71316741e-02 -7.61555016e-01 6.42009318e-01 -1.84831038e-01
-1.04553246e+00 8.19587409e-01 -1.17833741e-01 -1.21548951e+00
1.99517608e-03 -6.99477077e-01 -4.27018672e-01 4.24335480e-01
-1.41939902e+00 -9.37998712e-01 -4.99086022e-01 3.46899301e-01
5.18277168e-01 2.00734243e-01 5.34576714e-01 5.60667634e-01
-8.09139967e-01 7.19081998e-01 4.81404066e-02 2.37016678e-01
7.35722482e-01 -1.55946052e+00 6.07586503e-01 5.22800624e-01
1.11054063e-01 3.10984969e-01 6.48218811e-01 -3.93543839e-01
-1.10613620e+00 -1.35406852e+00 4.63444293e-01 -6.40774369e-01
5.75803101e-01 -3.53549957e-01 -7.02226758e-01 1.91602573e-01
-3.63685563e-02 4.74638134e-01 5.41795909e-01 -1.01727292e-01
2.00363904e-01 -3.39258183e-03 -1.22404277e+00 5.68248391e-01
1.09630311e+00 1.62226379e-01 -1.77449390e-01 7.76173055e-01
5.67496955e-01 -8.76939178e-01 -7.65357673e-01 6.49667323e-01
2.86826909e-01 -8.43617678e-01 1.05023468e+00 -3.63800824e-01
1.87470600e-01 -3.78487527e-01 1.83490813e-01 -9.49316025e-01
-3.03545862e-01 -5.57407260e-01 9.93551835e-02 8.69371355e-01
6.35265946e-01 -5.62333524e-01 9.03168797e-01 3.02054137e-01
-1.68968171e-01 -1.20165610e+00 -9.78279352e-01 -5.32626390e-01
-2.08703518e-01 -3.17755401e-01 3.31206471e-01 7.74323106e-01
-4.65078354e-01 1.27136782e-01 1.09648310e-01 1.32903352e-01
6.02180898e-01 2.03074202e-01 5.89444876e-01 -1.08030939e+00
-5.51867306e-01 -8.41108322e-01 -6.01816595e-01 -7.17794836e-01
-4.04388279e-01 -1.03819382e+00 2.53532588e-01 -1.70849633e+00
3.92597407e-01 -6.72393799e-01 -6.62140012e-01 6.65663958e-01
-4.82902199e-01 8.42260480e-01 1.66257367e-01 4.23457414e-01
-1.13088799e+00 -8.66752043e-02 1.33994496e+00 -3.77451241e-01
-2.00249195e-01 3.31407338e-01 -6.67898238e-01 7.28439927e-01
9.30417120e-01 -5.76689124e-01 -6.22655340e-02 -1.97969228e-01
-6.99733645e-02 -2.45220754e-02 5.55650115e-01 -1.24697053e+00
1.54785961e-01 6.31356239e-02 2.11850792e-01 -5.66807389e-01
1.67906284e-01 -6.82121336e-01 -1.51373968e-01 7.33276427e-01
-3.54070306e-01 2.44945381e-02 1.93000615e-01 3.84503186e-01
-3.67803872e-02 -2.65740395e-01 1.03064561e+00 -2.38118157e-01
-6.67576075e-01 4.00008708e-01 -1.03233255e-01 7.96570256e-02
1.12795043e+00 -2.10660905e-01 -2.66200662e-01 1.47087157e-01
-1.01292217e+00 4.50623959e-01 3.38424385e-01 3.25372249e-01
4.26363647e-01 -8.42022538e-01 -8.21691394e-01 -1.99344337e-01
2.72506267e-01 3.72395009e-01 3.07338685e-01 1.37494683e+00
-6.39037669e-01 2.50908107e-01 1.63434207e-01 -1.09353328e+00
-1.38869476e+00 4.95428920e-01 5.90955138e-01 -3.41837347e-01
-8.66142571e-01 8.91142964e-01 3.12297046e-01 -2.66552567e-01
2.02312142e-01 -5.11486650e-01 -3.32390934e-01 1.49707180e-02
4.88840908e-01 2.88894504e-01 5.05106509e-01 -2.50155181e-01
-5.59155226e-01 2.77315289e-01 -4.67933208e-01 1.71285957e-01
1.12900686e+00 7.30466098e-02 -7.07068294e-02 1.42499685e-01
9.91466522e-01 -3.77920479e-01 -1.04755175e+00 -3.34121108e-01
1.49785891e-01 -2.55554944e-01 2.44882673e-01 -1.06377172e+00
-1.26653790e+00 5.93941927e-01 1.01943803e+00 2.80560791e-01
9.76548731e-01 2.91238308e-01 6.28881812e-01 4.18033659e-01
1.77518025e-01 -1.06163538e+00 -2.18804389e-01 7.45080188e-02
5.81570745e-01 -1.70742416e+00 1.05781645e-01 -4.63567287e-01
-4.16605741e-01 8.05425286e-01 6.03816748e-01 1.34344995e-01
6.68340385e-01 2.02748716e-01 3.40839654e-01 -2.93495536e-01
-5.52860975e-01 -3.56796473e-01 6.30446196e-01 1.36935323e-01
3.78589153e-01 3.81137580e-01 -3.34988654e-01 3.41972172e-01
7.66596496e-02 -9.59733725e-02 3.55634749e-01 1.00107419e+00
-4.84853685e-01 -8.32717776e-01 -4.31345433e-01 8.25085282e-01
-7.92181075e-01 -1.36816025e-01 -2.28571311e-01 6.90231383e-01
3.18264663e-01 8.58372390e-01 8.22521076e-02 4.88788709e-02
2.59186745e-01 -5.78031301e-01 2.48711243e-01 -7.55954862e-01
-7.65509963e-01 2.46824756e-01 -3.76815312e-02 -6.60997689e-01
-3.30429226e-01 -7.85410941e-01 -1.34702027e+00 1.52230918e-01
-3.15063685e-01 1.15808144e-01 7.14316487e-01 7.31207609e-01
4.14757282e-01 7.26169586e-01 4.19447899e-01 -8.02326918e-01
-5.82121730e-01 -8.10102999e-01 -1.78727716e-01 5.16440988e-01
3.54318112e-01 -4.92533088e-01 -1.28813935e-02 -7.39347860e-02] | [14.972787857055664, -2.42423677444458] |
8738a3a3-72b2-45bf-9cd2-0e8e61eb46c9 | lifted-symmetry-detection-and-breaking-for | null | null | http://papers.nips.cc/paper/5691-lifted-symmetry-detection-and-breaking-for-map-inference | http://papers.nips.cc/paper/5691-lifted-symmetry-detection-and-breaking-for-map-inference.pdf | Lifted Symmetry Detection and Breaking for MAP Inference | Symmetry breaking is a technique for speeding up propositional satisfiability testing by adding constraints to the theory that restrict the search space while preserving satisfiability. In this work, we extend symmetry breaking to the problem of model finding in weighted and unweighted relational theories, a class of problems that includes MAP inference in Markov Logic and similar statistical-relational languages. We introduce term symmetries, which are induced by an evidence set and extend to symmetries over a relational theory. We provide the important special case of term equivalent symmetries, showing that such symmetries can be found in low-degree polynomial time. We show how to break an exponential number of these symmetries with added constraints whose number is linear in the size of the domain. We demonstrate the effectiveness of these techniques through experiments in two relational domains. We also discuss the connections between relational symmetry breaking and work on lifted inference in statistical-relational reasoning. | ['Parag Singla', 'Henry Kautz', 'Timothy Kopp'] | 2015-12-01 | null | null | null | neurips-2015-12 | ['symmetry-detection'] | ['computer-vision'] | [ 6.89666033e-01 7.41934299e-01 -7.92965829e-01 -4.63966638e-01
-6.28506839e-01 -6.27989948e-01 3.81647527e-01 9.73285958e-02
2.04234004e-01 6.41842186e-01 1.71416372e-01 -7.69015968e-01
-9.22060907e-01 -1.26924443e+00 -1.17629015e+00 -2.54649967e-01
-5.82353652e-01 1.02351880e+00 7.39054024e-01 -3.36619139e-01
3.25846255e-01 5.95454514e-01 -1.64068139e+00 7.60269344e-01
3.50536823e-01 4.62530851e-01 -6.05557084e-01 5.67465782e-01
9.18945447e-02 8.72390389e-01 1.68041363e-01 -3.86534691e-01
5.17133057e-01 -1.51065812e-01 -1.46687710e+00 5.07731270e-03
7.77593315e-01 -9.36815739e-02 -8.21256995e-01 1.13562870e+00
-4.26596045e-01 1.15691155e-01 1.05696209e-01 -2.10117364e+00
-3.74311775e-01 1.15671551e+00 -3.87608647e-01 1.31995246e-01
6.72428846e-01 -5.19383192e-01 1.51104629e+00 5.96220903e-02
8.84809494e-01 1.65624964e+00 4.94308829e-01 5.45700967e-01
-1.75405097e+00 -6.33944452e-01 8.28930736e-02 6.86798513e-01
-1.45561731e+00 -5.61113834e-01 2.85869241e-01 -2.02698439e-01
1.39828372e+00 7.64105618e-01 8.84147733e-02 5.00187576e-01
3.76180321e-01 1.68169528e-01 1.13453686e+00 -6.19406044e-01
4.33351174e-02 -1.00180008e-01 5.01369417e-01 1.07697463e+00
7.90797710e-01 6.15385100e-02 -7.23729193e-01 -6.56592727e-01
3.61021489e-01 -2.53037304e-01 1.45676835e-02 -6.73642337e-01
-1.10461771e+00 9.08003569e-01 -5.42178303e-02 1.27082273e-01
4.60204482e-01 3.93294424e-01 3.71389270e-01 7.35454619e-01
-5.91218024e-02 4.75338966e-01 -4.26280111e-01 2.59758651e-01
-4.45959657e-01 5.70069492e-01 1.16549075e+00 1.28668356e+00
6.34068012e-01 -5.28629899e-01 -4.64542322e-02 8.87001678e-02
2.03088909e-01 6.80935740e-01 -2.98089862e-01 -1.27147114e+00
6.57626092e-01 6.98926508e-01 2.52668560e-02 -8.86283040e-01
-4.09630924e-01 2.33537659e-01 -3.25954229e-01 -1.47898108e-01
2.80266166e-01 5.92084706e-01 -4.85761762e-01 2.03092957e+00
3.33742142e-01 4.73571643e-02 1.75801903e-01 1.53407156e-01
2.83509493e-01 5.13306975e-01 -5.90003550e-01 -3.64188343e-01
1.48457074e+00 -5.73756158e-01 -6.18723154e-01 -1.15864731e-01
9.72691834e-01 -1.70433477e-01 2.72206128e-01 5.50004959e-01
-1.34406996e+00 1.25374585e-01 -1.08423519e+00 -3.90371948e-01
-1.53042778e-01 -7.19877422e-01 1.29562032e+00 4.16122317e-01
-1.06927216e+00 5.25531054e-01 -8.64672720e-01 -2.01841161e-01
-8.82026404e-02 8.67726862e-01 -1.01276028e+00 -5.91740131e-01
-1.31057596e+00 9.97614384e-01 5.08317590e-01 -3.61710966e-01
-6.68048620e-01 -6.31407857e-01 -1.23335147e+00 1.27155691e-01
1.03339314e+00 -5.03504574e-01 1.36668348e+00 7.49940202e-02
-8.55612099e-01 8.54617536e-01 -6.33445680e-01 -6.42056644e-01
-9.09055024e-02 5.99831343e-01 -5.57576597e-01 1.03824638e-01
5.23743592e-02 4.43210565e-02 1.14143915e-01 -8.35945070e-01
-6.03005767e-01 -6.87751412e-01 8.46858144e-01 -4.09160227e-01
2.88969308e-01 2.19727501e-01 -6.93513379e-02 3.51491392e-01
9.41205502e-01 -1.40362179e+00 -1.76128000e-01 -3.39992821e-01
-5.14818072e-01 -4.76779908e-01 6.22981727e-01 -1.82279736e-01
1.16037190e+00 -1.89008343e+00 4.88320678e-01 8.80651295e-01
3.03574026e-01 -4.52352792e-01 -1.54720098e-01 7.06552029e-01
-3.49479645e-01 4.03856844e-01 1.26922857e-02 2.13050291e-01
5.12447119e-01 7.24236250e-01 -6.07637584e-01 5.67620754e-01
1.13778949e-01 9.77353990e-01 -5.96858442e-01 -5.78893840e-01
-1.93419367e-01 -3.10448050e-01 -8.30081701e-01 -5.07049799e-01
-5.32354832e-01 -3.45923871e-01 -3.42579633e-01 5.33708096e-01
9.26804900e-01 -1.64585620e-01 6.48356795e-01 8.79989788e-02
3.69147547e-02 8.17535222e-01 -1.49085224e+00 1.40865552e+00
-2.18154013e-01 2.61092722e-01 5.57606779e-02 -7.43607938e-01
4.15764540e-01 1.01406828e-01 2.32451648e-01 -5.84325373e-01
-1.39263481e-01 2.21977811e-02 3.59028250e-01 -5.02262592e-01
3.58233720e-01 -5.76691747e-01 -3.65468472e-01 7.95077622e-01
-2.00171262e-01 -3.26188684e-01 5.78977942e-01 3.87512892e-01
1.46346939e+00 -2.32984781e-01 1.61418512e-01 -3.84522527e-01
4.66773212e-01 4.80994321e-02 8.08161378e-01 8.76814961e-01
5.01338243e-01 -1.27285019e-01 1.18101764e+00 -3.84078473e-01
-9.69004750e-01 -8.88385475e-01 -4.01567012e-01 9.75311637e-01
1.17236033e-01 -1.15616393e+00 -3.36543351e-01 -4.54455912e-01
3.13357919e-01 7.60599315e-01 -7.41948247e-01 -5.80775678e-01
-3.77103835e-01 -4.10061479e-01 7.10162759e-01 5.58095753e-01
1.43732205e-01 -4.15187061e-01 -1.49552569e-01 -2.42171407e-01
-1.24174222e-01 -1.37009275e+00 2.02482045e-01 3.81739855e-01
-1.04594338e+00 -1.46904731e+00 1.07084513e+00 -3.79274487e-01
7.41484761e-01 4.06304598e-01 8.23144019e-01 3.03878427e-01
6.61951350e-03 1.20014884e-01 -2.84315813e-02 -3.22871655e-01
-6.64876640e-01 -4.49417941e-02 2.01857686e-01 -6.27459049e-01
5.80251038e-01 -4.88952816e-01 5.69783330e-01 4.51254129e-01
-1.21403658e+00 1.04775846e-01 1.05100609e-01 6.04553759e-01
5.71951509e-01 8.55336070e-01 -5.50839722e-01 -1.02962112e+00
1.37105346e-01 -3.16957563e-01 -1.23810744e+00 7.19304860e-01
-2.93560684e-01 8.60135436e-01 7.73886815e-02 -5.98566830e-02
-5.56070149e-01 -1.31727427e-01 6.95177376e-01 4.54615690e-02
3.31426173e-01 7.51231909e-01 -3.41655374e-01 -4.71664131e-01
4.15157109e-01 -1.69216216e-01 -6.66959137e-02 2.09441647e-01
5.82417138e-02 2.16564998e-01 3.14529151e-01 -1.20667148e+00
1.17007685e+00 6.59148991e-01 1.35488582e+00 -2.35694394e-01
-9.07627225e-01 -1.23956509e-01 -3.42514992e-01 6.24208450e-01
2.87450373e-01 -4.58524138e-01 -1.21984661e+00 -3.19000959e-01
-9.94440973e-01 -2.02747732e-01 -2.76121557e-01 4.39596355e-01
-7.88763165e-01 5.02589285e-01 -4.10490155e-01 -8.47254813e-01
4.07922506e-01 -1.13970912e+00 9.12546575e-01 -5.37517607e-01
-4.79566336e-01 -6.83630466e-01 3.61228526e-01 4.68192250e-01
-6.06077239e-02 -7.48359561e-02 1.63839030e+00 -9.10953283e-01
-7.99249172e-01 -2.88868308e-01 -2.19152197e-01 -1.49386823e-01
-2.05805361e-01 1.71057105e-01 -4.99541461e-01 -1.46240398e-01
-6.93941340e-02 -1.71029985e-01 6.70999229e-01 -3.58448476e-02
1.08800197e+00 -3.54345173e-01 -6.28579557e-01 3.16348195e-01
1.27556896e+00 4.76201288e-02 8.19357514e-01 3.38198900e-01
3.14761072e-01 5.75258732e-01 4.41924542e-01 7.01832399e-02
2.81341165e-01 7.90452957e-01 2.68861443e-01 3.73345912e-01
4.78725642e-01 -2.45669499e-01 1.04922540e-02 3.13409060e-01
-4.52353477e-01 7.90975019e-02 -9.07490909e-01 2.48374507e-01
-2.05129981e+00 -1.41495073e+00 -4.14311051e-01 2.42031240e+00
1.01331317e+00 3.38513464e-01 -1.27514124e-01 4.96815801e-01
5.07008433e-01 -3.28438669e-01 1.00270666e-01 -1.04001820e+00
-7.53080696e-02 5.95816791e-01 1.02615893e+00 1.28061092e+00
-8.03928912e-01 7.78013945e-01 7.12652016e+00 3.46279234e-01
-2.92833328e-01 1.60170674e-01 -3.59151036e-01 -1.67306468e-01
-8.28282297e-01 6.61815584e-01 -8.22009921e-01 -2.76781946e-01
1.11858594e+00 -2.50765920e-01 7.89203763e-01 7.33075082e-01
-3.36449713e-01 -2.26866007e-01 -1.75328386e+00 2.01558620e-01
4.59339656e-02 -1.14958072e+00 -4.97443043e-02 4.62588698e-01
8.52604389e-01 -2.95180678e-01 -1.40863627e-01 2.75926203e-01
5.82726240e-01 -1.20238316e+00 3.84447455e-01 2.57720143e-01
6.54010296e-01 -9.22635972e-01 7.11903691e-01 1.95750326e-01
-9.60813522e-01 -3.18960756e-01 -6.43061936e-01 -3.78598750e-01
-1.44794598e-01 1.16732053e-01 -6.79076433e-01 7.46887624e-01
3.12253386e-01 1.58247590e-01 -2.60933876e-01 8.82721364e-01
-2.60449499e-01 1.40663013e-01 -6.01234376e-01 4.12820399e-01
4.80810218e-02 1.38640832e-02 4.62131381e-01 8.35716665e-01
1.81978032e-01 4.41177726e-01 -5.98861650e-03 6.91851854e-01
2.42930315e-02 -7.55965650e-01 -9.26622868e-01 -1.10073037e-04
4.65809673e-01 6.45588994e-01 -4.71051753e-01 -1.84558630e-01
-4.87885624e-01 2.44875580e-01 1.14435874e-01 -1.31681621e-01
-7.66018271e-01 -2.02966839e-01 6.02389455e-01 4.20157343e-01
3.45363975e-01 -1.23186558e-01 1.29049435e-01 -1.10224533e+00
1.10436000e-01 -1.09179628e+00 6.14088953e-01 -6.04084074e-01
-8.84316623e-01 7.23797232e-02 9.07984436e-01 -5.40000975e-01
-2.33803004e-01 -8.86555016e-01 -4.03811149e-02 5.43338060e-01
-1.28577065e+00 -9.95119393e-01 8.70008543e-02 6.37095392e-01
-3.04709733e-01 2.19857767e-01 1.17164838e+00 -2.06918225e-01
-2.48782337e-01 4.70349222e-01 -3.74473602e-01 -4.60194439e-01
2.94498414e-01 -1.28082442e+00 3.74658734e-01 9.26554143e-01
2.24449053e-01 1.13895452e+00 9.65347290e-01 -6.46001339e-01
-2.07053256e+00 -7.58716226e-01 1.48468828e+00 -4.54106182e-01
1.10198426e+00 -5.63458562e-01 -7.92697012e-01 1.57131767e+00
-1.44312724e-01 9.26462859e-02 7.09180653e-01 7.07605660e-01
-1.12452316e+00 -2.23160148e-01 -1.17998087e+00 4.98886198e-01
1.58221591e+00 -9.70924377e-01 -9.34333026e-01 6.69810414e-01
8.04915965e-01 -5.90116084e-01 -9.17897403e-01 5.87111533e-01
5.20754755e-01 -5.69749713e-01 7.23717153e-01 -1.41883898e+00
1.98687300e-01 -6.43264234e-01 -5.98865390e-01 -5.02021074e-01
-7.48340666e-01 -7.70050406e-01 -3.07132900e-01 6.07283831e-01
6.55959070e-01 -7.17584670e-01 6.21241689e-01 1.08285666e+00
9.21832100e-02 -3.70098919e-01 -1.35540736e+00 -9.50009823e-01
-8.03324655e-02 -6.98705792e-01 9.21764255e-01 8.92619073e-01
9.73420739e-01 1.98400438e-01 -1.04629956e-01 5.29123008e-01
4.82607156e-01 4.14496481e-01 6.53370380e-01 -1.60218465e+00
-5.73845446e-01 7.19653349e-03 -1.00134623e+00 -2.18209520e-01
7.86242187e-01 -1.25422740e+00 -2.67272264e-01 -1.26552606e+00
8.21563363e-01 -3.00675988e-01 -1.68636188e-01 9.93255556e-01
6.45416200e-01 -6.98005781e-02 -1.05951585e-01 -2.73905724e-01
-7.78557420e-01 -1.75437614e-01 1.05821061e+00 -3.41286898e-01
2.77141541e-01 3.76856513e-02 -9.18855906e-01 5.08752882e-01
5.82539082e-01 -8.09189618e-01 -6.72979593e-01 -1.26471326e-01
1.25353658e+00 2.93206364e-01 5.56538165e-01 -7.69360483e-01
4.56419110e-01 -6.99896932e-01 -4.94982570e-01 -3.02319556e-01
1.84687823e-01 -9.69008863e-01 6.58654094e-01 6.47299707e-01
-5.76712847e-01 3.23880911e-01 4.28500146e-01 2.79806405e-01
4.61249091e-02 -5.68033993e-01 3.19015980e-01 5.80292977e-02
-3.97599667e-01 3.98790501e-02 -1.22306220e-01 -5.09373248e-02
1.14725161e+00 -1.01245393e-03 -6.54991984e-01 -1.57300740e-01
-7.77777016e-01 2.91064680e-01 5.81076622e-01 5.98523393e-02
3.85799021e-01 -1.23432076e+00 -4.24082994e-01 1.49829417e-01
5.00054181e-01 -1.31565690e-01 1.71416089e-01 1.23231184e+00
-2.29191706e-01 1.03048193e+00 -5.57419099e-02 -1.27933025e-01
-1.77155054e+00 9.03014958e-01 2.49771401e-01 -3.55934083e-01
-2.43373573e-01 6.83968544e-01 -1.46870598e-01 -4.01796848e-01
1.45714507e-01 -9.78907526e-01 5.16773641e-01 -5.05103230e-01
7.95337141e-01 4.25178736e-01 2.53857076e-01 -3.90423536e-01
-7.28025913e-01 5.56184232e-01 -2.04279825e-01 -2.20090225e-01
1.33732224e+00 3.46468747e-01 -1.17301404e+00 2.19036236e-01
7.94063926e-01 3.51578861e-01 -2.02109650e-01 -5.02072394e-01
1.68126017e-01 -4.22089815e-01 -1.42176360e-01 -2.31632441e-01
-5.10599554e-01 3.03351581e-01 -2.82867849e-01 5.14167607e-01
7.35650897e-01 5.95886409e-01 2.36150801e-01 1.13530982e+00
8.86872888e-01 -7.63642132e-01 -7.93388605e-01 5.93229234e-01
6.74386799e-01 -7.73182213e-01 5.04617512e-01 -1.01840985e+00
-6.07723482e-02 1.03705537e+00 3.53080899e-01 1.60034612e-01
2.74923980e-01 7.81136513e-01 -1.00393784e+00 -3.51215720e-01
-1.28024065e+00 7.68587366e-02 1.65350139e-01 3.97124082e-01
-6.08813241e-02 2.27366626e-01 -2.38305211e-01 5.66310525e-01
-5.31970918e-01 5.34658208e-02 6.70466721e-01 1.14330029e+00
-4.35961217e-01 -1.61678147e+00 -4.82066214e-01 2.45186925e-01
-4.55422014e-01 -9.56244320e-02 -5.43663263e-01 1.13632977e+00
1.58931255e-01 1.03963089e+00 -2.12232135e-02 -4.96534586e-01
2.01505169e-01 1.58981666e-01 1.30226827e+00 -8.39264929e-01
3.24968272e-03 -6.59287274e-01 6.20746970e-01 -8.39615047e-01
-4.25870538e-01 -8.08461726e-01 -1.31065321e+00 -1.08270621e+00
-5.33318102e-01 4.08190489e-01 1.93838209e-01 1.16146791e+00
-1.08093964e-02 2.66187519e-01 2.22743213e-01 1.79913253e-01
-8.30958843e-01 -5.89149177e-01 -7.37584293e-01 -1.46011740e-01
2.80632138e-01 -7.23257780e-01 -5.08002043e-01 -3.34857672e-01] | [8.633851051330566, 6.737736701965332] |
3904a971-4dec-406b-b132-b32ae3061e15 | generic-event-boundary-detection-in-video | 2301.04288 | null | https://arxiv.org/abs/2301.04288v1 | https://arxiv.org/pdf/2301.04288v1.pdf | Generic Event Boundary Detection in Video with Pyramid Features | Generic event boundary detection (GEBD) aims to split video into chunks at a broad and diverse set of actions as humans naturally perceive event boundaries. In this study, we present an approach that considers the correlation between neighbor frames with pyramid feature maps in both spatial and temporal dimensions to construct a framework for localizing generic events in video. The features at multiple spatial dimensions of a pre-trained ResNet-50 are exploited with different views in the temporal dimension to form a temporal pyramid feature map. Based on that, the similarity between neighbor frames is calculated and projected to build a temporal pyramid similarity feature vector. A decoder with 1D convolution operations is used to decode these similarities to a new representation that incorporates their temporal relationship for later boundary score estimation. Extensive experiments conducted on the GEBD benchmark dataset show the effectiveness of our system and its variations, in which we outperformed the state-of-the-art approaches. Additional experiments on TAPOS dataset, which contains long-form videos with Olympic sport actions, demonstrated the effectiveness of our study compared to others. | ['Soo-Hyung Kim', 'Guee-Sang Lee', 'Hyung-Jeong Yang', 'Van Thong Huynh'] | 2023-01-11 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [ 1.14209376e-01 -4.84517306e-01 -9.63794813e-02 -2.35173523e-01
-4.51656729e-01 -1.71446323e-01 4.70598400e-01 -2.94984411e-02
-5.11222959e-01 2.82396227e-01 8.50183070e-01 5.10133207e-01
2.42392477e-02 -7.10768700e-01 -6.88084841e-01 -3.58561158e-01
-5.23781657e-01 -1.81445837e-01 1.03456843e+00 -1.82409868e-01
4.02647674e-01 3.09136301e-01 -1.78928697e+00 8.78789008e-01
3.11310023e-01 1.28093481e+00 3.36702168e-01 5.92672050e-01
2.23850399e-01 1.13458776e+00 -5.04713774e-01 -1.60380796e-01
5.27021348e-01 -5.95510185e-01 -6.13631248e-01 2.14362234e-01
4.74561334e-01 -5.61907351e-01 -8.95917892e-01 7.59947717e-01
5.19349813e-01 5.97008884e-01 2.90547907e-01 -9.29772079e-01
-4.38351065e-01 4.51917946e-01 -6.20464325e-01 9.12265837e-01
7.82678604e-01 1.12663638e-02 9.59894300e-01 -1.11192918e+00
9.60213006e-01 1.27131343e+00 7.67807662e-01 1.55460730e-01
-7.48678088e-01 -5.75199842e-01 2.92575777e-01 8.99274528e-01
-1.35550261e+00 -1.35384291e-01 6.82452261e-01 -4.73758966e-01
1.17521620e+00 -2.08268519e-02 1.05164635e+00 9.68201816e-01
5.36862791e-01 8.50998402e-01 5.96628726e-01 -1.40771315e-01
5.45879416e-02 -5.81347346e-01 1.05351582e-02 6.88081145e-01
-2.60183871e-01 1.55574158e-01 -1.10201907e+00 2.10472703e-01
9.18394268e-01 3.26902688e-01 -2.72029757e-01 -2.99887925e-01
-1.52647638e+00 5.51383734e-01 4.49234217e-01 3.52051705e-01
-7.69206882e-01 -4.17969488e-02 7.27406383e-01 2.62636930e-01
3.83820444e-01 -5.08599356e-02 -9.31005105e-02 -2.74698704e-01
-1.06788361e+00 2.51695395e-01 5.60277581e-01 6.96877301e-01
6.21272147e-01 -1.78439468e-01 -7.00165212e-01 6.34945154e-01
-4.02215049e-02 -5.63926399e-02 6.80766523e-01 -1.13768339e+00
5.38947999e-01 6.49539113e-01 -4.00747955e-02 -1.32249153e+00
-5.16999781e-01 6.21300703e-03 -5.79834819e-01 5.15209846e-02
3.09082806e-01 -7.96271302e-03 -7.29872227e-01 1.51222253e+00
3.80820274e-01 8.86644661e-01 1.25362324e-02 1.24508476e+00
9.28923249e-01 9.04470444e-01 7.85948634e-02 -6.53355569e-02
1.50295949e+00 -1.05531144e+00 -5.92102826e-01 -4.55634408e-02
2.72666454e-01 -6.96724772e-01 7.58599281e-01 3.29106152e-01
-1.24544322e+00 -1.10596085e+00 -1.14145672e+00 -9.55196545e-02
-3.32140177e-02 1.36232004e-01 3.17926258e-01 1.28785431e-01
-1.04232728e+00 8.24540257e-01 -9.57902014e-01 -5.70545614e-01
1.56995818e-01 1.09824575e-01 -4.79356706e-01 1.72651231e-01
-1.42010617e+00 7.42844582e-01 7.75931537e-01 -5.09383157e-02
-1.02571869e+00 -5.51409543e-01 -1.06722438e+00 8.56362358e-02
3.57466429e-01 -4.74858552e-01 1.11885488e+00 -9.78751957e-01
-1.37641597e+00 6.70012712e-01 -2.91448593e-01 -8.24494183e-01
2.64479756e-01 -4.44951147e-01 -5.75339079e-01 6.69763565e-01
2.31956333e-01 8.26270163e-01 7.15502679e-01 -5.56818366e-01
-1.26968694e+00 -1.78328395e-01 2.45748371e-01 7.53671110e-01
-2.02381656e-01 3.42884839e-01 -8.05889726e-01 -8.83301079e-01
2.63901412e-01 -6.72595382e-01 -6.29837364e-02 -3.11407685e-01
-3.14858439e-03 -2.44824752e-01 8.42987895e-01 -9.33154166e-01
1.60752213e+00 -2.29755235e+00 5.15650690e-01 -6.60608336e-02
1.81997612e-01 5.97181097e-02 8.99354927e-03 4.81703997e-01
-1.72389954e-01 -3.81523669e-01 2.24496275e-01 -1.02176771e-01
-1.55558661e-01 -6.53574318e-02 -2.46641397e-01 4.88609135e-01
1.62359521e-01 6.66750908e-01 -9.08237338e-01 -6.82225645e-01
4.47411686e-01 3.98538232e-01 -5.60658753e-01 2.71529824e-01
1.30141065e-01 3.49313587e-01 -3.47442478e-01 3.58284891e-01
3.57900202e-01 -9.20213992e-04 -1.26370806e-02 -4.03524488e-01
-2.85455704e-01 1.75463572e-01 -1.35492396e+00 2.03227615e+00
5.75719178e-02 7.81834066e-01 -5.51087081e-01 -1.00162113e+00
8.15194130e-01 3.89305115e-01 7.15049088e-01 -8.17364335e-01
-1.90399345e-02 -1.15957268e-01 -2.73883075e-01 -8.13259363e-01
7.22419858e-01 1.58992141e-01 -2.04025328e-01 3.38062495e-02
3.28037351e-01 4.41643327e-01 6.68885112e-01 6.87042922e-02
1.13991952e+00 5.26243448e-01 3.74671042e-01 -7.67629743e-02
7.00361669e-01 -7.21516758e-02 9.66273129e-01 7.14380264e-01
-4.64472175e-01 9.06084538e-01 5.00599623e-01 -9.44755435e-01
-8.99744928e-01 -1.17524028e+00 2.92860091e-01 1.18466198e+00
6.00764930e-01 -7.29931712e-01 -8.11373949e-01 -6.15538120e-01
-3.44688445e-01 3.55421305e-01 -6.74417913e-01 -2.15063587e-01
-9.78053689e-01 -3.08878005e-01 3.61268580e-01 7.59652197e-01
1.01683354e+00 -1.19702244e+00 -9.70331490e-01 4.44372833e-01
-4.67145175e-01 -1.27402389e+00 -8.13432157e-01 -2.47628957e-01
-6.46281004e-01 -1.09694743e+00 -8.35442781e-01 -1.04186928e+00
7.69515261e-02 4.27905083e-01 9.44017529e-01 -3.81682098e-01
-2.46526614e-01 2.61289656e-01 -6.85056567e-01 -2.70786136e-02
-1.10453412e-01 -3.48762542e-01 -4.55997549e-02 3.34821612e-01
6.99153781e-01 -5.69116771e-01 -9.40116465e-01 5.45980036e-01
-8.10401022e-01 1.42399564e-01 3.73286694e-01 6.35395348e-01
5.95368326e-01 3.42026562e-03 3.69617313e-01 -1.58075184e-01
3.98940176e-01 -4.65004265e-01 -3.28313321e-01 2.72553235e-01
2.70060509e-01 -2.06386402e-01 4.15981978e-01 -4.54740167e-01
-1.28785264e+00 -9.00791842e-04 5.57547063e-02 -5.69358230e-01
-8.66597742e-02 4.08200264e-01 1.08987935e-01 2.88643569e-01
5.85703254e-01 3.72503370e-01 -3.55733156e-01 -3.57287943e-01
1.83667570e-01 4.48011458e-01 7.74198294e-01 -3.44162136e-01
2.97140568e-01 5.84765851e-01 -1.59602582e-01 -6.51877105e-01
-8.51130068e-01 -7.16384530e-01 -9.47673082e-01 -6.71718776e-01
1.33988583e+00 -1.22463572e+00 -4.86858606e-01 5.90686262e-01
-1.11044073e+00 -2.89748702e-02 -4.52215284e-01 9.81757820e-01
-7.53811240e-01 6.08115017e-01 -8.56034756e-01 -1.99330434e-01
-4.84756343e-02 -1.19226027e+00 1.03593051e+00 5.08599997e-01
-2.27448866e-01 -6.41718447e-01 1.75958022e-01 1.68360099e-01
-7.72810057e-02 4.10414368e-01 3.07567686e-01 -5.38746834e-01
-5.05932689e-01 9.13700014e-02 -1.10154882e-01 3.07251722e-01
-1.01155359e-02 -1.20091237e-01 -4.38175797e-01 -1.00426354e-01
1.54466033e-01 -8.21320191e-02 1.09531903e+00 5.68086505e-01
1.06594062e+00 1.38644204e-01 -2.26251379e-01 6.01296127e-01
1.07956970e+00 4.51029718e-01 8.17550242e-01 4.42674458e-01
4.91851628e-01 3.60533535e-01 9.92728353e-01 7.97089338e-01
3.61876935e-01 8.59084606e-01 8.60270262e-02 5.11690862e-02
-3.70517820e-01 -2.63073355e-01 7.24065661e-01 6.33654952e-01
-5.01659036e-01 -2.47935839e-02 -5.68418324e-01 4.90331203e-01
-1.99541342e+00 -1.57788944e+00 1.46717235e-01 2.07033205e+00
5.96147537e-01 2.69060999e-01 2.94616103e-01 -6.73922151e-02
1.11293006e+00 4.84634966e-01 -3.54323208e-01 -2.97822595e-01
-1.25618175e-01 1.15585491e-01 2.89202273e-01 -1.88380796e-02
-1.48931575e+00 9.75963593e-01 6.08172750e+00 8.81198764e-01
-9.70672846e-01 1.78588957e-01 4.47939664e-01 -2.54114062e-01
4.50255692e-01 -1.27293915e-01 -7.16145694e-01 4.28132474e-01
8.62711847e-01 -1.95609242e-01 1.73382550e-01 5.60095727e-01
4.14061934e-01 -3.03606957e-01 -1.09808493e+00 1.07892370e+00
3.73070091e-01 -1.41086411e+00 1.34249777e-02 -3.33570659e-01
7.79197037e-01 2.64427923e-02 -1.43207818e-01 4.13165480e-01
1.14609964e-01 -6.22600853e-01 1.05851185e+00 8.22725594e-01
4.08731788e-01 -8.45899999e-01 4.97493893e-01 4.50934619e-02
-1.78510642e+00 -2.20522314e-01 -5.50423443e-01 -3.06928039e-01
4.58640665e-01 1.98378861e-01 -4.69154954e-01 6.91528559e-01
1.10428715e+00 1.39617658e+00 -4.42687124e-01 1.11966503e+00
-1.19136237e-01 4.21999782e-01 -2.59753495e-01 3.37456316e-01
4.03139651e-01 -1.09369725e-01 6.09208107e-01 1.39146006e+00
6.41074657e-01 2.78127283e-01 3.37118447e-01 3.89245391e-01
1.74697444e-01 2.72079203e-02 -5.07348597e-01 3.74579787e-01
2.75688499e-01 1.02985668e+00 -8.10004592e-01 -5.43181300e-01
-7.79781938e-01 1.21212637e+00 1.15644142e-01 3.15810204e-01
-1.31002820e+00 -3.00934404e-01 6.69303358e-01 -6.48915842e-02
8.16178322e-01 -2.13754505e-01 2.33469397e-01 -1.35280895e+00
1.32855728e-01 -7.92508841e-01 9.31073129e-01 -8.28455806e-01
-1.04298449e+00 6.13737881e-01 2.93072611e-01 -1.70906925e+00
-2.01689333e-01 -1.85467198e-01 -6.44879282e-01 5.05915165e-01
-1.11382616e+00 -9.03044701e-01 -4.39600348e-01 9.38358366e-01
9.93160486e-01 -2.65409976e-01 3.59089464e-01 3.33400667e-01
-5.58535337e-01 2.49182582e-01 -3.19081917e-02 2.32385293e-01
6.96031928e-01 -8.20389867e-01 5.08945823e-01 1.10905612e+00
2.04304069e-01 1.02233492e-01 4.53837723e-01 -8.67085874e-01
-9.81941640e-01 -1.06084824e+00 8.15904915e-01 -4.15503308e-02
6.19254649e-01 -1.10287927e-01 -7.69134939e-01 8.36020470e-01
4.00395006e-01 9.65138525e-02 6.36236727e-01 -3.58882248e-01
-1.05940752e-01 -1.08478457e-01 -8.30727994e-01 7.04992890e-01
1.51657999e+00 -3.37729990e-01 -1.13777411e+00 1.00252749e-02
6.90320373e-01 -6.33978784e-01 -1.09784484e+00 4.33383107e-01
5.69043398e-01 -1.49388611e+00 1.13807940e+00 -4.44748223e-01
5.38945794e-01 -4.30761337e-01 -2.56124496e-01 -1.10862863e+00
-6.22465134e-01 -5.96120894e-01 -6.59729242e-02 9.34483469e-01
-2.44386092e-01 -1.92609996e-01 5.58576941e-01 1.21460550e-01
-2.94559956e-01 -5.46714604e-01 -1.08499515e+00 -7.41871715e-01
-5.11201084e-01 -5.01826346e-01 5.00635862e-01 4.75512713e-01
3.37668478e-01 7.55222365e-02 -5.12641370e-01 1.89157948e-01
3.29836786e-01 1.61526754e-01 4.85865921e-01 -8.65436256e-01
-1.08665437e-01 -2.82626420e-01 -1.02411175e+00 -1.39209950e+00
5.44304214e-02 -7.30246842e-01 -8.01928863e-02 -1.37311900e+00
3.69610339e-01 3.06636393e-01 -5.76807022e-01 1.30789116e-01
-2.40034178e-01 5.09801924e-01 4.30511862e-01 2.25327834e-01
-1.08436918e+00 5.16917169e-01 1.18657923e+00 3.79065648e-02
-3.07793409e-01 -3.00439119e-01 -1.62617177e-01 9.57692921e-01
5.66316962e-01 -3.63273978e-01 -3.63466591e-01 -3.54379922e-01
-2.20557258e-01 3.54331076e-01 3.52635473e-01 -1.73514903e+00
3.61902863e-01 -4.33101021e-02 6.02068424e-01 -9.31396902e-01
5.08611858e-01 -6.52767122e-01 2.88592070e-01 4.18202192e-01
-4.28583890e-01 3.05357158e-01 9.35570449e-02 5.86274624e-01
-6.47271454e-01 -2.45688995e-03 5.47107816e-01 -1.30051792e-01
-1.50961089e+00 2.58223146e-01 -4.82334048e-01 1.26825199e-01
1.51828015e+00 -6.35640204e-01 6.03833795e-02 -4.00215030e-01
-9.62070644e-01 1.69195846e-01 3.45699906e-01 6.84653699e-01
1.00757492e+00 -1.58627701e+00 -8.24437916e-01 2.67184883e-01
-7.25013092e-02 -4.51138645e-01 7.24218965e-01 8.72108161e-01
-6.52453840e-01 2.75434941e-01 -7.45581448e-01 -7.59409010e-01
-1.38804638e+00 6.27693415e-01 3.16024989e-01 -3.34916294e-01
-1.09148967e+00 7.59153843e-01 5.82405865e-01 3.40427369e-01
3.16340476e-01 -4.76827353e-01 -6.36153817e-01 2.49320120e-01
9.74025607e-01 4.22242969e-01 -1.55176952e-01 -1.04866004e+00
-3.67884815e-01 6.93832755e-01 3.48955393e-02 -1.80308670e-01
1.38007796e+00 -3.51925641e-01 1.96846485e-01 3.66113603e-01
1.24000371e+00 -3.81107956e-01 -1.77150786e+00 -5.37296474e-01
3.96401621e-02 -6.62934363e-01 -1.96303874e-01 -2.63237059e-01
-1.23061168e+00 6.36986554e-01 5.85195243e-01 2.92080529e-02
1.54501247e+00 -2.67211460e-02 1.03057992e+00 1.24437474e-01
3.14120740e-01 -1.18115139e+00 3.31143916e-01 5.67356229e-01
8.38164687e-01 -9.52775538e-01 -9.00320411e-02 -1.77347928e-01
-1.01091552e+00 1.40987527e+00 5.39847136e-01 -4.13057029e-01
6.09226227e-01 -1.39084896e-02 -2.70014763e-01 -2.26154789e-01
-6.14820182e-01 -3.18268776e-01 4.93700981e-01 4.39335436e-01
2.20253438e-01 -2.89314389e-01 -5.07498384e-01 5.86245060e-01
-2.35754140e-02 2.92738676e-01 3.84556949e-01 9.42394614e-01
-6.32513821e-01 -7.60054350e-01 -3.76699567e-01 2.00016394e-01
-5.85958540e-01 6.64330497e-02 -8.62281397e-03 7.46120811e-01
4.51821834e-01 9.14481521e-01 4.08433586e-01 -6.11678123e-01
6.65464878e-01 1.64161865e-02 4.51066643e-01 -3.99626553e-01
-7.86541224e-01 1.81820750e-01 9.68905166e-02 -1.17385781e+00
-7.08702862e-01 -9.62690353e-01 -1.34757245e+00 -1.58951990e-02
3.28194559e-01 -1.56851076e-02 6.41820282e-02 7.60768712e-01
4.44070011e-01 6.70887411e-01 5.97253263e-01 -1.30707848e+00
-2.51009315e-01 -8.03590000e-01 -6.13486469e-01 7.58692503e-01
2.12681456e-03 -8.17494571e-01 2.95725111e-02 5.72841883e-01] | [8.470132827758789, 0.40288248658180237] |
f0b74be6-e522-4a70-a0bd-fc4219d45a77 | microscopy-image-restoration-with-deep-wiener | 1911.10989 | null | https://arxiv.org/abs/1911.10989v3 | https://arxiv.org/pdf/1911.10989v3.pdf | Microscopy Image Restoration with Deep Wiener-Kolmogorov filters | Microscopy is a powerful visualization tool in biology, enabling the study of cells, tissues, and the fundamental biological processes; yet, the observed images typically suffer from blur and background noise. In this work, we propose a unifying framework of algorithms for Gaussian image deblurring and denoising. These algorithms are based on deep learning techniques for the design of learnable regularizers integrated into the Wiener-Kolmogorov filter. Our extensive experimentation line showcases that the proposed approach achieves a superior quality of image reconstruction and surpasses the solutions that rely either on deep learning or on optimization schemes alone. Augmented with the variance stabilizing transformation, the proposed reconstruction pipeline can also be successfully applied to the problem of Poisson image deblurring, surpassing the state-of-the-art methods. Moreover, several variants of the proposed framework demonstrate competitive performance at low computational complexity, which is of high importance for real-time imaging applications. | ['Stamatios Lefkimmiatis', 'Valeriya Pronina', 'Dmitry V. Dylov', 'Filippos Kokkinos'] | 2019-11-25 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3405_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123650188.pdf | eccv-2020-8 | ['image-deconvolution'] | ['computer-vision'] | [ 3.92419606e-01 -3.52128953e-01 3.95707637e-01 5.25766723e-02
-5.60814977e-01 -2.29111746e-01 6.31363213e-01 1.71813756e-01
-7.85513699e-01 8.19025338e-01 -1.93875268e-01 -2.41765812e-01
-2.17659891e-01 -3.18229526e-01 -6.28926516e-01 -1.41584945e+00
7.83981159e-02 2.63313204e-01 1.65539667e-01 2.51439214e-01
4.41857606e-01 7.36065328e-01 -1.32953370e+00 -1.80151939e-01
6.69471383e-01 9.28977311e-01 3.68573934e-01 7.15358496e-01
-8.06262940e-02 6.84887826e-01 -2.10848480e-01 -1.57391742e-01
-1.13956250e-01 -2.49995008e-01 -5.02334893e-01 7.95681030e-02
4.37550485e-01 -1.95858493e-01 -2.88055062e-01 1.05569041e+00
6.55398905e-01 3.95510718e-02 6.42049730e-01 -5.76430142e-01
-5.51741600e-01 5.02437539e-02 -6.23985052e-01 4.99898106e-01
-1.08352475e-01 1.90640777e-01 5.36153972e-01 -7.93860257e-01
5.64494133e-01 9.33328271e-01 5.97217917e-01 4.16012943e-01
-1.61556876e+00 -2.28466466e-01 -1.03003636e-01 3.83677065e-01
-1.10183573e+00 -6.05774343e-01 6.19127154e-01 -6.44268095e-01
4.45473522e-01 8.26138332e-02 3.89019042e-01 9.76406097e-01
4.41917837e-01 6.10149324e-01 1.41542053e+00 -1.86574772e-01
3.20921302e-01 -1.27556026e-01 1.27527311e-01 6.10161781e-01
3.21843565e-01 -2.71384306e-02 -3.59280199e-01 -7.57861659e-02
1.01271665e+00 1.61035247e-02 -7.67884612e-01 -3.37849021e-01
-1.28168762e+00 3.57654810e-01 3.51410836e-01 4.78209317e-01
-3.74971837e-01 2.73616880e-01 2.53778934e-01 8.68285149e-02
4.77729768e-01 2.54051715e-01 -1.08699694e-01 7.87763670e-02
-1.17235434e+00 1.66205451e-01 5.62094331e-01 3.75147998e-01
6.29151464e-01 1.12123348e-01 -1.52701363e-01 4.60734516e-01
1.53505161e-01 4.35826898e-01 3.28057200e-01 -9.83566761e-01
-1.92979321e-01 1.46351531e-01 3.63852441e-01 -9.97495294e-01
-4.89853352e-01 -7.14507222e-01 -1.38374949e+00 4.08046722e-01
6.52229011e-01 2.06635535e-01 -7.24057436e-01 1.30813146e+00
3.34406585e-01 7.14635551e-01 -5.96982427e-02 8.57279897e-01
6.64351463e-01 4.63997483e-01 -2.29520813e-01 -5.28858840e-01
1.20765650e+00 -6.80697262e-01 -9.64083433e-01 6.70201257e-02
1.24580100e-01 -7.89184272e-01 8.45014870e-01 6.72662914e-01
-9.76380706e-01 -3.80426943e-01 -8.70906413e-01 -1.75076336e-01
1.51552521e-02 1.79864720e-01 3.14670920e-01 5.11174560e-01
-1.18662381e+00 9.66343224e-01 -1.14726996e+00 -2.79883176e-01
6.25693381e-01 1.62786648e-01 -3.90721798e-01 -8.41508806e-02
-3.38482916e-01 9.30464149e-01 -4.49641198e-02 3.42492193e-01
-9.92551446e-01 -8.71147454e-01 -5.24746418e-01 1.36966735e-01
-2.13193055e-02 -8.42910588e-01 7.17788041e-01 -4.74158585e-01
-1.70483339e+00 1.05001843e+00 -3.16425025e-01 -5.92237413e-01
8.48223031e-01 -4.06802893e-01 1.66007593e-01 2.09992260e-01
-4.31984752e-01 9.08083093e-05 1.13198102e+00 -1.23209369e+00
-2.64144599e-01 -4.54494894e-01 -4.50242460e-01 -1.51804820e-01
-3.27793777e-01 -8.78935382e-02 -2.73999691e-01 -6.22667909e-01
3.83709185e-02 -5.99038124e-01 -2.98795044e-01 2.87195027e-01
-2.94209272e-01 2.48251915e-01 7.18473077e-01 -7.77943134e-01
8.02615225e-01 -2.13431597e+00 6.73333347e-01 -1.93991274e-01
3.58204037e-01 3.20785493e-01 -1.21444508e-01 2.86986560e-01
1.20428596e-02 -2.50990331e-01 -4.20238346e-01 -7.51388729e-01
-2.35278681e-01 6.99771848e-03 -2.41806403e-01 1.07872939e+00
3.81058678e-02 7.37484694e-01 -7.71804512e-01 -1.50794834e-01
5.32598674e-01 8.93207073e-01 -1.26569644e-01 2.96380699e-01
1.83074735e-02 1.07314205e+00 -7.09122280e-03 2.15523541e-01
8.75706315e-01 -4.68909681e-01 1.99313741e-02 -4.24273700e-01
-3.09512258e-01 -1.37346312e-01 -1.00404346e+00 1.53287065e+00
-3.89267653e-01 6.98462129e-01 6.64152265e-01 -1.26382887e+00
6.82489455e-01 1.07047774e-01 5.40258944e-01 -1.14018828e-01
2.50745982e-01 3.15935701e-01 -2.04036281e-01 -6.08623385e-01
8.58820975e-02 -1.59156397e-01 6.63593888e-01 3.31116587e-01
9.63702723e-02 -2.12581515e-01 1.06107794e-01 -1.27022550e-01
1.08324707e+00 8.27425048e-02 1.87799618e-01 -5.52273214e-01
8.77023280e-01 -6.23729765e-01 2.11475089e-01 7.49590695e-01
-9.35078859e-02 7.09592760e-01 4.34399068e-01 -4.58496958e-01
-1.13279355e+00 -9.31950092e-01 -4.15548533e-01 5.05591750e-01
2.48112097e-01 1.50498182e-01 -9.48026240e-01 -2.00011641e-01
-2.40726963e-01 3.24974358e-01 -5.61047018e-01 1.79691538e-01
-4.85759884e-01 -1.04935050e+00 4.24914837e-01 5.97928874e-02
3.33958179e-01 -8.11606884e-01 -6.23832762e-01 2.25714698e-01
-1.24845058e-01 -1.38071001e+00 -8.86262730e-02 1.27045363e-01
-8.81259680e-01 -1.04331040e+00 -1.21560073e+00 -6.24852061e-01
5.21781206e-01 3.82456362e-01 8.26037645e-01 2.09034368e-01
-3.70586395e-01 3.39468330e-01 -2.38133292e-03 1.09713368e-01
-4.82130408e-01 -1.20955996e-01 -1.26067653e-01 2.83488393e-01
-2.39664599e-01 -9.56811309e-01 -8.52266848e-01 1.70716852e-01
-1.10607278e+00 4.67617810e-02 5.38005531e-01 1.08228874e+00
6.77852571e-01 1.13866076e-01 2.88290679e-01 -5.56309283e-01
4.23206478e-01 -1.29610047e-01 -8.84877324e-01 1.24917723e-01
-4.63585883e-01 2.03517690e-01 5.95662653e-01 -3.86110991e-01
-9.38799322e-01 -5.98327033e-02 -2.60779232e-01 -3.56635809e-01
-2.98661649e-01 3.01638901e-01 2.77152918e-02 -5.70590019e-01
4.40062672e-01 5.61983228e-01 2.30073974e-01 -6.67657852e-01
1.43290117e-01 2.70823091e-01 7.88385570e-01 -3.20184857e-01
6.86580420e-01 1.06692994e+00 4.19445604e-01 -1.23208559e+00
-5.28933823e-01 -5.68039656e-01 -4.18200314e-01 -1.66672140e-01
8.60003710e-01 -6.00810170e-01 -8.31279516e-01 9.99705553e-01
-1.34951615e+00 -3.45190495e-01 3.08597162e-02 3.25398088e-01
-7.92517126e-01 9.49935257e-01 -9.90482092e-01 -7.02631056e-01
-3.32112938e-01 -1.33649123e+00 1.07542360e+00 2.19157115e-01
3.55254650e-01 -1.12491012e+00 1.01511024e-01 3.43831658e-01
5.89271903e-01 1.30040407e-01 9.32136953e-01 -1.85762554e-01
-7.02727437e-01 -6.66795298e-02 -4.69958782e-01 3.58820170e-01
1.24594666e-01 -3.66438180e-02 -1.18715727e+00 -6.40874863e-01
4.50554311e-01 -5.68966866e-02 1.08417046e+00 7.36801088e-01
1.27096260e+00 2.62260456e-02 -1.21109143e-01 9.36506689e-01
1.49492502e+00 -3.69552314e-01 7.09977269e-01 2.98060894e-01
5.60981929e-01 5.52416444e-01 -2.39513554e-02 4.94539797e-01
1.19560454e-02 7.60139763e-01 6.30674601e-01 -2.40749642e-01
-2.36414537e-01 4.07887250e-01 7.37008974e-02 6.99639976e-01
-2.08124936e-01 -2.50106990e-01 -7.07489848e-01 4.76631403e-01
-1.90681279e+00 -9.07237291e-01 -3.52657914e-01 2.40624452e+00
6.49411440e-01 -1.86193258e-01 -2.89491057e-01 2.37841338e-01
6.24561191e-01 1.48109555e-01 -5.12719572e-01 1.03455745e-01
-2.57421494e-01 2.50323117e-01 4.11574662e-01 5.62913537e-01
-1.20529008e+00 6.11585319e-01 6.17399359e+00 8.66137862e-01
-1.24202693e+00 1.01252481e-01 5.41178286e-01 2.99323916e-01
-8.98715481e-02 -4.33361858e-01 -5.72718978e-01 5.09657443e-01
5.45750082e-01 4.50108312e-02 6.41152501e-01 2.72331804e-01
5.07472277e-01 -1.72548741e-01 -9.72474635e-01 1.27104378e+00
-7.22716600e-02 -1.48281085e+00 -2.46821582e-01 1.64498881e-01
6.31043971e-01 -1.01117790e-02 3.46422315e-01 -4.79087979e-01
-6.23628721e-02 -1.01399219e+00 3.16585183e-01 7.15189338e-01
4.10708964e-01 -4.30882633e-01 9.66420174e-01 4.62229699e-01
-5.75994730e-01 6.02357648e-02 -4.47233319e-01 1.19161822e-01
4.19592768e-01 1.14904451e+00 -3.18819344e-01 5.25868952e-01
6.10793889e-01 8.81649613e-01 -2.90367126e-01 1.33083880e+00
-1.55874416e-01 4.80838120e-01 -2.22774401e-01 2.67097324e-01
-2.37083025e-02 -4.30799842e-01 7.46485829e-01 1.32053077e+00
5.65734625e-01 -1.25257120e-01 -1.97717950e-01 7.43586600e-01
7.60696083e-02 1.11344390e-01 -3.61723542e-01 5.28647751e-02
6.69331383e-03 1.40869045e+00 -9.30514574e-01 -2.76379526e-01
-2.32029334e-01 1.10068834e+00 3.01103294e-01 4.87392664e-01
-6.51869893e-01 8.81816819e-03 7.14577198e-01 2.12940425e-01
5.48757613e-01 -3.70648205e-01 -2.39086092e-01 -1.40611982e+00
-6.62383139e-02 -8.31191182e-01 -1.11062750e-01 -6.06172264e-01
-1.35335052e+00 4.78546232e-01 -3.52506071e-01 -7.97747672e-01
2.46744439e-01 -8.60294104e-01 -5.63976943e-01 7.90738344e-01
-1.96609223e+00 -8.80150437e-01 -4.53706145e-01 3.69239658e-01
3.44161421e-01 1.38008669e-01 6.91156924e-01 3.84913534e-01
-5.81773400e-01 9.18217301e-02 7.13377416e-01 -3.13904136e-01
7.13024557e-01 -1.19855964e+00 2.15522483e-01 1.01078713e+00
-5.60537763e-02 5.69427133e-01 1.07639086e+00 -2.05210015e-01
-1.26536763e+00 -7.86343038e-01 4.59793240e-01 6.63037272e-03
8.37060094e-01 -9.47826579e-02 -1.20720971e+00 1.99441582e-01
3.61608148e-01 2.24934965e-01 5.57858467e-01 -2.96502948e-01
-3.95462252e-02 -1.70218036e-01 -1.03842854e+00 4.99190688e-01
5.85693181e-01 -3.00229996e-01 -2.53645092e-01 3.54759634e-01
1.07669771e-01 -3.20085168e-01 -6.55541420e-01 2.78504968e-01
5.65791190e-01 -1.19162953e+00 1.12864852e+00 -3.28730971e-01
4.74145770e-01 -4.93282557e-01 1.66565433e-01 -1.24568224e+00
-4.49668080e-01 -7.84907818e-01 -3.14702004e-01 9.68529582e-01
-2.00488672e-01 -5.78264773e-01 6.41354203e-01 8.37586448e-02
-1.53891472e-02 -5.13882995e-01 -1.12062299e+00 -6.27082229e-01
4.95936275e-02 -7.55558461e-02 -3.16131115e-02 6.27655447e-01
-5.17371953e-01 -1.80442333e-02 -5.68273544e-01 2.97661752e-01
1.24027336e+00 3.11560072e-02 6.64431214e-01 -1.15829456e+00
-4.56496328e-01 -7.01297998e-01 -6.69500530e-01 -1.27025259e+00
2.62037307e-01 -5.32726884e-01 2.68180430e-01 -1.25116837e+00
3.47534925e-01 -1.31612673e-01 -4.03823733e-01 -5.41265607e-02
-3.43344659e-01 4.56642807e-01 6.62738755e-02 2.23972037e-01
-2.92019486e-01 4.54836667e-01 1.23471344e+00 -2.22348154e-01
6.89249560e-02 -2.48646792e-02 -3.04128677e-01 7.65177488e-01
7.50231326e-01 -2.87243307e-01 -9.78084356e-02 -5.96820712e-01
9.23065841e-02 -1.63007438e-01 6.47455335e-01 -1.06699514e+00
4.10979241e-01 5.57992607e-02 2.05106735e-01 -2.57589251e-01
3.15108776e-01 -7.32256114e-01 2.56341070e-01 5.66003621e-01
-2.32826352e-01 -2.88986683e-01 1.35191828e-01 8.10801566e-01
-8.97735804e-02 -2.83270180e-01 1.31183040e+00 -6.71501383e-02
-1.78703383e-01 1.45151138e-01 -7.33072340e-01 -1.85053676e-01
7.08063483e-01 4.32809349e-03 -2.93395191e-01 -2.80518621e-01
-7.06251919e-01 -2.30524898e-01 6.90709770e-01 -2.46305466e-01
6.29629016e-01 -8.44440281e-01 -8.01271617e-01 8.81021991e-02
-3.94142628e-01 -6.58702627e-02 4.04944479e-01 1.44611406e+00
-7.50626802e-01 1.80115148e-01 -2.44442925e-01 -7.67324328e-01
-1.18679321e+00 5.49058974e-01 4.45784986e-01 -4.15096611e-01
-6.81831419e-01 8.40047777e-01 4.87272143e-01 4.61766683e-02
1.67307451e-01 -3.24625313e-01 -1.66762561e-01 -2.22472206e-01
7.76345849e-01 5.82128406e-01 2.27414846e-01 -5.49091637e-01
-2.00196788e-01 7.72610843e-01 -7.06640109e-02 1.39915541e-01
1.52013457e+00 -3.30895573e-01 -4.59011078e-01 3.87684762e-01
9.89316761e-01 -2.31258329e-02 -1.44201386e+00 -1.94224507e-01
4.57620211e-02 -3.44292730e-01 4.12274659e-01 -3.10559660e-01
-1.08204424e+00 1.28372133e+00 7.29397118e-01 1.55289784e-01
9.99800026e-01 -1.27286419e-01 6.92380369e-01 3.24832737e-01
3.29276949e-01 -4.79192287e-01 1.12038016e-01 1.88230470e-01
7.49151528e-01 -1.14166892e+00 1.85799748e-01 -3.73836935e-01
-2.66251825e-02 1.18730760e+00 -1.36307171e-02 -2.79256523e-01
6.02267444e-01 5.23492098e-01 1.97097898e-01 4.37409617e-02
-5.05402923e-01 -1.10763825e-01 5.26361167e-02 6.53993905e-01
4.75045055e-01 -4.00211573e-01 -4.19373214e-01 1.35696054e-01
4.63722229e-01 1.90009788e-01 4.97746825e-01 4.34158117e-01
-4.36536998e-01 -9.23221767e-01 -4.07175660e-01 1.51821479e-01
-6.85590923e-01 -1.47125185e-01 9.35758203e-02 2.91039169e-01
-6.50153533e-02 6.62191093e-01 -1.56671032e-01 3.22746247e-01
-3.09971664e-02 -3.19700360e-01 7.75712371e-01 -1.48661762e-01
-3.28712761e-01 4.02176052e-01 -5.47136664e-01 -5.26883900e-01
-7.58896828e-01 -6.11701250e-01 -8.06996584e-01 -2.07583368e-01
-2.72760242e-01 -5.84730245e-02 4.34046477e-01 9.34993505e-01
2.87037075e-01 4.61502761e-01 3.54919672e-01 -1.27231050e+00
-5.59213936e-01 -7.75566280e-01 -6.32343709e-01 3.22998554e-01
7.05273390e-01 -5.87303817e-01 -5.09529412e-01 4.73190188e-01] | [12.102730751037598, -2.5872907638549805] |
d3bf6cb8-8eb4-45e3-afbd-3c4134f4ffa6 | contrastive-video-question-answering-via | 2302.13668 | null | https://arxiv.org/abs/2302.13668v2 | https://arxiv.org/pdf/2302.13668v2.pdf | Contrastive Video Question Answering via Video Graph Transformer | We propose to perform video question answering (VideoQA) in a Contrastive manner via a Video Graph Transformer model (CoVGT). CoVGT's uniqueness and superiority are three-fold: 1) It proposes a dynamic graph transformer module which encodes video by explicitly capturing the visual objects, their relations and dynamics, for complex spatio-temporal reasoning. 2) It designs separate video and text transformers for contrastive learning between the video and text to perform QA, instead of multi-modal transformer for answer classification. Fine-grained video-text communication is done by additional cross-modal interaction modules. 3) It is optimized by the joint fully- and self-supervised contrastive objectives between the correct and incorrect answers, as well as the relevant and irrelevant questions respectively. With superior video encoding and QA solution, we show that CoVGT can achieve much better performances than previous arts on video reasoning tasks. Its performances even surpass those models that are pretrained with millions of external data. We further show that CoVGT can also benefit from cross-modal pretraining, yet with orders of magnitude smaller data. The results demonstrate the effectiveness and superiority of CoVGT, and additionally reveal its potential for more data-efficient pretraining. We hope our success can advance VideoQA beyond coarse recognition/description towards fine-grained relation reasoning of video contents. Our code is available at https://github.com/doc-doc/CoVGT. | ['Tat-Seng Chua', 'Shuicheng Yan', 'Richang Hong', 'Yicong Li', 'Angela Yao', 'Pan Zhou', 'Junbin Xiao'] | 2023-02-27 | null | null | null | null | ['video-question-answering'] | ['computer-vision'] | [-2.03687042e-01 -7.48242736e-02 -3.29621173e-02 -1.26219228e-01
-9.26872790e-01 -6.70372784e-01 5.15082717e-01 -3.47922355e-01
-1.09235317e-01 3.68002892e-01 6.65741026e-01 -2.95334041e-01
-2.08580837e-01 -6.38013005e-01 -1.01580703e+00 -4.09580320e-01
-3.07524484e-02 6.26564622e-01 3.07856172e-01 -3.74090463e-01
-1.82804298e-02 6.31921068e-02 -1.63667524e+00 1.01015615e+00
6.38893425e-01 1.20521462e+00 1.01524308e-01 1.12575245e+00
-2.10956320e-01 1.66911185e+00 -4.49462742e-01 -8.72628808e-01
-8.51867162e-03 -4.50962842e-01 -1.05996788e+00 2.38019869e-01
9.24030185e-01 -6.94959700e-01 -1.00897861e+00 7.03609884e-01
1.24015599e-01 1.60777479e-01 5.05010962e-01 -1.36026204e+00
-1.05196404e+00 2.64186800e-01 -2.97876626e-01 3.74423027e-01
8.55747700e-01 5.07942855e-01 1.36137462e+00 -9.52052176e-01
7.32964277e-01 1.45893133e+00 4.38497454e-01 5.71415782e-01
-8.08681369e-01 -4.79930371e-01 4.12231743e-01 9.33813810e-01
-1.27305436e+00 -4.90110546e-01 6.41579807e-01 -4.55444694e-01
1.09190500e+00 4.09804195e-01 8.38596582e-01 1.33182311e+00
8.67273733e-02 1.17620814e+00 6.62268102e-01 1.73621044e-01
1.36210313e-02 -3.88088256e-01 -1.32333368e-01 1.18507457e+00
-3.84295970e-01 -2.73874223e-01 -8.46767128e-01 9.09307078e-02
7.12344766e-01 1.11881256e-01 -5.62413275e-01 -4.77530211e-01
-1.36582434e+00 7.15930402e-01 3.57470840e-01 9.85717103e-02
-1.93627074e-01 5.33322990e-01 5.02614081e-01 6.57654524e-01
1.74969852e-01 5.33064641e-02 -4.69599396e-01 -3.94609898e-01
-7.08767891e-01 1.26669884e-01 8.10006857e-01 1.16341758e+00
6.98350906e-01 1.20767795e-01 -3.81970078e-01 6.01442277e-01
2.66349494e-01 5.53336620e-01 3.77152920e-01 -1.42174780e+00
9.14635360e-01 6.78314030e-01 -1.24593534e-01 -1.32050180e+00
-1.17432028e-01 -1.38885990e-01 -7.60513365e-01 -3.05413991e-01
4.46318448e-01 7.08552822e-02 -7.84469485e-01 1.55982411e+00
1.81845739e-01 4.62682635e-01 1.27938751e-03 1.04921782e+00
1.30483556e+00 8.18199933e-01 7.10129812e-02 -8.09547082e-02
1.56124890e+00 -1.37311244e+00 -7.81449020e-01 -8.56097713e-02
7.14794457e-01 -5.01875222e-01 1.48468971e+00 3.06970149e-01
-1.48620772e+00 -6.27649248e-01 -6.63285971e-01 -5.52154541e-01
-2.77299285e-01 -5.39982393e-02 6.51900113e-01 1.94838330e-01
-1.35376513e+00 8.44888166e-02 -7.12100983e-01 -2.05538496e-01
6.64081097e-01 2.82571852e-01 -4.39673245e-01 -4.62630749e-01
-1.21994412e+00 4.28169966e-01 -6.45797849e-02 6.61526471e-02
-1.07844889e+00 -6.68426037e-01 -1.02962935e+00 1.64072007e-01
7.16835320e-01 -1.22656262e+00 1.22832167e+00 -1.14207125e+00
-1.34919918e+00 8.23164821e-01 -3.47213954e-01 -2.45268628e-01
5.35942852e-01 -2.57274210e-01 -3.24139237e-01 9.63668644e-01
3.17569971e-02 7.16117203e-01 1.06166255e+00 -1.02332246e+00
-5.35573184e-01 -2.38294482e-01 6.64906681e-01 3.75058800e-01
-3.88116002e-01 -1.65870100e-01 -1.20176387e+00 -6.94661975e-01
-1.24320649e-01 -5.31046510e-01 2.95340687e-01 2.53447235e-01
5.32160439e-02 -2.62166530e-01 1.10153377e+00 -8.92003894e-01
1.13285577e+00 -2.06924415e+00 6.11213267e-01 -3.07304770e-01
7.23540843e-01 1.18811570e-01 -4.37743217e-01 5.52236974e-01
3.56747732e-02 -8.43481496e-02 9.23474282e-02 -2.16858596e-01
6.63959682e-02 4.22536582e-01 -2.97560304e-01 2.73802578e-01
3.96052092e-01 1.53407073e+00 -8.89701962e-01 -6.58078134e-01
1.73722595e-01 5.59050262e-01 -9.20064449e-01 2.37944409e-01
-5.50005376e-01 2.23896280e-01 -5.86208940e-01 8.22918475e-01
3.94954950e-01 -9.47991610e-01 2.52688766e-01 -7.24195123e-01
5.33410490e-01 9.02907848e-02 -7.88339972e-01 1.76206326e+00
-2.34432340e-01 8.42439115e-01 1.76905453e-01 -1.09793627e+00
2.43623123e-01 3.65616530e-01 5.46724200e-01 -1.12025011e+00
2.53507402e-02 -2.21555948e-01 -3.39585185e-01 -1.03328598e+00
4.01808292e-01 2.08559200e-01 6.46399986e-03 2.86916614e-01
3.82585794e-01 -2.37133000e-02 3.89191091e-01 9.02440310e-01
1.30184567e+00 4.23845917e-01 -1.52666718e-01 1.20145492e-01
6.91428661e-01 7.00343475e-02 1.18029438e-01 5.74553490e-01
-2.17507452e-01 4.67449814e-01 6.62929893e-01 -4.48908210e-01
-7.28465736e-01 -1.15116560e+00 5.30407727e-01 1.26474392e+00
4.14219201e-01 -8.03210020e-01 -5.81837535e-01 -8.98567677e-01
9.87393083e-04 1.84377894e-01 -5.83673477e-01 -1.04310222e-01
-6.86572313e-01 -1.13106892e-01 4.24084544e-01 5.65317214e-01
6.87299132e-01 -8.67175639e-01 -2.61393249e-01 -2.20294729e-01
-9.72884357e-01 -1.45707774e+00 -6.62029803e-01 -3.30839872e-01
-7.56318867e-01 -1.33008206e+00 -6.07477784e-01 -6.91283047e-01
4.99233425e-01 6.02051497e-01 1.41032040e+00 5.21529019e-01
-2.30964124e-02 1.29069448e+00 -5.95627844e-01 4.02317405e-01
-6.03571050e-02 -4.00722414e-01 -2.85293281e-01 4.43258621e-02
1.41980201e-01 -5.10467827e-01 -8.71151865e-01 5.64488173e-01
-9.45984602e-01 2.30380848e-01 3.01754892e-01 8.66974652e-01
4.49096978e-01 -1.00547552e-01 2.14539140e-01 -7.36905575e-01
3.93308580e-01 -5.53679407e-01 -2.28159264e-01 4.14796174e-01
-9.15211365e-02 -6.58539385e-02 5.97187877e-01 -3.84624898e-01
-1.03065217e+00 -3.45786661e-01 -1.44575045e-01 -9.36452389e-01
7.10799694e-02 3.64290327e-01 8.88480432e-03 -6.18413985e-02
3.35481405e-01 1.76628515e-01 -9.99179035e-02 3.10633797e-02
4.62093413e-01 1.53200969e-01 5.95435739e-01 -7.64182627e-01
8.15789640e-01 6.48021102e-01 -2.72466779e-01 -6.24012113e-01
-8.78304958e-01 -5.37843645e-01 -3.24661523e-01 -6.54639304e-01
1.14455807e+00 -1.25810516e+00 -1.26471591e+00 2.44925693e-01
-1.06982625e+00 -6.43043935e-01 -1.92290902e-01 2.84386754e-01
-8.24847460e-01 7.54850030e-01 -1.06286550e+00 -4.13745254e-01
-3.32490318e-02 -1.10374069e+00 1.24605703e+00 -2.04066828e-01
3.11281253e-02 -1.10908508e+00 -4.17257667e-01 1.28434575e+00
1.34569973e-01 -2.10091725e-01 7.68835962e-01 -1.31461993e-01
-1.23325181e+00 1.99072391e-01 -5.72695434e-01 1.16318986e-01
-1.46552846e-01 -4.45494466e-02 -7.66287923e-01 -3.87723148e-01
-3.55104282e-02 -6.96344972e-01 9.56265926e-01 1.55439883e-01
1.39113915e+00 -4.69868124e-01 -1.08556218e-01 8.72951865e-01
1.25612772e+00 5.88514321e-02 8.12442899e-01 1.29969716e-01
1.17073715e+00 5.24260461e-01 6.56320572e-01 2.53691077e-01
1.02396357e+00 5.76645255e-01 7.61455834e-01 6.31026402e-02
-5.23539960e-01 -3.23368907e-01 6.59899294e-01 1.19748044e+00
-2.76914507e-01 -6.54033184e-01 -8.20946574e-01 4.43507075e-01
-2.15475321e+00 -1.35094190e+00 -2.24162385e-01 1.52227950e+00
3.90638649e-01 -1.94775477e-01 2.45504469e-01 3.44337113e-02
2.39530653e-01 3.20881456e-01 -4.79510844e-01 -1.30452260e-01
-1.80381700e-01 -2.51361802e-02 7.70437345e-02 4.44813848e-01
-1.00192094e+00 9.64661658e-01 6.06819916e+00 8.84545445e-01
-7.46231318e-01 2.57949710e-01 5.80100119e-01 -2.54095554e-01
-5.84883809e-01 -3.11594635e-01 -3.61632675e-01 2.86106467e-01
8.90808105e-01 2.12633029e-01 6.30849063e-01 2.96407551e-01
1.13140410e-02 9.55687091e-02 -1.04353690e+00 1.32636034e+00
4.53686923e-01 -1.80380380e+00 5.05258620e-01 -2.47284457e-01
5.35118103e-01 -9.85631868e-02 8.93099755e-02 5.88221848e-01
1.24256082e-01 -7.93333292e-01 9.88215625e-01 5.79149783e-01
8.19816172e-01 -4.17018056e-01 4.16149139e-01 1.51163250e-01
-1.56704056e+00 -3.94038498e-01 -1.05111599e-01 -1.01424709e-01
1.36133909e-01 -4.91434745e-02 -1.36694670e-01 7.30000436e-01
1.07509184e+00 1.16054916e+00 -6.74384177e-01 5.27172744e-01
-9.62677896e-02 4.91097718e-01 9.54114422e-02 1.24547727e-01
3.49319756e-01 -1.39021143e-01 3.27151209e-01 1.07297301e+00
2.15532452e-01 4.09753501e-01 3.29392101e-03 5.09896219e-01
-2.25508362e-01 -5.41670918e-02 -4.35030371e-01 -1.69151172e-01
6.47024363e-02 9.09206271e-01 -4.20774341e-01 -6.01389110e-01
-8.19561839e-01 1.32828712e+00 5.87389290e-01 7.87130535e-01
-1.23061860e+00 1.56894326e-01 6.87257290e-01 6.61568493e-02
7.24426508e-01 -1.60624117e-01 2.42703646e-01 -1.71627223e+00
2.12603837e-01 -1.30408406e+00 9.99671102e-01 -1.35004866e+00
-1.32353032e+00 6.20702446e-01 -1.08441509e-01 -1.15348148e+00
-1.07109353e-01 -9.36387300e-01 -1.90992400e-01 2.41768479e-01
-1.53155184e+00 -1.37956381e+00 -5.95405161e-01 1.36763203e+00
8.21696520e-01 -7.15973154e-02 6.01934671e-01 6.52980804e-01
-4.04802650e-01 6.00282371e-01 -1.81884095e-01 2.14613065e-01
6.56464159e-01 -1.09431100e+00 1.00411065e-01 5.99520981e-01
3.39854807e-01 1.33295730e-01 4.19913083e-01 -4.56310272e-01
-2.05064082e+00 -1.01696837e+00 4.67498690e-01 -7.61149406e-01
1.01591957e+00 -3.55415374e-01 -7.02933192e-01 9.57996368e-01
3.15755576e-01 1.24975666e-01 3.74334395e-01 -3.18021588e-02
-7.58328378e-01 -1.10621296e-01 -7.75355518e-01 8.22876096e-01
1.42144740e+00 -1.03166533e+00 -5.38296878e-01 5.27213752e-01
1.07609832e+00 -5.06276369e-01 -9.24411178e-01 1.97744355e-01
5.88409126e-01 -1.17697585e+00 1.36845851e+00 -6.61290050e-01
8.73588800e-01 -1.45273909e-01 -3.86523485e-01 -7.00062573e-01
-4.18990761e-01 -6.20393276e-01 -7.58421719e-01 1.00395858e+00
1.06816791e-01 -3.33397001e-01 8.86655927e-01 2.50075400e-01
-1.81741625e-01 -7.47246981e-01 -7.40817964e-01 -5.23257732e-01
-2.71381825e-01 -6.60856783e-01 3.58317286e-01 9.04923081e-01
-3.20456736e-02 4.42345262e-01 -6.18708611e-01 2.20034912e-01
3.87550086e-01 1.81763574e-01 7.13507533e-01 -7.27176607e-01
-5.81273019e-01 -4.06682253e-01 -5.22511363e-01 -1.63875687e+00
1.51251361e-01 -7.34423757e-01 -3.06309044e-01 -1.76263905e+00
3.22166413e-01 1.78064033e-01 -1.03056587e-01 3.37991774e-01
-6.26864955e-02 6.00693345e-01 3.71567547e-01 2.51062036e-01
-1.36356020e+00 6.19579911e-01 1.90606749e+00 -4.12449330e-01
2.91859031e-01 -4.75854129e-01 -4.59826887e-01 6.35912716e-01
4.65923786e-01 -2.23582909e-02 -7.54236221e-01 -9.48935211e-01
5.06545842e-01 5.47469914e-01 7.45064795e-01 -8.63467276e-01
3.85971129e-01 -9.55894589e-02 3.09484363e-01 -5.67574441e-01
7.63226926e-01 -8.58607352e-01 6.93746805e-02 2.26127729e-01
-1.80041149e-01 3.45110923e-01 2.19313592e-01 8.92542481e-01
-5.08785486e-01 3.43414009e-01 1.95113242e-01 -3.23025078e-01
-1.06408119e+00 6.51522040e-01 -4.28530216e-01 3.63646746e-01
9.86606419e-01 -2.78359264e-01 -7.46214390e-01 -1.02589369e+00
-8.80501091e-01 5.34849703e-01 2.41920620e-01 3.79627556e-01
9.86746609e-01 -1.42409647e+00 -4.82285053e-01 -1.98995378e-02
1.53303996e-01 -2.81003624e-01 8.66983414e-01 1.10528040e+00
-5.94487131e-01 3.53282541e-01 -9.08726230e-02 -7.79037535e-01
-1.42891693e+00 7.96256661e-01 3.53308976e-01 -1.82401016e-01
-7.53076434e-01 1.04317510e+00 7.04674661e-01 -2.40963966e-01
3.99952918e-01 -2.40111053e-01 -9.39927548e-02 -3.84464376e-02
5.10343492e-01 2.80016392e-01 -2.10487321e-01 -7.02249050e-01
-2.22417548e-01 7.65161276e-01 9.41855237e-02 2.37375468e-01
1.21783245e+00 -5.11171877e-01 -5.83029985e-02 2.34416172e-01
1.17336249e+00 -1.19010091e-01 -1.41996026e+00 -2.20718130e-01
-4.36868727e-01 -5.23567200e-01 -1.04012229e-01 -6.85061038e-01
-1.33531618e+00 9.41851437e-01 1.30838409e-01 2.72809654e-01
1.38822043e+00 3.66506070e-01 9.06300128e-01 4.35407460e-01
1.46438316e-01 -7.81940639e-01 8.95200670e-01 5.59179306e-01
9.72670734e-01 -1.24968445e+00 -2.10641176e-01 -4.71148282e-01
-7.51104414e-01 1.01418364e+00 6.70211196e-01 4.60020900e-02
4.87397552e-01 1.11001469e-01 3.30222808e-02 -7.06205428e-01
-1.38944137e+00 -2.69253612e-01 5.97044230e-01 5.75132191e-01
2.27918759e-01 -2.19680265e-01 2.24299699e-01 2.86832929e-01
4.28218730e-02 -1.95734370e-02 2.14304596e-01 6.24412656e-01
-7.65693560e-03 -7.66905665e-01 -2.00093582e-01 3.75640273e-01
-2.51024693e-01 -1.18486203e-01 -2.87139148e-01 8.80199015e-01
-1.56634599e-02 1.14590836e+00 1.50017515e-01 -5.58252752e-01
2.86998957e-01 -1.07750870e-01 7.01242268e-01 -1.83781207e-01
-5.05764365e-01 -8.46042708e-02 2.72999436e-01 -1.22747898e+00
-6.03542089e-01 -5.26821911e-01 -1.08694446e+00 -6.85917616e-01
1.29616484e-01 -4.62164264e-03 -1.01223290e-02 9.78578508e-01
4.93910700e-01 7.78322101e-01 2.57897139e-01 -6.97339475e-01
-1.06846422e-01 -3.25425953e-01 -1.18821353e-01 5.91387212e-01
5.23793101e-01 -6.09280229e-01 -4.12742943e-01 3.42719227e-01] | [10.328262329101562, 1.0474095344543457] |
edf1ffbb-e48b-44de-a03c-32bc01537de9 | identifying-relevant-positions-in-proteins-by | 1503.03815 | null | http://arxiv.org/abs/1503.03815v2 | http://arxiv.org/pdf/1503.03815v2.pdf | Identifying relevant positions in proteins by Critical Variable Selection | Evolution in its course found a variety of solutions to the same optimisation
problem. The advent of high-throughput genomic sequencing has made available
extensive data from which, in principle, one can infer the underlying structure
on which biological functions rely. In this paper, we present a new method
aimed at extracting sites encoding structural and func- tional properties from
a set of protein primary sequences, namely a Multiple Sequence Alignment. The
method, called Critical Variable Selection, is based on the idea that subsets
of relevant sites cor- respond to subsequences that occur with a particularly
broad frequency distribution in the dataset. By applying this algorithm to in
silico sequences, to the Response Regulator Receiver and to the Voltage Sensor
Domain of Ion Channels, we show that this procedure recovers not only
information encoded in single site statistics and pairwise correlations but it
also captures dependencies going beyond pairwise correlations. The method
proposed here is complementary to Statistical Coupling Analysis, in that the
most relevant sites predicted by the two methods markedly differ. We find
robust and consistent results for datasets as small as few hundred sequences,
that reveal a hidden hierarchy of sites that is consistent with present
knowledge on biologically relevant sites and evo- lutionary dynamics. This
suggests that Critical Variable Selection is able to identify in a Multiple
Sequence Alignment a core of sites encoding functional and structural
information. | [] | 2016-01-19 | null | null | null | null | ['multiple-sequence-alignment'] | ['medical'] | [ 8.08556616e-01 -2.94378281e-01 -4.16218936e-02 -2.55797982e-01
-6.22550309e-01 -9.85054314e-01 3.83995801e-01 5.69944322e-01
-4.85840499e-01 1.28483725e+00 1.71380296e-01 -4.52052295e-01
-5.57014287e-01 -4.61233467e-01 -6.80995941e-01 -1.28576076e+00
-3.44628513e-01 7.45118380e-01 5.02718151e-01 -5.07147133e-01
5.82190633e-01 3.29254687e-01 -1.56722856e+00 4.45700914e-01
7.35419571e-01 4.77434516e-01 6.72288358e-01 7.49224961e-01
-1.08877562e-01 -1.48345038e-01 -4.20973092e-01 -2.29421165e-02
-2.53895186e-02 -9.58187342e-01 -7.98770249e-01 -3.87923837e-01
-1.62844695e-02 5.58739722e-01 3.66825610e-01 7.64162302e-01
6.94871485e-01 -1.82232767e-01 7.59879291e-01 -6.23695016e-01
-1.89491734e-01 4.56300855e-01 -3.71514916e-01 4.69421685e-01
5.65944850e-01 2.03255057e-01 1.52581680e+00 -4.91981208e-01
1.23429012e+00 9.72235858e-01 5.99851847e-01 3.50453675e-01
-1.81740201e+00 -6.93263188e-02 -2.32641637e-01 1.06840357e-01
-1.15440285e+00 -1.03300795e-01 4.16706413e-01 -4.86844778e-01
1.33022571e+00 5.23123860e-01 7.29498267e-01 1.08656085e+00
8.41816902e-01 5.04162647e-02 1.13149428e+00 -5.37430584e-01
4.25079703e-01 -3.97730112e-01 3.39242406e-02 2.84903556e-01
2.76267201e-01 -1.10574206e-02 -6.39179170e-01 -9.16408718e-01
3.19175273e-01 -2.03325942e-01 -3.37591439e-01 -3.57077807e-01
-1.26226974e+00 6.56823337e-01 -2.55924404e-01 5.53096771e-01
-4.17207837e-01 -1.19849429e-01 6.55131936e-01 3.82385969e-01
1.25938758e-01 4.98201668e-01 -1.08430517e+00 -2.91882187e-01
-5.34645259e-01 1.55361712e-01 1.05248535e+00 4.25834328e-01
9.48914409e-01 -6.27398908e-01 4.12548333e-01 6.53605044e-01
3.78959745e-01 1.97859928e-01 6.63850725e-01 -5.97236156e-01
9.71830785e-02 4.32987362e-01 -6.48014694e-02 -7.53501534e-01
-5.60180783e-01 -1.22894026e-01 -6.35688603e-01 -9.95523781e-02
4.85858530e-01 2.59944587e-03 -5.74993551e-01 2.00967002e+00
4.06120032e-01 -1.11015014e-01 -9.06734616e-02 7.11865246e-01
1.15380645e-01 4.51865405e-01 8.82999077e-02 -6.83082759e-01
1.37497914e+00 2.34285016e-02 -3.89107734e-01 1.54212549e-01
5.32097697e-01 -6.99647725e-01 6.45327866e-01 5.01786411e-01
-7.28859723e-01 -2.85093516e-01 -1.02070558e+00 3.17277879e-01
-1.96536765e-01 -7.80105114e-01 6.67361319e-01 4.95288521e-01
-1.02131617e+00 1.03323710e+00 -7.57978857e-01 -7.70411491e-01
-2.44549870e-01 5.52941501e-01 -4.84739691e-01 4.89005506e-01
-1.22624385e+00 8.85060072e-01 5.11926651e-01 -9.88938883e-02
-5.62304318e-01 -3.03071827e-01 -2.81087995e-01 -4.36532423e-02
1.18100785e-01 -6.26464427e-01 5.97968161e-01 -1.04999542e+00
-1.24649870e+00 8.40137124e-01 -4.49618101e-01 -2.50968844e-01
2.95569887e-04 2.53019959e-01 -2.66998857e-01 1.38628468e-01
-6.73708320e-02 1.21049374e-01 4.33628917e-01 -7.33750761e-01
-3.08732897e-01 -6.29244268e-01 -6.23285890e-01 -1.67878438e-02
2.43277431e-01 3.02756041e-01 -3.63039635e-02 -3.98550421e-01
3.65080178e-01 -1.00174701e+00 -6.47603869e-01 -3.54899168e-01
-1.42557114e-01 -1.44533440e-01 1.75648496e-01 -5.46172738e-01
1.20841491e+00 -1.85016048e+00 9.67832029e-01 6.58018529e-01
2.77880915e-02 -1.63899809e-02 -9.09700617e-02 1.25091374e+00
-4.81190056e-01 2.61477023e-01 -6.42133892e-01 6.14612043e-01
-2.81161755e-01 5.19450068e-01 -7.33180791e-02 8.20553780e-01
1.70707613e-01 6.01200342e-01 -7.53525615e-01 -1.76984414e-01
-8.97675157e-02 4.95908082e-01 -7.43836939e-01 4.38728891e-02
-4.01850581e-01 6.28259003e-01 -6.36742771e-01 4.07651365e-01
4.53211069e-01 -2.87733048e-01 1.00297570e+00 2.86315262e-01
-3.17298412e-01 5.14457524e-01 -1.02944505e+00 1.78072596e+00
2.32290030e-01 4.57103223e-01 -1.08487442e-01 -1.26070988e+00
1.03576720e+00 2.97523677e-01 7.55293071e-01 -5.65963984e-01
-5.09132594e-02 3.72437745e-01 6.08225942e-01 -5.42380393e-01
6.38497919e-02 -3.26789856e-01 -1.00311756e-01 3.11922252e-01
2.12202191e-01 2.84801573e-01 1.18830502e-01 5.50834946e-02
1.36425769e+00 3.99726868e-01 8.62091660e-01 -6.77451909e-01
5.34058392e-01 3.03963255e-02 8.99656653e-01 5.69817007e-01
-1.81563631e-01 5.33371091e-01 8.29058945e-01 -3.26481253e-01
-1.48685086e+00 -7.22154558e-01 -4.64346379e-01 1.01888394e+00
-6.28202483e-02 -7.89037824e-01 -7.09877193e-01 -1.55917965e-02
-5.65488217e-03 1.34365186e-01 -7.33398914e-01 -2.34118551e-01
-6.31791890e-01 -1.08659136e+00 5.26587129e-01 -3.16053629e-02
-2.49958858e-01 -1.05050921e+00 -7.61163056e-01 6.05070829e-01
-1.11188173e-01 -3.82726610e-01 -4.22053672e-02 9.55771446e-01
-1.04893064e+00 -1.09366035e+00 -4.43963915e-01 -3.35745066e-01
1.88181579e-01 -8.52230191e-02 1.01336646e+00 1.81512117e-01
-7.43377864e-01 -1.37618601e-01 -3.12108576e-01 -1.54485971e-01
-4.64824945e-01 7.29811266e-02 2.06391767e-01 -2.86348999e-01
5.65615475e-01 -1.10607803e+00 -5.02139270e-01 4.77562845e-01
-9.03662205e-01 -4.49468732e-01 7.11743414e-01 1.06450868e+00
6.86506093e-01 -1.86867833e-01 4.60697442e-01 -8.92513454e-01
4.12688375e-01 -6.20381892e-01 -4.42010134e-01 2.55645812e-01
-5.86193740e-01 5.90495348e-01 5.68671405e-01 -8.02468434e-02
-8.03312480e-01 2.58409798e-01 -3.97269011e-01 4.48079109e-01
-4.13008422e-01 6.74859643e-01 -3.36905897e-01 1.25223130e-01
8.32751393e-01 5.83363533e-01 1.03281178e-01 -5.82478404e-01
1.77999735e-02 4.45222080e-01 1.44000977e-01 -6.70400202e-01
2.52153546e-01 4.30521369e-01 3.46561611e-01 -1.16329885e+00
-7.32207894e-02 -7.89201558e-01 -9.89382803e-01 8.24819580e-02
7.83160269e-01 -4.08766478e-01 -9.54688191e-01 2.34328926e-01
-8.34828734e-01 -8.59197676e-02 3.42791826e-02 3.09432060e-01
-9.28666413e-01 7.40739167e-01 -4.42871660e-01 -8.08408320e-01
6.57421425e-02 -9.74963248e-01 1.23734045e+00 -1.87638864e-01
-5.57117105e-01 -9.56960201e-01 9.00026560e-01 3.02223060e-02
1.72154494e-02 2.89823204e-01 1.08599532e+00 -8.04584086e-01
-3.61724854e-01 2.59750903e-01 4.82833505e-01 -1.25638753e-01
8.48548114e-02 3.46946299e-01 -5.30624747e-01 -3.30810577e-01
-8.89655948e-02 5.01300162e-03 1.03422320e+00 2.88786113e-01
5.93193114e-01 1.77184671e-01 -5.23565054e-01 5.20661294e-01
1.69512820e+00 3.55362862e-01 8.21748197e-01 4.39925730e-01
2.54666686e-01 8.40321720e-01 5.60579836e-01 4.51748729e-01
-4.50365067e-01 1.01037109e+00 2.52414584e-01 3.01372766e-01
5.96965969e-01 -8.03451166e-02 3.63218129e-01 6.46363497e-01
-5.21887720e-01 -1.50139809e-01 -1.02018332e+00 3.20027500e-01
-2.01748872e+00 -1.29286218e+00 -3.20916623e-01 2.28621769e+00
1.13089347e+00 1.13940658e-02 3.25653195e-01 -8.65180604e-03
7.12977648e-01 -1.19109556e-01 -6.12323880e-01 -6.74162209e-01
-5.11419773e-01 4.55736399e-01 7.26843894e-01 4.75967467e-01
-5.98783195e-01 5.28352380e-01 7.37681341e+00 6.40221536e-01
-7.71605968e-01 -2.80158639e-01 2.30570897e-01 9.91352871e-02
-5.69980264e-01 4.42748487e-01 -8.12645733e-01 5.93041718e-01
1.41224277e+00 1.07202365e-03 2.14003697e-01 4.99180347e-01
4.46390301e-01 -1.50305152e-01 -8.16449881e-01 4.65568691e-01
-2.62963623e-01 -1.35457635e+00 -1.79635778e-01 5.34452140e-01
4.02943850e-01 1.83821023e-01 -3.51822168e-01 -5.75970948e-01
3.59102190e-01 -1.02053881e+00 1.95456713e-01 5.49831092e-01
3.81295711e-01 -8.46642852e-01 5.47150910e-01 4.47486281e-01
-9.83656347e-01 1.25357732e-01 -6.37959957e-01 -1.68804720e-01
2.17449218e-01 6.84859872e-01 -7.75292695e-01 5.63272476e-01
6.06820166e-01 7.33538449e-01 -2.65656978e-01 9.54881847e-01
-1.56558648e-01 7.22549558e-01 -3.27246279e-01 -2.66069800e-01
5.10375351e-02 -6.28244102e-01 6.95235312e-01 1.31142378e+00
1.27589554e-01 1.70040354e-01 -1.86617732e-01 5.06560326e-01
6.29953504e-01 2.74242640e-01 -6.41623676e-01 -5.99714220e-02
2.38022313e-01 1.04376853e+00 -9.15620029e-01 -7.69394934e-02
-2.96100765e-01 1.08780146e+00 4.59713310e-01 2.16895074e-01
-6.24056041e-01 -2.89525211e-01 1.01433516e+00 -1.93507910e-01
5.01640856e-01 -3.30317140e-01 4.08501215e-02 -9.86912131e-01
-7.84698501e-02 -1.04487145e+00 5.25029778e-01 -3.83379161e-01
-1.12391865e+00 2.58487344e-01 -2.17036128e-01 -7.84791648e-01
-4.87315893e-01 -6.78015351e-01 -1.91229507e-01 1.25323641e+00
-9.60267425e-01 -4.05783594e-01 4.64315176e-01 2.74790525e-01
3.54142606e-01 7.92027637e-02 9.29543853e-01 5.84580861e-02
-5.25005400e-01 6.08459450e-02 7.52251923e-01 -5.73246300e-01
7.05113292e-01 -1.34690630e+00 4.37150151e-01 3.86067063e-01
1.20286807e-01 1.16104686e+00 1.21104610e+00 -8.87700438e-01
-1.51227582e+00 -4.02880192e-01 1.05995381e+00 -4.81481701e-01
7.14610100e-01 -4.75942940e-01 -1.09407032e+00 4.90573853e-01
1.96976960e-01 -5.73119998e-01 1.20722580e+00 2.86462367e-01
-2.13369474e-01 4.45178330e-01 -7.90247858e-01 3.54150414e-01
1.26043069e+00 -5.25988340e-01 -7.73200393e-01 2.83944547e-01
4.67583030e-01 1.54491901e-01 -8.78978372e-01 1.62455052e-01
8.01420927e-01 -1.17993069e+00 8.94388914e-01 -1.17704582e+00
2.32842371e-01 -3.63511175e-01 -3.98237497e-01 -1.20388496e+00
-6.65278792e-01 -8.71786118e-01 5.53658009e-01 9.29812193e-01
7.97993302e-01 -7.70365596e-01 4.91153628e-01 9.42792818e-02
-8.96368474e-02 -4.40094173e-01 -1.27777302e+00 -6.84363782e-01
1.69884283e-02 -6.54080957e-02 1.49845764e-01 9.86207664e-01
3.76192540e-01 5.06016254e-01 -2.22254470e-01 -7.58976117e-02
4.25585359e-01 2.22699732e-01 4.28524971e-01 -1.39826798e+00
-8.21323931e-01 -2.80615866e-01 -6.86435640e-01 -7.44488418e-01
2.53366325e-02 -7.85595655e-01 5.63864596e-02 -9.99853909e-01
5.21938324e-01 -1.54962718e-01 -5.00507474e-01 1.95057929e-01
-1.54575318e-01 -2.04302147e-02 -4.80979741e-01 3.30219924e-01
-2.51686364e-01 1.84518054e-01 7.30074286e-01 3.32976758e-01
-1.99110210e-01 -3.05626839e-01 -4.77194458e-01 4.88556653e-01
6.73925757e-01 -5.71645260e-01 -1.11499295e-01 2.75049835e-01
7.94317782e-01 2.93903742e-02 6.05195090e-02 -4.65536863e-01
-4.58695218e-02 -3.85396302e-01 2.65641183e-01 -3.05810422e-01
1.01558462e-01 -7.68399894e-01 7.32753932e-01 8.19923341e-01
-4.84053791e-01 1.97513729e-01 8.03638473e-02 8.26312125e-01
3.76599915e-02 -3.71961802e-01 7.03273833e-01 -3.14559132e-01
-8.09540510e-01 -2.07947761e-01 -7.56941378e-01 -2.33067989e-01
9.47026849e-01 -3.04671764e-01 1.30914373e-03 2.92846151e-02
-9.10728693e-01 -3.35338503e-01 7.16517568e-01 1.24855436e-01
2.92325228e-01 -8.09824705e-01 -6.86807036e-01 2.13039368e-01
2.11189702e-01 -7.48681784e-01 1.08530961e-01 9.10281658e-01
-6.19279563e-01 6.76014006e-01 -4.91159499e-01 -8.77035916e-01
-1.62392163e+00 7.15302527e-01 2.11737975e-01 5.12593612e-02
-4.65071142e-01 6.41295850e-01 7.26526454e-02 -3.47618043e-01
-5.44270158e-01 -2.08155420e-02 -1.95555151e-01 3.12847346e-01
2.85359621e-01 -8.52687657e-03 4.24978882e-02 -7.98126936e-01
-6.10713184e-01 7.38944113e-01 2.29172583e-04 -1.51162103e-01
1.53059196e+00 -2.52582461e-01 -6.38920188e-01 6.78674519e-01
1.19819212e+00 3.75895016e-02 -8.76776993e-01 1.49685249e-01
3.95440847e-01 -3.60907227e-01 -5.66885352e-01 -5.58803499e-01
-4.18569952e-01 5.28849483e-01 3.14563513e-01 1.92663804e-01
1.04895639e+00 2.62565434e-01 2.45399043e-01 5.41658103e-01
6.18581653e-01 -9.32285309e-01 -3.63712609e-01 3.88848513e-01
4.80311215e-01 -8.53790402e-01 -3.94180976e-02 -3.49916071e-01
-2.81992316e-01 1.07317245e+00 2.78520957e-02 -4.05438989e-02
2.73529381e-01 2.24946991e-01 -2.73037046e-01 -3.11246574e-01
-1.32660401e+00 -2.73284525e-01 -2.11841259e-02 5.11235356e-01
8.10096443e-01 -3.79203372e-02 -1.18529856e+00 -2.17227228e-02
1.92118846e-02 -4.25823063e-01 3.47181499e-01 9.58666384e-01
-8.14178824e-01 -1.75094628e+00 -2.05437645e-01 1.79380745e-01
-5.65548837e-01 -2.65124023e-01 -9.77565825e-01 7.12632179e-01
-9.56732184e-02 7.00823843e-01 -9.40735117e-02 -2.62277573e-01
1.01362579e-01 4.04443026e-01 4.88592714e-01 -4.81749415e-01
-6.76715434e-01 4.71381932e-01 2.53360122e-01 -5.97463965e-01
-4.78337854e-01 -1.09851539e+00 -1.35168302e+00 -5.62759414e-02
-6.06559336e-01 6.88217044e-01 7.64197111e-01 8.74199688e-01
4.43287730e-01 5.01858830e-01 4.68621969e-01 -5.75725853e-01
-5.60437024e-01 -7.30105221e-01 -8.09729338e-01 3.93026739e-01
1.57647029e-01 -4.09362763e-01 -3.82045120e-01 -6.87107863e-03] | [4.871285438537598, 5.213534355163574] |
9ee703d8-7347-4e9f-9a0a-f96e173085ac | modeling-long-and-short-term-temporal | 1703.07015 | null | http://arxiv.org/abs/1703.07015v3 | http://arxiv.org/pdf/1703.07015v3.pdf | Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networks | Multivariate time series forecasting is an important machine learning problem
across many domains, including predictions of solar plant energy output,
electricity consumption, and traffic jam situation. Temporal data arise in
these real-world applications often involves a mixture of long-term and
short-term patterns, for which traditional approaches such as Autoregressive
models and Gaussian Process may fail. In this paper, we proposed a novel deep
learning framework, namely Long- and Short-term Time-series network (LSTNet),
to address this open challenge. LSTNet uses the Convolution Neural Network
(CNN) and the Recurrent Neural Network (RNN) to extract short-term local
dependency patterns among variables and to discover long-term patterns for time
series trends. Furthermore, we leverage traditional autoregressive model to
tackle the scale insensitive problem of the neural network model. In our
evaluation on real-world data with complex mixtures of repetitive patterns,
LSTNet achieved significant performance improvements over that of several
state-of-the-art baseline methods. All the data and experiment codes are
available online. | ['Yiming Yang', 'Wei-Cheng Chang', 'Hanxiao Liu', 'Guokun Lai'] | 2017-03-21 | null | null | null | null | ['univariate-time-series-forecasting'] | ['time-series'] | [ 5.69204465e-02 -5.70930064e-01 1.59654282e-02 -3.03460270e-01
-3.78827512e-01 -4.08037871e-01 8.65194559e-01 -1.44156426e-01
1.54282168e-01 6.56914651e-01 2.03998685e-01 -6.90945864e-01
-3.49442810e-01 -8.53483677e-01 -6.96196198e-01 -8.43924284e-01
-4.44851220e-01 8.18084031e-02 -4.84271646e-02 -1.29606098e-01
5.62062077e-02 5.97645521e-01 -1.25001705e+00 5.58402985e-02
7.79605508e-01 1.35150230e+00 1.35254845e-01 6.70575976e-01
-2.95595080e-01 1.17628074e+00 -6.00175917e-01 2.47490957e-01
1.26971111e-01 -1.02354325e-01 -2.22391963e-01 -1.78871080e-01
-3.08733165e-01 1.22821815e-02 -7.07407773e-01 7.70727873e-01
4.27853793e-01 4.29483831e-01 7.78688371e-01 -1.27261627e+00
-9.54956055e-01 5.68330050e-01 -7.43974626e-01 5.57287514e-01
-4.00452375e-01 3.93405527e-01 4.40989316e-01 -7.81576872e-01
-1.65265366e-01 1.11868429e+00 8.89866054e-01 -1.12647921e-01
-8.96158218e-01 -8.06354284e-01 4.26561594e-01 2.26357982e-01
-1.06584167e+00 -1.95622101e-01 1.16169131e+00 -5.84764600e-01
1.28738546e+00 2.73342896e-02 2.67940193e-01 1.24245584e+00
6.40963316e-01 5.85736632e-01 9.97850716e-01 -1.45553380e-01
2.15608165e-01 -5.34826875e-01 1.15217663e-01 8.34001154e-02
-3.33628774e-01 3.30807239e-01 -2.05704868e-02 -1.56895891e-01
8.14366460e-01 6.23855233e-01 1.63477436e-01 3.09924126e-01
-1.06876099e+00 6.12928092e-01 5.82086384e-01 6.77740276e-01
-9.23058748e-01 2.15342849e-01 5.61821997e-01 3.09917748e-01
8.70760083e-01 7.68408626e-02 -8.78642499e-01 -2.32695922e-01
-7.91631699e-01 -3.65355276e-02 7.99364269e-01 7.48098969e-01
2.73602843e-01 6.79069757e-01 -2.36442685e-01 8.20650339e-01
-3.15535464e-03 5.97636998e-01 1.05064094e+00 -5.02249360e-01
5.21048546e-01 5.23658454e-01 -1.48672871e-02 -1.13325942e+00
-6.53814256e-01 -7.00270832e-01 -1.68728757e+00 -2.79535837e-02
1.78646177e-01 -5.49498320e-01 -1.09982097e+00 1.48353660e+00
-9.46201663e-03 7.51161516e-01 2.03763723e-01 4.80369180e-01
4.41038936e-01 1.25227010e+00 1.46286458e-01 -5.50789416e-01
9.05636132e-01 -1.12983143e+00 -9.02095616e-01 -1.49805084e-01
2.38418967e-01 -7.45782852e-01 6.89652264e-01 1.17493913e-01
-7.60828316e-01 -6.95261657e-01 -7.13632643e-01 1.68564379e-01
-7.79236197e-01 3.29811424e-01 6.25419557e-01 2.88096387e-02
-8.31883848e-01 9.61336970e-01 -1.12466443e+00 -2.48969778e-01
1.29076302e-01 1.20977506e-01 1.42154068e-01 2.87778467e-01
-1.07867634e+00 6.31360054e-01 3.03810865e-01 7.39253938e-01
-7.94716656e-01 -7.93959081e-01 -6.59863353e-01 2.73507565e-01
2.52474904e-01 -3.74265730e-01 1.48717892e+00 -9.30209875e-01
-1.72857738e+00 -9.53852758e-02 -4.96473938e-01 -5.19515872e-01
1.45615086e-01 -1.36127830e-01 -1.04560649e+00 -5.92060030e-01
-2.26910442e-01 -4.07252222e-01 8.18326294e-01 -5.18986464e-01
-5.59019089e-01 -4.05235827e-01 -6.63665771e-01 -2.37109810e-01
-3.29680473e-01 8.11435729e-02 9.37063694e-02 -8.81180227e-01
2.54579112e-02 -8.99772704e-01 -6.00180864e-01 -5.90813458e-01
-3.84007663e-01 -7.70283759e-01 1.23340619e+00 -8.36804152e-01
1.37200356e+00 -2.25325298e+00 -2.14148089e-01 1.29916504e-01
-2.06329197e-01 2.40110904e-01 -9.81124341e-02 6.40646577e-01
-6.69215262e-01 1.58996526e-02 -1.50743335e-01 -2.32956916e-01
5.35908192e-02 2.82994270e-01 -8.97042990e-01 4.54689622e-01
3.15354347e-01 1.05108452e+00 -7.40053415e-01 2.07725883e-01
3.53942335e-01 4.32207853e-01 5.24969816e-01 2.19386160e-01
-4.17695731e-01 3.71634573e-01 -3.07025313e-01 5.73949993e-01
5.66750467e-01 -5.24000943e-01 -1.83432847e-01 -8.06802362e-02
-5.17337918e-01 4.75828461e-02 -7.46570647e-01 1.26811206e+00
-6.17369056e-01 7.64982760e-01 -5.23575604e-01 -1.18200636e+00
1.12328243e+00 5.11075616e-01 7.36721694e-01 -7.95692861e-01
5.91041595e-02 2.02703595e-01 -1.51591003e-01 -5.20750403e-01
3.00632447e-01 -6.73855245e-02 7.10210949e-02 1.69604972e-01
-1.49645507e-01 2.49799743e-01 -4.33939360e-02 -5.26836693e-01
1.10455513e+00 -3.49445157e-02 2.62114018e-01 -8.88403505e-02
3.12186956e-01 -2.76384592e-01 8.00985515e-01 4.17178065e-01
6.64705411e-02 3.35755587e-01 3.57478291e-01 -8.89427722e-01
-1.03970397e+00 -7.38256037e-01 1.49027333e-01 1.12789941e+00
-4.51579392e-01 -1.19205885e-01 2.06520051e-01 -3.40832710e-01
1.86558347e-02 9.66372192e-01 -5.25220633e-01 2.63275858e-02
-6.50633335e-01 -1.07160604e+00 3.10870618e-01 9.87955809e-01
3.03600937e-01 -1.37368643e+00 -1.69121742e-01 5.93549490e-01
1.80745035e-01 -9.10719633e-01 -3.28123599e-01 4.54810262e-01
-9.24742937e-01 -7.65202045e-01 -7.45203733e-01 -5.08365810e-01
1.35530636e-01 1.27883747e-01 1.09333181e+00 -7.75873184e-01
-8.89474899e-03 -2.72164881e-01 -9.30568427e-02 -8.72613490e-01
-5.16982703e-03 1.25077292e-01 -4.06490117e-02 -2.98126433e-02
4.75980431e-01 -1.22928059e+00 -7.27741539e-01 2.27033481e-01
-8.01044285e-01 -2.81611562e-01 5.77239633e-01 6.83172226e-01
6.04567468e-01 6.74785495e-01 8.17150295e-01 -5.29465675e-01
8.24172497e-01 -9.38825250e-01 -8.59926462e-01 2.50253856e-01
-7.06597149e-01 -1.95912138e-01 1.21186364e+00 -7.72308409e-01
-9.58247602e-01 -1.18941493e-01 7.97597468e-02 -8.92727911e-01
-1.42609596e-01 1.03493011e+00 2.53809959e-01 4.34677362e-01
3.13718259e-01 4.79661882e-01 -4.04243082e-01 -5.66460788e-01
1.08431302e-01 6.12732649e-01 7.41757095e-01 -2.07857445e-01
7.44639277e-01 2.49070078e-01 1.41782194e-01 -6.75959706e-01
-7.06775546e-01 -4.97211128e-01 -4.80483621e-01 -2.98993643e-02
6.35969579e-01 -1.14574301e+00 -7.26517916e-01 7.94874012e-01
-1.28961921e+00 -5.05702615e-01 -3.61348480e-01 5.10560215e-01
-4.16474223e-01 -1.58170998e-01 -6.35849059e-01 -1.18141758e+00
-5.28919101e-01 -5.68326592e-01 9.49474037e-01 5.37570119e-01
5.99768050e-02 -1.29892743e+00 1.28729463e-01 -4.88996863e-01
9.39917386e-01 6.64625287e-01 1.00988698e+00 -8.20164621e-01
-2.27031693e-01 -3.60215604e-01 -1.63008645e-01 3.64837587e-01
4.27287728e-01 3.48139882e-01 -9.51679945e-01 -2.18190730e-01
2.21233711e-01 1.56018123e-01 7.01509714e-01 7.96447754e-01
1.65965116e+00 -3.51666689e-01 -2.82605529e-01 7.44628847e-01
1.23274457e+00 7.29272604e-01 5.39038837e-01 1.09688148e-01
8.01541686e-01 4.15577352e-01 8.13132748e-02 5.17291009e-01
3.92914474e-01 1.05561398e-01 2.44732961e-01 -1.88177362e-01
5.75022697e-01 -2.52870411e-01 3.61797243e-01 1.31106794e+00
-7.75737315e-02 -2.61863679e-01 -1.04240859e+00 6.52437568e-01
-2.24189806e+00 -1.17623699e+00 -2.27487698e-01 1.84689200e+00
1.99827492e-01 1.45545810e-01 1.37879416e-01 -5.37212081e-02
6.32184803e-01 4.57080901e-01 -1.11790657e+00 -2.71024078e-01
-2.90994883e-01 3.16444077e-02 6.01244509e-01 -1.34956330e-01
-1.34780586e+00 4.86558437e-01 6.51194811e+00 8.00817788e-01
-1.65056860e+00 8.43810756e-03 9.32018876e-01 1.16004191e-01
-7.59983137e-02 -3.68123621e-01 -5.51142812e-01 7.06342578e-01
1.56488466e+00 -4.72379625e-01 4.28172588e-01 9.12684262e-01
6.77766383e-01 3.49512875e-01 -8.33056211e-01 1.01051736e+00
-2.69258410e-01 -1.21967471e+00 -4.59904343e-01 -1.07718565e-01
9.39298153e-01 6.32275224e-01 2.77913302e-01 7.43632555e-01
6.52592480e-01 -1.36389649e+00 1.08128130e-01 9.86077845e-01
3.94025087e-01 -7.90010691e-01 7.89045095e-01 5.57402015e-01
-1.65111077e+00 -5.41855872e-01 -3.63278896e-01 -1.62706882e-01
1.60903424e-01 1.09357107e+00 -4.00822490e-01 8.31485391e-01
7.81898916e-01 1.28064334e+00 -2.40557790e-01 9.31697369e-01
1.76971048e-01 9.78311181e-01 -6.19814754e-01 7.25643784e-02
4.01393801e-01 -3.92915189e-01 4.51183975e-01 9.85840678e-01
7.62942016e-01 -1.07610799e-01 2.97648787e-01 7.22320676e-01
9.79444385e-02 -1.57419324e-01 -8.01949680e-01 -5.47740579e-01
2.96250522e-01 1.19777548e+00 -5.42920589e-01 -2.44172171e-01
-4.93830651e-01 7.18368053e-01 8.28979239e-02 8.10209274e-01
-7.79623270e-01 -3.57717007e-01 5.16054392e-01 -2.62342989e-01
5.42038679e-01 -4.34080899e-01 -2.57975250e-01 -1.05030441e+00
2.66814977e-01 -5.58960497e-01 5.57820797e-01 -9.92745399e-01
-1.97155750e+00 7.55136609e-01 -1.87330857e-01 -1.24316525e+00
-6.66410089e-01 -5.31658411e-01 -1.35161161e+00 1.29505253e+00
-1.62073815e+00 -1.26911688e+00 -2.08860874e-01 6.34097457e-01
9.80673730e-01 -3.27469230e-01 6.71500623e-01 1.03602469e-01
-8.40551794e-01 1.49416793e-02 8.34800184e-01 -2.97507048e-02
4.02664125e-01 -1.24827266e+00 7.81585693e-01 9.25115466e-01
-3.14262986e-01 4.91256237e-01 5.84231079e-01 -6.38252556e-01
-1.24241126e+00 -1.66010892e+00 7.91426659e-01 -3.82679850e-02
1.15798783e+00 1.72944134e-03 -1.12501860e+00 8.20133269e-01
3.76324862e-01 1.50767908e-01 4.68933642e-01 2.43567914e-01
-9.39035490e-02 -3.89461905e-01 -6.39340281e-01 3.72552067e-01
5.84521234e-01 -5.55428505e-01 -1.53062880e-01 4.55384433e-01
8.94082427e-01 -1.68004692e-01 -9.32370901e-01 7.08546042e-01
4.92626429e-01 -5.59830129e-01 7.06762016e-01 -7.50419915e-01
6.35481417e-01 -2.46193171e-01 -2.03178868e-01 -1.65100610e+00
-6.80114746e-01 -9.12289262e-01 -6.28779233e-01 1.23895717e+00
3.55100542e-01 -7.92140305e-01 4.39862520e-01 5.34632921e-01
-3.10902625e-01 -8.63553345e-01 -9.16209817e-01 -7.98151433e-01
2.31471479e-01 -4.43503320e-01 9.71645355e-01 9.72398221e-01
-3.52493703e-01 3.20430785e-01 -5.22932827e-01 4.44515347e-01
1.48518458e-01 4.50194567e-01 5.45394421e-01 -1.10747838e+00
-1.31741077e-01 -6.13728285e-01 -1.56660065e-01 -8.93882930e-01
2.57871836e-01 -4.20527726e-01 5.84626347e-02 -1.39280367e+00
-3.01224887e-01 -5.92261478e-02 -8.99103820e-01 2.14048788e-01
-1.23360343e-01 -4.82858926e-01 -5.92988767e-02 2.62654215e-01
-2.65878618e-01 9.96173501e-01 7.91881263e-01 -1.96184367e-01
-3.92819405e-01 5.36943078e-01 -2.82178938e-01 7.14462638e-01
1.01835406e+00 -2.51569331e-01 -5.48481464e-01 -2.57202953e-01
-2.14579385e-02 3.50317150e-01 3.28043073e-01 -9.76381063e-01
6.33631289e-01 -4.69079256e-01 7.37741411e-01 -1.18447530e+00
2.05940276e-01 -1.05191946e+00 4.77419347e-01 1.20364770e-01
-2.22588494e-01 6.92330241e-01 3.97246748e-01 8.25322807e-01
-2.81931251e-01 3.99665296e-01 2.13542998e-01 3.80176008e-02
-7.29654670e-01 6.02195799e-01 -4.90801394e-01 -5.93667030e-01
1.04419804e+00 1.79677501e-01 -5.48791826e-01 -4.71357107e-01
-6.04844689e-01 5.12785554e-01 -3.59534115e-01 6.05611026e-01
3.95248771e-01 -1.45007694e+00 -5.91621280e-01 2.07987815e-01
-1.40215531e-01 2.35448986e-01 4.23018217e-01 9.69510555e-01
7.74314404e-02 4.69447106e-01 1.04724161e-01 -5.74966192e-01
-7.68869579e-01 8.64839017e-01 5.76395392e-01 -5.83031118e-01
-6.44282162e-01 4.03220445e-01 1.16383158e-01 -4.64137346e-01
1.42473787e-01 -7.72695541e-01 -3.54619950e-01 1.68490224e-02
4.80809361e-01 4.63560998e-01 -6.92004198e-03 -3.57066721e-01
-5.94613068e-02 3.13882321e-01 6.30086362e-02 3.38210046e-01
1.86286521e+00 -2.02730417e-01 -3.20034087e-01 1.31548786e+00
1.24644375e+00 -5.92430949e-01 -1.22622693e+00 -4.64612812e-01
2.69499183e-01 4.96194325e-02 1.13417216e-01 -6.85695529e-01
-1.06212234e+00 7.78968573e-01 6.58529878e-01 6.81616247e-01
1.41285610e+00 -3.94541144e-01 8.83028269e-01 4.05499667e-01
-5.09348139e-02 -9.84440506e-01 -2.23476976e-01 9.59126770e-01
9.44219053e-01 -1.22273624e+00 -4.61973011e-01 8.66818130e-02
-3.53656888e-01 1.46608341e+00 5.46396077e-01 -3.94599229e-01
1.20410502e+00 4.30165380e-01 4.21942677e-03 -1.23403296e-01
-1.28255582e+00 -3.10110059e-02 3.18694502e-01 2.29573458e-01
5.11039853e-01 3.30311447e-01 3.32289934e-01 8.06796074e-01
-2.14387327e-01 2.90087461e-01 -2.91771838e-03 7.62498260e-01
3.11925430e-02 -4.88776118e-01 -2.94467360e-01 7.36538053e-01
-4.41426188e-01 6.35900721e-02 2.73023788e-02 4.44923282e-01
-1.97691217e-01 9.99166906e-01 4.58975792e-01 -3.63026142e-01
5.17376602e-01 2.25762591e-01 -3.36033136e-01 -3.47082093e-02
-5.63484311e-01 4.26691473e-01 -3.44231606e-01 -4.67166603e-01
-3.38430047e-01 -5.42150795e-01 -7.80271590e-01 -4.48179156e-01
-1.15300857e-01 -1.77100748e-01 7.26526558e-01 1.13529789e+00
4.51854855e-01 1.23244619e+00 1.06228232e+00 -9.07382786e-01
-6.56827211e-01 -1.40828538e+00 -5.64032137e-01 2.52171755e-01
6.24636233e-01 -2.29341924e-01 -2.45746374e-01 1.20009802e-01] | [6.862717628479004, 2.9417078495025635] |
f020f776-359d-4f3b-9ffc-bae955c88fa0 | sooner-than-expected-hitting-the-wall-of | 1609.07722 | null | http://arxiv.org/abs/1609.07722v1 | http://arxiv.org/pdf/1609.07722v1.pdf | Sooner than Expected: Hitting the Wall of Complexity in Evolution | In evolutionary robotics an encoding of the control software, which maps
sensor data (input) to motor control values (output), is shaped by stochastic
optimization methods to complete a predefined task. This approach is assumed to
be beneficial compared to standard methods of controller design in those cases
where no a-priori model is available that could help to optimize performance.
Also for robots that have to operate in unpredictable environments, an
evolutionary robotics approach is favorable. We demonstrate here that such a
model-free approach is not a free lunch, as already simple tasks can represent
unsolvable barriers for fully open-ended uninformed evolutionary computation
techniques. We propose here the 'Wankelmut' task as an objective for an
evolutionary approach that starts from scratch without pre-shaped controller
software or any other informed approach that would force the behavior to be
evolved in a desired way. Our focal claim is that 'Wankelmut' represents the
simplest set of problems that makes plain-vanilla evolutionary computation
fail. We demonstrate this by a series of simple standard evolutionary
approaches using different fitness functions and standard artificial neural
networks as well as continuous-time recurrent neural networks. All our tested
approaches failed. We claim that any other evolutionary approach will also fail
that does per-se not favor or enforce modularity and does not freeze or protect
already evolved functionalities. Thus we propose a hard-to-pass benchmark and
make a strong statement for self-complexifying and generative approaches in
evolutionary computation. We anticipate that defining such a 'simplest task to
fail' is a valuable benchmark for promoting future development in the field of
artificial intelligence, evolutionary robotics and artificial life. | ['Thomas Schmickl', 'Payam Zahadat', 'Heiko Hamann'] | 2016-09-25 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [ 5.06375372e-01 3.25082868e-01 5.68697035e-01 -1.15270108e-01
2.08055004e-01 -5.01263678e-01 7.10595369e-01 -2.38521978e-01
-4.29068297e-01 9.23916399e-01 -3.77680600e-01 -2.65402466e-01
-6.01695001e-01 -9.48240101e-01 -7.68940330e-01 -8.15171421e-01
-1.48524180e-01 4.70443070e-01 3.22771341e-01 -1.00165677e+00
3.26052636e-01 4.91836637e-01 -2.08104157e+00 -1.57187685e-01
8.57451618e-01 5.15105784e-01 4.69321489e-01 7.10722804e-01
2.94821829e-01 3.66222471e-01 -6.47566378e-01 -1.98544085e-01
2.58973688e-01 -8.49079072e-01 -6.55622244e-01 -1.80340901e-01
-6.03713810e-01 7.08833694e-01 2.47142553e-01 9.56985295e-01
5.89081347e-01 1.24558814e-01 7.92663276e-01 -1.16283560e+00
-6.30432367e-01 6.44354641e-01 1.42213225e-01 -3.97099584e-01
2.13120401e-01 5.28140247e-01 4.94394004e-01 -2.83914834e-01
9.45464313e-01 8.38131011e-01 7.50512958e-01 8.62825096e-01
-1.09681118e+00 -1.85683921e-01 -1.81696713e-01 -8.32366347e-02
-1.24211478e+00 -1.54587805e-01 6.26412272e-01 -3.32592428e-01
1.00164294e+00 4.90692794e-01 1.10680473e+00 1.18633449e+00
5.26349366e-01 3.55911434e-01 1.07043540e+00 -6.43241286e-01
7.12436795e-01 2.39882126e-01 -4.42537278e-01 5.50260484e-01
6.78494275e-01 6.52114749e-01 -1.79315478e-01 8.05330127e-02
3.91257614e-01 -3.84970963e-01 -2.60597795e-01 -5.88776469e-01
-1.15931511e+00 5.87331533e-01 1.99167773e-01 6.37592673e-01
-4.41202939e-01 3.49381894e-01 3.02025229e-01 6.02280200e-01
-3.66435438e-01 1.25554967e+00 -7.01024652e-01 -3.41203868e-01
-4.80168283e-01 4.50053364e-01 1.03548539e+00 5.46724975e-01
4.83925819e-01 5.15017927e-01 3.46090943e-01 4.50988054e-01
1.10730149e-01 9.01078284e-02 7.47656941e-01 -1.11042094e+00
-4.33629721e-01 8.13974619e-01 5.67607880e-02 -7.87618876e-01
-4.28716123e-01 -6.51605964e-01 -5.39468646e-01 1.07445097e+00
1.85861185e-01 -4.61839139e-01 -7.18274295e-01 1.74115896e+00
2.81979501e-01 -3.92602801e-01 2.83694506e-01 8.35841775e-01
3.82351503e-02 3.91211331e-01 -4.95597363e-01 -2.34680369e-01
9.02384639e-01 -6.63168132e-01 -3.59301925e-01 -1.97127312e-02
4.07778084e-01 -5.01500070e-01 1.05237997e+00 7.14548171e-01
-9.77908850e-01 -3.00555676e-01 -1.57747376e+00 5.99622905e-01
-6.21664166e-01 -1.71059698e-01 6.77523255e-01 9.78521824e-01
-1.17472267e+00 9.95319426e-01 -8.71450841e-01 -6.40860200e-01
-1.84525758e-01 4.96180981e-01 -1.49764270e-01 7.72500396e-01
-1.00008178e+00 1.41513467e+00 7.35690176e-01 1.47052452e-01
-6.77703500e-01 -1.24222688e-01 -3.44546616e-01 -7.51723945e-02
5.75037241e-01 -9.66606975e-01 1.03841960e+00 -1.33227611e+00
-1.96784711e+00 6.22298181e-01 3.07315558e-01 -5.72632968e-01
7.30103970e-01 2.78799415e-01 -1.62701339e-01 -4.57963556e-01
-5.52119613e-01 6.66052997e-01 8.15658212e-01 -1.39108062e+00
-3.22311848e-01 -1.11040965e-01 2.73684394e-02 4.94691469e-02
-2.17439339e-01 -2.24148221e-02 2.45860726e-01 -6.39810622e-01
-1.48757115e-01 -1.30542040e+00 -5.15397549e-01 -1.40317250e-02
-1.48879558e-01 -1.81906596e-02 5.67602754e-01 1.35013774e-01
1.08800089e+00 -1.62926865e+00 5.69807172e-01 1.60856694e-01
-1.74149156e-01 2.08370626e-01 7.95193166e-02 5.24589360e-01
-1.13712959e-01 2.62256294e-01 -4.69042450e-01 1.99121594e-01
3.40365738e-01 2.74923474e-01 -6.51989458e-03 2.79234380e-01
3.88288945e-01 8.47585678e-01 -7.47887492e-01 -9.16241184e-02
-1.50571102e-02 5.42776823e-01 -6.56547189e-01 -1.49653032e-01
-5.23876607e-01 4.29848194e-01 -3.97060961e-01 3.73411089e-01
-2.26915330e-01 1.11214317e-01 1.22694954e-01 3.30524772e-01
-5.36063373e-01 -8.92702341e-02 -1.22893953e+00 1.50872862e+00
-3.94667596e-01 6.02977872e-01 -2.07910091e-02 -1.16511512e+00
1.32578099e+00 1.38308942e-01 4.19999123e-01 -4.87016380e-01
5.90773463e-01 6.07480049e-01 6.01811171e-01 -4.18527573e-01
5.00938475e-01 -1.48928866e-01 -2.25196809e-01 4.08815444e-01
-1.71046972e-01 -7.63389885e-01 2.17237681e-01 -6.42881334e-01
1.35378861e+00 8.46478522e-01 2.58558214e-01 -4.87103879e-01
4.26758081e-01 4.03430223e-01 5.97531140e-01 7.36730397e-01
-6.82434589e-02 5.63280344e-01 3.44837338e-01 -4.23066139e-01
-1.34022629e+00 -7.86579907e-01 9.65136886e-02 7.67159939e-01
6.74242387e-03 -8.51818025e-02 -8.61919820e-01 -1.35200024e-01
-4.05241877e-01 9.82412279e-01 -7.37761915e-01 -6.27044320e-01
-5.35302699e-01 -8.32141459e-01 5.30999064e-01 -1.82734177e-01
9.80729908e-02 -1.42858505e+00 -1.74283099e+00 3.98735762e-01
5.21429598e-01 -3.60727519e-01 3.02111506e-01 7.94253051e-01
-7.24255621e-01 -8.49052608e-01 -4.74552482e-01 -6.13039017e-01
4.81695116e-01 -2.51113266e-01 9.15161788e-01 3.12688023e-01
-4.32943285e-01 1.13231681e-01 -5.19827008e-01 -7.26465762e-01
-8.20962787e-01 -2.79492121e-02 2.44672894e-01 -3.66162419e-01
-3.86784226e-02 -1.09087455e+00 -4.26264733e-01 3.21486950e-01
-1.01844263e+00 -8.43973607e-02 5.84865332e-01 1.03435230e+00
3.25257093e-01 4.27319884e-01 6.37346923e-01 -3.09760153e-01
8.74625325e-01 -2.53054857e-01 -5.65988302e-01 4.38415051e-01
-8.54038298e-01 4.13630068e-01 7.06257284e-01 -7.72497416e-01
-6.25924706e-01 3.91034335e-01 -5.18854298e-02 -1.55616343e-01
-6.34672046e-02 3.65553528e-01 -1.42438337e-01 -2.54072428e-01
1.02180910e+00 5.16593397e-01 2.88562775e-01 -9.84253213e-02
1.59048989e-01 4.44092065e-01 4.68838960e-01 -8.64257276e-01
7.47339308e-01 2.29734585e-01 2.33766556e-01 -7.04833210e-01
2.21564427e-01 3.76342475e-01 -4.54693466e-01 -3.28112602e-01
7.01033711e-01 -5.96043505e-02 -1.06376648e+00 3.15982699e-01
-1.20506871e+00 -4.45276052e-01 -7.47282088e-01 1.94218621e-01
-1.08473241e+00 -1.73969254e-01 1.52896971e-01 -1.19687724e+00
-1.79809809e-01 -1.28618312e+00 6.22369468e-01 3.95794630e-01
-6.35262072e-01 -5.30231833e-01 1.98799208e-01 -4.09512967e-01
8.27057540e-01 5.70525825e-01 8.31139922e-01 -6.07579529e-01
-3.22727382e-01 -1.18303113e-01 6.48172140e-01 2.01988950e-01
-2.31997538e-02 5.92634320e-01 -4.81264085e-01 -3.02733406e-02
2.37767071e-01 -9.84153971e-02 3.82617325e-01 7.48929307e-02
5.74862659e-01 -1.39940366e-01 -1.66630998e-01 4.15996283e-01
1.56542814e+00 7.99524665e-01 8.66520524e-01 8.38639617e-01
-9.69544705e-03 1.03189635e+00 4.16580290e-01 3.60980809e-01
4.39905301e-02 7.81755388e-01 6.39479816e-01 3.36358190e-01
2.17678890e-01 1.30259365e-01 6.70973241e-01 6.43988431e-01
-3.57477725e-01 -1.91264704e-01 -1.05455387e+00 5.59166849e-01
-1.87428319e+00 -1.08775949e+00 -5.89070730e-02 2.23198509e+00
8.43152106e-01 3.96057963e-01 2.23726913e-01 2.52497673e-01
7.92623997e-01 -5.76102376e-01 -5.40573955e-01 -9.81160879e-01
-1.88949540e-01 4.79089618e-01 1.60069689e-01 9.64766890e-02
-7.33561933e-01 6.86158657e-01 6.20832729e+00 5.26510119e-01
-1.29993677e+00 -9.65505317e-02 2.69605130e-01 -1.64736792e-01
-2.98165143e-01 2.11772829e-01 -4.29383397e-01 7.23452210e-01
1.24060845e+00 -4.05727774e-01 6.43318892e-01 8.97088110e-01
3.18265587e-01 -1.64834663e-01 -8.62163007e-01 4.69432175e-01
-1.89132303e-01 -1.21840310e+00 -3.39294374e-01 1.00050822e-01
7.84942091e-01 -1.76861003e-01 -5.64092100e-02 2.51822799e-01
5.89062929e-01 -1.22276032e+00 1.16361284e+00 8.59039664e-01
1.79422632e-01 -8.30043554e-01 3.70639622e-01 6.77112937e-01
-7.46402562e-01 -2.92452067e-01 -2.68126935e-01 -1.32749528e-01
2.72396654e-01 2.37294301e-01 -4.87278312e-01 3.92721146e-01
6.53573751e-01 -8.27056617e-02 -4.25145924e-01 1.25733924e+00
-8.53677467e-02 1.18953772e-01 -5.11646986e-01 -8.47654164e-01
1.36941880e-01 -2.14020789e-01 1.03775883e+00 9.54976797e-01
7.99064457e-01 -3.01831335e-01 -4.69500780e-01 1.13362837e+00
6.49296343e-01 -1.86032299e-02 -9.96798754e-01 -2.16397554e-01
3.44486713e-01 8.94574761e-01 -1.02749193e+00 8.86108950e-02
2.75789946e-01 9.25496340e-01 -1.38157487e-01 -3.36639769e-02
-1.05070853e+00 -6.17953837e-01 5.65417409e-01 -1.17242694e-01
3.77963603e-01 -3.65546286e-01 -5.26007831e-01 -8.78065109e-01
-8.33698288e-02 -1.04864764e+00 -2.61125743e-01 -9.29747283e-01
-7.65846670e-01 7.55978048e-01 -2.33913630e-01 -1.14480567e+00
-8.40737998e-01 -6.10877633e-01 -5.87620318e-01 5.27697265e-01
-9.13745642e-01 -7.90362120e-01 1.45635977e-02 2.13216901e-01
3.77378076e-01 -3.68108153e-01 8.30933750e-01 -2.06808865e-01
-3.63840878e-01 2.29459554e-01 1.45339206e-01 -6.64217055e-01
1.93843156e-01 -1.19120622e+00 1.26970246e-01 8.40618253e-01
1.18290354e-03 8.08855772e-01 1.38680482e+00 -6.31003678e-01
-1.79720056e+00 -7.48314261e-01 6.00987792e-01 -3.95273983e-01
6.73384666e-01 -2.45718360e-01 -6.14999294e-01 1.25108033e-01
2.47804925e-01 -4.66321319e-01 4.56901044e-02 -1.87979519e-01
2.82492310e-01 7.43673220e-02 -9.21059787e-01 9.75940466e-01
1.16726828e+00 -2.30220594e-02 -8.64251018e-01 8.87288079e-02
7.46528804e-01 -5.28902523e-02 -6.85878396e-01 6.16216362e-01
8.45003664e-01 -1.09117377e+00 8.22393179e-01 -4.34090972e-01
5.23703039e-01 -5.11173308e-01 -8.52873623e-02 -1.31734037e+00
4.70758118e-02 -1.12862551e+00 1.66661724e-01 1.07586741e+00
6.88877642e-01 -8.06728303e-01 6.96926475e-01 3.91621470e-01
-4.12681818e-01 -9.30362105e-01 -8.61174881e-01 -1.19327569e+00
2.88932234e-01 -4.80670422e-01 6.59945011e-01 8.71824741e-01
-1.23671396e-02 3.99186686e-02 2.11783964e-02 -4.10353154e-01
1.64723337e-01 5.99666452e-03 7.01618433e-01 -1.24558663e+00
-7.26242661e-01 -1.07514012e+00 -5.18716395e-01 -1.28901839e-01
3.66114988e-03 -7.39345312e-01 3.80335927e-01 -1.12622559e+00
-3.95632297e-01 -4.88350958e-01 -7.00554773e-02 3.37221235e-01
2.99627066e-01 -3.50911841e-02 9.25441533e-02 -6.05000043e-03
-1.76613465e-01 5.35964549e-01 1.05743563e+00 1.92063469e-02
-4.02398944e-01 -4.55391519e-02 -7.23071992e-01 6.45934939e-01
9.27958190e-01 -5.90030730e-01 -5.86187601e-01 -4.73211426e-03
7.54770517e-01 -1.62983179e-01 3.88353109e-01 -1.26528645e+00
3.08849841e-01 -4.57257777e-01 1.16666537e-02 1.53825119e-01
2.82106012e-01 -9.84378934e-01 8.82215321e-01 1.05779374e+00
-1.95540383e-01 2.70148605e-01 -5.54298330e-03 3.64490896e-01
8.76089260e-02 -6.84745669e-01 6.19051456e-01 -4.75917995e-01
-5.93543887e-01 -4.49619144e-01 -6.90844417e-01 -4.85902607e-01
1.35953188e+00 -8.85013640e-01 -1.68723509e-01 -5.55020273e-02
-7.82496572e-01 9.69128758e-02 1.02180719e+00 4.57134008e-01
3.22448701e-01 -8.04440916e-01 -6.58881307e-01 2.24942178e-03
-1.75565958e-01 -3.92213225e-01 -2.79770344e-01 7.19797313e-01
-7.35126615e-01 2.16764897e-01 -6.42338216e-01 -3.62368435e-01
-9.58271563e-01 5.44044197e-01 6.71982467e-01 1.26421228e-01
-5.00324965e-01 7.20091343e-01 -4.64756101e-01 -6.65425539e-01
-2.56407671e-02 -1.80547297e-01 -1.36390075e-01 -1.19786441e-01
-2.31825635e-02 1.32860318e-01 1.27027512e-01 -3.00956011e-01
-2.51045614e-01 3.68116587e-01 7.61455476e-01 -4.14860964e-01
1.72566581e+00 1.01591937e-01 -2.86078870e-01 5.13430536e-01
5.65155506e-01 -1.95505783e-01 -9.78366971e-01 7.64834344e-01
3.02960634e-01 2.91788969e-02 -3.17220390e-01 -1.02392066e+00
-6.50249422e-01 5.48667789e-01 5.49902201e-01 4.38167006e-01
1.30873430e+00 -4.66544271e-01 2.68352330e-01 7.37820923e-01
8.70274305e-01 -1.31120169e+00 -3.94188799e-02 5.25499940e-01
1.21379125e+00 -7.29149401e-01 -6.83150440e-02 8.59949291e-02
-4.74008679e-01 1.36915612e+00 5.77070832e-01 -4.04676825e-01
2.82584935e-01 6.70465708e-01 -3.09340358e-01 -5.16632274e-02
-1.12351155e+00 -4.09879506e-01 1.13253510e-02 7.73998857e-01
2.64688611e-01 -1.82361007e-01 -9.01022613e-01 4.78187740e-01
-6.17073298e-01 1.78212643e-01 5.86569786e-01 1.28687036e+00
-6.83422446e-01 -1.49827588e+00 -4.83919680e-01 1.54631594e-02
-1.99313745e-01 3.36388350e-01 -7.81079352e-01 1.13568044e+00
5.25128126e-01 5.82148969e-01 -3.11799705e-01 -7.06945956e-01
1.93680421e-01 1.78225324e-01 6.57709777e-01 -4.08733547e-01
-1.07088137e+00 -2.73384988e-01 1.39174238e-01 -3.48571062e-01
-3.31796318e-01 -6.76702797e-01 -1.29945731e+00 -2.15290442e-01
-4.19637650e-01 2.16320559e-01 1.26961625e+00 7.19990075e-01
4.91018593e-01 7.48515546e-01 3.25976819e-01 -9.65189874e-01
-6.48876786e-01 -4.73478287e-01 -7.98080117e-02 -1.35914057e-01
-1.69696167e-01 -7.98124611e-01 -1.47608176e-01 1.86907172e-01] | [5.683590412139893, 3.8805885314941406] |
020f7e8a-40c3-4be3-93eb-30e11f6d3b7d | cluster-based-deep-ensemble-learning-for | 2302.08343 | null | https://arxiv.org/abs/2302.08343v1 | https://arxiv.org/pdf/2302.08343v1.pdf | Cluster-based Deep Ensemble Learning for Emotion Classification in Internet Memes | Memes have gained popularity as a means to share visual ideas through the Internet and social media by mixing text, images and videos, often for humorous purposes. Research enabling automated analysis of memes has gained attention in recent years, including among others the task of classifying the emotion expressed in memes. In this paper, we propose a novel model, cluster-based deep ensemble learning (CDEL), for emotion classification in memes. CDEL is a hybrid model that leverages the benefits of a deep learning model in combination with a clustering algorithm, which enhances the model with additional information after clustering memes with similar facial features. We evaluate the performance of CDEL on a benchmark dataset for emotion classification, proving its effectiveness by outperforming a wide range of baseline models and achieving state-of-the-art performance. Further evaluation through ablated models demonstrates the effectiveness of the different components of CDEL. | ['Arkaitz Zubiaga', 'Jing Ma', 'XIAOYU GUO'] | 2023-02-16 | null | null | null | null | ['emotion-classification', 'emotion-classification'] | ['computer-vision', 'natural-language-processing'] | [-3.70480955e-01 -3.31810772e-01 1.35169283e-01 -1.74688771e-01
-2.60148615e-01 -4.25216347e-01 9.17546332e-01 2.86175936e-01
-3.09853524e-01 3.09784502e-01 5.58144629e-01 2.98234284e-01
2.44694799e-01 -4.99060422e-01 -2.27961391e-01 -5.77884078e-01
6.20367229e-02 -5.30598462e-02 -3.27815592e-01 -2.97953695e-01
2.35874549e-01 3.25483441e-01 -1.89514589e+00 8.81972134e-01
5.12304068e-01 1.38616538e+00 -3.14283878e-01 3.71583939e-01
-5.90001106e-01 1.45200849e+00 -4.69194651e-01 -8.29146326e-01
-6.24677874e-02 -5.59783041e-01 -6.98258936e-01 3.21926773e-01
3.14905435e-01 -1.39000684e-01 -3.26201677e-01 5.08990943e-01
5.06493211e-01 1.22330084e-01 5.90146363e-01 -1.52381980e+00
-4.84892219e-01 3.73234212e-01 -7.85839021e-01 -1.65219530e-01
4.36909497e-01 -2.87267357e-01 8.91656101e-01 -1.13997173e+00
8.41536462e-01 1.17191017e+00 8.27451348e-01 4.61056590e-01
-1.21279633e+00 -8.90494645e-01 -7.58701488e-02 2.89143860e-01
-1.36855173e+00 -5.28032780e-01 1.04693532e+00 -6.16950095e-01
8.03229868e-01 -1.90292369e-03 7.90393591e-01 1.40066242e+00
2.69071698e-01 1.09001124e+00 1.52586830e+00 -3.46648127e-01
1.95759401e-01 3.99947196e-01 -1.35724112e-01 8.33029687e-01
-6.68955028e-01 -3.15849423e-01 -1.10754120e+00 -2.16456503e-01
4.30622399e-02 2.19622254e-01 6.96366429e-02 -1.79081753e-01
-6.96888387e-01 8.83819163e-01 4.53894854e-01 3.34520876e-01
-5.08771598e-01 1.89454243e-01 5.73931932e-01 4.43037599e-01
1.28978503e+00 2.84648597e-01 1.88603684e-01 -2.16911525e-01
-1.32619679e+00 1.08513273e-01 9.24008310e-01 3.41986269e-01
6.90099299e-01 -2.46475548e-01 -2.11545557e-01 1.21205986e+00
1.90539986e-01 2.16694474e-01 3.56418788e-01 -7.01664984e-01
-1.06186926e-01 1.08966267e+00 -1.00872658e-01 -1.53984070e+00
-5.88225424e-01 -2.78181523e-01 -8.94997001e-01 2.87517726e-01
-7.42435828e-02 -2.28514358e-01 -5.44616759e-01 1.76552486e+00
2.63661146e-01 2.98134148e-01 -3.24034154e-01 6.18424952e-01
1.04729807e+00 7.70949960e-01 2.92537719e-01 3.75684090e-02
1.00964832e+00 -1.01665890e+00 -6.93678498e-01 1.88988015e-01
3.67897272e-01 -7.25904047e-01 6.99460804e-01 4.49424326e-01
-9.91126001e-01 -3.74716759e-01 -7.08969891e-01 2.55973917e-02
-6.59405351e-01 2.88661987e-01 5.61950445e-01 3.94134283e-01
-1.26555264e+00 5.37216604e-01 -7.24181354e-01 -5.98079979e-01
9.10187602e-01 2.13060841e-01 -7.37887204e-01 2.52423108e-01
-6.74129844e-01 6.18544877e-01 1.24784149e-01 -1.62259534e-01
-5.45312047e-01 -6.21594310e-01 -6.45475566e-01 -7.29194134e-02
9.57602262e-03 -4.53134507e-01 7.28090048e-01 -1.74268210e+00
-1.57478809e+00 1.30884242e+00 -1.34835631e-01 -3.31799418e-01
6.46798909e-01 -3.81272197e-01 -6.62147760e-01 4.59249675e-01
-1.98004469e-01 1.06364262e+00 1.02950287e+00 -1.24551761e+00
-3.27897727e-01 -3.82852972e-01 -8.35710689e-02 -1.27404466e-01
-1.10072505e+00 3.31753165e-01 -3.55963051e-01 -6.58255458e-01
-2.75685310e-01 -1.04659975e+00 1.53535306e-01 -4.95083034e-02
-3.25122267e-01 -2.38542140e-01 1.22276235e+00 -5.87638855e-01
1.38653588e+00 -2.33592987e+00 3.89027029e-01 1.35884136e-01
8.00541520e-01 -1.05072306e-02 -2.79253751e-01 8.50973189e-01
2.05170102e-02 1.21133968e-01 7.39974808e-03 -9.31750596e-01
1.65120587e-01 -4.06102270e-01 -2.03237951e-01 3.39320570e-01
4.67029482e-01 1.03405988e+00 -5.74513912e-01 -5.09153366e-01
3.66428830e-02 7.22228527e-01 -4.91738707e-01 3.31167787e-01
-1.05561592e-01 4.40754324e-01 -6.29602559e-03 4.57476050e-01
5.74656606e-01 -5.85162401e-01 2.77045995e-01 -2.20260739e-01
-1.17544465e-01 -2.47200027e-01 -8.13882828e-01 1.52562940e+00
-5.23069561e-01 1.11744332e+00 1.83947042e-01 -7.98129559e-01
9.48993742e-01 2.75551826e-01 8.10923696e-01 -6.71168447e-01
5.73049486e-01 -1.32268956e-02 -5.50250411e-01 -5.77674508e-01
7.46321142e-01 -1.33635566e-01 -2.28544623e-02 9.89689887e-01
3.08486015e-01 3.36147279e-01 3.08206826e-02 5.62553763e-01
1.05480731e+00 1.67559206e-01 2.25563884e-01 -2.18560129e-01
4.05851871e-01 -3.47965479e-01 7.01564848e-02 4.39610034e-01
-2.97553450e-01 4.38144684e-01 4.89198178e-01 -4.52397227e-01
-8.12191486e-01 -7.68248200e-01 3.90731096e-02 1.45898652e+00
1.12384111e-01 -6.41920447e-01 -8.16494882e-01 -6.06661975e-01
-3.99721339e-02 1.49567291e-01 -1.16260731e+00 -9.03862156e-03
2.83645187e-02 -7.99513340e-01 4.11489815e-01 4.25935984e-01
8.46500874e-01 -1.20424747e+00 -6.06190562e-01 2.02018410e-01
-3.70262980e-01 -1.15585458e+00 2.29811184e-02 -1.61264017e-01
-3.75929952e-01 -7.73176432e-01 -7.58065522e-01 -5.43314159e-01
4.24655050e-01 2.35747129e-01 1.43745494e+00 2.70949811e-01
-2.27493137e-01 9.30745602e-01 -6.34750187e-01 -5.14701068e-01
-3.61237288e-01 5.62032312e-02 -2.04671323e-01 8.23475897e-01
5.72868288e-01 -6.59409344e-01 -6.19595647e-01 1.48888439e-01
-1.06882381e+00 3.69505018e-01 2.04848170e-01 7.41821885e-01
6.15239404e-02 -3.23935807e-01 6.97924316e-01 -6.95816219e-01
7.50392437e-01 -1.06243181e+00 7.78033510e-02 1.23073205e-01
-9.52888653e-02 -3.86197329e-01 1.84948251e-01 -3.68531734e-01
-9.68470693e-01 -1.49011865e-01 6.02048310e-03 -5.82567275e-01
-9.05370489e-02 5.71283340e-01 2.99809217e-01 -2.46737108e-01
4.22762483e-01 -9.88554507e-02 3.28779519e-01 -3.08424085e-01
2.98747480e-01 8.83912385e-01 2.78769672e-01 -3.43211740e-01
1.97550297e-01 8.15729082e-01 -1.79907948e-01 -8.87049139e-01
-8.75205994e-01 -5.69555283e-01 -4.06518549e-01 -7.75888383e-01
8.96327615e-01 -9.79557455e-01 -8.09194505e-01 8.96451175e-01
-1.06369174e+00 -2.32096940e-01 1.56990245e-01 1.12948127e-01
-3.45099002e-01 4.17890511e-02 -9.50347245e-01 -9.30714488e-01
-2.39781216e-01 -7.64259517e-01 1.09426832e+00 3.06491554e-01
-4.43842053e-01 -1.11738133e+00 1.48386165e-01 3.33696187e-01
6.46603703e-01 7.55670130e-01 5.37234128e-01 -6.74597204e-01
-2.78734326e-01 -3.11119556e-02 -3.93918276e-01 2.06372350e-01
-1.78416789e-01 1.78956389e-01 -1.26937640e+00 -1.25995949e-01
-3.75161290e-01 -9.26459134e-01 1.31300831e+00 1.56018347e-01
1.25483525e+00 -7.70980865e-02 -3.25889200e-01 3.48030627e-01
1.36195290e+00 -1.86473727e-01 6.28977954e-01 3.76723826e-01
6.18294597e-01 8.27680528e-01 -5.03694871e-03 9.02892709e-01
5.11254370e-01 5.46390593e-01 5.39766073e-01 -4.40024406e-01
4.21618298e-02 3.13119665e-02 3.15365106e-01 8.07443857e-01
-1.23403288e-01 -4.00260538e-01 -9.26376343e-01 4.45209324e-01
-2.00137568e+00 -1.27179003e+00 2.27229178e-01 1.53042901e+00
4.30046916e-01 -4.34630513e-01 4.63827133e-01 6.78954348e-02
6.55136585e-01 3.82999182e-01 -3.58893692e-01 -5.13504684e-01
-2.87681133e-01 2.61272520e-01 -3.72124553e-01 -1.07830008e-02
-1.35086346e+00 8.68877769e-01 6.18732882e+00 7.76845276e-01
-1.40246487e+00 2.99978167e-01 9.89266098e-01 -2.70885438e-01
-1.01448566e-01 -6.10041499e-01 -3.42796654e-01 5.08853197e-01
8.57325196e-01 1.69709295e-01 5.38544595e-01 5.37167668e-01
2.26637740e-02 -1.15119293e-01 -7.49990880e-01 1.21384430e+00
5.94781101e-01 -1.73564374e+00 -1.28168762e-01 2.36822776e-02
1.11977053e+00 7.64133632e-02 3.86755913e-01 3.12352836e-01
1.83756016e-02 -1.05630016e+00 8.66266608e-01 6.59724116e-01
5.09017766e-01 -8.38775158e-01 6.92175448e-01 -9.13032964e-02
-9.58021224e-01 -2.60958761e-01 7.39054456e-02 -2.33203501e-01
3.78171005e-03 4.72720027e-01 -3.91641110e-01 3.56819272e-01
9.07971323e-01 1.24477291e+00 -8.16805661e-01 7.32050598e-01
1.59902722e-01 6.72333658e-01 -1.10395961e-01 -2.38207906e-01
3.18365961e-01 -1.34825543e-01 4.51674998e-01 1.74987161e+00
1.05858013e-01 -7.33470619e-02 7.68473595e-02 1.04094052e+00
-5.48842907e-01 5.15433311e-01 -4.70034510e-01 -5.92417002e-01
3.16355973e-01 1.82301128e+00 -9.57553267e-01 -2.10875735e-01
-3.45906287e-01 1.18709540e+00 6.21986032e-01 2.78456926e-01
-9.14664984e-01 -1.09345987e-01 4.35485005e-01 -1.09379061e-01
1.55638427e-01 -1.21227287e-01 6.84458464e-02 -9.87629533e-01
-1.83719143e-01 -9.44082856e-01 2.55218685e-01 -9.31282222e-01
-1.63998282e+00 9.71434295e-01 -4.21073258e-01 -1.10507596e+00
8.42803158e-03 -5.28061688e-01 -5.76758921e-01 3.46290380e-01
-1.07337224e+00 -1.49828088e+00 -7.30314195e-01 7.13265479e-01
1.39196396e-01 -4.20535713e-01 8.88215423e-01 3.18526357e-01
-5.77351630e-01 5.67526996e-01 1.65343508e-01 2.50599414e-01
8.58233988e-01 -1.01362026e+00 -1.89760998e-01 4.06943947e-01
4.13316488e-01 2.87604988e-01 4.18802798e-01 -2.53997296e-01
-1.01984489e+00 -9.86057103e-01 6.58250332e-01 -5.13414323e-01
7.94565558e-01 -6.82946146e-01 -6.49380624e-01 4.98876274e-01
7.91105092e-01 -6.22230731e-02 1.20321095e+00 2.87039965e-01
-5.84341049e-01 1.07516805e-02 -1.06897426e+00 4.88563865e-01
7.81732738e-01 -8.53581309e-01 -2.50152379e-01 3.88886705e-02
3.13988805e-01 -9.79276970e-02 -8.76495004e-01 6.12189397e-02
9.60358083e-01 -1.51926601e+00 5.32576621e-01 -7.07472324e-01
1.30528605e+00 1.97505765e-02 -1.66708454e-01 -1.36006117e+00
-1.26772597e-01 -5.54138720e-01 -2.21067101e-01 1.56705940e+00
1.05116464e-01 -2.43382916e-01 6.94362998e-01 2.89581180e-01
2.51797199e-01 -8.43735397e-01 -6.37188852e-01 -2.35081613e-01
-1.45838588e-01 -5.30985057e-01 4.24200028e-01 1.30316222e+00
1.58777550e-01 4.58800435e-01 -7.24545658e-01 -4.98013169e-01
1.92883030e-01 1.05052769e-01 8.91419947e-01 -1.29465413e+00
-4.56429161e-02 -7.74098694e-01 -7.06869125e-01 -3.58488113e-01
6.92919433e-01 -9.92816746e-01 -4.19990331e-01 -1.33115721e+00
7.97286749e-01 -1.60810545e-01 -4.36633825e-01 4.80268300e-01
-1.72712859e-02 1.02373385e+00 5.34203172e-01 1.21749446e-01
-1.28008032e+00 6.56284213e-01 6.86544895e-01 -2.62319535e-01
-3.08560252e-01 -6.49772465e-01 -7.12850630e-01 7.33066201e-01
6.49842680e-01 -3.02419335e-01 -4.10118178e-02 -8.51520225e-02
6.61223888e-01 -3.07185739e-01 6.53479815e-01 -1.12398994e+00
1.03070013e-01 3.77349883e-01 5.17305672e-01 -4.60982651e-01
3.96566302e-01 -7.17206717e-01 2.25792155e-01 -4.17098887e-02
-6.11085773e-01 1.25338286e-01 5.69685876e-01 6.73147559e-01
-5.10789692e-01 4.08413500e-01 7.98250377e-01 1.70658857e-01
-9.06362116e-01 3.27269584e-01 -4.63163257e-01 -4.28571329e-02
1.05836523e+00 -8.04719180e-02 -3.87751967e-01 -7.53986716e-01
-9.11726415e-01 -7.05369934e-02 4.99739408e-01 7.50762105e-01
4.45462972e-01 -1.45517647e+00 -9.26426172e-01 -4.64909710e-02
3.91270518e-01 -9.74709570e-01 3.74891192e-01 1.12985682e+00
-1.83708102e-01 -2.53286779e-01 -4.64295059e-01 -7.08334386e-01
-1.38536847e+00 3.09576750e-01 3.21292222e-01 -1.88105673e-01
-3.81861806e-01 8.99490535e-01 8.00976679e-02 -3.38478148e-01
4.82368981e-03 5.08878946e-01 -4.57429379e-01 5.64781368e-01
8.38061452e-01 5.20936072e-01 6.70304969e-02 -8.49200547e-01
-2.66548663e-01 3.89240086e-01 2.67696735e-02 -7.61731118e-02
1.61956656e+00 -3.80977869e-01 -5.43564796e-01 6.64786398e-01
1.37965548e+00 1.21215299e-01 -9.85737622e-01 -1.14819281e-01
-1.00921020e-02 -2.82781929e-01 1.87177345e-01 -9.48375940e-01
-1.30478704e+00 9.30560887e-01 7.68891513e-01 2.65907973e-01
1.33893168e+00 -1.00444611e-02 6.90544128e-01 1.33076459e-01
1.84498504e-01 -1.27586448e+00 8.09129059e-01 3.27489406e-01
8.25201929e-01 -1.41786206e+00 -1.83008865e-01 1.53133914e-01
-8.27506185e-01 1.18966615e+00 3.30421418e-01 -1.58564046e-01
9.01195526e-01 3.53045583e-01 2.41969690e-01 -3.65115672e-01
-8.82159412e-01 -9.49233696e-02 5.33877254e-01 1.61593243e-01
7.60407627e-01 -1.70139819e-02 -2.38731474e-01 6.25087142e-01
8.15377086e-02 1.03871055e-01 2.26266772e-01 7.88547516e-01
-1.66679427e-01 -8.17880869e-01 -1.27434254e-01 4.50149238e-01
-8.20973933e-01 -6.94373772e-02 -9.78573203e-01 7.05402792e-01
1.99397281e-01 1.21634519e+00 4.18963015e-01 -8.51159990e-01
-3.19801956e-01 3.45267683e-01 2.45719165e-01 -9.02248695e-02
-9.90241349e-01 -7.01011643e-02 -3.99965122e-02 -8.96942496e-01
-1.14103866e+00 -4.42521960e-01 -8.22883189e-01 -5.90861499e-01
-6.18520416e-02 2.31339522e-02 7.51304030e-01 1.06839621e+00
8.58457625e-01 4.27353472e-01 8.15325558e-01 -1.07765698e+00
2.18953416e-01 -8.90540242e-01 -7.24901974e-01 8.05342495e-01
2.04002962e-01 -7.76136160e-01 -2.53383875e-01 1.16924427e-01] | [8.490242004394531, 10.688573837280273] |
4267c762-fc87-48d8-b913-3d6995232732 | person-search-by-text-attribute-query-as-zero | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Dong_Person_Search_by_Text_Attribute_Query_As_Zero-Shot_Learning_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Dong_Person_Search_by_Text_Attribute_Query_As_Zero-Shot_Learning_ICCV_2019_paper.pdf | Person Search by Text Attribute Query As Zero-Shot Learning | Existing person search methods predominantly assume the availability of at least one-shot imagery sample of the queried person. This assumption is limited in circumstances where only a brief textual (or verbal) description of the target person is available. In this work, we present a deep learning method for attribute text description based person search without any query imagery. Whilst conventional cross-modality matching methods, such as global visual-textual embedding based zero-shot learning and local individual attribute recognition, are functionally applicable, they are limited by several assumptions invalid to person search in deployment scale, data quality, and/or category name semantics. We overcome these issues by formulating an Attribute-Image Hierarchical Matching (AIHM) model. It is able to more reliably match text attribute descriptions with noisy surveillance person images by jointly learning global category-level and local attribute-level textual-visual embedding as well as matching. Extensive evaluations demonstrate the superiority of our AIHM model over a wide variety of state-of-the-art methods on three publicly available attribute labelled surveillance person search benchmarks: Market-1501, DukeMTMC, and PA100K.
| [' Xiatian Zhu', ' Shaogang Gong', 'Qi Dong'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['person-search'] | ['computer-vision'] | [ 3.09140861e-01 -3.12301099e-01 -3.27499539e-01 -7.92444944e-01
-1.03601575e+00 -3.99012536e-01 1.25628233e+00 3.57037961e-01
-6.27912343e-01 5.95137060e-01 5.60594559e-01 2.92435765e-01
-2.84923106e-01 -7.51919389e-01 -4.07194972e-01 -4.82467145e-01
1.35149479e-01 9.69913721e-01 -7.08591193e-02 -1.90082733e-02
-2.89069891e-01 1.35573283e-01 -1.80480182e+00 2.06051156e-01
4.99597788e-01 1.31820452e+00 -2.48323396e-01 4.80337709e-01
2.17682570e-01 8.87199342e-01 -5.70098341e-01 -8.72659326e-01
3.36842865e-01 -1.53288767e-01 -9.01078463e-01 2.29789078e-01
1.24039733e+00 -7.79811919e-01 -8.07303250e-01 8.49610388e-01
8.31329882e-01 1.67212591e-01 9.49128866e-01 -1.62593198e+00
-9.93575096e-01 1.09627910e-01 -1.47643641e-01 2.37969518e-01
9.08697844e-01 3.42936546e-01 1.00423753e+00 -6.32538676e-01
6.46824181e-01 1.33495820e+00 9.95460153e-01 5.83980739e-01
-1.29839277e+00 -6.15062773e-01 7.77743384e-02 4.64961827e-01
-1.81159568e+00 -5.97890198e-01 3.79868031e-01 -6.64612234e-01
1.07018280e+00 3.01464766e-01 5.97689271e-01 1.59488261e+00
-1.02755308e-01 8.84256482e-01 8.82059813e-01 -5.05099036e-02
-2.58071311e-02 2.33039603e-01 1.76024064e-01 6.83785677e-01
6.39142394e-02 3.41339976e-01 -7.33201504e-01 -7.85403371e-01
3.42870504e-01 4.60610598e-01 -1.31690070e-01 -6.38691604e-01
-1.36392748e+00 9.72950637e-01 5.24080992e-01 -7.62642324e-02
-5.45767903e-01 1.58734784e-01 8.09752464e-01 4.19388235e-01
5.27199626e-01 1.78284734e-01 5.00631146e-02 1.88951075e-01
-1.19861424e+00 5.97861171e-01 7.05657661e-01 1.29699087e+00
7.54964530e-01 -1.00790337e-01 -8.79525542e-01 8.34522307e-01
6.14640899e-02 1.07486916e+00 1.85359448e-01 -7.94200003e-01
3.77851576e-01 5.47769070e-01 1.61700130e-01 -1.07232440e+00
-1.58869162e-01 -1.19465813e-01 -9.01264012e-01 -2.60521948e-01
1.14971533e-01 2.41336390e-01 -1.08515835e+00 1.80434978e+00
1.74486682e-01 2.03974769e-02 1.93600506e-01 1.08343160e+00
1.30603540e+00 2.91062683e-01 5.53314328e-01 1.89590126e-01
1.62420797e+00 -7.76961625e-01 -6.97318971e-01 -5.00545561e-01
1.78574964e-01 -3.09213549e-01 7.97957242e-01 -5.10577679e-01
-7.64112294e-01 -5.19151628e-01 -7.46201038e-01 1.37792174e-02
-7.51921892e-01 -1.06569983e-01 5.11494577e-01 5.82433224e-01
-1.16564882e+00 -5.38747758e-02 -2.26933271e-01 -1.07962227e+00
4.96355504e-01 4.03449506e-01 -7.71900952e-01 -4.21541780e-01
-1.64322519e+00 8.15201461e-01 1.18245646e-01 -3.20990801e-01
-1.30279446e+00 -5.88604808e-01 -1.17043769e+00 7.13521540e-02
3.88132662e-01 -1.23153508e+00 1.11905420e+00 -7.36616731e-01
-5.40193021e-01 1.48270547e+00 -2.34166712e-01 -6.29930258e-01
6.11063302e-01 -5.78934103e-02 -5.43448865e-01 3.06152552e-01
6.25474691e-01 8.21578562e-01 9.52542722e-01 -9.29952860e-01
-5.51348925e-01 -5.77292323e-01 7.75260404e-02 2.63769954e-01
-4.63267177e-01 3.90881956e-01 -5.58482289e-01 -6.43044531e-01
-4.86225396e-01 -6.58228040e-01 3.69782485e-02 4.42575932e-01
-3.25537980e-01 -4.05124575e-01 7.24006176e-01 -7.41729438e-01
8.09329569e-01 -2.00507164e+00 -7.73687959e-02 2.90123839e-02
2.40360767e-01 1.00230686e-01 -2.09208757e-01 6.13053858e-01
3.23724061e-01 -1.46517307e-01 -1.64622933e-01 -6.76936209e-01
3.81968468e-01 1.07058093e-01 -2.36337274e-01 7.96666265e-01
1.17711043e-02 1.20680034e+00 -9.26930428e-01 -1.02890003e+00
1.25823975e-01 6.53915107e-01 -1.57110259e-01 3.78352076e-01
-2.65358505e-03 -3.84495445e-02 -4.95281875e-01 1.11820233e+00
2.16473937e-01 -1.74277410e-01 -3.38512778e-01 -2.12139353e-01
3.04396778e-01 -2.69194573e-01 -7.68988132e-01 1.64293051e+00
-4.54635738e-04 6.95001245e-01 7.71848112e-02 -8.66193473e-01
7.77128279e-01 5.59731424e-01 5.09799659e-01 -8.91424537e-01
-2.04794958e-01 -1.08765647e-01 -8.90383184e-01 -4.08199787e-01
4.32018936e-01 1.08126193e-01 -5.57946265e-01 3.09321433e-01
2.03859985e-01 3.52571338e-01 -3.36772390e-02 3.16711754e-01
9.84011889e-01 -1.70469463e-01 4.82822508e-01 -2.05517754e-01
5.26178420e-01 1.54797539e-01 4.57713723e-01 1.34959292e+00
-5.63832164e-01 6.16584122e-01 -8.92202370e-03 -8.72062027e-01
-1.19732261e+00 -1.16687644e+00 -2.07475916e-01 1.39346445e+00
3.15163463e-01 -3.85524809e-01 -6.98155582e-01 -5.43705523e-01
2.47543231e-01 3.90235841e-01 -8.87941778e-01 -2.36660585e-01
-1.62810341e-01 -4.11253005e-01 9.60996568e-01 5.87313652e-01
9.00099218e-01 -1.04767764e+00 -7.03947067e-01 -3.59061807e-02
-4.68533635e-01 -1.38257468e+00 -8.98827493e-01 -3.85266900e-01
-2.87674870e-02 -1.08310080e+00 -9.95697439e-01 -7.82675028e-01
3.98541540e-01 1.29693290e-02 1.33585644e+00 8.88617337e-02
-6.91337168e-01 1.06388628e+00 -2.16258019e-01 -3.21007043e-01
-2.54651699e-02 1.30998008e-02 4.09430802e-01 2.07432702e-01
9.23129916e-01 8.41377527e-02 -6.17490113e-01 3.29839259e-01
-5.93610168e-01 -2.77613401e-01 4.63210195e-01 9.79190052e-01
5.50093114e-01 -3.86831820e-01 3.18258643e-01 -4.11466897e-01
6.47680342e-01 -4.25029844e-01 -2.54448354e-01 7.13114619e-01
-5.00806272e-01 -1.12198189e-01 2.58108079e-01 -4.86923695e-01
-8.30541313e-01 2.08628967e-01 1.87034950e-01 -5.71890712e-01
-5.52712560e-01 1.58927754e-01 -1.90398395e-01 3.49176675e-02
7.33159542e-01 6.68850362e-01 1.25413751e-02 -2.06778497e-01
1.81624055e-01 8.72500002e-01 1.06196320e+00 -2.99211890e-01
8.10309649e-01 6.32177830e-01 -2.81190783e-01 -8.00161958e-01
-8.25974643e-01 -9.58969593e-01 -6.78112745e-01 -2.51353294e-01
1.25436342e+00 -1.28212988e+00 -7.43542969e-01 6.15305722e-01
-9.22891736e-01 -9.98666659e-02 -1.08158685e-01 7.14240670e-02
-6.94818556e-01 3.35066229e-01 -4.44568813e-01 -8.05398166e-01
-7.48479128e-01 -9.47109163e-01 1.57083404e+00 -8.80511627e-02
-2.51610011e-01 -8.55498910e-01 -7.92418197e-02 5.79952419e-01
5.95941305e-01 4.42867130e-01 4.61009502e-01 -1.07520318e+00
-3.93840492e-01 -7.94533670e-01 -4.41716194e-01 -3.56957018e-01
1.47481263e-02 -7.13614285e-01 -1.07229030e+00 -7.43373513e-01
-4.61795777e-01 -7.17020750e-01 9.97708142e-01 1.40593424e-01
7.76915669e-01 -5.66098392e-01 -7.48738110e-01 8.10904026e-01
1.45705855e+00 -2.72232413e-01 3.00778747e-01 5.13432384e-01
5.80929935e-01 6.14158452e-01 6.34827673e-01 5.30090272e-01
6.47603035e-01 1.18243146e+00 2.69655317e-01 -1.55496627e-01
-9.65266153e-02 -3.41041774e-01 4.60278951e-02 -3.56565416e-01
3.76812369e-02 -3.96173537e-01 -1.20580029e+00 8.46607566e-01
-1.95993745e+00 -1.55277061e+00 4.67949569e-01 2.17864013e+00
6.29156709e-01 -3.82132053e-01 4.03165102e-01 -3.38655829e-01
7.70416141e-01 3.91295731e-01 -6.36969984e-01 2.13538751e-01
-4.23050135e-01 -5.82122505e-01 6.23219669e-01 2.41817266e-01
-1.80220056e+00 8.19154441e-01 6.34151268e+00 6.04919136e-01
-4.54114854e-01 3.16015095e-01 3.89095068e-01 -1.42535657e-01
-8.13547242e-03 -4.28722143e-01 -1.01855147e+00 4.71941739e-01
8.00926030e-01 -2.06894293e-01 2.59541780e-01 8.83951902e-01
-3.66856068e-01 2.53503859e-01 -1.47502923e+00 1.47466314e+00
5.44435740e-01 -1.21397495e+00 3.11826736e-01 3.05651687e-02
4.74747300e-01 5.54753356e-02 1.67024329e-01 4.61296916e-01
1.51201978e-01 -1.28646934e+00 7.57734716e-01 8.33011150e-01
1.23541880e+00 -4.95010614e-01 1.01459765e+00 2.18264222e-01
-1.52199328e+00 -3.76173228e-01 -3.43992412e-01 3.82581800e-01
1.63424402e-01 -2.01452032e-01 -5.68484962e-01 4.48693365e-01
1.27678645e+00 6.53727591e-01 -8.35752845e-01 1.03056824e+00
3.28945845e-01 1.24478251e-01 -3.10903788e-01 1.09571993e-01
3.32694918e-01 2.94330299e-01 6.09251022e-01 1.46544170e+00
-2.52663121e-02 2.11762488e-01 6.05549514e-01 9.68368173e-01
-2.09738221e-02 -8.86796266e-02 -9.11939144e-01 4.27786410e-02
6.75931275e-01 7.73436964e-01 -5.84093630e-02 -5.76476693e-01
-6.84469044e-01 1.38899624e+00 2.28596449e-01 5.46744645e-01
-6.93208635e-01 -1.84891269e-01 1.04340041e+00 2.07408443e-01
1.43564120e-01 2.35237330e-01 3.35743308e-01 -9.04153585e-01
3.44292372e-02 -1.00675881e+00 1.09262264e+00 -6.49625838e-01
-1.74920225e+00 6.86135292e-01 4.12407577e-01 -1.08033252e+00
-7.53505766e-01 -1.91456228e-01 -3.67378920e-01 9.22461569e-01
-1.26669419e+00 -2.05329466e+00 -7.08004415e-01 1.06007493e+00
5.98737240e-01 -8.92631233e-01 1.15884292e+00 3.93420845e-01
-8.13277811e-02 1.03145158e+00 -1.68058217e-01 5.92125595e-01
9.44316983e-01 -1.03412604e+00 5.62023520e-01 5.47189772e-01
-6.86612949e-02 5.44298232e-01 8.07855785e-01 -7.92059302e-01
-1.25337327e+00 -1.14991796e+00 1.14923990e+00 -8.27333570e-01
3.58874261e-01 -6.47691548e-01 -6.97677493e-01 7.04644144e-01
2.27331549e-01 1.85449556e-01 6.03699148e-01 -9.10034105e-02
-6.59681499e-01 -1.91372395e-01 -1.32106185e+00 3.05737138e-01
1.12969434e+00 -1.10039854e+00 -5.44842124e-01 5.95775843e-01
7.59701014e-01 -9.60125960e-03 -9.90582764e-01 4.14647937e-01
4.80084181e-01 -6.86411679e-01 1.70577598e+00 -7.40672827e-01
-1.36464387e-01 -8.88301358e-02 -4.70151514e-01 -8.43461096e-01
-4.76476312e-01 -2.19624355e-01 1.69347413e-02 1.34292877e+00
9.44452882e-02 -4.83549058e-01 5.48692107e-01 1.00934887e+00
2.56459355e-01 -3.41223210e-01 -1.02484989e+00 -9.91778493e-01
-2.90497482e-01 2.30886554e-03 8.34827006e-01 9.34412003e-01
-3.43824208e-01 1.68015301e-01 -9.19517875e-01 3.07291478e-01
1.26360619e+00 3.14710997e-02 6.75574839e-01 -1.39328408e+00
7.23851323e-02 -1.91911966e-01 -9.75873828e-01 -5.44987202e-01
5.07600069e-01 -6.91627443e-01 -2.49680337e-02 -1.82225430e+00
9.16355789e-01 -7.95394257e-02 -2.07776785e-01 6.62627518e-01
-1.67916223e-01 2.79348671e-01 8.43671784e-02 5.81387818e-01
-1.01609993e+00 5.46666443e-01 2.96464294e-01 -8.15365791e-01
2.77024150e-01 -1.11951202e-01 -2.72259414e-01 1.53317124e-01
4.44746852e-01 -3.08345973e-01 -2.85138607e-01 -5.06651282e-01
-2.10279137e-01 -2.57355091e-03 1.19760883e+00 -1.05979586e+00
7.45867014e-01 8.95368494e-03 6.46704078e-01 -5.47279775e-01
8.35312486e-01 -9.16663945e-01 1.80382311e-01 3.72388303e-01
-6.34243071e-01 1.49160713e-01 -3.84623185e-02 1.01032245e+00
-2.50727683e-01 1.86794952e-01 5.40673316e-01 -2.75312811e-01
-1.28471839e+00 9.09324467e-01 -2.07000151e-01 1.30226910e-01
9.59342837e-01 -5.23638725e-01 -4.85504061e-01 -7.72528529e-01
-4.85593766e-01 6.10973179e-01 7.14672804e-01 6.25057638e-01
8.49405348e-01 -1.51755881e+00 -1.12003052e+00 1.35626420e-01
9.69972551e-01 -7.05057681e-01 2.14892715e-01 5.02569139e-01
-7.59298801e-02 8.37374926e-01 -1.85803950e-01 -7.04845488e-01
-1.53549361e+00 1.00390589e+00 5.19234240e-01 2.34952748e-01
-7.57484913e-01 7.19940841e-01 3.11649412e-01 -5.44169009e-01
5.94985366e-01 5.88342249e-01 -1.33229345e-01 -2.20041759e-02
7.39952028e-01 1.57802790e-01 -3.43036473e-01 -1.29532504e+00
-8.00291359e-01 5.92148483e-01 4.88939211e-02 -1.42822126e-02
9.79957223e-01 -2.12517217e-01 8.04665610e-02 -2.25214474e-02
1.13019753e+00 -8.53686810e-01 -1.11933613e+00 -7.53078222e-01
-7.95759261e-02 -5.48293352e-01 -1.35532916e-01 -7.26181328e-01
-6.74985349e-01 5.59596658e-01 1.04400814e+00 -6.26256242e-02
1.00323117e+00 5.09733319e-01 7.39422500e-01 7.13709891e-01
3.95610988e-01 -1.13736069e+00 -2.12399103e-02 9.76023898e-02
8.78069758e-01 -1.76148713e+00 9.57081690e-02 -2.97246743e-02
-7.45260298e-01 5.45411885e-01 6.67401254e-01 4.35584426e-01
3.23425770e-01 -1.95376471e-01 8.27772636e-03 -4.94216919e-01
-6.91000700e-01 -4.85514224e-01 6.61035180e-01 1.03058279e+00
-4.47489694e-02 1.15413284e-02 4.08778936e-01 4.07104284e-01
-2.87914537e-02 -4.18089539e-01 -2.33508661e-01 8.08080733e-01
-3.11690927e-01 -5.24405956e-01 -4.36758190e-01 5.50782979e-01
-3.39404851e-01 -2.26140931e-01 -7.77271569e-01 7.74670362e-01
-6.36084974e-02 1.02662516e+00 3.01763624e-01 -2.44354084e-01
3.79728705e-01 2.35177711e-01 6.86777309e-02 -3.10736388e-01
-5.64289391e-01 -4.61247146e-01 2.06565142e-01 -6.62094533e-01
-6.02978885e-01 -9.13906395e-01 -5.15216172e-01 -4.17810500e-01
1.37300789e-01 -1.03500552e-01 1.42690897e-01 8.73954058e-01
1.94903851e-01 -1.68826371e-01 4.31078374e-02 -6.07510388e-01
-6.40198171e-01 -8.11123312e-01 -3.66140246e-01 1.03004479e+00
6.31576836e-01 -7.23754585e-01 -1.79559380e-01 2.70215780e-01] | [14.624531745910645, 0.9259564876556396] |
fcd00e02-9da1-42f2-8516-d58c1c19ff76 | reranking-overgenerated-responses-for-end-to | 2211.03648 | null | https://arxiv.org/abs/2211.03648v2 | https://arxiv.org/pdf/2211.03648v2.pdf | Reranking Overgenerated Responses for End-to-End Task-Oriented Dialogue Systems | End-to-end (E2E) task-oriented dialogue (ToD) systems are prone to fall into the so-called "likelihood trap", resulting in generated responses which are dull, repetitive, and often inconsistent with dialogue history. Comparing ranked lists of multiple generated responses against the "gold response" (from evaluation data) reveals a wide diversity in response quality, with many good responses placed lower in the ranked list. The main challenge, addressed in this work, is then how to reach beyond greedily generated system responses, that is, how to obtain and select such high-quality responses from the list of overgenerated responses at inference without availability of the gold response. To this end, we propose a simple yet effective reranking method which aims to select high-quality items from the lists of responses initially overgenerated by the system. The idea is to use any sequence-level (similarity) scoring function to divide the semantic space of responses into high-scoring versus low-scoring partitions. At training, the high-scoring partition comprises all generated responses whose similarity to the gold response is higher than the similarity of the greedy response to the gold response. At inference, the aim is to estimate the probability that each overgenerated response belongs to the high-scoring partition, given only previous dialogue history. We validate the robustness and versatility of our proposed method on the standard MultiWOZ dataset: our methods improve a state-of-the-art E2E ToD system by 2.0 BLEU, 1.6 ROUGE, and 1.3 METEOR scores, achieving new peak results. Additional experiments on the BiTOD dataset and human evaluation further ascertain the generalisability and effectiveness of the proposed framework. | ['Anna Korhonen', 'Fangyu Liu', 'Ivan Vulić', 'Songbo Hu'] | 2022-11-07 | null | null | null | null | ['task-oriented-dialogue-systems'] | ['natural-language-processing'] | [ 2.74675190e-01 3.13778102e-01 1.08936712e-01 -4.56692278e-01
-1.30211866e+00 -7.81875730e-01 6.23859107e-01 2.27728724e-01
-6.45061851e-01 1.03718531e+00 4.59985077e-01 -2.93742977e-02
-4.25219417e-01 -5.92084825e-01 -6.91678971e-02 -5.64106762e-01
2.51100451e-01 1.01865423e+00 5.14121056e-01 -7.20232069e-01
6.03057802e-01 -1.12350762e-01 -1.39137065e+00 6.65446997e-01
1.00456131e+00 6.77780807e-01 2.99171567e-01 1.02965486e+00
-2.03340545e-01 6.64034426e-01 -9.38606441e-01 -4.90859866e-01
-1.54631451e-01 -7.42636502e-01 -1.35797834e+00 -1.83307249e-02
1.60905182e-01 -3.18718314e-01 5.26320003e-02 8.58299971e-01
5.47950745e-01 4.42004979e-01 6.33636057e-01 -8.54544222e-01
-2.37422392e-01 7.63191640e-01 -1.09314077e-01 -2.19889963e-03
8.36087823e-01 -7.98593462e-03 1.31969535e+00 -1.00154233e+00
8.44496489e-01 1.27477086e+00 2.87404656e-01 9.16387618e-01
-1.12618613e+00 -1.46568105e-01 -1.77877501e-01 -1.80377215e-02
-1.03658545e+00 -5.50728679e-01 4.73169237e-01 -3.89923751e-01
8.56808901e-01 6.35420084e-01 1.45397019e-02 1.05324447e+00
-8.05984288e-02 8.21330309e-01 1.28711414e+00 -4.79662001e-01
2.89032608e-01 4.53301817e-01 4.05293554e-01 3.22734594e-01
-3.45651656e-01 -3.99906367e-01 -7.78812766e-01 -3.75255018e-01
-7.71788135e-03 -4.32020336e-01 -4.24575984e-01 5.71213327e-02
-1.18791115e+00 8.61233890e-01 7.19876513e-02 3.77275229e-01
-6.60328090e-01 -3.65290374e-01 3.75319242e-01 4.92848605e-01
5.59778750e-01 5.88946223e-01 -4.37112182e-01 -4.61393744e-01
-9.13379729e-01 5.30815601e-01 1.10196054e+00 6.73651516e-01
7.20814884e-01 -4.00734395e-01 -7.20541716e-01 1.23991287e+00
5.60827442e-02 2.36147419e-01 6.62662983e-01 -7.61897683e-01
7.96021581e-01 6.94043159e-01 5.74760735e-01 -8.55420828e-01
-2.51992464e-01 -1.87682629e-01 -6.80723190e-01 -1.77399844e-01
7.17176557e-01 -3.20419699e-01 -3.60416532e-01 1.69332504e+00
3.38370323e-01 -6.62817895e-01 2.89474100e-01 1.07043576e+00
7.30676413e-01 6.57167435e-01 -1.00930266e-01 -4.13494110e-01
1.22083151e+00 -9.23719883e-01 -6.38518453e-01 -1.40718997e-01
7.23423719e-01 -9.28544819e-01 1.31511319e+00 4.74046588e-01
-1.06049514e+00 -5.48313975e-01 -7.44039953e-01 2.85377264e-01
6.02767542e-02 2.19898298e-01 2.02151239e-01 4.66799408e-01
-1.10784996e+00 5.50237119e-01 3.51919830e-02 -4.80763704e-01
-4.01585490e-01 3.21888477e-01 -3.43191922e-01 1.51804343e-01
-1.54703164e+00 9.15449500e-01 5.19981682e-01 -1.02135755e-01
-6.56431258e-01 -2.23805383e-01 -3.02989125e-01 -3.34836692e-02
6.70500159e-01 -4.62460130e-01 1.66207898e+00 -9.27080214e-01
-1.72954524e+00 7.51031995e-01 -1.74158514e-01 -3.84674996e-01
8.55886221e-01 -2.36396074e-01 -1.75310656e-01 1.76107258e-01
7.98300281e-02 4.77780849e-01 4.78958517e-01 -1.10116935e+00
-9.04110968e-01 -2.28633061e-01 1.89529419e-01 5.19841731e-01
-2.74558902e-01 1.33850202e-01 -2.62681276e-01 -2.28272647e-01
3.00725996e-02 -9.63126361e-01 2.46957056e-02 -7.36109197e-01
-4.75351602e-01 -9.16922569e-01 1.03160858e-01 -5.50402522e-01
1.54172230e+00 -1.77051902e+00 9.95124355e-02 1.51885465e-01
2.41244510e-01 2.76746154e-01 -1.16453245e-01 8.24626565e-01
2.62128383e-01 7.86651969e-02 1.08669654e-01 3.28385853e-03
1.21274538e-01 -1.49866641e-01 -2.20085129e-01 -1.62171908e-02
5.68440696e-03 4.92184520e-01 -1.16633284e+00 -3.76903355e-01
-1.53704926e-01 -2.50069499e-01 -4.66526687e-01 6.35815740e-01
-2.99092740e-01 2.74984151e-01 -5.38279951e-01 1.19886033e-01
2.86135137e-01 -1.40276849e-01 3.92570853e-01 -6.68103918e-02
-9.66403186e-02 8.51222336e-01 -1.16696107e+00 1.30205166e+00
-3.79746825e-01 1.41444921e-01 -1.28502771e-01 -5.59505939e-01
1.22207129e+00 4.19924617e-01 1.64362371e-01 -6.51418686e-01
-5.80861755e-02 5.19361973e-01 2.11248189e-01 -5.34278333e-01
1.03048670e+00 -1.65421851e-02 -4.81468558e-01 8.32558036e-01
1.34768471e-01 2.19428111e-02 4.14631546e-01 5.33698738e-01
9.85602021e-01 -7.23453388e-02 4.35684741e-01 -1.27227515e-01
7.26307571e-01 9.79016945e-02 1.94285005e-01 1.01692593e+00
-4.73196581e-02 3.90431672e-01 7.50832260e-01 -1.96973309e-01
-9.75145578e-01 -7.26442337e-01 2.83866584e-01 1.53732765e+00
-5.85735850e-02 -3.05339187e-01 -9.52084064e-01 -8.68016303e-01
-3.42525929e-01 7.90143311e-01 -3.87909800e-01 3.96581553e-03
-5.06437719e-01 -4.92418915e-01 6.13054812e-01 -5.23004681e-02
4.59220916e-01 -1.10808957e+00 -4.69564587e-01 5.79065204e-01
-9.78937626e-01 -7.33149588e-01 -5.52195191e-01 4.76172306e-02
-5.12568891e-01 -9.03925598e-01 -7.44089365e-01 -4.01448280e-01
3.43475223e-01 2.18119845e-01 1.21070349e+00 9.31891203e-02
4.41631079e-01 -9.89489164e-03 -7.64075100e-01 8.23955461e-02
-8.43126595e-01 2.15317950e-01 1.32827461e-01 9.56818759e-02
2.90716738e-01 8.80772471e-02 -5.71352184e-01 6.91640794e-01
-6.58049643e-01 -3.11370920e-02 4.21016961e-01 1.08827066e+00
1.56881779e-01 -1.38277262e-01 1.02109456e+00 -1.10369885e+00
1.25515282e+00 -5.39741933e-01 -4.92472351e-02 4.63154972e-01
-7.76213050e-01 2.09603027e-01 9.32721078e-01 -4.42955822e-01
-1.13056612e+00 -2.84268767e-01 -3.39176327e-01 3.83572727e-01
-1.34220570e-01 4.11551833e-01 7.79326931e-02 3.98297578e-01
8.20673287e-01 3.88168454e-01 -2.28537843e-01 -4.69032735e-01
2.71542490e-01 1.15044868e+00 5.63231289e-01 -7.24336803e-01
2.70131499e-01 -1.95062980e-01 -5.10119438e-01 -5.50907195e-01
-7.79165030e-01 -7.85882354e-01 -2.97791928e-01 -5.28815329e-01
4.14763957e-01 -4.24064308e-01 -7.81732500e-01 3.47000748e-01
-1.27570939e+00 -1.70413673e-01 8.85784477e-02 3.24888289e-01
-5.53280056e-01 5.91225803e-01 -6.04189813e-01 -1.29909277e+00
-6.08541489e-01 -9.06078458e-01 8.37464929e-01 1.96915105e-01
-9.11040902e-01 -7.83693135e-01 1.67377010e-01 6.95096552e-01
2.16090754e-01 -2.02345490e-01 1.00480711e+00 -1.31741929e+00
-9.46368054e-02 -2.14895383e-01 5.34882769e-02 2.76101708e-01
-9.85100120e-02 -1.96762830e-01 -8.41007590e-01 -2.24511608e-01
1.90453008e-02 -6.95688128e-01 6.67621672e-01 -1.90460071e-01
4.32258099e-01 -6.04335129e-01 7.40012825e-02 -2.58830518e-01
1.04451716e+00 2.92232424e-01 4.56033528e-01 3.02326977e-01
7.60628358e-02 1.10948324e+00 1.10064387e+00 5.39451241e-01
3.82407039e-01 9.96863842e-01 1.30286381e-01 2.38750607e-01
1.48747966e-01 -3.77199769e-01 5.11992037e-01 9.82026994e-01
2.21633166e-01 -7.71870494e-01 -7.59765983e-01 6.50651991e-01
-1.93697393e+00 -1.02334952e+00 -4.60274160e-01 2.44692898e+00
1.18099761e+00 3.92157346e-01 4.67266232e-01 2.01113462e-01
8.05075765e-01 4.59312834e-03 -1.39778957e-01 -7.20143855e-01
5.74421175e-02 5.29296696e-02 -2.80310232e-02 7.14078188e-01
-7.14765549e-01 9.47516084e-01 5.78560972e+00 1.03383696e+00
-7.71823645e-01 -6.36926740e-02 5.14471948e-01 1.61219180e-01
-4.06765878e-01 -9.71098803e-03 -9.29164410e-01 5.35998523e-01
1.07169354e+00 -4.84675705e-01 3.24588150e-01 6.99357808e-01
3.77854824e-01 -3.29730451e-01 -1.08544040e+00 4.26843673e-01
-5.05373292e-02 -8.74911249e-01 -4.96389205e-03 -2.59344757e-01
5.09809494e-01 -3.81556720e-01 -1.13332316e-01 5.09755850e-01
6.29027724e-01 -6.37989104e-01 6.09947562e-01 4.16175902e-01
7.52858698e-01 -6.65990531e-01 1.02155864e+00 7.74250746e-01
-6.72271729e-01 1.62005827e-01 -3.53005230e-01 5.96098676e-02
-2.07928866e-02 3.83107543e-01 -1.55786586e+00 6.76224530e-01
3.16983849e-01 -4.47998010e-02 -3.08630377e-01 7.85148978e-01
-2.28714913e-01 5.55579007e-01 -8.20710585e-02 -6.47672176e-01
5.49260497e-01 3.36028226e-02 5.82151413e-01 1.35414350e+00
3.06493491e-01 1.91087604e-01 2.56306946e-01 5.14782071e-01
-3.42352758e-03 4.56133962e-01 -3.61957371e-01 1.72287419e-01
6.65955305e-01 1.23927224e+00 -4.76472944e-01 -5.10672569e-01
4.39264774e-02 1.03599370e+00 3.41162443e-01 1.29168376e-01
-4.66642320e-01 -4.87183750e-01 1.76073506e-01 -1.01794377e-01
-1.30335882e-01 2.55142301e-01 -1.48252591e-01 -7.57176459e-01
1.07781693e-01 -1.17289376e+00 5.06433308e-01 -6.05002582e-01
-1.33393335e+00 1.02240098e+00 -1.50835454e-01 -1.33227801e+00
-9.10172522e-01 -2.48655677e-01 -3.38818312e-01 1.11444664e+00
-9.89621341e-01 -5.11171639e-01 -1.94007024e-01 4.89907682e-01
1.12002766e+00 -8.55949223e-02 9.12884057e-01 9.19609815e-02
-2.65951782e-01 7.24941730e-01 1.91537127e-01 -4.17576805e-02
9.64232028e-01 -1.38029170e+00 1.10678263e-01 6.22460902e-01
-7.71959871e-02 7.99717784e-01 9.19393539e-01 -6.91468596e-01
-9.63974893e-01 -7.75490820e-01 1.56701767e+00 -4.42885339e-01
6.23232782e-01 -1.47742480e-01 -9.99541104e-01 2.18226537e-02
1.94417343e-01 -9.14513528e-01 5.31945109e-01 2.36185402e-01
-1.25912845e-01 8.18437710e-02 -1.12238443e+00 6.08175516e-01
5.75753510e-01 -4.84041244e-01 -8.83755744e-01 3.63705099e-01
3.42593402e-01 -3.61371815e-01 -6.04165494e-01 1.81353390e-01
5.62635422e-01 -1.20322001e+00 5.16159177e-01 -5.82456052e-01
5.24250388e-01 -7.71522522e-02 1.46911817e-03 -1.60304499e+00
-2.76003152e-01 -9.41189766e-01 2.55459338e-01 1.38995111e+00
5.30770600e-01 -3.15370798e-01 6.71716213e-01 5.55750072e-01
6.82602637e-04 -6.95914745e-01 -7.94619143e-01 -6.05593860e-01
-1.57070994e-01 -6.91802874e-02 3.17379743e-01 6.47932529e-01
4.28271145e-01 8.33145559e-01 -7.44218826e-01 -3.49614978e-01
2.61239260e-01 1.36961192e-01 9.00067329e-01 -1.13552642e+00
-3.63463610e-01 -4.89600420e-01 1.09676614e-01 -1.39954269e+00
-1.21330656e-01 -6.54375672e-01 6.16475940e-01 -1.42393863e+00
2.87749559e-01 -4.11415070e-01 -1.13372318e-01 2.96993226e-01
-6.52550817e-01 9.60528944e-03 1.65053770e-01 4.21266586e-01
-9.65052903e-01 4.27479446e-01 1.03185880e+00 5.37119284e-02
-2.91464984e-01 2.84893066e-01 -6.09517336e-01 4.50248927e-01
7.07428038e-01 -6.89160645e-01 -4.04442996e-01 -1.04630053e-01
2.98076898e-01 6.90367937e-01 -1.39707644e-02 -5.87077498e-01
4.20959741e-01 -2.21553594e-01 -2.22172827e-01 -5.11554182e-01
9.69953313e-02 -3.25131625e-01 -4.06090952e-02 2.86549181e-01
-9.54333961e-01 1.46351621e-01 -3.49760711e-01 4.64937538e-01
-2.20872328e-01 -6.32551730e-01 6.58087254e-01 -1.90551713e-01
-7.03202665e-01 -2.51523226e-01 -6.29230380e-01 3.62272441e-01
6.43046200e-01 -2.90246814e-01 -4.03734177e-01 -6.38767719e-01
-5.34349382e-01 1.98056564e-01 1.32007539e-01 4.60834682e-01
6.59841657e-01 -1.09299767e+00 -1.06797361e+00 -1.77219793e-01
4.07736629e-01 -4.24276620e-01 2.36753628e-01 7.40888476e-01
-1.40535682e-01 6.09863102e-01 -5.13198860e-02 -4.52941895e-01
-1.50364459e+00 7.28905026e-04 1.56260908e-01 -7.37895548e-01
-2.86388457e-01 9.52544093e-01 -1.05344700e-02 -5.03976047e-01
1.92613140e-01 1.35691375e-01 -4.73009616e-01 1.47352010e-01
4.99042839e-01 5.15882015e-01 2.59876132e-01 -5.35139918e-01
-2.29882613e-01 5.67073934e-02 -3.50033432e-01 -4.50728416e-01
9.64115202e-01 -2.57753789e-01 -1.08161859e-01 5.91595948e-01
9.39251304e-01 1.74106747e-01 -8.57728302e-01 -3.77354413e-01
4.64677006e-01 -5.39322317e-01 -4.04911816e-01 -1.14860237e+00
-2.57172585e-01 5.38467824e-01 1.70074612e-01 6.98109388e-01
9.09903884e-01 -1.59457117e-01 8.13094616e-01 5.92113674e-01
5.33543825e-01 -1.29470098e+00 4.16006058e-01 7.87574291e-01
8.41035485e-01 -1.04352069e+00 -5.13872266e-01 -1.97441772e-01
-1.20840204e+00 1.12254179e+00 7.47912884e-01 2.15209648e-01
-2.66718090e-01 -1.82142988e-01 2.16720343e-01 -1.05367981e-01
-1.21026075e+00 -9.50272158e-02 3.15999418e-01 2.03150541e-01
6.20686650e-01 1.23878315e-01 -8.63747001e-01 5.50873399e-01
-3.26093912e-01 -2.18635648e-01 6.50152624e-01 4.35341090e-01
-7.81544685e-01 -1.31022644e+00 -2.44420588e-01 4.18781489e-01
-5.82745194e-01 2.08147652e-02 -8.91175866e-01 4.13691580e-01
-3.32992762e-01 1.54281664e+00 -4.75552499e-01 -6.85396731e-01
5.95765352e-01 4.06489730e-01 7.07526430e-02 -6.67701364e-01
-1.01673520e+00 9.98258367e-02 6.66657984e-01 -2.40953282e-01
-1.36352196e-01 -3.72092605e-01 -1.02025068e+00 -4.15125012e-01
-5.16098261e-01 7.12850690e-01 4.02992457e-01 9.26389694e-01
2.84888856e-02 2.14215294e-01 1.12779498e+00 -2.94840217e-01
-1.33307266e+00 -1.16418898e+00 -3.22730243e-01 8.73441398e-01
1.16003349e-01 -3.23519289e-01 -4.21768457e-01 -2.21720725e-01] | [12.636003494262695, 8.096800804138184] |
102f31f8-82da-43cb-9f41-8203fad3737a | leti-learning-to-generate-from-textual | 2305.10314 | null | https://arxiv.org/abs/2305.10314v1 | https://arxiv.org/pdf/2305.10314v1.pdf | LeTI: Learning to Generate from Textual Interactions | Finetuning pre-trained language models (LMs) enhances the models' capabilities. Prior techniques fine-tune a pre-trained LM on input-output pairs (e.g., instruction fine-tuning), or with numerical rewards that gauge the quality of its outputs (e.g., reinforcement learning from human feedback). We explore LMs' potential to learn from textual interactions (LeTI) that not only check their correctness with binary labels, but also pinpoint and explain errors in their outputs through textual feedback. Our investigation focuses on the code generation task, where the model produces code pieces in response to natural language instructions. This setting invites a natural and scalable way to acquire the textual feedback: the error messages and stack traces from code execution using a Python interpreter. LeTI iteratively fine-tunes the model, using the LM objective, on a concatenation of natural language instructions, LM-generated programs, and textual feedback, which is only provided when the generated program fails to solve the task. Prepended to this fine-tuning text, a binary reward token is used to differentiate correct and buggy solutions. On MBPP, a code generation dataset, LeTI substantially improves the performance of two base LMs of different scales. LeTI requires no ground-truth outputs for training and even outperforms a fine-tuned baseline that does. LeTI's strong performance generalizes to other datasets. Trained on MBPP, it achieves comparable or better performance than the base LMs on unseen problems in HumanEval. Furthermore, compared to binary feedback, we observe that textual feedback leads to improved generation quality and sample efficiency, achieving the same performance with fewer than half of the gradient steps. LeTI is equally applicable in natural language tasks when they can be formulated as code generation, which we empirically verified on event argument extraction. | ['Heng Ji', 'Reyhaneh Jabbarvand', 'Hao Peng', 'Xingyao Wang'] | 2023-05-17 | null | null | null | null | ['code-generation'] | ['computer-code'] | [ 3.43733996e-01 3.74792576e-01 -4.20837402e-01 -3.20704430e-01
-1.30133665e+00 -9.49500263e-01 5.74753284e-01 1.88785091e-01
-2.61162013e-01 7.22228289e-01 1.27683237e-01 -9.50971067e-01
3.96609247e-01 -9.17421162e-01 -1.39216721e+00 -5.20507395e-02
1.61275752e-02 3.06349993e-01 2.50731975e-01 -1.82307318e-01
4.34695512e-01 -1.90244988e-01 -1.40691066e+00 6.32184386e-01
9.18253481e-01 5.05440831e-01 1.94118485e-01 1.12368929e+00
-2.22062320e-01 1.05185115e+00 -6.91035151e-01 -3.20418030e-01
-3.19329537e-02 -2.69647568e-01 -1.03352869e+00 -3.89808148e-01
2.49372900e-01 -3.35804522e-01 1.44195020e-01 1.00241804e+00
3.01715527e-02 -2.05141768e-01 5.18779457e-01 -1.28381717e+00
-6.27990365e-01 1.25489533e+00 -3.76327306e-01 1.30484179e-01
5.70264101e-01 6.12638831e-01 1.26424086e+00 -6.82840645e-01
6.32724345e-01 1.38444984e+00 7.01582730e-01 8.66566896e-01
-1.60270905e+00 -4.40292031e-01 -8.37948930e-04 -2.95739293e-01
-6.66876316e-01 -2.32415155e-01 1.49196774e-01 -6.11142993e-01
1.57581627e+00 2.24588096e-01 -6.61389232e-02 1.30088830e+00
2.47752532e-01 8.14045370e-01 8.92162621e-01 -5.89109898e-01
2.00704023e-01 2.57838458e-01 3.50786239e-01 8.88749957e-01
4.32027839e-02 4.69669819e-01 -3.72181624e-01 -4.06580567e-01
3.47274423e-01 -3.01717460e-01 -1.46210179e-01 1.57588661e-01
-1.33751619e+00 8.54021788e-01 1.81846365e-01 2.80610412e-01
4.94093895e-02 5.57081461e-01 6.26898050e-01 5.31836867e-01
-9.31554660e-02 1.07176232e+00 -9.21454370e-01 -6.27047896e-01
-8.09957683e-01 2.33098656e-01 1.01642144e+00 9.68813002e-01
9.80957091e-01 8.84271190e-02 -4.34840590e-01 3.90626669e-01
2.53679663e-01 4.86451298e-01 9.44126606e-01 -9.31304514e-01
8.69350135e-01 8.29593301e-01 4.10977937e-02 -4.98556256e-01
-2.85779446e-01 -1.35631979e-01 -9.71800834e-02 3.83530706e-01
6.17448568e-01 -4.91221696e-01 -6.16156995e-01 2.04303598e+00
-2.25531124e-02 3.54086719e-02 1.16432332e-01 5.54131806e-01
5.31473100e-01 7.54973114e-01 -6.34725578e-03 1.50472000e-01
1.18091834e+00 -9.88697171e-01 -9.24795493e-02 -5.41961491e-01
1.17159450e+00 -6.07508659e-01 1.75302017e+00 4.26891118e-01
-1.07747507e+00 -5.35817325e-01 -9.80223715e-01 1.30830780e-01
-1.30504936e-01 2.32063457e-01 5.15250146e-01 5.02365589e-01
-1.21193182e+00 7.22913802e-01 -1.04188669e+00 4.72077057e-02
-4.32577403e-03 3.93824100e-01 2.50297878e-02 2.62581050e-01
-1.06394684e+00 6.54389799e-01 5.00973403e-01 -4.41013575e-01
-1.29372716e+00 -7.83868313e-01 -1.15969646e+00 1.57893673e-01
4.78518635e-01 -5.80455184e-01 1.93986070e+00 -1.09660339e+00
-1.60814607e+00 7.41130054e-01 -2.14499742e-01 -6.11019015e-01
3.86642098e-01 -2.02553391e-01 -5.99648803e-02 -3.79087687e-01
2.41786882e-01 6.23348653e-01 8.31159890e-01 -1.04865313e+00
-6.06729507e-01 2.53224313e-01 4.50131208e-01 -4.91804987e-01
-1.97288185e-01 1.40182644e-01 -1.80628628e-01 -3.83332908e-01
-6.56802356e-01 -9.37470138e-01 -8.60164538e-02 -5.67637563e-01
-7.14109838e-01 -2.87064046e-01 4.67738122e-01 -5.07175803e-01
1.43323314e+00 -2.09724379e+00 8.35226774e-02 1.33396387e-01
1.63876742e-01 1.25929520e-01 -4.90917087e-01 3.82919103e-01
-1.50346607e-01 6.03644550e-01 -3.89284492e-01 -1.18553236e-01
5.15579939e-01 2.00627565e-01 -6.05766177e-01 -1.00693628e-01
7.75625408e-01 1.11366558e+00 -1.15199649e+00 -2.99589574e-01
-1.50232419e-01 -2.21629255e-02 -1.07453787e+00 5.18404961e-01
-1.00130892e+00 2.25227341e-01 -4.69392270e-01 3.96638364e-01
3.56345661e-02 -7.12667048e-01 7.81225115e-02 2.94372112e-01
-5.68961985e-02 7.70172119e-01 -9.96775389e-01 1.41816092e+00
-9.58292663e-01 7.29346156e-01 -3.48403096e-01 -7.31851220e-01
7.06200480e-01 2.41301805e-01 -3.54154021e-01 -6.38065279e-01
-1.38922095e-01 4.70716596e-01 1.07257664e-01 -6.59522831e-01
3.96618217e-01 2.10348397e-01 -4.98279929e-01 9.02083457e-01
1.47665069e-01 -2.04691708e-01 4.70361263e-01 2.73457527e-01
1.62302506e+00 5.26372910e-01 3.67576629e-01 4.24758717e-02
5.82670629e-01 8.78691450e-02 2.81965315e-01 1.10601795e+00
3.22594702e-01 3.13059568e-01 9.35683429e-01 -8.83745253e-02
-1.05689549e+00 -8.34760129e-01 1.25458598e-01 1.65403092e+00
-3.28861713e-01 -6.87184155e-01 -1.06909037e+00 -1.09128630e+00
4.54890542e-02 1.11849964e+00 -6.52407944e-01 -3.88765007e-01
-9.55229640e-01 -5.77820778e-01 8.00454021e-01 6.07001066e-01
1.12668872e-01 -1.37266541e+00 -6.54467165e-01 3.09200078e-01
-2.22821608e-01 -7.77145982e-01 -7.27316618e-01 5.48575401e-01
-8.60785902e-01 -1.15449500e+00 -1.43895950e-02 -6.44886255e-01
7.28723884e-01 -4.47836161e-01 1.67269409e+00 5.09583235e-01
-2.07120359e-01 1.79071814e-01 -2.47395754e-01 -6.45173118e-02
-1.20042801e+00 3.72556120e-01 -2.19718084e-01 -5.15016854e-01
2.35299230e-01 -4.04789716e-01 -1.22943074e-01 3.10371429e-01
-8.73614192e-01 -3.84345464e-02 6.62653208e-01 1.14706278e+00
2.49491915e-01 -2.92918593e-01 5.39468646e-01 -1.09875715e+00
7.42773712e-01 -6.20512187e-01 -9.08177018e-01 2.58187324e-01
-6.68095767e-01 9.47380245e-01 1.04541838e+00 -5.95342815e-01
-9.01572943e-01 -1.40008211e-01 -1.51429281e-01 -2.38879560e-03
-2.04876408e-01 5.05460441e-01 1.24285482e-01 6.25765920e-02
1.29798830e+00 6.37481585e-02 -4.66152191e-01 -2.11307392e-01
4.72223163e-01 4.57034379e-01 7.27153480e-01 -1.31382179e+00
9.28312004e-01 -2.45105326e-01 -5.01444042e-01 5.81617747e-03
-7.11926937e-01 4.80528846e-02 -1.12006098e-01 3.65838319e-01
5.59107363e-01 -4.46025789e-01 -1.09392607e+00 6.86714752e-03
-1.27139258e+00 -1.09123480e+00 -2.50697345e-01 1.17183879e-01
-6.48802638e-01 8.77611116e-02 -1.02466893e+00 -7.36004174e-01
-2.53263593e-01 -1.47313166e+00 1.41552043e+00 3.16140540e-02
-7.05836535e-01 -1.03849876e+00 1.14920668e-01 -3.21289934e-02
3.08787763e-01 1.75873876e-01 1.37914336e+00 -7.89337873e-01
-7.24195600e-01 1.13667741e-01 -4.13006432e-02 3.87797624e-01
-1.52603518e-02 2.59827822e-01 -1.04737425e+00 -8.40288773e-02
-2.28689134e-01 -7.73153424e-01 6.83760405e-01 2.43408289e-02
1.09086716e+00 -6.45623565e-01 -2.53092021e-01 4.80276823e-01
1.19392323e+00 -1.86737012e-02 3.02421749e-01 5.67654848e-01
3.88611943e-01 2.98532516e-01 5.00852525e-01 3.34917158e-01
2.45667905e-01 4.23080772e-01 5.73281229e-01 2.12254882e-01
1.07936423e-04 -6.24521315e-01 1.02931714e+00 4.35007244e-01
3.76432002e-01 -1.26617983e-01 -1.04393899e+00 4.10319209e-01
-1.70706332e+00 -7.21370399e-01 -7.21226335e-02 2.25486588e+00
1.55037367e+00 5.25513530e-01 6.33175671e-02 6.72387928e-02
3.49615484e-01 -2.93665767e-01 -6.67860985e-01 -7.16788769e-01
1.82365596e-01 4.40894723e-01 2.86667973e-01 8.08730781e-01
-7.78998554e-01 9.64819789e-01 5.99043417e+00 7.03734219e-01
-1.25430238e+00 6.84306398e-02 6.18301988e-01 1.48077458e-01
-6.61816835e-01 6.80742264e-02 -1.12263680e+00 4.75598752e-01
1.50333047e+00 -2.30162829e-01 9.04317737e-01 9.44311798e-01
5.74737489e-02 -7.46235102e-02 -1.90808308e+00 4.93605465e-01
-2.89221019e-01 -1.32597411e+00 -2.44674101e-01 -1.58990115e-01
8.87710869e-01 1.29128247e-01 1.51115701e-01 9.93915200e-01
9.52261388e-01 -1.14821708e+00 9.55059767e-01 2.56619304e-01
7.16673493e-01 -4.50178683e-01 6.47710860e-01 7.38889098e-01
-8.29632342e-01 -1.97987169e-01 -1.48134977e-02 -2.89772987e-01
-1.12249583e-01 3.87013018e-01 -1.53497398e+00 4.62111495e-02
3.12612176e-01 4.37140644e-01 -7.78929830e-01 5.99357903e-01
-7.74454415e-01 1.15809202e+00 -1.77092597e-01 -2.71239311e-01
3.34019214e-01 3.50360602e-01 3.67612541e-01 1.54904115e+00
1.44878432e-01 -4.53080744e-01 2.14256510e-01 1.52618229e+00
-2.97602832e-01 -1.69354782e-01 -3.88121814e-01 -3.16713631e-01
3.77101213e-01 1.18274856e+00 -2.70763606e-01 -6.04155064e-01
-3.50899875e-01 7.53809154e-01 5.67301691e-01 4.26181525e-01
-9.60244358e-01 -3.55247855e-01 4.35844183e-01 -9.75359976e-02
1.44668803e-01 4.04592082e-02 -4.47688311e-01 -1.00371695e+00
3.69177908e-02 -1.38165832e+00 2.65904158e-01 -7.58577526e-01
-1.01543558e+00 6.16862833e-01 -1.08625978e-01 -8.00722122e-01
-1.17791569e+00 -8.08172762e-01 -6.64659262e-01 9.86767590e-01
-1.47551930e+00 -5.98874271e-01 -2.10423619e-02 2.38601461e-01
5.66909432e-01 1.38910905e-01 8.73932779e-01 1.17015384e-01
-5.29363513e-01 9.84766960e-01 -2.33220041e-01 1.36688218e-01
6.79111719e-01 -1.65264344e+00 9.03748870e-01 8.43337655e-01
3.86032113e-03 9.53208387e-01 7.60060072e-01 -6.97384477e-01
-1.47647941e+00 -1.29703033e+00 8.31996918e-01 -8.84222031e-01
1.03427482e+00 -3.72074366e-01 -1.05857873e+00 1.08411014e+00
4.55581285e-02 -1.39891461e-01 2.97239572e-01 2.27859728e-02
-6.54029965e-01 3.85072380e-01 -9.01576102e-01 5.39784372e-01
7.27979362e-01 -7.15107262e-01 -7.46814430e-01 4.17408198e-01
1.09553993e+00 -7.19946206e-01 -6.88673735e-01 9.18058157e-02
3.00300449e-01 -7.24896371e-01 6.65749013e-01 -1.01178586e+00
9.26434755e-01 -3.04974467e-01 -1.99302286e-01 -1.36792469e+00
1.08384624e-01 -8.85684967e-01 -3.03076059e-01 1.29834366e+00
1.02323771e+00 -6.27344489e-01 4.43068415e-01 6.24671280e-01
-2.23467857e-01 -7.15977609e-01 -2.40873277e-01 -7.13684916e-01
1.87641218e-01 -7.27520049e-01 6.87949061e-01 6.06356800e-01
2.21421793e-01 3.29809099e-01 -2.80984249e-02 1.70236722e-01
2.86266118e-01 1.72430068e-01 9.22229111e-01 -7.75689125e-01
-1.10863948e+00 -5.93172669e-01 1.68271050e-01 -1.03375590e+00
5.75800776e-01 -1.22182453e+00 4.07457232e-01 -9.14338708e-01
3.46051037e-01 -3.09497744e-01 2.01778393e-02 9.46452260e-01
-5.22077382e-01 -1.00452453e-01 8.00702907e-03 -7.88289532e-02
-5.73056459e-01 6.34977520e-02 8.52430463e-01 -1.91562831e-01
-2.48590231e-01 1.29901422e-02 -6.24216378e-01 7.85668910e-01
6.98495448e-01 -6.02912366e-01 -1.82695463e-01 -5.34777701e-01
7.48572528e-01 3.48398685e-01 4.97087359e-01 -7.54257083e-01
-1.89754236e-02 -1.54808968e-01 -5.90234771e-02 3.18466686e-02
-3.82044852e-01 -2.07794100e-01 -1.95747226e-01 5.88711679e-01
-8.51232290e-01 2.36704007e-01 5.29757380e-01 2.82232344e-01
-8.68278220e-02 -7.09897995e-01 4.62715924e-01 -2.13351950e-01
-5.11359096e-01 -1.68654293e-01 -3.94219875e-01 6.05682850e-01
5.27583420e-01 4.78337295e-02 -5.85027397e-01 -1.82792276e-01
-5.41811824e-01 3.00856590e-01 5.11256218e-01 3.78571719e-01
2.88173527e-01 -1.00116587e+00 -6.65813982e-01 2.62583673e-01
2.10151657e-01 -1.55866414e-01 -3.92172933e-01 7.37365961e-01
-1.85371280e-01 4.82295960e-01 2.35521868e-01 -7.02163994e-01
-1.04350483e+00 4.98412758e-01 3.84901673e-01 -7.22142994e-01
-2.51992911e-01 8.27397287e-01 4.05748069e-01 -9.27411318e-01
1.42099649e-01 -1.12772715e+00 2.21974775e-01 -5.33054411e-01
6.59107149e-01 1.16309963e-01 1.80590987e-01 8.57617185e-02
-2.10384130e-01 2.59879678e-01 1.42314136e-01 -1.25790268e-01
1.12728477e+00 3.36777687e-01 -1.16501123e-01 6.55883849e-01
1.13473678e+00 2.42439047e-01 -1.34683025e+00 -2.04822943e-01
4.13316101e-01 -1.04920074e-01 -3.77698600e-01 -1.14868832e+00
-5.72593212e-01 9.59508538e-01 9.53933522e-02 1.55576229e-01
7.64416397e-01 6.59145266e-02 7.24312484e-01 6.72320843e-01
5.56997716e-01 -5.70047975e-01 4.66285765e-01 7.75627017e-01
7.33598232e-01 -1.32056570e+00 -7.33176887e-01 1.66931711e-02
-2.92887956e-01 1.34325206e+00 9.18761194e-01 -1.26355700e-02
-2.21516475e-01 7.73107350e-01 -2.90996969e-01 1.91919163e-01
-1.15777409e+00 2.15213269e-01 1.57402217e-01 3.92005265e-01
7.94862688e-01 7.21946582e-02 2.54068505e-02 7.71951735e-01
-5.80797493e-01 8.24908465e-02 5.95790803e-01 8.08839142e-01
-4.50624466e-01 -1.26906216e+00 -5.48054695e-01 5.81615925e-01
-4.98912454e-01 -4.84401047e-01 -2.57580668e-01 6.57586753e-01
-8.41437932e-03 9.59443569e-01 -2.83962458e-01 -4.20511961e-01
2.16426164e-01 2.39216253e-01 2.74991781e-01 -1.01300609e+00
-1.08928943e+00 -4.03301179e-01 2.72280037e-01 -8.32807600e-01
3.03867668e-01 -5.40120661e-01 -1.72417057e+00 -1.79635480e-01
-2.76174545e-01 2.76509613e-01 4.72157389e-01 9.23772633e-01
2.43677735e-01 8.52598429e-01 4.74991590e-01 -8.22989285e-01
-1.14939368e+00 -8.88479352e-01 1.97553262e-01 3.87579620e-01
5.93114853e-01 -2.20062047e-01 -5.37182212e-01 3.69435847e-01] | [7.9670562744140625, 7.6883158683776855] |
07997bda-77be-4f86-8ea0-74862bc8e237 | forget-free-continual-learning-with-soft | 2303.14962 | null | https://arxiv.org/abs/2303.14962v1 | https://arxiv.org/pdf/2303.14962v1.pdf | Forget-free Continual Learning with Soft-Winning SubNetworks | Inspired by Regularized Lottery Ticket Hypothesis (RLTH), which states that competitive smooth (non-binary) subnetworks exist within a dense network in continual learning tasks, we investigate two proposed architecture-based continual learning methods which sequentially learn and select adaptive binary- (WSN) and non-binary Soft-Subnetworks (SoftNet) for each task. WSN and SoftNet jointly learn the regularized model weights and task-adaptive non-binary masks of subnetworks associated with each task whilst attempting to select a small set of weights to be activated (winning ticket) by reusing weights of the prior subnetworks. Our proposed WSN and SoftNet are inherently immune to catastrophic forgetting as each selected subnetwork model does not infringe upon other subnetworks in Task Incremental Learning (TIL). In TIL, binary masks spawned per winning ticket are encoded into one N-bit binary digit mask, then compressed using Huffman coding for a sub-linear increase in network capacity to the number of tasks. Surprisingly, in the inference step, SoftNet generated by injecting small noises to the backgrounds of acquired WSN (holding the foregrounds of WSN) provides excellent forward transfer power for future tasks in TIL. SoftNet shows its effectiveness over WSN in regularizing parameters to tackle the overfitting, to a few examples in Few-shot Class Incremental Learning (FSCIL). | ['Chang D. Yoo', 'Sung Ju Hwang', 'Sultan Rizky Madjid', 'Jaehong Yoon', 'Haeyong Kang'] | 2023-03-27 | null | null | null | null | ['class-incremental-learning', 'few-shot-class-incremental-learning'] | ['computer-vision', 'methodology'] | [ 1.10096073e+00 5.07942379e-01 6.63797744e-03 -2.35922411e-01
1.77540123e-01 -1.24687046e-01 6.15076900e-01 -1.49379596e-01
-7.26699054e-01 1.20129931e+00 -3.17425281e-01 1.16182044e-01
-7.18668401e-01 -8.48292172e-01 -1.04121625e+00 -9.83009934e-01
-7.25748360e-01 5.44161379e-01 1.01080382e+00 -6.56597763e-02
-1.51928633e-01 3.76158923e-01 -1.83196115e+00 3.83375287e-01
4.88925576e-01 1.18237448e+00 6.02943242e-01 9.15342748e-01
-5.19294180e-02 1.11892354e+00 -8.35073948e-01 -2.27288976e-01
4.66260105e-01 -3.01062167e-01 -8.29245806e-01 -6.20720014e-02
1.93542525e-01 2.39732027e-01 -5.66083670e-01 9.33658898e-01
2.93703705e-01 1.52801171e-01 6.84030116e-01 -1.39753759e+00
-1.16482630e-01 1.12220383e+00 -3.48211378e-01 4.08909142e-01
-4.07083213e-01 6.38811588e-01 5.23693144e-01 -5.67724168e-01
5.58789968e-01 1.13493609e+00 9.38509881e-01 5.70286989e-01
-1.28039742e+00 -9.41714883e-01 2.49868840e-01 1.35555834e-01
-1.24788749e+00 -5.98518670e-01 7.19447553e-01 3.28759179e-02
9.41116035e-01 3.47484142e-01 9.93535459e-01 1.36291528e+00
6.83668792e-01 6.66877449e-01 1.06110501e+00 -2.08699673e-01
6.60126507e-01 1.84634298e-01 3.43285874e-02 7.57248223e-01
3.20282638e-01 1.49908438e-01 -9.18884277e-01 1.44216016e-01
8.61703396e-01 1.86943412e-01 8.86026323e-02 6.27249703e-02
-1.17711377e+00 4.11433965e-01 6.37046456e-01 3.17409396e-01
-7.95364082e-01 4.82697874e-01 4.19587851e-01 8.44648242e-01
4.38442737e-01 2.32675970e-01 -8.40472817e-01 2.53441840e-01
-1.15675104e+00 -2.99273729e-01 5.79753339e-01 8.83094490e-01
1.02316308e+00 5.75641394e-01 -3.73898387e-01 6.53795362e-01
-1.36459291e-01 2.71489531e-01 6.96406722e-01 -8.63202989e-01
-4.36688475e-02 5.16402006e-01 -4.19878632e-01 -6.05736494e-01
-4.33347821e-01 -8.60226750e-01 -1.43907988e+00 3.76605868e-01
2.68321428e-02 -4.04032081e-01 -1.25603449e+00 1.64475155e+00
2.16030061e-01 6.34140193e-01 1.26918435e-01 9.72660258e-02
3.07643592e-01 7.94937789e-01 3.19302440e-01 -4.88907605e-01
1.14638937e+00 -8.11755061e-01 -3.44094962e-01 -5.86903989e-01
-1.25260890e-01 -2.10198611e-01 9.42765892e-01 5.08228481e-01
-9.70897257e-01 -9.78629768e-01 -1.07501459e+00 8.58482957e-01
-4.10968363e-01 -5.55145025e-01 6.61230087e-01 4.72194493e-01
-1.29852700e+00 6.73141599e-01 -8.93056214e-01 -3.14247578e-01
8.03215504e-01 7.93679833e-01 1.60634905e-01 -1.78573027e-01
-1.22050810e+00 5.85138261e-01 8.15018594e-01 -2.01674819e-01
-1.61600542e+00 -6.11290038e-01 -5.77660441e-01 6.82443082e-02
6.09766781e-01 -7.59702623e-01 8.35804939e-01 -1.48428404e+00
-1.47383356e+00 3.36181760e-01 6.65196255e-02 -1.06710792e+00
3.20996940e-01 2.43548945e-01 -3.65637571e-01 5.05244732e-02
-1.43815205e-01 1.25096834e+00 1.62536287e+00 -1.16684341e+00
-8.61973882e-01 -4.72047329e-02 -1.09393895e-01 1.38563991e-01
-8.00987542e-01 -5.83374560e-01 -1.57031134e-01 -8.10615718e-01
1.30785778e-02 -6.51020646e-01 -3.82177323e-01 7.26679936e-02
-3.19602191e-01 -3.69250387e-01 1.21248162e+00 -1.03678331e-01
1.05749524e+00 -2.00578141e+00 5.51434159e-02 1.66895196e-01
3.26380581e-01 2.58616626e-01 -1.81732699e-01 -2.94308484e-01
1.82017893e-01 -1.45039901e-01 -4.70924765e-01 -2.62188137e-01
-4.86571729e-01 8.77462149e-01 2.06500404e-02 2.56060183e-01
2.98567712e-01 1.04101884e+00 -9.66303408e-01 -6.92367733e-01
3.77235338e-02 2.72543967e-01 -4.13830616e-02 -1.13930227e-02
-3.22497666e-01 1.82734337e-02 -2.17518240e-01 8.02299321e-01
4.37787890e-01 -4.00583565e-01 -6.71747103e-02 -1.51277035e-01
2.44086221e-01 -2.34477133e-01 -1.03138649e+00 1.22946119e+00
-2.96084851e-01 5.17833889e-01 2.82186538e-01 -1.30452716e+00
8.27373505e-01 1.54804826e-01 4.92209584e-01 -6.27946079e-01
-6.19893856e-02 -9.19285715e-02 7.74966925e-03 -2.48139232e-01
2.82126248e-01 -3.71678174e-01 -2.73956895e-01 2.76046813e-01
7.47817039e-01 1.47845829e-02 1.18377104e-01 1.76131830e-01
1.67642868e+00 -3.27392310e-01 1.19549349e-01 -2.88917571e-01
3.09862494e-01 1.08196959e-01 6.48344576e-01 1.34321296e+00
-2.48323962e-01 4.65534441e-02 4.41113152e-02 -6.51954949e-01
-1.01356757e+00 -1.14715326e+00 -4.54315953e-02 1.53711236e+00
-4.68614846e-02 3.57875496e-01 -4.09288824e-01 -8.31950307e-01
-4.94184569e-02 7.79323578e-01 -6.90110683e-01 -4.81580913e-01
-6.21529818e-01 -9.69759226e-01 7.45888591e-01 1.86929688e-01
7.71185160e-01 -1.60358524e+00 -9.44850385e-01 5.51058829e-01
3.63704354e-01 -7.50306904e-01 -2.50638127e-01 1.26086712e+00
-1.00118375e+00 -9.59934533e-01 -7.60654390e-01 -1.05182910e+00
8.09414685e-01 9.63475555e-02 9.32237685e-01 -7.72084519e-02
-5.10464728e-01 6.14697516e-01 -1.11908585e-01 -5.59616983e-01
-4.36169207e-01 1.94859520e-01 3.02047133e-01 1.38763294e-01
-2.30246529e-01 -8.22913766e-01 -3.50988775e-01 2.06264332e-01
-1.08193648e+00 -9.46211256e-03 1.05437124e+00 8.50076020e-01
7.10153639e-01 9.38780308e-01 8.54971707e-01 -9.76902962e-01
4.37006652e-01 -6.67478383e-01 -3.78361642e-01 3.56007814e-01
-6.78607881e-01 8.96908268e-02 8.01677883e-01 -1.08837938e+00
-1.01524317e+00 3.29816759e-01 3.54664832e-01 -5.96316099e-01
8.22447538e-02 8.74429196e-02 3.16224396e-01 -4.28963751e-01
6.70682669e-01 6.00618422e-01 9.52880904e-02 6.66379696e-03
1.33484572e-01 2.00562686e-01 7.09185600e-01 -2.84245759e-01
9.98535573e-01 5.01327157e-01 1.57121778e-01 -7.79270709e-01
-8.17652583e-01 -6.58171624e-02 -7.23809004e-01 -7.33977377e-01
6.63643181e-01 -8.85711133e-01 -4.66700226e-01 6.98873639e-01
-7.28565395e-01 -7.92636573e-01 -1.08601391e+00 -4.01480198e-02
-2.72163004e-01 -1.08111136e-01 -3.51877034e-01 -1.04216623e+00
-3.69905204e-01 -6.41735017e-01 5.08013129e-01 2.92655736e-01
-9.68624428e-02 -1.01883936e+00 -4.03412610e-01 -2.37545043e-01
8.74848902e-01 8.17304850e-02 8.01407874e-01 -6.85977936e-01
-4.36733186e-01 -8.36010352e-02 9.87482220e-02 6.49833977e-01
1.51425943e-01 -4.32100445e-01 -9.17207956e-01 -5.84245026e-01
2.58272439e-01 -6.23388112e-01 1.22546554e+00 4.71365273e-01
1.19180381e+00 -7.13749111e-01 -5.16736209e-01 3.60981673e-01
1.30998111e+00 1.52743831e-01 3.84464294e-01 7.00635910e-02
3.88131350e-01 2.80719668e-01 2.21034452e-01 3.40283453e-01
-4.06772308e-02 -1.25458911e-01 6.39701545e-01 -1.50099188e-01
-6.28861427e-01 7.14303255e-02 6.28132880e-01 7.50846505e-01
9.81042236e-02 -3.70914817e-01 -6.94090903e-01 5.64244926e-01
-1.64755750e+00 -9.04231727e-01 3.30544740e-01 2.00798345e+00
1.13058269e+00 1.16553569e+00 -2.90826291e-01 2.32071787e-01
7.87130475e-01 1.79326534e-01 -1.07036948e+00 1.33127213e-01
-4.68575656e-01 6.00466132e-01 9.11592007e-01 2.52273649e-01
-1.09973359e+00 9.56984162e-01 6.35850239e+00 1.23718953e+00
-1.02988529e+00 3.50563794e-01 9.82811332e-01 -2.75446653e-01
-9.13030356e-02 -3.02136302e-01 -8.50267708e-01 4.61744905e-01
1.32099104e+00 -6.67910799e-02 6.73779190e-01 7.79867947e-01
-2.30848268e-01 -2.85423279e-01 -7.93976724e-01 3.90691161e-01
-1.50721714e-01 -1.47179472e+00 1.66418791e-01 -3.81767154e-01
9.79273140e-01 2.99655139e-01 1.15540646e-01 7.62193263e-01
8.53485346e-01 -8.60943556e-01 7.72287786e-01 6.60691082e-01
1.02778316e+00 -5.43630719e-01 7.54678845e-01 6.32341683e-01
-1.28775418e+00 -7.00098574e-01 -7.08546281e-01 3.47210094e-02
-7.17990249e-02 7.43498385e-01 -1.20483577e+00 1.98843941e-01
8.96291018e-01 4.54831272e-01 -6.90108180e-01 9.49468613e-01
2.04237595e-01 1.11297214e+00 -6.57009780e-01 -1.02112936e-02
4.28323448e-01 5.28971672e-01 7.59377420e-01 1.22601807e+00
2.48300329e-01 -2.55610079e-01 2.12429479e-01 6.37199879e-01
-2.85412401e-01 -6.32995367e-01 -5.70724189e-01 2.51402199e-01
6.86208665e-01 1.04203415e+00 -1.22809052e+00 -5.63503325e-01
2.01007328e-03 1.21699274e+00 1.47118375e-01 3.19932044e-01
-6.39364660e-01 -3.66242230e-01 3.27800691e-01 1.49398655e-01
5.57953119e-01 1.05582349e-01 -3.93938959e-01 -3.93236697e-01
-3.93644303e-01 -4.78422344e-01 5.10686696e-01 -5.61824977e-01
-1.39480472e+00 5.93590915e-01 -8.00469238e-03 -1.00597692e+00
1.50865436e-01 -9.65222344e-02 -7.82885075e-01 3.07891250e-01
-1.44866574e+00 -1.03821874e+00 -3.71331125e-01 9.83238161e-01
1.19916487e+00 -5.71395099e-01 5.75362206e-01 1.02016237e-02
-4.75521773e-01 4.84900713e-01 -1.07083060e-02 -3.19888532e-01
3.83453757e-01 -9.34446633e-01 3.11113656e-01 6.00960374e-01
-3.21139060e-02 1.30068019e-01 4.69416082e-01 -9.98480916e-01
-1.25066650e+00 -1.67875957e+00 4.70075786e-01 2.76371371e-02
6.67742193e-01 -5.25929153e-01 -8.52009416e-01 5.46736896e-01
9.88116339e-02 2.98876017e-01 4.00325842e-02 -3.95142645e-01
2.31136948e-01 -5.85170329e-01 -1.39554524e+00 2.16090932e-01
1.11509407e+00 2.79825255e-02 -3.32046241e-01 4.31096554e-01
1.30032456e+00 4.94751409e-02 -3.64979476e-01 4.58506793e-01
1.56166151e-01 -5.99939466e-01 1.13289404e+00 -4.35575664e-01
-1.27663910e-01 -1.54339015e-01 1.30355373e-01 -1.00155509e+00
-6.56090200e-01 -7.70641088e-01 -2.67065048e-01 9.24617231e-01
4.92019981e-01 -5.27359426e-01 1.03608716e+00 -6.53544739e-02
-2.98202991e-01 -6.33884132e-01 -1.56686449e+00 -1.02214885e+00
-5.68418562e-01 -3.61719757e-01 2.69069731e-01 8.60244751e-01
-6.82577550e-01 4.08795267e-01 -5.46489120e-01 1.21810734e-01
1.04543424e+00 -7.90776849e-01 2.59958565e-01 -1.59562528e+00
-2.93938130e-01 -2.88037688e-01 -2.12045595e-01 -7.82997191e-01
-8.90848562e-02 -1.03276491e+00 2.78296649e-01 -9.68251348e-01
1.30766481e-01 -8.13009143e-01 -8.71424079e-01 9.81166542e-01
8.62771645e-02 2.97598928e-01 4.76703458e-02 3.99721771e-01
-1.11407578e+00 4.25759971e-01 1.16565180e+00 -5.26327312e-01
-2.10455805e-01 3.79983187e-01 -4.12947506e-01 6.80768728e-01
7.42429018e-01 -1.02597642e+00 -7.13616312e-01 -1.19075924e-01
1.36218786e-01 -2.71932986e-02 4.33743387e-01 -1.70379615e+00
7.55925477e-01 -1.27968282e-01 7.62803257e-01 -3.44397157e-01
2.75170267e-01 -9.87454474e-01 2.92726249e-01 1.04264653e+00
-3.94733518e-01 -1.96471468e-01 2.24101171e-01 1.14381099e+00
4.09489959e-01 -4.09972280e-01 8.76532376e-01 -5.65032840e-01
-7.63661921e-01 2.72106141e-01 -8.11431050e-01 -1.01494886e-01
1.18184912e+00 -7.10377455e-01 2.20995769e-02 -2.60999892e-02
-9.12551105e-01 3.48813206e-01 -1.11925542e-01 1.21410571e-01
8.46212029e-01 -9.42207038e-01 -6.51721358e-01 4.20986831e-01
-5.35310328e-01 4.64859128e-01 1.93222895e-01 5.25792897e-01
1.42541170e-01 -1.02659725e-01 -4.58500743e-01 -6.27900481e-01
-1.02031386e+00 2.52642810e-01 1.88934520e-01 -5.46433926e-01
-5.78284442e-01 1.57105231e+00 -1.22563817e-01 -2.23075494e-01
8.02768350e-01 -1.99886225e-02 -9.60462838e-02 2.45442111e-02
2.60982811e-01 5.48862994e-01 -4.94204834e-02 6.38110414e-02
-1.42301738e-01 -5.50949164e-02 -1.56277627e-01 2.54587561e-01
1.79985785e+00 -1.34623379e-01 -1.76701099e-01 7.30096161e-01
8.55124652e-01 -7.82631695e-01 -1.81662440e+00 -6.38070285e-01
8.73894617e-02 3.32883418e-01 2.62731880e-01 -9.79146123e-01
-1.11753941e+00 2.61721432e-01 9.51352477e-01 3.85374427e-01
1.29148221e+00 7.55077824e-02 7.73442984e-01 6.95934772e-01
5.12087226e-01 -1.24503410e+00 7.58103371e-01 7.61793017e-01
5.95647216e-01 -8.71676922e-01 4.02640775e-02 1.66147873e-01
-5.63728094e-01 1.07775187e+00 6.73892736e-01 -1.01397835e-01
9.60593343e-01 8.00645530e-01 -5.75354517e-01 -3.71642649e-01
-1.15974605e+00 2.75314245e-02 -1.53912678e-01 8.88212025e-01
-3.82673502e-01 4.60623093e-02 4.54725921e-01 4.76718336e-01
-8.05876404e-02 3.98476757e-02 2.34483078e-01 1.07997310e+00
-1.07699978e+00 -4.92630720e-01 -3.09931457e-01 1.08204365e+00
6.40115663e-02 -1.97860748e-01 -1.06234290e-01 5.18068850e-01
6.21222973e-01 4.96091634e-01 3.62943947e-01 -6.81316972e-01
5.00887185e-02 3.23276669e-01 2.99660355e-01 -6.27033293e-01
-9.07089651e-01 -1.58506468e-01 -2.18886331e-01 -2.07493067e-01
-2.82372892e-01 -5.37748754e-01 -1.19398904e+00 -1.40417755e-01
-1.81626156e-01 -2.34031647e-01 2.30454624e-01 8.12630117e-01
4.38773558e-02 1.11103404e+00 6.16616130e-01 -1.14732718e+00
-5.08612335e-01 -1.07338703e+00 -4.81655896e-01 -8.19148049e-02
4.37193483e-01 -5.35117805e-01 -5.93733966e-01 3.78745675e-01] | [9.80077075958252, 3.3889567852020264] |
b34884bf-d4ef-4031-8396-ab3197196be7 | multimodal-behavioral-markers-exploring | null | null | https://dl.acm.org/doi/abs/10.1145/3340555.3353718 | https://dl.acm.org/doi/pdf/10.1145/3340555.3353718 | Multimodal Behavioral Markers Exploring Suicidal Intent in Social Media Videos | Suicide is one of the leading causes of death in the modern world. In this digital age, individuals are increasingly using social media to express themselves and often use these platforms to express suicidal intent. Various studies have inspected suicidal intent behavioral markers in controlled environments but it is still unexplored if such markers will generalize to suicidal intent expressed on social media. In this work, we set out to study multimodal behavioral markers related to suicidal intent when expressed on social media videos. We explore verbal, acoustic and visual behavioral markers in the context of identifying individuals at higher risk of suicidal attempt. Our analysis reveals that frequent silences, slouched shoulders, rapid hand movements and profanity are predominant multimodal behavioral markers indicative of suicidal intent | ['Jeffrey M. Girard', 'Louis Philippe Morency', 'Vaibhav Vaibhav', 'Mahmoud Al Ismail', 'Ankit Parag Shah', 'Vasu Sharma'] | 2019-10-01 | null | null | null | international-conference-on-multimodal | ['multimodal-emotion-recognition', 'multimodal-emotion-recognition'] | ['computer-vision', 'speech'] | [ 1.92243859e-01 -7.02669546e-02 -1.01674698e-01 -2.31508657e-01
-8.89104068e-01 -2.28752345e-01 6.34381831e-01 4.62476432e-01
-5.84222317e-01 4.19627547e-01 9.30331528e-01 4.76305693e-01
1.04626231e-02 -2.67108738e-01 2.71209359e-01 -3.77729565e-01
-3.41132164e-01 -8.46078843e-02 -1.97784662e-01 -2.68393874e-01
5.55981159e-01 1.89245105e-01 -9.08905447e-01 3.06521922e-01
2.69296229e-01 6.35261297e-01 -2.56800950e-01 8.96428585e-01
6.61919219e-03 8.46896052e-01 -7.65539527e-01 -5.96662879e-01
-5.05124390e-01 -3.03806424e-01 -4.04569030e-01 -1.49913147e-01
2.53129363e-01 -6.75170004e-01 -5.76389968e-01 8.61083150e-01
9.61434305e-01 -3.59302759e-02 6.77215636e-01 -1.10455370e+00
-1.00663602e-01 7.48320222e-01 -5.11326492e-01 4.40401912e-01
1.57828736e+00 4.54408042e-02 4.81357872e-01 -5.69667220e-01
4.98317540e-01 1.23450339e+00 1.17937422e+00 9.70043421e-01
-1.05425143e+00 -8.64727497e-01 -3.81176740e-01 1.72631338e-01
-1.55976844e+00 -8.89444888e-01 1.09089339e+00 -3.38232517e-01
7.82121599e-01 1.35148749e-01 9.42998827e-01 2.00102854e+00
4.02257085e-01 5.02872229e-01 4.83144760e-01 -1.05467193e-01
2.63264537e-01 -2.67398376e-02 1.63881868e-01 4.22763973e-01
6.80143163e-02 -5.41398883e-01 -1.19211376e+00 -7.53680289e-01
2.84259692e-02 3.48747224e-01 -1.73417300e-01 4.42532778e-01
-1.25283837e+00 7.86360562e-01 -2.50471979e-01 3.70087028e-01
-5.79144001e-01 4.01123732e-01 8.17141473e-01 -3.45118076e-01
4.64377046e-01 -7.29614357e-03 4.33364630e-01 -1.32343364e+00
-9.83902276e-01 -1.10317320e-01 6.25389218e-01 4.31126952e-01
2.46379822e-01 -5.06400727e-02 1.27747878e-01 1.06622696e+00
4.84733313e-01 7.12890387e-01 6.34491980e-01 -9.08721507e-01
5.22310019e-01 7.24996984e-01 -3.73301804e-01 -1.56441081e+00
-9.33642685e-01 5.96692383e-01 -7.26507008e-01 -4.76702362e-01
-1.82027802e-01 -5.98219752e-01 2.27566555e-01 1.40927124e+00
6.81775883e-02 -5.99917434e-02 -2.58032531e-02 4.84279126e-01
1.16688168e+00 2.99764961e-01 4.17282671e-01 -2.29530856e-01
1.31105912e+00 3.07625234e-01 -9.66888964e-01 -3.78009051e-01
9.66297209e-01 -4.77259874e-01 7.48073816e-01 3.41676921e-01
-9.07247424e-01 1.02025777e-01 -6.87999308e-01 2.68961430e-01
-5.00130728e-02 -4.53704968e-02 4.62652892e-01 1.37826955e+00
-8.27736795e-01 4.59276199e-01 -7.15040982e-01 -1.03880763e+00
2.82589972e-01 3.28274071e-01 -8.90088618e-01 3.40043426e-01
-1.04395080e+00 4.58418846e-01 -6.47373945e-02 -1.78428859e-01
-3.60588759e-01 -4.44221377e-01 -1.02800703e+00 -7.11010754e-01
-2.21228078e-01 -1.77524850e-01 9.79240477e-01 -7.36460984e-01
-1.40609348e+00 1.01728666e+00 -1.11879632e-02 -1.98888049e-01
2.66956836e-01 -2.35139042e-01 -7.82826900e-01 7.88235247e-01
1.46366611e-01 5.34624815e-01 7.03274310e-01 -5.25233090e-01
-1.18951440e-01 -6.99858665e-01 -2.76099205e-01 2.44040608e-01
-1.05029213e+00 7.20513821e-01 3.05481672e-01 -4.58670259e-01
-6.83995187e-02 -7.87417173e-01 2.86241084e-01 -2.22636219e-02
-5.89677572e-01 -1.02965631e-01 1.23794150e+00 -6.29106522e-01
1.42203927e+00 -2.46231890e+00 -2.34965727e-01 -1.26801908e-01
3.56401443e-01 -9.21118930e-02 1.94217309e-01 1.05853891e+00
1.95425496e-01 5.38164675e-01 1.59809932e-01 -6.83167338e-01
2.10578471e-01 -1.93010449e-01 -1.40355453e-01 1.14895451e+00
-3.22832257e-01 5.32260060e-01 -8.96165311e-01 -7.43737757e-01
2.11297974e-01 6.70442522e-01 -4.22850549e-01 2.51814753e-01
8.11960340e-01 1.54094577e-01 -4.99435991e-01 7.73738146e-01
5.13033509e-01 3.01624268e-01 7.42269233e-02 1.09887943e-02
-2.10626096e-01 4.94293161e-02 -7.22481430e-01 1.17869282e+00
-5.21475747e-02 9.08752203e-01 1.92520857e-01 -4.47498769e-01
8.92227292e-01 5.41945159e-01 8.60852242e-01 -5.11028290e-01
2.99228430e-01 -1.89907581e-01 -3.78988445e-01 -9.35106397e-01
6.61463201e-01 -3.96033466e-01 -7.05892444e-01 4.76657271e-01
-2.09855229e-01 7.38208592e-02 -1.92499444e-01 4.24269468e-01
1.27760875e+00 -4.87542361e-01 6.13520443e-01 8.94135311e-02
1.69288382e-01 -4.56131309e-01 1.68021679e-01 6.35009110e-01
-5.61765075e-01 7.38630176e-01 5.46875298e-01 -2.00105608e-01
-2.77061582e-01 -9.01593208e-01 -5.00941388e-02 9.27909672e-01
-1.62240818e-01 -1.03229403e+00 -6.66579962e-01 -3.79737258e-01
-3.29970747e-01 6.35905981e-01 -4.68644619e-01 -4.50682759e-01
-2.97857046e-01 -5.64981520e-01 1.31599879e+00 3.55827272e-01
6.46203637e-01 -1.14173961e+00 -8.07045043e-01 3.30437273e-02
-2.55013555e-01 -1.11508965e+00 -4.78246659e-01 -3.53341460e-01
-6.49516225e-01 -9.12942946e-01 -9.06492472e-01 -2.00073108e-01
2.80014873e-01 9.89174396e-02 4.43517119e-01 -6.71858191e-02
-4.20472831e-01 1.36103702e+00 -5.92297077e-01 -8.24266002e-02
-1.88048661e-01 -1.69989571e-01 5.94283462e-01 3.28169346e-01
4.01709944e-01 -5.33889115e-01 -5.94649613e-01 6.67544156e-02
-6.15054846e-01 -3.69557858e-01 -2.09817335e-01 -1.08641423e-01
-2.41589099e-01 -4.71477628e-01 2.87612647e-01 -2.73957342e-01
1.01460433e+00 -9.69351947e-01 4.73068774e-01 -4.72638518e-01
2.75076777e-01 -9.85020518e-01 3.71842474e-01 -5.39968371e-01
-6.49875820e-01 -1.27044261e-01 -3.47247660e-01 5.89723326e-02
-8.73891890e-01 2.89363593e-01 1.49057582e-01 -1.75558969e-01
3.44443113e-01 2.50362575e-01 -7.40981996e-02 -6.08864538e-02
-4.81543601e-01 1.24802673e+00 1.66102707e-01 2.54787151e-02
1.90970540e-01 6.67934537e-01 -1.05889484e-01 -1.95169508e+00
-1.79332674e-01 -8.80329132e-01 -4.41554934e-01 -1.00010812e+00
1.24197340e+00 -5.68955958e-01 -1.30023921e+00 7.82829881e-01
-1.04254699e+00 -1.72021702e-01 2.19147235e-01 5.55217564e-01
-5.42815506e-01 1.10881627e+00 -7.24673748e-01 -1.37274277e+00
-4.72308546e-01 -7.34998345e-01 1.32869267e+00 1.92761391e-01
-1.26483250e+00 -1.09085393e+00 4.18012679e-01 4.45629150e-01
2.97540665e-01 1.75310522e-01 2.89608240e-01 -9.25845981e-01
6.88039899e-01 -6.82129741e-01 4.43499023e-03 -1.65451765e-01
3.27597469e-01 -2.02736501e-02 -6.78115487e-01 1.20086141e-01
6.97647855e-02 -6.00093603e-01 1.57606438e-01 1.36687815e-01
6.35101676e-01 -5.02864361e-01 -4.76202279e-01 2.73843974e-01
4.89323795e-01 4.16308433e-01 8.62168431e-01 2.37915620e-01
4.59835082e-01 7.54571617e-01 7.12861195e-02 1.26627362e+00
4.11773533e-01 7.92185545e-01 3.65831822e-01 6.67663872e-01
1.95457339e-01 -2.05865249e-01 9.97925878e-01 5.31660855e-01
-3.14344503e-02 -5.77615857e-01 -1.31994081e+00 1.62361220e-01
-1.16758192e+00 -1.42654467e+00 -4.17016536e-01 1.97914886e+00
1.57505333e-01 -3.50081831e-01 6.92045808e-01 4.12281662e-01
9.49349642e-01 3.23448598e-01 2.97249965e-02 -4.32638168e-01
-1.44775491e-02 -4.20248479e-01 -6.64228052e-02 1.68047339e-01
-1.03113830e+00 3.53094429e-01 7.08901215e+00 3.84179741e-01
-1.27807152e+00 3.03583383e-03 3.80139560e-01 -2.04209208e-01
-1.79441556e-01 -4.11550343e-01 -9.07106280e-01 6.56852722e-01
1.21096849e+00 1.33276060e-01 -3.35329771e-02 4.73277032e-01
5.97121000e-01 -4.43116188e-01 -7.98354387e-01 1.25594604e+00
8.44156027e-01 -5.78346491e-01 -2.82225341e-01 2.75331944e-01
-2.64905661e-01 -3.27603430e-01 2.00026169e-01 -1.40067443e-01
-4.39171255e-01 -9.48584378e-01 5.84620416e-01 7.83699095e-01
8.72090220e-01 -5.73371053e-01 6.22631550e-01 2.76332289e-01
-1.04644966e+00 -9.39247310e-02 1.38442442e-01 -1.09868385e-01
7.47156262e-01 1.33183703e-01 -9.85230148e-01 -2.10837051e-01
5.34103632e-01 1.10435438e+00 -6.53559506e-01 9.90667522e-01
2.00048670e-01 8.25060189e-01 -5.11885047e-01 -5.93550146e-01
8.27448517e-02 -7.63788223e-02 9.35397983e-01 1.43799078e+00
6.71420276e-01 6.12718500e-02 -1.58707872e-01 3.70986521e-01
3.58592570e-01 2.20040917e-01 -1.25474751e+00 -6.39921963e-01
-1.57498509e-01 1.10288513e+00 -6.68256462e-01 3.83994311e-01
-2.94018984e-01 1.19636822e+00 -1.70207605e-01 -7.40212426e-02
-8.00032675e-01 -2.53415823e-01 3.73802841e-01 4.34785843e-01
-6.34658396e-01 -4.28260505e-01 4.83358791e-03 -9.17698920e-01
-2.70810515e-01 -3.99258316e-01 5.64279616e-01 -7.39407241e-01
-1.14667511e+00 1.73900783e-01 4.68305126e-03 -1.23122942e+00
-3.80499899e-01 -2.64628585e-02 -1.07039905e+00 -7.60298083e-03
-1.76161706e-01 -7.04257965e-01 -4.12060529e-01 7.49634087e-01
5.41186750e-01 -1.49970502e-01 9.87849891e-01 2.05965549e-01
-7.66681910e-01 5.00785589e-01 -4.99450147e-01 8.85902643e-02
7.41968930e-01 -6.56972766e-01 -2.85215288e-01 2.04366609e-01
-3.91797841e-01 6.39984667e-01 6.81062102e-01 -1.10064650e+00
-1.77391613e+00 -6.30935669e-01 9.38604951e-01 -2.58774966e-01
9.25801158e-01 -3.92983496e-01 -2.96968728e-01 5.21204114e-01
3.39710802e-01 -4.27986085e-01 1.35736716e+00 -1.02119736e-01
1.29331231e-01 1.62720934e-01 -1.40760934e+00 8.77000391e-01
9.54735696e-01 -6.75162613e-01 -4.51882243e-01 3.73642176e-01
-2.75599957e-01 2.54799992e-01 -8.15858841e-01 3.52121145e-02
7.63918281e-01 -1.25639415e+00 8.50590229e-01 -3.08610853e-02
2.05029830e-01 6.07770085e-01 -2.69903421e-01 -9.57388818e-01
6.57869652e-02 -1.01806557e+00 5.65414488e-01 1.61121583e+00
-5.05709387e-02 -6.19567633e-01 7.16423690e-01 9.08138096e-01
2.08043563e-03 -6.87948763e-02 -1.42900097e+00 -2.61631310e-01
-3.28732193e-01 -7.25507021e-01 -1.29273176e-01 6.36521876e-01
1.19547939e+00 6.57335948e-03 -5.65664768e-01 -1.63630962e-01
4.58622217e-01 -8.73545945e-01 5.97075760e-01 -8.58839393e-01
3.13584685e-01 -4.63346064e-01 -1.06666529e+00 -3.79849374e-01
2.27020741e-01 -4.00173962e-01 -2.08961278e-01 -1.22282839e+00
5.47237098e-01 2.18092009e-01 4.14487779e-01 2.56833613e-01
4.38778490e-01 6.23911977e-01 9.01590586e-02 -1.57301515e-01
-1.01380134e+00 6.08435988e-01 4.36829180e-01 -3.28869998e-01
-1.96726903e-01 -9.87196118e-02 -3.50654006e-01 1.21841097e+00
8.65447283e-01 -4.61755991e-01 -2.56858468e-02 3.49118203e-01
6.33990288e-01 6.06511056e-01 4.61434990e-01 -1.27675354e+00
4.57351327e-01 8.40438809e-03 1.62749186e-01 -4.62167621e-01
8.91907811e-01 -5.02431273e-01 1.36990651e-01 5.06405652e-01
-2.42634505e-01 2.00904265e-01 2.27154404e-01 3.21425200e-01
1.93344075e-02 -5.50192058e-01 4.76203650e-01 1.51154771e-01
-2.70933151e-01 -7.83210918e-02 -1.65465868e+00 2.13612005e-01
1.19522226e+00 -5.49298942e-01 -1.82926834e-01 -1.17625260e+00
-6.58984303e-01 -2.76729316e-01 3.66037071e-01 1.28024817e-01
1.08547688e+00 -1.22295368e+00 -4.21698511e-01 -7.61063173e-02
1.77856281e-01 -1.07696211e+00 6.79775715e-01 1.42031288e+00
-3.74819368e-01 1.42297059e-01 -1.31167144e-01 -3.60988110e-01
-1.77298582e+00 5.93152866e-02 1.42383486e-01 5.04666090e-01
-8.40789378e-01 6.43928289e-01 -3.82387161e-01 5.11936694e-02
3.40284526e-01 3.84904265e-01 -4.90613312e-01 6.27973378e-01
9.08779263e-01 1.09347367e+00 -3.15566719e-01 -1.19538212e+00
-8.18950951e-01 6.88746035e-01 1.42248183e-01 -1.77990884e-01
1.00812757e+00 -3.90188754e-01 -6.48860708e-02 6.69023395e-01
1.23148739e+00 4.24449831e-01 -3.13487142e-01 8.33508909e-01
-1.53301820e-01 -2.32924312e-01 -1.23572767e-01 -6.64506555e-02
-7.39640832e-01 9.30550635e-01 5.41101813e-01 6.59034073e-01
7.47180939e-01 1.65476575e-01 8.65033567e-01 4.06677604e-01
3.93791765e-01 -9.62301970e-01 3.66004735e-01 4.33144510e-01
1.06660783e+00 -8.79489005e-01 -1.65744439e-01 -2.23589063e-01
-8.82446766e-01 1.36426735e+00 2.51873255e-01 -1.03967212e-01
6.77118242e-01 6.16093218e-01 -4.33776230e-02 -1.51127443e-01
-2.70845205e-01 2.56630570e-01 -1.20746665e-01 8.15809071e-01
5.05961776e-01 -4.62881811e-02 -3.56608808e-01 9.32545304e-01
-2.08812580e-01 -1.21475026e-01 8.71527195e-01 9.19823050e-01
-3.69534492e-01 -7.07913756e-01 -6.28657401e-01 5.08514166e-01
-9.59432364e-01 1.03640944e-01 -1.14204776e+00 4.90761906e-01
-1.87156543e-01 1.35789680e+00 3.85914259e-02 -7.14942873e-01
1.17792435e-01 2.98819959e-01 1.14323229e-01 -4.93563622e-01
-9.62275624e-01 -1.96682468e-01 5.69101214e-01 -6.62526309e-01
-5.43129861e-01 -1.27327037e+00 -1.25895488e+00 -3.22851717e-01
3.16870064e-01 -2.30150715e-01 6.01786852e-01 6.89579427e-01
1.85057864e-01 -9.86949652e-02 1.74599767e-01 -1.13788879e+00
4.36156169e-02 -1.07846630e+00 -9.12704170e-01 4.08456743e-01
3.51179570e-01 -5.27766049e-01 -7.57058322e-01 -2.73121208e-01] | [13.379636764526367, 2.2185497283935547] |
0138c26c-601f-4f10-a69b-eaa680a7100a | sos-stereo-matching-in-o-1-with-slanted | null | null | https://ieeexplore.ieee.org/document/8593800 | https://ieeexplore.ieee.org/document/8593800 | SOS: Stereo Matching in O(1) with Slanted Support Windows | Depth cameras have accelerated research in many areas of computer vision. Most triangulation-based depth cameras, whether structured light systems like the Kinect or active (assisted) stereo systems, are based on the principle of stereo matching. Depth from stereo is an active research topic dating back 30 years. Despite recent advances, algorithms usually trade-off accuracy for speed. In particular, efficient methods rely on fronto-parallel assumptions to reduce the search space and keep computation low. We present SOS (Slanted O(1) Stereo), the first algorithm capable of leveraging slanted support windows without sacrificing speed or accuracy. We use an active stereo configuration, where an illuminator textures the scene. Under this setting, local methods - such as PatchMatch Stereo - obtain state of the art results by jointly estimating disparities and slant, but at a large computational cost. We observe that these methods typically exploit local smoothness to simplify their initialization strategies. Our key insight is that local smoothness can in fact be used to amortize the computation not only within initialization, but across the entire stereo pipeline. Building on these insights, we propose a novel hierarchical initialization that is able to efficiently perform search over disparity and slants. We then show how this structure can be leveraged to provide high quality depth maps. Extensive quantitative evaluations demonstrate that the proposed technique yields significantly more precise results than current state of the art, but at a fraction of the computational cost. Our prototype implementation runs at 4000 fps on modern GPU architectures. | [] | 2018-01-01 | null | null | null | null | ['stereo-depth-estimation'] | ['computer-vision'] | [ 4.22970712e-01 -2.44000331e-01 -7.28633776e-02 -2.91360080e-01
-8.28212023e-01 -7.04833746e-01 5.03286660e-01 1.52737692e-01
-7.14723706e-01 3.79659474e-01 -7.35927448e-02 -1.96884483e-01
2.74921477e-01 -8.40181470e-01 -6.59260869e-01 -6.93270624e-01
2.17144877e-01 4.68323976e-01 6.23569906e-01 -3.01787946e-02
6.49506509e-01 5.91584384e-01 -1.91399074e+00 -1.85708348e-02
8.48628283e-01 9.79670882e-01 2.44588464e-01 7.19835818e-01
-8.13714340e-02 3.25203240e-01 -9.25210789e-02 -3.18340153e-01
6.54695213e-01 2.15904452e-02 -5.39619565e-01 2.00609863e-01
9.66321468e-01 -7.67572761e-01 -9.22865644e-02 9.40888762e-01
5.55654347e-01 -1.46747634e-01 4.93682995e-02 -9.81107891e-01
5.23253083e-01 -2.54999906e-01 -6.90742791e-01 -2.50331741e-02
7.72273421e-01 1.49035349e-01 8.69725466e-01 -1.01402605e+00
7.09348917e-01 9.40016210e-01 8.62409711e-01 3.19262803e-01
-1.35309589e+00 -3.69106144e-01 -1.56155322e-02 9.78012457e-02
-1.36788809e+00 -6.48636818e-01 6.73298776e-01 -3.47669542e-01
1.01314473e+00 2.88621277e-01 9.38537419e-01 7.92640030e-01
4.45208959e-02 5.87851167e-01 1.18083620e+00 -4.52908874e-01
4.13385361e-01 -1.59654140e-01 4.36233431e-02 6.91340864e-01
3.93283695e-01 2.23863661e-01 -1.00985491e+00 -1.70108408e-01
1.03724670e+00 1.02425210e-01 -4.38450515e-01 -8.77761662e-01
-1.21496201e+00 7.94307470e-01 2.52457470e-01 -2.44003445e-01
-1.48109332e-01 1.79343268e-01 2.46287629e-01 2.44571373e-01
5.14377892e-01 4.57820296e-02 -2.34679371e-01 -4.53065783e-01
-9.18275535e-01 2.37661511e-01 7.52426863e-01 8.91094744e-01
1.33472347e+00 -5.41674316e-01 5.97939670e-01 5.64081192e-01
2.49976680e-01 7.19092548e-01 6.58084154e-02 -1.29929399e+00
5.93702972e-01 5.60165882e-01 1.13057062e-01 -9.09419596e-01
-3.66574913e-01 9.69624445e-02 -5.51019251e-01 6.27424181e-01
3.84001672e-01 1.64585724e-01 -7.69174278e-01 1.10734534e+00
7.82261729e-01 1.62955403e-01 -2.35440493e-01 8.11662316e-01
4.09453064e-01 1.74266830e-01 -6.87206149e-01 -1.14283070e-01
1.03318143e+00 -7.45364547e-01 -1.60816327e-01 -3.64882529e-01
5.72737753e-01 -7.86593854e-01 8.55712891e-01 6.16052687e-01
-1.08323836e+00 -1.75079510e-01 -1.09558487e+00 -4.36067641e-01
1.65378004e-01 -4.87367839e-01 8.78273129e-01 8.39629829e-01
-1.36281550e+00 7.15251088e-01 -1.26429832e+00 -5.21471500e-01
1.82902679e-01 5.40951490e-01 -3.93472970e-01 -4.07827437e-01
-2.84735560e-01 5.64687073e-01 -3.66086140e-03 -1.67323828e-01
-5.09868085e-01 -7.69899607e-01 -9.26903248e-01 -2.77626783e-01
3.94291550e-01 -1.03955340e+00 1.23239028e+00 -6.82064772e-01
-1.67710829e+00 1.24290943e+00 -6.35483623e-01 -4.32129651e-01
7.80510604e-01 -3.65073740e-01 4.70340043e-01 4.81850296e-01
9.11206007e-02 6.95310473e-01 5.40061057e-01 -1.05724013e+00
-8.26073587e-01 -7.53315210e-01 2.76765615e-01 4.65612471e-01
-1.30152315e-01 -2.00765997e-01 -8.69107366e-01 -4.49115187e-02
7.67135859e-01 -1.07371473e+00 -3.50166410e-01 5.65948009e-01
-1.39816031e-01 1.88533470e-01 3.90383840e-01 -5.07497322e-03
5.14965117e-01 -1.95224631e+00 3.61994021e-02 2.14068115e-01
3.52071971e-01 -9.22672302e-02 3.74821723e-01 4.01488245e-01
3.77052933e-01 -4.21257049e-01 -2.29830295e-01 -8.81056905e-01
-2.83358574e-01 3.99268299e-01 -3.05716485e-01 9.17550981e-01
-4.01828617e-01 4.89436775e-01 -8.55653882e-01 -4.14225072e-01
7.59118795e-01 5.41702509e-01 -8.72082174e-01 1.17804296e-01
1.07916765e-01 5.30423522e-01 -1.17456175e-01 6.39581740e-01
9.82645869e-01 -2.11145565e-01 1.00325070e-01 -9.92155299e-02
-6.35640919e-01 5.76946259e-01 -1.37906623e+00 2.22354126e+00
-5.10149539e-01 7.04645157e-01 3.88499349e-01 -6.33756816e-01
5.75868785e-01 -1.34270638e-02 5.57542682e-01 -7.00280726e-01
2.92306151e-02 5.61490476e-01 -5.11766553e-01 -1.52132466e-01
5.49388647e-01 -2.25802921e-02 4.96763527e-01 2.69069493e-01
-4.81083483e-01 -3.51866901e-01 -5.84393740e-02 -1.96004380e-02
1.24529862e+00 3.73865962e-01 4.28226471e-01 -3.22607160e-01
3.18638653e-01 2.73613453e-01 5.02304852e-01 6.48305953e-01
4.97864448e-02 8.97623301e-01 4.05203626e-02 -4.22754169e-01
-1.09502184e+00 -1.07134461e+00 -3.25910628e-01 6.73986077e-01
6.56827927e-01 -4.79498088e-01 -4.70358074e-01 -7.49658123e-02
9.63252485e-02 1.11489685e-03 -3.24838936e-01 4.93196338e-01
-6.11700594e-01 -3.29658836e-01 1.74310520e-01 5.30449808e-01
5.24455667e-01 -3.78115624e-01 -1.37125552e+00 3.61290351e-02
1.36015460e-01 -1.16511869e+00 -4.98073138e-02 1.38383061e-01
-1.41737163e+00 -1.09487605e+00 -7.78962493e-01 -3.74990821e-01
6.66425526e-01 9.62667406e-01 1.07212615e+00 1.28688544e-01
-2.41383523e-01 7.03816354e-01 -2.12057337e-01 -8.38012472e-02
2.14972347e-01 -1.00317441e-01 2.57520769e-02 -1.95414200e-01
2.76467443e-01 -1.02599478e+00 -9.26273942e-01 3.59562695e-01
-7.98461676e-01 4.11584198e-01 1.53511375e-01 6.87653422e-01
7.90105462e-01 -5.28213382e-01 -5.78779757e-01 -7.70676732e-01
-1.77582964e-01 -2.03275364e-02 -1.18471992e+00 -2.87208825e-01
-7.31910646e-01 1.03186555e-01 3.28068584e-01 3.92543972e-02
-7.87269592e-01 3.47676933e-01 -1.54536247e-01 -5.66078961e-01
-9.06174406e-02 1.79950029e-01 -5.96312573e-03 -6.52218461e-01
6.51841402e-01 2.79666424e-01 3.39002497e-02 -4.67171550e-01
1.83632612e-01 4.65384960e-01 7.34658897e-01 -6.77006900e-01
7.54364491e-01 1.50599241e+00 3.70041847e-01 -8.77162933e-01
-5.30535400e-01 -1.03182626e+00 -7.67890513e-01 -1.29293710e-01
4.97279555e-01 -1.13332534e+00 -1.00329602e+00 6.23276770e-01
-1.25349748e+00 -2.89593458e-01 -5.24102859e-02 4.75312084e-01
-6.71962440e-01 6.39491737e-01 -3.89389277e-01 -8.75719905e-01
-1.25378445e-01 -1.23730588e+00 1.59108758e+00 4.78119627e-02
5.53377047e-02 -9.55191433e-01 3.35279107e-01 4.03977692e-01
2.25696608e-01 3.33380759e-01 1.29053146e-01 1.43625274e-01
-1.14053273e+00 -2.31084563e-02 -2.81535387e-01 -1.26180008e-01
2.71285709e-04 -1.25034824e-01 -1.36584055e+00 -3.98498386e-01
2.25627840e-01 -1.54008627e-01 8.12537491e-01 2.75996596e-01
7.59643435e-01 1.68902010e-01 -3.53819847e-01 1.24425399e+00
1.82893884e+00 -1.54684037e-01 7.26437390e-01 7.30189085e-01
9.02555883e-01 6.37897789e-01 6.20007396e-01 6.93522394e-01
6.23946965e-01 1.01255608e+00 6.63032234e-01 -5.71021587e-02
7.36586424e-03 4.11357619e-02 2.23689824e-01 6.87984228e-01
-3.28948975e-01 2.41765738e-01 -1.12600780e+00 5.30806005e-01
-1.74009776e+00 -7.79936254e-01 -3.91003639e-01 2.80098510e+00
6.84220195e-01 1.24404356e-01 4.88778688e-02 4.15790170e-01
3.27805966e-01 1.71742573e-01 -5.82434356e-01 -1.87652543e-01
-1.10470079e-01 3.00038218e-01 9.82053876e-01 9.28840876e-01
-8.83836627e-01 7.54353821e-01 5.75192213e+00 3.43906343e-01
-1.26713276e+00 -1.47265950e-02 1.10953152e-01 -3.66075665e-01
-4.12903398e-01 3.13082755e-01 -8.67543638e-01 2.47047722e-01
5.39588511e-01 -4.71180640e-02 4.43695575e-01 1.04835832e+00
1.64946362e-01 -7.00418890e-01 -1.39653134e+00 1.63110733e+00
1.36873230e-01 -1.37610972e+00 -3.79890054e-01 4.47995275e-01
7.01176405e-01 5.19796669e-01 -2.24963620e-01 -4.11680907e-01
5.19926064e-02 -6.58454895e-01 6.50919616e-01 3.75901796e-02
8.06637883e-01 -5.83115935e-01 4.63464975e-01 4.09829110e-01
-1.25220656e+00 3.56803060e-01 -4.55478579e-01 -4.61542100e-01
3.06654364e-01 8.99746835e-01 -7.02617884e-01 6.00952864e-01
7.53402233e-01 8.28197956e-01 -2.88624734e-01 1.17399681e+00
-1.03076145e-01 9.42676887e-02 -8.83881927e-01 3.59828830e-01
7.01249838e-02 -3.45475644e-01 6.00066423e-01 9.11981463e-01
3.79588038e-01 6.38404861e-02 1.42894179e-01 5.04667103e-01
2.54967004e-01 -1.38051167e-01 -8.36901248e-01 6.71538830e-01
4.45087940e-01 9.52948153e-01 -8.18528950e-01 -2.42035031e-01
-6.62284017e-01 1.33430958e+00 2.89024025e-01 1.49390027e-02
-5.51716506e-01 -9.56594348e-02 1.01840258e+00 3.38157475e-01
2.69155174e-01 -6.94515526e-01 -7.26331770e-01 -1.49369586e+00
4.09740448e-01 -6.48091435e-01 2.32795030e-02 -5.41244030e-01
-8.38463128e-01 2.49852329e-01 -1.34900585e-01 -1.43036795e+00
-3.33140701e-01 -6.65543139e-01 -1.52946115e-01 7.28880644e-01
-1.93192005e+00 -7.22844601e-01 -8.61599863e-01 8.17851484e-01
5.66712558e-01 5.73435009e-01 6.73766971e-01 3.74095917e-01
-2.09535033e-01 2.62833774e-01 1.75293803e-01 -2.75330275e-01
6.15149498e-01 -1.15046883e+00 5.44478118e-01 9.65241313e-01
1.83187976e-01 7.20275462e-01 6.45003438e-01 -4.64510024e-01
-1.94262409e+00 -3.99937481e-01 6.89358711e-01 -4.94952857e-01
4.16602939e-01 -4.20098066e-01 -5.52240729e-01 5.54989755e-01
-1.36697367e-01 2.95124918e-01 3.14143538e-01 1.31697297e-01
-4.58574802e-01 -2.18322247e-01 -1.12552178e+00 6.03633046e-01
1.37408245e+00 -7.71264911e-01 -3.14032376e-01 2.54176766e-01
5.11106968e-01 -1.01907575e+00 -4.71022397e-01 3.55504692e-01
7.92594612e-01 -1.99536955e+00 1.11165893e+00 2.76496738e-01
3.00654441e-01 -3.00573260e-01 -4.31173384e-01 -6.88889205e-01
1.96132347e-01 -8.29666078e-01 -1.41283059e-02 7.68773735e-01
-2.31840059e-01 -9.07339573e-01 1.21622586e+00 8.02524269e-01
-2.36366794e-01 -6.01690412e-01 -1.29009366e+00 -8.28093350e-01
-4.61384594e-01 -7.93929696e-01 2.07360163e-01 6.94924235e-01
-8.28650594e-02 7.62862638e-02 -2.12270208e-02 3.74797463e-01
1.15736783e+00 3.16113174e-01 1.16874039e+00 -1.38170838e+00
-3.78287494e-01 -2.45340347e-01 -8.48300159e-01 -1.71389961e+00
-2.25280806e-01 -4.43695158e-01 -3.07299988e-03 -1.35184789e+00
7.99576268e-02 -5.95693886e-01 3.47875923e-01 2.92837501e-01
-1.44454781e-02 4.89913642e-01 4.73299436e-02 4.71095711e-01
-3.85561705e-01 2.36762419e-01 7.78504729e-01 3.01613480e-01
-1.84960127e-01 -1.52827606e-01 -1.37830466e-01 1.05317688e+00
3.36516023e-01 -3.37023616e-01 -3.31656516e-01 -9.13156748e-01
5.33846259e-01 6.64439201e-02 5.63035727e-01 -1.28035080e+00
5.24443090e-01 -1.81583822e-01 -8.42435509e-02 -6.90214336e-01
8.37979853e-01 -9.30844069e-01 1.78406715e-01 4.89165395e-01
2.58950919e-01 5.65851182e-02 7.78429285e-02 5.21842241e-01
-1.68838337e-01 -4.50758152e-02 8.46518338e-01 -1.03936031e-01
-7.56016970e-01 3.50513697e-01 8.83263722e-02 -1.95410490e-01
8.26750576e-01 -8.12208652e-01 -1.82727218e-01 -2.67107189e-01
-1.86756119e-01 -9.20703933e-02 1.31567335e+00 -1.43976092e-01
6.50277019e-01 -1.05155897e+00 -3.67326945e-01 3.34875733e-01
1.85304046e-01 5.27525246e-01 1.00071251e-01 1.14513946e+00
-1.17028880e+00 4.44338113e-01 1.22201294e-02 -1.37004805e+00
-1.42375481e+00 1.18978798e-01 1.61219537e-01 1.51017383e-02
-9.43695545e-01 9.71007705e-01 4.04484868e-01 -6.01996370e-02
2.32710779e-01 -3.96649778e-01 4.74845916e-01 -4.85421158e-02
5.81249475e-01 5.96830845e-01 3.83417934e-01 -2.97418058e-01
-5.81347704e-01 1.22716331e+00 1.49327293e-01 -4.25293684e-01
1.28547168e+00 -4.96356308e-01 -1.95129529e-01 4.03923482e-01
1.15959907e+00 4.57624823e-01 -1.50431240e+00 -1.63194418e-01
-1.41314998e-01 -1.08145297e+00 2.18509078e-01 -4.74870279e-02
-8.88335764e-01 1.05131769e+00 5.75694978e-01 -8.58381484e-03
1.13886964e+00 -2.78396517e-01 8.56815040e-01 3.59043449e-01
1.20396614e+00 -7.03529060e-01 -2.87036866e-01 4.45962369e-01
3.19920689e-01 -1.47959125e+00 3.74686271e-01 -8.05760860e-01
-1.23372972e-01 1.23262143e+00 2.36070588e-01 -3.01539302e-01
2.33179048e-01 5.51217020e-01 9.56026465e-02 -2.01016873e-01
-3.86192143e-01 -2.64049232e-01 -1.16280958e-01 5.75209498e-01
2.44369879e-01 -3.28710943e-01 -9.31361765e-02 -3.79378915e-01
-1.31864503e-01 -1.96182393e-02 4.93667543e-01 1.12783277e+00
-5.77278256e-01 -1.35159171e+00 -5.00825465e-01 -4.38073389e-02
-5.55001684e-02 -2.51966864e-01 -1.18260503e-01 5.66667140e-01
-1.44821495e-01 9.44742858e-01 1.76809683e-01 -2.83706099e-01
1.91724524e-01 -3.57613027e-01 8.33307624e-01 -6.94428146e-01
-3.86325091e-01 1.27281323e-01 9.87451430e-03 -1.37195528e+00
-8.09985876e-01 -7.63425410e-01 -1.12660301e+00 -6.07461870e-01
-1.58343047e-01 -2.77935356e-01 9.23320532e-01 7.98251212e-01
5.13965368e-01 -3.13721150e-01 6.97236896e-01 -1.41338849e+00
-2.45206326e-01 -3.68450493e-01 -4.02993888e-01 8.63798261e-02
5.26049435e-01 -6.16760969e-01 -5.82989812e-01 -7.96458796e-02] | [8.879694938659668, -2.566859245300293] |
52574bbb-0abd-45df-8d04-00e5ed5c6c3a | precision-aware-latency-and-energy-balancing | 2306.05060 | null | https://arxiv.org/abs/2306.05060v1 | https://arxiv.org/pdf/2306.05060v1.pdf | Precision-aware Latency and Energy Balancing on Multi-Accelerator Platforms for DNN Inference | The need to execute Deep Neural Networks (DNNs) at low latency and low power at the edge has spurred the development of new heterogeneous Systems-on-Chips (SoCs) encapsulating a diverse set of hardware accelerators. How to optimally map a DNN onto such multi-accelerator systems is an open problem. We propose ODiMO, a hardware-aware tool that performs a fine-grain mapping across different accelerators on-chip, splitting individual layers and executing them in parallel, to reduce inference energy consumption or latency, while taking into account each accelerator's quantization precision to maintain accuracy. Pareto-optimal networks in the accuracy vs. energy or latency space are pursued for three popular dataset/DNN pairs, and deployed on the DIANA heterogeneous ultra-low power edge AI SoC. We show that ODiMO reduces energy/latency by up to 33%/31% with limited accuracy drop (-0.53%/-0.32%) compared to manual heuristic mappings. | ['Daniele Jahier Pagliari', 'Marian Verhelst', 'Massimo Poncino', 'Enrico Macii', 'Luca Benini', 'Giuseppe Maria Sarda', 'Alessio Burrello', 'Matteo Risso'] | 2023-06-08 | null | null | null | null | ['quantization'] | ['methodology'] | [-1.80914775e-01 -1.05950803e-01 -2.02380717e-01 -5.23282409e-01
-4.06930834e-01 -5.67148566e-01 3.26396644e-01 2.12169439e-01
-6.83581650e-01 5.53775847e-01 -3.95545773e-02 -4.12539870e-01
-1.91938534e-01 -9.01826918e-01 -8.36442471e-01 -5.69104552e-01
1.10873789e-01 5.07863939e-01 3.33435148e-01 -2.20299736e-02
-1.51721969e-01 7.60664225e-01 -1.87037587e+00 1.27816036e-01
6.78635538e-01 1.45461071e+00 2.41788626e-01 8.47018182e-01
2.41477251e-01 8.05920541e-01 -6.66773856e-01 -4.23352808e-01
5.90353429e-01 6.63510412e-02 -4.18352187e-02 -9.41209078e-01
9.36361611e-01 -4.90415514e-01 -2.05970123e-01 1.46734381e+00
6.12436295e-01 -2.50878572e-01 3.63984734e-01 -1.64355075e+00
-2.04688609e-02 8.21769774e-01 -3.68619174e-01 5.42532504e-01
-5.89396954e-01 1.47088185e-01 7.66179323e-01 -3.06406885e-01
4.66876954e-01 9.73947525e-01 7.26779580e-01 5.71967304e-01
-1.32310140e+00 -1.09785581e+00 -1.55346960e-01 4.85139102e-01
-1.57984006e+00 -7.74765253e-01 3.91096741e-01 -2.62978256e-01
1.69799864e+00 1.85102344e-01 8.45738351e-01 1.04748368e+00
8.65665257e-01 2.58765638e-01 6.11126065e-01 -1.27443507e-01
1.06069171e+00 -1.04127079e-02 3.83254558e-01 6.34277821e-01
8.40177000e-01 3.57562751e-02 -7.99641848e-01 6.20905235e-02
2.54008770e-01 -4.10283148e-01 4.21079308e-01 1.14855461e-01
-6.77807927e-01 6.71216905e-01 6.76354408e-01 1.68380797e-01
-4.43069220e-01 7.03827679e-01 1.01288879e+00 -1.44236237e-01
1.93090886e-01 4.71119791e-01 -7.45006621e-01 -4.14817005e-01
-1.24735951e+00 4.56092507e-01 7.09590018e-01 1.15600204e+00
4.96233493e-01 4.19624329e-01 -6.99010119e-02 4.28311557e-01
2.64710933e-01 4.77248400e-01 4.80910838e-01 -1.02780902e+00
3.76455396e-01 6.15143418e-01 -4.27644342e-01 -5.26001871e-01
-8.63981366e-01 -4.87418473e-01 -1.19846642e+00 5.03770530e-01
1.39137596e-01 -3.22570503e-01 -9.77032959e-01 1.83803260e+00
1.90535635e-01 7.78440535e-02 -1.49960611e-02 9.20149028e-01
5.54898441e-01 6.37282729e-01 3.76762837e-01 1.62068874e-01
1.69999135e+00 -1.03476644e+00 -6.26546860e-01 -5.20539939e-01
6.22575879e-01 -3.13233912e-01 8.87814879e-01 4.65872288e-01
-1.11520326e+00 -6.59804940e-01 -1.83584750e+00 -5.94773531e-01
-5.73400974e-01 3.85519564e-01 5.90175688e-01 8.16900074e-01
-1.41733634e+00 7.84870625e-01 -1.46426749e+00 4.72726440e-03
6.27089500e-01 8.80197644e-01 2.45006442e-01 2.69140989e-01
-9.69019651e-01 8.47391129e-01 6.50460958e-01 -5.58579108e-03
-8.26970458e-01 -1.31220376e+00 -3.08512777e-01 4.10100341e-01
-4.66632470e-02 -1.04256856e+00 1.21529531e+00 -8.72671068e-01
-1.66382349e+00 2.89436430e-01 2.62205929e-01 -1.00277293e+00
1.66126460e-01 -1.13850757e-01 -7.07501709e-01 -3.72511387e-01
-2.89237127e-02 1.17640173e+00 5.44530332e-01 -2.73942977e-01
-9.65353966e-01 -6.96735501e-01 -9.63493586e-02 -1.42598584e-01
-6.93316936e-01 -2.00001702e-01 -1.81025550e-01 -2.58575886e-01
-2.68703014e-01 -1.23434019e+00 -1.92032769e-01 1.37349635e-01
-3.20535868e-01 -1.77419722e-01 8.74319613e-01 -3.58919442e-01
1.00039601e+00 -1.87695158e+00 1.07014306e-01 3.90445031e-02
3.01614225e-01 1.80063725e-01 2.09846973e-01 -5.32529294e-01
3.10036361e-01 -1.98894575e-01 1.77307829e-01 -3.25814337e-01
2.37038225e-01 3.32948655e-01 -3.84390652e-02 3.05732936e-01
-4.26261276e-02 6.43498778e-01 -6.23041034e-01 -1.85563147e-01
2.42284730e-01 8.25683117e-01 -8.01729858e-01 -1.26822770e-01
-2.82308459e-01 -3.52258354e-01 -9.26003680e-02 7.13228643e-01
7.63744712e-01 -1.06204346e-01 3.15374404e-01 -8.97984564e-01
-3.54076207e-01 3.28667134e-01 -9.45815444e-01 1.92124426e+00
-6.11562014e-01 1.17142045e+00 3.01309705e-01 -4.34025109e-01
6.96798742e-01 5.36426939e-02 1.11628570e-01 -1.18621981e+00
6.20938301e-01 5.27321398e-01 3.89471114e-01 1.68164060e-01
7.53161788e-01 2.90693581e-01 -1.50540993e-01 -4.09419008e-05
2.87046164e-01 3.83798212e-01 3.99195820e-01 -1.84223816e-01
1.36851788e+00 -1.47681966e-01 6.59283996e-02 -1.09569252e+00
9.21297669e-02 1.83923230e-01 6.57669365e-01 4.18967903e-01
-3.76419544e-01 -1.58385396e-01 4.38983113e-01 -6.22570097e-01
-1.40439570e+00 -1.05161512e+00 -4.93105888e-01 1.24696624e+00
-1.07068986e-01 -5.49831450e-01 -1.22409630e+00 -1.08985111e-01
-7.08839223e-02 9.27254677e-01 -4.92936552e-01 -3.76015931e-01
-6.08314991e-01 -9.44402874e-01 8.32391202e-01 8.49962890e-01
7.04176128e-01 -5.53638399e-01 -1.49238145e+00 3.49374145e-01
8.18196058e-01 -1.35368013e+00 -2.76626926e-02 9.64871109e-01
-9.69737291e-01 -3.60362351e-01 6.58227205e-02 -2.36363634e-01
5.27692258e-01 -3.21418941e-01 1.42760623e+00 -4.75953192e-01
-3.69177610e-01 -6.18373990e-01 2.10840419e-01 -4.91802633e-01
-3.38665962e-01 7.81397521e-01 5.51312029e-01 -6.18522763e-01
4.26081628e-01 -8.12793136e-01 -6.43745601e-01 -1.87325522e-01
-4.38696682e-01 4.01217163e-01 6.07997239e-01 4.60432142e-01
6.77339137e-01 1.33415863e-01 2.56797671e-01 -5.64165413e-01
3.38417530e-01 -5.18489599e-01 -1.44045949e+00 -7.32883066e-02
-1.01128542e+00 5.44852138e-01 1.29439008e+00 -3.27270269e-01
-5.43153346e-01 3.80651146e-01 -3.00419033e-01 -6.15485370e-01
-3.48654203e-02 -1.74416840e-01 -4.35703814e-01 -2.12501720e-01
8.70523214e-01 -4.17583138e-01 -5.55992305e-01 -1.75798118e-01
2.97710717e-01 3.58363003e-01 9.14112329e-01 -2.29052395e-01
3.02707583e-01 2.03714445e-01 4.77020860e-01 -4.59174693e-01
-6.82376981e-01 3.38265896e-01 -4.36090678e-01 -1.59533024e-01
1.04719925e+00 -1.23342907e+00 -9.96469378e-01 4.92222086e-02
-9.47202802e-01 -4.83898044e-01 -1.16205804e-01 2.63173074e-01
-9.09048766e-02 -7.84981012e-01 -6.45849526e-01 -4.12495911e-01
-1.12469518e+00 -1.58197117e+00 9.91559625e-01 7.14622319e-01
-3.37847859e-01 -5.85844457e-01 -1.13544725e-01 -1.00685395e-01
7.37907231e-01 -3.34049128e-02 8.02765429e-01 -5.02792060e-01
-7.76242852e-01 1.74065009e-01 -5.59984565e-01 1.95574999e-01
-3.80520880e-01 1.46449491e-01 -1.24761391e+00 -3.71067315e-01
-1.56336546e-01 -7.91581068e-03 5.79154313e-01 6.63371027e-01
1.36957753e+00 -3.15451741e-01 -4.70054388e-01 1.11286819e+00
1.66201913e+00 4.28005099e-01 1.44226462e-01 3.20892245e-01
7.74117053e-01 -2.93214142e-01 -2.93850899e-03 4.71258253e-01
1.70208097e-01 9.47958767e-01 6.58843815e-01 2.64936984e-01
-5.74776381e-02 6.34562597e-02 2.48541608e-01 8.76564682e-01
4.31698948e-01 -3.08114231e-01 -8.72839749e-01 2.88555801e-01
-1.51197124e+00 -2.89249092e-01 1.16343968e-01 1.72175193e+00
5.92820168e-01 7.27340877e-01 1.40826972e-02 1.85998783e-01
2.34052867e-01 -1.18206136e-01 -1.06826842e+00 -9.15876508e-01
1.88999608e-01 3.77869695e-01 1.25809968e+00 2.83858448e-01
-8.67213786e-01 5.45253813e-01 5.37645960e+00 9.84996200e-01
-1.31032300e+00 4.34410006e-01 8.56514812e-01 -7.42168009e-01
1.31742761e-01 -3.25369895e-01 -1.46565294e+00 8.66658270e-01
2.15539217e+00 -3.43745619e-01 6.03107095e-01 1.39205861e+00
-1.08516730e-01 -9.37596783e-02 -1.40656579e+00 1.04382253e+00
-3.22483033e-01 -1.81963050e+00 -2.99783379e-01 9.48050395e-02
7.10564792e-01 5.72812259e-01 -8.53249356e-02 2.78250992e-01
4.05959219e-01 -8.40336919e-01 1.20476067e+00 1.35952652e-01
8.84124815e-01 -1.43803310e+00 7.32998848e-01 1.67080134e-01
-1.29404533e+00 -2.20481038e-01 -5.97241104e-01 -7.60997161e-02
-7.32385181e-03 8.17131400e-01 -8.23472500e-01 -3.86698022e-02
1.18692362e+00 2.08840489e-01 -5.68647563e-01 5.40761352e-01
2.11560756e-01 2.86305904e-01 -6.27000213e-01 -4.52412456e-01
1.61893323e-01 -1.41636044e-01 1.11345649e-01 1.27333188e+00
5.03021479e-01 -3.24857719e-02 -3.71621966e-01 9.82509255e-01
-4.06207055e-01 -4.19580072e-01 -1.26042321e-01 1.95692077e-01
9.41535413e-01 1.75265455e+00 -9.34071600e-01 -5.99941671e-01
-4.76410426e-02 7.96808362e-01 2.23549932e-01 -4.45048124e-01
-1.15495634e+00 -4.45048124e-01 1.24664235e+00 -2.15437002e-02
-8.02217424e-02 -1.24071896e-01 -9.07494247e-01 -6.04384840e-01
-4.69859801e-02 -4.96812284e-01 4.27593291e-02 -6.55326605e-01
-7.55249798e-01 1.12179267e+00 -2.17987746e-01 -7.89479017e-01
-3.55932325e-01 -7.41318107e-01 -1.11098386e-01 6.18368208e-01
-1.04645538e+00 -7.32011378e-01 -6.15334213e-01 -4.98265103e-02
4.37564522e-01 -4.23709452e-01 7.65980959e-01 9.75907683e-01
-1.01608837e+00 9.60053682e-01 1.56779617e-01 -2.93229818e-01
1.03465147e-01 -1.00635481e+00 9.98368919e-01 8.78577530e-01
-1.09042689e-01 1.72034949e-01 9.91582751e-01 -3.89668882e-01
-1.77601361e+00 -1.30443275e+00 4.29037273e-01 2.57946849e-02
7.46502042e-01 -9.31416214e-01 -7.50163853e-01 4.45506930e-01
3.44095975e-01 3.49972486e-01 3.19762290e-01 -1.17462993e-01
-2.44024679e-01 -7.19717503e-01 -1.33293319e+00 7.91684270e-01
1.16887569e+00 -3.74295831e-01 3.22499067e-01 3.59206766e-01
7.89605916e-01 -7.96486437e-01 -8.97365630e-01 3.41714829e-01
5.12861788e-01 -1.04049492e+00 9.70720708e-01 -7.48480409e-02
2.92784631e-01 -2.13731408e-01 -4.11954075e-01 -1.22756398e+00
-3.63144964e-01 -2.29966909e-01 -5.65058470e-01 9.45917845e-01
2.38069594e-01 -3.74087572e-01 1.15119743e+00 5.15593290e-01
-6.49140894e-01 -6.77415371e-01 -1.47057748e+00 -7.78934717e-01
-1.91387221e-01 -5.33794463e-01 7.09125698e-01 4.77676243e-01
-3.99357170e-01 4.62477595e-01 2.16807306e-01 3.01439375e-01
7.90788114e-01 -4.53010708e-01 2.01664388e-01 -1.23373115e+00
3.75429820e-03 -7.75336564e-01 -6.60657227e-01 -4.25527900e-01
2.20382765e-01 -8.80100667e-01 -5.83368763e-02 -9.93565023e-01
-1.31921247e-01 -3.94156724e-01 -2.14352220e-01 5.82899392e-01
4.95394498e-01 2.95404196e-01 7.54861459e-02 -2.46587560e-01
-5.03261387e-01 3.06988746e-01 2.71764755e-01 1.09364353e-02
5.64322434e-02 -6.64763272e-01 -4.87806857e-01 7.39029706e-01
6.94189847e-01 -6.32650495e-01 -5.43172359e-01 -7.94595122e-01
3.95048469e-01 -2.19168857e-01 2.40463108e-01 -1.97168636e+00
7.67774820e-01 2.24775478e-01 6.36525393e-01 -6.29406035e-01
4.07469988e-01 -9.31611180e-01 6.23024523e-01 8.06288600e-01
-1.18972182e-01 7.32161105e-01 9.03699219e-01 3.65264751e-02
3.07314217e-01 3.00177261e-02 1.17882311e+00 4.52870458e-01
-9.56646621e-01 1.70403287e-01 -3.25898468e-01 -1.73538506e-01
1.04282415e+00 2.38543198e-01 -6.84574246e-01 5.36328554e-01
-2.35816315e-01 -4.74025123e-02 2.53153503e-01 3.37985039e-01
-1.34591963e-02 -1.52759767e+00 -2.80585796e-01 4.81121182e-01
-2.99786925e-01 -3.22786607e-02 4.77710187e-01 3.52555692e-01
-9.73332167e-01 7.81882703e-01 -8.77270997e-01 -7.64630854e-01
-1.09369504e+00 2.43419483e-01 6.38067544e-01 -1.15209095e-01
-5.98513663e-01 1.05262637e+00 -4.48490471e-01 2.54996419e-02
2.04949901e-01 -8.21877718e-01 4.43691075e-01 8.55760202e-02
5.72300732e-01 7.30893552e-01 8.86971056e-01 -1.05591252e-01
-8.10252547e-01 1.87071174e-01 -6.76942468e-02 -4.31079678e-02
1.25141823e+00 4.05538112e-01 -1.53198028e-02 4.43960242e-02
1.31718493e+00 -6.48237765e-01 -1.32861292e+00 2.22297892e-01
1.23122811e-01 6.26096278e-02 1.00656819e+00 -8.13899279e-01
-1.53654361e+00 5.25064826e-01 1.21919322e+00 2.16878951e-02
1.48458922e+00 -3.11797976e-01 8.77719522e-01 4.01398599e-01
3.21336687e-01 -1.45272791e+00 -5.53297460e-01 4.65289265e-01
3.52992922e-01 -7.55513132e-01 2.12933302e-01 1.32081822e-01
5.19505404e-02 1.10865033e+00 1.09881580e+00 -2.42544681e-01
6.37511551e-01 1.50364399e+00 -3.63648355e-01 -4.24383804e-02
-1.32443929e+00 4.81229663e-01 1.17348112e-01 1.70575693e-01
1.40062556e-01 3.64055604e-01 1.47493988e-01 6.62817597e-01
-7.57937014e-01 2.11201489e-01 2.97111064e-01 7.92410374e-01
-2.13338614e-01 -6.52627528e-01 -7.30747730e-02 8.36362004e-01
-1.56599879e-01 -4.15904373e-02 2.49372199e-01 5.99073112e-01
5.31059027e-01 3.94624174e-01 1.00781715e+00 -9.09348011e-01
3.49185705e-01 -2.04205498e-01 2.91224331e-01 -9.83330980e-02
-8.91547620e-01 -1.87519833e-01 5.52206859e-02 -8.08912218e-01
4.02332604e-01 -4.78621066e-01 -1.41385734e+00 -8.80415142e-01
1.84679434e-01 -3.65702838e-01 1.57253063e+00 6.43420935e-01
8.53955269e-01 1.41440213e+00 6.03838265e-02 -1.14525962e+00
-4.08362508e-01 -6.00590527e-01 -3.88728231e-01 -6.20254993e-01
2.05617905e-01 -5.57209790e-01 -3.09091628e-01 -4.33335900e-01] | [8.386846542358398, 2.876842975616455] |
8cc33119-3df2-4e5a-9719-d2f4c12d51f2 | efficient-personalized-federated-learning-via | 2305.02776 | null | https://arxiv.org/abs/2305.02776v2 | https://arxiv.org/pdf/2305.02776v2.pdf | Efficient Personalized Federated Learning via Sparse Model-Adaptation | Federated Learning (FL) aims to train machine learning models for multiple clients without sharing their own private data. Due to the heterogeneity of clients' local data distribution, recent studies explore the personalized FL that learns and deploys distinct local models with the help of auxiliary global models. However, the clients can be heterogeneous in terms of not only local data distribution, but also their computation and communication resources. The capacity and efficiency of personalized models are restricted by the lowest-resource clients, leading to sub-optimal performance and limited practicality of personalized FL. To overcome these challenges, we propose a novel approach named pFedGate for efficient personalized FL by adaptively and efficiently learning sparse local models. With a lightweight trainable gating layer, pFedGate enables clients to reach their full potential in model capacity by generating different sparse models accounting for both the heterogeneous data distributions and resource constraints. Meanwhile, the computation and communication efficiency are both improved thanks to the adaptability between the model sparsity and clients' resources. Further, we theoretically show that the proposed pFedGate has superior complexity with guaranteed convergence and generalization error. Extensive experiments show that pFedGate achieves superior global accuracy, individual accuracy and efficiency simultaneously over state-of-the-art methods. We also demonstrate that pFedGate performs better than competitors in the novel clients participation and partial clients participation scenarios, and can learn meaningful sparse local models adapted to different data distributions. | ['Yaliang Li', 'Bolin Ding', 'Dawei Gao', 'Liuyi Yao', 'Daoyuan Chen'] | 2023-05-04 | null | null | null | null | ['personalized-federated-learning'] | ['methodology'] | [-2.51218468e-01 -1.74349263e-01 -7.68012702e-01 -5.60519040e-01
-1.04239511e+00 -2.27600813e-01 1.93475962e-01 -4.49591994e-01
3.28613780e-02 6.44292355e-01 1.91350892e-01 2.11067662e-01
-3.42774570e-01 -9.18668389e-01 -5.78027546e-01 -9.49319184e-01
9.78382900e-02 9.68751729e-01 5.62946610e-02 2.24989235e-01
-1.69526860e-01 5.39565742e-01 -1.44485748e+00 6.36306405e-01
9.31344271e-01 1.26976645e+00 3.65829498e-01 1.53854266e-01
-3.21035236e-01 8.99517715e-01 -3.35700840e-01 -4.03566480e-01
2.79218316e-01 -2.51749545e-01 -5.72362363e-01 1.11020736e-01
2.84908563e-01 -6.12914979e-01 -6.19202197e-01 6.52510703e-01
6.39645815e-01 6.00271448e-02 1.43117443e-01 -1.43353951e+00
-7.13192821e-01 7.94000506e-01 -3.63435298e-01 7.01605994e-03
-4.72077355e-02 3.24013859e-01 1.04677486e+00 -8.14484417e-01
4.23512578e-01 1.18877196e+00 6.25258625e-01 6.08158350e-01
-1.31446147e+00 -9.36081409e-01 6.87534094e-01 5.11475563e-01
-1.45187604e+00 -5.32630026e-01 5.68063438e-01 1.37002259e-01
6.87219143e-01 2.70018160e-01 1.80275992e-01 1.17339635e+00
-5.29664814e-01 1.21262670e+00 7.98929870e-01 7.07051530e-02
3.89313579e-01 3.84108126e-01 -1.52468756e-01 6.14071488e-01
1.61805507e-02 -2.77114004e-01 -8.59477103e-01 -5.62454522e-01
7.98953593e-01 5.92637181e-01 -2.87252516e-01 -5.00526190e-01
-7.82718122e-01 9.25361395e-01 4.60130781e-01 2.80579865e-01
-5.12973070e-01 2.60511246e-02 3.03136230e-01 2.38951936e-01
9.93766710e-02 -2.46798038e-01 -7.05938160e-01 -1.15407296e-01
-8.30910504e-01 8.04675445e-02 1.00772500e+00 1.25393343e+00
1.06256104e+00 2.43854076e-02 -3.78401250e-01 7.99985051e-01
1.33186221e-01 5.68271518e-01 6.91913068e-01 -9.70393836e-01
9.87587929e-01 7.61410296e-01 -7.10744038e-02 -9.00443912e-01
-4.62306812e-02 -9.12850678e-01 -1.04917216e+00 -5.62534750e-01
1.49315491e-01 -1.15193158e-01 -3.95147115e-01 1.89218438e+00
5.20174801e-01 4.80795324e-01 -4.25033756e-02 8.21754515e-01
6.98491514e-01 5.46260953e-01 3.46142873e-02 -1.86667144e-01
7.04814553e-01 -1.33127832e+00 -2.77030200e-01 -1.47716820e-01
8.46487164e-01 -3.01692516e-01 1.09086287e+00 2.14534655e-01
-1.01595688e+00 -1.69327214e-01 -2.55277544e-01 1.51120603e-01
3.23599726e-01 2.30127677e-01 9.47246730e-01 5.08012414e-01
-7.64833391e-01 3.48497808e-01 -8.46894860e-01 -1.41360894e-01
1.07116520e+00 6.78745031e-01 -2.75732756e-01 -5.57483017e-01
-7.37537622e-01 2.59224504e-01 -5.32332845e-02 -5.05937785e-02
-1.04476428e+00 -8.78394186e-01 -3.65443170e-01 6.68403625e-01
4.36447859e-01 -7.94845521e-01 1.34285927e+00 -9.78604913e-01
-1.57279027e+00 3.70660901e-01 -3.46258730e-01 -2.99560279e-01
7.18605042e-01 1.69820517e-01 -1.87548026e-01 4.29537632e-02
4.85261064e-03 2.28647552e-02 6.00816190e-01 -1.09827578e+00
-9.19375598e-01 -6.94038033e-01 -2.41806335e-03 1.53937399e-01
-8.50679040e-01 -3.94249976e-01 -8.92251134e-01 -3.27226400e-01
2.12893978e-01 -7.09171474e-01 -3.99143517e-01 -2.28626896e-02
-9.12688747e-02 -2.81113446e-01 1.07000780e+00 -2.68766195e-01
1.13491881e+00 -2.11352849e+00 -5.03795519e-02 5.38478613e-01
3.60674798e-01 2.23637432e-01 -5.37909329e-01 4.08662617e-01
7.11113632e-01 -1.83017224e-01 1.04753815e-01 -6.05760813e-01
1.20345145e-01 6.80841863e-01 -1.98060036e-01 3.66508275e-01
-3.35640103e-01 1.07174420e+00 -7.05793083e-01 -5.21054864e-01
-1.60518378e-01 6.89050674e-01 -8.74979496e-01 3.38327527e-01
-1.67812243e-01 5.99143565e-01 -1.14007139e+00 8.46893072e-01
8.99484217e-01 -8.20728779e-01 6.92924082e-01 -1.82566911e-01
5.15994370e-01 1.82811514e-01 -1.30455875e+00 1.67055631e+00
-9.99084890e-01 -1.10931411e-01 6.70165122e-01 -1.27948880e+00
7.78142154e-01 2.84805208e-01 6.02660418e-01 -1.00931549e+00
-3.15058976e-02 6.21708572e-01 -6.39006019e-01 -4.47312802e-01
-7.19814226e-02 -3.73242125e-02 1.33791104e-01 6.70904398e-01
1.73200592e-01 6.93080425e-01 -2.75205553e-01 3.05161297e-01
1.01244450e+00 -3.15508604e-01 -3.30085903e-01 4.27072234e-02
3.99143517e-01 -6.16767764e-01 9.05359983e-01 9.81465757e-01
-4.10455279e-02 4.68703568e-01 2.31616467e-01 -4.79277164e-01
-5.23247004e-01 -9.52959478e-01 1.92967340e-01 1.67209268e+00
2.69296736e-01 -1.53091758e-01 -4.02128726e-01 -1.00300527e+00
3.77671957e-01 3.35389555e-01 -3.16044986e-01 -6.71850843e-03
-7.53184736e-01 -6.59233093e-01 1.85156599e-01 6.00981236e-01
4.58550066e-01 -9.03202891e-01 -1.41085893e-01 3.08243454e-01
-3.15169007e-01 -1.19528997e+00 -7.79359818e-01 -3.93059142e-02
-1.06528282e+00 -8.69227231e-01 -6.05677903e-01 -7.62843668e-01
6.49519145e-01 7.36770689e-01 9.90384579e-01 1.35968387e-01
8.92363582e-03 4.34869051e-01 -2.52374619e-01 1.75350502e-01
6.90475255e-02 6.62412465e-01 -1.83709636e-01 5.62584281e-01
1.73077941e-01 -1.11096215e+00 -8.58727813e-01 4.54442114e-01
-9.58503485e-01 5.52865677e-03 6.96930647e-01 8.80551457e-01
7.51696825e-01 -9.96273160e-02 8.88731420e-01 -1.18838930e+00
3.06487054e-01 -1.01884127e+00 -2.52099723e-01 4.99082148e-01
-7.10065186e-01 6.11549942e-03 9.35369432e-01 -5.57522893e-01
-1.27051032e+00 -6.50735125e-02 1.86582327e-01 -5.77625096e-01
1.68704569e-01 4.18194979e-01 -6.81261778e-01 -2.17466608e-01
3.76589477e-01 4.03238773e-01 -5.84843270e-02 -9.12474811e-01
4.49256152e-01 8.00052285e-01 3.15167844e-01 -9.50277209e-01
5.38098574e-01 6.32377207e-01 -2.04093754e-01 -2.25334615e-01
-8.51432979e-01 -3.02370548e-01 -3.74187306e-02 2.66199201e-01
-9.00342837e-02 -1.10570693e+00 -1.00895083e+00 5.01692057e-01
-6.79149091e-01 -3.78655761e-01 -4.22377765e-01 3.62186044e-01
-4.22533065e-01 3.45618606e-01 -6.78433001e-01 -6.78409994e-01
-6.70186341e-01 -1.04324245e+00 8.93779457e-01 2.80331075e-01
3.78122807e-01 -8.99287343e-01 -3.36316556e-01 7.69130051e-01
9.59582448e-01 -2.04873890e-01 7.76839197e-01 -7.29091108e-01
-1.03607881e+00 -2.64417678e-01 -4.41033185e-01 2.74051577e-01
1.66358009e-01 -7.12444484e-01 -9.05472398e-01 -6.46811485e-01
-4.16806377e-02 -5.50340533e-01 6.74525559e-01 1.03537746e-01
1.61723113e+00 -1.03751290e+00 -4.71872360e-01 9.11850452e-01
1.43648100e+00 -2.36770362e-01 1.70806527e-01 3.50971371e-02
8.58072817e-01 5.36241710e-01 1.69364110e-01 1.05788720e+00
7.86703825e-01 8.08343828e-01 6.05915666e-01 1.00923486e-01
5.94288148e-02 -4.23956990e-01 2.36331463e-01 6.65481448e-01
2.07012504e-01 -2.25176007e-01 -5.64882874e-01 6.81446016e-01
-2.26112437e+00 -1.00086927e+00 2.00775295e-01 2.19765925e+00
6.93293929e-01 -5.18762171e-01 1.04515843e-01 -3.65640014e-01
5.92330098e-01 3.60785844e-03 -1.01378298e+00 2.54419670e-02
-2.46100396e-01 1.84013292e-01 5.42090476e-01 3.26182134e-02
-5.54925203e-01 7.12880313e-01 5.33725452e+00 1.34921134e+00
-1.27626812e+00 6.77300692e-01 9.00559187e-01 -7.54186749e-01
-6.78103089e-01 -1.79864854e-01 -9.79481518e-01 7.39471674e-01
8.93037021e-01 -1.17599785e-01 8.05108190e-01 1.30182338e+00
5.43887801e-02 4.20079201e-01 -1.05932570e+00 1.06803775e+00
-1.04872912e-01 -1.56986606e+00 1.83837652e-01 1.23097986e-01
1.02770472e+00 5.31235099e-01 1.84182495e-01 4.95988727e-01
2.80051112e-01 -9.72886801e-01 5.42174637e-01 4.98347759e-01
5.68841457e-01 -8.68085682e-01 6.57215655e-01 7.36556292e-01
-1.25647700e+00 -5.89442790e-01 -2.73318678e-01 2.42595553e-01
-4.50579869e-03 4.10666198e-01 -4.63285685e-01 4.96106893e-01
7.68784940e-01 3.93378764e-01 -2.21881896e-01 8.46137345e-01
2.41049439e-01 6.12028241e-01 -6.75140619e-01 2.75753856e-01
1.82951823e-01 -2.41014421e-01 -1.95899811e-02 9.72669303e-01
5.13285816e-01 1.24847218e-01 5.33606291e-01 7.09902048e-01
-4.85324949e-01 3.14816356e-01 5.66855408e-02 1.86441422e-01
9.56027091e-01 1.11195087e+00 -2.69793114e-03 -3.52911860e-01
-6.46701157e-01 7.47835040e-01 7.04972565e-01 5.69963872e-01
-6.40951157e-01 9.49955881e-02 8.17440033e-01 2.59154022e-01
6.22933447e-01 2.50445962e-01 -1.49503365e-01 -1.45586205e+00
4.28028107e-01 -9.58307147e-01 7.43743479e-01 -1.29308254e-01
-1.82460928e+00 5.67523777e-01 -4.06623423e-01 -9.36502993e-01
-3.87304910e-02 8.99424553e-02 -6.23794019e-01 8.21123898e-01
-1.63066781e+00 -1.67577946e+00 -5.30611932e-01 1.12334919e+00
2.73074418e-01 -4.15916294e-01 7.37338364e-01 6.16391718e-01
-7.36814737e-01 1.25331724e+00 6.37969851e-01 -3.46053362e-01
3.94592136e-01 -5.01394510e-01 -2.12216854e-01 2.45561078e-01
8.18172023e-02 4.50744867e-01 1.03547893e-01 -1.31266057e-01
-1.66755962e+00 -1.42309999e+00 8.44369173e-01 3.89296114e-02
3.83526862e-01 -2.37199560e-01 -1.10360253e+00 8.07391822e-01
-4.77913052e-01 5.99077463e-01 8.81837368e-01 2.54427195e-01
-5.98951101e-01 -7.05668151e-01 -1.55988955e+00 2.56655991e-01
1.21794546e+00 -6.24320030e-01 2.97619849e-01 5.80898702e-01
5.75144351e-01 -2.87229151e-01 -8.44296515e-01 2.55355060e-01
5.63406527e-01 -1.05878663e+00 7.94314563e-01 -6.62629664e-01
-2.64657885e-01 1.62116498e-01 -4.27712440e-01 -6.19018853e-01
-2.66567618e-01 -6.19185925e-01 -7.40768254e-01 1.41923618e+00
2.10567817e-01 -1.03863192e+00 1.37097692e+00 9.30831313e-01
1.03359371e-01 -1.19965529e+00 -1.15542543e+00 -7.49635637e-01
-1.61465228e-01 -4.03836727e-01 1.28237092e+00 8.10287774e-01
-3.20363820e-01 -5.96812777e-02 -6.28858924e-01 2.21492559e-01
6.11159384e-01 6.87402010e-01 9.70134497e-01 -1.05964744e+00
-8.16688061e-01 -3.16953659e-01 -1.11530833e-01 -1.46051455e+00
2.54574656e-01 -9.53914523e-01 -4.96887863e-01 -1.46956038e+00
6.89949334e-01 -1.15953827e+00 -6.17543101e-01 7.57883012e-01
4.60316427e-03 6.69615641e-02 3.85082483e-01 7.73804784e-01
-1.00183213e+00 8.90047014e-01 1.08333468e+00 4.26333435e-02
-4.50586975e-01 4.63350892e-01 -9.06467319e-01 2.36153051e-01
5.86008847e-01 -4.32758093e-01 -5.22879303e-01 -7.31793284e-01
-1.84519216e-01 1.39442667e-01 2.41207078e-01 -6.45425558e-01
2.78819114e-01 -5.12992144e-01 1.31574953e-02 -1.20310910e-01
2.13813782e-01 -9.50543404e-01 2.29205742e-01 2.23078072e-01
-1.83599010e-01 -3.83663148e-01 -2.98518747e-01 7.41808176e-01
-2.23959133e-01 1.89284116e-01 6.75624967e-01 -9.68295336e-02
-3.14911604e-01 1.09740937e+00 2.02477589e-01 -5.77974431e-02
9.35478806e-01 1.43296691e-03 -2.57437617e-01 -6.33051872e-01
-6.29124999e-01 5.48026979e-01 4.10286456e-01 2.41948083e-01
3.74977410e-01 -1.39222062e+00 -5.66971123e-01 3.20199668e-01
-9.69229192e-02 2.20520124e-01 9.48556900e-01 1.10132432e+00
-1.11270897e-01 4.62823033e-01 1.91559240e-01 -4.66288328e-01
-9.70570624e-01 5.05802155e-01 2.57155985e-01 -5.55439413e-01
-4.72907871e-01 8.88749838e-01 4.44623142e-01 -6.12271905e-01
5.42506397e-01 6.01414256e-02 3.64142358e-01 -2.06552595e-01
5.40023327e-01 6.21387005e-01 1.50709480e-01 -6.42168701e-01
-2.84545720e-01 2.16038167e-01 -1.40489295e-01 5.95321059e-01
1.61300540e+00 -2.71006674e-01 1.32593617e-01 -2.01987281e-01
1.35853243e+00 -3.83067578e-02 -1.39679384e+00 -9.38401639e-01
-4.01162088e-01 -8.77395988e-01 5.62675558e-02 -7.09051847e-01
-1.96191406e+00 6.23146355e-01 5.91815650e-01 -2.31961891e-01
1.30479169e+00 2.04383612e-01 1.25682461e+00 2.44017959e-01
9.03650701e-01 -7.85592079e-01 5.63852228e-02 1.63819268e-01
4.80745375e-01 -1.24073851e+00 -2.59751737e-01 -3.03571284e-01
-6.93929195e-01 7.07273066e-01 7.68631458e-01 8.27540606e-02
4.88008112e-01 -5.74336760e-02 -1.63097635e-01 6.29046261e-02
-9.56348658e-01 1.98341146e-01 -2.63062138e-02 5.88943899e-01
-3.28697823e-02 1.39794484e-01 2.18704324e-02 1.28619933e+00
2.51245439e-01 1.17134869e-01 -2.14071035e-01 8.11014831e-01
-1.61185697e-01 -1.33754086e+00 -1.19623326e-01 6.36317194e-01
-3.08140039e-01 1.75380200e-01 1.76725700e-01 3.73170614e-01
2.26556957e-01 8.96907032e-01 -1.75433695e-01 -1.80320174e-01
2.43184447e-01 -2.41318315e-01 2.78020918e-01 -5.61651230e-01
-7.53802836e-01 -1.75807048e-02 -1.94757059e-01 -9.70721602e-01
-1.43187523e-01 -4.71420765e-01 -1.32748115e+00 -6.61731422e-01
-3.88523221e-01 3.62777263e-01 3.60511869e-01 9.37633991e-01
1.13079941e+00 -1.49281202e-02 1.13842154e+00 -7.80949712e-01
-1.02368748e+00 -6.99159026e-01 -8.40737283e-01 3.09299409e-01
5.31396233e-02 -3.39886636e-01 -3.52789611e-01 -3.43887150e-01] | [5.820129871368408, 6.254515171051025] |
31fbee3d-1c72-48d1-bbf4-631b76bc08cf | cp-cnn-core-periphery-principle-guided | 2304.10515 | null | https://arxiv.org/abs/2304.10515v1 | https://arxiv.org/pdf/2304.10515v1.pdf | CP-CNN: Core-Periphery Principle Guided Convolutional Neural Network | The evolution of convolutional neural networks (CNNs) can be largely attributed to the design of its architecture, i.e., the network wiring pattern. Neural architecture search (NAS) advances this by automating the search for the optimal network architecture, but the resulting network instance may not generalize well in different tasks. To overcome this, exploring network design principles that are generalizable across tasks is a more practical solution. In this study, We explore a novel brain-inspired design principle based on the core-periphery property of the human brain network to guide the design of CNNs. Our work draws inspiration from recent studies suggesting that artificial and biological neural networks may have common principles in optimizing network architecture. We implement the core-periphery principle in the design of network wiring patterns and the sparsification of the convolution operation. The resulting core-periphery principle guided CNNs (CP-CNNs) are evaluated on three different datasets. The experiments demonstrate the effectiveness and superiority compared to CNNs and ViT-based methods. Overall, our work contributes to the growing field of brain-inspired AI by incorporating insights from the human brain into the design of neural networks. | ['Tianming Liu', 'Dajiang Zhu', 'Zihao Wu', 'Haixing Dai', 'Lin Zhao'] | 2023-03-27 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 2.08232149e-01 1.84257969e-01 1.17512830e-01 -3.80047202e-01
6.94131494e-01 -5.15452504e-01 4.45476413e-01 -4.27299470e-01
-2.76290119e-01 3.70190978e-01 6.55191690e-02 -3.11111987e-01
-3.67077619e-01 -6.84287608e-01 -5.91599464e-01 -6.65556908e-01
2.60858238e-01 5.36241662e-03 8.20330828e-02 -4.78399754e-01
4.53365535e-01 7.10185587e-01 -1.50583243e+00 2.72023171e-01
6.30782962e-01 1.03684306e+00 4.39313382e-01 3.98503393e-01
-2.23749042e-01 5.33223927e-01 -4.69478697e-01 -2.91924655e-01
3.69198531e-01 -6.88552022e-01 -8.67155015e-01 -2.09776357e-01
6.81854188e-02 2.71982312e-01 -3.49626929e-01 8.80080462e-01
4.59442019e-01 1.20721184e-01 5.93575537e-01 -1.24014997e+00
-8.75607729e-01 7.57603049e-01 -1.95902348e-01 3.03624213e-01
-6.09152913e-01 4.50174451e-01 1.23075449e+00 -7.22180486e-01
3.26392978e-01 1.10096443e+00 6.04196191e-01 9.90732372e-01
-1.18464315e+00 -5.22417605e-01 2.37796634e-01 3.07352871e-01
-1.41874540e+00 -4.34489101e-01 8.78953755e-01 -3.64646196e-01
9.50474322e-01 9.06392932e-02 1.03601229e+00 8.30342948e-01
3.65752637e-01 6.58921003e-01 6.31979883e-01 -5.56087613e-01
4.65359181e-01 2.23742798e-02 -2.50610650e-01 7.69471109e-01
4.54713553e-01 2.36960560e-01 -6.13170564e-01 2.56588608e-01
1.03100491e+00 -1.39058024e-01 -1.94525361e-01 -3.48066568e-01
-1.10496974e+00 7.34052956e-01 1.06105912e+00 5.76096952e-01
-4.65834320e-01 3.36690217e-01 1.43893793e-01 2.71104544e-01
-4.14598063e-02 1.36438036e+00 -6.29114389e-01 2.21371636e-01
-9.01490450e-01 4.98605892e-02 7.33884990e-01 8.41254652e-01
8.67653131e-01 3.47198397e-01 -2.38177419e-01 8.28908503e-01
3.10525835e-01 -2.16617227e-01 6.87323451e-01 -1.08228314e+00
-5.42526506e-02 9.93804395e-01 -5.51747561e-01 -9.11486149e-01
-6.71532393e-01 -8.97692680e-01 -1.07399166e+00 1.57446384e-01
2.28601828e-01 -3.51679921e-01 -8.13458741e-01 1.97265232e+00
6.49121553e-02 -1.01913333e-01 -1.00864563e-03 9.08335686e-01
7.02352405e-01 4.08193052e-01 1.59847979e-02 3.16187918e-01
1.35804892e+00 -1.16601586e+00 -3.79011750e-01 -5.32019317e-01
5.00090659e-01 -4.58894730e-01 8.87505770e-01 1.40728667e-01
-9.80051994e-01 -7.70422161e-01 -1.26651168e+00 3.33656996e-01
-3.30018163e-01 8.96109566e-02 7.63062954e-01 6.70097530e-01
-1.38002169e+00 7.55989194e-01 -5.73485136e-01 -5.63284993e-01
7.50374079e-01 5.12135923e-01 -1.37543930e-02 3.38005155e-01
-9.48212981e-01 9.33374286e-01 9.88771677e-01 3.25050324e-01
-7.92825043e-01 -8.38212907e-01 -4.20886725e-01 3.66482794e-01
1.66632757e-01 -1.17075074e+00 1.13931990e+00 -1.31857848e+00
-1.79656613e+00 5.79077542e-01 -1.13317231e-02 -6.06187522e-01
-1.22741193e-01 1.87187031e-01 -2.35065296e-02 -1.16552517e-01
-5.28824806e-01 1.14384699e+00 8.93860459e-01 -9.64043558e-01
-5.31814635e-01 -8.79694894e-02 -4.37338762e-02 -1.66824721e-02
-8.11290681e-01 -3.36308688e-01 -2.64897227e-01 -7.70775676e-01
1.37961730e-01 -9.72823501e-01 -4.37360972e-01 1.33745641e-01
-4.19612378e-01 -2.26630375e-01 4.97397572e-01 1.98763013e-02
1.32289505e+00 -2.18794632e+00 4.55813169e-01 3.52826029e-01
4.56859142e-01 3.64586979e-01 -4.91880149e-01 3.25905859e-01
-1.84301183e-01 2.77307391e-01 -7.09960908e-02 -5.37747182e-02
-1.87673628e-01 3.62994939e-01 -5.27389608e-02 7.37852752e-02
4.84031707e-01 1.35318542e+00 -5.84648550e-01 -2.37949580e-01
-2.03740209e-01 2.64409214e-01 -8.02116573e-01 2.63768882e-01
-1.73146933e-01 4.30148065e-01 -4.27522391e-01 5.56460738e-01
1.62189305e-01 -5.15867233e-01 2.33817175e-01 -3.70454460e-01
-2.32243121e-01 1.31200120e-01 -5.95616937e-01 1.63018513e+00
-3.00978720e-01 1.03020740e+00 -1.89884230e-01 -1.35469961e+00
1.30527639e+00 2.66889095e-01 1.47516429e-01 -8.20471942e-01
3.58798295e-01 2.97015131e-01 8.73108983e-01 -4.21004295e-01
6.30565360e-02 -1.16858091e-02 4.36144769e-01 5.36355972e-01
3.37920874e-01 -1.99697077e-01 1.01823978e-01 -1.20737016e-01
1.12149477e+00 -1.13947332e-01 2.35518917e-01 -8.53186607e-01
5.52746475e-01 2.98938695e-02 5.60481250e-01 7.51822352e-01
-1.81963950e-01 5.47575772e-01 4.69004631e-01 -9.42158997e-01
-1.28179181e+00 -6.78781152e-01 -6.75714910e-02 1.12582886e+00
-2.12855831e-01 -2.65738875e-01 -1.05067313e+00 -2.24232227e-01
-3.19096118e-01 4.55190837e-01 -1.00155604e+00 -5.61885178e-01
-6.87452674e-01 -6.25804782e-01 6.70972466e-01 4.78870004e-01
6.87360764e-01 -1.41083860e+00 -1.03857994e+00 2.23032147e-01
1.82205424e-01 -7.78654456e-01 -3.76610756e-01 3.75429094e-01
-1.12480879e+00 -8.62878919e-01 -6.80352092e-01 -9.21407938e-01
8.31606090e-01 1.21706598e-01 1.11066329e+00 6.07107878e-01
-4.60539341e-01 1.52593642e-01 -2.12161735e-01 -6.53692424e-01
-2.27948412e-01 6.39217138e-01 1.29123265e-03 1.48907527e-01
2.29468659e-01 -8.12104166e-01 -9.02276635e-01 4.31825668e-01
-9.00550961e-01 3.40747088e-01 9.64445293e-01 9.55314100e-01
4.21108723e-01 2.67683208e-01 5.68084002e-01 -5.69957554e-01
7.75118053e-01 -4.38492119e-01 -3.74778837e-01 3.82742524e-01
-7.74581969e-01 4.84476388e-01 7.23123968e-01 -4.63507473e-01
-7.64836729e-01 -4.15915735e-02 9.75123346e-02 -2.75415957e-01
-3.29429209e-02 4.89283949e-01 -8.83679762e-02 -2.74214178e-01
7.66873300e-01 2.80525744e-01 1.48930237e-01 -3.75259131e-01
1.73659191e-01 1.58653915e-01 3.62644404e-01 -6.74816370e-01
6.30072057e-01 2.44585812e-01 1.92913279e-01 -8.59387755e-01
-6.16304994e-01 -8.13814811e-03 -7.34677136e-01 -1.60262600e-01
7.49512792e-01 -3.84748340e-01 -6.43565476e-01 3.83272201e-01
-1.40398026e+00 -3.91715765e-01 -3.90429199e-01 1.44380331e-01
-3.95242721e-01 -2.82212049e-01 -2.44474113e-01 -5.22520721e-01
-5.27503371e-01 -1.25200081e+00 4.24759001e-01 5.29591560e-01
-5.50910890e-01 -1.08433640e+00 -1.61856040e-01 -1.90297171e-01
9.72317159e-01 8.71786177e-02 1.30467510e+00 -5.85262775e-01
-5.15424013e-01 3.61785620e-01 -2.76609689e-01 4.05705214e-01
9.41029415e-02 6.64024279e-02 -7.61741459e-01 -9.85303745e-02
7.69375544e-03 2.10279170e-02 9.56692159e-01 5.51625907e-01
1.19187117e+00 -4.54132795e-01 -1.98260084e-01 9.01453972e-01
1.37237763e+00 3.81212562e-01 8.90398562e-01 7.43852317e-01
5.82279742e-01 1.06866765e+00 -2.04843953e-01 3.01496327e-01
1.88078694e-02 4.92896885e-01 5.38923681e-01 -1.01839572e-01
-1.93504319e-01 -6.86906874e-02 3.21349539e-02 6.99426413e-01
-1.95797801e-01 -2.00762376e-01 -1.09758627e+00 6.53609693e-01
-1.77873683e+00 -7.28603244e-01 2.35895082e-01 1.65610778e+00
7.25575209e-01 2.42191684e-02 -3.77748013e-02 4.29630280e-02
7.22070813e-01 -1.13149390e-01 -8.48877728e-01 -6.07198954e-01
-2.64236808e-01 4.22967553e-01 2.45173290e-01 -1.69961780e-01
-5.63239634e-01 8.79812658e-01 6.81465101e+00 5.92168570e-01
-1.23992777e+00 -5.40082306e-02 7.17933178e-01 -1.87975809e-01
-2.82662988e-01 -4.15734686e-02 -8.18900943e-01 2.26558074e-01
8.02573800e-01 -8.22154209e-02 6.97410643e-01 8.10129642e-01
3.80223989e-02 4.67126012e-01 -1.18816292e+00 8.81817758e-01
-2.96256542e-01 -1.80898964e+00 2.81156152e-01 1.71273291e-01
9.51487601e-01 1.64945066e-01 2.10762784e-01 -1.34098038e-01
3.76450922e-03 -1.36591184e+00 9.55203354e-01 5.44827878e-01
4.20937628e-01 -7.09248304e-01 5.57004392e-01 7.29768425e-02
-1.15651011e+00 -5.02802849e-01 -5.11851728e-01 -1.09971881e-01
-3.03727776e-01 4.04262930e-01 -6.91524804e-01 -1.67624895e-02
8.28682482e-01 5.95706344e-01 -6.94053769e-01 1.17561197e+00
-1.77638963e-01 4.81746525e-01 -2.84266565e-02 -4.77678150e-01
5.37453353e-01 -2.71135792e-02 4.57610816e-01 1.08580887e+00
3.67336094e-01 8.55170339e-02 -5.33724785e-01 1.62832260e+00
-1.37965322e-01 -1.74326420e-01 -4.66400117e-01 -3.23915005e-01
6.91435993e-01 1.27856350e+00 -8.72169614e-01 1.20706238e-01
-1.21588156e-01 6.06299520e-01 3.88543695e-01 5.02411664e-01
-6.13275766e-01 -4.66111660e-01 8.46504688e-01 5.26068471e-02
5.18214881e-01 -2.33597234e-01 -8.34117651e-01 -6.11950934e-01
-1.62002742e-01 -9.10320222e-01 -2.76286732e-02 -5.11826038e-01
-1.08534455e+00 9.89320040e-01 -3.15156907e-01 -8.52915108e-01
1.35531291e-01 -8.18763793e-01 -1.08058727e+00 8.23046923e-01
-1.46806371e+00 -9.64061320e-01 -3.74639899e-01 5.09270549e-01
5.23600221e-01 -6.78459227e-01 6.03201270e-01 -5.80721349e-02
-9.52490807e-01 7.22675681e-01 -1.50846168e-01 9.43395570e-02
2.01286197e-01 -8.74984086e-01 6.59230769e-01 6.97013915e-01
1.05237618e-01 1.01057017e+00 5.13553619e-01 -9.47125331e-02
-1.49755061e+00 -9.93063211e-01 4.93981332e-01 -1.38050005e-01
5.01686931e-01 -1.77089378e-01 -8.86602938e-01 3.14072311e-01
3.98611099e-01 -3.15594405e-01 6.99409783e-01 1.02708064e-01
-2.88292229e-01 -2.15103179e-01 -9.04406428e-01 8.27689230e-01
1.27366006e+00 -1.19723246e-01 -4.80239809e-01 -2.05760837e-01
8.97214830e-01 1.40876710e-01 -5.88016391e-01 3.95473093e-01
8.25387418e-01 -8.83681834e-01 9.21045601e-01 -7.73406863e-01
7.04604626e-01 -2.32804522e-01 -1.74642894e-02 -1.53246152e+00
-8.63458753e-01 -7.28375673e-01 -1.86572522e-01 9.15109932e-01
6.64682686e-01 -8.52672815e-01 8.68229270e-01 6.65744305e-01
-3.02104980e-01 -1.16353571e+00 -8.45677972e-01 -6.30280972e-01
1.40511081e-01 -2.08563432e-01 1.02063155e+00 5.56284904e-01
-2.75493026e-01 2.72538006e-01 1.16588823e-01 -1.20480634e-01
3.91660333e-01 -2.04969123e-01 5.17896533e-01 -1.41601133e+00
-4.35567737e-01 -1.25995743e+00 -3.96546453e-01 -1.02122211e+00
-2.43214378e-03 -6.88625872e-01 -2.51703244e-02 -1.21088934e+00
-5.31516299e-02 -6.65069818e-01 -5.23721516e-01 6.78203642e-01
2.13832423e-01 2.40794267e-03 3.15665334e-01 4.35663581e-01
-1.88254923e-01 6.36924326e-01 1.46872342e+00 -1.20271668e-01
-3.32708508e-01 -2.00797379e-01 -1.16179943e+00 5.86993515e-01
1.19847429e+00 -3.87804985e-01 -5.39868772e-01 -9.46795762e-01
5.99542975e-01 -6.64080858e-01 2.63769418e-01 -1.17157817e+00
6.44679129e-01 -1.89672127e-01 3.81746978e-01 -3.01055284e-03
-8.10556673e-03 -8.60192418e-01 1.08532362e-01 8.02950501e-01
-5.45125604e-01 2.42090091e-01 4.24465984e-01 3.50775301e-01
-1.36469572e-03 -4.14760441e-01 9.17461276e-01 -3.76805037e-01
-8.38974595e-01 2.11848393e-01 -3.47584933e-01 -2.38163378e-02
7.65136600e-01 -5.77448130e-01 -4.10466254e-01 1.10013798e-01
-3.90332669e-01 1.44098490e-01 1.43413708e-01 6.06449366e-01
7.78848827e-01 -1.36059153e+00 -6.17553651e-01 4.36219484e-01
-3.75776626e-02 -5.90708852e-03 6.03877269e-02 7.39800394e-01
-6.21295929e-01 6.69006050e-01 -7.25818694e-01 -5.98265469e-01
-7.59676933e-01 1.94814861e-01 7.96757936e-01 8.03578123e-02
-2.33189896e-01 1.24357402e+00 5.40897369e-01 -3.35885674e-01
2.43808925e-01 -1.22788310e-01 -3.07678550e-01 -2.09250361e-01
5.88364363e-01 1.71168089e-01 -9.73782912e-02 -2.11364776e-01
-2.53169239e-01 5.34482598e-01 -5.37989400e-02 1.23349495e-01
1.65139258e+00 -1.92094110e-02 -4.77103978e-01 1.49585865e-03
1.06936705e+00 -8.72946799e-01 -1.23921907e+00 -3.22775900e-01
1.50532559e-01 -1.12457955e-02 1.95601776e-01 -5.83008587e-01
-1.63325083e+00 8.25624585e-01 4.34171796e-01 1.20056450e-01
1.33939159e+00 -8.89859647e-02 6.60180330e-01 5.75941026e-01
3.00149143e-01 -1.07032633e+00 4.46288496e-01 7.67521679e-01
9.67039704e-01 -9.05814767e-01 -2.62269318e-01 1.43292934e-01
-4.19614196e-01 1.52925313e+00 9.67362523e-01 -1.48089886e-01
8.56510937e-01 1.45051647e-02 -2.75417209e-01 -5.93317926e-01
-9.14431214e-01 -6.50639236e-02 5.18686533e-01 5.59070826e-01
5.25668561e-01 -1.22086883e-01 -7.23234415e-02 7.05242038e-01
-4.52994198e-01 -1.56944454e-01 1.13624610e-01 8.38851690e-01
-5.07225633e-01 -1.05527532e+00 -1.06657356e-01 2.37401515e-01
-1.31430820e-01 -3.29459727e-01 -5.82489729e-01 5.28439820e-01
3.73380125e-01 7.48614013e-01 2.13191703e-01 -6.03033364e-01
3.41209263e-01 6.21062592e-02 4.94627804e-01 -6.11463428e-01
-7.78390408e-01 -4.21068192e-01 -3.25073123e-01 -4.22671914e-01
-3.06911230e-01 -2.45748654e-01 -1.04530525e+00 -5.63385725e-01
-3.25267822e-01 8.77557695e-03 7.47186303e-01 9.73157525e-01
8.32128227e-01 9.17182326e-01 4.57056671e-01 -9.95434940e-01
-3.52306455e-01 -6.73323631e-01 -3.43241215e-01 -6.22839406e-02
1.20248638e-01 -4.86834943e-01 -1.18374318e-01 2.64224149e-02] | [8.458636283874512, 3.1661124229431152] |
41b54be7-068b-4ec8-8149-da79b175816f | contour-integration-using-graph-cut-and-non | 2010.14561 | null | https://arxiv.org/abs/2010.14561v2 | https://arxiv.org/pdf/2010.14561v2.pdf | Contour Integration using Graph-Cut and Non-Classical Receptive Field | Many edge and contour detection algorithms give a soft-value as an output and the final binary map is commonly obtained by applying an optimal threshold. In this paper, we propose a novel method to detect image contours from the extracted edge segments of other algorithms. Our method is based on an undirected graphical model with the edge segments set as the vertices. The proposed energy functions are inspired by the surround modulation in the primary visual cortex that help suppressing texture noise. Our algorithm can improve extracting the binary map, because it considers other important factors such as connectivity, smoothness, and length of the contour beside the soft-values. Our quantitative and qualitative experimental results show the efficacy of the proposed method. | ['Zahra Mousavi Kouzehkanan', 'Babak Nadjar Araabi', 'Reshad Hosseini'] | 2020-10-27 | null | null | null | null | ['contour-detection'] | ['computer-vision'] | [ 2.60258645e-01 -2.37063747e-02 -2.88981825e-01 -2.14316264e-01
2.21580446e-01 -3.45465988e-01 3.08362424e-01 2.18359932e-01
-6.01017952e-01 3.47746342e-01 8.16449150e-02 -3.94490287e-02
-1.42231464e-01 -9.54846919e-01 -1.91551924e-01 -6.61141038e-01
-6.28387854e-02 -4.52225387e-01 8.60718429e-01 -1.37206286e-01
7.02561677e-01 5.36429405e-01 -1.25688493e+00 1.41517147e-01
9.55765188e-01 1.08270860e+00 8.28090683e-02 4.62124616e-01
-3.99170429e-01 5.09499848e-01 -3.54501694e-01 -1.84317544e-01
2.03562930e-01 -6.05568230e-01 -2.24449709e-01 2.74697512e-01
-1.90634519e-01 -9.46917757e-02 -7.40846917e-02 1.55891240e+00
2.86204606e-01 -9.32338610e-02 8.32812428e-01 -1.07359254e+00
-6.44328833e-01 3.32644939e-01 -9.96836185e-01 2.53144979e-01
-8.55009928e-02 -9.65176448e-02 4.09146041e-01 -7.55813777e-01
6.73639357e-01 9.54841137e-01 5.19094646e-01 2.04466395e-02
-7.99240291e-01 -3.67671281e-01 1.95521876e-01 3.97604436e-01
-1.56272817e+00 -1.94223166e-01 1.34880865e+00 -2.30848819e-01
3.48547965e-01 3.37146550e-01 7.31044769e-01 1.99321687e-01
6.21850908e-01 5.96362174e-01 1.33220053e+00 -6.52482986e-01
2.28401646e-01 1.04106478e-01 1.79301634e-01 8.57654214e-01
3.65969121e-01 -3.26317213e-02 -1.08462103e-01 -2.47027501e-02
1.12465358e+00 -1.14340737e-01 -4.70566452e-01 -3.55477929e-01
-8.01372409e-01 6.65530503e-01 4.92364973e-01 6.06987298e-01
-3.30897480e-01 1.30495224e-02 9.09814239e-02 -1.21988684e-01
3.02075058e-01 -3.00851405e-01 -2.56822966e-02 4.03258264e-01
-1.14250481e+00 -4.15020943e-01 5.45462072e-01 5.69483876e-01
6.53892756e-01 -4.02632803e-02 -6.21488094e-02 5.45710564e-01
5.95443845e-01 4.07479942e-01 4.08248276e-01 -7.40910232e-01
-1.77405849e-01 8.71450841e-01 -4.96003404e-02 -1.58226585e+00
-4.10719872e-01 -3.21665525e-01 -9.22371864e-01 8.44360232e-01
3.00598592e-01 -2.39328086e-01 -1.04295468e+00 1.27525306e+00
3.03411663e-01 8.97767842e-02 -1.99495673e-01 1.00081909e+00
8.41688931e-01 5.42495251e-01 3.25135365e-02 -5.23627758e-01
1.25183082e+00 -6.81820035e-01 -1.28091860e+00 -1.09274730e-01
-3.65058705e-02 -9.44066286e-01 5.05829096e-01 4.98547643e-01
-9.62255120e-01 -4.70601201e-01 -1.19703841e+00 1.95231363e-01
-5.06627142e-01 4.76647586e-01 5.04449248e-01 5.42578042e-01
-1.13760352e+00 4.47673142e-01 -7.87488937e-01 -1.68586642e-01
2.02092484e-01 2.11660624e-01 -2.65185684e-01 3.71360600e-01
-7.96510994e-01 7.73194909e-01 6.02670908e-01 5.33221841e-01
2.02218264e-01 2.98740685e-01 -7.56613493e-01 2.71789040e-02
1.20554239e-01 -1.83097735e-01 5.57550788e-01 -1.11993229e+00
-1.53204989e+00 7.37820566e-01 -8.97945091e-02 -1.92978144e-01
4.53166038e-01 3.04239541e-01 -4.38323647e-01 7.09112585e-01
-3.90186161e-01 5.63617706e-01 7.73353159e-01 -1.30946362e+00
-4.00474727e-01 -2.88960576e-01 -4.34837341e-01 1.42555714e-01
-3.17963988e-01 2.36623790e-02 -6.46798432e-01 -7.56939948e-01
7.44092703e-01 -3.72572839e-01 -1.70109451e-01 2.48916909e-01
-3.13689530e-01 -3.82531360e-02 1.01044559e+00 -8.63759756e-01
1.45329869e+00 -2.24869561e+00 -1.74094841e-01 6.69459760e-01
1.96456403e-01 1.55710340e-01 2.45880902e-01 1.27296761e-01
1.71618208e-01 1.82079151e-01 -6.61828279e-01 4.03436095e-01
-4.38591450e-01 -1.75481796e-01 1.62116006e-01 6.25952482e-01
3.49165827e-01 6.83762014e-01 -6.57971799e-01 -1.13402295e+00
3.42081070e-01 6.50736988e-01 -9.30710956e-02 -1.70997918e-01
1.55466065e-01 -9.87692270e-03 -5.47837555e-01 5.88639319e-01
1.08200300e+00 6.49361238e-02 1.61411509e-01 -3.45798731e-01
-4.16961402e-01 -4.87838417e-01 -1.65447974e+00 1.23644221e+00
8.37874711e-02 6.87971354e-01 3.70798379e-01 -7.53753960e-01
1.52102828e+00 2.28065923e-02 3.94360870e-01 -7.08327472e-01
7.00687408e-01 1.54438555e-01 1.38780698e-01 -5.61906874e-01
3.51174474e-01 1.96474381e-02 3.96694750e-01 1.27999842e-01
-2.23760620e-01 -7.84589052e-02 9.89744663e-02 1.15328275e-01
6.18685305e-01 3.41166586e-01 5.32755852e-01 -5.76483786e-01
6.19342744e-01 -6.06933497e-02 6.19782150e-01 2.63184816e-01
-4.13913578e-01 5.58383346e-01 5.59745967e-01 -6.57264143e-02
-6.50574088e-01 -9.01116967e-01 -5.14864504e-01 4.12189722e-01
6.46590054e-01 -3.57076563e-02 -1.05477405e+00 -4.22594935e-01
-3.16466212e-01 3.90102625e-01 -5.76098382e-01 -1.17498875e-01
-5.42477310e-01 -7.37423182e-01 1.37758479e-01 3.81197393e-01
8.53399813e-01 -1.25464249e+00 -8.63227367e-01 2.35845745e-01
-5.62468078e-03 -8.15460742e-01 -4.50685978e-01 7.11115077e-02
-1.08568406e+00 -8.89850378e-01 -7.14633763e-01 -1.17237222e+00
9.91248786e-01 1.67181835e-01 6.32946014e-01 4.47369784e-01
-2.62328565e-01 -2.32284427e-01 -3.91487420e-01 -2.56215990e-01
-1.42552769e-02 -3.38160038e-01 -5.68713367e-01 2.75740355e-01
2.19439149e-01 -4.17426735e-01 -7.97181726e-01 2.98233390e-01
-1.02780819e+00 6.97537437e-02 8.06262732e-01 4.76587623e-01
7.01942742e-01 2.81837076e-01 4.21984196e-01 -7.00037360e-01
7.53937960e-01 -1.14510968e-01 -6.04585171e-01 2.16173694e-01
-5.07315636e-01 2.29480162e-01 3.97262216e-01 -4.63816345e-01
-1.10299754e+00 2.64835984e-01 2.61864252e-03 -6.40162602e-02
-8.52600187e-02 2.92915165e-01 -1.00952178e-01 -5.07175863e-01
4.51138467e-01 3.40180397e-01 -1.01215197e-02 -3.38596821e-01
3.65110189e-01 7.65163839e-01 7.67004550e-01 -4.60372306e-02
4.67313886e-01 6.78554714e-01 7.51105696e-02 -8.41936946e-01
-7.66666681e-02 -3.65213662e-01 -6.25808120e-01 -6.31864429e-01
8.74851227e-01 -7.65174404e-02 -7.62640655e-01 6.16807759e-01
-1.19447887e+00 7.09788352e-02 2.80637562e-01 5.92028379e-01
-2.82410383e-01 6.84168458e-01 -6.90209031e-01 -1.23205030e+00
-4.97160971e-01 -8.38814259e-01 5.95139444e-01 8.13008189e-01
2.27774993e-01 -1.12053931e+00 -8.20266083e-02 -4.45597857e-01
4.37673718e-01 6.01063907e-01 8.56940150e-01 -1.23105653e-01
-3.08673233e-01 -2.15888508e-02 -4.46871549e-01 1.44543633e-01
1.00439839e-01 5.36149204e-01 -7.71218181e-01 1.95168287e-01
1.81668356e-01 2.60482669e-01 1.15697300e+00 8.01155806e-01
9.83702183e-01 -9.62624624e-02 -4.31181133e-01 5.90135217e-01
1.81809497e+00 5.06638169e-01 9.69820023e-01 2.25543648e-01
2.68179983e-01 5.81344724e-01 6.18633211e-01 1.83129951e-01
-1.48884788e-01 2.54269481e-01 4.94217932e-01 -7.62829661e-01
-2.34361023e-01 -5.67925256e-03 1.08547956e-01 9.45515871e-01
-2.84978032e-01 -2.24284157e-01 -6.43567622e-01 4.34509039e-01
-1.81795168e+00 -8.54507029e-01 -6.45055652e-01 2.09388256e+00
8.84519935e-01 6.43260539e-01 -2.73363888e-01 3.65549743e-01
1.06198764e+00 2.72752419e-02 -2.78251708e-01 -6.16808236e-01
-3.55966955e-01 3.07581186e-01 6.56841457e-01 6.64319336e-01
-1.06083059e+00 8.02421093e-01 6.99355745e+00 7.60734379e-01
-1.34155107e+00 -2.58035451e-01 7.00283527e-01 4.22228724e-01
-2.08501637e-01 -2.49333354e-03 -4.05430287e-01 4.52175289e-01
1.12682283e-01 -6.71005696e-02 5.58694303e-02 5.26498139e-01
2.59180695e-01 -6.03062212e-01 -1.19262747e-01 8.72034550e-01
-1.51338279e-02 -9.58724558e-01 -1.68110475e-01 -1.31652519e-01
6.24329209e-01 -5.38903892e-01 -3.63440928e-03 -4.60581601e-01
-8.64854231e-02 -8.45319331e-01 7.36377597e-01 8.93097818e-01
6.26996458e-01 -7.36050129e-01 9.35860097e-01 2.39254922e-01
-1.46277773e+00 2.18655810e-01 -4.47728366e-01 -8.83227736e-02
1.18511058e-01 7.60505855e-01 -5.80973804e-01 4.75040197e-01
3.53419721e-01 3.51984680e-01 -6.73137486e-01 1.46514392e+00
-4.23636734e-01 4.45307374e-01 -5.07762551e-01 -3.02523255e-01
1.24093853e-01 -7.88990200e-01 4.04391855e-01 1.47603810e+00
8.23105201e-02 2.82555252e-01 -3.92263494e-02 1.06356168e+00
2.58543640e-01 5.79812527e-01 -6.35487139e-01 1.71861276e-01
3.21039021e-01 1.63536799e+00 -1.76126373e+00 -2.63894528e-01
-4.29520845e-01 9.01917040e-01 -7.80530050e-02 4.87850338e-01
-6.58003092e-01 -8.62180293e-01 -2.01581746e-01 6.74966797e-02
2.41631165e-01 -3.27229083e-01 -8.33944559e-01 -5.45645714e-01
5.85361533e-02 -2.75449455e-01 1.24880753e-01 -6.63624167e-01
-7.56089807e-01 6.29921019e-01 -1.51001677e-01 -1.06922960e+00
2.54923761e-01 -6.35316551e-01 -1.03422463e+00 7.88698316e-01
-1.34110188e+00 -7.35707223e-01 -4.05371040e-01 4.28085178e-01
2.06225574e-01 3.04920822e-01 3.41740787e-01 -8.05422757e-03
-4.07457769e-01 2.48110697e-01 -2.49388199e-02 4.34227109e-01
4.62466478e-01 -1.09365582e+00 1.95010807e-02 1.29486644e+00
6.80665448e-02 4.95041668e-01 6.62102520e-01 -8.95189285e-01
-8.26445818e-01 -4.05632466e-01 6.81757987e-01 3.41327965e-01
4.15505618e-01 -1.94442973e-01 -1.05628717e+00 1.68701947e-01
5.09572327e-01 -1.16397860e-03 1.01163082e-01 -5.23432374e-01
2.16576785e-01 1.28772631e-01 -1.26977074e+00 4.74118024e-01
7.95723319e-01 1.53577188e-02 -6.19240284e-01 -2.22547069e-01
3.54733884e-01 -9.62706804e-02 -5.97642541e-01 4.85791624e-01
5.68896770e-01 -1.15122509e+00 6.64811790e-01 1.62052959e-01
1.10196486e-01 -6.49102807e-01 3.76724869e-01 -1.02144206e+00
-5.49930573e-01 -2.50451982e-01 2.50448048e-01 1.17880344e+00
3.39689255e-01 -5.26655614e-01 6.97825134e-01 1.63629219e-01
1.32898092e-01 -8.04422617e-01 -6.48695767e-01 -4.10147399e-01
-4.20135379e-01 -4.12934944e-02 1.55559346e-01 8.41467202e-01
1.06613465e-01 1.13589698e-02 5.97627871e-02 6.24653473e-02
7.60793686e-01 6.81506097e-02 -4.01002914e-02 -1.33312964e+00
9.35262442e-02 -8.24790239e-01 -6.85523808e-01 -7.72493482e-01
-3.40826273e-01 -6.50601685e-01 9.43680201e-03 -1.71756756e+00
2.02256382e-01 -2.73832321e-01 -5.27093470e-01 4.08649981e-01
-2.83747584e-01 3.68842304e-01 7.98242241e-02 -6.35643378e-02
-2.81075090e-01 3.40754330e-01 1.54967058e+00 2.57597137e-02
-4.48754609e-01 -2.10738182e-01 -4.96933877e-01 1.08445501e+00
9.96064484e-01 -3.62460107e-01 -2.26164192e-01 -1.31498411e-01
9.45142880e-02 -1.54354542e-01 1.95036903e-01 -8.91128957e-01
4.53985631e-01 -2.99419999e-01 6.40665472e-01 -5.35370648e-01
-2.80618835e-02 -9.18275595e-01 5.34621021e-03 7.57915616e-01
7.04867691e-02 -9.81907547e-02 1.97014809e-01 4.45174009e-01
-2.74669826e-01 -5.21996140e-01 9.83793855e-01 4.59855124e-02
-8.26699197e-01 -3.05789918e-01 -2.80716151e-01 -2.44096443e-01
1.12209058e+00 -5.16388476e-01 -3.96300316e-01 -2.97705203e-01
-6.56318009e-01 -6.93964884e-02 5.46259761e-01 -1.11925136e-02
8.93915415e-01 -1.31980932e+00 -6.94757283e-01 4.18507636e-01
-1.89557090e-01 -3.92991126e-01 -8.51279348e-02 9.94408846e-01
-8.51804614e-01 4.07231860e-02 -6.65200055e-01 -4.32858139e-01
-1.30900073e+00 5.38324952e-01 3.54855090e-01 1.55964226e-01
-5.47061920e-01 5.60606599e-01 1.84132874e-01 3.66200298e-01
9.36643556e-02 -5.17552257e-01 -7.79470742e-01 -1.32577375e-01
4.43270653e-01 5.45483112e-01 -1.07989572e-01 -6.52228117e-01
-4.38251942e-01 1.03563726e+00 3.49737972e-01 -3.11792195e-01
8.47156405e-01 -6.01108419e-03 -3.02768260e-01 2.60814071e-01
1.04403615e+00 4.06349957e-01 -1.17627418e+00 5.47011979e-02
-6.60453504e-03 -4.29662138e-01 5.18854201e-01 -7.10969031e-01
-1.19691372e+00 7.22473264e-01 8.56444001e-01 4.01299357e-01
1.53601551e+00 -2.09958196e-01 5.94392300e-01 -2.63157964e-01
2.33011723e-01 -1.24789488e+00 -1.97243840e-01 -4.74313162e-02
7.00458407e-01 -9.91272867e-01 -1.99531764e-02 -8.62754285e-01
-4.74704832e-01 1.43413603e+00 3.82410914e-01 -3.07856202e-01
7.31441200e-01 6.82092845e-01 2.20105439e-01 -1.18144713e-01
-2.38370597e-01 -3.99321049e-01 4.27499861e-01 5.64348936e-01
3.99941325e-01 3.07970010e-02 -1.20241463e+00 5.99413157e-01
2.31833622e-01 7.80303851e-02 4.68538284e-01 9.76635337e-01
-1.11687183e+00 -9.00450110e-01 -6.78353190e-01 2.80449420e-01
-4.85128462e-01 -6.48602378e-03 -6.23910427e-01 5.68233132e-01
2.94412434e-01 1.08432770e+00 1.47153571e-01 -3.26550215e-01
1.65430427e-01 -4.31003086e-02 4.60060596e-01 2.06279624e-02
-3.17893654e-01 4.41197962e-01 -3.87164533e-01 -2.98557132e-01
-4.58198816e-01 -3.00184876e-01 -1.82067561e+00 -4.30238247e-02
-5.85184336e-01 8.42585713e-02 8.85690928e-01 6.01764619e-01
8.54652971e-02 4.67797816e-01 5.64917684e-01 -5.10893643e-01
2.98857000e-02 -9.60370123e-01 -6.69113696e-01 4.59077537e-01
7.25780278e-02 -6.77384317e-01 -4.78099585e-01 2.56697506e-01] | [10.943737030029297, -2.4188239574432373] |
2e167487-ac23-4405-8958-b292442c73a9 | prediction-guided-multi-objective | null | null | https://proceedings.icml.cc/static/paper_files/icml/2020/1114-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/1114-Paper.pdf | Prediction-Guided Multi-Objective Reinforcement Learning for Continuous Robot Control | Many real-world control problems involve conflicting objectives where we desire a dense and high-quality set of control policies that are optimal for different objective preferences (called Pareto-optimal). While extensive research in multi-objective reinforcement learning (MORL) has been conducted to tackle such problems, multi-objective optimization for complex continuous robot control is still under-explored. In this work, we propose an efficient evolutionary learning algorithm to find the Pareto set approximation for continuous robot control problems, by extending a state-of-the-art RL algorithm and presenting a novel prediction model to guide the learning process. In addition to efficiently discovering the individual policies on the Pareto front, we construct a continuous set of Pareto-optimal solutions by Pareto analysis and interpolation. Furthermore, we design six multi-objective RL environments with continuous action space, which is the first benchmark platform to evaluate MORL algorithms on various robot control problems. We test the previous methods on the proposed benchmark problems, and the experiments show that our approach is able to find a much denser and higher-quality set of Pareto policies than the existing algorithms. | ['Daniela Rus', 'Pingchuan Ma', 'Yunsheng Tian', 'Wojciech Matusik', 'Jie Xu', 'Shinjiro Sueda'] | null | null | https://proceedings.icml.cc/static/paper_files/icml/2020/1114-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/1114-Paper.pdf | icml-2020-1 | ['multi-objective-reinforcement-learning'] | ['methodology'] | [-1.39511809e-01 -3.37361664e-01 -4.78152186e-01 -5.20254113e-02
-8.00110698e-01 -1.43060774e-01 4.80787791e-02 2.42786452e-01
-5.05010188e-01 1.34457636e+00 -2.92036738e-02 3.12464833e-02
-8.19873393e-01 -5.53494513e-01 -6.18816555e-01 -8.30359399e-01
-3.71189862e-01 9.35105324e-01 1.93395272e-01 -4.56783652e-01
6.15580559e-01 2.77952343e-01 -2.09433961e+00 -8.49358097e-04
1.33700716e+00 9.44623768e-01 6.35922790e-01 4.16446179e-01
1.25932265e-02 5.59496462e-01 -7.24070370e-01 8.94749835e-02
1.52699158e-01 -3.78495574e-01 -7.63428092e-01 7.79321184e-03
-3.34432811e-01 2.54175216e-01 4.69609946e-01 9.58137691e-01
6.73342168e-01 4.61620092e-01 8.39182496e-01 -1.62809622e+00
-6.07290387e-01 4.06948000e-01 -5.85446119e-01 -3.97212863e-01
1.73052609e-01 1.06616214e-01 1.05359602e+00 -3.77189934e-01
4.63296533e-01 1.60505867e+00 2.43991077e-01 4.60844696e-01
-1.04677284e+00 -3.88712883e-01 1.27650484e-01 4.69417989e-01
-1.12119365e+00 2.48998664e-02 6.13774776e-01 -1.20639093e-01
6.45592511e-01 1.03177622e-01 7.66589463e-01 9.82105315e-01
8.22036386e-01 1.00257158e+00 1.22105050e+00 -3.91022295e-01
6.76990926e-01 -4.32471884e-03 -3.81700307e-01 7.80484617e-01
3.51205379e-01 5.59723556e-01 -5.50023429e-02 -2.40348369e-01
2.90731311e-01 -7.25841746e-02 -1.17411606e-01 -8.48786235e-01
-1.02777386e+00 1.02791083e+00 2.25718543e-01 2.23213330e-01
-6.80650175e-01 1.14379831e-01 2.71387815e-01 4.28394139e-01
-9.63284820e-02 1.07235301e+00 -5.88909984e-01 -4.95880507e-02
-4.29454029e-01 5.17996550e-01 7.93457389e-01 8.19109380e-01
8.34643185e-01 1.90502971e-01 -3.76298517e-01 9.38809633e-01
3.56513053e-01 3.92609924e-01 4.49492991e-01 -1.26131749e+00
1.63800851e-01 5.66310704e-01 6.25326991e-01 -9.40374911e-01
-4.78462517e-01 -6.78588331e-01 -6.33937478e-01 8.74678850e-01
1.95797205e-01 -4.12576526e-01 -3.87349099e-01 1.33381283e+00
2.21110329e-01 -1.15137853e-01 3.87844473e-01 9.51570272e-01
-1.00022675e-02 9.06091213e-01 -1.32915899e-01 -7.03556001e-01
7.98218906e-01 -1.18714070e+00 -6.43020034e-01 -7.34166950e-02
4.55567241e-01 -5.89218020e-01 1.13696587e+00 6.75394416e-01
-7.75515199e-01 -5.09944379e-01 -1.10873520e+00 7.14101553e-01
-1.26161724e-01 1.54677570e-01 4.84201014e-01 4.28050071e-01
-8.20875049e-01 7.70103633e-01 -3.10545325e-01 -2.05644071e-01
3.00201565e-01 3.84340286e-01 1.47356778e-01 -1.62174553e-01
-1.01034951e+00 1.30414593e+00 9.73772407e-01 -2.85996616e-01
-1.16439617e+00 -3.43482733e-01 -5.59242725e-01 -4.51619737e-02
1.05019653e+00 -4.22419965e-01 1.19829357e+00 -1.19903600e+00
-2.02553916e+00 1.62380368e-01 3.40161800e-01 -5.15116751e-02
2.40790009e-01 -7.92005733e-02 -5.04190385e-01 -1.71381310e-01
-6.55761808e-02 5.75108886e-01 9.66682971e-01 -1.82478046e+00
-1.22220826e+00 -1.14148945e-01 9.75567997e-02 2.73994088e-01
-3.69077921e-01 -6.90132305e-02 -7.61062130e-02 -3.38738769e-01
-7.00030863e-01 -7.70307720e-01 -6.13017917e-01 -4.94243145e-01
3.61178070e-02 -5.31963229e-01 6.32108688e-01 -2.19200224e-01
1.24490917e+00 -1.73171163e+00 7.38604784e-01 2.77391940e-01
-2.26681232e-01 2.23218143e-01 -4.31031644e-01 3.91640186e-01
4.01487172e-01 -8.66180584e-02 -1.86808422e-01 -1.95140168e-01
2.98697263e-01 6.03761971e-01 6.31925762e-02 3.42752457e-01
3.85556258e-02 6.02562666e-01 -1.21762764e+00 -7.14853406e-01
2.39664707e-02 -1.49382785e-01 -7.25796044e-01 2.84297138e-01
-7.16602147e-01 4.78531837e-01 -8.22998166e-01 9.89649296e-01
2.43071780e-01 1.42311931e-01 2.79872984e-01 2.81654596e-01
-4.12076920e-01 -6.40524983e-01 -1.28732646e+00 1.59548354e+00
-5.62012970e-01 1.63834598e-02 2.17896193e-01 -1.24834704e+00
1.46340954e+00 3.04771122e-03 8.68645310e-01 -7.89235830e-01
5.11267066e-01 5.70595205e-01 1.01891689e-01 -5.08180499e-01
7.13490605e-01 -1.42529026e-01 -2.48198450e-01 1.84531584e-01
4.30258922e-02 -7.74053708e-02 4.42043573e-01 -8.03109229e-01
7.67821908e-01 4.29067373e-01 5.23482561e-01 -5.13702691e-01
7.76190460e-01 2.81751275e-01 8.99821699e-01 6.22430921e-01
-1.80959672e-01 -8.22545588e-02 4.37588990e-01 -3.47117245e-01
-8.83449018e-01 -6.75812066e-01 3.43018144e-01 1.20507348e+00
4.43936527e-01 -9.66239795e-02 -2.19044462e-01 -6.37184918e-01
1.61252439e-01 1.11449432e+00 -3.08673590e-01 -2.19287321e-01
-6.02446496e-01 -6.35396361e-01 1.54312000e-01 1.60062909e-01
2.29076028e-01 -1.52595270e+00 -1.12545514e+00 5.73130071e-01
1.64438710e-02 -4.96483117e-01 -3.07355255e-01 2.30116412e-01
-7.18035340e-01 -1.23452497e+00 -7.03216195e-01 -1.12286139e+00
2.93094665e-01 -1.76911339e-01 1.09488845e+00 -1.11906931e-01
-2.51702696e-01 2.79911309e-01 -5.15481114e-01 -6.57353759e-01
-5.98580062e-01 -5.64652048e-02 1.35543048e-01 4.08523791e-02
-1.06491238e-01 -3.89194399e-01 -9.39036831e-02 6.50598466e-01
-6.25570714e-01 -4.89303589e-01 1.04174483e+00 9.06770170e-01
1.01566386e+00 7.35083699e-01 8.89683902e-01 -3.22510228e-02
1.13035607e+00 -5.93471348e-01 -1.12053561e+00 6.08547032e-01
-7.91383147e-01 7.00080216e-01 1.12581146e+00 -5.70802569e-01
-8.81560147e-01 -1.71152085e-01 1.24519192e-01 -6.03302598e-01
8.61207256e-04 4.43794727e-01 -2.09814847e-01 7.13881105e-02
3.88009012e-01 2.01114357e-01 2.72102028e-01 -3.48734140e-01
1.91911653e-01 4.84189659e-01 2.77269155e-01 -1.08429730e+00
3.39651525e-01 -6.54917508e-02 4.07579243e-01 -4.92307574e-01
-5.74544489e-01 -3.97191256e-01 -2.86060154e-01 -4.17314559e-01
7.77683437e-01 -1.59099042e-01 -1.23989415e+00 4.80529293e-02
-9.21421230e-01 -2.52698928e-01 -3.39414120e-01 5.74662387e-01
-1.33438671e+00 4.35325950e-02 2.77639508e-01 -1.15241635e+00
-1.67158067e-01 -1.47999680e+00 7.61876583e-01 2.08199054e-01
-9.85668320e-03 -7.75623441e-01 4.64553893e-01 -2.35283375e-01
2.62493730e-01 5.12528181e-01 1.14758670e+00 -3.02876413e-01
-1.76369399e-01 3.13315213e-01 4.12765026e-01 2.00423777e-01
3.49378772e-02 -3.73325050e-02 -2.42178421e-02 -6.07758403e-01
1.35920299e-02 -7.05502033e-01 4.95279104e-01 4.50828493e-01
1.20516896e+00 -3.46315891e-01 -4.19577718e-01 3.27354252e-01
1.78785598e+00 9.28481638e-01 4.80886787e-01 7.69922137e-01
6.24487922e-02 7.22133636e-01 1.48025358e+00 7.69935727e-01
2.19676822e-01 7.37271667e-01 9.88935173e-01 2.95990080e-01
5.43350101e-01 -1.79742619e-01 5.59535444e-01 4.43793178e-01
-2.61799514e-01 -5.15275419e-01 -8.24528813e-01 4.39663976e-01
-2.40264821e+00 -1.02925146e+00 3.84856045e-01 2.12805223e+00
5.23758650e-01 -2.90557723e-02 4.62078780e-01 1.25635013e-01
8.31738770e-01 -1.39981091e-01 -6.78859532e-01 -6.66544795e-01
5.40703908e-02 2.59433463e-02 3.53728831e-01 4.51156169e-01
-1.08266997e+00 7.60331392e-01 5.97162962e+00 1.06340253e+00
-9.23001111e-01 -1.74522594e-01 2.76320904e-01 -1.06659815e-01
-1.09013304e-01 -3.36585313e-01 -7.38576233e-01 4.05596435e-01
6.86808646e-01 -1.64411753e-01 8.47061694e-01 1.02686656e+00
3.14359486e-01 -7.95979798e-02 -7.24566281e-01 1.01108587e+00
-2.67017215e-01 -1.16972136e+00 -1.55076832e-01 1.45060942e-01
1.09135211e+00 -3.02620977e-01 2.87620537e-02 5.51165283e-01
7.32823491e-01 -1.21122265e+00 8.20116162e-01 7.12748647e-01
5.14704108e-01 -1.52978325e+00 7.96844780e-01 6.63409948e-01
-1.02511072e+00 -9.56802964e-01 -7.34677672e-01 1.70278788e-01
5.49964681e-02 1.65172920e-01 -5.70567489e-01 1.03670990e+00
6.44388080e-01 7.73715138e-01 -1.15837373e-01 1.37322330e+00
-1.22303739e-01 4.06199135e-02 1.15228988e-01 -8.47684979e-01
7.20762491e-01 -3.83275062e-01 8.11668038e-01 6.77948475e-01
7.48597801e-01 -1.60805628e-01 7.12542355e-01 6.83367670e-01
3.15404415e-01 3.92256469e-01 -5.23447454e-01 -8.92660320e-02
1.50589138e-01 1.14909685e+00 -4.85934377e-01 2.47100040e-01
2.78270971e-02 6.03772402e-01 5.36089420e-01 1.86775550e-01
-1.02915406e+00 -3.87516975e-01 7.38653541e-01 -4.94317710e-01
4.72755700e-01 -2.27329031e-01 8.27508792e-02 -5.79276264e-01
-2.61840373e-01 -1.11963165e+00 4.36926961e-01 -3.98133159e-01
-1.31778443e+00 4.92695272e-01 2.25431826e-02 -1.49418485e+00
-2.96662837e-01 -8.19359839e-01 -5.16073704e-01 4.03506458e-01
-1.59738588e+00 -5.91075957e-01 5.85551448e-02 6.08573258e-01
7.15183854e-01 -8.72573674e-01 7.49166310e-01 -4.48354408e-02
-2.18660057e-01 2.01310590e-01 8.10274422e-01 -8.36815536e-01
6.16005242e-01 -1.17778957e+00 -8.02832425e-01 1.75050035e-01
-4.19328779e-01 -1.21927440e-01 7.99011350e-01 -5.05029619e-01
-1.72935724e+00 -1.08079839e+00 1.83463916e-01 -2.46341173e-02
6.33100390e-01 2.46875957e-01 -4.66101050e-01 4.32857238e-02
3.60955417e-01 -4.71132129e-01 4.25824136e-01 2.28113253e-02
4.86777991e-01 -3.51938993e-01 -1.17434180e+00 6.71543598e-01
9.03208613e-01 7.08333552e-01 -5.87605059e-01 -1.45495795e-02
7.12379515e-01 3.10806121e-04 -8.55435073e-01 5.73342144e-01
4.73627090e-01 -8.29019785e-01 8.44140232e-01 -7.59086370e-01
5.14553666e-01 -4.98767108e-01 -3.72516006e-01 -2.14658809e+00
-6.98324919e-01 -6.42371118e-01 -1.23705581e-01 9.35481131e-01
1.69747233e-01 -6.04401231e-01 4.51057255e-01 -4.85840142e-01
-4.97475564e-01 -1.35609949e+00 -8.74729276e-01 -1.28135931e+00
1.19246721e-01 7.18209799e-03 8.16613257e-01 4.65659708e-01
-1.89516917e-01 3.07725310e-01 -6.06938243e-01 -4.77411859e-02
8.15829456e-01 4.55596656e-01 5.29252470e-01 -1.32967079e+00
-4.89822805e-01 -8.20872843e-01 1.16070591e-01 -5.46444178e-01
5.19087851e-01 -7.14634418e-01 4.87158269e-01 -1.64419758e+00
-2.60020763e-01 -7.20706224e-01 -4.58436519e-01 2.23067835e-01
3.71705741e-02 -5.49838662e-01 2.81690896e-01 3.13601121e-02
-1.11146832e+00 1.22159624e+00 1.73651314e+00 -4.01348889e-01
-4.27245051e-01 2.05864608e-01 -6.92971766e-01 7.29087889e-01
1.13749540e+00 -6.03492677e-01 -6.14849031e-01 -1.04054242e-01
4.49197926e-02 4.47866648e-01 -1.46831647e-01 -1.13414299e+00
-3.49126495e-02 -9.82130706e-01 8.22188631e-02 -6.71206951e-01
1.78378150e-01 -1.12313211e+00 8.53314549e-02 7.97484040e-01
-1.61002681e-01 2.73905456e-01 1.37347311e-01 7.14232147e-01
-2.40651086e-01 -6.18110418e-01 1.02145553e+00 -2.10755467e-01
-9.92214859e-01 2.02683166e-01 -4.28620785e-01 6.48947209e-02
1.51039028e+00 -2.47247014e-02 1.98692679e-02 -5.88349961e-02
-3.48094136e-01 8.64009917e-01 3.37861359e-01 6.16721451e-01
8.14964056e-01 -1.33077669e+00 -8.65941942e-01 -3.67615998e-01
2.45599654e-02 -3.55959862e-01 -2.78041333e-01 4.67389375e-01
-2.06966624e-01 3.81070256e-01 -8.10628235e-01 -3.12272012e-01
-1.01981926e+00 7.95505106e-01 4.23397839e-01 -7.19550788e-01
-1.58152983e-01 2.95604408e-01 -4.87495601e-01 -6.11437023e-01
3.58354330e-01 -1.07124507e-01 -6.75371170e-01 1.25155449e-01
1.07070275e-01 1.03671646e+00 -4.61060435e-01 -2.54755646e-01
-3.25839013e-01 6.40805483e-01 5.58489621e-01 -4.24896590e-02
1.55004680e+00 1.29513353e-01 -1.15977690e-01 3.09945613e-01
9.43805456e-01 -2.34596252e-01 -1.30249381e+00 1.99540257e-01
2.74207920e-01 -4.61196184e-01 -1.86751261e-01 -9.73210514e-01
-8.10515523e-01 3.39528292e-01 5.20241439e-01 9.59319994e-03
1.33277929e+00 -3.45211446e-01 4.43841100e-01 7.45466471e-01
9.44549561e-01 -1.56711876e+00 7.07380593e-01 8.41617703e-01
1.33296752e+00 -1.07959533e+00 -1.26240134e-01 8.92291591e-02
-9.46187735e-01 1.21308172e+00 8.52176905e-01 -2.27480888e-01
3.22022378e-01 1.09659858e-01 -2.71816075e-01 8.81754160e-02
-8.60449195e-01 -3.94083351e-01 1.41328439e-01 8.01465750e-01
-1.26629964e-01 1.48172140e-01 -6.39923632e-01 4.43406314e-01
-7.18272617e-03 -4.06266302e-02 2.00808212e-01 1.05757761e+00
-1.03401458e+00 -1.44197512e+00 -8.02163243e-01 3.22324574e-01
-1.34975761e-01 5.41589677e-01 -1.91805568e-02 7.51177549e-01
3.52412581e-01 1.07370639e+00 -2.97795177e-01 -3.07901710e-01
4.66062546e-01 -2.90499687e-01 6.78041160e-01 -4.03347433e-01
-3.36381257e-01 1.41359493e-02 1.19115137e-01 -5.93062580e-01
-4.40476775e-01 -4.91013438e-01 -1.54950309e+00 8.76622275e-02
-2.06881985e-02 5.53526223e-01 4.15446341e-01 8.25919926e-01
2.65690029e-01 8.75687420e-01 8.39152038e-01 -8.54235411e-01
-9.15003717e-01 -5.87027848e-01 -6.48377776e-01 -1.05376162e-01
1.20403603e-01 -1.32756817e+00 1.20350122e-01 -5.58727622e-01] | [4.280002593994141, 2.378115653991699] |
69e9437d-314b-4632-bdca-aba0cdd0fd35 | balanced-and-hierarchical-relation-learning | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Yang_Balanced_and_Hierarchical_Relation_Learning_for_One-Shot_Object_Detection_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Yang_Balanced_and_Hierarchical_Relation_Learning_for_One-Shot_Object_Detection_CVPR_2022_paper.pdf | Balanced and Hierarchical Relation Learning for One-Shot Object Detection | Instance-level feature matching is significantly important to the success of modern one-shot object detectors. Recently, the methods based on the metric-learning paradigm have achieved an impressive process. Most of these works only measure the relations between query and target objects on a single level, resulting in suboptimal performance overall. In this paper, we introduce the balanced and hierarchical learning for our detector. The contributions are two-fold: firstly, a novel Instance-level Hierarchical Relation (IHR) module is proposed to encode the contrastive-level, salient-level, and attention-level relations simultaneously to enhance the query-relevant similarity representation. Secondly, we notice that the batch training of the IHR module is substantially hindered by the positive-negative sample imbalance in the one-shot scenario. We then introduce a simple but effective Ratio-Preserving Loss (RPL) to protect the learning of rare positive samples and suppress the effects of negative samples. Our loss can adjust the weight for each sample adaptively, ensuring the desired positive-negative ratio consistency and boosting query-related IHR learning. Extensive experiments show that our method outperforms the state-of-the-art method by 1.6% and 1.3% on PASCAL VOC and MS COCO datasets for unseen classes, respectively. The code will be available at https://github.com/hero-y/BHRL. | ['Yu Zhang', 'Yong Tang', 'Xian-Sheng Hua', 'Jianqiang Huang', 'Bing Deng', 'Hualian Sheng', 'Sijia Cai', 'Hanqing Yang'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['one-shot-object-detection'] | ['computer-vision'] | [ 6.77125752e-02 -2.77567863e-01 -3.87359619e-01 -6.04726553e-01
-9.25093412e-01 -8.43214989e-02 5.74837863e-01 4.36288834e-01
-5.01984179e-01 3.06276172e-01 -1.80510730e-01 3.43561918e-01
-1.51035354e-01 -6.44970477e-01 -6.00154281e-01 -7.27166355e-01
9.21131596e-02 1.57590583e-01 9.03375268e-01 -1.78321868e-01
2.90774733e-01 4.12318558e-01 -2.08602953e+00 4.65297133e-01
7.36065507e-01 1.44945192e+00 1.65387660e-01 3.81951571e-01
-2.87882965e-02 8.16569269e-01 -4.74938810e-01 -5.51667094e-01
3.29174250e-01 -3.52692872e-01 -4.52781796e-01 -9.61393788e-02
5.70208013e-01 -1.34714887e-01 -4.52258617e-01 1.22987461e+00
7.63836443e-01 2.26328894e-01 5.37101448e-01 -1.45598459e+00
-5.31734645e-01 2.98408568e-01 -8.20387006e-01 5.20727336e-01
6.16882108e-02 6.48616254e-02 1.36550176e+00 -1.37457287e+00
4.16456044e-01 1.20218205e+00 3.06223094e-01 3.21936339e-01
-1.01527047e+00 -8.48514557e-01 1.44309863e-01 8.04469049e-01
-1.60485089e+00 -3.86572570e-01 9.10573304e-01 -3.66028011e-01
7.49595404e-01 2.93894827e-01 4.71042335e-01 6.54167116e-01
4.05747443e-02 1.02487206e+00 7.91045904e-01 -4.35458720e-01
1.24525905e-01 2.75629640e-01 3.39229882e-01 5.83226621e-01
1.94454104e-01 1.09404229e-01 -7.20243812e-01 -3.14506628e-02
2.91507363e-01 2.15844437e-01 -6.94681853e-02 -7.89718747e-01
-8.83412123e-01 8.20657432e-01 7.59766161e-01 3.22160691e-01
-1.86942056e-01 -1.02803439e-01 4.39504355e-01 2.14603409e-01
4.21528757e-01 2.31823981e-01 -2.66850024e-01 1.98526874e-01
-8.42900038e-01 2.11155042e-01 5.34281194e-01 1.11285901e+00
8.81578743e-01 -3.82992655e-01 -6.79814517e-01 1.09735548e+00
2.37632811e-01 1.99175775e-01 3.80571693e-01 -5.13544977e-01
4.17183578e-01 7.98832417e-01 -1.56183124e-01 -1.07313645e+00
-1.25427142e-01 -7.46325076e-01 -5.71586311e-01 -3.17036361e-02
1.63559020e-01 4.69285041e-01 -7.49942243e-01 1.57858229e+00
6.04285538e-01 1.65945977e-01 -2.94507354e-01 8.90698552e-01
9.25355315e-01 3.41719508e-01 9.42874998e-02 -1.49969801e-01
1.43137801e+00 -1.07430303e+00 -6.17549241e-01 -2.23294362e-01
5.03479242e-01 -7.90814102e-01 1.15220487e+00 4.89408933e-02
-9.75339830e-01 -6.19315743e-01 -1.26435626e+00 -1.13596268e-01
-4.33770686e-01 1.24630809e-01 4.82901603e-01 3.89142454e-01
-4.79824364e-01 4.33635145e-01 -5.37888288e-01 -2.03912228e-01
7.50244260e-01 2.66491413e-01 -1.51877671e-01 -2.92504042e-01
-1.27854276e+00 8.29296947e-01 3.98727745e-01 -5.20590767e-02
-8.00580680e-01 -8.04045737e-01 -8.11470747e-01 1.97920054e-01
5.92071950e-01 -2.69792229e-01 1.19941354e+00 -6.85611010e-01
-8.87884021e-01 1.12587929e+00 -2.07298234e-01 -3.59023660e-01
4.87469673e-01 -2.59621263e-01 -2.65980273e-01 9.36171189e-02
1.96331933e-01 7.08962739e-01 7.54864395e-01 -1.13438511e+00
-9.18536901e-01 -5.29137254e-01 -5.17005101e-02 2.82078981e-01
-4.95201051e-01 2.77322084e-01 -7.34465778e-01 -6.29871190e-01
1.23101570e-01 -5.52582026e-01 6.93897381e-02 4.18783516e-01
-1.16487339e-01 -6.21509433e-01 7.28202760e-01 -2.29489312e-01
1.30899310e+00 -2.42472768e+00 -2.22235501e-01 -1.12509698e-01
1.03264853e-01 5.12397289e-01 -2.24992558e-01 2.73150355e-01
-4.50939052e-02 -2.27777198e-01 -1.93559989e-01 -2.75037616e-01
-2.19687019e-02 -5.41178919e-02 -2.05593795e-01 5.29136419e-01
5.73728323e-01 9.13937271e-01 -9.93793488e-01 -6.69214010e-01
1.74751669e-01 3.97488356e-01 -4.63172972e-01 3.86714935e-01
2.10035816e-02 3.74813974e-02 -2.72308499e-01 9.19392645e-01
6.43318772e-01 -2.10269630e-01 -1.34966075e-01 -4.13092464e-01
-2.47895475e-02 2.73496509e-01 -1.24048722e+00 1.52426219e+00
-2.48969961e-02 4.34149653e-01 -1.65511310e-01 -1.08149326e+00
1.08097684e+00 -1.66980142e-03 4.07944769e-01 -1.16407490e+00
1.99882254e-01 2.50758260e-01 7.48234168e-02 -3.82566214e-01
3.57723951e-01 -3.18387821e-02 3.99733149e-02 -5.80158383e-02
1.21111274e-01 9.06743109e-02 2.57074982e-01 2.94756323e-01
9.79261875e-01 -9.19749662e-02 5.17258465e-01 -2.34784707e-01
6.49345458e-01 -2.97865987e-01 8.55610669e-01 6.36717439e-01
-7.12003350e-01 6.03943944e-01 3.91277790e-01 -1.23966530e-01
-8.14668179e-01 -1.00771785e+00 -3.70812565e-01 1.37682343e+00
5.83500147e-01 -3.07429731e-01 -4.00872171e-01 -6.44706368e-01
2.24000499e-01 6.03580952e-01 -6.09768629e-01 -5.29522359e-01
-5.23125708e-01 -8.24609995e-01 2.11207405e-01 5.97956896e-01
4.94223833e-01 -1.03364623e+00 -5.23004532e-01 6.96371123e-02
1.52537785e-02 -9.71345067e-01 -5.20218134e-01 1.82736948e-01
-6.53052330e-01 -1.13894784e+00 -5.73733151e-01 -8.92543554e-01
5.39428949e-01 6.18526340e-01 1.09107602e+00 1.40485197e-01
-7.04869688e-01 1.31294385e-01 -4.19793844e-01 -5.49390614e-01
1.83878720e-01 -5.86860217e-02 -1.26771316e-01 2.98027813e-01
7.65402973e-01 -4.05925512e-01 -7.26929665e-01 5.60859203e-01
-7.14971840e-01 -1.65794462e-01 5.75290859e-01 1.00364661e+00
8.03897202e-01 -2.07555652e-01 6.05287910e-01 -6.61154449e-01
2.60814965e-01 -4.01553482e-01 -5.36925197e-01 3.52150649e-01
-6.80831492e-01 -6.37096539e-03 2.11083278e-01 -5.57299674e-01
-8.65082979e-01 2.65352428e-02 9.07019302e-02 -5.17235637e-01
1.29244015e-01 -1.10019436e-02 -3.84599864e-01 -7.84060135e-02
6.19475186e-01 1.15874417e-01 -3.57691169e-01 -3.42616856e-01
2.99624950e-01 7.12896526e-01 4.78421479e-01 -4.55937654e-01
7.72143722e-01 5.52163184e-01 -1.22961357e-01 -5.55928826e-01
-1.19870675e+00 -1.02185512e+00 -5.99986196e-01 -2.24620685e-01
5.19549072e-01 -9.99567509e-01 -5.09064257e-01 5.02783477e-01
-9.39626336e-01 -3.41683580e-03 -4.96436298e-01 3.77372533e-01
-3.39704871e-01 1.69762209e-01 -5.12944698e-01 -8.76603246e-01
-3.90785128e-01 -9.94264543e-01 1.02074218e+00 4.87158626e-01
1.78957075e-01 -4.17654783e-01 -3.89156975e-02 3.06462407e-01
4.57840472e-01 3.80042270e-02 5.88113606e-01 -8.70326877e-01
-6.47532701e-01 -3.83891314e-01 -5.69252849e-01 3.79731715e-01
4.10435675e-03 -1.58785403e-01 -1.12075889e+00 -3.20606053e-01
1.72981218e-01 -5.97377181e-01 1.12430036e+00 -2.16663647e-02
1.15431380e+00 4.85060439e-02 -3.32683593e-01 4.17898655e-01
1.39143813e+00 5.20315655e-02 5.86069763e-01 2.19775438e-01
6.72062099e-01 7.01078892e-01 1.20789564e+00 4.57152814e-01
3.85829091e-01 8.38203251e-01 4.44048494e-01 7.98666209e-04
-2.86015511e-01 -1.93316594e-01 8.21049660e-02 5.86054385e-01
4.24728006e-01 -8.79462063e-03 -7.64650524e-01 7.63548374e-01
-1.98034012e+00 -8.88816774e-01 -8.92198533e-02 2.40675211e+00
8.63389909e-01 3.65780711e-01 6.59970716e-02 2.44227186e-01
9.00105238e-01 2.51188517e-01 -6.44173801e-01 1.37654543e-01
-2.02193782e-01 7.40104914e-02 3.02141756e-01 3.24599862e-01
-1.30878699e+00 8.22445214e-01 4.74299622e+00 1.09422278e+00
-9.21196818e-01 2.49266863e-01 4.98155326e-01 -4.12540585e-01
-1.81500930e-02 -1.20947197e-01 -1.21911001e+00 4.11711186e-01
5.51311433e-01 -2.63498425e-01 -7.36554787e-02 9.19376373e-01
-1.02175802e-01 -9.98548195e-02 -1.15566945e+00 1.17874610e+00
2.15141907e-01 -9.34292912e-01 -5.47055602e-02 -2.27141723e-01
4.93841380e-01 6.94208220e-02 4.69534136e-02 5.59851766e-01
-2.62108177e-01 -5.48586726e-01 8.58999670e-01 4.67598498e-01
6.01634264e-01 -8.33903968e-01 7.65921056e-01 3.17159116e-01
-1.49142432e+00 -1.95486501e-01 -6.16695285e-01 1.54211998e-01
-1.99601322e-01 7.95609176e-01 -4.56489027e-01 5.39698064e-01
7.68524945e-01 7.58245409e-01 -8.19629610e-01 1.37370181e+00
-2.45501295e-01 4.48801816e-01 -1.99297771e-01 -1.10582292e-01
-4.08853404e-03 1.69731051e-01 5.92750371e-01 1.25695562e+00
-1.29419237e-01 5.52524477e-02 1.87039629e-01 7.10292339e-01
-1.93136364e-01 2.79088885e-01 -1.93977684e-01 2.85391986e-01
6.25026762e-01 1.34043872e+00 -4.76062417e-01 -3.35362285e-01
-5.13815999e-01 8.02448571e-01 6.78668976e-01 1.30084157e-02
-8.43609631e-01 -6.66562974e-01 5.94906807e-01 3.52697819e-02
5.59870303e-01 2.27912799e-01 -2.05799460e-01 -1.03837442e+00
5.26376545e-01 -6.70312703e-01 6.94083810e-01 -2.57425696e-01
-1.62672222e+00 3.29696834e-01 -7.18130469e-02 -1.36932290e+00
1.95791677e-01 -3.49732608e-01 -5.53318501e-01 7.59602308e-01
-1.78649700e+00 -9.57686603e-01 -4.37296361e-01 4.09655511e-01
4.88542199e-01 -4.17586602e-02 4.68471885e-01 7.44832456e-01
-6.37105823e-01 9.52078938e-01 -1.77808136e-01 1.91706732e-01
8.96199048e-01 -1.05110371e+00 1.84174821e-01 8.32835019e-01
1.65395424e-01 4.37945366e-01 5.15004814e-01 -3.50592524e-01
-1.11638618e+00 -1.11225092e+00 1.02208149e+00 -2.96681553e-01
6.09548092e-01 -6.01022184e-01 -1.30645013e+00 1.39578104e-01
-2.05550000e-01 5.79352856e-01 5.60435355e-01 -8.80240798e-02
-8.38773310e-01 -6.60945714e-01 -1.17802918e+00 3.80803019e-01
1.12202728e+00 -5.92949748e-01 -6.52686059e-01 2.70199329e-01
6.86112881e-01 -2.22136423e-01 -7.03251600e-01 7.45194614e-01
5.53331137e-01 -1.08834207e+00 9.14476335e-01 -4.88970429e-01
1.49041757e-01 -4.06733662e-01 -3.41915548e-01 -7.14010596e-01
-4.63996619e-01 -2.20654696e-01 -4.25041318e-01 1.42122841e+00
1.95513800e-01 -4.52239454e-01 6.56494915e-01 3.72893989e-01
-2.10364088e-02 -1.14537644e+00 -9.43506777e-01 -9.94556010e-01
-2.14500219e-01 -3.29514265e-01 3.11451048e-01 7.56844699e-01
-1.80987179e-01 5.40894747e-01 -2.04969913e-01 9.26640704e-02
7.65488863e-01 1.39205620e-01 6.37586832e-01 -1.22029436e+00
-2.98520535e-01 -5.41190803e-01 -7.14105904e-01 -9.68655944e-01
-2.27556467e-01 -9.15613711e-01 2.75078744e-01 -1.25047028e+00
6.58212543e-01 -4.21266973e-01 -8.20841908e-01 3.01145375e-01
-6.55874550e-01 2.79345483e-01 3.07754636e-01 3.13819945e-01
-1.01116002e+00 8.10330272e-01 9.90947485e-01 -1.74319208e-01
-9.27465186e-02 4.75680828e-02 -5.88239133e-01 4.38447624e-01
7.87702978e-01 -8.37139308e-01 -3.31961274e-01 -7.08822235e-02
-6.00463934e-02 -4.85994786e-01 4.41800684e-01 -1.06957567e+00
5.04745305e-01 7.97747150e-02 2.86777943e-01 -8.05531025e-01
2.78752953e-01 -5.87998033e-01 -4.08826590e-01 5.33230543e-01
-4.62386698e-01 -2.75207788e-01 8.10351595e-02 7.20330894e-01
-4.02310938e-01 -1.53936669e-01 1.25479746e+00 1.55205205e-01
-8.03560853e-01 5.14876246e-01 4.56036299e-01 4.09054011e-01
1.20951986e+00 -2.70525888e-02 -3.30415726e-01 1.19125396e-02
-1.67669967e-01 5.24504066e-01 2.12884530e-01 6.65407896e-01
6.29591346e-01 -1.49171519e+00 -8.27334166e-01 1.27128169e-01
6.85599625e-01 -5.72484918e-02 2.90026218e-01 9.69462991e-01
-1.07149715e-02 2.65747160e-01 -1.12236097e-01 -6.93355083e-01
-1.42579162e+00 7.56915987e-01 3.27233702e-01 -3.58072609e-01
-4.25118715e-01 1.20126975e+00 4.22589421e-01 -1.79915920e-01
6.26910865e-01 8.93796384e-02 -1.11480817e-01 2.94780672e-01
9.21367168e-01 4.63737756e-01 1.39780298e-01 -5.77896714e-01
-7.30823636e-01 4.91976738e-01 -4.17303473e-01 1.25240624e-01
1.13505685e+00 8.56265649e-02 -1.28344242e-02 5.71203589e-01
1.35587454e+00 -2.76639700e-01 -1.21717012e+00 -6.85922205e-01
2.16222450e-01 -6.10415637e-01 -8.02934766e-02 -4.23793703e-01
-9.78895664e-01 1.01855779e+00 8.60981584e-01 3.64908539e-02
1.19249010e+00 2.60836273e-01 6.92327261e-01 1.63289279e-01
2.71784067e-01 -1.02567029e+00 3.48163217e-01 3.12582284e-01
8.81192446e-01 -1.38403261e+00 3.81266288e-02 -5.38057685e-01
-4.40217614e-01 6.26580298e-01 8.59435260e-01 -1.17608860e-01
5.93613207e-01 9.56421494e-02 -2.05284223e-01 -1.68759689e-01
-8.47857058e-01 -4.96261477e-01 6.87693000e-01 3.97693455e-01
4.29004580e-01 -8.85591283e-02 -4.39944625e-01 4.98457104e-01
2.59801179e-01 -1.51648492e-01 -1.07734211e-01 1.07961035e+00
-7.59626627e-01 -9.64583993e-01 -1.64729923e-01 4.59137499e-01
-4.40114111e-01 -1.33126050e-01 -3.76400292e-01 6.44114733e-01
2.09180593e-01 9.99606371e-01 1.72027200e-01 -3.60137761e-01
6.69904530e-01 1.22481426e-02 4.73252445e-01 -6.94999397e-01
-5.72822869e-01 -5.31546772e-02 -2.31920153e-01 -7.28548467e-01
-3.17293227e-01 -6.31137788e-01 -1.22822034e+00 1.36265427e-01
-5.36193192e-01 -5.76650687e-02 2.70653605e-01 6.67242169e-01
4.02290165e-01 4.25082803e-01 7.96908021e-01 -6.60142362e-01
-8.47041845e-01 -9.50125694e-01 -4.97044623e-01 5.55243969e-01
3.04048300e-01 -8.58569443e-01 -4.62610245e-01 -3.34123403e-01] | [9.368559837341309, 1.2142620086669922] |
f780a1d8-93c5-44b8-8557-d4f2f5ebb5f8 | simple-yet-effective-code-switching-language | 2305.19759 | null | https://arxiv.org/abs/2305.19759v1 | https://arxiv.org/pdf/2305.19759v1.pdf | Simple yet Effective Code-Switching Language Identification with Multitask Pre-Training and Transfer Learning | Code-switching, also called code-mixing, is the linguistics phenomenon where in casual settings, multilingual speakers mix words from different languages in one utterance. Due to its spontaneous nature, code-switching is extremely low-resource, which makes it a challenging problem for language and speech processing tasks. In such contexts, Code-Switching Language Identification (CSLID) becomes a difficult but necessary task if we want to maximally leverage existing monolingual tools for other tasks. In this work, we propose two novel approaches toward improving language identification accuracy on an English-Mandarin child-directed speech dataset. Our methods include a stacked Residual CNN+GRU model and a multitask pre-training approach to use Automatic Speech Recognition (ASR) as an auxiliary task for CSLID. Due to the low-resource nature of code-switching, we also employ careful silver data creation using monolingual corpora in both languages and up-sampling as data augmentation. We focus on English-Mandarin code-switched data, but our method works on any language pair. Our best model achieves a balanced accuracy of 0.781 on a real English-Mandarin code-switching child-directed speech corpus and outperforms the previous baseline by 55.3%. | ['Bismarck Odoom', 'Tianjian Li', 'Cihan Xiao', 'Shuyue Stella Li'] | 2023-05-31 | null | null | null | null | ['automatic-speech-recognition'] | ['speech'] | [ 1.90426111e-01 -4.39480692e-01 -1.92073286e-01 -3.89025003e-01
-1.17161834e+00 -6.83405578e-01 3.51088017e-01 -1.69543233e-02
-2.78836310e-01 2.11591899e-01 1.34053081e-01 -9.83546257e-01
5.34779310e-01 -6.18810095e-02 -6.31811678e-01 -4.67257679e-01
1.43360049e-01 4.09848511e-01 3.82421575e-02 -3.27383399e-01
-6.65174350e-02 3.62165906e-02 -1.49436557e+00 3.86067986e-01
1.26464593e+00 3.92261565e-01 6.34345889e-01 5.96556187e-01
-4.06167299e-01 1.02221441e+00 -4.10310388e-01 -5.25936604e-01
1.70375556e-01 -4.17841703e-01 -6.91420913e-01 -3.42427604e-02
5.09403169e-01 2.64390279e-02 -2.09104374e-01 1.32406700e+00
3.88940871e-01 -3.12648043e-02 2.93472797e-01 -9.27930355e-01
-8.10661554e-01 1.20521629e+00 -6.47496164e-01 2.68075794e-01
5.17066538e-01 -1.14369750e-01 8.89505804e-01 -1.07835531e+00
3.63497972e-01 1.31170607e+00 6.25842810e-01 8.59213591e-01
-1.39957154e+00 -1.19372630e+00 2.27700219e-01 -1.33088499e-01
-1.58699596e+00 -8.48753154e-01 8.46051455e-01 -8.39656115e-01
1.14930367e+00 1.47154152e-01 2.04005837e-01 1.34620595e+00
-3.15989763e-01 1.02505219e+00 1.00425684e+00 -7.93846846e-01
-1.49017453e-01 2.93978095e-01 1.09881155e-01 6.31216288e-01
-2.67849773e-01 -1.34444296e-01 -4.97735262e-01 8.94495398e-02
2.86857903e-01 9.81388390e-02 -2.21792340e-01 -1.04524363e-02
-1.24369037e+00 8.65124941e-01 -2.99428590e-02 7.01001048e-01
3.01790833e-01 -2.80826688e-01 6.02773130e-01 6.63899839e-01
5.49831510e-01 1.64968088e-01 -5.06420493e-01 -4.26497608e-01
-9.08353984e-01 -1.01000182e-01 4.86087888e-01 1.37507617e+00
5.85666597e-01 1.72099203e-01 1.45732582e-01 1.86657548e+00
2.20958769e-01 7.00466573e-01 9.62707996e-01 -3.77542078e-01
8.82011235e-01 5.43118834e-01 -6.76143765e-01 -4.20557618e-01
-2.33529285e-02 -4.58953798e-01 -6.41962290e-01 -2.83286840e-01
2.75669992e-01 -8.61074850e-02 -8.29393566e-01 1.94475448e+00
7.52005652e-02 -1.03344239e-01 1.02097437e-01 3.75690192e-01
7.43307054e-01 6.34778023e-01 -6.26515746e-02 -4.55331877e-02
1.28307760e+00 -1.34793663e+00 -6.03542924e-01 -3.49042892e-01
9.07139421e-01 -1.06716037e+00 1.35166025e+00 2.48159364e-01
-8.42512786e-01 -5.21063983e-01 -8.45905483e-01 -1.17560506e-01
-3.29775453e-01 3.67937326e-01 6.00953043e-01 7.91682005e-01
-1.07403123e+00 -1.78392082e-01 -6.49130166e-01 -4.46697265e-01
1.16063058e-01 2.49472871e-01 -5.71167469e-01 -3.03064287e-01
-9.64872181e-01 6.44252360e-01 2.30088860e-01 -3.38548273e-01
-8.56570482e-01 -5.69793701e-01 -1.06884563e+00 -7.95186162e-02
3.22364837e-01 2.47631595e-01 1.42842400e+00 -1.22330844e+00
-1.47397196e+00 1.33898294e+00 -2.29147762e-01 -1.71974540e-01
2.31122479e-01 8.07731599e-02 -8.37948859e-01 -4.39712137e-01
3.29046667e-01 3.04134160e-01 7.12640107e-01 -1.04358101e+00
-6.74526393e-01 -1.99372590e-01 -1.36611521e-01 1.27278385e-03
-3.52997065e-01 6.24343038e-01 -5.83509445e-01 -9.55760717e-01
1.75518632e-01 -1.22106588e+00 1.35303631e-01 -7.36772239e-01
-3.54768246e-01 -3.21321636e-01 4.10451800e-01 -9.04029489e-01
1.56758249e+00 -2.53263998e+00 8.64896923e-02 -2.10927296e-02
-2.74038035e-03 5.24128854e-01 -2.58655667e-01 3.77551883e-01
-2.59159297e-01 -4.56113182e-03 -5.36524534e-01 -8.10049713e-01
-3.03106457e-01 2.98885088e-02 -1.84057489e-01 1.51521116e-01
3.78768235e-01 4.90600646e-01 -8.25714052e-01 -2.94015616e-01
-2.79051624e-02 3.76374632e-01 -7.71139443e-01 3.94362360e-01
1.06589116e-01 6.13611758e-01 1.16050616e-01 1.06676292e+00
5.38383722e-01 1.91261619e-01 1.75560549e-01 6.66845739e-01
-4.26878452e-01 6.41066730e-01 -8.15592885e-01 1.91953397e+00
-1.10646641e+00 7.34340370e-01 2.99293131e-01 -8.59627664e-01
8.30434322e-01 2.85854161e-01 -3.96771263e-03 -6.15222096e-01
1.84016868e-01 7.96061575e-01 4.69455808e-01 -3.69882703e-01
1.72599241e-01 -2.31635664e-02 -2.63674587e-01 3.43895048e-01
2.85595894e-01 -8.92223194e-02 1.26437292e-01 1.40525281e-01
1.23496938e+00 -4.35087204e-01 2.81486243e-01 -5.27629972e-01
9.49416161e-01 -3.12962383e-01 6.16138816e-01 6.38015151e-01
-1.90241590e-01 7.57111311e-01 2.30227187e-01 -2.52114922e-01
-7.47097194e-01 -9.69386756e-01 -2.44053483e-01 1.61705112e+00
-4.43122238e-01 -3.26021165e-01 -8.50529373e-01 -6.19565368e-01
-4.94084060e-02 5.78334808e-01 -2.21335754e-01 -6.94021359e-02
-8.23251009e-01 -3.83925915e-01 8.40243936e-01 1.79735348e-01
2.25826621e-01 -7.72715926e-01 3.21621925e-01 3.80329221e-01
-3.42917114e-01 -1.30268681e+00 -1.12812912e+00 3.26633275e-01
-2.80447721e-01 -9.32411611e-01 -7.16184199e-01 -1.42803931e+00
5.48828900e-01 4.89758909e-01 1.06418884e+00 3.87284547e-01
7.40994513e-02 -1.72621131e-01 -6.07514739e-01 -1.63174972e-01
-1.08660221e+00 4.95069802e-01 2.39812657e-01 3.01365316e-01
6.34609044e-01 -5.08166432e-01 6.35297820e-02 1.04854867e-01
-6.97249949e-01 5.44192120e-02 4.23604786e-01 8.51816595e-01
-4.73524770e-03 -4.48148161e-01 3.85980636e-01 -8.86143804e-01
4.95338142e-01 -6.76398098e-01 -7.44765222e-01 3.98852468e-01
-2.68988967e-01 2.77871918e-02 7.66616404e-01 -8.71165276e-01
-8.91666234e-01 3.50027412e-01 -4.32736546e-01 -3.92705381e-01
-9.02571753e-02 2.85155326e-01 -3.37770909e-01 1.65372633e-03
3.96860063e-01 4.51448768e-01 -1.27878621e-01 -7.72817194e-01
7.90209174e-02 1.44396448e+00 6.59238160e-01 -7.30769157e-01
6.23785019e-01 -5.72358482e-02 -8.94175887e-01 -1.05381858e+00
-5.16223967e-01 -6.77609861e-01 -7.42017627e-01 1.78083759e-02
8.46063912e-01 -1.43435836e+00 -5.14894187e-01 7.71199405e-01
-1.29762733e+00 -3.95300210e-01 3.94579738e-01 5.01511693e-01
-1.49650574e-02 1.88459024e-01 -6.16803885e-01 -8.23008299e-01
-1.20682947e-01 -1.75892675e+00 1.04213047e+00 -3.70788634e-01
-9.56858397e-02 -8.28024805e-01 2.20116094e-01 6.23075366e-01
6.02042735e-01 -3.55456918e-01 9.64727402e-01 -1.02503228e+00
-3.51660460e-01 7.02610537e-02 -4.18310463e-02 6.18321538e-01
3.23663354e-01 1.35294512e-01 -1.03674150e+00 -5.47627866e-01
-5.70270494e-02 -4.71017718e-01 6.51529014e-01 -8.67044851e-02
9.90958273e-01 -1.80716179e-02 5.88641837e-02 7.41574824e-01
1.02339077e+00 5.78692555e-01 -8.12626556e-02 -1.00696258e-01
1.08493268e+00 6.96010768e-01 -6.84430962e-03 2.67133527e-02
7.83029735e-01 8.11977625e-01 -1.96247816e-01 1.97789371e-01
-3.24247152e-01 -3.89988661e-01 6.80803120e-01 1.99774265e+00
3.59741151e-01 1.69555873e-01 -1.58988869e+00 8.52075458e-01
-1.47868240e+00 -6.84177339e-01 -1.61842182e-01 2.26999736e+00
1.25981462e+00 -7.91377798e-02 1.53839504e-02 2.56551690e-02
1.19640481e+00 -2.65937466e-02 -3.13868761e-01 -5.35457850e-01
-1.79208159e-01 2.48547569e-01 3.37637782e-01 7.42361784e-01
-1.05647779e+00 1.33655739e+00 5.32522058e+00 1.12342668e+00
-1.22921145e+00 7.12184668e-01 4.34834957e-01 -1.49789592e-02
-3.12755853e-01 -1.60460547e-02 -8.53719532e-01 8.98161113e-01
1.01135361e+00 1.93884149e-02 9.83318448e-01 9.27177906e-01
-3.38557690e-01 1.36958882e-01 -1.25478923e+00 1.46779323e+00
2.86616027e-01 -8.20290208e-01 -2.24056378e-01 -9.39412788e-02
8.25972199e-01 4.70969766e-01 -9.73815024e-02 7.32035935e-01
6.91699266e-01 -7.77848780e-01 9.92128074e-01 -3.70005637e-01
1.25974929e+00 -6.64727926e-01 3.31350476e-01 4.43239063e-01
-1.37325287e+00 -2.01061726e-01 7.15926439e-02 2.06100736e-02
-1.96507454e-01 6.13942206e-01 -5.62386155e-01 -4.15010452e-02
5.92481017e-01 5.95782161e-01 -6.71215296e-01 5.69236279e-01
-1.95051834e-01 8.58851314e-01 -1.70848712e-01 7.28133917e-02
1.02211379e-01 4.64610122e-02 3.64472300e-01 1.48772442e+00
3.98505718e-01 -3.80573750e-01 2.88533807e-01 6.64139688e-01
-4.17305201e-01 3.03356498e-01 -8.90962899e-01 -1.48682281e-01
8.43169332e-01 9.69589233e-01 -4.81707096e-01 -1.87184364e-01
-8.21495116e-01 1.02727950e+00 7.79615283e-01 3.48137431e-02
-5.52864134e-01 -3.90994757e-01 9.79275644e-01 -3.46744150e-01
-8.51872843e-03 -3.40483159e-01 6.19095415e-02 -1.64504504e+00
1.12333834e-01 -1.31221890e+00 8.63923728e-02 -1.79210529e-01
-1.18960488e+00 9.23011959e-01 -4.50206578e-01 -1.19754446e+00
-3.55431855e-01 -4.31128323e-01 -5.10310531e-01 8.28052759e-01
-1.39275277e+00 -1.25807762e+00 7.24699497e-02 7.41552234e-01
1.11884499e+00 -8.34243774e-01 6.93880141e-01 9.26035225e-01
-8.13467205e-01 1.11558425e+00 2.19700068e-01 6.49315953e-01
7.64901698e-01 -1.13645732e+00 7.35464096e-01 1.18906319e+00
3.52651685e-01 8.81920934e-01 2.53891587e-01 -5.35176754e-01
-1.40404522e+00 -1.00027955e+00 1.25698256e+00 -5.63948214e-01
1.00204206e+00 -1.25774145e+00 -9.32176530e-01 7.02562392e-01
2.13191882e-01 -7.39670321e-02 8.26319814e-01 2.84716725e-01
-6.39536321e-01 -4.95477952e-02 -7.20147133e-01 6.12718403e-01
1.18850720e+00 -1.26782179e+00 -4.61878479e-01 2.08780602e-01
1.11302102e+00 -3.39378089e-01 -5.68085313e-01 9.30365473e-02
2.72370249e-01 -7.74148703e-01 4.86845940e-01 -3.40243340e-01
3.39711189e-01 -1.09288722e-01 -4.17179614e-01 -1.45153809e+00
-8.10056403e-02 -7.79244304e-01 3.94715816e-01 1.77855599e+00
5.22910178e-01 -5.91558754e-01 2.33384431e-03 2.61235237e-01
-3.77371550e-01 -2.98224270e-01 -1.16162479e+00 -9.82236564e-01
3.48257750e-01 -6.94018185e-01 7.96783090e-01 1.50023568e+00
1.52179301e-01 3.98397595e-01 -2.18618691e-01 1.33346736e-01
3.35296154e-01 -1.29302084e-01 6.84472382e-01 -9.85215306e-01
-3.06730539e-01 -5.55170596e-01 -1.30951703e-01 -8.56388628e-01
8.13174903e-01 -1.50631011e+00 2.29142085e-01 -4.75587755e-01
2.07681909e-01 -6.79993212e-01 -1.56598419e-01 5.68205595e-01
-1.08254150e-01 -3.53274494e-02 2.62855619e-01 2.15275854e-01
-3.40131044e-01 4.30633724e-01 6.00950062e-01 -3.24126899e-01
-1.90023616e-01 1.21912256e-01 -6.72192574e-01 5.13374090e-01
6.69354916e-01 -6.38313830e-01 -2.78818995e-01 -8.81565750e-01
-2.45029237e-02 1.88232720e-01 -3.99220139e-01 -1.06399131e+00
2.60866821e-01 1.95800483e-01 -5.33497751e-01 -2.50043333e-01
-1.25773132e-01 -8.29565883e-01 -1.19884722e-01 4.69433486e-01
-4.07887131e-01 3.18167359e-01 1.79004446e-02 -1.31795391e-01
-4.32927102e-01 -2.79417604e-01 9.22680140e-01 -7.19702542e-02
-4.28799719e-01 2.17506483e-01 -6.65562332e-01 4.54526484e-01
7.81699955e-01 1.54771775e-01 -2.03730658e-01 -7.90512264e-02
-3.10178041e-01 2.22173303e-01 4.87784624e-01 1.11843824e+00
2.54552901e-01 -1.41228521e+00 -7.58999765e-01 9.44787204e-01
6.08862042e-01 -1.63411081e-01 -6.68334737e-02 6.65591419e-01
-4.68036145e-01 3.50661308e-01 1.60555497e-01 -6.85822606e-01
-1.31876612e+00 4.70751822e-01 2.05950424e-01 1.01253614e-01
-8.43603238e-02 1.13635910e+00 1.44853428e-01 -1.17854130e+00
4.37573612e-01 -4.02775675e-01 3.14302780e-02 6.58760965e-02
4.95843530e-01 7.73086697e-02 3.37983787e-01 -1.13894904e+00
-5.64427912e-01 3.95805180e-01 -3.66180360e-01 -1.56661179e-02
1.08852482e+00 -2.10662946e-01 -3.66439223e-01 7.35420406e-01
1.52401471e+00 5.78466535e-01 -7.53192067e-01 -4.61464375e-01
2.75839329e-01 -4.83708650e-01 -1.40227154e-01 -4.16761875e-01
-1.12960327e+00 9.51684594e-01 3.96245927e-01 1.05969936e-01
7.41654873e-01 3.20080340e-01 8.52793396e-01 1.89686075e-01
5.76349258e-01 -1.24842238e+00 -2.88988858e-01 1.03968561e+00
7.07316220e-01 -1.68923378e+00 -8.42976213e-01 -2.67785698e-01
-6.13193631e-01 7.23542988e-01 6.37714565e-01 3.02419811e-01
7.69421101e-01 4.15593177e-01 4.41297293e-01 2.87378073e-01
-6.67644203e-01 -3.80711526e-01 1.33738682e-01 3.47881258e-01
9.35165584e-01 2.97590762e-01 -1.31598055e-01 5.15273392e-01
-5.00121713e-01 -5.46750605e-01 4.34559971e-01 8.32666993e-01
-7.79414847e-02 -1.39590645e+00 -3.21465969e-01 2.84259588e-01
-6.54424548e-01 -5.64260840e-01 -2.55583853e-01 4.15013522e-01
2.28326902e-01 1.20066881e+00 1.43999875e-01 -6.98810279e-01
1.76001221e-01 3.68515640e-01 4.38775234e-02 -1.08257592e+00
-8.87016892e-01 2.20074505e-03 -1.43747598e-01 -3.24398011e-01
-1.70293257e-01 -1.00847876e+00 -1.15556157e+00 -2.28356570e-01
-3.46512109e-01 -4.77878787e-02 8.33097756e-01 8.51843417e-01
1.14154480e-01 4.05968249e-01 9.18708563e-01 -5.06653905e-01
-2.53713280e-01 -9.47239757e-01 -4.27608073e-01 4.42889929e-01
7.88043380e-01 -5.59836149e-01 -5.57576835e-01 -1.16713479e-01] | [14.349710464477539, 6.959644794464111] |
13c79135-8630-4354-b600-33dc8a95465f | multi-view-matrix-completion-for-multi-label | 1904.03901 | null | http://arxiv.org/abs/1904.03901v1 | http://arxiv.org/pdf/1904.03901v1.pdf | Multi-View Matrix Completion for Multi-Label Image Classification | There is growing interest in multi-label image classification due to its
critical role in web-based image analytics-based applications, such as
large-scale image retrieval and browsing. Matrix completion has recently been
introduced as a method for transductive (semi-supervised) multi-label
classification, and has several distinct advantages, including robustness to
missing data and background noise in both feature and label space. However, it
is limited by only considering data represented by a single-view feature, which
cannot precisely characterize images containing several semantic concepts. To
utilize multiple features taken from different views, we have to concatenate
the different features as a long vector. But this concatenation is prone to
over-fitting and often leads to very high time complexity in MC based image
classification. Therefore, we propose to weightedly combine the MC outputs of
different views, and present the multi-view matrix completion (MVMC) framework
for transductive multi-label image classification. To learn the view
combination weights effectively, we apply a cross validation strategy on the
labeled set. In the learning process, we adopt the average precision (AP) loss,
which is particular suitable for multi-label image classification. A least
squares loss formulation is also presented for the sake of efficiency, and the
robustness of the algorithm based on the AP loss compared with the other losses
is investigated. Experimental evaluation on two real world datasets (PASCAL
VOC' 07 and MIR Flickr) demonstrate the effectiveness of MVMC for transductive
(semi-supervised) multi-label image classification, and show that MVMC can
exploit complementary properties of different features and output-consistent
labels for improved multi-label image classification. | ['DaCheng Tao', 'Yong Luo', 'Tongliang Liu', 'Chao Xu'] | 2019-04-08 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 5.20550609e-01 -5.69193721e-01 -2.64707565e-01 -4.89605814e-01
-1.08312464e+00 -4.81153071e-01 4.08237010e-01 2.63847679e-01
-4.49499607e-01 4.58129048e-01 -2.17042223e-01 6.61908388e-02
-3.72005433e-01 -5.06832540e-01 -5.28249621e-01 -1.16348386e+00
4.85920250e-01 1.54975504e-01 -1.57044873e-01 1.17498182e-01
3.01815961e-02 1.79614633e-01 -1.70102811e+00 5.91922820e-01
8.93121243e-01 1.17684960e+00 1.87288582e-01 2.45826855e-01
1.31288329e-02 7.91973054e-01 -1.56220853e-01 -5.97627878e-01
6.22126423e-02 -3.32918882e-01 -6.52615607e-01 6.58772469e-01
4.67866927e-01 3.02508473e-02 2.65052736e-01 1.35271645e+00
3.41006964e-01 -5.07968739e-02 8.19572508e-01 -1.47968197e+00
-3.68504971e-01 4.53310981e-02 -8.00698757e-01 -3.18994612e-01
1.64552346e-01 -3.50768805e-01 1.12609935e+00 -1.18096530e+00
5.58071256e-01 1.26189303e+00 5.15722036e-01 3.03507328e-01
-1.40323067e+00 -6.71297550e-01 2.37053514e-01 1.46944776e-01
-1.42109740e+00 -1.86922163e-01 8.74024987e-01 -6.05024934e-01
9.84254554e-02 5.16656518e-01 3.87634695e-01 8.92462730e-01
2.17574105e-01 1.04401624e+00 1.63175571e+00 -4.48955148e-01
-4.91404347e-02 5.22544444e-01 1.14645429e-01 9.95021224e-01
-1.05525695e-01 -2.40729287e-01 -3.53565753e-01 -4.04906750e-01
4.41228867e-01 4.32168663e-01 -3.15118760e-01 -9.87443149e-01
-1.39709270e+00 1.05396748e+00 4.52615082e-01 1.78732827e-01
-1.73925728e-01 -1.47635043e-01 5.90273559e-01 4.24508393e-01
6.86639130e-01 -1.14611886e-01 -2.73519367e-01 3.24973255e-01
-5.22891164e-01 -9.20020342e-02 3.37495714e-01 7.52126753e-01
9.42077816e-01 -3.95492733e-01 3.32297012e-02 1.41342556e+00
4.56316292e-01 6.59542561e-01 4.37591285e-01 -8.52357984e-01
5.81932366e-01 8.15240085e-01 -1.57614872e-01 -1.13190293e+00
-3.71932834e-01 -3.51358324e-01 -1.19284713e+00 1.80268750e-01
2.96141922e-01 2.88450480e-01 -5.58428764e-01 1.57824910e+00
5.32116354e-01 -4.35985923e-02 8.72083753e-02 9.28311706e-01
6.87673748e-01 5.41388512e-01 1.58770636e-01 -6.66258276e-01
1.40354741e+00 -9.27167833e-01 -5.96193552e-01 2.10374650e-02
7.77404249e-01 -7.71991551e-01 9.38916922e-01 4.35732096e-01
-6.28332078e-01 -4.72612321e-01 -1.01691854e+00 1.73518136e-01
-2.84120262e-01 3.27958405e-01 5.42972267e-01 6.30403280e-01
-5.08930743e-01 3.43599945e-01 -3.51038486e-01 -3.27863127e-01
4.67708796e-01 3.48019719e-01 -8.36747944e-01 -6.60082400e-01
-9.40854073e-01 6.12184286e-01 3.33514661e-01 6.14868030e-02
-4.79585230e-01 -4.33562756e-01 -8.61149251e-01 -1.93926856e-01
4.93748695e-01 -4.82569456e-01 7.47356594e-01 -1.18941283e+00
-1.20261133e+00 1.05202186e+00 7.15377778e-02 1.85686797e-01
4.72547233e-01 1.64553687e-01 -5.79994321e-01 3.57879728e-01
1.93647102e-01 4.81148273e-01 1.14271593e+00 -1.57896566e+00
-8.48128259e-01 -6.95056617e-01 1.14953056e-01 3.75150383e-01
-5.56531429e-01 -1.67034610e-04 -3.84806514e-01 -5.02142906e-01
5.93335629e-01 -1.23021901e+00 -6.13297075e-02 -3.34294289e-02
-2.14329854e-01 -1.77748784e-01 6.10864878e-01 -3.94015044e-01
8.84463966e-01 -2.23020697e+00 3.39349747e-01 2.55736083e-01
2.45086625e-01 2.11687699e-01 -1.60299733e-01 2.75676966e-01
-1.34127200e-01 -1.46969259e-01 -2.84438759e-01 -4.37014312e-01
-2.49701738e-01 1.58755198e-01 8.20224211e-02 7.53770947e-01
3.92647237e-02 7.00221837e-01 -9.23053980e-01 -8.05608034e-01
4.41996962e-01 4.20495778e-01 -1.77329302e-01 8.93596560e-03
9.46291834e-02 4.86120433e-01 -5.18750012e-01 8.45286310e-01
7.08480716e-01 -5.29024601e-01 2.57521898e-01 -6.24540091e-01
2.50512838e-01 -6.01498961e-01 -1.32122242e+00 1.70920599e+00
-6.03610873e-01 1.76651112e-03 3.63701358e-02 -1.15172839e+00
6.61441803e-01 4.18111265e-01 7.88983166e-01 -4.81157392e-01
1.67266592e-01 3.78171712e-01 -5.50318956e-01 -4.73846614e-01
2.59478744e-02 -2.94034809e-01 -6.97625577e-02 4.61587250e-01
1.02315441e-01 1.68708622e-01 1.33550406e-01 9.25716087e-02
5.80009997e-01 -8.20300207e-02 4.53349084e-01 -1.49363369e-01
9.56683636e-01 -2.07014769e-01 7.53513515e-01 2.35483244e-01
-8.78279284e-02 5.66988766e-01 3.84905666e-01 -2.97638476e-01
-6.78780735e-01 -8.32524359e-01 -3.26380908e-01 1.10330713e+00
4.95942771e-01 -2.61816770e-01 -4.59793180e-01 -1.14283752e+00
-2.30565537e-02 1.94198176e-01 -5.30071080e-01 -5.28470166e-02
-2.36008734e-01 -1.10699701e+00 8.76548737e-02 1.50786728e-01
4.79805678e-01 -6.52985990e-01 -1.33542776e-01 1.48479208e-01
-5.60578406e-01 -1.09985483e+00 -4.74521518e-01 1.92148730e-01
-9.03452933e-01 -1.28912723e+00 -7.77071059e-01 -7.85761595e-01
8.92941773e-01 7.69407153e-01 6.95856214e-01 -9.64697823e-03
-1.78704575e-01 6.61750853e-01 -4.75709409e-01 1.52444109e-01
-5.64938188e-01 -2.51414448e-01 5.31149879e-02 8.85348141e-01
7.44629651e-02 -3.34329695e-01 -3.89078975e-01 6.52601242e-01
-1.18897617e+00 2.34804139e-01 6.93018019e-01 1.32480073e+00
8.99147630e-01 7.14876950e-02 6.54791832e-01 -1.27233040e+00
2.66609639e-01 -3.39624137e-01 -4.37454581e-01 7.53601611e-01
-8.91450703e-01 -1.69795454e-01 5.29094636e-01 -4.93398130e-01
-8.98592472e-01 2.77285635e-01 2.32613742e-01 -5.70823789e-01
6.89122081e-02 5.54456592e-01 -4.58393872e-01 -3.95734191e-01
2.75383353e-01 2.49382198e-01 3.61815572e-01 -2.90090859e-01
4.57190216e-01 7.08542943e-01 6.07187450e-02 -3.52942854e-01
6.18838966e-01 5.68714142e-01 2.74874985e-01 -6.26094759e-01
-1.04018724e+00 -9.85196173e-01 -5.81116378e-01 -4.61728185e-01
8.71705413e-01 -1.06412923e+00 -7.77409971e-01 4.69614714e-01
-8.02726686e-01 4.65173900e-01 1.51615471e-01 5.94248533e-01
-5.93343675e-01 6.27484918e-01 -6.21144593e-01 -5.63493133e-01
-2.88931727e-01 -1.49455440e+00 1.06913030e+00 7.84997642e-03
3.66063952e-01 -1.01882386e+00 -2.05622911e-01 8.99807453e-01
-1.03020065e-01 1.86386004e-01 1.12104654e+00 -4.27479446e-01
-4.73921537e-01 -4.27470595e-01 -4.28953230e-01 7.66527593e-01
1.05014943e-01 -3.26011956e-01 -1.03954566e+00 -5.96832097e-01
9.55115333e-02 -7.33955085e-01 8.83428752e-01 1.13854460e-01
1.03214550e+00 -7.00321868e-02 -3.52942318e-01 3.45748752e-01
1.77978182e+00 -1.41656682e-01 3.45477521e-01 8.86414871e-02
1.02912641e+00 7.91305244e-01 1.07680297e+00 2.56142616e-01
3.82616848e-01 6.73557460e-01 5.35390139e-01 -1.16085552e-01
2.78332651e-01 4.52390090e-02 2.48546019e-01 1.18320191e+00
7.44783878e-02 -1.77907258e-01 -3.87043864e-01 6.89141601e-02
-1.99973631e+00 -7.63939202e-01 -1.88768357e-01 2.39562130e+00
5.52502930e-01 -2.53056079e-01 4.72339336e-03 3.04563940e-01
8.69057298e-01 1.41446576e-01 -4.83743519e-01 -8.41001421e-03
-2.03923777e-01 -3.47820163e-01 4.45561588e-01 1.90235138e-01
-1.51101172e+00 4.69873220e-01 4.59001303e+00 1.36002862e+00
-9.52323020e-01 3.63937169e-01 7.64338672e-01 2.00438201e-01
-2.05580279e-01 -1.46490455e-01 -6.82762921e-01 2.65650541e-01
3.99554998e-01 2.77626216e-01 3.67612720e-01 6.76862121e-01
-1.20757848e-01 -1.79128826e-01 -1.01004481e+00 1.40534198e+00
3.90974194e-01 -9.65205312e-01 2.13158950e-01 1.77256197e-01
7.92764843e-01 -3.34919840e-01 2.36421719e-01 1.64036870e-01
-2.04935402e-01 -5.50378323e-01 5.17121673e-01 3.07514906e-01
8.86552930e-01 -8.29538405e-01 7.40546227e-01 4.90858644e-01
-1.21694875e+00 -2.39445537e-01 -3.62940282e-01 4.77654755e-01
-3.47068496e-02 8.15074742e-01 -4.23695594e-01 9.27947998e-01
3.42521936e-01 9.13758457e-01 -7.02349305e-01 7.49759674e-01
2.80403465e-01 1.25405192e-01 4.23408151e-02 2.04425573e-01
8.53549689e-02 -6.46267116e-01 3.41772795e-01 7.92533278e-01
1.55616894e-01 -1.14348553e-01 5.81436276e-01 3.09723318e-01
-5.37672043e-02 7.19451785e-01 -5.42842984e-01 1.79543182e-01
1.73345506e-01 1.70536482e+00 -8.34472299e-01 -1.85125783e-01
-7.25218594e-01 1.13286841e+00 3.41285378e-01 2.56625265e-01
-7.65733182e-01 1.08511999e-01 3.24866176e-01 -3.00583273e-01
1.10962771e-01 1.66885033e-01 -4.24858220e-02 -1.34645057e+00
7.50403628e-02 -9.49651539e-01 5.14280200e-01 -4.80405867e-01
-1.38490820e+00 3.68793786e-01 -1.54649511e-01 -1.88660944e+00
9.92761925e-02 -4.51559782e-01 1.61110327e-01 5.58817744e-01
-1.52919328e+00 -1.71882582e+00 -2.10627601e-01 6.50060177e-01
4.20720041e-01 -1.30009234e-01 9.42806721e-01 6.95801079e-01
-6.66890025e-01 6.78548992e-01 5.01089931e-01 -1.08687311e-01
9.58921134e-01 -1.17707980e+00 -6.32498026e-01 3.59579802e-01
3.63416970e-01 2.84656256e-01 1.82170704e-01 -2.76862681e-01
-1.40658224e+00 -1.31644928e+00 6.23435676e-01 -2.01370984e-01
4.16093618e-01 -1.52024150e-01 -7.14704692e-01 3.35763633e-01
-2.21950278e-01 3.74555200e-01 1.08636642e+00 -3.38940248e-02
-5.26583970e-01 -4.86078441e-01 -1.15557826e+00 3.90463114e-01
6.11576736e-01 -5.82460523e-01 2.23277751e-02 6.38626099e-01
6.00016475e-01 5.08520193e-02 -1.27309656e+00 6.23256087e-01
6.68153942e-01 -1.01126206e+00 1.08576369e+00 -2.43039131e-01
2.92024612e-01 -3.74262720e-01 -4.84647334e-01 -1.33218944e+00
-2.95793831e-01 1.51062563e-01 3.15256804e-01 1.47339344e+00
2.51312107e-01 -7.05489278e-01 5.41338623e-01 3.35424989e-01
2.33825639e-01 -8.71454298e-01 -8.12593043e-01 -6.97751284e-01
-2.30569810e-01 -2.54523844e-01 1.56622723e-01 1.09444511e+00
-1.81286454e-01 5.23373783e-01 -5.63306510e-01 1.79139376e-01
1.04572582e+00 5.09850621e-01 5.25894284e-01 -1.39167464e+00
-2.30823725e-01 -1.06110938e-01 -4.98032391e-01 -6.28617883e-01
2.26162508e-01 -1.09540629e+00 -7.46288821e-02 -1.16515970e+00
5.81080735e-01 -7.42297530e-01 -6.32249475e-01 1.37936383e-01
-3.16138685e-01 6.06422365e-01 3.12759310e-01 5.13992429e-01
-9.05487239e-01 6.00603044e-01 1.43237615e+00 -4.15029168e-01
1.70314074e-01 1.70512527e-01 -4.16494191e-01 7.04467773e-01
4.25976574e-01 -5.66523910e-01 -4.31760162e-01 -5.65702543e-02
1.70012936e-01 3.62256885e-01 3.41813624e-01 -7.96204269e-01
1.11226412e-02 -1.58606395e-01 2.35385299e-01 -3.94675791e-01
5.98305047e-01 -1.13919163e+00 2.88460314e-01 3.97594839e-01
-4.56537604e-01 -2.90388316e-01 -4.12200063e-01 1.07927954e+00
-4.81781602e-01 -3.06285322e-01 9.85399008e-01 -1.16656549e-01
-6.64229035e-01 4.00147498e-01 -8.41121897e-02 -2.43274659e-01
1.21304548e+00 -7.43569806e-02 -9.70436186e-02 -4.79190461e-02
-6.23270154e-01 2.36252889e-01 4.06083435e-01 3.92786443e-01
6.92410350e-01 -1.77486169e+00 -6.13919735e-01 2.35411972e-01
7.79301941e-01 -3.70587885e-01 4.99078095e-01 1.16722274e+00
-2.74600834e-01 1.68127909e-01 1.80616453e-02 -9.43376601e-01
-1.69412065e+00 8.15430999e-01 -3.56727615e-02 -4.86589462e-01
-3.27723414e-01 6.22876167e-01 4.01455462e-01 -6.15066051e-01
3.17006186e-02 1.48601010e-02 -5.04132569e-01 4.47786361e-01
4.54921007e-01 3.57431203e-01 1.33041248e-01 -8.71782124e-01
-1.83178216e-01 9.48974431e-01 -1.99030414e-01 8.30646381e-02
1.02655125e+00 -5.01103461e-01 -4.16296661e-01 8.98120105e-01
1.75201309e+00 -3.27652603e-01 -8.56747627e-01 -6.65194690e-01
-6.54404387e-02 -5.41524053e-01 1.59189284e-01 -4.61629599e-01
-1.21687567e+00 7.92607903e-01 8.94349694e-01 5.82901351e-02
1.19025159e+00 -3.22655141e-01 6.57328963e-01 2.98602283e-01
6.05727851e-01 -1.09727836e+00 2.37191036e-01 -1.89879034e-02
6.36424243e-01 -1.72694910e+00 1.24977089e-01 -7.95653999e-01
-8.30562711e-01 1.03069866e+00 3.30847383e-01 2.67530859e-01
7.13918030e-01 -4.14773464e-01 2.03059852e-01 -1.77698269e-01
-5.31781077e-01 -7.19287619e-02 3.04936171e-01 2.50120491e-01
1.23111121e-01 2.49296471e-01 -3.57569993e-01 1.52678892e-01
7.37322092e-01 -1.91328123e-01 2.11144462e-01 8.48270118e-01
-3.15409094e-01 -1.21959722e+00 -5.31261027e-01 6.73002839e-01
-4.38136578e-01 1.83826819e-01 1.57326609e-01 4.63520855e-01
1.61230460e-01 1.15225136e+00 -5.00244498e-01 -5.05738318e-01
1.76002890e-01 1.17299840e-01 5.29866397e-01 -4.58712578e-01
-3.95667374e-01 3.96656066e-01 -1.31841689e-01 -5.07211149e-01
-9.70928967e-01 -7.74557710e-01 -8.22185993e-01 -1.27030939e-01
-7.39190102e-01 6.80155084e-02 7.66977370e-01 9.01901126e-01
4.54430990e-02 2.53795058e-01 1.17013812e+00 -5.77400327e-01
-7.59792805e-01 -7.26789713e-01 -9.98973131e-01 8.81661773e-01
8.33253935e-02 -8.63961577e-01 -3.96495938e-01 9.37341079e-02] | [8.681797981262207, 4.427880764007568] |
ca0e968b-005b-413a-8761-32fc273b3492 | building-on-huang-et-al-glossbert-for-word | 2112.07089 | null | https://arxiv.org/abs/2112.07089v1 | https://arxiv.org/pdf/2112.07089v1.pdf | Building on Huang et al. GlossBERT for Word Sense Disambiguation | We propose to take on the problem ofWord Sense Disambiguation (WSD). In language, words of the same form can take different meanings depending on context. While humans easily infer the meaning or gloss of such words by their context, machines stumble on this task.As such, we intend to replicated and expand upon the results of Huang et al.GlossBERT, a model which they design to disambiguate these words (Huang et al.,2019). Specifically, we propose the following augmentations: data-set tweaking(alpha hyper-parameter), ensemble methods, and replacement of BERT with BART andALBERT. The following GitHub repository contains all code used in this report, which extends on the code made available by Huang et al. | ['Yichun Yu', 'Apoorva Sharma', 'Kanika Jindal', 'James Hale', 'Nikhil Patel'] | 2021-12-14 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 2.83220112e-01 8.57532993e-02 -1.90984443e-01 -2.72449464e-01
-4.69307393e-01 -9.16912615e-01 9.93142605e-01 5.29786706e-01
-7.58745015e-01 7.86591709e-01 5.94599366e-01 -4.90859747e-01
-2.48149052e-01 -6.31960154e-01 -9.44264680e-02 -4.03910249e-01
3.37209761e-01 4.21198249e-01 6.73315674e-02 -5.81586897e-01
5.26992142e-01 2.01655313e-01 -1.60848153e+00 1.73685417e-01
9.18485641e-01 7.44105875e-01 4.04870152e-01 2.47552097e-01
-3.35223317e-01 2.37060040e-01 -7.53944755e-01 -2.75621980e-01
1.73098091e-02 -1.72268838e-01 -8.95856559e-01 -2.03764156e-01
1.71333700e-01 2.60105699e-01 3.83556411e-02 9.82786536e-01
5.34599483e-01 4.22588736e-01 6.69498503e-01 -9.89168942e-01
-8.26414287e-01 9.81765568e-01 -2.04765230e-01 3.23069066e-01
5.34322858e-01 -6.53339624e-02 1.30656636e+00 -9.01709616e-01
5.78771293e-01 1.05293202e+00 6.09852433e-01 5.39556205e-01
-1.20496321e+00 -7.10242569e-01 3.82568836e-01 2.11920723e-01
-1.49278128e+00 -6.75845385e-01 5.14513671e-01 -4.18253064e-01
1.22414970e+00 4.15184557e-01 6.05726719e-01 1.38774729e+00
-1.87720403e-01 5.26858985e-01 1.21114278e+00 -8.06375921e-01
3.08637172e-01 -3.95513996e-02 4.50435787e-01 1.40840501e-01
7.20141709e-01 -6.22512549e-02 -4.88526732e-01 -3.27788740e-01
2.57059783e-01 -1.25043526e-01 -2.79025853e-01 3.39133590e-02
-1.04728448e+00 6.54335916e-01 3.59771885e-02 8.40795577e-01
-3.26172650e-01 -1.31928012e-01 2.91127384e-01 2.92062074e-01
6.19050205e-01 9.85145450e-01 -6.49789393e-01 -2.79068828e-01
-6.24659300e-01 4.70318377e-01 9.27732348e-01 8.20162594e-01
6.66895032e-01 -3.60858500e-01 -1.51670024e-01 9.87820208e-01
2.61561930e-01 4.40712541e-01 8.52293491e-01 -7.13821054e-01
7.98590854e-02 5.80967128e-01 3.21446419e-01 -9.62771833e-01
-4.31118846e-01 -6.52941242e-02 -4.95705754e-01 -3.23792726e-01
1.18655510e-01 -2.26588413e-01 -9.72285390e-01 1.73179126e+00
2.27685422e-01 2.48289183e-01 3.71555798e-02 8.39748144e-01
9.16336834e-01 1.14024051e-01 4.73003447e-01 -2.32842892e-01
1.61608076e+00 -3.76029640e-01 -9.27571952e-01 -3.46503228e-01
7.10448623e-01 -9.94095325e-01 1.12117875e+00 3.16512883e-01
-6.73513591e-01 -3.16508770e-01 -9.75817144e-01 -1.26835957e-01
-7.65718877e-01 -2.22035244e-01 8.34155798e-01 6.50482833e-01
-1.00255811e+00 4.53824103e-01 -6.31717801e-01 -9.35229123e-01
8.33007414e-03 -5.90868220e-02 -1.00615054e-01 1.89162672e-01
-1.64750922e+00 1.27526283e+00 6.74500644e-01 -1.23025082e-01
-1.95655376e-02 -5.29205382e-01 -7.56560445e-01 -2.38709107e-01
8.06207418e-01 -7.52091825e-01 1.29275763e+00 -8.66764784e-01
-1.17234373e+00 1.22399271e+00 -2.37484485e-01 -3.99524927e-01
-8.90688598e-02 -3.93883973e-01 -7.74577796e-01 -1.64468795e-01
3.30379874e-01 2.55885124e-01 4.80552107e-01 -1.18525326e+00
-5.71774662e-01 -3.62693101e-01 1.75411060e-01 1.93776667e-01
-1.58757612e-01 1.25014320e-01 -1.26406506e-01 -1.06401527e+00
2.59416737e-02 -9.87318814e-01 -5.15155196e-02 -6.30161762e-01
-4.25188392e-01 -5.38824379e-01 2.67829657e-01 -3.90810132e-01
1.72136962e+00 -2.11059880e+00 1.27429543e-02 1.04219221e-01
2.97682524e-01 4.49261665e-01 -1.67538360e-01 7.30342329e-01
-2.24332199e-01 6.91751242e-01 -4.54603791e-01 -3.75998646e-01
2.70259768e-01 4.82105017e-01 -4.26790446e-01 1.32123679e-01
-7.77496174e-02 7.01505661e-01 -1.10251975e+00 -2.44872227e-01
1.76182315e-01 2.47399047e-01 -2.46529251e-01 -1.76825926e-01
-3.29615921e-02 9.06538963e-03 -5.40026248e-01 4.55475926e-01
3.14899057e-01 -1.69438526e-01 3.30792964e-01 -1.52716100e-01
-3.17135185e-01 7.39677727e-01 -1.27332997e+00 1.59649539e+00
-5.39325476e-01 4.44765359e-01 -1.19610101e-01 -6.80800974e-01
6.68058395e-01 3.30522269e-01 2.99169123e-01 -2.50038445e-01
1.77663952e-01 2.49725103e-01 -1.27794251e-01 -3.63287419e-01
9.53629017e-01 -3.50672513e-01 -3.06250304e-01 5.00410318e-01
-2.17589214e-01 -5.11893153e-01 4.87811804e-01 1.42900631e-01
8.96345615e-01 6.37751147e-02 9.67772484e-01 -4.49065924e-01
3.55932921e-01 7.22035244e-02 2.82317281e-01 8.72879028e-01
-2.35136151e-02 4.14372742e-01 1.39442533e-01 -2.12568939e-01
-6.96408153e-01 -8.34480762e-01 -5.29413462e-01 1.21807647e+00
-6.32201284e-02 -7.91194558e-01 -4.96421486e-01 -4.27236527e-01
1.14761606e-01 1.34509528e+00 -7.64838099e-01 7.22202240e-03
-2.18264371e-01 -6.37890875e-01 6.68835223e-01 1.92485645e-01
3.38630863e-02 -9.84582245e-01 -6.50082052e-01 2.33179957e-01
-2.53102183e-01 -9.54736173e-01 -4.16723222e-01 2.32850686e-01
-3.50989640e-01 -1.07804167e+00 -1.61578923e-01 -4.28767860e-01
1.71569914e-01 3.55865806e-01 1.28830051e+00 2.08772823e-01
-1.15672633e-01 5.74086607e-01 -9.65441823e-01 -7.89508700e-01
-4.11729544e-01 2.97032952e-01 2.00571552e-01 -3.64028543e-01
9.16045845e-01 -5.60369253e-01 -3.87629449e-01 6.31558597e-02
-1.29887486e+00 -1.20783791e-01 1.68544739e-01 5.73696196e-01
5.14027834e-01 -2.80865014e-01 3.98205370e-01 -1.14896464e+00
9.92641389e-01 -6.99909270e-01 -2.40804940e-01 2.87530839e-01
-1.01652956e+00 2.01624259e-01 1.43724948e-01 -2.65656888e-01
-6.78013802e-01 -3.56401801e-01 -1.29461244e-01 -9.06744003e-02
-3.25818270e-01 6.56776905e-01 5.80109842e-02 1.80906028e-01
6.06486619e-01 -5.97531572e-02 -3.75394255e-01 -5.79384923e-01
6.61271513e-01 7.86402643e-01 2.09775895e-01 -7.60702133e-01
4.49231356e-01 1.25791356e-01 -3.44258845e-01 -7.17850626e-01
-9.96410489e-01 -6.22651279e-01 -6.71602845e-01 8.61805975e-02
6.93258762e-01 -7.75842190e-01 -2.09395245e-01 6.06340282e-02
-9.88396943e-01 -4.56680730e-02 -2.25374967e-01 2.73352474e-01
-2.21671194e-01 3.21219742e-01 -9.39264446e-02 -7.13219702e-01
-2.87432879e-01 -6.74147308e-01 9.66265500e-01 -1.99145600e-02
-1.06980300e+00 -1.24816310e+00 1.16950519e-01 5.82627654e-02
5.43134570e-01 3.77013564e-01 9.25666153e-01 -1.18925357e+00
1.23714402e-01 5.92132322e-02 1.85225382e-01 4.08313312e-02
4.88695890e-01 1.90861952e-02 -9.92570698e-01 -1.44172058e-01
-1.79110747e-02 -4.52542417e-02 9.84371066e-01 2.61404693e-01
9.31637526e-01 -4.23564434e-01 -4.20971245e-01 1.48133397e-01
1.32233858e+00 1.19105861e-01 1.51006863e-01 5.79443336e-01
4.63341326e-01 6.50624335e-01 5.46962857e-01 5.78202367e-01
5.42355001e-01 6.21502519e-01 1.74628032e-04 3.07669163e-01
3.87498103e-02 -2.51520127e-01 -1.16107196e-01 7.83971250e-01
-1.77434847e-01 -4.79526997e-01 -1.06973648e+00 6.23228610e-01
-1.68888831e+00 -8.57038260e-01 -5.91509752e-02 2.25340414e+00
9.82407749e-01 1.33046523e-01 -4.76238318e-02 3.60991135e-02
5.93636274e-01 4.79543030e-01 -2.73228973e-01 -2.80387163e-01
-3.29040825e-01 3.64898443e-01 3.23886365e-01 8.88283849e-01
-8.40638340e-01 1.21592176e+00 6.39630365e+00 7.14499772e-01
-7.96622455e-01 6.50367364e-02 4.26687598e-01 -3.19510465e-03
-7.78478026e-01 1.36892512e-01 -5.67662597e-01 4.17282283e-01
6.95832610e-01 -5.59019029e-01 6.50585234e-01 3.78940254e-01
7.76259229e-02 -3.89656007e-01 -9.34315741e-01 8.67233157e-01
1.66292474e-01 -8.38483989e-01 9.26069692e-02 -2.52788216e-01
5.71048081e-01 -4.81202491e-02 -1.76440373e-01 2.77384520e-01
5.32981992e-01 -9.65717435e-01 5.27367890e-01 4.86220449e-01
5.81766903e-01 -4.43026870e-01 6.47029757e-01 2.40785666e-02
-9.31332767e-01 1.29497766e-01 -1.40537649e-01 -2.92251736e-01
3.99555452e-02 8.29240799e-01 -8.49196494e-01 6.12435162e-01
4.93042439e-01 6.78301394e-01 -6.39368355e-01 5.21588624e-01
-5.62638342e-01 6.98122561e-01 -2.30465174e-01 -2.66632140e-01
6.12060120e-03 -1.40765667e-01 7.84110129e-01 1.22551036e+00
3.60487670e-01 4.64067906e-01 2.12867647e-01 5.75017154e-01
2.55457193e-01 1.69505149e-01 -6.15811288e-01 -4.23845410e-01
1.09005547e+00 1.11502826e+00 -6.41859353e-01 -4.80614394e-01
-2.86876112e-01 7.95300901e-01 8.04971606e-02 4.33620751e-01
-2.35320121e-01 -3.83924276e-01 9.57848847e-01 -8.37933123e-02
2.61536669e-02 -1.86373234e-01 -8.00208509e-01 -1.00532901e+00
-9.44222584e-02 -7.60140836e-01 5.17590761e-01 -8.16605568e-01
-1.50186002e+00 5.55572987e-01 3.40671539e-01 -1.01514709e+00
-3.94499034e-01 -6.53314352e-01 -1.06060185e-01 9.48455989e-01
-1.24111128e+00 -7.86502779e-01 -7.00669037e-03 1.44717216e-01
3.82000118e-01 1.11372791e-01 1.12307715e+00 -5.18846139e-02
-2.84408122e-01 2.75113702e-01 -6.33848924e-03 -7.68958852e-02
1.07097304e+00 -1.24687684e+00 5.75558960e-01 6.42865539e-01
3.18702698e-01 1.19407368e+00 1.20556700e+00 -7.10151553e-01
-9.30085003e-01 -6.46524668e-01 1.51773322e+00 -8.33268940e-01
1.07525873e+00 -1.40371099e-01 -8.76210272e-01 7.96079397e-01
4.39788967e-01 -5.26012659e-01 9.46243107e-01 5.15636325e-01
-4.19044673e-01 2.13869408e-01 -1.05696654e+00 7.37614334e-01
1.13493538e+00 -4.95893925e-01 -1.07391703e+00 1.86126560e-01
7.85298884e-01 -3.81160021e-01 -8.40492368e-01 1.83015600e-01
4.31169569e-01 -7.91415036e-01 7.58134186e-01 -7.35516787e-01
1.02327041e-01 -3.62468272e-01 -3.92719597e-01 -1.78928792e+00
-3.42028499e-01 -6.34858429e-01 5.58091290e-02 1.14789104e+00
5.12920320e-01 -7.91730702e-01 -7.77952150e-02 8.38005185e-01
-1.54979736e-01 -5.72484195e-01 -1.00962400e+00 -6.00614786e-01
1.41824484e-01 -8.07187080e-01 1.07465291e+00 1.30767298e+00
4.40425873e-01 4.60485131e-01 1.23865150e-01 6.26286641e-02
1.69062644e-01 8.92090425e-02 3.62569988e-01 -1.39361668e+00
-7.85569400e-02 -7.29923964e-01 -3.52203771e-02 -8.25597227e-01
2.41535261e-01 -8.18527222e-01 4.46064398e-02 -1.45352459e+00
-1.35222688e-01 -4.55616444e-01 -6.47221029e-01 8.69177699e-01
-7.29370296e-01 2.63425782e-02 2.88195699e-01 1.62589297e-01
-5.31780303e-01 2.42467538e-01 9.03282762e-01 9.03247371e-02
-1.21143609e-01 -1.61618099e-01 -1.13176417e+00 5.14177740e-01
9.15840626e-01 -5.75197220e-01 -3.30063552e-01 -4.51440543e-01
4.58256781e-01 -2.30606511e-01 2.90374756e-01 -5.28138876e-01
1.87366251e-02 -5.37893713e-01 3.54419905e-03 -1.49640620e-01
1.40709668e-01 -7.98718512e-01 1.79586843e-01 3.69209200e-01
-5.19716978e-01 3.94545943e-01 3.60923290e-01 3.82955492e-01
-5.25357947e-02 -3.32418174e-01 2.50607789e-01 -1.94267929e-01
-8.03784907e-01 -6.38934374e-02 -5.65697014e-01 3.93058449e-01
6.34545743e-01 -1.24239489e-01 -2.67924666e-01 -4.13168371e-01
-6.89321518e-01 7.94853792e-02 6.30096853e-01 8.27919662e-01
2.26089716e-01 -1.31093144e+00 -7.35301852e-01 -9.94667560e-02
4.40634459e-01 -3.03197831e-01 -4.73282896e-02 6.76335931e-01
4.88292836e-02 4.15351540e-01 1.60459071e-01 1.39144778e-01
-1.01111400e+00 5.03129005e-01 2.42509563e-02 3.00386064e-02
-1.71523660e-01 7.03135371e-01 -1.57604754e-01 -3.45689833e-01
-3.14187050e-01 -5.36088884e-01 -4.41521257e-01 3.99838746e-01
6.82709157e-01 3.09075981e-01 1.09090224e-01 -5.81645370e-01
-6.32478833e-01 2.17957050e-01 1.98768564e-02 -1.43432528e-01
1.16005158e+00 -3.74528825e-01 -2.29806438e-01 1.01890790e+00
7.14444101e-01 3.74055535e-01 -4.15541351e-01 -3.64774287e-01
3.68666321e-01 -2.94186652e-01 3.59174199e-02 -1.21730471e+00
-3.21310401e-01 3.03684354e-01 4.34938073e-01 7.13611960e-01
1.20484674e+00 2.56743673e-02 5.88791013e-01 3.29672515e-01
3.82651746e-01 -1.11476159e+00 -5.66039860e-01 9.93590534e-01
8.77415240e-01 -1.08928084e+00 1.33654565e-01 -2.06472591e-01
-6.31649673e-01 1.03528512e+00 3.09693307e-01 9.93125662e-02
6.35104120e-01 1.44919246e-01 1.63154408e-01 -1.56132832e-01
-8.15169334e-01 -6.50308371e-01 3.60848457e-01 7.20589161e-01
7.63121426e-01 3.84057164e-01 -1.04407644e+00 8.11387479e-01
-4.79262739e-01 -1.17496707e-01 3.50485414e-01 9.07494724e-01
-4.68055636e-01 -1.30725932e+00 -4.04585660e-01 5.22178531e-01
-2.28408724e-01 -7.66036093e-01 -8.05313110e-01 7.49739528e-01
2.21065179e-01 1.27606940e+00 3.67753841e-02 -5.52512586e-01
4.03971076e-01 3.71140867e-01 1.73965931e-01 -8.00035655e-01
-6.17750525e-01 -3.46430629e-01 5.05423129e-01 -1.30558327e-01
-6.94677472e-01 -8.44325781e-01 -1.06661844e+00 -2.40645036e-01
-9.50541273e-02 2.22751573e-01 4.86591101e-01 1.36744201e+00
4.35941368e-01 3.92006397e-01 2.62732893e-01 -4.44607615e-01
-4.29942936e-01 -1.44190240e+00 -6.63456976e-01 6.08275473e-01
1.01054125e-01 -7.76604772e-01 -3.91788632e-01 1.07707493e-02] | [10.28553581237793, 9.039926528930664] |
abec81a9-c351-40ad-9f02-9441883d8112 | lgnn-a-context-aware-line-segment-detector | 2008.05892 | null | https://arxiv.org/abs/2008.05892v2 | https://arxiv.org/pdf/2008.05892v2.pdf | LGNN: A Context-aware Line Segment Detector | We present a novel real-time line segment detection scheme called Line Graph Neural Network (LGNN). Existing approaches require a computationally expensive verification or postprocessing step. Our LGNN employs a deep convolutional neural network (DCNN) for proposing line segment directly, with a graph neural network (GNN) module for reasoning their connectivities. Specifically, LGNN exploits a new quadruplet representation for each line segment where the GNN module takes the predicted candidates as vertexes and constructs a sparse graph to enforce structural context. Compared with the state-of-the-art, LGNN achieves near real-time performance without compromising accuracy. LGNN further enables time-sensitive 3D applications. When a 3D point cloud is accessible, we present a multi-modal line segment classification technique for extracting a 3D wireframe of the environment robustly and efficiently. | ['Qiang Hu', 'Quan Meng', 'Jingyi Yu', 'Xuming He', 'Jiakai Zhang'] | 2020-08-13 | null | null | null | null | ['line-segment-detection'] | ['computer-vision'] | [ 2.10516840e-01 2.68935770e-01 -2.27255121e-01 -2.47309402e-01
-3.64129692e-01 -6.27510130e-01 2.50446290e-01 3.09492946e-01
1.64280906e-01 -4.95477405e-04 -3.63716990e-01 -7.61027634e-01
5.05022099e-03 -1.21000838e+00 -1.10098219e+00 1.54918388e-01
-4.71797585e-01 4.18798476e-01 4.23324049e-01 -2.23014832e-01
4.01233882e-01 1.51018584e+00 -1.22472358e+00 -1.37842253e-01
5.90431154e-01 1.52468681e+00 -3.15062404e-01 7.07930624e-01
-2.18206346e-01 4.07061517e-01 -2.33302176e-01 -1.67657226e-01
7.83674300e-01 2.42600232e-01 -5.25953114e-01 1.52589783e-01
8.80013883e-01 -5.25248826e-01 -8.03382456e-01 9.32305396e-01
5.67581177e-01 -1.10139161e-01 4.64147419e-01 -1.45493865e+00
-1.27035350e-01 4.09167141e-01 -9.07888889e-01 -2.26062730e-01
6.47277892e-01 1.54413860e-02 7.51109004e-01 -8.23509336e-01
7.95698524e-01 7.66360998e-01 1.34340262e+00 -3.03369880e-01
-9.57574666e-01 -5.97671151e-01 -2.78635025e-02 -2.54498422e-01
-1.52708030e+00 -1.26174581e-03 1.25663507e+00 -3.88356686e-01
1.34160185e+00 6.28297403e-02 9.65012789e-01 4.80701298e-01
2.46809021e-01 6.81573689e-01 2.49560133e-01 -3.38273108e-01
2.45907214e-02 -6.85775042e-01 -4.20990363e-02 1.23170388e+00
2.35329151e-01 1.15741231e-01 -2.39299670e-01 1.30753284e-02
1.41191411e+00 -6.96536601e-02 -2.66592771e-01 -8.49441469e-01
-1.18345129e+00 4.94836599e-01 9.22062635e-01 -1.25553235e-01
-4.09123093e-01 3.65939677e-01 4.21114117e-01 1.43133268e-01
2.87089586e-01 3.94293904e-01 -3.36247027e-01 1.71565741e-01
-1.23300314e+00 7.81718418e-02 6.67853773e-01 1.36464727e+00
1.10013556e+00 1.73572481e-01 8.89681727e-02 4.19351906e-01
2.64892936e-01 5.01625121e-01 -4.08578753e-01 -9.21898007e-01
5.99188626e-01 1.16247940e+00 -2.63732791e-01 -1.48627353e+00
-9.45897818e-01 -6.23486042e-01 -7.06070006e-01 4.36016411e-01
7.63862208e-03 -1.38221160e-01 -9.25610602e-01 9.29508030e-01
5.78895152e-01 1.61361560e-01 -3.19393754e-01 6.90273345e-01
9.45108116e-01 4.94800121e-01 -6.58016086e-01 4.82447505e-01
8.97329986e-01 -9.44090605e-01 -5.27005978e-02 -2.70949870e-01
7.78096914e-01 -8.75249624e-01 5.42431593e-01 1.96922392e-01
-1.01812863e+00 -4.69376713e-01 -1.35124683e+00 -3.64115715e-01
-2.56186008e-01 2.54013181e-01 9.34452951e-01 1.79684430e-01
-9.53781366e-01 4.71001983e-01 -5.60828567e-01 -3.37849408e-01
4.80475217e-01 4.98438984e-01 -6.05516553e-01 -4.96357214e-03
-6.16708577e-01 3.12036932e-01 3.42918515e-01 4.24416810e-01
-4.74029541e-01 -6.56827331e-01 -1.36676598e+00 9.23027247e-02
4.36833680e-01 -7.67722368e-01 1.05126095e+00 -4.68140870e-01
-1.22329032e+00 9.35836732e-01 2.72722870e-01 -5.00655770e-01
6.24692261e-01 -1.95468441e-01 -3.18989903e-01 5.36294103e-01
2.33427778e-01 5.90546846e-01 6.55972004e-01 -1.40815783e+00
-6.65139675e-01 -7.26064958e-04 2.12799728e-01 -5.41788228e-02
5.02884150e-01 -3.15773517e-01 -8.68023932e-01 -4.86488640e-01
6.86361194e-01 -7.64907062e-01 -2.82248586e-01 4.62248534e-01
-1.10634398e+00 -1.89944103e-01 1.09904397e+00 -3.73104513e-01
1.17122114e+00 -1.97643149e+00 -5.06512284e-01 9.73229468e-01
4.15257305e-01 9.42332447e-02 -1.85724810e-01 6.55782700e-01
-2.59181857e-01 -1.81064025e-01 -2.23521516e-01 -2.85672903e-01
6.87552616e-02 -1.12658352e-01 -2.45025292e-01 7.72251129e-01
1.10113025e-01 1.00382149e+00 -7.01760054e-01 -2.57899880e-01
6.16160870e-01 3.19250584e-01 -4.58707660e-01 1.00161284e-01
-1.85243353e-01 -2.69000471e-01 -3.93732220e-01 1.17314494e+00
9.45375800e-01 -4.15763855e-01 -2.24837229e-01 -8.04169416e-01
-4.72579598e-01 -9.98035818e-03 -1.15523767e+00 2.00375080e+00
-1.17048502e-01 8.70161831e-01 -1.99025139e-01 -8.14665198e-01
1.42680955e+00 -2.41685838e-01 7.70534396e-01 -6.44811690e-01
3.11355293e-01 1.22836262e-01 -5.51236510e-01 -3.82949561e-01
7.44256496e-01 7.71661818e-01 -2.95370907e-01 3.74036461e-01
-2.36348361e-01 -4.23535287e-01 -2.06873730e-01 2.85545439e-01
1.37057340e+00 5.12648463e-01 1.79763168e-01 4.50536609e-02
2.56912529e-01 3.46613824e-01 4.33604598e-01 7.14128435e-01
3.56720835e-02 7.44332969e-01 4.85622436e-01 -8.16109121e-01
-1.27815497e+00 -1.01934958e+00 2.87944674e-01 2.52924740e-01
5.89252174e-01 -4.73026603e-01 -5.26168406e-01 -7.61651516e-01
3.14748406e-01 3.31316769e-01 -3.98304194e-01 1.07762590e-01
-7.80831754e-01 3.85969669e-01 7.48893440e-01 6.48504853e-01
7.90221214e-01 -3.84108901e-01 -5.53534746e-01 8.16549808e-02
2.82327235e-01 -1.37032247e+00 -3.80700171e-01 1.68569297e-01
-6.83317363e-01 -1.17274272e+00 -5.20257391e-02 -1.02210116e+00
9.72935081e-01 5.07198155e-01 1.11647594e+00 2.35806361e-01
-3.20376188e-01 4.27987963e-01 -9.34721157e-02 -3.68575662e-01
-3.79062723e-03 2.16800585e-01 -2.91798234e-01 -3.19745004e-01
2.04037696e-01 -6.09059870e-01 -7.96857417e-01 1.94390431e-01
-4.02617931e-01 4.59856838e-01 4.81502682e-01 2.93534517e-01
9.90220189e-01 -3.07223219e-02 -1.19628169e-01 -6.74421310e-01
2.80861497e-01 -3.10802996e-01 -1.15049684e+00 6.48227111e-02
-4.66307282e-01 -3.04458141e-01 2.95786053e-01 1.01841465e-01
-4.00137156e-01 4.49411184e-01 -3.09800863e-01 -7.99173057e-01
-3.51041526e-01 6.38396919e-01 6.51245043e-02 -6.11754000e-01
4.16433245e-01 -1.05432697e-01 -1.65962741e-01 -1.98728114e-01
3.02363634e-01 2.90172637e-01 9.28172767e-01 -4.31221575e-01
1.27702999e+00 6.30878568e-01 5.84546328e-01 -6.67729914e-01
-8.27399015e-01 -4.92994636e-01 -9.52823639e-01 -7.33091235e-01
6.41891837e-01 -8.72601926e-01 -8.15488517e-01 4.65750635e-01
-1.35546887e+00 -4.39299494e-01 -3.14381748e-01 3.28857899e-01
-6.73367143e-01 4.13236171e-01 -6.69224262e-01 -4.61968124e-01
-3.53917748e-01 -9.98528183e-01 1.48356831e+00 2.21358404e-01
8.15356895e-02 -7.78179586e-01 -1.30455926e-01 -3.42543647e-02
3.86926942e-02 1.05522883e+00 7.58781016e-01 -2.20447764e-01
-9.86332059e-01 -7.16421247e-01 -7.96605408e-01 -1.68458104e-01
-8.61457884e-02 3.95443946e-01 -7.29168773e-01 -5.31849079e-02
-6.84326351e-01 6.64394349e-02 3.18409294e-01 4.23108697e-01
1.26814449e+00 -1.62592474e-02 -5.55942953e-01 1.26668179e+00
1.70273948e+00 9.95242745e-02 5.53251505e-01 7.00284243e-01
1.24398577e+00 1.80802181e-01 4.57028210e-01 4.55409795e-01
6.84476793e-01 3.15075159e-01 8.75511825e-01 -7.40905225e-01
-1.33430943e-01 -5.40668011e-01 -4.28182930e-02 5.94331443e-01
-1.50663391e-01 -3.78460437e-01 -1.26482141e+00 2.73492336e-01
-1.93698657e+00 -7.34851241e-01 -4.44550961e-01 1.75427473e+00
5.51509857e-03 4.59302902e-01 4.17895988e-03 2.08247244e-01
6.77412152e-01 2.02531084e-01 -6.50840580e-01 -3.40251148e-01
-1.40108779e-01 5.82998171e-02 1.15708256e+00 1.42466754e-01
-1.32283175e+00 9.21642482e-01 6.30200481e+00 6.86729729e-01
-1.21976280e+00 -6.39603257e-01 3.61387253e-01 3.37965399e-01
-3.06039989e-01 -1.10750878e-02 -5.72856724e-01 -8.83481950e-02
3.27426970e-01 1.79302320e-01 9.02986526e-02 1.13469791e+00
1.43370181e-01 -1.61836036e-02 -9.33268785e-01 1.22586429e+00
1.39809370e-01 -1.87659264e+00 -6.45475760e-02 8.71409476e-02
6.29487693e-01 6.36189580e-01 -4.37852710e-01 -7.92813897e-02
1.49963126e-01 -7.39859879e-01 9.70713735e-01 6.93626761e-01
1.00468779e+00 -1.16027820e+00 4.88720745e-01 1.17925510e-01
-1.65398371e+00 2.35268906e-01 -2.16671184e-01 1.23444788e-01
4.94931936e-01 7.78931856e-01 -1.00061178e+00 9.64004695e-01
5.61872900e-01 8.22673142e-01 -5.91869831e-01 1.43462229e+00
-3.31291854e-01 5.46163678e-01 -8.28491747e-01 3.55405569e-01
5.32614112e-01 -2.37928852e-01 7.07706511e-01 1.11565757e+00
5.71401238e-01 -3.07292521e-01 5.42185605e-01 8.38861704e-01
-2.01700002e-01 -7.01088086e-02 -1.06259751e+00 1.43770710e-01
6.07080758e-01 1.31269169e+00 -1.25680685e+00 -2.75772903e-02
-6.15667462e-01 9.13596034e-01 4.49708700e-01 3.69840622e-01
-6.94163322e-01 -9.19420540e-01 4.61760074e-01 2.39766404e-01
5.08220792e-01 -8.55687976e-01 -4.38689739e-01 -8.04425120e-01
-9.98547375e-02 -1.81708544e-01 1.89239725e-01 -9.14861381e-01
-1.08983469e+00 6.91771924e-01 -3.44433188e-01 -1.77523983e+00
-7.95758739e-02 -5.08293509e-01 -6.50858939e-01 3.92059714e-01
-1.69137907e+00 -1.81509781e+00 -6.18378401e-01 7.62179673e-01
1.45035386e-01 -1.84390543e-03 4.71916765e-01 2.93374419e-01
-5.92460811e-01 4.96165067e-01 -2.94090658e-01 7.16496885e-01
2.77384460e-01 -1.09316909e+00 1.05008876e+00 1.03805780e+00
-5.75775690e-02 4.55565065e-01 3.42978030e-01 -9.63431895e-01
-1.65637314e+00 -1.41403377e+00 5.91694832e-01 3.61864939e-02
5.87027013e-01 -4.76066113e-01 -4.87285912e-01 9.26372170e-01
-3.05462610e-02 2.24834725e-01 4.71372336e-01 -2.47970670e-01
-4.36409742e-01 6.25288039e-02 -1.00479233e+00 4.64130342e-01
1.54692054e+00 -6.55381501e-01 -2.43777514e-01 3.28132808e-01
9.73362148e-01 -1.11279440e+00 -1.09621251e+00 6.56403482e-01
5.28162122e-01 -1.11283994e+00 1.15561450e+00 -2.78196335e-02
3.75228614e-01 -7.72659838e-01 -4.54623848e-02 -9.10139918e-01
-2.91655064e-01 -6.79268479e-01 -3.37639153e-01 9.06323612e-01
3.06564927e-01 -3.75926912e-01 1.07583594e+00 3.79541188e-01
-9.16171551e-01 -8.35988760e-01 -7.18253672e-01 -6.60980165e-01
-7.44442999e-01 -1.00296926e+00 1.22940469e+00 1.03764784e+00
-4.53820229e-01 6.81346357e-02 -6.95257708e-02 8.16568792e-01
7.23498523e-01 6.98117912e-01 1.18229198e+00 -1.11812794e+00
2.06360579e-01 -5.56320727e-01 -9.38727140e-01 -1.36724281e+00
8.70449876e-04 -9.72803235e-01 -1.23874761e-01 -1.87472010e+00
-7.47012258e-01 -6.03159070e-01 7.78969675e-02 4.62484360e-01
5.29231369e-01 4.15134728e-01 -3.28391716e-02 -6.41119331e-02
-6.27252519e-01 1.61784902e-01 1.10577309e+00 -2.29250312e-01
-2.41792277e-01 -8.89672935e-02 -1.59414023e-01 9.88228858e-01
7.04131544e-01 -2.91015469e-02 -2.27929205e-01 -7.37558961e-01
6.46325350e-01 -2.31360663e-02 4.28606212e-01 -1.28155470e+00
7.02358425e-01 1.23355985e-01 6.71219707e-01 -1.66813648e+00
1.66919574e-01 -1.04002726e+00 1.02493025e-01 1.77766025e-01
7.25070536e-02 2.25992277e-01 4.10863429e-01 3.07689250e-01
-1.01356089e-01 -3.44580635e-02 3.20047647e-01 7.85452724e-02
-1.05499268e+00 9.63305116e-01 -1.29746750e-01 -5.87462664e-01
1.14336431e+00 -7.17197955e-01 -2.97042578e-01 -2.04238057e-01
-3.13634217e-01 4.32379335e-01 8.04686129e-01 1.78300142e-01
1.05828702e+00 -1.51184881e+00 -2.92130172e-01 5.91115355e-01
4.19373423e-01 7.29720116e-01 2.45007560e-01 3.78235221e-01
-1.42570639e+00 8.38582739e-02 -1.32081285e-01 -8.87396216e-01
-8.74838889e-01 4.10903662e-01 4.85185623e-01 1.58277288e-01
-1.07366049e+00 9.01658237e-01 -2.81706989e-01 -6.38424754e-01
1.27355188e-01 -4.85507637e-01 1.15406647e-01 -1.46296710e-01
5.06804548e-02 4.47184443e-01 1.37995571e-01 -6.52904093e-01
-2.28563011e-01 1.05472362e+00 3.10650021e-01 4.61174935e-01
1.55446923e+00 -1.27620593e-01 -3.90874743e-02 1.13030161e-04
1.16943312e+00 1.23270698e-01 -1.29768503e+00 -8.63144547e-02
-8.22016597e-02 -3.98738742e-01 4.02312636e-01 -3.21356148e-01
-1.22532845e+00 5.09740055e-01 3.17382544e-01 3.23782295e-01
9.35926020e-01 -7.92519301e-02 1.16906822e+00 4.84854043e-01
6.95401907e-01 -1.06766570e+00 -3.61109465e-01 5.91918945e-01
7.96058357e-01 -1.04239535e+00 2.20730066e-01 -8.07313085e-01
1.21977188e-01 1.76428330e+00 6.25602067e-01 -4.32153136e-01
7.14807570e-01 5.55430710e-01 1.52820991e-02 -8.78290117e-01
6.22629263e-02 -2.28709236e-01 3.69329572e-01 6.11754954e-01
-2.02225119e-01 -2.06786934e-02 5.92167258e-01 4.39852141e-02
-4.71203625e-01 -1.04604952e-01 2.54432201e-01 1.08813238e+00
-2.95314759e-01 -7.66673207e-01 -2.57559896e-01 5.55566251e-01
2.46448189e-01 1.78631723e-01 -3.21641654e-01 1.00063384e+00
5.26918806e-02 4.52919483e-01 4.18045521e-01 -7.52500355e-01
7.09433436e-01 -6.64088130e-01 2.60835171e-01 -3.51404667e-01
-5.01019835e-01 1.00807324e-01 2.27144301e-01 -1.27644026e+00
-3.91517520e-01 -5.07638812e-01 -1.49592602e+00 -4.00759101e-01
-3.75237763e-01 -3.53313178e-01 8.67300510e-01 6.54145777e-01
7.00253844e-01 3.47898394e-01 9.41865683e-01 -1.33375359e+00
5.49750566e-01 -1.87102795e-01 -4.58619744e-01 -2.06168611e-02
3.85329336e-01 -5.64076960e-01 1.25641152e-02 -3.14370126e-01] | [8.10633373260498, -1.901037335395813] |
1380061a-6874-4bb3-b220-50a8c298a189 | enriching-the-e2e-dataset | null | null | https://aclanthology.org/2021.inlg-1.18 | https://aclanthology.org/2021.inlg-1.18.pdf | Enriching the E2E dataset | This study introduces an enriched version of the E2E dataset, one of the most popular language resources for data-to-text NLG. We extract intermediate representations for popular pipeline tasks such as discourse ordering, text structuring, lexicalization and referring expression generation, enabling researchers to rapidly develop and evaluate their data-to-text pipeline systems. The intermediate representations are extracted by aligning non-linguistic and text representations through a process called delexicalization, which consists in replacing input referring expressions to entities/attributes with placeholders. The enriched dataset is publicly available. | ['Adriana Pagano', 'Brian Davis', 'Helena Vaz', 'Thiago castro Ferreira'] | null | null | null | null | inlg-acl-2021-8 | ['referring-expression-generation'] | ['computer-vision'] | [ 2.75786489e-01 8.81379545e-01 -4.13945079e-01 -4.18768972e-01
-6.80292904e-01 -7.96558142e-01 1.11242008e+00 7.33801067e-01
-4.26173121e-01 8.09723139e-01 1.32831085e+00 -2.36727074e-01
1.33398727e-01 -7.39461362e-01 -3.76743793e-01 1.89084843e-01
4.41979229e-01 6.58553541e-01 -1.55256882e-01 -5.92522800e-01
3.61806780e-01 5.63833639e-02 -1.43712771e+00 9.20023322e-01
7.96939492e-01 5.87639868e-01 -2.63492793e-01 4.13700759e-01
-9.72202122e-01 1.24946845e+00 -8.14688921e-01 -6.17881715e-01
-3.95775974e-01 -5.15297294e-01 -1.43937433e+00 -1.61079824e-01
-5.88695295e-02 2.09395457e-02 -9.05343890e-02 7.46344328e-01
5.27426124e-01 4.00515109e-01 8.15603077e-01 -9.21513557e-01
-1.08233511e+00 1.59683681e+00 -1.42183557e-01 2.08278328e-01
9.00750995e-01 -2.31947765e-01 1.34860814e+00 -1.09628177e+00
1.46857524e+00 1.63384926e+00 1.27469331e-01 8.47060919e-01
-1.27603889e+00 -1.62595004e-01 -1.44594476e-01 1.45874709e-01
-1.32045805e+00 -9.20110464e-01 6.23718739e-01 -5.45457661e-01
1.63119960e+00 4.09446329e-01 3.18656832e-01 1.22627521e+00
-5.25150239e-01 9.15926874e-01 5.29316247e-01 -6.93983555e-01
-2.19919965e-01 -7.45055154e-02 3.44854534e-01 3.19098502e-01
8.77385288e-02 -4.73525167e-01 -5.96872509e-01 1.10408880e-01
5.17282546e-01 -6.77954435e-01 -1.85253173e-01 4.85841870e-01
-1.37684298e+00 8.33220959e-01 3.12286049e-01 4.64149743e-01
-5.87248802e-01 -9.44698304e-02 8.35012197e-01 1.02390580e-01
6.39503658e-01 9.85489964e-01 -1.53092220e-01 -3.03761929e-01
-5.30681610e-01 5.45716524e-01 8.14761937e-01 1.31168640e+00
4.69345778e-01 -2.05981478e-01 -9.64764297e-01 8.45887065e-01
3.04344535e-01 1.68192387e-01 6.74454689e-01 -9.11002517e-01
1.04289103e+00 1.17891181e+00 2.23396763e-01 -9.54589367e-01
-4.68331963e-01 1.86981827e-01 -5.51547050e-01 -6.57089114e-01
1.93351716e-01 -3.50392520e-01 -2.56193340e-01 1.48050439e+00
4.18485999e-01 -3.89980406e-01 5.63068032e-01 4.58515167e-01
1.97239697e+00 8.66542280e-01 3.14420432e-01 -2.06713364e-01
1.56486428e+00 -6.28898025e-01 -1.34333289e+00 -1.98603585e-01
1.16559541e+00 -8.33091617e-01 1.24693859e+00 -2.36080766e-01
-1.26950228e+00 -3.49980652e-01 -7.62739658e-01 -1.08602536e+00
-7.77388692e-01 1.95429921e-01 3.04553986e-01 8.07406083e-02
-7.78207421e-01 4.26583797e-01 -4.54829633e-01 -5.12012959e-01
3.60384136e-01 -2.19631031e-01 -5.11228502e-01 3.26126099e-01
-1.42475891e+00 9.99949038e-01 8.98377717e-01 -8.04678872e-02
-7.22691510e-03 -7.22872376e-01 -1.29416871e+00 -3.14844996e-01
3.51623803e-01 -6.16807163e-01 1.13635337e+00 -5.67145646e-01
-1.72472870e+00 1.60105205e+00 -3.79121810e-01 -5.31574607e-01
2.08428368e-01 -7.59344220e-01 -4.55970883e-01 -2.67228603e-01
2.76447326e-01 8.41448069e-01 2.66830802e-01 -6.82218492e-01
-3.54259789e-01 -3.03239763e-01 5.90257980e-02 2.31449530e-01
-2.54257233e-03 9.24329698e-01 -4.00437057e-01 -7.87971735e-01
-1.37973234e-01 -4.08544004e-01 8.67122337e-02 -5.89941025e-01
-1.07957387e+00 -9.51177835e-01 4.27719325e-01 -9.90232646e-01
1.99157226e+00 -1.86121857e+00 5.24332881e-01 -2.19330996e-01
2.94179291e-01 -1.03399307e-01 2.49929018e-02 8.45686495e-01
-3.45894188e-01 7.24150181e-01 8.69772360e-02 -3.24807942e-01
2.69365102e-01 7.86428675e-02 -6.32103205e-01 -8.82673487e-02
7.24095762e-01 1.33782697e+00 -1.02307940e+00 -7.80050099e-01
1.00407526e-01 2.30663791e-01 -5.19859195e-01 4.13095683e-01
-6.79635465e-01 4.85468894e-01 -6.77316725e-01 2.92900890e-01
-6.43907562e-02 -1.67467564e-01 1.96326852e-01 -3.11553419e-01
-3.76386881e-01 1.23720574e+00 -9.78235602e-01 1.63071716e+00
-6.13560438e-01 7.98229873e-01 -4.62373078e-01 -5.23120046e-01
1.02787662e+00 4.53258872e-01 1.60536960e-01 -5.18423438e-01
2.97539145e-01 3.65088098e-02 -4.37615991e-01 -8.42320561e-01
1.32904005e+00 2.55425453e-01 -7.97947109e-01 5.36484122e-01
3.65325585e-02 -2.92223454e-01 5.74247718e-01 3.64925295e-01
8.81813824e-01 4.40089077e-01 9.86585796e-01 -2.10877910e-01
5.27269721e-01 2.50479370e-01 2.91708767e-01 1.97052538e-01
5.73732972e-01 2.30693921e-01 9.00427759e-01 -3.37234735e-01
-9.55776036e-01 -7.35756636e-01 -1.79809377e-01 1.46382654e+00
-3.09486985e-01 -1.21142852e+00 -7.59519756e-01 -5.18778086e-01
-1.61460772e-01 1.46686685e+00 -8.11738253e-01 1.05878726e-01
-1.03421926e+00 -1.98182523e-01 1.03413677e+00 4.98099595e-01
1.54106274e-01 -1.44739926e+00 -5.80993712e-01 4.74566191e-01
-6.66210711e-01 -1.13154638e+00 -1.28136247e-01 -7.98964053e-02
-4.73644286e-01 -9.19236004e-01 -1.07880168e-01 -5.12778282e-01
6.03420079e-01 -5.80949128e-01 1.65976846e+00 1.10491112e-01
7.64697865e-02 -1.87107250e-01 -7.74971366e-01 -4.31682050e-01
-7.74931669e-01 3.40580225e-01 -3.97189379e-01 -3.91414464e-01
5.15254259e-01 -5.59271220e-03 -6.51903078e-02 -1.79612264e-01
-1.04114747e+00 4.32477027e-01 -2.72436380e-01 4.78698641e-01
8.01927328e-01 -8.77841175e-01 4.49784040e-01 -1.41511738e+00
1.09229946e+00 -7.11114943e-01 -2.71669120e-01 3.35736483e-01
-1.35426251e-02 5.79003632e-01 5.28336704e-01 -3.56316147e-03
-1.26031864e+00 -2.50076979e-01 -4.06001151e-01 1.90053657e-01
7.53846243e-02 9.23856080e-01 -4.05298650e-01 1.13436568e+00
8.85107160e-01 -4.51780707e-01 -5.04306436e-01 -6.62855446e-01
1.35356843e+00 8.79202008e-01 8.51780295e-01 -9.67024148e-01
3.33831787e-01 -2.39380579e-02 -2.91495025e-01 -8.51717651e-01
-1.02218390e+00 -1.77210242e-01 -7.85909653e-01 6.00344539e-02
1.10230505e+00 -7.01103866e-01 -3.70124161e-01 2.77960841e-02
-1.76340067e+00 -2.21825510e-01 -9.18514490e-01 -1.17353491e-01
-3.74668062e-01 -2.86402088e-03 -5.64024746e-01 -3.51760328e-01
-5.33731997e-01 -5.77198923e-01 1.26582885e+00 2.11563811e-01
-1.14313328e+00 -1.08834517e+00 -4.66626249e-02 1.53053015e-01
2.82942653e-01 7.68404484e-01 1.32899523e+00 -1.18737602e+00
8.45931172e-02 1.51540473e-01 -1.72360361e-01 -2.45374426e-01
2.07149819e-01 5.93908429e-01 -5.94333291e-01 2.57404149e-01
-5.06988466e-01 -6.52155280e-01 5.20641387e-01 9.99374539e-02
1.15505600e+00 -8.46458018e-01 -3.17637205e-01 6.07475460e-01
7.53543913e-01 -2.52873540e-01 7.43026614e-01 5.85253954e-01
7.29203343e-01 1.16710687e+00 5.52487075e-01 7.17098594e-01
8.04882526e-01 6.76815808e-01 1.93779394e-01 1.12965040e-01
-3.59417319e-01 -4.48528707e-01 2.20012397e-01 1.09774148e+00
1.05729125e-01 -4.87326026e-01 -1.17673206e+00 5.26790738e-01
-1.75526333e+00 -9.10167277e-01 -3.41072172e-01 1.63080776e+00
1.44365311e+00 -2.40569457e-01 -1.09451726e-01 -1.85614020e-01
7.68531322e-01 4.12820607e-01 -1.96070358e-01 -3.58778119e-01
-4.75823581e-01 4.35094863e-01 -3.65631878e-02 6.14985406e-01
-1.01729596e+00 1.63712966e+00 6.02945375e+00 5.36631107e-01
-7.65094936e-01 1.83831137e-02 3.96637768e-01 5.70061728e-02
-7.07451999e-01 1.50564648e-02 -1.23207867e+00 4.05110084e-02
1.04509938e+00 -9.48206902e-01 1.36263505e-01 6.50825620e-01
2.25177214e-01 3.49782676e-01 -1.61581755e+00 7.56074011e-01
-1.14812478e-02 -1.79744637e+00 5.29406428e-01 -5.67972600e-01
3.71317863e-01 -1.54384404e-01 -3.50855231e-01 4.80862021e-01
6.48146272e-01 -1.25983036e+00 8.09062421e-01 6.63275838e-01
1.21221101e+00 -3.65566939e-01 6.19842470e-01 -4.08546478e-02
-1.17748749e+00 3.24091375e-01 -1.08088233e-01 9.34770852e-02
4.83098835e-01 3.62738013e-01 -9.70132589e-01 6.35896325e-01
2.58819163e-01 9.23703671e-01 -7.75860071e-01 2.39641532e-01
-7.91248441e-01 2.94551671e-01 -2.48734608e-01 -3.33548307e-01
-4.02985848e-02 -6.37195706e-02 7.22685516e-01 1.64273870e+00
2.23389685e-01 1.37581214e-01 -6.63640816e-03 1.27189159e+00
-5.40052414e-01 5.88410735e-01 -7.83270955e-01 -4.93286848e-01
1.12459123e+00 1.20533144e+00 -5.71019292e-01 -5.65773845e-01
-1.14117548e-01 6.65228486e-01 5.92552185e-01 1.92071959e-01
-5.17723441e-01 -8.01362038e-01 4.34151679e-01 4.31493931e-02
-1.87893927e-01 -9.07626078e-02 -4.76560183e-02 -1.09345329e+00
-6.75891638e-02 -9.64046955e-01 6.63231611e-01 -8.86076331e-01
-1.26253247e+00 8.15046906e-01 4.39745277e-01 -9.08914626e-01
-1.07419932e+00 -3.80037993e-01 -5.51712215e-01 9.21361744e-01
-1.09866977e+00 -1.17979801e+00 -1.52063519e-01 5.36141753e-01
6.74341023e-01 -1.21761598e-02 1.14017713e+00 1.28486291e-01
-9.24797893e-01 3.80105346e-01 -3.30192894e-01 6.13448322e-01
6.36308491e-01 -1.17958701e+00 9.14127707e-01 7.50946462e-01
1.22529693e-01 1.00907850e+00 7.33198225e-01 -8.75077069e-01
-1.12228858e+00 -1.21984649e+00 1.49057329e+00 -7.35214293e-01
1.05412614e+00 -3.93156588e-01 -1.11200261e+00 1.10825264e+00
6.89136386e-01 -8.58508721e-02 8.67133021e-01 2.28048638e-01
-3.16782668e-02 5.02605140e-01 -8.08834136e-01 8.20160031e-01
1.15104425e+00 -7.04116464e-01 -1.41218555e+00 2.79445231e-01
1.20818484e+00 -9.88832295e-01 -1.00867069e+00 1.50835946e-01
6.96776202e-03 -2.20141008e-01 5.89578271e-01 -1.17159593e+00
8.30986798e-01 -2.05616847e-01 -2.41039678e-01 -1.26131141e+00
8.80064890e-02 -1.03307116e+00 -3.16611916e-01 2.19342732e+00
7.62124121e-01 -1.65102586e-01 7.88847581e-02 6.49307489e-01
-3.53324085e-01 -4.40743327e-01 -7.93787301e-01 -2.77824283e-01
1.96254596e-01 -3.59159261e-01 1.05578423e+00 1.09467280e+00
5.37448406e-01 9.10514057e-01 2.67214894e-01 -4.83722806e-01
-1.06513679e-01 -1.44713931e-02 7.66458690e-01 -1.22854519e+00
6.13019615e-02 -6.50911033e-01 3.21746245e-02 -1.04340255e+00
6.16245568e-01 -1.41054213e+00 -1.08344153e-01 -1.84413397e+00
-2.36363977e-01 -2.17296451e-01 3.27107072e-01 5.88646650e-01
-2.17606336e-01 -4.09584582e-01 -2.66706757e-02 1.71858847e-01
-6.77444816e-01 8.13057423e-01 9.01786745e-01 2.65921839e-02
-5.04879177e-01 -5.42291224e-01 -9.34342980e-01 8.26021552e-01
4.47129875e-01 -6.13918483e-01 -1.14766143e-01 -7.98260093e-01
7.71163821e-01 7.57299811e-02 -1.10315718e-01 -4.08656836e-01
2.31645592e-02 -3.39308888e-01 1.02829628e-01 -7.18850195e-01
-2.77664252e-02 1.22893136e-02 5.77229112e-02 -1.51176319e-01
-1.03152478e+00 5.28415978e-01 3.33062977e-01 -3.81966501e-01
-4.09675509e-01 -4.57134217e-01 3.79346102e-01 -8.99262801e-02
-6.35685503e-01 -2.36679509e-01 -3.70285243e-01 8.79286766e-01
8.13277125e-01 6.26790822e-02 -1.04001844e+00 -3.11238259e-01
-6.18637621e-01 2.79342592e-01 3.13630551e-01 6.78725779e-01
5.88072598e-01 -1.43686497e+00 -1.09136999e+00 7.98825454e-03
3.09764087e-01 3.56435508e-01 -3.03765029e-01 3.45845699e-01
-5.57823002e-01 3.49452049e-01 -2.37153918e-01 -1.04505174e-01
-1.14017320e+00 3.45402747e-01 3.89828384e-02 -5.04601240e-01
-7.78753638e-01 9.77302492e-01 -2.81162471e-01 -5.06815970e-01
-8.35146382e-02 -6.09986424e-01 -9.56207216e-01 5.15392005e-01
7.23199844e-01 3.77580851e-01 1.53460493e-02 -1.00767350e+00
-3.51050138e-01 1.15599342e-01 -1.19208105e-01 -2.78662950e-01
1.38096690e+00 -3.37360591e-01 -5.56656957e-01 6.97176456e-01
8.90970528e-01 3.79581869e-01 -6.35576427e-01 -2.50441968e-01
9.36025441e-01 4.89401594e-02 -1.15455762e-01 -3.42831880e-01
-4.51037198e-01 8.50472867e-01 -3.48682761e-01 3.83791447e-01
6.61941588e-01 5.39807141e-01 5.76647282e-01 8.03963721e-01
-3.00688833e-01 -1.22402608e+00 -2.94351261e-02 1.27565384e+00
1.56240284e+00 -8.25514674e-01 2.46355291e-02 -4.65584338e-01
-8.14242840e-01 1.19632423e+00 7.19525099e-01 9.85613614e-02
2.52369463e-01 2.32239127e-01 4.42293771e-02 -6.09474957e-01
-9.48398352e-01 -2.57896453e-01 3.89532208e-01 6.51578903e-01
1.38236308e+00 2.60754842e-02 -4.27882671e-01 1.25279248e+00
-8.02258551e-01 -3.03115517e-01 5.85273325e-01 6.87665343e-01
-5.36853820e-02 -1.35713565e+00 -7.94876069e-02 4.25247103e-01
-4.95491594e-01 -4.85089511e-01 -9.53340948e-01 9.08698320e-01
-2.30919734e-01 9.17065024e-01 4.54834312e-01 -4.48365435e-02
6.87295675e-01 2.50393331e-01 1.05159745e-01 -1.19008482e+00
-9.03551102e-01 -2.97519177e-01 1.00896990e+00 -6.54924870e-01
-5.31455278e-01 -6.89014316e-01 -1.86304533e+00 -2.07502514e-01
3.27604450e-02 1.35958910e-01 3.43690962e-01 1.02985275e+00
6.87758982e-01 8.77871156e-01 1.90729856e-01 -7.73099840e-01
-7.45248348e-02 -1.28242075e+00 -4.76627499e-02 8.42750907e-01
-1.40252322e-01 -5.00071108e-01 2.91276425e-02 2.67363787e-01] | [11.041111946105957, 9.08449649810791] |
cb6e97f4-bbed-4f3a-be4a-96bddcf1cfc3 | face-sketch-synthesis-with-style-transfer | 2009.08679 | null | https://arxiv.org/abs/2009.08679v1 | https://arxiv.org/pdf/2009.08679v1.pdf | Face Sketch Synthesis with Style Transfer using Pyramid Column Feature | In this paper, we propose a novel framework based on deep neural networks for face sketch synthesis from a photo. Imitating the process of how artists draw sketches, our framework synthesizes face sketches in a cascaded manner. A content image is first generated that outlines the shape of the face and the key facial features. Textures and shadings are then added to enrich the details of the sketch. We utilize a fully convolutional neural network (FCNN) to create the content image, and propose a style transfer approach to introduce textures and shadings based on a newly proposed pyramid column feature. We demonstrate that our style transfer approach based on the pyramid column feature can not only preserve more sketch details than the common style transfer method, but also surpasses traditional patch based methods. Quantitative and qualitative evaluations suggest that our framework outperforms other state-of-the-arts methods, and can also generalize well to different test images. Codes are available at https://github.com/chaofengc/Face-Sketch | ['Kwan-Yee K. Wong', 'Xiao Tan', 'Chaofeng Chen'] | 2020-09-18 | null | null | null | null | ['face-sketch-synthesis'] | ['computer-vision'] | [ 1.64688259e-01 -1.50296584e-01 4.61981632e-02 -3.93747360e-01
-2.33333722e-01 -6.38691902e-01 8.12376201e-01 -6.71437919e-01
1.81385309e-01 5.32948196e-01 2.49522045e-01 8.84066597e-02
3.10049027e-01 -1.11695647e+00 -8.23970795e-01 -3.37359130e-01
5.17291069e-01 9.46908444e-02 -2.20665932e-02 -2.32141435e-01
2.80425191e-01 9.58546758e-01 -1.44885206e+00 5.02434731e-01
5.22285521e-01 1.15439820e+00 -7.71687180e-02 3.67781729e-01
-4.31624085e-01 5.33672810e-01 -4.10668105e-01 -7.54890680e-01
4.20104623e-01 -4.44093317e-01 -4.32155013e-01 3.05151552e-01
7.73267806e-01 -9.33217049e-01 -7.08310664e-01 8.69710803e-01
3.96576434e-01 -1.10480182e-01 7.72083640e-01 -1.30130541e+00
-1.15953493e+00 1.35252520e-01 -5.70249498e-01 -6.23588979e-01
3.54737937e-01 7.49928430e-02 7.32111216e-01 -1.56457341e+00
7.59326398e-01 1.46839070e+00 6.08495355e-01 8.21654320e-01
-1.21312129e+00 -1.09552312e+00 6.04718737e-02 -2.11270899e-01
-1.42291987e+00 -5.01355350e-01 1.32928145e+00 -2.04138100e-01
2.38457307e-01 9.30630192e-02 8.60299230e-01 1.10607791e+00
-1.85871627e-02 9.70676064e-01 9.10163939e-01 -3.24636072e-01
1.33524910e-01 -1.71797872e-01 -6.59733832e-01 1.10459328e+00
-9.56684276e-02 1.54873565e-01 -6.07405007e-01 -2.32314602e-01
1.52411449e+00 4.86206025e-01 -1.06567793e-01 -3.81749153e-01
-8.78084183e-01 6.81988597e-01 4.77818400e-01 1.38949603e-01
-3.73379678e-01 4.54648018e-01 -2.07995679e-02 1.00912921e-01
5.73319852e-01 -1.12725114e-02 3.87712680e-02 2.59091854e-01
-1.18934166e+00 4.13405865e-01 7.11382389e-01 1.17438149e+00
1.01532686e+00 1.87926754e-01 -4.15765703e-01 1.07779789e+00
2.04582319e-01 6.40656114e-01 -2.55326331e-01 -1.41680968e+00
9.71275568e-02 7.02812672e-01 1.07699767e-01 -9.66282606e-01
2.42762566e-01 1.15785927e-01 -9.91628110e-01 4.50431615e-01
1.01661429e-01 -1.19854063e-01 -9.70226467e-01 1.55004036e+00
2.23912373e-01 2.95077235e-01 -4.98553395e-01 7.24707365e-01
1.16205335e+00 5.32524645e-01 -9.14541557e-02 2.63938278e-01
1.25346327e+00 -1.04955554e+00 -7.82550454e-01 1.25966236e-01
-3.21752459e-01 -8.70775998e-01 1.26791751e+00 2.60238588e-01
-1.38583267e+00 -7.10393429e-01 -9.08388555e-01 -4.69033688e-01
-1.31258026e-01 5.81104398e-01 4.89947170e-01 5.16685724e-01
-1.27092397e+00 7.53583968e-01 -4.69745606e-01 -3.56074750e-01
1.01132059e+00 1.04974128e-01 -3.97893667e-01 -2.13278443e-01
-6.96246922e-01 3.01886648e-01 -2.08937690e-01 1.71221137e-01
-8.91774595e-01 -8.80353034e-01 -8.12301934e-01 2.47785687e-01
8.80954340e-02 -8.32487822e-01 1.14741170e+00 -1.18921685e+00
-2.08640981e+00 7.01207042e-01 -3.63273352e-01 3.35643739e-01
5.71063817e-01 -2.44707778e-01 -1.03933334e-01 4.04521853e-01
-1.18695840e-01 1.17705178e+00 1.24134731e+00 -1.54080927e+00
-3.72368842e-01 2.63417047e-02 1.53841197e-01 -9.28944573e-02
-4.72195417e-01 -2.26878021e-02 -8.32058728e-01 -1.14717317e+00
-1.99694067e-01 -7.07725465e-01 1.14555642e-01 9.09539104e-01
-2.55815327e-01 -1.48830011e-01 1.02463341e+00 -6.02065623e-01
1.11727107e+00 -2.14314485e+00 3.26483250e-02 2.68533170e-01
2.59296924e-01 3.42927337e-01 -4.69851404e-01 8.53486121e-01
1.14025757e-01 1.23678483e-01 -4.54211563e-01 -5.93465328e-01
1.14804357e-01 1.45635188e-01 -5.69628060e-01 5.86496033e-02
5.97517908e-01 1.10268474e+00 -7.52388775e-01 -3.40596259e-01
2.24377021e-01 8.93620551e-01 -7.26409197e-01 3.46082628e-01
-3.43356639e-01 4.99736339e-01 -4.20925826e-01 8.33500743e-01
1.04331720e+00 -2.03975350e-01 1.49375856e-01 -3.51616591e-01
-9.65256430e-03 -1.49606645e-01 -9.58839357e-01 1.93086267e+00
-3.66667926e-01 5.21949947e-01 1.48466811e-01 -5.11239886e-01
1.21270013e+00 2.65932769e-01 1.88858494e-01 -4.86161798e-01
5.97880594e-02 2.34427199e-01 -5.49699426e-01 -1.74481556e-01
2.18369424e-01 -1.45576462e-01 3.56765360e-01 4.94793892e-01
1.19656794e-01 -3.82554024e-01 -3.13944556e-02 2.05981135e-01
7.75456309e-01 5.75738847e-01 -9.62756500e-02 -2.72829562e-01
7.16675103e-01 -6.21957541e-01 3.38642478e-01 3.02839041e-01
2.16683090e-01 9.13529158e-01 6.15618765e-01 -8.12829971e-01
-1.23566830e+00 -1.19837976e+00 9.84268636e-02 8.60796452e-01
-7.55945453e-03 -4.91359293e-01 -9.67106700e-01 -6.75369680e-01
2.25288957e-01 2.35805124e-01 -8.23796988e-01 1.40779421e-01
-4.60920066e-01 2.23408669e-01 5.63863456e-01 6.76741898e-01
8.00949156e-01 -1.40768361e+00 -9.14255008e-02 3.77440006e-02
6.28362074e-02 -9.56194580e-01 -9.08941448e-01 -7.92262256e-01
-6.92062616e-01 -7.95149207e-01 -1.19773519e+00 -9.56331670e-01
1.05386233e+00 1.56479061e-01 9.56390321e-01 6.60081506e-01
-3.13113630e-01 5.81109583e-01 -9.75052416e-02 -2.90741146e-01
-2.02887833e-01 -1.94992945e-01 -2.80842096e-01 4.84982938e-01
-4.10296707e-05 -9.00691450e-01 -1.00778413e+00 1.90429389e-01
-1.13403749e+00 3.24637592e-01 5.95079005e-01 9.13731158e-01
5.17012060e-01 -3.76296282e-01 5.10485649e-01 -7.76033878e-01
6.36759758e-01 -6.36518449e-02 -5.06093681e-01 2.79334694e-01
-1.85722172e-01 -6.33961149e-03 6.28193319e-01 -3.95514190e-01
-1.30642045e+00 2.50217795e-01 -3.01444322e-01 -7.47044742e-01
-2.09068879e-01 -1.32725179e-01 -2.79345363e-01 -3.83849502e-01
2.39637107e-01 2.04972908e-01 1.75848454e-01 -6.07291937e-01
5.75023532e-01 3.98262829e-01 3.78103018e-01 -9.67251003e-01
1.07580137e+00 8.48212719e-01 1.01721875e-01 -8.06349039e-01
-4.18381065e-01 2.73289323e-01 -8.00083578e-01 -3.65742415e-01
5.00845790e-01 -7.58897722e-01 -8.84556830e-01 5.57710230e-01
-1.43222713e+00 -5.33265352e-01 -1.73693776e-01 -1.47871897e-01
-6.24787927e-01 3.23281169e-01 -8.73519242e-01 -5.48566043e-01
-4.97076005e-01 -9.93695498e-01 1.42689085e+00 3.58774275e-01
8.13478604e-02 -7.87550032e-01 -1.17597736e-01 -2.76070833e-02
5.39998829e-01 2.86470979e-01 8.89423490e-01 2.28387162e-01
-6.98507786e-01 5.74388579e-02 -6.01151884e-01 3.95437360e-01
3.91495585e-01 3.99560094e-01 -1.07411277e+00 -2.58752257e-01
-5.20986557e-01 -2.30087206e-01 8.01961124e-01 1.53869390e-01
1.58714306e+00 -5.05587637e-01 -2.14958787e-01 8.56970370e-01
1.36958194e+00 1.33557826e-01 8.40701401e-01 -1.96878061e-01
8.37378442e-01 5.79223037e-01 3.43015678e-02 4.70415384e-01
3.44233245e-01 5.10740042e-01 4.03392315e-02 -5.06582141e-01
-4.75053281e-01 -7.93527722e-01 1.98100403e-01 5.08947253e-01
-3.22137207e-01 -1.84411183e-03 -5.48349977e-01 3.82094473e-01
-1.62034178e+00 -8.01396668e-01 3.34467322e-01 1.82109749e+00
8.31559777e-01 -3.57761890e-01 1.19904950e-01 -1.10631421e-01
5.84611058e-01 2.40037903e-01 -4.21023816e-01 -4.00544614e-01
1.28608122e-01 7.60322034e-01 -9.14888307e-02 5.20024717e-01
-7.95040309e-01 1.25945473e+00 5.99393559e+00 8.54914486e-01
-1.26468039e+00 -8.84680972e-02 7.02241182e-01 2.10886262e-03
-6.86756492e-01 -7.19559565e-02 -4.19751793e-01 3.37706119e-01
1.19176209e-01 2.13396903e-02 7.86747277e-01 7.13863254e-01
-7.65437111e-02 3.41202438e-01 -1.13493896e+00 1.07377636e+00
1.95207700e-01 -1.70638657e+00 5.63150108e-01 4.03185114e-02
9.26460803e-01 -6.24273360e-01 2.71814287e-01 3.91499363e-02
1.43785819e-01 -1.08981538e+00 7.77797818e-01 8.20072591e-01
1.29872119e+00 -8.03245425e-01 2.07047328e-01 -8.30884501e-02
-1.46106112e+00 1.01699211e-01 -4.13641751e-01 1.08057808e-03
-1.40569463e-01 3.33901882e-01 -3.61732841e-01 4.10083652e-01
6.41979516e-01 8.27890515e-01 -4.85667348e-01 7.68270850e-01
-4.99720514e-01 3.26681852e-01 -8.80685896e-02 9.39861313e-02
5.25094531e-02 -3.25734824e-01 1.59198374e-01 1.18564320e+00
5.87968290e-01 3.57975721e-01 -8.85977894e-02 1.48377717e+00
-6.13890588e-01 2.24026591e-01 -8.37772489e-01 -1.03937380e-01
6.67011917e-01 1.50179839e+00 -6.27545416e-01 -5.29810786e-01
-4.75818276e-01 1.34099782e+00 2.73303479e-01 6.69323802e-01
-5.67326963e-01 -5.07262230e-01 5.30775428e-01 1.91389471e-01
6.24236763e-01 -6.44248724e-02 -2.18680426e-01 -1.13034046e+00
1.93592995e-01 -8.04031610e-01 -2.72395104e-01 -9.87918377e-01
-1.45453227e+00 6.68959022e-01 -2.55131036e-01 -1.21319270e+00
2.05129743e-01 -6.48758054e-01 -1.05446327e+00 9.16747928e-01
-1.52070963e+00 -1.74125564e+00 -6.34860694e-01 6.44187689e-01
5.47836423e-01 -3.23634535e-01 7.97053635e-01 3.37844580e-01
-3.52348357e-01 6.78077340e-01 -2.33506933e-01 4.18787539e-01
8.04149270e-01 -8.42484891e-01 6.84797227e-01 5.46413779e-01
6.92452714e-02 8.07089925e-01 8.43934789e-02 -7.38024712e-01
-1.45062411e+00 -1.06484747e+00 5.05581021e-01 -2.57191896e-01
2.78148979e-01 -6.17150128e-01 -8.75635624e-01 5.69032729e-01
5.30660868e-01 2.81899124e-01 3.80799353e-01 -4.22883898e-01
-5.75558364e-01 -3.46115291e-01 -1.14692414e+00 8.28306198e-01
1.33408654e+00 -6.84638083e-01 -3.37923676e-01 -1.19897574e-02
3.65758002e-01 -3.13840449e-01 -7.76930034e-01 2.97354639e-01
1.13250327e+00 -9.59479034e-01 9.64571655e-01 -3.54796052e-01
8.94420326e-01 -3.37049484e-01 -2.33307332e-02 -1.08823287e+00
-5.22846997e-01 -7.73675084e-01 2.18084604e-02 1.46230602e+00
4.53051776e-02 -4.97239172e-01 8.25002015e-01 4.50474024e-01
1.61485001e-01 -8.15820575e-01 -5.63220501e-01 -5.37060022e-01
2.89535820e-01 5.28249070e-02 1.06333935e+00 7.14005053e-01
-4.63678092e-01 -8.92661959e-02 -4.17747945e-01 -7.25647062e-02
7.33931124e-01 2.15679005e-01 9.95068550e-01 -1.17813289e+00
7.48870820e-02 -6.73103869e-01 7.87741542e-02 -1.03342533e+00
3.52380842e-01 -7.28349090e-01 -2.24350855e-01 -1.52343369e+00
1.11478128e-01 -4.07725662e-01 -1.48117002e-02 6.80118263e-01
-5.22640124e-02 7.00223327e-01 4.50293571e-01 4.04103063e-02
-1.08186625e-01 8.17581713e-01 1.69497859e+00 -1.68048799e-01
7.34319091e-02 -3.21693093e-01 -7.13739157e-01 6.57954991e-01
6.91811502e-01 -1.71577528e-01 -2.67657280e-01 -4.75571036e-01
-3.00421380e-02 7.46199340e-02 5.17904103e-01 -8.10526431e-01
1.94878206e-01 -1.80640116e-01 8.79501224e-01 -4.79513288e-01
5.26211321e-01 -7.92077899e-01 3.00892889e-01 3.45957786e-01
-2.62574613e-01 -1.05898626e-01 3.72012913e-01 3.52329373e-01
-1.57594740e-01 1.54251426e-01 6.87923133e-01 -1.64332449e-01
-2.75502592e-01 7.84361601e-01 5.48157953e-02 -3.94177198e-01
6.50244594e-01 -9.12405252e-02 -1.88524157e-01 -5.54884911e-01
-4.65463877e-01 -9.63681936e-02 7.73178041e-01 5.75779796e-01
9.64050889e-01 -1.98513949e+00 -7.55464792e-01 6.98517561e-01
-2.39458494e-02 -1.53702319e-01 2.59800375e-01 3.16483140e-01
-7.38714755e-01 1.04109123e-01 -6.57454908e-01 -2.88256079e-01
-1.03004968e+00 4.52212602e-01 2.29917616e-01 3.03812861e-01
-6.97493851e-01 7.54700124e-01 8.34069490e-01 -3.99287939e-01
2.20292702e-01 -2.03731373e-01 1.93527371e-01 -2.50730604e-01
7.44497955e-01 1.32512569e-01 -3.25282633e-01 -3.74016583e-01
-1.69375226e-01 1.02393782e+00 1.06386647e-01 -1.73835427e-01
1.46599841e+00 2.44988516e-01 -3.75889510e-01 -8.74225143e-03
1.05760229e+00 2.97771424e-01 -1.77930140e+00 -3.11957836e-01
-4.89890218e-01 -7.67073333e-01 -2.07374007e-01 -7.68184841e-01
-1.47728121e+00 1.24434686e+00 2.99596518e-01 -2.07246408e-01
1.22405529e+00 -2.81893909e-01 8.94948840e-01 1.18868515e-01
2.74483830e-01 -7.99720168e-01 4.89084870e-01 2.59939313e-01
1.57209885e+00 -9.04921532e-01 -7.87618905e-02 -5.19181311e-01
-5.00904381e-01 1.56809700e+00 5.59975088e-01 -5.69502831e-01
7.67472684e-01 3.74049425e-01 -5.75918071e-02 -1.83154315e-01
-5.15015066e-01 2.29924880e-02 5.41044950e-01 4.60980535e-01
4.54309493e-01 -4.28056382e-02 -2.15327814e-01 5.55772781e-01
8.97694603e-02 4.93221968e-01 1.87714010e-01 7.20990360e-01
-7.58462101e-02 -1.46440363e+00 -3.13378185e-01 2.12279841e-01
-1.39791787e-01 -1.63331538e-01 -6.63249016e-01 6.41211450e-01
2.67092973e-01 5.14448404e-01 1.40392557e-01 -2.95315444e-01
3.50312859e-01 1.07689835e-02 7.47331381e-01 -4.66044962e-01
-3.61184895e-01 2.40674824e-01 -2.26215333e-01 -6.79496408e-01
-3.24284703e-01 -2.42895722e-01 -9.05125678e-01 -5.85190058e-01
2.26915330e-01 -2.81120896e-01 5.12628019e-01 5.92173338e-01
5.67127287e-01 3.04034233e-01 8.30378175e-01 -1.39096868e+00
6.50798678e-02 -8.05125117e-01 -5.30464113e-01 3.88328373e-01
3.30518872e-01 -7.64292538e-01 -1.02269307e-01 3.54006708e-01] | [12.348952293395996, -0.12106359004974365] |
7df1fac5-20f5-4567-ad21-1d7cfa143d65 | end-to-end-trainable-self-attentive-shallow | 2008.06146 | null | https://arxiv.org/abs/2008.06146v1 | https://arxiv.org/pdf/2008.06146v1.pdf | End-to-End Trainable Self-Attentive Shallow Network for Text-Independent Speaker Verification | Generalized end-to-end (GE2E) model is widely used in speaker verification (SV) fields due to its expandability and generality regardless of specific languages. However, the long-short term memory (LSTM) based on GE2E has two limitations: First, the embedding of GE2E suffers from vanishing gradient, which leads to performance degradation for very long input sequences. Secondly, utterances are not represented as a properly fixed dimensional vector. In this paper, to overcome issues mentioned above, we propose a novel framework for SV, end-to-end trainable self-attentive shallow network (SASN), incorporating a time-delay neural network (TDNN) and a self-attentive pooling mechanism based on the self-attentive x-vector system during an utterance embedding phase. We demonstrate that the proposed model is highly efficient, and provides more accurate speaker verification than GE2E. For VCTK dataset, with just less than half the size of GE2E, the proposed model showed significant performance improvement over GE2E of about 63%, 67%, and 85% in EER (Equal error rate), DCF (Detection cost function), and AUC (Area under the curve), respectively. Notably, when the input length becomes longer, the DCF score improvement of the proposed model is about 17 times greater than that of GE2E. | ['Jungbae Park', 'Sang Wan Lee', 'Hyeonmook Park'] | 2020-08-14 | null | null | null | null | ['text-independent-speaker-verification'] | ['speech'] | [-3.25995870e-02 1.57920159e-02 1.16685636e-01 -5.96008241e-01
-8.57483327e-01 -2.33555049e-01 3.08835834e-01 -1.83884740e-01
-4.97242242e-01 3.95048082e-01 3.33628654e-01 -5.00896156e-01
2.62396544e-01 -2.01343372e-01 -3.52852285e-01 -6.51579738e-01
-1.63492244e-02 -2.69994438e-01 1.58144124e-02 -2.47789323e-01
-2.25828812e-01 3.12535495e-01 -1.16555846e+00 4.42483872e-02
9.40767109e-01 1.17143345e+00 6.51407838e-02 5.14543533e-01
-5.58997020e-02 3.44808310e-01 -7.56980062e-01 -5.53970635e-01
-1.57066822e-01 -3.69175941e-01 -4.72472072e-01 -9.36192945e-02
3.81934345e-01 -3.36856693e-01 -6.59976363e-01 1.11156809e+00
9.47786152e-01 2.44473577e-01 4.53307897e-01 -1.12634754e+00
-9.00357187e-01 7.11416721e-01 -2.49559805e-01 1.65989131e-01
7.93644711e-02 2.60534972e-01 7.84758270e-01 -1.23394895e+00
1.66327834e-01 1.26883662e+00 7.25956023e-01 9.85886931e-01
-9.18235660e-01 -8.39903057e-01 4.16575551e-01 2.65659034e-01
-1.56937397e+00 -7.96867549e-01 6.66345000e-01 -1.25039414e-01
1.01945078e+00 1.97003648e-01 2.27322623e-01 1.15643823e+00
1.14359573e-01 9.07476306e-01 7.67576516e-01 -3.12715441e-01
1.54453009e-01 5.02454638e-01 4.35740411e-01 6.56065822e-01
-3.88587683e-01 3.61670196e-01 -5.34656823e-01 5.83712831e-02
3.57425272e-01 -1.81890413e-01 -4.16922599e-01 3.52024823e-01
-1.01412344e+00 7.36891031e-01 6.10880435e-01 3.75336349e-01
-2.21830428e-01 -1.89237535e-01 6.69216335e-01 4.18588668e-01
5.15255153e-01 8.69066920e-03 -3.90237957e-01 -1.57359883e-01
-8.97451460e-01 -1.21285610e-01 3.50588530e-01 6.87973976e-01
1.66260704e-01 8.32449555e-01 -4.89255667e-01 1.04822314e+00
5.26312590e-01 6.13484144e-01 9.14732695e-01 -1.83378577e-01
6.34092569e-01 1.66107059e-01 -2.68592834e-01 -7.87695408e-01
-2.92331964e-01 -9.05092895e-01 -1.07834268e+00 -1.68448925e-01
-7.40865767e-02 -3.63508642e-01 -8.75827849e-01 2.11249876e+00
4.03947458e-02 1.71996266e-01 5.69956303e-01 9.67658699e-01
1.25311708e+00 9.69527125e-01 3.35055083e-01 -2.37135813e-01
1.42014360e+00 -1.12195873e+00 -9.86764193e-01 -4.50134486e-01
4.41851705e-01 -4.97526348e-01 1.00868523e+00 4.95341681e-02
-8.26229513e-01 -7.33646095e-01 -1.30570161e+00 4.95412685e-02
-1.74366087e-01 2.59844840e-01 2.13922173e-01 9.40583110e-01
-1.28069329e+00 2.86196113e-01 -5.22795379e-01 -2.68371254e-01
2.05095187e-01 3.77922177e-01 -3.33250672e-01 9.34951827e-02
-1.58943951e+00 7.94535279e-01 2.66471475e-01 5.81754565e-01
-8.82539213e-01 -6.02991462e-01 -1.03155863e+00 3.08648437e-01
-3.02979983e-02 -2.30603561e-01 1.15230620e+00 -6.83024645e-01
-2.02034235e+00 4.90874052e-01 -4.67867136e-01 -5.11752903e-01
2.51221508e-01 -4.64652851e-02 -9.43336785e-01 -3.32614779e-01
-3.93427044e-01 5.66503525e-01 6.94161296e-01 -7.40541279e-01
-1.18812263e-01 -5.91665685e-01 -3.06012690e-01 1.63340777e-01
-8.06007147e-01 3.46015573e-01 -2.60078669e-01 -5.79210877e-01
1.96328074e-01 -8.30025554e-01 2.49210168e-02 -1.53673097e-01
-3.98610145e-01 -4.52215165e-01 1.01674998e+00 -1.22798347e+00
1.40195358e+00 -2.68620276e+00 -1.77085042e-01 -8.30737129e-02
5.27346767e-02 7.66368866e-01 -3.05755079e-01 9.39152241e-02
-4.56989817e-02 8.46852064e-02 -3.31826925e-01 -5.04565597e-01
1.42384976e-01 -2.30912000e-01 -2.66520828e-01 3.99204254e-01
1.07870854e-01 9.29879963e-01 -6.72253788e-01 -2.37446696e-01
1.13963649e-01 7.80305326e-01 -6.19112253e-02 1.20116420e-01
2.10254952e-01 2.24332690e-01 -1.48173153e-01 3.92478138e-01
9.43900943e-01 1.75642505e-01 -8.71467888e-02 -6.00457527e-02
-7.46438876e-02 3.27998161e-01 -1.08435452e+00 1.59167862e+00
-6.03426278e-01 9.29840863e-01 1.17428347e-01 -5.41832983e-01
1.31906366e+00 7.95003951e-01 -9.36853811e-02 -7.61727333e-01
2.20943376e-01 3.16899151e-01 -3.74675281e-02 -3.02559227e-01
2.68399239e-01 -3.74849260e-01 5.92895271e-03 2.70354420e-01
2.26229444e-01 2.93574095e-01 -5.02612054e-01 4.89684008e-02
6.17739201e-01 -4.42425787e-01 4.40219007e-02 -5.70966415e-02
8.45522940e-01 -9.27295625e-01 6.75443292e-01 3.44517499e-01
-6.11487091e-01 4.46264893e-01 8.64522755e-02 -8.41777492e-03
-7.55225301e-01 -9.61173475e-01 -2.92053759e-01 8.70763004e-01
-8.48436430e-02 -1.59372866e-01 -8.48044515e-01 -5.47008455e-01
-1.35193542e-01 8.33823621e-01 -2.19083443e-01 -5.11026084e-01
-4.98817593e-01 -6.20951474e-01 1.03658128e+00 6.32992387e-01
9.90766168e-01 -9.50786114e-01 1.95166662e-01 3.59032065e-01
-2.01900750e-01 -1.28770721e+00 -9.32807148e-01 -2.15131715e-01
-7.93199241e-01 -3.06515604e-01 -9.54055548e-01 -1.03300631e+00
4.31710869e-01 1.47129565e-01 5.36315382e-01 -1.78901851e-01
1.74095303e-01 -1.80792153e-01 -1.21257752e-01 -3.45418870e-01
-4.16340739e-01 -1.37008997e-02 3.73150200e-01 2.85564423e-01
5.80447078e-01 -1.40901536e-01 -4.49254394e-01 4.01686758e-01
-5.14921546e-01 -2.55048275e-01 4.00458664e-01 1.06421661e+00
8.74222219e-02 -1.47981033e-01 1.04101562e+00 -2.52414703e-01
6.87561333e-01 -1.07518822e-01 -4.85993624e-01 2.68049330e-01
-7.41700172e-01 -7.53222331e-02 6.70657694e-01 -3.65213424e-01
-1.23455763e+00 -2.91375160e-01 -6.85457170e-01 -3.83581251e-01
1.72929272e-01 3.98871750e-01 -3.99099648e-01 -5.20045720e-02
4.87208098e-01 5.79436541e-01 1.66863754e-01 -4.98060644e-01
1.76788792e-01 1.29447007e+00 4.40668553e-01 2.04620864e-02
6.46698475e-01 -2.54816562e-01 -5.99174678e-01 -1.05418372e+00
-4.33693588e-01 -1.89410627e-01 -2.08110213e-01 -1.19980231e-01
7.10962653e-01 -1.13571060e+00 -9.67827618e-01 7.81094015e-01
-1.17930532e+00 7.81901851e-02 -2.97582410e-02 9.21366513e-01
6.19278476e-02 3.64336342e-01 -8.24234784e-01 -1.18652475e+00
-9.31502759e-01 -1.13862741e+00 6.92201376e-01 3.84132802e-01
1.25862837e-01 -8.50745499e-01 -3.01274657e-01 4.28167582e-01
8.93835783e-01 -1.13980949e-01 7.57119715e-01 -7.56591201e-01
-2.05054730e-01 -3.61208647e-01 -2.52018929e-01 8.88635814e-01
-1.04401886e-01 -1.96918011e-01 -1.42402864e+00 -6.10630929e-01
2.40465403e-01 -1.11825608e-01 8.69577944e-01 3.75453800e-01
1.03568864e+00 -4.68810290e-01 -3.15608047e-02 5.55531144e-01
1.22961879e+00 4.21981752e-01 5.53493321e-01 -6.94746897e-02
4.52844471e-01 4.72509354e-01 2.44308800e-01 8.13606828e-02
4.44878429e-01 8.94701421e-01 -2.62477025e-02 -1.35529652e-01
-2.78059155e-01 -3.22027713e-01 9.26096320e-01 1.22805762e+00
3.26413125e-01 -3.44610125e-01 -6.48644507e-01 5.71486771e-01
-1.39761066e+00 -9.31082606e-01 1.10777952e-01 2.29366875e+00
6.19751930e-01 1.47198051e-01 6.38607889e-02 2.49506995e-01
9.61739361e-01 3.69583786e-01 -7.51075387e-01 -7.35296786e-01
-3.03192824e-01 -1.30494133e-01 1.20949395e-01 5.51807702e-01
-9.19004738e-01 9.22521889e-01 5.77833986e+00 8.42727423e-01
-1.62539732e+00 4.50979531e-01 5.79517722e-01 8.74274597e-02
-6.46462962e-02 -5.14582217e-01 -1.05566704e+00 5.72765827e-01
1.41759264e+00 -2.18381092e-01 2.22470134e-01 8.57801974e-01
1.59193680e-01 6.11023366e-01 -8.71983469e-01 1.08734882e+00
2.52818167e-01 -9.06953752e-01 -2.48888746e-01 -6.67881817e-02
3.70007306e-01 3.53835672e-02 5.34848928e-01 7.35838175e-01
-1.85204417e-01 -9.36039269e-01 6.05309486e-01 -3.89454886e-02
1.10522652e+00 -7.57796168e-01 1.06708145e+00 2.59670079e-01
-1.21232557e+00 -2.05186814e-01 -4.12780106e-01 2.22558632e-01
2.61455268e-01 4.90250498e-01 -9.62686658e-01 4.60753471e-01
3.07055950e-01 3.74306053e-01 -3.17768693e-01 8.84456456e-01
-2.57482171e-01 7.84535587e-01 -1.53691143e-01 -3.16230863e-01
2.94181705e-01 1.45388037e-01 6.20969236e-01 1.38015449e+00
3.72426033e-01 -1.44347595e-02 -3.09988379e-01 6.34012520e-01
-3.07887614e-01 2.28628904e-01 -3.31670076e-01 -3.78617756e-02
5.20380676e-01 9.53180671e-01 4.93779369e-02 -2.70615041e-01
-2.76137650e-01 9.97489095e-01 1.17636934e-01 6.60357416e-01
-8.87905598e-01 -7.68666923e-01 6.37160540e-01 -2.60474324e-01
3.16009194e-01 -3.69340368e-02 -1.91747472e-01 -1.19731843e+00
4.09366667e-01 -8.90268385e-01 1.11364029e-01 -4.46053505e-01
-1.09943330e+00 1.29975295e+00 -4.52778012e-01 -1.18295097e+00
-2.04244748e-01 -4.13251936e-01 -7.72929728e-01 1.16106248e+00
-1.50147319e+00 -1.07940078e+00 -1.91743374e-01 4.26558524e-01
6.40613973e-01 -4.84360486e-01 1.02284241e+00 5.81667721e-01
-1.02922583e+00 1.51263309e+00 3.39249760e-01 3.08228761e-01
4.37127173e-01 -8.76416147e-01 6.31709337e-01 1.04821134e+00
-4.05823626e-02 7.04610288e-01 3.78237218e-01 -1.70888335e-01
-1.16690218e+00 -1.23949337e+00 1.20378828e+00 6.86123595e-02
1.68942302e-01 -6.26192033e-01 -1.00203788e+00 5.14785111e-01
1.28403589e-01 2.21666433e-02 6.31764591e-01 2.28341386e-01
-5.41099548e-01 -4.26149607e-01 -1.33391488e+00 4.81867284e-01
7.63703823e-01 -9.09050822e-01 -4.54110056e-01 6.77474141e-02
1.02673018e+00 -2.88607806e-01 -8.08019936e-01 3.81221563e-01
5.93902946e-01 -6.25510335e-01 7.73222983e-01 -3.41799855e-01
-1.05295926e-01 -1.76277384e-01 -2.80559301e-01 -1.35406494e+00
-2.06496492e-01 -5.23200929e-01 -1.08790964e-01 1.52900600e+00
7.94015288e-01 -9.45816815e-01 5.82674801e-01 6.08504295e-01
-4.47705865e-01 -8.54854286e-01 -1.26214886e+00 -1.05382013e+00
-6.65114149e-02 -3.37946683e-01 8.62427235e-01 7.15722084e-01
-2.17798427e-02 5.85856378e-01 -5.03357410e-01 3.46770585e-01
5.01801908e-01 -3.31098646e-01 3.17799628e-01 -9.02004719e-01
-1.39705300e-01 -3.19209516e-01 -5.61022699e-01 -1.39308655e+00
2.30930671e-01 -9.12823737e-01 1.80525064e-01 -1.31278992e+00
1.03191175e-01 -2.89895117e-01 -6.19204640e-01 3.61140758e-01
-4.27366614e-01 2.44963598e-02 3.02846491e-01 -1.02948859e-01
-1.71173900e-01 1.08195639e+00 9.54149246e-01 -3.21426988e-01
-3.41087908e-01 1.81842774e-01 -5.15071034e-01 3.29482377e-01
6.71055317e-01 -2.20560506e-01 -2.55980790e-01 -5.56489289e-01
-6.62193298e-01 2.69563973e-01 2.91086379e-02 -1.01374173e+00
2.94629693e-01 5.35104454e-01 2.50563145e-01 -5.26424766e-01
6.88751280e-01 -3.47261339e-01 -1.43510848e-01 5.59234917e-01
-3.67349207e-01 -1.42743587e-01 3.87905508e-01 3.84049058e-01
-6.01312160e-01 -1.19637392e-01 8.14035535e-01 3.72430354e-01
-6.70366645e-01 5.09270072e-01 -3.46657604e-01 -2.86818504e-01
7.15630531e-01 -2.37902254e-01 -2.27829680e-01 -5.38262784e-01
-6.02680504e-01 1.63875639e-01 -1.52548835e-01 6.52576029e-01
8.91845345e-01 -1.52719748e+00 -9.86129999e-01 5.15018165e-01
2.24810421e-01 -3.83772343e-01 8.84205520e-01 6.41300678e-01
-3.61023694e-02 8.33565354e-01 2.54489481e-01 -5.68757176e-01
-1.43153334e+00 3.49158019e-01 4.49857205e-01 1.27337137e-02
-5.72005868e-01 1.30871212e+00 2.07478642e-01 -7.26531863e-01
6.67290866e-01 -1.89794868e-01 -2.06762850e-01 -1.93485633e-01
8.05622458e-01 2.25782588e-01 2.76954710e-01 -8.54410946e-01
-6.96989536e-01 4.36027646e-01 -2.10975677e-01 -2.12376744e-01
1.11336195e+00 -1.18788533e-01 3.18936884e-01 3.43964636e-01
1.53754067e+00 -1.72873273e-01 -1.01847041e+00 -6.08970344e-01
-3.15763235e-01 -1.55703157e-01 4.28281099e-01 -8.22538376e-01
-1.14792526e+00 1.26140761e+00 1.06586444e+00 -1.22915789e-01
1.06585550e+00 -2.62309343e-01 1.35404122e+00 1.00538373e-01
1.01270974e-01 -7.86221564e-01 -1.62960216e-01 5.24891973e-01
9.77649331e-01 -1.28022444e+00 -4.57058370e-01 -8.52363780e-02
-8.72951746e-01 8.85491073e-01 6.41324520e-01 4.33647841e-01
7.26583660e-01 -1.36957079e-01 2.66794175e-01 1.53785586e-01
-6.54521227e-01 2.17799261e-01 3.31642151e-01 3.80159646e-01
3.51345927e-01 2.12475389e-01 -1.81093648e-01 7.99050331e-01
-2.69987345e-01 -2.54970968e-01 2.07135141e-01 4.10220236e-01
-2.99328506e-01 -7.40850329e-01 -2.47765332e-01 1.13736056e-01
-3.82775962e-01 -2.18382478e-01 3.34453420e-04 3.15927953e-01
-1.60350695e-01 1.29503989e+00 -4.91609573e-02 -8.46263826e-01
5.14220774e-01 2.67558604e-01 7.92894736e-02 -3.53461295e-01
-6.28809631e-01 -1.23775832e-01 2.76268031e-02 -1.94337860e-01
1.04152329e-01 -3.38515401e-01 -1.06881690e+00 -3.36174697e-01
-7.09694028e-01 2.42842227e-01 1.17625189e+00 9.27070498e-01
5.35244524e-01 5.97421050e-01 1.05185938e+00 -3.10615540e-01
-8.75727475e-01 -1.41225898e+00 -6.53218150e-01 8.31222460e-02
6.82308435e-01 -4.25262421e-01 -5.38292825e-01 -3.67842376e-01] | [14.340799331665039, 6.07287073135376] |
17c57ed4-b0ff-4071-b457-4f1dec04a4d4 | soil-moisture-estimation-from-sentinel-1 | 2210.10665 | null | https://arxiv.org/abs/2210.10665v1 | https://arxiv.org/pdf/2210.10665v1.pdf | Soil moisture estimation from Sentinel-1 interferometric observations over arid regions | We present a methodology based on interferometric synthetic aperture radar (InSAR) time series analysis that can provide surface (top 5 cm) soil moisture (SSM) estimations. The InSAR time series analysis consists of five processing steps. A co-registered Single Look Complex (SLC) SAR stack as well as meteorological information are required as input of the proposed workflow. In the first step, ice/snow-free and zero-precipitation SAR images are identified using meteorological data. In the second step, construction and phase extraction of distributed scatterers (DSs) (over bare land) is performed. In the third step, for each DS the ordering of surface soil moisture (SSM) levels of SAR acquisitions based on interferometric coherence is calculated. In the fourth step, for each DS the coherence due to SSM variations is calculated. In the fifth step, SSM is estimated by a constrained inversion of an analytical interferometric model using coherence and phase closure information. The implementation of the proposed approach is provided as an open-source software toolbox (INSAR4SM) available at www.github.com/kleok/INSAR4SM. A case study over an arid region in California/Arizona is presented. The proposed workflow was applied in Sentinel- 1 (C-band) VV-polarized InSAR observations. The estimated SSM results were assessed with independent SSM observations from a station of the International Soil Moisture Network (ISMN) (RMSE: 0.027 $m^3/m^3$ R: 0.88) and ERA5-Land reanalysis model data (RMSE: 0.035 $m^3/m^3$ R: 0.71). The proposed methodology was able to provide accurate SSM estimations at high spatial resolution (~250 m). A discussion of the benefits and the limitations of the proposed methodology highlighted the potential of interferometric observables for SSM estimation over arid regions. | ['Vassilia Karathanassi', 'Kleanthis Karamvasis'] | 2022-10-18 | null | null | null | null | ['soil-moisture-estimation'] | ['computer-vision'] | [ 4.59808975e-01 -2.39171416e-01 4.36201006e-01 -1.26622275e-01
-6.97224140e-01 -5.48679709e-01 5.75598598e-01 2.10270450e-01
-2.28304267e-01 1.32554436e+00 -1.80562407e-01 -5.27802646e-01
-4.73636180e-01 -1.18795919e+00 -3.21442276e-01 -1.05634356e+00
-7.41118550e-01 4.96776551e-01 -1.59429051e-02 -7.57503927e-01
-2.85824910e-02 8.77941012e-01 -1.46238792e+00 -1.75308913e-01
1.35411704e+00 7.26758838e-01 5.45759201e-01 6.20209932e-01
2.78654635e-01 1.52765408e-01 -1.42699137e-01 5.83478391e-01
3.75064224e-01 -3.83291483e-01 -3.29831511e-01 -1.27586916e-01
2.57015437e-01 -3.11921388e-01 4.98608410e-01 1.07222009e+00
4.26974595e-01 -1.76038250e-01 5.32819450e-01 -4.86058533e-01
5.96828759e-01 2.70930588e-01 -7.79247522e-01 9.67869386e-02
3.69293317e-02 -7.50366300e-02 6.06914461e-01 -9.46996450e-01
5.21340370e-01 7.06224561e-01 5.89838982e-01 -7.01892912e-01
-1.42777216e+00 -6.48219109e-01 -3.42654437e-01 -6.89412430e-02
-1.50810671e+00 -4.05166745e-01 3.12158495e-01 -5.72524071e-01
7.24626422e-01 5.47072470e-01 1.17220736e+00 1.84776023e-01
5.78323066e-01 4.86555696e-03 1.68941057e+00 -4.38840479e-01
4.01801914e-01 -2.43893057e-01 3.66410464e-01 2.00037301e-01
1.06443202e+00 5.02872288e-01 -2.53570795e-01 -1.67796478e-01
4.72328633e-01 -2.17307672e-01 -2.76845574e-01 -1.17782287e-01
-8.77562702e-01 9.60927844e-01 5.59381902e-01 3.97562295e-01
-1.10117817e+00 -2.27997050e-01 4.90939524e-03 4.15038884e-01
7.65790403e-01 2.47534603e-01 -4.47279274e-01 3.14288348e-01
-1.50578344e+00 6.83066726e-01 5.24527073e-01 4.37752217e-01
8.96425247e-01 6.48437977e-01 5.14687955e-01 3.75277758e-01
7.17142045e-01 1.68580389e+00 -1.51734799e-01 -5.33596873e-01
1.24916941e-01 3.22599947e-01 5.84223032e-01 -9.17003572e-01
-4.48529989e-01 -8.62025321e-01 -1.10630751e+00 3.93864334e-01
-1.08188197e-01 -5.60395837e-01 -8.75228286e-01 1.23583817e+00
2.21086830e-01 -6.50012046e-02 5.66442311e-01 7.80433714e-01
2.70822585e-01 7.93787241e-01 4.16512154e-02 -6.54187083e-01
1.60169780e+00 1.22698620e-01 -8.84661674e-01 -6.21932268e-01
5.88177145e-01 -1.04480672e+00 -4.08425480e-02 1.22661948e-01
-7.51986861e-01 -2.83182710e-01 -1.16021109e+00 1.10166538e+00
-4.11704868e-01 3.64761472e-01 5.31011522e-01 3.93084079e-01
-9.97380257e-01 5.10583639e-01 -1.15203464e+00 -5.86879075e-01
4.42452952e-02 1.48635969e-01 -2.25600317e-01 4.56043370e-02
-1.36511695e+00 1.02436268e+00 3.60477775e-01 9.54060674e-01
-3.68905872e-01 -5.60346723e-01 -1.02475512e+00 -1.27326965e-01
-4.04493153e-01 -4.65758264e-01 4.10374522e-01 -9.94273305e-01
-1.04165733e+00 7.41655231e-01 -3.00370455e-01 -7.56034970e-01
2.23349124e-01 -4.57434833e-01 -6.15105927e-01 2.98040092e-01
3.71889859e-01 1.99727416e-02 4.83570635e-01 -1.44410646e+00
-4.27191138e-01 -7.30663776e-01 -6.05491221e-01 2.38515198e-01
6.52183712e-01 -8.83793086e-02 7.09072649e-01 -3.96508366e-01
8.96622896e-01 -1.00076604e+00 -3.11855465e-01 -6.94454253e-01
-3.32323276e-02 1.03965259e+00 6.04187846e-01 -1.00346732e+00
1.04574966e+00 -1.90421319e+00 -2.04035461e-01 6.48116469e-01
-2.87567049e-01 3.64079177e-01 1.98416058e-02 8.82698894e-01
-1.69394568e-01 -2.56283909e-01 -8.51000011e-01 2.84268349e-01
-5.32598257e-01 1.32160768e-01 -5.13796508e-01 7.48051286e-01
1.33687317e-01 4.04084206e-01 -6.70586646e-01 -8.35290700e-02
5.16241133e-01 3.45048189e-01 -4.70244139e-02 -4.64371368e-02
2.42682900e-02 5.49646080e-01 -4.69961941e-01 7.03434467e-01
1.93228328e+00 2.57466763e-01 6.24436975e-01 -1.53352544e-01
-9.70523655e-01 -4.49090526e-02 -1.46194255e+00 9.60956573e-01
-3.19299281e-01 6.04878485e-01 4.62962806e-01 -9.71786857e-01
1.22090340e+00 3.32321584e-01 3.31671119e-01 -6.22907996e-01
-3.54185671e-01 7.78601944e-01 3.90851855e-01 -4.52580363e-01
6.14849389e-01 -4.23562855e-01 2.72507995e-01 1.12127913e-02
-3.96480143e-01 -5.07892251e-01 -1.29666748e-02 -1.44521907e-01
5.27014256e-01 3.50378394e-01 6.72640562e-01 -8.77463639e-01
6.32610202e-01 5.59496224e-01 3.11615944e-01 4.60427165e-01
2.19265461e-01 1.81373850e-01 1.58061162e-01 -3.24227065e-01
-8.73698592e-01 -1.07596183e+00 -7.58439600e-01 -8.04037694e-03
-1.09365806e-01 2.12243110e-01 -1.35478929e-01 3.55758548e-01
8.41118097e-02 7.57827103e-01 -3.93940687e-01 5.03012002e-01
-3.83682430e-01 -1.50893199e+00 1.22640364e-01 -7.60971680e-02
8.10211957e-01 -7.54266918e-01 -8.38295281e-01 4.45283681e-01
-2.27573693e-01 -8.88151705e-01 9.42646384e-01 3.68087649e-01
-1.57238126e+00 -7.41678178e-01 -4.95955616e-01 -1.52828291e-01
5.91383517e-01 3.81353617e-01 1.01714349e+00 -4.10650998e-01
-1.01600952e-01 -2.14685634e-01 -4.44099218e-01 -3.61830175e-01
-2.34691739e-01 -2.97502756e-01 -2.14183703e-01 -4.41892259e-02
1.75176978e-01 -8.37651193e-01 -7.78119147e-01 2.94413865e-01
-6.87019110e-01 2.88956575e-02 6.42448425e-01 5.69995940e-01
3.69647324e-01 -1.81822494e-01 4.73369211e-01 -7.67131984e-01
-1.45812826e-02 -6.83758080e-01 -1.36779356e+00 -1.66493610e-01
-4.58339959e-01 -2.78637081e-01 -1.97829120e-02 3.72877330e-01
-1.12557280e+00 1.42479325e-02 -4.82345745e-02 3.76361340e-01
-2.95092523e-01 1.42365158e+00 1.69240877e-01 -2.59822682e-02
5.32592297e-01 5.00677228e-01 1.96455061e-01 -4.25972581e-01
-5.10190167e-02 4.94366050e-01 1.74828380e-01 -1.63537264e-01
1.06117380e+00 6.58168077e-01 3.94672573e-01 -1.57537544e+00
-2.89015800e-01 -5.18857777e-01 -6.13870203e-01 -2.63001591e-01
5.58950543e-01 -1.29864120e+00 -8.91360417e-02 8.30366671e-01
-7.82601833e-01 -2.06100062e-01 -3.45173851e-02 9.50056016e-01
-2.47192159e-01 2.37419084e-01 1.53615102e-01 -1.37283051e+00
-6.42433345e-01 -7.75260329e-01 7.73191988e-01 6.99906871e-02
-3.02151501e-01 -8.15612316e-01 3.29050779e-01 1.97222590e-01
7.98855901e-01 7.64245033e-01 5.64191461e-01 -7.16699064e-02
-7.74370074e-01 -3.30016047e-01 -3.07200730e-01 3.07294458e-01
2.77631462e-01 -3.23430845e-03 -8.51813972e-01 -3.70090365e-01
-1.79581344e-02 1.32920787e-01 8.11695457e-01 1.09950125e+00
-1.97867602e-01 -1.55286089e-01 -2.26051107e-01 6.01250410e-01
2.15500808e+00 2.20936999e-01 8.91644835e-01 5.81697226e-01
-1.34113088e-01 4.36047077e-01 1.21634483e+00 6.46083117e-01
-1.54632851e-01 3.99344027e-01 6.52643561e-01 -2.74229079e-01
2.31193751e-01 5.48603296e-01 3.17615271e-01 2.95292109e-01
-5.51124573e-01 1.26561224e-01 -1.09475303e+00 6.64418876e-01
-1.59383905e+00 -1.20305586e+00 -8.81255269e-01 2.48916030e+00
1.41383484e-01 -3.09399720e-02 -5.28779984e-01 2.26555154e-01
3.27391922e-01 6.70420945e-01 1.50670975e-01 -2.79523134e-01
-3.54887307e-01 8.02947819e-01 1.06786346e+00 8.92451108e-01
-1.04644382e+00 8.16800594e-01 4.03041697e+00 1.06737182e-01
-1.32453251e+00 -9.28847715e-02 -2.09294453e-01 5.36655903e-01
-3.39852810e-01 4.34006542e-01 -5.38801014e-01 2.61472464e-01
1.19338548e+00 4.30214033e-02 -1.03991255e-01 2.85871446e-01
1.08142686e+00 -1.06210065e+00 5.68045117e-02 2.62504816e-01
-5.69104493e-01 -1.31026173e+00 -1.23051330e-01 2.48905256e-01
7.04517663e-01 5.21164834e-01 -3.80900025e-01 -1.09891146e-01
1.07152328e-01 -5.80774426e-01 3.57713342e-01 8.49665284e-01
8.61125946e-01 -6.39766276e-01 1.30704153e+00 1.26919135e-01
-1.29162133e+00 2.78610438e-01 -7.36991093e-02 -3.98368984e-01
3.16404372e-01 1.26172185e+00 -7.57381201e-01 1.33947384e+00
4.75832999e-01 8.89945149e-01 4.19771783e-02 8.97538245e-01
-2.07250834e-01 8.31828475e-01 -6.34504676e-01 4.35177624e-01
3.44495028e-01 -8.76940131e-01 8.96886885e-01 8.98080349e-01
9.76673245e-01 4.74454790e-01 -2.98992127e-01 6.23048663e-01
7.35828936e-01 6.35427684e-02 -8.38003457e-01 1.53839916e-01
5.10128677e-01 1.12292266e+00 -3.63861531e-01 -1.23265639e-01
1.59898996e-02 4.37554657e-01 -8.46197069e-01 6.01631343e-01
-2.37509355e-01 -1.24453448e-01 9.52944994e-01 5.07368803e-01
3.31224144e-01 -5.49605250e-01 -2.37843573e-01 -8.78907323e-01
-8.35288987e-02 -7.20955789e-01 -1.47013199e-02 -9.85538065e-01
-3.85547131e-01 4.31016654e-01 3.01558703e-01 -1.76471829e+00
-3.47920865e-01 -2.16689110e-01 -6.08696401e-01 1.74568629e+00
-1.89872313e+00 -1.06303191e+00 -5.74789762e-01 -1.91188380e-01
1.73386261e-02 -8.05711225e-02 1.38434160e+00 2.88181063e-02
-7.02756792e-02 -6.26123130e-01 7.81302571e-01 -4.15180504e-01
3.34768295e-01 -1.00277627e+00 2.05668494e-01 1.01879978e+00
-7.45289683e-01 5.25561213e-01 1.12190461e+00 -1.00430357e+00
-1.17126811e+00 -1.06827748e+00 1.21231461e+00 5.73531926e-01
7.92316496e-01 2.43553847e-01 -7.00177610e-01 7.40770996e-01
1.20789669e-01 1.89155251e-01 5.23028612e-01 -1.08595222e-01
3.91283870e-01 -5.25730491e-01 -1.25340009e+00 1.77017733e-01
3.92443761e-02 -4.77621630e-02 -3.33893090e-01 2.58681357e-01
-4.86486219e-02 -3.67226422e-01 -1.05292499e+00 9.65061784e-01
9.15935397e-01 -1.23570883e+00 7.49690056e-01 2.02399734e-02
2.31789440e-01 -5.79089820e-01 -5.90690732e-01 -1.29040098e+00
-1.21739969e-01 -2.62632698e-01 5.57434499e-01 5.35376728e-01
2.43712395e-01 -1.03756595e+00 3.61886293e-01 -4.68150347e-01
6.93802908e-02 -1.46671399e-01 -1.08086097e+00 -5.64933300e-01
-2.21842825e-01 -2.32265458e-01 7.85305277e-02 7.02843487e-01
-6.85754895e-01 -5.14605045e-02 -7.89395422e-02 1.25337720e+00
1.00274181e+00 5.11405528e-01 5.52764773e-01 -1.64410973e+00
2.16834009e-01 2.42088988e-01 -2.65320033e-01 -1.63093776e-01
-3.75512123e-01 -4.09951806e-01 -2.16860414e-01 -1.49744236e+00
-2.21341461e-01 -5.66322982e-01 3.62868275e-04 1.34355426e-01
1.67505741e-01 5.34550771e-02 -6.68274686e-02 4.23104614e-01
8.44228446e-01 4.20424074e-01 6.47591054e-01 6.89014345e-02
-2.60759741e-01 3.00760746e-01 2.02549011e-01 6.11176014e-01
7.36449897e-01 -5.07012069e-01 -1.11233093e-01 -1.38146326e-01
4.73655254e-01 6.64327085e-01 6.51404917e-01 -1.50117612e+00
-4.62323546e-01 -2.21838832e-01 2.42041469e-01 -1.37828326e+00
3.04624051e-01 -9.82872546e-01 8.23457062e-01 1.05313122e+00
5.60013115e-01 -2.07138285e-01 6.01295114e-01 8.29553977e-02
-5.15469134e-01 -2.45296985e-01 1.01833344e+00 -9.42730233e-02
-5.94441772e-01 -6.58060834e-02 -7.12772965e-01 -7.03814864e-01
7.84097731e-01 -3.09705645e-01 -3.63842487e-01 -2.89975673e-01
-7.90633559e-01 2.00994670e-01 3.67176622e-01 -4.92428273e-01
5.01060486e-01 -7.66999006e-01 -1.15506959e+00 5.67495763e-01
1.15036473e-01 -2.46961266e-01 6.59979880e-01 1.14592433e+00
-1.04278898e+00 4.18068290e-01 -4.40497339e-01 -6.92594349e-01
-1.19512832e+00 -4.43176270e-01 6.41918123e-01 -4.32053000e-01
-1.25605255e-01 1.98835090e-01 -3.30348819e-01 -3.49773496e-01
-9.75049853e-01 -2.94913232e-01 -3.76697540e-01 7.08193898e-01
1.32164791e-01 2.39501089e-01 3.33015800e-01 -7.26075113e-01
-6.18402183e-01 8.30672264e-01 4.91329014e-01 -4.79902446e-01
1.88178277e+00 -3.65722239e-01 -5.18251717e-01 4.80787545e-01
7.63637304e-01 3.20810229e-01 -6.13154709e-01 1.12584932e-02
-7.13966116e-02 -1.94032803e-01 4.65633363e-01 -8.30218196e-01
-7.28552818e-01 5.14258623e-01 9.91841793e-01 5.22931665e-02
1.23961782e+00 -7.10600972e-01 1.68655276e-01 2.72257775e-01
1.22554094e-01 -6.51082158e-01 -1.04720473e+00 6.00010216e-01
1.08241475e+00 -1.16051948e+00 7.55986392e-01 -5.11708677e-01
-2.32570067e-01 1.19001710e+00 -3.82680207e-01 -1.36635721e-01
9.30598497e-01 2.43394956e-01 1.82753861e-01 -5.03783047e-01
-8.43151584e-02 -5.33760667e-01 -2.33844906e-01 2.75377333e-01
6.87374830e-01 6.50436878e-01 -8.68211329e-01 -1.00632906e-01
-1.80186018e-01 2.83334583e-01 7.37429678e-01 1.25187421e+00
-7.35673904e-01 -8.43832552e-01 -8.67748201e-01 4.25397098e-01
1.58814654e-01 -3.68299484e-01 4.68182623e-01 1.02218533e+00
-1.37401387e-01 6.79687202e-01 1.85752615e-01 1.25681192e-01
1.17085025e-01 1.95308551e-02 1.68195322e-01 -5.13612688e-01
-2.13212162e-01 2.39560843e-01 4.63638425e-01 -3.28040570e-01
-9.17528272e-01 -1.16321599e+00 -1.01110947e+00 -1.52760193e-01
-1.45415381e-01 5.12497067e-01 1.11026299e+00 7.72360265e-01
1.77441001e-01 2.28042871e-01 8.16845894e-01 -9.06811595e-01
-2.59895295e-01 -1.27957535e+00 -1.36384368e+00 -5.02978742e-01
6.21271193e-01 -6.25261664e-01 -4.90727961e-01 -2.09268168e-01] | [9.471580505371094, -1.662549376487732] |
e1a33e71-b058-483d-a780-a662f9effe56 | run-time-monitors-design-for-adaptive-radar | 2302.09985 | null | https://arxiv.org/abs/2302.09985v1 | https://arxiv.org/pdf/2302.09985v1.pdf | Run-Time Monitors Design for Adaptive Radar Systems: A Practical Framework | Adaptivity in multi-function radar systems is rapidly increasing, especially when moving towards fully adaptive, cognitive radar systems. However, the large number of available system configurations makes the rigorous verification and certification process during the testing phase, deployment, and after hardware and software upgrades, challenging, if not infeasible. To alleviate the verification process, run-time verification can be applied to oversee the correct function of a system during its operation as done in applications where on-the-fly reconfiguration/adaptation is pervasive, e.g., spacecrafts and self-driving cars. Though possible, the application of run-time verification into a radar system is not straightforward, e.g., when verifying (adaptive) radar resource managers or performance measures, such as track initiation time in dynamic environments. The goal of this paper is to introduce a framework to identify, characterize, and map the various aspects necessary for implementing run-time verification for (components of) multi-function radar systems. The proposed framework can be used by radar practitioners and researchers for applying run-time-verification to adaptive, re-configurable radar systems. In addition, we discuss how run-time verification can be leveraged to gather new insights from operational data to improve functionalities in upcoming update cycles and present an example of a verifier designed using the introduced framework. | ['Laura Anitori', 'Ahmad Mouri Sardarabadi', 'Giuseppe Papari', 'Mario Coutino', 'Pepijn Cox'] | 2023-02-20 | null | null | null | null | ['self-driving-cars'] | ['computer-vision'] | [ 2.35142484e-01 -2.17333600e-01 1.76431447e-01 -5.33202648e-01
-2.67113373e-02 -1.05098307e+00 5.02777338e-01 2.40818813e-01
-5.68293873e-03 5.69732308e-01 -5.07440865e-01 -8.31821442e-01
-6.52257621e-01 -8.56328309e-01 -4.58354264e-01 -4.69102412e-01
-4.16340977e-01 6.97580278e-01 3.14380407e-01 -2.16010734e-01
1.03672154e-01 1.14021039e+00 -1.80332828e+00 -2.19729692e-01
4.81774777e-01 1.01890326e+00 3.61176953e-03 8.24419737e-01
3.00008595e-01 1.52485937e-01 -8.20762694e-01 1.27041750e-02
5.56922555e-01 -1.36432439e-01 -1.96020119e-02 2.16914445e-01
7.25653172e-02 -3.80860239e-01 1.19387224e-01 9.80622709e-01
2.78288037e-01 -2.28572905e-01 1.75608784e-01 -1.66459191e+00
-9.57169663e-03 5.57983279e-01 -2.73959395e-02 -4.50183288e-04
1.27419576e-01 3.67664546e-01 5.56340218e-01 -3.32286566e-01
3.44479650e-01 7.47601449e-01 1.25225827e-01 -1.82864264e-01
-1.12677896e+00 -8.17467570e-01 2.28934810e-01 1.55154631e-01
-1.43903196e+00 -6.34370506e-01 5.68666518e-01 -5.73441327e-01
6.92916691e-01 5.10567665e-01 5.32096624e-01 6.13684893e-01
5.59983730e-01 -6.13516197e-02 1.01066399e+00 -3.70468408e-01
6.10215783e-01 1.78302288e-01 3.35126966e-01 3.81475270e-01
9.01532173e-01 5.96262038e-01 1.50295660e-01 -2.32193515e-01
3.59354824e-01 -1.16558187e-01 -1.28846332e-01 -4.56452489e-01
-8.96007836e-01 3.19179624e-01 -1.29642770e-01 3.07257295e-01
-4.73422021e-01 3.50052118e-02 2.97291666e-01 6.77908182e-01
-2.08596006e-01 5.31573534e-01 -4.06462759e-01 -2.98885822e-01
-1.08534431e+00 1.68159768e-01 8.04135919e-01 1.15901089e+00
6.84898615e-01 6.20098531e-01 8.41457099e-02 -2.24658191e-01
3.11965942e-01 7.33021736e-01 -1.58648659e-02 -4.52641070e-01
-1.18013099e-01 6.23768508e-01 3.78340721e-01 -5.60776591e-01
-7.16117680e-01 -8.66048038e-01 -7.11171269e-01 7.34382331e-01
-1.97743535e-01 -2.26765722e-01 -8.34957898e-01 1.38912892e+00
4.29131061e-01 1.21642858e-01 2.62851238e-01 8.46400261e-01
-1.99948996e-01 2.26071835e-01 -2.58737087e-01 -4.17766124e-01
1.45733535e+00 -1.16320342e-01 -5.26315153e-01 -2.90204406e-01
1.09991953e-01 -8.42069328e-01 3.54529262e-01 5.15332937e-01
-5.63186646e-01 -4.90151465e-01 -1.63871658e+00 1.19329536e+00
-1.23519897e-01 -1.69063322e-02 3.42071533e-01 1.22941768e+00
-7.60869801e-01 1.87574215e-02 -8.89996648e-01 -3.77618641e-01
-2.15427637e-01 4.17923510e-01 8.07634220e-02 1.30723611e-01
-1.09837043e+00 1.15370393e+00 1.34060860e-01 2.09039673e-01
-1.20508742e+00 -6.88980758e-01 -3.59537870e-01 1.02199860e-01
4.17788535e-01 -4.81399715e-01 1.20757151e+00 -5.88119805e-01
-1.28484702e+00 1.43384248e-01 4.60468084e-01 -7.91499853e-01
4.91821736e-01 1.84874739e-02 -1.10362375e+00 -1.81045964e-01
-2.84539074e-01 -1.99274242e-01 1.01428485e+00 -9.37371910e-01
-6.89759791e-01 -1.95566058e-01 4.09330964e-01 -4.25118923e-01
2.68177122e-01 2.26857830e-02 3.24687421e-01 -2.15880260e-01
-1.84416026e-01 -1.15461218e+00 -3.26926671e-02 -4.22518522e-01
-1.02406271e-01 7.24135697e-01 8.85092318e-01 -2.00707242e-01
9.68859315e-01 -1.94112372e+00 -1.76430836e-01 7.44316638e-01
-2.95864254e-01 2.74742335e-01 2.15391248e-01 6.15198553e-01
-1.20493107e-01 -4.08977658e-01 -5.47308102e-02 4.79659915e-01
2.68224359e-01 3.52320336e-02 -4.02353168e-01 5.57080090e-01
1.06769495e-01 2.19913960e-01 -2.29660317e-01 7.62225315e-02
4.81181502e-01 8.66074041e-02 -9.99776050e-02 1.00608587e-01
-2.42342919e-01 5.01214504e-01 -5.13657629e-01 9.12889421e-01
5.44979513e-01 1.98839083e-01 4.05473858e-01 -3.04948717e-01
-5.71732044e-01 -5.39624691e-02 -1.46678376e+00 7.29039967e-01
-7.13069320e-01 6.29807532e-01 3.02170306e-01 -8.75765085e-01
1.04918683e+00 3.17209393e-01 3.26629639e-01 -7.50377893e-01
3.39975357e-01 -6.53043669e-03 5.19639850e-01 -4.99089509e-02
6.86830103e-01 -3.87882181e-02 -3.19123715e-01 6.86475098e-01
-5.13560295e-01 9.48963314e-02 3.19747746e-01 -2.51253188e-01
1.34528446e+00 -1.54179558e-01 6.32721424e-01 -2.67182082e-01
8.78897011e-01 1.22209013e-01 4.13509965e-01 5.62593818e-01
-7.30298087e-02 -2.13799596e-01 1.34129077e-01 -2.59236991e-01
-9.99503255e-01 -8.60511839e-01 -2.82328218e-01 5.88567078e-01
8.33520386e-03 -1.71104819e-01 -1.74422652e-01 -7.95834437e-02
1.80592477e-01 9.23508942e-01 -5.42134829e-02 -4.35894102e-01
-4.49523300e-01 -5.65026879e-01 6.10674083e-01 3.25737834e-01
2.55268246e-01 -4.34636265e-01 -1.52007508e+00 6.23085976e-01
4.62759018e-01 -1.09860218e+00 1.19992867e-01 2.72002488e-01
-6.02333724e-01 -1.09331596e+00 2.19035625e-01 -2.19190288e-02
5.55571735e-01 2.78133184e-01 7.75702596e-01 2.14069366e-01
-4.81067806e-01 6.27637863e-01 -4.67134804e-01 -3.93870175e-01
-5.54971755e-01 -3.78853470e-01 4.95113909e-01 2.88722999e-02
-2.09653720e-01 -6.77002430e-01 -2.61917263e-01 8.49084854e-01
-9.15836453e-01 -2.93558091e-01 8.71147871e-01 5.61167181e-01
3.18217814e-01 4.47822928e-01 7.60613382e-01 -6.51440740e-01
6.06385827e-01 -6.72098473e-02 -1.66494429e+00 8.24533463e-01
-9.55735922e-01 7.97555745e-02 7.95885980e-01 -4.36100602e-01
-6.73032880e-01 3.56649980e-02 1.46239728e-01 -1.93807989e-01
-4.07113619e-02 7.75608063e-01 -4.39075947e-01 -4.64286447e-01
5.42213082e-01 9.65764374e-02 -1.22051284e-01 1.61118689e-03
2.36342773e-01 5.67902267e-01 5.99901259e-01 -5.62662959e-01
1.38989007e+00 2.04148456e-01 3.57251257e-01 -6.99903369e-01
-4.00178358e-02 -8.68830010e-02 -1.47302195e-01 -5.23404717e-01
6.09918945e-02 -6.78476751e-01 -1.00647044e+00 5.91138527e-02
-6.77801847e-01 -1.68421149e-01 -2.75162429e-01 4.51927930e-01
-3.01502496e-01 -1.53377671e-02 3.34608525e-01 -1.06194079e+00
-3.94639343e-01 -1.26892364e+00 5.04826963e-01 2.71275371e-01
-2.22511917e-01 -5.61386347e-01 -2.59227715e-02 9.49858725e-02
9.93070900e-01 2.32708648e-01 9.21840549e-01 -7.00928271e-01
-1.15608764e+00 -7.04281390e-01 2.15912566e-01 5.51779754e-02
4.09938619e-02 2.97898173e-01 -6.86486781e-01 -7.45157003e-01
-2.48399414e-02 2.94999421e-01 -1.62570164e-01 1.91519752e-01
4.50741082e-01 -1.44752562e-01 -2.18454346e-01 3.53523701e-01
1.20662415e+00 5.64108253e-01 4.50517446e-01 4.21129435e-01
-4.42821503e-01 3.20813000e-01 1.39659381e+00 8.09255958e-01
-2.30965000e-02 9.69042659e-01 6.55759454e-01 5.52553236e-01
3.21056724e-01 4.90475297e-01 6.17415071e-01 5.53543232e-02
1.43814906e-01 -9.15987864e-02 -8.27826619e-01 4.82874922e-02
-1.44986117e+00 -9.07933772e-01 1.39627874e-01 2.63798141e+00
8.51405933e-02 7.14282095e-01 -1.34004265e-01 1.79640308e-01
7.81346858e-01 -2.61391133e-01 -6.18936837e-01 -3.40227753e-01
3.58923167e-01 2.76021689e-01 7.61224449e-01 4.46020454e-01
-7.74937809e-01 5.06985724e-01 5.45729065e+00 1.07732281e-01
-1.25807166e+00 -2.22406443e-02 -2.55504131e-01 7.38384798e-02
-4.04580310e-02 7.54227161e-01 -8.11192811e-01 1.60574928e-01
1.45743620e+00 -7.32164979e-01 3.33110571e-01 9.75257277e-01
3.77366185e-01 -2.97687799e-01 -1.13957107e+00 4.21079814e-01
-6.61960170e-02 -1.21966422e+00 -5.35949945e-01 6.74842447e-02
4.14203592e-02 -1.20156489e-01 -2.75277108e-01 3.91570091e-01
2.32983306e-01 -5.33746362e-01 9.01487470e-01 5.81073999e-01
6.95882916e-01 -9.62468445e-01 8.67777646e-01 2.46248230e-01
-1.32805800e+00 -1.18975535e-01 -7.95841590e-02 -1.05660535e-01
4.99890715e-01 8.04364443e-01 -1.18922877e+00 9.69267309e-01
2.92527348e-01 -4.14894164e-01 -6.27687395e-01 1.22542572e+00
-1.69181332e-01 2.64322370e-01 -3.82505357e-01 -7.21653998e-02
-3.50211531e-01 -1.20976344e-01 8.35324466e-01 7.87266552e-01
5.83774090e-01 -1.44118160e-01 2.44588643e-01 5.50405085e-01
6.53093815e-01 -6.29069149e-01 -4.18023854e-01 -3.40679824e-01
8.01619709e-01 1.59274364e+00 -7.30905414e-01 -6.33512884e-02
-7.62844235e-02 1.19934537e-01 -4.81426507e-01 1.25847951e-01
-1.19666541e+00 -4.39365119e-01 9.14338052e-01 1.89852938e-01
2.60379761e-01 -8.08498263e-01 -3.50169539e-02 -5.10912955e-01
4.38606404e-02 -8.78187180e-01 7.50382543e-02 -6.24474466e-01
-7.26096630e-01 7.65051842e-01 2.68837810e-01 -1.68612075e+00
-6.74914479e-01 -2.15695247e-01 -4.24050570e-01 5.72233498e-01
-1.19039607e+00 -1.15837932e+00 -4.30390805e-01 3.41910928e-01
-3.14047448e-02 -5.65240741e-01 7.90531278e-01 4.24184650e-01
-3.99767011e-01 4.76807535e-01 -2.86817044e-01 -4.53562319e-01
6.84389472e-01 -6.00216568e-01 2.33970419e-01 1.43030035e+00
-2.28188440e-01 6.59390807e-01 1.26978362e+00 -8.89768183e-01
-1.96435726e+00 -1.12604308e+00 1.37137860e-01 2.18248814e-01
8.90537500e-01 -3.76728266e-01 -4.43708301e-01 7.28414774e-01
1.10712558e-01 -2.09845766e-01 6.10029042e-01 1.63219541e-01
-3.62140201e-02 -5.59405267e-01 -1.11606610e+00 4.58552837e-01
3.76705199e-01 -2.20868483e-01 -4.58052397e-01 1.74773589e-01
4.15792167e-01 -3.91166747e-01 -9.35015917e-01 5.45277119e-01
6.81100368e-01 -8.34273458e-01 6.74722612e-01 -2.37731591e-01
-7.16624200e-01 -1.04562402e+00 -3.45763087e-01 -1.08838820e+00
-3.02673787e-01 -6.88914001e-01 -4.89158854e-02 1.36672211e+00
2.11163789e-01 -8.88351977e-01 3.83597374e-01 5.06781816e-01
-4.46241885e-01 1.89065398e-03 -1.17932343e+00 -1.11287796e+00
-8.62573862e-01 -6.53398991e-01 8.92361343e-01 5.12011945e-01
-2.15246066e-01 1.12659737e-01 -2.03640014e-01 1.07300365e+00
4.37962174e-01 5.87986529e-01 1.11882579e+00 -1.29834032e+00
-5.71165264e-01 -1.10363781e-01 -1.00439942e+00 -2.22866029e-01
-1.18675299e-01 -5.34111857e-01 6.80374503e-02 -1.14419484e+00
-5.49083173e-01 -6.06898427e-01 3.70195881e-02 3.51416767e-01
5.08661568e-01 -3.80197644e-01 2.07519859e-01 9.64346528e-02
-1.74981207e-01 3.00261497e-01 3.59169394e-01 -2.19558045e-01
-5.38258739e-02 6.40885532e-01 -4.08676505e-01 2.10113958e-01
8.21597636e-01 -4.37413752e-01 -4.85962361e-01 2.35285833e-01
2.59340227e-01 3.68086815e-01 5.99864364e-01 -1.59153581e+00
7.65130281e-01 -5.00733256e-01 1.77050103e-02 -7.44866610e-01
2.69612134e-01 -1.50927842e+00 9.90691006e-01 7.40835309e-01
3.67221296e-01 4.95142609e-01 2.96855658e-01 4.28082287e-01
-6.32002503e-02 -2.49737635e-01 5.78630209e-01 4.41938668e-01
-7.13647842e-01 -1.38941407e-02 -6.76775455e-01 -4.90598619e-01
1.59606767e+00 -2.33497307e-01 -7.17440248e-01 -2.69535601e-01
-5.91323435e-01 5.16959429e-01 5.51408827e-01 2.81583548e-01
6.07846975e-01 -9.43424344e-01 -4.31266248e-01 5.30709743e-01
2.27090091e-01 -4.73307043e-01 3.73821914e-01 9.61537123e-01
-4.46253777e-01 3.97382528e-01 -4.98862773e-01 -5.58626890e-01
-1.35784996e+00 5.30646563e-01 3.34472388e-01 -2.67643686e-02
-5.51757514e-01 -9.65676755e-02 -3.10103476e-01 -3.12079675e-03
-9.93977785e-02 -1.71641201e-01 2.32172921e-01 7.05164671e-03
5.56735575e-01 8.96292925e-02 6.71378255e-01 -2.95448173e-02
-8.63480389e-01 2.76841909e-01 -2.70535778e-02 -4.56691235e-01
1.21745014e+00 -2.54042223e-02 -7.33747259e-02 2.08122119e-01
7.04779699e-02 3.21819410e-02 -9.06133950e-01 1.49497390e-01
2.27150396e-01 -3.24034005e-01 9.02504921e-02 -9.16928947e-01
-9.20697987e-01 3.03265393e-01 9.25198555e-01 3.29211354e-01
1.28257549e+00 -4.61297870e-01 1.15023732e-01 6.79082811e-01
9.38535392e-01 -1.04598451e+00 -4.83858913e-01 3.87822926e-01
7.07657754e-01 -4.34883446e-01 5.00023961e-01 -4.16034386e-02
-5.62091410e-01 1.18833542e+00 3.37653905e-01 7.93536287e-03
8.33995998e-01 1.05207384e+00 -2.84686863e-01 -1.23618416e-01
-1.02907121e+00 -2.62733638e-01 -1.51205018e-01 6.31708086e-01
8.32159370e-02 4.24184740e-01 -1.89328343e-01 4.85332906e-01
-3.01810086e-01 -7.02455640e-02 1.02613461e+00 1.31121349e+00
-6.58706665e-01 -1.26840603e+00 -7.26398170e-01 4.00009543e-01
2.91952729e-01 3.36294174e-01 -2.13549837e-01 1.12385499e+00
9.57002211e-03 1.02846062e+00 -2.91516702e-03 -5.61337590e-01
6.11493230e-01 2.18791381e-01 3.55098695e-01 -3.55581224e-01
-5.64693809e-01 -7.99099132e-02 5.68809152e-01 -3.77004683e-01
-3.63872558e-01 -8.51516485e-01 -1.07927597e+00 -1.37319356e-01
-3.43372881e-01 1.60661131e-01 1.03378820e+00 9.71729875e-01
5.09091854e-01 1.07095540e+00 7.55642176e-01 -2.73661345e-01
-8.10894012e-01 -6.87109828e-01 -4.51007634e-01 -5.82140923e-01
1.45435393e-01 -1.03953946e+00 -2.70821661e-01 -2.70818293e-01] | [5.035825729370117, 2.2367210388183594] |
d63a1f64-359f-4794-a4d7-2709f032d63d | greedy-offset-guided-keypoint-grouping-for | 2107.03098 | null | https://arxiv.org/abs/2107.03098v2 | https://arxiv.org/pdf/2107.03098v2.pdf | Greedy Offset-Guided Keypoint Grouping for Human Pose Estimation | We propose a simple yet reliable bottom-up approach with a good trade-off between accuracy and efficiency for the problem of multi-person pose estimation. Given an image, we employ an Hourglass Network to infer all the keypoints from different persons indiscriminately as well as the guiding offsets connecting the adjacent keypoints belonging to the same persons. Then, we greedily group the candidate keypoints into multiple human poses (if any), utilizing the predicted guiding offsets. And we refer to this process as greedy offset-guided keypoint grouping (GOG). Moreover, we revisit the encoding-decoding method for the multi-person keypoint coordinates and reveal some important facts affecting accuracy. Experiments have demonstrated the obvious performance improvements brought by the introduced components. Our approach is comparable to the state of the art on the challenging COCO dataset under fair conditions. The source code and our pre-trained model are publicly available online. | ['Zengfu Wang', 'Jiwei Chen', 'Linhua Xiang', 'Jia Li'] | 2021-07-07 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [-7.16926754e-02 -1.05001532e-01 -8.87062103e-02 -2.95907646e-01
-9.51735556e-01 -4.03563321e-01 5.49467266e-01 4.42665517e-02
-4.76321518e-01 4.95666534e-01 3.63118023e-01 3.79256248e-01
-2.11934507e-01 -3.78385127e-01 -8.95809114e-01 -5.72310627e-01
-2.35677138e-01 6.04950607e-01 1.58661366e-01 -5.00694588e-02
2.40657568e-01 3.87421429e-01 -1.51209915e+00 -6.83948919e-02
6.47491753e-01 8.53557706e-01 -7.67186582e-02 6.73670053e-01
5.92719376e-01 2.55646080e-01 -5.56219161e-01 -7.01375842e-01
3.70186418e-01 -2.12439969e-01 -6.08619511e-01 1.41698211e-01
9.87172425e-01 -4.38359261e-01 -5.20448208e-01 8.31597567e-01
7.29942322e-01 2.48390526e-01 3.70880842e-01 -1.24419570e+00
-3.44254702e-01 3.63274872e-01 -7.63970077e-01 -2.13421192e-02
6.85288548e-01 -1.86255813e-01 9.10897017e-01 -1.13812852e+00
5.76418817e-01 1.21575522e+00 8.44069660e-01 4.13739532e-01
-8.96039307e-01 -4.77735817e-01 1.92096993e-01 4.08701926e-01
-1.83874607e+00 -5.18798292e-01 5.49371123e-01 -2.29541898e-01
4.54800606e-01 4.21204895e-01 7.80025184e-01 1.01955986e+00
1.71960704e-02 7.61090994e-01 6.69834077e-01 -4.20595437e-01
-3.95987332e-02 -1.16601825e-01 4.33643274e-02 7.94885397e-01
3.52370948e-01 -1.31926909e-01 -9.00893509e-01 -2.69746184e-01
7.12814629e-01 6.11119606e-02 -1.51945993e-01 -6.73825204e-01
-1.59226143e+00 5.05177081e-01 5.48617244e-01 -1.85195625e-01
-3.96720678e-01 4.93060857e-01 2.56231695e-01 -1.89393550e-01
1.98844686e-01 2.48184681e-01 -2.03806162e-01 -1.36407718e-01
-9.23980236e-01 7.41994441e-01 4.97386962e-01 1.15445817e+00
6.03800416e-01 -6.50260031e-01 -2.73635268e-01 8.01076233e-01
3.63055587e-01 3.25089216e-01 -7.10010678e-02 -1.01723719e+00
7.37181723e-01 3.57806444e-01 4.92949724e-01 -1.38484693e+00
-4.08255070e-01 -3.65011513e-01 -7.10917592e-01 -2.29015350e-01
6.04868114e-01 -1.26111925e-01 -8.52857172e-01 1.55320144e+00
6.16682053e-01 1.21307410e-01 -3.70495915e-01 1.04904413e+00
6.56445920e-01 3.82153451e-01 -1.22263148e-01 8.74154866e-02
1.62993693e+00 -1.24503028e+00 -6.22177064e-01 -2.90589869e-01
1.66741163e-01 -5.85669398e-01 7.27342069e-01 4.09124911e-01
-1.11279213e+00 -7.16395736e-01 -9.71064210e-01 -1.54405922e-01
-1.41890496e-01 5.92306256e-01 6.76870584e-01 5.11914790e-01
-8.54152858e-01 6.04023397e-01 -8.08040738e-01 -4.59074438e-01
2.18923837e-01 3.57466847e-01 -4.93788540e-01 -2.34698448e-02
-1.00239289e+00 6.02988482e-01 3.75392467e-01 5.07123709e-01
-6.73250079e-01 -3.98642838e-01 -8.52305353e-01 -2.06221998e-01
7.49751210e-01 -7.79784501e-01 1.01693857e+00 -4.18130159e-01
-1.21870196e+00 7.87518620e-01 -4.70437735e-01 -4.43182379e-01
1.00148797e+00 -8.49579394e-01 -1.63595036e-01 4.05995101e-01
3.45052779e-01 9.05115545e-01 8.29194665e-01 -1.35682046e+00
-8.75990152e-01 -3.69628817e-01 5.81460558e-02 4.72553849e-01
-1.09828621e-01 7.30716884e-02 -1.39712155e+00 -7.08084881e-01
3.39160204e-01 -1.26741672e+00 -1.73512354e-01 1.42952828e-02
-9.88782227e-01 -3.25589597e-01 2.57165819e-01 -9.15562093e-01
1.28676963e+00 -1.99320090e+00 4.26091611e-01 4.04015064e-01
2.96082616e-01 -1.53456867e-01 1.94680989e-01 5.26084244e-01
6.29948750e-02 -2.57296026e-01 7.38186166e-02 -6.34308398e-01
1.79353014e-01 -2.26782635e-01 -1.56673510e-02 6.67779863e-01
-2.31172398e-01 9.31229591e-01 -7.70378053e-01 -4.61201280e-01
2.35867798e-01 5.24707019e-01 -4.00572777e-01 1.39102951e-01
2.63505638e-01 3.43661249e-01 -2.64832586e-01 7.15108931e-01
6.14380598e-01 -2.00003073e-01 8.58990550e-02 -4.57110912e-01
5.67066297e-02 2.96134390e-02 -1.56421542e+00 1.79602242e+00
1.21026106e-01 4.17193055e-01 -3.18439752e-01 -3.47165167e-01
5.96527755e-01 1.47906721e-01 4.59854364e-01 -2.15094477e-01
-6.67575654e-03 9.06389132e-02 -2.79454678e-01 -3.08139116e-01
8.81416380e-01 3.80209923e-01 -2.50217229e-01 7.46465102e-02
-1.06692806e-01 5.44177413e-01 2.11190149e-01 2.64857590e-01
5.41183889e-01 4.34745222e-01 3.84772688e-01 -1.78364329e-02
4.24387604e-01 -1.99887484e-01 3.58690083e-01 9.22670245e-01
-3.20625275e-01 9.73427713e-01 4.09350902e-01 -5.72109878e-01
-1.07726455e+00 -9.18256342e-01 2.31441453e-01 1.04536784e+00
5.37467480e-01 -1.02225339e+00 -9.34353650e-01 -5.85043073e-01
-1.93148144e-02 1.82389110e-01 -7.85865366e-01 7.69423693e-02
-7.26229072e-01 -5.34567654e-01 5.43320715e-01 5.97907245e-01
5.98802745e-01 -6.01991594e-01 -6.53434575e-01 -4.13902327e-02
-5.60038745e-01 -1.14397681e+00 -7.80346513e-01 -2.77860373e-01
-3.38615507e-01 -1.06701934e+00 -1.15366518e+00 -7.40296602e-01
8.41696978e-01 4.50094670e-01 7.40829170e-01 2.57315278e-01
-1.08058348e-01 3.15097451e-01 -2.37008184e-01 -1.94507018e-02
2.52092361e-01 1.51860684e-01 3.17033798e-01 6.59163073e-02
4.03743714e-01 -1.17859155e-01 -8.89252484e-01 5.27918935e-01
-1.84735432e-01 1.98801398e-01 5.40756285e-01 5.22255063e-01
8.23487759e-01 -1.00120548e-02 6.59051239e-02 -5.71044505e-01
2.81229794e-01 -3.30638550e-02 -3.28130901e-01 3.53195906e-01
-1.52329087e-01 -1.74865842e-01 2.76290178e-01 -2.04868793e-01
-5.99012911e-01 2.43481681e-01 -1.23355687e-01 -2.46951953e-01
-3.13966095e-01 6.68068752e-02 -2.76424021e-01 -1.98645294e-01
2.46830672e-01 1.13768607e-01 -4.25357670e-01 -5.37585557e-01
5.05768776e-01 4.63157058e-01 8.23989213e-01 -6.93365633e-01
9.52077270e-01 6.37394607e-01 -3.47347818e-02 -6.84602261e-01
-8.04405153e-01 -5.58062196e-01 -9.15209889e-01 -3.86992514e-01
9.22933161e-01 -1.12889683e+00 -9.59767044e-01 6.96700394e-01
-1.21215606e+00 1.76922962e-01 1.01395495e-01 4.10692245e-01
-5.30719340e-01 5.47645986e-01 -5.82787871e-01 -7.92120099e-01
-2.76690632e-01 -1.09724140e+00 1.38468373e+00 2.76962936e-01
-2.71886498e-01 -4.10366833e-01 -9.70308110e-02 4.56319094e-01
-2.31195047e-01 4.00236338e-01 4.15479392e-01 -5.45807660e-01
-6.81298614e-01 -4.46459264e-01 -2.19434902e-01 -2.07626864e-01
-2.64565796e-01 -2.14814082e-01 -5.90699673e-01 -4.99285728e-01
-5.59190631e-01 -9.53098312e-02 7.36268640e-01 4.28078085e-01
1.00283408e+00 -2.35768884e-01 -6.74406528e-01 7.08982289e-01
1.12685061e+00 -1.61637023e-01 6.12974346e-01 6.88213408e-01
1.17534769e+00 7.52543390e-01 7.66853869e-01 4.29814547e-01
8.48578513e-01 1.14115632e+00 1.60657823e-01 4.98185493e-02
6.60925210e-02 -6.29793644e-01 4.84285206e-02 4.73924100e-01
-5.09963989e-01 -2.65343279e-01 -7.46972501e-01 4.85470831e-01
-2.10504937e+00 -9.40483987e-01 -1.04176268e-01 2.43198633e+00
3.85963768e-01 1.54773861e-01 5.02338231e-01 -4.61582989e-02
9.82362747e-01 2.40315810e-01 -3.58576804e-01 2.47069061e-01
8.70151520e-02 -2.83104151e-01 6.43322349e-01 4.41483259e-01
-1.39839494e+00 9.12499607e-01 6.53072405e+00 7.21664011e-01
-5.22510350e-01 -1.52784199e-01 5.37194490e-01 -2.84025639e-01
1.71060696e-01 -7.81382099e-02 -1.23037684e+00 4.53826427e-01
5.29765308e-01 1.18276075e-01 4.54587638e-01 7.43089080e-01
9.80171487e-02 -1.53858423e-01 -1.13974977e+00 1.24054694e+00
4.15143639e-01 -1.05125690e+00 2.48504132e-02 1.18753865e-01
5.67775548e-01 -3.72469693e-01 -9.37675908e-02 -5.63160852e-02
-1.64674923e-01 -7.38971114e-01 1.12879312e+00 6.85806096e-01
5.57503521e-01 -9.98438597e-01 5.71913481e-01 1.35832995e-01
-1.47113812e+00 3.32801156e-02 -3.04790288e-01 7.84562826e-02
3.83527577e-01 2.33412758e-01 -4.34688568e-01 8.70345533e-01
9.41965818e-01 5.59628963e-01 -8.33926558e-01 1.27835035e+00
-4.37211573e-01 1.47229731e-01 -2.69891292e-01 1.15372226e-01
3.91860902e-02 -1.05718777e-01 6.35893643e-01 9.84177768e-01
4.34458762e-01 -7.48183206e-02 2.71739244e-01 3.72713625e-01
5.68945929e-02 7.47170448e-02 -1.43560186e-01 3.47202033e-01
7.55185902e-01 1.20210111e+00 -9.29142416e-01 -5.45653164e-01
-2.65219748e-01 1.30130374e+00 3.20398748e-01 3.06110173e-01
-1.05481291e+00 -4.56306130e-01 6.19515181e-01 2.29693383e-01
4.64641780e-01 -3.99909735e-01 7.99551234e-02 -1.14245915e+00
3.65493923e-01 -9.10995960e-01 4.64417636e-01 -7.70859957e-01
-9.58739042e-01 5.16900659e-01 3.19317997e-01 -1.40209901e+00
-2.32726753e-01 -4.03656453e-01 -3.34356517e-01 6.65360868e-01
-1.12178409e+00 -1.34659660e+00 -3.00533175e-01 4.67132330e-01
3.89870614e-01 1.19284667e-01 4.59485263e-01 3.50297153e-01
-6.27512693e-01 8.28152239e-01 -6.67254478e-02 3.46489996e-01
8.67108166e-01 -1.25860119e+00 7.55467951e-01 1.19052303e+00
1.87081665e-01 9.96506214e-01 7.61350930e-01 -7.53570259e-01
-1.34139514e+00 -8.50963235e-01 1.11005759e+00 -5.34692526e-01
3.09889287e-01 -7.34267950e-01 -5.43274343e-01 8.92865479e-01
9.59609915e-03 -1.94155261e-01 4.74246025e-01 2.83428520e-01
-2.44298950e-01 7.38633499e-02 -6.76827550e-01 8.45250487e-01
1.25688016e+00 -1.91866115e-01 -6.81875467e-01 4.27764177e-01
6.36847973e-01 -8.62566650e-01 -6.01960719e-01 1.55327380e-01
8.31094444e-01 -1.02809095e+00 1.37872577e+00 -2.76298285e-01
2.72805393e-01 -4.65854526e-01 -1.36505947e-01 -1.09030294e+00
-5.53395092e-01 -6.84817016e-01 -2.46374473e-01 1.08169067e+00
1.59037635e-02 -3.66950721e-01 9.44975853e-01 6.24560714e-01
6.78042024e-02 -8.00686657e-01 -9.07985508e-01 -5.37859857e-01
-4.75209057e-01 -3.56864691e-01 6.29935920e-01 4.91598666e-01
-7.05148503e-02 1.20498694e-01 -1.08124852e+00 5.48643947e-01
9.40303743e-01 7.73875639e-02 1.24192941e+00 -1.02517390e+00
-4.19658512e-01 -1.86975598e-02 -4.98180658e-01 -1.53006732e+00
-3.47083449e-01 -4.46992606e-01 2.11289540e-01 -1.44138718e+00
3.96482855e-01 -2.21875638e-01 -1.09110892e-01 3.87033969e-01
-6.82438970e-01 5.56959748e-01 4.67376441e-01 4.19699579e-01
-9.30818319e-01 2.78564483e-01 9.83605146e-01 1.65808901e-01
-1.36815459e-01 2.77317706e-02 -8.12556565e-01 8.27598631e-01
3.70265931e-01 -3.67776901e-01 -3.59422229e-02 -4.67118353e-01
3.32744569e-01 -9.31664482e-02 8.21674526e-01 -1.27521420e+00
5.04769921e-01 1.02204315e-01 7.22570300e-01 -1.00167024e+00
6.21091306e-01 -6.28507257e-01 3.25935692e-01 4.62518543e-01
-2.72255838e-01 2.65318185e-01 -1.12058707e-01 7.25120604e-01
2.05629040e-02 5.29019199e-02 3.76160860e-01 -4.59577283e-03
-9.00467098e-01 4.31166291e-01 1.02997109e-01 -2.52169192e-01
1.08506477e+00 -3.89848500e-01 -2.55682975e-01 -5.04707277e-01
-7.87541926e-01 2.64386743e-01 4.45056677e-01 5.04968107e-01
5.48994780e-01 -1.57064462e+00 -6.22957945e-01 2.05804691e-01
2.85576463e-01 -1.44988894e-01 3.90424192e-01 9.47985709e-01
-4.65987086e-01 4.58408237e-01 -1.84967332e-02 -6.24463379e-01
-1.41478944e+00 4.05154020e-01 2.91751474e-01 -7.39844702e-03
-8.44313383e-01 9.76912081e-01 5.59327491e-02 -2.86054760e-01
3.87161285e-01 7.25027025e-02 -1.71915859e-01 1.22716941e-01
8.36709619e-01 5.86618543e-01 -2.01467901e-01 -1.06661296e+00
-6.64297640e-01 9.58339334e-01 -1.88185155e-01 -1.58463269e-01
1.18424010e+00 -4.22711253e-01 4.86183986e-02 2.37619624e-01
1.06524193e+00 2.20800295e-01 -1.43450093e+00 -1.13185093e-01
-1.57944754e-01 -8.33217800e-01 -5.44232905e-01 -5.39324820e-01
-7.44123936e-01 4.50315803e-01 5.06343424e-01 -3.39579545e-02
8.25703979e-01 1.01154380e-01 8.47240806e-01 2.84962863e-01
6.33545220e-01 -1.15809703e+00 -4.19022180e-02 2.45676026e-01
7.35797584e-01 -9.86710370e-01 2.88496464e-01 -7.48199642e-01
-6.10912263e-01 9.68018949e-01 5.86052597e-01 -2.81545728e-01
1.29432783e-01 -2.12961733e-01 5.62253743e-02 -2.37862676e-01
-3.37347478e-01 -3.99058387e-02 5.94818056e-01 5.22034883e-01
2.29995802e-01 3.42357941e-02 -3.01259339e-01 4.59199250e-01
-4.54646826e-01 -1.05268538e-01 2.36139953e-01 7.20640719e-01
-3.26739073e-01 -8.88222933e-01 -8.26325238e-01 1.84992343e-01
-3.50398988e-01 7.96662346e-02 -3.08702499e-01 9.11843538e-01
2.33969301e-01 7.82195568e-01 -4.81271334e-02 -4.78278160e-01
6.13812983e-01 -1.95188344e-01 5.63410938e-01 -3.19849819e-01
-3.16631675e-01 2.94863105e-01 7.50521496e-02 -8.15429688e-01
-4.36164945e-01 -8.16852212e-01 -9.95392323e-01 -3.94895941e-01
-2.04569444e-01 8.31628367e-02 3.83189321e-01 9.20206487e-01
4.15852368e-01 4.00444508e-01 2.81089187e-01 -1.33538020e+00
-3.15291107e-01 -7.14047492e-01 -3.13610107e-01 5.15017271e-01
2.66331792e-01 -7.04139650e-01 -1.14727184e-01 3.77936624e-02] | [7.197500705718994, -0.7557703256607056] |
5c8493fe-acf4-401a-abf7-7565b4047b7c | deepbeat-a-multi-task-deep-learning-approach | 2001.00155 | null | https://arxiv.org/abs/2001.00155v2 | https://arxiv.org/pdf/2001.00155v2.pdf | DeepBeat: A multi-task deep learning approach to assess signal quality and arrhythmia detection in wearable devices | Wearable devices enable theoretically continuous, longitudinal monitoring of physiological measurements like step count, energy expenditure, and heart rate. Although the classification of abnormal cardiac rhythms such as atrial fibrillation from wearable devices has great potential, commercial algorithms remain proprietary and tend to focus on heart rate variability derived from green spectrum LED sensors placed on the wrist where noise remains an unsolved problem. Here, we develop a multi-task deep learning method to assess signal quality and arrhythmia event detection in wearable photoplethysmography devices for real-time detection of atrial fibrillation (AF). We train our algorithm on over one million simulated unlabeled physiological signals and fine-tune on a curated dataset of over 500K labeled signals from over 100 individuals from 3 different wearable devices. We demonstrate that in comparison with a traditional random forest-based approach (precision:0.24, recall:0.58, f1:0.34, auPRC:0.44) and a single task CNN (precision:0.59, recall:0.69, f1:0.64, auPRC:0.68) our architecture using unsupervised transfer learning through convolutional denoising autoencoders dramatically improves the performance of AF detection in participants at rest (pr:0.94, rc:0.98, f1:0.96, auPRC:0.96). In addition, we validate algorithm performance on a prospectively derived replication cohort of ambulatory subjects using data derived from an independently engineered device. We show that two-stage training can help address the unbalanced data problem common to biomedical applications where large well-annotated datasets are scarce. In conclusion, though a combination of simulation and transfer learning and we develop and apply a multitask architecture to the problem of AF detection from wearable wrist sensors demonstrating high levels of accuracy and a solution for the vexing challenge of mechanical noise. | ['Euan Ashley', 'Jessica Torres Soto'] | 2020-01-01 | null | null | null | null | ['heart-rate-variability', 'arrhythmia-detection'] | ['medical', 'medical'] | [ 3.12044442e-01 -3.08106542e-01 1.92868114e-01 -2.39490569e-01
-1.12108171e+00 -8.17668259e-01 -2.09602699e-01 9.24688056e-02
-3.75406146e-01 1.00974739e+00 2.15467319e-01 -4.25145775e-01
-1.56390712e-01 -3.58785331e-01 -5.65068483e-01 -7.31635332e-01
-4.77494091e-01 7.66996369e-02 -6.79009080e-01 2.52614468e-01
-4.38021809e-01 1.76689029e-01 -8.99841607e-01 1.41436830e-01
6.73604071e-01 1.30581987e+00 -6.85501099e-01 1.06444561e+00
8.46386671e-01 1.11948758e-01 -9.27976549e-01 1.16234563e-01
2.55966872e-01 -6.36713386e-01 -2.06210822e-01 -5.34621954e-01
5.04670680e-01 -3.68749946e-01 2.24483430e-01 3.88807774e-01
1.48380482e+00 -3.48960787e-01 4.06798095e-01 -7.65325546e-01
-3.42300981e-01 2.13008493e-01 -3.51321548e-01 5.31399310e-01
3.22884440e-01 5.52822649e-01 5.10096192e-01 -4.05289620e-01
7.99210370e-02 3.55632186e-01 1.66922092e+00 5.46993732e-01
-1.70935977e+00 -8.34312975e-01 -6.56818926e-01 -6.04876578e-01
-1.26500082e+00 -5.41834295e-01 4.74272877e-01 -5.96755922e-01
9.29619551e-01 4.16354746e-01 1.00288200e+00 1.49827516e+00
5.98124444e-01 -1.26564391e-02 1.40053427e+00 -1.29703224e-01
1.67762861e-01 -2.05639154e-01 2.09818259e-01 5.01649976e-01
4.85409498e-01 2.35229105e-01 -5.32315671e-01 -7.36170888e-01
8.08775902e-01 -3.40194106e-02 -5.33458054e-01 3.75066817e-01
-1.37442803e+00 4.02689964e-01 2.44134981e-02 4.07248527e-01
-5.08238316e-01 3.43978405e-01 5.41421175e-01 4.78557825e-01
5.14545560e-01 5.89906394e-01 -9.07940388e-01 -5.24847865e-01
-1.00258243e+00 9.35992450e-02 8.87379229e-01 2.51261979e-01
1.74463227e-01 2.52321124e-01 -4.99238491e-01 6.94267809e-01
1.41464278e-01 8.91510904e-01 6.48366809e-01 -1.02880847e+00
-5.17572910e-02 3.49442214e-01 3.44294101e-01 -6.13246500e-01
-9.66282785e-01 -8.38806987e-01 -1.16759956e+00 5.33699915e-02
6.31949306e-01 -8.25884163e-01 -8.35518599e-01 1.83455455e+00
1.60125662e-02 2.60551691e-01 -5.77430204e-02 9.41243887e-01
8.42799902e-01 1.69639513e-01 3.70639533e-01 -5.62970221e-01
1.46098840e+00 -5.74063808e-02 -6.67651355e-01 -1.28654018e-01
4.02472556e-01 -5.31333864e-01 1.09481800e+00 7.54966676e-01
-1.01813447e+00 -6.48477972e-01 -1.08910978e+00 2.19867274e-01
1.35943204e-01 4.27628756e-01 4.77604866e-01 1.32472813e+00
-1.06074989e+00 7.22815990e-01 -1.06681073e+00 -2.63375908e-01
7.51247168e-01 5.43794096e-01 -7.73121864e-02 2.36292139e-01
-1.15858865e+00 6.09244704e-01 -1.63799062e-01 3.21120590e-01
-9.29347754e-01 -1.16862965e+00 -3.97903770e-01 -5.28441407e-02
-3.64958823e-01 -1.17535436e+00 8.17574024e-01 -7.92507708e-01
-1.50557089e+00 9.18739617e-01 1.24996535e-01 -5.59170544e-01
6.16960168e-01 -5.73812246e-01 -5.80704272e-01 5.77474423e-02
-1.45011097e-02 4.80589531e-02 7.78095841e-01 -7.05238700e-01
-5.64495847e-02 -5.51883876e-01 -5.35632789e-01 -2.32979774e-01
-3.36246639e-01 -5.78582548e-02 4.72661912e-01 -8.34373355e-01
-2.55861133e-01 -8.08962107e-01 -1.03694454e-01 -4.33264151e-02
-7.05114901e-02 2.58421659e-01 2.28762373e-01 -9.72550511e-01
1.12792933e+00 -2.00261736e+00 -2.53257543e-01 3.55749041e-01
6.72563970e-01 4.00802493e-01 2.10207775e-01 3.88985686e-02
-9.23823938e-02 3.27940226e-01 -5.60858667e-01 3.81987207e-02
-4.25574273e-01 -1.52227059e-01 -4.90581337e-03 6.97794557e-01
-1.52465492e-03 1.24256122e+00 -7.17730463e-01 -5.45095764e-02
9.73972976e-02 7.73169696e-01 -9.10262689e-02 3.58485132e-01
2.63073742e-01 8.25870395e-01 -1.26090571e-01 6.72542632e-01
3.15717846e-01 -3.10232937e-01 2.64932573e-01 -5.00716627e-01
1.91384956e-01 3.69520187e-02 -9.21096504e-01 1.95770454e+00
-3.74548674e-01 3.63756686e-01 -3.57690863e-02 -1.06185102e+00
1.14798915e+00 7.80203342e-01 9.86496568e-01 -5.40226102e-01
2.21778795e-01 1.63738266e-01 2.98453033e-01 -6.07006848e-01
-6.67946875e-01 -6.66558027e-01 -9.33941081e-03 7.22488821e-01
1.50630966e-01 2.82719553e-01 -4.38162118e-01 -4.02213484e-01
1.70256376e+00 1.65011823e-01 1.99597135e-01 -6.44064426e-01
1.43384770e-01 -2.73788035e-01 6.59936488e-01 1.03556895e+00
-5.12541413e-01 8.39825988e-01 2.29619399e-01 -9.93006408e-01
-7.29668915e-01 -1.39196587e+00 -4.44955438e-01 5.89618921e-01
-5.42960346e-01 -5.17425239e-01 -6.88738644e-01 -2.84567893e-01
1.47533253e-01 -3.41169983e-02 -6.72712088e-01 -3.88202250e-01
-6.43507659e-01 -1.53959596e+00 1.31153083e+00 7.99589932e-01
4.32614237e-01 -9.45526779e-01 -1.31400466e+00 4.89316493e-01
-3.14990342e-01 -9.15870190e-01 -1.74554691e-01 3.60141695e-01
-1.04342294e+00 -9.79648173e-01 -8.73017967e-01 -4.00437951e-01
1.41215846e-01 -6.52561545e-01 1.43883562e+00 -9.43060070e-02
-9.03496802e-01 6.09182119e-01 1.30151212e-01 -6.65800929e-01
-5.48500828e-02 -1.17880151e-01 4.04764831e-01 6.21649846e-02
3.93822163e-01 -9.50524449e-01 -1.25172281e+00 1.37860715e-01
-1.23591490e-01 -4.89262551e-01 4.44788963e-01 8.35870266e-01
6.15416646e-01 -6.20878160e-01 1.12453592e+00 -5.16839743e-01
1.02568555e+00 -2.94169664e-01 -1.71873793e-01 -9.82577819e-03
-9.68492150e-01 -3.80503923e-01 5.00507891e-01 -6.65527403e-01
-5.46699047e-01 2.50138909e-01 -5.24836360e-03 -2.41399258e-01
-2.42284894e-01 2.35455677e-01 3.13894510e-01 1.16235919e-01
1.39943719e+00 4.01110090e-02 3.11567187e-01 -7.53887668e-02
-5.06183133e-02 5.42025924e-01 7.39662468e-01 -7.35480309e-01
4.09480125e-01 3.37492228e-01 7.83959329e-02 -7.67592311e-01
-5.48781633e-01 -2.52684534e-01 -3.14396143e-01 -2.52371430e-01
9.68475580e-01 -1.33027756e+00 -1.12611413e+00 4.08331931e-01
-6.00035250e-01 -8.01660776e-01 -5.23513436e-01 5.82312286e-01
-5.27045250e-01 2.26219714e-01 -5.26948273e-01 -1.01205277e+00
-1.36155403e+00 -5.59038639e-01 1.08240306e+00 -6.16609082e-02
-8.32312047e-01 -8.53227556e-01 4.50221628e-01 1.56942248e-01
8.15995514e-01 1.14393497e+00 3.56910914e-01 -4.13874745e-01
2.12878570e-01 -2.44892046e-01 4.53044921e-02 4.88752961e-01
6.33532584e-01 -4.75520372e-01 -1.39758861e+00 -4.66070741e-01
3.08868766e-01 -4.62574869e-01 5.64658821e-01 8.94289255e-01
1.07420945e+00 -9.11064669e-02 -2.29706928e-01 6.10884309e-01
1.09443080e+00 7.28021041e-02 7.96911776e-01 -9.17963013e-02
5.43265581e-01 -8.16907212e-02 -1.90295845e-01 4.10993606e-01
-6.02622516e-02 2.38649189e-01 -2.06857592e-01 -4.95581567e-01
-1.15370519e-01 3.04501772e-01 3.72925550e-01 2.45509103e-01
-5.93523681e-01 1.26217604e-01 -9.54142213e-01 3.84462953e-01
-1.39872181e+00 -6.85420573e-01 -2.31928244e-01 2.52467060e+00
9.44445252e-01 9.07248259e-02 5.29308796e-01 3.25563699e-01
5.49197614e-01 -3.27868611e-01 -6.73426569e-01 -2.58689642e-01
-2.29165763e-01 7.92336226e-01 2.33395621e-01 1.11599855e-01
-1.09780252e+00 2.47401781e-02 6.38846111e+00 -1.00961879e-01
-1.21612239e+00 3.99764925e-01 9.28040624e-01 -3.14964771e-01
2.31168240e-01 -6.10167801e-01 -1.22425444e-01 6.83224499e-01
1.64837682e+00 3.44232209e-02 4.53647465e-01 2.04716042e-01
3.91925305e-01 1.69628933e-01 -1.09422112e+00 1.38817358e+00
8.55516717e-02 -9.87787426e-01 -7.76347280e-01 -7.56843984e-02
3.92039299e-01 2.34836116e-01 -9.59066898e-02 4.86520538e-03
-3.03874731e-01 -1.30512655e+00 1.35588840e-01 7.46940553e-01
1.42318273e+00 -2.45824963e-01 6.89941645e-01 -9.89731550e-02
-8.98681641e-01 -1.13134108e-01 1.73272029e-01 -2.83652246e-01
1.65250152e-02 1.29493546e+00 -6.44552112e-01 1.85047910e-01
1.10412347e+00 7.01136649e-01 -1.98196337e-01 9.49120522e-01
3.63074355e-02 1.14405215e+00 -6.15347564e-01 2.02808708e-01
-5.59859157e-01 1.54385209e-01 3.48732740e-01 1.33520770e+00
4.59013760e-01 2.75847375e-01 -4.19311523e-02 9.41391945e-01
-5.64946607e-02 -1.66107733e-02 -5.39467633e-01 2.24233255e-01
2.27493662e-02 1.35416102e+00 -4.52057362e-01 -2.45633990e-01
-1.76179737e-01 5.62661469e-01 -2.38029405e-01 3.73824000e-01
-8.00034165e-01 -3.44388634e-01 5.57035506e-01 3.15976232e-01
-2.87910432e-01 -6.45923838e-02 -9.04064000e-01 -1.16510594e+00
3.87244731e-01 -1.01836944e+00 5.14355600e-01 -5.43675661e-01
-1.41009390e+00 5.25078535e-01 -4.76799935e-01 -1.16145134e+00
-9.21083540e-02 -4.75544006e-01 -6.44144773e-01 1.30247498e+00
-9.69185829e-01 -8.03441584e-01 -6.78830504e-01 5.87487042e-01
8.22357833e-02 4.04729210e-02 1.44029796e+00 5.45010805e-01
-3.61775607e-01 5.82177281e-01 -2.81331301e-01 1.90987125e-01
1.02280545e+00 -1.43500316e+00 3.73762816e-01 5.67674518e-01
-1.24714509e-01 7.72118628e-01 3.10077429e-01 -6.51222408e-01
-1.34308338e+00 -1.11278903e+00 4.53984231e-01 -8.84541154e-01
3.69918913e-01 -3.48893225e-01 -8.12662840e-01 4.07928377e-01
-1.08654991e-01 4.62641925e-01 1.11661947e+00 3.71125340e-01
-1.50679231e-01 -5.43620706e-01 -1.26208103e+00 4.16332483e-02
8.69050682e-01 -5.04054248e-01 -3.83174151e-01 5.33725142e-01
-2.95075960e-03 -5.15200555e-01 -1.66896522e+00 7.66254842e-01
1.33070874e+00 -6.22036994e-01 1.04307616e+00 -5.95881224e-01
-4.81619462e-02 -1.33671433e-01 2.57848084e-01 -1.19138145e+00
-2.16527566e-01 -1.12303913e+00 -4.39372391e-01 8.82851720e-01
4.48005974e-01 -7.45283008e-01 6.25728428e-01 7.23067164e-01
-9.23415199e-02 -6.90100133e-01 -1.00992036e+00 -5.98749518e-01
1.05186015e-01 -3.94186020e-01 2.55471747e-02 7.68068016e-01
-5.89942746e-02 3.84469658e-01 -4.72473890e-01 -2.97848228e-03
7.41232336e-01 1.80914178e-02 3.78583610e-01 -1.64940441e+00
-3.25862974e-01 5.40621243e-02 -2.45910153e-01 -1.88879490e-01
-3.65079433e-01 -7.87708938e-01 -8.82106200e-02 -1.23376048e+00
6.24783449e-02 -4.72038180e-01 -8.43324780e-01 7.75777698e-01
-2.59767532e-01 6.69668436e-01 -5.33206820e-01 5.74226119e-02
-3.74679640e-02 -1.41427368e-01 5.89610457e-01 -1.88142136e-01
-6.32497847e-01 -1.27036851e-02 -7.67211616e-01 4.40207034e-01
1.20588934e+00 -5.24895787e-01 -2.44666353e-01 -2.38197729e-01
2.20139116e-01 8.87911990e-02 7.34876990e-01 -1.40159369e+00
-1.97507337e-01 5.89554429e-01 1.14836228e+00 2.95030564e-01
1.16234705e-01 -6.00661278e-01 4.70208794e-01 9.49510694e-01
-1.37491599e-01 1.74037099e-01 4.14643109e-01 4.75526452e-01
4.54918891e-01 7.16128767e-01 5.34695387e-01 -1.82305485e-01
2.89075583e-01 1.74824491e-01 -5.06422102e-01 3.44936669e-01
5.74335933e-01 -2.00282097e-01 -3.56491536e-01 -2.27055147e-01
-1.17807794e+00 -7.02404976e-02 -2.02680349e-01 2.73508072e-01
5.28491497e-01 -1.08692765e+00 -9.62125003e-01 3.91860068e-01
6.34972900e-02 -2.04731762e-01 1.41794726e-01 1.26270998e+00
-2.70769328e-01 1.50848068e-02 -4.25716698e-01 -9.61569250e-01
-9.27337170e-01 1.06790245e-01 9.12689626e-01 1.58787400e-01
-8.27809215e-01 3.74014616e-01 -3.30464959e-01 3.60742062e-02
2.67705861e-02 -6.12907946e-01 2.79249072e-01 -2.14945674e-02
4.71587986e-01 4.90964115e-01 5.78186691e-01 1.03230514e-01
-5.13765931e-01 7.77821243e-01 6.20501816e-01 7.77279288e-02
1.34529102e+00 1.99084178e-01 2.48058494e-02 5.47137499e-01
8.06592345e-01 -1.08932130e-01 -1.03429723e+00 4.10635233e-01
-3.75851631e-01 8.26482028e-02 6.73200993e-05 -1.56191921e+00
-1.03861153e+00 6.92412376e-01 1.53781474e+00 1.59720853e-01
1.50716484e+00 -3.94484162e-01 7.69421756e-01 3.55464786e-01
1.21833809e-01 -7.98999667e-01 -4.51378375e-02 -3.95951450e-01
6.90806031e-01 -9.69024956e-01 5.02524562e-02 1.55868128e-01
-2.80111253e-01 1.01933885e+00 2.24965930e-01 -1.94633096e-01
8.01535368e-01 5.45973301e-01 4.43805337e-01 -2.62810826e-01
-3.15696806e-01 2.98742771e-01 6.44691586e-02 8.11827362e-01
7.91070938e-01 1.43663198e-01 -6.06725276e-01 9.95311558e-01
-6.12845719e-02 6.15624070e-01 2.38859370e-01 8.43619823e-01
-4.40444127e-02 -7.97126591e-01 -1.60253659e-01 1.00776219e+00
-1.04397929e+00 -1.96676478e-02 -1.58886135e-01 3.74863893e-01
1.89510331e-01 1.05307531e+00 -2.14355454e-01 -2.63307184e-01
4.49840069e-01 5.27268529e-01 5.05559564e-01 -3.90468746e-01
-1.16153896e+00 1.65296927e-01 1.96142182e-01 -5.34649909e-01
-6.59354329e-01 -5.80329716e-01 -7.95473874e-01 4.23164248e-01
-6.13238066e-02 -4.28834818e-02 4.98405993e-01 6.55934572e-01
7.59096861e-01 7.26395130e-01 3.56747478e-01 -5.02024949e-01
-3.55528235e-01 -1.08915520e+00 -5.35483062e-01 3.84823561e-01
7.87925601e-01 -3.32848966e-01 -3.35064411e-01 4.55103248e-01] | [14.15150260925293, 3.1740121841430664] |
62abec11-1a5f-43e2-abb0-aa3dbeb00753 | d2im-net-learning-detail-disentangled | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Li_D2IM-Net_Learning_Detail_Disentangled_Implicit_Fields_From_Single_Images_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Li_D2IM-Net_Learning_Detail_Disentangled_Implicit_Fields_From_Single_Images_CVPR_2021_paper.pdf | D2IM-Net: Learning Detail Disentangled Implicit Fields From Single Images | We present the first single-view 3D reconstruction network aimed at recovering geometric details from an input image which encompass both topological shape structures and surface features. Our key idea is to train the network to learn a detail disentangled reconstruction consisting of two functions, one implicit field representing the coarse 3D shape and the other capturing the details. Given an input image, our network, coined D^2IM_Net, encodes it into global and local features which are respectively fed into two decoders. The base decoder uses the global features to reconstruct a coarse implicit field, while the detail decoder reconstructs, from the local features, two displacement maps, defined over the front and back sides of the captured object. The final 3D reconstruction is a fusion between the base shape and the displacement maps, with three losses enforcing the recovery of coarse shape, overall structure, and surface details via a novel Laplacian term. | ['Hao Zhang', 'Manyi Li'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['single-view-3d-reconstruction'] | ['computer-vision'] | [ 3.75783086e-01 4.28541839e-01 2.28152975e-01 -3.86806160e-01
-9.26449180e-01 -5.58515370e-01 6.33731782e-01 -8.27755854e-02
-6.50217980e-02 5.13088226e-01 3.81756157e-01 2.27148846e-01
-4.33133990e-02 -1.03128600e+00 -1.05045879e+00 -9.86540735e-01
1.95152715e-01 7.32581675e-01 -5.93096428e-02 9.80595946e-02
3.46945554e-01 6.45614028e-01 -1.15593112e+00 2.27183655e-01
4.38095838e-01 1.26454818e+00 1.17884405e-01 6.07849479e-01
9.34861675e-02 6.86317265e-01 6.58945963e-02 -3.90327454e-01
3.68464500e-01 -1.35038212e-01 -6.03061736e-01 1.82750136e-01
5.57287037e-01 -5.77290893e-01 -6.90453053e-01 8.68400097e-01
2.50618756e-01 -1.94119230e-01 8.32094610e-01 -5.79951763e-01
-8.34784627e-01 1.28673622e-02 -4.17248756e-01 -2.55500168e-01
4.27654177e-01 1.96150720e-01 1.16610122e+00 -1.06288230e+00
9.97100472e-01 1.16755116e+00 6.17450953e-01 3.41147453e-01
-1.72143090e+00 -2.66798258e-01 -9.44875628e-02 -3.17301571e-01
-1.10638356e+00 -4.46959943e-01 1.15313792e+00 -7.78695822e-01
6.55602336e-01 1.44357579e-02 6.14950895e-01 1.01079202e+00
3.76821697e-01 4.30636853e-01 1.17501438e+00 4.02910402e-03
1.36960521e-01 -8.33655298e-02 -1.07361980e-01 1.01803958e+00
1.23058714e-01 4.05203193e-01 -5.26739061e-01 -1.54714495e-01
1.35380530e+00 4.39296722e-01 -6.25345051e-01 -7.77779758e-01
-1.15627825e+00 7.45514512e-01 5.99339664e-01 9.68550369e-02
-4.77848023e-01 1.13537796e-01 -1.26600429e-01 1.58218846e-01
8.24377477e-01 1.74712062e-01 -3.15594107e-01 2.74033666e-01
-6.77371442e-01 3.00740272e-01 6.37383699e-01 6.29055202e-01
1.32623506e+00 -3.43147576e-01 2.93221623e-02 2.91540235e-01
5.34945846e-01 7.63811111e-01 -2.38077998e-01 -1.09114361e+00
5.52493989e-01 9.06376898e-01 1.06642224e-01 -1.16582823e+00
-2.80135404e-02 -3.05983037e-01 -1.08423245e+00 4.15380090e-01
1.85894921e-01 1.77891836e-01 -9.76664364e-01 1.92428887e+00
7.53066987e-02 -7.90607184e-02 -2.99268067e-01 8.71852815e-01
7.34740436e-01 4.68415707e-01 -5.16378880e-01 2.56651819e-01
1.04921305e+00 -5.03007174e-01 -3.32437456e-01 -2.20285252e-01
-1.44365281e-01 -5.04444838e-01 3.96935910e-01 8.25887639e-03
-1.62448287e+00 -3.53151828e-01 -1.08008385e+00 -6.20506823e-01
-9.25559402e-02 1.58095449e-01 3.93480342e-03 -2.35771716e-01
-1.15974081e+00 8.03975165e-01 -8.96653712e-01 1.82941243e-01
5.85987806e-01 3.32588524e-01 -8.45403075e-01 -2.16733828e-01
-7.14604557e-01 6.90095663e-01 -2.60913789e-01 9.99054834e-02
-1.13509333e+00 -7.90628791e-01 -1.16471255e+00 1.49105802e-01
8.26330185e-02 -1.30200410e+00 7.77468741e-01 -4.94246453e-01
-1.47214842e+00 1.32634699e+00 -2.18647465e-01 1.72721431e-01
5.38438737e-01 2.82373965e-01 2.64592826e-01 2.97478944e-01
3.11906159e-01 3.64294708e-01 1.12085247e+00 -1.59748209e+00
-1.03004806e-01 -9.46260929e-01 -2.93496773e-02 1.41247168e-01
4.74004000e-01 -9.20842767e-01 -4.38177317e-01 -5.48671246e-01
5.14818966e-01 -5.54236233e-01 1.41702006e-02 5.75168967e-01
-6.19901299e-01 4.63344991e-01 5.08118451e-01 -1.09108639e+00
6.62176728e-01 -2.32074809e+00 1.06083822e+00 3.12126487e-01
8.38501036e-01 -4.13853675e-01 -3.24310541e-01 3.27218354e-01
-1.04762986e-01 2.52670571e-02 -6.73115671e-01 -7.60264337e-01
-6.89769238e-02 7.92368725e-02 -2.47253045e-01 7.12125540e-01
4.86025780e-01 1.22458518e+00 -8.22904170e-01 2.62884289e-01
2.89354026e-01 8.95635188e-01 -7.38158286e-01 6.04103982e-01
-2.28286356e-01 9.03290153e-01 -7.24563599e-01 1.82380572e-01
9.15021420e-01 -3.85172516e-01 1.54628037e-02 -3.66353273e-01
-2.09266752e-01 4.98242170e-01 -8.42238247e-01 2.10655308e+00
-5.48571825e-01 3.25097829e-01 7.11212516e-01 -1.00583088e+00
9.68702257e-01 3.17622602e-01 5.73719621e-01 -6.39956534e-01
1.32421017e-01 2.80117542e-01 -6.85196221e-01 -3.68374795e-01
6.17521480e-02 -5.41176200e-01 -4.26130369e-02 4.74559009e-01
2.62607664e-01 -4.81213808e-01 -6.87101305e-01 2.85884678e-01
1.18080831e+00 2.90955573e-01 -1.08668827e-01 -1.64333973e-02
6.21659994e-01 -5.72383702e-01 1.61891788e-01 9.11019966e-02
3.99400324e-01 1.08520877e+00 7.87786961e-01 -6.90854311e-01
-1.35946369e+00 -1.54258657e+00 1.89888407e-03 2.39313409e-01
2.55612880e-01 -3.97587940e-02 -5.29496253e-01 -6.25377297e-01
3.34674895e-01 2.72994220e-01 -9.51451838e-01 -3.94751579e-01
-5.83476841e-01 1.07967198e-01 8.42836648e-02 6.04287125e-02
4.69150841e-01 -8.80004466e-01 -4.85952795e-01 -2.62192413e-02
-2.84060150e-01 -9.50475335e-01 -8.94747436e-01 1.22708611e-01
-1.00608861e+00 -1.13012624e+00 -7.13791370e-01 -6.36950135e-01
9.62946773e-01 2.46349454e-01 1.19686234e+00 1.56623721e-01
-2.42750958e-01 2.13953719e-01 2.31171042e-01 4.31293398e-01
-2.37444714e-01 -1.83309972e-01 -4.01137561e-01 6.18277907e-01
-3.81175339e-01 -1.16240847e+00 -6.22322619e-01 -9.57472192e-04
-9.22529876e-01 4.87538427e-01 6.02169871e-01 8.99837673e-01
9.99048650e-01 -4.13960636e-01 -1.04502318e-02 -8.07654858e-01
-3.55910994e-02 -3.93954933e-01 -3.90162021e-01 -3.68625149e-02
-1.51266828e-01 5.17852545e-01 4.92971539e-01 1.49362937e-01
-8.64906251e-01 1.30586743e-01 -2.98092008e-01 -5.83689272e-01
-1.05480701e-01 1.27558380e-01 -5.26294589e-01 -1.30745158e-01
2.92966753e-01 4.98430043e-01 3.19370419e-01 -9.24067676e-01
3.89651090e-01 1.79120153e-01 5.04990637e-01 -4.73531544e-01
8.25196624e-01 8.20706546e-01 6.58105388e-02 -5.91797471e-01
-9.45094585e-01 -1.80523813e-01 -8.55547547e-01 7.30753019e-02
1.11430562e+00 -9.56800461e-01 -7.61516392e-01 5.13482571e-01
-1.29372716e+00 -2.59747207e-01 -6.38271272e-01 1.12400554e-01
-9.25767958e-01 7.25258142e-02 -6.38773322e-01 -2.24281102e-01
-3.15392733e-01 -1.14281118e+00 1.80190480e+00 -3.19852084e-01
2.57336438e-01 -1.01147330e+00 1.81880355e-01 3.10139745e-01
3.93263906e-01 6.94497168e-01 1.31432164e+00 1.05599962e-01
-1.31062055e+00 -8.27783942e-02 -4.07672346e-01 3.47116947e-01
1.32381216e-01 -4.12231773e-01 -9.66709554e-01 -2.95578420e-01
4.44217861e-01 -2.12630600e-01 9.87442911e-01 4.60720301e-01
8.70834172e-01 -3.51002902e-01 -1.23849079e-01 1.16093862e+00
1.68364286e+00 -3.03960711e-01 5.16828537e-01 -3.69183153e-01
9.26900387e-01 3.13933015e-01 -3.84461015e-01 2.71150380e-01
5.71917772e-01 5.18272996e-01 7.31419861e-01 -2.71762069e-02
-3.70892406e-01 -7.77959883e-01 3.44913810e-01 9.53496516e-01
-2.49737248e-01 4.00474407e-02 -5.55101037e-01 2.99181938e-01
-1.51013255e+00 -8.26273203e-01 1.37532026e-01 2.35079169e+00
5.86518228e-01 1.61810406e-02 -3.88265938e-01 -1.79295182e-01
3.62097949e-01 5.30293643e-01 -8.92467678e-01 6.64284229e-02
-8.32256824e-02 3.07809860e-01 1.31629854e-01 1.00864506e+00
-6.24760866e-01 4.54206228e-01 5.64325857e+00 3.15642536e-01
-1.29981589e+00 -6.67895898e-02 4.69166636e-01 1.93232615e-02
-1.10795951e+00 1.24176823e-01 -2.37239361e-01 2.85141855e-01
3.74695957e-01 1.62560612e-01 7.21880555e-01 1.56412959e-01
-1.14136804e-02 1.77514866e-01 -1.32208169e+00 8.69519413e-01
1.15186736e-01 -1.67301190e+00 3.97804499e-01 2.90897131e-01
6.14569545e-01 1.36020198e-01 1.18115947e-01 -2.67667621e-01
7.05604702e-02 -1.03124356e+00 1.02599835e+00 1.04956377e+00
1.49879026e+00 -4.22651201e-01 3.39457303e-01 6.75632060e-01
-1.02679741e+00 3.51173818e-01 -2.73940153e-03 -2.37880219e-02
2.79032826e-01 6.24476314e-01 -2.31615022e-01 7.51207769e-01
3.02027941e-01 1.15325081e+00 3.29292007e-03 3.11451405e-01
-2.74070382e-01 8.72736722e-02 -1.25443414e-02 6.80637896e-01
-8.95256177e-02 -6.84651911e-01 9.61753845e-01 5.75266063e-01
1.87801033e-01 2.58371174e-01 -4.34254818e-02 1.58696485e+00
-3.80541056e-01 -5.42886913e-01 -8.47814620e-01 1.87678054e-01
5.96078783e-02 1.16195488e+00 -3.50540191e-01 -2.10361391e-01
-2.97845185e-01 1.14032197e+00 7.41240680e-01 5.49713910e-01
-3.59043300e-01 3.64171946e-03 8.45057428e-01 4.64693278e-01
4.70979035e-01 -4.15670633e-01 -6.77579820e-01 -1.62866747e+00
2.31232449e-01 -5.21932542e-01 -2.23747984e-01 -8.39480996e-01
-1.37186980e+00 4.86103356e-01 -3.48650008e-01 -9.27635431e-01
1.36790916e-01 -6.24752641e-01 -4.82423693e-01 1.37855136e+00
-1.43035257e+00 -1.23362398e+00 -3.02427590e-01 4.18294042e-01
1.98893711e-01 2.13202149e-01 7.67515242e-01 2.53281504e-01
-2.31201917e-01 1.02247231e-01 -1.24099500e-01 1.12992711e-01
2.25161433e-01 -1.29155385e+00 5.50040543e-01 3.52900177e-01
-9.02639031e-02 5.07591307e-01 1.69102415e-01 -5.74621797e-01
-1.79185331e+00 -8.98829281e-01 1.09995806e+00 -6.47500098e-01
1.63323075e-01 -6.32827997e-01 -7.66086578e-01 8.86028945e-01
-1.21717721e-01 3.01938504e-01 1.53450239e-02 -5.79405248e-01
-6.28331244e-01 1.37309879e-01 -1.26466286e+00 1.10608585e-01
1.14595044e+00 -1.00617051e+00 -4.97003853e-01 -1.33549258e-01
7.94513941e-01 -5.87805569e-01 -1.00443232e+00 1.41270325e-01
6.03865564e-01 -1.15361524e+00 1.12388158e+00 -2.89937884e-01
9.08470988e-01 -1.95006788e-01 -4.46376383e-01 -1.36701703e+00
-5.66550970e-01 -4.78842944e-01 -1.39117256e-01 6.72238827e-01
3.76929909e-01 -5.71347713e-01 7.80039251e-01 3.79392624e-01
-1.32103816e-01 -1.04533041e+00 -9.92762208e-01 -9.12268460e-02
1.49787381e-01 -8.14215690e-02 4.93151695e-01 6.32780015e-01
-6.32654250e-01 7.70330012e-01 -2.80680686e-01 6.35920689e-02
9.15280521e-01 4.86045331e-01 3.64653260e-01 -1.24238133e+00
-3.17668885e-01 -3.08247477e-01 -3.49667102e-01 -1.55178404e+00
7.00648576e-02 -1.35056520e+00 -2.07728660e-03 -1.39372718e+00
4.83040780e-01 -2.56112158e-01 2.74675786e-01 2.61311203e-01
1.58502504e-01 2.76897311e-01 1.63360998e-01 1.48299336e-01
-5.64333983e-02 1.00409389e+00 1.84682834e+00 -1.15264811e-01
1.18728643e-02 -7.96347260e-02 -7.53060281e-01 6.95022583e-01
6.97810724e-02 -3.87041211e-01 2.98659448e-02 -8.99818778e-01
3.69078428e-01 6.83717549e-01 9.02653813e-01 -4.93314773e-01
9.43307132e-02 1.36770055e-01 5.57914257e-01 -3.90941590e-01
4.87708002e-01 -9.36145604e-01 3.93693358e-01 2.41252169e-01
-4.18477595e-01 -1.24693185e-01 -2.76132882e-01 7.44361222e-01
-1.71034902e-01 3.31741571e-01 8.38982463e-01 -2.02006727e-01
5.65476418e-02 1.01561940e+00 3.88648212e-01 1.39526889e-01
5.69567859e-01 -6.26191869e-02 5.28278835e-02 -3.92323822e-01
-1.02435589e+00 -1.46412313e-01 1.02353287e+00 1.68997496e-01
1.02802205e+00 -1.49310136e+00 -8.67348552e-01 9.97952521e-01
-2.16156840e-01 6.15542591e-01 3.79636198e-01 6.89440310e-01
-5.72903275e-01 2.43496642e-01 -2.12342843e-01 -5.84848166e-01
-6.65614486e-01 3.62332463e-01 5.66407919e-01 -3.96084815e-01
-9.92813110e-01 7.94358373e-01 7.36460328e-01 -8.05884540e-01
-5.31478450e-02 -4.57922220e-01 1.51490942e-01 -2.70973772e-01
2.68021286e-01 2.10164815e-01 -1.02183511e-02 -8.81991923e-01
-1.70052618e-01 1.21207607e+00 4.03195620e-01 -3.12511116e-01
1.80541062e+00 -1.16338968e-01 -5.25026500e-01 4.50758368e-01
1.86890912e+00 1.19500324e-01 -1.82989430e+00 -4.28700835e-01
-3.83104414e-01 -5.07055521e-01 9.49828178e-02 -7.19747484e-01
-1.27560604e+00 1.13968503e+00 9.93038639e-02 -5.58281736e-03
1.03524089e+00 3.05641055e-01 9.05093789e-01 -2.73540556e-01
5.85757554e-01 -1.27419412e-01 1.41810641e-01 6.93520427e-01
1.24177897e+00 -8.64760041e-01 -1.23557948e-01 -3.57684463e-01
-1.40193418e-01 1.04892743e+00 1.15112603e-01 -7.67737448e-01
8.84870589e-01 1.75391614e-01 -4.63146836e-01 -7.24521816e-01
-8.16132486e-01 1.47161946e-01 6.94225848e-01 3.04038286e-01
1.46063209e-01 9.84479934e-02 2.14609727e-01 5.49725115e-01
8.44831858e-03 -4.87360954e-02 1.22012995e-01 5.02570868e-01
-2.14379087e-01 -9.16610062e-01 -9.88510344e-03 4.34901416e-01
-1.65956169e-01 2.24247411e-01 -5.25831640e-01 3.70721489e-01
4.54216868e-01 1.34236276e-01 3.75920117e-01 -4.79804486e-01
6.60875380e-01 -1.83053061e-01 7.03074932e-01 -8.08717668e-01
-2.23040625e-01 -2.18046814e-01 -2.98838586e-01 -8.56790960e-01
-1.34339139e-01 -5.05385816e-01 -1.14539599e+00 -2.82451302e-01
1.15477808e-01 -1.33393332e-01 6.09891117e-01 8.77844632e-01
5.19955993e-01 3.41787964e-01 9.69387412e-01 -1.46655118e+00
-5.73325753e-01 -6.49349093e-01 -7.65727997e-01 6.68625534e-01
1.13238740e+00 -5.95573545e-01 -6.17623866e-01 6.62051514e-02] | [8.844468116760254, -3.420180559158325] |
7835b4b3-4607-4fe1-b024-2b9d16e5d420 | leveraging-language-identification-to-enhance | 2306.04964 | null | https://arxiv.org/abs/2306.04964v1 | https://arxiv.org/pdf/2306.04964v1.pdf | Leveraging Language Identification to Enhance Code-Mixed Text Classification | The usage of more than one language in the same text is referred to as Code Mixed. It is evident that there is a growing degree of adaption of the use of code-mixed data, especially English with a regional language, on social media platforms. Existing deep-learning models do not take advantage of the implicit language information in the code-mixed text. Our study aims to improve BERT-based models performance on low-resource Code-Mixed Hindi-English Datasets by experimenting with language augmentation approaches. We propose a pipeline to improve code-mixed systems that comprise data preprocessing, word-level language identification, language augmentation, and model training on downstream tasks like sentiment analysis. For language augmentation in BERT models, we explore word-level interleaving and post-sentence placement of language information. We have examined the performance of vanilla BERT-based models and their code-mixed HingBERT counterparts on respective benchmark datasets, comparing their results with and without using word-level language information. The models were evaluated using metrics such as accuracy, precision, recall, and F1 score. Our findings show that the proposed language augmentation approaches work well across different BERT models. We demonstrate the importance of augmenting code-mixed text with language information on five different code-mixed Hindi-English downstream datasets based on sentiment analysis, hate speech detection, and emotion detection. | ['Mukta S. Takalikar', 'Raviraj Joshi', 'Aryan Patil', 'Varad Patwardhan', 'Abhishek Phaltankar', 'Gauri Takawane'] | 2023-06-08 | null | null | null | null | ['hate-speech-detection', 'sentiment-analysis'] | ['natural-language-processing', 'natural-language-processing'] | [-2.52237350e-01 -2.56582797e-01 -1.95100367e-01 -3.93749446e-01
-7.20812857e-01 -5.46972930e-01 4.85273659e-01 4.91167486e-01
-7.38346934e-01 2.48137340e-01 2.23504156e-01 -7.02101231e-01
5.01448333e-01 -4.02469456e-01 -3.49888444e-01 -1.52865842e-01
1.62564367e-02 -1.92561485e-02 -2.80439883e-01 -6.17764652e-01
3.11364055e-01 1.65461540e-01 -9.81097162e-01 7.45869875e-01
1.15284443e+00 4.40378398e-01 2.22448662e-01 7.36779571e-01
-5.04457355e-01 1.46671247e+00 -5.12695312e-01 -8.57939601e-01
6.92022145e-02 -1.45836577e-01 -6.41091108e-01 -1.37152344e-01
-1.15146525e-01 1.56781098e-04 -1.40470579e-01 1.09854662e+00
2.93160975e-01 -2.70660311e-01 3.49577218e-01 -1.36635983e+00
-9.14603412e-01 1.02605331e+00 -7.04780161e-01 1.64472178e-01
2.49361247e-01 7.47673735e-02 1.00516415e+00 -1.21064687e+00
4.99777734e-01 9.88254070e-01 1.07088244e+00 4.10719275e-01
-1.12213731e+00 -8.37354720e-01 1.78133860e-01 -3.66716385e-02
-1.52006817e+00 -1.93654895e-01 1.01852393e+00 -1.03746784e+00
1.49053729e+00 -1.17404982e-01 2.33059257e-01 1.19411564e+00
3.86473954e-01 1.08184564e+00 1.00045991e+00 -6.91388965e-01
-1.74685866e-01 7.01209724e-01 3.72359991e-01 7.12554812e-01
-1.96499988e-01 -4.34864849e-01 -3.52621824e-01 -1.21957086e-01
-3.12583297e-01 4.96528782e-02 6.88093528e-02 8.94670114e-02
-9.67591465e-01 1.23195207e+00 2.58575052e-01 5.33597946e-01
-9.31937248e-02 -5.88574782e-02 1.04510319e+00 2.76102513e-01
8.61290395e-01 6.25818849e-01 -8.19005370e-01 -4.51912224e-01
-1.05413222e+00 3.22700441e-02 6.64034247e-01 1.09341788e+00
8.12878966e-01 1.50112405e-01 -8.27742592e-02 1.18676507e+00
4.11608785e-01 3.50322038e-01 9.49755013e-01 -3.32029343e-01
8.22709143e-01 9.89369094e-01 -3.92799556e-01 -7.48717427e-01
-4.99282390e-01 -5.07462382e-01 -6.91204965e-01 -4.89047579e-02
-8.76104683e-02 -5.40468812e-01 -6.83107615e-01 1.75898600e+00
-3.11197430e-01 -2.28962675e-01 2.78390646e-01 1.34688556e-01
9.21635330e-01 5.71029007e-01 2.36752555e-01 2.23978147e-01
1.13214517e+00 -1.41044307e+00 -7.52165496e-01 -6.20055139e-01
1.45591402e+00 -1.06678116e+00 1.50686359e+00 1.64437383e-01
-6.01694882e-01 -4.95759726e-01 -9.50500607e-01 -2.99696416e-01
-9.21308398e-01 5.28342962e-01 3.23716372e-01 1.02093768e+00
-9.64482307e-01 2.93077454e-02 -5.11459589e-01 -3.88077617e-01
1.89219370e-01 1.69553205e-01 -4.42899883e-01 -1.91332065e-02
-1.17742479e+00 7.42302775e-01 2.02251613e-01 -1.25907421e-01
-5.71804404e-01 -6.33412600e-01 -1.30429792e+00 -1.28402486e-01
5.46065494e-02 1.34890482e-01 1.18566000e+00 -1.14025116e+00
-9.82338369e-01 1.07026625e+00 -9.99859497e-02 -4.37544882e-01
3.38854343e-01 -4.65802968e-01 -6.01029873e-01 -5.00427485e-01
3.03951114e-01 4.18854922e-01 5.11248946e-01 -1.10907745e+00
-3.56769770e-01 -4.45479266e-02 2.63676733e-01 -1.75116435e-02
-8.52241695e-01 4.74366784e-01 -2.03719482e-01 -7.77144790e-01
-6.67991340e-01 -1.16143036e+00 -1.19870320e-01 -4.42943156e-01
-4.56516504e-01 -6.63190261e-02 9.33454752e-01 -9.59581375e-01
1.89336693e+00 -2.65400910e+00 -2.14055657e-01 -6.08794354e-02
1.08473197e-01 4.27273959e-01 -4.86851603e-01 7.43682563e-01
-1.58329740e-01 5.53488851e-01 -4.62991744e-01 -9.08878267e-01
-2.15369314e-01 1.77688152e-01 -1.92841351e-01 2.50782132e-01
5.50490975e-01 8.67425561e-01 -8.20871532e-01 -3.32364500e-01
-3.35642807e-02 4.80348527e-01 -1.10286295e+00 2.34047711e-01
1.31259531e-01 1.24522105e-01 -1.74817629e-02 5.54069579e-01
5.60106158e-01 4.53596339e-02 -7.19952434e-02 3.21648926e-01
-4.04743463e-01 5.02678394e-01 -4.97757465e-01 1.72161782e+00
-1.06384480e+00 9.71441805e-01 2.03838989e-01 -5.45197666e-01
8.65710735e-01 2.79438645e-01 1.66856974e-01 -4.61109906e-01
1.38082564e-01 3.24393123e-01 2.38353536e-01 -7.41466343e-01
7.72812426e-01 -5.34404814e-02 -4.98958349e-01 2.31330857e-01
7.32165128e-02 -2.57364929e-01 1.93306580e-01 4.36190426e-01
1.10411012e+00 -1.55967161e-01 2.79143840e-01 -2.93079168e-01
9.36649978e-01 3.77064608e-02 2.05510840e-01 5.52610338e-01
-3.13403994e-01 5.94061196e-01 6.99073613e-01 3.47351804e-02
-9.42800939e-01 -4.13112193e-01 -9.14611369e-02 1.63615310e+00
-4.96301353e-01 -8.73264492e-01 -7.44144976e-01 -9.57398117e-01
-1.42910317e-01 9.05465066e-01 -7.01970756e-01 -3.33572447e-01
-4.53279972e-01 -9.27558720e-01 1.05328929e+00 4.02791977e-01
3.76388341e-01 -1.00376856e+00 -1.86501682e-01 1.90102041e-01
-2.64156431e-01 -1.25142121e+00 -3.99563789e-01 7.06908166e-01
-2.75651813e-01 -7.85505235e-01 -3.69493306e-01 -9.05024946e-01
4.26896483e-01 -1.87823437e-02 1.16080391e+00 4.35175151e-01
-2.08538752e-02 2.52026886e-01 -7.63275385e-01 -5.62265277e-01
-9.39534009e-01 6.08697236e-01 -2.11351916e-01 -5.14853261e-02
6.22824550e-01 -1.26780987e-01 -7.01320767e-02 -1.29180104e-01
-9.86819983e-01 1.63219005e-01 5.50312102e-01 7.87768662e-01
-3.43901634e-01 -3.79905850e-01 4.31311160e-01 -1.22144210e+00
9.05102372e-01 -9.02613938e-01 -1.11526862e-01 4.60902713e-02
-3.95674050e-01 -2.17560995e-02 8.37551713e-01 -4.98728573e-01
-9.60242510e-01 3.86624113e-02 -7.64860153e-01 -2.29401477e-02
-6.50763959e-02 1.12112379e+00 3.88008878e-02 8.51687267e-02
7.22533941e-01 -1.77218523e-02 -1.67502947e-02 -5.24488986e-01
3.34282845e-01 1.30388784e+00 1.74121112e-01 -3.96278054e-01
7.52143800e-01 1.72271162e-01 -8.07345092e-01 -9.64583695e-01
-7.33921707e-01 -8.76689911e-01 -9.30588186e-01 1.00092813e-01
1.13059092e+00 -1.18611491e+00 -7.58253932e-02 8.13782275e-01
-1.28031373e+00 -4.31906581e-01 1.55543283e-01 3.01508814e-01
1.99090261e-02 3.69642824e-01 -1.01843655e+00 -9.11361098e-01
-2.39280507e-01 -1.45411551e+00 1.05682802e+00 -4.47586328e-01
-3.77637446e-01 -1.15661430e+00 2.32429549e-01 4.92832720e-01
5.62298656e-01 -1.40274793e-01 1.37990379e+00 -1.05063689e+00
2.33458117e-01 -3.66105616e-01 -1.72082707e-01 7.05743968e-01
1.62453391e-02 3.06659728e-01 -1.14290953e+00 -2.83237785e-01
-3.49134603e-03 -5.33222437e-01 6.53915286e-01 -2.29633912e-01
9.76479411e-01 -3.29821795e-01 1.49918988e-01 5.01230717e-01
1.42950726e+00 1.52526110e-01 3.71225059e-01 4.84015584e-01
1.07426703e+00 5.66759646e-01 3.55710894e-01 5.12705386e-01
5.21460533e-01 3.94324034e-01 4.00804222e-01 -1.68589324e-01
9.00203958e-02 -1.27044618e-01 7.36385822e-01 1.53498876e+00
5.55039108e-01 -1.16428345e-01 -1.56527102e+00 8.11409891e-01
-1.60737288e+00 -5.44824004e-01 -4.92791623e-01 1.61630762e+00
9.71775234e-01 2.40765810e-01 1.06115686e-02 9.29143056e-02
5.30678034e-01 7.46312812e-02 -1.34426668e-01 -9.25439477e-01
-9.26535279e-02 -9.17853639e-02 3.55843514e-01 6.24077797e-01
-1.30256128e+00 1.06008577e+00 5.62784481e+00 8.27219367e-01
-1.27515411e+00 4.39487457e-01 6.94008410e-01 7.88653269e-02
-3.15629780e-01 -1.71725497e-01 -6.95700228e-01 4.91481364e-01
1.12327754e+00 -7.79360160e-02 4.61319178e-01 1.19689035e+00
7.42104501e-02 8.88359994e-02 -9.97806549e-01 8.94883037e-01
2.75597095e-01 -1.03163469e+00 -1.80649042e-01 -1.13527834e-01
8.45386565e-01 7.27920949e-01 1.01384431e-01 1.12216163e+00
3.32289398e-01 -9.03399527e-01 1.04769433e+00 -2.35793456e-01
7.73413777e-01 -7.83895731e-01 1.23550010e+00 5.43893933e-01
-1.25310636e+00 -4.03048277e-01 2.05558911e-01 -3.98689449e-01
-2.14568987e-01 2.88892597e-01 -7.34471142e-01 1.81595713e-01
5.70254445e-01 9.66167271e-01 -1.29432583e+00 5.17735720e-01
-1.53340444e-01 7.30992198e-01 2.49799751e-02 -8.97776559e-02
6.15127325e-01 5.48034087e-02 1.66533202e-01 2.01089644e+00
1.40739426e-01 -5.47725022e-01 2.80369341e-01 8.07999849e-01
-2.61738628e-01 5.00046253e-01 -8.14545095e-01 -4.74637896e-01
2.00740501e-01 1.40073609e+00 -5.63440621e-01 -2.39689603e-01
-1.18743360e+00 8.65972519e-01 4.83312726e-01 2.25562260e-01
-9.59269822e-01 -5.62427282e-01 7.46645391e-01 7.85640106e-02
-1.53235540e-01 -3.98098409e-01 -4.56297845e-01 -1.20725846e+00
-9.50074345e-02 -1.07673895e+00 2.35817015e-01 -7.85496771e-01
-1.33879304e+00 9.30470169e-01 -2.41524965e-01 -1.03685594e+00
-1.55933291e-01 -7.79691696e-01 -4.86578822e-01 8.14888299e-01
-1.59226930e+00 -1.27966118e+00 -8.14284161e-02 4.61132646e-01
8.53088379e-01 -5.92210710e-01 6.99549437e-01 6.58721924e-01
-7.40803599e-01 7.63843298e-01 2.62324929e-01 8.46466780e-01
7.20001221e-01 -1.19598413e+00 5.42651355e-01 1.00522816e+00
7.04677328e-02 8.00825953e-01 3.90907377e-01 -6.61881506e-01
-1.06127393e+00 -1.18814266e+00 1.14131153e+00 -6.48125827e-01
1.26431060e+00 -1.01735866e+00 -9.67738271e-01 7.14786232e-01
4.78735089e-01 -1.92777708e-01 1.01387703e+00 2.98695803e-01
-5.10600507e-01 3.60771596e-01 -1.00528443e+00 5.88115513e-01
6.03419840e-01 -1.23331678e+00 -4.80035812e-01 1.66378871e-01
9.26718593e-01 2.26784945e-02 -5.05912304e-01 2.93090910e-01
1.42489076e-01 -7.78575718e-01 4.54021096e-01 -6.26160800e-01
9.79353607e-01 3.79457287e-02 -2.08476692e-01 -1.29807770e+00
-8.87108520e-02 -4.82515931e-01 3.78104955e-01 1.49827123e+00
7.22252786e-01 -3.17543089e-01 3.47367525e-01 5.86241543e-01
-2.17831627e-01 -7.09003448e-01 -6.23796403e-01 -4.49200302e-01
5.91761768e-01 -7.69180000e-01 2.57071406e-01 1.38270581e+00
2.58417010e-01 4.40586746e-01 -1.60484210e-01 -1.13564454e-01
-2.08850250e-01 -3.60443056e-01 7.33527422e-01 -9.28113222e-01
-1.11481383e-01 -5.15384376e-01 -2.71046311e-01 -5.35009325e-01
7.50671089e-01 -1.28885210e+00 2.64067482e-02 -1.28491521e+00
4.08566803e-01 -3.49984765e-01 -2.08968937e-01 6.95006788e-01
-2.63733625e-01 4.07313257e-01 3.31790119e-01 4.15769294e-02
-4.11290824e-01 5.16119182e-01 4.49661821e-01 -2.24344149e-01
-2.14516327e-01 -3.18272471e-01 -5.20001233e-01 1.02060378e+00
8.37031722e-01 -6.14117324e-01 -2.99631059e-01 -5.73725343e-01
7.88710654e-01 -4.26130652e-01 1.89287253e-02 -9.93076980e-01
-1.60505831e-01 1.73014581e-01 -1.63800284e-01 -3.47021669e-01
2.19599549e-02 -8.95151854e-01 -4.98515218e-01 4.81578290e-01
-6.15402758e-01 4.96993124e-01 6.52206957e-01 9.72835496e-02
-3.59970897e-01 -6.28288567e-01 8.09639513e-01 -1.58304334e-01
-7.57452548e-01 -1.89632803e-01 -1.11068618e+00 3.52464765e-01
8.29896450e-01 6.34370819e-02 -2.01427937e-01 -2.80890405e-01
-4.34654295e-01 1.78481042e-02 5.22321403e-01 9.54313159e-01
2.97513008e-01 -1.03732324e+00 -6.73791170e-01 5.15515268e-01
6.85785592e-01 -6.21443093e-01 -8.84910151e-02 9.26175833e-01
-6.77256644e-01 2.00372428e-01 -9.82000772e-03 -3.79600316e-01
-1.42455542e+00 7.05097079e-01 2.88568318e-01 -4.85851437e-01
-9.81889963e-02 9.20638680e-01 9.28351879e-02 -1.15029609e+00
1.40166003e-02 -4.44546163e-01 -3.17600071e-01 1.92917675e-01
2.04442263e-01 2.28914693e-02 2.52463698e-01 -1.04166269e+00
-4.86465901e-01 2.12478891e-01 -2.49087676e-01 -8.25488269e-02
1.32042181e+00 -9.52103734e-02 -4.61579114e-01 7.74343550e-01
1.58203483e+00 6.85705960e-01 -6.37884080e-01 3.87263410e-02
4.61181641e-01 -2.42794961e-01 -5.15020750e-02 -8.43733907e-01
-9.55118239e-01 1.12946892e+00 3.50711197e-01 1.08959019e-01
8.35526168e-01 -2.00241730e-01 6.53752148e-01 3.47915381e-01
2.25306213e-01 -1.20293820e+00 6.41404763e-02 1.14913166e+00
7.10663557e-01 -1.51023281e+00 -3.79438013e-01 -2.06233904e-01
-9.01124120e-01 7.87185550e-01 8.61638963e-01 1.40050828e-01
9.91154373e-01 7.18649030e-01 5.84794164e-01 6.81789741e-02
-7.17043996e-01 -1.71155527e-01 1.95782825e-01 2.82370120e-01
1.29851699e+00 -8.70063677e-02 -2.04555169e-01 7.86164880e-01
-2.86203235e-01 -3.77171993e-01 7.71925092e-01 1.15982032e+00
-3.09400670e-02 -1.08858025e+00 -1.46219894e-01 3.65879565e-01
-9.69680786e-01 -7.46251941e-01 -5.89335620e-01 8.52512300e-01
2.46260539e-01 1.17057014e+00 -1.72468558e-01 -6.51087403e-01
-1.95457153e-02 4.49103475e-01 -3.50018233e-01 -1.20579350e+00
-1.34206533e+00 -1.48901641e-01 8.35702866e-02 -2.02790216e-01
-2.38174543e-01 -4.06369507e-01 -1.17381716e+00 -2.20797285e-01
-2.19589129e-01 2.26463556e-01 8.85128617e-01 8.83380413e-01
3.38957071e-01 6.08244300e-01 3.26452047e-01 -6.65593147e-01
-1.51690543e-01 -1.13568997e+00 -3.39695692e-01 6.65039957e-01
6.14920378e-01 -2.35100403e-01 -5.31440854e-01 -3.70993875e-02] | [9.411457061767578, 10.45862102508545] |
8370e049-b965-4584-af7f-49c3b5fee77c | faceverse-a-fine-grained-and-detail | 2203.14057 | null | https://arxiv.org/abs/2203.14057v3 | https://arxiv.org/pdf/2203.14057v3.pdf | FaceVerse: a Fine-grained and Detail-controllable 3D Face Morphable Model from a Hybrid Dataset | We present FaceVerse, a fine-grained 3D Neural Face Model, which is built from hybrid East Asian face datasets containing 60K fused RGB-D images and 2K high-fidelity 3D head scan models. A novel coarse-to-fine structure is proposed to take better advantage of our hybrid dataset. In the coarse module, we generate a base parametric model from large-scale RGB-D images, which is able to predict accurate rough 3D face models in different genders, ages, etc. Then in the fine module, a conditional StyleGAN architecture trained with high-fidelity scan models is introduced to enrich elaborate facial geometric and texture details. Note that different from previous methods, our base and detailed modules are both changeable, which enables an innovative application of adjusting both the basic attributes and the facial details of 3D face models. Furthermore, we propose a single-image fitting framework based on differentiable rendering. Rich experiments show that our method outperforms the state-of-the-art methods. | ['Yebin Liu', 'Liang Li', 'Chenguang Ma', 'Tao Yu', 'ZhiYuan Chen', 'Lizhen Wang'] | 2022-03-26 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Wang_FaceVerse_A_Fine-Grained_and_Detail-Controllable_3D_Face_Morphable_Model_From_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Wang_FaceVerse_A_Fine-Grained_and_Detail-Controllable_3D_Face_Morphable_Model_From_CVPR_2022_paper.pdf | cvpr-2022-1 | ['3d-face-reconstruction', 'face-model', 'face-reconstruction'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-2.98867077e-01 3.38333726e-01 1.99130550e-01 -8.40303004e-01
-4.74871725e-01 -3.08094341e-02 5.36479414e-01 -1.06695330e+00
1.66159689e-01 2.91761875e-01 8.33853558e-02 1.97590753e-01
8.90398324e-02 -9.43173349e-01 -8.04818749e-01 -6.05323315e-01
1.90124914e-01 7.34527290e-01 -2.11917937e-01 -3.38702470e-01
-1.09563261e-01 1.00193787e+00 -1.86309314e+00 2.92985104e-02
6.40923321e-01 1.48840392e+00 -3.04343462e-01 3.08877856e-01
-2.62638628e-01 1.62482172e-01 -1.04714595e-01 -7.56575584e-01
4.79471356e-01 -3.04194689e-01 -2.53931254e-01 3.09664816e-01
7.39464641e-01 -7.20309377e-01 -3.08525681e-01 7.38131702e-01
6.58527672e-01 -1.79712012e-01 7.57888854e-01 -1.29799938e+00
-9.16158140e-01 9.26333889e-02 -7.07817793e-01 -6.91373229e-01
3.45895499e-01 7.57640377e-02 1.93871439e-01 -1.14054656e+00
4.74492580e-01 1.93588603e+00 8.53010833e-01 1.05911136e+00
-9.68365610e-01 -1.22120404e+00 2.33412445e-01 -1.93500265e-01
-1.47183585e+00 -6.88520610e-01 1.08956897e+00 -2.29877010e-01
3.18148494e-01 7.33419806e-02 9.54171360e-01 1.22723019e+00
-3.51706706e-03 3.64079744e-01 1.41551161e+00 -1.98897272e-01
-6.52380139e-02 4.12820512e-03 -4.70498234e-01 1.22320783e+00
-1.26544297e-01 2.77853131e-01 -5.98495245e-01 -1.68625936e-01
1.35337842e+00 1.76628783e-01 7.82576799e-02 -4.19266522e-01
-7.57295310e-01 5.64937353e-01 4.75149423e-01 -3.99625115e-02
-2.97816634e-01 6.08982965e-02 -3.54283512e-01 1.57257635e-03
7.97700822e-01 -2.54068971e-01 -6.40738487e-01 1.38722315e-01
-9.76990044e-01 3.74483943e-01 5.37643492e-01 1.13232136e+00
1.16258788e+00 1.76542073e-01 -2.76682638e-02 7.39330590e-01
7.78808057e-01 9.09949839e-01 2.82381684e-01 -1.16412973e+00
-4.41491194e-02 7.85830736e-01 -2.88881958e-01 -8.30725789e-01
-4.35066640e-01 -1.99989930e-01 -1.04715753e+00 2.83957303e-01
1.27415776e-01 1.62809506e-01 -1.36054540e+00 1.66468358e+00
8.06518674e-01 4.08930659e-01 -3.45185280e-01 7.41617620e-01
1.24790692e+00 2.92355537e-01 -1.60071060e-01 -7.41073443e-03
1.38588870e+00 -7.36592233e-01 -5.92178285e-01 1.11426860e-01
-5.70903812e-03 -4.98343378e-01 1.00364292e+00 1.93728343e-01
-1.28832579e+00 -8.19634736e-01 -6.72233760e-01 -4.65649337e-01
-3.96032780e-01 1.16108254e-01 7.04188526e-01 9.85741436e-01
-1.35096753e+00 4.87002641e-01 -5.92720509e-01 -1.57268673e-01
8.42819452e-01 6.51856959e-01 -6.91791832e-01 -1.96108222e-01
-9.41030145e-01 6.48003399e-01 -6.38561249e-02 5.20355940e-01
-9.31744218e-01 -9.67385948e-01 -9.98824418e-01 -2.42359772e-01
1.27854496e-02 -9.35269713e-01 9.00800705e-01 -8.01890433e-01
-2.11521316e+00 1.33349586e+00 -2.19199389e-01 4.39381212e-01
6.54363692e-01 -1.96449528e-03 -3.05853635e-01 -5.69505282e-02
-4.05727118e-01 8.06601942e-01 1.48147476e+00 -1.34207439e+00
-5.14085926e-02 -9.57689703e-01 -1.89568385e-01 1.75723899e-02
-2.32091874e-01 1.10350192e-01 -1.00798643e+00 -6.18786871e-01
1.96696654e-01 -8.43602180e-01 -1.63643062e-01 5.35231054e-01
-1.77957326e-01 1.07025430e-02 8.13945591e-01 -8.32598805e-01
6.70057654e-01 -2.12061954e+00 3.20118904e-01 4.69603539e-01
4.24949586e-01 -2.85873003e-02 -2.05824167e-01 -4.55110282e-01
-3.61880995e-02 2.01438680e-01 -1.40181571e-01 -1.03697968e+00
3.31403971e-01 2.20012084e-01 1.95709430e-02 3.65524650e-01
4.71748143e-01 9.48485911e-01 -2.77595550e-01 -6.22364521e-01
1.05448335e-01 1.06400275e+00 -6.63895607e-01 3.82453591e-01
-8.62402469e-02 8.53970885e-01 -6.69807315e-01 1.17741477e+00
1.31235051e+00 -1.40281647e-01 -2.29441836e-01 -2.85678595e-01
2.09836990e-01 -4.05175447e-01 -1.11656284e+00 1.93573368e+00
-3.75890881e-01 -2.29178458e-01 3.40674490e-01 -3.65022391e-01
1.32355142e+00 9.99365076e-02 3.21953773e-01 -7.36940920e-01
3.61979574e-01 2.32387021e-01 -6.66530550e-01 -2.15865195e-01
2.23856211e-01 -1.28974661e-01 -1.07966075e-02 2.42640078e-01
1.01197220e-01 -5.76143682e-01 -6.42416894e-01 -2.37404987e-01
3.14546466e-01 7.91327059e-01 -1.85283586e-01 -1.40305027e-01
5.82892120e-01 -7.79321790e-01 7.02804446e-01 1.11346871e-01
2.61172317e-02 1.07794380e+00 3.75159264e-01 -7.51359344e-01
-9.12761092e-01 -9.14363801e-01 -2.63762534e-01 8.58521938e-01
-1.56804785e-01 -2.88726628e-01 -1.11551511e+00 -5.78942001e-01
2.47096762e-01 -1.03768855e-01 -1.22360671e+00 -2.23169103e-02
-6.18348241e-01 -6.55586660e-01 5.03416657e-01 3.88209283e-01
8.01835537e-01 -8.32426846e-01 1.62294134e-02 -3.06966156e-01
2.87247807e-01 -1.06713104e+00 -5.19301891e-01 -4.69882727e-01
-8.77897620e-01 -8.96911025e-01 -8.07051182e-01 -5.86807609e-01
9.33463216e-01 -3.18325609e-01 1.13731480e+00 3.39726508e-01
-2.20288202e-01 3.37589085e-01 -9.46328118e-02 -4.71505463e-01
-3.16641688e-01 -1.14465684e-01 3.28019291e-01 4.48360503e-01
1.87502742e-01 -6.96494043e-01 -5.17965317e-01 3.52797061e-01
-7.36486733e-01 2.44096413e-01 5.55377483e-01 5.09788036e-01
8.58051896e-01 -4.43487108e-01 1.28841400e-01 -7.85713851e-01
-4.17233482e-02 -1.57478228e-01 -6.47887588e-01 3.87859732e-01
-5.50464094e-01 1.09200120e-01 2.83087820e-01 -4.71728146e-01
-1.29076469e+00 1.72807083e-01 -4.97425199e-01 -8.15733671e-01
-2.81028837e-01 -2.95990855e-01 -8.37840974e-01 -5.77319086e-01
2.10462481e-01 7.69400075e-02 3.24798852e-01 -8.14427793e-01
4.60973382e-01 6.04865074e-01 4.72741336e-01 -9.44323897e-01
1.12539089e+00 5.60804248e-01 3.30644846e-01 -6.59560561e-01
-5.00388861e-01 4.82131362e-01 -1.08844197e+00 -2.26482913e-01
7.28568256e-01 -1.05366802e+00 -8.93826544e-01 1.10656869e+00
-1.02897775e+00 -4.92258251e-01 -1.84186533e-01 3.52109922e-03
-6.24560416e-01 -2.57421881e-02 -6.90622389e-01 -6.25566006e-01
-2.45577946e-01 -1.33321297e+00 1.71697879e+00 5.64508259e-01
3.95896792e-01 -6.66613936e-01 -1.79592371e-01 3.42927992e-01
4.24947232e-01 4.85129118e-01 6.72645450e-01 8.76210481e-02
-6.41360939e-01 5.96380159e-02 -2.85709471e-01 2.61458188e-01
1.65460497e-01 2.46924892e-01 -1.38667500e+00 -7.22743496e-02
-1.09523185e-01 -3.38679880e-01 6.16673172e-01 3.60781640e-01
1.60911632e+00 -1.40027717e-01 -2.12599218e-01 1.51185822e+00
9.80028331e-01 -1.90574169e-01 8.84400606e-01 -1.45773739e-01
1.06446588e+00 6.56768441e-01 3.75747472e-01 5.14734507e-01
8.01736236e-01 6.46155298e-01 5.20747662e-01 -3.72967094e-01
-2.04705924e-01 -4.55891639e-01 2.02837914e-01 6.50527000e-01
-6.20778501e-01 5.10282278e-01 -4.99506384e-01 -1.32711530e-01
-1.23126173e+00 -4.78660405e-01 3.01954150e-01 1.86515641e+00
8.63368928e-01 -2.96744078e-01 7.70406798e-02 -1.90700218e-01
4.65168625e-01 4.04396243e-02 -7.19385266e-01 -1.08561225e-01
-2.86334455e-01 6.22400820e-01 4.65583168e-02 3.54253441e-01
-7.10839868e-01 1.09259462e+00 6.27769947e+00 8.36232543e-01
-1.11578751e+00 -1.50283461e-03 1.02282512e+00 -5.08538261e-02
-6.53243780e-01 -3.14722985e-01 -1.14545703e+00 4.40611094e-01
7.25093365e-01 4.38917577e-01 4.99824077e-01 7.71412492e-01
-1.52867198e-01 4.69470352e-01 -8.92651081e-01 1.21488488e+00
3.07055503e-01 -1.19761026e+00 3.35148960e-01 4.70052809e-01
7.04384685e-01 -4.91702795e-01 4.11684066e-01 4.15779293e-01
1.16535023e-01 -1.34130144e+00 8.72029960e-01 1.10860610e+00
1.51591706e+00 -7.64057279e-01 2.67252445e-01 8.57605189e-02
-1.10907304e+00 1.84611380e-01 -4.04686183e-01 3.22802186e-01
-1.57316834e-01 4.31183219e-01 -1.94300696e-01 6.49581254e-01
1.14694929e+00 6.05876446e-01 -7.53268480e-01 3.35960716e-01
-5.17624319e-02 -6.08846433e-02 -4.21737820e-01 3.60390455e-01
-2.75210559e-01 -3.40024143e-01 -9.21189487e-02 7.27840126e-01
7.28100300e-01 4.90188956e-01 -1.57211155e-01 9.57555473e-01
-3.92483622e-01 -5.56622110e-02 -4.60080773e-01 3.03477526e-01
4.27108318e-01 1.47365201e+00 -3.14602435e-01 -1.36473656e-01
-2.74059653e-01 9.83573675e-01 3.06372553e-01 1.82579547e-01
-7.47784853e-01 1.46201447e-01 8.15774143e-01 2.39278302e-01
2.82009363e-01 -1.85819238e-01 -3.01828217e-02 -1.24284077e+00
-1.52715132e-01 -8.90473068e-01 -1.50906723e-02 -9.30163920e-01
-1.17921221e+00 7.95758307e-01 -1.70116410e-01 -8.44188988e-01
-2.31081635e-01 -9.46026623e-01 -2.29176030e-01 1.10214543e+00
-1.78402340e+00 -2.15173650e+00 -6.52837396e-01 1.00433552e+00
3.33252028e-02 -2.70101845e-01 1.03987467e+00 2.34523907e-01
-5.70084333e-01 9.37251389e-01 -2.74801135e-01 1.28794298e-01
8.07066083e-01 -7.78018236e-01 6.00615203e-01 3.95203143e-01
-2.23761022e-01 4.81616199e-01 3.16898078e-02 -6.18478477e-01
-1.77206910e+00 -1.13591802e+00 4.78656232e-01 -6.46434307e-01
-3.51942442e-02 -6.57078147e-01 -7.69138694e-01 7.32794344e-01
-3.81909549e-01 5.52870810e-01 5.87664425e-01 -4.20682970e-03
-5.77359557e-01 -5.68587184e-01 -1.66190851e+00 4.61481541e-01
1.39906752e+00 -5.11233985e-01 -1.44614935e-01 -9.30279717e-02
7.70198166e-01 -9.06257272e-01 -1.17519605e+00 5.44553220e-01
1.00769114e+00 -1.14917409e+00 1.10407853e+00 -4.83803511e-01
2.91207671e-01 -1.38356283e-01 -2.59923756e-01 -1.07651842e+00
-2.94378370e-01 -6.76568210e-01 -3.97428423e-01 1.32894015e+00
-1.16862319e-01 -5.28609931e-01 9.13352013e-01 7.86595285e-01
-4.31360379e-02 -8.46801162e-01 -8.88901889e-01 -2.37710103e-01
2.10639670e-01 -3.95556331e-01 1.77684414e+00 7.05876172e-01
-8.76828015e-01 -2.39368871e-01 -4.71127093e-01 1.04221582e-01
8.60852659e-01 9.67646167e-02 1.01964211e+00 -1.71058202e+00
4.59765792e-02 -3.79136682e-01 -2.53133446e-01 -9.41550732e-01
6.09247148e-01 -5.59716582e-01 -3.14802736e-01 -7.90959477e-01
1.02495469e-01 -6.81550086e-01 2.74821296e-02 6.08791173e-01
4.51901229e-03 8.90621305e-01 5.54437488e-02 -5.41800074e-02
-3.93570885e-02 9.48051512e-01 1.71669471e+00 4.47834171e-02
-3.54136974e-02 -1.84962377e-02 -7.89010108e-01 7.46993959e-01
3.53586525e-01 1.14644840e-02 -5.22366129e-02 -5.04236877e-01
1.04799747e-01 1.29241832e-02 5.38824975e-01 -6.68550193e-01
-1.23684824e-01 -1.67174622e-01 1.09724987e+00 -5.24238467e-01
7.90008783e-01 -9.03588891e-01 4.33461994e-01 -1.62714664e-02
2.18984470e-01 -6.45060390e-02 2.19608590e-01 2.03546956e-01
2.36056000e-01 4.54890490e-01 9.95968282e-01 -3.26245576e-01
-3.86445791e-01 1.27150583e+00 5.82506180e-01 -3.78138870e-01
8.09767544e-01 -4.69690919e-01 5.46188094e-02 -1.41384378e-01
-7.47582436e-01 -6.41532317e-02 8.87528419e-01 4.97546822e-01
6.93453431e-01 -1.71550846e+00 -8.03804874e-01 1.14282846e+00
9.13107023e-03 4.10624653e-01 5.62350929e-01 4.17243093e-01
-3.60441297e-01 -3.14978771e-02 -5.35510480e-01 -6.19822145e-01
-1.05151415e+00 2.71461576e-01 5.17057121e-01 2.09976599e-01
-3.92146081e-01 9.38853621e-01 2.35986367e-01 -9.93136346e-01
1.48465931e-01 -2.93628961e-01 -1.79930344e-01 -4.33911458e-02
7.02500343e-01 6.33000284e-02 1.15750067e-01 -1.09652483e+00
-3.15469503e-01 1.38501048e+00 2.90126085e-01 6.97244182e-02
1.57502079e+00 -2.03656688e-01 -4.43901390e-01 1.59609750e-01
1.08749866e+00 -5.87323681e-02 -1.76054156e+00 -1.94599211e-01
-7.78774500e-01 -6.36609554e-01 -3.77993695e-02 -5.10285735e-01
-1.49653852e+00 9.80247080e-01 7.30585933e-01 -4.34801131e-01
1.49780989e+00 6.99149817e-02 6.07347369e-01 -1.39762536e-01
5.67426145e-01 -6.51412427e-01 -1.15243094e-02 5.33344328e-01
1.01697767e+00 -1.19247091e+00 -3.54264826e-02 -4.26403701e-01
-2.27691799e-01 1.20571852e+00 7.49434292e-01 7.19725192e-02
9.70082879e-01 3.09022605e-01 3.56366783e-02 -1.87898964e-01
-2.79630333e-01 1.23351887e-01 4.61354256e-01 8.67118895e-01
1.29852667e-01 -3.10436133e-02 4.39498901e-01 8.83073390e-01
-5.71099401e-01 1.17323712e-01 -3.03812213e-02 4.07317072e-01
1.15547679e-01 -1.21778202e+00 -6.58150733e-01 1.20418601e-01
-2.26006627e-01 2.03328311e-01 -1.61227778e-01 8.58944178e-01
4.85911816e-01 2.49801233e-01 2.86015213e-01 -5.70516467e-01
3.80557537e-01 1.47009745e-01 9.61888015e-01 -3.12418133e-01
-3.66394073e-01 1.10717885e-01 -4.85746384e-01 -9.13243234e-01
-4.80576456e-01 -5.31427085e-01 -8.90495181e-01 -6.54275835e-01
-2.31884383e-02 -2.49559075e-01 8.76053810e-01 7.80454636e-01
5.97146451e-01 2.85843939e-01 8.49489391e-01 -1.51351571e+00
-2.42207259e-01 -9.12626266e-01 -9.32735860e-01 1.81233853e-01
4.76426661e-01 -9.07738388e-01 -1.82255492e-01 1.46686718e-01] | [12.994573593139648, -0.10028346627950668] |
439990e1-9cc3-41e4-922b-578156822843 | unscene3d-unsupervised-3d-instance | 2303.14541 | null | https://arxiv.org/abs/2303.14541v1 | https://arxiv.org/pdf/2303.14541v1.pdf | UnScene3D: Unsupervised 3D Instance Segmentation for Indoor Scenes | 3D instance segmentation is fundamental to geometric understanding of the world around us. Existing methods for instance segmentation of 3D scenes rely on supervision from expensive, manual 3D annotations. We propose UnScene3D, the first fully unsupervised 3D learning approach for class-agnostic 3D instance segmentation of indoor scans. UnScene3D first generates pseudo masks by leveraging self-supervised color and geometry features to find potential object regions. We operate on a basis of geometric oversegmentation, enabling efficient representation and learning on high-resolution 3D data. The coarse proposals are then refined through self-training our model on its predictions. Our approach improves over state-of-the-art unsupervised 3D instance segmentation methods by more than 300% Average Precision score, demonstrating effective instance segmentation even in challenging, cluttered 3D scenes. | ['Angela Dai', 'Or Litany', 'David Rozenberszki'] | 2023-03-25 | null | null | null | null | ['3d-instance-segmentation-1'] | ['computer-vision'] | [ 4.66073602e-01 5.64496696e-01 -3.81852627e-01 -5.20981550e-01
-9.97778535e-01 -8.41622531e-01 6.29606307e-01 3.05750400e-01
-8.03487077e-02 4.84692417e-02 -2.91641593e-01 -4.78996724e-01
8.65026414e-02 -7.95778215e-01 -8.78783762e-01 -7.95606673e-02
-1.55317754e-01 1.31503260e+00 6.70028925e-01 2.26634651e-01
2.86660999e-01 1.08914483e+00 -1.53731787e+00 -3.66082937e-02
8.55154216e-01 9.89856303e-01 -4.95280102e-02 7.28540599e-01
-6.51647747e-01 -1.00970879e-01 -2.25075170e-01 -1.33233279e-01
7.88188398e-01 8.49656612e-02 -9.88005519e-01 8.11341465e-01
7.49040365e-01 -2.47948021e-01 1.99449390e-01 7.43088186e-01
1.16436362e-01 1.07662492e-01 1.02269042e+00 -7.82192647e-01
-2.24913494e-03 9.67783928e-02 -9.07669008e-01 -1.02918237e-01
4.41069573e-01 7.07975700e-02 8.60563159e-01 -9.51567113e-01
7.42575705e-01 1.23199344e+00 7.79158056e-01 3.67070377e-01
-1.52296531e+00 -4.20168072e-01 4.62746590e-01 -5.71222663e-01
-1.45843697e+00 4.77929600e-02 9.44060862e-01 -5.75934172e-01
1.05841768e+00 3.53212446e-01 7.35897005e-01 5.82609415e-01
-2.73203522e-01 1.06791914e+00 1.30082715e+00 -3.87493193e-01
4.37687159e-01 -5.37205525e-02 1.10675968e-01 9.85062718e-01
1.89357251e-01 -1.04936622e-01 -1.20280594e-01 1.13883629e-01
1.15091956e+00 -1.32652760e-01 4.81895328e-01 -1.06635892e+00
-9.00383592e-01 4.44829136e-01 7.10810065e-01 -1.10771269e-01
-2.77424365e-01 6.52187839e-02 -9.54000875e-02 -2.78825372e-01
8.68572235e-01 4.73954171e-01 -7.34788120e-01 9.62370485e-02
-1.20630038e+00 2.88781315e-01 5.04521728e-01 1.16431463e+00
1.07340026e+00 -2.87359774e-01 2.80216336e-01 6.75155699e-01
3.12654197e-01 6.03619874e-01 -3.08070183e-01 -1.19355440e+00
4.31064218e-01 1.03726220e+00 9.84816849e-02 -5.23579419e-01
-6.16057456e-01 -4.12041962e-01 -3.32712263e-01 3.54355693e-01
3.50042611e-01 3.48071665e-01 -1.83964300e+00 8.08671832e-01
9.64065492e-01 -4.68505993e-02 -3.70758533e-01 7.20578313e-01
8.63006949e-01 2.85425603e-01 1.08087994e-02 3.65069598e-01
8.79232228e-01 -7.86821485e-01 1.94992483e-01 -4.58210021e-01
5.09646237e-01 -5.74409068e-01 9.85144377e-01 4.49160397e-01
-1.14884281e+00 -5.86064458e-01 -7.63297558e-01 -9.68268961e-02
-5.20856261e-01 -1.26876473e-01 7.50106394e-01 9.95724261e-01
-6.38533950e-01 3.34780514e-01 -1.03514051e+00 -1.86627224e-01
1.27781987e+00 5.84522605e-01 -3.38485807e-01 -2.26145118e-01
-1.67386219e-01 6.01964116e-01 5.74917197e-01 -3.18506837e-01
-9.77240026e-01 -9.94337142e-01 -1.06778109e+00 -5.68314195e-01
6.09953165e-01 -5.83885610e-01 1.23581374e+00 -4.27555293e-01
-1.29871273e+00 1.73183870e+00 -5.39774187e-02 -4.16785657e-01
7.54112899e-01 -4.52906013e-01 2.12712109e-01 1.49681285e-01
3.52138132e-01 8.93205047e-01 7.33033061e-01 -1.81431961e+00
-6.61020815e-01 -7.05469966e-01 5.76828532e-02 1.27617687e-01
5.40413022e-01 -6.53733432e-01 -9.74716246e-01 -2.95490324e-01
9.00538921e-01 -8.24443698e-01 -1.00055444e+00 -2.81302016e-02
-9.24933553e-01 -1.22399271e-01 9.03149784e-01 -2.93281555e-01
5.43407619e-01 -1.80167735e+00 -2.28583977e-01 8.20490539e-01
2.47268990e-01 8.27909112e-02 1.28444523e-01 -1.96513310e-01
2.07736343e-01 4.13918287e-01 -6.68863833e-01 -4.46479052e-01
2.51781136e-01 3.26769888e-01 2.72203404e-02 4.80598480e-01
5.79707503e-01 9.39971387e-01 -1.05391121e+00 -7.62246847e-01
7.76921690e-01 3.43535483e-01 -7.34286189e-01 2.28413314e-01
-7.50147700e-01 7.08668470e-01 -8.98487389e-01 1.19017589e+00
9.23305511e-01 -3.35793644e-01 -7.68700168e-02 -1.22370176e-01
-1.99852526e-01 3.23480934e-01 -1.29282641e+00 2.29201794e+00
-3.85239840e-01 2.93088481e-02 -4.65205051e-02 -9.38146830e-01
1.08267570e+00 -3.30924392e-01 7.66282618e-01 -5.93294978e-01
6.45109117e-02 3.72909158e-01 -6.08608842e-01 -1.91787943e-01
2.83441424e-01 -3.82466726e-02 -4.20685202e-01 6.92787886e-01
9.02492926e-02 -1.38034809e+00 -1.67865813e-01 2.99469262e-01
8.97575080e-01 7.69598901e-01 2.71313846e-01 -2.59442478e-01
2.32420444e-01 4.03756022e-01 1.61881372e-01 9.57333922e-01
-2.67548999e-03 8.41510952e-01 1.90170288e-01 -5.46184361e-01
-1.07323563e+00 -1.47036660e+00 -3.48029256e-01 6.12899005e-01
5.60071051e-01 -1.76024154e-01 -8.83368075e-01 -1.34481740e+00
2.16760203e-01 6.64580703e-01 -5.77390790e-01 5.52659214e-01
-6.08301103e-01 -6.28395498e-01 1.41494244e-01 5.33950627e-01
5.79465747e-01 -5.72607517e-01 -6.94361925e-01 1.72651723e-01
4.25058424e-01 -1.44822955e+00 1.43342223e-02 6.34460330e-01
-1.13070774e+00 -1.11973834e+00 -5.55267930e-01 -5.82045615e-01
1.13663793e+00 2.33342797e-01 1.58660769e+00 -5.66437766e-02
-8.76308680e-01 9.20719683e-01 -1.67807266e-01 -5.11447906e-01
-2.01005876e-01 3.52476567e-01 -2.09702387e-01 -3.94108862e-01
2.47277334e-01 -6.22379661e-01 -5.35707891e-01 3.43859702e-01
-5.16676009e-01 1.27118945e-01 7.30975211e-01 1.29744038e-01
1.56278443e+00 7.11248219e-02 -7.56736100e-02 -1.46474242e+00
-2.86374062e-01 -1.79748520e-01 -8.47120881e-01 -3.26854400e-02
-3.18475753e-01 1.02510154e-01 -1.31300718e-01 3.45813320e-03
-1.06905735e+00 7.04924464e-01 -2.19144523e-01 -3.91379952e-01
-9.73674953e-01 -3.98008704e-01 -2.89718628e-01 -2.82812327e-01
6.07095480e-01 -5.53718023e-02 -4.89906788e-01 -6.00941896e-01
7.97802210e-01 2.61540294e-01 4.71393257e-01 -9.10221100e-01
1.41926515e+00 7.86733985e-01 2.02850699e-01 -8.62537026e-01
-1.26146853e+00 -6.56506658e-01 -1.57476687e+00 -2.40674913e-01
1.10596752e+00 -7.28730857e-01 -6.02456629e-02 6.03259578e-02
-8.19378793e-01 -8.51683855e-01 -6.17215097e-01 1.71472162e-01
-7.45869398e-01 6.33569434e-02 -1.91324070e-01 -1.06710243e+00
1.27144098e-01 -8.95672739e-01 1.77182794e+00 1.08926907e-01
-1.26702651e-01 -7.94414878e-01 -1.57739371e-01 6.83933258e-01
-3.27481508e-01 6.18328691e-01 8.23549569e-01 -4.83650774e-01
-1.12314284e+00 -2.52324700e-01 -2.98914760e-01 1.53529897e-01
1.94916546e-01 -8.49656463e-02 -1.14749646e+00 3.99676502e-01
-5.21463931e-01 -3.23878676e-01 7.91604757e-01 4.30071294e-01
1.75863326e+00 1.77277252e-01 -6.53526306e-01 6.61015749e-01
1.41192222e+00 -2.87228256e-01 2.87641197e-01 2.33115077e-01
9.47018743e-01 6.18215144e-01 9.06394720e-01 2.36906216e-01
4.74788994e-01 2.18529776e-01 8.28725994e-01 -5.62424302e-01
-1.53057903e-01 -4.81419444e-01 -4.19508517e-01 1.60497595e-02
-1.70470327e-01 1.42482772e-01 -1.25778699e+00 6.35254622e-01
-1.25492561e+00 -3.59565079e-01 -4.12888646e-01 2.08897233e+00
5.78050911e-01 8.07216704e-01 3.42495084e-01 2.50656575e-01
4.10340369e-01 1.29549399e-01 -8.42691183e-01 -2.27187425e-01
2.26125240e-01 7.72765756e-01 8.65127623e-01 6.10167623e-01
-1.63837707e+00 1.30884182e+00 6.17916346e+00 7.67998040e-01
-5.13720870e-01 -1.49979636e-01 1.07148230e+00 1.04657970e-01
-4.28108394e-01 -1.60466775e-01 -8.53414953e-01 -1.66145340e-01
1.03822164e-01 5.47468305e-01 -2.15541832e-02 1.08833778e+00
1.33086622e-01 -3.53339553e-01 -9.98889744e-01 1.02784312e+00
-4.17808741e-02 -1.40880871e+00 5.66105880e-02 2.00538367e-01
1.19024169e+00 1.87312171e-01 -8.44863504e-02 8.56319536e-03
7.26035416e-01 -1.05798733e+00 6.64955914e-01 2.98233688e-01
7.97236383e-01 -8.13067257e-01 1.41767353e-01 5.82261980e-01
-1.36859322e+00 2.81775862e-01 8.68827011e-03 3.24530721e-01
2.46741176e-01 9.59779263e-01 -1.21757305e+00 4.89315808e-01
8.21859777e-01 5.82614124e-01 -6.90979004e-01 9.64516342e-01
-3.66883516e-01 4.45844948e-01 -6.05218410e-01 4.26763356e-01
5.88549316e-01 -1.69362202e-01 4.62821215e-01 1.39766812e+00
-6.60476387e-02 3.78741801e-01 7.02534378e-01 1.08186185e+00
-8.44324008e-02 -1.68499544e-01 -7.02700436e-01 2.68217355e-01
2.75961637e-01 1.22184348e+00 -1.51846993e+00 -3.13760012e-01
3.06207548e-05 1.03890872e+00 -5.38754417e-03 3.22752669e-02
-6.68086052e-01 3.04944534e-02 4.19981122e-01 4.70114738e-01
5.64280033e-01 -7.40020514e-01 -7.28199065e-01 -6.25816941e-01
-5.59830368e-02 -2.12611690e-01 2.03672215e-01 -4.96539861e-01
-1.20743430e+00 1.69270024e-01 1.34531543e-01 -1.06725430e+00
-8.30273330e-02 -7.75308251e-01 -2.65533566e-01 3.20845366e-01
-1.78618121e+00 -1.58922696e+00 -3.68518859e-01 3.33609194e-01
9.91821826e-01 4.66661543e-01 7.82913625e-01 -1.18890949e-01
-2.77713895e-01 7.04746991e-02 -4.49859440e-01 1.04569629e-01
5.10871746e-02 -1.76647091e+00 8.27717721e-01 6.06671453e-01
3.12502533e-01 1.73649088e-01 2.79920161e-01 -8.19384992e-01
-1.19945598e+00 -1.45962536e+00 2.81268030e-01 -8.93959165e-01
1.69462234e-01 -7.66223133e-01 -5.23000121e-01 5.23331046e-01
-5.97707570e-01 3.97319943e-01 5.68928778e-01 1.53925018e-02
-4.22785044e-01 2.51894653e-01 -1.44052410e+00 3.54412913e-01
1.77069306e+00 -3.53984058e-01 -5.07969439e-01 5.18375039e-01
7.25486815e-01 -8.37202907e-01 -9.11297381e-01 6.16641939e-01
3.19437653e-01 -1.05208659e+00 1.45020568e+00 -4.06017363e-01
2.44333431e-01 -4.16239828e-01 -2.26015285e-01 -7.03878105e-01
6.53637350e-02 -5.05103648e-01 -1.37989059e-01 9.26483393e-01
5.32620072e-01 -5.38086183e-02 1.54348099e+00 6.87340021e-01
-4.22695100e-01 -6.77010894e-01 -7.81753719e-01 -7.39299357e-01
-2.89665349e-02 -1.08932793e+00 6.78137422e-01 7.93177903e-01
-9.14273381e-01 1.44128464e-02 5.12754500e-01 5.36018908e-01
1.18424010e+00 4.76734817e-01 1.28299403e+00 -1.66010451e+00
5.42704947e-02 -4.44695771e-01 -5.39985597e-01 -1.58852029e+00
1.75786108e-01 -8.87842715e-01 1.78164050e-01 -1.71084714e+00
-8.67895409e-02 -1.14677227e+00 2.87490398e-01 4.23060596e-01
8.12120736e-02 7.67657995e-01 -2.71765381e-01 -5.91989942e-02
-9.50648427e-01 1.58181012e-01 1.41541576e+00 -3.30845803e-01
-6.26833260e-01 2.13672683e-01 -3.74433100e-01 1.11795700e+00
7.02319741e-01 -2.41267651e-01 -3.85480106e-01 -3.93708527e-01
-3.07275683e-01 -5.83865762e-01 5.11304080e-01 -1.05935705e+00
-3.25942159e-01 -4.17163819e-01 9.41533506e-01 -1.41672146e+00
5.67458510e-01 -9.16218460e-01 -3.43932748e-01 1.41814858e-01
-7.31710941e-02 -6.15675449e-01 3.37520272e-01 4.91224766e-01
4.31782156e-01 5.03403060e-02 7.13401020e-01 -8.26161683e-01
-9.34750021e-01 6.51936471e-01 -2.72051860e-02 1.73445612e-01
1.14703977e+00 -8.69224310e-01 4.16035920e-01 1.94347605e-01
-9.04680908e-01 2.61273235e-01 7.03976929e-01 9.42181274e-02
7.47732699e-01 -8.58602405e-01 -3.86332154e-01 5.32867074e-01
7.45084584e-02 1.14009237e+00 1.04443341e-01 2.37030208e-01
-5.92084110e-01 3.00056487e-01 2.20135644e-01 -1.23874402e+00
-1.04113972e+00 7.83979073e-02 4.71169353e-01 -2.61768192e-01
-8.64139676e-01 1.03537047e+00 2.81675518e-01 -1.02610242e+00
1.46998182e-01 -5.50682187e-01 1.78589270e-01 -3.09193701e-01
-9.88272652e-02 1.72702804e-01 2.27752686e-01 -5.35123944e-01
-5.64051688e-01 1.00532436e+00 1.61224604e-02 -1.88446939e-01
1.54246581e+00 7.61199323e-03 2.87933111e-01 2.94239461e-01
9.00060177e-01 3.30470875e-02 -1.46938932e+00 -1.49306044e-01
2.12680534e-01 -7.25457907e-01 2.01099455e-01 -8.04700375e-01
-7.26970136e-01 8.36004496e-01 5.17150879e-01 6.51517585e-02
6.95211411e-01 6.82797253e-01 5.97132325e-01 3.76740843e-01
7.01712549e-01 -9.99166548e-01 2.08865479e-01 4.24806416e-01
3.89476746e-01 -1.38881826e+00 4.17126983e-01 -9.09449339e-01
-1.69076368e-01 1.06228602e+00 7.03778625e-01 -3.96924466e-01
8.46480370e-01 6.84478879e-02 -3.88897061e-02 -6.66783392e-01
2.34846681e-01 -5.37821829e-01 6.91339314e-01 1.01158762e+00
5.94577678e-02 1.92577317e-01 4.45588857e-01 2.89041340e-01
-3.32338482e-01 -5.58068335e-01 -1.81887504e-02 1.02599084e+00
-4.91316319e-01 -1.19592094e+00 -6.32955253e-01 5.57139516e-01
3.98182310e-02 3.47584277e-01 -6.09327912e-01 1.14480388e+00
4.19306517e-01 4.10872430e-01 3.40819120e-01 -1.62050113e-01
4.48004216e-01 -1.24430269e-01 8.05389583e-01 -1.13862252e+00
-1.08824477e-01 2.72167474e-01 -2.55032927e-02 -9.02940273e-01
-5.74457169e-01 -7.82278955e-01 -1.65082181e+00 3.91058266e-01
-2.41466925e-01 -2.13828862e-01 8.59830081e-01 9.97593820e-01
2.12348878e-01 2.75041759e-01 7.72298992e-01 -1.61818945e+00
2.24640086e-01 -2.53256381e-01 -4.28971767e-01 1.39945015e-01
2.08689913e-01 -7.37021744e-01 -1.69013575e-01 8.97751302e-02] | [7.9958577156066895, -3.25665020942688] |
509efe49-df0f-4e5a-9c7e-e1e46acc2cfd | transferring-textual-knowledge-for-visual | 2207.01297 | null | https://arxiv.org/abs/2207.01297v4 | https://arxiv.org/pdf/2207.01297v4.pdf | Revisiting Classifier: Transferring Vision-Language Models for Video Recognition | Transferring knowledge from task-agnostic pre-trained deep models for downstream tasks is an important topic in computer vision research. Along with the growth of computational capacity, we now have open-source vision-language pre-trained models in large scales of the model architecture and amount of data. In this study, we focus on transferring knowledge for video classification tasks. Conventional methods randomly initialize the linear classifier head for vision classification, but they leave the usage of the text encoder for downstream visual recognition tasks undiscovered. In this paper, we revise the role of the linear classifier and replace the classifier with the different knowledge from pre-trained model. We utilize the well-pretrained language model to generate good semantic target for efficient transferring learning. The empirical study shows that our method improves both the performance and the training speed of video classification, with a negligible change in the model. Our simple yet effective tuning paradigm achieves state-of-the-art performance and efficient training on various video recognition scenarios, i.e., zero-shot, few-shot, general recognition. In particular, our paradigm achieves the state-of-the-art accuracy of 87.8% on Kinetics-400, and also surpasses previous methods by 20~50% absolute top-1 accuracy under zero-shot, few-shot settings on five popular video datasets. Code and models can be found at https://github.com/whwu95/Text4Vis . | ['Wanli Ouyang', 'Zhun Sun', 'Wenhao Wu'] | 2022-07-04 | null | null | null | null | ['zero-shot-action-recognition', 'action-classification'] | ['computer-vision', 'computer-vision'] | [ 8.93017054e-02 -3.95498693e-01 -3.88181686e-01 -3.67867291e-01
-7.30855107e-01 -4.77525860e-01 5.67711532e-01 -5.19683242e-01
-6.31376565e-01 3.23479831e-01 1.11743324e-01 -3.77671659e-01
4.12263900e-01 -5.39778173e-01 -1.01710761e+00 -7.53111243e-01
3.31311464e-01 5.57149984e-02 5.24312675e-01 -3.29600088e-02
9.99860764e-02 1.01529650e-01 -1.62974846e+00 5.85456431e-01
4.68288958e-01 1.14235854e+00 4.08925116e-01 8.23283553e-01
8.97612888e-03 1.08382785e+00 -3.02597344e-01 -5.11273563e-01
3.13460082e-01 -3.85167539e-01 -7.69420624e-01 8.01685452e-02
5.95626891e-01 -6.68839037e-01 -7.18748629e-01 9.02363002e-01
4.25579011e-01 4.41504735e-03 6.50131464e-01 -1.28235221e+00
-1.06593144e+00 4.37643886e-01 -6.25555456e-01 4.62361902e-01
-7.54145682e-02 3.55968565e-01 7.07707942e-01 -1.18895054e+00
4.91082728e-01 8.91174316e-01 6.77178323e-01 8.80771101e-01
-8.45178306e-01 -8.38657379e-01 2.72369891e-01 6.84658468e-01
-1.51401484e+00 -7.79430747e-01 3.22736561e-01 -6.24361813e-01
1.18975294e+00 -1.90821454e-01 4.78862673e-01 1.34823930e+00
5.78672625e-02 7.73445427e-01 7.98609257e-01 -5.23193121e-01
1.12407506e-02 7.32507110e-02 3.78780097e-01 9.07761157e-01
1.58256605e-01 6.04906902e-02 -5.84274590e-01 2.61431545e-01
6.17141545e-01 2.19042227e-01 -4.11247879e-01 -1.83134034e-01
-1.05485320e+00 7.22310841e-01 3.48648697e-01 6.82966933e-02
-1.92699075e-01 3.92811716e-01 5.76962709e-01 3.83402735e-01
3.86406392e-01 -1.72036245e-01 -6.16921842e-01 -2.99180984e-01
-8.72645557e-01 -3.70234191e-01 7.15968966e-01 1.28419960e+00
6.86543226e-01 2.03991398e-01 -4.35620517e-01 8.95061433e-01
1.26526982e-01 5.79289734e-01 5.68953037e-01 -8.82703424e-01
4.48286742e-01 1.73331574e-01 -2.71866262e-01 -4.18115348e-01
4.84135114e-02 -4.77492988e-01 -8.12034965e-01 -2.00803712e-01
3.81560504e-01 -2.09042251e-01 -1.45104635e+00 1.64503658e+00
7.02277943e-03 6.73931360e-01 2.67066211e-01 9.08243895e-01
9.91714180e-01 8.81924331e-01 2.36649811e-01 -1.03377551e-01
1.37536514e+00 -1.57572556e+00 -2.25228891e-01 -3.43777388e-01
6.69915378e-01 -6.36268139e-01 1.10427451e+00 1.54304251e-01
-7.04801679e-01 -7.63240576e-01 -8.25113893e-01 -1.60011321e-01
-2.68907309e-01 4.72146988e-01 4.65812743e-01 3.59135181e-01
-1.18896759e+00 2.91209728e-01 -8.32836092e-01 -6.29668355e-01
7.81727552e-01 1.58798307e-01 -3.47560674e-01 -5.98127365e-01
-9.39901054e-01 8.10685217e-01 2.95797199e-01 -1.13103934e-01
-1.44812012e+00 -8.46238315e-01 -6.73450589e-01 -3.00264303e-02
5.41690171e-01 -7.76409626e-01 1.42853630e+00 -1.21520770e+00
-1.53373694e+00 9.43388462e-01 -2.17143789e-01 -4.47481751e-01
4.10277784e-01 -2.97783226e-01 -1.86844900e-01 2.22240657e-01
-1.65866949e-02 9.29364085e-01 1.09993398e+00 -1.03434515e+00
-6.76534951e-01 -1.06889375e-01 1.90607637e-01 1.08752243e-01
-6.67070925e-01 -5.82047785e-03 -1.18889046e+00 -5.04209697e-01
-4.03910190e-01 -9.72370744e-01 6.36031553e-02 1.17711008e-01
2.73241028e-02 -2.05835655e-01 8.62189293e-01 -6.88146412e-01
9.73052800e-01 -2.26045752e+00 -4.90148067e-02 -3.42298687e-01
9.12757069e-02 6.96631014e-01 -4.87595409e-01 2.35513628e-01
2.21983530e-02 -1.15220830e-01 -3.97564583e-02 -2.86987036e-01
-2.90301025e-01 2.16951966e-01 -2.75015056e-01 3.22794467e-01
2.65624136e-01 1.09684277e+00 -7.30232775e-01 -3.21717322e-01
1.70078710e-01 5.38872659e-01 -6.43530130e-01 4.25630033e-01
6.01232275e-02 1.42367855e-01 -2.92298079e-01 6.37557268e-01
3.34695756e-01 -6.04909062e-01 -9.23054591e-02 -3.92242014e-01
1.00218885e-01 -1.67921886e-01 -5.94882965e-01 1.83643913e+00
-3.75648737e-01 7.84427762e-01 -1.95099488e-01 -1.28333867e+00
6.34156823e-01 1.71101898e-01 2.49900609e-01 -5.70097625e-01
1.19486243e-01 -1.16520129e-01 1.06301993e-01 -6.49410784e-01
5.98658547e-02 7.16842785e-02 2.40931287e-01 1.40420064e-01
5.27282178e-01 4.14953202e-01 1.65938258e-01 1.34752050e-01
1.11501110e+00 2.18099430e-01 2.14386910e-01 -7.29448423e-02
4.32389498e-01 1.23402439e-01 4.55183923e-01 7.60216355e-01
-3.22561085e-01 5.95101058e-01 8.36358964e-02 -2.79828668e-01
-9.83118176e-01 -8.28032494e-01 2.50995420e-02 1.56205535e+00
4.42218259e-02 -4.53430623e-01 -8.84743929e-01 -5.59343100e-01
-8.22682977e-02 5.48863888e-01 -5.38672149e-01 -5.27101219e-01
-3.20667267e-01 -6.46117270e-01 7.19325542e-01 7.78720081e-01
7.93025374e-01 -8.11541975e-01 -4.87155527e-01 -1.55107275e-01
-1.07240744e-01 -1.50568271e+00 -5.51125348e-01 -3.02050007e-03
-6.88203156e-01 -1.01891613e+00 -9.21058834e-01 -1.01481438e+00
7.00128496e-01 7.44108856e-01 9.85859394e-01 1.03087783e-01
-3.35477650e-01 7.56835639e-01 -6.38853431e-01 -2.91346699e-01
-1.63256243e-01 3.67613845e-02 -2.91921780e-03 4.38087434e-02
6.79076374e-01 -8.24298114e-02 -5.83915055e-01 2.81321287e-01
-6.61870062e-01 3.25994670e-01 6.45755112e-01 1.02038658e+00
4.46595371e-01 -3.25817317e-01 5.28815985e-01 -6.06284380e-01
2.13657618e-01 -5.98649859e-01 -3.95988673e-01 4.08632100e-01
-5.56801915e-01 -6.86800256e-02 6.77175760e-01 -6.34414136e-01
-1.01844740e+00 1.53266326e-01 -5.69040067e-02 -1.09069133e+00
-1.84283495e-01 3.88041198e-01 8.04419518e-02 -1.30765691e-01
5.70567012e-01 4.75264996e-01 -4.06113565e-02 -3.28614205e-01
5.72869003e-01 9.11334991e-01 4.34971541e-01 -4.59161997e-01
6.18364990e-01 4.82154191e-01 -4.73802805e-01 -9.55300570e-01
-1.03309989e+00 -5.44690132e-01 -5.70667326e-01 -2.80441910e-01
9.00327325e-01 -1.41006517e+00 -3.52741838e-01 8.11246395e-01
-1.10132051e+00 -7.23090172e-01 -5.73341325e-02 4.24090505e-01
-4.15296406e-01 1.87433720e-01 -8.84216905e-01 -3.84830534e-01
-4.22645569e-01 -1.28019297e+00 8.81180823e-01 2.00870857e-01
2.63570130e-01 -8.22845995e-01 -2.83851236e-01 6.37730062e-01
5.21191180e-01 -5.14225364e-01 5.91573238e-01 -5.32586396e-01
-6.26103044e-01 1.18595675e-01 -6.00302756e-01 6.66516960e-01
-8.73482004e-02 -7.15001523e-02 -1.28120005e+00 -4.49351937e-01
-2.37864599e-01 -7.09029317e-01 1.40568244e+00 4.22178537e-01
1.40398598e+00 -1.03464790e-01 -2.95325041e-01 8.97557378e-01
1.47433889e+00 2.55715996e-01 6.79533660e-01 1.45116538e-01
9.26403046e-01 3.01434964e-01 4.95539576e-01 3.04419130e-01
5.41464508e-01 5.78865588e-01 9.86765921e-02 6.68926686e-02
-4.69379395e-01 -2.61156082e-01 6.78737223e-01 9.14446533e-01
-2.18109235e-01 -2.39876688e-01 -1.08191633e+00 4.91963536e-01
-1.91748953e+00 -9.71084893e-01 2.03174263e-01 1.97780395e+00
8.25737596e-01 8.58931467e-02 -4.20891568e-02 -3.92695338e-01
7.61128902e-01 4.62266691e-02 -7.15985000e-01 -2.93353237e-02
2.28692994e-01 1.53383147e-03 7.04436719e-01 3.37796420e-01
-1.21451294e+00 1.42298603e+00 5.87824726e+00 9.35209751e-01
-1.39042473e+00 4.62020874e-01 7.38565743e-01 -2.20936418e-01
2.65100956e-01 -1.11191511e-01 -1.10514498e+00 5.45693398e-01
1.21716833e+00 -1.71656758e-01 4.37768817e-01 9.94717360e-01
5.85579276e-02 1.13655016e-01 -1.28353202e+00 1.32512748e+00
5.43515682e-01 -1.48660970e+00 3.29967827e-01 -1.81455687e-01
7.31420398e-01 6.16215944e-01 -1.18569203e-01 6.93979263e-01
4.73106764e-02 -8.86744142e-01 6.73064470e-01 5.84251165e-01
1.16800737e+00 -3.79962802e-01 6.73644960e-01 2.69078702e-01
-1.14968526e+00 -2.28412092e-01 -4.49896693e-01 -1.03993826e-01
-1.05378151e-01 2.28287891e-01 -4.40759599e-01 2.31221646e-01
9.35851753e-01 9.75408971e-01 -5.16735554e-01 9.24778461e-01
-1.24619402e-01 9.74996448e-01 -5.22378571e-02 1.59395725e-01
2.94783056e-01 1.02779325e-02 1.40579529e-02 1.47463858e+00
2.50582457e-01 3.46398592e-01 3.69856030e-01 4.69224870e-01
-3.17311853e-01 -1.79639041e-01 -5.91643691e-01 -8.80437791e-02
4.41881984e-01 1.05270493e+00 -5.39016306e-01 -7.24622667e-01
-8.57461274e-01 1.13830304e+00 4.49642301e-01 6.27274990e-01
-1.14027929e+00 -1.85025141e-01 6.88128054e-01 -1.28445506e-01
7.00321913e-01 -2.70869017e-01 1.01505890e-01 -1.46275949e+00
3.50167751e-02 -8.61683071e-01 3.22266966e-01 -7.54484057e-01
-1.24430239e+00 5.34103751e-01 8.84065181e-02 -1.11118531e+00
-2.32848808e-01 -9.75025535e-01 -5.68751693e-01 4.43388671e-01
-1.67278552e+00 -1.29860306e+00 -6.61953330e-01 8.46639454e-01
1.02875471e+00 -5.58104932e-01 6.46957397e-01 4.98990774e-01
-7.59552598e-01 7.88664758e-01 1.49934396e-01 6.14352643e-01
8.83983433e-01 -6.46062374e-01 2.72841543e-01 1.01019180e+00
1.79633692e-01 3.65477651e-01 2.41916627e-01 -3.61057848e-01
-1.68228912e+00 -1.39442205e+00 4.21398699e-01 -3.95880580e-01
7.95096099e-01 -3.56265843e-01 -8.74209225e-01 8.01464379e-01
2.81333059e-01 3.99860650e-01 6.61026537e-01 -9.74380448e-02
-7.22562611e-01 -3.60860258e-01 -6.16696239e-01 5.82429886e-01
1.35083997e+00 -6.95067227e-01 -5.25319636e-01 2.79387295e-01
8.36532414e-01 -2.42248893e-01 -6.04520977e-01 3.13428521e-01
6.08119309e-01 -7.16045201e-01 9.92776573e-01 -7.89581180e-01
6.43273771e-01 -1.15980253e-01 -2.33195767e-01 -1.06502759e+00
-5.84227979e-01 -1.54068679e-01 -3.29586953e-01 1.11354244e+00
4.05561447e-01 -4.81713831e-01 7.45515645e-01 5.05444407e-01
-2.86286384e-01 -8.53304327e-01 -6.24808013e-01 -8.67048740e-01
1.62819903e-02 -5.96095085e-01 -4.81594913e-02 8.90148580e-01
-3.46052706e-01 6.24488771e-01 -4.61273402e-01 2.70219450e-03
6.17884994e-01 5.56154959e-02 7.13987112e-01 -6.97697043e-01
-4.76830453e-01 -3.50078672e-01 -4.61572468e-01 -1.34784043e+00
3.32920194e-01 -9.45654571e-01 1.56004988e-02 -1.52710617e+00
6.92558229e-01 -1.83979809e-01 -5.14683902e-01 8.97855222e-01
-1.37971655e-01 4.84498292e-01 4.16927636e-01 4.16463852e-01
-8.89628649e-01 5.55714130e-01 1.00124145e+00 -3.02936912e-01
8.61851200e-02 -2.51819700e-01 -7.22746551e-01 6.87796295e-01
8.51947904e-01 -3.06933194e-01 -7.57185161e-01 -9.13219690e-01
-4.29107994e-01 -2.81915486e-01 4.68001753e-01 -1.03719831e+00
4.62090313e-01 -1.21777259e-01 3.72953027e-01 -1.27554342e-01
4.85412151e-01 -6.60167575e-01 -2.23885044e-01 4.80893493e-01
-3.47487986e-01 -1.63749337e-01 3.80647749e-01 5.76734960e-01
-1.22273102e-01 -2.13061765e-01 8.93924594e-01 -9.57365036e-02
-1.46991515e+00 5.58735073e-01 -2.67215788e-01 1.59560248e-01
1.19168317e+00 -1.78296819e-01 -7.63265610e-01 -3.66048127e-01
-4.70677823e-01 1.64152965e-01 3.68304670e-01 7.36338198e-01
8.22926283e-01 -1.16097701e+00 -7.79129565e-01 1.52961552e-01
4.09294128e-01 -3.23876321e-01 3.06096077e-01 8.55200529e-01
-3.74141604e-01 4.73098129e-01 -3.17453027e-01 -8.07933033e-01
-1.32269084e+00 5.36745846e-01 2.67556697e-01 2.50611782e-01
-5.97906411e-01 1.03793502e+00 4.72199768e-01 9.10998955e-02
4.08429742e-01 -2.08380312e-01 -2.27143299e-02 -1.60540882e-02
7.91476786e-01 1.36357844e-01 -1.01276167e-01 -5.40046871e-01
-4.49994862e-01 6.86115324e-01 -3.20506424e-01 2.95921177e-01
1.20720577e+00 -6.27093911e-02 2.46152222e-01 4.57049489e-01
1.40686333e+00 -6.26258910e-01 -1.57554078e+00 -2.85888642e-01
-2.84329474e-01 -3.47386479e-01 2.92590141e-01 -6.64618373e-01
-1.30840230e+00 1.03774846e+00 6.52061343e-01 -3.76185894e-01
1.14005351e+00 1.49773329e-01 6.68535531e-01 6.96423531e-01
4.86200511e-01 -1.06883490e+00 1.89818814e-01 7.94512033e-01
6.67500317e-01 -1.60961175e+00 -1.98349208e-01 -2.81248778e-01
-8.95885289e-01 9.08663332e-01 9.79853511e-01 -5.55374809e-02
7.63572514e-01 2.37376481e-01 1.50095388e-01 1.19705893e-01
-1.15106642e+00 -4.02396888e-01 2.63728619e-01 5.79559684e-01
4.22972828e-01 -1.04965948e-01 7.32822791e-02 5.25349855e-01
2.50150383e-01 4.06534553e-01 2.68350661e-01 8.52347970e-01
-5.27328908e-01 -6.66118026e-01 2.26951912e-02 5.69727957e-01
-2.90292382e-01 -4.31579798e-01 -9.57314894e-02 6.32744372e-01
8.58368650e-02 8.90320957e-01 1.35201156e-01 -5.92069745e-01
1.17343940e-01 2.14323685e-01 5.21266758e-01 -6.44238234e-01
-9.16054398e-02 -1.14055939e-01 9.55103785e-02 -6.62245929e-01
-4.99417603e-01 -3.65652412e-01 -1.06852341e+00 -2.50575393e-01
-2.23830581e-01 -1.22914754e-01 4.77751046e-01 9.79175448e-01
6.47271633e-01 4.45440620e-01 3.39198738e-01 -8.05860341e-01
-6.81508601e-01 -9.15171683e-01 -1.08861081e-01 3.31223011e-01
1.89478695e-01 -6.70129478e-01 -2.39539504e-01 5.97884595e-01] | [9.751955032348633, 1.3264479637145996] |
0868da6e-927a-4d9b-bdae-dd244f420b1e | learning-canonical-representations-for-scene | 1912.07414 | null | https://arxiv.org/abs/1912.07414v5 | https://arxiv.org/pdf/1912.07414v5.pdf | Learning Canonical Representations for Scene Graph to Image Generation | Generating realistic images of complex visual scenes becomes challenging when one wishes to control the structure of the generated images. Previous approaches showed that scenes with few entities can be controlled using scene graphs, but this approach struggles as the complexity of the graph (the number of objects and edges) increases. In this work, we show that one limitation of current methods is their inability to capture semantic equivalence in graphs. We present a novel model that addresses these issues by learning canonical graph representations from the data, resulting in improved image generation for complex visual scenes. Our model demonstrates improved empirical performance on large scene graphs, robustness to noise in the input scene graph, and generalization on semantically equivalent graphs. Finally, we show improved performance of the model on three different benchmarks: Visual Genome, COCO, and CLEVR. | ['Roei Herzig', 'Amir Globerson', 'Gal Chechik', 'Amir Bar', 'Trevor Darrell', 'Huijuan Xu'] | 2019-12-16 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/5328_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123710205.pdf | eccv-2020-8 | ['layout-to-image-generation', 'scene-generation'] | ['computer-vision', 'computer-vision'] | [ 3.17434788e-01 2.77185827e-01 1.37211353e-01 -1.49737746e-01
-1.73092753e-01 -8.79516959e-01 7.62395263e-01 2.36485481e-01
-1.16556369e-01 5.84335804e-01 2.10073829e-01 -1.78487629e-01
1.07010081e-01 -1.09515798e+00 -8.41853917e-01 -1.21707879e-01
1.55595150e-02 4.74357903e-01 4.99984831e-01 -3.11794788e-01
8.14141333e-02 5.44394255e-01 -1.50423956e+00 3.51961881e-01
7.16944993e-01 3.83599460e-01 2.16757014e-01 9.62620735e-01
-2.17099667e-01 1.03766382e+00 -6.73516572e-01 -4.40828711e-01
2.74197102e-01 -6.15505815e-01 -7.75047898e-01 5.80632389e-01
7.66220868e-01 7.44559318e-02 -4.30236012e-01 1.12681615e+00
2.04289705e-01 1.48062482e-01 7.51105189e-01 -1.76479495e+00
-8.52479160e-01 5.16593814e-01 -4.71408665e-01 -2.27751140e-03
5.09200513e-01 2.38672540e-01 9.61190879e-01 -5.91398537e-01
1.09060669e+00 1.59135628e+00 2.78573573e-01 5.22704244e-01
-1.62968898e+00 -3.50610316e-01 3.30166370e-01 6.68722242e-02
-1.31971133e+00 -3.28427166e-01 5.98465264e-01 -5.27634263e-01
1.03960681e+00 2.03626126e-01 6.73514366e-01 8.73572350e-01
-6.49717301e-02 4.70956624e-01 1.00876999e+00 -5.27031243e-01
3.07188362e-01 9.39900130e-02 -8.54499713e-02 8.79065692e-01
5.51397502e-01 -1.01320863e-01 -2.55205870e-01 -6.54897019e-02
9.09930408e-01 -3.98251146e-01 -3.25393915e-01 -8.45517516e-01
-1.17256796e+00 8.90037358e-01 7.72944570e-01 1.62269026e-01
-2.25340016e-02 3.92264307e-01 1.78308517e-01 1.52378768e-01
5.29798642e-02 9.18829560e-01 2.17726864e-02 3.40342611e-01
-5.54306626e-01 4.08306301e-01 9.04395878e-01 1.50337434e+00
6.60626531e-01 1.76264852e-01 -3.61952960e-01 5.42145014e-01
3.95780317e-02 5.07367253e-01 -3.81177850e-02 -1.04076958e+00
4.70151037e-01 8.44522357e-01 -4.56124917e-02 -1.35463107e+00
-3.09850067e-01 -2.07281202e-01 -8.42757046e-01 2.15422839e-01
4.38196957e-01 7.47327805e-02 -1.37263536e+00 1.87826085e+00
1.69873863e-01 5.96784055e-02 1.11110844e-01 5.83755672e-01
8.71121764e-01 7.16010869e-01 3.11864942e-01 7.87679404e-02
1.18547320e+00 -1.01598060e+00 -5.92799306e-01 -6.08634830e-01
4.57035720e-01 -6.08518481e-01 1.17959249e+00 -1.20737962e-02
-1.09700859e+00 -4.89450455e-01 -9.79403317e-01 -1.43955603e-01
-6.01394415e-01 -1.01703383e-01 8.58342767e-01 4.68114674e-01
-1.29856849e+00 2.05670208e-01 -3.79954278e-01 -6.58554435e-01
4.74975586e-01 3.36182237e-01 -5.45874834e-01 -3.22732359e-01
-8.17850471e-01 6.38457417e-01 6.64553344e-01 -4.00921255e-01
-8.13107491e-01 -5.97666860e-01 -1.33925104e+00 2.49632984e-01
4.01624382e-01 -1.10156190e+00 9.03000236e-01 -1.04329395e+00
-1.00533557e+00 9.50726330e-01 -2.94320881e-02 -3.76787037e-01
4.96410459e-01 2.42194846e-01 -1.56059816e-01 2.69806534e-01
1.38550252e-01 9.89634335e-01 7.25343049e-01 -1.66807652e+00
-2.85648346e-01 -1.75709918e-01 5.76347232e-01 3.12791049e-01
1.83005519e-02 -8.81467089e-02 -7.37136483e-01 -5.07090926e-01
-2.37841889e-01 -1.06005824e+00 -6.88861251e-01 -4.41707931e-02
-5.59130490e-01 7.56619126e-02 7.63958097e-01 -1.88557893e-01
8.35316956e-01 -2.26041484e+00 2.85750210e-01 2.09947065e-01
3.67648542e-01 1.00807130e-01 -5.07190406e-01 5.91641963e-01
-2.11073786e-01 5.13499081e-01 -6.69459179e-02 -1.85537979e-01
-1.51061445e-01 3.31691891e-01 -1.95014715e-01 9.92365628e-02
3.72042328e-01 1.14191818e+00 -1.03488708e+00 -7.81240642e-01
3.01156372e-01 3.44736308e-01 -7.30571389e-01 3.17143530e-01
-4.16876495e-01 2.24898726e-01 -3.71304691e-01 3.41474354e-01
5.59414089e-01 -7.13116348e-01 4.24490929e-01 -2.33511254e-01
3.91862899e-01 -1.22785270e-01 -1.31697297e+00 1.43743920e+00
-3.18943053e-01 5.57201624e-01 -2.42775157e-01 -6.24392211e-01
6.56858325e-01 -4.65307459e-02 5.73352873e-02 -6.41373038e-01
-8.79871249e-02 -3.30032408e-01 -9.12718326e-02 -2.68211514e-01
4.62931156e-01 8.55182260e-02 -1.26035362e-01 1.90383092e-01
4.21688966e-02 -8.71929526e-01 8.07668865e-01 8.12980831e-01
1.01061404e+00 -2.32187770e-02 4.60323334e-01 -1.40815109e-01
2.42606491e-01 3.27174366e-01 1.02611303e-01 9.36435223e-01
1.72409609e-01 7.19396234e-01 9.13583934e-01 -2.56925583e-01
-1.21152663e+00 -1.21885705e+00 3.36627662e-01 7.01707959e-01
3.32896918e-01 -8.28360736e-01 -7.51021564e-01 -6.90552056e-01
-1.64042249e-01 8.39347064e-01 -6.54489815e-01 -1.76877737e-01
-4.72513616e-01 -5.86810589e-01 2.83927113e-01 5.30033946e-01
2.89786786e-01 -1.06372738e+00 -4.39008802e-01 -1.21097192e-01
-1.27295807e-01 -1.72969210e+00 -4.48330462e-01 -3.10676873e-01
-6.36351407e-01 -1.26427197e+00 -2.59612262e-01 -8.50681424e-01
1.18397439e+00 4.55605209e-01 1.60206532e+00 1.90859333e-01
-5.31666279e-01 6.66821480e-01 -2.29489878e-01 -3.48883659e-01
-6.30368114e-01 -7.18933493e-02 -5.04216552e-01 -1.16957404e-01
-2.32697204e-01 -4.44107622e-01 -3.10530096e-01 1.73165232e-01
-1.13900638e+00 4.96561915e-01 3.22480500e-01 7.83715129e-01
6.17581069e-01 8.77796337e-02 1.87112853e-01 -1.19485271e+00
6.56678617e-01 -2.81758875e-01 -7.99025834e-01 4.67073143e-01
-3.12054247e-01 2.55358309e-01 7.26064682e-01 -3.18301976e-01
-9.56369579e-01 3.59382957e-01 4.38764483e-01 -3.45539331e-01
-1.00563981e-01 2.61872649e-01 -1.30000442e-01 -6.48614466e-02
8.81752610e-01 6.77905828e-02 -1.98023006e-01 1.20976195e-01
8.49464059e-01 9.61607043e-03 4.84835714e-01 -4.14960295e-01
9.71388459e-01 5.22849441e-01 4.57937717e-01 -7.67460644e-01
-7.30523765e-01 -2.21839100e-01 -6.32437706e-01 -1.14227846e-01
8.56210709e-01 -9.81505215e-01 -3.83154809e-01 1.23720467e-01
-1.15638375e+00 -4.43516195e-01 -3.86220306e-01 1.57168522e-01
-7.27151573e-01 4.10688221e-01 -2.97662854e-01 -3.89461607e-01
9.28895199e-04 -1.09312284e+00 1.22303939e+00 1.95697263e-01
-2.69505560e-01 -1.04499960e+00 -7.92712420e-02 1.24879345e-01
3.05143267e-01 7.05611825e-01 1.21925104e+00 -1.22720361e-01
-1.04653430e+00 -2.06681013e-01 -5.10363817e-01 3.61865051e-02
1.72124431e-01 2.48639897e-01 -6.05604529e-01 -2.96951503e-01
-5.67508399e-01 -4.18543041e-01 6.55749619e-01 1.65079623e-01
1.13255417e+00 -2.58285046e-01 -4.40323681e-01 6.54351830e-01
1.76679540e+00 -4.33245450e-02 8.21102917e-01 -3.04987375e-02
1.05977607e+00 6.70903683e-01 4.20007378e-01 8.27564597e-02
4.05030012e-01 6.28488839e-01 5.29028416e-01 -3.28081906e-01
-5.50076306e-01 -6.32337987e-01 7.75026008e-02 3.47348064e-01
3.59964482e-02 -7.34874606e-01 -8.99613500e-01 7.66187191e-01
-1.91656148e+00 -9.84790027e-01 -3.90975296e-01 1.90672421e+00
4.01883215e-01 -2.31745280e-02 -3.34429070e-02 -2.67660469e-01
6.77619576e-01 2.48082548e-01 -2.83757120e-01 -2.73187488e-01
-3.86863172e-01 7.56684914e-02 4.83285666e-01 5.40985227e-01
-9.44846153e-01 1.26844251e+00 7.58431149e+00 3.89845431e-01
-7.78321505e-01 -2.46353596e-01 7.12399960e-01 -1.62254609e-02
-4.01476383e-01 2.34996125e-01 -4.95297104e-01 8.39318782e-02
5.27471304e-01 -4.99342263e-01 4.88659799e-01 9.77250934e-01
1.49414781e-02 -1.84860118e-02 -1.28531742e+00 1.05350530e+00
2.92442381e-01 -1.51976657e+00 7.16053963e-01 -4.42719497e-02
9.44946647e-01 -1.70569733e-01 -1.02658376e-01 1.08811520e-01
8.91149640e-01 -1.36062062e+00 6.19400322e-01 2.86266088e-01
7.66119897e-01 -6.34551346e-01 3.21932077e-01 5.82634024e-02
-1.22036433e+00 1.77719191e-01 -4.76296484e-01 5.05772419e-02
1.83771521e-01 2.85661310e-01 -1.08084702e+00 5.19631386e-01
5.06792247e-01 5.94236732e-01 -1.13517416e+00 9.60991263e-01
-3.46472263e-01 1.83131024e-01 -1.10977136e-01 -4.72927094e-02
1.64613292e-01 -1.46071151e-01 3.94513577e-01 1.28374898e+00
1.81438252e-01 1.77759714e-02 4.44091320e-01 1.04281545e+00
-2.10055590e-01 1.21109746e-01 -1.28834605e+00 -2.66334623e-01
1.38435602e-01 1.37519836e+00 -9.84809279e-01 -4.19589281e-01
-3.17646027e-01 9.40920055e-01 7.59071469e-01 5.70906639e-01
-8.30411255e-01 -2.14242265e-01 3.43428433e-01 3.56012791e-01
1.86078191e-01 -4.96383578e-01 -1.37912631e-01 -1.27578044e+00
-1.19691730e-01 -1.06497025e+00 6.02065742e-01 -1.24994028e+00
-1.04724193e+00 7.15770423e-01 2.38165781e-01 -8.46837401e-01
-2.80100405e-01 -6.55832112e-01 -4.30074781e-01 5.27231097e-01
-1.11847901e+00 -1.37045717e+00 -6.73696816e-01 7.52070546e-01
4.35909182e-01 1.91805050e-01 7.30467498e-01 -4.96026203e-02
-1.57923281e-01 2.38507748e-01 -3.02775562e-01 1.64595544e-01
4.07128632e-01 -1.51339650e+00 8.08568954e-01 1.02236104e+00
4.26019728e-01 3.55746478e-01 6.62169099e-01 -5.23472428e-01
-1.42947996e+00 -1.34782815e+00 7.19516397e-01 -6.45043254e-01
4.91757065e-01 -7.65048504e-01 -6.11138165e-01 9.09024417e-01
4.15555567e-01 1.56079620e-01 3.36817265e-01 -1.71992615e-01
-5.71684062e-01 3.04127127e-01 -1.01056290e+00 1.11602080e+00
1.42965674e+00 -3.89710486e-01 -1.97985798e-01 5.99550903e-01
8.62140059e-01 -4.62106228e-01 -5.32062113e-01 3.01417768e-01
1.59844279e-01 -9.10856724e-01 1.01841569e+00 -9.44498658e-01
4.98908132e-01 -5.29521763e-01 -1.08713374e-01 -1.55741894e+00
-5.03374279e-01 -4.76373136e-01 1.88217968e-01 9.97978985e-01
4.26649749e-01 -3.03239673e-01 5.95257103e-01 5.03641367e-01
2.77465165e-01 -1.82640284e-01 -3.60054880e-01 -7.74396837e-01
-2.92810202e-01 -2.04120114e-01 6.59984946e-01 9.12484527e-01
-2.65842289e-01 8.02383542e-01 -2.33839467e-01 1.76592633e-01
6.55701518e-01 2.40635067e-01 1.21879649e+00 -9.75894928e-01
-3.79463971e-01 -2.66110122e-01 -8.44220161e-01 -7.08504856e-01
1.84019804e-01 -8.96718025e-01 -1.49887517e-01 -1.93620586e+00
4.17721033e-01 -1.95480272e-01 2.19228148e-01 3.59228075e-01
-3.37851852e-01 2.87256390e-01 6.81533933e-01 -2.61535227e-01
-7.67024159e-01 2.95753717e-01 1.55641782e+00 -3.32807809e-01
-5.79886325e-02 -4.72838879e-01 -7.21967459e-01 6.51272058e-01
7.24666893e-01 -3.77422929e-01 -8.68942857e-01 -5.69615185e-01
3.88141155e-01 -9.86308604e-02 5.46919227e-01 -8.21817279e-01
5.78111038e-03 -4.22206074e-01 3.64534438e-01 -1.93657950e-01
2.56690651e-01 -6.58727825e-01 5.18620014e-01 3.28578532e-01
-3.12216282e-01 3.31503332e-01 2.93033421e-01 6.87708020e-01
-2.45517448e-01 1.41114590e-03 9.23170865e-01 -4.45816934e-01
-7.61749864e-01 3.13908905e-01 -1.02801658e-01 2.81972587e-01
1.31340671e+00 -1.74802825e-01 -5.84797084e-01 -7.10820675e-01
-7.63752341e-01 3.15867513e-01 9.10962701e-01 6.82446301e-01
6.83367193e-01 -1.44040310e+00 -7.54474521e-01 -1.03287371e-02
4.58233893e-01 5.13459854e-02 6.56265467e-02 2.20469669e-01
-8.39387536e-01 5.15251383e-02 -2.52420604e-01 -6.81479871e-01
-1.60711479e+00 1.06155610e+00 4.62990582e-01 -4.29291308e-01
-4.67517287e-01 5.52597165e-01 8.30857396e-01 -2.39140034e-01
-3.53877917e-02 -2.96257794e-01 -1.06047250e-01 -2.83529878e-01
3.30445051e-01 9.29182395e-02 -3.07322890e-01 -8.10742199e-01
-1.51742727e-01 5.89823544e-01 -5.43099158e-02 -3.36993635e-02
1.05201113e+00 -5.40765300e-02 -2.52706707e-01 1.22406781e-01
9.99481976e-01 8.24380293e-02 -1.06689870e+00 4.15494712e-03
-1.19792439e-01 -7.07269073e-01 -2.93273151e-01 -5.27164578e-01
-1.05947757e+00 6.14137173e-01 2.80131370e-01 4.27533805e-01
1.11886585e+00 2.49572903e-01 2.33327582e-01 4.66355145e-01
3.36391449e-01 -6.67255342e-01 4.18349117e-01 3.82973611e-01
1.09012580e+00 -1.17041171e+00 1.00681879e-01 -1.03864014e+00
-8.76608968e-01 9.21002388e-01 8.15942287e-01 -1.38557076e-01
4.52801049e-01 2.01075599e-01 -1.21654198e-01 -5.69148898e-01
-7.99383581e-01 -4.47415352e-01 3.51032168e-01 9.90347505e-01
2.36800343e-01 1.53760821e-01 -2.09974758e-02 -1.58488363e-01
-3.02389711e-01 -2.86678463e-01 8.94110084e-01 7.54612863e-01
-1.49247304e-01 -1.00394285e+00 -1.83383837e-01 2.92827427e-01
-2.83026814e-01 -2.28498727e-01 -9.55277324e-01 1.03654313e+00
-3.59994173e-02 1.00293159e+00 1.02587961e-01 3.80785465e-02
5.99272430e-01 -3.28385592e-01 7.63091624e-01 -9.80763614e-01
-3.02344888e-01 -1.96479425e-01 3.01302165e-01 -7.67236769e-01
-4.64626402e-01 -3.00049305e-01 -1.09697700e+00 -1.23689607e-01
-1.22574985e-01 -9.39457677e-03 3.91157746e-01 4.09337759e-01
5.89929760e-01 5.71131527e-01 4.18966830e-01 -5.26098132e-01
4.68160119e-03 -5.02926469e-01 -4.66190487e-01 1.05574727e+00
9.94313508e-02 -3.85449469e-01 -1.76679224e-01 4.91201967e-01] | [10.51976203918457, 1.4424207210540771] |
be1e1ae7-feaf-4ef2-a565-7fd9da2555ba | deep-attention-based-supernovae | 2201.08482 | null | https://arxiv.org/abs/2201.08482v3 | https://arxiv.org/pdf/2201.08482v3.pdf | Deep Attention-Based Supernovae Classification of Multi-Band Light-Curves | In astronomical surveys, such as the Zwicky Transient Facility, supernovae (SNe) are relatively uncommon objects compared to other classes of variable events. Along with this scarcity, the processing of multi-band light-curves is a challenging task due to the highly irregular cadence, long time gaps, missing-values, few observations, etc. These issues are particularly detrimental to the analysis of transient events: SN-like light-curves. We offer three main contributions: 1) Based on temporal modulation and attention mechanisms, we propose a Deep attention model (TimeModAttn) to classify multi-band light-curves of different SN types, avoiding photometric or hand-crafted feature computations, missing-value assumptions, and explicit imputation/interpolation methods. 2) We propose a model for the synthetic generation of SN multi-band light-curves based on the Supernova Parametric Model, allowing us to increase the number of samples and the diversity of cadence. Thus, the TimeModAttn model is first pre-trained using synthetic light-curves. Then, a fine-tuning process is performed. The TimeModAttn model outperformed other Deep Learning models, based on Recurrent Neural Networks, in two scenarios: late-classification and early-classification. Also, the TimeModAttn model outperformed a Balanced Random Forest (BRF) classifier (trained with real data), increasing the balanced-$F_1$score from $\approx.525$ to $\approx.596$. When training the BRF with synthetic data, this model achieved similar performance to the TimeModAttn model proposed while still maintaining extra advantages. 3) We conducted interpretability experiments. High attention scores were obtained for observations earlier than and close to the SN brightness peaks. This also correlated with an early highly variability of the learned temporal modulation. | ['Francisco Förster', 'Pablo A. Estévez', 'Óscar Pimentel'] | 2022-01-20 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [-1.26505420e-02 -5.16116858e-01 1.94972128e-01 -2.25065887e-01
-6.91835940e-01 -2.66519547e-01 8.78757238e-01 -2.70209253e-01
-5.36298990e-01 8.07406545e-01 -4.13114965e-01 -4.42000091e-01
-4.35549527e-01 -7.99066722e-01 -7.83222497e-01 -1.05966616e+00
1.52652457e-01 4.44524378e-01 2.40420908e-01 -1.12838484e-01
1.33460924e-01 6.08592510e-01 -2.05685019e+00 2.03086026e-02
7.09678352e-01 1.09437966e+00 5.05107522e-01 5.43531716e-01
-2.26165310e-01 6.31879389e-01 -6.44511700e-01 -1.71730280e-01
3.07002753e-01 -4.55170006e-01 -2.88350582e-01 -2.27825493e-01
1.35267496e-01 8.10638219e-02 -2.19679296e-01 7.91203916e-01
4.86250430e-01 4.74672347e-01 6.05775237e-01 -1.33412170e+00
-2.68968791e-01 1.35779098e-01 -5.63360929e-01 4.82185245e-01
-1.40268773e-01 4.14269835e-01 8.93461168e-01 -7.07565546e-01
5.75373828e-01 7.77219117e-01 7.93034613e-01 3.10845703e-01
-1.28595829e+00 -6.94915473e-01 -2.94217527e-01 7.68643796e-01
-1.06544006e+00 -9.16614458e-02 9.49032485e-01 -6.40072346e-01
1.05079544e+00 2.97613025e-01 8.33520651e-01 1.31445146e+00
1.31808082e-02 4.79557574e-01 1.26547527e+00 -5.36642253e-01
1.85763046e-01 -3.79518494e-02 2.16202676e-01 1.56375155e-01
3.27150039e-02 6.93200171e-01 -3.70207489e-01 9.57837403e-02
5.96458673e-01 -4.65124920e-02 -3.29738438e-01 -4.33173701e-02
-1.03556752e+00 8.18554997e-01 2.39610299e-01 3.93320531e-01
-4.39693391e-01 2.13508345e-02 4.28483665e-01 4.27415341e-01
4.59303319e-01 4.66385931e-01 -6.31163776e-01 -1.93846300e-01
-1.03765380e+00 4.60958898e-01 2.81058371e-01 4.70825225e-01
8.11534822e-01 4.87085253e-01 -7.91610256e-02 1.03929436e+00
3.59429568e-02 6.01823986e-01 6.43640161e-01 -6.05836153e-01
1.37076765e-01 2.31228098e-01 7.37572610e-02 -5.57698011e-01
-7.99736619e-01 -8.93760324e-01 -8.24056864e-01 5.43901443e-01
7.31803536e-01 1.04555637e-01 -8.90483856e-01 1.84335375e+00
5.67152239e-02 1.59644559e-01 -5.18716797e-02 8.99713576e-01
7.78799236e-01 8.54865909e-01 -9.11494810e-03 -5.24945378e-01
1.26006329e+00 -8.48570347e-01 -6.28174782e-01 -2.52889335e-01
1.91974774e-01 -7.98092544e-01 1.08446777e+00 4.20233279e-01
-8.48731637e-01 -8.02395284e-01 -9.35138226e-01 4.02755141e-02
-2.66988784e-01 2.77354494e-02 7.40642548e-01 5.66508770e-01
-7.69830883e-01 6.82421565e-01 -7.34728456e-01 -1.19814709e-01
8.19417685e-02 -7.90088326e-02 7.92502165e-02 4.15976644e-01
-9.56091404e-01 6.46913052e-01 2.09200963e-01 1.84528992e-01
-8.35341930e-01 -9.07139301e-01 -4.81432557e-01 2.34259702e-02
1.45222858e-01 -5.51361501e-01 1.24625397e+00 -1.10162961e+00
-1.41075075e+00 8.55891883e-01 -2.08288044e-01 -6.18198514e-01
4.79948252e-01 3.12308911e-02 -7.91674316e-01 -1.18657179e-01
-5.57719432e-02 1.87602997e-01 9.95454907e-01 -1.13765609e+00
-4.57264364e-01 -4.50983010e-02 -3.25322419e-01 -3.16093266e-01
3.39846134e-01 2.65847355e-01 -9.74861458e-02 -8.54886949e-01
2.11270928e-01 -8.90217781e-01 -3.14477906e-02 -2.30403110e-01
4.71241120e-03 -3.22443753e-01 8.26869965e-01 -5.75560689e-01
8.16579401e-01 -2.28305030e+00 -1.36327788e-01 1.19522974e-01
-8.21997505e-03 3.33048046e-01 1.70770407e-01 6.24934286e-02
-5.31295359e-01 -2.38562971e-01 -3.62204254e-01 -2.53274649e-01
-6.15379401e-03 1.93019032e-01 -4.46764916e-01 2.99673140e-01
5.88264056e-02 7.07642555e-01 -6.47079945e-01 1.44134676e-02
3.64134699e-01 3.10867637e-01 -3.05375844e-01 3.48473758e-01
-5.18616915e-01 6.11618578e-01 2.01644391e-01 5.00243306e-01
8.23177278e-01 4.50574830e-02 -3.52614462e-01 1.79304574e-02
-6.29086077e-01 2.07870275e-01 -1.06530392e+00 1.34356129e+00
-7.98483551e-01 9.28053260e-01 -1.08332686e-01 -9.33838844e-01
1.30595863e+00 3.36552620e-01 3.55738699e-01 -9.97434974e-01
6.26857504e-02 3.44686508e-01 1.55097291e-01 -5.89539289e-01
4.66410875e-01 -4.45838511e-01 2.88112372e-01 5.00524700e-01
7.00897425e-02 -3.27862710e-01 2.74928391e-01 -2.52843112e-01
8.22376370e-01 3.77242386e-01 3.74301001e-02 -2.54090905e-01
4.44760829e-01 -2.31763363e-01 8.77327740e-01 8.11081827e-01
-9.17632058e-02 7.49092817e-01 5.59423685e-01 -8.22940052e-01
-1.22924268e+00 -1.01230121e+00 -4.91657883e-01 9.87162828e-01
-1.58780336e-01 -1.15079165e-01 -2.72460401e-01 -3.31571013e-01
-3.44628960e-01 1.13073003e+00 -4.60548282e-01 -1.02209806e-01
-4.78450477e-01 -1.36886001e+00 3.23861837e-01 3.84416878e-01
3.94436628e-01 -1.53092194e+00 -7.42272913e-01 2.13208228e-01
-2.36900911e-01 -8.21890056e-01 8.78445655e-02 5.08848429e-01
-7.07974494e-01 -8.73356879e-01 -3.89640272e-01 -4.79267269e-01
1.97200641e-01 1.27115279e-01 1.19797194e+00 -2.69560535e-02
-4.15796578e-01 -1.01021469e-01 -3.68557066e-01 -7.68118858e-01
-3.33591163e-01 -2.48178154e-01 -8.73658806e-02 2.03102112e-01
5.26273549e-01 -7.07558632e-01 -3.05131197e-01 4.14279670e-01
-9.08520401e-01 4.14439738e-02 4.90932316e-01 1.22185671e+00
6.19147956e-01 1.22521736e-01 6.89090371e-01 -4.50344682e-01
4.26985584e-02 -5.22965491e-01 -1.12065089e+00 -7.45615959e-02
-6.63855076e-01 1.85037971e-01 7.50140309e-01 -4.75366175e-01
-1.46093881e+00 -2.87532985e-01 -5.24837136e-01 -4.67524618e-01
-3.34993482e-01 8.40575472e-02 -6.12113578e-03 1.30436078e-01
8.03288698e-01 2.86802620e-01 3.21885524e-03 -5.97727835e-01
-1.91505149e-01 5.36818564e-01 7.30282068e-01 -6.00878775e-01
9.07423854e-01 4.23608661e-01 5.66767529e-02 -1.00824571e+00
-8.18404436e-01 -3.62895906e-01 -3.03125262e-01 -4.77675349e-01
7.41593778e-01 -7.12185144e-01 -5.96182048e-01 7.80769467e-01
-1.01966524e+00 -5.06747186e-01 -4.14134383e-01 8.17886293e-01
-6.05511487e-01 2.18891039e-01 -4.42711145e-01 -1.11091101e+00
-2.93582320e-01 -1.08790660e+00 7.62051880e-01 3.72727245e-01
9.79720503e-02 -8.02438736e-01 2.25324243e-01 3.09809834e-01
4.68168080e-01 3.40920031e-01 1.00655091e+00 -4.34379548e-01
-5.73285341e-01 -1.67732108e-02 -2.89883882e-01 2.60720551e-01
-3.51718634e-01 2.87681222e-01 -1.36392224e+00 -9.10024270e-02
5.77731252e-01 -2.92846590e-01 9.96177077e-01 5.79813182e-01
1.46628320e+00 2.08487332e-01 1.13048814e-01 1.05128503e+00
1.36514843e+00 5.71917832e-01 7.53503382e-01 7.81577349e-01
2.33601585e-01 4.40065503e-01 4.04953271e-01 5.77838242e-01
2.05116142e-02 8.20788562e-01 5.80499470e-01 -1.33987144e-01
-1.24556072e-01 9.53464881e-02 2.60090441e-01 7.07026422e-01
-3.29488188e-01 -2.84287753e-03 -7.15241909e-01 6.76900148e-01
-1.60060084e+00 -1.38867414e+00 -8.25503290e-01 2.42426181e+00
5.31295896e-01 3.52718979e-01 4.36451808e-02 3.80377352e-01
6.46407545e-01 3.29337299e-01 -4.07993048e-01 -5.62978446e-01
-5.39288640e-01 4.29797292e-01 2.81561166e-01 2.11261958e-01
-8.44621122e-01 3.77300054e-01 5.49803448e+00 8.59873354e-01
-1.22279751e+00 2.35892355e-01 2.76628196e-01 -2.52715409e-01
-4.82493967e-01 -6.15840254e-04 -6.98876441e-01 7.90576458e-01
1.04479170e+00 -1.07871415e-03 6.69689059e-01 6.84522510e-01
2.58679390e-01 -1.33423880e-01 -6.77731931e-01 1.12456465e+00
-7.81732351e-02 -1.18663883e+00 -5.28040946e-01 -2.57073611e-01
6.52635336e-01 3.54944855e-01 -2.33587235e-01 6.35933459e-01
7.93741494e-02 -6.30869627e-01 9.43126202e-01 8.12612236e-01
6.14455044e-01 -6.40356123e-01 7.60607064e-01 4.22478318e-01
-9.01379943e-01 -1.77678972e-01 -5.14268517e-01 -5.55786341e-02
3.18277717e-01 1.09083772e+00 -6.27705872e-01 8.83261323e-01
9.76411462e-01 3.96185011e-01 -6.47441149e-01 1.26445246e+00
-3.31344306e-01 8.30815732e-01 -4.82510805e-01 3.82272527e-03
-4.91959862e-02 -5.61291575e-01 6.73174918e-01 1.00471461e+00
4.96689498e-01 -3.23827595e-01 -3.67107570e-01 1.00699413e+00
3.03801566e-01 -1.24769799e-01 -3.45537633e-01 3.58073592e-01
5.45444302e-02 1.36753011e+00 -5.31223297e-01 -1.13658398e-01
-6.49303615e-01 3.87299806e-01 3.59978005e-02 4.62896049e-01
-1.06547999e+00 -4.14119661e-01 4.59272593e-01 1.03776284e-01
2.88811386e-01 -1.61978200e-01 -4.60821539e-01 -9.81822491e-01
1.61963150e-01 -5.76898873e-01 5.18469155e-01 -1.20543420e+00
-1.25253427e+00 8.92183900e-01 -5.71174808e-02 -1.28882599e+00
-3.39599967e-01 -5.16776323e-01 -8.79971921e-01 8.34586263e-01
-1.61119735e+00 -9.03557122e-01 -5.72597682e-01 3.84407401e-01
5.30445158e-01 -3.91477317e-01 5.41688144e-01 3.80182624e-01
-6.98232830e-01 2.69693553e-01 4.79881406e-01 -2.93121219e-01
5.46220005e-01 -1.24074900e+00 2.96700239e-01 9.81391490e-01
2.49455363e-01 9.88116413e-02 9.26580012e-01 -2.83477545e-01
-8.32259893e-01 -1.03501654e+00 8.71041179e-01 -1.89018130e-01
7.25100696e-01 -3.59296292e-01 -1.20401978e+00 5.44154763e-01
1.05344437e-01 3.24090868e-02 4.04798239e-01 1.03664994e-01
-3.16837043e-01 -3.87725025e-01 -9.77095664e-01 3.72936755e-01
8.65962446e-01 -5.41737974e-01 -7.72462249e-01 3.53237808e-01
4.56188262e-01 -1.06363952e-01 -5.93804598e-01 4.98334169e-01
3.48967463e-01 -1.32116556e+00 8.43798280e-01 -2.49030426e-01
2.18648806e-01 -4.30270553e-01 5.18688895e-02 -1.29460859e+00
-3.24834943e-01 -5.64677000e-01 1.22754529e-01 1.16916287e+00
3.45801264e-01 -7.22044587e-01 5.09746671e-01 1.83796212e-01
-4.20554429e-01 -3.07429194e-01 -1.14258599e+00 -9.99578238e-01
-5.44614829e-02 -9.48740959e-01 5.80698907e-01 7.15214550e-01
-7.36234069e-01 -3.84672661e-03 -4.11766827e-01 1.25389546e-01
6.87701762e-01 5.60028493e-01 7.36958027e-01 -1.53637731e+00
-6.61925316e-01 -5.03538072e-01 -5.78816012e-02 -5.36842763e-01
1.93609610e-01 -9.28785682e-01 4.51631099e-02 -9.69320238e-01
-1.10201247e-01 -6.17907286e-01 -4.35399890e-01 3.06248575e-01
1.04180425e-02 2.40288064e-01 2.60562394e-02 1.65739462e-01
-9.15601552e-02 6.54406309e-01 8.80797505e-01 1.22736394e-01
-5.39768860e-02 3.47752333e-01 1.09847881e-01 6.44351542e-01
8.89602184e-01 -5.64304471e-01 -7.70615339e-02 -2.31587082e-01
4.32007641e-01 2.50149190e-01 6.71368122e-01 -1.23833668e+00
-3.84228528e-02 -1.56114563e-01 3.04181308e-01 -8.59331429e-01
3.87163907e-01 -5.78897953e-01 6.35731220e-01 2.56503165e-01
1.61682025e-01 -2.67058220e-02 2.07520142e-01 3.04693073e-01
-2.19485730e-01 -6.81187212e-01 1.05617702e+00 2.12501977e-02
-6.30419433e-01 1.05514787e-01 -5.49299717e-01 -4.10440825e-02
9.41690445e-01 1.31337601e-03 -2.74647474e-01 -2.37838238e-01
-5.53211510e-01 -1.83769837e-01 3.14698517e-01 4.03289378e-01
1.53909475e-01 -1.10280836e+00 -5.95403671e-01 7.94610083e-01
5.34968711e-02 1.57088369e-01 6.51915371e-01 9.17474389e-01
-5.37616491e-01 2.39700258e-01 -3.84001344e-01 -7.93656588e-01
-7.72844672e-01 6.79208279e-01 5.02895772e-01 -1.96722358e-01
-6.18860424e-01 6.74850941e-01 1.55953288e-01 -5.56488097e-01
-4.35230732e-02 -3.03714663e-01 -1.22733094e-01 2.27481574e-01
4.46351826e-01 5.10923266e-01 5.46099007e-01 -3.63302469e-01
9.30300169e-03 5.00451088e-01 1.96117356e-01 3.78067270e-02
1.62107360e+00 1.60759017e-01 -1.70997486e-01 6.47231698e-01
8.97328854e-01 5.19841425e-02 -1.43690383e+00 -2.35299632e-01
5.31232841e-02 -4.62047935e-01 1.28236338e-01 -8.41946304e-01
-1.13729882e+00 8.20418358e-01 6.20300293e-01 4.46797401e-01
1.34343612e+00 -7.83248618e-02 6.04957283e-01 1.93720251e-01
3.08247000e-01 -1.09919500e+00 -2.37270653e-01 5.61106920e-01
9.82534230e-01 -1.07625329e+00 -3.06949317e-01 1.67543560e-01
-4.10562307e-01 1.28820992e+00 5.69925785e-01 1.62616521e-01
5.18259346e-01 2.03371689e-01 2.50060022e-01 -7.36014172e-02
-9.60770845e-01 -2.58431464e-01 8.24096948e-02 4.20244336e-01
6.50829151e-02 -8.45453143e-02 -4.13676798e-01 7.57389784e-01
-4.04826701e-01 -9.02521834e-02 6.63755655e-01 3.58902127e-01
-2.39844069e-01 -1.01825619e+00 -7.96050549e-01 4.84368205e-01
-3.60579789e-01 5.48580326e-02 3.34130853e-01 7.10910916e-01
2.83648133e-01 7.97754586e-01 3.69659454e-01 -9.51330587e-02
2.63754785e-01 4.41298723e-01 5.88870086e-02 -1.23653412e-02
-6.53515458e-01 1.37939245e-01 5.28749116e-02 -3.13634306e-01
-3.40459675e-01 -1.01574588e+00 -9.12740648e-01 -1.73549592e-01
-3.04130018e-01 1.03678286e-01 9.39113498e-01 9.38403249e-01
-7.90898278e-02 7.18404591e-01 8.45714509e-01 -9.64501202e-01
-4.06568289e-01 -1.11086798e+00 -6.08098507e-01 6.20655358e-01
3.95565927e-01 -9.30157065e-01 -8.38480234e-01 -1.74781725e-01] | [7.5849127769470215, 3.123342752456665] |
5fcabfa0-4747-4b59-ab83-fcbd1a9b085d | reactive-human-to-robot-handovers-of | 2011.08961 | null | https://arxiv.org/abs/2011.08961v2 | https://arxiv.org/pdf/2011.08961v2.pdf | Reactive Human-to-Robot Handovers of Arbitrary Objects | Human-robot object handovers have been an actively studied area of robotics over the past decade; however, very few techniques and systems have addressed the challenge of handing over diverse objects with arbitrary appearance, size, shape, and rigidity. In this paper, we present a vision-based system that enables reactive human-to-robot handovers of unknown objects. Our approach combines closed-loop motion planning with real-time, temporally-consistent grasp generation to ensure reactivity and motion smoothness. Our system is robust to different object positions and orientations, and can grasp both rigid and non-rigid objects. We demonstrate the generalizability, usability, and robustness of our approach on a novel benchmark set of 26 diverse household objects, a user study with naive users (N=6) handing over a subset of 15 objects, and a systematic evaluation examining different ways of handing objects. More results and videos can be found at https://sites.google.com/nvidia.com/handovers-of-arbitrary-objects. | ['Dieter Fox', 'Maya Cakmak', 'Yu-Wei Chao', 'Arsalan Mousavian', 'Chris Paxton', 'Wei Yang'] | 2020-11-17 | null | null | null | null | ['grasp-generation'] | ['computer-vision'] | [-2.02965364e-01 -1.28543392e-01 1.24013133e-01 -2.03857988e-01
-2.95678198e-01 -9.64844942e-01 9.99580026e-02 -1.60558328e-01
-8.23152885e-02 5.04172504e-01 -1.68367520e-01 -5.29664047e-02
-2.25770354e-01 -1.93325609e-01 -6.81633770e-01 -5.06042004e-01
-3.82250369e-01 8.85418892e-01 6.36525512e-01 -5.73404431e-01
3.54959667e-01 9.85555172e-01 -1.50608385e+00 6.49497285e-02
6.65884078e-01 7.13951468e-01 7.71729827e-01 8.52326751e-01
6.32746994e-01 2.75667459e-01 -3.19341719e-01 3.45507152e-02
4.72146004e-01 5.00806987e-01 -8.92692924e-01 2.54594982e-01
2.61919200e-01 -7.94660211e-01 -3.02128673e-01 5.54649830e-01
9.09401953e-01 2.68258959e-01 4.47700232e-01 -1.56152320e+00
-7.97625840e-01 2.57441580e-01 -4.05881852e-01 -1.68842480e-01
9.86341834e-01 6.97868407e-01 5.56658685e-01 -6.45589232e-01
6.86954498e-01 1.53259814e+00 5.05030334e-01 7.19708025e-01
-1.07088065e+00 -4.04359013e-01 2.12564647e-01 8.08346048e-02
-1.34285593e+00 -4.45201993e-01 3.76935393e-01 -5.61535418e-01
1.08167434e+00 2.48300731e-01 6.25683606e-01 1.13356853e+00
3.67381692e-01 7.50586033e-01 5.79112470e-01 -3.50751191e-01
2.32260749e-01 -1.37780467e-02 -6.83358982e-02 4.33170140e-01
4.85920519e-01 -1.20366022e-01 -9.13371146e-02 -2.29370788e-01
1.31639004e+00 -3.92362252e-02 -3.99903595e-01 -1.17798686e+00
-1.57806313e+00 1.23588599e-01 4.38757181e-01 8.22624099e-03
-7.16267765e-01 4.25849378e-01 3.94673012e-02 -1.32286623e-01
-2.71385401e-01 3.91272545e-01 -6.70165777e-01 -4.16414440e-01
-8.35623126e-03 7.35001922e-01 1.05556893e+00 1.79505706e+00
-1.47844208e-02 -3.34074974e-01 -2.36122042e-01 5.50408125e-01
2.86083996e-01 7.41019130e-01 -4.35822010e-02 -1.41026783e+00
4.59185213e-01 1.26670480e-01 1.12529421e+00 -9.48070467e-01
-8.31796825e-01 4.55573887e-01 -1.86779737e-01 5.16003311e-01
4.60694939e-01 -2.52157986e-01 -8.42818975e-01 1.27103817e+00
5.55662870e-01 -6.63364530e-01 -7.15352222e-02 1.41355610e+00
5.50007999e-01 2.35267207e-01 -2.88168807e-02 1.67311504e-01
1.21588671e+00 -1.13286507e+00 -6.25346839e-01 -9.89526287e-02
-5.32962494e-02 -9.02541220e-01 1.18413270e+00 6.08603597e-01
-1.31831384e+00 -5.02239347e-01 -6.82168782e-01 4.78453748e-02
2.00507287e-02 1.12854943e-01 7.95635879e-01 3.69336724e-01
-1.18794346e+00 5.43122649e-01 -1.03877068e+00 -9.32376325e-01
1.54913381e-01 8.93326700e-01 -1.51068330e-01 -1.51675016e-01
-3.70209038e-01 1.03069305e+00 2.94304967e-01 2.13064536e-01
-7.56939471e-01 -2.17343509e-01 -2.76302606e-01 -2.75453418e-01
4.63576555e-01 -7.13967800e-01 1.69568801e+00 -4.84840274e-01
-1.90866983e+00 5.14643013e-01 2.27350220e-01 1.63850233e-01
7.20528066e-01 -6.96681559e-01 3.73311900e-02 1.73269540e-01
4.27760184e-02 8.93284798e-01 8.69762242e-01 -1.59468496e+00
-4.67857629e-01 -2.20764384e-01 3.51927072e-01 5.07930219e-01
5.53203076e-02 -5.00065573e-02 -5.78979552e-01 -4.83932793e-01
2.64080077e-01 -1.57697177e+00 -1.93596587e-01 3.88813406e-01
-4.15149748e-01 -1.29657820e-01 6.69855535e-01 -4.11864161e-01
3.80169004e-01 -2.03986669e+00 5.54247737e-01 1.27833486e-01
-3.31680775e-01 -1.22841492e-01 -1.79132029e-01 5.28405666e-01
2.88680404e-01 -2.78771967e-01 1.71093613e-01 9.98030677e-02
1.14856429e-01 9.55559611e-02 -1.07797213e-01 4.04939651e-01
-4.56579886e-02 7.44702339e-01 -1.12153709e+00 -3.12009931e-01
1.70802012e-01 4.82951581e-01 -4.83561099e-01 3.38191718e-01
-2.97326744e-01 6.44968331e-01 -5.88730752e-01 1.13229871e+00
6.50306821e-01 -5.84852584e-02 4.43540186e-01 -2.57592201e-01
-7.18361093e-03 -2.79249877e-01 -1.36378920e+00 1.64066386e+00
-1.61931768e-01 2.85421848e-01 4.80176419e-01 -4.90634367e-02
5.75871050e-01 3.89791101e-01 5.95558107e-01 -2.73444861e-01
3.40434253e-01 4.21987951e-01 -1.14001771e-02 -1.01453137e+00
7.19027460e-01 4.95397329e-01 3.23081240e-02 1.28055379e-01
-3.55778366e-01 -6.01006091e-01 4.94524129e-02 -2.00859651e-01
1.12322986e+00 6.67624593e-01 9.84097719e-02 -3.12626630e-01
-1.43179879e-01 2.33648211e-01 -7.25467801e-02 7.36939490e-01
-2.82744288e-01 8.38425636e-01 -1.97217733e-01 -3.48466128e-01
-1.01824701e+00 -1.24617934e+00 -4.15483154e-02 1.27999425e+00
8.57869923e-01 3.04350406e-01 -5.59634387e-01 -1.48317084e-01
5.39330304e-01 5.86121678e-01 -2.05302775e-01 3.08051139e-01
-7.65843868e-01 -2.37819254e-01 -3.27302841e-03 7.80299366e-01
1.34869277e-01 -1.64373338e+00 -1.42320085e+00 3.08282852e-01
-3.03016305e-01 -1.21070254e+00 -5.81740975e-01 -1.62618369e-01
-8.39773357e-01 -9.97191012e-01 -1.03693354e+00 -9.47983742e-01
9.14786577e-01 6.58676744e-01 8.74982178e-01 -1.87092852e-02
-5.12642264e-01 1.09791636e+00 -5.41407883e-01 -3.43975961e-01
-1.24965906e-01 -4.43468690e-02 5.87969303e-01 -3.66994172e-01
-3.54848027e-01 -4.57402349e-01 -7.10587800e-01 8.90807450e-01
-4.48943287e-01 9.63930413e-02 6.15298212e-01 1.53054640e-01
2.02021360e-01 -3.87415409e-01 5.71634769e-02 1.50160789e-01
7.27752268e-01 -3.54825318e-01 -6.18317842e-01 2.52183497e-01
4.56948020e-02 -3.41587096e-01 -8.91269967e-02 -9.91409004e-01
-8.61372650e-01 2.77956009e-01 3.42164189e-01 -4.14600432e-01
-2.89830923e-01 -5.79810776e-02 -2.68015638e-02 -4.11110073e-01
6.72844648e-01 -3.49997789e-01 -1.32241920e-01 -3.88038814e-01
5.07449865e-01 7.69718230e-01 5.39108217e-01 -1.01470280e+00
5.70555389e-01 5.95401227e-01 -2.94451892e-01 -7.59226620e-01
1.15336448e-01 -4.16686982e-01 -9.28652227e-01 -3.97656083e-01
6.86377466e-01 -5.43725073e-01 -1.40803242e+00 7.93373525e-01
-1.53254735e+00 -9.68107700e-01 3.09245139e-02 5.21998048e-01
-1.05024290e+00 8.01029131e-02 -6.05339766e-01 -1.00332916e+00
-3.13240081e-01 -1.24515831e+00 1.26422441e+00 3.51868987e-01
-6.13474309e-01 -1.86127514e-01 -3.61249447e-01 1.21290453e-01
5.42830348e-01 3.10377896e-01 4.67772126e-01 -8.32661688e-02
-8.76656413e-01 -1.29503027e-01 -2.05977112e-01 -4.24133956e-01
4.91449296e-01 3.54811028e-02 -3.85727733e-01 -9.02015805e-01
-5.20358205e-01 -4.66131270e-01 -1.06410317e-01 4.30164546e-01
6.98343039e-01 -2.76452899e-01 -6.25466883e-01 -3.72148231e-02
1.12668276e+00 5.19562364e-01 6.03320539e-01 4.84660536e-01
7.99841940e-01 6.42601073e-01 1.11357224e+00 6.32158637e-01
4.33219373e-01 1.20172548e+00 6.45248950e-01 5.22507250e-01
2.45899037e-01 5.38992465e-01 2.21684828e-01 2.50885457e-01
-7.51144528e-01 -4.18929189e-01 -9.99051809e-01 6.77284062e-01
-1.90108800e+00 -6.50494456e-01 3.39873247e-02 2.07403898e+00
5.47758222e-01 1.30285278e-01 4.06179398e-01 -3.24644893e-01
7.79318511e-01 -5.61808109e-01 -8.52938414e-01 -2.41167814e-01
6.06720209e-01 -2.38968253e-01 6.65766835e-01 3.66038710e-01
-8.24900150e-01 9.71950829e-01 5.87276554e+00 1.39506692e-02
-1.08137321e+00 -2.11414173e-01 -3.55202526e-01 -3.60799104e-01
2.24134743e-01 -2.43605599e-01 -5.77575803e-01 2.37165391e-01
1.05362721e-01 2.16416061e-01 7.05392122e-01 1.00461376e+00
2.22287282e-01 -2.61323124e-01 -1.23324013e+00 8.32310259e-01
-2.05801800e-01 -5.94495654e-01 -3.49984348e-01 -2.23958433e-01
4.38833892e-01 1.11531980e-01 1.54571431e-02 1.75996721e-01
3.64608914e-01 -6.62036121e-01 1.41745234e+00 5.91916621e-01
5.95638037e-01 -2.15211079e-01 3.74163389e-01 2.51814425e-01
-1.11971760e+00 -4.05105293e-01 -9.47928429e-02 -5.23673892e-02
4.83917296e-01 -2.16756970e-01 -8.23545337e-01 3.62286031e-01
1.36681461e+00 6.54808283e-02 -1.96766466e-01 1.19273353e+00
2.01427355e-01 -2.83671141e-01 -4.41171020e-01 -3.70643646e-01
-1.68293297e-01 1.51446164e-01 8.00821424e-01 9.67246234e-01
8.58670250e-02 6.41609311e-01 5.46925128e-01 5.95170319e-01
4.98255461e-01 -2.23743364e-01 -3.48977923e-01 1.26871511e-01
4.29249227e-01 1.01543427e+00 -1.06447864e+00 1.17265712e-02
-1.05205309e-02 1.32892132e+00 1.15376830e-01 5.66833615e-01
-9.64176714e-01 -5.24196923e-01 6.38272643e-01 1.60282895e-01
4.48943287e-01 -9.62855875e-01 -1.08194731e-01 -8.18569124e-01
5.65291703e-01 -8.76964271e-01 -2.39238098e-01 -1.19697368e+00
-1.02393746e+00 5.76938689e-01 3.58254910e-01 -1.38226259e+00
8.14551022e-03 -6.43730342e-01 -1.97580144e-01 5.54854274e-01
-8.12588274e-01 -1.14264286e+00 -9.30727601e-01 3.15472603e-01
8.05034876e-01 2.08203509e-01 6.77078485e-01 -4.87524122e-02
-7.12504983e-02 1.41194150e-01 -1.21699393e-01 -3.66149515e-01
8.50801706e-01 -8.01837683e-01 5.05259454e-01 1.12508088e-01
-8.04491282e-01 7.01813936e-01 8.78660500e-01 -9.58655953e-01
-1.97761703e+00 -5.92126966e-01 -7.89996888e-03 -7.90620327e-01
3.56991708e-01 -3.57243747e-01 -7.27312326e-01 8.65588188e-01
2.64954362e-02 5.03342822e-02 -1.72031134e-01 -3.68303716e-01
1.89536899e-01 2.47725710e-01 -1.45096946e+00 9.55395281e-01
1.40855753e+00 1.43983945e-01 -3.74626637e-01 3.73180926e-01
8.00115108e-01 -1.18743753e+00 -7.35765934e-01 5.91566443e-01
1.26340616e+00 -6.44973397e-01 1.06526864e+00 -3.35722774e-01
-7.31996074e-02 -5.15424371e-01 -4.82591331e-01 -9.09094393e-01
-7.69878626e-01 -9.62352872e-01 -1.53367579e-01 7.36451209e-01
1.95595138e-02 -4.17861611e-01 5.52842796e-01 9.19447005e-01
-1.22185118e-01 -6.61105812e-01 -5.71759641e-01 -1.03504062e+00
-4.81514484e-01 -1.15840912e-01 3.62483412e-01 5.86015642e-01
2.87715942e-01 -1.76147208e-01 -2.41492182e-01 7.45749712e-01
4.59658027e-01 4.46668714e-02 9.44600642e-01 -1.12516630e+00
-2.75165021e-01 -3.18925619e-01 -2.85608500e-01 -1.01984823e+00
-1.80559203e-01 -2.98005223e-01 6.74909472e-01 -1.90560770e+00
4.80039492e-02 -7.09947526e-01 4.35764074e-01 6.17633283e-01
1.43631890e-01 2.93095279e-02 5.50724387e-01 6.09133244e-01
-6.53417587e-01 3.33332032e-01 1.42262435e+00 5.10812439e-02
-6.77380800e-01 1.44472957e-01 -1.10786520e-01 6.39589369e-01
9.62041259e-01 -1.63050652e-01 -2.47184396e-01 -6.52052760e-01
-1.77812099e-01 2.09684089e-01 5.35630107e-01 -1.21904016e+00
1.23420261e-01 -6.69531405e-01 2.25508079e-01 -3.12749356e-01
5.54105937e-01 -1.15315545e+00 3.85638565e-01 7.56772697e-01
1.09210901e-01 3.36347729e-01 5.95205784e-01 4.37613428e-01
8.13995481e-01 -1.33532686e-02 4.83232409e-01 -1.26677349e-01
-8.81269395e-01 9.50256363e-02 -4.06402290e-01 -2.39873603e-01
1.39986992e+00 -3.35135639e-01 -1.33038610e-01 -3.63466829e-01
-7.71911919e-01 4.48580772e-01 7.78723657e-01 1.06086385e+00
7.99428940e-01 -1.18086910e+00 -4.84451354e-01 -1.72942311e-01
-1.13863898e-02 2.10560888e-01 -4.15720381e-02 7.06588686e-01
-9.28474903e-01 2.90372908e-01 -5.66988230e-01 -8.47678065e-01
-1.27251470e+00 6.91195786e-01 3.37900557e-02 6.85356617e-01
-5.43067992e-01 5.78515351e-01 -1.62959665e-01 -5.70203245e-01
5.24565578e-01 -4.96074110e-01 1.75032377e-01 -6.50771499e-01
1.25877336e-01 8.64298940e-01 -2.09840328e-01 -4.82213885e-01
-6.67593718e-01 8.82443905e-01 7.72583336e-02 -2.19310865e-01
1.04119694e+00 -2.76055008e-01 6.95812106e-02 2.15706065e-01
4.60034162e-01 -1.48972020e-01 -1.61983204e+00 2.17917472e-01
-8.87525454e-02 -5.04498780e-01 -8.07016790e-01 -8.18935394e-01
-6.04207039e-01 1.89739138e-01 6.94754839e-01 1.62507892e-02
6.96022272e-01 1.84970796e-01 5.49182773e-01 8.57273161e-01
1.18393517e+00 -8.64349782e-01 5.15070438e-01 5.77563107e-01
1.60117984e+00 -1.01086628e+00 -7.17686936e-02 -7.29754388e-01
-7.58294940e-01 1.15817928e+00 1.03155923e+00 -9.85300727e-03
2.88878858e-01 3.64659935e-01 1.32791502e-02 -1.59626007e-01
-1.78960219e-01 8.22511390e-02 6.01139069e-02 8.69501114e-01
8.93867165e-02 1.88036904e-01 9.36623588e-02 2.59273887e-01
-1.96363358e-03 2.85859048e-01 6.33644938e-01 1.75511730e+00
-4.87574428e-01 -6.06407404e-01 -6.59095585e-01 1.06286034e-02
6.43766969e-02 4.30043101e-01 -4.27566729e-02 8.89482856e-01
-2.56120384e-01 9.76154387e-01 -7.31650963e-02 -2.54664600e-01
8.41360092e-01 -3.26663017e-01 1.06878233e+00 -5.94458401e-01
-2.91940600e-01 -2.40987271e-01 -1.02840960e-02 -9.47969854e-01
-2.07793951e-01 -7.29834020e-01 -1.50405633e+00 -2.52021492e-01
-2.05360591e-01 -4.77428854e-01 8.24146032e-01 5.08942306e-01
6.04949296e-01 2.97593683e-01 2.96801239e-01 -2.15349674e+00
-7.90920556e-01 -7.60192752e-01 -3.50407034e-01 4.66444969e-01
3.42282593e-01 -1.00036347e+00 1.01248726e-01 1.46216573e-02] | [5.653860092163086, -0.6644056439399719] |
727fa164-b697-4045-9932-cc7e66d404ef | white-box-membership-attack-against-machine | 2206.03584 | null | https://arxiv.org/abs/2206.03584v1 | https://arxiv.org/pdf/2206.03584v1.pdf | White-box Membership Attack Against Machine Learning Based Retinopathy Classification | The advances in machine learning (ML) have greatly improved AI-based diagnosis aid systems in medical imaging. However, being based on collecting medical data specific to individuals induces several security issues, especially in terms of privacy. Even though the owner of the images like a hospital put in place strict privacy protection provisions at the level of its information system, the model trained over his images still holds disclosure potential. The trained model may be accessible to an attacker as: 1) White-box: accessing to the model architecture and parameters; 2) Black box: where he can only query the model with his own inputs through an appropriate interface. Existing attack methods include: feature estimation attacks (FEA), membership inference attack (MIA), model memorization attack (MMA) and identification attacks (IA). In this work we focus on MIA against a model that has been trained to detect diabetic retinopathy from retinal images. Diabetic retinopathy is a condition that can cause vision loss and blindness in the people who have diabetes. MIA is the process of determining whether a data sample comes from the training data set of a trained ML model or not. From a privacy perspective in our use case where a diabetic retinopathy classification model is given to partners that have at their disposal images along with patients' identifiers, inferring the membership status of a data sample can help to state if a patient has contributed or not to the training of the model. | ['Gouenou Coatrieux', 'Gwenolé Quellec', 'Reda Bellafqira', 'Mounia Hamidouche'] | 2022-05-30 | null | null | null | null | ['inference-attack', 'membership-inference-attack'] | ['adversarial', 'computer-vision'] | [ 5.64162552e-01 4.76450980e-01 -1.65330797e-01 -5.81286848e-01
-2.93649554e-01 -5.15212715e-01 3.73365402e-01 1.79272011e-01
-5.12186944e-01 6.55588090e-01 -2.78338641e-01 -6.03996813e-01
-8.17868561e-02 -9.65827346e-01 -8.33393216e-01 -6.78703547e-01
-1.32445887e-01 5.71721256e-01 -3.02393019e-01 6.03484869e-01
2.98355222e-01 8.66808832e-01 -1.18629527e+00 4.10141587e-01
6.33966565e-01 9.91184890e-01 -5.32126904e-01 8.68890047e-01
2.56169885e-01 6.22159302e-01 -9.74764407e-01 -8.45972240e-01
8.12996149e-01 -8.92018601e-02 -9.10646439e-01 -5.68490587e-02
6.14258230e-01 -1.00767934e+00 -1.44236684e-01 1.48639989e+00
2.40840092e-01 -7.17871547e-01 3.96635473e-01 -1.69404125e+00
-1.31727025e-01 3.65036607e-01 -3.56152624e-01 -9.29545760e-02
3.01771969e-01 3.45336646e-01 1.18148379e-01 -1.09262966e-01
6.97198510e-01 1.09834826e+00 3.38616371e-01 7.68024981e-01
-1.35504436e+00 -9.50051486e-01 -5.82636237e-01 -1.34092778e-01
-1.54835963e+00 -6.14614248e-01 2.57250279e-01 -6.18737698e-01
5.34817696e-01 6.09836698e-01 5.47088981e-01 9.51785624e-01
3.55339170e-01 3.60050827e-01 1.14654672e+00 -3.59897912e-01
4.11732018e-01 6.68516874e-01 5.92314899e-01 6.97906792e-01
8.33702981e-01 4.51714993e-01 -3.75164419e-01 -9.99440253e-01
1.10588598e+00 -1.84144273e-01 -1.60039067e-01 -4.95453089e-01
-6.90773785e-01 9.27007914e-01 1.32679865e-01 -4.35997605e-01
-4.44172263e-01 -3.99450883e-02 2.59201735e-01 4.94858533e-01
-2.53589839e-01 3.45627218e-01 -3.96693617e-01 3.40826988e-01
-4.96225327e-01 9.84870344e-02 1.04232812e+00 8.51315141e-01
7.91957319e-01 -4.51407582e-01 3.55538875e-02 -6.57120943e-02
3.88211370e-01 3.18833798e-01 2.21485019e-01 -6.80287302e-01
3.32564622e-01 9.04552877e-01 2.53773749e-01 -9.86066699e-01
-3.02573349e-02 -7.61109367e-02 -8.27478051e-01 7.10994065e-01
5.29321551e-01 -4.98932838e-01 -8.50679636e-01 1.47158492e+00
4.68894899e-01 1.20851047e-01 4.33596611e-01 7.89283335e-01
7.56517410e-01 1.93906859e-01 1.37176529e-01 -5.01310676e-02
1.53238547e+00 -1.32698432e-01 -6.12395406e-01 -1.55041829e-01
5.98466158e-01 -5.11467636e-01 2.59665042e-01 3.89625907e-01
-9.14448798e-01 -3.16268027e-01 -1.04887843e+00 1.70528948e-01
-4.74006772e-01 1.87748030e-01 6.12283349e-01 1.34608448e+00
-7.96836317e-01 7.17469230e-02 -7.81149387e-01 -5.54297626e-01
1.09091055e+00 1.09336627e+00 -8.24515283e-01 -1.92736700e-01
-9.95318115e-01 7.33919561e-01 -5.04594222e-02 -6.95654228e-02
-7.05528021e-01 -7.28539944e-01 -7.96814382e-01 -1.85080916e-01
2.00591922e-01 -1.08955395e+00 6.15625262e-01 -1.14726889e+00
-8.11632395e-01 1.30339539e+00 -1.33157030e-01 -9.95710671e-01
8.70943427e-01 -1.08687878e-01 -5.51193833e-01 1.28433421e-01
-6.92689270e-02 5.76873243e-01 1.04006684e+00 -1.01759493e+00
-5.95964551e-01 -8.69114399e-01 6.04134277e-02 -3.86200696e-01
7.45906010e-02 2.76849538e-01 -5.79912141e-02 -1.61762342e-01
-1.56204164e-01 -7.43481934e-01 -2.52697647e-01 5.32637298e-01
-9.06048775e-01 4.43221927e-01 1.01391757e+00 -7.27519989e-01
1.18082738e+00 -2.04993320e+00 -6.62804365e-01 7.81277418e-01
1.70351073e-01 8.79693329e-01 2.63183922e-01 3.32372814e-01
-1.92858223e-02 5.53630590e-01 -1.28280506e-01 -5.27239554e-02
-3.97997230e-01 1.63247198e-01 -3.36739868e-01 5.65234005e-01
2.07663506e-01 5.94762802e-01 -3.44957896e-02 -3.87826264e-01
1.57025144e-01 4.59236830e-01 -4.49671298e-01 3.66200656e-01
1.69242784e-01 4.56313610e-01 -7.01820016e-01 6.93577588e-01
1.03664100e+00 -1.66583017e-01 2.39578754e-01 -3.12319934e-01
1.83694899e-01 -2.57255472e-02 -1.24873745e+00 8.11326504e-01
1.86176375e-01 3.23497206e-01 1.73556075e-01 -6.57020211e-01
6.94802940e-01 4.13323462e-01 -2.54069511e-02 -1.70780703e-01
2.77236551e-01 -1.28193185e-01 2.67550021e-01 -9.45263445e-01
-2.70620584e-02 3.95388842e-01 2.70495147e-01 6.15365505e-01
-4.43343222e-01 7.28534400e-01 -4.65577573e-01 7.64179379e-02
1.11446798e+00 -3.80384475e-01 5.74740350e-01 -3.32928710e-02
7.00584769e-01 -6.28155693e-02 4.28951651e-01 1.00760865e+00
-3.19092661e-01 2.51848012e-01 5.69619894e-01 -6.63244843e-01
-8.81178081e-01 -7.83034503e-01 -5.14557183e-01 -2.30825506e-02
-6.85565174e-02 -2.08624452e-01 -8.12945187e-01 -8.54816973e-01
4.26387370e-01 4.28773522e-01 -6.69530690e-01 -1.52738050e-01
-2.49575689e-01 -4.09575671e-01 8.58282030e-01 1.49648964e-01
8.00080776e-01 -6.39129341e-01 -1.14429140e+00 -1.41974822e-01
1.23654239e-01 -7.77589977e-01 -2.45247319e-01 -3.73033434e-01
-8.01210821e-01 -1.50968158e+00 -1.28346533e-01 -5.27197421e-01
1.31529164e+00 -4.32575941e-01 4.88295227e-01 2.77833730e-01
-9.01934624e-01 4.37190175e-01 1.89810216e-01 -8.20138812e-01
-7.45057702e-01 -3.79281104e-01 -8.21020901e-02 5.15675783e-01
9.81628120e-01 -1.37163132e-01 -5.06636441e-01 1.38580963e-01
-9.23186839e-01 -2.31149584e-01 4.63409275e-01 4.45311099e-01
2.07639024e-01 2.06262797e-01 1.93254650e-01 -1.57012856e+00
3.28382760e-01 -3.69568020e-01 -9.77849364e-01 4.28661793e-01
-6.64613962e-01 -1.93409041e-01 2.49432474e-02 -4.72212881e-01
-5.61031878e-01 5.37259936e-01 3.83017421e-01 -3.87919813e-01
-6.16086245e-01 3.65361243e-01 -6.27768457e-01 -5.12136102e-01
4.18219090e-01 -7.46281594e-02 7.67225921e-01 -4.14262563e-01
-2.14284360e-01 1.13571119e+00 4.28229213e-01 -3.32276188e-02
6.94162190e-01 6.58633888e-01 2.78780311e-01 -6.03068292e-01
-1.05567947e-01 -8.77477303e-02 -4.97939438e-01 -9.19199958e-02
7.69325197e-01 -7.09634423e-01 -1.38578355e+00 8.48319292e-01
-1.01648688e+00 3.41745824e-01 7.72970542e-02 5.43124199e-01
2.94865556e-02 1.70170650e-01 -2.47822344e-01 -1.05302203e+00
-2.77397782e-01 -1.08755541e+00 3.36579770e-01 3.17337453e-01
-3.93068820e-01 -7.58876801e-01 -4.75437611e-01 8.52279723e-01
4.37489897e-01 4.63767588e-01 1.29686964e+00 -9.10932839e-01
-8.82396042e-01 -7.61918366e-01 -1.06784090e-01 3.44962686e-01
2.00381249e-01 2.34150551e-02 -1.15918326e+00 -5.00491679e-01
2.73832291e-01 -1.58133879e-01 2.12409258e-01 5.22215724e-01
8.91084909e-01 -9.51103151e-01 -6.65732622e-01 5.32892525e-01
1.32236505e+00 4.32902664e-01 1.01365101e+00 2.39325598e-01
3.28080088e-01 8.84067357e-01 3.17518115e-01 4.22988325e-01
-1.39159476e-02 3.53902340e-01 5.04676044e-01 -3.17676336e-01
4.84144837e-01 -1.55071720e-01 1.51165202e-01 -6.81703269e-01
2.81030178e-01 1.61705655e-03 -8.48746479e-01 1.35564998e-01
-1.63411534e+00 -6.88523829e-01 -1.42863870e-01 2.86133027e+00
7.85412788e-01 -7.45906979e-02 7.23570660e-02 -1.66649610e-01
8.01866949e-01 -5.68938434e-01 -8.87896657e-01 -4.84641969e-01
1.35959387e-01 2.33579412e-01 1.02292550e+00 4.32322890e-01
-1.12303936e+00 5.82445681e-01 5.89168072e+00 1.77032072e-02
-1.05044305e+00 -3.16507220e-01 9.68768895e-01 -1.13925479e-01
9.66674238e-02 3.24490368e-01 -9.31267262e-01 2.69700140e-01
8.05705667e-01 -2.78458774e-01 2.01340288e-01 6.22427404e-01
2.31952593e-02 -4.61774051e-01 -1.55024457e+00 9.83393550e-01
1.97488844e-01 -1.38173902e+00 3.60581189e-01 7.93429017e-01
8.21990445e-02 -4.50664878e-01 3.25776249e-01 -3.53818595e-01
2.72067368e-01 -1.40176499e+00 1.94227412e-01 8.57220948e-01
8.63170147e-01 -1.04352856e+00 8.54048669e-01 5.20230711e-01
-4.14166391e-01 -1.23368643e-01 -4.76659924e-01 1.12268835e-01
-1.77799761e-01 1.81414783e-01 -1.31864989e+00 4.87971485e-01
6.07380390e-01 5.44853061e-02 -6.66421115e-01 1.08869135e+00
1.09306961e-01 6.79605126e-01 -2.43129492e-01 3.70743960e-01
-2.91079134e-01 -2.90853173e-01 6.71050906e-01 4.86587673e-01
3.33842337e-02 2.80563504e-01 -1.99996561e-01 1.08436918e+00
6.91604614e-02 -1.64450750e-01 -7.38464832e-01 -2.55174160e-01
5.63037038e-01 7.51566827e-01 9.60509628e-02 4.89944406e-02
-3.30266625e-01 7.28041232e-01 -1.72160894e-01 2.30041072e-01
-1.29002631e-01 -1.88358516e-01 8.97538304e-01 5.08753717e-01
6.17016740e-02 3.30684215e-01 -1.26348421e-01 -4.86938059e-01
2.43717413e-02 -1.18213427e+00 5.94458759e-01 -6.42956316e-01
-1.25633693e+00 4.24732953e-01 -2.75987118e-01 -1.04633772e+00
-3.46636057e-01 -4.63214427e-01 -2.80054092e-01 1.45521593e+00
-1.09479296e+00 -1.41391790e+00 7.32731298e-02 9.93400931e-01
-5.36810040e-01 -7.03456402e-01 1.21194804e+00 1.24501608e-01
-6.71994746e-01 9.09815311e-01 -4.42935407e-01 7.86817908e-01
5.60052931e-01 -5.97869635e-01 3.17793153e-02 1.00648892e+00
-8.63477141e-02 1.03596807e+00 4.39472556e-01 -9.96368647e-01
-1.57051039e+00 -1.07966304e+00 1.04310000e+00 -5.51478624e-01
5.51054142e-02 -3.18525463e-01 -8.95477712e-01 1.03984022e+00
-2.13892028e-01 -7.59023651e-02 1.10501659e+00 -3.11184734e-01
-1.36990890e-01 -2.39436388e-01 -1.95007491e+00 3.71987879e-01
1.43642440e-01 -6.57278419e-01 -1.11304693e-01 1.65323034e-01
1.62159681e-01 -2.06725031e-01 -8.54724586e-01 2.18133405e-01
7.08975852e-01 -9.04578984e-01 8.11759055e-01 -1.15568542e+00
-1.15936905e-01 -3.38230014e-01 -1.80559419e-02 -3.04415286e-01
2.50310242e-01 -8.68406594e-01 1.11678550e-02 1.40821612e+00
2.48036742e-01 -1.09197366e+00 1.04219222e+00 1.77604532e+00
9.68850076e-01 -2.73388833e-01 -1.13106883e+00 -4.69825208e-01
-1.59457132e-01 -1.30124435e-01 1.02072072e+00 1.04842329e+00
-2.67668605e-01 -1.17201291e-01 -4.87109214e-01 1.12861490e+00
9.07453001e-01 -3.78383815e-01 1.12445378e+00 -1.34668219e+00
-6.47340193e-02 1.43136159e-01 -8.53375793e-01 -3.73300076e-01
-2.31549487e-01 -6.28237367e-01 -7.73296893e-01 -9.76817727e-01
2.07399085e-01 -5.45456707e-01 -3.56486477e-02 5.59518754e-01
2.86351681e-01 -2.34560579e-01 3.59454192e-02 4.62806016e-01
4.56865489e-01 -4.95761365e-01 6.97457194e-01 7.04812035e-02
-1.24643609e-01 7.69765973e-01 -6.30659223e-01 5.60967386e-01
7.44465828e-01 -8.11480284e-01 -3.49956512e-01 -1.29546031e-01
6.61336333e-02 3.00743639e-01 1.04081452e+00 -9.40422416e-01
8.22539628e-01 7.46222138e-02 5.51456451e-01 -4.03887451e-01
1.50997058e-01 -1.42562997e+00 8.04869950e-01 8.97191405e-01
-3.68925929e-01 -4.66910094e-01 3.88145633e-02 4.65030909e-01
6.95244521e-02 -3.47086310e-01 7.33671844e-01 -1.54649422e-01
-1.87404856e-01 3.81774843e-01 -4.95670199e-01 -6.54233813e-01
1.23359859e+00 -5.45050323e-01 -6.39158785e-01 -2.58647621e-01
-8.37673485e-01 2.37653136e-01 6.89123809e-01 -4.82520461e-02
6.86023414e-01 -7.56637812e-01 -6.43619239e-01 1.06575048e+00
1.84992388e-01 -2.23841161e-01 3.60565493e-03 5.93923509e-01
-3.79724473e-01 1.07954100e-01 -3.26788783e-01 -3.18279088e-01
-2.10466123e+00 6.12844467e-01 5.49660087e-01 1.53837755e-01
-4.69166160e-01 5.06003082e-01 1.53526738e-01 -1.04111969e-01
5.84314704e-01 -1.04094915e-01 -2.52963483e-01 -1.21211246e-01
8.64418745e-01 3.45190316e-01 -1.35538429e-01 -6.40936255e-01
-3.73881310e-01 1.84529305e-01 -4.73180920e-01 -2.30173487e-02
8.92278790e-01 2.93388925e-02 -3.64340723e-01 -1.29644990e-01
1.06643486e+00 -3.72536592e-02 -9.58274126e-01 -2.55838394e-01
-3.08831960e-01 -9.38330591e-01 -1.80893186e-02 -1.30754066e+00
-1.14929664e+00 7.87576735e-01 1.09200227e+00 1.55312255e-01
9.20426786e-01 -2.27215081e-01 4.91943896e-01 3.37135583e-01
4.98998374e-01 -5.86985648e-01 -7.15539753e-01 -3.37867647e-01
5.96264184e-01 -1.24224401e+00 1.75491765e-01 -4.70574647e-01
-7.36557305e-01 9.63120162e-01 4.25518572e-01 2.41126239e-01
1.08129644e+00 3.33699971e-01 4.35325980e-01 -1.44574761e-01
-4.75399315e-01 3.87097746e-01 1.72139496e-01 1.14254081e+00
-3.52405012e-01 8.77499133e-02 2.28473432e-02 7.72515059e-01
-1.63469404e-01 4.95445102e-01 7.07935631e-01 9.99941945e-01
-9.91764888e-02 -1.42882836e+00 -4.58396375e-01 7.71555662e-01
-5.28098881e-01 1.00593632e-02 -8.11281741e-01 7.85036862e-01
3.40277493e-01 8.47172797e-01 3.41280311e-01 -8.36479217e-02
2.39796773e-01 1.14890702e-01 6.96680099e-02 -4.97418702e-01
-9.80029881e-01 -3.39557201e-01 3.29421401e-01 -5.57885885e-01
-3.92123967e-01 -8.73343110e-01 -1.07041633e+00 -3.82827371e-01
-1.24462821e-01 -1.70755446e-01 7.08923399e-01 6.69188499e-01
6.57886565e-01 -2.88884550e-01 6.18798554e-01 2.79403687e-01
-4.94747728e-01 -5.66018403e-01 -6.96350396e-01 2.27178663e-01
5.73865354e-01 3.58563252e-02 -2.99254715e-01 3.32999170e-01] | [5.975453853607178, 7.121121883392334] |
d1fbb0a8-22b8-40d0-87f6-65b3f377f951 | searching-with-consistent-prioritization-for | 1812.06356 | null | http://arxiv.org/abs/1812.06356v1 | http://arxiv.org/pdf/1812.06356v1.pdf | Searching with Consistent Prioritization for Multi-Agent Path Finding | We study prioritized planning for Multi-Agent Path Finding (MAPF). Existing
prioritized MAPF algorithms depend on rule-of-thumb heuristics and random
assignment to determine a fixed total priority ordering of all agents a priori.
We instead explore the space of all possible partial priority orderings as part
of a novel systematic and conflict-driven combinatorial search framework. In a
variety of empirical comparisons, we demonstrate state-of-the-art solution
qualities and success rates, often with similar runtimes to existing
algorithms. We also develop new theoretical results that explore the
limitations of prioritized planning, in terms of completeness and optimality,
for the first time. | ['Peter J. Stuckey', 'Sven Koenig', 'Daniel Harabor', 'Jiaoyang Li', 'Hang Ma'] | 2018-12-15 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [ 3.46062928e-01 3.36921930e-01 -4.24303353e-01 -1.24544285e-01
-6.60466969e-01 -8.35534275e-01 5.68665385e-01 3.02391469e-01
-4.12174731e-01 1.19103265e+00 2.45136380e-01 -5.62536180e-01
-1.08855486e+00 -1.00588822e+00 -3.70647609e-01 -3.55230361e-01
-8.37691486e-01 1.21058381e+00 8.41437936e-01 -5.55583417e-01
4.27397758e-01 6.63672447e-01 -1.04737937e+00 3.52107398e-02
6.76117063e-01 4.60711151e-01 2.78546453e-01 5.95722318e-01
1.05615295e-01 5.20912290e-01 -5.32351732e-01 7.09701627e-02
5.83956301e-01 -2.42290020e-01 -1.27805865e+00 1.22093663e-01
-4.02178258e-01 -3.50490481e-01 -3.02985668e-01 7.41146028e-01
1.13217324e-01 2.40362629e-01 1.98609844e-01 -1.93901455e+00
-1.40290204e-02 6.86765730e-01 -5.80274761e-01 2.48640582e-01
7.24028587e-01 6.28817528e-02 1.08978856e+00 -1.54306903e-01
9.88566518e-01 1.18139124e+00 3.60436708e-01 3.09473872e-01
-1.22880971e+00 -1.65873468e-01 7.81506956e-01 3.78467053e-01
-1.28800178e+00 -3.77969742e-01 2.74094641e-01 2.23152637e-02
1.41364598e+00 5.63738346e-01 6.22119784e-01 4.02498990e-01
2.76175410e-01 6.49192214e-01 9.26513910e-01 -6.13485932e-01
3.40116233e-01 -4.31966722e-01 4.21821438e-02 6.25348330e-01
4.81342435e-01 4.77935344e-01 -3.09457183e-01 -5.52299023e-01
5.97018361e-01 -2.70185024e-01 -1.93424627e-01 -7.50397503e-01
-1.43058145e+00 8.32294166e-01 2.39433691e-01 -1.25300214e-01
-7.07930028e-01 3.49765748e-01 2.38420621e-01 2.55001307e-01
-1.86561495e-01 6.97231710e-01 -6.29139662e-01 -2.16987252e-01
-5.16254842e-01 6.53865695e-01 9.14083362e-01 1.14032722e+00
7.94450641e-01 -4.09849048e-01 -1.46714936e-03 2.95013785e-01
2.48178661e-01 2.89081454e-01 -3.94028664e-01 -1.38304925e+00
6.03952587e-01 4.50125158e-01 5.84918618e-01 -9.22869921e-01
-8.89208615e-01 -1.30749449e-01 1.39339045e-02 5.94399452e-01
1.17636666e-01 -1.81895599e-01 -6.11797512e-01 1.68931830e+00
5.17013550e-01 -3.05805653e-01 1.97575837e-01 8.89589727e-01
-1.00997172e-01 8.68360400e-01 -3.71338367e-01 -7.78369486e-01
8.55616391e-01 -1.51207411e+00 -4.07770008e-01 -2.51235992e-01
5.26824951e-01 -4.18978214e-01 7.45691001e-01 4.42546219e-01
-1.22407174e+00 4.16756362e-01 -8.68741214e-01 6.17437303e-01
-1.37633190e-01 -7.50802994e-01 9.23191249e-01 6.46827102e-01
-1.28440726e+00 3.12845796e-01 -9.75916803e-01 -5.22924364e-01
-3.29021551e-02 5.07037282e-01 -2.06829220e-01 -3.15621704e-01
-7.73638844e-01 1.08424616e+00 3.49973202e-01 -5.26727326e-02
-9.68918800e-01 -3.20231318e-01 -5.76620400e-01 1.47952707e-02
1.23515892e+00 -5.20694852e-01 1.56656170e+00 -2.25590125e-01
-1.44128275e+00 -1.28572173e-02 -1.66957527e-01 -3.39444250e-01
4.56408024e-01 2.33201668e-01 -1.77084312e-01 1.85317770e-01
4.73687053e-01 5.47674417e-01 -1.90666914e-01 -1.40208614e+00
-1.26329708e+00 1.57716841e-01 7.47924149e-01 4.26151633e-01
1.99594930e-01 1.37015283e-01 -2.81778246e-01 4.07582037e-02
1.94463767e-02 -1.20199859e+00 -1.07015204e+00 -4.01969463e-01
-4.75928038e-01 -1.22769862e-01 1.47955000e-01 5.42538948e-02
1.36795878e+00 -1.49569821e+00 3.25857401e-01 5.78012526e-01
3.98348235e-02 -4.12112683e-01 -5.62390029e-01 1.18177915e+00
3.84908795e-01 -9.23365355e-03 -5.16237952e-02 1.94238380e-01
3.56071353e-01 4.10518944e-01 -1.37692764e-01 3.68503332e-01
-1.77322492e-01 6.60127342e-01 -1.33741069e+00 -3.18648845e-01
1.70758247e-01 -4.10884351e-01 -6.95967555e-01 -2.66372055e-01
-5.43259919e-01 -4.08471264e-02 -5.55725813e-01 6.41590238e-01
3.88885796e-01 -1.01635076e-01 7.42589533e-01 4.66535032e-01
-6.88052714e-01 5.19393861e-01 -1.19667828e+00 1.59102380e+00
-1.81714907e-01 1.91224679e-01 2.05165088e-01 -6.00749969e-01
3.15437615e-01 8.07185769e-02 9.19490576e-01 -7.87161231e-01
-2.40639031e-01 3.33100677e-01 1.69244915e-01 4.17852551e-02
7.26298571e-01 1.17379680e-01 -3.95863801e-01 9.24283028e-01
-7.69840300e-01 3.77273321e-01 7.60466933e-01 2.40340516e-01
1.65858388e+00 -7.49904066e-02 5.89678049e-01 -4.79602993e-01
1.82107836e-01 8.41749847e-01 5.35369396e-01 1.03309798e+00
-2.36869365e-01 -1.23618312e-01 4.96988595e-01 -6.94522798e-01
-6.72122002e-01 -8.60163927e-01 2.92584509e-01 9.99761522e-01
6.55263782e-01 -6.64832592e-01 -2.03524560e-01 -5.41392267e-01
-1.65247455e-01 9.25503552e-01 -5.01486301e-01 4.26999837e-01
-7.35376656e-01 -7.59136856e-01 -1.74795985e-02 2.10213408e-01
-2.04630243e-03 -1.06193674e+00 -1.46268904e+00 6.89390957e-01
1.72818825e-02 -1.20391655e+00 -3.12550396e-01 1.23718075e-01
-3.88718784e-01 -1.41316545e+00 -3.12172323e-02 -5.08374631e-01
6.94827557e-01 8.84172082e-01 9.02865410e-01 2.57186264e-01
1.85163751e-01 6.03802025e-01 -5.57102203e-01 -7.64881298e-02
-2.41629750e-01 9.40694660e-02 4.10618111e-02 -7.30245888e-01
-2.72457123e-01 -4.50689316e-01 -4.06780899e-01 6.34732664e-01
-3.68305176e-01 6.01124540e-02 6.50522232e-01 4.47420120e-01
4.54948992e-01 7.86918640e-01 3.68812710e-01 -4.92233843e-01
9.15003896e-01 -4.50947851e-01 -8.94779563e-01 5.13128936e-01
-8.07584107e-01 -1.73015907e-01 3.26927423e-01 -5.96110150e-02
-8.14798057e-01 2.66891718e-03 3.26344103e-01 1.55745506e-01
-4.01193798e-02 8.94078314e-01 1.80776678e-02 -3.70018363e-01
2.98022985e-01 -1.87842234e-03 -9.24669802e-02 -6.40320685e-03
3.99771899e-01 -7.17194006e-02 3.83729547e-01 -8.30785930e-01
5.24458706e-01 2.89547086e-01 3.94204229e-01 -3.23831677e-01
-1.05407648e-01 -2.45712340e-01 -2.03130111e-01 -3.40867221e-01
1.95038259e-01 -2.33993828e-01 -1.03990316e+00 -3.81219313e-02
-1.08268487e+00 -8.29127014e-01 -4.04265057e-03 2.54782826e-01
-8.04898262e-01 2.68137455e-01 -3.24469894e-01 -1.01218414e+00
2.52606571e-01 -1.30669081e+00 6.26935661e-01 -7.73524269e-02
-4.40718949e-01 -7.14303911e-01 3.08829516e-01 -4.70927209e-02
4.43211704e-01 5.30371904e-01 8.45978737e-01 -5.34092307e-01
-1.04728508e+00 9.50429812e-02 -1.52780697e-01 -9.84689653e-01
1.47178829e-01 -1.09337233e-01 2.54707664e-01 -4.25867945e-01
-6.82405174e-01 1.84718966e-01 1.39541656e-01 5.41400313e-01
2.55064249e-01 -7.94430375e-01 -1.09839070e+00 -1.17193066e-01
1.60242212e+00 8.41289878e-01 5.22360802e-01 1.14779305e+00
-2.91632414e-02 7.53077805e-01 1.04709482e+00 6.44069791e-01
8.86905193e-01 9.50787842e-01 7.40727723e-01 4.55790728e-01
2.21838325e-01 -5.98225696e-03 4.98904102e-02 5.58821075e-02
5.21485321e-02 -7.53062487e-01 -1.27157915e+00 9.83340263e-01
-2.35285687e+00 -8.94895971e-01 -1.00988159e-02 2.05068278e+00
4.84183252e-01 3.41330081e-01 7.31890202e-01 2.21611694e-01
8.17932487e-01 -8.78283754e-02 -2.45562464e-01 -6.06909096e-01
3.93564850e-01 -2.03099042e-01 8.98146152e-01 1.13866806e+00
-8.11662495e-01 8.88121307e-01 7.72388601e+00 2.93862820e-01
-6.23689711e-01 1.36347646e-02 -1.07362553e-01 -4.14383441e-01
-4.50371981e-01 4.18994159e-01 -3.76955509e-01 1.25984222e-01
8.10765803e-01 -4.24784809e-01 9.56269801e-01 6.31397665e-01
2.55206406e-01 -3.32713574e-01 -8.89253914e-01 2.65133232e-01
-4.21622187e-01 -1.53064322e+00 -3.46574187e-01 5.08614779e-01
7.89028406e-01 -1.07361950e-01 -4.44683135e-01 -9.20895040e-02
8.95020485e-01 -8.65347385e-01 1.04312861e+00 1.22182509e-02
1.79245308e-01 -1.02666497e+00 4.60257053e-01 3.51616144e-01
-1.19993055e+00 -4.86492217e-01 1.17009930e-01 -7.20977545e-01
9.39497232e-01 1.19328268e-01 -8.60216558e-01 8.22492063e-01
5.69799721e-01 3.51189151e-02 2.17290178e-01 1.43948698e+00
-2.16566950e-01 -5.45867756e-02 -7.17289567e-01 -3.42286885e-01
7.65191317e-01 2.85171792e-02 9.03466463e-01 8.24292958e-01
1.97808407e-02 5.17724037e-01 8.22562575e-01 2.88996398e-01
7.29727685e-01 -3.68886203e-01 -2.93542117e-01 8.02523494e-02
1.07329988e+00 8.24097931e-01 -1.22808421e+00 6.43586367e-02
-2.75999367e-01 1.95473060e-01 1.35916188e-01 2.59839654e-01
-8.21131706e-01 -2.56060809e-01 6.21803761e-01 2.27574422e-03
3.15760583e-01 -6.70594394e-01 -3.71694207e-01 -2.76647925e-01
3.88448883e-04 -7.47552574e-01 6.06811821e-01 -3.74045342e-01
-8.00951898e-01 7.42670774e-01 6.76107883e-01 -8.06014299e-01
-4.18215901e-01 -2.81924903e-01 -6.57866597e-01 3.09499860e-01
-1.54681945e+00 -8.90393436e-01 -2.74780113e-02 2.44306281e-01
6.16476476e-01 -8.99836943e-02 8.91543508e-01 7.47267678e-02
-4.36967105e-01 9.82533768e-02 -2.62829870e-01 -5.80571115e-01
-3.67841907e-02 -8.55493724e-01 4.65395033e-01 1.16344762e+00
-4.38506216e-01 5.19358635e-01 1.00056267e+00 -7.92699039e-01
-1.70774341e+00 -5.39868772e-01 6.99881554e-01 -1.09833412e-01
7.22807646e-01 1.60065413e-01 -1.25720605e-01 8.72042120e-01
2.18117595e-01 -5.97914159e-01 3.43026757e-01 4.34764981e-01
2.28543624e-01 8.25915188e-02 -1.19829106e+00 8.85909557e-01
1.35190880e+00 4.21750098e-01 -2.89118677e-01 4.73136425e-01
8.80922318e-01 -4.82431203e-01 -4.25846130e-01 6.24495327e-01
6.60142243e-01 -8.76004755e-01 8.86571229e-01 -4.16035831e-01
2.47187237e-03 -7.24939346e-01 -4.75496113e-01 -1.24777937e+00
-8.35745394e-01 -1.10924315e+00 1.79872200e-01 6.46973133e-01
6.67632043e-01 -9.23809767e-01 9.06791985e-01 7.95530379e-01
-5.13909996e-01 -8.70016396e-01 -1.07144165e+00 -1.10379899e+00
-3.37743491e-01 -3.50302488e-01 9.12851155e-01 6.71497285e-01
5.40290475e-01 7.30133504e-02 -2.38160089e-01 6.23106599e-01
6.95466936e-01 3.43351036e-01 6.68711007e-01 -8.86662483e-01
-1.89812347e-01 -7.81212807e-01 -2.09573824e-02 -7.50856102e-01
-2.05243118e-02 -3.59036893e-01 3.57991904e-01 -2.25393987e+00
-4.39986996e-02 -9.39009428e-01 -1.28895044e-01 8.63481045e-01
3.88266444e-01 -4.19554800e-01 4.20092911e-01 1.62810206e-01
-1.15084541e+00 2.90878862e-01 1.07089841e+00 -6.33879155e-02
-5.07105947e-01 -1.85591266e-01 -8.03025246e-01 2.85650194e-01
8.85541499e-01 -4.97369885e-01 -7.96365738e-01 -5.07062197e-01
3.79894197e-01 6.34803414e-01 7.20171779e-02 -7.08626628e-01
5.12956202e-01 -1.32333517e+00 -6.74893916e-01 -6.46703243e-01
2.80937940e-01 -8.80951405e-01 7.05839574e-01 8.41350377e-01
-2.82373667e-01 6.06480777e-01 2.67795742e-01 5.90021431e-01
2.46262982e-01 -3.08469981e-01 1.39803410e-01 -2.22517788e-01
-1.09722626e+00 1.20768949e-01 -7.48626113e-01 -3.36555809e-01
1.45831943e+00 -3.84212554e-01 -7.54777133e-01 -3.92172426e-01
-2.52489835e-01 8.29260826e-01 6.02092743e-01 1.40836477e-01
6.35109186e-01 -1.10918415e+00 -5.67981124e-01 -3.75483215e-01
-7.05105141e-02 -9.51297060e-02 6.04937002e-02 8.57997894e-01
-7.07244158e-01 7.63737082e-01 -5.38646936e-01 -4.94691171e-02
-9.90470886e-01 9.60393369e-01 5.72092459e-02 -4.97913837e-01
-5.91113091e-01 3.74357998e-01 -1.74222976e-01 -4.03769344e-01
2.21202940e-01 -4.51649576e-02 3.16676974e-01 -3.60613048e-01
4.85988617e-01 7.11735070e-01 -2.73806900e-01 -1.58680618e-01
-1.10657704e+00 3.08696985e-01 9.51699391e-02 -6.25233710e-01
1.41474938e+00 -3.88389945e-01 -3.41424048e-01 -3.42007786e-01
2.72976965e-01 8.15724880e-02 -1.04034996e+00 8.86266455e-02
2.49526054e-01 -7.32805550e-01 -5.35322092e-02 -1.19692898e+00
-7.45821059e-01 -2.26970464e-01 -3.20196450e-01 4.58859533e-01
1.11611032e+00 -6.58456236e-02 6.28958642e-01 5.06658018e-01
1.22735643e+00 -9.27900910e-01 -1.61690176e-01 5.82961202e-01
7.66331613e-01 -4.47728425e-01 1.99564651e-01 -7.83587456e-01
-5.61117768e-01 9.31133747e-01 5.46824038e-01 -8.38865042e-02
2.05277413e-01 5.47033191e-01 -2.95857966e-01 -2.30936289e-01
-1.09158468e+00 -4.42176431e-01 -5.10367632e-01 8.19125235e-01
-5.08928895e-01 2.37151012e-01 -6.95627689e-01 3.25645506e-01
-2.75278211e-01 -2.30659515e-01 9.82980430e-01 1.64707971e+00
-7.99936533e-01 -1.35842407e+00 -4.22277957e-01 5.69774657e-02
3.02239895e-01 2.77165860e-01 -3.96294117e-01 9.59028423e-01
-4.03819025e-01 1.43461263e+00 -1.23106726e-01 -2.67434031e-01
4.57879096e-01 -4.13753033e-01 8.14055383e-01 -2.99946934e-01
-2.90612489e-01 1.16499728e-02 1.02058387e+00 -6.88198447e-01
-5.06021082e-01 -8.74140084e-01 -1.41753066e+00 -5.79900205e-01
-1.90924823e-01 3.00869495e-01 3.53731990e-01 1.07335556e+00
3.69007498e-01 5.59493423e-01 5.52656174e-01 -8.57362330e-01
-2.74992049e-01 -2.31190786e-01 -1.34727895e-01 -3.34116608e-01
2.21898690e-01 -9.84021246e-01 1.49150670e-01 -8.24907184e-01] | [4.94943904876709, 1.8612700700759888] |
4b69c814-96bb-4453-ad93-7ac735d1ac2d | margin-based-few-shot-class-incremental | 2210.04524 | null | https://arxiv.org/abs/2210.04524v1 | https://arxiv.org/pdf/2210.04524v1.pdf | Margin-Based Few-Shot Class-Incremental Learning with Class-Level Overfitting Mitigation | Few-shot class-incremental learning (FSCIL) is designed to incrementally recognize novel classes with only few training samples after the (pre-)training on base classes with sufficient samples, which focuses on both base-class performance and novel-class generalization. A well known modification to the base-class training is to apply a margin to the base-class classification. However, a dilemma exists that we can hardly achieve both good base-class performance and novel-class generalization simultaneously by applying the margin during the base-class training, which is still under explored. In this paper, we study the cause of such dilemma for FSCIL. We first interpret this dilemma as a class-level overfitting (CO) problem from the aspect of pattern learning, and then find its cause lies in the easily-satisfied constraint of learning margin-based patterns. Based on the analysis, we propose a novel margin-based FSCIL method to mitigate the CO problem by providing the pattern learning process with extra constraint from the margin-based patterns themselves. Extensive experiments on CIFAR100, Caltech-USCD Birds-200-2011 (CUB200), and miniImageNet demonstrate that the proposed method effectively mitigates the CO problem and achieves state-of-the-art performance. | ['Ruixuan Li', 'Yuhua Li', 'Shanghang Zhang', 'Yixiong Zou'] | 2022-10-10 | null | null | null | null | ['few-shot-class-incremental-learning'] | ['methodology'] | [ 4.03945327e-01 5.18834405e-02 -2.42120221e-01 -5.87812662e-01
-5.50111175e-01 -1.93861827e-01 4.27941620e-01 1.23670772e-01
-3.97037029e-01 6.04145408e-01 -3.04271221e-01 -3.46365720e-01
-3.76179099e-01 -7.87952781e-01 -6.63925171e-01 -7.46585190e-01
-2.77987365e-02 1.30708262e-01 7.63538003e-01 -1.50843039e-01
1.58493951e-01 4.55285013e-01 -1.74445617e+00 6.24479413e-01
1.11495602e+00 1.15630352e+00 7.81444833e-02 3.38584363e-01
-1.01351418e-01 8.11379313e-01 -5.43783367e-01 -8.39788392e-02
4.14965361e-01 -6.56295657e-01 -5.91559410e-01 3.35403293e-01
4.34164912e-01 -1.10490404e-01 -1.35346785e-01 1.10530090e+00
3.29581887e-01 3.36362690e-01 6.07560754e-01 -1.41038644e+00
-6.90087199e-01 5.28678894e-01 -5.23612499e-01 4.22012627e-01
-2.75895670e-02 2.09875870e-02 7.01363564e-01 -1.26700509e+00
4.38493937e-01 9.13085997e-01 6.92371845e-01 7.66692877e-01
-9.10793960e-01 -6.91759408e-01 3.88928533e-01 5.13714850e-01
-1.49057698e+00 -3.52513939e-01 9.08838749e-01 -3.71027559e-01
7.89053202e-01 2.21363977e-01 5.28218567e-01 7.20280349e-01
3.61615270e-02 1.00255156e+00 1.18211544e+00 -5.95966756e-01
5.76166332e-01 3.85137141e-01 8.13403845e-01 6.02861464e-01
9.21583772e-02 2.82189578e-01 -2.79866248e-01 2.44494304e-02
4.48853970e-01 3.03820819e-01 -3.17820847e-01 -3.61205399e-01
-5.98706067e-01 5.70164979e-01 7.48519897e-01 4.22610134e-01
-1.37554556e-01 -5.37898183e-01 3.13043684e-01 4.35563028e-01
4.13878113e-01 2.47271255e-01 -4.38772470e-01 1.38600454e-01
-8.90793443e-01 2.75406297e-02 5.19379854e-01 1.01111400e+00
1.02801514e+00 1.89695567e-01 -2.69246370e-01 9.70150411e-01
-2.12467283e-01 2.86081254e-01 8.69173706e-01 -2.46944338e-01
2.68494815e-01 8.43302965e-01 -1.95771471e-01 -9.04592037e-01
-3.07200789e-01 -8.62155080e-01 -7.71379650e-01 3.74363601e-01
3.86721611e-01 -8.41401070e-02 -1.28154278e+00 1.67577088e+00
3.58108759e-01 4.52747345e-01 2.02870950e-01 6.07265592e-01
8.26070726e-01 6.66050017e-01 -8.01616982e-02 -6.07980788e-01
8.94823551e-01 -1.06241548e+00 -3.96966130e-01 -3.86486441e-01
6.09198034e-01 -1.88597292e-01 1.25502861e+00 4.11595851e-01
-5.26172996e-01 -8.66147101e-01 -1.30819988e+00 5.28454065e-01
-4.51790631e-01 -4.41771112e-02 5.47130823e-01 6.05610669e-01
-4.68581975e-01 6.82216406e-01 -5.89806199e-01 -2.26133227e-01
5.99572361e-01 3.65035050e-02 -2.43018374e-01 -1.31629243e-01
-9.74912524e-01 7.23118603e-01 6.92603648e-01 7.40495101e-02
-5.56242824e-01 -9.00605738e-01 -6.37696683e-01 2.16681883e-01
5.71865499e-01 6.91160187e-02 1.10865724e+00 -1.35827124e+00
-1.25575364e+00 5.67758977e-01 -1.83418393e-01 -5.75749218e-01
4.08840597e-01 -2.66078534e-03 -6.29103661e-01 -4.85313646e-02
-1.60428286e-01 4.38207895e-01 8.59655440e-01 -1.27962744e+00
-9.37049747e-01 -3.00375253e-01 -1.36896834e-01 1.28243238e-01
-5.66831708e-01 -4.89090502e-01 -9.17665809e-02 -5.45047998e-01
3.43475282e-01 -8.11560094e-01 -9.49367806e-02 -1.09817728e-01
-1.56438813e-01 -5.26101649e-01 1.04762876e+00 -1.89522937e-01
1.37632906e+00 -2.40058398e+00 -4.05434608e-01 1.00611225e-01
1.42689946e-03 7.79802382e-01 -1.62641197e-01 -2.65042502e-02
-3.20358604e-01 -1.86922684e-01 -3.03078830e-01 -4.21326719e-02
-4.08910990e-01 1.85620055e-01 -6.18393421e-01 1.74447343e-01
2.69561827e-01 7.96022475e-01 -1.07760227e+00 -1.82165995e-01
7.12388307e-02 -2.83611212e-02 -5.41768670e-01 9.71678495e-02
-1.32970780e-01 5.43348379e-02 -8.44482109e-02 7.46497691e-01
8.11758339e-01 -5.12190349e-02 -2.74623409e-02 -2.26465657e-01
-3.73884961e-02 -3.15690070e-01 -1.23890996e+00 1.26524329e+00
-6.16912469e-02 4.48837757e-01 -6.35934114e-01 -1.28674853e+00
1.09468329e+00 3.48109901e-02 2.34423041e-01 -7.18908012e-01
6.12851344e-02 3.54678392e-01 3.76356870e-01 -4.02525008e-01
1.65039733e-01 -4.59754050e-01 1.39102653e-01 1.63030297e-01
2.91697592e-01 3.71282250e-01 1.33725017e-01 -8.42043087e-02
8.61654162e-01 2.84581818e-03 5.54531753e-01 -3.07183474e-01
5.17217159e-01 -1.06661804e-02 9.84675229e-01 9.54556584e-01
-5.44454038e-01 4.73218858e-01 2.46461481e-01 -7.17307746e-01
-7.34601974e-01 -1.10556161e+00 -3.17065597e-01 9.57252443e-01
1.73370779e-01 -1.95869476e-01 -5.32005787e-01 -1.06181967e+00
-6.00008518e-02 8.15952063e-01 -8.33305240e-01 -6.51056588e-01
-4.95150626e-01 -9.49825704e-01 1.50594696e-01 7.04383790e-01
6.87834084e-01 -1.16510141e+00 -6.06961191e-01 3.42888653e-01
2.32977614e-01 -8.46483767e-01 -3.06555659e-01 4.60634410e-01
-9.51788008e-01 -1.27355087e+00 -6.88482940e-01 -1.00288308e+00
9.24360812e-01 5.12212574e-01 5.76377094e-01 1.18917167e-01
-3.93568367e-01 1.25002176e-01 -5.94446242e-01 -6.07869923e-01
-2.27582231e-01 -1.24648809e-02 2.15415463e-01 3.11177492e-01
5.18306792e-01 -6.03013337e-01 -4.66038644e-01 4.20494258e-01
-7.79167891e-01 -4.76824529e-02 7.29818165e-01 1.09022009e+00
5.21733284e-01 3.83937240e-01 1.21915543e+00 -1.03723514e+00
3.51452798e-01 -4.89705056e-01 -3.64009321e-01 5.61184645e-01
-9.22580898e-01 -2.92487830e-01 8.41035903e-01 -9.48563099e-01
-1.07998216e+00 1.91576257e-02 -7.14422241e-02 -5.88413537e-01
-8.99949074e-02 5.02318740e-01 -1.93418488e-01 -1.44219697e-01
9.81409609e-01 5.98374128e-01 -4.68441211e-02 -3.68719012e-01
2.72498369e-01 7.66683519e-01 6.14159584e-01 -3.41980278e-01
8.06285322e-01 4.61981952e-01 -1.75588951e-01 -9.32963312e-01
-1.25503588e+00 -4.77286398e-01 -6.41789615e-01 -2.40472853e-01
4.00074035e-01 -6.69914484e-01 -2.33606681e-01 6.59064889e-01
-6.13696456e-01 -4.09483165e-01 -6.68416321e-01 3.51856977e-01
-3.02573323e-01 2.81376302e-01 -2.97532678e-01 -9.35788095e-01
-3.99231106e-01 -8.10422480e-01 2.54806042e-01 6.69565201e-01
9.11773965e-02 -8.13999414e-01 -1.42603025e-01 -2.14154914e-01
4.30145353e-01 1.17428400e-01 9.24795985e-01 -9.17796195e-01
-2.70140797e-01 -2.99830526e-01 -2.47724742e-01 7.00500011e-01
4.13564891e-01 -3.84792835e-01 -1.14161849e+00 -5.27319610e-01
1.69460505e-01 -3.96123201e-01 1.02030635e+00 7.56320730e-02
1.22682536e+00 -2.18571186e-01 -2.49215990e-01 7.09433436e-01
1.55839109e+00 7.17265487e-01 6.95205927e-01 2.88553864e-01
4.76803690e-01 4.13922489e-01 8.01024020e-01 3.71606678e-01
1.59778092e-02 4.93907690e-01 4.36744504e-02 -1.34775089e-02
-3.22217792e-01 -2.83638984e-01 1.21239565e-01 6.06043875e-01
2.64277369e-01 8.60033482e-02 -8.02413702e-01 5.05513132e-01
-1.94811213e+00 -1.05688572e+00 2.19986320e-01 2.26662397e+00
8.79336357e-01 5.08411586e-01 -2.85967607e-02 5.12227118e-01
6.62780762e-01 -2.22624093e-01 -8.47492456e-01 -1.70219943e-01
-1.23630375e-01 1.14725836e-01 1.50876150e-01 4.46062237e-01
-1.15761769e+00 9.01383698e-01 5.80308867e+00 1.25826371e+00
-1.35901964e+00 -4.74000573e-02 8.37955058e-01 3.94897759e-02
3.47617231e-02 6.49765655e-02 -1.09240770e+00 3.99663985e-01
5.73966563e-01 -3.99494171e-01 3.03212315e-01 1.17559528e+00
-2.65072405e-01 -7.38940611e-02 -1.20620823e+00 1.11130846e+00
3.55226129e-01 -1.30758739e+00 1.43468782e-01 -3.75722796e-01
7.98524022e-01 -2.61866391e-01 -1.29288230e-02 1.00826645e+00
-3.50992084e-01 -5.25003493e-01 6.42596662e-01 4.81453717e-01
7.60148466e-01 -7.59643614e-01 5.48216820e-01 6.82339728e-01
-1.15815449e+00 -4.94583249e-01 -6.86058104e-01 -1.74928367e-01
-2.45789513e-01 7.39314735e-01 -9.30653751e-01 5.05956888e-01
5.62391281e-01 7.67090619e-01 -7.87962079e-01 1.35613286e+00
-1.47935003e-01 7.90848255e-01 -2.01234490e-01 -5.07706851e-02
1.73033997e-01 2.75223218e-02 4.97715980e-01 1.06755209e+00
2.25881815e-01 3.49424422e-01 3.76137823e-01 7.48365402e-01
1.13573983e-01 1.32552860e-02 -4.32027429e-01 3.27026397e-01
4.79746133e-01 9.56914544e-01 -9.50590849e-01 -5.83653152e-01
-3.22017193e-01 8.72383535e-01 5.00432670e-01 4.14646089e-01
-6.23563647e-01 -5.75153887e-01 2.20066786e-01 7.35177994e-02
4.14169341e-01 2.05655992e-01 -4.14968818e-01 -1.14651465e+00
2.11940065e-01 -7.89728045e-01 4.67363507e-01 -4.45705891e-01
-1.35778451e+00 7.33447433e-01 4.60727550e-02 -1.58473444e+00
4.29293625e-02 -4.25180405e-01 -8.93974245e-01 4.83412832e-01
-1.60638189e+00 -9.02063549e-01 -3.66214931e-01 5.22917867e-01
8.05664301e-01 -4.01256084e-01 6.98249698e-01 3.31992537e-01
-6.20411038e-01 1.08999026e+00 3.83428708e-02 7.43228570e-02
5.10236681e-01 -1.04940653e+00 5.45337237e-02 9.47044492e-01
1.21502668e-01 5.59231222e-01 4.40026909e-01 -6.19964778e-01
-1.02001679e+00 -1.37057853e+00 7.73448288e-01 -1.61606461e-01
3.63959610e-01 -3.37689936e-01 -1.28361309e+00 6.03381932e-01
-2.43225262e-01 3.61238211e-01 6.52316213e-01 5.15800603e-02
-4.68770891e-01 -5.25269866e-01 -1.20830154e+00 4.85563487e-01
1.03829038e+00 -2.68682450e-01 -9.87496495e-01 2.66004205e-01
8.33537698e-01 -1.78404063e-01 -3.66514146e-01 7.23737478e-01
5.13839841e-01 -9.29558754e-01 7.91850150e-01 -6.34903729e-01
1.27002716e-01 -3.14883351e-01 -2.41321504e-01 -1.35483468e+00
-5.69555640e-01 -1.55285284e-01 -1.88580751e-01 1.23296332e+00
4.62787688e-01 -7.46371865e-01 8.17868710e-01 3.04723650e-01
-2.95650512e-01 -1.16242278e+00 -9.39799488e-01 -1.34145534e+00
1.18701741e-01 -4.37115490e-01 4.29698825e-01 1.05031133e+00
2.88908761e-02 2.07634762e-01 -3.44466299e-01 6.52654395e-02
6.48677170e-01 2.55601257e-01 6.57181740e-01 -1.15068197e+00
-4.52867061e-01 -3.39239389e-01 -4.64079648e-01 -9.31581497e-01
-1.35110840e-01 -8.66209388e-01 2.72364914e-01 -1.11382043e+00
4.02900934e-01 -7.16713011e-01 -7.35604048e-01 7.67845690e-01
-5.21006227e-01 7.98161104e-02 4.20076579e-01 3.13535899e-01
-7.66152084e-01 5.79997241e-01 1.03141272e+00 -4.08050790e-02
-4.32967156e-01 1.22590683e-01 -7.35367000e-01 6.41056776e-01
6.44717157e-01 -4.68869507e-01 -7.77756989e-01 8.99146404e-03
-1.37662277e-01 -2.54464984e-01 5.24705872e-02 -1.43798077e+00
3.85958672e-01 -2.33678490e-01 4.73199606e-01 -6.91324532e-01
2.57492326e-02 -8.12427819e-01 -3.04710984e-01 6.96443081e-01
-2.77788430e-01 -5.18978536e-01 3.20470572e-01 7.44634986e-01
-1.84695482e-01 -3.97818148e-01 1.09222746e+00 -3.14176641e-03
-1.19351220e+00 4.36349511e-01 -1.31970271e-01 3.82476412e-02
1.28718686e+00 -4.65362430e-01 -3.94429207e-01 -1.78567413e-02
-8.96620154e-01 2.42858946e-01 1.45892054e-01 5.31514287e-01
8.51058125e-01 -1.48545980e+00 -6.38727963e-01 6.56057060e-01
3.98743451e-01 -5.34480810e-02 4.29565370e-01 5.86216152e-01
-5.50868697e-02 9.54206437e-02 -1.85424134e-01 -6.76865816e-01
-1.05558848e+00 7.81406164e-01 3.64985764e-01 -1.38018683e-01
-7.76739657e-01 1.04269826e+00 2.64324367e-01 -4.21939701e-01
4.45985436e-01 -2.36347377e-01 -2.01476216e-01 1.01542331e-01
9.62294281e-01 2.34036118e-01 1.91987276e-01 -6.45558015e-02
-3.47603172e-01 4.04941350e-01 -3.10277104e-01 2.65533119e-01
1.24085057e+00 -3.66004780e-02 2.60007590e-01 8.95837784e-01
1.15209949e+00 -4.02222961e-01 -1.40263605e+00 -6.81392908e-01
1.37558833e-01 -4.02709812e-01 -2.70312846e-01 -9.25764620e-01
-8.67570519e-01 7.97302485e-01 8.82195473e-01 5.51365912e-02
1.28965950e+00 -2.99986571e-01 7.23443627e-01 5.84678054e-01
4.89032447e-01 -1.20197630e+00 2.92061687e-01 6.34565234e-01
9.04260576e-01 -1.30509639e+00 -8.59401822e-02 -3.66356939e-01
-5.54062486e-01 1.23507071e+00 8.97159100e-01 -3.52136582e-01
1.04560339e+00 -1.04808621e-01 -9.61945280e-02 1.67895630e-01
-8.04356635e-01 -1.46972820e-01 4.60750580e-01 5.49641073e-01
5.01949340e-02 5.94231077e-02 -2.29867488e-01 9.86699760e-01
2.21617403e-04 1.19757898e-01 3.73734564e-01 1.18519056e+00
-9.04032469e-01 -8.11476707e-01 -1.06015071e-01 5.93741298e-01
2.51507312e-02 -7.17064291e-02 -8.09167027e-02 6.67161226e-01
3.99995029e-01 8.76670718e-01 1.62269339e-01 -6.33290052e-01
5.09938061e-01 3.42588186e-01 4.18805897e-01 -8.41438532e-01
-1.82843864e-01 -2.01531082e-01 -3.49428803e-01 -2.81977594e-01
-1.12532943e-01 -4.29394394e-01 -1.28277326e+00 1.57704338e-01
-5.12793422e-01 2.69219548e-01 2.43450806e-01 1.15328765e+00
2.14360625e-01 5.97321570e-01 9.77251172e-01 -4.55305964e-01
-9.04112279e-01 -1.04409742e+00 -6.82900250e-01 5.81023812e-01
3.53955269e-01 -6.53142393e-01 -6.45820498e-01 -1.09546781e-01] | [9.831647872924805, 3.280817747116089] |
ee0d861d-03a0-45ca-b1ad-302510188a8a | robust-semi-supervised-anomaly-detection-via | 2303.03925 | null | https://arxiv.org/abs/2303.03925v1 | https://arxiv.org/pdf/2303.03925v1.pdf | Robust Semi-Supervised Anomaly Detection via Adversarially Learned Continuous Noise Corruption | Anomaly detection is the task of recognising novel samples which deviate significantly from pre-establishednormality. Abnormal classes are not present during training meaning that models must learn effective rep-resentations solely across normal class data samples. Deep Autoencoders (AE) have been widely used foranomaly detection tasks, but suffer from overfitting to a null identity function. To address this problem, weimplement a training scheme applied to a Denoising Autoencoder (DAE) which introduces an efficient methodof producing Adversarially Learned Continuous Noise (ALCN) to maximally globally corrupt the input priorto denoising. Prior methods have applied similar approaches of adversarial training to increase the robustnessof DAE, however they exhibit limitations such as slow inference speed reducing their real-world applicabilityor producing generalised obfuscation which is more trivial to denoise. We show through rigorous evaluationthat our ALCN method of regularisation during training improves AUC performance during inference whileremaining efficient over both classical, leave-one-out novelty detection tasks with the variations-: 9 (normal)vs. 1 (abnormal) & 1 (normal) vs. 9 (abnormal); MNIST - AUCavg: 0.890 & 0.989, CIFAR-10 - AUCavg: 0.670& 0.742, in addition to challenging real-world anomaly detection tasks: industrial inspection (MVTEC-AD -AUCavg: 0.780) and plant disease detection (Plant Village - AUC: 0.770) when compared to prior approaches. | ['Toby P Breckon', 'Yona Falinie A Gaus', 'Neelanjan Bhowmik', 'Jack W Barker'] | 2023-03-02 | null | null | null | null | ['supervised-anomaly-detection', 'semi-supervised-anomaly-detection'] | ['computer-vision', 'computer-vision'] | [ 3.75508726e-01 1.07932463e-01 4.27685916e-01 -1.09742023e-01
-6.39659882e-01 -8.01142156e-01 8.46092403e-01 1.53454095e-01
-3.32521379e-01 7.03136265e-01 -2.82944918e-01 -4.59696293e-01
-1.60830006e-01 -8.76023054e-01 -9.76239741e-01 -9.73302782e-01
-3.19289416e-01 1.04631722e-01 1.04404669e-02 -2.30704889e-01
2.52586845e-02 6.86479807e-01 -1.61185491e+00 3.36201817e-01
8.33349466e-01 1.08450472e+00 -6.85957015e-01 1.04466319e+00
3.66969943e-01 6.93929434e-01 -1.18594265e+00 -4.26512361e-01
5.50885677e-01 -5.69173336e-01 -5.92293382e-01 1.19843662e-01
6.58462644e-01 -4.13880050e-01 -1.55859977e-01 1.32507038e+00
4.30656314e-01 2.36158058e-01 9.31456685e-01 -1.36892450e+00
-8.01931858e-01 1.11825190e-01 -4.85410511e-01 4.69076157e-01
1.57418430e-01 3.52050960e-01 5.94528317e-01 -6.50502920e-01
2.88794726e-01 1.02842116e+00 1.00732648e+00 6.23058081e-01
-1.65405977e+00 -4.56871092e-01 -1.61849752e-01 9.77336392e-02
-1.12933087e+00 -2.78114825e-01 3.97283226e-01 -3.10247838e-01
9.82254386e-01 6.72324002e-01 2.56001770e-01 1.64253509e+00
5.89786708e-01 5.13257265e-01 1.02532876e+00 -1.50155231e-01
4.95155722e-01 -1.79000542e-01 -2.66795576e-01 3.15931380e-01
3.46344262e-01 4.07560796e-01 -3.41291912e-02 -3.36256474e-01
8.31521809e-01 1.93399146e-01 -1.53212443e-01 2.59611845e-01
-8.77483845e-01 8.24735641e-01 4.50028777e-01 3.96871835e-01
-8.17905843e-01 1.28484383e-01 7.62623906e-01 1.07606125e+00
3.06733370e-01 7.83315003e-01 -4.95376170e-01 4.68952470e-02
-7.77058721e-01 4.09534007e-01 6.76822126e-01 6.22294068e-01
4.36474353e-01 9.03596878e-01 -1.30543992e-01 7.43351996e-01
-1.09422050e-01 4.83420640e-01 6.83272898e-01 -9.20531273e-01
-4.13760990e-02 2.61512756e-01 -1.31213501e-01 -9.56747591e-01
-1.21265769e-01 -5.50754666e-01 -1.33557940e+00 7.19735205e-01
5.19971490e-01 -3.78829658e-01 -1.41721654e+00 1.41552424e+00
2.00129226e-01 3.53647918e-01 2.97613055e-01 4.23500359e-01
4.58683729e-01 6.35490656e-01 1.08307816e-01 -7.99663886e-02
1.08005345e+00 -3.15334201e-01 -6.46915555e-01 -1.16111085e-01
4.72679496e-01 -7.32813179e-01 8.90345514e-01 1.01620722e+00
-8.51519406e-01 -4.25056189e-01 -1.04591238e+00 4.59625989e-01
-7.32132673e-01 -3.62855285e-01 3.52705210e-01 7.71150410e-01
-1.00982368e+00 8.74887168e-01 -8.44007015e-01 -1.80296153e-01
6.25589907e-01 1.95898712e-01 -6.91177607e-01 -1.72871854e-02
-1.02492118e+00 9.81233001e-01 5.27502894e-01 7.74041712e-02
-1.35879683e+00 -8.77528310e-01 -7.35603750e-01 -1.78926349e-01
3.41028929e-01 -4.42603439e-01 1.12623012e+00 -1.22154176e+00
-1.19332337e+00 7.02389002e-01 4.31497782e-01 -9.83115196e-01
6.19198561e-01 -4.27537382e-01 -1.01518297e+00 1.00582935e-01
-6.33059889e-02 3.28583837e-01 1.15214741e+00 -1.12246001e+00
-4.94728029e-01 -3.75372320e-01 8.77278764e-03 -2.96828389e-01
-2.08556816e-01 -7.20711797e-02 5.27943730e-01 -1.24006188e+00
6.46723583e-02 -6.00653410e-01 -3.06337744e-01 6.93771988e-02
-6.74067497e-01 -4.79361005e-02 1.19253135e+00 -9.10916269e-01
8.74221087e-01 -2.15544319e+00 -2.14850128e-01 3.56225014e-01
1.07715905e-01 7.03422725e-01 -1.05670631e-01 3.83594394e-01
-5.65611780e-01 -6.77028224e-02 -7.25248933e-01 9.78388190e-02
-3.15806754e-02 5.96578121e-01 -4.74630207e-01 5.84885180e-01
6.70667827e-01 7.50500023e-01 -9.50677574e-01 1.30474567e-01
2.46078208e-01 3.99297863e-01 -5.60968161e-01 2.85202116e-01
-1.81556553e-01 1.08075172e-01 4.38271984e-02 9.29497600e-01
5.29745042e-01 3.87755722e-01 -4.90172714e-01 1.08039550e-01
3.32611740e-01 -4.22038645e-01 -1.25070870e+00 1.15385115e+00
-2.23399207e-01 5.56959152e-01 1.75863907e-01 -1.28608239e+00
9.05869246e-01 4.75377977e-01 2.91378617e-01 -1.74091086e-01
1.37536705e-01 3.03196996e-01 2.31433928e-01 -4.01466668e-01
1.00198209e-01 -3.53711903e-01 2.30138674e-02 4.63544857e-03
3.83220702e-01 -1.11956418e-01 1.51584446e-02 -1.18561119e-01
1.64987886e+00 -2.28325546e-01 3.59950155e-01 -3.02426189e-01
4.22570586e-01 -2.00934932e-01 4.64068055e-01 1.06664968e+00
-2.04886302e-01 6.47881925e-01 5.66243827e-01 -5.57647109e-01
-1.02891648e+00 -1.48044026e+00 -3.51520360e-01 7.05310941e-01
-6.19105458e-01 2.40996517e-02 -7.34731197e-01 -9.70248461e-01
1.95997909e-01 1.02272570e+00 -8.19443941e-01 -6.14540458e-01
-2.97780365e-01 -1.06126153e+00 1.14846313e+00 7.38638222e-01
5.01181483e-01 -1.23315752e+00 -5.16956329e-01 2.50408918e-01
3.26254457e-01 -8.53290021e-01 -1.30089531e-02 5.84769726e-01
-8.08971226e-01 -9.64528561e-01 -6.34930134e-01 -4.66948479e-01
7.05360830e-01 -5.52141070e-01 1.07739770e+00 -1.33934483e-01
-5.72393715e-01 3.90045524e-01 -2.85491109e-01 -8.03914249e-01
-8.59896421e-01 -4.63231623e-01 3.30982715e-01 -3.09561919e-02
5.69858730e-01 -7.96690166e-01 -2.90894747e-01 6.69867769e-02
-1.39582574e+00 -9.01582539e-01 7.29931235e-01 1.09648275e+00
4.59601432e-01 2.78146267e-01 8.66853774e-01 -9.36456621e-01
7.31070578e-01 -6.65416181e-01 -5.09080648e-01 -2.72457749e-02
-6.41031384e-01 -1.75678313e-01 1.06797802e+00 -6.39869392e-01
-6.86812222e-01 -7.34043643e-02 -5.98884404e-01 -7.54619777e-01
-8.34433556e-01 3.27656806e-01 -7.04227611e-02 6.95905685e-02
1.27979636e+00 2.11920276e-01 6.60134181e-02 -3.53007823e-01
1.73764676e-01 4.24338609e-01 1.06821978e+00 -2.70334870e-01
1.15382361e+00 2.48821571e-01 1.46108717e-01 -1.01067436e+00
-5.10589480e-01 -2.12286696e-01 -3.67980331e-01 1.80541992e-01
6.04113042e-01 -7.48763442e-01 -4.34531093e-01 7.30850995e-01
-9.17433143e-01 -9.38276276e-02 -8.07115674e-01 1.91420898e-01
-3.36796463e-01 5.00973821e-01 -6.99592710e-01 -7.63595462e-01
-5.00830531e-01 -5.69724739e-01 6.77857995e-01 -1.69528782e-01
-2.96919018e-01 -9.77133751e-01 -7.27912709e-02 -3.88671160e-01
5.83139718e-01 1.08721471e+00 7.65718520e-01 -1.25189149e+00
7.21803531e-02 -7.40762711e-01 9.81963053e-02 1.18920553e+00
2.26443499e-01 9.65493557e-04 -1.38093746e+00 -3.52447331e-01
2.79935896e-01 -1.79077566e-01 7.25649357e-01 1.35991976e-01
1.46864104e+00 -6.76949799e-01 2.07812309e-01 4.92885679e-01
1.33118129e+00 2.92944819e-01 9.80126739e-01 4.14613187e-01
4.28841472e-01 2.56779224e-01 2.04132885e-01 3.34405988e-01
-6.13043666e-01 1.58952668e-01 7.45206773e-01 -1.38123214e-01
1.80757299e-01 2.03528911e-01 6.96603358e-01 1.83381408e-01
-9.39158648e-02 -2.72156268e-01 -8.21298420e-01 6.12800181e-01
-1.35057425e+00 -1.20175064e+00 -2.08570883e-01 2.23747945e+00
7.51395464e-01 4.42575276e-01 1.62998885e-01 7.81124651e-01
5.24507344e-01 -1.39908761e-01 -6.22276068e-01 -8.35448921e-01
-1.61409155e-01 5.04487514e-01 3.67921352e-01 2.06163242e-01
-1.41368973e+00 4.01918769e-01 5.79219770e+00 6.66628301e-01
-9.06293273e-01 -5.66603839e-02 6.59969032e-01 1.01144396e-01
1.18096307e-01 -5.52337468e-01 -2.58467823e-01 5.52816391e-01
1.26877725e+00 9.45169628e-02 4.58821207e-01 9.62661803e-01
-2.20997006e-01 2.21725538e-01 -1.04928756e+00 7.46542990e-01
1.46411747e-01 -8.48726332e-01 -3.11028026e-02 -5.19548357e-02
9.00143266e-01 9.97593179e-02 1.46075726e-01 5.81464410e-01
5.13480246e-01 -1.15033114e+00 1.51331112e-01 5.26738226e-01
6.54350936e-01 -1.01839256e+00 1.11856830e+00 2.87463456e-01
-5.28518498e-01 -2.48906821e-01 -4.16852504e-01 1.50971234e-01
-2.97239900e-01 9.15354073e-01 -9.71992254e-01 6.18874550e-01
9.52521265e-01 4.82925981e-01 -5.41375458e-01 8.82280171e-01
-2.41858512e-01 8.34013879e-01 -4.20270145e-01 4.31241065e-01
3.88183206e-01 1.51344433e-01 8.33960295e-01 1.24198294e+00
3.74614030e-01 -2.62247711e-01 4.84581366e-02 7.93767035e-01
2.91595273e-02 -3.31526399e-01 -8.84111404e-01 -1.17360570e-01
1.78827211e-01 9.40277398e-01 -4.88165051e-01 -2.19154596e-01
-1.66442290e-01 1.27464843e+00 -3.48854840e-01 4.40145433e-01
-7.52610326e-01 -9.64899480e-01 7.73964643e-01 6.89990520e-02
4.64863598e-01 3.45839083e-01 -1.18184365e-01 -9.74885285e-01
8.73522237e-02 -1.31549561e+00 6.97974384e-01 -4.39314336e-01
-1.81809175e+00 7.06091404e-01 4.79394337e-03 -1.27886307e+00
-5.66958189e-01 -8.82256567e-01 -9.42735314e-01 8.12387764e-01
-9.60953176e-01 -8.97765458e-01 -7.43224174e-02 5.05032301e-01
4.54792321e-01 -5.02814412e-01 1.17282367e+00 1.89174339e-01
-4.07860070e-01 8.45710099e-01 2.87057668e-01 2.69440293e-01
5.04326522e-01 -1.67210364e+00 5.52644253e-01 1.36807561e+00
1.71470679e-02 5.24363220e-01 9.45454597e-01 -6.51998043e-01
-8.93770278e-01 -1.45121491e+00 4.90163267e-01 -6.84455872e-01
6.37276053e-01 -1.50966063e-01 -1.45486009e+00 7.29850829e-01
1.06832035e-01 4.82058406e-01 4.15275604e-01 -9.66977775e-02
-4.57124323e-01 -2.86760833e-02 -1.68481743e+00 5.67950666e-01
6.55137360e-01 -5.38809597e-01 -7.62576938e-01 3.09078366e-01
4.24756795e-01 -4.57334667e-01 -1.16447008e+00 5.25013566e-01
2.47636795e-01 -9.18526649e-01 1.15774250e+00 -7.91690767e-01
4.48047191e-01 -2.84525305e-01 -1.49626181e-01 -1.35048616e+00
-6.33861199e-02 -6.86834276e-01 -6.25840127e-01 1.05645275e+00
1.14578471e-01 -9.25565243e-01 6.33597851e-01 1.90122247e-01
-4.39678937e-01 -6.93522990e-01 -1.04301405e+00 -1.10115421e+00
2.44547412e-01 -4.88347948e-01 5.08634984e-01 1.25583160e+00
-5.99360287e-01 -1.79988608e-01 -2.39652961e-01 5.49395621e-01
4.59593296e-01 -7.05611348e-01 6.29017174e-01 -1.26180899e+00
-3.65744233e-01 -2.97573447e-01 -1.02451992e+00 -2.57681727e-01
-1.48027375e-01 -6.83508039e-01 6.80729151e-02 -8.84758651e-01
-5.07680774e-01 1.80282280e-01 -6.86598003e-01 8.01269174e-01
-2.33584478e-01 3.67646843e-01 -3.19743842e-01 -2.57469535e-01
-1.00848742e-01 3.06831300e-01 6.66699529e-01 -8.53553191e-02
7.51706064e-02 2.63567477e-01 -5.25690794e-01 7.81465113e-01
9.99148786e-01 -6.87371492e-01 -4.42017794e-01 -4.67120074e-02
4.23038043e-02 -4.27523166e-01 9.89895165e-01 -1.27259696e+00
-2.12829828e-01 8.53797719e-02 9.73119199e-01 -4.06237006e-01
2.79309154e-02 -9.69060302e-01 -4.36149508e-04 7.65417814e-01
-1.66224182e-01 4.71211821e-01 4.75280136e-01 9.43106472e-01
-1.84177488e-01 -4.58810866e-01 7.99189806e-01 -1.89462379e-01
-6.20077133e-01 1.40767619e-01 -7.13492930e-01 5.43961413e-02
1.08992314e+00 -2.88867801e-01 -1.18887246e-01 -4.06873286e-01
-9.44631279e-01 -2.23901749e-01 2.41257355e-01 3.06583494e-01
6.22566223e-01 -1.17806280e+00 -8.66592169e-01 6.01091325e-01
-3.14145535e-02 1.93296432e-01 1.26142040e-01 6.63707554e-01
-5.51732957e-01 -1.96163118e-01 -3.34568769e-01 -5.53779721e-01
-1.07426488e+00 5.52073598e-01 5.60724616e-01 -4.41376045e-02
-7.54486084e-01 8.92347097e-01 -6.45634457e-02 -5.86506546e-01
2.59440124e-01 -3.75228494e-01 1.44428223e-01 -1.43917069e-01
6.10850155e-01 6.34786785e-01 5.54530382e-01 -2.41209880e-01
-1.71340510e-01 -1.66332096e-01 -2.05573916e-01 3.60952437e-01
1.43007040e+00 6.09361649e-01 -3.49300429e-02 3.69060457e-01
9.99106586e-01 -2.95753062e-01 -1.06630540e+00 4.80278954e-02
1.78128287e-01 -3.43641073e-01 4.78591062e-02 -9.73401308e-01
-7.90470898e-01 6.99886024e-01 1.05194867e+00 6.00543976e-01
1.35595345e+00 -4.94574755e-01 5.71113050e-01 6.84082150e-01
-1.78718165e-01 -9.99858081e-01 1.81021050e-01 4.27669942e-01
1.12523758e+00 -1.18484867e+00 4.15075533e-02 4.00607884e-02
-5.82188070e-01 1.13374829e+00 6.33795559e-01 -3.75657588e-01
5.40307462e-01 2.60516882e-01 1.78756416e-01 -2.35751599e-01
-4.97213542e-01 2.30164453e-01 3.51931125e-01 8.51636410e-01
1.22680828e-01 -1.14389233e-01 2.78068900e-01 3.97542983e-01
-1.85021847e-01 -3.99013549e-01 4.56760734e-01 1.10818565e+00
-1.29672289e-01 -7.66613722e-01 -6.31617010e-01 9.23919678e-01
-8.94248486e-01 1.25327464e-02 -3.15072656e-01 9.05567467e-01
1.81910262e-01 8.45719397e-01 8.47622454e-02 -2.37981811e-01
5.44958711e-01 3.00977468e-01 1.00773657e-02 -5.25478601e-01
-8.71042728e-01 -1.08436443e-01 -7.59806558e-02 -4.35532689e-01
-1.15549192e-02 -6.89191580e-01 -8.62039506e-01 -2.88021713e-01
-4.57066685e-01 -7.21594170e-02 3.29852670e-01 8.42636347e-01
2.45458931e-01 7.91557491e-01 5.01583934e-01 -4.53125507e-01
-1.02816689e+00 -1.05024266e+00 -6.54883623e-01 7.55324960e-01
6.99563622e-01 -2.91539103e-01 -7.24249661e-01 8.32320377e-02] | [7.632523536682129, 2.35868501663208] |
fd0923f3-ff24-4e35-ab4b-225d8886ef9e | bisenet-bilateral-segmentation-network-for | 1808.00897 | null | http://arxiv.org/abs/1808.00897v1 | http://arxiv.org/pdf/1808.00897v1.pdf | BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation | Semantic segmentation requires both rich spatial information and sizeable
receptive field. However, modern approaches usually compromise spatial
resolution to achieve real-time inference speed, which leads to poor
performance. In this paper, we address this dilemma with a novel Bilateral
Segmentation Network (BiSeNet). We first design a Spatial Path with a small
stride to preserve the spatial information and generate high-resolution
features. Meanwhile, a Context Path with a fast downsampling strategy is
employed to obtain sufficient receptive field. On top of the two paths, we
introduce a new Feature Fusion Module to combine features efficiently. The
proposed architecture makes a right balance between the speed and segmentation
performance on Cityscapes, CamVid, and COCO-Stuff datasets. Specifically, for a
2048x1024 input, we achieve 68.4% Mean IOU on the Cityscapes test dataset with
speed of 105 FPS on one NVIDIA Titan XP card, which is significantly faster
than the existing methods with comparable performance. | ['Changxin Gao', 'Jingbo Wang', 'Changqian Yu', 'Nong Sang', 'Gang Yu', 'Chao Peng'] | 2018-08-02 | bisenet-bilateral-segmentation-network-for-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Changqian_Yu_BiSeNet_Bilateral_Segmentation_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Changqian_Yu_BiSeNet_Bilateral_Segmentation_ECCV_2018_paper.pdf | eccv-2018-9 | ['thermal-image-segmentation', 'dichotomous-image-segmentation'] | ['computer-vision', 'computer-vision'] | [ 8.49010199e-02 -4.93031293e-01 1.05442330e-01 -4.25974756e-01
-4.83146757e-01 -4.63340789e-01 2.44080007e-01 7.15889037e-02
-8.08056891e-01 5.81685066e-01 -8.18955153e-02 -2.40728423e-01
6.32861331e-02 -1.00616741e+00 -5.36566794e-01 -6.76913738e-01
3.43730748e-01 -1.37465611e-01 8.81380618e-01 1.14265725e-01
4.19664830e-01 3.96478921e-01 -1.70314622e+00 2.05699623e-01
1.19514334e+00 1.27323246e+00 5.98831832e-01 4.33247268e-01
-1.86318576e-01 3.64755750e-01 -4.58796144e-01 -4.44638170e-03
4.00732547e-01 -5.42382635e-02 -7.74140894e-01 -1.68748319e-01
3.24669212e-01 -3.55895549e-01 -1.92795351e-01 1.09291661e+00
5.88063240e-01 3.34038764e-01 3.56137872e-01 -8.67078841e-01
-1.69484988e-01 3.72150481e-01 -1.03192139e+00 4.28232938e-01
-9.33834091e-02 9.64169130e-02 7.29961574e-01 -5.94022095e-01
4.65813398e-01 1.17551696e+00 4.63528395e-01 2.58781135e-01
-1.03590512e+00 -7.89457083e-01 2.25690722e-01 5.91400005e-02
-1.57571304e+00 -1.55204028e-01 2.45030299e-01 -3.36787775e-02
8.11880946e-01 2.84467012e-01 7.53435254e-01 5.56817114e-01
1.43967360e-01 7.90664017e-01 1.11857200e+00 -1.51076028e-02
2.80977964e-01 -1.27439171e-01 3.69730294e-02 6.22541308e-01
3.40228170e-01 -2.71912456e-01 -3.89063925e-01 2.38958552e-01
1.24045396e+00 2.12001890e-01 -1.01004906e-01 8.44463110e-02
-1.14926684e+00 5.86016357e-01 8.11114192e-01 4.12781447e-01
-2.59715945e-01 5.52319400e-02 3.68223757e-01 -9.15799588e-02
2.35915016e-02 1.60251260e-01 -3.42026711e-01 -8.72159377e-02
-1.00500715e+00 3.00514519e-01 2.74553925e-01 7.77906120e-01
7.94087112e-01 -2.43153684e-02 -9.12575796e-02 7.59394109e-01
1.34344697e-01 6.33583188e-01 5.54067969e-01 -1.03558588e+00
5.68769634e-01 5.64746976e-01 1.50891149e-03 -1.25192368e+00
-4.94858861e-01 -3.80704135e-01 -1.07227409e+00 2.31309086e-01
5.00167668e-01 -1.21242151e-01 -1.13335383e+00 1.48443377e+00
5.04613221e-01 2.42338702e-01 -1.04397006e-01 1.21861815e+00
7.58943498e-01 9.08485055e-01 1.04365297e-01 1.65726766e-01
1.65283132e+00 -1.08294141e+00 -3.71421963e-01 -2.64260113e-01
3.70325118e-01 -6.83478117e-01 1.13963854e+00 5.16562223e-01
-1.02239549e+00 -8.49107563e-01 -1.24364913e+00 -3.73508841e-01
-3.03988278e-01 1.62649766e-01 6.84797347e-01 4.03597146e-01
-8.36728096e-01 5.29414713e-01 -9.66385424e-01 -3.20347339e-01
4.70188886e-01 4.64408726e-01 -1.77596018e-01 -6.87370226e-02
-9.32800472e-01 3.82165849e-01 4.52596098e-01 4.81507704e-02
-2.09400162e-01 -4.97488976e-01 -5.87577164e-01 1.80193320e-01
2.65737325e-01 -4.75891829e-01 9.39199269e-01 -7.96647727e-01
-1.43578935e+00 4.32768643e-01 -8.04864317e-02 -3.62068772e-01
4.65965152e-01 -9.86017063e-02 -2.94420242e-01 2.86582708e-01
2.47815713e-01 1.14714861e+00 7.21147656e-01 -7.71871388e-01
-1.15177929e+00 -5.16340554e-01 -9.99098271e-02 3.71239722e-01
-3.41760635e-01 -2.36116812e-01 -7.75597394e-01 -6.44241810e-01
4.96878326e-01 -6.92313254e-01 -4.18571979e-01 -2.45081913e-02
-2.57434517e-01 -1.39956810e-02 7.82012284e-01 -4.80408996e-01
1.19301212e+00 -2.48317862e+00 -2.14625254e-01 3.17630291e-01
1.37369528e-01 6.35986328e-01 2.41196007e-02 -2.71005034e-01
3.39388222e-01 -2.00901385e-02 -2.73611903e-01 -7.72070838e-03
-2.34499961e-01 1.28840312e-01 -9.91519168e-02 2.89996058e-01
2.88852323e-02 7.21826136e-01 -6.38384461e-01 -6.89144015e-01
3.17301661e-01 5.54502070e-01 -7.10067272e-01 -1.41190350e-01
1.31319866e-01 4.27279621e-01 -7.58691728e-01 5.27141809e-01
1.21132684e+00 -2.07696661e-01 -4.52765776e-03 -1.31256372e-01
-4.94725466e-01 1.79784279e-02 -1.48343170e+00 1.88934445e+00
-3.66898417e-01 4.25199866e-01 3.10046196e-01 -7.43980587e-01
1.08734286e+00 -1.52629405e-01 2.32462183e-01 -1.26015306e+00
3.63336891e-01 3.85161668e-01 -2.44556755e-01 -2.07298473e-01
6.64641261e-01 7.82562196e-02 -1.17493056e-01 1.91665307e-01
-2.83228189e-01 -1.98052730e-02 1.50069296e-01 -1.80379421e-01
1.03857052e+00 3.22971717e-02 1.47544369e-01 -3.93811345e-01
4.96178329e-01 -2.89825890e-02 9.44706559e-01 5.36239028e-01
-2.86760539e-01 7.46580064e-01 3.82107437e-01 -5.57190835e-01
-8.39603662e-01 -1.12633872e+00 -3.23530942e-01 9.04997289e-01
6.34500682e-01 -3.56263876e-01 -9.63347852e-01 -3.85323793e-01
-1.60422221e-01 2.65749842e-01 -4.00797486e-01 2.63129443e-01
-6.72115445e-01 -6.55139506e-01 5.10603130e-01 8.63967299e-01
1.32275403e+00 -1.02901900e+00 -1.31039202e+00 2.39042267e-01
-9.41283777e-02 -1.05328143e+00 -5.33960760e-01 -1.22458607e-01
-1.05614555e+00 -7.86218524e-01 -6.31141841e-01 -9.38036859e-01
5.79133630e-01 5.37363529e-01 6.17214918e-01 -1.20491115e-02
-4.30198789e-01 -5.61427951e-01 -1.95770845e-01 -4.67047393e-02
4.64824140e-01 2.91371733e-01 -2.26172835e-01 -2.09214509e-01
2.67792076e-01 -4.86194432e-01 -1.06361961e+00 2.80621678e-01
-1.02694869e+00 4.22840595e-01 7.63859212e-01 7.15854466e-01
9.25448298e-01 2.64679253e-01 4.28875208e-01 -5.71789384e-01
2.23083869e-01 -3.09606940e-01 -8.55142415e-01 1.04573900e-02
-3.81487608e-01 -9.37009379e-02 8.75189483e-01 -2.43065834e-01
-1.25760913e+00 2.65666038e-01 -2.91593790e-01 -2.43515968e-01
-1.30160019e-01 1.73790023e-01 -1.15181953e-01 -3.09851244e-02
4.60065097e-01 3.14322710e-01 -2.54594423e-02 -4.57577765e-01
1.08366221e-01 8.11921179e-01 6.70028925e-01 -5.13215303e-01
3.84233952e-01 7.40521729e-01 -1.59910515e-01 -8.18584561e-01
-3.78330171e-01 -2.95649260e-01 -5.87433755e-01 1.92742378e-01
9.94592607e-01 -9.14242864e-01 -8.91129911e-01 5.91375172e-01
-8.51272821e-01 -3.14694613e-01 2.78517604e-02 5.48257113e-01
-2.27074936e-01 9.24903750e-02 -6.45950854e-01 -4.78008389e-01
-5.84335089e-01 -1.22198939e+00 1.00046313e+00 8.71894121e-01
1.13578178e-02 -4.79006499e-01 -2.99182415e-01 2.36971766e-01
5.29661655e-01 1.55295759e-01 5.34143090e-01 -2.54538516e-03
-7.00311363e-01 1.50832027e-01 -1.00121486e+00 1.56927168e-01
8.17117468e-02 -6.59868792e-02 -9.27202821e-01 -2.46229440e-01
-9.76576954e-02 -4.97824736e-02 9.42310631e-01 5.30176580e-01
1.46023107e+00 3.63445692e-02 -4.63765144e-01 8.72059524e-01
1.63709950e+00 3.42214257e-01 6.37288332e-01 3.05208534e-01
7.35425174e-01 4.56774652e-01 7.31192887e-01 4.90408063e-01
4.38900530e-01 4.38817024e-01 1.53938919e-01 -2.98437744e-01
5.24402447e-02 -1.80975109e-01 -1.91716403e-01 7.07352579e-01
-5.50647415e-02 -5.08945249e-02 -8.58525395e-01 5.25316358e-01
-1.81015170e+00 -6.59430146e-01 -3.09564956e-02 2.14953184e+00
5.81725895e-01 2.51087755e-01 -5.71476109e-02 1.47531807e-01
7.05951691e-01 2.03985319e-01 -7.21759021e-01 -3.34376127e-01
1.82981193e-02 2.93302625e-01 5.89849114e-01 3.29230964e-01
-1.20141041e+00 9.75516915e-01 5.80127811e+00 1.24611080e+00
-1.51394999e+00 -1.35507276e-02 8.07180047e-01 -2.08972186e-01
-1.16793960e-01 -2.83830285e-01 -8.81531298e-01 6.82771921e-01
6.49187148e-01 1.40632063e-01 2.42864132e-01 7.40644038e-01
9.20661911e-02 -4.03007239e-01 -3.46433580e-01 1.11936080e+00
-2.10271612e-01 -1.26139355e+00 -2.65992284e-01 6.11219555e-02
6.39762282e-01 4.20056023e-02 7.91587979e-02 1.87335983e-01
1.03751071e-01 -1.03993165e+00 5.28290927e-01 1.29235640e-01
8.44876111e-01 -9.82751012e-01 6.85013413e-01 3.19623411e-01
-1.44250727e+00 -5.36531843e-02 -5.60619056e-01 -4.62824181e-02
-1.94904301e-02 4.97792780e-01 -3.55876178e-01 4.51137453e-01
1.04930830e+00 5.40272236e-01 -4.70349312e-01 1.02945614e+00
3.97578143e-02 3.22193623e-01 -6.97759986e-01 -4.29898389e-02
6.14244819e-01 -1.58293724e-01 3.85714434e-02 1.15072274e+00
4.34229940e-01 2.77145714e-01 1.05980240e-01 6.55652583e-01
4.08042315e-03 1.69353545e-01 -1.83179557e-01 3.95855993e-01
6.17952406e-01 1.49241984e+00 -1.29216254e+00 -3.71372402e-01
-5.09989202e-01 9.77207303e-01 3.36380988e-01 1.60020187e-01
-1.06686080e+00 -7.69203901e-01 5.98060131e-01 -2.46306561e-04
5.73205769e-01 -2.01729327e-01 -6.85248554e-01 -1.09753931e+00
5.50687909e-02 -5.87555766e-01 2.35291243e-01 -4.93228316e-01
-7.96962142e-01 6.24602735e-01 -1.85436159e-01 -1.11190736e+00
1.85573980e-01 -4.98799562e-01 -5.29226959e-01 8.33284736e-01
-1.56628656e+00 -9.16016221e-01 -6.48372829e-01 5.98289311e-01
5.00502646e-01 3.05450708e-01 4.48140204e-01 5.14420867e-01
-7.49001265e-01 4.61431086e-01 1.54035002e-01 8.31493288e-02
4.94951546e-01 -1.10395300e+00 7.18036354e-01 7.84737647e-01
-1.92876756e-01 6.16050899e-01 2.42168307e-01 -4.85460728e-01
-1.03081679e+00 -1.17073691e+00 5.52080393e-01 2.78147787e-01
1.09962232e-01 -2.52645761e-01 -1.13574779e+00 3.16449493e-01
8.73290300e-02 2.79432178e-01 3.75278920e-01 -1.83762625e-01
-2.94901729e-01 -4.13193583e-01 -1.32302237e+00 7.48859406e-01
1.04046297e+00 -7.98462257e-02 -3.80348384e-01 -1.15801148e-01
4.58773613e-01 -5.44327259e-01 -6.10907853e-01 3.82695735e-01
7.53550529e-01 -1.14863896e+00 8.27678621e-01 9.15826410e-02
1.75258696e-01 -6.44553185e-01 -1.67373240e-01 -8.91459048e-01
-4.08354342e-01 -3.70752454e-01 5.67967176e-01 1.07016361e+00
1.74167916e-01 -7.98534334e-01 9.06460524e-01 3.37085187e-01
-2.43409574e-02 -1.01029766e+00 -9.88546789e-01 -6.04571223e-01
9.03502703e-02 -2.21818879e-01 8.69899511e-01 5.41771710e-01
-1.97518617e-01 2.44740382e-01 1.18136175e-01 3.45813446e-02
4.77884293e-01 3.78414959e-01 5.73134005e-01 -1.06154215e+00
6.60116179e-03 -5.10072410e-01 -4.28650647e-01 -1.34927666e+00
-2.49499246e-01 -3.75195891e-01 4.74517420e-02 -1.45403218e+00
1.82138339e-01 -7.93992400e-01 -3.13874722e-01 4.63179469e-01
-1.76560163e-01 5.84734738e-01 1.08289853e-01 8.65427852e-02
-6.08025670e-01 3.70136172e-01 1.50412703e+00 1.73102155e-01
-2.77729362e-01 -2.84124643e-01 -6.67041361e-01 8.01699758e-01
9.78694499e-01 -1.00749843e-01 -5.68397462e-01 -8.55832696e-01
-7.28668049e-02 -5.32737523e-02 1.56352445e-01 -1.37629676e+00
3.24889630e-01 -1.60135075e-01 6.79693103e-01 -7.69033909e-01
3.24116796e-01 -6.37092829e-01 -2.55835466e-02 6.41253412e-01
8.32164735e-02 1.07315294e-01 4.46562141e-01 3.45743358e-01
-3.52128714e-01 5.57114929e-02 9.62635040e-01 -4.50761952e-02
-9.98169184e-01 2.96661615e-01 -3.40025902e-01 -3.40223797e-02
1.29412556e+00 -4.27273810e-01 -4.03646737e-01 1.92673773e-01
-2.60622591e-01 4.17202532e-01 5.21289587e-01 2.90680736e-01
6.52772963e-01 -1.15711427e+00 -4.32862848e-01 6.78948700e-01
-3.47033083e-01 4.57917541e-01 7.27121532e-01 7.79318035e-01
-1.06970596e+00 4.83879507e-01 -6.50035322e-01 -5.48284888e-01
-1.16099441e+00 3.31328899e-01 4.58483398e-02 -8.12926441e-02
-7.76361406e-01 9.33130443e-01 3.00934970e-01 -1.65917993e-01
-8.49229321e-02 -5.53161383e-01 -3.14455003e-01 1.95421278e-01
7.23706365e-01 4.87483531e-01 4.80250493e-02 -5.35314023e-01
-5.00703514e-01 9.33899283e-01 -1.28082946e-01 -2.03058645e-01
9.44195569e-01 -2.25945234e-01 1.16960287e-01 5.81431156e-03
1.11736238e+00 -2.06815433e-02 -1.50232172e+00 -1.41470850e-01
-4.11461413e-01 -7.54285395e-01 3.06612790e-01 -4.41217035e-01
-1.28290331e+00 1.05762815e+00 7.34427392e-01 1.14331029e-01
1.36633670e+00 -3.75273526e-01 1.31289625e+00 3.42900306e-02
4.46613669e-01 -1.27041030e+00 -1.37975961e-01 4.34345156e-01
3.74244422e-01 -1.26525712e+00 -4.22564819e-02 -5.78629732e-01
-6.74061179e-01 9.21170354e-01 9.53389645e-01 -3.72910887e-01
4.85488355e-01 4.04058129e-01 2.77947150e-02 8.42444077e-02
-3.59710097e-01 -3.06569219e-01 4.50738445e-02 2.81950206e-01
1.70264170e-01 1.49572968e-01 -3.74292523e-01 5.27818561e-01
-2.56189972e-01 -3.46644707e-02 6.15468398e-02 1.00464380e+00
-6.53387964e-01 -7.37497330e-01 -3.14801633e-01 4.47908759e-01
-5.75535893e-01 -2.91271675e-02 3.23977530e-01 7.16619074e-01
3.49470526e-01 8.80983293e-01 7.02093542e-01 -4.91810173e-01
3.31668407e-01 -4.54565972e-01 2.80462950e-01 -3.89649451e-01
-5.88008165e-01 2.71691352e-01 -3.83330315e-01 -8.07568014e-01
-2.14256063e-01 -5.55636883e-01 -1.69528210e+00 -6.45047843e-01
-9.98920575e-02 1.70394536e-02 6.33213460e-01 7.11277902e-01
6.24334931e-01 5.93613327e-01 3.85731339e-01 -6.08196557e-01
-3.11609674e-02 -8.28674018e-01 -6.78726256e-01 7.42911696e-02
-4.64670686e-03 -5.66628456e-01 -6.56547472e-02 -1.64756477e-01] | [9.350322723388672, -0.5574501752853394] |
0be5f031-d131-4493-b254-92a6113c1a2d | kfnet-learning-temporal-camera-relocalization | 2003.10629 | null | https://arxiv.org/abs/2003.10629v1 | https://arxiv.org/pdf/2003.10629v1.pdf | KFNet: Learning Temporal Camera Relocalization using Kalman Filtering | Temporal camera relocalization estimates the pose with respect to each video frame in sequence, as opposed to one-shot relocalization which focuses on a still image. Even though the time dependency has been taken into account, current temporal relocalization methods still generally underperform the state-of-the-art one-shot approaches in terms of accuracy. In this work, we improve the temporal relocalization method by using a network architecture that incorporates Kalman filtering (KFNet) for online camera relocalization. In particular, KFNet extends the scene coordinate regression problem to the time domain in order to recursively establish 2D and 3D correspondences for the pose determination. The network architecture design and the loss formulation are based on Kalman filtering in the context of Bayesian learning. Extensive experiments on multiple relocalization benchmarks demonstrate the high accuracy of KFNet at the top of both one-shot and temporal relocalization approaches. Our codes are released at https://github.com/zlthinker/KFNet. | ['Long Quan', 'Lei Zhou', 'Yao Yao', 'Mingmin Zhen', 'Zixin Luo', 'Tianwei Shen', 'Tian Fang', 'Jiahui Zhang'] | 2020-03-24 | kfnet-learning-temporal-camera-relocalization-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Zhou_KFNet_Learning_Temporal_Camera_Relocalization_Using_Kalman_Filtering_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhou_KFNet_Learning_Temporal_Camera_Relocalization_Using_Kalman_Filtering_CVPR_2020_paper.pdf | cvpr-2020-6 | ['camera-relocalization'] | ['computer-vision'] | [-1.63147524e-01 -3.21087509e-01 -2.16684356e-01 -3.29599202e-01
-7.65132964e-01 -4.46527481e-01 5.30952871e-01 -1.69846028e-01
-6.10276341e-01 4.71648395e-01 1.09085172e-01 1.06901474e-01
-3.46479058e-01 -4.76911098e-01 -8.70351076e-01 -5.00715196e-01
2.51924515e-01 1.06291175e-01 5.08272886e-01 1.32786348e-01
1.65442973e-01 4.84330028e-01 -1.04827821e+00 -4.03510898e-01
4.40472424e-01 7.82438219e-01 1.97629511e-01 7.70887673e-01
5.17137229e-01 7.78445661e-01 3.29644233e-02 -1.51835963e-01
4.94822294e-01 -5.42634800e-02 -5.70114791e-01 2.01724187e-01
1.08958364e+00 -6.09329641e-01 -7.25933075e-01 1.17298460e+00
3.95087838e-01 5.49377799e-01 8.49695206e-02 -1.16159940e+00
-2.57988155e-01 3.50830287e-01 -7.03204513e-01 1.25686139e-01
6.26762986e-01 -3.03269401e-02 1.07775140e+00 -1.20997345e+00
7.73233831e-01 1.17488790e+00 1.04097128e+00 1.42368764e-01
-1.40891063e+00 -5.56576967e-01 4.79023010e-01 3.77578139e-01
-1.95028698e+00 -4.90125030e-01 5.59464514e-01 -6.77401125e-01
8.67112100e-01 -3.51937413e-01 7.60513186e-01 9.50553656e-01
2.75348425e-01 7.82991290e-01 7.01607287e-01 -2.33628437e-01
-4.92955036e-02 -4.63633984e-01 3.68740886e-01 7.51189590e-01
1.36201173e-01 1.86267152e-01 -8.59716117e-01 6.78596422e-02
1.00238919e+00 3.35463405e-01 -3.01669478e-01 -9.98894274e-01
-1.60166049e+00 5.29628992e-01 3.98826003e-01 -6.72711954e-02
-3.27887416e-01 6.57825768e-01 3.01511228e-01 7.47107938e-02
4.66811240e-01 2.61848629e-01 -3.91084671e-01 -2.74803460e-01
-1.15884531e+00 3.11195292e-02 6.14687443e-01 1.22828317e+00
1.22824442e+00 -5.09770885e-02 -6.38512373e-02 4.95728791e-01
2.92826593e-01 5.17779291e-01 1.73271835e-01 -1.29421496e+00
2.20621288e-01 2.46087447e-01 6.89082220e-02 -1.07559240e+00
-4.18331265e-01 -2.25227684e-01 -5.57611406e-01 -2.04408634e-02
3.64862502e-01 -1.82592615e-01 -6.95954084e-01 1.56921017e+00
4.36380833e-01 9.20108676e-01 -1.01125032e-01 1.14522779e+00
5.96241713e-01 5.46532273e-01 -3.65799457e-01 -9.35776606e-02
1.11462712e+00 -1.22618318e+00 -5.37373960e-01 -1.42135307e-01
5.37355304e-01 -7.92552650e-01 7.33390927e-01 5.93151152e-01
-8.36104453e-01 -6.75473213e-01 -1.01011765e+00 -1.26308128e-01
1.80046202e-03 3.29144478e-01 3.82587582e-01 3.27558815e-01
-1.21133399e+00 4.90589142e-01 -1.14557850e+00 -6.06475949e-01
-2.77117118e-02 3.17079246e-01 -6.56895280e-01 -2.30151996e-01
-9.33956563e-01 7.33556330e-01 4.90474761e-01 2.22904980e-01
-1.14662063e+00 -8.54168534e-01 -9.99730408e-01 -1.50435582e-01
9.35506821e-01 -7.47043550e-01 1.38240612e+00 -7.10292816e-01
-1.60915232e+00 3.62977982e-01 -1.66872352e-01 -6.04141772e-01
6.30371809e-01 -7.75562167e-01 7.99421445e-02 3.18731159e-01
1.26150146e-01 7.76081979e-01 8.89512122e-01 -8.99870336e-01
-8.41852784e-01 3.12469006e-02 4.25123692e-01 2.82359391e-01
-4.04671356e-02 -2.04970255e-01 -1.14258254e+00 -5.73740125e-01
3.60170305e-01 -1.29605019e+00 -2.80760705e-01 2.47351706e-01
-4.33934957e-01 -1.86164211e-02 9.07139719e-01 -5.25108933e-01
1.02964711e+00 -2.08901453e+00 4.41958964e-01 6.56931102e-02
3.04435372e-01 -8.10870752e-02 -5.68969063e-02 3.12286496e-01
-1.93091139e-01 -4.66877520e-01 1.24470986e-01 -7.06822991e-01
-1.55437574e-01 5.60376467e-03 -7.63371736e-02 1.06262481e+00
-3.38728905e-01 6.55210495e-01 -9.92163718e-01 -3.60447049e-01
7.08608747e-01 5.97473741e-01 -7.13236749e-01 -2.21825968e-02
-7.51128346e-02 4.81792480e-01 -2.13757664e-01 5.07012367e-01
5.86157858e-01 -1.94256559e-01 -3.87510099e-02 -6.31247580e-01
-3.68887752e-01 5.51575385e-02 -1.34842885e+00 2.51869416e+00
-3.29114318e-01 7.67706931e-01 2.74833515e-02 -5.51297665e-01
6.25663877e-01 3.79204959e-01 8.81629288e-01 -3.13215911e-01
5.04273549e-02 -1.46540970e-01 -5.09235144e-01 -8.75554904e-02
9.90466356e-01 1.42246678e-01 -1.13949284e-01 3.48051071e-01
3.03498626e-01 -8.95522013e-02 2.66016781e-01 5.42955756e-01
9.11320031e-01 7.04367340e-01 4.51245785e-01 -2.35801727e-01
5.10388315e-01 7.66805634e-02 9.66762841e-01 8.02711368e-01
-1.61864966e-01 7.57633090e-01 1.20005772e-01 -3.71317059e-01
-9.24778223e-01 -9.97802138e-01 1.51719898e-01 8.89535606e-01
7.07990050e-01 -8.15646827e-01 -6.63664401e-01 -6.06322169e-01
-5.27354702e-02 5.00781476e-01 -3.68870378e-01 -5.48727252e-02
-4.27125275e-01 -2.78404862e-01 4.31961119e-01 5.41442692e-01
4.87115562e-01 -2.94545770e-01 -6.02555692e-01 8.57658461e-02
-2.84444064e-01 -1.25977182e+00 -8.18103790e-01 -1.07683584e-01
-6.33577645e-01 -1.17550635e+00 -5.18831193e-01 -2.21414968e-01
5.94895720e-01 7.49873519e-01 7.18123674e-01 -3.31210494e-01
-2.18070194e-01 1.10563254e+00 -4.70084220e-01 -2.22898014e-02
1.27612561e-01 1.13945410e-01 3.35597843e-01 2.85992213e-02
2.70607501e-01 -5.06278396e-01 -6.63222313e-01 5.15872657e-01
-6.94193959e-01 8.84256512e-02 4.07699108e-01 6.65033042e-01
7.64440656e-01 -1.13728359e-01 -2.19919123e-02 -5.33393860e-01
5.44016771e-02 -2.78143853e-01 -1.02573454e+00 1.54442281e-01
-6.02442086e-01 -8.13309103e-02 4.69169281e-02 -4.63235259e-01
-1.08158851e+00 5.20712256e-01 2.73183793e-01 -9.80567336e-01
4.46232557e-02 4.58761722e-01 1.12883717e-01 -1.87385619e-01
4.00106877e-01 7.26087168e-02 -5.12737222e-02 -4.02097046e-01
5.09561300e-01 1.01743549e-01 5.92105508e-01 -5.58876932e-01
8.68557572e-01 7.45526850e-01 -1.14708934e-02 -6.14501595e-01
-1.02758050e+00 -9.29392755e-01 -1.15611720e+00 -6.39397502e-01
9.82645094e-01 -1.41638231e+00 -7.67536998e-01 4.29791719e-01
-1.20802844e+00 -3.23401451e-01 -2.28490517e-01 1.02176785e+00
-7.72487402e-01 5.26796997e-01 -4.27095771e-01 -3.60610753e-01
8.24394356e-03 -1.25660145e+00 1.16836286e+00 1.63435675e-02
-1.20512441e-01 -9.71060514e-01 3.26725066e-01 1.87706724e-01
-2.78278757e-02 -1.09881438e-01 -4.25764769e-02 -1.83363885e-01
-1.01846254e+00 -2.20736235e-01 -2.10464612e-01 3.18050116e-01
-2.72054281e-02 1.34412646e-01 -7.91292548e-01 -6.89776003e-01
-2.38801926e-01 -5.23048565e-02 9.53823686e-01 6.16843164e-01
5.91937602e-01 2.80396789e-01 -2.21696049e-01 8.54654670e-01
1.56620193e+00 -1.51721179e-01 4.23740655e-01 3.71896088e-01
1.04067373e+00 2.34197140e-01 1.01238394e+00 5.61463475e-01
4.41362917e-01 9.00076509e-01 6.19040608e-01 7.82923028e-02
-1.31995261e-01 -3.77654225e-01 5.36049008e-01 7.24515915e-01
4.06930447e-02 -2.47039735e-01 -9.36604738e-01 5.57039499e-01
-2.35706949e+00 -7.61332154e-01 -1.92479230e-02 2.19279432e+00
3.54475468e-01 -9.48438942e-02 -1.52514726e-01 -5.21237910e-01
9.13914800e-01 3.66312623e-01 -4.96507466e-01 3.15264493e-01
1.19183064e-01 -5.07569432e-01 1.04407811e+00 8.01063895e-01
-1.27276087e+00 1.27618587e+00 5.80405712e+00 7.01968610e-01
-1.05684268e+00 3.71057957e-01 2.15325747e-02 -2.15487719e-01
2.98755586e-01 4.76143688e-01 -1.17498028e+00 8.25736448e-02
6.48753822e-01 -1.31845057e-01 5.06691277e-01 8.77740562e-01
3.83857876e-01 -4.21251804e-01 -1.19670224e+00 1.20059049e+00
3.00122261e-01 -1.26266468e+00 -3.18361759e-01 -1.58240825e-01
7.58472681e-01 4.36733127e-01 -1.89040124e-01 2.70245373e-02
4.69302058e-01 -4.40465719e-01 1.08355415e+00 9.69531000e-01
6.00600660e-01 -6.18668795e-01 4.73952949e-01 2.40345865e-01
-1.48787773e+00 -3.40062962e-03 -3.58651966e-01 -1.82185657e-02
4.95532781e-01 6.81427300e-01 -8.19403112e-01 7.49408185e-01
9.27272916e-01 1.44175386e+00 -5.08940101e-01 1.06076646e+00
-4.34576988e-01 3.79550278e-01 -4.27533805e-01 5.29445946e-01
2.38006487e-01 -2.96103597e-01 8.78509820e-01 9.80054200e-01
4.44077551e-01 -8.83146077e-02 6.22579515e-01 5.05310357e-01
4.27425765e-02 -2.37392053e-01 -4.93812263e-01 3.77345592e-01
4.32504445e-01 1.17666924e+00 -6.00444794e-01 -2.84629166e-01
-7.61930883e-01 9.85270441e-01 1.56231433e-01 3.83493125e-01
-8.22768271e-01 -1.14262454e-01 6.90606833e-01 4.91437651e-02
5.22450984e-01 -6.69735551e-01 3.99674475e-01 -1.45481348e+00
3.19359731e-03 -4.77282822e-01 3.01466376e-01 -1.01004958e+00
-1.04764080e+00 1.82980239e-01 3.97619039e-01 -1.50496614e+00
-2.08719671e-01 -4.38040316e-01 -2.19781309e-01 3.75501335e-01
-1.37551177e+00 -1.24127007e+00 -5.13446093e-01 7.10156381e-01
7.40827203e-01 8.96084607e-02 2.77177423e-01 3.90037835e-01
-6.08864248e-01 1.52617350e-01 1.11895069e-01 1.39274731e-01
1.17736137e+00 -1.01632106e+00 2.08320141e-01 1.33886576e+00
1.96874216e-01 7.78144419e-01 6.04504347e-01 -6.46446943e-01
-1.65411901e+00 -1.22323680e+00 3.88045281e-01 -4.02926654e-01
9.89661157e-01 -3.40566933e-01 -4.73851472e-01 1.17938614e+00
2.94376295e-02 3.61194760e-01 2.15196982e-01 1.09761328e-01
-3.60322207e-01 -2.96842605e-01 -5.85797369e-01 5.75841904e-01
1.06928670e+00 -5.79545259e-01 -2.88889468e-01 2.01853275e-01
8.09769988e-01 -6.05085909e-01 -9.19161558e-01 2.51309872e-01
6.68675482e-01 -9.93900418e-01 1.14075935e+00 7.16577545e-02
1.90637656e-03 -7.84401536e-01 -3.04244757e-01 -1.01827633e+00
-3.88798863e-01 -6.93956256e-01 1.89654529e-03 1.04850698e+00
7.08484128e-02 -5.60718656e-01 7.67718971e-01 2.85866290e-01
-1.23888940e-01 -3.01765501e-01 -9.75710809e-01 -9.61537123e-01
-4.42938924e-01 -5.96288323e-01 1.53864220e-01 6.78126395e-01
-4.01774406e-01 2.69441307e-01 -8.05855811e-01 5.07611275e-01
8.15609396e-01 1.66621637e-02 1.10948765e+00 -1.09573829e+00
-3.59530091e-01 4.71889451e-02 -7.01118588e-01 -1.58527803e+00
3.13970357e-01 -6.66498482e-01 3.59748542e-01 -1.38714743e+00
3.05638760e-01 -2.14836091e-01 -2.88494945e-01 3.40391964e-01
1.42819941e-01 3.79111975e-01 3.36360484e-01 3.30606163e-01
-1.11780536e+00 5.00441432e-01 8.15311372e-01 -7.41453618e-02
-1.22784205e-01 -2.54720330e-01 -2.27964327e-01 9.71862435e-01
6.31281316e-01 -5.09854317e-01 -5.55760026e-01 -7.41221726e-01
3.61406118e-01 2.41197810e-01 7.90313184e-01 -1.21053457e+00
7.95687437e-01 -1.17333762e-01 4.47451882e-02 -8.79867971e-01
6.65936112e-01 -1.02044487e+00 4.32814568e-01 1.90037727e-01
-2.41679430e-01 1.61533028e-01 1.03310071e-01 1.11663425e+00
-2.38219425e-01 -9.34277773e-02 6.52986944e-01 6.08846880e-02
-1.23354733e+00 6.28054082e-01 -4.16698635e-01 -2.64625639e-01
1.10448015e+00 -4.33677882e-01 -9.21877623e-02 -4.77809697e-01
-8.53169203e-01 2.08070934e-01 7.72534430e-01 4.04311717e-01
6.44713044e-01 -1.16617119e+00 -4.40575927e-01 -8.50929841e-02
2.06395686e-01 1.68325454e-01 3.37766349e-01 1.47148204e+00
-5.79845130e-01 4.17342395e-01 9.56826210e-02 -1.05029845e+00
-1.42067778e+00 4.63784814e-01 3.71264398e-01 -7.68477917e-02
-8.14573944e-01 7.14512944e-01 4.25313652e-01 -4.48129594e-01
2.45397702e-01 -1.45571142e-01 2.26925910e-01 -5.69548234e-02
3.08078974e-01 6.28252387e-01 -5.09330854e-02 -7.77066529e-01
-3.02389324e-01 8.82393420e-01 -1.63617432e-01 -2.81973243e-01
1.20056057e+00 -6.73696876e-01 -1.92944154e-01 5.77226877e-01
1.10736406e+00 -2.88136825e-02 -1.83904064e+00 -6.23723745e-01
-5.79618402e-02 -5.99300444e-01 4.35393035e-01 -1.73722431e-01
-1.13196754e+00 5.05048037e-01 4.37710553e-01 -5.32592654e-01
9.03450012e-01 -1.22152567e-01 6.24522090e-01 8.48816812e-01
6.08497560e-01 -1.16391873e+00 2.43026808e-01 8.02248001e-01
4.86530095e-01 -1.05605364e+00 5.45119524e-01 -5.12102902e-01
-3.44797134e-01 1.14394259e+00 5.03754854e-01 -3.50747705e-01
8.64321649e-01 2.66397029e-01 -1.26377493e-01 -1.21925883e-01
-6.97392046e-01 -1.54050381e-03 9.94237512e-02 2.30994076e-01
1.24775916e-01 -1.84190303e-01 1.23312593e-01 8.80681574e-02
1.78779930e-01 3.00362147e-02 5.33209682e-01 9.92018282e-01
-2.33461753e-01 -8.78759086e-01 -4.61708874e-01 -5.44555262e-02
-6.47836030e-02 -6.74440786e-02 -1.44577876e-01 8.26615393e-01
-6.04533125e-03 7.97915041e-01 9.62096751e-02 -4.04584885e-01
2.56808996e-01 -3.58936757e-01 5.18696725e-01 -6.37619317e-01
-2.28919536e-01 3.38264644e-01 -1.24916658e-01 -1.10640883e+00
-7.22100616e-01 -9.23862517e-01 -1.08848429e+00 -2.88754284e-01
-6.77518547e-01 -9.74403322e-02 6.23742342e-01 8.29183996e-01
3.58850539e-01 4.52784479e-01 3.11043233e-01 -1.19345462e+00
-1.52611628e-01 -7.53153682e-01 -5.45576572e-01 -5.65831922e-02
4.48144376e-01 -9.15880799e-01 -2.43624866e-01 3.24210107e-01] | [8.090576171875, -2.1282448768615723] |
1ca1a9dd-0de6-4c77-8cb6-1f05b9a9963e | non-intrusive-load-monitoring-in-chaotic | 1801.05363 | null | http://arxiv.org/abs/1801.05363v1 | http://arxiv.org/pdf/1801.05363v1.pdf | Non Intrusive Load Monitoring in Chaotic Switching Networks | In this work, a non intrusive load disaggregation scheme is proposed. By
using a kernel based nonlinear regression strategy, the switching dynamic of an
electric network, simulated as a set of RLC circuits with chaotic switching, is
approximated using a time series of the total power consumption. The results
suggest that the employed methodology can be useful in the design of efficient
load disaggregation schemes. | [] | 2018-01-12 | null | null | null | null | ['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring'] | ['knowledge-base', 'miscellaneous', 'time-series'] | [-1.09471850e-01 -8.40697140e-02 4.80698934e-03 3.02783726e-03
1.86496034e-01 -8.40411246e-01 5.09223282e-01 -7.87339360e-02
2.03348976e-02 9.79630768e-01 -4.97211754e-01 -4.15073395e-01
-3.35178524e-01 -6.00409985e-01 -2.27021743e-02 -1.20167339e+00
-1.89073026e-01 2.78316051e-01 -1.83174610e-02 -4.77086693e-01
1.75476581e-01 1.33749592e+00 -1.19232202e+00 -3.34160864e-01
9.68704998e-01 9.29889798e-01 -4.44464564e-01 4.69050288e-01
5.57261467e-01 9.32487667e-01 -1.21556067e+00 1.59879342e-01
-1.64726362e-01 -7.08485723e-01 -5.06734192e-01 -3.19292694e-01
-8.07027936e-01 -2.38173902e-01 -4.84047234e-01 9.78547812e-01
3.04498464e-01 2.37224191e-01 1.02408278e+00 -1.16271114e+00
1.98496744e-01 8.22905719e-01 -1.96891949e-01 5.88237464e-01
2.00076327e-01 -1.14369944e-01 2.53690511e-01 3.02391872e-02
1.87940091e-01 5.62821865e-01 3.85308325e-01 -2.54743248e-02
-1.87950993e+00 -7.12934136e-01 -4.35271621e-01 1.81391671e-01
-1.72528410e+00 1.02241576e-01 1.31817818e+00 -1.36119083e-01
1.14504492e+00 5.13279915e-01 9.00380254e-01 5.63328207e-01
8.23485553e-01 1.36314556e-01 1.50511432e+00 -2.89346546e-01
6.30065262e-01 5.34056246e-01 2.66655028e-01 3.57899934e-01
5.14005601e-01 6.06354736e-02 4.22890007e-01 -6.78211987e-01
3.16791326e-01 -3.12994272e-01 -1.92707539e-01 -1.44035235e-01
-3.87306422e-01 6.84913158e-01 1.24615692e-01 1.04611516e+00
-1.77450567e-01 1.11295037e-01 5.82934439e-01 4.68966961e-01
5.84654272e-01 5.05767167e-01 -1.43159053e-03 -1.29338950e-01
-1.03134191e+00 8.96548480e-02 1.41833305e+00 5.19974232e-01
3.96883070e-01 5.85475504e-01 4.29287255e-01 3.46687958e-02
-1.09229282e-01 7.89935648e-01 5.20266235e-01 -7.82215774e-01
-3.65149170e-01 6.30068481e-01 4.85060224e-03 -8.53898168e-01
-8.47995460e-01 -3.36971074e-01 -7.44499385e-01 2.93175340e-01
1.74391285e-01 -5.60047269e-01 1.86243057e-01 1.21726525e+00
1.19814985e-01 -6.15392216e-02 2.00891390e-01 1.70341030e-01
-3.24208327e-02 1.11349118e+00 -7.10918531e-02 -5.85623503e-01
9.79612291e-01 -2.20622033e-01 -1.24813116e+00 8.39038432e-01
5.65469205e-01 -5.02797604e-01 2.29069784e-01 5.65585434e-01
-9.99628305e-01 -1.50311723e-01 -1.36159325e+00 6.48772418e-01
-6.41019821e-01 3.22858483e-01 1.82984114e-01 7.84147799e-01
-9.62501943e-01 7.94258952e-01 -7.20587254e-01 -2.57601917e-01
-8.86886194e-02 6.06203198e-01 1.58200964e-01 7.04829872e-01
-1.09612310e+00 1.33655179e+00 5.27632713e-01 3.16845477e-01
-3.81503612e-01 -5.52933931e-01 -4.63378817e-01 2.07888380e-01
-1.69240721e-02 1.16090998e-01 9.09015357e-01 -7.15494215e-01
-2.00018477e+00 1.50279611e-01 3.45205247e-01 -2.38927543e-01
4.22225267e-01 3.51773143e-01 -8.93991888e-01 4.45737094e-01
-5.99730849e-01 -5.93849123e-01 8.93230557e-01 -9.40921605e-01
-1.21801302e-01 -5.88583574e-02 -3.17135334e-01 -2.56853282e-01
2.29398143e-02 -1.62978619e-01 7.09533274e-01 -5.16061246e-01
-4.20813918e-01 -8.89259517e-01 -7.54522458e-02 -7.73535192e-01
-5.29352307e-01 -3.40128332e-01 1.17813766e+00 -6.58976018e-01
1.39962137e+00 -2.07992482e+00 2.42382810e-01 1.07256413e+00
-2.08091572e-01 2.17290610e-01 5.80282688e-01 1.13979983e+00
-1.94857001e-01 -3.25639248e-01 -2.59525627e-01 3.96092385e-01
6.26514629e-02 -6.62169829e-02 -6.12317562e-01 1.00971544e+00
1.28750771e-03 7.10189104e-01 -4.45818245e-01 -1.20958522e-01
5.76726258e-01 8.64279151e-01 8.06868151e-02 7.21959621e-02
2.57689685e-01 1.77272215e-01 -7.26323366e-01 2.46793240e-01
4.55429912e-01 1.16887555e-01 3.63851666e-01 -4.59509522e-01
-4.51222837e-01 -1.72350153e-01 -7.47896850e-01 7.15450704e-01
-5.23523569e-01 8.42894137e-01 3.54154818e-02 -1.36474454e+00
1.22317541e+00 4.58428115e-01 7.95005798e-01 -6.11260116e-01
7.85757840e-01 1.68428048e-01 -1.82105731e-02 -1.87516451e-01
1.06453344e-01 3.57729532e-02 -3.81694108e-01 9.24202383e-01
7.80624226e-02 -4.35984105e-01 5.97471893e-02 -6.84697703e-02
1.15290749e+00 -1.89696342e-01 3.08946311e-01 -1.17679381e+00
9.35014844e-01 1.85930133e-01 -8.54802355e-02 1.82822257e-01
-7.42435455e-03 -7.43295550e-01 1.04625607e+00 -1.69357419e-01
-1.13746536e+00 -7.16316938e-01 -6.53416693e-01 2.35560000e-01
2.57087767e-01 2.61058062e-01 -1.24443185e+00 -2.52237823e-02
-8.26488361e-02 1.22186542e+00 -3.63861412e-01 -6.87509060e-01
-8.23675096e-01 -1.21971810e+00 7.49231875e-01 7.89005682e-02
2.89404839e-01 -1.05715728e+00 -8.30949545e-01 6.00061476e-01
2.72618651e-01 -9.10182834e-01 -1.36119619e-01 6.16575837e-01
-1.19440591e+00 -9.10495877e-01 -3.28060597e-01 -5.81578016e-01
6.50694668e-01 -4.98529375e-01 6.96990073e-01 -1.95422173e-02
-6.37667537e-01 3.93496186e-01 -1.38197407e-01 8.53994191e-02
-8.90127301e-01 3.06001335e-01 5.49349487e-02 3.00794677e-03
3.32812190e-01 -9.73877847e-01 -4.45477843e-01 4.54447418e-01
-8.89015317e-01 -5.58140755e-01 2.38325074e-01 2.84508497e-01
1.09799944e-01 7.35770941e-01 9.28805649e-01 -5.48234701e-01
1.06463468e+00 -5.17017782e-01 -1.40101373e+00 7.79219195e-02
-9.74525869e-01 2.36944512e-01 1.29017663e+00 -7.76893318e-01
-1.16630566e+00 -1.63330305e-02 2.27668032e-01 1.44802436e-01
-3.93516235e-02 -1.70483127e-01 -5.36294468e-02 -7.44294107e-01
2.99207747e-01 5.24188399e-01 2.86154091e-01 -3.57963800e-01
-2.60293186e-02 6.65683091e-01 3.81098717e-01 -3.50935042e-01
9.12068069e-01 3.04898322e-01 6.07397497e-01 -1.11001015e+00
4.92926203e-02 1.06923468e-01 -2.81172097e-01 -4.53063369e-01
3.98757637e-01 -4.88893956e-01 -1.56799972e+00 6.58498347e-01
-7.30005920e-01 -4.04573381e-01 -5.17447889e-01 3.74531031e-01
-7.01894224e-01 4.30449173e-02 -8.97525311e-01 -1.06295288e+00
-4.83301848e-01 -1.02795219e+00 2.52069712e-01 4.34556037e-01
-2.65938699e-01 -1.26094139e+00 4.20662820e-01 -2.26411089e-01
6.98678315e-01 5.34449160e-01 1.09000993e+00 -7.18183458e-01
-2.48887554e-01 -6.32740080e-01 3.12359422e-01 4.26850528e-01
-4.02940735e-02 2.21453175e-01 -8.13779950e-01 -4.60694730e-01
6.47426188e-01 2.01689869e-01 -8.54314640e-02 3.21545601e-01
7.24142015e-01 -2.76114672e-01 -4.31196749e-01 4.63440150e-01
1.88885748e+00 7.43089139e-01 4.98189598e-01 -8.83558616e-02
5.10257706e-02 2.91753978e-01 -1.72758549e-01 4.25734401e-01
-2.61644833e-02 4.22523409e-01 -9.18855295e-02 -1.85659856e-01
8.59889090e-01 2.83702642e-01 3.87468725e-01 7.73087800e-01
-1.37455940e-01 -2.59635657e-01 -2.91051239e-01 -7.06955558e-03
-1.28901267e+00 -8.53712022e-01 -2.16344707e-02 1.65635324e+00
3.13765645e-01 1.63499564e-01 2.44478613e-01 5.38252413e-01
7.23324537e-01 -9.84317958e-02 -6.98923349e-01 -1.02833092e+00
-3.33042175e-01 2.79388577e-01 8.13710153e-01 5.13064921e-01
-4.42979187e-01 1.82603106e-01 7.46828842e+00 8.74164701e-01
-1.32347643e+00 -3.97430137e-02 4.21261460e-01 1.08174607e-01
4.44889367e-02 -1.54705852e-01 -3.41131151e-01 8.10296297e-01
1.41917300e+00 -6.72201514e-01 7.73145199e-01 5.66253662e-01
2.68484443e-01 -5.76290250e-01 -5.18956840e-01 6.37645721e-01
-3.70536447e-02 -6.71981156e-01 -3.20805520e-01 8.28584507e-02
5.51570892e-01 -4.21954334e-01 -1.09557271e-01 1.17297843e-01
-6.47361353e-02 -7.82409370e-01 3.17613721e-01 9.21749711e-01
4.64012057e-01 -1.30875313e+00 9.13217425e-01 2.78374195e-01
-1.14347541e+00 -1.83557838e-01 2.27263674e-01 5.94587587e-02
2.18394935e-01 4.90020365e-01 -1.31621704e-01 4.17101741e-01
2.15850949e-01 2.81409740e-01 -3.40311378e-01 5.31546295e-01
3.17759871e-01 9.33062911e-01 -6.55584514e-01 -6.07864797e-01
-1.27743870e-01 -9.81245577e-01 4.63790596e-01 1.02806783e+00
5.40280528e-02 3.07774454e-01 -4.95857924e-01 1.03095961e+00
3.41264635e-01 -2.97467858e-02 -6.15247607e-01 5.07880189e-02
2.51309186e-01 1.40502882e+00 -1.25005972e+00 -2.52177656e-01
5.95304109e-02 6.90714657e-01 -3.61442447e-01 6.13563001e-01
-8.27702522e-01 -9.80111957e-01 3.42379332e-01 1.02461815e-01
1.16176680e-01 -7.00804498e-03 -9.45922956e-02 -9.48433578e-01
-3.12662661e-01 -2.28253633e-01 1.37632014e-03 -4.82076168e-01
-8.15135658e-01 8.58463466e-01 5.22620738e-01 -8.60086858e-01
-5.90077102e-01 -3.09974819e-01 -9.21434641e-01 7.67120242e-01
-9.79338706e-01 -4.17086571e-01 2.11519733e-01 7.05328465e-01
-1.23158135e-01 -3.56998950e-01 7.41703153e-01 2.19967991e-01
-6.85657144e-01 3.77579629e-01 7.16368437e-01 -2.87994623e-01
-3.55119020e-01 -1.14224112e+00 -2.35977441e-01 3.17118466e-01
-7.53695309e-01 -2.05967709e-01 1.22870910e+00 -4.42372203e-01
-1.43115187e+00 -5.35615802e-01 5.08556485e-01 2.12015495e-01
8.52928281e-01 -5.82827747e-01 -9.70411241e-01 3.67753416e-01
4.60145414e-01 -1.48831248e-01 4.28649902e-01 -6.87661111e-01
5.50591171e-01 -5.78733087e-01 -1.61934137e+00 2.99165398e-01
-7.29345307e-02 -7.25124896e-01 -4.41562563e-01 1.50061011e-01
9.78522450e-02 1.35497898e-01 -1.29064643e+00 1.01128638e-01
2.98497945e-01 -6.89257205e-01 4.61843938e-01 2.18548357e-01
-6.19789124e-01 -2.25051045e-01 4.72490638e-01 -1.26663613e+00
-1.33438095e-01 -1.19187474e+00 -4.91497874e-01 9.01347935e-01
1.74213707e-01 -1.25574493e+00 1.28664702e-01 2.92995155e-01
4.86586094e-01 -6.38670802e-01 -1.09990776e+00 -6.56592548e-01
5.67668155e-02 3.65066469e-01 5.14287293e-01 6.60831630e-01
7.03894675e-01 -9.73031968e-02 1.27144396e-01 9.39283893e-02
7.14860559e-01 -1.09864995e-01 -3.54036808e-01 -1.11778748e+00
1.25399128e-01 -6.08298182e-01 -5.23783743e-01 -1.46769714e-02
5.10321736e-01 -5.11056483e-01 -2.77883619e-01 -6.36356473e-01
-3.86561394e-01 -1.77022949e-01 -4.21518803e-01 -7.05080628e-02
6.20860875e-01 1.90469533e-01 -1.77570120e-01 -1.91271454e-02
2.84900120e-03 5.08010268e-01 5.47607303e-01 1.99252293e-01
-2.73009449e-01 3.63811433e-01 7.84121081e-02 4.70414579e-01
1.19650328e+00 -5.17695904e-01 -4.43209022e-01 7.93287396e-01
1.47036552e-01 3.30144942e-01 4.03801985e-02 -1.21341062e+00
2.35013664e-01 3.34770948e-01 1.20985910e-01 -7.61973381e-01
1.15217648e-01 -1.35447168e+00 7.92081773e-01 1.09027839e+00
-9.05327946e-02 4.23578262e-01 2.60079384e-01 3.78054947e-01
-5.68313859e-02 -2.90253580e-01 7.76250780e-01 5.75011432e-01
3.39121372e-01 -5.11191905e-01 -1.26943612e+00 -2.18619645e-01
1.64411473e+00 4.71818149e-02 -1.78708598e-01 -1.92503214e-01
-5.57974398e-01 -8.94262828e-03 3.27784508e-01 -2.62727171e-01
-8.85533616e-02 -1.09585702e+00 -1.64293706e-01 4.10872787e-01
-4.92373794e-01 -7.23237336e-01 2.71668643e-01 9.01922584e-01
-8.54010582e-01 4.83428329e-01 -3.55094969e-01 -3.08671683e-01
-7.63518631e-01 5.04007101e-01 8.56714606e-01 -2.01153591e-01
-6.31075621e-01 -5.55472672e-01 -6.13652647e-01 1.74753457e-01
-3.95713723e-05 -3.00911278e-01 -4.78613913e-01 2.88609773e-01
3.98571491e-01 1.02784944e+00 3.79852690e-02 -5.68943202e-01
-1.04987316e-01 7.48271525e-01 4.42175210e-01 6.25238195e-02
1.21942687e+00 -1.32959172e-01 -6.23873234e-01 7.89381623e-01
1.35093379e+00 -1.67579919e-01 -1.03443336e+00 1.65425479e-01
-1.30192712e-01 4.04167712e-01 2.97279477e-01 -6.57305777e-01
-1.10845590e+00 1.31757647e-01 8.19876611e-01 1.03270090e+00
1.62734854e+00 -2.00326413e-01 4.41113681e-01 5.32176554e-01
2.00439036e-01 -1.30585158e+00 -7.30578482e-01 -1.00730471e-02
4.96837914e-01 -3.42787176e-01 7.52363950e-02 -1.48566216e-01
-8.55894014e-02 1.62174892e+00 -2.88609657e-02 -8.08440566e-01
1.26375616e+00 1.06876731e+00 -3.01343858e-01 -1.20679662e-01
-6.67425334e-01 3.11959594e-01 1.16898708e-01 3.26739371e-01
-5.19115850e-02 2.31367141e-01 -6.57051206e-01 3.88915628e-01
-1.53529122e-02 9.12818462e-02 9.31319773e-01 9.55074787e-01
-9.99694914e-02 -5.88209569e-01 -3.53609860e-01 3.95239770e-01
-3.25830936e-01 4.30658400e-01 1.84632353e-02 1.06492460e+00
-2.68736601e-01 7.58040190e-01 2.43304670e-01 -1.24987051e-01
6.65645182e-01 -1.12491250e-01 2.12450743e-01 4.60724562e-01
-8.93652320e-01 1.84818842e-02 -3.01893443e-01 -6.42595649e-01
-1.34635955e-01 -4.35457468e-01 -1.14978027e+00 -8.14412594e-01
-4.16662842e-01 5.16830981e-01 9.30793166e-01 9.41522539e-01
-1.18053019e-01 5.71290016e-01 1.27154052e+00 -7.24834859e-01
-5.92634439e-01 -8.25579703e-01 -1.36754000e+00 -2.07262456e-01
4.60904598e-01 -4.14234161e-01 -1.05924344e+00 -4.95138496e-01] | [5.811440944671631, 2.805062770843506] |
5b6cdec3-de4a-44ac-84df-e12ba33f74f2 | improving-segmentation-of-objects-with | 2304.06229 | null | https://arxiv.org/abs/2304.06229v1 | https://arxiv.org/pdf/2304.06229v1.pdf | Improving Segmentation of Objects with Varying Sizes in Biomedical Images using Instance-wise and Center-of-Instance Segmentation Loss Function | In this paper, we propose a novel two-component loss for biomedical image segmentation tasks called the Instance-wise and Center-of-Instance (ICI) loss, a loss function that addresses the instance imbalance problem commonly encountered when using pixel-wise loss functions such as the Dice loss. The Instance-wise component improves the detection of small instances or ``blobs" in image datasets with both large and small instances. The Center-of-Instance component improves the overall detection accuracy. We compared the ICI loss with two existing losses, the Dice loss and the blob loss, in the task of stroke lesion segmentation using the ATLAS R2.0 challenge dataset from MICCAI 2022. Compared to the other losses, the ICI loss provided a better balanced segmentation, and significantly outperformed the Dice loss with an improvement of $1.7-3.7\%$ and the blob loss by $0.6-5.0\%$ in terms of the Dice similarity coefficient on both validation and test set, suggesting that the ICI loss is a potential solution to the instance imbalance problem. | ['Henrik Skibbe', 'Charissa Poon', 'Muhammad Febrian Rachmadi'] | 2023-04-13 | null | null | null | null | ['lesion-segmentation'] | ['medical'] | [ 2.96188533e-01 1.87895596e-01 -1.16146013e-01 -4.85193998e-01
-1.03679645e+00 -4.75831151e-01 2.97697604e-01 5.17221570e-01
-6.61714137e-01 7.24865735e-01 -4.54621017e-01 -1.48919657e-01
-1.15426399e-01 -5.69223881e-01 -6.40929639e-01 -8.07063878e-01
-2.44869798e-01 1.42441303e-01 4.26948339e-01 1.93627626e-01
2.58232296e-01 3.06777716e-01 -1.21578622e+00 3.00160974e-01
1.18829906e+00 1.40593898e+00 1.31638482e-01 2.06188202e-01
-2.01984867e-02 6.52372777e-01 -6.44774199e-01 -4.49838579e-01
3.26802582e-01 -3.83359998e-01 -6.86766326e-01 -3.20614994e-01
6.54128551e-01 -1.73082188e-01 -1.28715605e-01 1.08800471e+00
8.61732423e-01 1.13464287e-02 8.75609100e-01 -1.17660058e+00
-9.52241272e-02 2.76402980e-01 -1.09656739e+00 5.49096644e-01
-2.27209032e-01 2.94931605e-02 8.46681178e-01 -4.69294429e-01
5.44070542e-01 8.79729807e-01 9.11375105e-01 3.38222623e-01
-1.06366181e+00 -6.60844564e-01 2.40161717e-02 2.21190453e-01
-1.32720578e+00 8.17857534e-02 4.59503293e-01 -5.60887158e-01
7.00543165e-01 3.98386717e-01 2.79113710e-01 3.33812982e-01
1.68879077e-01 1.13780856e+00 1.20435381e+00 -2.11190641e-01
5.08666523e-02 1.17574139e-02 2.12589696e-01 5.55823922e-01
3.27270746e-01 -2.05808014e-01 1.45078227e-01 -5.49276583e-02
7.26021588e-01 -1.54728472e-01 -3.83089513e-01 -2.55470067e-01
-9.97835934e-01 5.98403752e-01 8.96716833e-01 2.28870690e-01
-2.13561177e-01 7.25652203e-02 6.65524006e-01 -1.39081940e-01
6.86945915e-01 4.33858842e-01 -2.86879152e-01 -2.56106984e-02
-1.30724955e+00 2.13104814e-01 2.98380136e-01 5.36771894e-01
1.27209067e-01 -4.32006359e-01 -5.16134560e-01 1.28553295e+00
-2.18230523e-02 1.92283541e-01 4.76675779e-01 -9.26370025e-01
6.71345234e-01 7.21406996e-01 -1.57196641e-01 -9.65754211e-01
-6.73683167e-01 -8.40157330e-01 -7.98917949e-01 2.85494149e-01
8.17899227e-01 -3.76911052e-02 -1.08623350e+00 1.86657262e+00
3.14801127e-01 -1.05465194e-02 -4.11612630e-01 9.46338594e-01
9.00901377e-01 3.47394586e-01 5.68884760e-02 -7.26236030e-02
1.33919764e+00 -8.65897536e-01 -6.74222708e-01 -1.97921902e-01
7.05276251e-01 -6.17279470e-01 1.14658141e+00 2.52674341e-01
-1.33545446e+00 -2.89870888e-01 -1.12705266e+00 1.02889210e-01
-1.04637176e-01 1.12836704e-01 4.09770161e-01 6.27752066e-01
-8.16907525e-01 6.69198513e-01 -8.79988730e-01 1.99142154e-02
8.66280377e-01 1.45232722e-01 -2.34683096e-01 -3.28873545e-01
-7.14641750e-01 8.29715848e-01 -2.09797304e-02 -1.74175963e-01
-5.25079608e-01 -1.35311639e+00 -6.15569532e-01 7.73262233e-02
2.33177617e-01 -2.47162715e-01 9.62202787e-01 -7.46657908e-01
-8.98861110e-01 1.17172766e+00 5.10516167e-02 -6.01225674e-01
1.05796230e+00 -1.24125276e-02 -3.94831635e-02 3.09800863e-01
3.18294644e-01 8.26425910e-01 1.75094441e-01 -1.22154689e+00
-5.48966050e-01 -8.43022048e-01 -2.69899666e-02 1.74472779e-01
3.45976949e-02 -2.03104079e-01 -4.09959078e-01 -9.60587084e-01
1.98064551e-01 -8.39495182e-01 -1.30323708e-01 2.57499784e-01
-3.13503653e-01 -1.31007269e-01 7.46603489e-01 -1.01623011e+00
1.25817180e+00 -2.14081478e+00 -2.90763199e-01 1.99882865e-01
3.64555031e-01 3.67862225e-01 -1.22658256e-02 -2.44795278e-01
-2.36688569e-01 1.67194501e-01 -7.75006771e-01 -1.04305953e-01
-3.92505884e-01 -1.31031215e-01 3.07759315e-01 4.10407484e-01
1.80546671e-01 6.56691849e-01 -8.63423526e-01 -2.98916131e-01
-3.36212702e-02 5.54154992e-01 -4.30863857e-01 -1.09069504e-01
3.10360551e-01 3.16466957e-01 -3.29243043e-03 3.56430590e-01
8.60246480e-01 -2.78993875e-01 -1.25575647e-01 -3.54735762e-01
2.08471626e-01 7.25382473e-03 -9.20899689e-01 1.63709939e+00
-2.95107126e-01 7.65366077e-01 6.24632426e-02 -1.17074478e+00
7.40644634e-01 2.82056838e-01 7.03193426e-01 -1.04711556e+00
-4.44263481e-02 3.88324618e-01 2.86777318e-01 -2.08320752e-01
1.07275648e-02 -3.28163534e-01 8.04183558e-02 1.30049691e-01
-2.12084830e-01 -8.04632977e-02 3.72470677e-01 2.50882268e-01
1.17762899e+00 -1.52487889e-01 -3.29321809e-02 -4.47100312e-01
3.75831783e-01 -2.07126081e-01 7.36366034e-01 6.12821460e-01
-6.06548309e-01 1.20534527e+00 1.04142451e+00 -1.66211993e-01
-7.41155386e-01 -1.11674368e+00 -5.27092755e-01 4.29947972e-01
2.86932975e-01 -6.13037311e-02 -1.01662910e+00 -9.94906843e-01
7.68316537e-02 6.85653269e-01 -4.55149263e-01 -1.21848390e-01
-4.67839479e-01 -1.22177029e+00 6.66430712e-01 7.42463350e-01
7.73463547e-01 -6.58775926e-01 -7.47378886e-01 1.51423821e-02
-4.04277712e-01 -1.10671926e+00 -6.09215200e-01 7.28302151e-02
-9.17138577e-01 -1.22507036e+00 -1.28888476e+00 -5.68560600e-01
8.26248527e-01 -4.52461885e-03 1.11049330e+00 1.13842465e-01
-8.28019381e-01 -4.55881916e-02 -4.11432892e-01 -5.19106328e-01
-2.37520821e-02 -1.75310165e-01 -5.08710384e-01 -2.10700259e-01
-6.92969337e-02 -2.99207985e-01 -1.22139275e+00 5.92735291e-01
-9.17412579e-01 -1.58361033e-01 3.36657822e-01 9.22005475e-01
8.42577398e-01 9.73705351e-02 6.66089296e-01 -9.00794923e-01
4.29873526e-01 -2.07703680e-01 -4.17870969e-01 2.08161667e-01
-8.38874161e-01 -1.36869609e-01 2.40918860e-01 -6.38024956e-02
-6.88473761e-01 -2.02692971e-01 -4.40648273e-02 -1.98865756e-01
1.49520144e-01 3.74478459e-01 6.54965115e-04 -2.56774500e-02
5.33426225e-01 -1.28612742e-01 2.28248924e-01 -5.34860671e-01
5.33111729e-02 5.54875076e-01 4.88031507e-01 -3.71402532e-01
-1.20757124e-03 5.42353511e-01 1.18572369e-01 -4.58943993e-01
-6.43629789e-01 -6.13296688e-01 -3.86976302e-01 -1.66399240e-01
9.78149652e-01 -6.94961071e-01 -4.24224734e-01 6.71781600e-01
-8.85918260e-01 -3.38546216e-01 -5.09352088e-01 4.02074575e-01
-4.50673997e-01 5.19321442e-01 -4.54311401e-01 -4.76322979e-01
-5.99783659e-01 -1.63165474e+00 9.41479862e-01 2.76277035e-01
-1.15031146e-01 -6.91015840e-01 -2.13504896e-01 5.40478945e-01
4.38710392e-01 8.22680891e-01 1.14215589e+00 -5.17183363e-01
5.76337166e-02 -6.66810274e-01 -6.68006361e-01 7.26064742e-01
1.15926892e-01 -2.15122879e-01 -7.38786638e-01 -3.65603298e-01
-1.44810259e-01 -7.50285061e-03 1.09583318e+00 8.72453511e-01
1.57841647e+00 5.24083637e-02 -1.70870423e-01 4.67793435e-01
1.41634595e+00 5.44792354e-01 9.52519417e-01 2.46554092e-01
4.80260909e-01 6.57422066e-01 3.70974302e-01 1.26685575e-01
3.19663823e-01 9.84425128e-01 3.88387173e-01 -4.60598439e-01
-6.26402080e-01 1.27856791e-01 -1.58676907e-01 2.84498513e-01
5.71597517e-02 -7.97313079e-02 -1.08310342e+00 7.30474055e-01
-1.69790697e+00 -6.32668793e-01 -2.99800128e-01 2.45834398e+00
7.88358748e-01 2.21530274e-01 3.79634082e-01 3.55630159e-01
9.20597136e-01 -2.19982620e-02 -6.47409260e-01 -4.77870941e-01
-5.58280796e-02 2.84408391e-01 7.14835048e-01 3.78627181e-01
-1.05720699e+00 4.41270322e-01 5.54034424e+00 1.22469962e+00
-1.07228231e+00 3.00708979e-01 1.31791067e+00 -4.10903871e-01
1.30427286e-01 -3.01081181e-01 -5.03068984e-01 8.22923899e-01
7.34845102e-01 1.83876365e-01 -4.80720662e-02 5.41231334e-01
1.98700085e-01 -5.13577521e-01 -8.39432478e-01 9.17632282e-01
2.44674068e-02 -1.13055050e+00 -9.18015242e-02 9.52935368e-02
7.02392161e-01 1.91395618e-02 2.03872532e-01 -1.15935124e-01
-3.52566004e-01 -1.08767116e+00 7.05014586e-01 2.08118275e-01
1.07925260e+00 -7.97470510e-01 8.75191927e-01 5.98959178e-02
-9.40374255e-01 1.12342000e-01 -1.81982175e-01 2.66295791e-01
1.61488175e-01 1.05667889e+00 -4.46561575e-01 4.05986786e-01
9.75368977e-01 5.02728105e-01 -4.79679912e-01 1.73661673e+00
-2.14523706e-03 5.46704471e-01 -3.45826983e-01 4.86378103e-01
3.83406579e-01 -2.65498638e-01 6.28464639e-01 1.27496874e+00
1.57039315e-01 4.76652989e-03 -2.46745467e-01 8.75107050e-01
-2.14573324e-01 9.19479504e-03 7.51733501e-03 4.29972172e-01
1.68367937e-01 1.23837376e+00 -1.06907701e+00 -2.53864616e-01
-1.66018158e-01 8.96247685e-01 -1.71915535e-02 1.13959439e-01
-8.50166261e-01 -8.70794892e-01 3.69118810e-01 4.73902047e-01
3.64453912e-01 3.72994274e-01 -8.27732265e-01 -5.74343085e-01
3.99455488e-01 -6.71586931e-01 5.58569431e-01 -5.01044452e-01
-1.26307750e+00 4.70580250e-01 8.36327896e-02 -1.14037907e+00
3.68741274e-01 -6.26097918e-01 -7.15408504e-01 8.30813468e-01
-1.48236430e+00 -6.13084376e-01 -4.59139585e-01 1.17601365e-01
2.18296334e-01 1.58877969e-01 2.69330502e-01 7.71068811e-01
-7.84028172e-01 1.12521768e+00 2.47103915e-01 2.41244763e-01
6.52150333e-01 -1.35454094e+00 4.20707203e-02 6.37684286e-01
-2.91858137e-01 2.23661184e-01 4.12560970e-01 -2.88397819e-01
-3.96077752e-01 -9.20475185e-01 5.79670131e-01 -3.12328070e-01
6.43265024e-02 1.17943905e-01 -8.10181081e-01 2.16388062e-01
-1.80175722e-01 2.74143726e-01 4.62958843e-01 -3.33352685e-01
-3.59792262e-01 -2.25703269e-01 -1.83625436e+00 3.34378868e-01
7.42003381e-01 -3.46821453e-03 -2.64573455e-01 3.88741791e-01
4.58692312e-01 -4.63077009e-01 -1.23216271e+00 9.34573174e-01
5.95875204e-01 -9.46703672e-01 1.09385347e+00 -4.42428678e-01
6.80440187e-01 -1.36396915e-01 -2.37313155e-02 -1.17060852e+00
-1.62929967e-01 -1.09479517e-01 3.88908267e-01 1.14210284e+00
8.05059016e-01 -4.51108903e-01 8.43388498e-01 6.50219858e-01
-3.45962912e-01 -1.28218973e+00 -1.32985401e+00 -8.09537768e-01
7.09429502e-01 -2.84831077e-01 3.30685824e-01 7.09204137e-01
-9.05341953e-02 -1.72645137e-01 1.01684116e-01 -4.13329989e-01
6.64385259e-01 -3.16574425e-01 2.39122361e-01 -9.12661493e-01
-6.88780174e-02 -8.72339725e-01 -6.26579821e-01 -7.11284161e-01
-3.35218132e-01 -1.23647058e+00 -7.99946338e-02 -1.75527525e+00
4.89992052e-01 -5.42798400e-01 -6.53475702e-01 3.42222720e-01
-5.32605886e-01 4.73659724e-01 3.13990414e-01 1.27169430e-01
-2.91682214e-01 2.51333416e-01 1.28451276e+00 -2.39503667e-01
-1.88836321e-01 7.91420713e-02 -6.18874609e-01 6.39494002e-01
8.42828512e-01 -5.74760318e-01 -2.01834261e-01 -3.61230373e-01
-1.18666008e-01 -2.31892198e-01 5.03409564e-01 -1.06032813e+00
-2.13517353e-01 5.14384806e-01 3.83482486e-01 -5.34139395e-01
8.77659246e-02 -5.89587450e-01 -3.16405326e-01 6.43151581e-01
-4.41374481e-01 5.68343624e-02 3.39630604e-01 4.46290940e-01
-1.88020453e-01 -7.94467479e-02 1.34857619e+00 -1.60928011e-01
-3.70130211e-01 7.95989558e-02 1.25818506e-01 5.25867641e-01
1.20099950e+00 -4.74818379e-01 -4.48865265e-01 -1.54715031e-01
-5.96184552e-01 3.01517397e-01 9.88633335e-02 1.83574319e-01
4.46148455e-01 -1.17325985e+00 -8.35488439e-01 6.67525083e-02
6.83104992e-02 2.15152581e-03 3.29782039e-01 1.54808259e+00
-7.52749145e-01 2.38434434e-01 -1.68492913e-01 -8.00654471e-01
-1.25447035e+00 -3.28351185e-02 5.81358194e-01 -5.77010632e-01
-7.07425177e-01 1.20439112e+00 4.95686054e-01 -3.38013977e-01
3.89426470e-01 -4.70821291e-01 -6.52006790e-02 7.57578164e-02
6.00909173e-01 8.39091897e-01 4.44739103e-01 -5.89240432e-01
-6.85902238e-01 5.74769557e-01 -2.28255019e-01 -2.58447994e-02
1.11543369e+00 1.09887004e-01 -1.86611176e-01 4.96009178e-02
1.47905695e+00 -4.70357150e-01 -1.29295647e+00 -1.85860783e-01
-2.13471968e-02 -4.95393038e-01 3.77436697e-01 -1.43477643e+00
-1.59633279e+00 9.73019361e-01 1.21267509e+00 -2.22307518e-02
1.06588387e+00 -4.19224612e-02 1.15843177e+00 -5.67948997e-01
9.98425204e-03 -1.11464775e+00 -2.09209189e-01 1.29047483e-01
7.54559636e-01 -1.22549701e+00 -1.12002492e-02 -3.82313341e-01
-8.13769042e-01 6.33668959e-01 4.12927568e-01 -2.85860188e-02
6.33425355e-01 2.80474573e-01 1.19817052e-02 -2.33490258e-01
-1.48126304e-01 -2.83027953e-03 7.17448831e-01 5.46113014e-01
5.39450049e-01 2.49967843e-01 -1.10551226e+00 6.44735456e-01
5.86747192e-03 -8.14072564e-02 2.00288013e-01 6.54666483e-01
-2.29257271e-01 -7.38755345e-01 -9.65998396e-02 9.16042686e-01
-8.10178638e-01 -1.55324871e-02 -2.75856018e-01 7.12581158e-01
2.85512567e-01 8.97118151e-01 3.44814658e-01 -1.21528208e-01
7.13728547e-01 -1.29093215e-01 5.50205171e-01 -2.78721035e-01
-6.67630732e-01 2.68877726e-02 -1.30051121e-01 -6.46260440e-01
-1.96583331e-01 -5.65563619e-01 -1.51981652e+00 -7.65064508e-02
-3.63886476e-01 -2.94785779e-02 7.49094844e-01 6.52038813e-01
3.88566732e-01 5.98545194e-01 4.26822186e-01 -5.17588973e-01
-4.94135886e-01 -8.02244544e-01 -7.09368646e-01 7.11807489e-01
1.78982675e-01 -6.74578071e-01 -4.59817499e-01 -2.90274978e-01] | [14.656875610351562, -2.357570171356201] |
7d1601c8-da18-4553-8f8e-911a02c6256c | unleashing-the-power-of-neural-discourse-1 | null | null | https://aclanthology.org/2020.coling-main.337 | https://aclanthology.org/2020.coling-main.337.pdf | Unleashing the Power of Neural Discourse Parsers - A Context and Structure Aware Approach Using Large Scale Pretraining | RST-based discourse parsing is an important NLP task with numerous downstream applications, such as summarization, machine translation and opinion mining. In this paper, we demonstrate a simple, yet highly accurate discourse parser, incorporating recent contextual language models. Our parser establishes the new state-of-the-art (SOTA) performance for predicting structure and nuclearity on two key RST datasets, RST-DT and Instr-DT. We further demonstrate that pretraining our parser on the recently available large-scale {``}silver-standard{''} discourse treebank MEGA-DT provides even larger performance benefits, suggesting a novel and promising research direction in the field of discourse analysis. | ['Giuseppe Carenini', 'Patrick Huber', 'Grigorii Guz'] | 2020-12-01 | null | null | null | coling-2020-8 | ['discourse-parsing'] | ['natural-language-processing'] | [ 3.41964990e-01 9.31521416e-01 -5.35267353e-01 -3.93976718e-01
-1.35460329e+00 -7.41772354e-01 8.59166682e-01 5.26999831e-01
-4.24423873e-01 1.08431816e+00 1.06746447e+00 -8.82155061e-01
3.27010512e-01 -6.08497262e-01 -5.18105268e-01 -3.93587649e-01
-2.13186800e-01 5.74527681e-01 5.22848964e-01 -5.84992111e-01
4.70275819e-01 -2.21461207e-01 -9.52321351e-01 9.71524477e-01
8.08364332e-01 5.08264959e-01 -1.85629930e-02 9.56622243e-01
-4.21556354e-01 1.22568977e+00 -1.15034366e+00 -7.27588236e-01
-4.70876575e-01 -5.78325093e-01 -1.45825720e+00 -2.47032434e-01
1.05567701e-01 -2.47158244e-01 -7.30079934e-02 5.99336565e-01
4.54916686e-01 7.37697706e-02 4.40783381e-01 -3.81457925e-01
-6.49568141e-01 1.36102474e+00 -7.32288301e-01 5.98617911e-01
7.65746832e-01 -2.54689693e-01 1.47336304e+00 -4.10452724e-01
1.26667404e+00 1.52771461e+00 4.34471518e-01 5.88755846e-01
-8.52425098e-01 -1.27783909e-01 5.18762767e-01 2.82892525e-01
-3.08685720e-01 -4.24015909e-01 8.15794706e-01 -2.87203699e-01
1.52625430e+00 4.43861157e-01 4.49188977e-01 1.38870215e+00
6.90612346e-02 1.28142822e+00 1.10058522e+00 -8.25540483e-01
1.11545891e-01 -4.45297420e-01 5.77372491e-01 4.44313794e-01
-1.31478593e-01 -5.87275982e-01 -7.31089413e-01 -3.29958022e-01
2.16026917e-01 -1.12677073e+00 -1.85186476e-01 5.93703508e-01
-1.26675904e+00 9.44583356e-01 -8.34467784e-02 7.05142021e-01
-1.39439121e-01 -1.05015664e-04 9.36203659e-01 4.41195756e-01
8.09966862e-01 5.21035433e-01 -4.12785590e-01 -7.72331238e-01
-4.32813287e-01 2.38712534e-01 1.21128047e+00 7.42845416e-01
8.47108141e-02 -9.81050879e-02 -3.18394005e-01 6.32342756e-01
1.67576492e-01 1.04431540e-01 5.09169340e-01 -1.23228908e+00
1.08398783e+00 4.65143830e-01 -8.85119140e-02 -6.84391379e-01
-4.04311687e-01 -3.63641838e-03 -5.51866710e-01 -4.65604722e-01
5.20813882e-01 -5.18786073e-01 -4.01898444e-01 1.67678857e+00
5.18420815e-01 -2.37164855e-01 3.13775480e-01 4.75372314e-01
1.21336365e+00 8.93972993e-01 2.30756417e-01 -8.49128127e-01
1.63520205e+00 -8.36037755e-01 -8.63023102e-01 -3.02916735e-01
1.08424914e+00 -7.59570122e-01 9.01396394e-01 3.05154532e-01
-1.39195704e+00 1.70012787e-02 -8.64680767e-01 -5.11205137e-01
1.94531888e-01 6.44320995e-02 7.65036523e-01 5.57059526e-01
-7.79716253e-01 5.38581669e-01 -9.24255610e-01 -1.36279017e-01
4.30190593e-01 -5.92807401e-03 -1.31919429e-01 2.59333491e-01
-1.29181480e+00 1.05121064e+00 5.46974003e-01 -5.11036292e-02
-3.79233986e-01 -4.71439183e-01 -9.27046299e-01 -1.89892501e-01
5.73756814e-01 -4.16131169e-01 1.79021823e+00 -5.82701385e-01
-1.99029887e+00 1.16615617e+00 -7.20894039e-01 -8.30082417e-01
3.46420825e-01 -6.66343212e-01 -1.81484461e-01 2.59263575e-01
1.05741210e-01 8.96273702e-02 2.85811603e-01 -8.34478557e-01
-5.40634334e-01 -3.23537409e-01 3.02576631e-01 2.95222163e-01
-1.77816465e-01 7.20013976e-01 1.16689038e-02 -6.46394849e-01
-5.39922453e-02 -4.86498982e-01 -6.93324804e-02 -7.35911548e-01
-6.53552949e-01 -8.92443240e-01 4.78520721e-01 -1.03515768e+00
1.72716665e+00 -1.64508891e+00 3.94859433e-01 -3.57827246e-01
-2.65625287e-02 4.13792402e-01 5.69009595e-02 7.09304035e-01
1.85534582e-01 3.87628078e-01 -2.84097999e-01 -2.72169769e-01
1.80268735e-02 2.14384720e-01 -3.55346441e-01 1.08703025e-01
3.80245447e-01 1.13661408e+00 -1.01045084e+00 -6.32612348e-01
-1.18850870e-02 -1.14328645e-01 -3.68287027e-01 2.53780842e-01
-8.95830452e-01 5.84734499e-01 -7.26457834e-01 3.40981126e-01
-3.76212522e-02 -2.21911430e-01 9.99462545e-01 2.06534401e-01
-3.04076314e-01 1.13397980e+00 -4.97124493e-01 1.83406317e+00
-6.93787821e-03 1.09051585e+00 9.58604291e-02 -1.18224299e+00
6.80334926e-01 5.08158624e-01 6.11913055e-02 -6.74769819e-01
3.87252748e-01 1.18134506e-01 2.61321425e-01 -8.56541991e-01
8.80139768e-01 -1.45114049e-01 -5.54147184e-01 7.29070485e-01
-1.40236497e-01 6.05219156e-02 5.53184688e-01 2.21600085e-01
1.25616920e+00 1.29589871e-01 7.60699809e-01 -3.28653067e-01
5.87198675e-01 3.90068412e-01 5.78482330e-01 3.02921593e-01
-6.73027262e-02 3.58479023e-01 9.90886867e-01 -8.03747550e-02
-1.03092027e+00 -5.25527358e-01 -7.95799196e-02 1.58797526e+00
-4.11424547e-01 -6.74639702e-01 -8.29021633e-01 -7.73078442e-01
-4.05397713e-01 9.91821527e-01 -5.97357750e-01 6.58024013e-01
-1.76993775e+00 -8.96787167e-01 8.07227075e-01 6.25745237e-01
4.33227479e-01 -1.06673455e+00 -3.98179322e-01 5.08101761e-01
-5.70955336e-01 -1.28837931e+00 7.08120912e-02 -1.60879120e-01
-8.18880856e-01 -1.21089196e+00 -4.72807080e-01 -9.89667356e-01
5.02899438e-02 -1.29087955e-01 1.29572523e+00 2.10152119e-02
3.25504631e-01 -1.79065451e-01 -5.20076811e-01 -4.59601760e-01
-9.65726018e-01 4.09670889e-01 -5.02995908e-01 -7.34471619e-01
2.99940616e-01 -4.20475274e-01 -2.23320127e-01 -2.98098683e-01
-4.02002275e-01 3.15902233e-02 1.31902769e-01 7.88852036e-01
3.36251706e-01 -5.05793631e-01 8.55485320e-01 -1.49137807e+00
1.10944235e+00 -4.11983579e-01 -1.75948352e-01 5.00468194e-01
1.61499873e-01 1.05721027e-01 4.57613826e-01 -1.29625067e-01
-1.76268566e+00 -8.53740036e-01 -7.71219492e-01 6.26044154e-01
-1.81494113e-02 1.01169825e+00 -1.02296025e-01 7.95536935e-01
7.14426637e-01 -2.17177719e-01 -3.69663954e-01 -5.21886826e-01
6.32832706e-01 6.51092529e-01 7.73386538e-01 -8.68898928e-01
2.90076286e-01 2.86608428e-01 -2.37331331e-01 -1.19960368e+00
-1.18931246e+00 -3.48516583e-01 -7.58418143e-01 7.40627572e-02
1.05904293e+00 -8.26031208e-01 -7.35205889e-01 2.64388353e-01
-1.72665203e+00 -4.44282204e-01 -2.31807798e-01 1.03976026e-01
-3.89828444e-01 8.33902240e-01 -1.14893365e+00 -9.51018393e-01
-6.84049547e-01 -5.89537919e-01 1.01049876e+00 1.99724644e-01
-7.45042741e-01 -1.16462362e+00 3.61369461e-01 5.96749902e-01
-1.36443257e-01 7.01106548e-01 1.22928321e+00 -8.75039220e-01
-2.95904636e-01 4.15682226e-01 2.49510445e-03 4.29579243e-02
-2.02817664e-01 2.42588893e-01 -7.50508130e-01 -6.79935515e-02
-5.60151087e-03 -6.42080784e-01 8.87253582e-01 3.71160716e-01
6.63242579e-01 -4.72023547e-01 -2.03700319e-01 8.19515157e-03
8.51343930e-01 9.55693498e-02 6.31472826e-01 6.18350983e-01
3.62429917e-01 7.73152590e-01 8.61329734e-01 3.01301748e-01
7.64811337e-01 2.00114742e-01 -2.16497496e-01 4.88724202e-01
-1.88545913e-01 -1.08791485e-01 5.85865796e-01 1.45805800e+00
-3.43202800e-01 -4.73503828e-01 -1.04035723e+00 7.90873706e-01
-2.10056400e+00 -1.14038420e+00 -4.72567320e-01 1.42251420e+00
1.38322878e+00 5.06332695e-01 5.89487143e-02 4.65260185e-02
5.69617212e-01 7.65243769e-01 -1.44863769e-01 -8.45740318e-01
-4.59332049e-01 4.83989149e-01 -2.47700289e-01 5.77266812e-01
-1.12722850e+00 1.09331536e+00 6.67908096e+00 7.17534244e-01
-7.27675796e-01 2.56935328e-01 6.91313684e-01 4.77904268e-02
-3.77513945e-01 -6.25458285e-02 -8.74084234e-01 1.69406071e-01
1.40126860e+00 -5.29455960e-01 -3.58244807e-01 7.28834987e-01
1.48057312e-01 -4.77189481e-01 -1.05103719e+00 1.80315092e-01
5.67460358e-02 -1.97399819e+00 -9.15594026e-02 -1.51569933e-01
6.89117193e-01 1.22872248e-01 -4.01462972e-01 4.06751573e-01
6.56111717e-01 -8.10604811e-01 6.69400454e-01 -1.09405763e-01
7.86391497e-01 -5.61407089e-01 6.80707633e-01 5.45665085e-01
-9.99089181e-01 1.42976448e-01 -1.22233301e-01 -2.98657328e-01
7.32368767e-01 3.58132958e-01 -8.36250722e-01 8.42300773e-01
2.11893767e-01 6.08698547e-01 -1.46927565e-01 3.84312600e-01
-7.65117586e-01 1.12800956e+00 -1.61814600e-01 -3.14554214e-01
2.70729125e-01 1.36508971e-01 8.12586010e-01 1.56757092e+00
-1.75921515e-01 8.07900608e-01 1.81112796e-01 2.41677478e-01
-4.36004847e-01 5.84922619e-02 -2.18186080e-01 -1.24044113e-01
6.05690598e-01 7.27634013e-01 -5.25420249e-01 -5.21328688e-01
-2.07252815e-01 4.58333373e-01 7.49778390e-01 9.27165058e-03
-5.31918228e-01 1.24990717e-02 3.42659980e-01 -2.88183153e-01
4.10715669e-01 -4.70917523e-01 -3.37166935e-01 -1.19021869e+00
9.70545560e-02 -1.01931322e+00 6.07936621e-01 -1.34350181e-01
-1.25652373e+00 7.03816533e-01 2.45555967e-01 -6.53510571e-01
-7.47529328e-01 -6.00940645e-01 -9.78224695e-01 4.56692874e-01
-1.74869633e+00 -1.16199148e+00 2.83896744e-01 -9.13292821e-03
9.80888307e-01 -4.94014705e-03 9.58873630e-01 -1.62301302e-01
-8.23476136e-01 3.21284503e-01 2.38555998e-01 4.78460014e-01
5.96780181e-01 -1.43775082e+00 7.66565979e-01 7.34310567e-01
-5.49415573e-02 4.98781502e-01 8.28186572e-01 -6.23336256e-01
-1.22796142e+00 -7.27723777e-01 1.36948645e+00 -6.01803124e-01
1.15217733e+00 -2.13193640e-01 -1.14999545e+00 8.92290294e-01
7.69115388e-01 -8.88606310e-01 8.71785581e-01 6.27342403e-01
-9.27450806e-02 6.47562504e-01 -7.80961454e-01 4.80483234e-01
1.01498342e+00 -5.69753766e-01 -1.54077196e+00 4.25286829e-01
1.03653646e+00 -9.86242592e-01 -1.27127302e+00 2.49942869e-01
4.15496290e-01 -9.01273668e-01 7.23915517e-01 -9.05799687e-01
9.99278128e-01 4.22326624e-01 -3.49339209e-02 -9.73763883e-01
4.49251123e-02 -1.07819045e+00 -5.94554961e-01 1.57870555e+00
6.51046515e-01 -3.65046978e-01 1.00236344e+00 6.39668107e-01
-4.14732814e-01 -8.60809326e-01 -1.20721233e+00 -4.02595967e-01
7.07613826e-01 -4.74735737e-01 3.70892644e-01 1.03620744e+00
8.61975253e-01 9.27351058e-01 2.06914451e-02 -3.72894704e-01
2.98626244e-01 3.10393363e-01 5.57964742e-01 -1.12573957e+00
-3.37429553e-01 -4.56969768e-01 3.22702080e-01 -1.49716341e+00
5.20560384e-01 -7.05897272e-01 -7.78093562e-02 -1.70756233e+00
9.04255807e-02 -2.01812685e-01 4.27790374e-01 8.29511788e-03
-4.15536940e-01 -3.24338794e-01 1.26806021e-01 3.66852880e-02
-8.88256729e-01 6.74448013e-01 1.21307158e+00 -3.84931080e-02
-3.34821641e-01 -1.04884226e-02 -6.62473857e-01 9.79432046e-01
7.09184468e-01 -2.23986685e-01 -3.90423350e-02 -7.38308668e-01
2.10081339e-01 4.89445478e-01 -3.57469648e-01 -1.31766453e-01
1.90828070e-01 -2.55959719e-01 -9.58557799e-02 -9.35621440e-01
2.78038234e-01 2.69363075e-01 -6.05381727e-01 1.55334324e-01
-6.31110132e-01 2.10838646e-01 2.89123386e-01 4.33908880e-01
-3.40771347e-01 -4.82527047e-01 2.47024953e-01 -4.02098596e-01
-7.10073829e-01 -4.48209137e-01 -5.60952902e-01 7.66545594e-01
8.07283938e-01 8.48909467e-02 -1.15924072e+00 -1.56285107e-01
-5.77256560e-01 3.63414586e-01 6.05148748e-02 1.31018326e-01
3.49811912e-01 -7.35743105e-01 -1.14655602e+00 -6.11705005e-01
-3.26149762e-01 4.22239155e-01 1.28773144e-02 6.65562391e-01
-3.60409439e-01 6.79293275e-01 1.84100389e-01 -5.59920073e-01
-1.68550932e+00 -3.67497243e-02 -2.99861848e-01 -9.56064343e-01
-7.28136361e-01 1.09791946e+00 -1.34899974e-01 -7.51155093e-02
1.20330945e-01 -5.26934624e-01 -5.30221760e-01 3.37456644e-01
6.55372441e-01 4.11950380e-01 -7.96375498e-02 -6.36443496e-01
-2.58859217e-01 1.13739140e-01 -2.11539865e-01 -2.96253711e-01
1.52628815e+00 -2.21007809e-01 -4.93932724e-01 6.83411837e-01
8.23489785e-01 7.43081048e-02 -7.57610381e-01 -3.48908991e-01
8.44768703e-01 -8.93582851e-02 -3.55169266e-01 -5.66648662e-01
-5.14878668e-02 7.80687034e-01 -5.91841340e-01 5.21460354e-01
6.76103890e-01 2.97486305e-01 9.88130867e-01 7.76914060e-01
2.58723021e-01 -1.38833487e+00 1.84016928e-01 1.09240305e+00
1.06696987e+00 -1.09794343e+00 3.64679068e-01 -5.94327927e-01
-8.63501012e-01 1.01031291e+00 4.03281122e-01 -6.08294494e-02
2.67334521e-01 2.98260242e-01 -8.26569423e-02 -3.63955259e-01
-1.20973694e+00 -4.40317057e-02 1.43558219e-01 5.90089262e-01
1.12480104e+00 2.67324932e-02 -7.16472983e-01 6.93178535e-01
-6.45522594e-01 -3.87841851e-01 6.93039834e-01 1.11695862e+00
-7.06097484e-01 -1.46434927e+00 -1.73485532e-01 3.04169923e-01
-8.98874104e-01 -1.41449824e-01 -7.65619040e-01 9.46118891e-01
-4.99516428e-01 1.09660697e+00 -3.34212072e-02 1.40724719e-01
2.89573342e-01 1.37058511e-01 7.52985299e-01 -9.39319432e-01
-1.01850331e+00 1.36637732e-01 1.47383821e+00 -1.24303833e-01
-9.90513563e-01 -1.04287052e+00 -1.60309720e+00 -6.06500626e-01
-3.67460072e-01 3.04273397e-01 2.76507050e-01 1.30366468e+00
1.45784810e-01 6.31957114e-01 2.10276455e-01 -7.04840302e-01
-4.62656975e-01 -1.26816916e+00 -9.77161108e-04 2.07671657e-01
1.90477058e-01 -1.61080480e-01 2.00275332e-01 7.12276772e-02] | [10.81224250793457, 9.471895217895508] |
08c83109-2fd2-4abf-92cd-f8fed1077338 | expand-rerank-and-retrieve-query-reranking | 2305.17080 | null | https://arxiv.org/abs/2305.17080v1 | https://arxiv.org/pdf/2305.17080v1.pdf | Expand, Rerank, and Retrieve: Query Reranking for Open-Domain Question Answering | We propose EAR, a query Expansion And Reranking approach for improving passage retrieval, with the application to open-domain question answering. EAR first applies a query expansion model to generate a diverse set of queries, and then uses a query reranker to select the ones that could lead to better retrieval results. Motivated by the observation that the best query expansion often is not picked by greedy decoding, EAR trains its reranker to predict the rank orders of the gold passages when issuing the expanded queries to a given retriever. By connecting better the query expansion model and retriever, EAR significantly enhances a traditional sparse retrieval method, BM25. Empirically, EAR improves top-5/20 accuracy by 3-8 and 5-10 points in in-domain and out-of-domain settings, respectively, when compared to a vanilla query expansion model, GAR, and a dense retrieval model, DPR. | ['James Glass', 'Wen-tau Yih', 'Shang-Wen Li', 'Wei Fang', 'Yung-Sung Chuang'] | 2023-05-26 | null | null | null | null | ['passage-retrieval', 'open-domain-question-answering'] | ['natural-language-processing', 'natural-language-processing'] | [ 8.07960704e-02 -2.37329140e-01 -4.21016932e-01 8.85151401e-02
-1.87287366e+00 -7.22480834e-01 4.53755528e-01 4.23078537e-01
-5.44435143e-01 7.78467953e-01 6.33980811e-01 -1.60059333e-01
-5.50321937e-01 -7.59202600e-01 -5.44999957e-01 -1.52638137e-01
-5.58358058e-02 1.19367540e+00 5.09704947e-01 -7.24175870e-01
4.80697393e-01 8.01175162e-02 -1.22788382e+00 4.73090380e-01
1.01648247e+00 7.72465527e-01 3.43412697e-01 8.28905344e-01
-2.23226473e-01 5.23901880e-01 -5.81969619e-01 -1.50108665e-01
4.96930368e-02 -2.50364691e-01 -1.23504865e+00 -4.76873606e-01
2.75014281e-01 -5.10121346e-01 -7.31600702e-01 5.56187212e-01
6.16918266e-01 6.24185979e-01 6.93036079e-01 -4.92683530e-01
-1.01708889e+00 8.60444605e-01 -3.79185438e-01 5.42655587e-01
1.16970456e+00 -5.74506164e-01 1.58691418e+00 -8.08168292e-01
7.00134456e-01 1.19242871e+00 3.42993438e-02 3.02806020e-01
-1.06523836e+00 -3.89554501e-01 2.14693733e-02 9.61607993e-02
-1.67030561e+00 -2.42110521e-01 3.81109834e-01 1.69977367e-01
1.02423942e+00 6.17183149e-01 2.47064665e-01 7.48675048e-01
-3.64068270e-01 1.14744449e+00 4.47858810e-01 -5.24492979e-01
3.00865293e-01 -8.64057466e-02 4.37652498e-01 6.73822314e-02
1.00113325e-01 -1.53889939e-01 -4.35260057e-01 -8.30772758e-01
3.95861894e-01 5.97300716e-02 -3.80064666e-01 1.11220382e-01
-9.50568199e-01 9.95365441e-01 6.74202502e-01 3.94052029e-01
-5.78538597e-01 -1.02756016e-01 1.53829589e-01 5.63374758e-01
3.78679186e-01 1.32210016e+00 -6.11680746e-01 6.89525530e-03
-1.14947176e+00 7.36996472e-01 9.33293521e-01 8.55845332e-01
9.52767849e-01 -7.70979762e-01 -9.53424811e-01 1.23213446e+00
3.67170274e-01 8.22906137e-01 5.31653643e-01 -8.97356689e-01
5.19695699e-01 6.29113674e-01 3.76125365e-01 -7.47732103e-01
2.65471581e-02 -4.76918191e-01 -3.07257354e-01 -1.00392044e+00
-2.34229630e-03 1.21330835e-01 -8.99617732e-01 1.33683980e+00
6.93401620e-02 -1.00260004e-01 1.89709142e-01 8.95434737e-01
9.30514276e-01 9.24731791e-01 -5.06558605e-02 -1.15076244e-01
1.41450691e+00 -9.16242719e-01 -1.65464520e-01 -2.99490362e-01
8.51363242e-01 -1.04583502e+00 1.11046326e+00 3.16981226e-02
-1.02638006e+00 -2.50247300e-01 -5.30231535e-01 -3.11753809e-01
-4.26535010e-02 3.51686105e-02 3.09618205e-01 -1.16494820e-02
-1.23474646e+00 2.07853571e-01 -2.80727148e-01 -4.48828340e-01
2.28686798e-02 2.31515259e-01 -1.13792550e-02 -6.62231505e-01
-1.61903870e+00 5.88771522e-01 2.82055467e-01 -5.90193272e-01
-8.79667521e-01 -8.79792690e-01 -4.34098125e-01 3.98591131e-01
3.81592453e-01 -1.03358150e+00 1.44214201e+00 -1.06415659e-01
-1.03067791e+00 6.32739007e-01 -7.39574194e-01 -5.93562543e-01
-9.02214348e-02 -6.84551418e-01 -3.73863250e-01 5.68185568e-01
3.15348506e-01 7.36619115e-01 6.72441840e-01 -9.87057924e-01
-5.43036819e-01 -2.08702996e-01 4.40587014e-01 5.89484155e-01
-1.77057922e-01 1.14634231e-01 -9.46621776e-01 -4.90864515e-01
2.46254489e-01 -8.25198472e-01 -4.37748253e-01 -7.14391768e-01
-4.03206378e-01 -6.84489965e-01 4.59105939e-01 -5.68461001e-01
1.64393008e+00 -2.07600474e+00 7.68210143e-02 6.03302360e-01
3.39750022e-01 1.81200415e-01 -6.15633786e-01 8.69032264e-01
3.89482409e-01 1.15393177e-01 3.26357275e-01 8.25291127e-02
1.57503448e-02 2.47586407e-02 -8.48907828e-01 -1.62390739e-01
-2.68963903e-01 1.19031453e+00 -1.12199008e+00 -5.57984293e-01
-3.26793224e-01 2.24943995e-01 -8.54468048e-01 3.86788517e-01
-6.05936468e-01 1.78499833e-01 -1.04512870e+00 4.62545335e-01
2.88863987e-01 -7.34978437e-01 -2.25900725e-01 2.01002240e-01
5.85150838e-01 7.59166479e-01 -7.64921367e-01 1.66918349e+00
-4.03194189e-01 3.87595266e-01 -3.75088423e-01 -4.73456055e-01
9.43495572e-01 3.72668386e-01 5.87135673e-01 -1.21920300e+00
-3.58442724e-01 4.25606012e-01 -4.56144810e-01 -5.17481804e-01
1.14907956e+00 3.03061455e-01 -2.25908026e-01 7.36863911e-01
-2.39922553e-01 -1.33240759e-01 5.59442878e-01 9.28083777e-01
1.33881593e+00 -4.32631284e-01 -2.44504045e-04 1.03376821e-01
5.03102720e-01 1.12466410e-01 -5.75565845e-02 1.36272943e+00
2.67848790e-01 6.22816265e-01 1.11340582e-01 2.61856671e-02
-9.06145096e-01 -9.78264034e-01 -3.95510532e-02 1.93769932e+00
2.50055343e-01 -4.51954782e-01 -2.86613971e-01 -6.05240583e-01
3.41076165e-01 7.17709124e-01 -3.54312122e-01 -3.86246145e-01
-5.60464859e-01 -1.77580222e-01 5.77441216e-01 1.80072725e-01
5.93207479e-02 -1.02785850e+00 8.68151933e-02 1.56260937e-01
-6.58486605e-01 -7.85325110e-01 -8.15196931e-01 -1.31772831e-01
-7.92537689e-01 -8.91206503e-01 -1.37166333e+00 -9.18696404e-01
5.27938783e-01 6.00980282e-01 1.57709396e+00 3.13127697e-01
1.64797053e-01 5.73669612e-01 -1.06310725e+00 1.34181678e-01
-2.44062513e-01 8.01675320e-01 -2.04492077e-01 -5.06443501e-01
5.97305954e-01 -1.78399652e-01 -1.05853188e+00 2.94643670e-01
-1.13079202e+00 -7.10102737e-01 5.59724450e-01 8.05927157e-01
4.93436247e-01 -4.83731657e-01 8.87834847e-01 -6.68467343e-01
1.19998717e+00 -7.66304135e-01 -3.77603740e-01 7.72827506e-01
-7.59308219e-01 4.19184387e-01 1.82784617e-01 -4.85863596e-01
-6.37692273e-01 -5.74103534e-01 -3.53843063e-01 -2.36827359e-01
1.85419276e-01 7.79091835e-01 5.48592508e-01 1.16748035e-01
1.01061106e+00 4.67885494e-01 -5.22568882e-01 -6.51131034e-01
6.40924513e-01 7.01283455e-01 1.32678762e-01 -6.65590286e-01
6.96422040e-01 -1.31282918e-02 -6.02551401e-01 -6.82727039e-01
-9.40474451e-01 -1.36676204e+00 -6.16404861e-02 2.90802211e-01
4.76436853e-01 -1.09933031e+00 -2.40751520e-01 -2.75806129e-01
-9.98547614e-01 1.12009324e-01 -4.74916458e-01 3.91095042e-01
-1.13480188e-01 4.69696909e-01 -6.63986683e-01 -5.57957232e-01
-7.25285351e-01 -9.64371383e-01 1.59742928e+00 2.26596728e-01
-4.96043772e-01 -7.13808537e-01 4.58799422e-01 5.25553703e-01
5.87946951e-01 -7.72441804e-01 1.13833368e+00 -1.38313663e+00
-7.07374692e-01 -5.57197511e-01 -2.10199267e-01 -1.58264607e-01
-2.24107057e-01 -5.98588765e-01 -6.28395677e-01 -5.91620624e-01
-6.34546041e-01 -5.42693853e-01 1.15646267e+00 3.45846921e-01
7.34937847e-01 -5.07918954e-01 -5.68931520e-01 1.66491538e-01
1.17206895e+00 1.30882666e-01 8.30901444e-01 3.12387466e-01
1.41635418e-01 2.30059755e-04 6.51546717e-01 5.09179652e-01
2.98705757e-01 7.11947083e-01 -6.99771345e-02 1.06659718e-01
1.44196287e-01 -7.56540835e-01 1.26808584e-01 7.63653994e-01
4.68516260e-01 -6.22386932e-01 -6.02443874e-01 7.18456686e-01
-1.43111873e+00 -8.11332583e-01 4.39654499e-01 2.42689085e+00
8.93591940e-01 -2.05598339e-01 2.56206095e-01 -2.25492299e-01
3.86881441e-01 2.03181952e-01 -5.85146248e-01 2.37986585e-03
-1.04933299e-01 5.89325964e-01 2.25620106e-01 5.97694278e-01
-6.89659536e-01 1.19886839e+00 7.39729881e+00 1.04063082e+00
-7.73095548e-01 -2.93057650e-01 4.99436915e-01 -1.05837002e-01
-8.27991247e-01 9.14141238e-02 -1.00031447e+00 1.68296561e-01
8.04357886e-01 -6.27826452e-01 7.25033343e-01 7.01009691e-01
-1.66465506e-01 -8.32709298e-02 -1.11009359e+00 7.29823351e-01
1.28166676e-01 -1.22390521e+00 7.51495302e-01 -1.76336333e-01
8.32649767e-01 2.72760987e-01 -4.68401518e-03 7.83126175e-01
5.57524323e-01 -8.96438122e-01 6.71621412e-02 5.06226838e-01
4.40386713e-01 -4.73206073e-01 6.91696823e-01 4.04129922e-01
-9.46383893e-01 -1.47698849e-01 -5.65418363e-01 2.57341444e-01
2.01549798e-01 4.03728932e-01 -1.18699741e+00 4.59661603e-01
5.81227541e-01 1.87155530e-01 -5.32034576e-01 1.45608735e+00
6.97684139e-02 8.46606255e-01 -5.62263668e-01 -4.44462627e-01
3.92879665e-01 1.48272231e-01 7.28723884e-01 1.01653051e+00
3.61369073e-01 2.87392706e-01 2.72652656e-01 4.06071007e-01
-4.88989532e-01 4.96930867e-01 -4.10522640e-01 -2.39295304e-01
9.49208260e-01 7.21162438e-01 -1.06107228e-01 -5.71832418e-01
-6.16428889e-02 9.43664372e-01 2.69577980e-01 8.71133506e-01
-3.80123287e-01 -5.88454485e-01 3.42069238e-01 1.89591438e-01
4.88722444e-01 1.81712747e-01 3.25351208e-01 -1.17986560e+00
6.11233823e-02 -1.06082785e+00 9.19772148e-01 -9.02429700e-01
-1.28732049e+00 6.26860261e-01 4.61135134e-02 -1.09862578e+00
-6.97408974e-01 1.85284644e-01 -1.87764600e-01 1.02555990e+00
-1.50345623e+00 -5.11017978e-01 2.56833375e-01 3.99831414e-01
4.60081428e-01 -9.09803957e-02 9.68130171e-01 5.09101391e-01
7.37142339e-02 8.79165590e-01 5.06714106e-01 1.57720253e-01
6.87123001e-01 -9.78040516e-01 3.21841180e-01 4.19699252e-01
5.08729458e-01 9.31179166e-01 5.38448334e-01 -4.27814186e-01
-1.42591190e+00 -8.04636717e-01 1.33069110e+00 -4.32518125e-01
6.05108202e-01 2.67369032e-01 -1.15828192e+00 4.17579293e-01
1.62501574e-01 -4.32188660e-01 6.66865766e-01 4.55926508e-01
-5.73083878e-01 -1.28197238e-01 -9.60710347e-01 5.11458457e-01
7.33830810e-01 -9.16402400e-01 -1.14060783e+00 6.40508533e-01
1.20542514e+00 -3.94917339e-01 -8.69332492e-01 2.76717186e-01
4.52669382e-01 -2.69678921e-01 1.43748713e+00 -8.42881799e-01
1.71359211e-01 -1.04332812e-01 -4.20973420e-01 -1.19703853e+00
-6.02304757e-01 -6.92315936e-01 -3.01479250e-01 8.40393126e-01
8.46995890e-01 -5.44362307e-01 7.58878648e-01 4.81101215e-01
2.86751062e-01 -9.08992112e-01 -7.61677027e-01 -4.96403635e-01
2.21018091e-01 -1.72593534e-01 7.70541131e-01 3.95137399e-01
6.98031858e-02 7.49680459e-01 -4.46640328e-02 1.37031317e-01
1.03068337e-01 4.36542809e-01 5.88275135e-01 -1.10008776e+00
-4.11541164e-01 -4.05108899e-01 5.91027141e-02 -2.26443720e+00
-8.66496339e-02 -1.04811227e+00 1.29989997e-01 -1.74544501e+00
3.49246711e-01 -6.32298410e-01 -5.52230239e-01 7.99049363e-02
-5.58949888e-01 1.01438731e-01 5.89788780e-02 4.86238867e-01
-1.44420183e+00 5.81294477e-01 1.21460736e+00 -2.81486541e-01
-4.48590219e-01 2.15393186e-01 -1.15746021e+00 4.12761569e-02
2.43725732e-01 -5.18205762e-01 -5.48383892e-01 -6.81890130e-01
6.68143511e-01 4.79349583e-01 -7.70362793e-03 -6.24588728e-01
4.41797137e-01 1.15805715e-01 1.18892714e-01 -8.47655773e-01
1.97852001e-01 -4.51237351e-01 -3.64612728e-01 1.22630395e-01
-1.16993642e+00 2.53724068e-01 -1.04208000e-01 7.10283279e-01
-4.62503552e-01 -4.07008290e-01 2.14932010e-01 -1.44145504e-01
-6.57947302e-01 4.54296142e-01 -1.98345572e-01 7.79137015e-01
3.26786608e-01 2.03664482e-01 -3.19428116e-01 -8.66257608e-01
-5.96650958e-01 7.60392487e-01 8.52413103e-02 5.11433601e-01
7.24706292e-01 -1.24291873e+00 -1.03796208e+00 7.48930722e-02
5.33544123e-01 -3.71124484e-02 9.61392149e-02 3.41359019e-01
-1.82652965e-01 1.01564491e+00 8.79173875e-01 -4.70468640e-01
-1.03039122e+00 3.54433298e-01 1.97882056e-02 -9.83146966e-01
-3.30741435e-01 1.15884471e+00 -4.93927784e-02 -2.64508426e-01
3.23691875e-01 -2.24834532e-01 -3.69292498e-01 -1.16622061e-01
7.89611459e-01 4.13262278e-01 1.76705554e-01 -3.37220371e-01
-2.03595057e-01 4.51047659e-01 -5.34080684e-01 -3.04501891e-01
8.66938770e-01 -2.71429956e-01 -2.01914713e-01 -2.10412638e-03
1.53932846e+00 2.22088516e-01 -3.32828045e-01 -7.88827240e-01
1.73220530e-01 -4.12145376e-01 -1.09541424e-01 -8.96637976e-01
-4.65871453e-01 2.25031137e-01 2.24689871e-01 3.90417427e-01
1.17251027e+00 5.51241159e-01 1.31431031e+00 1.04381490e+00
5.17658472e-01 -7.62832582e-01 2.07035214e-01 1.03451252e+00
9.36397910e-01 -1.04975569e+00 -5.81429750e-02 7.13445544e-02
-4.54158813e-01 5.37709415e-01 3.18900734e-01 -5.84208146e-02
5.52443862e-01 -4.27257299e-01 5.09657823e-02 -2.98034310e-01
-9.91172314e-01 -3.83747607e-01 8.43458533e-01 2.07899228e-01
3.92107904e-01 -6.67486340e-02 -2.17127413e-01 1.26170099e-01
-3.28393936e-01 2.27768600e-01 -2.77481169e-01 6.04618192e-01
-8.55978727e-01 -1.01782787e+00 -3.49461168e-01 8.78198385e-01
-5.18006802e-01 -5.23938179e-01 -3.23458314e-01 2.63631344e-01
-6.87832296e-01 1.19802022e+00 -2.83266380e-02 -4.09101993e-01
2.54240960e-01 1.61983177e-01 6.43064007e-02 -6.88123107e-01
-4.75742072e-01 6.34831190e-02 1.19220383e-01 -3.95265460e-01
1.71319712e-02 -3.15025508e-01 -1.11424243e+00 -3.77325527e-02
-6.40755832e-01 1.15122890e+00 1.64321437e-03 9.56693053e-01
9.16315019e-01 6.14730269e-02 7.85045087e-01 -9.10542160e-02
-9.11067426e-01 -1.15829730e+00 -4.38619137e-01 6.02544546e-01
3.91337544e-01 -2.49144435e-01 -3.52333784e-01 -5.51445961e-01] | [11.504715919494629, 7.653847694396973] |
75350e10-369f-4466-9062-8d831bf01c3b | combining-strategic-learning-and-tactical | 1709.03480 | null | http://arxiv.org/abs/1709.03480v1 | http://arxiv.org/pdf/1709.03480v1.pdf | Combining Strategic Learning and Tactical Search in Real-Time Strategy Games | A commonly used technique for managing AI complexity in real-time strategy
(RTS) games is to use action and/or state abstractions. High-level abstractions
can often lead to good strategic decision making, but tactical decision quality
may suffer due to lost details. A competing method is to sample the search
space which often leads to good tactical performance in simple scenarios, but
poor high-level planning.
We propose to use a deep convolutional neural network (CNN) to select among a
limited set of abstract action choices, and to utilize the remaining
computation time for game tree search to improve low level tactics. The CNN is
trained by supervised learning on game states labelled by Puppet Search, a
strategic search algorithm that uses action abstractions. The network is then
used to select a script --- an abstract action --- to produce low level actions
for all units. Subsequently, the game tree search algorithm improves the
tactical actions of a subset of units using a limited view of the game state
only considering units close to opponent units.
Experiments in the microRTS game show that the combined algorithm results in
higher win-rates than either of its two independent components and other
state-of-the-art microRTS agents.
To the best of our knowledge, this is the first successful application of a
convolutional network to play a full RTS game on standard game maps, as
previous work has focused on sub-problems, such as combat, or on very small
maps. | ['Marius Stanescu', 'Michael Buro', 'Nicolas A. Barriga'] | 2017-09-11 | null | null | null | null | ['real-time-strategy-games'] | ['playing-games'] | [ 3.71467680e-01 2.62510896e-01 -1.33852765e-01 2.19865441e-02
-4.81119603e-01 -6.34041548e-01 6.15985334e-01 -1.36512548e-01
-7.33875096e-01 8.11328888e-01 2.39305589e-02 -4.72540855e-01
-2.75428981e-01 -1.02110076e+00 -2.36916736e-01 -4.15467620e-01
-4.60774988e-01 1.07552457e+00 7.06662714e-01 -1.15928698e+00
5.97101986e-01 5.94993055e-01 -1.74531221e+00 4.42079216e-01
6.09788775e-01 8.72865319e-01 1.93305507e-01 9.69171405e-01
-9.40154865e-02 1.28864467e+00 -8.82053614e-01 -1.88548416e-02
6.88804328e-01 -5.60532689e-01 -1.12409770e+00 -3.62173200e-01
-2.82960892e-01 -4.20273364e-01 6.32548556e-02 7.49420583e-01
4.75029349e-01 4.01488155e-01 1.93800256e-01 -1.03485084e+00
5.41521370e-01 9.87004519e-01 -9.11595300e-02 3.32663178e-01
4.69235778e-01 5.42294621e-01 9.56073940e-01 -6.65330589e-02
6.27325416e-01 1.23800468e+00 5.30970633e-01 6.46616936e-01
-1.14191890e+00 -7.12199986e-01 1.11818209e-01 7.54538998e-02
-1.11794126e+00 -1.31167471e-01 6.87179446e-01 -9.76346582e-02
1.29022217e+00 2.37811565e-01 1.11919367e+00 7.25772202e-01
3.64209324e-01 8.87190282e-01 1.33261025e+00 -2.37524837e-01
7.15600014e-01 -4.46327776e-01 -4.59916204e-01 7.00801671e-01
-2.56366462e-01 6.51287675e-01 -4.45027709e-01 -1.63730681e-01
9.40364182e-01 -5.00550449e-01 1.52771488e-01 -3.84424686e-01
-9.47237194e-01 9.77946043e-01 6.70047820e-01 4.74229902e-01
-8.25962245e-01 3.28232974e-01 4.89701837e-01 6.52846217e-01
1.87246334e-02 1.42561448e+00 -4.17492449e-01 -7.92731941e-01
-1.10355163e+00 8.70139003e-01 1.14440227e+00 2.07215026e-01
6.81871831e-01 2.68341124e-01 -8.48296210e-02 3.29649568e-01
-2.80273318e-01 -1.21767439e-01 1.48491502e-01 -1.17269397e+00
3.39790225e-01 6.87877238e-01 1.86454654e-01 -6.80017233e-01
-7.86226988e-01 -5.98000348e-01 -2.19270781e-01 1.15154529e+00
2.67784894e-01 -3.20520759e-01 -9.59606946e-01 1.60640812e+00
1.75296098e-01 1.96091719e-02 1.92952797e-01 1.01799703e+00
3.22775394e-01 4.70657498e-01 -1.96326412e-02 1.67860523e-01
1.07353294e+00 -7.21584439e-01 -2.46822774e-01 -7.40057945e-01
6.88536108e-01 -2.12449551e-01 7.04916596e-01 7.21590877e-01
-1.41955185e+00 -4.00965601e-01 -1.15629077e+00 5.45589089e-01
-4.11501467e-01 -2.30176508e-01 1.06783891e+00 5.43444991e-01
-1.21692908e+00 9.96623218e-01 -8.75415564e-01 -2.26085290e-01
3.64810675e-01 8.27398717e-01 -9.47053209e-02 1.95998192e-01
-1.40989137e+00 1.41637576e+00 8.19374204e-01 1.26335531e-01
-1.22515464e+00 -3.13350469e-01 -7.64676571e-01 1.89574689e-01
1.09044898e+00 -3.56918007e-01 1.63487828e+00 -1.02332139e+00
-2.05569553e+00 4.61905181e-01 5.62994778e-01 -7.64864802e-01
4.98971820e-01 7.30622979e-03 1.77681655e-01 6.03867415e-03
2.50247568e-01 8.57316971e-01 5.17589211e-01 -1.00491500e+00
-1.00392199e+00 -6.21278770e-02 1.01257956e+00 8.04791689e-01
3.99959087e-01 9.01900679e-02 -1.34724870e-01 -5.30970022e-02
-6.37859330e-02 -1.12171328e+00 -8.51951361e-01 -6.70492291e-01
-5.42942062e-02 -6.69372529e-02 2.33739644e-01 -3.07159603e-01
1.11658359e+00 -1.57447636e+00 8.38203788e-01 1.88801005e-01
7.94928819e-02 4.18142498e-01 -7.09963068e-02 5.68171203e-01
-1.16739407e-01 -6.69652298e-02 -9.56283212e-02 9.21238810e-02
-7.48195350e-02 3.65251571e-01 -1.80203080e-01 7.25834491e-03
3.90972048e-02 9.76429045e-01 -1.11196768e+00 -5.45987822e-02
2.68027604e-01 -3.39818716e-01 -7.19799280e-01 1.05773620e-01
-5.50985873e-01 2.91338176e-01 -7.48913884e-01 3.54635298e-01
2.37388164e-01 4.76465553e-01 2.58654296e-01 3.29055667e-01
-4.60263401e-01 7.13519394e-01 -1.18026686e+00 1.72699952e+00
-4.92507815e-01 4.42099571e-01 1.73420712e-01 -1.05404878e+00
7.10737228e-01 -1.20796973e-03 4.11541373e-01 -7.38492131e-01
5.03032088e-01 4.13144320e-01 6.32031024e-01 1.09925047e-01
6.41873300e-01 -2.67565906e-01 -7.35354722e-01 5.00839889e-01
-2.21635401e-02 -7.71564305e-01 4.01056916e-01 9.18109529e-03
1.44529307e+00 5.17753482e-01 5.01000762e-01 -2.49355525e-01
2.39608973e-01 6.80077732e-01 3.70854080e-01 1.17301989e+00
-9.95784253e-02 4.33819182e-02 9.81631339e-01 -9.19315696e-01
-7.66199529e-01 -6.17179334e-01 6.40016139e-01 1.23919868e+00
8.70732442e-02 -6.05612576e-01 -7.60061920e-01 -6.83833480e-01
-5.42190015e-01 1.06557930e+00 -8.94385993e-01 -3.91294926e-01
-9.66739416e-01 -3.63796175e-01 6.73750162e-01 3.93669903e-01
8.01940382e-01 -1.70357203e+00 -1.66262984e+00 6.63703084e-01
1.68555871e-01 -5.98818660e-01 9.27661136e-02 7.51211345e-01
-6.18610322e-01 -1.02897871e+00 -1.79992512e-01 -4.06395525e-01
2.94363033e-02 -1.84611380e-01 1.09378541e+00 2.09439009e-01
-8.98226202e-02 -9.79342833e-02 -4.69986528e-01 -4.42283273e-01
-5.47120512e-01 4.17141914e-01 -1.16285928e-01 -6.11866713e-01
1.98114872e-01 -4.88140255e-01 -2.55900800e-01 2.18966648e-01
-6.03050530e-01 5.25178611e-01 7.13250577e-01 9.71477747e-01
1.08231775e-01 4.01239902e-01 3.65529966e-04 -6.85756624e-01
1.06394958e+00 1.38320336e-02 -7.98359811e-01 -1.74696833e-01
-1.62178636e-01 3.57954204e-01 7.12458670e-01 -4.53644365e-01
-7.63868749e-01 -3.26856710e-02 -5.69262505e-02 -2.71055788e-01
-1.64802387e-01 7.74699807e-01 1.87636271e-01 -7.71664158e-02
9.49271739e-01 2.24940389e-01 6.59035072e-02 -4.92919376e-03
1.46122009e-01 2.02109888e-01 1.80247381e-01 -7.63208270e-01
6.63667977e-01 1.52910069e-01 1.96974069e-01 -3.93666565e-01
-5.64351022e-01 -6.84507415e-02 -5.62418640e-01 -3.40521604e-01
7.49873042e-01 -4.07040536e-01 -9.50867772e-01 3.99686962e-01
-8.70155632e-01 -1.02483654e+00 -4.60676640e-01 1.62894532e-01
-9.61055100e-01 -2.66142637e-01 -4.48616654e-01 -8.64333272e-01
3.84985246e-02 -1.58978391e+00 9.69881117e-01 4.14098710e-01
-4.33012158e-01 -5.42978346e-01 1.78847998e-01 8.28455016e-02
4.72698152e-01 3.06458890e-01 3.93652856e-01 -8.09962630e-01
-5.13387203e-01 -2.87091881e-01 4.38663125e-01 -2.09133893e-01
-1.34087607e-01 -3.67804796e-01 -4.76372957e-01 -1.53434068e-01
-1.82799593e-01 -5.87294400e-01 6.05761647e-01 4.35259700e-01
6.72480762e-01 -3.93141359e-02 -3.75498325e-01 3.96184146e-01
1.05731869e+00 6.89976633e-01 8.40797544e-01 8.86173725e-01
2.09623516e-01 7.11246312e-01 1.18057466e+00 4.20710891e-01
8.04608315e-03 1.00565934e+00 8.29102397e-01 -2.01668963e-02
2.27675274e-01 -1.19538650e-01 3.81599963e-01 -1.57817692e-01
-5.08933306e-01 -3.84722836e-03 -1.06262732e+00 2.67014027e-01
-1.96377420e+00 -1.28364873e+00 4.77681190e-01 1.94549692e+00
6.69126749e-01 8.04806769e-01 4.94457513e-01 1.45649344e-01
2.94617057e-01 3.78574848e-01 -4.45010573e-01 -9.84513700e-01
3.04111928e-01 8.14628541e-01 6.56454623e-01 6.09726131e-01
-9.85306323e-01 1.75430048e+00 5.96952581e+00 1.02458549e+00
-1.00104058e+00 -1.79817334e-01 2.46602669e-01 -4.30188209e-01
1.06679082e-01 2.66561121e-01 -5.05540907e-01 -8.88071768e-03
8.80678773e-01 -3.18851061e-02 9.24352169e-01 9.76971567e-01
8.73529911e-02 -4.21084404e-01 -8.55724692e-01 3.53787631e-01
-2.31703699e-01 -1.71324492e+00 -2.24728465e-01 2.47474357e-01
2.81245589e-01 -1.50697440e-01 -1.06667854e-01 9.19720471e-01
1.02949655e+00 -1.34761810e+00 9.61431026e-01 1.38126850e-01
6.21361375e-01 -1.26871777e+00 9.35582221e-01 6.51048779e-01
-1.27397037e+00 -4.36858743e-01 -1.52981490e-01 -7.04426944e-01
1.30134642e-01 -5.96576869e-01 -1.06831813e+00 5.05958676e-01
5.65433145e-01 2.06705377e-01 -2.71541536e-01 7.49121904e-01
-4.80941296e-01 3.27920824e-01 -4.97386366e-01 -5.69541931e-01
1.03659010e+00 -7.43313357e-02 7.12611735e-01 7.02477753e-01
-6.35102838e-02 5.60246825e-01 5.93064845e-01 7.48636603e-01
7.52545297e-01 -3.39458734e-01 -6.50256157e-01 -7.89728910e-02
2.90189236e-01 1.08759010e+00 -1.12435114e+00 -2.13278159e-01
2.38893092e-01 9.78023350e-01 4.85335976e-01 9.24531445e-02
-7.77167320e-01 -3.64938438e-01 6.93211675e-01 8.85480121e-02
3.11095178e-01 -1.99234188e-01 -3.08373898e-01 -7.15570092e-01
-5.44775665e-01 -1.31920826e+00 4.60903227e-01 -7.63447225e-01
-2.54288614e-01 9.61594462e-01 4.12693381e-01 -1.11873651e+00
-8.94890368e-01 -5.36597133e-01 -9.23197508e-01 1.02388132e+00
-8.74837875e-01 -9.05726612e-01 -4.32039239e-02 3.21056068e-01
8.10560882e-01 -6.19346738e-01 8.36812139e-01 -5.44951260e-01
-1.39652923e-01 1.07411571e-01 -7.12681472e-01 -5.34916855e-02
-1.17724352e-01 -1.32445908e+00 5.12252510e-01 7.45161712e-01
-2.50294417e-01 2.89174557e-01 1.00941420e+00 -7.70701349e-01
-1.12216246e+00 -3.10086459e-01 2.28460968e-01 -1.91549376e-01
6.06742859e-01 -3.29492539e-01 -3.69195908e-01 4.93250698e-01
1.55128002e-01 -6.12134933e-01 2.99849212e-01 2.17348367e-01
3.36203337e-01 2.11313277e-01 -9.19590592e-01 1.24084091e+00
1.05087912e+00 -1.71151221e-01 -9.08357620e-01 -6.42288402e-02
5.54236531e-01 -8.32512319e-01 -3.29700947e-01 2.16109440e-01
4.88999575e-01 -1.30993986e+00 9.48439181e-01 -8.14921141e-01
4.14333165e-01 -3.28749627e-01 1.11498639e-01 -1.76081097e+00
-5.00552118e-01 -9.29222822e-01 3.39855134e-01 3.42722684e-01
2.25028098e-01 -4.24211055e-01 1.17030728e+00 3.73093128e-01
-2.29849756e-01 -8.61247659e-01 -1.14587653e+00 -5.18372118e-01
2.04380434e-02 -5.63934922e-01 6.52104974e-01 3.97520393e-01
1.37846902e-01 3.91152382e-01 -3.12814921e-01 2.75707208e-02
3.69074881e-01 2.55255908e-01 9.28344488e-01 -1.20743215e+00
-4.69303906e-01 -8.83270919e-01 -4.05479759e-01 -8.05003762e-01
2.65758038e-01 -5.05823851e-01 3.36579978e-01 -1.55010009e+00
-4.54927206e-01 -5.96591532e-01 2.25232821e-02 7.63042927e-01
9.47148725e-02 -7.84606412e-02 4.08521086e-01 -7.83472359e-02
-5.56191742e-01 3.19940388e-01 1.34367394e+00 -2.56108642e-02
-6.65212572e-01 2.85802662e-01 -6.66875780e-01 7.12405503e-01
1.12145352e+00 -3.67229819e-01 -5.23376465e-01 1.76731532e-03
4.10102278e-01 4.88046885e-01 9.02876928e-02 -1.37767851e+00
3.21492881e-01 -6.59238458e-01 3.38583589e-02 -3.80834609e-01
6.15148962e-01 -6.10711277e-01 1.55566037e-01 1.06271636e+00
-5.09309947e-01 1.86224822e-02 4.92131442e-01 8.32592323e-02
-3.58956084e-02 -4.13486749e-01 5.10930002e-01 -6.73262477e-01
-1.03092790e+00 2.90474165e-02 -1.07149303e+00 -2.09278002e-01
1.11214912e+00 -7.58659601e-01 8.32534358e-02 -6.63682818e-01
-7.20882773e-01 4.03578788e-01 3.62129569e-01 2.19683245e-01
3.61598492e-01 -9.18880284e-01 -6.15402818e-01 7.06451479e-03
-2.47144729e-01 6.06667884e-02 -1.74617930e-03 6.37942493e-01
-8.87927413e-01 3.47798377e-01 -8.62080514e-01 -2.43236810e-01
-1.13012671e+00 3.71787459e-01 8.36142838e-01 -8.58100832e-01
-5.75246394e-01 9.08264160e-01 -1.99615490e-02 -5.69547951e-01
-7.53933340e-02 -2.61740059e-01 -5.69993317e-01 6.67147711e-02
4.59944487e-01 8.38609189e-02 -4.38565128e-02 -4.10785466e-01
-5.28901398e-01 1.97242677e-01 -5.84649853e-03 -7.03046143e-01
1.38873243e+00 4.70004976e-01 1.10045671e-01 9.48975682e-02
3.14301789e-01 -2.06925303e-01 -1.46648371e+00 1.48268715e-01
2.16222298e-03 -4.37111646e-01 1.63787425e-01 -9.87935543e-01
-6.94706142e-01 6.48355782e-01 1.63130745e-01 4.87652570e-01
1.22297955e+00 -2.36275509e-01 2.71717221e-01 5.05286574e-01
1.02667284e+00 -1.09476912e+00 2.28639707e-01 1.04067612e+00
8.73250782e-01 -7.14705229e-01 -1.10241041e-01 -7.10652769e-03
-1.07575011e+00 1.26940560e+00 1.03197193e+00 -4.79859233e-01
-6.24868758e-02 5.36301136e-01 -6.22932799e-02 -4.66125220e-01
-1.02167976e+00 -6.60363495e-01 -1.01721071e-01 6.95935190e-01
-1.70421898e-01 2.12223753e-02 -1.68031693e-01 4.75131303e-01
-7.52650976e-01 6.73034191e-02 6.40779495e-01 1.37049019e+00
-7.33004689e-01 -1.13596058e+00 -3.24725986e-01 4.75481212e-01
-2.39677429e-01 -7.80951530e-02 -7.01857984e-01 9.94899809e-01
2.15807691e-01 8.11459243e-01 1.30893320e-01 -7.11688936e-01
5.62215507e-01 -1.36266232e-01 7.03632712e-01 -8.97982657e-01
-1.32061768e+00 3.84727642e-02 6.05155706e-01 -1.18338335e+00
-1.06721260e-01 -7.06142068e-01 -1.35165119e+00 -4.94501293e-01
-8.98396689e-03 3.62230748e-01 3.45677167e-01 1.00490296e+00
-2.68895850e-02 9.07258630e-01 2.58382708e-01 -1.62407112e+00
-5.44088244e-01 -6.96477890e-01 -3.61927301e-01 -1.19284905e-01
-1.75840873e-02 -1.01251471e+00 -1.14599075e-02 -8.86144638e-01] | [3.5574915409088135, 1.475885272026062] |
964cd64b-4892-4a38-8849-9422a50240b0 | knowledge-distillation-for-detection | 2211.08071 | null | https://arxiv.org/abs/2211.08071v2 | https://arxiv.org/pdf/2211.08071v2.pdf | Knowledge Distillation for Detection Transformer with Consistent Distillation Points Sampling | DETR is a novel end-to-end transformer architecture object detector, which significantly outperforms classic detectors when scaling up the model size. In this paper, we focus on the compression of DETR with knowledge distillation. While knowledge distillation has been well-studied in classic detectors, there is a lack of researches on how to make it work effectively on DETR. We first provide experimental and theoretical analysis to point out that the main challenge in DETR distillation is the lack of consistent distillation points. Distillation points refer to the corresponding inputs of the predictions for student to mimic, and reliable distillation requires sufficient distillation points which are consistent between teacher and student. Based on this observation, we propose a general knowledge distillation paradigm for DETR(KD-DETR) with consistent distillation points sampling. Specifically, we decouple detection and distillation tasks by introducing a set of specialized object queries to construct distillation points. In this paradigm, we further propose a general-to-specific distillation points sampling strategy to explore the extensibility of KD-DETR. Extensive experiments on different DETR architectures with various scales of backbones and transformer layers validate the effectiveness and generalization of KD-DETR. KD-DETR boosts the performance of DAB-DETR with ResNet-18 and ResNet-50 backbone to 41.4$\%$, 45.7$\%$ mAP, respectively, which are 5.2$\%$, 3.5$\%$ higher than the baseline, and ResNet-50 even surpasses the teacher model by $2.2\%$. | ['Errui Ding', 'Junyu Han', 'Haocheng Feng', 'Gang Zhang', 'Wanping Zhang', 'Fukui Yang', 'Shengzhao Wen', 'Xin Li', 'Yu Wang'] | 2022-11-15 | null | null | null | null | ['general-knowledge'] | ['miscellaneous'] | [-3.51407170e-01 -8.09528977e-02 -1.41849175e-01 -2.42245778e-01
-6.28054321e-01 -3.50491792e-01 5.45441151e-01 -1.62083451e-02
-6.67657256e-01 5.12221456e-01 -2.48152450e-01 -2.35015005e-01
1.21108837e-01 -8.63086522e-01 -9.39213395e-01 -3.34794313e-01
2.60156929e-01 5.58831155e-01 7.90402174e-01 -1.87379763e-01
-3.44916910e-01 2.60477394e-01 -1.41252124e+00 3.37276846e-01
8.65103006e-01 1.16340983e+00 4.80932593e-01 2.81558365e-01
-9.64795649e-02 8.05148542e-01 -7.46293485e-01 -6.99738622e-01
3.69725108e-01 -2.25376338e-01 -5.27984619e-01 -4.76796627e-01
7.58401752e-01 -6.49192870e-01 -5.77094078e-01 1.27106380e+00
6.68795764e-01 9.08162370e-02 5.41839838e-01 -1.13424814e+00
-1.01779425e+00 1.23888457e+00 -5.43682098e-01 5.62710762e-01
-2.06181094e-01 3.30343723e-01 1.13193119e+00 -1.36824954e+00
3.04900497e-01 1.31705320e+00 6.84440315e-01 4.53717262e-01
-1.25943124e+00 -1.25181246e+00 2.89398521e-01 5.30394554e-01
-1.73727286e+00 -1.76230609e-01 4.56835896e-01 -2.41028175e-01
1.12521636e+00 2.33437687e-01 7.15659320e-01 9.64366257e-01
-4.18906927e-01 1.13824916e+00 9.34243202e-01 -2.55776882e-01
9.45397764e-02 3.40826660e-01 2.84767121e-01 7.52169609e-01
4.26597148e-01 -3.10867224e-02 -4.71202284e-01 1.47922739e-01
7.35643566e-01 3.10628057e-01 -1.90429226e-01 -1.45464405e-01
-9.40358043e-01 6.84191942e-01 8.46609533e-01 1.07196189e-01
-3.03736806e-01 2.94260621e-01 4.23544616e-01 3.13925117e-01
1.21525064e-01 2.55767643e-01 -3.98431927e-01 -5.33862077e-02
-1.12341559e+00 2.36726448e-01 5.57298303e-01 1.36076963e+00
7.88823128e-01 3.28429312e-01 -4.09137964e-01 9.00322855e-01
1.36612594e-01 6.49436057e-01 7.19128251e-01 -5.59526622e-01
4.95410591e-01 7.05491841e-01 -4.84105825e-01 -6.47298038e-01
9.41740796e-02 -9.01618540e-01 -9.08681929e-01 -1.55519709e-01
7.45554874e-03 -4.36120182e-02 -1.02098691e+00 1.84046197e+00
1.82440802e-01 5.57333767e-01 -1.32173961e-02 7.36920714e-01
1.06536412e+00 4.54476625e-01 2.60304242e-01 6.22641249e-03
1.58552527e+00 -1.10685706e+00 -2.25739688e-01 -3.36324900e-01
5.34140766e-01 -5.56751549e-01 1.14587021e+00 2.98957676e-01
-1.09236503e+00 -8.00287664e-01 -1.13248682e+00 -1.96596608e-01
-3.10782373e-01 2.04757139e-01 3.91190708e-01 2.38275930e-01
-8.29654157e-01 5.55867791e-01 -9.05966401e-01 -7.89842457e-02
4.52772617e-01 3.53755504e-01 6.68650540e-03 7.01226667e-02
-1.27384043e+00 7.74771869e-01 8.47432554e-01 -6.85936362e-02
-1.20683730e+00 -9.28455830e-01 -5.16319931e-01 4.80033517e-01
4.93448049e-01 -7.61032939e-01 1.76954937e+00 -4.31959271e-01
-1.08278775e+00 5.69673896e-01 1.57755256e-01 -9.77865636e-01
5.90017378e-01 -4.60444778e-01 -4.81412888e-01 -2.56376546e-02
2.23479569e-02 7.64171422e-01 7.75098264e-01 -7.19647169e-01
-9.82352078e-01 -2.18882963e-01 6.70233890e-02 1.02241538e-01
-6.94566071e-01 -3.55023175e-01 -7.87617743e-01 -6.52876556e-01
7.96978399e-02 -7.13134944e-01 -3.35872285e-02 -1.82394147e-01
-5.74774921e-01 -5.08937001e-01 9.88562286e-01 -2.62849420e-01
1.41261828e+00 -2.25725269e+00 -2.77999073e-01 -1.17269196e-01
8.05021167e-01 7.26223409e-01 -1.00731149e-01 6.08002543e-02
-1.83820203e-02 -1.47767946e-01 -4.77008000e-02 -4.44324464e-01
2.35701010e-01 2.48734668e-01 -6.25240922e-01 9.28222612e-02
8.15122798e-02 1.04461420e+00 -9.79731202e-01 -5.68493366e-01
1.63985014e-01 4.77509916e-01 -6.71708226e-01 8.48346725e-02
-4.06729817e-01 -3.45626056e-01 -5.60592115e-01 6.49110794e-01
8.82060885e-01 -3.20711941e-01 -1.61655024e-02 -3.88424754e-01
-4.06933129e-02 5.50548255e-01 -1.25696564e+00 1.51121497e+00
-1.62274823e-01 6.33674979e-01 -1.17295623e-01 -9.71594691e-01
9.23441827e-01 3.41840275e-02 2.69729108e-01 -7.09442616e-01
5.94309047e-02 3.58572751e-01 -7.33220354e-02 5.38044497e-02
6.97926164e-01 -1.60789534e-01 -3.49000469e-02 1.19377285e-01
4.12245691e-01 3.07481196e-02 8.90184641e-02 3.76200497e-01
1.04368639e+00 -1.66541204e-01 4.17093225e-02 -2.11996481e-01
3.71298969e-01 -1.35209084e-01 8.20677042e-01 9.39968288e-01
-2.26048231e-01 3.11490268e-01 2.76818097e-01 -2.83560187e-01
-6.89477026e-01 -1.33138919e+00 -2.64383972e-01 1.38959384e+00
2.01973334e-01 -7.77838051e-01 -7.79936016e-01 -8.82269382e-01
2.07003787e-01 7.24813044e-01 -2.33174920e-01 -3.22824448e-01
-6.54953718e-01 -7.42263258e-01 8.03548694e-01 8.72742355e-01
1.11935937e+00 -8.28304470e-01 -6.08457208e-01 2.55884767e-01
3.85337360e-02 -1.16377783e+00 -4.27666843e-01 4.07908827e-01
-8.71446371e-01 -8.63641143e-01 -4.74308044e-01 -7.86198497e-01
3.08872253e-01 5.09787679e-01 1.26901424e+00 -1.37443542e-01
-1.40709907e-01 8.59977864e-03 -3.83627675e-02 -5.81826806e-01
-1.27557635e-01 4.69554603e-01 1.73941165e-01 -5.07810831e-01
9.12531257e-01 -7.40939796e-01 -8.64145041e-01 3.62662524e-01
-8.69827926e-01 1.93281621e-01 9.91287529e-01 8.58445406e-01
9.12569225e-01 -1.61129665e-02 4.80854154e-01 -9.23870683e-01
5.99414706e-01 -4.72852141e-01 -6.66264951e-01 9.11015645e-02
-1.03373456e+00 2.57015377e-01 8.46684813e-01 -7.16145873e-01
-6.35053515e-01 -1.32219285e-01 -1.61414415e-01 -1.07709134e+00
3.08594316e-01 3.64524394e-01 7.03335330e-02 1.81938812e-01
9.52855647e-01 4.99278367e-01 -3.95628422e-01 -8.14251363e-01
6.40990794e-01 6.21641755e-01 7.51258075e-01 -6.62065327e-01
8.60711753e-01 3.18677127e-01 -4.61138755e-01 -4.79576558e-01
-8.04299831e-01 -2.84427196e-01 -1.20127581e-01 3.11172426e-01
3.55922759e-01 -1.35406053e+00 -7.90976882e-01 1.72821119e-01
-7.38413751e-01 -5.64802587e-02 -8.00224960e-01 4.86258715e-01
-1.85639843e-01 1.90017059e-01 -7.79952884e-01 -3.07320476e-01
-7.34609783e-01 -1.33343220e+00 7.92456508e-01 2.38898292e-01
2.50497818e-01 -4.03149098e-01 -1.63261041e-01 1.18401945e-01
5.83444715e-01 -1.91527218e-01 7.35054076e-01 -8.91836584e-01
-8.78674328e-01 -5.23869134e-02 -5.40352464e-01 7.03040421e-01
-3.50646347e-01 -3.63461077e-01 -1.15021503e+00 -3.31389338e-01
-1.66177619e-02 -3.54235709e-01 1.31262016e+00 -9.74064413e-03
1.40171790e+00 -3.88738483e-01 -4.16919917e-01 6.70201898e-01
1.35376084e+00 -1.50146997e-02 3.19626898e-01 1.81987450e-01
6.74556553e-01 -2.51231134e-01 1.83901131e-01 3.88690472e-01
5.58129430e-01 5.36503553e-01 6.37505949e-01 1.60811469e-02
-4.84072030e-01 -8.11830282e-01 4.49035108e-01 1.14655149e+00
3.23394388e-02 1.84914306e-01 -7.39093065e-01 6.62373900e-01
-1.70081663e+00 -9.42787409e-01 1.19965367e-01 2.07157159e+00
1.18567061e+00 6.03996038e-01 1.79752648e-01 -1.34263322e-01
7.14666963e-01 2.91311406e-02 -8.09642315e-01 -1.30265608e-01
1.71521846e-02 4.43406343e-01 6.70988739e-01 1.40207455e-01
-8.65606964e-01 1.27011681e+00 5.28793240e+00 1.42870390e+00
-1.25626731e+00 2.90429711e-01 4.22834158e-01 -3.18902850e-01
-7.35301301e-02 7.74305537e-02 -1.49697268e+00 5.53118706e-01
9.13046837e-01 -3.76977235e-01 2.87411183e-01 1.48999000e+00
-3.00102681e-01 2.79938072e-01 -1.40862489e+00 1.02624619e+00
-3.08191806e-01 -1.33039796e+00 1.95252612e-01 -1.47214308e-01
5.85497618e-01 4.27818835e-01 3.33999097e-01 1.21995449e+00
7.18053043e-01 -7.60469258e-01 8.89615357e-01 9.02629569e-02
8.71912897e-01 -6.38485551e-01 4.29071397e-01 5.88556826e-01
-1.46432340e+00 -7.61928707e-02 -6.82486236e-01 1.08734258e-01
-2.20102072e-01 6.89385533e-01 -1.28925443e+00 2.47464627e-01
8.96035016e-01 6.12261176e-01 -7.05815315e-01 9.46815968e-01
-5.15935004e-01 7.40738928e-01 -6.60325766e-01 -8.40745568e-02
3.16566885e-01 2.99233764e-01 3.57281059e-01 1.50070751e+00
3.08407605e-01 -1.04541205e-01 3.14809322e-01 1.15911293e+00
-7.04427600e-01 -1.49143681e-01 -1.74982250e-01 1.46739647e-01
9.61429775e-01 1.17911494e+00 -2.17820123e-01 -7.40176678e-01
-3.40173632e-01 8.25299084e-01 5.29842257e-01 1.97964862e-01
-1.13215113e+00 -4.94915307e-01 8.01852226e-01 3.03854078e-01
5.76421499e-01 -9.84838307e-02 1.32911175e-01 -1.25372899e+00
7.64120966e-02 -7.50305295e-01 6.09085381e-01 -5.68053305e-01
-1.17887700e+00 6.31426990e-01 3.15989137e-01 -1.11535788e+00
-2.09058568e-01 -3.58901680e-01 -5.22604287e-01 9.18861449e-01
-1.85662389e+00 -9.93505895e-01 -3.96059513e-01 7.08231747e-01
6.35939240e-01 1.62198190e-02 5.54427922e-01 4.58655447e-01
-7.42620885e-01 1.12637150e+00 1.76366225e-01 3.16449523e-01
6.04971826e-01 -1.24900913e+00 4.98551011e-01 9.60164905e-01
2.92422533e-01 8.34755003e-01 4.00054991e-01 -2.99082249e-01
-1.31844831e+00 -1.36504173e+00 5.64111054e-01 -2.06609443e-01
6.74777091e-01 -4.34727877e-01 -1.01241136e+00 8.77940476e-01
1.13188438e-01 3.25389057e-01 3.23260367e-01 6.37348741e-02
-7.88587570e-01 -5.54190993e-01 -1.13550735e+00 7.04339564e-01
1.11447835e+00 -6.24041557e-01 -8.40715408e-01 2.39430875e-01
1.02314603e+00 -4.17371541e-01 -8.41044903e-01 4.66404676e-01
3.13912570e-01 -9.62391198e-01 1.18071783e+00 -3.98229212e-01
1.67781994e-01 -3.35898757e-01 -1.63034230e-01 -1.06190932e+00
-5.57790160e-01 -5.27523398e-01 -7.36764193e-01 1.16792798e+00
3.39141458e-01 -6.72694147e-01 9.13451433e-01 2.29843736e-01
-3.21377784e-01 -8.54737401e-01 -9.65127826e-01 -1.08308780e+00
2.31330007e-01 -6.51188493e-01 8.83724213e-01 8.77968907e-01
-1.97216645e-01 5.37503660e-01 -3.11756786e-03 1.39785111e-01
5.64439535e-01 1.86206892e-01 6.15239978e-01 -1.06185710e+00
-7.61738002e-01 -5.93442082e-01 -4.27820176e-01 -1.79033363e+00
-1.62333995e-01 -1.07507837e+00 -2.63600320e-01 -1.31418490e+00
3.38714927e-01 -7.01036096e-01 -7.52498865e-01 8.02629769e-01
-1.90842405e-01 1.60224408e-01 4.12134498e-01 5.80520928e-01
-6.72530472e-01 6.74533963e-01 1.12819672e+00 -3.26339960e-01
-2.73957670e-01 -1.07856542e-01 -9.44383800e-01 7.31665611e-01
5.87391794e-01 -4.73902673e-01 -6.88484967e-01 -7.26941943e-01
5.04180836e-03 -3.00183177e-01 4.01206642e-01 -1.20041227e+00
5.54851532e-01 1.10608414e-01 3.26327384e-01 -7.57038116e-01
4.17193621e-01 -7.06411719e-01 -6.41201586e-02 4.56870884e-01
-1.60348132e-01 1.56089351e-01 3.54079515e-01 5.79237819e-01
-2.34025374e-01 -2.10381329e-01 8.29036057e-01 -2.81616300e-01
-1.02561057e+00 3.97470206e-01 2.32286185e-01 3.42153192e-01
1.10388339e+00 -2.18236253e-01 -4.48637098e-01 1.04647219e-01
-6.17685080e-01 3.08979303e-01 -4.50233594e-02 2.40393296e-01
7.25813091e-01 -1.41533780e+00 -6.95004404e-01 4.10764784e-01
9.49585661e-02 3.71708840e-01 1.41661540e-01 7.42832661e-01
-3.04255843e-01 4.68973845e-01 1.33245021e-01 -6.77303553e-01
-9.40329552e-01 6.23204589e-01 3.79894376e-01 -2.83598632e-01
-8.25762153e-01 1.24080193e+00 4.27467108e-01 -4.14660797e-02
4.52126175e-01 -8.30631733e-01 1.40835285e-01 -1.00526854e-01
5.07365227e-01 3.88288766e-01 1.87355995e-01 -2.00558022e-01
-2.44141266e-01 4.87022474e-02 -6.96672976e-01 1.37633383e-01
1.18901086e+00 3.60902518e-01 2.76527792e-01 4.96448800e-02
9.55054939e-01 -1.78125650e-01 -1.27074945e+00 -8.40875030e-01
-1.22351341e-01 -3.04354042e-01 2.09115133e-01 -6.24818802e-01
-1.31214833e+00 1.03562474e+00 6.86470449e-01 1.95451334e-01
1.31106579e+00 1.13593623e-01 1.02814114e+00 5.15581846e-01
3.10859501e-01 -9.28571880e-01 1.21426983e-02 6.36206686e-01
5.41550517e-01 -8.99273694e-01 -7.99170136e-02 -2.61496454e-01
-3.07701349e-01 6.92995369e-01 1.09939909e+00 -1.90703869e-01
6.13016665e-01 2.41666868e-01 -5.88514984e-01 -1.22219615e-01
-1.06796122e+00 -2.80376464e-01 1.43731609e-01 1.96493819e-01
2.15378404e-01 2.37557739e-01 -1.40889272e-01 8.71892691e-01
-3.88094932e-01 6.24958985e-02 1.33382261e-01 8.00524473e-01
-7.51900136e-01 -8.34154487e-01 -5.74307144e-02 4.84155357e-01
-4.12126154e-01 -4.90696520e-01 -4.58305748e-03 8.27257514e-01
4.74779606e-01 4.94579643e-01 -5.93333803e-02 -7.13910878e-01
6.45571470e-01 1.21558249e-01 1.99858338e-01 -7.36727953e-01
-8.03719163e-01 -8.40735510e-02 -2.96864331e-01 -3.58217448e-01
2.09651530e-01 -2.02759475e-01 -1.39251018e+00 -4.96622831e-01
-5.97237051e-01 2.81662405e-01 5.13074219e-01 5.24332821e-01
6.45175934e-01 5.33526599e-01 1.56621858e-01 -1.31776497e-01
-1.21916974e+00 -1.07149315e+00 -5.10022759e-01 2.66584516e-01
1.47671193e-01 -7.10312605e-01 -2.33626857e-01 -4.98549193e-02] | [9.258536338806152, 1.2587190866470337] |
f9751d14-9083-4164-8620-0a4a947448c1 | inference-on-extreme-quantiles-of-unobserved | 2210.08524 | null | https://arxiv.org/abs/2210.08524v3 | https://arxiv.org/pdf/2210.08524v3.pdf | Inference on Extreme Quantiles of Unobserved Individual Heterogeneity | We develop a methodology for conducting inference on extreme quantiles of unobserved individual heterogeneity (heterogeneous coefficients, heterogeneous treatment effects, etc.) in a panel data or meta-analysis setting. Inference in such settings is challenging: only noisy estimates of unobserved heterogeneity are available, and approximations based on the central limit theorem work poorly for extreme quantiles. For this situation, under weak assumptions we derive an extreme value theorem and an intermediate order theorem for noisy estimates and appropriate rate and moment conditions. Both theorems are then used to construct confidence intervals for extremal quantiles. The intervals are simple to construct and require no optimization. Inference based on the intermediate order theorem involves a novel self-normalized intermediate order theorem. In simulations, our extremal confidence intervals have favorable coverage properties in the tail. Our methodology is illustrated with an application to firm productivity in denser and less dense areas. | ['Vladislav Morozov'] | 2022-10-16 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 1.19185142e-01 2.48609826e-01 -8.46809447e-01 -2.72642702e-01
-8.02964568e-01 -5.65300941e-01 2.55953789e-01 5.58579206e-01
-1.72521159e-01 1.23859155e+00 2.27638558e-01 -6.04038477e-01
-6.15168154e-01 -7.35593677e-01 -5.76560378e-01 -6.13725364e-01
-2.35851243e-01 3.72922033e-01 -2.37511337e-01 2.37315908e-01
5.24472892e-01 2.65808076e-01 -1.35665512e+00 -2.26373523e-01
1.25792933e+00 7.16072023e-01 -3.69725496e-01 5.75795412e-01
-5.60178421e-02 4.05317068e-01 -5.69919348e-01 -4.07086700e-01
9.52064544e-02 -3.60015482e-01 -4.11787927e-01 -6.30737394e-02
3.06556702e-01 -3.40319782e-01 4.96766001e-01 1.13905811e+00
7.11913645e-01 -4.45277654e-02 1.30119300e+00 -1.68106782e+00
-7.45725870e-01 7.93497562e-01 -1.04388130e+00 2.42428347e-01
3.83625388e-01 -2.61433452e-01 8.70023191e-01 -5.97451031e-01
3.61940503e-01 1.31503356e+00 9.10748363e-01 -4.58472930e-02
-1.53657758e+00 -6.14815652e-01 5.18404227e-03 -5.66791117e-01
-1.36230648e+00 -3.98873329e-01 2.44068913e-02 -5.77907681e-01
4.50745940e-01 1.75153583e-01 2.25839019e-01 9.88799810e-01
5.86111486e-01 2.94180632e-01 1.02468944e+00 -4.80002165e-01
5.75057328e-01 2.42694065e-01 3.32710832e-01 3.48016769e-01
1.20297348e+00 5.60907647e-02 -8.20547566e-02 -8.28218818e-01
9.97664690e-01 2.65877426e-01 6.50092447e-03 -1.56199068e-01
-1.08599627e+00 1.15370595e+00 -1.63739726e-01 -6.66852966e-02
-7.06169367e-01 1.54390996e-02 3.90068829e-01 2.24377885e-01
7.55203962e-01 -2.89140195e-02 -3.22705030e-01 1.93149559e-02
-8.85539114e-01 5.23757279e-01 1.16083050e+00 1.01340783e+00
3.44557256e-01 -1.07996196e-01 -5.88399887e-01 6.29882216e-01
-1.29276486e-02 9.21381474e-01 -2.49146484e-02 -1.29181850e+00
6.96096420e-01 -4.48580831e-02 7.50501513e-01 -8.39875817e-01
-3.71324658e-01 -4.28753912e-01 -9.70705032e-01 -4.36986275e-02
9.59408462e-01 -6.14605010e-01 -5.79666793e-01 2.07780337e+00
1.86142191e-01 8.36565346e-02 -1.41910929e-02 3.01171660e-01
1.51701868e-01 5.26196301e-01 2.87876040e-01 -7.43003666e-01
1.35245979e+00 -1.84076205e-01 -9.00846124e-01 3.59604001e-01
7.07042396e-01 -3.49717051e-01 8.48012447e-01 1.94335982e-01
-1.44140005e+00 -2.23238602e-01 -4.46818292e-01 2.24182457e-01
-1.16830727e-03 -2.81192243e-01 4.32187229e-01 8.95979285e-01
-8.23554039e-01 4.83138859e-01 -3.83435220e-01 -1.35756016e-01
1.89091608e-01 3.56921107e-01 -4.32353914e-02 -3.46947014e-02
-9.48443890e-01 3.37775677e-01 -2.90981159e-02 6.51906803e-02
-3.37480396e-01 -1.12483370e+00 -8.47606421e-01 5.44899523e-01
2.10252866e-01 -1.05676901e+00 1.24095559e+00 -7.28065670e-01
-1.19635081e+00 5.65428257e-01 -2.23081544e-01 -2.01952577e-01
5.75182736e-01 2.27949828e-01 -2.25987270e-01 -1.29619436e-02
6.24600828e-01 -1.78820536e-01 4.07789111e-01 -1.03215158e+00
-6.00053072e-01 -5.34295619e-01 -2.21044481e-01 -1.31120667e-01
3.33496220e-02 -3.66377756e-02 4.25770193e-01 -4.42125857e-01
-1.05135385e-02 -5.70603073e-01 -4.85213786e-01 -5.99724054e-01
-5.25553644e-01 -3.18576604e-01 -9.06073451e-02 -3.80639881e-01
1.23379123e+00 -2.00763893e+00 -4.91821647e-01 6.25141203e-01
6.63036928e-02 -5.19550920e-01 1.20193325e-01 6.65924609e-01
-1.32351825e-02 4.36514229e-01 -1.79499686e-01 3.30124386e-02
3.94387424e-01 -2.75102019e-01 -1.67680323e-01 8.08465719e-01
4.21423428e-02 4.81336445e-01 -7.29462266e-01 -6.16229773e-01
-8.50353315e-02 -1.83499921e-02 -7.94023514e-01 -8.60218182e-02
3.07679415e-01 2.40536392e-01 -4.19692695e-01 6.27552807e-01
1.01884627e+00 -3.79314333e-01 2.49952078e-01 4.26374495e-01
-5.41880667e-01 -1.10556133e-01 -1.19284129e+00 7.12406099e-01
-6.60242319e-01 2.34712839e-01 1.30898699e-01 -1.24377954e+00
8.11748207e-01 4.05180842e-01 3.71548563e-01 -2.29280531e-01
2.39035636e-01 3.77664775e-01 -1.52184188e-01 -6.02436483e-01
2.54203498e-01 -7.62031317e-01 -4.83611882e-01 2.74550915e-01
-2.99242526e-01 1.65605560e-01 1.24463223e-01 1.45755447e-02
9.46781039e-01 -4.45142299e-01 5.38932085e-01 -8.52207661e-01
-5.26415883e-03 -3.02757651e-01 5.30645728e-01 1.32393241e+00
-1.04362585e-01 3.96536499e-01 1.33919835e+00 5.34652062e-02
-1.18022132e+00 -1.38900757e+00 -6.28804445e-01 7.18468487e-01
-1.52025163e-01 2.67644912e-01 -6.55070126e-01 -1.91512629e-01
4.79566723e-01 7.31006682e-01 -8.84349227e-01 5.13422973e-02
1.10027499e-01 -8.00353765e-01 2.86512166e-01 6.15088582e-01
3.41158986e-01 -4.53253090e-01 -3.32701176e-01 8.75021070e-02
4.03219163e-02 -7.84914732e-01 -2.06808776e-01 -6.87862486e-02
-9.40233052e-01 -8.20002437e-01 -1.25338900e+00 -4.38296795e-01
5.20068347e-01 -1.33472532e-02 1.16204309e+00 -2.79896438e-01
3.29793096e-01 3.28260124e-01 1.05587728e-01 -7.87195683e-01
-1.84638381e-01 -8.36274773e-02 9.33363885e-02 -1.34702995e-01
2.80883074e-01 -4.72450465e-01 -6.61242008e-01 1.72527507e-01
-6.71731710e-01 -7.03147829e-01 4.18263942e-01 9.17000234e-01
3.18363845e-01 2.04257309e-01 1.29884934e+00 -1.10243034e+00
8.02143514e-01 -1.10460544e+00 -9.39414740e-01 2.58229375e-01
-4.05302197e-01 3.30218226e-02 3.64301443e-01 -5.75267196e-01
-1.34163535e+00 -3.99692655e-01 2.79612035e-01 5.58441877e-02
-9.17453319e-02 7.83339560e-01 -1.79206088e-01 5.75987875e-01
5.61755359e-01 -5.95741332e-01 -6.44643754e-02 -2.77725667e-01
-5.08573726e-02 8.98871243e-01 5.55674255e-01 -9.29263949e-01
1.59687698e-01 5.84196448e-01 3.16547692e-01 -9.44200039e-01
-9.68104541e-01 -3.57452929e-01 -2.69413203e-01 2.28760034e-01
5.94500244e-01 -8.98528099e-01 -1.26469266e+00 -2.15562388e-01
-6.65684640e-01 -3.77743512e-01 -3.98668677e-01 9.98196483e-01
-9.12834764e-01 1.85076699e-01 -4.35173303e-01 -1.61460197e+00
1.68067947e-01 -9.14117396e-01 1.17770970e+00 4.52417821e-01
-1.79933906e-01 -1.47889531e+00 2.11237371e-01 -7.92408437e-02
1.10234559e-01 6.22380137e-01 9.53605592e-01 -4.84839559e-01
2.02617310e-02 -3.85813534e-01 -3.87767822e-01 -2.09949672e-01
5.07128872e-02 3.68416786e-01 -6.40433133e-01 -1.21414766e-01
-2.39550337e-01 9.75120515e-02 8.00534844e-01 1.50900042e+00
1.24397159e+00 -4.46669757e-01 -5.04292607e-01 2.62967259e-01
1.36619651e+00 -5.07660583e-02 5.06347239e-01 2.86185443e-02
4.74645942e-02 8.11269283e-01 5.21779716e-01 1.07348740e+00
2.92652488e-01 2.05838874e-01 7.11977929e-02 4.15356532e-02
7.97128260e-01 -9.44033116e-02 2.14619011e-01 2.55689919e-01
-2.60034978e-01 -4.12851483e-01 -7.81879306e-01 8.32712471e-01
-1.71964383e+00 -1.41824448e+00 -6.60251200e-01 2.91254735e+00
7.38757491e-01 5.19732051e-02 6.97178841e-01 -1.54312238e-01
1.26601052e+00 -5.76002121e-01 -4.22722399e-01 -7.32841849e-01
-9.66498181e-02 1.18080378e-01 8.96977484e-01 6.87444806e-01
-6.49122894e-01 8.32578242e-02 7.63838577e+00 6.35424316e-01
-3.80438358e-01 3.88552547e-02 8.20950747e-01 2.05249503e-01
-6.80141747e-01 2.30773702e-01 -8.65905583e-01 6.51704609e-01
1.41522324e+00 -5.89991391e-01 -3.32450211e-01 6.30391359e-01
5.63608468e-01 -5.92583299e-01 -9.14043546e-01 6.37325048e-01
-6.43648267e-01 -9.18514848e-01 -5.09023666e-01 7.05160618e-01
1.13364649e+00 -4.96579230e-01 4.27471071e-01 2.50732005e-01
7.38721490e-01 -9.40802932e-01 3.67972940e-01 5.64681947e-01
8.85614753e-01 -1.21097469e+00 1.04899526e+00 3.36655617e-01
-6.75706625e-01 -3.26870233e-01 -7.97921538e-01 -4.30057824e-01
1.03979602e-01 1.34022212e+00 -5.61147273e-01 4.54531848e-01
2.39500195e-01 1.44135922e-01 1.82137117e-02 1.25804400e+00
2.67415881e-01 6.62894845e-01 -4.74406272e-01 3.91458273e-02
1.54400051e-01 -2.99644142e-01 1.21872827e-01 1.23306334e+00
8.71970057e-01 1.78603411e-01 -2.56969154e-01 8.43751550e-01
-7.13504776e-02 1.69668794e-01 -8.62535000e-01 1.66198343e-01
7.67853856e-01 7.41591156e-01 -8.05419385e-01 -5.41526437e-01
-5.50100625e-01 3.49604130e-01 -3.97035778e-02 6.71204686e-01
-8.26998651e-01 -4.21667397e-01 3.92953634e-01 3.28863189e-02
1.94842115e-01 1.45333558e-01 -6.78146601e-01 -1.16725087e+00
-1.43686920e-01 -4.69390780e-01 8.06463301e-01 -2.51033962e-01
-1.51158452e+00 -3.55644345e-01 6.79947138e-01 -8.42989981e-01
-5.72018445e-01 -4.25707847e-01 -7.26132631e-01 1.07453072e+00
-1.05431819e+00 -4.50123131e-01 9.95764881e-02 3.48929852e-01
-3.33999395e-02 2.96302021e-01 5.59460878e-01 9.73172544e-04
-7.97866464e-01 5.95340788e-01 7.20650494e-01 -4.14912522e-01
4.32386786e-01 -1.36605525e+00 -1.82936147e-01 3.81076366e-01
-5.73560715e-01 6.50370717e-01 1.04034877e+00 -8.45434189e-01
-7.92261064e-01 -6.45358026e-01 9.07073796e-01 -2.38587350e-01
7.19255745e-01 -7.59960338e-02 -6.87475383e-01 7.59293139e-01
5.99191263e-02 -6.46793768e-02 9.24108207e-01 7.40840316e-01
-1.67374924e-01 2.01636240e-01 -1.37039936e+00 4.41933453e-01
7.09967911e-01 -1.13004990e-01 -6.45505339e-02 6.18579149e-01
3.93672317e-01 1.37118265e-01 -1.43761241e+00 3.39617282e-01
9.87835824e-01 -1.02367342e+00 7.45618343e-01 -8.93407345e-01
4.32576805e-01 2.90506214e-01 -3.42423439e-01 -1.19768894e+00
-3.87683898e-01 -6.49384260e-01 3.03028941e-01 1.33826017e+00
6.42595291e-01 -9.12338793e-01 3.59569997e-01 8.94594550e-01
2.39457712e-01 -5.03204048e-01 -7.96614528e-01 -1.14394915e+00
5.39087594e-01 -8.71650204e-02 7.34130323e-01 1.08101630e+00
2.67499357e-01 5.07147014e-02 -1.56653062e-01 3.47251862e-01
1.02315950e+00 3.38815629e-01 6.15163147e-01 -1.56776810e+00
-1.74568817e-01 -3.20702553e-01 -2.79204935e-01 -5.15419245e-01
4.15833563e-01 -1.90604061e-01 -2.04207018e-01 -1.20733821e+00
6.82499170e-01 -4.60327327e-01 -1.15456812e-01 -2.50648689e-02
-3.01236480e-01 -4.95564789e-02 -3.17756802e-01 -7.43335113e-02
-1.63175792e-01 2.97633082e-01 8.98266137e-01 2.19382957e-01
-3.41397047e-01 5.24855554e-01 -8.91835868e-01 7.74601042e-01
1.15580761e+00 -5.29644370e-01 -5.16320407e-01 2.50165999e-01
4.46129620e-01 7.83311844e-01 4.78780448e-01 -4.44964796e-01
-1.66621596e-01 -6.56419516e-01 4.60890770e-01 -6.35023892e-01
-7.68941715e-02 -4.03835654e-01 2.79308528e-01 4.11312401e-01
-4.71618176e-01 1.14785410e-01 -3.59257013e-02 5.63509703e-01
1.07390039e-01 -5.21886945e-01 6.62911117e-01 -3.56984418e-03
2.97336698e-01 1.91833451e-02 -5.14931619e-01 3.66573453e-01
8.68965864e-01 -2.11414114e-01 -4.15975630e-01 -8.42618048e-01
-8.53696644e-01 3.92824978e-01 3.76589239e-01 -3.98713857e-01
1.39777914e-01 -1.31519306e+00 -1.02009535e+00 -7.22704455e-02
-7.76448995e-02 -4.15908188e-01 4.82583672e-01 1.37245357e+00
-1.00916564e-01 4.01959866e-01 -8.35086703e-02 -5.17553806e-01
-6.48758352e-01 6.89950287e-01 1.13552161e-01 -2.02610061e-01
-1.69629112e-01 3.46193820e-01 6.91287220e-01 -1.29910871e-01
-1.32211626e-01 -4.66266721e-01 1.77576527e-01 1.83877289e-01
5.98923624e-01 8.51902485e-01 -2.06050962e-01 -2.49776855e-01
-2.25856394e-01 2.82919258e-01 3.74093503e-01 -4.59893227e-01
1.22188520e+00 -6.06586874e-01 2.07928512e-02 7.81800210e-01
1.27066851e+00 2.87399441e-01 -1.12414372e+00 1.85907841e-01
7.05630854e-02 -4.00781214e-01 -1.58139482e-01 -2.72188634e-01
-5.60755551e-01 6.85615480e-01 1.26559779e-01 9.46790755e-01
9.31354403e-01 -2.40970892e-03 8.25038273e-03 7.24704787e-02
5.11240512e-02 -9.96901929e-01 -5.53035676e-01 5.12298606e-02
5.55108368e-01 -1.09687555e+00 1.56843618e-01 -5.09118021e-01
-2.99486905e-01 8.77447009e-01 5.53463679e-03 -8.63464847e-02
8.19530070e-01 3.28585058e-01 -4.20373321e-01 -4.63535488e-02
-7.65056729e-01 -1.64531425e-01 -1.17609680e-01 6.44652545e-01
6.97430253e-01 3.28949839e-01 -7.86929607e-01 4.92335439e-01
-2.87268758e-01 -7.52312467e-02 9.00689662e-01 5.93392789e-01
-4.80018795e-01 -5.24524450e-01 -6.70925260e-01 8.17714214e-01
-1.00606787e+00 -2.47984752e-02 1.00021087e-01 1.10545766e+00
-3.44299614e-01 1.12687159e+00 4.87806171e-01 4.23307300e-01
3.16728711e-01 -2.07637418e-02 4.41420794e-01 -3.67184550e-01
-7.46211316e-03 3.64774287e-01 1.51607573e-01 -1.65771052e-01
-4.77911323e-01 -1.05130780e+00 -8.18231881e-01 -9.36437488e-01
-6.28067195e-01 5.08324742e-01 4.41299707e-01 7.54488885e-01
1.50628537e-01 1.07644148e-01 8.21534514e-01 -7.27340758e-01
-1.05632567e+00 -9.17197406e-01 -1.28289187e+00 -5.96231967e-02
4.25572067e-01 -8.40336442e-01 -6.51997447e-01 -2.06746042e-01] | [7.619297027587891, 4.82524299621582] |
8562eabe-23eb-4782-ba4c-db0564b84947 | tribe-or-not-critical-inspection-of-group | 2303.09664 | null | https://arxiv.org/abs/2303.09664v1 | https://arxiv.org/pdf/2303.09664v1.pdf | Tribe or Not? Critical Inspection of Group Differences Using TribalGram | With the rise of AI and data mining techniques, group profiling and group-level analysis have been increasingly used in many domains including policy making and direct marketing. In some cases, the statistics extracted from data may provide insights to a group's shared characteristics; in others, the group-level analysis can lead to problems including stereotyping and systematic oppression. How can analytic tools facilitate a more conscientious process in group analysis? In this work, we identify a set of accountable group analytics design guidelines to explicate the needs for group differentiation and preventing overgeneralization of a group. Following the design guidelines, we develop TribalGram, a visual analytic suite that leverages interpretable machine learning algorithms and visualization to offer inference assessment, model explanation, data corroboration, and sense-making. Through the interviews with domain experts, we showcase how our design and tools can bring a richer understanding of "groups" mined from the data. | ['Rebecca Hwa', 'Wen-Ting Chung', 'Yu-Ru Lin', 'Muheng Yan', 'Yongsu Ahn'] | 2023-03-16 | null | null | null | null | ['interpretable-machine-learning', 'marketing'] | ['methodology', 'miscellaneous'] | [ 2.11027607e-01 7.44336009e-01 -6.55868709e-01 -7.18298554e-01
1.33145601e-01 -3.00713897e-01 2.48091981e-01 1.02635276e+00
2.84620225e-02 2.36980781e-01 8.58640611e-01 -8.54576468e-01
-5.92908263e-01 -4.79799628e-01 1.20021641e-01 -2.39919037e-01
-5.31732328e-02 3.30881655e-01 -5.39196432e-01 2.05888879e-02
8.26326489e-01 3.65700692e-01 -1.48333848e+00 4.40758973e-01
1.18418527e+00 7.98406422e-01 -4.50583667e-01 -1.43082470e-01
-2.07115218e-01 1.18985474e+00 -6.79232419e-01 -4.50326204e-01
1.14309482e-01 -2.93907374e-01 -8.31600547e-01 3.21482986e-01
1.28552541e-01 -3.24732095e-01 7.89941370e-01 1.19278634e+00
5.63597046e-02 -1.25063598e-01 4.76005584e-01 -1.33846021e+00
-6.64735377e-01 1.11918688e+00 -6.54770970e-01 2.99007725e-02
4.98925090e-01 2.19455704e-01 9.94240880e-01 -2.36402676e-01
6.94576383e-01 1.79483080e+00 5.43377817e-01 1.50143892e-01
-1.46498525e+00 -1.14972031e+00 5.25339723e-01 -1.12452060e-01
-1.16469693e+00 -3.07305694e-01 7.02191114e-01 -1.03169525e+00
3.65953028e-01 5.15074193e-01 9.60093260e-01 7.11922109e-01
-1.17504328e-01 3.96708041e-01 1.51242995e+00 -2.75012523e-01
5.14471114e-01 4.68649477e-01 4.76020694e-01 3.72629195e-01
7.84664094e-01 -1.52500421e-01 -5.49834371e-01 -4.80481267e-01
5.25819361e-01 3.24740976e-01 1.49635106e-01 -8.21070522e-02
-9.84062791e-01 1.15526295e+00 3.21570963e-01 1.43284589e-01
-5.18367052e-01 -3.44259799e-01 2.98579961e-01 2.01638654e-01
7.88499296e-01 9.99042571e-01 3.04392409e-02 -2.10547686e-01
-7.95246363e-01 4.52607423e-01 9.11804497e-01 3.64511430e-01
8.23106527e-01 -1.95195526e-01 1.04870260e-01 5.23445308e-01
6.17847919e-01 -5.92712313e-04 2.24382371e-01 -1.10334170e+00
1.41126975e-01 1.47104692e+00 2.63286382e-01 -1.41971457e+00
-5.36762118e-01 -3.72581452e-01 -3.01003367e-01 5.92197299e-01
5.57093859e-01 -2.15410575e-01 -6.43440068e-01 1.37859058e+00
3.72982770e-01 -5.04312754e-01 -2.91585445e-01 7.61880994e-01
4.64230090e-01 -2.12471798e-01 5.93243361e-01 -1.14807270e-01
1.24882877e+00 -5.39206453e-02 -8.86914730e-01 -2.62010783e-01
9.36499417e-01 -4.29283708e-01 1.07434094e+00 5.39579570e-01
-8.53720844e-01 -2.36120939e-01 -8.44503462e-01 1.14783928e-01
-3.50608021e-01 -5.07713079e-01 9.83895183e-01 9.26187098e-01
-5.06359816e-01 7.37816870e-01 -7.06012130e-01 -6.63790703e-01
7.18044281e-01 2.13027835e-01 1.84112694e-02 2.40269080e-01
-7.34906316e-01 5.73591173e-01 2.33070076e-01 -9.03327540e-02
-1.29151285e-01 -9.09518063e-01 -5.00015080e-01 4.58487757e-02
6.06802464e-01 -4.95198697e-01 8.93009007e-01 -1.24209023e+00
-7.35839665e-01 8.48853052e-01 2.34269843e-01 -2.28307068e-01
4.50869262e-01 -2.03716457e-01 -4.31528747e-01 -3.01778376e-01
2.18172908e-01 2.19891280e-01 4.02438849e-01 -1.17719686e+00
-6.05201781e-01 -8.55519176e-01 -2.53268890e-02 -9.06310976e-02
-2.64455736e-01 6.23497903e-01 9.14745748e-01 -3.00237477e-01
3.02642077e-01 -3.84895235e-01 -4.52771634e-01 -1.44920349e-01
-5.52424490e-01 -1.83181286e-01 6.12209797e-01 -6.23284757e-01
1.81645215e+00 -2.18085599e+00 -3.01343381e-01 7.21685410e-01
7.87931919e-01 -4.48723473e-02 6.55963480e-01 7.67839968e-01
4.75556478e-02 8.69028330e-01 4.63790148e-01 -8.86910409e-02
3.38280439e-01 -8.88771415e-02 -3.39817494e-01 1.33676574e-01
-1.96307283e-02 3.15645397e-01 -9.78984594e-01 -5.09423554e-01
5.20436391e-02 -2.35716298e-01 -7.28990793e-01 -4.33156546e-03
-2.55471058e-02 4.18930411e-01 -5.38019717e-01 7.82348692e-01
3.75707507e-01 -4.79918629e-01 7.56424367e-01 8.83418843e-02
-4.83021557e-01 4.78970468e-01 -9.98837471e-01 6.73119485e-01
1.66513517e-01 4.50530767e-01 4.15689260e-01 -5.77844441e-01
1.27256107e+00 -2.86207527e-01 2.08176330e-01 -4.06548530e-01
2.11882338e-01 1.90386161e-01 6.69126213e-01 -5.27044177e-01
5.26331604e-01 -7.00484589e-03 3.51977460e-02 1.05530417e+00
-6.25979066e-01 1.31736651e-01 -1.39498562e-01 1.68983594e-01
7.23241448e-01 -3.80652636e-01 7.12496042e-01 -6.08578920e-01
-2.85978556e-01 3.12018245e-01 7.97883928e-01 4.80567902e-01
-2.43259132e-01 -1.87859327e-01 7.56067276e-01 -5.61186612e-01
-7.86527693e-01 -5.35511732e-01 8.82790908e-02 1.10632193e+00
-3.97506841e-02 -7.81809330e-01 -5.36352754e-01 -4.20523971e-01
4.08460230e-01 8.57477605e-01 -7.11840510e-01 -7.13879019e-02
-3.12055834e-02 -3.91229421e-01 -1.96009711e-03 3.06329489e-01
4.12494928e-01 -7.08223820e-01 -1.00488150e+00 5.35341911e-02
2.75279552e-01 -5.63245952e-01 1.06270872e-01 -2.49782547e-01
-8.10829341e-01 -1.15187132e+00 4.50898230e-01 -2.82719582e-01
8.36766839e-01 2.64335394e-01 9.21965182e-01 4.22337085e-01
-3.10983211e-02 3.07587795e-02 -4.23267335e-01 -1.24763453e+00
-7.91207731e-01 -1.39859140e-01 8.38876814e-02 -9.34873987e-03
1.03963029e+00 -8.49778593e-01 -5.55532396e-01 4.33589131e-01
-6.11663818e-01 4.40758377e-01 2.66468495e-01 1.35434121e-01
-4.18606512e-02 -1.03716753e-01 6.32077038e-01 -1.19315219e+00
1.24401224e+00 -7.48883188e-01 -2.26402596e-01 1.30407766e-01
-1.17489171e+00 -3.52312863e-01 2.68450022e-01 -5.19823670e-01
-6.89273596e-01 -7.67744124e-01 6.19720876e-01 -1.73204631e-01
-1.66257322e-01 8.69091272e-01 7.00302497e-02 2.50559479e-01
9.80959713e-01 -6.84679627e-01 6.75468862e-01 -5.73767483e-01
7.34144300e-02 8.07311296e-01 3.44887041e-02 -5.82369566e-01
4.56125528e-01 5.51285505e-01 -4.06401336e-01 -4.84150887e-01
-7.59926856e-01 -2.39509717e-01 -5.35557687e-01 -5.64988971e-01
6.02267265e-01 -7.15583861e-01 -1.30527842e+00 -2.01832652e-01
-4.21438187e-01 -5.72160959e-01 -3.18543345e-01 4.71928954e-01
-1.70399249e-01 -5.25197722e-02 1.66229263e-01 -1.07054484e+00
-1.13250643e-01 -7.28732646e-01 4.24021930e-01 4.53970730e-01
-1.41627681e+00 -1.09292245e+00 -2.95070648e-01 7.27280378e-01
4.06356961e-01 8.29184711e-01 1.20863104e+00 -1.33889222e+00
-4.58224118e-01 -1.00435689e-01 3.57247354e-03 -1.33944768e-02
3.64711910e-01 3.21534634e-01 -6.71302438e-01 1.89261585e-02
-1.21000499e-01 -2.26725012e-01 -3.76565754e-02 1.35563672e-01
1.27425325e+00 -9.04358089e-01 -4.32718188e-01 2.41103888e-01
8.03938329e-01 5.00457823e-01 2.54177541e-01 4.83551294e-01
5.91692805e-01 1.46186471e+00 8.26648474e-01 8.58469844e-01
4.11893606e-01 2.18996525e-01 2.85096020e-01 -1.84277073e-01
5.44436097e-01 -4.93779778e-01 4.35004421e-02 1.45500317e-01
-1.87346086e-01 4.66592193e-01 -1.25446951e+00 2.29278058e-01
-2.02928543e+00 -9.93613899e-01 -3.89686897e-02 2.11265707e+00
3.89238268e-01 2.75814742e-01 6.13562167e-01 1.79701656e-01
6.06041133e-01 1.63394362e-01 -4.66296971e-01 -8.82987857e-01
4.91278589e-01 -2.87809610e-01 2.77534008e-01 3.83705735e-01
-3.58836681e-01 6.12615764e-01 6.97956419e+00 2.19084740e-01
-9.47135031e-01 -3.97145927e-01 8.97537470e-01 -9.90201235e-02
-8.52174461e-01 2.97691494e-01 -6.21636450e-01 4.03615475e-01
5.95318794e-01 -6.43360913e-01 1.99156925e-01 8.42372894e-01
7.84008145e-01 -2.31269181e-01 -1.37974656e+00 5.99197686e-01
-2.98469454e-01 -1.42058277e+00 -1.58672705e-01 7.71627307e-01
3.85468870e-01 -5.94694078e-01 5.30174747e-02 -2.23768707e-02
1.01034737e+00 -1.32588720e+00 6.96747839e-01 2.83642262e-01
4.26368922e-01 -6.34290159e-01 3.96815330e-01 3.32144380e-01
-2.66366005e-01 -6.53449237e-01 -4.96462360e-02 -1.01393509e+00
4.15284932e-03 5.46601355e-01 -1.33518028e+00 2.31681824e-01
4.45583135e-01 6.08048856e-01 -5.37056565e-01 4.99931782e-01
-5.83843365e-02 6.70712411e-01 -1.39350789e-02 -1.61203638e-01
-1.49482965e-01 -5.24755001e-01 4.46541399e-01 7.40631104e-01
6.15663975e-02 1.01383127e-01 4.06806499e-01 1.27767026e+00
4.41294134e-01 1.32838950e-01 -5.83733022e-01 -5.92658639e-01
1.04236722e+00 1.34261656e+00 -7.45565772e-01 -2.91391641e-01
-1.83622748e-01 -1.45838290e-01 8.43013227e-02 4.07032460e-01
-3.08589227e-02 -5.54581173e-02 1.15516651e+00 8.63958955e-01
-3.96051466e-01 -1.68329835e-01 -9.04636681e-01 -7.23832846e-01
-5.60061276e-01 -1.41459441e+00 5.06985605e-01 -4.64197576e-01
-1.23666358e+00 -1.90971538e-01 2.97073483e-01 -6.57396853e-01
-4.74096835e-01 -3.65018427e-01 -8.43418121e-01 7.07812428e-01
-4.91137803e-01 -1.05076444e+00 -4.34363723e-01 3.99727710e-02
-1.35928735e-01 -8.03236067e-02 6.94540322e-01 -1.56449988e-01
-6.55977130e-01 3.65167975e-01 -2.41477653e-01 -1.70384310e-02
6.59674287e-01 -1.11035132e+00 2.20359445e-01 4.72637385e-01
-4.89449739e-01 1.06354833e+00 1.00182438e+00 -9.98193562e-01
-8.51495981e-01 -5.18380702e-01 6.96038604e-01 -3.13319176e-01
1.02601504e+00 -5.80256820e-01 -1.01920617e+00 8.51718485e-01
-9.19814706e-02 -7.23022461e-01 1.52236891e+00 8.47945631e-01
-3.06696743e-01 -8.88363868e-02 -1.45443451e+00 8.44483316e-01
1.16789091e+00 -4.58083034e-01 -4.83708739e-01 7.43351504e-02
5.21478832e-01 5.49806692e-02 -1.22830403e+00 -9.64446813e-02
6.90685689e-01 -1.37385106e+00 4.97025400e-01 -7.99611330e-01
3.13981473e-01 -5.23022227e-02 3.11467070e-02 -1.12411606e+00
-4.64278609e-01 -1.03835785e+00 3.39965135e-01 1.53650975e+00
1.24285229e-01 -8.52768660e-01 6.87935233e-01 1.52811289e+00
1.13373056e-01 -5.97515941e-01 -6.46283105e-02 -2.64662892e-01
-1.13804288e-01 -2.94727743e-01 1.16368973e+00 1.48272324e+00
6.45206094e-01 9.33557749e-03 9.22695026e-02 -5.84411174e-02
5.93791008e-01 2.50364482e-01 1.12549496e+00 -1.83003950e+00
-5.74940294e-02 -4.84636545e-01 -5.01782835e-01 -1.19976610e-01
-1.53514430e-01 -6.24944866e-01 -7.88154542e-01 -1.55453575e+00
1.89076260e-01 -5.94918847e-01 1.66790545e-01 5.46049297e-01
-8.13933183e-03 -4.80556339e-01 3.65786374e-01 3.24188590e-01
-3.20320845e-01 -2.47927994e-01 9.50562418e-01 3.68283451e-01
-7.69080102e-01 -1.69027254e-01 -1.85795414e+00 1.03159022e+00
8.45187664e-01 -2.66211599e-01 -5.21187961e-01 7.17969835e-02
7.80229151e-01 -4.33745503e-01 4.95228708e-01 -6.58551216e-01
2.49883998e-02 -9.46172297e-01 3.51500869e-01 -2.04852134e-01
-2.21784994e-01 -6.78620815e-01 6.98587298e-01 5.24823785e-01
-5.35512745e-01 -7.56181702e-02 -1.06504612e-01 1.21191055e-01
-2.34605675e-03 1.65585786e-01 4.03173745e-01 -5.06547615e-02
-3.67970794e-01 -1.86681300e-01 -5.25301814e-01 -2.26925552e-01
1.04478097e+00 -6.00755453e-01 -7.47744560e-01 -6.82609141e-01
-7.34159887e-01 6.04686260e-01 9.46066916e-01 3.56578678e-01
1.78556487e-01 -1.03691781e+00 -5.43138623e-01 3.80366623e-01
6.64782152e-02 2.63836104e-02 -1.75551325e-02 9.19824719e-01
-1.42236486e-01 -1.09667279e-01 -4.74953383e-01 -3.39830697e-01
-1.42621815e+00 3.64678890e-01 4.27279957e-02 1.42668471e-01
-3.93323660e-01 2.97427982e-01 3.55733275e-01 5.68099245e-02
1.24538839e-02 -3.72541368e-01 -4.84740406e-01 5.98750174e-01
7.27087140e-01 8.88543308e-01 -5.71948767e-01 -2.76818156e-01
-3.82882237e-01 4.69112583e-02 -3.57126981e-01 -1.31664813e-01
1.28806853e+00 -3.01509142e-01 -2.25441769e-01 7.00955272e-01
5.20377040e-01 2.08795875e-01 -1.13127398e+00 7.29700699e-02
4.98000383e-01 -9.36218083e-01 -4.25335199e-01 -9.02329504e-01
-2.51847267e-01 7.26569772e-01 -5.07646278e-02 1.03294051e+00
8.61512423e-01 5.96223734e-02 -2.01704979e-01 -1.40504077e-01
1.89256340e-01 -1.30834830e+00 -3.02157621e-03 -1.68346971e-01
8.86913538e-01 -1.11395633e+00 5.04476905e-01 -6.01351798e-01
-8.79136801e-01 9.43176925e-01 6.27435386e-01 1.19952038e-01
4.67385590e-01 5.12924157e-02 3.20245594e-01 -6.41723692e-01
-7.51842439e-01 1.29788369e-01 -7.23653883e-02 8.14536989e-01
6.47808433e-01 4.78702933e-01 -4.96063918e-01 7.90694416e-01
-6.50865138e-01 5.52618690e-02 6.10749424e-01 7.06845164e-01
-7.73373127e-01 -9.88210559e-01 -6.96215212e-01 7.78214216e-01
-6.75458074e-01 2.77553469e-01 -1.11985803e+00 1.06191623e+00
1.49502382e-01 1.20972276e+00 4.29780781e-01 -6.47317886e-01
7.30814040e-02 -2.04391647e-02 -6.65865764e-02 -6.91430628e-01
-8.93592656e-01 9.16357487e-02 7.28030086e-01 -6.47774816e-01
-5.02374887e-01 -9.23618674e-01 -1.07548845e+00 -7.80750513e-01
4.88326401e-02 1.39182061e-01 5.73710620e-01 7.75904179e-01
7.88621247e-01 2.08891287e-01 3.88616771e-01 -2.86636382e-01
-5.00675738e-01 -9.80119824e-01 -6.11565292e-01 6.59511805e-01
1.41471066e-02 -5.73894262e-01 -4.29387212e-01 -3.06796759e-01] | [8.90078353881836, 5.678644180297852] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.