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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a7f644f2-84b3-46bf-8b2d-e226661a1de6
|
ultra-scalable-spectral-clustering-and
|
1903.01057
| null |
http://arxiv.org/abs/1903.01057v2
|
http://arxiv.org/pdf/1903.01057v2.pdf
|
Ultra-Scalable Spectral Clustering and Ensemble Clustering
|
This paper focuses on scalability and robustness of spectral clustering for
extremely large-scale datasets with limited resources. Two novel algorithms are
proposed, namely, ultra-scalable spectral clustering (U-SPEC) and
ultra-scalable ensemble clustering (U-SENC). In U-SPEC, a hybrid representative
selection strategy and a fast approximation method for K-nearest
representatives are proposed for the construction of a sparse affinity
sub-matrix. By interpreting the sparse sub-matrix as a bipartite graph, the
transfer cut is then utilized to efficiently partition the graph and obtain the
clustering result. In U-SENC, multiple U-SPEC clusterers are further integrated
into an ensemble clustering framework to enhance the robustness of U-SPEC while
maintaining high efficiency. Based on the ensemble generation via multiple
U-SEPC's, a new bipartite graph is constructed between objects and base
clusters and then efficiently partitioned to achieve the consensus clustering
result. It is noteworthy that both U-SPEC and U-SENC have nearly linear time
and space complexity, and are capable of robustly and efficiently partitioning
ten-million-level nonlinearly-separable datasets on a PC with 64GB memory.
Experiments on various large-scale datasets have demonstrated the scalability
and robustness of our algorithms. The MATLAB code and experimental data are
available at https://www.researchgate.net/publication/330760669.
|
['Chee-Keong Kwoh', 'Jian-Huang Lai', 'Jian-Sheng Wu', 'Chang-Dong Wang', 'Dong Huang']
|
2019-03-04
| null | null | null | null |
['imagedocument-clustering']
|
['computer-vision']
|
[-2.04295307e-01 -4.42875743e-01 -3.12129799e-02 -1.60439879e-01
-9.76105154e-01 -4.92879778e-01 -9.18009356e-02 1.82907823e-02
-4.25402448e-02 4.92711604e-01 1.54256895e-01 -8.04810151e-02
-6.08560205e-01 -8.01162183e-01 -4.98637438e-01 -1.27250934e+00
-2.87445873e-01 5.58039546e-01 3.40322226e-01 2.06232920e-01
1.98977515e-01 2.13256106e-01 -1.51665258e+00 3.49936485e-01
1.19823480e+00 7.98176110e-01 2.37449259e-01 4.77638453e-01
1.68979079e-01 1.21597342e-01 -1.92884669e-01 4.00444977e-02
5.20080686e-01 -3.66530269e-01 -8.03378224e-01 3.95919472e-01
-2.30543435e-01 1.62380889e-01 -2.74155349e-01 1.12777519e+00
5.93677163e-01 3.08064878e-01 5.90528309e-01 -1.41135633e+00
-4.21229452e-01 8.13760996e-01 -1.06868744e+00 -1.47526473e-01
2.94110984e-01 4.96345460e-02 9.77963209e-01 -9.40425038e-01
5.61946988e-01 1.09424937e+00 6.24320984e-01 3.83397192e-02
-1.51135910e+00 -9.11450744e-01 5.80237433e-02 2.51156092e-01
-2.18560171e+00 -2.55401045e-01 6.40229046e-01 -2.56880343e-01
5.26199222e-01 6.88166738e-01 5.35980582e-01 4.63573039e-01
-3.14819068e-01 5.67838371e-01 1.17472553e+00 -5.92956133e-02
5.12868822e-01 -5.82585074e-02 4.31829870e-01 5.36424875e-01
4.45480883e-01 -2.24581048e-01 -2.38050595e-01 -6.51469648e-01
2.86330730e-01 1.80743545e-01 -3.66589755e-01 -4.01969761e-01
-1.29360104e+00 6.64408684e-01 6.69154227e-01 3.15445155e-01
-4.76857185e-01 -2.15182021e-01 4.85396773e-01 -2.86312122e-02
2.43317753e-01 -9.24661905e-02 -8.59936401e-02 2.50123143e-01
-1.00278342e+00 -1.30390838e-01 7.47042954e-01 1.25482619e+00
9.74293113e-01 -2.63058484e-01 7.32835829e-02 8.76907587e-01
7.50636607e-02 7.20343113e-01 3.04600358e-01 -1.04741192e+00
3.37950140e-01 1.01461625e+00 -1.62467897e-01 -1.34506369e+00
-4.49664801e-01 -4.64505970e-01 -1.45075452e+00 -5.34268379e-01
-6.02278486e-02 -2.56316178e-02 -5.80748081e-01 1.48712254e+00
7.67468631e-01 5.94069004e-01 -1.95110368e-03 1.03581059e+00
6.42705321e-01 7.88381100e-01 -2.38970742e-01 -6.19491696e-01
1.25633311e+00 -6.48576379e-01 -3.08859110e-01 3.88825893e-01
5.15901506e-01 -4.57228988e-01 6.52487338e-01 1.70710534e-01
-8.26245964e-01 -4.40588087e-01 -9.00128186e-01 4.12248582e-01
-1.59932584e-01 4.97878045e-02 3.22018266e-01 4.84652102e-01
-1.07110381e+00 2.49116793e-01 -7.19784737e-01 -4.43460613e-01
3.19836676e-01 4.71804470e-01 -4.54131812e-01 -3.05772036e-01
-9.70092118e-01 -1.50418747e-02 8.32074165e-01 -1.00536786e-01
-4.12261069e-01 -5.18819153e-01 -4.61383700e-01 1.63373306e-01
4.73723859e-01 -6.11277699e-01 2.92393655e-01 -7.37124443e-01
-9.27804470e-01 5.16691208e-01 -2.09893137e-01 -9.82935075e-03
6.76719174e-02 4.79669750e-01 -6.64174736e-01 4.69979495e-01
4.13439572e-01 3.58960152e-01 4.46902633e-01 -1.66272807e+00
-7.66468823e-01 -6.35497391e-01 -6.72086835e-01 2.33701959e-01
-6.69068813e-01 -8.91855359e-02 -9.06803668e-01 -5.56202412e-01
5.55030763e-01 -1.20288622e+00 -4.99401271e-01 -7.44625449e-01
-5.44019639e-01 -2.16882765e-01 9.31506753e-01 -5.18696070e-01
1.82161725e+00 -2.41303897e+00 4.15020257e-01 1.06357765e+00
2.96189189e-01 1.33702204e-01 -2.71060079e-01 6.51016235e-01
-4.26717848e-02 6.63251728e-02 -5.44796288e-01 2.26315111e-02
-1.59940496e-01 1.06477782e-01 6.70004860e-02 6.81918025e-01
-5.43519139e-01 4.96929675e-01 -9.81064379e-01 -6.83941722e-01
5.98839261e-02 1.80861071e-01 -4.70293611e-01 -4.61696461e-02
3.61163676e-01 1.61444366e-01 -5.07086098e-01 8.17498803e-01
1.08546531e+00 -5.95634460e-01 7.60425091e-01 -3.60666841e-01
2.29872111e-02 -4.82544273e-01 -1.80893898e+00 1.41533566e+00
2.15646867e-02 -1.93728730e-01 4.34457362e-01 -1.13186717e+00
8.93043041e-01 2.18718737e-01 8.14100206e-01 -4.23570454e-01
9.20699835e-02 5.70372820e-01 -4.17555422e-02 -1.11145005e-01
4.74250853e-01 1.74623296e-01 -2.62015343e-01 7.40890920e-01
-2.49193102e-01 3.51646870e-01 6.42193019e-01 7.31924295e-01
1.03264546e+00 -3.30241024e-01 1.85612574e-01 -8.21441889e-01
8.37948143e-01 1.01653943e-02 9.26171482e-01 4.57236350e-01
-1.98954597e-01 4.70070332e-01 7.13622347e-02 -1.53175831e-01
-9.14875448e-01 -1.10382175e+00 -1.54332109e-02 9.35005426e-01
5.06915987e-01 -7.66624033e-01 -8.95358503e-01 -4.92176443e-01
1.79630131e-01 3.52243841e-01 -2.77880877e-01 -1.64966285e-01
-2.37004369e-01 -1.02084041e+00 2.96066225e-01 4.68651921e-01
4.76184338e-01 -5.76443434e-01 3.84966135e-02 1.83432370e-01
-4.18061405e-01 -8.68479908e-01 -7.60192394e-01 -2.03346536e-01
-8.10951650e-01 -1.18752015e+00 -3.60833496e-01 -8.33284259e-01
1.00613642e+00 8.76189351e-01 7.57334530e-01 3.21341425e-01
-3.32900345e-01 3.70262086e-01 -4.84772503e-01 1.98451623e-01
-1.35517925e-01 4.65054847e-02 5.69754779e-01 4.79674697e-01
3.89231265e-01 -9.20998454e-01 -8.03990781e-01 6.00435078e-01
-9.00399804e-01 -4.91507426e-02 5.80506742e-01 7.34055459e-01
9.03341293e-01 5.63769221e-01 6.31117404e-01 -8.47365141e-01
4.68380570e-01 -9.18321371e-01 -4.51763809e-01 3.04631770e-01
-7.35436082e-01 -2.99157917e-01 8.43458951e-01 -2.53548384e-01
-1.01701224e+00 2.90721476e-01 3.75129610e-01 -5.65078855e-01
3.25804800e-02 4.96665835e-01 -2.70099074e-01 2.03131810e-01
4.35427010e-01 4.85723644e-01 -5.47352098e-02 -3.13826621e-01
4.42616940e-01 1.12878346e+00 5.21454692e-01 -7.66514122e-01
9.38811600e-01 6.95941210e-01 -7.15238303e-02 -8.15997720e-01
-5.16936600e-01 -9.94641960e-01 -7.27990627e-01 -2.21977711e-01
6.91439390e-01 -1.05703485e+00 -7.77364969e-01 3.97598028e-01
-4.64128524e-01 -9.67962854e-03 1.03049994e-01 4.23659116e-01
-2.06188485e-01 4.54673707e-01 -6.31453097e-01 -7.34550595e-01
-4.41805691e-01 -1.02296853e+00 9.62593973e-01 1.96821898e-01
-4.57587317e-02 -7.75615335e-01 1.16242364e-03 5.28412521e-01
-6.21206872e-02 2.84733236e-01 6.62973642e-01 -6.25820458e-01
-6.08088076e-01 -1.12889901e-01 -3.61368001e-01 8.37846696e-02
1.83160231e-01 1.93336323e-01 -5.40039420e-01 -9.32069302e-01
-4.22293007e-01 -5.03408164e-02 7.22082734e-01 2.55493939e-01
1.30435133e+00 -2.39393756e-01 -6.80533588e-01 7.19944656e-01
1.60271096e+00 1.82325140e-01 4.08389091e-01 -1.85019709e-02
8.05690944e-01 5.10891140e-01 6.82223141e-01 6.21691823e-01
5.89108884e-01 3.08791906e-01 3.71993193e-03 3.69541943e-02
2.30542332e-01 2.44302619e-02 2.49382019e-01 1.60931039e+00
-2.84583241e-01 -1.24918632e-01 -1.04845202e+00 6.60704792e-01
-2.11268234e+00 -1.00209594e+00 -6.02695286e-01 2.19394231e+00
6.06150448e-01 -3.39475095e-01 3.46534640e-01 2.92530417e-01
1.24453461e+00 -1.87710881e-01 -5.05298972e-01 4.77537550e-02
-1.40588477e-01 -2.35247850e-01 4.58656043e-01 2.02694520e-01
-1.01411283e+00 5.72815180e-01 5.46005487e+00 1.21057105e+00
-6.17980957e-01 3.98998037e-02 7.17216432e-01 -2.22419262e-01
-1.57224491e-01 7.54422843e-02 -4.47266191e-01 8.30784321e-01
9.58531022e-01 -4.38954055e-01 5.63186347e-01 7.07095802e-01
2.14329600e-01 -1.06752157e-01 -6.26076519e-01 1.09843087e+00
-1.35266945e-01 -1.24899971e+00 2.86667813e-02 2.50226498e-01
1.21516085e+00 -1.10800162e-01 -1.32122785e-01 1.00807227e-01
6.85145736e-01 -5.44976413e-01 2.95794040e-01 2.81906337e-01
8.88371646e-01 -1.27285767e+00 6.31781936e-01 4.25217569e-01
-1.78262067e+00 -3.60081106e-01 -4.85491097e-01 3.78279597e-01
-5.10319695e-02 9.80302870e-01 -2.21414819e-01 1.31382334e+00
1.19700003e+00 6.97200418e-01 -5.62083840e-01 8.03313375e-01
3.27365786e-01 7.05194831e-01 -5.72056115e-01 1.99112222e-01
1.51365593e-01 -8.15382123e-01 6.15720510e-01 1.21916950e+00
4.77749050e-01 6.64356649e-01 6.33006334e-01 5.34193933e-01
-1.28012881e-01 3.34449887e-01 -1.61838710e-01 1.80736959e-01
1.14306319e+00 1.63150036e+00 -1.17042875e+00 -6.37784421e-01
-2.24441439e-01 8.32330227e-01 3.50245744e-01 3.65285307e-01
-8.47911358e-01 -4.80664730e-01 5.49878418e-01 -7.07998872e-02
3.33096385e-01 -3.09033669e-03 -1.86641708e-01 -9.81616259e-01
-1.25455722e-01 -1.03176749e+00 9.50725734e-01 -5.55967152e-01
-1.44680500e+00 4.60299224e-01 -1.76319659e-01 -1.34560609e+00
3.94800484e-01 -8.52280483e-02 -5.15713573e-01 5.56657672e-01
-6.28773034e-01 -1.02703941e+00 -4.27552164e-01 1.05939996e+00
-1.26947314e-01 -7.46290460e-02 6.15230620e-01 3.03565383e-01
-1.00313616e+00 4.42887604e-01 7.81521678e-01 9.68068279e-03
6.99067235e-01 -1.11277926e+00 -2.36593306e-01 9.82090712e-01
-1.68085903e-01 9.57806766e-01 2.98921347e-01 -7.49345303e-01
-1.66844714e+00 -1.41816545e+00 2.57201046e-01 -6.87718615e-02
6.34832501e-01 -3.55178654e-01 -1.02826059e+00 3.77520472e-01
-7.23793462e-04 1.95667092e-02 1.03379607e+00 1.20218940e-01
-4.57775712e-01 -5.10803640e-01 -1.11454082e+00 5.15693307e-01
1.19962382e+00 -4.52440560e-01 -1.71293557e-01 3.90309423e-01
4.65135336e-01 -5.46231605e-02 -1.26115859e+00 4.41281974e-01
1.97402701e-01 -1.08481026e+00 9.27533269e-01 -9.38268080e-02
-6.78960308e-02 -7.51095414e-01 -2.33143046e-01 -1.11122394e+00
-8.19812596e-01 -6.08736813e-01 1.16589986e-01 1.41901326e+00
9.05363783e-02 -8.29306543e-01 5.54270864e-01 2.00492322e-01
5.73154794e-05 -6.94065094e-01 -7.91272342e-01 -8.77590358e-01
-2.39853784e-01 -1.02733329e-01 8.51822019e-01 1.20709074e+00
1.11842006e-01 2.79389650e-01 -2.06375733e-01 4.53303903e-01
1.06764388e+00 7.74039447e-01 7.86548436e-01 -1.23093593e+00
2.74204444e-02 -3.29053462e-01 -2.06425712e-01 -4.86002654e-01
6.92539141e-02 -1.33595252e+00 -1.77486867e-01 -1.42732573e+00
8.51934731e-01 -5.47701299e-01 -5.52325606e-01 3.02430183e-01
-4.92300987e-01 4.51224029e-01 7.07004070e-02 6.47178113e-01
-1.12962890e+00 4.36013073e-01 9.49185371e-01 1.21416464e-01
-3.79585445e-01 -1.69991508e-01 -7.51026452e-01 4.29877192e-01
7.99053192e-01 -3.41519743e-01 -4.73789424e-01 2.03043431e-01
-2.03486875e-01 3.57033499e-02 -1.60357673e-02 -9.42797244e-01
4.63791788e-01 -2.09564328e-01 2.53348351e-01 -8.11620891e-01
-1.07217818e-01 -1.03858972e+00 7.52766311e-01 5.72164118e-01
1.38020203e-01 -5.52271074e-03 -1.52226016e-01 8.51461112e-01
-1.91094190e-01 4.45522249e-01 8.71578813e-01 1.21678822e-01
-7.15747416e-01 6.02252245e-01 -6.76674917e-02 -8.95035714e-02
1.54796958e+00 -3.67342800e-01 -2.18022615e-01 -6.75749183e-02
-6.18292987e-01 8.86311829e-01 6.89298749e-01 1.41357705e-01
5.63876152e-01 -1.70595348e+00 -9.58541334e-01 7.19292983e-02
2.49698803e-01 9.11802799e-02 7.14144289e-01 1.05370760e+00
-2.97355026e-01 2.21285805e-01 5.93946651e-02 -8.87918293e-01
-1.52118981e+00 8.72326791e-01 -8.44592974e-02 -9.68398675e-02
-4.89750862e-01 5.28056264e-01 6.81524575e-02 -6.70603931e-01
-1.07479952e-01 3.74468148e-01 3.11898619e-01 1.31984010e-01
4.62608308e-01 8.89658630e-01 -1.85871616e-01 -9.97225344e-01
-6.54376805e-01 5.04821002e-01 2.06003278e-01 1.78308651e-01
1.39857566e+00 -4.29041862e-01 -8.30483198e-01 1.03541203e-01
1.23190653e+00 7.99166560e-02 -7.84046173e-01 -2.70860404e-01
-2.04026382e-02 -4.61137176e-01 -2.66985863e-01 -2.68837929e-01
-1.33631039e+00 2.10649654e-01 3.75837177e-01 2.41992950e-01
1.59892011e+00 9.67846066e-03 7.35288143e-01 1.68358907e-01
4.83330429e-01 -1.37194514e+00 -1.21903464e-01 1.14125364e-01
5.01152813e-01 -8.45010519e-01 2.52356857e-01 -6.70313001e-01
-5.99564493e-01 7.65034437e-01 6.57185674e-01 -6.57124072e-02
6.86806738e-01 3.44856530e-01 -2.01208830e-01 -2.09423780e-01
-6.20416880e-01 -1.84870794e-01 1.35509491e-01 4.02062923e-01
4.67305966e-02 4.69659567e-01 -3.79454762e-01 6.71465874e-01
1.03360370e-01 -4.00839001e-01 2.09166214e-01 6.71055675e-01
-4.44102347e-01 -9.14238930e-01 -7.77579069e-01 6.12566769e-01
2.79568192e-02 1.23046294e-01 -3.48565221e-01 5.32973051e-01
1.73258245e-01 1.09405136e+00 1.26090094e-01 -7.68325925e-01
8.51229429e-02 -7.42810667e-02 -5.80081381e-02 -3.89113367e-01
-5.16133428e-01 3.58818501e-01 -1.99911147e-01 -8.38395953e-01
-5.14986455e-01 -6.05120778e-01 -1.45737576e+00 -6.66245937e-01
-3.78577471e-01 6.29371941e-01 1.37843281e-01 3.30330282e-01
7.71054149e-01 3.37137938e-01 1.11687779e+00 -7.19163895e-01
-2.88948625e-01 -5.74764371e-01 -1.01767802e+00 7.51244247e-01
-2.42664114e-01 -3.76130402e-01 -5.49454689e-01 -5.47655076e-02]
|
[7.56129264831543, 4.743781089782715]
|
d7f02ca6-2d34-43c2-b142-bc17459076bf
|
knowledge-augmented-language-model-prompting
|
2306.04136
| null |
https://arxiv.org/abs/2306.04136v1
|
https://arxiv.org/pdf/2306.04136v1.pdf
|
Knowledge-Augmented Language Model Prompting for Zero-Shot Knowledge Graph Question Answering
|
Large Language Models (LLMs) are capable of performing zero-shot closed-book question answering tasks, based on their internal knowledge stored in parameters during pre-training. However, such internalized knowledge might be insufficient and incorrect, which could lead LLMs to generate factually wrong answers. Furthermore, fine-tuning LLMs to update their knowledge is expensive. To this end, we propose to augment the knowledge directly in the input of LLMs. Specifically, we first retrieve the relevant facts to the input question from the knowledge graph based on semantic similarities between the question and its associated facts. After that, we prepend the retrieved facts to the input question in the form of the prompt, which is then forwarded to LLMs to generate the answer. Our framework, Knowledge-Augmented language model PromptING (KAPING), requires no model training, thus completely zero-shot. We validate the performance of our KAPING framework on the knowledge graph question answering task, that aims to answer the user's question based on facts over a knowledge graph, on which ours outperforms relevant zero-shot baselines by up to 48% in average, across multiple LLMs of various sizes.
|
['Amir Saffari', 'Alham Fikri Aji', 'Jinheon Baek']
|
2023-06-07
| null | null | null | null |
['graph-question-answering']
|
['graphs']
|
[ 1.10339798e-01 8.77872884e-01 -8.99788737e-02 -1.47988558e-01
-8.32327783e-01 -6.92393482e-01 5.65245032e-01 6.46170437e-01
-5.08161008e-01 6.07543826e-01 2.96430826e-01 -4.55641419e-01
-5.75070642e-02 -1.22332561e+00 -9.50854838e-01 1.28437757e-01
6.09531105e-01 7.77919352e-01 7.56099284e-01 -5.86001158e-01
9.44919214e-02 -2.59331942e-01 -1.34216380e+00 4.90349114e-01
1.14696097e+00 9.22593117e-01 3.75290692e-01 8.07894528e-01
-9.22090828e-01 1.35509038e+00 -5.84715188e-01 -1.08355939e+00
-4.93554056e-01 -4.95233715e-01 -1.41878009e+00 -3.37972313e-01
5.72250426e-01 -3.85851860e-01 -4.07964140e-01 8.27310920e-01
9.26172137e-02 6.22059226e-01 4.39044356e-01 -8.44148636e-01
-1.12487566e+00 8.10689807e-01 2.43194565e-01 2.35814378e-01
6.50719166e-01 3.33019719e-02 1.23098862e+00 -1.06749403e+00
6.49781585e-01 1.38308394e+00 2.07454652e-01 6.90351069e-01
-9.45212364e-01 -7.28603229e-02 3.72605413e-01 5.98469317e-01
-1.03193438e+00 -4.14732665e-01 5.90329826e-01 -1.31845653e-01
1.29955149e+00 2.18169972e-01 4.27127570e-01 8.38161409e-01
-2.80817211e-01 7.25192428e-01 5.96871018e-01 -5.33142805e-01
2.24915445e-01 3.67524922e-01 4.97722566e-01 9.51048732e-01
1.95576206e-01 -3.49260598e-01 -6.73061490e-01 -3.95226061e-01
3.38131040e-01 -3.10353637e-01 -2.61980265e-01 -5.43986671e-02
-7.91165650e-01 8.50990772e-01 4.44049656e-01 1.90950751e-01
-5.10546565e-01 3.59031595e-02 3.89889851e-02 4.07419771e-01
2.47211277e-01 7.20086813e-01 -5.19826174e-01 5.21145537e-02
-4.52460378e-01 3.74986202e-01 1.20178533e+00 9.16240633e-01
9.54004228e-01 -5.39524972e-01 -7.52607703e-01 7.85617471e-01
2.84400463e-01 3.93843383e-01 4.10566688e-01 -9.49012280e-01
5.62651813e-01 9.89746928e-01 3.01552296e-01 -1.05818558e+00
-2.64880270e-01 -2.97609389e-01 -3.52560401e-01 -7.72645295e-01
3.39681000e-01 -1.25917599e-01 -7.82540739e-01 1.65300989e+00
6.35672510e-01 3.53157520e-01 3.67213994e-01 7.41184413e-01
1.48851502e+00 7.00206101e-01 3.12856972e-01 2.64720023e-02
1.55094874e+00 -1.12054718e+00 -8.15021753e-01 -5.63328624e-01
9.18665469e-01 -3.96891207e-01 1.54496682e+00 4.92650224e-03
-9.34624374e-01 -4.98131037e-01 -4.98374641e-01 -4.05871302e-01
-4.24528301e-01 -2.26081207e-01 2.68791586e-01 2.56931931e-01
-1.03719199e+00 2.79671669e-01 -3.06305110e-01 -3.81128639e-01
2.47957990e-01 2.25185882e-02 1.34495839e-01 -4.71703827e-01
-1.73251438e+00 8.61090243e-01 6.48286223e-01 -1.75695509e-01
-7.23313391e-01 -9.09465134e-01 -9.66184318e-01 4.50444221e-01
1.11424112e+00 -1.14081526e+00 1.64129329e+00 -4.97529954e-01
-1.53056300e+00 5.69889486e-01 -4.91870135e-01 -3.90860736e-01
-9.03540999e-02 -1.58234999e-01 -4.17163789e-01 3.24334174e-01
1.81128792e-02 5.58438540e-01 7.99857795e-01 -1.24061012e+00
-5.35574734e-01 -1.22930035e-01 8.02212298e-01 1.95710182e-01
-1.88496307e-01 -3.87438834e-01 -8.43172371e-01 -2.32834697e-01
-7.59260952e-02 -4.64622349e-01 -4.67170216e-02 -7.44498849e-01
-2.71148443e-01 -5.84274411e-01 5.12547731e-01 -8.06002319e-01
1.46449530e+00 -1.73518550e+00 -1.10684363e-02 2.12372988e-01
3.79539847e-01 4.88566816e-01 -4.57302809e-01 5.12094796e-01
4.48587269e-01 3.31823975e-02 -3.94768678e-02 -8.60989392e-02
3.59972529e-02 4.58551258e-01 -7.38282263e-01 -4.91194725e-01
9.46706012e-02 1.69647026e+00 -1.32576048e+00 -5.47679245e-01
-1.27395868e-01 6.37118192e-03 -7.16035485e-01 4.64671195e-01
-9.78719354e-01 9.06363353e-02 -5.44330001e-01 3.22377682e-01
2.17081368e-01 -7.89375424e-01 5.84068373e-02 -2.84488380e-01
7.26926625e-01 5.93811214e-01 -8.04933250e-01 1.61741614e+00
-6.57210588e-01 1.43949524e-01 -3.13506305e-01 -7.36123562e-01
7.88768470e-01 2.84116268e-01 -1.48489296e-01 -8.71548533e-01
-8.18718821e-02 3.09610590e-02 -2.58285642e-01 -7.46684730e-01
7.17943728e-01 -5.71754500e-02 3.97089422e-02 5.36802649e-01
5.35994232e-01 -3.36718529e-01 3.17876637e-01 1.02284515e+00
1.12244499e+00 -8.80773887e-02 1.77178785e-01 1.58179566e-01
7.91220486e-01 7.25600794e-02 -1.81226939e-01 1.10110152e+00
2.37003863e-01 -6.52645081e-02 5.72204232e-01 2.25785542e-02
-3.91165227e-01 -9.66062486e-01 6.29878461e-01 1.34165466e+00
2.26756617e-01 -6.52006269e-01 -7.72011459e-01 -1.11594725e+00
9.08548087e-02 1.32327950e+00 -4.16481942e-01 -5.91406941e-01
-3.05792958e-01 5.17882779e-03 4.62658077e-01 3.52941006e-01
4.09894288e-01 -1.17687285e+00 -4.60870743e-01 4.81969833e-01
-5.54058969e-01 -1.29999363e+00 -4.43463504e-01 -3.81026506e-01
-7.45065451e-01 -1.26575708e+00 -4.23640370e-01 -6.08876288e-01
7.88672924e-01 2.08889425e-01 1.56045413e+00 5.10173023e-01
2.83919126e-01 1.03974116e+00 -5.63360453e-01 -2.40989849e-01
-4.64314371e-01 1.30560204e-01 -4.45330620e-01 1.98513381e-02
5.13389170e-01 -3.33192855e-01 -5.94140232e-01 4.64297757e-02
-1.07384443e+00 1.67922974e-02 3.14832807e-01 5.91158688e-01
5.35754502e-01 -2.06052899e-01 1.14655256e+00 -1.21020436e+00
1.16230977e+00 -6.57290936e-01 -6.39158547e-01 9.13525283e-01
-4.48598623e-01 3.76337886e-01 7.05278695e-01 -2.65272707e-01
-1.43282008e+00 -4.36765105e-01 -2.82840282e-01 -2.41085604e-01
2.10994810e-01 9.56525862e-01 -1.54682004e-03 -3.20583619e-02
9.72794712e-01 1.99010506e-01 -4.70073342e-01 -4.82678741e-01
1.03849697e+00 4.29806203e-01 8.03844452e-01 -7.46229351e-01
7.25801885e-01 1.28548846e-01 -4.29509223e-01 -6.18378699e-01
-1.59127426e+00 -6.89533949e-01 -1.49172410e-01 -1.13916822e-01
5.49103916e-01 -6.06972396e-01 -7.79423594e-01 1.74005255e-01
-1.33618701e+00 -4.41192925e-01 -5.77717066e-01 1.25795454e-01
-3.14618975e-01 4.27342921e-01 -5.26688755e-01 -6.85287654e-01
-4.71111655e-01 -5.96605062e-01 7.53462493e-01 3.73766840e-01
-3.03226382e-01 -1.29863572e+00 1.71602610e-02 7.08029449e-01
4.93827343e-01 -4.33569700e-01 1.60846317e+00 -9.60978329e-01
-5.62663138e-01 -1.84812829e-01 -1.65543184e-01 1.97278321e-01
4.08468843e-02 -4.86223310e-01 -9.37279880e-01 1.56875238e-01
-3.87220345e-02 -7.64946282e-01 8.81313682e-01 -1.03677675e-01
1.09014857e+00 -6.83306038e-01 -2.35346705e-01 -6.29973710e-02
1.15692127e+00 -2.88194329e-01 3.56889546e-01 -1.27465978e-01
6.57654583e-01 7.63001919e-01 5.57272613e-01 1.82593822e-01
1.05189419e+00 4.04715806e-01 2.79681623e-01 4.11420077e-01
-2.59024143e-01 -8.54098439e-01 6.67427108e-02 8.93987715e-01
2.53614157e-01 -3.64123821e-01 -7.50769913e-01 6.82377517e-01
-1.86366653e+00 -6.66152239e-01 -8.44994932e-02 2.11286259e+00
1.17821419e+00 -7.57000409e-03 -2.95914143e-01 -4.24818158e-01
3.55053395e-01 7.27520809e-02 -6.92748725e-01 -2.60342628e-01
5.93502000e-02 4.72855508e-01 1.00585490e-01 1.04955637e+00
-3.75170141e-01 1.38821590e+00 5.22426605e+00 8.09823990e-01
-5.43134093e-01 2.63333470e-01 2.17428058e-01 7.09730387e-02
-9.47413921e-01 3.98409069e-01 -8.39395821e-01 1.18900299e-01
1.08607793e+00 -6.83190048e-01 5.76571167e-01 6.57084107e-01
-2.98403174e-01 -3.91102850e-01 -9.15021539e-01 7.04118729e-01
2.85219938e-01 -1.44997358e+00 6.18952692e-01 -6.14221454e-01
7.43333638e-01 -3.14201891e-01 -3.13400149e-01 1.05387568e+00
4.45761681e-01 -6.93718910e-01 3.46448809e-01 8.46336007e-01
3.68932843e-01 -6.48348629e-01 6.09359682e-01 8.29441607e-01
-9.84474540e-01 8.37372914e-02 -4.33828741e-01 -9.39962547e-03
3.86520952e-01 5.07783651e-01 -1.01795340e+00 6.92989171e-01
3.61462891e-01 5.16568683e-02 -7.28013813e-01 6.03947222e-01
-8.87443125e-01 7.71720767e-01 -2.24017963e-01 -1.91757113e-01
2.10815713e-01 2.49726534e-01 1.92223877e-01 7.84433603e-01
1.09757662e-01 6.05308294e-01 8.99207070e-02 9.72049356e-01
-5.86464643e-01 2.39003792e-01 -2.76354492e-01 -3.50185037e-01
5.50815523e-01 1.17131329e+00 -1.81618288e-01 -1.03390920e+00
-5.88176847e-01 1.05496383e+00 6.64296925e-01 7.55048156e-01
-5.11322498e-01 -4.61601496e-01 -2.73231696e-02 4.10126299e-02
9.99766290e-02 1.89759299e-01 3.08707833e-01 -1.34278560e+00
1.18007965e-01 -8.54425013e-01 8.21890056e-01 -9.62717116e-01
-1.24202096e+00 4.30863857e-01 4.90790084e-02 -2.09822416e-01
-6.19239211e-01 -2.46764109e-01 -4.94997054e-01 1.04244328e+00
-1.84099162e+00 -8.64902675e-01 -4.94519770e-01 7.80798554e-01
3.83824885e-01 1.59286916e-01 8.64765525e-01 1.46500558e-01
-1.20795593e-01 5.57697594e-01 -5.57326794e-01 -8.91599283e-02
5.69284201e-01 -1.10803640e+00 6.94460571e-01 6.52460277e-01
5.10838807e-01 6.59324765e-01 5.32402515e-01 -9.60726380e-01
-1.40514851e+00 -1.18941784e+00 1.38075328e+00 -7.94760585e-01
7.44821310e-01 -1.22321919e-01 -1.63785493e+00 7.53840208e-01
1.39798671e-01 -1.05984606e-01 6.75822318e-01 1.50884137e-01
-4.48317558e-01 3.44308466e-01 -8.83056760e-01 6.33772314e-01
8.88454437e-01 -7.82294810e-01 -1.32170665e+00 5.43909013e-01
1.34354222e+00 -4.67015237e-01 -6.67535186e-01 2.34360605e-01
-6.38849288e-02 -4.57215518e-01 7.91802347e-01 -1.01120698e+00
2.90422708e-01 -1.36846840e-01 5.72976209e-02 -1.36192286e+00
-2.07116783e-01 -4.75860327e-01 -9.76403236e-01 1.09413683e+00
6.37448311e-01 -5.88634789e-01 5.82485139e-01 8.17596674e-01
-2.94575989e-02 -7.74604201e-01 -8.25176239e-01 -6.22872412e-01
-2.60102898e-01 -4.62657899e-01 7.30955303e-01 6.60618544e-01
1.83809400e-01 9.67941225e-01 5.95038272e-02 2.00268045e-01
2.27747411e-01 2.05656514e-01 7.87851214e-01 -1.09020901e+00
-4.99294758e-01 -1.35691479e-01 4.09235775e-01 -1.53103304e+00
3.63266379e-01 -1.06956816e+00 -3.90443020e-02 -2.08120298e+00
1.07259907e-01 -2.30607301e-01 -1.57755777e-01 6.13966584e-01
-9.87569034e-01 -3.20879787e-01 4.55153733e-02 -3.33882384e-02
-1.03494918e+00 6.33911133e-01 1.34776592e+00 -1.40241846e-01
-2.92760193e-01 -2.14496776e-01 -9.00892913e-01 7.27283418e-01
5.18214285e-01 -3.81310254e-01 -8.83745253e-01 -4.90654469e-01
7.86613226e-01 2.38901556e-01 5.58348000e-01 -5.73637903e-01
6.12495005e-01 -1.18783854e-01 -1.81417331e-01 -3.65087897e-01
3.29461277e-01 -5.71170509e-01 -4.65434462e-01 2.62724787e-01
-4.40833926e-01 -2.96534806e-01 2.75728703e-01 6.75570786e-01
-3.11179489e-01 -5.66293955e-01 2.66284406e-01 -2.95344830e-01
-9.96667027e-01 3.39098871e-01 2.77454685e-02 7.60378659e-01
4.98418033e-01 2.54953444e-01 -5.86573720e-01 -8.07275951e-01
-6.17499113e-01 5.62176287e-01 1.76291447e-02 4.97658461e-01
7.29552686e-01 -1.07637846e+00 -4.85130072e-01 -1.69430330e-01
5.31372607e-01 1.26930669e-01 6.51002407e-01 5.77426970e-01
-2.51316354e-02 7.89783537e-01 5.83662570e-01 -1.28324255e-01
-8.09401095e-01 8.87167573e-01 3.09501350e-01 -4.68234748e-01
-3.90541166e-01 1.05249417e+00 1.39072016e-01 -6.21083498e-01
1.70669526e-01 -4.25943077e-01 -4.73991752e-01 9.83043835e-02
6.71101511e-01 1.80737361e-01 2.69408703e-01 -1.74563915e-01
-9.82013866e-02 2.36095414e-01 -8.32836330e-02 -1.78364620e-01
6.72417641e-01 -2.28616357e-01 -8.41461420e-02 2.01275393e-01
7.79204547e-01 1.14656024e-01 -6.98174953e-01 -9.04989362e-01
2.95089394e-01 -2.56659240e-01 -2.14360207e-01 -1.13150573e+00
-6.25233054e-01 6.20095789e-01 -2.10088804e-01 1.14291452e-01
8.78675461e-01 5.73161006e-01 1.15720594e+00 1.21339810e+00
2.99265921e-01 -9.26953793e-01 3.54582578e-01 8.85196090e-01
8.86188090e-01 -1.19541800e+00 -4.34962273e-01 -3.41418624e-01
-6.97868526e-01 7.79111683e-01 8.12331259e-01 3.19590360e-01
1.95676953e-01 -4.42040980e-01 1.94945671e-02 -4.81165320e-01
-1.13785219e+00 -4.18264270e-01 6.14749193e-01 3.93205673e-01
1.48684502e-01 -1.53038338e-01 -1.07010253e-01 9.90671992e-01
-2.79591948e-01 2.40483537e-01 2.68774271e-01 8.67024601e-01
-7.81072795e-01 -8.94911826e-01 -6.79332092e-02 6.96334362e-01
-2.06959367e-01 -5.36421180e-01 -6.18395209e-01 3.15545052e-01
-2.49888837e-01 1.34956872e+00 -2.85374284e-01 -1.87600195e-01
7.60702252e-01 6.20950520e-01 4.80552167e-01 -1.01539385e+00
-5.90400934e-01 -7.41919041e-01 3.27324718e-01 -5.32815278e-01
5.98457307e-02 1.96162611e-01 -1.59092283e+00 -7.35531598e-02
-3.05914313e-01 5.95377326e-01 1.83445379e-01 1.24072266e+00
6.08776391e-01 5.23557186e-01 3.19808945e-02 2.00157925e-01
-6.36055231e-01 -9.24450040e-01 -3.57073843e-02 4.69064713e-01
1.05989479e-01 -4.76142436e-01 -2.35698387e-01 -2.20076437e-03]
|
[10.90010929107666, 7.9243550300598145]
|
9c691dde-1fee-4f3f-a0b3-789dcbd097a8
|
most-multiple-object-localization-with-self
|
2304.05387
| null |
https://arxiv.org/abs/2304.05387v1
|
https://arxiv.org/pdf/2304.05387v1.pdf
|
MOST: Multiple Object localization with Self-supervised Transformers for object discovery
|
We tackle the challenging task of unsupervised object localization in this work. Recently, transformers trained with self-supervised learning have been shown to exhibit object localization properties without being trained for this task. In this work, we present Multiple Object localization with Self-supervised Transformers (MOST) that uses features of transformers trained using self-supervised learning to localize multiple objects in real world images. MOST analyzes the similarity maps of the features using box counting; a fractal analysis tool to identify tokens lying on foreground patches. The identified tokens are then clustered together, and tokens of each cluster are used to generate bounding boxes on foreground regions. Unlike recent state-of-the-art object localization methods, MOST can localize multiple objects per image and outperforms SOTA algorithms on several object localization and discovery benchmarks on PASCAL-VOC 07, 12 and COCO20k datasets. Additionally, we show that MOST can be used for self-supervised pre-training of object detectors, and yields consistent improvements on fully, semi-supervised object detection and unsupervised region proposal generation.
|
['Abhinav Shrivastava', 'Rama Chellappa', 'Ishan Misra', 'Sai Saketh Rambhatla']
|
2023-04-11
| null | null | null | null |
['unsupervised-object-localization', 'object-discovery', 'semi-supervised-object-detection', 'object-localization']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 3.63592565e-01 1.11614026e-01 -1.12131231e-01 -3.16354394e-01
-1.23712838e+00 -6.67396307e-01 6.47285163e-01 3.72142494e-01
-1.50244296e-01 4.04149115e-01 -3.01641434e-01 2.30678404e-03
1.50776178e-01 -6.28594816e-01 -1.10020494e+00 -7.62541056e-01
-2.59585708e-01 8.64829063e-01 1.18476403e+00 3.65776420e-01
2.72048682e-01 5.17272890e-01 -1.67386961e+00 6.59788728e-01
5.18519878e-01 1.36047316e+00 4.40593541e-01 7.58453786e-01
-3.94034177e-01 9.58734274e-01 -7.49655187e-01 4.99170795e-02
2.49802887e-01 -9.22918990e-02 -8.84939313e-01 3.91491681e-01
1.10433245e+00 -1.86186936e-02 2.49748811e-01 8.46049666e-01
3.05614918e-01 -2.94476181e-01 9.15674329e-01 -1.32849479e+00
-3.72851491e-01 6.52834058e-01 -4.87321496e-01 4.85278487e-01
1.14505835e-01 1.84065793e-02 1.01509941e+00 -1.41079533e+00
5.83789885e-01 1.24768507e+00 8.26047003e-01 2.38265738e-01
-1.63903046e+00 -6.11174345e-01 -3.09907287e-01 3.99403041e-03
-1.62407422e+00 -4.24913436e-01 5.80710828e-01 -5.19127727e-01
8.83653283e-01 9.13479999e-02 5.33873916e-01 5.57483554e-01
-3.48253903e-04 1.07284105e+00 1.24159384e+00 -7.37861753e-01
3.85941714e-01 5.20691752e-01 -8.50282088e-02 1.15919852e+00
1.90045655e-01 -1.23856947e-01 -6.66460037e-01 -1.64402902e-01
7.83307433e-01 -1.58724487e-01 4.56120700e-01 -9.22971845e-01
-1.60107636e+00 7.09648669e-01 7.21416593e-01 5.14338791e-01
-7.83091038e-02 4.72024292e-01 3.80255669e-01 -1.91991001e-01
5.22657275e-01 5.12206733e-01 -4.46252882e-01 2.80812413e-01
-1.10252333e+00 3.78159508e-02 5.00335395e-01 1.27214372e+00
1.06194293e+00 -8.94108340e-02 -3.88078988e-01 5.42834878e-01
9.97373238e-02 6.51711524e-01 3.57913017e-01 -6.97016478e-01
2.74474144e-01 8.40477407e-01 -3.53142098e-02 -5.79642832e-01
-2.60237187e-01 -3.78460675e-01 -2.11864948e-01 2.62749732e-01
7.11785257e-01 4.45358425e-01 -1.06925189e+00 1.06703353e+00
5.62448204e-01 6.61636069e-02 5.50031625e-02 4.84778851e-01
8.37891877e-01 4.40957218e-01 2.28977315e-02 2.17820972e-01
1.53000915e+00 -1.18348181e+00 -3.11103374e-01 -1.80691496e-01
8.32290769e-01 -9.11396921e-01 9.09947753e-01 2.53863752e-01
-9.21377122e-01 -7.64936566e-01 -7.31587768e-01 1.74421385e-01
-7.64098704e-01 6.66392386e-01 7.95977712e-01 6.79627359e-01
-9.31310058e-01 4.09488320e-01 -7.40341306e-01 -5.42347014e-01
1.18744004e+00 3.52436870e-01 -2.06716537e-01 1.22524701e-01
-3.19995105e-01 8.02103102e-01 8.19090784e-01 -3.78479362e-01
-1.56066692e+00 -7.92059898e-01 -7.94548452e-01 -1.28890455e-01
5.08247137e-01 -7.03041628e-02 1.15351129e+00 -1.04704797e+00
-1.07636333e+00 1.07219517e+00 -7.86108077e-02 -8.12339425e-01
4.15321052e-01 -6.35933727e-02 -1.66344419e-01 3.05333763e-01
8.11228693e-01 1.20182490e+00 1.12260699e+00 -1.35683072e+00
-1.08655906e+00 -4.65235077e-02 -2.54661292e-01 -2.06963137e-01
-1.48856893e-01 2.82831013e-01 -3.13606083e-01 -5.50714552e-01
3.15223277e-01 -6.56090975e-01 -5.50072715e-02 2.32363775e-01
-6.53739929e-01 -6.67478859e-01 1.52484763e+00 -6.64810464e-02
3.47547531e-01 -2.03626299e+00 -4.88744438e-01 2.49087274e-01
1.94880575e-01 -1.36058489e-02 -3.86029445e-02 -1.64410114e-01
-7.48121664e-02 -1.60945117e-01 -6.55766800e-02 -4.19704467e-01
-4.79638651e-02 -2.72592120e-02 -3.79119873e-01 6.48960412e-01
4.90598053e-01 1.13441908e+00 -9.82905626e-01 -1.24909234e+00
5.18752396e-01 1.18712276e-01 -1.99809089e-01 1.48564473e-01
-4.97324258e-01 1.78199992e-01 -2.65531331e-01 1.17582822e+00
4.94629860e-01 -5.46306968e-01 -1.62750691e-01 -3.90775502e-01
-8.09515342e-02 1.66899845e-01 -1.07650495e+00 1.34028876e+00
-2.31010228e-01 8.60843658e-01 -2.65335590e-01 -1.00883842e+00
8.16763043e-01 2.64476873e-02 5.17310858e-01 -4.32902962e-01
1.89669207e-02 2.84741312e-01 -3.05777043e-01 -2.14547306e-01
2.98811466e-01 1.81523621e-01 -2.75025144e-02 4.76052731e-01
6.29939079e-01 -3.96614522e-01 5.20577669e-01 3.26213628e-01
1.11883783e+00 1.92771912e-01 2.59280205e-01 -4.56186742e-01
2.94005752e-01 4.94251668e-01 1.10915452e-01 9.67279732e-01
-1.69527128e-01 7.20640481e-01 3.77701640e-01 -4.10325408e-01
-1.07928336e+00 -1.47413576e+00 -4.84494001e-01 1.51126587e+00
1.55748129e-01 -4.46950376e-01 -8.06970239e-01 -1.16724706e+00
1.73583612e-01 3.47362369e-01 -7.12541282e-01 2.55496800e-01
-4.20566201e-01 -4.50312972e-01 7.23170638e-01 7.82571852e-01
3.76602650e-01 -1.21874130e+00 -9.71358955e-01 2.11436599e-01
8.93171355e-02 -1.36595857e+00 -2.31372595e-01 9.15197492e-01
-8.34551930e-01 -1.08686781e+00 -7.68287599e-01 -1.14146352e+00
1.09824431e+00 3.59524101e-01 1.28276038e+00 -1.57544866e-01
-9.11284685e-01 5.57348192e-01 -2.56897509e-01 -6.77639842e-01
-3.38802338e-01 -3.70671228e-02 1.32020831e-01 6.70640264e-03
2.11752862e-01 -7.47073069e-02 -2.88505137e-01 7.19612539e-01
-6.59758151e-01 -7.90045783e-02 8.11328053e-01 7.47283340e-01
8.40462685e-01 2.64959484e-01 2.48946741e-01 -7.33703911e-01
-1.89373240e-01 -1.03241056e-01 -9.28194106e-01 3.76321733e-01
-1.45301893e-01 2.16380544e-02 3.66923392e-01 -6.32107019e-01
-7.77379453e-01 7.49169409e-01 5.27426422e-01 -5.21345139e-01
-5.36363363e-01 -4.03769851e-01 -5.43301627e-02 -6.68894053e-01
9.01222765e-01 3.83067191e-01 -4.28085029e-01 -2.34440356e-01
6.29297674e-01 4.13461208e-01 5.45845628e-01 -5.23526430e-01
1.15486753e+00 8.80430162e-01 -2.11994231e-01 -7.74747968e-01
-1.07679951e+00 -9.53342080e-01 -1.07958078e+00 -2.12047607e-01
9.57196236e-01 -8.60321105e-01 -2.86921382e-01 1.12173915e-01
-1.14172733e+00 -4.12710398e-01 -7.52861083e-01 1.47422850e-01
-7.57959485e-01 1.00422382e-01 -1.39007106e-01 -9.36764181e-01
-6.99059218e-02 -8.80095482e-01 1.66868889e+00 1.64264217e-01
5.26134335e-02 -7.92798579e-01 -1.58281878e-01 4.37543243e-01
2.97501862e-01 4.64175381e-02 5.24913907e-01 -8.17207694e-01
-1.01084149e+00 -2.07448155e-01 -4.48555887e-01 2.58771390e-01
2.07269341e-01 -1.28231019e-01 -1.14678562e+00 -1.29899874e-01
-6.11590028e-01 -6.85292602e-01 9.62217331e-01 2.75904357e-01
1.27211773e+00 -1.83185801e-01 -8.44536364e-01 3.75185013e-01
1.30154049e+00 -1.04702666e-01 2.58692950e-01 2.22492203e-01
6.60720944e-01 4.97950524e-01 8.05379450e-01 7.57697672e-02
3.83885652e-02 5.71013510e-01 3.97656798e-01 -4.61141765e-01
-2.55687594e-01 -1.70119688e-01 2.86100745e-01 -3.05306576e-02
2.17480734e-01 2.03099951e-01 -1.11383891e+00 8.99546385e-01
-1.62660050e+00 -7.87922621e-01 -6.61556572e-02 1.79989171e+00
7.04121053e-01 5.09543777e-01 4.74591315e-01 6.82754070e-02
8.09562981e-01 -2.38972262e-01 -3.60010743e-01 3.09487402e-01
-3.86140376e-01 4.82256681e-01 9.96333838e-01 -1.20824262e-01
-1.75678515e+00 1.44772148e+00 6.85062885e+00 1.12019181e+00
-8.40798914e-01 3.70564252e-01 6.67762339e-01 1.81166485e-01
4.30480659e-01 7.73290396e-02 -1.11854112e+00 1.86778769e-01
5.83636701e-01 3.32169086e-01 -9.42676961e-02 1.54827142e+00
-2.34931439e-01 -5.20100415e-01 -1.28653324e+00 7.97144294e-01
2.28313148e-01 -1.61640108e+00 -1.30910665e-01 -1.58539802e-01
9.17093158e-01 5.71015440e-02 8.51912498e-02 2.10839421e-01
3.50288838e-01 -8.80065918e-01 1.07863724e+00 7.02618212e-02
7.19447672e-01 -3.66229951e-01 6.00846648e-01 1.79162472e-01
-1.70135522e+00 -1.63804665e-01 -6.38925135e-01 4.50263351e-01
-9.29750949e-02 6.49794042e-01 -1.67424893e+00 6.55273944e-02
1.04864037e+00 5.30449569e-01 -1.33287001e+00 1.23539758e+00
-1.41295984e-01 7.58340955e-01 -5.87903321e-01 -1.64254412e-01
2.66322732e-01 3.27795357e-01 2.70493835e-01 1.54105425e+00
-3.12005617e-02 -5.20963728e-01 4.66227174e-01 1.33700633e+00
3.61908153e-02 2.06225619e-01 -6.86816573e-01 -1.16353355e-01
3.11912388e-01 1.56274796e+00 -1.65169024e+00 -7.47034013e-01
-1.29647478e-01 8.08081746e-01 3.95018518e-01 -5.60504012e-02
-1.01383090e+00 -4.49713409e-01 -9.70008522e-02 3.95307600e-01
8.53545427e-01 -8.36571380e-02 -1.88594371e-01 -7.30664909e-01
-1.69507280e-01 -4.92267966e-01 2.63982236e-01 -9.86628950e-01
-1.20074916e+00 2.20178872e-01 1.24040142e-01 -1.25645268e+00
1.81140438e-01 -8.58470201e-01 -7.50427783e-01 1.32771283e-01
-1.04191422e+00 -1.74238718e+00 -1.85737312e-01 4.75273609e-01
5.51618218e-01 -2.97934055e-01 5.39857626e-01 6.24078363e-02
-1.32218584e-01 3.23446244e-01 1.63086787e-01 4.31770980e-01
6.41004682e-01 -1.66302431e+00 4.06166703e-01 6.66830719e-01
8.58230650e-01 4.12710190e-01 3.92487079e-01 -6.25254869e-01
-1.14283192e+00 -1.58686626e+00 4.81146961e-01 -8.79866719e-01
6.68966591e-01 -9.72575605e-01 -6.10137284e-01 6.62004769e-01
-1.10408187e-01 7.61162221e-01 2.33022362e-01 -2.14061335e-01
-5.66741586e-01 -2.00170889e-01 -1.18011653e+00 1.75606728e-01
1.03387856e+00 -5.17256320e-01 -5.13548732e-01 1.11289346e+00
6.02104366e-01 -2.99260497e-01 -5.99625647e-01 2.80030131e-01
2.56446242e-01 -8.95161986e-01 1.16018867e+00 -3.25173229e-01
-6.17425963e-02 -6.20458722e-01 -1.69192642e-01 -7.47922897e-01
-1.60484627e-01 -4.04963553e-01 -5.30487821e-02 1.33436251e+00
4.01023269e-01 -3.09841901e-01 1.21289051e+00 -3.96333992e-01
-4.22602333e-02 -4.09591407e-01 -1.15702903e+00 -1.03483367e+00
-3.30567390e-01 -4.00279552e-01 2.54879564e-01 4.88714635e-01
-2.95907646e-01 2.00927362e-01 2.00990364e-01 2.26125345e-01
1.18396842e+00 2.55587846e-01 1.07469416e+00 -1.14365911e+00
-6.79316968e-02 -3.21840972e-01 -9.02987719e-01 -8.01447034e-01
7.75834024e-02 -8.47492218e-01 5.09782851e-01 -1.32334840e+00
4.25408721e-01 -7.44638443e-01 -1.49237186e-01 7.18507648e-01
1.85942024e-01 7.84583449e-01 -8.51832181e-02 2.28419036e-01
-1.50311422e+00 2.41406292e-01 7.06152558e-01 -3.95124495e-01
7.28658140e-02 -6.29428625e-02 -1.53355092e-01 8.21427763e-01
5.60875893e-01 -8.43954802e-01 2.08781108e-01 1.29265144e-01
-3.74960482e-01 -5.86296976e-01 7.69256234e-01 -1.57092464e+00
1.61151111e-01 1.11631066e-01 8.48526597e-01 -1.18431520e+00
2.78953820e-01 -8.94842863e-01 -3.76440734e-01 4.32696342e-01
-1.08633965e-01 -2.53078520e-01 2.35933080e-01 5.36968589e-01
9.01664793e-02 -9.10588875e-02 1.12886286e+00 -3.45159888e-01
-9.19541299e-01 -1.95053034e-02 -4.58530843e-01 1.80981811e-02
1.39994323e+00 -3.92739981e-01 -2.77526319e-01 7.29935095e-02
-6.00911081e-01 4.55955006e-02 3.72145057e-01 2.32290998e-01
6.82822168e-01 -1.32771242e+00 -4.00029182e-01 1.31334379e-01
6.47079408e-01 3.15809906e-01 -1.99261934e-01 8.44897211e-01
-6.21449828e-01 5.76715350e-01 -3.87149230e-02 -1.48801434e+00
-1.55977118e+00 6.62377000e-01 3.00598174e-01 -2.27604210e-01
-5.67992926e-01 1.04127991e+00 4.69627082e-01 -4.39202815e-01
4.23196614e-01 -6.85272574e-01 1.53601646e-01 -1.94574390e-02
3.74651462e-01 1.76506847e-01 3.50082144e-02 -7.46239126e-01
-6.38840616e-01 6.30259097e-01 -2.99970228e-02 -4.57454361e-02
1.16675532e+00 3.19066197e-01 -6.99953269e-03 5.08599162e-01
1.17120290e+00 -1.23558357e-01 -1.09344184e+00 -5.95618665e-01
5.83352506e-01 -2.68045783e-01 -9.85905379e-02 -6.29484177e-01
-7.47650743e-01 6.91569090e-01 1.03240561e+00 2.24751815e-01
6.44171834e-01 8.65785718e-01 2.11452186e-01 7.50771403e-01
5.29771864e-01 -1.12579799e+00 8.94358575e-01 3.61066967e-01
5.02677023e-01 -1.42115569e+00 7.25369528e-02 -5.11970282e-01
-3.54981869e-01 1.08246613e+00 8.05971801e-01 -2.08136871e-01
3.83439600e-01 6.31002426e-01 -7.07992911e-02 -4.81628776e-01
-3.58314455e-01 -5.56711316e-01 5.67137897e-01 9.15843070e-01
1.74676448e-01 -5.85295483e-02 7.35746622e-01 -7.14016671e-04
-3.30778509e-02 -4.30203795e-01 1.24520883e-01 1.20058680e+00
-9.01329458e-01 -6.17825568e-01 -9.98350799e-01 6.57343924e-01
-3.19499671e-01 -1.14265680e-01 -6.28755629e-01 7.94141412e-01
5.67656755e-01 6.56488299e-01 3.82215828e-01 7.07551911e-02
2.73493696e-02 9.12755169e-03 6.29685163e-01 -1.08878684e+00
-5.01480758e-01 1.88221425e-01 -1.11210316e-01 -4.66649711e-01
-5.40571630e-01 -6.96563840e-01 -1.32422554e+00 5.11379302e-01
-9.79816854e-01 1.94587126e-01 7.25147128e-01 8.54581237e-01
-3.76030169e-02 6.15646243e-01 3.89855802e-01 -1.16563213e+00
-2.51389325e-01 -1.06808639e+00 -5.02041698e-01 2.79164255e-01
2.16633409e-01 -9.24194634e-01 -3.63811195e-01 4.29338604e-01]
|
[9.394859313964844, 0.931625485420227]
|
140efed5-e877-411b-9b8e-ab57221def6a
|
call-for-papers-the-babylm-challenge-sample
|
2301.11796
| null |
https://arxiv.org/abs/2301.11796v1
|
https://arxiv.org/pdf/2301.11796v1.pdf
|
Call for Papers -- The BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus
|
We present the call for papers for the BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus. This shared task is intended for participants with an interest in small scale language modeling, human language acquisition, low-resource NLP, and cognitive modeling. In partnership with CoNLL and CMCL, we provide a platform for approaches to pretraining with a limited-size corpus sourced from data inspired by the input to children. The task has three tracks, two of which restrict the training data to pre-released datasets of 10M and 100M words and are dedicated to explorations of approaches such as architectural variations, self-supervised objectives, or curriculum learning. The final track only restricts the amount of text used, allowing innovation in the choice of the data, its domain, and even its modality (i.e., data from sources other than text is welcome). We will release a shared evaluation pipeline which scores models on a variety of benchmarks and tasks, including targeted syntactic evaluations and natural language understanding.
|
['Chengxu Zhuang', 'Ethan Wilcox', 'Adina Williams', 'Aaron Mueller', 'Leshem Choshen', 'Alex Warstadt']
|
2023-01-27
| null | null | null | null |
['language-acquisition']
|
['natural-language-processing']
|
[ 2.10753262e-01 7.31134534e-01 -3.05340797e-01 -8.49673569e-01
-8.90347660e-01 -8.02517533e-01 6.77995324e-01 4.12712067e-01
-8.04724395e-01 4.57412750e-01 7.07915843e-01 -4.65058625e-01
2.22393215e-01 -4.03637975e-01 -1.03048062e+00 1.99557200e-01
1.41602010e-01 1.14327967e+00 1.42600134e-01 -2.77708292e-01
-1.15266815e-01 -7.59264752e-02 -1.29992151e+00 8.07694197e-01
6.67413712e-01 3.35654229e-01 4.84083116e-01 6.87476337e-01
-3.68624300e-01 8.80622685e-01 -4.18814957e-01 -5.87866724e-01
-2.40591392e-01 -1.82497576e-01 -1.27704382e+00 -4.87856418e-01
7.42654562e-01 -3.30215305e-01 1.37060359e-01 5.59368491e-01
4.82559651e-01 8.78178552e-02 5.77531934e-01 -8.64217162e-01
-8.45041037e-01 1.63516295e+00 6.49603531e-02 2.50580579e-01
3.79032433e-01 1.09228566e-01 7.74689913e-01 -1.03234231e+00
7.94732928e-01 1.67133951e+00 4.96800065e-01 1.26966119e+00
-1.22661257e+00 -5.50446510e-01 5.95070124e-01 -5.96140437e-02
-8.65642369e-01 -8.23107421e-01 3.67007822e-01 -4.64528412e-01
1.28908908e+00 5.12388796e-02 2.90495217e-01 1.57558465e+00
-4.05150414e-01 1.25043070e+00 8.02723587e-01 -9.23940539e-01
-8.97963531e-03 2.87359774e-01 3.20028692e-01 5.31831086e-01
-1.45802153e-02 6.61388412e-02 -8.10606301e-01 1.24387205e-01
3.41407299e-01 -7.92020857e-01 -7.61929825e-02 -1.58276469e-01
-1.16643953e+00 6.40051425e-01 -1.05061159e-01 3.88256967e-01
2.26519719e-01 2.01140568e-01 5.01526237e-01 3.31649274e-01
5.33626199e-01 6.75662398e-01 -1.07463610e+00 -3.38038623e-01
-9.02822971e-01 4.58288372e-01 8.37555230e-01 1.42378974e+00
2.79603779e-01 2.04029754e-02 -3.75602804e-02 1.06894398e+00
5.36401629e-01 2.31881425e-01 6.59525990e-01 -9.01423991e-01
9.64393973e-01 3.44822139e-01 -3.17787141e-01 9.91943553e-02
-6.65007949e-01 -2.47483868e-02 -4.03264344e-01 -1.47906691e-01
7.08693802e-01 -4.34933811e-01 -9.20313478e-01 2.16651559e+00
3.31604809e-01 7.70869106e-02 1.94845259e-01 2.09579647e-01
1.42227316e+00 5.04626155e-01 6.08246624e-01 3.50740626e-02
1.06023538e+00 -1.06006742e+00 -2.10387751e-01 -6.29091561e-01
1.12224913e+00 -6.39760137e-01 1.25997412e+00 5.04563630e-01
-1.64351141e+00 -5.46856284e-01 -6.55180573e-01 -5.53658009e-01
-6.44839048e-01 -1.50696158e-01 5.28899729e-01 6.06166244e-01
-1.17086506e+00 4.39564675e-01 -8.45219314e-01 -4.73738432e-01
1.53103635e-01 2.16047630e-01 -4.15638000e-01 -3.35795671e-01
-1.13081896e+00 1.16629791e+00 8.04092407e-01 -4.03312802e-01
-8.29539895e-01 -1.22871220e+00 -1.15358567e+00 -9.38484520e-02
-1.06278151e-01 -5.01459718e-01 1.62395513e+00 -8.90388668e-01
-1.35734224e+00 1.40652215e+00 6.82498962e-02 -5.56952953e-01
4.34931427e-01 -4.41246808e-01 -3.35236527e-02 -4.76384908e-01
8.21538344e-02 1.23952091e+00 2.04259664e-01 -7.32956648e-01
-5.79645813e-01 -1.11330576e-01 2.50148494e-02 3.28416765e-01
-3.49343628e-01 7.11048782e-01 -3.50747675e-01 -6.10855520e-01
-2.23326966e-01 -6.62823260e-01 -2.59547472e-01 -6.21097982e-01
-1.63042903e-01 -5.68950832e-01 9.10894573e-02 -8.18008184e-01
9.95297372e-01 -1.84501016e+00 1.57927707e-01 -3.28705937e-01
-1.55658334e-01 1.96489528e-01 -4.12164748e-01 5.66540599e-01
-2.04350315e-02 5.87764263e-01 1.34089842e-01 -9.33128536e-01
2.25833625e-01 9.00897682e-02 -4.51367766e-01 -1.21346563e-01
3.36460650e-01 1.00082517e+00 -8.61351490e-01 -5.80858231e-01
9.01144668e-02 2.16710523e-01 -7.01986015e-01 4.50049192e-01
-6.41575158e-01 6.05037689e-01 -1.95206657e-01 2.30719969e-01
1.80657864e-01 6.32748753e-02 4.67494167e-02 2.59532839e-01
-2.96348512e-01 9.95038748e-01 -9.60498214e-01 2.00183845e+00
-7.97744334e-01 5.38615465e-01 4.24770191e-02 -4.34532046e-01
5.28889716e-01 5.33130765e-01 -8.82498622e-02 -4.78925765e-01
-1.48230717e-02 2.08648175e-01 5.73647507e-02 -2.93521851e-01
4.48106825e-01 1.94279611e-01 -1.29008457e-01 5.74054182e-01
7.55779803e-01 -2.84235358e-01 3.84582967e-01 1.41220897e-01
1.03558207e+00 5.41382730e-01 1.15848474e-01 -5.60146928e-01
2.32120648e-01 -1.55988440e-01 3.95362496e-01 7.35772371e-01
2.15497091e-01 5.26150823e-01 1.29822537e-01 -4.48532611e-01
-1.00460184e+00 -8.94194365e-01 -2.24830985e-01 1.95880795e+00
-4.88315850e-01 -5.75078428e-01 -8.28630924e-01 -5.62041581e-01
-3.09353322e-01 1.15977824e+00 -6.59746230e-01 4.16491069e-02
-1.02334130e+00 -4.63282913e-01 8.56040180e-01 8.58940244e-01
-1.25275642e-01 -1.33776248e+00 -3.58930528e-01 2.76114792e-01
-3.42993881e-03 -1.12764728e+00 -3.58813792e-01 4.39462900e-01
-7.70810664e-01 -6.38077557e-01 -4.56790894e-01 -1.28494191e+00
5.08265615e-01 -5.72123408e-01 1.62687135e+00 2.58113034e-02
2.39119697e-02 4.90462750e-01 -3.21167439e-01 -9.43544209e-01
-9.26648021e-01 3.61627668e-01 -3.16857157e-04 -7.97169745e-01
4.71458703e-01 -4.48585510e-01 -4.77172323e-02 -2.98277527e-01
-5.85778475e-01 5.78786373e-01 2.55861938e-01 5.23274839e-01
4.18220013e-01 -7.58593321e-01 5.12566209e-01 -1.18531072e+00
5.42946339e-01 -4.86370951e-01 -4.02822554e-01 5.27945936e-01
-3.58864188e-01 2.37397000e-01 4.81838048e-01 -6.82835698e-01
-1.13228810e+00 -4.65244316e-02 -3.11225116e-01 3.24860290e-02
-5.23875713e-01 4.94194210e-01 -4.33390826e-01 4.19295907e-01
6.33152843e-01 3.17685679e-02 -4.81159896e-01 -9.55409884e-01
9.01998401e-01 3.42048824e-01 6.29486501e-01 -1.22601914e+00
3.75996143e-01 -3.51578951e-01 -6.54370010e-01 -7.80065656e-01
-8.87185872e-01 -6.79626539e-02 -8.39878559e-01 3.07308257e-01
8.30306351e-01 -9.71736193e-01 -6.77269027e-02 4.07038271e-01
-1.58800924e+00 -9.36233580e-01 -3.53669286e-01 4.49029952e-01
-3.85435432e-01 -1.81496575e-01 -8.78001332e-01 -4.46728587e-01
-5.22531986e-01 -8.35987568e-01 7.66834319e-01 1.12082139e-01
-6.62243426e-01 -1.43966937e+00 2.68054992e-01 4.88982946e-01
4.48003501e-01 -1.16720803e-01 1.40243971e+00 -1.44814146e+00
-2.94449270e-01 1.83790177e-01 1.89480290e-01 3.87011081e-01
-4.16344315e-01 9.21633393e-02 -9.75633979e-01 -2.31226876e-01
-3.66519153e-01 -1.18146658e+00 5.43580532e-01 3.35557550e-01
9.86606598e-01 -3.65126312e-01 -3.10893953e-01 5.38491189e-01
8.45269918e-01 -4.49031405e-02 -2.78476235e-02 2.82834947e-01
6.93189204e-01 1.05372739e+00 2.06547871e-01 -7.14914352e-02
1.01675498e+00 3.59453917e-01 1.35164887e-01 5.83123676e-02
-3.14765096e-01 -6.30113840e-01 4.29767370e-01 1.10999811e+00
1.98485970e-01 -4.05395061e-01 -1.40416980e+00 9.71425056e-01
-1.55223036e+00 -4.11550373e-01 1.43041462e-01 2.06037211e+00
1.50491941e+00 3.98889363e-01 -6.30546408e-03 -5.45170367e-01
4.31393355e-01 1.53400805e-02 -3.90565097e-01 -7.93394983e-01
-6.40938198e-03 6.36931777e-01 1.25521958e-01 7.73482084e-01
-8.42129350e-01 1.33430052e+00 6.62790585e+00 6.37226522e-01
-8.59808505e-01 3.71868223e-01 7.77479827e-01 -1.50300503e-01
-6.12171352e-01 -1.46946445e-01 -1.42420292e+00 3.54914188e-01
1.56623602e+00 -3.70854288e-01 2.86776423e-01 7.74982929e-01
-7.92896673e-02 2.41155192e-01 -1.85410571e+00 3.97045732e-01
7.27545843e-02 -1.24993682e+00 2.04795286e-01 -4.09980923e-01
5.78841865e-01 6.57799006e-01 -6.76290989e-02 7.73859203e-01
7.65075564e-01 -1.37690115e+00 1.08670616e+00 2.33046517e-01
8.07657063e-01 -5.51490307e-01 2.98638195e-01 9.03190792e-01
-8.43989015e-01 3.21563452e-01 -1.51875839e-01 -2.95433134e-01
5.48628122e-02 -5.56137487e-02 -8.35789561e-01 -1.41945243e-01
6.61812246e-01 3.34678978e-01 -8.06338251e-01 6.17207944e-01
-4.75836277e-01 1.04770970e+00 -5.63598096e-01 -2.88954169e-01
4.03391235e-02 4.81504738e-01 2.22792923e-01 1.69065619e+00
-1.55643195e-01 8.45241174e-02 3.17265421e-01 8.32624972e-01
-2.31126919e-01 3.60530883e-01 -5.41691005e-01 5.80476187e-02
9.09930527e-01 9.09594655e-01 -3.11243951e-01 -2.24701419e-01
-6.24171734e-01 3.23400229e-01 7.82430649e-01 -6.52967114e-03
-2.55994529e-01 6.98712328e-03 2.92094797e-01 2.29578465e-01
-5.11953346e-02 -2.17795625e-01 -3.06737572e-01 -1.03985083e+00
-1.66683912e-01 -1.12130380e+00 4.15365189e-01 -6.35478795e-01
-1.10482764e+00 6.57168508e-01 5.59226632e-01 -3.29705447e-01
-8.32396209e-01 -8.14668000e-01 -6.57510102e-01 1.12569332e+00
-1.12851322e+00 -1.62249637e+00 2.28931338e-01 2.32436299e-01
9.26777244e-01 -3.34862828e-01 9.91209328e-01 7.45018423e-02
-3.74117553e-01 8.22569132e-01 -1.85890228e-01 2.21536919e-01
6.78876877e-01 -1.43078232e+00 1.18099642e+00 7.69054949e-01
4.03956741e-01 6.25883818e-01 7.67863452e-01 -5.62211812e-01
-8.22084367e-01 -1.04277027e+00 1.39462388e+00 -1.10249424e+00
7.93450534e-01 -7.25348711e-01 -8.99146259e-01 1.17946720e+00
4.40397680e-01 -2.88099438e-01 7.69935727e-01 4.44841594e-01
-2.25015938e-01 3.21913004e-01 -7.92936504e-01 5.08035541e-01
1.17226863e+00 -3.19563895e-01 -9.75642145e-01 3.98268193e-01
1.15912199e+00 -8.43591571e-01 -8.19952428e-01 5.76885164e-01
3.89844388e-01 -4.44812983e-01 6.48959875e-01 -1.13046861e+00
7.00091481e-01 4.89182025e-01 1.85617041e-02 -1.32192421e+00
-2.00579703e-01 -7.49785960e-01 -1.80972919e-01 1.82747614e+00
9.08581972e-01 -1.49514139e-01 7.71215916e-01 9.51212108e-01
-3.87171686e-01 -8.39329720e-01 -8.46446931e-01 -4.79873031e-01
9.89342093e-01 -8.69728625e-01 5.39199293e-01 7.26095855e-01
-1.12143084e-02 7.26550043e-01 1.20885253e-01 -8.91099349e-02
5.05716145e-01 -3.85521621e-01 6.95100129e-01 -1.08637893e+00
-3.76569837e-01 -4.46626097e-01 1.28150076e-01 -1.16931498e+00
3.58065456e-01 -1.20420253e+00 6.42782971e-02 -1.50443041e+00
-3.94873694e-02 -5.89935124e-01 -6.06352873e-02 7.37409592e-01
-4.54164408e-02 -3.62306148e-01 1.88903466e-01 -2.51595795e-01
-6.48595154e-01 2.21152380e-01 7.78839409e-01 5.69519438e-02
-1.99834302e-01 -6.24083579e-02 -8.17939937e-01 1.19408488e+00
5.84744811e-01 -3.75731349e-01 -6.35914028e-01 -1.32945812e+00
3.45083028e-01 -2.66595036e-01 5.43672740e-02 -8.51828694e-01
3.21470201e-01 -4.06433642e-01 2.66360343e-01 -5.79108834e-01
1.70830503e-01 -2.00463369e-01 -2.98010290e-01 3.00231352e-02
-1.02393949e+00 1.59608796e-01 5.48904657e-01 -1.68409571e-01
-3.19722034e-02 -6.02750361e-01 6.71576738e-01 -4.78869557e-01
-6.44798815e-01 1.19003393e-01 -1.48989866e-02 9.18796957e-01
5.14803886e-01 -5.53975487e-03 -4.81348366e-01 -2.46302262e-01
-8.25522900e-01 7.27940857e-01 1.49250239e-01 9.21827912e-01
3.92440349e-01 -1.08274913e+00 -1.14768970e+00 8.43872651e-02
1.61795944e-01 5.56704521e-01 -1.06452823e-01 1.88971385e-01
-2.70004094e-01 6.70358956e-01 9.32583064e-02 -2.13188142e-01
-1.40237474e+00 2.43974105e-01 2.39674538e-01 -5.08650303e-01
-3.09798270e-01 1.56168759e+00 2.11516351e-01 -1.02957535e+00
6.26706779e-01 -6.35499358e-01 -3.16097438e-01 -9.63736326e-02
7.79143989e-01 2.02003792e-01 3.17150215e-03 -3.71688217e-01
-3.03929955e-01 1.26581937e-01 -4.24202085e-01 -3.10110182e-01
1.58593273e+00 9.49152708e-02 4.65851016e-02 8.36052835e-01
8.56997013e-01 2.86107082e-02 -1.14274478e+00 -4.91255611e-01
5.08634090e-01 3.55269581e-01 -2.13584423e-01 -1.18069434e+00
-3.58712494e-01 9.67683613e-01 2.06912115e-01 -2.47216240e-01
6.41427100e-01 5.37471592e-01 6.15108073e-01 4.86645281e-01
3.73407900e-01 -1.15952659e+00 -5.09206951e-02 1.17079878e+00
1.09963834e+00 -1.15551996e+00 -2.48318389e-01 -1.30815074e-01
-4.03619140e-01 1.02181113e+00 1.11054707e+00 2.23349363e-01
5.31303465e-01 4.83255237e-01 1.22200347e-01 1.51140079e-01
-1.37690902e+00 1.04101010e-01 5.31261325e-01 7.62997150e-01
1.03030348e+00 2.91171186e-02 -1.69771805e-01 1.28314400e+00
-8.10628593e-01 -3.38615894e-01 2.62245834e-01 7.37353921e-01
-4.24319535e-01 -1.33280075e+00 1.29321873e-01 2.46288389e-01
-4.97355640e-01 -7.56112993e-01 -5.20074725e-01 8.15531194e-01
4.81342584e-01 7.31448948e-01 1.92044338e-03 4.36615311e-02
4.81755406e-01 4.84218627e-01 7.52749622e-01 -1.42027819e+00
-8.04106474e-01 -3.80013525e-01 6.32686079e-01 -3.57578903e-01
-1.27950624e-01 -7.13940620e-01 -1.31016576e+00 3.42444819e-03
-1.25263765e-01 2.71663275e-02 7.75970995e-01 1.19389582e+00
-5.11603616e-03 1.66350529e-01 -4.91298325e-02 -7.16966569e-01
-4.68075514e-01 -1.25124991e+00 9.54969674e-02 5.99558651e-02
2.70974506e-02 -1.14480890e-01 -7.01512843e-02 3.49859715e-01]
|
[10.870394706726074, 9.15762710571289]
|
7181e27d-9dca-4dd1-b894-7a6bf704df10
|
convolutional-neural-networks-with-intra
| null | null |
http://papers.nips.cc/paper/5634-convolutional-neural-networks-with-intra-layer-recurrent-connections-for-scene-labeling
|
http://papers.nips.cc/paper/5634-convolutional-neural-networks-with-intra-layer-recurrent-connections-for-scene-labeling.pdf
|
Convolutional Neural Networks with Intra-Layer Recurrent Connections for Scene Labeling
|
Scene labeling is a challenging computer vision task. It requires the use of both local discriminative features and global context information. We adopt a deep recurrent convolutional neural network (RCNN) for this task, which is originally proposed for object recognition. Different from traditional convolutional neural networks (CNN), this model has intra-layer recurrent connections in the convolutional layers. Therefore each convolutional layer becomes a two-dimensional recurrent neural network. The units receive constant feed-forward inputs from the previous layer and recurrent inputs from their neighborhoods. While recurrent iterations proceed, the region of context captured by each unit expands. In this way, feature extraction and context modulation are seamlessly integrated, which is different from typical methods that entail separate modules for the two steps. To further utilize the context, a multi-scale RCNN is proposed. Over two benchmark datasets, Standford Background and Sift Flow, the model outperforms many state-of-the-art models in accuracy and efficiency.
|
['Ming Liang', 'Xiaolin Hu', 'Bo Zhang']
|
2015-12-01
| null | null | null |
neurips-2015-12
|
['scene-labeling']
|
['computer-vision']
|
[ 3.00915211e-01 -4.50235218e-01 -5.26687682e-01 -4.88079160e-01
-6.64061680e-02 -3.01615447e-01 5.22019386e-01 -3.34699243e-01
-5.64042807e-01 3.65966588e-01 3.99243504e-01 -2.87903905e-01
4.04032379e-01 -9.00676727e-01 -3.79362583e-01 -7.07221806e-01
8.40020105e-02 -4.74027872e-01 6.43999994e-01 -7.96960816e-02
3.51924568e-01 6.98151529e-01 -1.40609419e+00 5.80493450e-01
4.14564371e-01 1.11572015e+00 2.89369136e-01 6.48587286e-01
-5.02558291e-01 1.48574746e+00 -4.12425995e-01 7.99676776e-02
4.83765863e-02 -4.10696596e-01 -7.46292591e-01 1.03928387e-01
4.80269521e-01 -2.73488283e-01 -8.77661765e-01 9.91704464e-01
3.15072030e-01 3.83279204e-01 2.69522458e-01 -7.42975712e-01
-9.16356087e-01 5.29736221e-01 -5.10605276e-01 4.31664467e-01
4.41333950e-02 -1.14770038e-02 1.00799727e+00 -1.04845941e+00
7.27891922e-01 1.12668669e+00 6.92670286e-01 4.51192349e-01
-9.52685654e-01 -6.14506960e-01 8.23393047e-01 3.01149249e-01
-1.15587139e+00 -8.46911669e-02 9.73145902e-01 -2.61983067e-01
1.03563726e+00 -2.09489581e-03 8.98146152e-01 1.02049601e+00
7.50877485e-02 1.01821804e+00 5.71669757e-01 -2.09762141e-01
1.90190803e-02 -2.65059292e-01 2.80388504e-01 6.32099688e-01
7.13130087e-02 5.80346845e-02 -3.08155358e-01 3.33296031e-01
1.07517350e+00 6.61114275e-01 -9.65265259e-02 -3.14591289e-01
-1.21793425e+00 7.84825146e-01 1.15889359e+00 4.34477538e-01
-2.81357139e-01 6.30359828e-01 6.46661639e-01 1.56036898e-01
1.87262893e-01 1.02339961e-01 -3.14676881e-01 1.05398118e-01
-8.77077997e-01 -1.17216982e-01 4.27040607e-01 9.23170507e-01
8.99271548e-01 3.64182085e-01 -5.29281497e-01 7.27190793e-01
2.47040257e-01 2.81372573e-02 5.66869915e-01 -6.55965865e-01
3.66036147e-01 1.15840256e+00 -2.31799796e-01 -1.04799831e+00
-3.73589188e-01 -7.44289458e-01 -1.05630744e+00 7.19257891e-02
7.04762042e-02 2.09680051e-02 -1.17246175e+00 1.39210665e+00
2.10073829e-01 8.04830074e-01 7.29599446e-02 9.80359256e-01
1.21325338e+00 6.93988562e-01 2.11013913e-01 9.05269310e-02
1.15811837e+00 -1.54987621e+00 -7.46250391e-01 -3.87774140e-01
5.14631808e-01 -8.15372467e-01 8.57326508e-01 1.08954990e-02
-7.34585524e-01 -1.01887453e+00 -1.16315973e+00 -4.40193623e-01
-7.54023254e-01 4.64309067e-01 9.60289717e-01 2.06669271e-01
-8.54375303e-01 4.22039896e-01 -5.95334709e-01 -1.87184885e-01
5.24769664e-01 1.67490050e-01 -1.50408596e-01 -3.35662872e-01
-9.85985279e-01 5.97461164e-01 2.65699804e-01 7.04916656e-01
-8.63165855e-01 -3.04662198e-01 -1.20237589e+00 -8.97572655e-03
2.12128088e-01 -3.06827158e-01 1.21531284e+00 -1.08042395e+00
-1.47081900e+00 5.54175138e-01 -3.44351381e-01 -3.68795663e-01
4.49907333e-01 -3.93715709e-01 -3.52675527e-01 -1.09781772e-02
5.59187075e-03 6.32510126e-01 7.63689697e-01 -8.67745697e-01
-7.75291145e-01 3.58281769e-02 5.76306917e-02 1.34132504e-01
-8.06955323e-02 1.03985287e-01 -8.16282272e-01 -1.00240946e+00
4.19193029e-01 -8.79693687e-01 -7.02648163e-01 1.22942254e-02
-2.04349712e-01 -3.16750646e-01 1.47774088e+00 -1.30183518e-01
1.38736725e+00 -2.27849555e+00 2.91576181e-02 -2.11202111e-02
8.65636244e-02 4.68796819e-01 -1.38716877e-01 5.27976453e-02
-2.47676507e-01 -1.68225124e-01 -1.53083488e-01 -3.57842445e-01
-2.67653704e-01 4.56029624e-01 -5.14032543e-01 5.05648434e-01
5.18911183e-01 1.35572863e+00 -1.02586234e+00 -2.18706846e-01
5.29483557e-01 5.43952465e-01 -4.76579994e-01 1.10996723e-01
-1.65192589e-01 3.12028855e-01 -3.50944847e-01 8.63534629e-01
5.20402908e-01 -4.21522349e-01 1.10861681e-01 -2.12641940e-01
-4.62995857e-01 2.72219509e-01 -1.22148860e+00 1.86205339e+00
-4.56335157e-01 9.57690299e-01 -4.50169563e-01 -1.22161376e+00
1.10852706e+00 8.88391361e-02 1.10814847e-01 -8.20169151e-01
2.30685130e-01 9.18201208e-02 -1.04183830e-01 -4.49628502e-01
7.57880986e-01 3.47910881e-01 -1.29866108e-01 2.68825054e-01
1.59972951e-01 3.97524744e-01 1.14490837e-02 -9.56474915e-02
9.22218621e-01 4.19970423e-01 2.73531169e-01 -7.71570653e-02
6.97611153e-01 -1.01466589e-01 1.04104662e+00 6.53750539e-01
-4.43386793e-01 7.46245205e-01 3.24177623e-01 -1.04252279e+00
-8.00523937e-01 -6.27632618e-01 -3.42198424e-02 1.15230536e+00
4.14431006e-01 -3.41757625e-01 -2.78747231e-01 -7.92071164e-01
-1.34195462e-01 1.55061007e-01 -9.42218125e-01 -1.71841308e-01
-9.76358354e-01 -3.13277990e-01 3.52540523e-01 9.73157823e-01
1.14056480e+00 -1.49026680e+00 -8.81521404e-01 4.57485050e-01
4.42086495e-02 -1.30751026e+00 -4.56324100e-01 4.18589741e-01
-8.33207905e-01 -1.07682037e+00 -6.41914964e-01 -1.10812235e+00
5.82686961e-01 6.22784853e-01 1.04364836e+00 2.21191570e-01
-6.38467968e-01 8.87374878e-02 -3.86271864e-01 4.86710779e-02
2.68819958e-01 2.76422024e-01 -4.39697266e-01 1.39187753e-01
5.74124992e-01 -4.08616781e-01 -9.00839269e-01 3.60595554e-01
-8.42765033e-01 1.08918674e-01 8.18770409e-01 9.76198912e-01
8.37245882e-01 -2.92002529e-01 6.01235688e-01 -8.93700838e-01
1.30464122e-01 -3.43794823e-01 -5.61110437e-01 3.35994631e-01
-2.02629380e-02 -2.14860618e-01 6.24947727e-01 -5.84637582e-01
-1.01159799e+00 3.86523366e-01 1.31765768e-01 -6.06151760e-01
-2.61310101e-01 4.40095395e-01 5.04773371e-02 -2.26418853e-01
3.61810535e-01 2.46403515e-01 -5.28481662e-01 -3.49403948e-01
5.45714974e-01 4.40848112e-01 6.57977343e-01 -1.55240506e-01
4.37713772e-01 4.44265842e-01 -6.33172914e-02 -6.36225879e-01
-9.50533569e-01 -6.12380564e-01 -8.56409550e-01 -3.24676782e-01
1.03907669e+00 -9.69600976e-01 -7.24023998e-01 6.13409519e-01
-1.16791451e+00 -4.30296898e-01 -5.59479415e-01 4.78259236e-01
-2.39312246e-01 -3.82361072e-03 -7.18454063e-01 -5.78917384e-01
-1.73496246e-01 -1.28499758e+00 8.38217497e-01 7.03665972e-01
1.75287500e-01 -1.04474223e+00 -1.54159442e-01 -3.31724375e-01
6.34093404e-01 2.62080103e-01 4.25002784e-01 -3.85524511e-01
-8.52173030e-01 -2.40395591e-01 -6.30911112e-01 4.27532077e-01
4.39572424e-01 5.65560423e-02 -1.09289563e+00 -9.17273238e-02
-1.53100684e-01 -3.17047089e-01 1.53886664e+00 3.66397947e-01
1.44115305e+00 -1.44226596e-01 -4.04788882e-01 9.20964479e-01
1.50211179e+00 1.78913876e-01 7.77012587e-01 4.09717679e-01
9.10583317e-01 2.94388324e-01 4.09795403e-01 1.41510427e-01
3.19611341e-01 2.39739701e-01 5.16164839e-01 -4.63241965e-01
-3.25355202e-01 -1.80337489e-01 2.74121970e-01 7.55614758e-01
5.50441034e-02 2.84625143e-01 -8.64780724e-01 6.07743919e-01
-2.17031193e+00 -1.13012612e+00 3.29473205e-02 1.75239789e+00
4.27741379e-01 2.73368925e-01 -1.92302927e-01 -1.19893871e-01
7.66144574e-01 5.45918107e-01 -8.31772983e-01 -4.32960659e-01
-3.08036089e-01 7.01659918e-02 3.75219703e-01 1.86726391e-01
-1.40180719e+00 1.30596340e+00 6.59534311e+00 5.12806535e-01
-1.41396713e+00 -2.44117841e-01 8.01674187e-01 7.05100670e-02
-2.15736181e-02 3.93197425e-02 -9.27355587e-01 2.40632042e-01
4.12179708e-01 4.64725405e-01 1.44123554e-01 1.13794708e+00
-7.02096596e-02 -1.21445037e-01 -9.12035465e-01 8.73854816e-01
6.57144189e-02 -1.70751798e+00 1.06091380e-01 -2.94258624e-01
1.00081313e+00 3.64189625e-01 1.43697008e-01 6.20198071e-01
4.58068162e-01 -1.25823009e+00 4.51517612e-01 6.12035513e-01
7.91591167e-01 -9.12395298e-01 7.84411013e-01 -8.23288485e-02
-1.97092164e+00 -4.29506928e-01 -5.45865297e-01 -6.65334985e-02
-1.11355297e-02 4.79009390e-01 -1.82469919e-01 5.91013312e-01
8.57788086e-01 1.38632667e+00 -6.81357861e-01 1.11319625e+00
-3.73167515e-01 5.30455768e-01 -4.17602248e-02 7.44296089e-02
7.60505617e-01 -2.29932442e-01 1.23018406e-01 1.76017666e+00
-9.33041051e-02 6.66664541e-02 4.77320284e-01 7.80017555e-01
-1.59751251e-01 -2.46037602e-01 -6.59805059e-01 2.52698451e-01
2.68659592e-01 1.48543131e+00 -9.25165653e-01 -5.70555687e-01
-7.46249437e-01 9.90715325e-01 4.66259629e-01 7.24470794e-01
-7.24764347e-01 -7.90866375e-01 8.25117886e-01 -4.73577261e-01
7.31670380e-01 -4.67258543e-01 -3.77852678e-01 -1.25309801e+00
-4.91296612e-02 -3.49383742e-01 4.15405661e-01 -4.57693905e-01
-1.02768421e+00 8.17925751e-01 -4.31394935e-01 -1.55080044e+00
-7.65774921e-02 -7.23628938e-01 -1.02681220e+00 8.18897069e-01
-2.16393113e+00 -1.35881853e+00 -5.95351696e-01 7.29031384e-01
8.51818502e-01 -1.34639636e-01 5.88086247e-01 2.55755544e-01
-9.15911913e-01 5.01499474e-01 -2.09644273e-01 6.37579679e-01
3.98031473e-01 -9.47142303e-01 6.07924342e-01 1.09304798e+00
2.21338589e-02 7.87649214e-01 -4.97153960e-03 -5.34948945e-01
-1.16977656e+00 -1.67897332e+00 6.92094505e-01 1.05927765e-01
6.91028297e-01 -3.64841223e-01 -9.37173784e-01 8.11812401e-01
2.62194604e-01 8.26923430e-01 6.15314722e-01 1.50939068e-02
-6.53525412e-01 -7.04314932e-02 -7.31332004e-01 7.48375714e-01
1.41688919e+00 -7.17408359e-01 -5.29893339e-01 -1.76445886e-01
9.76329088e-01 -4.15591568e-01 -4.87955928e-01 5.78327954e-01
5.71535826e-01 -8.97910297e-01 9.51648057e-01 -6.56791747e-01
4.05318052e-01 -5.84076822e-01 -3.84530216e-01 -8.46790731e-01
-4.45895761e-01 -4.74022657e-01 -2.00951114e-01 1.02037036e+00
1.36241212e-01 -5.01835585e-01 7.37010300e-01 1.14635348e-01
-3.26981634e-01 -9.93179142e-01 -6.75915420e-01 -5.15224814e-01
-1.44409835e-01 -4.63164061e-01 5.31024039e-01 1.04942417e+00
-5.01717746e-01 3.89121890e-01 -1.07135952e-01 2.45558023e-02
3.19459885e-01 5.27671635e-01 4.13941354e-01 -1.01254117e+00
1.51810065e-01 -6.89796150e-01 -5.63708007e-01 -1.53259432e+00
1.94964007e-01 -8.35260153e-01 -8.23109448e-02 -1.63172948e+00
3.49358767e-02 -5.32033980e-01 -5.82238853e-01 7.59600639e-01
-2.76652187e-01 4.09912258e-01 1.86484814e-01 1.39968142e-01
-8.26742470e-01 7.49234200e-01 1.27689254e+00 -1.92615345e-01
-2.86648631e-01 -9.04865284e-03 -4.51304674e-01 8.12900722e-01
8.12104225e-01 -1.47240549e-01 -3.52271885e-01 -4.59729582e-01
1.37434214e-01 -4.32520121e-01 4.55289155e-01 -1.03301847e+00
7.19688118e-01 -1.61319852e-01 8.59032452e-01 -9.83320415e-01
7.02329352e-02 -7.92397439e-01 -1.75968587e-01 4.62599427e-01
-4.19926614e-01 1.55336425e-01 2.43431434e-01 5.66302061e-01
-5.61485469e-01 -8.30681548e-02 6.14568472e-01 -3.70912224e-01
-1.28084838e+00 4.69877452e-01 -5.09662628e-01 -2.75675148e-01
9.36822474e-01 -3.84163529e-01 -4.46223378e-01 6.25062957e-02
-6.43974960e-01 2.63039470e-01 4.40650322e-02 5.66012442e-01
8.73099148e-01 -1.60348475e+00 -3.68864149e-01 2.49515772e-01
3.96320708e-02 6.07148409e-01 2.60603815e-01 5.81523657e-01
-3.67751449e-01 3.95144939e-01 -1.65263206e-01 -7.65131474e-01
-8.21685731e-01 5.22298574e-01 5.54331243e-01 -1.90882012e-01
-8.88271511e-01 9.19889271e-01 3.36557418e-01 -4.78246808e-01
4.71630096e-01 -6.61426604e-01 -6.17011845e-01 7.59992525e-02
8.01851332e-01 1.14753820e-01 -1.59027174e-01 -4.67179209e-01
-1.75476894e-01 8.38035405e-01 -3.56084973e-01 3.05003762e-01
1.25864923e+00 -3.93961295e-02 -2.17697218e-01 5.62337220e-01
1.42316818e+00 -3.58175486e-01 -1.57874739e+00 -6.49074554e-01
1.48620605e-01 -3.50219160e-01 4.24012542e-02 -4.43670988e-01
-1.41085911e+00 9.66378689e-01 4.67258126e-01 -4.82881851e-02
1.32774580e+00 -2.37673089e-01 5.96692085e-01 5.92224658e-01
1.24785053e-02 -1.06121910e+00 5.07714748e-01 1.05225098e+00
8.10333550e-01 -1.17882216e+00 -2.71305740e-02 -2.46600702e-01
-5.38660467e-01 1.36227930e+00 9.98284042e-01 -6.16872847e-01
8.00633192e-01 1.91997632e-01 8.92699733e-02 -1.39783278e-01
-7.85138071e-01 -4.21099931e-01 2.13137507e-01 2.68392444e-01
5.57667434e-01 -1.83898792e-01 1.38041541e-01 5.47469199e-01
1.96358323e-01 -5.44559658e-02 2.77230263e-01 1.24933743e+00
-4.80286688e-01 -8.87272835e-01 1.12672806e-01 1.46826774e-01
-2.88025111e-01 -1.09990798e-01 -1.59243166e-01 7.01040685e-01
1.24061547e-01 6.41796827e-01 3.94365638e-01 -4.91215795e-01
2.81395584e-01 -1.83280274e-01 -4.00070809e-02 -6.48494840e-01
-8.36607635e-01 2.91809049e-02 -2.37124696e-01 -8.99661362e-01
-6.56975806e-01 -3.82817507e-01 -1.35250366e+00 1.43449949e-02
-3.70810837e-01 -2.45290563e-01 5.28371453e-01 7.41863489e-01
2.82484680e-01 1.09176803e+00 7.95334041e-01 -9.42950070e-01
1.00733139e-01 -7.64761150e-01 -1.18375964e-01 2.18128413e-02
6.05204880e-01 -6.05499566e-01 -3.99588607e-02 -5.99411540e-02]
|
[9.524770736694336, 0.49369657039642334]
|
e7dfab64-bfc3-4738-98d6-c9872a5ed9d0
|
latent-dirichlet-allocation-based
|
1511.05076
| null |
http://arxiv.org/abs/1511.05076v1
|
http://arxiv.org/pdf/1511.05076v1.pdf
|
Latent Dirichlet Allocation Based Organisation of Broadcast Media Archives for Deep Neural Network Adaptation
|
This paper presents a new method for the discovery of latent domains in
diverse speech data, for the use of adaptation of Deep Neural Networks (DNNs)
for Automatic Speech Recognition. Our work focuses on transcription of
multi-genre broadcast media, which is often only categorised broadly in terms
of high level genres such as sports, news, documentary, etc. However, in terms
of acoustic modelling these categories are coarse. Instead, it is expected that
a mixture of latent domains can better represent the complex and diverse
behaviours within a TV show, and therefore lead to better and more robust
performance. We propose a new method, whereby these latent domains are
discovered with Latent Dirichlet Allocation, in an unsupervised manner. These
are used to adapt DNNs using the Unique Binary Code (UBIC) representation for
the LDA domains. Experiments conducted on a set of BBC TV broadcasts, with more
than 2,000 shows for training and 47 shows for testing, show that the use of
LDA-UBIC DNNs reduces the error up to 13% relative compared to the baseline
hybrid DNN models.
|
['Raymond W. M. Ng', 'Oscar Saz', 'Mortaza Doulaty', 'Thomas Hain']
|
2015-11-16
| null | null | null | null |
['acoustic-modelling']
|
['speech']
|
[ 3.71015146e-02 -4.65238579e-02 -4.42699715e-02 -5.79162776e-01
-1.07917309e+00 -7.05274582e-01 8.28703701e-01 -2.46944562e-01
-1.28961116e-01 7.06278324e-01 8.16694677e-01 -1.02584049e-01
-1.02335833e-01 -4.35650349e-01 -5.94303608e-01 -9.67009544e-01
8.47088024e-02 8.75350893e-01 3.69701952e-01 -8.31362754e-02
-8.18764642e-02 4.55402732e-01 -1.76175821e+00 6.86397612e-01
2.52606034e-01 8.53904903e-01 2.76791364e-01 6.31479084e-01
-2.76498556e-01 5.08232474e-01 -8.81686985e-01 -1.66821942e-01
-1.56322420e-01 -4.26032126e-01 -7.74851918e-01 2.42579773e-01
3.14967781e-01 -3.04612160e-01 -1.10188469e-01 9.13168311e-01
6.88954294e-01 2.91283935e-01 8.52881312e-01 -7.98587263e-01
-2.87502825e-01 9.14878190e-01 9.96096507e-02 2.73755312e-01
2.97551394e-01 -2.12218747e-01 8.99765909e-01 -6.11155808e-01
6.60595000e-01 1.44818866e+00 6.06180429e-01 7.35192657e-01
-1.22677374e+00 -7.72196710e-01 -7.67624378e-03 2.59488463e-01
-1.25338733e+00 -9.25832689e-01 8.93567383e-01 -3.25210363e-01
7.98471510e-01 1.59070164e-01 5.19587278e-01 1.91877675e+00
-2.83387303e-01 8.65037799e-01 9.37732279e-01 -6.62316740e-01
5.66623211e-01 2.72403806e-01 1.34373584e-03 -3.64053398e-02
-4.07632828e-01 -2.13747561e-01 -6.15695953e-01 -2.84733742e-01
4.75421101e-01 -2.96978027e-01 6.55139163e-02 -2.91491181e-01
-9.39699054e-01 1.11284590e+00 -3.96793514e-01 8.33280504e-01
-6.90845370e-01 -2.65334904e-01 7.02887297e-01 1.90857485e-01
9.12020147e-01 8.59371349e-02 -2.25290477e-01 -7.46828675e-01
-1.16516793e+00 1.86912090e-01 1.04671919e+00 8.71451378e-01
4.41657603e-01 3.24367136e-01 1.79267935e-02 1.61580956e+00
4.89762574e-01 3.40863466e-01 8.37703645e-01 -8.58553469e-01
2.60132909e-01 2.36136764e-01 -1.98493123e-01 -7.57733703e-01
-1.87334001e-01 -3.21967334e-01 -7.27565885e-01 -4.21527267e-01
3.18443403e-02 -1.81083754e-01 -1.28672671e+00 1.57516348e+00
2.54466474e-01 7.29668736e-02 3.38342041e-01 5.61275423e-01
7.71728098e-01 1.23038316e+00 1.96387079e-02 -3.25928003e-01
9.68931437e-01 -7.96744943e-01 -9.17561948e-01 -9.31605250e-02
4.26345348e-01 -9.00482476e-01 9.11599755e-01 6.76971853e-01
-1.03838265e+00 -6.20184660e-01 -8.22107136e-01 2.49595314e-01
-3.63355875e-01 -8.05903226e-02 1.73340917e-01 9.26533937e-01
-1.27100527e+00 2.17221320e-01 -8.33017588e-01 -7.85168529e-01
4.04593758e-02 3.31671089e-01 -4.69446406e-02 -1.00884914e-01
-1.39494848e+00 6.17559314e-01 3.76649022e-01 -2.39770159e-01
-1.37625456e+00 -1.21456265e-01 -6.24277771e-01 9.32572037e-02
-7.99684692e-03 1.86921194e-01 1.30388153e+00 -1.11780131e+00
-1.87636864e+00 6.45543694e-01 -1.25055894e-01 -7.73493350e-01
2.10943371e-02 -3.05245146e-02 -8.43774259e-01 1.80440173e-01
4.71763946e-02 6.88076377e-01 8.00217986e-01 -1.33334196e+00
-6.83679044e-01 -3.24423090e-02 -1.22094266e-01 9.44453403e-02
-7.35581517e-01 2.67013580e-01 -5.46900511e-01 -6.39570534e-01
3.90072823e-01 -1.13609743e+00 1.97186962e-01 -8.44132602e-01
-6.49755895e-02 -4.97841477e-01 1.03061390e+00 -8.25463116e-01
1.33296359e+00 -2.40010715e+00 2.07492039e-01 3.46625417e-01
-1.24067806e-01 2.36549273e-01 1.84044093e-01 4.31484669e-01
-1.04351819e-01 -6.17843913e-03 -3.43178995e-02 -5.63032925e-01
2.49312103e-01 7.76520193e-01 -2.88066179e-01 1.66303203e-01
-3.88973296e-01 1.78048968e-01 -4.31654721e-01 -2.87886143e-01
7.83029422e-02 4.94042724e-01 -4.79527444e-01 7.47443736e-02
-1.93706706e-01 3.26490849e-01 -1.87907323e-01 4.57719386e-01
4.94386405e-01 2.71776915e-01 4.43861872e-01 9.49706137e-02
-2.36724868e-01 7.63794243e-01 -1.03622258e+00 1.66162920e+00
-4.29463416e-01 9.47774351e-01 1.52011856e-01 -1.25827754e+00
1.24390507e+00 9.06622529e-01 4.74961758e-01 -4.78213608e-01
-6.95334598e-02 3.46628338e-01 6.07012250e-02 -2.72098869e-01
4.08544183e-01 -2.28353530e-01 -2.50118762e-01 3.22181582e-01
3.84699911e-01 -1.37268066e-01 1.73197970e-01 -6.98763803e-02
1.03505945e+00 -2.03036755e-01 -1.39064059e-01 -2.28945419e-01
2.86905080e-01 -8.92943665e-02 5.38725734e-01 5.91563046e-01
-2.66981512e-01 7.19565868e-01 5.35407126e-01 -5.30298769e-01
-1.34017432e+00 -9.24861670e-01 -2.00028718e-01 1.34209168e+00
-5.50048530e-01 -8.03851262e-02 -8.59064758e-01 -4.97082323e-01
-4.12742019e-01 9.66976285e-01 -1.18060283e-01 -8.49024020e-03
-4.75308985e-01 -6.57623410e-01 7.63443351e-01 2.24862367e-01
3.55538070e-01 -1.20418596e+00 1.42345324e-01 5.43671191e-01
-5.16873658e-01 -1.28534234e+00 -6.07054792e-02 5.39515972e-01
-8.19988668e-01 -1.63491547e-01 -9.68766510e-01 -1.01580060e+00
1.14320256e-01 -1.28558904e-01 9.91782546e-01 -5.22371531e-01
3.91873717e-01 4.74826127e-01 -8.18340302e-01 -2.62666464e-01
-1.14106107e+00 2.88595647e-01 4.38916802e-01 2.65184760e-01
7.71705389e-01 -9.35617864e-01 -1.06230661e-01 3.23312759e-01
-9.07373011e-01 -2.18804911e-01 5.50172865e-01 7.11620331e-01
3.42914820e-01 1.39206365e-01 6.40275657e-01 -6.61085546e-01
7.86101341e-01 -5.82545996e-01 -3.10572475e-01 -7.64815956e-02
-2.02020302e-01 -1.92730635e-01 3.87977928e-01 -7.16254830e-01
-1.15053093e+00 -1.88368314e-03 -6.78415358e-01 -3.84111613e-01
-8.63381088e-01 3.87350351e-01 -3.33479941e-01 2.79342681e-01
7.03580320e-01 4.60702360e-01 -2.90435612e-01 -7.16288507e-01
-7.03509385e-03 1.38107812e+00 1.96227089e-01 -5.95076323e-01
3.03611428e-01 2.16424167e-01 -7.67195880e-01 -1.19886875e+00
-4.66742039e-01 -6.78887486e-01 -6.94570184e-01 -4.10706818e-01
9.62157845e-01 -1.00140965e+00 5.74331582e-02 4.05471772e-01
-1.26797271e+00 -3.69448721e-01 -1.86733872e-01 8.26586306e-01
-6.25652432e-01 1.88972741e-01 -6.24710500e-01 -9.13297892e-01
1.48184612e-01 -1.22863460e+00 1.01985288e+00 -1.00282721e-01
-5.16132176e-01 -1.05846906e+00 2.33627915e-01 4.80234295e-01
3.66921693e-01 -1.18260443e-01 1.17114615e+00 -1.32076979e+00
-2.05955505e-01 -9.53835994e-02 3.95357370e-01 1.00494516e+00
1.36671036e-01 -2.00662240e-01 -1.34035492e+00 -2.81735789e-02
1.75288022e-01 -2.58277774e-01 4.85897958e-01 5.33127725e-01
9.34409440e-01 -2.77110547e-01 -2.40509793e-01 1.86827362e-01
8.98926079e-01 8.63017201e-01 8.38085592e-01 4.00527954e-01
4.34990168e-01 5.22549629e-01 1.98086083e-01 4.10322428e-01
2.02048630e-01 8.48441303e-01 1.37347296e-01 -3.26236081e-03
-1.03028513e-01 1.90326318e-01 6.54967129e-01 1.66629839e+00
-4.44442853e-02 -4.31376129e-01 -1.20313108e+00 1.00233006e+00
-1.49409711e+00 -9.37593699e-01 1.62691921e-01 1.99709284e+00
9.66210902e-01 4.73197311e-01 2.79009402e-01 1.72999129e-01
9.14137840e-01 4.78078686e-02 -1.65129453e-01 -8.48658025e-01
-1.94918424e-01 7.38125741e-02 2.19262838e-01 2.66191304e-01
-1.18110502e+00 8.16951931e-01 6.44551897e+00 1.26089621e+00
-1.03291500e+00 4.59055156e-01 7.55452573e-01 -9.38401222e-02
-1.29861400e-01 -2.44987980e-01 -1.08378017e+00 6.36676192e-01
1.65325212e+00 4.90568280e-01 2.56111413e-01 9.44236279e-01
3.27014178e-01 1.97398916e-01 -9.92372394e-01 8.49793315e-01
2.93414950e-01 -1.19743538e+00 1.16979279e-01 2.05367073e-01
9.20064926e-01 1.24616250e-01 9.86398607e-02 5.86030424e-01
4.63394850e-01 -5.57663321e-01 8.88841212e-01 2.88354844e-01
3.82508188e-01 -8.42894971e-01 8.65756035e-01 4.89557832e-01
-6.43013954e-01 -7.47185796e-02 -4.08650726e-01 2.14921609e-01
1.62764668e-01 3.43256354e-01 -1.19403124e+00 2.22066104e-01
9.53798473e-01 3.70659620e-01 -1.18082063e-02 7.83498824e-01
1.57865494e-01 1.26515675e+00 -5.38909435e-01 -1.38770834e-01
6.84716880e-01 1.69907466e-01 7.40602314e-01 1.48885226e+00
5.82103372e-01 -2.22471610e-01 4.76043485e-02 3.72981668e-01
8.10177103e-02 1.37915865e-01 -6.07986510e-01 -1.97567672e-01
4.68560338e-01 9.10763800e-01 -8.78735662e-01 -5.16693890e-01
-2.23852009e-01 8.02252650e-01 -1.74518391e-01 3.71161312e-01
-8.42442334e-01 -5.64570259e-03 3.92259419e-01 1.68455034e-01
5.14141142e-01 -2.90592998e-01 9.38165113e-02 -7.43178248e-01
-2.21648261e-01 -1.34351242e+00 9.57696661e-02 -7.15731919e-01
-1.23365819e+00 9.60686803e-01 3.96355689e-01 -1.25349689e+00
-7.17560112e-01 -3.35962474e-01 -2.36080468e-01 7.37235904e-01
-9.83997941e-01 -1.13853312e+00 2.71276683e-01 4.29685056e-01
1.12598586e+00 -7.67701566e-01 1.13460803e+00 6.91808701e-01
-1.72133073e-01 3.72110248e-01 8.17809522e-01 4.35970165e-02
6.11800790e-01 -1.10163820e+00 3.98659348e-01 6.52823806e-01
4.24212873e-01 3.90345395e-01 7.69768357e-01 -4.19406414e-01
-6.97007179e-01 -8.48184526e-01 1.01679647e+00 -3.97542834e-01
5.01742661e-01 -8.29063058e-01 -9.03211355e-01 5.37419677e-01
4.82352883e-01 -6.24336362e-01 8.40059996e-01 3.12571466e-01
9.12732258e-03 -3.52324337e-01 -8.35300505e-01 3.63727063e-01
6.17567718e-01 -6.44224465e-01 -7.04795361e-01 5.06321549e-01
6.79014981e-01 -2.04966784e-01 -7.31470525e-01 1.04355253e-01
4.83232945e-01 -9.16505396e-01 8.39085460e-01 -2.22503796e-01
1.97809920e-01 1.06939554e-01 -5.56564867e-01 -1.39720082e+00
-1.61868304e-01 -5.13657272e-01 7.84401372e-02 1.70869291e+00
4.85328645e-01 -3.32209259e-01 8.39707911e-01 1.79087907e-01
-4.51616615e-01 -1.40368924e-01 -1.51407409e+00 -7.13475525e-01
-8.10981318e-02 -7.74327040e-01 5.08773088e-01 9.08055723e-01
-3.90641510e-01 3.67223859e-01 -5.40976405e-01 8.43719617e-02
2.02968195e-01 -4.44474429e-01 6.95457101e-01 -1.30423224e+00
-1.03346661e-01 -2.27572232e-01 -6.14580572e-01 -1.28354347e+00
2.58764416e-01 -4.92016941e-01 3.85882795e-01 -1.34763956e+00
-2.51052529e-01 -4.22068566e-01 -2.08364472e-01 2.69066066e-01
8.31714094e-01 2.09901795e-01 -8.68007392e-02 2.22253978e-01
-5.48317671e-01 3.96967709e-01 6.82562590e-01 -2.10906297e-01
-2.59019524e-01 2.73023158e-01 -2.79610246e-01 9.06845033e-01
6.19669974e-01 -8.51585567e-01 -2.69329935e-01 -3.35774034e-01
-1.59658417e-01 -1.85288899e-02 8.23356491e-03 -1.40096533e+00
2.19164401e-01 1.86238691e-01 1.21516012e-01 -9.16214347e-01
7.53219962e-01 -9.36511517e-01 2.18958259e-01 3.73191163e-02
-4.80856419e-01 -2.73913950e-01 2.92565376e-01 3.92928094e-01
-6.56212986e-01 -4.36042249e-01 5.90731323e-01 -1.44979328e-01
-7.17120349e-01 -2.30784446e-01 -1.06670749e+00 -4.23356771e-01
5.74537814e-01 -3.79511595e-01 3.24175268e-01 -9.23731685e-01
-1.13394606e+00 -3.93992186e-01 -5.47873452e-02 5.11740088e-01
3.22528154e-01 -1.49756253e+00 -5.87888837e-01 2.66875356e-01
-8.62451494e-02 -1.30221412e-01 3.76947135e-01 5.10108292e-01
-3.88036788e-01 5.78934371e-01 -9.71715227e-02 -9.36532617e-01
-1.29903543e+00 -2.72837169e-02 9.31506753e-02 -2.16767043e-01
-3.66908103e-01 8.41787100e-01 2.44715437e-01 -4.60694522e-01
6.31491661e-01 -3.33657891e-01 -6.23245358e-01 4.36462671e-01
1.49959683e-01 1.73669755e-01 1.70580342e-01 -9.99772668e-01
-2.25887582e-01 1.59070432e-01 -1.87297195e-01 -5.03623188e-01
1.51398909e+00 -4.07132357e-01 -2.49116533e-02 9.11123455e-01
1.23642623e+00 -8.32521170e-02 -1.01971281e+00 -2.00535566e-01
2.37427652e-01 8.19365121e-03 8.14134106e-02 -6.44131362e-01
-6.17641091e-01 9.83197212e-01 9.44148362e-01 7.59424984e-01
1.08221412e+00 1.58746317e-01 6.88831747e-01 3.55380952e-01
3.80374528e-02 -1.47184491e+00 2.54450236e-02 8.08616161e-01
7.78591752e-01 -7.11467147e-01 -3.85780156e-01 2.59497732e-01
-7.79999435e-01 1.30651498e+00 2.25910828e-01 1.83531046e-01
5.94599068e-01 2.01796725e-01 2.35455349e-01 9.75191407e-03
-7.87418604e-01 3.13968249e-02 1.16108097e-01 7.28259504e-01
4.05704975e-01 6.35648072e-02 -1.08425751e-01 6.83356583e-01
-2.24033475e-01 -3.58616114e-01 4.93083030e-01 7.60270834e-01
-6.69460475e-01 -1.20089734e+00 -6.61100328e-01 1.62586927e-01
-7.04375088e-01 -6.21688105e-02 -1.72701851e-01 6.39654160e-01
2.51578152e-01 1.01107991e+00 2.11878985e-01 -3.68169785e-01
1.53367668e-01 6.42915905e-01 -1.54125959e-01 -7.27451384e-01
-3.12748462e-01 7.48080850e-01 4.73164141e-01 2.64824815e-02
-8.58196318e-01 -8.22626650e-01 -6.08117104e-01 -4.30746302e-02
-1.50050431e-01 4.75499153e-01 1.05233288e+00 1.11373770e+00
7.65862837e-02 4.84714866e-01 5.99728107e-01 -8.38622987e-01
-3.18702251e-01 -1.42546046e+00 -7.03181028e-01 3.55308980e-01
2.31833890e-01 -7.73046851e-01 -5.67066967e-01 5.23639917e-01]
|
[14.476990699768066, 6.652678489685059]
|
808298b7-0437-4f2c-abe5-56ccff74385d
|
convolutional-mesh-regression-for-single
|
1905.03244
| null |
https://arxiv.org/abs/1905.03244v1
|
https://arxiv.org/pdf/1905.03244v1.pdf
|
Convolutional Mesh Regression for Single-Image Human Shape Reconstruction
|
This paper addresses the problem of 3D human pose and shape estimation from a single image. Previous approaches consider a parametric model of the human body, SMPL, and attempt to regress the model parameters that give rise to a mesh consistent with image evidence. This parameter regression has been a very challenging task, with model-based approaches underperforming compared to nonparametric solutions in terms of pose estimation. In our work, we propose to relax this heavy reliance on the model's parameter space. We still retain the topology of the SMPL template mesh, but instead of predicting model parameters, we directly regress the 3D location of the mesh vertices. This is a heavy task for a typical network, but our key insight is that the regression becomes significantly easier using a Graph-CNN. This architecture allows us to explicitly encode the template mesh structure within the network and leverage the spatial locality the mesh has to offer. Image-based features are attached to the mesh vertices and the Graph-CNN is responsible to process them on the mesh structure, while the regression target for each vertex is its 3D location. Having recovered the complete 3D geometry of the mesh, if we still require a specific model parametrization, this can be reliably regressed from the vertices locations. We demonstrate the flexibility and the effectiveness of our proposed graph-based mesh regression by attaching different types of features on the mesh vertices. In all cases, we outperform the comparable baselines relying on model parameter regression, while we also achieve state-of-the-art results among model-based pose estimation approaches.
|
['Nikos Kolotouros', 'Georgios Pavlakos', 'Kostas Daniilidis']
|
2019-05-08
|
convolutional-mesh-regression-for-single-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Kolotouros_Convolutional_Mesh_Regression_for_Single-Image_Human_Shape_Reconstruction_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Kolotouros_Convolutional_Mesh_Regression_for_Single-Image_Human_Shape_Reconstruction_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['3d-human-pose-and-shape-estimation', 'monocular-3d-human-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[-3.70939709e-02 5.09864748e-01 -2.00842977e-01 2.64306273e-02
-6.10767722e-01 -4.75507319e-01 4.06166166e-01 3.87744121e-02
-2.45766416e-01 3.22360456e-01 6.33950755e-02 1.88840091e-01
6.86594844e-02 -6.39570713e-01 -1.08324385e+00 -3.88437718e-01
-1.02123164e-01 1.25989521e+00 3.65421027e-01 -2.38110110e-01
9.75857526e-02 7.82749176e-01 -1.28327429e+00 -1.13196380e-01
2.33934820e-01 1.08139181e+00 -1.57746941e-01 7.42475510e-01
2.61509687e-01 -3.75515991e-03 -2.52808690e-01 -3.14091235e-01
4.37135965e-01 -3.13605592e-02 -7.20045865e-01 3.23504150e-01
8.74283075e-01 -2.54905581e-01 -3.69625211e-01 8.05744886e-01
3.97763580e-01 -4.16021328e-03 9.03726637e-01 -1.01314211e+00
-1.04976876e-03 8.36860985e-02 -7.10285187e-01 -3.92578959e-01
4.42458510e-01 9.08102542e-02 1.06588972e+00 -9.69400465e-01
1.06654155e+00 1.32165956e+00 9.66215849e-01 5.97807586e-01
-1.54464614e+00 -4.93659139e-01 1.34027749e-01 -2.86621571e-01
-1.67032158e+00 -3.44375670e-01 1.04081929e+00 -7.13012278e-01
6.31222367e-01 3.79211791e-02 9.75522935e-01 9.87905264e-01
3.67832720e-01 4.54818875e-01 7.34793007e-01 -3.30675989e-01
1.98098831e-02 -2.66591072e-01 -2.93374807e-01 1.15331542e+00
9.02736410e-02 -3.22460495e-02 -4.82993096e-01 -6.92603365e-02
1.40173924e+00 -1.03486538e-01 -9.76713076e-02 -1.05241430e+00
-1.19494963e+00 7.72289336e-01 6.69740796e-01 -1.80747941e-01
-4.85475183e-01 5.11217177e-01 2.57358998e-01 -8.28664228e-02
6.22097313e-01 3.94069374e-01 -5.93809009e-01 1.75711736e-01
-1.19929779e+00 5.92071533e-01 9.59542036e-01 9.37104702e-01
9.09214556e-01 -2.39110366e-01 7.41444752e-02 5.63892543e-01
4.09238398e-01 4.52252597e-01 -2.80792177e-01 -1.19429564e+00
3.47707659e-01 7.55179763e-01 -2.66106352e-02 -1.38882899e+00
-6.68023467e-01 -4.43796992e-01 -7.99091637e-01 3.00000519e-01
5.47906220e-01 -1.59102492e-02 -1.16640723e+00 1.63441181e+00
6.64009929e-01 2.30603412e-01 -6.20771587e-01 1.02824581e+00
6.61560416e-01 2.74473637e-01 -1.68480590e-01 2.48731270e-01
1.29367363e+00 -8.83942306e-01 -9.58584175e-02 -3.08176875e-01
3.84756476e-01 -5.43596804e-01 8.29178870e-01 3.95956278e-01
-1.20463550e+00 -4.34648216e-01 -1.01819396e+00 -1.42596290e-01
-1.84218232e-02 6.00757785e-02 3.31025481e-01 2.55485773e-01
-1.08092868e+00 8.48741710e-01 -1.02298415e+00 -3.44426841e-01
3.17918479e-01 7.92122424e-01 -7.02432215e-01 1.08387537e-01
-7.38068759e-01 1.01026845e+00 3.63222510e-02 3.37836444e-01
-7.71673560e-01 -6.68954849e-01 -1.05224133e+00 -1.14397041e-01
6.83972239e-01 -1.18099236e+00 9.94426608e-01 -7.27174759e-01
-1.43235159e+00 1.13682961e+00 -4.32714038e-02 -2.23593846e-01
8.37403119e-01 -8.08679760e-02 2.72796065e-01 3.11723113e-01
-1.09181613e-01 8.24314833e-01 1.10934842e+00 -1.51343834e+00
-6.48479015e-02 -4.23843175e-01 1.18580207e-01 9.48521420e-02
1.64665997e-01 -4.65299875e-01 -9.07356381e-01 -5.80529809e-01
3.87081265e-01 -1.39817989e+00 -4.21471089e-01 5.09846151e-01
-6.32179677e-01 -9.58044529e-02 5.52325249e-01 -8.23924303e-01
9.34104323e-01 -1.81229961e+00 6.21567070e-01 6.86829209e-01
5.14893770e-01 -1.51640192e-01 7.17074238e-03 3.21010619e-01
6.29521534e-02 8.64282772e-02 -2.58422136e-01 -7.01542616e-01
1.49924289e-02 2.70729542e-01 7.88323805e-02 9.40833867e-01
2.77438700e-01 1.07484949e+00 -5.46864152e-01 -7.63856351e-01
4.68353570e-01 6.68322802e-01 -9.70726371e-01 1.73211336e-01
-5.47196865e-01 7.23893523e-01 -5.77480137e-01 5.57428539e-01
5.05605996e-01 -4.79677320e-01 3.56172115e-01 -5.77728152e-01
2.62927890e-01 2.71714590e-02 -1.27854705e+00 1.98150659e+00
-5.00890613e-01 1.01153500e-01 3.38562340e-01 -9.44948316e-01
7.53419936e-01 1.55560285e-01 7.25675166e-01 -1.96812496e-01
2.54005998e-01 6.57961741e-02 -1.49548337e-01 -1.13892011e-01
1.50045335e-01 -1.99541256e-01 -1.63465112e-01 1.85052261e-01
8.23353454e-02 -4.19060111e-01 -3.71483803e-01 -8.90356116e-03
9.61482346e-01 6.80356681e-01 2.35941470e-01 -2.61475295e-01
3.83509934e-01 5.10649718e-02 2.81525880e-01 2.53418416e-01
3.17833602e-01 9.30333793e-01 5.82855582e-01 -5.20713449e-01
-1.26444340e+00 -1.05071545e+00 2.21268982e-02 7.51411319e-01
1.22894034e-01 -5.50467849e-01 -8.50733757e-01 -7.69532382e-01
2.37600520e-01 8.32096711e-02 -9.55344737e-01 1.30227255e-03
-1.08561778e+00 -2.16203593e-02 2.64422357e-01 3.94177526e-01
7.94790015e-02 -8.18374097e-01 -4.44572806e-01 1.19508654e-01
-6.75544068e-02 -1.19103193e+00 -5.54364443e-01 7.75925592e-02
-8.81588578e-01 -1.18620408e+00 -5.15690506e-01 -6.12770438e-01
9.38604057e-01 -3.81126374e-01 1.29574251e+00 4.92240489e-01
-4.13654596e-01 4.20509458e-01 -1.25693856e-02 -3.99525687e-02
-2.41714433e-01 3.22917014e-01 -6.48144912e-03 -4.32093106e-02
-4.92082089e-01 -8.47854555e-01 -8.45092475e-01 3.06032568e-01
-5.56348026e-01 1.13516256e-01 4.33007628e-01 6.71430051e-01
9.89677668e-01 -2.47512743e-01 -5.99985681e-02 -9.80578244e-01
-6.84950575e-02 -2.75907665e-01 -4.71867830e-01 -6.06340915e-02
-2.81815261e-01 2.07931951e-01 5.09582222e-01 -5.55424213e-01
-4.10215050e-01 5.96481085e-01 -1.17448874e-01 -9.31087255e-01
-3.16979513e-02 3.29394519e-01 -9.85026509e-02 -3.29115182e-01
5.57269156e-01 -3.02436978e-01 3.59017313e-01 -6.52513862e-01
3.52968156e-01 -5.88351041e-02 5.25427043e-01 -8.63844395e-01
9.40554857e-01 6.03644729e-01 6.80003285e-01 -6.49533510e-01
-8.16062510e-01 -2.79918313e-01 -1.13409281e+00 -3.71417820e-01
9.71186757e-01 -8.51121485e-01 -8.63295794e-01 3.41552682e-02
-1.30757105e+00 -4.10679907e-01 -1.98317230e-01 1.37290135e-01
-8.71827066e-01 3.29391956e-01 -6.04967952e-01 -5.74895382e-01
-1.89657599e-01 -1.19665778e+00 1.60630584e+00 -4.79972035e-01
-5.29668689e-01 -1.14965904e+00 -1.66011192e-02 1.96617052e-01
2.63246801e-03 7.84242749e-01 9.65070248e-01 -3.59702319e-01
-6.54734135e-01 -4.44202155e-01 8.00599456e-02 4.95689623e-02
-2.19308659e-01 1.30071091e-02 -7.50668585e-01 -3.90742123e-01
-3.08080494e-01 -1.68329686e-01 6.28974497e-01 6.14210010e-01
1.23271430e+00 -2.31320411e-01 -4.72920537e-01 8.61878633e-01
1.26954961e+00 -8.58356178e-01 3.42808545e-01 1.12793408e-01
1.15871656e+00 7.61180103e-01 4.76199090e-01 3.31779361e-01
4.79636312e-01 1.01810598e+00 7.60834038e-01 -3.08089256e-01
-3.50011587e-01 -6.24311090e-01 1.33692443e-01 7.23858833e-01
-4.44847286e-01 3.52809317e-02 -8.91829312e-01 2.40068465e-01
-1.88745928e+00 -3.86560649e-01 -1.23479135e-01 2.34636879e+00
6.57584548e-01 2.43403554e-01 4.02401000e-01 -1.97430953e-01
5.44716060e-01 2.00905755e-01 -4.90637004e-01 -9.38924700e-02
2.83310056e-01 4.34252530e-01 6.58295870e-01 6.08287156e-01
-1.07187033e+00 1.05426919e+00 6.68969250e+00 5.95351994e-01
-1.14707220e+00 -1.03913181e-01 3.08655113e-01 -1.84184387e-01
-7.69051090e-02 6.52682260e-02 -7.58718252e-01 1.87789619e-01
6.69746161e-01 3.47694844e-01 4.59365070e-01 6.89328313e-01
1.68266684e-01 -2.96472814e-02 -1.55543828e+00 9.79978919e-01
7.13453963e-02 -1.20944023e+00 1.94521129e-01 5.12722135e-01
4.36029106e-01 -1.70256123e-01 -1.02603465e-01 -2.69284863e-02
5.96242882e-02 -1.30490911e+00 9.48206425e-01 8.05194557e-01
1.01136935e+00 -6.72932029e-01 4.31083858e-01 5.07923067e-01
-1.23824275e+00 4.07434165e-01 -1.98187977e-01 -3.60051021e-02
1.37660146e-01 4.68865395e-01 -9.52987611e-01 5.08952379e-01
5.06588221e-01 6.53698564e-01 -5.71864665e-01 7.72106647e-01
-2.38081679e-01 3.89005423e-01 -5.01514018e-01 3.97539645e-01
-1.25866115e-01 -1.07678853e-01 5.31865239e-01 9.41998124e-01
1.75800219e-01 -2.22063288e-01 6.07847154e-01 9.21025038e-01
-2.01322168e-01 1.61372542e-01 -5.37635088e-01 3.36868614e-01
1.56162947e-01 1.26269007e+00 -9.16068316e-01 -2.01943174e-01
-1.49076149e-01 8.90663564e-01 5.77866495e-01 1.12348996e-01
-7.41633534e-01 2.87109137e-01 4.97206032e-01 7.50660837e-01
4.70276326e-01 -5.20132005e-01 -2.71801889e-01 -1.01141047e+00
2.12433875e-01 -7.66929150e-01 1.33168787e-01 -6.28683388e-01
-1.16860104e+00 2.69387513e-01 1.58617422e-01 -1.13527954e+00
-4.17445302e-01 -6.45342350e-01 -2.72409081e-01 8.10573041e-01
-1.09666348e+00 -1.58977365e+00 -9.88718048e-02 6.35081232e-01
3.38419437e-01 4.57145810e-01 7.85471857e-01 -1.93234421e-02
-1.94549561e-01 4.51453328e-01 -6.40322447e-01 2.37031743e-01
6.65140986e-01 -1.13118744e+00 4.90073949e-01 3.69964391e-01
1.77244768e-01 6.53833091e-01 8.07084799e-01 -9.74966764e-01
-1.66543102e+00 -9.75582063e-01 5.76207519e-01 -8.04337323e-01
5.42570472e-01 -8.03491950e-01 -7.05425441e-01 8.42614353e-01
-4.39842582e-01 3.36524934e-01 3.94544899e-02 2.18463108e-01
-4.27690774e-01 1.68604180e-01 -9.09418404e-01 5.72918892e-01
1.24304283e+00 -4.60357577e-01 -3.73556882e-01 3.93703371e-01
5.65610826e-01 -7.84738898e-01 -1.08119655e+00 5.06213903e-01
8.01717997e-01 -6.49281800e-01 1.32998812e+00 -5.25483429e-01
4.64491338e-01 -3.69951993e-01 7.57783279e-02 -1.12084329e+00
-2.48591825e-01 -4.32260573e-01 -2.66600847e-01 7.24845827e-01
3.49132180e-01 -1.45902574e-01 1.21565485e+00 6.63864911e-01
-4.06393260e-02 -9.85295177e-01 -1.01374006e+00 -5.12778223e-01
2.03997374e-01 -4.03446823e-01 3.27507764e-01 7.14228034e-01
-4.23459888e-01 2.36917362e-01 -4.92527544e-01 3.95011187e-01
6.79274321e-01 -3.59823667e-02 1.15066350e+00 -1.51727271e+00
-3.92825663e-01 -3.09263498e-01 -5.92295110e-01 -1.21418130e+00
4.07930851e-01 -9.13799167e-01 2.39994109e-01 -1.45944190e+00
8.17544982e-02 -3.89784187e-01 2.54900992e-01 4.56732035e-01
1.46057606e-01 4.54403430e-01 1.88936129e-01 2.31860280e-01
-4.39201027e-01 2.60764956e-01 1.58726645e+00 -1.96687207e-02
-7.17093349e-02 -8.38451646e-03 -7.45240077e-02 1.03983366e+00
3.79748672e-01 -4.11537349e-01 -9.37327296e-02 -2.77355820e-01
2.64334649e-01 3.60762119e-01 8.72379005e-01 -8.00496399e-01
3.62460345e-01 2.63563097e-02 6.97177887e-01 -4.91961092e-01
8.14765096e-01 -1.04913640e+00 4.05128658e-01 3.58742356e-01
-1.74772128e-01 1.16311572e-01 -2.76945392e-03 5.78693867e-01
1.99565381e-01 8.45119879e-02 8.14023137e-01 -3.03160191e-01
-2.52145946e-01 7.33896613e-01 1.26207069e-01 8.55086893e-02
6.90607905e-01 -3.96699250e-01 3.75144243e-01 -3.73893231e-01
-1.18411374e+00 4.49730046e-02 1.00498092e+00 2.28540093e-01
6.69986308e-01 -1.25350428e+00 -5.20239413e-01 2.46338770e-01
-8.81159678e-02 5.56583941e-01 5.51161394e-02 1.04344869e+00
-6.31034672e-01 -3.58205587e-02 -1.25456778e-02 -9.33383703e-01
-1.15533733e+00 4.70789164e-01 5.86987734e-01 -3.51856977e-01
-9.45989907e-01 6.65184081e-01 4.13045824e-01 -5.60730457e-01
2.04561532e-01 -3.31318915e-01 1.22039154e-01 -1.13331735e-01
-2.01226935e-01 1.75034627e-01 1.09303892e-01 -9.59640384e-01
-3.55937660e-01 1.20812285e+00 2.47972742e-01 -1.86066702e-02
1.40001357e+00 1.82788260e-03 -1.76147610e-01 3.42310548e-01
1.28085721e+00 2.36606568e-01 -1.38462257e+00 -1.03086129e-01
-1.29508778e-01 -3.02904814e-01 -7.31498078e-02 -5.24886370e-01
-1.21349537e+00 8.74261558e-01 9.28633884e-02 -1.68534920e-01
5.03011525e-01 3.04758549e-01 7.29002833e-01 4.82088774e-02
6.20437741e-01 -9.59008992e-01 7.76480287e-02 4.48814422e-01
1.03778410e+00 -9.92247343e-01 4.05366838e-01 -9.20559347e-01
-1.67486832e-01 1.16673434e+00 3.97445232e-01 -7.47188091e-01
8.79210532e-01 1.31796286e-01 -2.36981884e-01 -6.51210904e-01
-3.87632042e-01 -5.87921515e-02 8.58403087e-01 4.34477746e-01
3.07957232e-01 6.99346960e-02 1.43751666e-01 2.08953664e-01
-5.42015553e-01 -3.13028157e-01 3.49617340e-02 6.28659070e-01
-2.68358141e-01 -1.15819454e+00 -3.80323917e-01 4.00010288e-01
-3.68954778e-01 1.06690377e-01 -5.05346119e-01 1.15030277e+00
6.94065243e-02 4.53480065e-01 6.18254840e-02 -4.52448666e-01
6.58352733e-01 -1.28138110e-01 8.15843344e-01 -8.94385159e-01
-4.73846674e-01 4.07509029e-01 1.27478287e-01 -9.45307672e-01
-3.79268676e-01 -5.77675283e-01 -1.30890572e+00 -2.76216418e-01
-2.32040137e-01 -2.06857160e-01 7.12567210e-01 8.97324324e-01
2.36631319e-01 2.85351515e-01 2.65266508e-01 -1.52165174e+00
-3.44179749e-01 -6.49386644e-01 -5.70063889e-01 5.24433315e-01
3.95563275e-01 -1.19307089e+00 -3.03537846e-01 4.97305319e-02]
|
[7.030370712280273, -1.2561959028244019]
|
790d2fe3-b96f-4b22-a3e8-72e970f7c456
|
semi-supervised-object-detection-with-object
|
2212.02747
| null |
https://arxiv.org/abs/2212.02747v1
|
https://arxiv.org/pdf/2212.02747v1.pdf
|
Semi-Supervised Object Detection with Object-wise Contrastive Learning and Regression Uncertainty
|
Semi-supervised object detection (SSOD) aims to boost detection performance by leveraging extra unlabeled data. The teacher-student framework has been shown to be promising for SSOD, in which a teacher network generates pseudo-labels for unlabeled data to assist the training of a student network. Since the pseudo-labels are noisy, filtering the pseudo-labels is crucial to exploit the potential of such framework. Unlike existing suboptimal methods, we propose a two-step pseudo-label filtering for the classification and regression heads in a teacher-student framework. For the classification head, OCL (Object-wise Contrastive Learning) regularizes the object representation learning that utilizes unlabeled data to improve pseudo-label filtering by enhancing the discriminativeness of the classification score. This is designed to pull together objects in the same class and push away objects from different classes. For the regression head, we further propose RUPL (Regression-Uncertainty-guided Pseudo-Labeling) to learn the aleatoric uncertainty of object localization for label filtering. By jointly filtering the pseudo-labels for the classification and regression heads, the student network receives better guidance from the teacher network for object detection task. Experimental results on Pascal VOC and MS-COCO datasets demonstrate the superiority of our proposed method with competitive performance compared to existing methods.
|
['Tae-Kyun Kim', 'Xuepeng Shi', 'Zhixiang Chen', 'Honggyu Choi']
|
2022-12-06
| null | null | null | null |
['semi-supervised-object-detection']
|
['computer-vision']
|
[ 2.98849702e-01 3.71285886e-01 -3.33231121e-01 -7.86049545e-01
-9.74460244e-01 -4.73963857e-01 4.41907704e-01 1.95445135e-01
-5.37112892e-01 4.93175030e-01 -3.19657832e-01 -2.62563564e-02
6.32021502e-02 -4.88453746e-01 -7.77637303e-01 -9.81074035e-01
4.26954538e-01 4.57657218e-01 6.72211111e-01 3.93528819e-01
9.84027758e-02 4.87872392e-01 -1.78258157e+00 3.20663631e-01
1.05887175e+00 1.28545880e+00 5.48652887e-01 4.38077807e-01
-8.95166621e-02 7.33667433e-01 -4.71424848e-01 -1.86898708e-02
2.72491276e-01 -2.87482709e-01 -5.17614365e-01 5.28996646e-01
7.51581430e-01 -2.89485604e-01 3.61611471e-02 1.33395815e+00
3.90880108e-01 2.60610253e-01 6.82495236e-01 -1.18589008e+00
-4.55335617e-01 6.73233747e-01 -7.76487589e-01 8.75066295e-02
-2.01593414e-01 -6.83024004e-02 9.45699871e-01 -1.47437239e+00
3.72683465e-01 1.33388233e+00 4.70213622e-01 7.37071455e-01
-1.20503139e+00 -8.70449007e-01 4.72243309e-01 5.95863611e-02
-1.45276845e+00 -1.40469268e-01 6.11064732e-01 -4.24393684e-01
1.52254462e-01 1.87939748e-01 2.81194001e-01 7.27381110e-01
-3.90701324e-01 1.29913640e+00 1.18264294e+00 -5.69229007e-01
2.86358237e-01 6.34522676e-01 6.05306983e-01 7.89214671e-01
2.46852875e-01 3.19228172e-01 -5.23078203e-01 8.38450938e-02
2.80518532e-01 6.74392730e-02 -1.53661758e-01 -6.32246733e-01
-7.47431576e-01 6.51029944e-01 8.15986991e-01 -1.84594810e-01
-2.61113733e-01 -6.11479916e-02 2.48632822e-02 -6.03183806e-02
5.02466083e-01 2.30350077e-01 -5.09650886e-01 5.25463283e-01
-1.03455663e+00 1.87724847e-02 3.55525643e-01 1.14768815e+00
9.61257219e-01 -1.00638069e-01 -7.37369537e-01 9.62877035e-01
7.47585118e-01 4.40946758e-01 2.49399483e-01 -8.39160800e-01
2.04746127e-01 9.22125101e-01 1.03650615e-01 -2.89069623e-01
-1.85918108e-01 -7.55374491e-01 -2.03087062e-01 4.76929247e-01
6.22918487e-01 -8.99778828e-02 -1.45690691e+00 1.59920835e+00
8.79380524e-01 4.31099653e-01 -7.78808892e-02 1.05565047e+00
9.91687715e-01 4.46362644e-01 2.00521007e-01 -1.80640936e-01
1.34822595e+00 -1.36930609e+00 -4.44298625e-01 -3.05350691e-01
6.62093759e-01 -6.61953926e-01 7.59676397e-01 2.83107281e-01
-7.71556139e-01 -8.06212187e-01 -1.01909804e+00 6.33168742e-02
-2.57441908e-01 5.78504980e-01 1.70665622e-01 6.47860885e-01
-5.76504886e-01 2.45425925e-01 -8.55607510e-01 -5.08287661e-02
8.08745980e-01 3.39177042e-01 4.55292240e-02 -3.21774840e-01
-7.61056066e-01 5.15441179e-01 7.20152676e-01 1.28182352e-01
-1.35208225e+00 -6.18864357e-01 -8.27217162e-01 4.21806797e-02
8.71779919e-01 3.02385725e-02 1.18465090e+00 -9.00392830e-01
-1.23580468e+00 7.95595467e-01 2.62786243e-02 -4.27993953e-01
4.41535413e-01 -6.80434704e-02 -5.47179133e-02 1.46250710e-01
1.68953925e-01 1.30384028e+00 1.18759513e+00 -1.50758028e+00
-1.31886625e+00 -4.54159737e-01 -1.03842527e-01 3.23596478e-01
-3.92984539e-01 -1.63254753e-01 -7.34119952e-01 -5.97165823e-01
5.62932611e-01 -1.01290476e+00 -1.74013779e-01 1.88893989e-01
-4.97622788e-01 -6.89331055e-01 1.28548908e+00 -1.53351381e-01
9.42958295e-01 -2.23640275e+00 -1.81796312e-01 1.94440022e-01
2.65037596e-01 5.08611381e-01 -2.96202689e-01 -3.65596056e-01
-7.77762011e-02 -3.69808495e-01 -6.52635843e-02 -6.14818990e-01
-1.83459029e-01 2.85392702e-01 -1.47814468e-01 5.03242791e-01
6.03511274e-01 6.97617650e-01 -1.00633025e+00 -6.78584874e-01
2.00760677e-01 3.36298078e-01 -4.34895277e-01 4.67448413e-01
-4.32994664e-01 5.95267653e-01 -5.55457354e-01 8.18750262e-01
8.81801665e-01 -2.87681639e-01 -2.03670710e-01 -1.41454563e-01
-1.64869726e-02 1.77553207e-01 -1.52448046e+00 1.19685340e+00
-1.19185932e-01 2.71737784e-01 2.94075727e-01 -9.56888974e-01
8.83897543e-01 6.15484677e-02 1.10018715e-01 -2.12789923e-01
2.22371176e-01 1.63797915e-01 -1.15177870e-01 -2.97922492e-01
3.15271527e-01 2.59296536e-01 2.54323214e-01 3.96978170e-01
3.59277397e-01 -1.21636331e-01 3.95416558e-01 3.86165529e-01
6.33060277e-01 3.28810364e-01 -3.64562646e-02 -2.51899064e-01
6.74397647e-01 -1.47114620e-01 8.47396016e-01 9.68893588e-01
-4.22687739e-01 5.32474399e-01 -2.00196332e-03 8.88714101e-03
-4.81005579e-01 -1.00547445e+00 -4.10619885e-01 1.62328708e+00
3.51611912e-01 -2.69454159e-02 -8.34344149e-01 -1.48354661e+00
1.04568966e-01 6.42836511e-01 -6.86375439e-01 -2.13731602e-01
-3.15850079e-01 -6.58313334e-01 2.68364549e-01 7.54270554e-01
2.88854808e-01 -9.51875329e-01 -3.77953619e-01 2.31123101e-02
1.87086433e-01 -1.17450285e+00 -6.73428059e-01 8.40962708e-01
-7.07857668e-01 -1.14956284e+00 -6.02915525e-01 -1.04645586e+00
1.26079547e+00 5.14028132e-01 5.97339392e-01 1.09308399e-01
-2.82611430e-01 2.06233889e-01 -3.33210766e-01 -6.08239949e-01
-4.70801651e-01 -2.34304681e-01 7.84499869e-02 3.82268608e-01
4.40528631e-01 5.32407425e-02 -4.12516624e-01 6.27569914e-01
-7.94527113e-01 -2.09992975e-01 7.00170577e-01 1.04362583e+00
9.13988292e-01 3.34378444e-02 5.99424124e-01 -1.08567274e+00
-2.69829668e-02 -2.83840358e-01 -9.74474013e-01 2.17420131e-01
-9.41518247e-01 9.93480533e-02 3.29509646e-01 -7.20772862e-01
-1.25696599e+00 5.76843977e-01 2.20989674e-01 -5.71714401e-01
-2.95793921e-01 7.06376582e-02 -9.33090970e-02 -1.98872462e-01
7.26225376e-01 2.70579588e-02 -2.22983554e-01 -6.05211437e-01
3.98321539e-01 8.04310083e-01 5.63251793e-01 -5.77494979e-01
8.49303961e-01 5.13190866e-01 -1.61822885e-01 -4.86772329e-01
-1.51556742e+00 -8.16916943e-01 -6.33076549e-01 -3.48870128e-01
8.46548557e-01 -1.13473940e+00 -4.05115098e-01 3.46632868e-01
-8.03071976e-01 -1.36035979e-01 -5.88251114e-01 6.92862332e-01
-1.00331716e-01 1.52463332e-01 -3.58164221e-01 -1.03810394e+00
-6.69881552e-02 -1.37256384e+00 1.21977496e+00 7.74693310e-01
3.46353263e-01 -6.04120433e-01 -4.22224700e-01 5.37430823e-01
-9.85981449e-02 -3.16841066e-01 3.75753611e-01 -1.06096601e+00
-6.98281050e-01 -3.18551004e-01 -5.38199425e-01 6.69819593e-01
-1.72692239e-01 -1.28899604e-01 -1.37131822e+00 -4.23492879e-01
-1.42112792e-01 -8.24076355e-01 1.25602436e+00 2.81249821e-01
1.30353749e+00 1.39084414e-01 -4.24379110e-01 3.05682480e-01
1.00586498e+00 -8.97914171e-02 -6.07150346e-02 -6.53776824e-02
8.29452872e-01 8.04138005e-01 1.13477361e+00 2.99763203e-01
2.48806760e-01 4.82260227e-01 5.65894365e-01 -1.15483090e-01
-5.41565776e-01 -3.19671124e-01 4.11063433e-01 2.88089395e-01
5.14137208e-01 -5.35989888e-02 -5.77988029e-01 4.28401142e-01
-1.82634354e+00 -3.87507170e-01 -3.21644127e-01 2.13930440e+00
8.40231061e-01 3.24574202e-01 4.61444072e-02 1.87236294e-01
9.71827388e-01 -1.47248700e-01 -7.12373495e-01 2.19984636e-01
5.53080849e-02 1.48409363e-02 5.17600656e-01 4.24905032e-01
-1.32200623e+00 1.04083133e+00 5.12607622e+00 1.14274740e+00
-8.68573010e-01 2.14434043e-01 5.49671113e-01 1.71323657e-01
1.99403092e-01 -1.59795936e-02 -1.57401288e+00 2.76012182e-01
3.86587650e-01 5.19723892e-01 7.86415264e-02 1.23583984e+00
1.48095876e-01 -4.77739722e-01 -1.21329868e+00 5.70873559e-01
7.16163218e-03 -8.90632153e-01 -8.87680799e-02 -1.59366995e-01
1.03106475e+00 1.25678703e-02 1.85667858e-01 5.35275817e-01
4.42747444e-01 -4.58387494e-01 9.92407322e-01 7.89631754e-02
4.69800323e-01 -5.53658009e-01 5.82217395e-01 7.08195865e-01
-1.17586064e+00 -3.81024927e-01 -3.94141883e-01 3.56216490e-01
-2.14240715e-01 6.92023993e-01 -1.13028276e+00 2.26975873e-01
6.91320121e-01 4.65896487e-01 -7.95526922e-01 1.13790667e+00
-7.72159636e-01 8.77094865e-01 -3.50189239e-01 1.37974846e-03
4.25790042e-01 -1.00167207e-01 6.26138210e-01 1.14665186e+00
-5.89786246e-02 -2.57543642e-02 7.08089530e-01 1.00787199e+00
-1.90764055e-01 -4.72826958e-02 2.90145576e-02 5.04796393e-02
6.65184736e-01 1.53344548e+00 -9.11527693e-01 -5.08319795e-01
-2.34879300e-01 7.27315366e-01 4.20358717e-01 3.35926205e-01
-7.36732662e-01 -2.54980475e-01 1.53839113e-02 -1.54539213e-01
5.58556259e-01 1.82578430e-01 -2.94356406e-01 -7.46824205e-01
-2.00162396e-01 -4.59330052e-01 7.14809179e-01 -5.89086771e-01
-1.04310656e+00 2.03220844e-01 -2.91323848e-02 -1.09262228e+00
2.06885815e-01 -6.49479568e-01 -4.35640901e-01 6.98160827e-01
-1.76147246e+00 -1.29602039e+00 -3.73583972e-01 4.07360852e-01
7.92117894e-01 -2.71787439e-02 3.38672847e-01 3.23600590e-01
-8.18360448e-01 6.71707869e-01 -2.74619535e-02 1.27363622e-01
7.97060788e-01 -1.58203959e+00 -2.35686690e-01 1.00712025e+00
4.38005298e-01 3.32404971e-01 4.70862478e-01 -6.27410531e-01
-9.86794114e-01 -1.49896944e+00 4.77980345e-01 -2.91802287e-01
2.99315214e-01 -3.79902840e-01 -8.44362259e-01 4.28408295e-01
-4.05946791e-01 6.22916579e-01 3.96924317e-01 -9.34427083e-02
-4.05417413e-01 -1.17944121e-01 -1.25428629e+00 2.59994924e-01
6.13157332e-01 -2.95829713e-01 -5.85021198e-01 4.44259465e-01
7.76896238e-01 -4.16499138e-01 -3.78652722e-01 6.06082261e-01
2.70483434e-01 -6.33423388e-01 7.64090776e-01 -3.77509564e-01
-1.01358987e-01 -8.95631015e-01 -1.98118556e-02 -1.09248769e+00
-3.10398247e-02 -2.17687950e-01 -2.57342488e-01 1.43937242e+00
3.57084125e-01 -2.20822439e-01 1.08356082e+00 4.55664217e-01
-2.93885916e-01 -7.23528028e-01 -6.90304875e-01 -8.48938763e-01
-4.01177794e-01 -4.62897718e-01 2.57578731e-01 8.00430536e-01
-5.06593227e-01 4.09445912e-01 -9.37816426e-02 7.28080153e-01
9.37434435e-01 1.65482193e-01 7.52635777e-01 -1.34336340e+00
-3.60581398e-01 -1.50645643e-01 -1.46394894e-01 -1.48930430e+00
9.76588354e-02 -1.03656197e+00 5.50513387e-01 -1.02882969e+00
3.35551620e-01 -8.56167972e-01 -5.63583374e-01 6.19920492e-01
-6.46249533e-01 5.87123334e-01 1.03880666e-01 9.86037999e-02
-1.09020948e+00 4.83679116e-01 1.40126252e+00 -9.29704085e-02
-2.81833202e-01 3.58027518e-01 -4.35894430e-01 8.68404627e-01
2.34397411e-01 -9.00119722e-01 -3.53724539e-01 -1.00623354e-01
-3.33284199e-01 -2.87392616e-01 2.10572705e-01 -8.97915900e-01
3.93116653e-01 1.09411076e-01 4.89468038e-01 -1.04121876e+00
2.49143213e-01 -9.56668437e-01 -6.43946707e-01 5.67667961e-01
-5.86189628e-01 -8.93068671e-01 -1.80358991e-01 9.25355077e-01
-7.85645992e-02 -6.19260252e-01 1.26643288e+00 2.22509429e-01
-7.70958662e-01 3.29245418e-01 -9.24748629e-02 8.14641491e-02
1.25010228e+00 -1.47849157e-01 -1.44997030e-01 6.02156892e-02
-9.38554466e-01 8.27119350e-01 -2.46750135e-02 3.65817875e-01
5.10347247e-01 -1.11549604e+00 -6.07135236e-01 4.04052079e-01
4.64540720e-01 5.63247561e-01 -7.20799044e-02 8.17753971e-01
9.52995941e-02 6.51290342e-02 2.47101888e-01 -1.05636835e+00
-1.42405236e+00 5.94918191e-01 2.53028750e-01 -6.08383268e-02
-2.58313239e-01 1.46321511e+00 4.50812757e-01 -6.57509804e-01
8.63917708e-01 -2.90534914e-01 -3.55882943e-01 2.66701788e-01
6.34528160e-01 4.70725477e-01 6.49770871e-02 -4.22239244e-01
-2.55733639e-01 2.51350701e-01 -5.35395443e-01 1.12012163e-01
1.06596303e+00 -2.04292178e-01 1.15076579e-01 3.29394370e-01
8.91283989e-01 -5.68643138e-02 -1.68976068e+00 -6.46751225e-01
3.39258343e-01 -3.58866066e-01 3.21918070e-01 -8.46457183e-01
-9.90444362e-01 8.75544131e-01 8.57288659e-01 -2.85379253e-02
9.14454043e-01 1.48178071e-01 3.22876245e-01 3.64280879e-01
1.92109376e-01 -1.27065170e+00 3.98773253e-01 4.00381863e-01
3.18927824e-01 -1.53845847e+00 -9.12520848e-03 -8.12981009e-01
-4.56379414e-01 9.65867579e-01 1.05428648e+00 -5.60873747e-03
5.95979393e-01 2.44656265e-01 2.10109875e-01 -5.36208972e-02
-4.89239097e-01 -5.01289010e-01 6.62472367e-01 3.73113960e-01
8.20421949e-02 9.13805049e-03 -1.37119770e-01 5.21193027e-01
5.57199955e-01 -2.50974864e-01 2.42999807e-01 1.03485835e+00
-1.08449399e+00 -1.06906438e+00 -6.99735761e-01 4.65052933e-01
-2.43760705e-01 7.98496529e-02 -3.47449780e-01 3.08642387e-01
4.89149004e-01 1.10015607e+00 -8.74743909e-02 -1.20963454e-01
1.24063522e-01 1.80327520e-01 2.92131960e-01 -1.18649697e+00
-5.25875866e-01 3.89628023e-01 -1.88157976e-01 -4.44837421e-01
-3.82194191e-01 -5.16522527e-01 -1.49500763e+00 6.42482340e-01
-1.13018584e+00 3.35613847e-01 7.65194535e-01 9.43223834e-01
5.57907484e-02 6.48171246e-01 8.26424181e-01 -7.96152711e-01
-9.55766916e-01 -1.00926399e+00 -7.18436122e-01 2.21278891e-01
3.56023729e-01 -9.62926686e-01 -4.67198551e-01 -4.27708812e-02]
|
[9.199033737182617, 1.3232842683792114]
|
8ae572ae-80e3-4ae9-aeaa-2cda858d3671
|
biomedical-language-models-are-robust-to-sub
|
2306.17649
| null |
https://arxiv.org/abs/2306.17649v3
|
https://arxiv.org/pdf/2306.17649v3.pdf
|
Biomedical Language Models are Robust to Sub-optimal Tokenization
|
As opposed to general English, many concepts in biomedical terminology have been designed in recent history by biomedical professionals with the goal of being precise and concise. This is often achieved by concatenating meaningful biomedical morphemes to create new semantic units. Nevertheless, most modern biomedical language models (LMs) are pre-trained using standard domain-specific tokenizers derived from large scale biomedical corpus statistics without explicitly leveraging the agglutinating nature of biomedical language. In this work, we first find that standard open-domain and biomedical tokenizers are largely unable to segment biomedical terms into meaningful components. Therefore, we hypothesize that using a tokenizer which segments biomedical terminology more accurately would enable biomedical LMs to improve their performance on downstream biomedical NLP tasks, especially ones which involve biomedical terms directly such as named entity recognition (NER) and entity linking. Surprisingly, we find that pre-training a biomedical LM using a more accurate biomedical tokenizer does not improve the entity representation quality of a language model as measured by several intrinsic and extrinsic measures such as masked language modeling prediction (MLM) accuracy as well as NER and entity linking performance. These quantitative findings, along with a case study which explores entity representation quality more directly, suggest that the biomedical pre-training process is quite robust to instances of sub-optimal tokenization.
|
['Yu Su', 'Huan Sun', 'Bernal Jiménez Gutiérrez']
|
2023-06-30
| null | null | null | null |
['entity-linking', 'named-entity-recognition-ner', 'cg']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 2.94317812e-01 5.56139231e-01 -3.90649766e-01 -2.95396686e-01
-1.17268002e+00 -5.37057817e-01 4.31503594e-01 8.45336974e-01
-9.53407228e-01 1.04484630e+00 3.91122162e-01 -5.34114778e-01
1.12845801e-01 -5.15987992e-01 -5.27282596e-01 -3.41222882e-01
2.73444038e-02 5.75956464e-01 -3.25594872e-01 7.13357106e-02
5.61344400e-02 2.93350160e-01 -6.36030674e-01 1.96157739e-01
1.28160655e+00 2.99325883e-01 3.87104116e-02 3.99183840e-01
-3.61112624e-01 4.54903692e-01 -8.13901067e-01 -6.07439518e-01
-2.05926016e-01 -4.10651028e-01 -1.11690760e+00 -3.46634150e-01
2.41973639e-01 2.82099158e-01 4.45583872e-02 9.84765351e-01
7.16134667e-01 6.99881418e-03 8.09708595e-01 -3.33748579e-01
-8.00330997e-01 9.74252701e-01 -1.12467788e-01 1.27747640e-01
3.30743402e-01 3.44353542e-02 1.10486281e+00 -7.34026730e-01
9.30086315e-01 1.04809546e+00 1.03037119e+00 8.18878651e-01
-1.46441734e+00 -7.01593697e-01 -1.59283459e-01 -3.95004004e-01
-1.48116052e+00 -6.33252203e-01 2.07884416e-01 -4.57640171e-01
1.23846173e+00 1.55525446e-01 8.59253481e-02 1.12578166e+00
4.67459679e-01 2.62135148e-01 8.05317938e-01 -5.14807284e-01
6.68452978e-02 4.05397147e-01 1.21117845e-01 6.37829661e-01
8.39409232e-01 -3.06350112e-01 -2.77872354e-01 -3.52732509e-01
4.31691915e-01 -5.79719424e-01 -1.56656891e-01 3.31686676e-01
-1.58501887e+00 6.83004081e-01 1.56047493e-01 8.62871289e-01
-5.27249277e-01 5.77266552e-02 6.64496779e-01 1.50766566e-01
5.15995324e-01 1.26099706e+00 -6.88091576e-01 4.38145176e-02
-1.30011415e+00 -3.73272926e-01 8.19080114e-01 6.50058687e-01
5.14283061e-01 2.61533912e-02 -2.05011353e-01 9.96411204e-01
2.30910167e-01 2.48268187e-01 1.08013964e+00 -6.36192799e-01
3.45367551e-01 4.96574759e-01 -2.27769509e-01 -6.35781825e-01
-6.54111028e-01 -5.97103775e-01 -5.88057458e-01 -4.97461587e-01
4.76672649e-01 -4.66433644e-01 -1.02302134e+00 2.11978340e+00
1.73854619e-01 2.67112907e-02 4.35265779e-01 2.66200691e-01
1.01277697e+00 3.67910385e-01 1.07649481e+00 -5.44649102e-02
1.80124867e+00 -2.42530614e-01 -7.32997000e-01 -3.96710247e-01
1.38189709e+00 -8.99836481e-01 7.88500905e-01 -4.66153608e-04
-1.09708071e+00 -3.61884981e-01 -1.03799963e+00 -3.78591418e-01
-6.88888609e-01 -5.40151894e-02 5.08978784e-01 7.91546285e-01
-9.07958627e-01 5.28636515e-01 -9.02651906e-01 -6.82328939e-01
3.42085540e-01 4.69850689e-01 -7.19860077e-01 -2.16507399e-03
-1.52779877e+00 1.48378408e+00 7.27570772e-01 -9.21324342e-02
-3.50157082e-01 -9.44262624e-01 -1.02757835e+00 -1.38527444e-02
-1.39046028e-01 -1.12591624e+00 8.74150097e-01 -7.69175172e-01
-1.01167083e+00 1.31501710e+00 -1.78153455e-01 -5.23659170e-01
2.32282076e-02 3.94229665e-02 -4.90701884e-01 1.16961531e-01
4.51828986e-01 9.72931385e-01 1.34772763e-01 -1.04308915e+00
-4.14084971e-01 -1.89120457e-01 -2.80751258e-01 1.79476425e-01
-5.86580992e-01 9.63572115e-02 -6.46402463e-02 -7.95594335e-01
-1.36786461e-01 -6.45042598e-01 -4.04372483e-01 -4.37062114e-01
-6.22760177e-01 -3.95379215e-01 -8.51967186e-02 -9.24256861e-01
1.22727334e+00 -2.02603602e+00 -2.70343751e-01 1.68269858e-01
1.83215812e-01 3.46067905e-01 -1.53211251e-01 3.23494941e-01
-4.62408543e-01 6.93018973e-01 -2.26884499e-01 -4.30804908e-01
-7.21682087e-02 3.02372158e-01 5.97777814e-02 4.39509273e-01
4.76132482e-01 1.03936851e+00 -1.07560337e+00 -9.94609058e-01
-1.35205790e-01 4.41354811e-01 -3.84384811e-01 -1.39978737e-01
-7.36449435e-02 3.67218375e-01 -3.93170387e-01 6.91164613e-01
2.62989640e-01 -3.92313063e-01 3.39314669e-01 -1.66275308e-01
9.24358815e-02 7.38044322e-01 -5.72095931e-01 1.81262910e+00
-5.30791759e-01 3.95435125e-01 -2.48409569e-01 -1.02115989e+00
6.46191001e-01 8.39123726e-01 7.05277979e-01 -3.67848605e-01
1.54926879e-02 5.44902742e-01 1.43872470e-01 -4.58873868e-01
4.81108338e-01 -7.60531664e-01 -1.42156124e-01 1.09312892e-01
2.54266918e-01 1.68210179e-01 1.15872957e-01 1.23521298e-01
1.22933817e+00 -1.31110519e-01 5.93939841e-01 -4.51079011e-01
5.67763925e-01 3.87459159e-01 6.79552197e-01 5.68211019e-01
-1.80396214e-02 4.35118377e-01 2.13388398e-01 2.82951236e-01
-8.83007288e-01 -1.06833625e+00 -7.40058243e-01 8.00126612e-01
-4.31506902e-01 -4.78476524e-01 -8.33062232e-01 -6.06638730e-01
-5.40763140e-02 9.77957249e-01 -4.73840594e-01 -3.59807342e-01
-6.46532774e-01 -1.05472088e+00 1.31883204e+00 4.38508987e-01
1.37146255e-02 -1.03505242e+00 -3.13368320e-01 4.86064047e-01
-2.87398815e-01 -1.32782853e+00 -5.15402317e-01 4.81883466e-01
-1.00413358e+00 -8.20024014e-01 -9.14644837e-01 -1.05305588e+00
8.45821619e-01 -5.17015457e-01 1.10945773e+00 4.26361896e-02
-1.86151817e-01 1.78754076e-01 -1.61329672e-01 -4.45477992e-01
-8.15258086e-01 5.01414359e-01 -2.57938746e-02 -5.35841167e-01
7.18482435e-01 -2.17064261e-01 -4.76605684e-01 -2.01928809e-01
-9.45646226e-01 -1.74623117e-01 1.03596747e+00 8.99929523e-01
4.57222074e-01 -4.36074287e-01 9.85613942e-01 -1.13918197e+00
7.53319502e-01 -6.43589079e-01 2.99669683e-01 3.56666565e-01
-7.99923062e-01 2.72700965e-01 5.30884624e-01 -5.52877963e-01
-7.92621493e-01 -2.60081321e-01 -5.15586793e-01 2.36068204e-01
-4.25610811e-01 9.43500042e-01 4.50224541e-02 1.53586179e-01
8.53956997e-01 6.56410158e-02 -1.90602675e-01 -3.95963520e-01
2.01706350e-01 6.96125627e-01 4.76272464e-01 -7.13095903e-01
5.42719543e-01 1.85719788e-01 -8.76004063e-03 -9.56938505e-01
-8.59875739e-01 -5.00790119e-01 -4.71736163e-01 3.95536840e-01
1.38562799e+00 -9.75944102e-01 -4.17325586e-01 -1.30058452e-01
-1.09676015e+00 -3.09254766e-01 -2.86018103e-01 8.43381286e-01
-1.35983691e-01 4.25290614e-01 -9.29485321e-01 -2.63852865e-01
-4.61598366e-01 -9.19836998e-01 1.03400528e+00 -1.43689697e-03
-8.57764423e-01 -1.42004538e+00 1.83256775e-01 2.84582406e-01
2.47658193e-01 3.76017720e-01 1.39812052e+00 -1.25163198e+00
2.57616848e-01 -2.11919636e-01 -3.99896130e-02 2.87971139e-01
4.35176820e-01 -3.63869220e-01 -8.46471369e-01 -7.67036527e-02
-1.42146438e-01 -6.66711032e-02 7.29449034e-01 2.88286150e-01
5.98934412e-01 -2.88407177e-01 -6.20087266e-01 3.88513982e-01
1.32351053e+00 1.03944570e-01 5.16836226e-01 3.54198009e-01
6.46334648e-01 7.44648874e-01 2.04845220e-01 -1.63756847e-01
3.62901390e-01 2.16394708e-01 -3.09770584e-01 -3.75792325e-01
-3.41204852e-02 -1.59360126e-01 4.11386073e-01 9.23419952e-01
3.20194244e-01 -1.61224946e-01 -1.23782063e+00 8.38413775e-01
-1.27054799e+00 -7.09185123e-01 1.77322865e-01 2.00563049e+00
1.63327968e+00 2.10487142e-01 -2.00584069e-01 -2.82221109e-01
6.32040560e-01 -3.78715754e-01 -3.21331531e-01 -4.89817291e-01
-1.61578655e-01 7.00727165e-01 7.15872705e-01 4.01090920e-01
-8.09633374e-01 1.07471716e+00 6.52315474e+00 7.83217072e-01
-1.07922220e+00 1.90169707e-01 7.91063666e-01 2.07561940e-01
-2.69522667e-01 -1.29788309e-01 -8.23210657e-01 3.85038704e-01
1.43677497e+00 -3.14207494e-01 -1.70337722e-01 4.42781568e-01
1.75660253e-01 1.95113778e-01 -1.42134368e+00 6.39110923e-01
-1.48417875e-01 -1.27364516e+00 2.53630698e-01 1.33476466e-01
4.90972877e-01 3.67369317e-02 -3.76209728e-02 3.40009421e-01
2.99286604e-01 -1.47938812e+00 4.13672268e-01 4.74132448e-01
1.00165677e+00 -4.41191047e-01 9.29342628e-01 1.13219008e-01
-9.03768778e-01 4.81604576e-01 -3.02884489e-01 5.71686566e-01
2.05192357e-01 7.60895133e-01 -1.17778993e+00 5.87875366e-01
2.21481949e-01 4.30430830e-01 -6.13307238e-01 8.96611869e-01
6.00337703e-03 8.29308927e-01 -2.06024289e-01 1.34215847e-01
3.08855206e-01 2.11860508e-01 4.68457460e-01 1.88104129e+00
1.79241866e-01 4.86340411e-02 -1.11059815e-01 9.27780271e-01
-3.79771262e-01 4.44554776e-01 -4.15241927e-01 -6.56840801e-01
5.62435091e-01 1.13957870e+00 -8.17735553e-01 -5.82241476e-01
-4.57497954e-01 7.07692146e-01 1.71571612e-01 1.84128135e-01
-6.50517642e-01 -6.25343621e-01 7.46662259e-01 -8.41612145e-02
-9.28481519e-02 -8.36188197e-02 -5.33766031e-01 -1.03707314e+00
-2.20156878e-01 -9.43522513e-01 4.66600001e-01 -3.23820800e-01
-1.53059733e+00 5.50808072e-01 -2.85664290e-01 -7.46828794e-01
-1.51916206e-01 -6.91282034e-01 -3.76614153e-01 1.04656219e+00
-1.47521472e+00 -1.06861007e+00 3.80863786e-01 2.76455104e-01
1.72390938e-01 2.15633243e-01 1.11546242e+00 5.88424146e-01
-5.61794400e-01 1.02463567e+00 1.16711268e-02 4.92005110e-01
1.27292740e+00 -1.43364108e+00 3.40375572e-01 5.62332571e-01
5.86498231e-02 1.44291115e+00 7.71472871e-01 -9.16120172e-01
-8.03313792e-01 -1.17708528e+00 1.60697448e+00 -5.45131624e-01
5.96099317e-01 -1.65241078e-01 -9.97232378e-01 7.90966153e-01
1.26956552e-01 -3.87282103e-01 1.23611319e+00 2.16236293e-01
-1.19104929e-01 3.16734076e-01 -1.19626820e+00 6.77793086e-01
8.24700296e-01 -7.48036325e-01 -1.02114820e+00 5.02856135e-01
7.62636244e-01 -9.76678208e-02 -1.52697170e+00 4.15393084e-01
2.31600747e-01 -4.66383584e-02 1.12313402e+00 -1.06508625e+00
2.75547892e-01 3.18412744e-02 9.60123911e-03 -1.14017904e+00
-3.09236556e-01 -4.21963423e-01 3.90381813e-01 1.34782982e+00
1.00761616e+00 -8.30484509e-01 6.02053642e-01 7.27522612e-01
-3.21167350e-01 -6.95174932e-01 -9.12890673e-01 -7.64612317e-01
6.65839970e-01 -2.12549523e-01 2.17209697e-01 1.40592873e+00
4.40345526e-01 5.42944968e-01 2.66543359e-01 1.23181298e-01
3.11330795e-01 -5.31622827e-01 1.28227383e-01 -1.27386665e+00
7.06416136e-03 -5.53129196e-01 -3.73365909e-01 -6.56238437e-01
4.24215049e-01 -1.36314261e+00 7.54910186e-02 -1.78599489e+00
1.24016590e-01 -6.55002892e-01 -6.08160675e-01 6.20132327e-01
-4.70751911e-01 1.78631455e-01 -3.05002034e-01 6.08817264e-02
-3.40419561e-01 -3.82088572e-02 8.52199554e-01 -1.34517904e-02
-1.42806396e-01 -4.83054519e-01 -1.17470872e+00 7.04716086e-01
7.40472376e-01 -8.82991135e-01 5.24167418e-02 -3.51398081e-01
2.12374747e-01 -2.05245912e-02 7.72796497e-02 -8.76378834e-01
1.92569971e-01 2.45889053e-02 5.77195525e-01 1.56935174e-02
8.44590962e-02 -3.75989467e-01 8.13990310e-02 6.61850572e-01
-6.06576145e-01 1.66959420e-01 4.37622309e-01 2.37761065e-01
-1.70267969e-02 -4.31126028e-01 7.32268333e-01 -2.69951105e-01
-3.92628998e-01 -1.95961520e-02 -6.44624233e-01 4.07338321e-01
7.53408730e-01 -3.50452781e-01 -2.09947675e-01 1.47665992e-01
-8.69777322e-01 9.42099094e-02 4.16383952e-01 6.54442608e-02
1.58732846e-01 -9.62677836e-01 -8.59716594e-01 -3.12365770e-01
7.29137287e-02 -2.88737237e-01 -8.95755962e-02 1.01067030e+00
-4.41220909e-01 9.16267931e-01 6.92445338e-02 -2.74176657e-01
-1.00768566e+00 4.14736837e-01 4.03457582e-01 -5.98666012e-01
-4.67240095e-01 8.55004728e-01 3.09974134e-01 -4.70511854e-01
5.00596203e-02 -6.68723047e-01 -2.13068798e-01 1.47965163e-01
1.91783756e-01 4.72653005e-03 1.92748606e-01 -6.99860454e-01
-6.57599032e-01 3.01943421e-01 -5.40389195e-02 -9.00353491e-02
1.16292989e+00 -5.32963611e-02 -1.31034106e-01 4.20450538e-01
1.28766978e+00 2.73827970e-01 -1.63344130e-01 -1.18373960e-01
6.79312050e-01 3.76202047e-01 -1.35164574e-01 -8.86756122e-01
-6.20259881e-01 6.01296008e-01 1.98956788e-01 -1.65642142e-01
6.91040635e-01 1.58923149e-01 1.05344427e+00 4.16864604e-01
2.38274392e-02 -1.12715900e+00 -2.82053143e-01 3.97477329e-01
2.42956713e-01 -1.00310266e+00 -1.22026116e-01 -1.41248420e-01
-4.66975480e-01 9.14823472e-01 4.26926494e-01 2.21656129e-01
5.73821425e-01 2.59870887e-01 1.13099121e-01 -3.84426147e-01
-4.84244674e-01 -2.44486272e-01 4.50047225e-01 3.34974378e-01
1.07237458e+00 2.65057068e-02 -6.20852888e-01 6.76384032e-01
-2.88820386e-01 -1.28825292e-01 2.17277110e-01 7.88259327e-01
-2.72365361e-01 -1.38972080e+00 -2.66646981e-01 6.41910195e-01
-1.18482757e+00 -6.09015644e-01 -3.44620883e-01 7.49430239e-01
3.50828826e-01 9.12472606e-01 -5.41000888e-02 -1.28336087e-01
1.19487807e-01 6.12876832e-01 2.82739341e-01 -1.18084502e+00
-9.24744248e-01 -1.06745377e-01 3.89642507e-01 -1.70172438e-01
-3.96157712e-01 -5.70159554e-01 -1.65987945e+00 -1.02790065e-01
-5.80041945e-01 5.17129838e-01 4.88401949e-01 1.16552746e+00
3.36592495e-01 7.31519222e-01 -2.88145065e-01 -1.73505768e-01
-4.65409487e-01 -1.07805371e+00 -3.16248208e-01 3.86078715e-01
1.28047943e-01 -2.90128708e-01 -1.64462015e-01 2.39104211e-01]
|
[8.537652015686035, 8.692413330078125]
|
cc987243-5ed8-480f-b586-ea182c8491d4
|
group-equivariant-convolutional-networks
|
1602.07576
| null |
http://arxiv.org/abs/1602.07576v3
|
http://arxiv.org/pdf/1602.07576v3.pdf
|
Group Equivariant Convolutional Networks
|
We introduce Group equivariant Convolutional Neural Networks (G-CNNs), a
natural generalization of convolutional neural networks that reduces sample
complexity by exploiting symmetries. G-CNNs use G-convolutions, a new type of
layer that enjoys a substantially higher degree of weight sharing than regular
convolution layers. G-convolutions increase the expressive capacity of the
network without increasing the number of parameters. Group convolution layers
are easy to use and can be implemented with negligible computational overhead
for discrete groups generated by translations, reflections and rotations.
G-CNNs achieve state of the art results on CIFAR10 and rotated MNIST.
|
['Taco S. Cohen', 'Max Welling']
|
2016-02-24
| null | null | null | null |
['rotated-mnist', 'colorectal-gland-segmentation', 'breast-tumour-classification', 'multi-tissue-nucleus-segmentation']
|
['computer-vision', 'medical', 'medical', 'medical']
|
[ 1.22952297e-01 4.68937427e-01 -1.25833958e-01 -5.10736227e-01
1.61905766e-01 -4.83647734e-01 8.05882752e-01 -5.09867668e-01
-7.77097940e-01 3.49759668e-01 4.59369481e-01 -5.25107682e-01
1.28988950e-02 -9.74335015e-01 -1.02861953e+00 -6.20837033e-01
-4.17037994e-01 -3.14242601e-01 -1.51918521e-02 -3.38851064e-01
-1.14350595e-01 8.03995609e-01 -1.19521904e+00 6.71122015e-01
-5.17607406e-02 9.32262480e-01 -2.48199999e-01 7.56748021e-01
2.78544962e-01 8.64126623e-01 -2.50186205e-01 -3.71544808e-01
6.45474613e-01 6.13994850e-03 -9.41594124e-01 -9.96589512e-02
8.07190537e-01 -4.52023625e-01 -9.06244278e-01 9.58807051e-01
4.54317838e-01 4.42069054e-01 3.64877909e-01 -1.17915273e+00
-9.94854271e-01 1.10264695e+00 4.63645235e-02 2.49288723e-01
-4.36263561e-01 -2.18800053e-01 1.31259763e+00 -1.11755323e+00
5.10122418e-01 1.61471486e+00 1.06427932e+00 6.85353816e-01
-1.25053489e+00 -8.87793005e-01 3.34355891e-01 9.43400338e-02
-1.32763898e+00 -4.92366821e-01 4.50173229e-01 -7.18593150e-02
1.50679207e+00 2.51836479e-01 6.21837199e-01 1.08146381e+00
6.55443221e-02 4.68782783e-01 3.63285244e-01 -2.38645971e-01
1.02153882e-01 -8.25551987e-01 3.56025584e-02 9.01453733e-01
4.22629625e-01 -2.42359042e-01 -5.63920140e-01 -1.58581182e-01
1.27288616e+00 4.03400064e-01 -1.99181288e-01 -1.89770564e-01
-1.64276755e+00 1.01174629e+00 1.01622319e+00 1.03337802e-01
1.13873720e-01 1.11098146e+00 4.52423096e-01 7.68805742e-01
3.92050356e-01 6.47428811e-01 -5.49192607e-01 4.39197153e-01
-2.38879308e-01 3.41804922e-01 5.12987196e-01 1.26646042e+00
7.69681871e-01 4.05952573e-01 -6.69996440e-03 4.96095777e-01
-5.87257417e-03 4.55662748e-03 6.13765895e-01 -9.29137886e-01
4.28334892e-01 4.88950819e-01 -5.87555528e-01 -7.00323999e-01
-7.61295140e-01 -6.23232305e-01 -1.21803093e+00 8.37949738e-02
2.19679132e-01 -2.43651137e-01 -9.15705800e-01 1.87888730e+00
-3.03546757e-01 1.07935324e-01 2.34316930e-01 3.85349512e-01
8.71125102e-01 2.84048587e-01 2.31473930e-02 7.04522133e-01
1.29730427e+00 -1.22436845e+00 -3.28184098e-01 -1.15997650e-01
9.69325185e-01 -5.42257547e-01 7.69297957e-01 -4.43016738e-02
-9.73970354e-01 -6.17673755e-01 -1.29753959e+00 -5.81150115e-01
-6.04847670e-01 2.78394222e-01 1.31239915e+00 7.91983008e-01
-1.38802803e+00 1.12274826e+00 -1.06940699e+00 6.93249330e-02
8.23716760e-01 9.61927950e-01 -7.75603890e-01 1.20781578e-01
-9.91781354e-01 2.77326852e-01 5.75378180e-01 -5.14045022e-02
-5.62040210e-01 -1.21464801e+00 -8.85213614e-01 2.22641513e-01
-2.67658263e-01 -7.29693532e-01 1.38554072e+00 -9.80263412e-01
-1.52786112e+00 6.44963980e-01 2.04603389e-01 -8.02127004e-01
4.37954128e-01 -3.19674984e-02 -2.84358621e-01 -4.15776707e-02
-3.33340704e-01 1.10954392e+00 8.04109633e-01 -2.50617146e-01
-2.89215833e-01 -1.05998501e-01 2.88612247e-01 3.26871015e-02
-4.39725757e-01 1.17227286e-02 -7.76904151e-02 -9.14452136e-01
4.93013859e-01 -1.34082234e+00 -4.49385673e-01 3.60765517e-01
-4.34160650e-01 -3.42562258e-01 1.08721817e+00 -2.05241628e-02
3.77834558e-01 -2.31395721e+00 -1.10895187e-01 2.36756772e-01
6.05778515e-01 3.52640599e-01 -3.34843010e-01 2.51200706e-01
-7.27615118e-01 5.31209946e-01 2.67834216e-01 -4.77453262e-01
1.41900718e-01 3.89234126e-01 -4.80057806e-01 5.70013642e-01
3.95085871e-01 1.17083812e+00 -6.78029537e-01 3.42391372e-01
2.79138703e-02 6.37610614e-01 -1.23258138e+00 -3.07334483e-01
-3.59624177e-02 8.36644322e-02 -1.21648714e-01 1.09103136e-01
6.85790241e-01 -3.11228842e-01 3.63316417e-01 -4.43869323e-01
-1.98915049e-01 5.76787472e-01 -1.09572434e+00 1.63532019e+00
-2.53680438e-01 8.84010255e-01 -5.49947917e-01 -1.07515383e+00
8.04238498e-01 2.81325012e-01 6.65425435e-02 -2.61171877e-01
2.38204524e-01 7.01511651e-02 2.94333726e-01 1.71450183e-01
5.20320535e-01 3.89073610e-01 -6.92310184e-02 5.83429098e-01
6.29146218e-01 2.46999487e-01 -5.20984177e-03 4.30832468e-02
9.44208503e-01 -2.50644237e-01 2.02043891e-01 -6.75566018e-01
1.06428107e-02 -7.79260874e-01 4.90457565e-01 7.81074643e-01
5.97137749e-01 5.52599847e-01 7.49208093e-01 -1.16821635e+00
-1.18868852e+00 -9.77231145e-01 1.37944415e-01 1.39662647e+00
-3.89434904e-01 -5.72493494e-01 -7.53472924e-01 -1.83582500e-01
-1.18673444e-01 -1.63095910e-02 -9.76641953e-01 -4.21468705e-01
-8.41936648e-01 -8.20771098e-01 1.15701318e+00 9.51982856e-01
1.14469552e+00 -8.73312831e-01 -4.34070915e-01 9.84542966e-02
1.98212102e-01 -1.43197417e+00 -6.91604376e-01 5.38206339e-01
-1.19944608e+00 -1.10860193e+00 -4.77327526e-01 -1.13347578e+00
1.10718811e+00 4.42028642e-01 9.10123527e-01 2.07120962e-02
-2.09212139e-01 -2.23288611e-01 -1.53235808e-01 -6.51141584e-01
2.88193896e-02 5.86194992e-01 3.51331532e-01 2.22031586e-02
-1.93694886e-02 -8.64562094e-01 -5.45866430e-01 5.07512093e-01
-9.08615470e-01 2.58376986e-01 4.56903577e-01 7.69190729e-01
5.62035501e-01 -2.88115591e-01 1.89427063e-01 -1.01547527e+00
3.10170293e-01 3.88593897e-02 -5.70182443e-01 -2.47102529e-01
1.33030862e-01 3.56131911e-01 8.51404369e-01 -6.77311659e-01
-5.88227689e-01 -2.68874709e-02 1.06280103e-01 -3.95913273e-01
2.95524180e-01 3.29403609e-01 -5.27032092e-02 -6.65991008e-01
9.02887464e-01 -3.24517459e-01 -1.32384464e-01 -5.21213770e-01
5.87105572e-01 1.37100309e-01 8.42486143e-01 -4.85046983e-01
8.54863882e-01 7.91611195e-01 4.80310351e-01 -8.88340712e-01
-7.06906974e-01 -4.45379264e-04 -7.91562021e-01 3.84322554e-01
8.49201083e-01 -1.25458133e+00 -9.35878932e-01 7.23007917e-01
-1.14640522e+00 -7.20378041e-01 -3.51638973e-01 6.90315783e-01
-3.58008295e-01 -1.16248257e-01 -6.67244911e-01 1.93931028e-01
-3.95802319e-01 -9.44985449e-01 5.71987927e-01 -2.73603909e-02
-1.36376187e-01 -9.56585944e-01 -6.96272731e-01 -3.79694521e-01
7.00823128e-01 4.97503042e-01 9.62457895e-01 -6.72095835e-01
-8.80063295e-01 -7.37662464e-02 -4.05755967e-01 6.20863318e-01
6.08622506e-02 -2.20417500e-01 -1.14574015e+00 -4.64688301e-01
-4.83929694e-01 -1.40444368e-01 1.16157424e+00 2.82941699e-01
1.93116891e+00 -7.11927891e-01 6.28833752e-03 1.36488175e+00
1.03636158e+00 -1.33656368e-01 7.71794736e-01 1.90812409e-01
9.30391490e-01 1.40681518e-02 -5.11042953e-01 2.92484134e-01
-1.19900219e-02 3.88247311e-01 6.77384615e-01 -3.78623486e-01
-3.76708388e-01 -2.34525025e-01 1.70012116e-01 9.47802365e-01
-7.68346071e-01 5.28473929e-02 -5.45868754e-01 3.64533395e-01
-1.82631660e+00 -1.01028359e+00 -6.25261515e-02 1.64935625e+00
5.16373575e-01 5.64138256e-02 -2.25486919e-01 3.03199679e-01
4.68181580e-01 4.59066659e-01 -2.49706849e-01 -4.63000417e-01
-3.04889947e-01 8.37916195e-01 1.36813867e+00 2.67277360e-01
-1.36031294e+00 1.00798631e+00 7.54414368e+00 4.81192380e-01
-1.12464011e+00 -3.86988223e-02 4.44863021e-01 -1.67126715e-01
-1.86452299e-01 -2.77008474e-01 -7.33305335e-01 -2.62414306e-01
7.43652582e-01 -1.12356715e-01 6.10587776e-01 1.02209604e+00
-2.76555479e-01 9.33987439e-01 -1.56723845e+00 9.05026257e-01
-1.57009065e-01 -2.03459072e+00 3.49097610e-01 1.34292200e-01
1.06195533e+00 7.74444997e-01 4.14952427e-01 1.65018857e-01
6.87405825e-01 -1.37106180e+00 8.30798149e-01 2.45950401e-01
1.03139138e+00 -9.40244794e-01 4.71558839e-01 -3.16087514e-01
-1.23360455e+00 -8.76394287e-03 -1.09316444e+00 -4.59953129e-01
-4.01825875e-01 7.32028484e-02 -6.72457278e-01 1.59699216e-01
9.80982780e-01 7.74373651e-01 -5.26674509e-01 6.69510841e-01
-2.42273167e-01 3.50822926e-01 -3.06586415e-01 -1.37954426e-03
7.15014994e-01 1.37777835e-01 1.45233735e-01 1.29135847e+00
2.90469140e-01 5.65744080e-02 1.70387775e-02 6.98187768e-01
-9.63570058e-01 -3.90432715e-01 -6.79944456e-01 -3.72355394e-02
3.45624775e-01 1.05872643e+00 -6.97730124e-01 -8.30431655e-02
-3.79157603e-01 6.67146742e-01 4.95132685e-01 3.93060356e-01
-5.32104254e-01 -8.23991358e-01 1.32901752e+00 -1.66207850e-01
7.47474074e-01 -6.30696416e-01 3.61256301e-02 -1.17965579e+00
-3.55260879e-01 -7.36847639e-01 1.17137916e-01 -4.59471792e-01
-7.18569875e-01 5.16845584e-01 -8.06284696e-02 -1.16375494e+00
-1.07776992e-01 -1.37311602e+00 -5.27154803e-01 6.61876738e-01
-1.42009532e+00 -1.20670247e+00 -5.07065415e-01 1.00499094e+00
4.02724408e-02 -4.66674119e-01 1.29125929e+00 2.37901598e-01
-1.82973638e-01 1.06312680e+00 -1.21295109e-01 7.80310750e-01
4.05607611e-01 -1.14418483e+00 1.44512796e+00 7.96264052e-01
4.10108641e-02 1.00412524e+00 1.66887507e-01 9.35186669e-02
-1.30749691e+00 -1.79827452e+00 8.66086900e-01 1.36113346e-01
6.14177048e-01 -7.82347918e-01 -4.77912843e-01 1.46333444e+00
6.54809177e-03 8.47778201e-01 8.03432882e-01 1.56932190e-01
-1.34277487e+00 -3.33470181e-02 -7.65847564e-01 1.14409876e+00
1.73753405e+00 -9.58591700e-01 -1.50276601e-01 4.57231581e-01
1.21537316e+00 -6.91886485e-01 -9.57060039e-01 1.00310974e-01
5.36359847e-01 -6.57549798e-01 1.20404804e+00 -9.37613368e-01
1.62433773e-01 -1.82058617e-01 -3.23178649e-01 -1.36699224e+00
-7.60960042e-01 -1.16304827e+00 2.34587371e-01 2.72574127e-01
3.97592962e-01 -1.17479467e+00 8.13525021e-01 1.25928715e-01
-5.52989364e-01 -3.47264647e-01 -7.56900311e-01 -1.06252432e+00
1.62309021e-01 -4.74191964e-01 1.14625490e+00 1.17329693e+00
-3.24739873e-01 -1.40229285e-01 -2.35398918e-01 1.39288411e-01
4.23826039e-01 -4.93202895e-01 7.11576343e-01 -1.32521534e+00
-2.62057006e-01 -5.47808707e-01 -7.87765086e-01 -9.35492694e-01
1.63935497e-01 -1.58045769e+00 -6.67164326e-01 -9.41105843e-01
-4.31871295e-01 -2.45235085e-01 -3.57872933e-01 1.19976199e+00
5.91663837e-01 1.00619161e+00 2.06517845e-01 -2.88361516e-02
-3.88763189e-01 2.40335688e-01 1.22080660e+00 -3.27572972e-02
-5.20841666e-02 -2.28738338e-01 -8.59859109e-01 1.00534034e+00
1.01549768e+00 -3.65073234e-01 -4.44417596e-01 -8.39207649e-01
5.71848869e-01 -7.97505140e-01 4.17062461e-01 -1.12670314e+00
1.58530727e-01 1.75049230e-01 5.00415981e-01 -1.49600387e-01
9.37778801e-02 -3.21212232e-01 3.06877673e-01 6.15235269e-01
-5.78433931e-01 3.78988296e-01 2.99733579e-01 2.99024403e-01
-2.30376702e-02 1.53618544e-01 8.06529641e-01 -2.03252599e-01
-4.34797317e-01 9.37597871e-01 -4.92311954e-01 -9.77653116e-02
4.11666721e-01 -1.39299378e-01 -7.16368437e-01 -2.19601959e-01
-7.18569577e-01 -4.03222531e-01 1.46305086e-02 5.05605876e-01
2.98689097e-01 -1.78005552e+00 -4.02070373e-01 4.94122297e-01
2.57453006e-02 4.10286814e-01 2.97434051e-02 5.04335880e-01
-8.76349270e-01 4.66823608e-01 -6.48589849e-01 -2.10995406e-01
-1.24470675e+00 3.28078978e-02 6.89947367e-01 -6.69961572e-02
-1.01914048e+00 1.06203473e+00 3.95544022e-01 -6.13978386e-01
2.40671381e-01 -9.98116076e-01 -1.65841747e-02 -9.96683016e-02
8.41722190e-01 3.57710868e-01 5.06674290e-01 -3.92170101e-01
-2.63821453e-01 2.90233046e-01 -1.87411383e-01 1.59055933e-01
1.68141258e+00 7.28303730e-01 -2.24067017e-01 -7.55533725e-02
1.64350474e+00 -4.64498550e-01 -1.34937644e+00 -5.44681847e-01
-2.44222105e-01 8.42044279e-02 2.03247711e-01 7.85200745e-02
-1.36071432e+00 7.65868008e-01 1.79665729e-01 -5.94616309e-02
6.54660642e-01 -3.23262036e-01 4.17094231e-01 1.38076735e+00
9.68726501e-02 -9.75895941e-01 3.24946404e-01 1.04970753e+00
1.00631440e+00 -5.25711238e-01 -5.69269573e-03 -4.92237777e-01
-1.51707847e-02 1.53766835e+00 1.94919065e-01 -7.61121094e-01
1.05660617e+00 3.53316367e-01 -7.92312995e-02 -3.47941443e-02
-7.23040521e-01 2.87534110e-02 4.14730310e-01 5.90584576e-01
3.86058748e-01 1.84775695e-01 7.28734136e-02 4.84549195e-01
-9.09241676e-01 -3.09921086e-01 5.93214750e-01 7.02438533e-01
-1.54980257e-01 -9.66876447e-01 3.83705907e-02 3.11480641e-01
-7.63854742e-01 -2.74432898e-01 -1.38085186e-01 1.01354778e+00
7.59754330e-02 4.26588506e-01 6.75275028e-01 -4.42962259e-01
2.70101905e-01 2.19896603e-02 6.07200563e-01 -5.78237295e-01
-6.13710523e-01 -3.80866319e-01 8.29858854e-02 -7.88498640e-01
-5.79103410e-01 -3.56988251e-01 -1.15102983e+00 -7.65300989e-01
8.70137066e-02 -4.78767991e-01 7.42124677e-01 7.31226146e-01
4.15683359e-01 9.19572413e-01 5.25491416e-01 -1.14025509e+00
-5.41015744e-01 -8.47631097e-01 -4.72933710e-01 1.98459789e-01
4.72983062e-01 -1.96095496e-01 -2.92999208e-01 1.05710737e-01]
|
[8.915361404418945, 2.3821351528167725]
|
65bc88d7-9045-4a1c-b9ed-d29a19e6999a
|
prod-progressive-distillation-for-dense
|
2209.13335
| null |
https://arxiv.org/abs/2209.13335v3
|
https://arxiv.org/pdf/2209.13335v3.pdf
|
PROD: Progressive Distillation for Dense Retrieval
|
Knowledge distillation is an effective way to transfer knowledge from a strong teacher to an efficient student model. Ideally, we expect the better the teacher is, the better the student. However, this expectation does not always come true. It is common that a better teacher model results in a bad student via distillation due to the nonnegligible gap between teacher and student. To bridge the gap, we propose PROD, a PROgressive Distillation method, for dense retrieval. PROD consists of a teacher progressive distillation and a data progressive distillation to gradually improve the student. We conduct extensive experiments on five widely-used benchmarks, MS MARCO Passage, TREC Passage 19, TREC Document 19, MS MARCO Document and Natural Questions, where PROD achieves the state-of-the-art within the distillation methods for dense retrieval. The code and models will be released.
|
['Nan Duan', 'Rangan Majumder', 'Daxin Jiang', 'Jingwen Lu', 'Jian Jiao', 'Anlei Dong', 'Chen Lin', 'Hang Zhang', 'Xiao Liu', 'Yeyun Gong', 'Zhenghao Lin']
|
2022-09-27
| null | null | null | null |
['natural-questions']
|
['miscellaneous']
|
[-8.90145227e-02 -5.27053066e-02 -1.54117122e-01 -1.83199883e-01
-1.12118161e+00 -7.44709313e-01 5.95857620e-01 3.64059895e-01
-7.61618257e-01 9.20971215e-01 2.40892753e-01 -4.92323488e-01
-4.81797606e-01 -6.37027860e-01 -7.72697806e-01 -7.25513041e-01
1.49548575e-01 8.55912089e-01 5.71829438e-01 -2.68482894e-01
2.45937288e-01 2.10681185e-01 -1.45854151e+00 4.23979372e-01
1.29905856e+00 6.88204169e-01 2.09671482e-01 7.91422367e-01
-2.11662516e-01 1.03597689e+00 -5.58672726e-01 -5.37174225e-01
8.04414153e-02 -3.44748795e-01 -1.26012504e+00 -4.65646386e-01
6.14858806e-01 -3.58940750e-01 -2.63160318e-01 9.69273984e-01
4.96168762e-01 5.27918637e-01 9.09406662e-01 -6.59493566e-01
-7.68554211e-01 1.02805877e+00 -5.41239262e-01 3.82447958e-01
3.20908248e-01 -8.84584188e-02 1.13110006e+00 -9.78474081e-01
5.37228584e-01 1.34606767e+00 3.70572269e-01 4.26068157e-01
-9.25974190e-01 -6.52429461e-01 -2.32933704e-02 3.54416490e-01
-1.55214667e+00 1.43400386e-01 4.31232214e-01 -1.96117595e-01
7.34248579e-01 2.69878149e-01 6.84782803e-01 7.14363635e-01
-2.46829256e-01 1.33386981e+00 1.19363451e+00 -4.87195134e-01
2.91066673e-02 3.33815724e-01 4.41291511e-01 5.51351726e-01
-2.80970591e-03 1.63253888e-01 -4.12613153e-01 -4.88579124e-02
2.67042756e-01 -1.61250948e-03 -5.31476855e-01 -2.50538528e-01
-6.50658429e-01 7.45994508e-01 5.77466726e-01 1.77097932e-01
-2.52218306e-01 -4.39576386e-03 2.15262800e-01 8.22768152e-01
2.95195907e-01 7.76801586e-01 -6.69551730e-01 -5.14866054e-01
-1.19334686e+00 6.45965815e-01 8.81633997e-01 9.20509040e-01
6.92007005e-01 -3.77845794e-01 -4.13668871e-01 9.59389091e-01
3.85506183e-01 4.33364362e-01 7.59344637e-01 -8.91751468e-01
4.07103479e-01 5.95656335e-01 -9.53881815e-02 -6.06876850e-01
1.26490578e-01 -5.11641085e-01 -5.67164242e-01 -1.24374151e-01
1.94002301e-01 -1.48573648e-02 -1.12773871e+00 1.54129517e+00
3.79126281e-01 3.34976375e-01 2.47116253e-01 7.74587691e-01
1.24405229e+00 8.96559477e-01 7.00534359e-02 -1.40655026e-01
1.30980217e+00 -1.30039930e+00 -5.76370120e-01 1.36645697e-02
9.40187871e-01 -1.08228636e+00 1.16546369e+00 6.15979135e-01
-1.31226265e+00 -5.22433460e-01 -6.97493911e-01 -4.92678523e-01
-5.71469367e-01 -2.86454916e-01 1.92072064e-01 3.54299486e-01
-8.48099828e-01 6.51357710e-01 -7.20467389e-01 -1.05346151e-01
2.25228935e-01 1.71517357e-01 -1.40603557e-01 -4.47850883e-01
-1.37145233e+00 1.01279891e+00 7.11486220e-01 -2.83933908e-01
-1.04553163e+00 -1.11390793e+00 -6.32023871e-01 4.49026644e-01
4.39484894e-01 -6.88086271e-01 1.62575591e+00 -4.12310839e-01
-1.42699456e+00 7.34224021e-01 4.47045505e-01 -5.89573681e-01
5.32567322e-01 -7.23742187e-01 8.49929005e-02 1.51722267e-01
-3.99598658e-01 9.92879570e-01 4.83778328e-01 -1.09637272e+00
-7.10705757e-01 -1.81557447e-01 1.88037917e-01 6.18022919e-01
-3.94454509e-01 -1.47251666e-01 -6.33874834e-01 -5.06831765e-01
-1.83469757e-01 -9.08401728e-01 -3.12178303e-02 -4.38569874e-01
-1.91643789e-01 -7.53690302e-01 8.68255258e-01 -3.23741972e-01
1.54387927e+00 -2.00041866e+00 1.48844481e-01 2.13441566e-01
2.32388169e-01 8.19966495e-01 -4.84755397e-01 4.03023154e-01
1.96571983e-02 -1.04364768e-01 1.65163264e-01 -2.92090088e-01
4.60622497e-02 3.47829551e-01 -4.31623191e-01 6.85322657e-02
-1.05310030e-01 8.93732309e-01 -1.32452881e+00 -7.03076065e-01
-1.43162450e-02 3.30866188e-01 -9.10486102e-01 3.54512542e-01
-3.70414704e-01 -2.26812623e-02 -5.56064487e-01 5.33311844e-01
6.43653274e-01 -4.37635064e-01 -1.89674854e-01 1.44387141e-01
2.85342876e-02 5.65906465e-01 -1.05663943e+00 1.78090513e+00
-2.15238065e-01 5.91549635e-01 -1.08842015e-01 -9.68610227e-01
6.90959930e-01 3.14789623e-01 2.63606012e-01 -6.49114072e-01
1.09294884e-01 1.49203017e-01 3.56819108e-02 -4.83578116e-01
1.13259625e+00 3.35243680e-02 1.42332211e-01 4.82571542e-01
1.22203328e-01 -7.08749712e-01 4.07791436e-01 6.76299155e-01
1.08635151e+00 -6.34519309e-02 -3.45209837e-02 -3.15875560e-01
3.20586652e-01 -6.16605580e-02 2.67002314e-01 9.58280802e-01
2.80107781e-02 5.18695295e-01 3.51081520e-01 -6.06016554e-02
-6.65217578e-01 -1.28270924e+00 -1.69490799e-02 1.29170680e+00
-1.09188698e-01 -6.79130852e-01 -8.07579100e-01 -9.00104463e-01
1.25767484e-01 5.60607374e-01 -4.82777566e-01 -3.82991433e-01
-3.97489041e-01 -1.71103373e-01 5.82818806e-01 4.65908736e-01
5.77428102e-01 -1.00244856e+00 -2.48622626e-01 -9.56712142e-02
-1.38585955e-01 -9.51564610e-01 -5.59500694e-01 7.88565576e-02
-8.47263038e-01 -9.80471849e-01 -1.02561724e+00 -7.74662197e-01
6.28805339e-01 4.59523529e-01 1.43806648e+00 4.42625791e-01
-6.06032275e-03 4.73195732e-01 -4.25472558e-01 -3.67090940e-01
-3.03116769e-01 4.82254863e-01 -7.94241321e-04 -7.54173458e-01
6.63031101e-01 -4.15265113e-01 -7.16384649e-01 2.33111195e-02
-1.24124908e+00 5.62479682e-02 5.58063745e-01 1.03698158e+00
5.90824425e-01 1.67996123e-01 3.49523544e-01 -1.05809915e+00
8.91627431e-01 -3.92355829e-01 -6.27513349e-01 5.72830379e-01
-9.77040589e-01 3.06902200e-01 4.47632670e-01 -5.68034351e-01
-9.30525243e-01 -3.09822172e-01 -1.41196638e-01 -6.11409426e-01
1.40997529e-01 1.04478157e+00 4.76214647e-01 5.70382252e-02
8.06240916e-01 2.16398954e-01 -3.93684268e-01 -7.59463131e-01
6.64144814e-01 6.73718750e-01 5.23728728e-01 -1.15729380e+00
9.06949222e-01 -1.26541033e-01 -3.62489790e-01 -6.62721574e-01
-1.34687555e+00 -7.25245595e-01 -3.02979738e-01 2.84443162e-02
4.34215099e-01 -9.48394954e-01 -4.85714257e-01 3.81706774e-01
-1.03599095e+00 -5.02155840e-01 -4.78558719e-01 5.93272805e-01
-1.95200935e-01 3.03056568e-01 -7.08248794e-01 -2.38631845e-01
-5.78120887e-01 -1.23055792e+00 6.73849106e-01 7.14474380e-01
-4.05643657e-02 -1.10573518e+00 3.76491845e-01 6.34247303e-01
5.38369834e-01 -4.57166404e-01 1.09710693e+00 -9.29930329e-01
-6.31101251e-01 -1.07953846e-01 -3.31854612e-01 5.40518701e-01
-8.99681151e-02 1.44986704e-01 -7.62681127e-01 -4.20568079e-01
-3.96644890e-01 -1.07799923e+00 1.17403758e+00 -1.47673078e-02
1.26540852e+00 -2.98596919e-01 -6.08647093e-02 3.33979398e-01
1.20557439e+00 -8.21465477e-02 7.71040618e-01 2.29816914e-01
4.53857005e-01 3.05085748e-01 5.98274589e-01 7.91527480e-02
5.54392159e-01 3.90867889e-01 2.30688858e-03 2.53385007e-02
-2.25499511e-01 -4.69543785e-01 2.47943506e-01 1.50166595e+00
-8.66788700e-02 -1.59312978e-01 -1.01153660e+00 7.13440537e-01
-1.60861862e+00 -7.13062167e-01 9.89364684e-02 2.08617425e+00
1.62070191e+00 1.02132559e-01 -7.57721961e-02 -1.56845227e-02
1.31615341e-01 -5.85473999e-02 -1.46853313e-01 -3.44037592e-01
2.22676367e-01 5.92518628e-01 1.36649489e-01 7.65816629e-01
-6.09501004e-01 1.35160160e+00 6.14666319e+00 1.26914179e+00
-9.64726865e-01 -3.54684815e-02 3.26599300e-01 -1.56280398e-01
-4.56739217e-01 -1.02379844e-01 -1.05969918e+00 2.66609699e-01
9.92532670e-01 -3.91624451e-01 2.43314698e-01 7.08324492e-01
-2.97975928e-01 -3.15818429e-01 -1.12452710e+00 7.92684734e-01
-1.34640574e-01 -1.12822664e+00 2.23491356e-01 -2.09988847e-01
1.23626292e+00 1.29070953e-01 3.22180063e-01 1.03174686e+00
8.26677740e-01 -1.02913558e+00 1.14896402e-01 3.61402750e-01
2.00902477e-01 -7.43105531e-01 5.60462773e-01 5.29648364e-01
-6.96429491e-01 2.36004412e-01 -3.42479616e-01 1.46110654e-01
-3.16316664e-01 4.41135108e-01 -1.02183461e+00 4.81665522e-01
6.72846913e-01 4.18047220e-01 -7.25791335e-01 1.22973037e+00
-5.93225837e-01 7.92129993e-01 -3.27810138e-01 -3.01113427e-01
6.09452248e-01 3.01691927e-02 2.92447895e-01 1.36167300e+00
2.39380419e-01 5.06787181e-01 4.12355423e-01 4.78828609e-01
-4.05024260e-01 2.15870544e-01 -2.49988616e-01 -1.53021559e-01
6.95415616e-01 1.04025936e+00 -2.87920296e-01 -7.00342953e-01
-3.85823309e-01 7.34913826e-01 6.15577340e-01 4.79619473e-01
-5.12906611e-01 -5.81775904e-01 3.72692555e-01 -8.63481313e-02
2.33371273e-01 -2.77757645e-02 1.11363389e-01 -1.06849742e+00
-2.73703516e-01 -1.29507744e+00 8.18084538e-01 -7.82767475e-01
-1.21650648e+00 2.28493243e-01 2.43516222e-01 -9.49722707e-01
-3.42854470e-01 -3.73306096e-01 -3.79343837e-01 8.07808459e-01
-2.00913882e+00 -5.36993086e-01 -2.10564122e-01 6.77072644e-01
5.89281678e-01 -1.33981705e-02 7.80597210e-01 5.15235901e-01
-2.01716766e-01 7.15517581e-01 3.32859784e-01 1.39389455e-01
1.00831199e+00 -1.50550067e+00 -1.64032832e-01 4.30812329e-01
3.15803438e-01 8.77356052e-01 7.92568028e-01 -4.09134835e-01
-1.29978347e+00 -9.02247012e-01 9.23627853e-01 -3.21397334e-01
7.91261494e-01 2.28066444e-01 -1.33009827e+00 5.92804551e-01
5.40085018e-01 -8.67797360e-02 6.33520186e-01 4.53237057e-01
-4.53073382e-01 -1.44765571e-01 -9.28401411e-01 4.94284123e-01
3.69512558e-01 -6.34892941e-01 -1.25413871e+00 5.41809678e-01
7.67502666e-01 -8.93446803e-01 -9.84117210e-01 4.58890975e-01
4.40287530e-01 -4.76659566e-01 1.02060568e+00 -8.32582891e-01
7.49170005e-01 -8.61702710e-02 1.63421065e-01 -1.69207144e+00
-1.91595614e-01 -4.86522168e-01 -5.77518821e-01 1.23389411e+00
5.23078442e-01 -1.05919510e-01 9.07639623e-01 6.24411345e-01
-1.63037345e-01 -1.07910097e+00 -6.35191321e-01 -1.01950383e+00
7.08420873e-01 -2.07383469e-01 4.87395167e-01 1.18065357e+00
8.89998153e-02 4.74614531e-01 -1.90730076e-02 -9.48166549e-02
3.67832869e-01 1.07124776e-01 7.15296149e-01 -1.19095361e+00
-3.03681016e-01 -4.84313160e-01 -3.16279195e-02 -1.90300214e+00
-1.43664260e-03 -8.60575676e-01 1.40990943e-01 -1.53090715e+00
4.59172577e-01 -6.31186664e-01 -5.92901170e-01 4.91427809e-01
-5.65076530e-01 2.62892572e-03 2.81005502e-02 -1.36568189e-01
-1.00395179e+00 6.53206944e-01 1.70428264e+00 -2.34859869e-01
-2.84361750e-01 -3.21852155e-02 -6.77404583e-01 5.80612838e-01
3.88729990e-01 -6.95568621e-01 -7.71345019e-01 -6.37836814e-01
4.27967787e-01 -1.62836015e-02 -1.03816904e-01 -6.33064389e-01
7.69936740e-01 -1.91610977e-01 1.15550153e-01 -8.58863056e-01
2.90782273e-01 -6.64205670e-01 -5.71842730e-01 4.82922584e-01
-7.05484211e-01 1.51885793e-01 1.47180289e-01 1.63511232e-01
-4.91070092e-01 -7.40014076e-01 8.43353271e-01 -1.06215142e-01
-5.40040076e-01 3.95916015e-01 -9.03338864e-02 6.37235284e-01
4.51381505e-01 3.66328180e-01 -6.67829812e-01 -3.90985698e-01
-4.60778296e-01 8.27511668e-01 -1.81037746e-03 3.89560431e-01
6.72528565e-01 -1.42399991e+00 -8.70520592e-01 -6.47769794e-02
-5.13584465e-02 4.25418466e-01 1.65147483e-01 7.48006761e-01
-4.37938452e-01 6.94668114e-01 3.20160300e-01 -5.23438931e-01
-1.63121319e+00 4.35238451e-01 1.41804665e-01 -7.79689610e-01
-3.70197535e-01 1.11278343e+00 3.47141325e-02 -6.21802092e-01
6.76477134e-01 -6.18809342e-01 -1.04772463e-01 6.25872612e-02
8.33575487e-01 2.97056645e-01 1.13845490e-01 8.65167528e-02
1.65871203e-01 3.82372320e-01 -8.50343287e-01 -4.31838892e-02
1.16062236e+00 -5.28929010e-03 7.93508142e-02 1.46071821e-01
1.16735888e+00 -9.95729193e-02 -9.20296729e-01 -7.48145759e-01
2.94420682e-02 -4.23417568e-01 2.89470285e-01 -9.74062085e-01
-9.39052820e-01 8.47797573e-01 3.82371664e-01 3.90749812e-01
1.07034922e+00 1.13902129e-02 8.05791795e-01 9.81964886e-01
3.48850712e-02 -1.18187404e+00 3.76345158e-01 1.14530170e+00
6.28433466e-01 -1.23238611e+00 2.91601658e-01 -1.46064565e-01
-3.56922418e-01 7.34118342e-01 8.20049763e-01 1.04144134e-01
6.86054528e-01 5.66565730e-02 -2.30639358e-03 -2.39237472e-01
-1.17655838e+00 -1.71691224e-01 6.18973672e-01 1.57938123e-01
6.83820188e-01 -1.32484585e-01 -3.43995661e-01 3.90421063e-01
-5.00063777e-01 1.91326305e-01 3.05846572e-01 9.30022180e-01
-7.44093537e-01 -1.13668323e+00 -1.61103442e-01 1.81364551e-01
-5.89749157e-01 -4.94527757e-01 -4.58293140e-01 8.47596884e-01
-1.64247259e-01 7.44719684e-01 -1.96016744e-01 -3.93197536e-02
2.48345628e-01 1.44615367e-01 8.45335722e-01 -7.27509022e-01
-8.00686359e-01 -1.72488943e-01 -1.33254781e-01 -1.92439049e-01
-2.53959656e-01 -2.31141895e-01 -1.35267961e+00 -5.24530709e-01
-5.51074147e-01 8.83132041e-01 3.23042810e-01 1.10525250e+00
1.44910678e-01 4.33064371e-01 4.79518533e-01 1.30183339e-01
-9.72167730e-01 -1.19864392e+00 -4.05653358e-01 2.40502402e-01
3.66274059e-01 -4.23747778e-01 -3.84227157e-01 -8.13017637e-02]
|
[11.389006614685059, 7.678753852844238]
|
b20403a8-b5e8-48e5-8682-f1e5501732e0
|
solving-novel-program-synthesis-problems-with
|
2306.04839
| null |
https://arxiv.org/abs/2306.04839v1
|
https://arxiv.org/pdf/2306.04839v1.pdf
|
Solving Novel Program Synthesis Problems with Genetic Programming using Parametric Polymorphism
|
Contemporary genetic programming (GP) systems for general program synthesis have been primarily concerned with evolving programs that can manipulate values from a standard set of primitive data types and simple indexed data structures. In contrast, human programmers do not limit themselves to a small finite set of data types and use polymorphism to express an unbounded number of types including nested data structures, product types, and generic functions. Code-building Genetic Programming (CBGP) is a recently introduced method that compiles type-safe programs from linear genomes using stack-based compilation and a formal type system. Although prior work with CBGP has shown initial demonstrations of polymorphism inside evolved programs, we have provided a deeper exploration of these capabilities through the evolution of programs which make use of generic data types such as key-value maps, tuples, and sets, as well as higher order functions and functions with polymorphic type signatures. In our experiments, CBGP is able to solve problems with all of these properties, where every other GP system that we know of has restrictions that make it unable to even consider problems with these properties. This demonstration provides a significant step towards fully aligning the expressiveness of GP to real world programming.
|
['Thomas Helmuth', 'Edward Pantridge']
|
2023-06-08
| null | null | null | null |
['program-synthesis']
|
['computer-code']
|
[ 1.95802972e-01 3.25861514e-01 -1.52086467e-01 -2.64453799e-01
-2.28801057e-01 -8.95849943e-01 5.52254856e-01 3.11929315e-01
-5.40879630e-02 7.09494412e-01 -2.77921647e-01 -8.83412361e-01
-1.07073281e-02 -1.43634176e+00 -7.02381670e-01 -3.69281650e-01
-8.01770806e-01 4.07299221e-01 6.35732174e-01 -7.41085887e-01
2.86842942e-01 2.77607739e-01 -2.30012107e+00 3.59563351e-01
9.56674218e-01 4.57722366e-01 -6.06416957e-03 8.29349458e-01
-5.84494889e-01 2.33267397e-01 -7.05862761e-01 -3.66662681e-01
2.21723318e-01 -4.01477396e-01 -6.76970422e-01 -4.05396372e-01
7.80686289e-02 8.21886677e-03 4.18818235e-01 8.98775935e-01
3.38352025e-01 -6.35869950e-02 2.13955104e-01 -1.74601555e+00
-7.65049815e-01 7.57828295e-01 -8.94550756e-02 -2.24908158e-01
4.29549187e-01 3.63012165e-01 1.03184414e+00 -2.69306570e-01
6.97589159e-01 1.23372436e+00 8.84119153e-01 7.80289114e-01
-1.51916480e+00 -1.85616061e-01 -9.63696912e-02 -6.68692231e-01
-1.17883193e+00 -1.81747824e-01 3.00475135e-02 -5.41410625e-01
1.41818631e+00 8.42812538e-01 9.19621527e-01 4.30204868e-01
1.89672738e-01 2.36526549e-01 9.56599951e-01 -6.28791213e-01
2.81416088e-01 3.47122222e-01 3.46476912e-01 7.37788796e-01
5.80745101e-01 2.72042453e-01 -2.23949417e-01 -7.11603284e-01
5.89942694e-01 -4.95907605e-01 -3.06024700e-01 -6.62317872e-01
-1.23669958e+00 7.85191238e-01 1.48626894e-01 3.12190682e-01
2.45359838e-01 3.26235682e-01 6.24427497e-01 4.61631030e-01
-2.00286031e-01 8.48261952e-01 -5.60936749e-01 -3.91073793e-01
-2.45253041e-01 6.62200391e-01 1.45731771e+00 1.35258842e+00
7.53446281e-01 1.40892416e-01 5.31030707e-02 6.43640935e-01
2.56655253e-02 2.82677770e-01 4.73003596e-01 -1.00003886e+00
-2.62539228e-03 8.97636890e-01 -4.71421182e-02 -7.05746710e-01
-1.91839859e-01 -1.54967979e-01 -5.36929928e-02 6.48502052e-01
3.16153735e-01 -2.86350660e-02 -5.69268227e-01 1.75389361e+00
3.63944262e-01 -4.86930937e-01 3.60936910e-01 3.50640684e-01
5.54729223e-01 7.05549061e-01 -2.30351776e-01 -5.45320474e-02
1.16307592e+00 -5.38047254e-01 -4.98598032e-02 1.78817958e-01
1.18036437e+00 -1.78981215e-01 1.20176864e+00 3.37113291e-01
-1.37726843e+00 -1.68943465e-01 -1.17609930e+00 -1.46927759e-01
-6.85147285e-01 -8.04901123e-01 1.28120196e+00 1.41507709e+00
-1.56177104e+00 4.68038231e-01 -6.93272948e-01 -4.54726219e-01
-2.24373881e-02 6.46552265e-01 -1.13483675e-01 4.12662596e-01
-8.41770530e-01 6.10988855e-01 8.18530142e-01 -2.08652496e-01
-4.16086197e-01 -8.76126409e-01 -8.67834628e-01 6.88280612e-02
4.18634742e-01 -9.05082285e-01 1.18568325e+00 -9.95807886e-01
-1.32529855e+00 9.22236025e-01 2.51228690e-01 -4.00548309e-01
-3.76203172e-02 4.31338251e-01 -3.94392312e-01 -5.14485657e-01
-2.21131936e-01 4.79103088e-01 2.43893340e-01 -1.14486313e+00
-6.94781363e-01 -1.60500884e-01 6.77501917e-01 6.02892190e-02
-1.80895478e-01 3.67339820e-01 -2.67403983e-02 -4.78239715e-01
-6.27506375e-02 -1.10033333e+00 4.94365133e-02 7.76827708e-03
-2.38929428e-02 -2.02473085e-02 4.49130327e-01 -1.39136061e-01
1.20859683e+00 -2.11634588e+00 4.53546196e-01 4.03432190e-01
7.42149204e-02 1.65895879e-01 1.67201325e-01 5.54163456e-01
2.18868256e-02 5.25812030e-01 -6.56807005e-01 6.26288474e-01
5.94741344e-01 6.44666314e-01 -7.68777207e-02 5.28006852e-02
1.09861769e-01 7.91566074e-01 -8.98196220e-01 -2.30430409e-01
-4.75865722e-01 2.41340101e-02 -1.35353827e+00 4.43503186e-02
-1.05150998e+00 -1.53967872e-01 -4.03955519e-01 7.29370356e-01
4.49167639e-01 4.13324870e-02 5.50483763e-01 5.09973407e-01
-5.88178039e-01 3.48018676e-01 -1.23869574e+00 1.57506847e+00
-4.07182008e-01 1.84723288e-01 3.03772807e-01 -5.30742049e-01
8.61069679e-01 4.49915137e-03 2.16807783e-01 -4.11916137e-01
-1.97130010e-01 5.05603075e-01 3.86021137e-01 -3.45413327e-01
8.27526093e-01 2.20017741e-04 -4.41035420e-01 6.48865938e-01
-3.85198653e-01 -7.76486099e-01 5.06202102e-01 9.72738862e-02
1.15770888e+00 4.73862886e-01 1.43700913e-01 -5.73298037e-01
4.67773318e-01 3.92255515e-01 7.19195306e-01 8.76441479e-01
5.02376199e-01 3.38779002e-01 9.51476038e-01 -3.85242224e-01
-1.23032367e+00 -1.06946468e+00 -3.37578803e-01 1.36223471e+00
-2.51262605e-01 -9.90018666e-01 -7.05341339e-01 -1.56418219e-01
9.12675485e-02 9.24571037e-01 -2.98647940e-01 2.08242964e-02
-7.88264096e-01 -1.00608027e+00 8.81696582e-01 2.75057763e-01
3.41967911e-01 -9.79879797e-01 -1.09031153e+00 2.68462270e-01
4.07483667e-01 -2.60401279e-01 -1.22893535e-01 2.70980895e-01
-7.82477975e-01 -8.46578121e-01 -2.19619751e-01 -8.54675412e-01
5.48050106e-01 -3.36942047e-01 1.41067207e+00 5.60832679e-01
-2.99992919e-01 3.33999157e-01 -2.49764949e-01 -4.10472989e-01
-9.63330865e-01 -6.45409301e-02 -4.97918189e-01 -7.37485707e-01
-2.49789413e-02 -6.18757248e-01 2.56407499e-01 2.18236551e-01
-1.25532258e+00 4.95884381e-02 1.12442993e-01 1.23472524e+00
1.82235479e-01 3.45521122e-01 1.40764073e-01 -1.23339272e+00
6.51655018e-01 -4.68707144e-01 -1.06364048e+00 3.55319917e-01
-5.24827302e-01 5.00101686e-01 3.05979937e-01 -2.34424382e-01
-1.01830196e+00 -2.66833246e-01 -2.42139548e-01 5.40181518e-01
2.27599293e-01 8.50999415e-01 -5.56776524e-01 -2.37868711e-01
9.89009619e-01 2.65909523e-01 3.57466847e-01 -1.37126178e-01
4.48308557e-01 3.72408658e-01 6.89833462e-01 -1.66773832e+00
4.41845328e-01 -6.19552061e-02 2.06645116e-01 -7.63047397e-01
3.95937711e-01 3.25261801e-01 -2.92605072e-01 4.73921359e-01
3.19483638e-01 -3.69673193e-01 -6.48648024e-01 6.39067829e-01
-7.92206287e-01 -6.61766827e-01 -5.98371387e-01 -3.18007648e-01
-8.04057598e-01 2.40881801e-01 -5.30639172e-01 -6.16669893e-01
8.00670087e-02 -1.35629702e+00 8.56641769e-01 -1.92280486e-01
-3.61398578e-01 -8.62506986e-01 9.49880704e-02 -3.84849846e-01
7.90971160e-01 4.77802217e-01 1.71336102e+00 -4.26120073e-01
-8.12506795e-01 1.28857657e-01 2.78806478e-01 -1.60462949e-02
-1.62269369e-01 4.75351602e-01 -3.28873545e-01 -2.52559513e-01
-1.24472469e-01 5.78338876e-02 1.60982698e-01 -1.48643702e-01
9.02025640e-01 -5.10675251e-01 -4.62886274e-01 9.07202899e-01
1.46805477e+00 5.62902987e-01 8.87919247e-01 7.81974912e-01
2.85307795e-01 6.19775653e-01 2.43718147e-01 3.82350177e-01
4.90361840e-01 7.99951315e-01 3.28234732e-01 3.32224369e-01
2.54143953e-01 1.14557855e-02 3.60170782e-01 2.84348845e-01
-1.89456026e-04 -1.88626572e-01 -1.24975967e+00 3.79519612e-01
-1.63241458e+00 -8.70242238e-01 -3.46670538e-01 2.55789614e+00
1.29549456e+00 1.00611433e-01 4.84716654e-01 1.78828821e-01
6.86474442e-01 -3.86235267e-01 -2.42044762e-01 -9.99933720e-01
-5.46911322e-02 4.60885614e-01 5.62868297e-01 3.06760103e-01
-6.59117520e-01 6.09790742e-01 7.28801441e+00 1.38523698e-01
-1.08368528e+00 -2.19871268e-01 1.29172787e-01 1.28444303e-02
-1.07174790e+00 3.86340648e-01 -8.44848096e-01 5.47014952e-01
1.11190522e+00 -8.05015743e-01 8.11314881e-01 7.58521140e-01
-5.96033931e-01 -7.78573304e-02 -1.31351781e+00 2.80749232e-01
-2.24547416e-01 -1.33848560e+00 -1.76962316e-01 2.34137356e-01
5.75235903e-01 -3.91300946e-01 9.64585915e-02 4.23446417e-01
6.65245414e-01 -1.13317764e+00 9.11139250e-01 2.37991020e-01
6.75159514e-01 -5.81428587e-01 1.72715828e-01 1.66687861e-01
-7.48469830e-01 -3.25079471e-01 -2.11982772e-01 -1.96502671e-01
-1.13453403e-01 2.27749899e-01 -6.18334353e-01 5.04509330e-01
7.37144589e-01 1.76658690e-01 -4.74231780e-01 1.10819924e+00
1.74713507e-01 1.44003451e-01 -5.86467743e-01 -1.46834791e-01
-7.87536427e-02 -8.01189095e-02 5.79245389e-01 1.14966285e+00
7.68159211e-01 2.39468336e-01 -9.46930274e-02 1.09399509e+00
5.25219500e-01 -1.37318760e-01 -8.41142118e-01 -2.52337724e-01
5.23859918e-01 7.17704535e-01 -5.38010418e-01 -3.57600510e-01
-5.51115692e-01 1.80230215e-02 8.55043232e-02 9.39294398e-02
-7.62486637e-01 -5.96824169e-01 8.54457378e-01 3.15031201e-01
2.55249560e-01 -5.27480483e-01 -3.80594224e-01 -1.14106703e+00
5.74912168e-02 -1.43612683e+00 3.95690680e-01 -7.10491180e-01
-6.67145312e-01 5.00217140e-01 4.33962554e-01 -5.15432894e-01
-6.37835801e-01 -5.49787700e-01 -4.11076397e-01 1.00133574e+00
-7.33775556e-01 -8.59727979e-01 1.14525445e-01 1.91258267e-01
-2.64149338e-01 -7.80081600e-02 1.04512739e+00 5.90275005e-02
-3.08444768e-01 5.89958608e-01 -7.27766007e-02 -4.21901196e-01
6.89515471e-02 -1.36392486e+00 8.52097869e-01 8.89475346e-01
-5.00791967e-01 1.32519221e+00 7.96894491e-01 -4.77733195e-01
-2.08356643e+00 -6.82436049e-01 5.53593695e-01 -3.67540866e-01
5.91462731e-01 -6.21360481e-01 -1.20367885e+00 8.95661354e-01
-1.74379628e-02 -1.49798483e-01 2.96370268e-01 2.20183909e-01
-4.60089684e-01 -1.06433481e-01 -1.27073550e+00 7.28706002e-01
1.35412204e+00 -3.35833997e-01 -4.62143153e-01 1.90733567e-01
8.53862643e-01 -7.87500381e-01 -1.19340312e+00 3.00735116e-01
7.10315347e-01 -1.12035716e+00 1.00344360e+00 -4.36119705e-01
7.97404051e-02 -5.36983192e-01 -3.51952672e-01 -1.26048911e+00
-2.54658937e-01 -8.11365008e-01 1.13656901e-01 1.57084835e+00
5.68580091e-01 -1.29663384e+00 6.43780410e-01 1.12927580e+00
-6.11270845e-01 -3.75814766e-01 -3.23501736e-01 -1.06313384e+00
4.49065000e-01 -2.85040587e-01 1.45892489e+00 8.32851350e-01
5.24978638e-01 -9.46235284e-02 2.33580291e-01 -2.23062746e-02
2.02306464e-01 3.88559550e-01 1.00338972e+00 -1.08278310e+00
-9.86765385e-01 -7.79942334e-01 -6.07482731e-01 -4.03880417e-01
-1.08616948e-02 -1.21646059e+00 1.21673830e-01 -1.12296581e+00
-1.41661748e-01 -1.35866392e+00 3.23973686e-01 7.09506214e-01
1.95599124e-01 -2.77321607e-01 -5.50303049e-02 -7.58623332e-02
1.06833749e-01 -7.33336359e-02 8.62566888e-01 -4.01997603e-02
-2.54546702e-01 -2.89269894e-01 -9.84766543e-01 3.25972676e-01
7.03773677e-01 -4.65284139e-01 -6.22036576e-01 -5.38348377e-01
8.68269205e-01 2.31867880e-01 3.79566669e-01 -9.60600436e-01
9.17726457e-02 -4.34078127e-01 -3.80895644e-01 -6.45494163e-02
1.47380993e-01 -5.48249125e-01 1.16729152e+00 5.10127187e-01
-3.70657086e-01 3.95792037e-01 5.18780947e-01 2.56997384e-02
6.01236746e-02 -5.66831648e-01 3.90357167e-01 -5.36600351e-01
-1.08206964e+00 -1.15021370e-01 -2.74248868e-01 1.17908917e-01
1.17017925e+00 -5.01764297e-01 -9.27345812e-01 2.95742869e-01
-5.64620018e-01 8.00026208e-02 1.44537687e+00 3.55036825e-01
8.40408951e-02 -1.06252491e+00 -4.73128766e-01 6.81470633e-01
2.68040597e-01 -1.80574954e-02 -4.36676294e-01 3.60198587e-01
-9.63957012e-01 1.95313573e-01 -6.57418489e-01 -5.96802771e-01
-1.15384960e+00 8.93876791e-01 3.97951603e-01 1.31347120e-01
-5.70333600e-01 7.53618836e-01 3.54786783e-01 -8.09191644e-01
-2.01445445e-01 -6.38308823e-01 4.29299444e-01 -3.78344893e-01
3.33850652e-01 1.76287875e-01 2.36798197e-01 -3.87833595e-01
-3.40964466e-01 2.71064043e-01 3.88809979e-01 -2.71457881e-01
1.31892169e+00 4.81556386e-01 -7.65567660e-01 3.53268176e-01
8.73984039e-01 6.28371257e-03 -8.04859698e-01 3.27933848e-01
2.09959354e-02 -6.35009587e-01 -5.01000941e-01 -7.46770084e-01
-6.95016682e-01 4.06610906e-01 1.04926594e-01 6.49310291e-01
1.12161028e+00 -2.50910133e-01 3.66979450e-01 4.40207750e-01
1.05339170e+00 -4.55427259e-01 -4.87267792e-01 5.17212033e-01
7.66978264e-01 -2.80764699e-01 -5.62200584e-02 -7.24870384e-01
-2.18517661e-01 1.09905422e+00 4.42121267e-01 1.79347128e-01
1.99690491e-01 9.47793901e-01 -4.83960539e-01 -1.07783712e-01
-1.13776565e+00 -1.50544897e-01 -2.96840638e-01 1.05996931e+00
5.07802784e-01 4.17074002e-02 -7.07196712e-01 2.40286633e-01
-5.75065851e-01 3.00465286e-01 8.92307520e-01 1.70560765e+00
-4.30093586e-01 -1.78595757e+00 -5.99885762e-01 4.75906253e-01
-4.55143720e-01 -2.44223952e-01 -9.33642015e-02 1.19309247e+00
3.58976394e-01 4.66767788e-01 2.72224575e-01 -4.33544554e-02
1.96672127e-01 4.99387085e-02 9.46267843e-01 -1.09482455e+00
-1.02321267e+00 -4.73983824e-01 6.65557325e-01 -2.96420962e-01
-1.42702146e-03 -8.87310088e-01 -1.32675278e+00 -5.84128678e-01
-1.60359293e-01 3.58870685e-01 7.71931350e-01 2.28080973e-01
4.33063656e-01 4.64778304e-01 -5.80945006e-03 -4.77870345e-01
-4.81785476e-01 5.97334020e-02 -6.29181862e-01 4.29271795e-02
7.93593526e-02 -4.56875414e-01 1.45798996e-02 8.30103382e-02]
|
[8.054279327392578, 7.291537284851074]
|
b16b3fd2-725a-4101-98ab-4959ba726907
|
mapping-and-localization-from-planar-markers
|
1606.00151
| null |
http://arxiv.org/abs/1606.00151v2
|
http://arxiv.org/pdf/1606.00151v2.pdf
|
Mapping and Localization from Planar Markers
|
Squared planar markers are a popular tool for fast, accurate and robust
camera localization, but its use is frequently limited to a single marker, or
at most, to a small set of them for which their relative pose is known
beforehand. Mapping and localization from a large set of planar markers is yet
a scarcely treated problem in favour of keypoint-based approaches. However,
while keypoint detectors are not robust to rapid motion, large changes in
viewpoint, or significant changes in appearance, fiducial markers can be
robustly detected under a wider range of conditions. This paper proposes a
novel method to simultaneously solve the problems of mapping and localization
from a set of squared planar markers. First, a quiver of pairwise relative
marker poses is created, from which an initial pose graph is obtained. The pose
graph may contain small pairwise pose errors, that when propagated, leads to
large errors. Thus, we distribute the rotational and translational error along
the basis cycles of the graph so as to obtain a corrected pose graph. Finally,
we perform a global pose optimization by minimizing the reprojection errors of
the planar markers in all observed frames. The experiments conducted show that
our method performs better than Structure from Motion and visual SLAM
techniques.
|
['Rafael Medina-Carnicer', 'Enrique Yeguas-Bolivar', 'Manuel J. Marín-Jimenez', 'Rafael Muñoz-Salinas']
|
2016-06-01
| null | null | null | null |
['camera-localization']
|
['computer-vision']
|
[ 1.82256073e-01 -3.63085926e-01 -7.03291371e-02 -6.33600056e-02
-7.39074409e-01 -7.55339026e-01 6.61391139e-01 4.50592488e-01
-4.87162262e-01 7.22397089e-01 -4.35966253e-01 1.71946421e-01
-6.54796362e-02 -4.72624302e-01 -8.11390519e-01 -5.49587429e-01
1.50774017e-01 7.44882524e-01 4.37976420e-01 3.93541791e-02
3.89362007e-01 8.35440159e-01 -1.05287135e+00 -9.69720602e-01
7.24071324e-01 6.94438517e-01 3.92754108e-01 4.38573211e-01
2.19615012e-01 2.53783762e-01 -6.14471912e-01 -1.70475841e-01
1.91577509e-01 -3.29176545e-01 -3.61386180e-01 3.74307096e-01
6.58856094e-01 -4.74122353e-02 6.63249865e-02 1.32421041e+00
3.74442875e-01 2.75450915e-01 2.11447433e-01 -1.00527227e+00
1.44217268e-01 -2.45703962e-02 -8.87763917e-01 -2.81524569e-01
8.84239852e-01 -3.74455720e-01 5.33238709e-01 -8.89443874e-01
9.20966566e-01 6.42368853e-01 8.89312327e-01 7.86792561e-02
-1.19540405e+00 -3.05442244e-01 -9.29587260e-02 -2.87595570e-01
-1.83726406e+00 -4.14434314e-01 9.03045297e-01 -4.25338864e-01
3.37756842e-01 3.98558199e-01 6.47699952e-01 6.00521445e-01
1.72041044e-01 -1.68040246e-01 8.02170694e-01 -6.76499128e-01
1.35412917e-01 1.83702737e-01 -1.67725980e-01 8.82724941e-01
6.78931892e-01 -1.92640707e-01 -4.21913177e-01 -3.08274090e-01
1.05304158e+00 1.88165516e-01 -5.73950231e-01 -1.28453326e+00
-1.58800280e+00 5.56712747e-01 5.55630445e-01 2.08854645e-01
-3.14542174e-01 1.50617018e-01 -1.15939327e-01 -1.49069637e-01
2.21091941e-01 5.27320564e-01 8.63037929e-02 -1.24428108e-01
-7.25841343e-01 1.11643523e-01 6.64754212e-01 1.10302460e+00
1.14613080e+00 -1.68933406e-01 5.66721618e-01 5.17283916e-01
2.74560004e-01 6.97569430e-01 1.55924618e-01 -8.61343741e-01
4.56365973e-01 5.72361052e-01 5.13044894e-01 -1.67499781e+00
-3.65708381e-01 -7.46972933e-02 -6.51840210e-01 3.25354375e-02
5.12633383e-01 -8.16412419e-02 -4.99561042e-01 1.48447430e+00
5.73922455e-01 -8.98812804e-03 -3.39759320e-01 1.12443423e+00
1.30722195e-01 3.22964281e-01 -5.26532531e-01 -2.89780468e-01
1.02733183e+00 -7.88254857e-01 -5.42962730e-01 -6.31418645e-01
3.91500086e-01 -1.06039548e+00 3.44239920e-01 1.70766413e-01
-8.15667748e-01 -3.10442537e-01 -1.13371980e+00 8.83875713e-02
-9.21610743e-02 1.73875481e-01 1.62580103e-01 3.93663347e-01
-9.05163825e-01 4.23248529e-01 -8.46134961e-01 -5.14646709e-01
-4.05315876e-01 4.85405654e-01 -7.12135375e-01 -1.26045436e-01
-6.22920334e-01 1.23175859e+00 1.12225957e-01 4.82279629e-01
-4.39887851e-01 -9.70807299e-02 -9.70049500e-01 -3.72445524e-01
6.56457007e-01 -5.08306861e-01 8.33926916e-01 -6.87490404e-01
-1.43692803e+00 7.78717637e-01 -4.43270743e-01 -1.28631279e-01
7.93954432e-01 -2.15298548e-01 4.26982483e-03 1.01090111e-01
2.48995736e-01 3.79306227e-01 8.71063411e-01 -1.50662816e+00
-2.27821589e-01 -3.49900663e-01 -1.37702124e-02 4.91724938e-01
1.93969205e-01 -1.00900806e-01 -8.48584175e-01 -1.62168175e-01
7.75483191e-01 -1.33820176e+00 -4.04421508e-01 8.49479139e-02
-3.48354429e-01 3.51830781e-01 5.15922308e-01 -4.43224370e-01
8.09330404e-01 -2.05109644e+00 3.52628887e-01 5.45392513e-01
1.24069102e-01 -6.28638193e-02 2.51049936e-01 3.73523057e-01
1.59457281e-01 -2.16731802e-01 -3.44107002e-02 -3.23689729e-01
-3.00975502e-01 1.51752442e-01 1.24155544e-01 1.00311422e+00
-4.10845399e-01 3.13921064e-01 -1.02424431e+00 -4.06577408e-01
5.37889481e-01 5.01130700e-01 -1.06376737e-01 -4.89293896e-02
1.25221074e-01 5.10162115e-01 -4.33527678e-01 5.20137429e-01
5.98748326e-01 7.21355202e-03 1.44776016e-01 -2.13196114e-01
-3.57418656e-01 4.64401096e-02 -1.61870992e+00 1.85443962e+00
-1.57342523e-01 6.12280786e-01 1.65137231e-01 -5.69542050e-01
1.24276352e+00 2.46203914e-01 6.28017902e-01 -1.88148871e-01
2.43752137e-01 4.43647504e-01 -1.88953400e-01 1.31446376e-01
7.87337363e-01 2.20667291e-02 -1.18789591e-01 5.17908596e-02
-3.74344647e-01 -3.82768452e-01 1.29487202e-01 3.03227138e-02
8.19614232e-01 2.02096030e-01 6.35267854e-01 7.12702200e-02
5.72784722e-01 1.57359451e-01 6.18576646e-01 3.29287469e-01
1.40725430e-02 8.86618257e-01 1.67867079e-01 -4.57821310e-01
-9.67345059e-01 -7.81951964e-01 -1.53740615e-01 2.22785920e-01
8.57793689e-01 -5.29870033e-01 -6.38275504e-01 -3.27945739e-01
6.32479787e-02 -6.56565055e-02 -2.94693917e-01 -1.81072913e-02
-6.57600582e-01 -3.70508015e-01 1.52905658e-02 1.83562085e-01
4.17548060e-01 -3.84187967e-01 -8.59467804e-01 2.04307899e-01
-1.34509206e-01 -1.14758277e+00 -6.50582016e-01 -8.90154019e-03
-8.68475080e-01 -1.27697396e+00 -8.48676622e-01 -6.62717044e-01
1.20368075e+00 7.39429653e-01 6.16177201e-01 6.82406351e-02
3.09577826e-02 3.60252947e-01 -2.72680253e-01 -8.88336599e-02
-1.49717003e-01 -2.61287272e-01 3.05299312e-01 1.63138628e-01
-1.47167578e-01 -2.03000546e-01 -2.62781590e-01 8.48111808e-01
-3.89052927e-01 -6.94293976e-02 2.42752850e-01 5.27037978e-01
8.52290928e-01 -2.14300066e-01 -1.67079553e-01 -4.27134693e-01
2.64495939e-01 3.58027034e-02 -1.09649622e+00 1.89505070e-01
-2.81645417e-01 -1.56657249e-01 2.12646320e-01 -5.11359036e-01
-6.01707876e-01 6.10424757e-01 3.68090212e-01 -5.19855559e-01
1.00391500e-01 5.80168843e-01 -2.01817185e-01 -7.22291410e-01
6.40586436e-01 -4.11045738e-02 1.31915644e-01 -4.24478233e-01
2.92897552e-01 3.00780207e-01 6.15872502e-01 -3.40130925e-01
1.03385806e+00 4.95473474e-01 3.23880643e-01 -8.32871497e-01
-3.90110105e-01 -5.79442799e-01 -1.10408556e+00 -3.53934079e-01
5.58280826e-01 -8.13449502e-01 -2.98521638e-01 3.79507244e-01
-1.23814070e+00 9.49455276e-02 4.30780509e-03 7.14463890e-01
-4.37859386e-01 7.92861760e-01 -8.81863534e-02 -5.50407469e-01
6.91324696e-02 -1.29942811e+00 9.50786471e-01 3.22098881e-01
-4.27261025e-01 -1.04156244e+00 3.18511665e-01 -7.43507892e-02
-8.25219452e-02 5.92091322e-01 1.12541184e-01 -1.07860610e-01
-8.29242110e-01 -7.77449250e-01 1.35591030e-01 -1.28549665e-01
3.10332417e-01 1.05472319e-01 -3.21831673e-01 -5.42544723e-01
2.07756478e-02 3.67441773e-01 1.68932021e-01 1.47021413e-01
2.13592127e-01 8.31096545e-02 -7.85079896e-01 6.65659070e-01
1.48501658e+00 3.58028680e-01 2.69111753e-01 7.02852964e-01
8.56303334e-01 3.22734982e-01 8.50307584e-01 2.34307334e-01
3.27643603e-01 1.05574632e+00 5.08410513e-01 8.71448517e-02
1.99915230e-01 -2.82963425e-01 2.03653112e-01 7.53185987e-01
-2.17898175e-01 2.99923532e-02 -1.03153670e+00 4.43978846e-01
-1.81021118e+00 -3.75681609e-01 -3.17835987e-01 2.80110908e+00
4.16651160e-01 -1.09591208e-01 -1.45107016e-01 8.26878287e-03
1.05064595e+00 1.79734066e-01 -2.66170532e-01 1.01768039e-02
4.33373079e-02 -4.25661772e-01 8.64661515e-01 8.37256074e-01
-8.35091233e-01 7.89362490e-01 5.79540443e+00 1.78929389e-01
-1.21035016e+00 -3.94847840e-02 -3.83507520e-01 1.88659385e-01
3.53152230e-02 5.19402802e-01 -8.52504432e-01 4.41146106e-01
3.59969229e-01 -8.15458223e-02 2.94793189e-01 8.12634349e-01
-1.22393802e-01 -7.10145831e-01 -9.42937374e-01 1.22064316e+00
3.84697855e-01 -9.88692880e-01 -3.13755572e-01 2.10206747e-01
7.30598271e-01 -1.45913335e-03 -3.39410305e-01 -4.77111876e-01
-3.86784561e-02 -4.13529724e-01 7.66753793e-01 3.43565404e-01
6.98013365e-01 -6.70780361e-01 5.34770072e-01 5.97500980e-01
-1.18971217e+00 4.21335399e-01 -5.92750490e-01 1.20928518e-01
3.65182459e-01 4.94704723e-01 -7.51554906e-01 7.20582306e-01
1.33094013e-01 4.41026270e-01 -4.77334410e-01 1.42347479e+00
-3.25761259e-01 -2.06694096e-01 -6.20681047e-01 -4.24940959e-02
5.14795296e-02 -5.54011285e-01 7.52464652e-01 6.22377574e-01
7.60429859e-01 -1.59132421e-01 3.74445289e-01 4.30790752e-01
1.24050744e-01 5.42916022e-02 -8.15912724e-01 4.31725532e-01
5.97079515e-01 1.38091862e+00 -9.15957928e-01 -7.20505267e-02
-3.79491627e-01 1.11579382e+00 3.14735472e-01 2.11521044e-01
-7.32490778e-01 -2.80394256e-01 3.22672725e-01 3.33706439e-01
-4.66130003e-02 -7.76187003e-01 6.65143803e-02 -1.38619673e+00
2.50153691e-01 -4.98205036e-01 3.36982869e-02 -9.03072298e-01
-3.72907877e-01 5.19188702e-01 1.08112708e-01 -1.45105875e+00
-4.98161077e-01 -2.36868814e-01 -4.46578354e-01 8.89511347e-01
-1.08552730e+00 -9.29951727e-01 -6.07082069e-01 5.14812589e-01
8.54385272e-02 2.66639590e-01 7.37960100e-01 2.03052923e-01
-3.73210311e-01 1.02945156e-01 2.42191702e-01 5.48585877e-02
9.51083660e-01 -1.06698656e+00 1.68491244e-01 1.10112429e+00
4.21895832e-01 7.59070516e-01 7.76635289e-01 -7.57908881e-01
-1.52041912e+00 -6.78303778e-01 1.00459433e+00 -5.22763133e-01
4.69300807e-01 -4.35448974e-01 -7.51009703e-01 8.70520473e-01
-3.34493488e-01 3.11002254e-01 -5.73773906e-02 -7.07973689e-02
-2.34601158e-03 -1.91516951e-01 -8.54187429e-01 4.47775573e-01
5.93055308e-01 -4.12981838e-01 -2.68500984e-01 3.40111047e-01
2.39840642e-01 -1.03836620e+00 -6.43297672e-01 2.35426679e-01
4.38827336e-01 -7.75266349e-01 9.94059324e-01 3.67994130e-01
-5.68494499e-01 -8.07599127e-01 7.18155056e-02 -1.39565003e+00
-1.13593355e-01 -9.09409881e-01 2.95664102e-01 1.22969031e+00
1.56091284e-02 -6.61467433e-01 7.67386794e-01 4.30800706e-01
9.77175459e-02 -1.87234581e-01 -1.10246408e+00 -8.18217576e-01
-6.10717237e-01 1.35168806e-02 3.59005988e-01 9.51935470e-01
-1.36385125e-03 2.35956028e-01 -5.15222609e-01 4.14254487e-01
7.11841524e-01 2.00198099e-01 1.15268934e+00 -1.26756203e+00
1.45055160e-01 6.15103450e-03 -7.31434405e-01 -1.06134963e+00
-1.29491329e-01 -3.90540212e-01 4.53483611e-01 -1.33915997e+00
-8.61822069e-02 -6.88505113e-01 2.03004882e-01 -3.20024975e-02
-7.24123791e-02 2.72066146e-01 2.46826231e-01 4.77270544e-01
-5.09240329e-01 5.33737950e-02 9.22392309e-01 2.83002108e-01
-3.08353513e-01 -1.75192561e-02 -1.80339351e-01 1.02755594e+00
4.61468071e-01 -5.85108101e-01 -1.50176942e-01 -5.53153694e-01
2.75588304e-01 3.93197089e-01 4.53396410e-01 -1.06120169e+00
6.02746606e-01 -1.38844386e-01 2.79704481e-01 -5.49840093e-01
5.66384554e-01 -9.06587124e-01 6.64632380e-01 2.98582554e-01
3.61132413e-01 3.56718481e-01 -1.73720762e-01 6.10496223e-01
-3.09037954e-01 -5.33826947e-01 6.83745325e-01 -2.01103494e-01
-6.31663144e-01 2.01671630e-01 2.52684931e-05 -3.38817120e-01
1.18037403e+00 -5.91978610e-01 -2.14547850e-02 -5.14687717e-01
-6.80813313e-01 -1.52460635e-01 1.18016911e+00 1.87611088e-01
5.72197378e-01 -1.38305652e+00 -1.86317682e-01 1.99106619e-01
1.46684006e-01 6.15510523e-01 -1.15606062e-01 1.17776787e+00
-9.20761108e-01 3.51152927e-01 1.45154195e-02 -9.65850830e-01
-1.46083260e+00 5.09808719e-01 3.44643682e-01 1.44451484e-01
-5.67208707e-01 5.90532064e-01 -1.12175353e-01 -3.80863130e-01
1.29568696e-01 -2.56303996e-01 1.40573531e-01 1.11804627e-01
1.54427454e-01 4.34760869e-01 4.11016569e-02 -1.28993607e+00
-5.72582126e-01 1.37682855e+00 2.91265041e-01 -1.49569586e-01
9.27764535e-01 -4.54875022e-01 -2.45662540e-01 3.40579093e-01
1.12975824e+00 5.74272335e-01 -1.22439003e+00 -1.64326146e-01
6.14910647e-02 -9.31622744e-01 -3.16855490e-01 -1.11473098e-01
-6.35917425e-01 6.38204157e-01 3.47778887e-01 -3.93963978e-02
7.30602980e-01 -1.33304462e-01 4.07793403e-01 3.15313846e-01
7.59881139e-01 -6.94706798e-01 -1.72808796e-01 4.48313028e-01
7.35809743e-01 -1.11678982e+00 2.67517537e-01 -6.19813442e-01
-3.21765363e-01 1.09928870e+00 3.06538105e-01 -2.34598368e-01
1.25303820e-01 2.36275062e-01 1.63355783e-01 7.13884681e-02
5.03358208e-02 4.03369851e-02 3.17218244e-01 6.03421152e-01
1.19719647e-01 -5.85307516e-02 -2.01591462e-01 -3.88144940e-01
-1.46940872e-01 -3.05743843e-01 5.99636555e-01 1.05784690e+00
-5.76453924e-01 -1.20981467e+00 -1.00022805e+00 -3.38108867e-01
-1.50579467e-01 3.46362263e-01 -4.78378415e-01 1.07252502e+00
-8.64720251e-03 6.12106442e-01 1.50342673e-01 -1.60900772e-01
4.04655695e-01 -2.18983218e-01 6.69070840e-01 -4.47713286e-01
-9.54058394e-02 4.36254978e-01 1.12460656e-02 -5.66476882e-01
-4.81450558e-01 -8.63612354e-01 -1.17201090e+00 -9.04448628e-02
-7.51298249e-01 2.71758735e-01 9.74260330e-01 6.46418869e-01
3.05802952e-02 -8.85702148e-02 5.89953423e-01 -1.13635719e+00
-2.82866329e-01 -5.36101699e-01 -5.82298458e-01 3.85963440e-01
5.00777066e-01 -8.01382184e-01 -3.90081495e-01 -1.50851130e-01]
|
[7.831070899963379, -2.2698867321014404]
|
6d0c02b6-6d90-4111-bac1-8aa719bb4304
|
dual-attention-network-for-scene-segmentation
|
1809.02983
| null |
http://arxiv.org/abs/1809.02983v4
|
http://arxiv.org/pdf/1809.02983v4.pdf
|
Dual Attention Network for Scene Segmentation
|
In this paper, we address the scene segmentation task by capturing rich
contextual dependencies based on the selfattention mechanism. Unlike previous
works that capture contexts by multi-scale features fusion, we propose a Dual
Attention Networks (DANet) to adaptively integrate local features with their
global dependencies. Specifically, we append two types of attention modules on
top of traditional dilated FCN, which model the semantic interdependencies in
spatial and channel dimensions respectively. The position attention module
selectively aggregates the features at each position by a weighted sum of the
features at all positions. Similar features would be related to each other
regardless of their distances. Meanwhile, the channel attention module
selectively emphasizes interdependent channel maps by integrating associated
features among all channel maps. We sum the outputs of the two attention
modules to further improve feature representation which contributes to more
precise segmentation results. We achieve new state-of-the-art segmentation
performance on three challenging scene segmentation datasets, i.e., Cityscapes,
PASCAL Context and COCO Stuff dataset. In particular, a Mean IoU score of 81.5%
on Cityscapes test set is achieved without using coarse data. We make the code
and trained model publicly available at https://github.com/junfu1115/DANet
|
['Zhiwei Fang', 'Yongjun Bao', 'Jing Liu', 'Hanqing Lu', 'Haijie Tian', 'Yong Li', 'Jun Fu']
|
2018-09-09
|
dual-attention-network-for-scene-segmentation-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Fu_Dual_Attention_Network_for_Scene_Segmentation_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Fu_Dual_Attention_Network_for_Scene_Segmentation_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['thermal-image-segmentation']
|
['computer-vision']
|
[ 9.80414897e-02 -1.84945107e-01 5.45598902e-02 -5.97381949e-01
-7.28407741e-01 -5.21357834e-01 3.72727692e-01 3.02181691e-01
-6.02519572e-01 5.47927260e-01 2.36812219e-01 7.77797550e-02
-4.26171571e-02 -7.72774398e-01 -7.42723048e-01 -7.82812476e-01
-1.19213352e-03 3.37517969e-02 6.21896684e-01 -2.31674314e-01
2.54322916e-01 2.44602978e-01 -1.45565557e+00 3.81209850e-01
1.30077696e+00 1.08645558e+00 5.92628956e-01 5.99233806e-01
-1.50764048e-01 4.33273256e-01 -3.82144719e-01 -2.20281288e-01
1.18190072e-01 -1.94791511e-01 -9.15159404e-01 -4.47885208e-02
4.02383447e-01 -2.07837194e-01 -2.55525529e-01 1.15889823e+00
3.50919664e-01 2.81943262e-01 5.20434141e-01 -9.51150000e-01
-9.43306208e-01 5.29068351e-01 -7.83064008e-01 4.41699922e-01
4.14010994e-02 1.83712006e-01 1.23189342e+00 -6.22824669e-01
3.01120192e-01 1.09539509e+00 4.23868299e-01 1.79439619e-01
-1.00807631e+00 -4.54424351e-01 7.26772070e-01 3.85085046e-01
-1.21397865e+00 3.60037684e-02 7.20309913e-01 -2.35089734e-01
8.40150118e-01 2.16412261e-01 5.57380974e-01 7.27370858e-01
-1.75207686e-02 1.07999969e+00 8.73242855e-01 -2.39360243e-01
-6.74296543e-02 -2.01872125e-01 3.83312672e-01 5.04266381e-01
4.18407470e-02 -3.04520249e-01 -8.76396000e-02 2.91612864e-01
9.18402731e-01 2.17901394e-01 -2.54086733e-01 -1.98381711e-02
-1.06120050e+00 8.87963355e-01 9.83427644e-01 5.40579736e-01
-3.67109805e-01 2.05372542e-01 2.75425583e-01 -1.14861555e-01
3.17150921e-01 2.27051929e-01 -6.60108089e-01 1.87070251e-01
-4.04879212e-01 7.31736794e-02 1.38778910e-01 1.04598129e+00
1.00871503e+00 -3.22698683e-01 -4.36957240e-01 9.99191880e-01
3.07885766e-01 3.58325511e-01 5.38167417e-01 -8.05061042e-01
5.45239091e-01 6.91932380e-01 -1.61534861e-01 -8.98193598e-01
-4.97490585e-01 -5.57865918e-01 -8.09610188e-01 -3.04077834e-01
2.86994994e-01 -2.92975187e-01 -1.29376745e+00 1.70214224e+00
3.65573198e-01 5.42878389e-01 -5.56323342e-02 1.08435345e+00
1.09355545e+00 7.82595694e-01 3.37698549e-01 2.91150838e-01
1.38368046e+00 -1.39611101e+00 -5.71195900e-01 -2.59550661e-01
6.06828332e-01 -8.68422449e-01 1.16824508e+00 8.78080353e-03
-9.88404870e-01 -8.56091499e-01 -9.35893774e-01 -4.36621010e-01
-6.18998170e-01 1.88192591e-01 8.05315435e-01 8.06515440e-02
-8.66677523e-01 4.34298128e-01 -8.21847975e-01 -2.54254103e-01
5.86181104e-01 3.88680130e-01 -7.55463615e-02 -5.15294187e-02
-1.22554219e+00 3.04057032e-01 2.70881593e-01 2.51203299e-01
-4.82670218e-01 -5.46301842e-01 -9.51646149e-01 1.41402200e-01
2.74116457e-01 -3.78526509e-01 1.16870368e+00 -1.05776787e+00
-1.19889784e+00 5.24835885e-01 -2.71682799e-01 -2.27620974e-01
2.12431580e-01 -3.82740289e-01 -2.89364189e-01 8.85685831e-02
3.96454036e-01 9.95989919e-01 3.59277070e-01 -1.20304132e+00
-9.53348577e-01 -3.26590955e-01 2.10308611e-01 4.07969207e-01
-1.25075951e-01 -2.76514918e-01 -1.01988518e+00 -5.96935570e-01
1.67922884e-01 -6.79453552e-01 -5.02531588e-01 -3.41852576e-01
-5.18585563e-01 -1.04166418e-01 9.19216514e-01 -4.78156090e-01
1.18493879e+00 -2.30687380e+00 5.38621210e-02 8.92954767e-02
6.69600815e-02 4.06461716e-01 -3.46748292e-01 4.31395043e-03
1.29027516e-01 2.08232522e-01 -4.26854521e-01 -1.91909507e-01
-1.73420891e-01 2.78096050e-01 -2.41140369e-02 1.54975086e-01
3.72649342e-01 1.17971027e+00 -8.75649333e-01 -5.19894481e-01
5.01325667e-01 5.56141794e-01 -6.13536000e-01 1.23114638e-01
-1.44724533e-01 5.99777341e-01 -7.29576886e-01 5.94403625e-01
8.30460668e-01 -3.12907070e-01 -1.39321625e-01 -1.83204040e-01
-2.19937101e-01 2.17346270e-02 -1.07457900e+00 1.96889055e+00
-5.16772449e-01 4.02441502e-01 7.60885179e-02 -9.24489558e-01
8.18306267e-01 7.51518970e-03 3.60126168e-01 -1.00115585e+00
2.71691859e-01 8.11262503e-02 8.47928226e-02 -4.32497948e-01
5.63949347e-01 3.31201702e-01 -2.31845483e-01 -2.44139880e-02
2.33771041e-01 5.82440607e-02 2.93003947e-01 1.62539914e-01
7.74524808e-01 6.55366760e-03 6.94694221e-02 -3.74871194e-01
6.77992702e-01 -1.27914757e-01 6.26495123e-01 6.01604939e-01
-4.43697125e-01 8.46680760e-01 4.72272396e-01 -4.25514638e-01
-6.91381037e-01 -9.72173035e-01 -2.65064567e-01 1.32643020e+00
6.91648722e-01 -2.59401143e-01 -8.53920221e-01 -7.43656218e-01
-5.94872087e-02 3.69756132e-01 -9.55617726e-01 1.33155107e-01
-4.98974115e-01 -7.75591969e-01 3.23743492e-01 9.87377524e-01
9.50987279e-01 -1.17300987e+00 -5.74680984e-01 2.18978077e-01
-2.69049436e-01 -1.06303430e+00 -6.83483601e-01 3.34304988e-01
-5.54960132e-01 -1.05820107e+00 -8.06021988e-01 -9.72771704e-01
5.70283532e-01 3.12281787e-01 8.15992236e-01 1.14878424e-01
-3.61525536e-01 9.73349903e-03 -5.15263617e-01 -2.62541592e-01
5.25797904e-01 3.56800824e-01 -7.06016362e-01 1.35664701e-01
2.83176064e-01 -4.09839690e-01 -8.59157264e-01 4.24191982e-01
-8.28724325e-01 1.85234338e-01 7.26252854e-01 7.98189282e-01
1.01274312e+00 -2.18653321e-01 5.37366688e-01 -1.02821314e+00
3.42492461e-01 -6.25553787e-01 -5.46677351e-01 1.84865922e-01
7.28231817e-02 4.80273040e-03 6.87801182e-01 -2.57721007e-01
-1.25425017e+00 2.21364871e-01 -3.87939900e-01 -1.84511110e-01
-5.05480051e-01 2.99195588e-01 -4.38938677e-01 1.85967490e-01
2.42437586e-01 1.31171197e-01 -6.56245887e-01 -3.69430542e-01
5.10188222e-01 6.29600048e-01 6.57798409e-01 -6.68158054e-01
2.36931711e-01 4.37290072e-01 -4.04151410e-01 -6.47088468e-01
-9.18528736e-01 -6.20585978e-01 -8.78140807e-01 5.13307080e-02
1.38759112e+00 -8.35805774e-01 -5.26959419e-01 6.10744774e-01
-1.00493419e+00 -5.34067690e-01 -1.76899582e-01 4.51737344e-01
-4.03960407e-01 1.58315733e-01 -6.17011249e-01 -3.95347893e-01
-2.34283119e-01 -1.33152735e+00 1.26928067e+00 7.79458821e-01
1.36577159e-01 -9.86367285e-01 -1.16257519e-01 1.83664680e-01
4.42597687e-01 3.34158868e-01 7.26526201e-01 -5.15320480e-01
-6.20915711e-01 8.38211700e-02 -6.97770059e-01 1.08919926e-01
1.43468469e-01 7.44970292e-02 -1.02717054e+00 5.63887544e-02
-4.21960026e-01 -8.36865604e-02 1.27276397e+00 7.81825721e-01
1.54404509e+00 1.41223222e-01 -4.01419044e-01 8.18742394e-01
1.47045815e+00 3.49465311e-01 5.55308700e-01 1.14635348e-01
9.57086444e-01 4.83416945e-01 5.19957900e-01 2.53448308e-01
5.55100024e-01 4.34815764e-01 5.53781867e-01 -4.71350700e-01
-1.11840062e-01 -1.36122525e-01 -1.74021229e-01 5.36047637e-01
-2.89441273e-02 -3.35052550e-01 -8.08835745e-01 8.07997644e-01
-2.08341002e+00 -6.45760477e-01 -4.29525375e-01 1.93204474e+00
4.38850403e-01 2.61667192e-01 -1.39391124e-01 -2.29618534e-01
9.31572318e-01 2.63331026e-01 -5.90381444e-01 -3.57974201e-01
-2.14423150e-01 2.14283168e-01 5.79851985e-01 6.19639337e-01
-1.51472902e+00 1.17211843e+00 5.16886282e+00 9.74882126e-01
-1.10614383e+00 1.71530426e-01 1.00470495e+00 -9.16255359e-03
-3.03017378e-01 -9.91866887e-02 -5.97910941e-01 6.54135466e-01
4.87684727e-01 2.88898200e-01 1.45898819e-01 7.23382235e-01
-2.48187203e-02 -2.31191292e-01 -5.42738199e-01 6.69426382e-01
-1.44981921e-01 -1.14662981e+00 -4.27554138e-02 -1.02098137e-01
9.78166938e-01 3.15646499e-01 2.20205367e-01 2.12667510e-01
5.00423849e-01 -1.03653562e+00 5.87370336e-01 4.21983719e-01
5.36693990e-01 -8.81120324e-01 8.94243836e-01 1.98648274e-02
-1.69168901e+00 -1.88348070e-01 -3.79462451e-01 4.61063795e-02
2.22713962e-01 4.93093699e-01 -1.92246631e-01 8.36160779e-01
8.63719881e-01 8.92388344e-01 -6.47347867e-01 1.22086430e+00
-3.66445452e-01 5.71494102e-01 -4.25424665e-01 1.68506786e-01
7.49119461e-01 -9.76587459e-02 1.28695175e-01 1.37010586e+00
1.36264786e-01 2.35842854e-01 2.27349922e-01 7.28086233e-01
-1.65206008e-02 2.55905867e-01 -2.09911764e-01 4.83873099e-01
3.42640966e-01 1.61092818e+00 -9.98530209e-01 -5.06567240e-01
-6.90457702e-01 1.03378415e+00 3.69644225e-01 5.87435603e-01
-1.04844654e+00 -6.65402114e-01 6.46366358e-01 -1.49315879e-01
7.24518299e-01 -7.41898045e-02 -4.69729871e-01 -1.02630746e+00
-1.32070899e-01 -1.11857310e-01 5.02474606e-01 -6.88126683e-01
-1.15868831e+00 8.33407402e-01 -1.85829267e-01 -9.61470068e-01
3.73961926e-01 -4.24917579e-01 -1.06536365e+00 9.58843768e-01
-1.78169024e+00 -1.29084623e+00 -5.40068865e-01 7.00467944e-01
7.54238129e-01 3.79951894e-01 5.02154946e-01 4.17486876e-01
-7.94734955e-01 5.18119395e-01 1.05737604e-01 2.92849302e-01
5.87558031e-01 -1.40027940e+00 3.99018615e-01 7.54921198e-01
4.14561406e-02 4.49946970e-01 1.15782626e-01 -5.19212008e-01
-5.81899107e-01 -1.37860489e+00 5.85954070e-01 -7.38147497e-02
3.64840776e-01 -2.56676555e-01 -1.04173625e+00 6.80372953e-01
3.58563602e-01 4.54204619e-01 4.31551278e-01 1.14815407e-01
-2.15124801e-01 1.79616939e-02 -9.93790030e-01 3.56318384e-01
1.08635044e+00 -1.19122162e-01 -4.41905469e-01 4.17208187e-02
8.87929201e-01 -4.47051197e-01 -6.77686930e-01 4.75020796e-01
3.88330042e-01 -9.50610518e-01 6.47523582e-01 -2.68332422e-01
5.15143573e-01 -5.31647921e-01 -2.78951108e-01 -1.23281324e+00
-5.87098002e-01 -1.76833019e-01 3.79071385e-01 1.29459548e+00
6.18499041e-01 -6.46995366e-01 5.85300624e-01 4.06221271e-01
-5.35059869e-01 -1.02981853e+00 -7.60462463e-01 -3.60454470e-01
1.72808886e-01 -3.51253897e-01 8.55262399e-01 8.22517097e-01
-2.76563764e-01 4.17332798e-01 2.54866015e-02 2.80418515e-01
2.43756428e-01 3.55069280e-01 3.88065726e-01 -1.01481271e+00
-2.31102079e-01 -5.97493887e-01 -2.85335213e-01 -1.29080427e+00
4.23033051e-02 -8.36125076e-01 7.48301819e-02 -1.65127778e+00
2.63058871e-01 -5.58860719e-01 -5.93603492e-01 5.38054824e-01
-5.71360469e-01 2.80490547e-01 2.63839632e-01 5.90971448e-02
-1.05274975e+00 5.46212912e-01 1.54487765e+00 -1.03340179e-01
-3.58522415e-01 -1.90087914e-01 -9.00074601e-01 7.23339796e-01
9.50470984e-01 -1.33169904e-01 -4.08500671e-01 -8.55278730e-01
-4.20834035e-01 -2.63241857e-01 2.68618405e-01 -1.08471596e+00
2.08117083e-01 -9.45368484e-02 7.18376458e-01 -6.31646574e-01
2.33710706e-01 -7.22096503e-01 -2.81156182e-01 2.12472662e-01
-4.26859289e-01 -1.23792119e-01 3.94888967e-01 5.66701591e-01
-4.02545124e-01 -4.62134629e-02 7.20762849e-01 -1.63062662e-01
-1.07075441e+00 4.98102576e-01 -2.25876644e-01 7.01083317e-02
1.20970547e+00 -1.88969243e-02 -3.72410059e-01 -8.36474076e-02
-8.91435862e-01 7.27235258e-01 2.73519546e-01 3.50188643e-01
3.77090961e-01 -1.25933897e+00 -5.65407753e-01 4.06737268e-01
1.69481054e-01 5.42783380e-01 7.78920591e-01 8.01725507e-01
-3.92305076e-01 5.24151802e-01 -3.69670242e-01 -6.99225783e-01
-1.04620683e+00 4.31369513e-01 2.95194328e-01 -2.69177675e-01
-3.22545707e-01 1.12538838e+00 7.96540976e-01 -3.36787790e-01
-1.60010196e-02 -7.82712936e-01 -4.29709375e-01 -5.11674657e-02
4.71608937e-01 1.37117222e-01 -1.53043672e-01 -8.72417569e-01
-3.29850078e-01 9.41166997e-01 -2.75229126e-01 1.19801573e-01
1.17577684e+00 -2.98072338e-01 1.60029411e-01 1.03174314e-01
1.22693503e+00 -1.49915084e-01 -1.63047516e+00 -2.51392484e-01
-3.61340255e-01 -4.79295522e-01 4.76333201e-02 -9.35957372e-01
-1.51834607e+00 1.12543797e+00 5.70623100e-01 4.72422168e-02
1.40582561e+00 2.78414309e-01 9.62489426e-01 -1.38790548e-01
-1.27811581e-01 -1.11226118e+00 -6.24075308e-02 6.25743330e-01
6.38795733e-01 -1.36674893e+00 -3.45632285e-01 -5.49789190e-01
-9.09644663e-01 8.12329352e-01 9.59560096e-01 -4.22373563e-01
7.23178029e-01 1.96802378e-01 5.20636253e-02 -1.81759864e-01
-4.64627266e-01 -7.78005302e-01 3.62995148e-01 4.87619787e-01
5.40733874e-01 2.19488293e-01 -3.78890187e-01 9.63118732e-01
1.54519975e-01 -3.91236424e-01 6.07014112e-02 7.67302394e-01
-5.41509151e-01 -9.81050193e-01 -1.60566404e-01 4.79471743e-01
-4.63565320e-01 -2.15619296e-01 -2.01867059e-01 6.69159651e-01
5.89842975e-01 8.76106322e-01 4.34515446e-01 -3.30082327e-01
3.97576541e-01 -2.26958096e-01 1.49317861e-01 -4.56375301e-01
-6.25408769e-01 4.50717121e-01 -2.81603456e-01 -5.77409267e-01
-3.11237186e-01 -5.82336426e-01 -1.71963477e+00 3.34343836e-02
-3.07309240e-01 9.16749984e-02 4.02834237e-01 8.16699982e-01
5.23241222e-01 8.60134840e-01 4.28387165e-01 -9.04559910e-01
2.52178341e-01 -9.80057359e-01 -4.74998891e-01 5.68865001e-01
3.19095463e-01 -6.55491769e-01 7.80187324e-02 5.45242243e-03]
|
[9.568659782409668, 0.2673945724964142]
|
c9494118-0f37-46d6-bd66-7b54e05e7a95
|
ecg-tcn-wearable-cardiac-arrhythmia-detection
|
2103.1374
| null |
https://arxiv.org/abs/2103.13740v2
|
https://arxiv.org/pdf/2103.13740v2.pdf
|
ECG-TCN: Wearable Cardiac Arrhythmia Detection with a Temporal Convolutional Network
|
Personalized ubiquitous healthcare solutions require energy-efficient wearable platforms that provide an accurate classification of bio-signals while consuming low average power for long-term battery-operated use. Single lead electrocardiogram (ECG) signals provide the ability to detect, classify, and even predict cardiac arrhythmia. In this paper, we propose a novel temporal convolutional network (TCN) that achieves high accuracy while still being feasible for wearable platform use. Experimental results on the ECG5000 dataset show that the TCN has a similar accuracy (94.2%) score as the state-of-the-art (SoA) network while achieving an improvement of 16.5% in the balanced accuracy score. This accurate classification is done with 27 times fewer parameters and 37 times less multiply-accumulate operations. We test our implementation on two publicly available platforms, the STM32L475, which is based on ARM Cortex M4F, and the GreenWaves Technologies GAP8 on the GAPuino board, based on 1+8 RISC-V CV32E40P cores. Measurements show that the GAP8 implementation respects the real-time constraints while consuming 0.10 mJ per inference. With 9.91 GMAC/s/W, it is 23.0 times more energy-efficient and 46.85 times faster than an implementation on the ARM Cortex M4F (0.43 GMAC/s/W). Overall, we obtain 8.1% higher accuracy while consuming 19.6 times less energy and being 35.1 times faster compared to a previous SoA embedded implementation.
|
['Luca Benini', 'Lukas Cavigelli', 'Alessio Burrello', 'Michael Hersche', 'Xiaying Wang', 'Thorir Mar Ingolfsson']
|
2021-03-25
| null | null | null | null |
['arrhythmia-detection']
|
['medical']
|
[ 9.58823636e-02 -2.25645915e-01 -2.31254861e-01 -3.37268889e-01
-3.05015624e-01 -1.30846754e-01 -4.19612467e-01 1.83704615e-01
-5.96581221e-01 6.09582365e-01 -2.87809372e-01 -6.82606816e-01
8.17441866e-02 -7.66947567e-01 -4.42304224e-01 -5.11881530e-01
-4.40789431e-01 -3.00287753e-01 -2.28955805e-01 2.47606099e-01
-4.85622853e-01 1.96145222e-01 -1.19452715e+00 2.89958656e-01
6.58481538e-01 1.75384331e+00 -3.64208579e-01 7.73829758e-01
7.72583961e-01 7.33948410e-01 -6.89875007e-01 -7.01977238e-02
4.42372859e-02 -3.11790466e-01 -4.24406856e-01 -8.47066641e-01
1.15339421e-01 -4.40618277e-01 -2.26374000e-01 4.95141119e-01
1.04489195e+00 -6.22745335e-01 7.43964463e-02 -1.10446608e+00
1.37093529e-01 6.63863897e-01 -3.28676581e-01 4.99603182e-01
7.11609498e-02 1.23565339e-01 2.40178615e-01 -3.22744876e-01
-2.46384814e-02 3.82996112e-01 1.21687734e+00 5.22182703e-01
-1.04883540e+00 -9.97776031e-01 -6.88579917e-01 2.54243016e-01
-1.81016111e+00 -6.61992133e-01 6.17577314e-01 2.46642753e-01
1.65076447e+00 6.02310181e-01 1.15798950e+00 1.18069923e+00
1.02400470e+00 6.42696992e-02 9.41435218e-01 -2.48836711e-01
6.54619634e-01 -3.41599345e-01 3.62636626e-01 5.97334743e-01
8.98617744e-01 -1.48174986e-01 -7.88659334e-01 -5.60884833e-01
4.75284368e-01 1.55530035e-01 -3.14878285e-01 5.09631634e-01
-1.08419037e+00 1.12716414e-01 5.06688893e-01 4.06235605e-01
-5.90114832e-01 1.13780892e+00 6.98747098e-01 -1.95329804e-02
1.13391399e-01 2.52923071e-01 -6.63850784e-01 -5.84776282e-01
-9.31671798e-01 -2.44797319e-02 8.68105054e-01 6.73134983e-01
1.31127119e-01 4.82931852e-01 -1.10475674e-01 1.44743472e-01
2.45103493e-01 7.98938572e-01 6.53603017e-01 -7.41818309e-01
2.26647288e-01 5.64693391e-01 4.34941947e-02 -6.33920074e-01
-1.11139858e+00 -6.95876122e-01 -1.37024987e+00 -1.80847198e-01
1.56287670e-01 -4.83885080e-01 -7.49958754e-01 1.49020970e+00
5.12970006e-03 1.30860835e-01 9.33714677e-03 6.44617260e-01
8.75558138e-01 5.21503448e-01 2.53952712e-01 -2.00855538e-01
1.91532409e+00 -4.19283777e-01 -9.15321767e-01 -2.63063550e-01
4.57725137e-01 -3.04160446e-01 6.25635147e-01 5.51078439e-01
-9.98857975e-01 -7.23534763e-01 -1.69098818e+00 -5.60030900e-02
2.09239990e-01 5.30612707e-01 6.89522862e-01 1.42073321e+00
-1.19607818e+00 6.95221126e-01 -1.39853597e+00 -2.16602802e-01
5.31680584e-01 6.71389282e-01 3.36563975e-01 3.53605896e-01
-1.07526040e+00 6.79465771e-01 3.09964754e-02 1.37132004e-01
-7.06930101e-01 -1.09806657e+00 -4.86597598e-01 4.60410208e-01
-2.97459632e-01 -9.35292542e-01 1.06000376e+00 -7.92032838e-01
-1.58053005e+00 4.40798312e-01 4.82020453e-02 -9.57367539e-01
8.29047114e-02 -1.55588582e-01 -1.05481791e+00 3.04846525e-01
-4.12852943e-01 3.35560858e-01 4.05461490e-01 1.59825943e-02
-2.24992529e-01 -4.62790042e-01 -2.44750574e-01 -3.82630825e-01
-5.81188560e-01 -1.50890648e-01 6.93970323e-02 -3.91660005e-01
-1.45485923e-01 -1.11002386e+00 -7.91388601e-02 2.74207503e-01
1.17570665e-02 1.00765526e-01 6.03346825e-01 -7.11912274e-01
1.38900256e+00 -2.00699401e+00 -7.20523834e-01 4.02167737e-01
3.00936311e-01 3.55701506e-01 4.45336014e-01 -2.37751585e-02
2.60988269e-02 3.05533558e-02 -1.95371404e-01 3.22381668e-02
-3.13614070e-01 9.95238591e-03 2.41612554e-01 7.92547941e-01
-2.21775606e-01 1.14101672e+00 -2.29313210e-01 -5.12193777e-02
6.81580231e-02 9.62615609e-01 -2.82647491e-01 -3.29728216e-01
6.29014909e-01 4.52808589e-02 -3.14815402e-01 7.79108226e-01
5.25134265e-01 -4.09215182e-01 8.93418312e-01 -8.17394257e-01
-3.97998393e-02 4.74624157e-01 -8.94722581e-01 2.13865733e+00
-6.22819722e-01 7.62826860e-01 7.14823157e-02 -6.82643294e-01
9.63243961e-01 6.66639268e-01 5.14469624e-01 -1.04385757e+00
7.43677914e-01 4.29416001e-01 2.87937582e-01 -5.31466603e-01
-6.97564557e-02 1.52384549e-01 -1.26072288e-01 3.44341844e-01
-5.01673520e-02 5.20776272e-01 -3.56799334e-01 -9.97720137e-02
1.69580460e+00 9.99296829e-02 3.03487957e-01 -1.12733138e+00
9.20560956e-03 -3.42322290e-01 9.00625527e-01 4.76196676e-01
-3.20280701e-01 -3.96892019e-02 2.53847055e-02 -9.83770132e-01
-7.57156968e-01 -9.57755983e-01 -4.90620553e-01 1.99770078e-01
-2.93731634e-02 -8.14336061e-01 -8.80977631e-01 9.91437212e-02
-2.44575068e-01 4.40301478e-01 -3.00054818e-01 -3.35564464e-01
-7.64457643e-01 -1.18157494e+00 1.29118133e+00 1.11864722e+00
9.89498496e-01 -5.59727967e-01 -2.28185773e+00 5.52038789e-01
-1.96490780e-01 -9.72576797e-01 -6.38739541e-02 4.06704068e-01
-1.23222125e+00 -7.16093838e-01 -2.74166048e-01 -3.23987126e-01
2.80085593e-01 -4.20930237e-01 1.29912806e+00 1.29102454e-01
-7.93585062e-01 2.03878745e-01 -2.30595116e-02 -7.34020472e-01
2.41502106e-01 1.38205066e-01 4.25401837e-01 -3.15727472e-01
6.04553699e-01 -8.44448805e-01 -1.26415205e+00 -1.26820244e-02
-3.37950617e-01 -4.05278616e-02 4.28738445e-01 4.77443486e-01
2.98016965e-01 -3.13018382e-01 8.00185144e-01 -2.76791751e-01
7.56664202e-02 -2.48393387e-01 -2.72816360e-01 -9.44505110e-02
-1.14385319e+00 -9.99743119e-02 7.74291158e-01 -3.19631755e-01
-7.03173757e-01 4.20263618e-01 -4.20027971e-01 -7.06015676e-02
3.35346818e-01 8.23022425e-02 -2.21931115e-02 -1.39894977e-01
9.05036986e-01 -1.47519574e-01 4.89217341e-02 -2.22163990e-01
-4.75635886e-01 8.45652819e-01 7.23526418e-01 -2.85671324e-01
-1.24124279e-02 4.72750008e-01 1.39965415e-01 -6.24110699e-01
-1.32661196e-03 5.75502738e-02 1.14437053e-02 -1.99951947e-01
9.85927284e-01 -1.49506104e+00 -1.36166358e+00 3.67440164e-01
-8.02001238e-01 -4.24219280e-01 2.74963956e-02 4.93629128e-01
2.64878515e-02 -2.98945233e-02 -7.82175303e-01 -7.88843036e-01
-1.74528146e+00 -7.33735323e-01 7.01675475e-01 3.78465921e-01
-7.42220581e-01 -4.13715184e-01 -3.56802404e-01 1.02493167e-01
9.42161202e-01 6.89355075e-01 4.38468218e-01 -2.10108246e-05
-1.04255937e-01 -1.39658347e-01 2.79578418e-01 1.89176843e-01
-2.45824121e-02 -3.01151931e-01 -1.35961246e+00 -5.71312964e-01
4.27120864e-01 3.43718641e-02 4.99624193e-01 6.69683516e-01
1.53074467e+00 -7.65889734e-02 -6.78069115e-01 7.82301486e-01
1.68010211e+00 4.42369282e-01 9.77333188e-01 -1.53256372e-01
3.39621246e-01 -3.79233360e-01 8.48761275e-02 5.39101660e-01
2.68061131e-01 5.98445058e-01 3.93111914e-01 -1.97019354e-01
1.27649993e-01 3.99044335e-01 3.59530389e-01 8.45845520e-01
-3.10057163e-01 -5.43889515e-02 -1.08137786e+00 3.86142313e-01
-1.45098543e+00 -4.66296494e-01 -5.58681130e-01 2.13331985e+00
7.96317518e-01 1.87546238e-01 -2.12943293e-02 5.97013533e-01
3.33862782e-01 -2.64963895e-01 -7.91535556e-01 -6.68496490e-01
2.77054071e-01 1.14634001e+00 7.82682419e-01 -1.12718619e-01
-7.65806019e-01 -3.19027454e-01 5.31725979e+00 3.68945092e-01
-1.39686179e+00 5.35977006e-01 8.12106431e-01 -6.94135547e-01
3.20529342e-01 -5.36188185e-01 -3.45850885e-01 8.65115166e-01
2.09893489e+00 -2.13935211e-01 1.67103671e-02 8.99966538e-01
5.40046766e-02 -1.61484122e-01 -1.32188368e+00 1.55207992e+00
-3.00273169e-02 -1.37821639e+00 -8.25854599e-01 2.02430710e-02
2.09298521e-01 4.66480032e-02 -4.45793957e-01 8.81333053e-02
-6.71424985e-01 -1.02580142e+00 6.74626112e-01 4.04218942e-01
1.48169279e+00 -1.13675046e+00 9.79060709e-01 7.21566677e-02
-1.50543439e+00 -1.45330086e-01 -1.94098294e-01 -5.89971781e-01
-1.42918035e-01 1.14933729e+00 -4.14227366e-01 4.73347217e-01
1.63294661e+00 3.11700821e-01 -3.95569950e-01 5.86363316e-01
1.39960855e-01 9.80659246e-01 -7.29776502e-01 -3.75672907e-01
-3.73838991e-01 3.35107267e-01 2.45439471e-03 1.15875614e+00
5.50709903e-01 4.56995845e-01 -3.63766521e-01 3.97963524e-01
-2.24409550e-01 -4.75462705e-01 -3.93625908e-02 6.84845626e-01
6.21845305e-01 1.53628671e+00 -6.89458907e-01 -6.05086446e-01
-7.11786151e-02 8.89901817e-01 -4.19731468e-01 -2.59172529e-01
-1.35544634e+00 -7.84571707e-01 6.79791689e-01 1.59673259e-01
-1.53120697e-01 -7.36765712e-02 -7.80300319e-01 -8.60170186e-01
5.12031436e-01 -6.68645799e-01 1.17738126e-02 -5.73789477e-01
-7.23512352e-01 7.64469028e-01 -5.40969670e-01 -9.84201312e-01
3.58836316e-02 -5.47324836e-01 -5.23450136e-01 7.07356274e-01
-1.10387540e+00 -6.43153071e-01 -9.55581486e-01 5.24302065e-01
2.97504906e-02 3.48445594e-01 1.36183631e+00 8.35865021e-01
-5.29616654e-01 9.05221879e-01 -1.65048152e-01 -7.64250755e-02
2.92169303e-01 -8.04559112e-01 6.73486292e-01 8.29404235e-01
-6.03000402e-01 7.05821753e-01 3.60159487e-01 -3.99745017e-01
-2.02098417e+00 -1.29115927e+00 7.81111717e-01 -4.93051596e-02
-2.32999511e-02 -5.39830804e-01 -6.02674961e-01 2.62612432e-01
5.09893477e-01 3.09416294e-01 8.70216012e-01 -2.75428206e-01
2.11662482e-02 -1.02308238e+00 -1.39369535e+00 4.01346952e-01
1.04270875e+00 -2.90385336e-01 3.46821882e-02 2.19384823e-02
2.08156705e-01 -4.84261662e-01 -1.33522356e+00 7.00974882e-01
1.12323749e+00 -6.36289358e-01 7.84513295e-01 2.59734660e-01
-2.75127161e-02 -3.27606767e-01 4.96163741e-02 -8.23633134e-01
-3.45935643e-01 -7.93709397e-01 -4.97330129e-01 8.48200202e-01
1.80666238e-01 -8.42018962e-01 6.03205740e-01 5.80735683e-01
-2.21014827e-01 -1.14447653e+00 -1.43312740e+00 -8.25436592e-01
-4.08928454e-01 -4.41560119e-01 5.21942139e-01 6.43086910e-01
3.24499696e-01 3.62281382e-01 -2.57391334e-01 -1.00524820e-01
6.59578741e-01 -2.04539031e-01 4.65582162e-02 -1.19164324e+00
-1.63990632e-01 1.31965071e-01 -6.55705392e-01 -4.07884955e-01
-6.14325106e-01 -6.38354540e-01 -2.53219813e-01 -1.17160165e+00
7.97317997e-02 -3.92424285e-01 -6.34903073e-01 9.33439553e-01
1.32645786e-01 7.56348252e-01 -2.48168364e-01 -3.31735611e-01
-1.61983773e-01 4.51586545e-02 9.24737230e-02 -1.91292316e-01
-3.92446853e-02 -4.59519863e-01 -4.47108179e-01 6.22541368e-01
1.26915753e+00 -5.63304424e-01 -3.31186384e-01 -4.49050516e-01
3.86508584e-01 5.60329743e-02 4.68307078e-01 -2.05090022e+00
3.40236008e-01 6.55002415e-01 9.80112016e-01 -3.28634530e-01
4.44781125e-01 -8.87430608e-01 8.67362916e-01 1.48484778e+00
3.31066489e-01 5.32223225e-01 6.64860010e-01 7.22954795e-02
5.75244606e-01 4.80629772e-01 7.22780764e-01 3.01543385e-01
-2.91602165e-01 -4.22920138e-02 -7.38785207e-01 -4.18403685e-01
1.09814048e+00 -9.69170034e-02 -5.55407226e-01 2.42857352e-01
-4.20510113e-01 -3.61575224e-02 1.55993253e-01 1.46203265e-01
5.48491120e-01 -1.35402477e+00 -5.47422409e-01 9.82379541e-02
-1.89725891e-01 -4.60702717e-01 6.60140038e-01 9.80631888e-01
-8.43489945e-01 4.42522556e-01 -5.81956565e-01 -8.35606039e-01
-1.35376441e+00 -5.44776805e-02 7.91615129e-01 1.29669448e-02
-1.09923744e+00 4.63629127e-01 -7.33566701e-01 4.71672297e-01
-4.01610471e-02 -8.24147165e-01 2.98015684e-01 -1.96522057e-01
7.08326995e-01 9.70091999e-01 7.61107147e-01 4.18175519e-01
-1.24052012e+00 4.74439979e-01 6.50774240e-01 2.35611588e-01
1.18346202e+00 3.86546552e-01 -1.29321620e-01 2.96710998e-01
1.12296093e+00 -3.27246666e-01 -6.77573740e-01 4.34664309e-01
-2.22494930e-01 1.65951997e-01 4.33888942e-01 -1.19763052e+00
-1.37170172e+00 6.57956123e-01 1.55886197e+00 -5.22385910e-02
1.79619002e+00 -4.84671921e-01 1.22222674e+00 3.18642706e-01
7.63764501e-01 -1.03485155e+00 -3.41952860e-01 -2.23179579e-01
3.57463330e-01 -4.04086411e-01 4.31850225e-01 -1.39043242e-01
4.41299118e-02 1.23130214e+00 5.02673924e-01 -2.24269465e-01
7.32091844e-01 9.45509315e-01 -9.83622000e-02 -1.40045881e-01
-7.67171919e-01 5.83560765e-01 -6.09822311e-02 2.96561509e-01
6.41605616e-01 4.97322232e-01 -7.44284570e-01 9.37685013e-01
-7.75740892e-02 6.28339946e-01 5.96465111e-01 1.22703373e+00
9.75258425e-02 -6.34117961e-01 -1.06700823e-01 8.90055239e-01
-1.06574953e+00 -1.06988825e-01 3.33406866e-01 2.53802180e-01
3.70661855e-01 1.21344852e+00 4.86134797e-01 -5.15227377e-01
2.30234250e-01 1.00781098e-01 5.14126122e-01 -5.32825999e-02
-1.31470978e+00 -2.33319309e-02 3.78614426e-01 -1.09653342e+00
-3.71695906e-01 -4.23840255e-01 -1.42380822e+00 -4.85851586e-01
-7.81150488e-03 -1.85696468e-01 1.23144579e+00 4.37309206e-01
1.10174286e+00 1.42065179e+00 1.77122548e-01 -5.16737044e-01
-2.23722398e-01 -9.60489988e-01 -4.89813983e-01 -5.25362909e-01
1.96509674e-01 -2.26227388e-01 3.35858986e-02 -2.45454777e-02]
|
[13.999740600585938, 3.1368298530578613]
|
78e91f83-908c-4a85-8554-7c6fd33a4880
|
privacy-enabled-biometric-search
|
1708.04726
| null |
http://arxiv.org/abs/1708.04726v1
|
http://arxiv.org/pdf/1708.04726v1.pdf
|
Privacy-Enabled Biometric Search
|
Biometrics have a long-held hope of replacing passwords by establishing a
non-repudiated identity and providing authentication with convenience.
Convenience drives consumers toward biometrics-based access management
solutions. Unlike passwords, biometrics cannot be script-injected; however,
biometric data is considered highly sensitive due to its personal nature and
unique association with users. Biometrics differ from passwords in that
compromised passwords may be reset. Compromised biometrics offer no such
relief. A compromised biometric offers unlimited risk in privacy (anyone can
view the biometric) and authentication (anyone may use the biometric).
Standards such as the Biometric Open Protocol Standard (BOPS) (IEEE 2410-2016)
provide a detailed mechanism to authenticate biometrics based on pre-enrolled
devices and a previous identity by storing the biometric in encrypted form.
This paper describes a biometric-agnostic approach that addresses the privacy
concerns of biometrics through the implementation of BOPS. Specifically, two
novel concepts are introduced. First, a biometric is applied to a neural
network to create a feature vector. This neural network alone can be used for
one-to-one matching (authentication), but would require a search in linear time
for the one-to-many case (identity lookup). The classifying algorithm described
in this paper addresses this concern by producing normalized floating-point
values for each feature vector. This allows authentication lookup to occur in
up to polynomial time, allowing for search in encrypted biometric databases
with speed, accuracy and privacy.
|
['Scott Streit', 'Stephen Suffian', 'Brian Streit']
|
2017-08-16
| null | null | null | null |
['novel-concepts']
|
['reasoning']
|
[ 3.50005537e-01 -1.63132519e-01 -1.64969668e-01 -4.97194320e-01
-4.09288138e-01 -1.11338866e+00 3.07855248e-01 3.00284952e-01
-8.54043245e-01 6.44466579e-01 -3.71002406e-01 -6.41896486e-01
-6.83539137e-02 -9.97027755e-01 -1.94843173e-01 -7.20651269e-01
1.52055144e-01 1.37534693e-01 -2.77563661e-01 -2.43891496e-02
2.30903149e-01 1.11980391e+00 -1.38890684e+00 -1.18393451e-01
3.85648131e-01 1.07208836e+00 -6.83980942e-01 6.24964237e-01
-2.26391833e-02 -4.34365958e-01 -6.58679724e-01 -8.04528415e-01
7.02733636e-01 -3.21193375e-02 -8.70768070e-01 -6.75130486e-01
3.57258677e-01 -7.12235451e-01 1.04380652e-01 7.85261452e-01
5.98812282e-01 -1.85720861e-01 4.80924994e-01 -1.33789384e+00
-3.26788217e-01 -2.16834657e-02 -3.54113042e-01 -5.89952350e-01
8.01044047e-01 4.16345417e-01 6.19100749e-01 -6.59660280e-01
1.11060336e-01 7.88860381e-01 1.02523029e+00 8.82984459e-01
-1.38914275e+00 -5.44244349e-01 -7.58536398e-01 -2.52446055e-01
-1.91527474e+00 -4.74511623e-01 3.15740854e-02 -1.62901714e-01
8.84567082e-01 7.81936288e-01 8.26762319e-01 7.07271576e-01
2.65055329e-01 -1.14539795e-01 1.19392169e+00 -5.25054514e-01
1.49583176e-01 4.52715129e-01 2.55510032e-01 3.88884664e-01
8.03840637e-01 1.64641038e-01 -2.94214755e-01 -6.31716788e-01
8.57142568e-01 1.36609182e-01 -2.42087379e-01 -3.48689705e-02
-8.26414347e-01 6.13258839e-01 -1.28628522e-01 3.45177591e-01
-6.32098019e-01 -1.18380673e-01 3.31629872e-01 2.71668971e-01
-4.15157139e-01 4.52877671e-01 -5.92408240e-01 -2.47752041e-01
-7.32549131e-01 2.10290790e-01 1.39462364e+00 5.36136568e-01
8.04287791e-01 -1.02565929e-01 2.14047611e-01 4.21639532e-01
3.87375295e-01 8.68251026e-01 4.62672234e-01 -4.31902707e-01
-1.82316795e-01 5.13524055e-01 2.77237296e-01 -8.43443155e-01
-2.41511241e-01 3.48020196e-01 -1.00955379e+00 4.28076148e-01
9.10457551e-01 -1.02136962e-01 -6.04955196e-01 1.48045123e+00
3.09303522e-01 -1.71170846e-01 2.45139211e-01 3.32625389e-01
5.10113299e-01 3.58856052e-01 3.90820801e-02 -7.12628439e-02
1.85131192e+00 1.55939519e-01 -6.14996374e-01 3.97067636e-01
-3.20360996e-02 -8.67031336e-01 1.05058193e+00 3.08847368e-01
-1.02437317e+00 -3.12705010e-01 -1.17395234e+00 1.23560958e-01
-9.21380639e-01 -4.54459876e-01 4.64846790e-01 1.87724650e+00
-1.19379270e+00 4.32163477e-01 -5.07817090e-01 -5.48081458e-01
2.18634337e-01 1.24781430e+00 -7.96816885e-01 3.47662389e-01
-1.33794188e+00 1.00966549e+00 5.08939885e-02 2.11984459e-02
2.22063467e-01 -5.51755846e-01 -9.11568701e-01 1.41299754e-01
-2.53691196e-01 -5.40543556e-01 8.50523055e-01 -3.61255437e-01
-1.49471879e+00 1.05291724e+00 -3.52038264e-01 -4.47898090e-01
2.68267304e-01 2.67320424e-01 -4.79236335e-01 1.68614369e-02
-2.00226575e-01 3.73476148e-01 7.29024112e-01 -7.26461828e-01
-1.98515117e-01 -5.30707598e-01 -2.47517135e-03 -2.78337568e-01
-4.59425420e-01 1.59150138e-01 5.68662621e-02 -1.82497025e-01
4.74702977e-02 -7.72485256e-01 6.39178092e-03 -1.04572132e-01
-1.78422645e-01 2.02354521e-01 7.29726911e-01 -6.35785222e-01
9.93093312e-01 -2.02519703e+00 -9.76714492e-01 8.60844195e-01
-7.04111382e-02 6.85861230e-01 1.95183516e-01 5.58261633e-01
-1.27076758e-02 4.57663149e-01 -6.05483465e-02 -1.23237111e-02
8.99178442e-03 1.09933496e-01 -2.14205772e-01 6.18454754e-01
1.07609704e-01 8.75807464e-01 -4.57088292e-01 -2.45280325e-01
2.02565730e-01 8.53448391e-01 -1.27656177e-01 -5.54799177e-02
7.43808150e-01 1.72697574e-01 -1.15077265e-01 9.52115774e-01
1.12571120e+00 1.71515957e-01 3.53640527e-01 -3.07935923e-01
-2.55438983e-01 2.02886492e-01 -1.50888312e+00 8.49220514e-01
-3.76328349e-01 3.07062604e-02 7.97698945e-02 -2.21541196e-01
1.00638330e+00 7.54435301e-01 5.07779241e-01 -1.92100808e-01
3.21565658e-01 4.52586830e-01 -1.71328872e-01 -1.10088497e-01
5.64868927e-01 -2.51463577e-02 -9.76684019e-02 7.84831941e-01
-1.43423930e-01 1.70992985e-01 -3.57385427e-01 -3.87029558e-01
6.99387014e-01 -4.07038331e-02 6.99471414e-01 -2.19857827e-01
9.94254410e-01 -5.79900563e-01 3.28956187e-01 8.36845279e-01
-1.21664219e-01 3.25745910e-01 -7.30084926e-02 -5.62100053e-01
-7.68783033e-01 -9.80854452e-01 -4.80864614e-01 3.50229263e-01
4.03780378e-02 -2.89254606e-01 -9.54735756e-01 -5.02428174e-01
2.86506355e-01 2.15850875e-01 -2.48351768e-01 4.40446027e-02
-3.83558512e-01 -2.77359515e-01 1.26079118e+00 2.20950201e-01
6.23450994e-01 -7.18988478e-01 -8.80577087e-01 1.89610243e-01
3.19917277e-02 -8.24538827e-01 -4.77406949e-01 5.51107712e-02
-7.31297851e-01 -1.02616072e+00 -5.37857771e-01 -4.46994931e-01
7.70809531e-01 3.43636982e-02 3.02564710e-01 3.13542455e-01
-3.98796469e-01 6.57256365e-01 2.21547917e-01 -6.18428946e-01
-4.91373599e-01 8.58736783e-02 5.56152940e-01 2.99314409e-01
9.86363232e-01 -5.88794172e-01 -6.74717724e-01 4.00913745e-01
-8.01957965e-01 -5.34925640e-01 3.54151487e-01 9.24449027e-01
2.54155695e-01 -3.24479938e-01 5.97632408e-01 -5.74400425e-01
7.49251366e-01 5.40205985e-02 -6.38794959e-01 4.41129208e-01
-1.15162981e+00 -1.51528552e-01 5.32713056e-01 -4.26660687e-01
-6.54924691e-01 4.43419605e-01 -3.94781321e-01 2.27872580e-01
-4.33347940e-01 2.43225813e-01 -4.91623193e-01 -9.70308125e-01
4.88142312e-01 4.31603879e-01 4.94076341e-01 -4.07504708e-01
1.20307833e-01 1.11057293e+00 7.44806409e-01 -6.20107174e-01
9.68220770e-01 3.82630318e-01 2.40151092e-01 -7.82666028e-01
2.51102954e-01 -5.68100154e-01 -9.14192080e-01 -1.11379683e-01
4.25286382e-01 -3.67657304e-01 -1.85866487e+00 8.85193884e-01
-1.19741285e+00 6.03426933e-01 -4.28209871e-01 3.27791393e-01
-2.01507226e-01 5.81002891e-01 -4.70099777e-01 -1.25367832e+00
-7.71398425e-01 -1.01410282e+00 6.06923044e-01 6.86967134e-01
-7.14661658e-01 -7.74985373e-01 -2.39504158e-01 5.46774492e-02
8.58919382e-01 3.23466003e-01 4.40450579e-01 -9.98395860e-01
-3.26789379e-01 -1.17392325e+00 -8.52892697e-02 3.04836988e-01
6.09769881e-01 -8.82127509e-02 -1.28498673e+00 -3.43646228e-01
2.89463133e-01 4.79898043e-02 -6.13527149e-02 -1.04437303e-02
5.88221431e-01 -7.25726902e-01 4.82423976e-02 6.92958593e-01
1.58520389e+00 4.30862516e-01 1.03826666e+00 2.75359124e-01
2.45097950e-01 6.04775131e-01 2.50600666e-01 4.52948093e-01
2.35581562e-01 5.01608193e-01 7.45958742e-03 -1.32071719e-01
5.94567001e-01 -7.57913142e-02 1.30543299e-02 -2.72996817e-02
-3.55798215e-01 3.42229158e-01 -7.74683952e-01 2.60965228e-01
-1.11353958e+00 -9.80713069e-01 -1.68761536e-01 2.92497182e+00
1.15243649e+00 -3.36830825e-01 2.54218787e-01 5.25950134e-01
6.08951330e-01 -4.89221305e-01 -2.96719015e-01 -8.36907506e-01
-1.81958247e-02 1.04112923e+00 9.97630596e-01 6.24259591e-01
-7.42779195e-01 7.14508295e-01 6.20156384e+00 3.55190575e-01
-1.42666817e+00 -4.37367409e-02 3.88329178e-01 3.29955637e-01
-1.12559043e-01 1.25956222e-01 -9.96678948e-01 2.32397139e-01
9.87608850e-01 -4.17471558e-01 3.87929559e-01 5.07202983e-01
4.06828895e-02 -2.31827140e-01 -1.04413903e+00 1.11421919e+00
-2.70137548e-01 -1.24767089e+00 7.08972141e-02 6.47629976e-01
1.00319706e-01 -7.33795702e-01 9.92917940e-02 -2.65319049e-01
-3.35579008e-01 -1.14230824e+00 4.83997017e-02 6.12594128e-01
1.26419461e+00 -8.21241021e-01 9.18731391e-01 -1.33862197e-01
-1.18016577e+00 2.61008620e-01 -1.92555457e-01 -1.16229309e-02
-2.29824334e-02 -7.22716972e-02 -9.73562539e-01 6.00727797e-01
3.78518075e-01 -4.30616707e-01 -3.35472345e-01 9.53950524e-01
2.54180789e-01 3.12841058e-01 -7.64942646e-01 -1.52952448e-01
-1.32073596e-01 -3.88734877e-01 1.48650512e-01 1.02188003e+00
3.76610190e-01 2.05203697e-01 -3.65243763e-01 7.24580467e-01
2.78724700e-01 3.87085050e-01 -6.37362957e-01 -1.22653052e-01
7.46449113e-01 1.24893153e+00 -5.90116382e-01 -1.60426889e-02
-2.46008456e-01 1.06867909e+00 -7.17363954e-01 7.06742182e-02
-3.43392342e-01 -9.51714694e-01 6.56795144e-01 2.03480646e-01
-8.03921446e-02 -1.67243704e-01 -7.11254358e-01 -8.03813398e-01
-1.04736490e-02 -1.10192525e+00 2.55793780e-01 -1.26975417e-01
-1.16053903e+00 4.28658724e-01 -2.50202090e-01 -1.06758916e+00
-4.91133869e-01 -5.11394024e-01 -3.46166134e-01 1.83584690e+00
-9.57202196e-01 -1.51126397e+00 -2.19724059e-01 8.79282832e-01
-6.32657588e-01 -1.94826469e-01 1.85938549e+00 2.45489076e-01
1.34280279e-01 1.28616524e+00 -1.50915429e-01 4.06441867e-01
7.26923108e-01 -9.39662993e-01 7.44477451e-01 6.20388269e-01
-2.61957645e-01 1.75878906e+00 3.30800980e-01 -5.67173064e-01
-1.77731073e+00 -3.32672715e-01 1.33901858e+00 -2.55512416e-01
1.78308129e-01 -2.71529764e-01 -8.58980417e-01 3.88010532e-01
-2.66127158e-02 -2.09556937e-01 1.23564172e+00 -3.78070846e-02
-6.06845260e-01 -3.02675128e-01 -2.01825261e+00 5.06728292e-01
1.27680719e-01 -9.56779301e-01 -1.60866067e-01 -2.69709259e-01
3.85695882e-02 -2.44510353e-01 -1.21524751e+00 1.16264038e-01
1.55566669e+00 -7.68903077e-01 1.04674864e+00 -3.76012892e-01
-5.98112583e-01 -2.90921390e-01 -8.55124071e-02 -2.81607419e-01
2.18628570e-01 -1.14231050e+00 1.87543094e-01 1.46333981e+00
2.84143388e-01 -1.44428515e+00 8.05168808e-01 1.42820346e+00
1.02721310e+00 -5.62259495e-01 -9.83907402e-01 -9.59822237e-01
-2.99708873e-01 -1.13016576e-01 1.29907322e+00 1.02102470e+00
1.96663007e-01 -1.55071974e-01 -4.15810555e-01 3.52251619e-01
7.06349015e-01 -3.80809486e-01 9.40747738e-01 -1.46665537e+00
-1.42323241e-01 -1.89952940e-01 -8.91542971e-01 -5.13686717e-01
-3.62836897e-01 -3.83168876e-01 -3.83073926e-01 -8.03405643e-01
-2.68533289e-01 -6.79086030e-01 -2.38799691e-01 8.14372122e-01
2.87997067e-01 5.95695496e-01 -4.56413953e-04 7.48464763e-02
7.13273644e-01 -4.84565556e-01 4.51264858e-01 -1.50071774e-02
-4.34768736e-01 5.19327700e-01 -7.53661931e-01 4.46347147e-01
9.59232390e-01 -3.94416898e-01 -7.01585412e-02 4.65133101e-01
1.17587499e-01 1.61521479e-01 4.18475062e-01 -9.11925972e-01
3.87783170e-01 -2.02825397e-01 3.73377502e-01 -3.56882840e-01
4.81567740e-01 -1.13140059e+00 5.45369267e-01 7.93045104e-01
1.92096800e-01 6.04205579e-02 2.94278532e-01 8.34024996e-02
1.66800156e-01 -6.03673577e-01 7.56553769e-01 1.73553213e-01
-4.15120386e-02 3.78564805e-01 -5.14426172e-01 -7.30776489e-01
7.94260085e-01 -9.91760075e-01 -1.29960924e-01 -3.71314675e-01
-7.03329086e-01 -3.56134385e-01 7.90297508e-01 -1.03486955e-01
7.01888978e-01 -9.90460157e-01 -2.94421583e-01 8.41682076e-01
-6.07006736e-02 -4.95397538e-01 1.68687552e-02 5.22135496e-01
-6.83550954e-01 3.54750216e-01 -4.92222339e-01 -3.33756775e-01
-1.81794167e+00 1.92380860e-01 5.28002739e-01 1.24722481e-01
-8.37652683e-02 5.76162994e-01 -6.01354599e-01 -3.99311453e-01
1.64323553e-01 2.95302957e-01 -1.50147500e-02 -3.16173621e-02
1.05704510e+00 1.78489730e-01 1.86906919e-01 -5.74767113e-01
-6.20304525e-01 5.14151335e-01 -1.46370575e-01 -5.35079777e-01
1.05646360e+00 -8.32910836e-03 -5.36103904e-01 1.69754833e-01
1.09890628e+00 5.16631901e-01 -4.25536692e-01 4.74173948e-02
-2.62823775e-02 -7.47907460e-01 -5.58315217e-01 -8.13287199e-01
-6.12878025e-01 5.91620743e-01 9.06990647e-01 4.50512558e-01
9.46804106e-01 -7.74549901e-01 8.63612771e-01 4.89567190e-01
5.70041299e-01 -7.10644662e-01 -9.12642777e-01 3.97476107e-01
3.29162359e-01 -8.96112680e-01 8.18248466e-02 -2.53713876e-01
-2.87199259e-01 1.49136114e+00 2.66379535e-01 3.68345380e-01
7.92868316e-01 4.83231455e-01 2.71136820e-01 5.65302260e-02
1.19717151e-01 1.52539030e-01 3.09565991e-01 1.13020813e+00
3.91244769e-01 -2.93932054e-02 -8.08631897e-01 7.93289781e-01
-3.82342577e-01 1.98236391e-01 5.07400036e-01 1.24682188e+00
7.42111951e-02 -1.89697897e+00 -6.73049569e-01 2.95951694e-01
-7.31636107e-01 -1.71610534e-01 -5.50749898e-01 5.48567235e-01
2.36695930e-02 9.32519019e-01 -1.58422142e-01 -4.54320699e-01
2.19911188e-01 5.71785390e-01 5.77474475e-01 -8.81197155e-02
-1.27096808e+00 -2.64112025e-01 -8.32337812e-02 -3.92783433e-01
-2.51717389e-01 -7.35460877e-01 -1.19762552e+00 -8.96103442e-01
-5.23942113e-01 2.79021323e-01 1.33956313e+00 7.52973735e-01
3.44835252e-01 -4.25176471e-01 7.79262006e-01 -3.87233287e-01
-8.58607113e-01 -4.60262060e-01 -7.57963657e-01 4.58551018e-04
2.19675869e-01 6.51573092e-02 -7.51517564e-02 1.78903297e-01]
|
[13.12362289428711, 1.2147568464279175]
|
96fdb2e2-5c79-4c2a-9ddf-3b6d4e5574a2
|
neural-parameter-calibration-for-large-scale
|
2209.13565
| null |
https://arxiv.org/abs/2209.13565v3
|
https://arxiv.org/pdf/2209.13565v3.pdf
|
Neural parameter calibration for large-scale multi-agent models
|
Computational models have become a powerful tool in the quantitative sciences to understand the behaviour of complex systems that evolve in time. However, they often contain a potentially large number of free parameters whose values cannot be obtained from theory but need to be inferred from data. This is especially the case for models in the social sciences, economics, or computational epidemiology. Yet many current parameter estimation methods are mathematically involved and computationally slow to run. In this paper we present a computationally simple and fast method to retrieve accurate probability densities for model parameters using neural differential equations. We present a pipeline comprising multi-agent models acting as forward solvers for systems of ordinary or stochastic differential equations, and a neural network to then extract parameters from the data generated by the model. The two combined create a powerful tool that can quickly estimate densities on model parameters, even for very large systems. We demonstrate the method on synthetic time series data of the SIR model of the spread of infection, and perform an in-depth analysis of the Harris-Wilson model of economic activity on a network, representing a non-convex problem. For the latter, we apply our method both to synthetic data and to data of economic activity across Greater London. We find that our method calibrates the model orders of magnitude more accurately than a previous study of the same dataset using classical techniques, while running between 195 and 390 times faster.
|
['Mark Girolami', 'Grigorios A. Pavliotis', 'Thomas Gaskin']
|
2022-09-27
| null | null | null | null |
['epidemiology']
|
['medical']
|
[-2.45704744e-02 -1.16389796e-01 2.08214894e-01 1.67192891e-01
-4.19004947e-01 -5.44457018e-01 6.95298672e-01 8.23414996e-02
-7.44426191e-01 1.00075769e+00 -3.10591180e-02 -6.60926700e-01
-4.68663931e-01 -8.31543863e-01 -4.57612544e-01 -7.98575222e-01
-5.20490885e-01 1.08103561e+00 4.58343215e-02 -3.45478743e-01
5.39287105e-02 4.48741585e-01 -1.06256270e+00 -4.82477456e-01
8.02286565e-01 4.68818456e-01 -7.81704783e-02 8.34902287e-01
1.85691699e-01 5.81457973e-01 -4.95077580e-01 -2.31657445e-01
1.99558049e-01 -3.98611367e-01 -5.11352122e-01 -1.51014656e-01
-4.26292777e-01 -1.96642116e-01 -2.74698377e-01 6.76615775e-01
4.51146454e-01 -3.57413031e-02 1.21289194e+00 -1.29611897e+00
-2.48909116e-01 4.18630511e-01 -5.92916071e-01 4.82217789e-01
1.20450526e-01 3.33343834e-01 5.82841098e-01 -3.50210637e-01
5.08236945e-01 1.38482845e+00 1.07896757e+00 1.19331136e-01
-1.65605366e+00 -4.64098364e-01 -1.71568722e-01 -3.16774487e-01
-1.49322987e+00 -1.54629946e-01 2.74183452e-01 -6.65743351e-01
1.02082014e+00 -1.27629249e-03 9.47946906e-01 9.91656125e-01
3.00917804e-01 3.10443163e-01 9.16030407e-01 -1.10519856e-01
4.07819837e-01 3.28680053e-02 -2.18985200e-01 6.13137960e-01
2.85362035e-01 3.22620094e-01 -3.45306396e-02 -7.84306347e-01
1.04875231e+00 1.25651121e-01 2.41083950e-02 -9.32304859e-02
-1.10976124e+00 1.17004180e+00 9.13140178e-02 6.73441067e-02
-6.53465927e-01 3.39798093e-01 1.36191115e-01 2.01031253e-01
9.50287104e-01 3.96113664e-01 -5.62122881e-01 -2.90864527e-01
-8.34768951e-01 6.95959926e-01 1.36874080e+00 3.67002726e-01
4.80316043e-01 -7.79303536e-02 3.04951608e-01 7.02990174e-01
3.16015542e-01 7.60209382e-01 -7.13180285e-03 -1.23764265e+00
-2.91997958e-02 4.23378855e-01 4.34628606e-01 -1.02504683e+00
-6.90322399e-01 -3.37579966e-01 -1.06694925e+00 5.08665703e-02
9.72368240e-01 -6.34354293e-01 -6.27538145e-01 1.70451450e+00
4.52931643e-01 3.70621115e-01 -2.11515307e-01 5.25102079e-01
1.55243063e-02 1.03461170e+00 -2.59560570e-02 -4.04511988e-01
1.16242313e+00 -3.89800966e-01 -3.61969113e-01 -1.36438295e-01
9.12384510e-01 -5.14548898e-01 3.70686412e-01 2.12450922e-01
-1.24353409e+00 3.25657606e-01 -3.59721065e-01 5.06124556e-01
-5.18483400e-01 -3.22934866e-01 7.19819784e-01 4.42327678e-01
-1.24187386e+00 7.40487695e-01 -1.10564625e+00 -4.90991592e-01
4.16068256e-01 4.95720685e-01 1.18411273e-01 1.64112166e-01
-1.23886466e+00 1.11179376e+00 3.31974179e-02 4.24475446e-02
-5.55468619e-01 -1.10764825e+00 -7.42520094e-01 3.62416536e-01
3.76173884e-01 -7.82179356e-01 1.22850764e+00 -9.22753394e-01
-1.16606724e+00 4.48884010e-01 7.58136809e-02 -6.66039705e-01
7.46278107e-01 3.80474240e-01 4.19912003e-02 -1.19861268e-01
-3.43019888e-02 2.62443870e-01 3.42285067e-01 -9.37570453e-01
-2.48955786e-01 -8.93089697e-02 -1.65213913e-01 7.65911490e-02
6.76777139e-02 2.81810492e-01 -6.41798005e-02 -5.65307379e-01
-5.40196836e-01 -9.86640453e-01 -6.16609812e-01 1.07648104e-01
-4.46450152e-02 -1.34872660e-01 4.13821280e-01 -1.00352633e+00
1.06083024e+00 -1.70728028e+00 2.52146512e-01 4.82192665e-01
2.30276704e-01 1.10625699e-01 9.30160470e-03 7.59416580e-01
2.08897412e-01 2.13382080e-01 -8.29168439e-01 -2.55555093e-01
1.62701383e-02 3.23576033e-01 1.11744709e-01 6.19611084e-01
3.42428476e-01 7.82202899e-01 -9.43607330e-01 -4.44214433e-01
-3.03648561e-02 7.84027576e-01 -7.67658234e-01 -1.48554891e-01
-1.75212979e-01 7.17343464e-02 -2.99027443e-01 2.50110865e-01
3.85978699e-01 -5.31882584e-01 3.25949848e-01 5.36639154e-01
-1.95745025e-02 -1.16228415e-02 -1.31614292e+00 9.31934953e-01
-5.93192041e-01 4.50173706e-01 4.03574705e-01 -1.28870523e+00
3.78108680e-01 2.33358532e-01 8.08841050e-01 -2.85701692e-01
3.53477478e-01 2.60485142e-01 4.58270311e-01 -2.41328746e-01
1.18847579e-01 -2.92603761e-01 -7.11440891e-02 9.53803539e-01
-8.40211734e-02 -4.70576853e-01 5.26463866e-01 1.59190208e-01
1.37360966e+00 -9.23139453e-02 2.39897817e-01 -5.93876839e-01
1.69439405e-01 2.50211388e-01 2.81828880e-01 6.44411981e-01
1.28637120e-01 2.16430724e-01 9.55525815e-01 -5.62311590e-01
-1.44408131e+00 -8.33689392e-01 -1.95239767e-01 6.32152736e-01
-3.95610482e-01 -1.27714291e-01 -9.68112707e-01 9.90375802e-02
2.91709065e-01 4.04808849e-01 -8.15399408e-01 -9.88020375e-02
-4.69523668e-01 -1.60319090e+00 4.98492301e-01 2.89517250e-02
1.34825259e-01 -9.44287479e-01 -4.89254445e-01 3.51808935e-01
1.15338981e-01 -8.58222604e-01 -2.71902591e-01 -7.42405429e-02
-5.60950100e-01 -1.09136260e+00 -9.84861016e-01 -4.74523991e-01
7.79234052e-01 -3.82691205e-01 1.11300206e+00 3.54900181e-01
-6.26507580e-01 3.33325565e-01 3.73575479e-01 -5.94952643e-01
-8.59264135e-01 1.78075150e-01 3.35797966e-01 -1.36158586e-01
1.37531504e-01 -5.33071995e-01 -5.32888949e-01 3.15536290e-01
-1.02791905e+00 -5.53395897e-02 3.01916778e-01 7.96260715e-01
6.38043582e-02 1.15811706e-01 7.61429310e-01 -6.97508812e-01
1.10118389e+00 -1.06088150e+00 -1.11511707e+00 7.12104961e-02
-4.18913156e-01 2.17606395e-01 6.80961311e-01 -6.47730172e-01
-7.73127854e-01 4.18736935e-02 1.36881664e-01 9.92968306e-02
8.93960148e-02 9.15725052e-01 6.95099056e-01 5.97809069e-02
4.49340463e-01 -7.20589682e-02 5.44255555e-01 -3.11731160e-01
8.36520176e-03 5.24608612e-01 8.92391056e-02 -6.05982602e-01
6.55269504e-01 4.67548460e-01 4.14778084e-01 -9.36659455e-01
-1.94510594e-01 -7.68718049e-02 -4.46749896e-01 -9.14983302e-02
3.44108969e-01 -7.04717398e-01 -9.13904667e-01 8.28776598e-01
-1.03310335e+00 -1.03172112e+00 -2.43976429e-01 4.67402190e-01
-5.83989918e-01 -9.97300819e-02 -7.57087052e-01 -1.08442557e+00
9.30139124e-02 -8.29446852e-01 7.46664047e-01 7.27818683e-02
-2.77255893e-01 -1.73097003e+00 6.33699954e-01 -1.76108792e-01
9.21056569e-01 3.36390257e-01 9.59716916e-01 -4.73560393e-01
-3.17940474e-01 -2.28598908e-01 -1.81022987e-01 -1.15774505e-01
-5.16653480e-03 5.17279267e-01 -3.85513216e-01 -1.67687193e-01
-6.76683560e-02 8.85095745e-02 6.67512655e-01 8.84659708e-01
8.22204232e-01 -5.15434325e-01 -5.20098567e-01 5.85792422e-01
1.29213178e+00 6.43506348e-02 3.19198608e-01 9.02983695e-02
2.58869112e-01 7.21969485e-01 1.09273292e-01 6.23180509e-01
5.38832784e-01 4.69700247e-01 9.84973162e-02 -2.69782275e-01
6.64686084e-01 1.36231914e-01 2.94350564e-01 5.93192637e-01
-2.88712442e-01 -3.16597223e-01 -1.43244791e+00 6.70290768e-01
-1.91020298e+00 -1.06876147e+00 -1.54972285e-01 2.12694979e+00
1.03909373e+00 2.72933897e-02 7.13930488e-01 -2.10342184e-01
6.13959074e-01 -3.48840892e-01 -6.01897418e-01 -4.86457586e-01
-2.61061992e-02 2.79934347e-01 7.85455346e-01 8.35998654e-01
-6.44935369e-01 5.46737134e-01 7.50820112e+00 5.91035008e-01
-7.38145292e-01 -4.76769507e-02 9.18285131e-01 -5.37679970e-01
5.47826178e-02 -3.46591808e-02 -3.64372522e-01 5.22530973e-01
1.52258503e+00 -5.73584139e-01 9.11033094e-01 1.82752326e-01
5.87607682e-01 -5.24602771e-01 -6.69951022e-01 5.12672067e-01
-3.34360182e-01 -1.32163906e+00 -5.33063233e-01 5.54684103e-01
7.77644873e-01 2.79185742e-01 -6.13591373e-02 4.19346169e-02
9.91571546e-01 -1.22475147e+00 1.41060755e-01 6.33614659e-01
4.34649706e-01 -8.69650364e-01 8.00761521e-01 7.09892750e-01
-7.43050098e-01 5.98088354e-02 -1.43270880e-01 -5.05639672e-01
4.57675874e-01 6.27705753e-01 -8.63833666e-01 -1.23810709e-01
5.63374639e-01 6.43003583e-01 -2.64243782e-01 1.12615693e+00
4.29057509e-01 7.44432211e-01 -1.01268971e+00 -3.54187578e-01
1.63896874e-01 -5.59688866e-01 3.34691465e-01 1.03169000e+00
4.92197156e-01 1.50220513e-01 -2.81503439e-01 1.16409934e+00
2.14773059e-01 -2.13929817e-01 -7.43228018e-01 -2.16593593e-01
2.16025829e-01 1.26134241e+00 -9.13584590e-01 -1.98646560e-01
-3.42340499e-01 3.77937496e-01 1.98359519e-01 6.61112964e-01
-8.34519506e-01 -2.34040260e-01 6.69869423e-01 3.53991270e-01
1.37344003e-01 -5.56870520e-01 2.44327962e-01 -9.69416320e-01
-3.42354536e-01 -7.75294960e-01 1.77901298e-01 -5.54953277e-01
-1.34000695e+00 9.23141092e-02 4.32281256e-01 -4.25785780e-01
-7.75958240e-01 -8.24352622e-01 -5.56656539e-01 1.06024218e+00
-1.08864331e+00 -4.26682800e-01 2.50136197e-01 2.43554980e-01
8.92293081e-03 1.13778412e-01 6.28993034e-01 2.01097593e-01
-8.30693960e-01 -2.79976308e-01 6.78837180e-01 5.29596023e-03
2.37483695e-01 -1.12105978e+00 7.86186099e-01 5.35871089e-01
-2.68283933e-01 4.95255530e-01 7.21528232e-01 -7.53796518e-01
-1.23325598e+00 -5.69754422e-01 6.73373640e-01 -4.89034534e-01
1.20102000e+00 -5.62482476e-01 -8.47330272e-01 6.14265800e-01
9.07416120e-02 -1.56552091e-01 4.49516922e-01 -9.02260616e-02
2.35224396e-01 4.06796992e-01 -1.18121433e+00 5.44068813e-01
7.83118665e-01 -2.43910179e-01 -5.51300682e-02 5.02181113e-01
2.60692149e-01 -3.17841694e-02 -9.37300086e-01 -1.74835965e-01
5.90982020e-01 -6.55251026e-01 1.05825961e+00 -5.65151393e-01
4.03094709e-01 -7.24507147e-04 3.05382818e-01 -1.71355033e+00
-1.39339745e-01 -9.66631711e-01 -1.91167295e-01 9.58743393e-01
5.66873670e-01 -1.07483339e+00 3.99148315e-01 8.17921281e-01
7.01111913e-01 -7.73494124e-01 -9.54970539e-01 -7.37165809e-01
4.59799856e-01 -1.35067865e-01 5.37920356e-01 9.68835473e-01
-1.17080323e-01 6.66198134e-02 -9.89264995e-02 -2.45550517e-02
7.61567831e-01 -5.41811526e-01 5.61968327e-01 -1.59289622e+00
-3.01010191e-01 -7.17030585e-01 -2.02480748e-01 -4.81996089e-01
3.40575546e-01 -5.04362643e-01 -4.10828814e-02 -1.49947643e+00
3.36633652e-01 -6.03391647e-01 2.59060502e-01 2.15315193e-01
-6.86282888e-02 1.59163121e-02 -2.08099753e-01 9.17825326e-02
-5.10245413e-02 4.15908754e-01 8.12534213e-01 1.60623714e-01
-2.58575201e-01 6.06268011e-02 -4.64472681e-01 9.33946133e-01
9.07991886e-01 -7.11694062e-01 -2.49396607e-01 -1.81011423e-01
6.63171291e-01 2.08885252e-01 8.23054135e-01 -5.64156115e-01
1.15569718e-01 -5.07821560e-01 1.37258366e-01 -1.34813622e-01
2.52276182e-01 -7.46155024e-01 4.03685659e-01 7.59341955e-01
-6.08204938e-02 4.74334449e-01 4.33114022e-01 4.33459103e-01
2.52543211e-01 -1.26477376e-01 5.85619390e-01 -2.08355263e-01
1.06124975e-01 3.08787078e-01 -7.83292592e-01 3.51324677e-01
8.37344348e-01 2.15040207e-01 -4.83774304e-01 -8.20533812e-01
-5.70361435e-01 5.11870623e-01 6.65778875e-01 -3.84541266e-02
1.50368899e-01 -8.52271199e-01 -1.01560104e+00 1.11642398e-01
-6.47043228e-01 9.50064063e-02 -1.64459981e-02 1.09375632e+00
-9.28736627e-01 1.63852528e-01 -2.41193175e-02 -4.49988574e-01
-7.73312449e-01 5.20399988e-01 8.66826057e-01 -4.66384321e-01
-1.81526288e-01 3.55844975e-01 1.48108704e-02 -5.81412435e-01
-2.58034050e-01 -4.07059312e-01 1.59117877e-01 1.45257890e-01
3.44480276e-01 5.39133132e-01 -5.62977612e-01 -6.71759844e-01
-3.06270838e-01 5.06425083e-01 2.42797688e-01 -5.24843216e-01
1.85578859e+00 -7.52770454e-02 -2.23495543e-01 4.65091676e-01
1.01867366e+00 -3.81609738e-01 -1.47243917e+00 -8.52078944e-02
-1.12692639e-01 6.92229867e-02 4.24722023e-02 -6.20459735e-01
-9.62362945e-01 6.78909421e-01 1.21786058e-01 9.63153303e-01
8.32726419e-01 4.54643965e-02 3.96282762e-01 2.76915997e-01
9.29201022e-02 -1.02186084e+00 -4.69027251e-01 3.12047541e-01
7.17836857e-01 -1.03340101e+00 2.55562186e-01 -1.51888400e-01
-2.38982707e-01 7.10995197e-01 4.65138331e-02 -4.24387604e-01
1.01925564e+00 6.37686670e-01 -2.69145250e-01 -2.58395404e-01
-9.52695727e-01 5.93420491e-02 -8.42515081e-02 6.85988605e-01
1.58524796e-01 -1.25939742e-01 -1.00082019e-02 1.84162974e-01
-1.60967574e-01 1.50560915e-01 8.41593027e-01 7.89855123e-01
-9.63740237e-03 -9.58220422e-01 -3.60748947e-01 7.23347306e-01
-5.70634186e-01 -1.96525216e-01 -3.66342336e-01 1.03286517e+00
-4.13310200e-01 5.86684465e-01 4.51211214e-01 3.67728919e-01
7.60698691e-03 8.67133308e-03 3.24927449e-01 -2.97612906e-01
-6.10763729e-01 -2.05419865e-02 4.64302152e-02 -1.99444190e-01
-5.18848598e-01 -1.13588643e+00 -9.76046085e-01 -9.40317690e-01
-2.15759620e-01 6.45433217e-02 6.72770202e-01 1.00198638e+00
3.10251683e-01 3.81082386e-01 4.16097522e-01 -1.20890057e+00
-4.87681001e-01 -8.95364761e-01 -7.44833887e-01 3.74111976e-03
4.57798719e-01 -6.16620779e-01 -8.67776394e-01 -1.27594233e-01]
|
[6.064451217651367, 4.335069179534912]
|
aab1645b-b1ba-441a-aadd-f519272e6997
|
syntax-representation-in-word-embeddings-and
|
2010.01063
| null |
https://arxiv.org/abs/2010.01063v1
|
https://arxiv.org/pdf/2010.01063v1.pdf
|
Syntax Representation in Word Embeddings and Neural Networks -- A Survey
|
Neural networks trained on natural language processing tasks capture syntax even though it is not provided as a supervision signal. This indicates that syntactic analysis is essential to the understating of language in artificial intelligence systems. This overview paper covers approaches of evaluating the amount of syntactic information included in the representations of words for different neural network architectures. We mainly summarize re-search on English monolingual data on language modeling tasks and multilingual data for neural machine translation systems and multilingual language models. We describe which pre-trained models and representations of language are best suited for transfer to syntactic tasks.
|
['David Mareček', 'Tomasz Limisiewicz']
|
2020-10-02
| null | null | null | null |
['syntax-representation']
|
['natural-language-processing']
|
[-2.12881733e-02 3.25644672e-01 -7.62268901e-01 -7.48226702e-01
-4.68938291e-01 -6.70427620e-01 7.66557753e-01 6.33822978e-02
-8.65337610e-01 7.79695332e-01 5.91677606e-01 -1.03219688e+00
2.99015135e-01 -5.44904411e-01 -7.10811675e-01 -5.75614609e-02
1.25730708e-01 6.82657778e-01 -4.60641503e-01 -5.28387845e-01
-1.17936358e-01 3.80061477e-01 -6.82863057e-01 7.59716868e-01
6.33702815e-01 3.65558058e-01 6.04513168e-01 4.51103836e-01
-6.89511895e-01 1.15961564e+00 -5.14089167e-01 -3.93103093e-01
-1.07655652e-01 -3.52956921e-01 -9.48835313e-01 -4.58566427e-01
5.67296326e-01 -1.65121853e-02 -2.80609220e-01 9.76480782e-01
1.98110893e-01 -3.29575717e-01 6.13391042e-01 -6.35206282e-01
-1.26824117e+00 1.31148458e+00 7.77122006e-02 7.03729033e-01
2.44186834e-01 -1.12847254e-01 1.22745812e+00 -1.18230808e+00
1.21525192e+00 1.55537164e+00 5.59711516e-01 5.39815843e-01
-1.31413889e+00 -3.43411088e-01 3.10413003e-01 -2.49571763e-02
-1.01119721e+00 -6.94893599e-01 6.76316261e-01 -5.08581221e-01
1.79150033e+00 -3.04944307e-01 2.80811876e-01 1.26865160e+00
6.87323391e-01 8.49247158e-01 1.01645637e+00 -1.00618196e+00
-4.19019818e-01 3.69043767e-01 6.42189622e-01 8.45684707e-01
4.01948184e-01 2.27188200e-01 -4.70559925e-01 1.19877577e-01
6.21149361e-01 -5.22199273e-01 -3.05198878e-02 -1.41950652e-01
-1.32189584e+00 1.25076210e+00 5.58463335e-01 8.81145954e-01
-4.25387055e-01 1.55452847e-01 8.59684348e-01 7.41829634e-01
5.44270694e-01 7.27522373e-01 -1.03960502e+00 3.69655937e-01
-6.83046401e-01 -6.19362481e-02 6.91737413e-01 8.49654973e-01
6.14140511e-01 7.72576928e-01 1.81196600e-01 1.17585850e+00
2.82676369e-01 7.63117373e-01 8.80911589e-01 -5.52115202e-01
7.15178013e-01 3.71287346e-01 -3.57004553e-01 -6.56813383e-01
-6.52858078e-01 -3.46259415e-01 -4.27187741e-01 -3.61076415e-01
4.03215975e-01 -3.19467574e-01 -6.78253174e-01 1.87370861e+00
-4.40594941e-01 -9.03425574e-01 7.04105258e-01 5.03911793e-01
1.00308836e+00 9.27094817e-01 5.70826948e-01 -4.78650257e-02
1.54267073e+00 -9.96623874e-01 -8.07648242e-01 -9.20250297e-01
1.13158703e+00 -8.61742198e-01 8.84080887e-01 2.49673445e-02
-1.15916467e+00 -5.60555041e-01 -9.34114277e-01 -4.83512014e-01
-8.51208270e-01 5.22748172e-01 6.55833662e-01 4.43477243e-01
-1.27302408e+00 3.07556123e-01 -6.80266798e-01 -6.02029026e-01
8.94178376e-02 3.17790359e-01 -5.82935750e-01 -2.31520459e-01
-1.54828656e+00 1.68537354e+00 7.95840800e-01 2.74635199e-03
-5.18149078e-01 -3.63223135e-01 -1.43942654e+00 -1.63779944e-01
-3.63402605e-01 -4.30518389e-01 1.47251427e+00 -1.43401897e+00
-9.54110324e-01 1.35023296e+00 -4.16269720e-01 -7.34247744e-01
-3.21345180e-01 -3.11289001e-02 -5.45997560e-01 -8.60316530e-02
1.94813088e-01 9.42304790e-01 3.14023554e-01 -1.09289157e+00
-4.49442059e-01 -3.95337313e-01 -1.00762643e-01 1.98365241e-01
-2.31844634e-01 7.75638461e-01 1.21472679e-01 -7.78200746e-01
-2.04331443e-01 -7.05992699e-01 -2.04734519e-01 -4.28663433e-01
-1.16617866e-01 -5.83306313e-01 3.31803769e-01 -1.18999922e+00
1.01127505e+00 -1.71440339e+00 2.86492109e-01 -2.21619785e-01
-2.08574444e-01 2.65917003e-01 -5.76284766e-01 3.88230294e-01
-2.95382947e-01 3.55843782e-01 9.64066908e-02 -2.66411841e-01
1.39126070e-02 7.67321527e-01 -5.08418739e-01 2.67756641e-01
4.96997863e-01 1.35068822e+00 -6.94818377e-01 -1.55966654e-01
1.71688020e-01 5.59071958e-01 -1.90882802e-01 -3.67529690e-02
-2.07466334e-01 1.26964018e-01 -2.86089748e-01 5.08194089e-01
2.17281386e-01 1.53387174e-01 4.24171209e-01 -7.44058490e-02
-2.34624520e-01 1.25783527e+00 -3.29405814e-01 1.88046503e+00
-8.04732203e-01 9.63836133e-01 1.60149083e-01 -1.04023349e+00
8.92426491e-01 7.34549463e-01 -3.00926626e-01 -1.01408851e+00
4.46373373e-01 6.04825795e-01 4.86733139e-01 -1.83792815e-01
4.95459586e-01 -4.24339354e-01 -3.18256468e-01 6.05157673e-01
7.79575825e-01 1.09078035e-01 2.57526219e-01 -1.30481645e-01
6.09275460e-01 1.58890083e-01 5.39838195e-01 -8.57979536e-01
5.39630234e-01 3.90250862e-01 3.40874135e-01 4.74024951e-01
2.53539085e-02 -4.58489768e-02 2.24630445e-01 -9.41421747e-01
-1.28576577e+00 -6.09251201e-01 -4.50046957e-01 1.69925272e+00
-8.09065461e-01 -2.66542017e-01 -6.65471554e-01 -5.27882636e-01
-2.97819376e-01 1.20840406e+00 -5.39186239e-01 -1.31693613e-02
-9.90560174e-01 -6.15503132e-01 9.34597135e-01 6.55517936e-01
-2.28337809e-01 -1.69174826e+00 -2.55256951e-01 3.28861713e-01
-5.90864047e-02 -1.19143808e+00 -1.12142742e-01 8.60449374e-01
-9.41152036e-01 -6.50570393e-01 -5.72522819e-01 -1.54507077e+00
5.87649405e-01 -5.69336861e-02 1.48818398e+00 -3.34649719e-03
2.33192742e-01 -3.06817871e-02 -7.42117316e-02 -6.34088933e-01
-8.18693280e-01 4.69862401e-01 1.37235448e-01 -7.58037925e-01
9.52559590e-01 -2.83901691e-01 1.70021266e-01 -2.35592753e-01
-6.68998897e-01 -1.91310093e-01 6.11092627e-01 9.66882706e-01
1.58303127e-01 -6.41805768e-01 5.01404762e-01 -9.47272539e-01
9.51001167e-01 -6.01211786e-01 -4.53881085e-01 2.16173053e-01
-4.08987105e-01 5.47318578e-01 8.39554429e-01 -4.17319357e-01
-1.17811894e+00 -1.33343831e-01 -1.48581535e-01 8.98919180e-02
-4.48371500e-01 8.86774123e-01 5.94842657e-02 1.32028237e-01
9.13591206e-01 6.04238175e-02 -1.49930954e-01 -6.23572648e-01
6.37610734e-01 4.22474891e-01 3.22552264e-01 -8.60392988e-01
3.74640524e-01 -1.53380617e-01 -4.77142125e-01 -9.63402271e-01
-6.91485584e-01 -1.51496097e-01 -9.57734942e-01 4.10657138e-01
1.03063643e+00 -1.01871383e+00 2.54102647e-01 2.06820816e-02
-1.88815582e+00 -1.88297942e-01 -1.05687246e-01 6.80601895e-01
-3.09182674e-01 -2.10500151e-01 -1.08100629e+00 -2.91440934e-01
-4.97633308e-01 -1.18745911e+00 7.73735046e-01 -3.50990027e-01
-4.88999665e-01 -1.59454310e+00 4.12515312e-01 5.65027408e-02
5.53724051e-01 -3.83413941e-01 1.47360945e+00 -1.09434474e+00
4.21473458e-02 -1.87318981e-01 -2.02718690e-01 3.50308448e-01
-9.60773602e-02 -3.05502713e-01 -9.74271715e-01 -1.21323235e-01
-7.60426149e-02 -6.73188984e-01 7.81819344e-01 4.94171143e-01
2.09937692e-01 -4.21197444e-01 -7.78256506e-02 3.26636612e-01
1.32980263e+00 2.77856737e-01 1.55182004e-01 5.14081657e-01
6.19578362e-01 1.11232388e+00 -2.12113187e-01 -5.42949736e-01
5.35594583e-01 3.45340371e-01 -2.31795862e-01 -2.54812121e-01
-1.44376114e-01 -3.01281810e-01 8.86781394e-01 1.47460938e+00
2.57097423e-01 -1.60404399e-01 -1.34760261e+00 7.93359935e-01
-1.54544151e+00 -4.76351857e-01 -4.87104841e-02 1.53145647e+00
1.05200803e+00 1.10955136e-02 -4.45119649e-01 -4.91561621e-01
6.01308405e-01 2.30651796e-01 -1.22228526e-01 -1.29067743e+00
-3.67693335e-01 5.94644904e-01 5.16504645e-01 7.23754883e-01
-9.66198444e-01 1.65168667e+00 7.53527498e+00 4.48008448e-01
-1.16757596e+00 3.08926821e-01 4.03817177e-01 4.67380345e-01
-3.07731420e-01 2.76352875e-02 -8.67449820e-01 -1.72436625e-01
1.81412733e+00 -1.85629860e-01 3.20331335e-01 1.00878930e+00
2.29221016e-01 2.97855616e-01 -1.51063514e+00 5.04510343e-01
3.26640904e-01 -1.27777839e+00 5.02549171e-01 -1.66709661e-01
5.39461434e-01 8.18109751e-01 -2.62151420e-01 6.66143656e-01
7.56291211e-01 -1.26276350e+00 7.01687157e-01 2.18906865e-01
6.03682399e-01 -5.65096259e-01 9.73292470e-01 3.60862136e-01
-9.02478397e-01 2.80656666e-01 -7.32122600e-01 -2.98655301e-01
2.89439231e-01 8.34480748e-02 -7.62227058e-01 2.91832477e-01
2.39969164e-01 7.34412432e-01 -7.98852146e-01 1.72966063e-01
-6.26368344e-01 6.83538735e-01 -1.37965366e-01 -1.68484405e-01
6.29108012e-01 1.50334872e-02 3.38367581e-01 1.69621670e+00
1.07133575e-01 -4.60488617e-01 3.75596344e-01 8.71207356e-01
-2.37395719e-01 7.95203507e-01 -1.32913601e+00 -3.04410458e-01
1.81421727e-01 8.07669520e-01 -3.77898246e-01 -4.06839550e-01
-7.99857616e-01 6.08001292e-01 7.86856830e-01 3.73867333e-01
-1.20916598e-01 -2.24099591e-01 4.13163900e-01 -1.32737696e-01
-2.09547319e-02 -5.51623344e-01 -3.90537918e-01 -1.42771471e+00
-2.39145279e-01 -1.04714096e+00 1.47527948e-01 -1.11819959e+00
-1.33762562e+00 1.11863005e+00 -1.03987634e-01 -4.30574387e-01
-8.44056129e-01 -1.37648499e+00 -3.46922457e-01 1.26057267e+00
-1.65133941e+00 -1.44315767e+00 7.83504605e-01 4.72209096e-01
7.89238036e-01 -7.35611379e-01 1.47312367e+00 1.96120322e-01
-3.85785341e-01 4.47859585e-01 4.70010154e-02 7.61799037e-01
5.84318757e-01 -9.40715909e-01 5.97916126e-01 7.47347116e-01
5.62884867e-01 1.11604345e+00 4.01549369e-01 -3.91066790e-01
-1.22767758e+00 -9.90405381e-01 1.80774605e+00 -6.51245475e-01
1.04834819e+00 -2.89688975e-01 -9.80158150e-01 1.29122996e+00
1.23556495e+00 -3.74060422e-01 7.19524741e-01 3.44620109e-01
-5.73185444e-01 2.66061276e-01 -6.18986607e-01 6.00609899e-01
4.60431546e-01 -9.61075604e-01 -1.39422774e+00 3.30847025e-01
8.75909209e-01 1.44487441e-01 -6.90486968e-01 2.53052920e-01
2.29444936e-01 -1.66624188e-01 7.68943012e-01 -1.26266313e+00
6.03586972e-01 1.24124229e-01 -3.67170066e-01 -1.43498826e+00
-3.35088491e-01 -1.05183654e-01 3.35636646e-01 9.13918793e-01
1.11991394e+00 -6.00426853e-01 1.39285907e-01 4.17580187e-01
-1.82031050e-01 -2.33597964e-01 -9.02412474e-01 -5.70995390e-01
1.05002725e+00 -5.56406856e-01 -3.27068344e-02 1.29222143e+00
1.94450870e-01 1.14256406e+00 3.70312817e-02 -3.16930592e-01
2.62541115e-01 -2.96972483e-01 2.24437684e-01 -1.17825854e+00
4.08904105e-02 -7.59025395e-01 -2.65682936e-01 -7.57406473e-01
9.74858701e-01 -1.67823327e+00 -3.15059423e-02 -1.53337109e+00
1.11828350e-01 6.18963316e-02 -2.40418911e-01 7.95209825e-01
3.64275128e-01 3.76539975e-02 7.83744976e-02 1.97820649e-01
-1.21999234e-01 3.63413870e-01 6.84278548e-01 -2.23570138e-01
6.30484745e-02 -6.37192130e-01 -6.61935270e-01 1.06509829e+00
1.01360846e+00 -5.57384312e-01 -1.14284486e-01 -1.35041833e+00
3.00688177e-01 1.20885335e-02 -1.12986714e-01 -5.17394841e-01
-1.20056249e-01 -1.32870495e-01 3.64649445e-01 -3.34621251e-01
-6.01327457e-02 -7.68451452e-01 -5.36213994e-01 4.54871953e-01
-6.55062556e-01 7.89121866e-01 5.44188499e-01 -9.03256610e-02
-4.98684585e-01 -4.87382114e-01 6.20857954e-01 -6.08564973e-01
-7.63585210e-01 -2.68013537e-01 -1.06570458e+00 9.39988568e-02
2.91894466e-01 2.66912967e-01 -3.51804972e-01 -9.66132954e-02
-9.48498547e-01 1.00681126e-01 1.68322697e-01 8.41127098e-01
2.90670455e-01 -1.43942022e+00 -7.26712644e-01 2.35933423e-01
1.72318101e-01 -5.90377748e-01 -5.87762773e-01 4.43311930e-01
-7.28504837e-01 1.27893615e+00 -3.77458841e-01 -2.65196472e-01
-8.75511050e-01 5.84601641e-01 3.33738118e-01 -5.85421979e-01
-3.54499280e-01 5.63567758e-01 1.68633386e-01 -1.14197493e+00
1.39674172e-01 -5.90619981e-01 -6.31963074e-01 2.27876790e-02
3.64294827e-01 -3.40630233e-01 2.38288909e-01 -1.16060257e+00
-3.36090714e-01 3.25405955e-01 -3.90362203e-01 -2.38950461e-01
1.32088518e+00 4.02992330e-02 -5.50978959e-01 8.97689164e-01
1.26143813e+00 -2.52007484e-01 -9.32226926e-02 -6.64538205e-01
6.77130282e-01 4.88170266e-01 1.62974462e-01 -8.41252744e-01
-7.54203677e-01 1.34229469e+00 2.45786056e-01 -4.06660199e-01
5.19502163e-01 8.46079513e-02 5.83668709e-01 1.16236055e+00
3.80954742e-01 -1.20025492e+00 -4.83422846e-01 1.46351385e+00
9.96603549e-01 -1.24085927e+00 -3.59931916e-01 3.46991606e-02
-5.99298656e-01 1.37660575e+00 5.08895934e-01 -4.61730272e-01
8.02157581e-01 4.33743358e-01 4.69240218e-01 -2.37865537e-01
-1.00193691e+00 -1.91861968e-02 3.07355493e-01 3.78310770e-01
1.13376355e+00 1.71823338e-01 -7.03956008e-01 7.95190692e-01
-4.60868299e-01 -3.67980480e-01 3.72958452e-01 1.00678527e+00
-4.60476458e-01 -1.39389157e+00 -2.74444550e-01 1.25351146e-01
-8.09944212e-01 -9.14007902e-01 -7.04866469e-01 9.02092159e-01
6.47802055e-02 6.72783554e-01 1.53316438e-01 6.19946681e-02
2.36392140e-01 8.85166824e-01 2.74911195e-01 -1.03692794e+00
-8.02903950e-01 5.95288500e-02 5.99170864e-01 -2.79198945e-01
-5.44999361e-01 -3.73903215e-01 -1.15297079e+00 -3.16006206e-02
-9.65981558e-02 2.68344998e-01 8.75942349e-01 1.01584673e+00
9.48855057e-02 4.75249857e-01 -4.05463353e-02 -8.62463117e-01
-2.33115226e-01 -1.35188305e+00 -1.45797774e-01 1.12174079e-01
1.38825119e-01 -2.66197562e-01 -1.62647426e-01 3.23806375e-01]
|
[11.000838279724121, 9.785962104797363]
|
90031198-f047-4b8d-98b1-5f6132ad7a74
|
translate-to-adapt-rgb-d-scene-recognition
|
2103.14672
| null |
https://arxiv.org/abs/2103.14672v2
|
https://arxiv.org/pdf/2103.14672v2.pdf
|
Multi-Modal RGB-D Scene Recognition Across Domains
|
Scene recognition is one of the basic problems in computer vision research with extensive applications in robotics. When available, depth images provide helpful geometric cues that complement the RGB texture information and help to identify discriminative scene image features. Depth sensing technology developed fast in the last years and a great variety of 3D cameras have been introduced, each with different acquisition properties. However, those properties are often neglected when targeting big data collections, so multi-modal images are gathered disregarding their original nature. In this work, we put under the spotlight the existence of a possibly severe domain shift issue within multi-modality scene recognition datasets. As a consequence, a scene classification model trained on one camera may not generalize on data from a different camera, only providing a low recognition performance. Starting from the well-known SUN RGB-D dataset, we designed an experimental testbed to study this problem and we use it to benchmark the performance of existing methods. Finally, we introduce a novel adaptive scene recognition approach that leverages self-supervised translation between modalities. Indeed, learning to go from RGB to depth and vice-versa is an unsupervised procedure that can be trained jointly on data of multiple cameras and may help to bridge the gap among the extracted feature distributions. Our experimental results confirm the effectiveness of the proposed approach.
|
['Tatiana Tommasi', 'Silvia Bucci', 'Andrea Ferreri']
|
2021-03-26
| null | null | null | null |
['scene-recognition']
|
['computer-vision']
|
[ 6.88819051e-01 -3.52541596e-01 -1.85802966e-01 -4.48403120e-01
-4.91464913e-01 -6.73539579e-01 7.14180291e-01 5.81643954e-02
-4.78710502e-01 4.24016148e-01 -1.50185466e-01 2.85571311e-02
-3.30960602e-01 -7.22479880e-01 -4.89226699e-01 -1.06213140e+00
4.70159888e-01 2.91613609e-01 3.55724961e-01 -4.20081541e-02
2.80647010e-01 6.38486862e-01 -1.94680607e+00 1.59091219e-01
6.54437542e-01 1.25609291e+00 5.32599866e-01 3.29424053e-01
-1.11909024e-01 7.11109042e-01 -1.57320425e-01 -1.62217960e-01
5.15637577e-01 -2.53002733e-01 -5.88469982e-01 5.83292067e-01
5.93998015e-01 -2.28332967e-01 -4.00156587e-01 1.20338404e+00
2.75667936e-01 -6.21496998e-02 3.84141445e-01 -1.21078467e+00
-2.31967032e-01 -4.88799401e-02 -5.12889445e-01 1.87812358e-01
5.19554734e-01 5.61396107e-02 8.32390904e-01 -7.16252804e-01
6.17611170e-01 8.94396544e-01 2.65740961e-01 4.97198880e-01
-1.07776868e+00 -3.20778601e-02 3.56207229e-02 3.18662882e-01
-1.21043098e+00 -3.87446344e-01 1.21741009e+00 -3.68721992e-01
6.36449873e-01 1.81717366e-01 6.81652248e-01 1.15470469e+00
-1.50678769e-01 8.70416641e-01 1.59948242e+00 -4.76783901e-01
3.32678139e-01 2.69378006e-01 -5.54341376e-02 4.71265197e-01
3.71882707e-01 1.06127389e-01 -7.37864196e-01 2.52351433e-01
6.31543398e-01 2.83760697e-01 -3.58016372e-01 -9.07658815e-01
-1.27327502e+00 6.25790954e-01 4.30842221e-01 3.74803185e-01
-2.91012496e-01 -2.46656582e-01 1.91388667e-01 3.05422992e-01
6.34100959e-02 1.29573464e-01 -2.07499251e-01 -2.00074539e-01
-6.39846921e-01 -9.44428965e-02 5.89030325e-01 7.87430406e-01
1.12144768e+00 -2.49388590e-01 4.65542167e-01 7.23573923e-01
2.16482684e-01 6.96824074e-01 5.83884418e-01 -7.34395623e-01
3.59266251e-01 8.29859614e-01 -1.39465809e-01 -1.14088547e+00
-4.85226929e-01 -9.42001045e-02 -9.71378982e-01 1.00760095e-01
7.93735147e-01 3.48926902e-01 -7.17950642e-01 1.28442621e+00
4.99899268e-01 -5.39302193e-02 9.87899378e-02 1.15913796e+00
5.44754684e-01 1.78958803e-01 -3.61246675e-01 1.84169598e-02
1.04043877e+00 -6.40292764e-01 -2.72259921e-01 -4.07052994e-01
4.45977658e-01 -7.13188708e-01 1.05230105e+00 6.20757878e-01
-4.71137166e-01 -4.58859950e-01 -1.01097655e+00 1.51486978e-01
-6.04818881e-01 -4.25122790e-02 8.12532961e-01 8.04600716e-01
-7.64630914e-01 4.08781886e-01 -7.52641559e-01 -8.40985000e-01
2.83428043e-01 1.35712370e-01 -7.36341655e-01 -6.70958042e-01
-6.03839636e-01 8.22115242e-01 5.18488705e-01 1.54039577e-01
-6.20003939e-01 -2.29027718e-01 -7.55975008e-01 -4.93337214e-01
5.13117611e-01 -4.73877251e-01 8.21616650e-01 -1.04995155e+00
-1.59618378e+00 1.11292875e+00 1.79842394e-02 -1.58494249e-01
5.06558001e-01 -1.19149700e-01 -3.16801906e-01 4.10604089e-01
9.29295551e-04 3.61757964e-01 8.76253426e-01 -1.35994148e+00
-7.44316459e-01 -7.34772146e-01 3.55225176e-01 3.18556070e-01
-3.73459816e-01 -3.64228725e-01 -3.50297093e-01 -1.95650548e-01
5.69737971e-01 -9.71700370e-01 -3.02273650e-02 -2.81841252e-02
-4.03835177e-01 2.00533852e-01 8.76239777e-01 -6.26745671e-02
5.52144706e-01 -2.20064497e+00 4.63373929e-01 2.69804746e-01
-1.52671039e-02 -1.22927010e-01 1.20363832e-01 4.71363872e-01
1.47996143e-01 -3.92256558e-01 -4.86505657e-01 -2.65335441e-01
-2.33245537e-01 4.41046476e-01 -2.07442582e-01 8.43086541e-01
1.20005682e-01 5.63730955e-01 -9.53735888e-01 -3.39019805e-01
6.75067902e-01 3.98458481e-01 -3.48807514e-01 2.39003330e-01
-1.65200904e-02 8.69096935e-01 -4.57075417e-01 9.44443762e-01
7.00677335e-01 -2.59747207e-01 1.44147366e-01 -3.06491017e-01
-1.03525937e-01 -2.60552187e-02 -1.25370562e+00 2.04245305e+00
-4.13025677e-01 5.14277160e-01 -6.78775087e-02 -1.49011636e+00
9.74633992e-01 -1.70097262e-01 7.66804338e-01 -9.45470393e-01
1.78066283e-01 4.27029341e-01 -1.14689954e-01 -6.42227590e-01
5.46259224e-01 -3.30333561e-01 -1.75328955e-01 2.38100290e-01
1.22408226e-01 -4.43408519e-01 6.33321330e-02 -2.90441483e-01
9.92535949e-01 1.61420479e-01 3.26450706e-01 5.15449792e-02
6.44000053e-01 1.71604097e-01 3.07470411e-01 6.91521227e-01
-2.54096627e-01 7.82602370e-01 8.89221579e-02 -3.31307292e-01
-9.01914477e-01 -9.58190680e-01 -3.28741580e-01 6.20821953e-01
6.04837596e-01 -4.49827909e-02 -3.77055794e-01 -5.79782784e-01
5.82319982e-02 1.85657531e-01 -4.96825576e-01 -1.33827895e-01
-2.34681740e-01 -7.15556204e-01 3.71684879e-01 2.14980334e-01
7.84072936e-01 -7.22710550e-01 -1.20385945e+00 -9.83718485e-02
-6.71085119e-02 -1.50714803e+00 3.05556059e-01 4.00449634e-01
-9.83476877e-01 -1.37572157e+00 -6.58713698e-01 -5.07293224e-01
6.23592138e-01 9.46251810e-01 8.64186168e-01 -2.24346653e-01
-2.59214371e-01 9.36683834e-01 -6.27630472e-01 -1.54390320e-01
-1.69961020e-01 2.43497603e-02 3.01708311e-01 5.24001777e-01
4.74867582e-01 -7.06163287e-01 -5.94916463e-01 3.15602869e-01
-1.24591351e+00 5.38300648e-02 7.27910399e-01 8.11499536e-01
5.17128348e-01 -1.53049305e-02 2.44925439e-01 -7.47088313e-01
-8.60695541e-03 -4.24909472e-01 -6.29361808e-01 2.62982488e-01
-3.35540473e-01 2.03085970e-02 5.61523080e-01 -2.73063123e-01
-1.01589775e+00 4.14355129e-01 1.30387947e-01 -4.78473395e-01
-6.08488917e-01 6.24704540e-01 -4.12354231e-01 -3.14500690e-01
5.40024817e-01 4.68306631e-01 1.41371727e-01 -5.08571267e-01
3.95515382e-01 7.00664878e-01 6.11598015e-01 -5.15345991e-01
9.39190507e-01 1.01625574e+00 1.67669907e-01 -1.14914536e+00
-8.01976383e-01 -8.23182642e-01 -1.04376531e+00 -4.18154687e-01
6.14243269e-01 -8.57146978e-01 -4.07186002e-01 7.30812788e-01
-8.57596815e-01 -1.95550755e-01 -2.50240207e-01 5.13865173e-01
-6.38681650e-01 5.43785036e-01 -9.30464417e-02 -7.40609825e-01
3.65358740e-01 -1.04191875e+00 1.18244827e+00 4.13863957e-01
2.93905735e-01 -9.88362849e-01 3.00819483e-02 4.45021719e-01
2.17035517e-01 2.99444914e-01 5.82906306e-01 -4.23280746e-01
-7.36596882e-01 -2.65077114e-01 -2.85298914e-01 3.70393068e-01
4.83187020e-01 -2.62619942e-01 -1.29544675e+00 -1.90407991e-01
2.47177303e-01 -3.32376301e-01 8.73917222e-01 8.49978328e-02
9.67595100e-01 3.32219452e-01 -6.54499680e-02 7.19784379e-01
1.77036238e+00 -6.65543005e-02 5.98261416e-01 7.51380801e-01
9.01966810e-01 7.74922311e-01 6.81850791e-01 4.98365939e-01
4.21047330e-01 7.48373210e-01 6.49450243e-01 1.55152183e-03
5.49917966e-02 -1.46075517e-01 3.37163925e-01 7.28864551e-01
-8.53359178e-02 1.54964969e-01 -9.84362841e-01 3.42125118e-01
-1.75434053e+00 -7.73839176e-01 7.45559260e-02 2.44188499e+00
5.66866994e-01 5.60564250e-02 8.51100013e-02 5.48446774e-01
3.89768749e-01 2.01325029e-01 -6.12426519e-01 1.62903607e-01
-6.05597675e-01 -3.17324474e-02 6.48635328e-01 1.03890024e-01
-1.18602741e+00 7.00958490e-01 5.31876993e+00 4.72534627e-01
-1.53712165e+00 -1.35776758e-01 2.10522830e-01 1.52230740e-01
-1.26054525e-01 9.76337492e-02 -4.56465423e-01 2.69862235e-01
4.97379035e-01 -6.41094297e-02 4.22491491e-01 9.67695594e-01
-7.86308348e-02 -6.34449422e-01 -1.19329214e+00 1.47262144e+00
3.72257978e-01 -9.16900873e-01 -5.97994514e-02 1.59600273e-01
6.24971330e-01 1.24762870e-01 2.61970386e-02 -1.23065323e-01
-1.47650272e-01 -6.36860788e-01 7.08712578e-01 4.93230373e-01
5.58419108e-01 -3.89907181e-01 5.15361309e-01 4.20178205e-01
-9.92091596e-01 -2.23541230e-01 -4.70220298e-01 -2.31510043e-01
3.78422551e-02 8.42194200e-01 -7.12438285e-01 9.11390066e-01
7.81051219e-01 1.12851906e+00 -7.94526577e-01 1.17072332e+00
-1.33156389e-01 1.05137341e-01 -4.94737655e-01 2.10305884e-01
1.74648054e-02 -4.01090860e-01 4.96022075e-01 7.81555057e-01
4.58321393e-01 -1.89732760e-01 1.26330227e-01 6.54154480e-01
2.11201370e-01 -1.99277624e-02 -1.20231390e+00 1.23342931e-01
2.01058716e-01 1.32836401e+00 -9.65534747e-01 -1.00828573e-01
-7.91095197e-01 1.11515546e+00 8.90162885e-02 2.85309553e-01
-5.54930508e-01 5.72650954e-02 5.33124566e-01 -1.37230590e-01
2.88840175e-01 -5.54094076e-01 -3.45113873e-01 -1.58263290e+00
1.85440660e-01 -6.69165969e-01 2.56440639e-01 -6.47113860e-01
-1.41181707e+00 4.06223118e-01 3.61583233e-02 -1.77071166e+00
-2.05055639e-01 -9.39706504e-01 -3.72314826e-02 3.86687279e-01
-1.97883558e+00 -1.05233192e+00 -7.09297836e-01 9.55184519e-01
4.63042974e-01 -5.25897555e-02 6.97330117e-01 3.49819064e-01
-5.03294885e-01 2.33667269e-01 3.63489479e-01 -4.60263155e-02
6.05476320e-01 -1.08185112e+00 -3.72186989e-01 9.83991027e-01
3.41257423e-01 4.03135955e-01 5.03434360e-01 -2.69055694e-01
-2.02783632e+00 -8.24119866e-01 2.93131471e-01 -4.63808775e-01
6.22743785e-01 -2.37664655e-01 -7.86224723e-01 3.26687157e-01
-1.57836288e-01 1.95026398e-01 5.91767728e-01 -7.81307891e-02
-5.06429255e-01 -3.71468335e-01 -9.88041580e-01 2.49927849e-01
9.32163954e-01 -8.37805927e-01 -4.66705829e-01 1.49168998e-01
1.36994332e-01 -4.30872768e-01 -7.99716473e-01 4.48352188e-01
4.78269428e-01 -1.45227921e+00 9.51901078e-01 -2.28679236e-02
4.05776709e-01 -3.91043007e-01 -7.02629566e-01 -1.12921822e+00
3.23824495e-01 -5.59879467e-03 1.36431754e-01 1.15203309e+00
-5.93804792e-02 -7.34828532e-01 8.02534580e-01 5.88163555e-01
5.83661138e-04 -2.46937186e-01 -1.11112344e+00 -9.50437546e-01
-1.89109445e-01 -6.46304607e-01 2.91942060e-01 9.85952675e-01
-1.23180486e-01 1.18585475e-01 -3.11187208e-01 1.83686897e-01
8.47690642e-01 5.33938229e-01 1.01238918e+00 -1.35259938e+00
-2.25601450e-01 -4.37184483e-01 -8.22464466e-01 -1.19406581e+00
-1.04311872e-02 -7.67320991e-01 1.31252527e-01 -1.11028159e+00
2.25945711e-01 -6.08456373e-01 -2.49292955e-01 2.33414263e-01
3.31469066e-02 4.29841399e-01 3.20134878e-01 3.92336756e-01
-5.68733394e-01 5.86127937e-01 1.01473129e+00 -2.51099110e-01
-4.70495932e-02 -4.38505113e-02 -4.40358847e-01 6.84909225e-01
6.21439040e-01 -1.45121351e-01 -4.01436925e-01 -4.53266323e-01
1.70339733e-01 -1.52610436e-01 6.31018341e-01 -1.21312571e+00
3.80009204e-01 -3.05128843e-01 1.33069769e-01 -2.84596831e-01
4.94047523e-01 -1.18077803e+00 4.54863831e-02 1.31247714e-01
9.57428589e-02 -3.15157652e-01 1.21911699e-02 7.53906786e-01
-5.51863790e-01 -2.37856414e-02 6.77245915e-01 -1.80395633e-01
-1.36207080e+00 1.32321283e-01 -1.57395706e-01 -2.52207428e-01
9.28845286e-01 -6.35743499e-01 -1.75461143e-01 -1.38560727e-01
-3.24255526e-01 -2.67065883e-01 9.28163648e-01 5.74396729e-01
7.65125036e-01 -1.31782877e+00 -2.04125226e-01 3.54998797e-01
7.00515032e-01 2.81189084e-01 2.81632602e-01 1.03307557e+00
-4.56053376e-01 4.53632116e-01 -4.29274440e-01 -1.17491877e+00
-1.07334173e+00 4.96938676e-01 2.76115119e-01 1.18284062e-01
-6.03652358e-01 5.02356946e-01 1.42379075e-01 -4.61955994e-01
1.01619102e-01 -3.69187117e-01 -7.74110109e-02 1.86701328e-01
3.38739872e-01 8.67984369e-02 2.57401764e-01 -9.04069304e-01
-4.79345232e-01 9.26408887e-01 3.35231066e-01 -7.12356996e-03
1.46662199e+00 -5.80160677e-01 -9.37216207e-02 8.90702724e-01
1.14213908e+00 -1.84759676e-01 -1.29261637e+00 -5.54131448e-01
2.33439863e-01 -8.74796689e-01 -1.21178403e-02 -4.03005987e-01
-1.12749398e+00 9.29671049e-01 6.41207695e-01 3.84886444e-01
1.44065487e+00 1.11209668e-01 3.95363510e-01 4.80782092e-01
9.43375349e-01 -1.07896030e+00 1.77156761e-01 4.24218774e-01
5.01889110e-01 -1.79269779e+00 -1.46216163e-02 -3.82106692e-01
-6.77042663e-01 1.29445064e+00 3.46514970e-01 -1.80684275e-03
4.26849753e-01 -6.11876324e-02 1.10188834e-01 -2.03257814e-01
-2.40869015e-01 -6.74276650e-01 6.05962984e-02 7.52388537e-01
3.08705084e-02 -9.52600017e-02 1.84687331e-01 2.15950996e-01
7.14697465e-02 2.23578438e-02 5.87396860e-01 1.13738704e+00
-3.75006825e-01 -1.30223680e+00 -5.18984675e-01 2.02341333e-01
4.66355996e-04 2.23422945e-01 -4.60650355e-01 8.15365911e-01
-3.81289646e-02 1.02456975e+00 -2.31021851e-01 -6.31369174e-01
3.70517582e-01 -1.26025230e-02 8.21956992e-01 -4.37719941e-01
-8.65077041e-03 -8.86769295e-02 -2.35132560e-01 -6.85713053e-01
-1.10666132e+00 -9.81552243e-01 -8.19748998e-01 -1.07188523e-01
-1.57600075e-01 -3.65877539e-01 8.93280029e-01 1.12886536e+00
9.04538929e-02 1.86568916e-01 9.61456299e-01 -1.18392920e+00
-2.65333802e-01 -6.18508518e-01 -8.01165402e-01 6.69238210e-01
4.59543973e-01 -9.72767830e-01 -4.07419473e-01 1.05505958e-01]
|
[8.290055274963379, -2.3077173233032227]
|
982993e9-9a9a-48ec-9320-1e40c712c5b0
|
supervised-contrastive-learning-for-accented
|
2107.00921
| null |
https://arxiv.org/abs/2107.00921v1
|
https://arxiv.org/pdf/2107.00921v1.pdf
|
Supervised Contrastive Learning for Accented Speech Recognition
|
Neural network based speech recognition systems suffer from performance degradation due to accented speech, especially unfamiliar accents. In this paper, we study the supervised contrastive learning framework for accented speech recognition. To build different views (similar "positive" data samples) for contrastive learning, three data augmentation techniques including noise injection, spectrogram augmentation and TTS-same-sentence generation are further investigated. From the experiments on the Common Voice dataset, we have shown that contrastive learning helps to build data-augmentation invariant and pronunciation invariant representations, which significantly outperforms traditional joint training methods in both zero-shot and full-shot settings. Experiments show that contrastive learning can improve accuracy by 3.66% (zero-shot) and 3.78% (full-shot) on average, comparing to the joint training method.
|
['Wei Han', 'Ziang Yang', 'Hantao Huang', 'Tao Han']
|
2021-07-02
| null | null | null | null |
['accented-speech-recognition']
|
['speech']
|
[ 2.49061942e-01 2.80104112e-02 -9.01342705e-02 -5.47883093e-01
-1.09624648e+00 -2.27584615e-01 7.37035513e-01 -3.55068147e-01
-5.05907178e-01 7.81555533e-01 6.42964542e-01 -2.81145424e-01
2.44600743e-01 -1.42889455e-01 -3.48652750e-01 -7.90376306e-01
8.87175649e-02 1.83555916e-01 -1.26600310e-01 -3.80598158e-01
-1.34262651e-01 4.30771500e-01 -1.63881934e+00 1.79193288e-01
6.40464783e-01 6.99312091e-01 2.65380770e-01 9.66157794e-01
-2.82225162e-01 8.01683307e-01 -1.16302371e+00 -1.95905522e-01
-2.85013206e-02 -3.75022203e-01 -8.08473289e-01 3.01445305e-01
6.21579409e-01 -3.19118947e-02 -3.92021269e-01 1.06555045e+00
9.73383427e-01 7.12047875e-01 3.38882208e-01 -1.09380698e+00
-6.76980674e-01 6.89847648e-01 -2.89492369e-01 4.49215561e-01
7.59457722e-02 1.05581544e-01 6.52368903e-01 -1.39797592e+00
1.69082835e-01 1.44265437e+00 4.05751318e-01 9.86939371e-01
-1.03996074e+00 -7.61140525e-01 2.42617488e-01 3.16293269e-01
-1.07087111e+00 -1.10527933e+00 9.90977883e-01 1.13206275e-01
1.09506762e+00 7.20440507e-01 2.51343966e-01 1.41297507e+00
-2.94828475e-01 1.05888999e+00 1.34607148e+00 -7.27958083e-01
3.89696270e-01 9.68204811e-02 2.82260746e-01 2.73281008e-01
-5.90090811e-01 2.71426201e-01 -8.78925502e-01 4.20389585e-02
4.73225504e-01 -2.76992172e-01 -4.02983427e-01 8.70996118e-02
-9.53418911e-01 7.27445722e-01 1.10802606e-01 3.89756441e-01
-4.98299152e-01 -2.74667144e-01 5.60371399e-01 3.44692141e-01
7.83249080e-01 3.93245101e-01 -6.97277665e-01 -4.64946240e-01
-7.68715978e-01 -8.40439647e-02 6.47238076e-01 8.35748971e-01
4.76434350e-01 1.08411515e+00 -3.06443363e-01 1.51113462e+00
5.42451553e-02 6.80504978e-01 1.08434558e+00 -4.56730157e-01
3.62620384e-01 -1.23247273e-01 -2.12898418e-01 -1.54971495e-01
-2.33570650e-01 -7.00920820e-01 -1.05734289e+00 1.25073344e-01
7.88227171e-02 -5.21448612e-01 -1.57125306e+00 1.79615164e+00
-2.84290034e-03 4.36948508e-01 5.12722433e-01 7.32440650e-01
8.82978797e-01 9.89415765e-01 2.38936767e-01 -6.93490505e-01
1.20533788e+00 -1.44929814e+00 -1.41894221e+00 -4.76018041e-01
3.03188682e-01 -1.14095616e+00 1.39815927e+00 3.07003796e-01
-1.07446444e+00 -9.05813098e-01 -1.04014552e+00 3.96046698e-01
-5.48100710e-01 -1.14875093e-01 1.49246439e-01 9.68046546e-01
-9.91590321e-01 1.22767903e-01 -4.79273796e-01 -1.02630794e-01
2.40507096e-01 2.04447791e-01 -4.34312105e-01 1.54716402e-01
-1.34977281e+00 9.76654887e-01 2.81396687e-01 -1.89291209e-01
-8.29514802e-01 -8.40766907e-01 -9.65214074e-01 1.50248259e-01
3.18229914e-01 -2.13128924e-01 1.60582078e+00 -9.86462653e-01
-2.00615501e+00 3.64126235e-01 -3.79985660e-01 -5.75142026e-01
6.12739148e-03 -5.39239645e-01 -9.24674988e-01 -2.49846056e-01
-3.60301197e-01 6.05063200e-01 1.07909918e+00 -1.20972800e+00
-4.07408029e-01 -3.05168629e-01 -4.55097377e-01 4.74619180e-01
-5.09276867e-01 3.52241129e-01 -1.45318702e-01 -1.19569039e+00
7.90085122e-02 -7.30425775e-01 -3.42631012e-01 -7.64277339e-01
-3.43285382e-01 -1.98173568e-01 1.15338039e+00 -8.64945173e-01
9.96301115e-01 -2.19811583e+00 -1.99948326e-02 -1.48592502e-01
-3.07674170e-01 9.82161641e-01 -2.40840510e-01 -4.43999562e-03
-4.81215119e-01 2.31595244e-02 -9.81028229e-02 -6.56898320e-01
-1.05070569e-01 4.08016115e-01 -3.28029573e-01 -4.39794958e-02
3.49722058e-01 7.31347203e-01 -7.91836858e-01 -4.62487601e-02
6.73025310e-01 7.29833603e-01 -1.40581563e-01 5.62236488e-01
3.34197104e-01 2.65438974e-01 4.28699821e-01 5.15839458e-01
7.80847013e-01 5.75202882e-01 1.72779143e-01 -2.82907307e-01
-5.62044000e-03 5.14168262e-01 -1.14592433e+00 1.41599488e+00
-7.64527917e-01 7.35128105e-01 1.70007914e-01 -8.29229832e-01
1.20750773e+00 8.30616117e-01 -2.89562166e-01 -7.60021091e-01
-5.19068465e-02 1.35099024e-01 2.29701791e-02 -1.16308235e-01
4.86548483e-01 -5.70979953e-01 1.49183691e-01 -5.15470579e-02
5.60847759e-01 -1.35956332e-01 -3.28853190e-01 -2.02952564e-01
5.83003819e-01 -4.44315046e-01 3.45205933e-01 -2.55183935e-01
6.44246936e-01 -6.79972351e-01 5.13173401e-01 7.06221879e-01
-5.08996189e-01 5.91490567e-01 -1.13660552e-01 -2.64986873e-01
-8.67698789e-01 -1.22084141e+00 -5.46805859e-02 1.45602095e+00
-3.78941655e-01 -2.21733794e-01 -9.74066436e-01 -6.39990330e-01
-5.05871475e-01 1.16179621e+00 -2.60570467e-01 -3.37286055e-01
-6.53597891e-01 -6.89813077e-01 5.02804339e-01 8.10119510e-01
5.25886655e-01 -1.15045941e+00 3.17692757e-02 1.87056020e-01
-1.46542713e-01 -1.17991734e+00 -6.88544691e-01 7.02352822e-01
-5.66102922e-01 -1.74577981e-01 -8.98347378e-01 -1.07606268e+00
3.70383143e-01 2.59951234e-01 7.03788102e-01 -5.56063175e-01
-2.78828256e-02 2.79462785e-01 -3.78324568e-01 -5.73715270e-01
-8.00070524e-01 -1.16822496e-01 5.82291305e-01 1.04627885e-01
3.26949388e-01 -5.57120383e-01 -2.89774202e-02 1.50814697e-01
-5.46305239e-01 -1.45759061e-01 7.07722545e-01 1.25655425e+00
4.27366257e-01 -2.77646005e-01 1.10221756e+00 -5.80524921e-01
7.96340823e-01 -1.22446761e-01 -1.79375112e-01 1.65380180e-01
-5.39963663e-01 6.33450449e-02 5.47750354e-01 -6.69140756e-01
-1.44494724e+00 -2.14461237e-02 -5.08715034e-01 -7.80333102e-01
-6.81719482e-01 3.34229380e-01 -4.16453689e-01 2.64378220e-01
6.43876374e-01 5.35566509e-01 3.48074019e-01 -7.39578903e-01
5.64413190e-01 1.17708898e+00 9.00730669e-01 -1.63714617e-01
6.18560314e-01 -2.96135664e-01 -6.09454274e-01 -1.40742505e+00
-6.80127442e-01 -5.93405604e-01 -6.60423934e-01 -2.33260110e-01
6.14435792e-01 -9.63914394e-01 -1.92590490e-01 7.33877301e-01
-1.04347181e+00 -1.83361575e-01 -5.01755595e-01 7.09975541e-01
-5.43399334e-01 3.03129971e-01 -6.60298824e-01 -1.32664216e+00
-5.87002635e-01 -1.14003658e+00 6.77817106e-01 2.83387035e-01
-2.74648935e-01 -8.70034337e-01 -1.35927955e-02 5.24935067e-01
8.03982794e-01 -4.74096835e-01 5.08651078e-01 -1.20119858e+00
2.03154292e-02 -6.54481864e-03 3.42704505e-01 1.03497255e+00
6.13362610e-01 -2.43227601e-01 -1.75371277e+00 -2.97553807e-01
3.04929912e-01 -3.78731102e-01 6.69951558e-01 3.12164456e-01
9.61097062e-01 -5.46722651e-01 1.80119336e-01 2.16926441e-01
8.15468907e-01 6.25980258e-01 7.77586281e-01 -1.29300412e-02
5.51588953e-01 4.59313124e-01 6.21315122e-01 1.70317531e-01
-3.03422481e-01 8.98749471e-01 7.12802559e-02 -4.29958433e-01
-6.86023176e-01 -7.33233169e-02 4.89060074e-01 1.39054990e+00
1.52446404e-01 -2.06297949e-01 -7.26050854e-01 6.67994082e-01
-1.52697241e+00 -1.05496013e+00 2.61641502e-01 2.28864169e+00
1.14913905e+00 2.36811951e-01 1.95430070e-01 4.25367713e-01
9.94242549e-01 4.06186372e-01 -2.33620241e-01 -7.80582607e-01
-2.82771558e-01 6.87320054e-01 2.21890733e-02 9.02222216e-01
-1.35288441e+00 1.07556057e+00 6.76330471e+00 1.08181632e+00
-1.29747331e+00 2.03825444e-01 8.31811786e-01 -5.76936416e-02
1.80810273e-01 -5.30844212e-01 -6.24941707e-01 3.62129450e-01
1.50316298e+00 -2.00141698e-01 2.87807882e-01 1.07381094e+00
7.50018731e-02 3.33829790e-01 -7.44685531e-01 1.15671515e+00
3.42640460e-01 -1.18302977e+00 6.43742979e-02 -2.31150791e-01
5.79850495e-01 8.77848938e-02 2.84414947e-01 6.53728485e-01
1.16537288e-01 -9.29008484e-01 4.39837128e-01 2.29624346e-01
6.96001947e-01 -1.12223029e+00 1.06400430e+00 2.01752543e-01
-9.58441496e-01 2.20207825e-01 -6.41036853e-02 -4.64668162e-02
3.01067591e-01 3.34309399e-01 -1.32803249e+00 5.36295056e-01
5.13454735e-01 2.45519504e-02 -2.05561861e-01 7.55709350e-01
4.48901616e-02 9.85701799e-01 -5.75705990e-02 -6.36572912e-02
7.42735863e-02 1.87863901e-01 6.21466756e-01 1.50637960e+00
5.57981543e-02 -1.07442789e-01 6.59021288e-02 3.09818894e-01
-1.23382628e-01 8.72161090e-02 -4.95063722e-01 9.55151916e-02
7.83866346e-01 1.11044407e+00 -1.24410167e-01 -6.46163404e-01
-3.84145468e-01 1.14842927e+00 2.44947836e-01 4.96048599e-01
-5.77718019e-01 -5.33963263e-01 1.02645528e+00 -4.39782292e-01
4.10903305e-01 -1.38369605e-01 -1.12873919e-01 -7.14875817e-01
-2.07142383e-01 -1.10177970e+00 -6.71464130e-02 -6.72660470e-01
-1.29847026e+00 8.99283350e-01 -1.71884730e-01 -9.92984772e-01
-6.92421556e-01 -5.67741752e-01 -8.27929378e-01 1.15428126e+00
-1.35383499e+00 -9.26359534e-01 2.89951116e-02 4.06606853e-01
1.37372899e+00 -7.00849950e-01 1.33578134e+00 2.86230892e-01
-7.33982205e-01 1.04007459e+00 3.37250903e-02 1.63256004e-01
6.95287287e-01 -1.28685856e+00 7.39966273e-01 9.43396151e-01
4.79274511e-01 5.35777748e-01 7.09255338e-01 -2.38647819e-01
-1.00535882e+00 -1.08609223e+00 9.45840716e-01 -1.34497643e-01
4.51467156e-01 -3.87512803e-01 -1.19759595e+00 5.04014313e-01
6.64684951e-01 -1.51663562e-02 9.68087614e-01 2.92722076e-01
-2.86015183e-01 -1.59286380e-01 -1.08709550e+00 7.76446402e-01
6.20594859e-01 -8.97835314e-01 -1.04458821e+00 6.10154085e-02
1.03439510e+00 -1.79670811e-01 -7.74216771e-01 5.48331857e-01
1.34559229e-01 -7.80690253e-01 9.17433977e-01 -8.00212681e-01
-3.01835209e-01 -9.52112526e-02 -4.04192597e-01 -1.83082652e+00
-2.64088094e-01 -7.59673953e-01 -2.43334949e-01 1.62329423e+00
5.42441428e-01 -5.08250535e-01 5.32450020e-01 9.56297815e-02
-6.19558871e-01 -5.59314549e-01 -1.26280141e+00 -1.17671239e+00
7.31420517e-02 -5.82703173e-01 4.28133339e-01 1.17434561e+00
1.30531102e-01 7.90971458e-01 -6.63486600e-01 1.13972448e-01
2.72556692e-01 -5.93142450e-01 5.85224211e-01 -7.82142997e-01
-1.22590147e-01 -3.32612544e-01 -4.63307977e-01 -1.02156866e+00
2.39013538e-01 -5.65824687e-01 2.12047398e-01 -1.07809353e+00
-1.53248340e-01 1.58741385e-01 -5.60292840e-01 5.74763238e-01
-6.36668324e-01 6.12620972e-02 2.55356073e-01 -2.83397287e-01
-1.36630967e-01 8.34415078e-01 8.59220386e-01 -2.42364958e-01
-2.26761207e-01 1.23150691e-01 -3.79767179e-01 6.74592555e-01
1.12408352e+00 5.82329817e-02 -5.85351229e-01 -1.63808800e-02
-1.00662446e+00 -8.85427445e-02 -2.14200169e-02 -1.13197565e+00
1.00030787e-01 1.91237181e-02 2.29774594e-01 -5.25187910e-01
8.99240196e-01 -4.20939863e-01 -3.81514460e-01 3.76679599e-01
-5.03826201e-01 -1.47628233e-01 7.15088904e-01 3.68726224e-01
-5.75727642e-01 -2.36621305e-01 1.03422034e+00 2.19137445e-02
-7.68276572e-01 -9.14556831e-02 -7.99266696e-01 -3.89549844e-02
5.14319658e-01 -8.31927825e-03 -2.84478426e-01 -6.51893139e-01
-1.00246704e+00 -2.77034134e-01 -3.10353518e-01 7.71415353e-01
8.33137512e-01 -1.57000041e+00 -7.99810290e-01 6.07124686e-01
-2.38268767e-02 -3.79356712e-01 3.92238885e-01 5.12156367e-01
2.76009381e-01 4.00203884e-01 -1.09518141e-01 -3.61907154e-01
-1.67808247e+00 4.65747774e-01 4.06366110e-01 2.03740641e-01
-2.25136057e-01 1.17185426e+00 1.41446277e-01 -7.76056945e-01
7.00730264e-01 -1.65981054e-01 -2.11991817e-01 6.44427165e-02
7.89628804e-01 4.63061392e-01 3.93416047e-01 -6.16838098e-01
-3.02648544e-01 -1.75955296e-02 -6.04620457e-01 -4.28634197e-01
1.06779504e+00 3.11276875e-02 3.70400667e-01 7.78447270e-01
1.06342733e+00 1.70277864e-01 -9.12899911e-01 -4.12258834e-01
2.44225953e-02 -2.94508547e-01 2.97994286e-01 -1.06580603e+00
-7.48222709e-01 9.24230099e-01 1.09661198e+00 2.72250235e-01
1.07361937e+00 -1.90845842e-03 6.57434881e-01 4.37992364e-01
-1.33782610e-01 -1.25828862e+00 2.01274082e-01 7.09771812e-01
1.12213647e+00 -1.21485758e+00 -5.87975740e-01 -4.62802351e-01
-9.25283253e-01 9.74373937e-01 8.26785326e-01 4.71531719e-01
6.28518999e-01 4.61130142e-01 6.39017999e-01 2.95683235e-01
-8.51490498e-01 -3.39285433e-01 3.30510736e-01 9.81848061e-01
6.79709494e-01 3.35043192e-01 1.06484540e-01 5.33860147e-01
-4.73952442e-01 -6.59299016e-01 4.25909489e-01 8.76116335e-01
-6.15329266e-01 -1.15410638e+00 -5.41882038e-01 2.19860867e-01
-2.96154350e-01 -3.81738394e-01 -4.18910712e-01 5.53263545e-01
-3.12157303e-01 1.24539471e+00 3.27855796e-01 -5.57387114e-01
5.60204089e-01 8.67484152e-01 1.86374590e-01 -6.64459884e-01
-4.82397974e-01 4.81181204e-01 5.05307078e-01 -2.82745585e-02
-3.38505894e-01 -5.54792881e-01 -1.12285697e+00 1.38753489e-01
-6.42294765e-01 2.10595578e-01 8.07036161e-01 1.00180602e+00
2.56036997e-01 9.67243731e-01 9.91403878e-01 -8.20994973e-01
-8.44633937e-01 -1.56225312e+00 -5.04825771e-01 3.58586490e-01
5.54990411e-01 -4.97866035e-01 -7.22517192e-01 8.84744972e-02]
|
[14.510257720947266, 6.474651336669922]
|
cd2b3ce0-49f4-4472-b7d1-dbb96c4a1e21
|
approaches-and-applications-of-early
|
2005.02595
| null |
https://arxiv.org/abs/2005.02595v2
|
https://arxiv.org/pdf/2005.02595v2.pdf
|
Approaches and Applications of Early Classification of Time Series: A Review
|
Early classification of time series has been extensively studied for minimizing class prediction delay in time-sensitive applications such as healthcare and finance. A primary task of an early classification approach is to classify an incomplete time series as soon as possible with some desired level of accuracy. Recent years have witnessed several approaches for early classification of time series. As most of the approaches have solved the early classification problem with different aspects, it becomes very important to make a thorough review of the existing solutions to know the current status of the area. These solutions have demonstrated reasonable performance in a wide range of applications including human activity recognition, gene expression based health diagnostic, industrial monitoring, and so on. In this paper, we present a systematic review of current literature on early classification approaches for both univariate and multivariate time series. We divide various existing approaches into four exclusive categories based on their proposed solution strategies. The four categories include prefix based, shapelet based, model based, and miscellaneous approaches. The authors also discuss the applications of early classification in many areas including industrial monitoring, intelligent transportation, and medical. Finally, we provide a quick summary of the current literature with future research directions.
|
['Tanima Dutta', 'Hari Prabhat Gupta', 'Bhaskar Biswas', 'Ashish Gupta']
|
2020-05-06
| null | null | null | null |
['miscellaneous']
|
['miscellaneous']
|
[ 4.63623106e-01 -6.09588504e-01 -3.52404773e-01 -4.01763678e-01
-3.29216719e-01 -3.11689794e-01 1.95148349e-01 5.73122561e-01
-5.26872836e-02 5.70589304e-01 -6.17123425e-01 -2.11693943e-01
-5.92570543e-01 -5.96094310e-01 3.41666453e-02 -9.47400331e-01
-4.79636610e-01 1.38665885e-01 1.62596568e-01 -8.56727883e-02
5.24071813e-01 7.13212252e-01 -2.00503349e+00 3.07230324e-01
9.18413699e-01 1.46582890e+00 -1.91219494e-01 8.12383831e-01
3.67444381e-02 8.73247325e-01 -9.61533606e-01 9.75812748e-02
2.82880843e-01 -4.61696953e-01 -5.56620359e-01 1.43361297e-02
-5.94613552e-01 1.56607568e-01 1.04311913e-01 6.63772583e-01
4.12395597e-01 3.10116142e-01 7.21949100e-01 -2.01534033e+00
-7.17957839e-02 2.15882003e-01 -5.84743500e-01 5.22873759e-01
3.40199769e-01 -3.06553930e-01 4.20646936e-01 -7.81937718e-01
2.51520813e-01 5.64450204e-01 9.98743057e-01 1.73570201e-01
-9.41245317e-01 -7.05050111e-01 1.06632467e-02 6.76826954e-01
-1.45283294e+00 -1.80046603e-01 1.01862168e+00 -5.81503689e-01
9.46228683e-01 6.36236012e-01 5.08253455e-01 4.66547221e-01
7.14492142e-01 7.20694065e-01 1.23700750e+00 -4.48930383e-01
4.40219104e-01 -1.16854370e-01 5.62439024e-01 3.40624213e-01
-8.56862217e-02 3.86598766e-01 -3.31444860e-01 -4.33372080e-01
2.35166028e-01 4.94908571e-01 5.61807938e-02 7.92405605e-02
-1.02748477e+00 5.87822378e-01 -1.66222513e-01 5.51247656e-01
-4.67414916e-01 -4.88688290e-01 7.37936497e-01 8.21620703e-01
9.14714992e-01 2.98565269e-01 -7.32010007e-01 -3.60366315e-01
-6.90899253e-01 1.35230854e-01 8.10803473e-01 8.71679127e-01
3.24839890e-01 5.21213785e-02 -6.00960515e-02 9.37319994e-01
5.91855757e-02 7.00440854e-02 4.50749695e-01 -7.80245602e-01
9.91212949e-03 3.45794678e-01 -1.55713975e-01 -1.05552161e+00
-3.68038535e-01 -4.71779943e-01 -9.58127797e-01 3.44123214e-01
3.60789239e-01 -8.65845010e-02 -7.61545360e-01 1.13670492e+00
4.08058912e-01 4.52684999e-01 4.35415059e-02 3.41263682e-01
4.99593973e-01 1.06900167e+00 1.23273127e-01 -1.12308335e+00
1.11571980e+00 -8.38600755e-01 -1.00519800e+00 2.11089566e-01
4.36112165e-01 -9.23091531e-01 1.91651151e-01 6.04202271e-01
-4.93782997e-01 -5.71134210e-01 -9.18740094e-01 3.64619613e-01
-7.11311281e-01 -1.45181492e-01 8.16028953e-01 4.94991571e-01
-4.06688362e-01 1.02364922e+00 -1.04238522e+00 -6.70359910e-01
1.22880638e-01 4.98041838e-01 2.07035825e-01 1.62287682e-01
-1.19371343e+00 7.91665912e-01 1.70401130e-02 2.63710082e-01
-4.20672596e-01 -7.04971075e-01 -6.56804383e-01 -4.82252121e-01
4.57071178e-02 -3.34778428e-01 1.44990492e+00 -6.56517029e-01
-1.52025688e+00 5.37925780e-01 -2.57002711e-01 -5.51522851e-01
3.51188332e-01 8.62746388e-02 -8.14871430e-01 -1.45011947e-01
-4.23847362e-02 -1.43741444e-01 4.80261713e-01 -4.16091323e-01
-1.18495548e+00 -4.49593097e-01 -2.85784632e-01 -1.13500401e-01
-1.95657000e-01 3.87259752e-01 3.25035542e-01 -8.48066807e-01
5.07793911e-02 -8.80206466e-01 -5.70652008e-01 9.20700207e-02
1.21031217e-01 -7.84046054e-01 1.24539030e+00 -5.27537405e-01
1.50413918e+00 -1.99693799e+00 -2.61324048e-01 9.94316638e-02
-1.25882760e-01 7.01247081e-02 2.62058318e-01 7.14312732e-01
-6.38862431e-01 -2.50961363e-01 -3.22344154e-01 2.62851149e-01
-4.78136986e-01 1.31650284e-01 -4.81717199e-01 7.18691468e-01
-2.72077750e-02 6.27550602e-01 -9.69551504e-01 -2.69056588e-01
6.12798750e-01 3.56205583e-01 3.12025934e-01 2.04416394e-01
1.13088436e-01 5.72815657e-01 -5.48203945e-01 1.00122845e+00
3.88960272e-01 4.67589013e-02 -3.10680896e-01 -2.22352654e-01
-4.62509781e-01 -3.11650038e-01 -8.83736670e-01 1.20312524e+00
-9.00062397e-02 5.17682374e-01 -3.49282354e-01 -1.76686311e+00
1.07077622e+00 6.88746750e-01 1.72616553e+00 -8.10079038e-01
2.64243633e-01 5.44102132e-01 -9.62269232e-02 -6.30225718e-01
9.64779556e-02 -2.20980316e-01 -1.41208455e-01 2.10605621e-01
-4.29648757e-01 1.38335034e-01 5.12595594e-01 -5.24039984e-01
1.14392769e+00 4.26809452e-02 7.94749439e-01 -1.16883829e-01
7.73339868e-01 3.34749937e-01 8.80785048e-01 2.37253681e-01
-7.23536849e-01 2.13755146e-01 1.05818465e-01 -6.43882215e-01
-7.09098995e-01 -6.64073944e-01 -2.21141040e-01 9.33670700e-01
-2.04243988e-01 -3.90608847e-01 -2.73316354e-01 -3.13750952e-01
6.62714168e-02 4.79599535e-01 -5.23586035e-01 -3.44477206e-01
-7.14973569e-01 -1.20654607e+00 2.09230646e-01 6.40021682e-01
2.54533172e-01 -6.91754162e-01 -6.44734204e-01 4.16886896e-01
-1.21438205e-01 -8.86922002e-01 4.45294492e-02 5.34823179e-01
-1.46340168e+00 -1.15013492e+00 -8.74486744e-01 -7.63767660e-01
5.55718958e-01 1.92266434e-01 7.98188567e-01 -2.30871111e-01
-6.18734121e-01 5.02206564e-01 -5.86357534e-01 -1.01904261e+00
-3.17854673e-01 -2.02973261e-01 1.76993161e-02 1.17839321e-01
4.98864084e-01 -5.66575289e-01 -4.03776139e-01 5.37501812e-01
-5.74978769e-01 -5.38383305e-01 3.20526659e-01 6.75740302e-01
7.39930332e-01 7.47592807e-01 8.78371298e-01 -6.35128200e-01
6.49825156e-01 -7.68573761e-01 -4.73864734e-01 3.61944944e-01
-1.13790941e+00 -3.47721547e-01 7.34424770e-01 -5.77104747e-01
-7.44802475e-01 2.39384323e-01 -3.14335257e-01 -2.65944421e-01
-1.45219743e-01 7.06325769e-01 3.69393706e-01 -2.23768011e-01
4.52667117e-01 3.53683203e-01 2.43239760e-01 -5.85141540e-01
-3.31609786e-01 9.55980778e-01 2.10206538e-01 -4.77887362e-01
3.54420841e-01 3.07591766e-01 3.78441870e-01 -1.11739218e+00
-5.68465352e-01 -9.85784709e-01 -7.21857548e-01 -6.17499471e-01
5.89456797e-01 -3.44800085e-01 -7.07760274e-01 9.67039585e-01
-9.13260639e-01 -1.48010477e-01 -5.47454953e-01 4.29660529e-01
-6.38245106e-01 3.03360045e-01 -6.20607555e-01 -1.14268756e+00
-4.36378807e-01 -7.56372452e-01 8.59397113e-01 1.45133913e-01
-4.52028662e-01 -1.16626370e+00 2.04266340e-01 -1.92413561e-03
4.80086833e-01 8.55723321e-01 8.56875181e-01 -6.30560637e-01
1.33382203e-02 -5.80822647e-01 4.24240440e-01 2.54885375e-01
5.22900045e-01 2.74686873e-01 -8.45068455e-01 -1.98974550e-01
4.72154915e-01 3.02288383e-01 4.34225619e-01 5.35148084e-01
1.25863361e+00 1.28167212e-01 -7.53901780e-01 3.28633726e-01
1.17716241e+00 1.26738274e+00 6.62511766e-01 1.72097728e-01
1.55698434e-01 7.10018396e-01 1.47363114e+00 4.70996350e-01
7.73797743e-03 4.02281314e-01 8.27839896e-02 1.92631297e-02
2.19447598e-01 4.77649361e-01 3.30553114e-01 1.10508835e+00
-3.31628531e-01 -2.12731406e-01 -7.98394084e-01 4.04261291e-01
-1.92048490e+00 -1.21306694e+00 -3.95447314e-01 2.27355742e+00
5.52607179e-01 1.02826236e-02 2.71852374e-01 1.03348112e+00
8.66223156e-01 -1.80961385e-01 -5.61828673e-01 -3.62550825e-01
3.42973500e-01 1.66289106e-01 2.82389551e-01 -6.45610839e-02
-1.33691013e+00 4.59345803e-02 7.48710489e+00 5.47898412e-01
-1.39634883e+00 -3.42398547e-02 7.58149505e-01 2.34758645e-01
6.26721978e-01 -1.02805875e-01 -4.78993446e-01 5.34961939e-01
1.32255733e+00 -7.30117023e-01 -3.57552641e-03 9.34431732e-01
4.36173141e-01 -1.57596126e-01 -1.18280339e+00 1.29046977e+00
-1.04624264e-01 -8.14008951e-01 -5.77269077e-01 -1.83225721e-01
5.45032918e-01 -2.89095640e-01 -1.26578197e-01 4.50383008e-01
-5.51327288e-01 -4.99111593e-01 4.57161069e-01 7.40070760e-01
6.43737733e-01 -8.52640331e-01 6.22157633e-01 3.85633469e-01
-1.74810803e+00 -6.59021735e-01 7.24010216e-03 -4.85802203e-01
3.73401076e-01 1.11503828e+00 -4.71552014e-01 7.32912660e-01
6.44326389e-01 1.12414074e+00 -1.72829732e-01 1.44534707e+00
4.55646604e-01 4.63312477e-01 -3.40654045e-01 -1.16240710e-01
-3.12553465e-01 3.92204076e-02 5.52505314e-01 7.77564883e-01
5.70267260e-01 4.17119622e-01 5.83221853e-01 2.95663446e-01
7.45678186e-01 1.10423155e-01 -6.01994991e-01 -3.69487882e-01
2.91361541e-01 9.36128438e-01 -9.65948880e-01 -4.62145209e-01
-5.08515954e-01 7.45244741e-01 -4.74375039e-01 1.84989497e-01
-7.38151610e-01 -9.27214265e-01 7.10950494e-01 -5.92009351e-03
-3.46148729e-01 -4.38850731e-01 -4.42244172e-01 -8.32704902e-01
-1.32339731e-01 -6.68236673e-01 1.19471073e+00 -6.58318639e-01
-1.22443318e+00 4.12969649e-01 3.87115836e-01 -1.81223321e+00
-3.60421479e-01 -7.34163105e-01 -6.14638448e-01 4.72852111e-01
-1.20160174e+00 -5.77889025e-01 -2.40052819e-01 2.96071619e-01
1.02996242e+00 -1.99066047e-02 7.38724172e-01 6.66530848e-01
-7.23585844e-01 -4.54068929e-02 1.48437172e-01 -1.52852818e-01
6.48115933e-01 -9.96942341e-01 -2.36093346e-02 8.59563351e-01
-5.54373264e-01 5.92603445e-01 9.80591357e-01 -6.52567506e-01
-1.23828530e+00 -1.16139865e+00 1.00484824e+00 -1.81168392e-01
3.54952782e-01 1.23042174e-01 -7.70456910e-01 3.57106894e-01
-1.85115263e-02 6.72587827e-02 8.23257565e-01 -2.51929820e-01
5.03151178e-01 -8.33703995e-01 -1.24973536e+00 2.39134207e-01
7.15155661e-01 -1.71731129e-01 -3.93650740e-01 5.63372135e-01
2.61866927e-01 -2.08234981e-01 -1.21135235e+00 7.73218989e-01
6.81526184e-01 -6.34732783e-01 8.55447710e-01 -3.85923445e-01
-2.41939455e-01 -2.50013441e-01 2.21825004e-01 -9.39391732e-01
-4.93100405e-01 -8.35959315e-01 -5.28573275e-01 1.07600987e+00
3.30136493e-02 -8.21919620e-01 6.82348728e-01 6.88429594e-01
-3.70296508e-01 -9.98996317e-01 -6.67966485e-01 -1.11919117e+00
-4.03394043e-01 -5.02510011e-01 4.80172664e-01 9.75586295e-01
2.40106851e-01 2.87188888e-01 -1.67073771e-01 -2.02767730e-01
6.87017679e-01 4.73135382e-01 3.76372457e-01 -1.64807439e+00
3.89214426e-01 -2.84929842e-01 -6.19562984e-01 -5.73092043e-01
-3.12478572e-01 -7.32392788e-01 1.79250106e-01 -1.38743687e+00
4.52678278e-03 -3.41083080e-01 -7.97770083e-01 4.50866103e-01
2.55482048e-01 8.37795809e-02 -5.17700970e-01 2.80257672e-01
-1.34636015e-01 1.30109206e-01 8.38957727e-01 -1.59045234e-01
-2.36092851e-01 6.14977658e-01 -2.63572395e-01 5.76300800e-01
8.98565888e-01 -6.26399100e-01 -7.14053690e-01 5.22121526e-02
-6.98758103e-03 3.63646775e-01 1.17542610e-01 -1.11745965e+00
5.10408700e-01 -6.29717648e-01 1.97336107e-01 -9.73505497e-01
3.46397571e-02 -1.26133275e+00 3.31026465e-01 8.59378338e-01
-1.29875481e-01 4.76121783e-01 -8.84030983e-02 6.47447765e-01
-3.42806876e-01 -3.48119484e-03 9.28113461e-01 2.37777978e-02
-1.15939045e+00 5.38561225e-01 -6.61935091e-01 -4.49838459e-01
1.67924404e+00 -5.20628691e-01 9.29043256e-03 -5.51791601e-02
-8.78397584e-01 9.63435471e-02 -1.55015811e-01 6.72415912e-01
5.44648230e-01 -1.35105765e+00 -3.93220574e-01 3.59957308e-01
2.55529761e-01 -5.45398235e-01 9.81509089e-02 1.31229794e+00
-3.46699744e-01 5.58865190e-01 -3.46382737e-01 -7.99078703e-01
-1.43920815e+00 8.45545173e-01 1.74579620e-01 3.54735032e-02
-3.34341049e-01 4.17578906e-01 -3.50152522e-01 2.92718410e-04
2.61071295e-01 -6.38830781e-01 -4.95209664e-01 1.52109757e-01
5.64607024e-01 1.07647669e+00 3.04206252e-01 -5.56209683e-01
-6.90372109e-01 7.95862257e-01 2.57411510e-01 2.80981004e-01
1.49781263e+00 -1.72863573e-01 -2.96529740e-01 9.62662399e-01
1.38407767e+00 -5.03228426e-01 -7.08521903e-01 1.27784833e-01
3.87423694e-01 -1.41198948e-01 -3.12010422e-02 -5.53446174e-01
-9.12968576e-01 7.32741892e-01 9.12716269e-01 3.92610818e-01
1.65684497e+00 -5.11234164e-01 9.36612904e-01 3.04135382e-01
6.61998987e-01 -1.35821664e+00 -4.96214516e-02 4.30592269e-01
8.49719942e-01 -8.18542719e-01 1.63862228e-01 -7.23481774e-01
-1.38473317e-01 1.52758682e+00 2.15566441e-01 -1.87860020e-02
1.16886568e+00 3.94431472e-01 -1.23683520e-01 -1.59671903e-03
-9.40871537e-01 -1.24803148e-01 3.26275170e-01 7.10772574e-01
7.12308824e-01 -2.76350647e-01 -8.05725336e-01 6.33363783e-01
3.09665591e-01 4.52237338e-01 5.47110215e-02 1.50699925e+00
-4.31862652e-01 -1.15345371e+00 -4.85920250e-01 5.48689723e-01
-7.27198899e-01 5.81120610e-01 -1.50976181e-01 4.02606189e-01
8.90588686e-02 1.30947530e+00 2.37160735e-02 -5.34369111e-01
4.67694253e-01 1.52384281e-01 3.33941370e-01 -4.48290735e-01
-2.16773167e-01 -2.37903302e-03 6.64603040e-02 -5.03447890e-01
-6.72825754e-01 -1.06604707e+00 -1.22726667e+00 -3.88331078e-02
-4.18418348e-01 3.59018773e-01 6.33752942e-01 9.96113837e-01
3.43581319e-01 6.52661800e-01 9.86140132e-01 -6.69687331e-01
-4.43984270e-01 -6.43142581e-01 -5.65147519e-01 1.43317670e-01
2.41555214e-01 -7.65257061e-01 -3.18308443e-01 4.01215494e-01]
|
[7.22817850112915, 3.137441396713257]
|
d5a96f10-500f-4a2d-a2a4-71c35841b0d7
|
time-aware-relational-graph-attention-network
| null | null |
https://openreview.net/forum?id=ShtJLsF7cbb
|
https://openreview.net/pdf?id=ShtJLsF7cbb
|
Time-aware Relational Graph Attention Network for Temporal Knowledge Graph Embeddings
|
Embedding-based representation learning approaches for knowledge graphs (KGs) have been mostly designed for static data. However, many KGs involve temporal data, which creates the need for new representation learning approaches that can characterize and reason over time. In this work, we propose a Time-aware Relational Graph ATtention Network (TR-GAT) for temporal knowledge graph (TKG) embeddings, in which the initial feature of each entity is represented by fusing its embedding and the embeddings of its connected relations and timestamps as well as its neighboring entities. Different from the existing temporal GNN models which discretize temporal graphs into multiple snapshots, we treat timestamps as properties of links between entities. To further incorporate relation and time information into the graph structures, we utilize a self-attention mechanism which specifies different weights to different nodes according to the corresponding link features, i.e., embeddings of the relevant relations and timestamps within one neighborhood. Experimental results show that our approach achieves state-of-the-art performances regarding TKG completion and entity alignment tasks on several well-established TKG datasets due to the effective and efficient integration of time information.
|
['Jens Lehmann', 'Fenglong Su', 'Chengjin Xu']
|
2021-09-29
| null | null | null | null |
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'methodology']
|
[-5.49313903e-01 9.03970152e-02 -5.09081066e-01 -4.05886352e-01
-8.48551467e-03 -4.48022842e-01 6.69130385e-01 8.00811708e-01
-3.48317236e-01 3.30268770e-01 4.62549508e-01 -3.53338659e-01
-5.35284162e-01 -1.23999918e+00 -5.47489703e-01 -4.94135261e-01
-6.15438104e-01 5.70450068e-01 2.50365615e-01 -2.20950216e-01
-2.50229806e-01 3.35069358e-01 -8.74144495e-01 -2.44236499e-01
7.57663488e-01 9.36642468e-01 -1.73236698e-01 2.47970819e-01
-3.07610393e-01 1.09406698e+00 -3.84083182e-01 -6.43381834e-01
-8.17446709e-02 -8.58134180e-02 -8.30883086e-01 -3.87101948e-01
8.16468000e-02 -2.54715234e-01 -1.41638052e+00 8.21935236e-01
2.43365943e-01 5.46824157e-01 5.10863423e-01 -1.67234266e+00
-1.38418913e+00 9.92900133e-01 -4.32673335e-01 4.12885338e-01
1.95697874e-01 -2.18044281e-01 1.45477927e+00 -4.50051069e-01
7.97807038e-01 1.06763077e+00 6.13890827e-01 1.15487397e-01
-9.46110129e-01 -4.70686346e-01 6.64220750e-01 9.54565167e-01
-1.45132959e+00 4.95667867e-02 8.93562019e-01 -3.51260990e-01
1.28729856e+00 -5.84727414e-02 7.60807097e-01 9.21936095e-01
-7.02142622e-03 5.22303820e-01 1.28441975e-01 -1.80197954e-01
-4.84651625e-02 -4.03901458e-01 4.98338193e-01 5.68354547e-01
3.66263241e-01 -2.23797694e-01 -4.04686451e-01 -7.91433379e-02
6.96117282e-01 3.08730423e-01 -2.28985965e-01 -5.54008961e-01
-1.41668260e+00 7.75895417e-01 9.51602638e-01 4.98946935e-01
-5.05934060e-01 5.90770125e-01 7.15902925e-01 2.28877977e-01
4.85629052e-01 1.27302080e-01 -4.90717977e-01 4.20901217e-02
-3.50260824e-01 -1.19751237e-01 5.71064234e-01 1.13772380e+00
6.59183264e-01 -1.68224536e-02 -2.23761886e-01 5.39942503e-01
2.33239815e-01 9.96017233e-02 5.20966530e-01 -3.44313890e-01
7.21519053e-01 1.12545025e+00 8.57905392e-03 -1.48859584e+00
-4.24752682e-01 -2.29470477e-01 -8.16944838e-01 -7.88911641e-01
5.33930697e-02 1.73386291e-01 -9.26441491e-01 1.93324530e+00
4.68958229e-01 7.21416652e-01 4.42442261e-02 6.31418347e-01
9.60949123e-01 7.67165005e-01 1.41711101e-01 -1.43340275e-01
1.22068882e+00 -8.53988051e-01 -9.23031271e-01 7.83846155e-02
8.36446702e-01 7.46492296e-02 6.74345791e-01 -3.94424379e-01
-5.08407652e-01 -3.31459522e-01 -9.14385200e-01 -3.28212589e-01
-9.69019532e-01 -3.23059887e-01 9.06743228e-01 -1.20567925e-01
-8.28413785e-01 7.96145678e-01 -1.11648881e+00 -5.77794969e-01
9.48743969e-02 1.90493286e-01 -4.84434307e-01 -8.96000713e-02
-1.87676394e+00 8.09526324e-01 8.68083239e-01 3.37724388e-01
-4.08187211e-01 -6.21292651e-01 -1.26285005e+00 2.49064967e-01
4.96862620e-01 -7.04957008e-01 7.75611341e-01 -1.95935518e-01
-7.94517219e-01 5.48289657e-01 -1.36670232e-01 -6.68788373e-01
-1.06562451e-02 -1.03243083e-01 -1.11074853e+00 6.22220486e-02
1.44095659e-01 7.05125257e-02 3.36004704e-01 -7.83408105e-01
-3.55680585e-01 -3.97931814e-01 3.37874711e-01 8.63828510e-02
-7.56578326e-01 -1.40218601e-01 -9.69574094e-01 -6.06727719e-01
1.48337007e-01 -6.28691256e-01 -2.09088162e-01 -2.20585585e-01
-4.14425999e-01 -5.37322640e-01 8.05204988e-01 -8.31194520e-01
1.87005055e+00 -2.02765751e+00 4.57592726e-01 2.37958074e-01
5.03451169e-01 1.09694302e-01 -1.97086751e-01 8.69076669e-01
-3.06867540e-01 7.58865625e-02 2.03020334e-01 -1.40738383e-01
2.39226893e-01 6.20533466e-01 -3.80037725e-01 4.09204066e-01
6.16857372e-02 1.31748426e+00 -1.39487600e+00 -4.52019244e-01
2.17415348e-01 5.85185945e-01 -3.34115662e-02 6.81680292e-02
-2.38344267e-01 8.86486024e-02 -5.97856998e-01 4.11613584e-01
2.32842043e-01 -6.25699818e-01 5.88214755e-01 -6.12466812e-01
1.39346108e-01 3.80490005e-01 -9.74006951e-01 1.73773932e+00
-3.27363700e-01 3.59219104e-01 -6.80171013e-01 -1.00241506e+00
8.45030665e-01 4.36699629e-01 6.91364527e-01 -7.03787029e-01
-1.26487459e-03 -6.03401512e-02 -4.98889275e-02 -3.92093480e-01
7.19428122e-01 4.10540342e-01 -2.90632635e-01 4.22204524e-01
2.67482787e-01 6.30929649e-01 3.50321442e-01 8.12394261e-01
1.25840199e+00 6.30337000e-02 3.65533262e-01 3.02001715e-01
3.01658213e-01 -2.04852119e-01 7.57683039e-01 1.40693620e-01
-1.32308111e-01 3.81590575e-02 6.73469663e-01 -7.35353291e-01
-9.04106975e-01 -8.92129064e-01 3.77700567e-01 9.00282741e-01
3.49112570e-01 -9.98808324e-01 -8.33590105e-02 -7.73850203e-01
3.82754803e-01 5.66133976e-01 -1.01133621e+00 -6.01711571e-01
-6.31132603e-01 -4.70921576e-01 3.09028804e-01 1.01932263e+00
1.51365936e-01 -9.55102682e-01 3.22372243e-02 6.00566268e-01
-2.95769900e-01 -1.26333761e+00 -6.64225399e-01 6.84391381e-03
-7.25660384e-01 -1.31711328e+00 -3.24986309e-01 -7.33278632e-01
5.87268412e-01 2.53242701e-01 9.74583626e-01 1.05556389e-02
-6.22656085e-02 5.08002162e-01 -6.59853995e-01 8.59924480e-02
2.66586542e-01 2.33058944e-01 1.85235769e-01 2.01360419e-01
4.00786519e-01 -9.09746945e-01 -5.49168587e-01 8.14934298e-02
-8.00814390e-01 -1.31513074e-01 1.60930827e-01 7.27393687e-01
7.32758164e-01 1.65418863e-01 5.60558558e-01 -9.23197567e-01
5.55095732e-01 -8.77780497e-01 -5.20101786e-01 7.18868315e-01
-8.00957203e-01 1.57697648e-01 7.58937299e-01 -6.36034489e-01
-5.61684310e-01 -4.79972005e-01 6.24970853e-01 -9.72904742e-01
5.26857793e-01 1.19582224e+00 -1.57388330e-01 1.75039187e-01
1.83113739e-01 2.37574458e-01 -4.89140868e-01 -3.95110875e-01
8.61567616e-01 1.42785683e-01 6.67628229e-01 -6.69171929e-01
1.01341558e+00 3.20623577e-01 2.77761044e-03 -3.02597314e-01
-7.33675599e-01 -5.42365611e-01 -8.18394423e-01 -1.54927894e-02
7.16510832e-01 -8.12419415e-01 -6.50783896e-01 3.01112920e-01
-1.17041647e+00 -1.18148871e-01 -3.12475711e-01 5.98404884e-01
-2.39263147e-01 5.47420442e-01 -8.49022388e-01 -5.28544307e-01
-3.44055831e-01 -5.38194776e-01 6.50078475e-01 1.91726834e-01
-1.21657774e-01 -1.43531585e+00 1.66843101e-01 -2.70114895e-02
3.04805875e-01 5.97007334e-01 1.24655557e+00 -8.43829751e-01
-7.25464642e-01 -2.80705243e-01 -4.59174991e-01 -1.77291811e-01
3.78712445e-01 1.16068073e-01 -3.22908640e-01 -3.32979620e-01
-6.12448096e-01 7.35550001e-02 8.39149237e-01 5.10285087e-02
9.10323262e-01 -4.57789034e-01 -7.08912432e-01 7.83404469e-01
1.43693161e+00 2.87464857e-01 4.09258306e-01 4.58414733e-01
1.34079969e+00 3.62624228e-01 7.01583922e-01 4.96326059e-01
9.82879579e-01 6.40215695e-01 5.62826693e-01 2.29231820e-01
7.38102719e-02 -6.48502171e-01 9.87155959e-02 1.30646574e+00
-2.64766723e-01 -4.22553092e-01 -9.03500021e-01 9.69016314e-01
-2.37536144e+00 -1.02940917e+00 -1.85953170e-01 1.97683907e+00
5.98483026e-01 -1.25890389e-01 -8.06433931e-02 -1.18902944e-01
9.64582562e-01 7.19718039e-01 -7.94574142e-01 -9.30313990e-02
1.71542019e-02 -8.02477747e-02 6.10614061e-01 1.87295660e-01
-1.05720186e+00 1.04242134e+00 4.81003904e+00 4.96501416e-01
-7.92802691e-01 5.73959686e-02 -2.15222891e-02 8.82820040e-02
-5.39458096e-01 1.81917056e-01 -4.65822875e-01 5.43632388e-01
1.08672225e+00 -8.04202795e-01 6.34029269e-01 6.43801987e-01
-1.57209843e-01 5.20837605e-01 -1.22132552e+00 8.94858301e-01
-1.25704482e-01 -1.30901659e+00 1.85162559e-01 -1.16648413e-02
5.99618733e-01 6.24034852e-02 -1.74452513e-01 5.49524546e-01
6.48534477e-01 -7.92701781e-01 4.69309300e-01 7.16713667e-01
8.07540178e-01 -7.51612127e-01 6.80479109e-01 -1.12657256e-01
-2.04551196e+00 -2.58651543e-02 -3.65595639e-01 2.01420262e-01
2.45445877e-01 4.54188615e-01 -8.85892987e-01 1.56490445e+00
6.01319969e-01 1.40119851e+00 -5.05703092e-01 8.62273633e-01
-4.68431532e-01 4.07912791e-01 -1.95006713e-01 1.84374973e-01
2.17465222e-01 -3.89225870e-01 2.07868442e-01 8.60367298e-01
3.21055800e-01 8.51910859e-02 6.07820004e-02 5.84460258e-01
-3.64130825e-01 7.19756400e-03 -6.41461611e-01 -5.70309460e-01
1.04603529e+00 1.13808310e+00 -5.64936161e-01 -3.41564715e-01
-6.48263276e-01 8.70745242e-01 8.33056271e-01 6.81594670e-01
-1.07796359e+00 -6.58077955e-01 8.22695136e-01 -8.71721506e-02
5.64368725e-01 -4.98308986e-01 3.93386453e-01 -1.39195955e+00
8.46824721e-02 -1.66971862e-01 1.14809716e+00 -7.83128202e-01
-1.49961925e+00 5.88555157e-01 1.47946507e-01 -1.15736318e+00
-2.07066685e-01 -1.60092354e-01 -6.92029417e-01 7.47637272e-01
-1.44831514e+00 -1.56868386e+00 -2.30147392e-01 7.89145708e-01
-2.79014647e-01 1.64343566e-01 8.18517268e-01 5.36037862e-01
-6.84835136e-01 5.27641118e-01 1.70938745e-01 7.17018545e-01
5.81220031e-01 -1.23866546e+00 8.97304416e-01 8.67525041e-01
4.21274036e-01 9.18436527e-01 2.35691011e-01 -8.14847410e-01
-1.61688256e+00 -1.49463868e+00 1.18778253e+00 -2.91759282e-01
1.38956201e+00 -1.91301480e-01 -1.26575458e+00 1.50266778e+00
-4.83511761e-02 5.88082075e-01 5.74447691e-01 5.38667262e-01
-8.89580309e-01 -3.72964859e-01 -5.56626439e-01 5.55906951e-01
1.45388210e+00 -8.53233159e-01 -6.82656586e-01 2.69014537e-01
1.35358799e+00 -3.93765807e-01 -1.38169181e+00 3.93480122e-01
2.79683113e-01 -2.00054467e-01 1.02790880e+00 -1.04966295e+00
-5.13745798e-03 -5.56692779e-01 1.30119324e-02 -1.34497786e+00
-6.44448698e-01 -3.76792282e-01 -8.25269938e-01 1.60067761e+00
1.71558559e-01 -8.60438049e-01 5.19158542e-01 5.52659690e-01
-1.30797595e-01 -7.16593146e-01 -9.51215386e-01 -8.96233737e-01
-4.48843360e-01 -2.91430593e-01 1.05600107e+00 1.66632915e+00
1.70047715e-01 4.38034058e-01 -5.21370769e-01 5.19546390e-01
4.30988073e-01 6.68042600e-01 4.22413617e-01 -1.36757612e+00
-1.16065508e-02 -2.13050053e-01 -9.17082429e-01 -5.81251264e-01
3.54267538e-01 -1.13583148e+00 -5.22858500e-01 -2.05511475e+00
-3.16383243e-02 -3.14747334e-01 -8.50436151e-01 7.65942812e-01
-2.53750294e-01 -4.59704787e-01 -9.59408879e-02 1.48885697e-01
-9.44009900e-01 8.80252600e-01 9.49242711e-01 -3.67173761e-01
-5.31627424e-02 -4.94044065e-01 -4.70771402e-01 2.49879450e-01
4.45195258e-01 -4.12578166e-01 -7.71554053e-01 -6.67717516e-01
3.96149844e-01 1.87829390e-01 2.36199602e-01 -6.01919472e-01
7.76928008e-01 -2.46993065e-01 5.30966325e-03 -6.47492409e-01
2.65668988e-01 -8.74762058e-01 5.94175816e-01 1.27221510e-01
-1.27089679e-01 2.19366699e-01 -5.59500419e-02 1.25806224e+00
-4.77160454e-01 3.49210024e-01 7.13664070e-02 1.56181589e-01
-1.15687513e+00 1.05353403e+00 2.83389717e-01 4.03582454e-02
1.04354000e+00 -4.16534953e-02 -6.18966281e-01 -3.61468792e-01
-8.09095919e-01 7.93607056e-01 3.82721424e-01 6.99836910e-01
5.94138563e-01 -1.82266915e+00 -2.63776273e-01 -2.65437573e-01
5.63193321e-01 2.51314789e-01 4.65904325e-01 6.98084891e-01
-6.84825554e-02 3.53021532e-01 -3.12332865e-02 -5.44954948e-02
-9.91273701e-01 1.08234870e+00 1.61703035e-01 -5.54465473e-01
-7.53218651e-01 6.76116407e-01 -2.49317549e-02 -3.22331280e-01
2.10946098e-01 -5.91328681e-01 -4.93973762e-01 4.00270402e-01
2.21709743e-01 2.76626408e-01 -3.47957090e-02 -6.04251206e-01
-6.41964495e-01 5.38305521e-01 -2.63405770e-01 2.62737513e-01
1.46683919e+00 3.17789661e-03 -3.51670712e-01 7.05974162e-01
1.21188629e+00 -2.79736310e-01 -7.55141437e-01 -8.20333779e-01
3.07456285e-01 -4.28443432e-01 -1.84262276e-01 -2.78881133e-01
-1.29664302e+00 5.72556257e-01 -1.90682054e-01 2.59689867e-01
8.85347068e-01 2.31038421e-01 1.06369305e+00 2.73172528e-01
5.53396463e-01 -7.09712863e-01 -5.82373962e-02 6.74826801e-01
5.64181745e-01 -8.33623469e-01 -4.04677726e-02 -3.14300209e-01
-4.87837046e-01 1.01203692e+00 6.81028426e-01 9.58018471e-03
7.16506302e-01 -3.83014709e-01 -4.15229589e-01 -5.27226865e-01
-9.22980189e-01 -3.28083694e-01 3.59735191e-01 6.44978464e-01
8.10049176e-02 2.70298243e-01 -1.61722451e-01 9.23337758e-01
1.55570164e-01 -1.36281922e-01 2.18051851e-01 7.47382283e-01
9.39061493e-02 -1.11526096e+00 3.16916823e-01 3.86634529e-01
3.14021632e-02 -7.63678476e-02 -2.29652062e-01 9.77302074e-01
-1.32027090e-01 5.73413908e-01 2.08315179e-01 -7.72389054e-01
4.50230628e-01 8.69167410e-03 2.26616099e-01 -6.97841585e-01
-3.29912931e-01 -5.41173160e-01 2.17453972e-01 -6.07194304e-01
-2.93577164e-01 -5.10268390e-01 -1.55243766e+00 -5.24716914e-01
-2.92584181e-01 3.93471271e-01 1.58214733e-01 6.34030104e-01
6.60166860e-01 8.73725533e-01 5.14177263e-01 -3.38620007e-01
-1.04442656e-01 -8.01804543e-01 -9.15370703e-01 6.19085133e-01
3.86400856e-02 -8.57187390e-01 -1.29928932e-01 -1.50199130e-01]
|
[8.578947067260742, 7.880882740020752]
|
030072e4-5b0f-486f-b275-0581ad89a2a6
|
graphonomy-universal-image-parsing-via-graph
|
2101.1062
| null |
https://arxiv.org/abs/2101.10620v1
|
https://arxiv.org/pdf/2101.10620v1.pdf
|
Graphonomy: Universal Image Parsing via Graph Reasoning and Transfer
|
Prior highly-tuned image parsing models are usually studied in a certain domain with a specific set of semantic labels and can hardly be adapted into other scenarios (e.g., sharing discrepant label granularity) without extensive re-training. Learning a single universal parsing model by unifying label annotations from different domains or at various levels of granularity is a crucial but rarely addressed topic. This poses many fundamental learning challenges, e.g., discovering underlying semantic structures among different label granularity or mining label correlation across relevant tasks. To address these challenges, we propose a graph reasoning and transfer learning framework, named "Graphonomy", which incorporates human knowledge and label taxonomy into the intermediate graph representation learning beyond local convolutions. In particular, Graphonomy learns the global and structured semantic coherency in multiple domains via semantic-aware graph reasoning and transfer, enforcing the mutual benefits of the parsing across domains (e.g., different datasets or co-related tasks). The Graphonomy includes two iterated modules: Intra-Graph Reasoning and Inter-Graph Transfer modules. The former extracts the semantic graph in each domain to improve the feature representation learning by propagating information with the graph; the latter exploits the dependencies among the graphs from different domains for bidirectional knowledge transfer. We apply Graphonomy to two relevant but different image understanding research topics: human parsing and panoptic segmentation, and show Graphonomy can handle both of them well via a standard pipeline against current state-of-the-art approaches. Moreover, some extra benefit of our framework is demonstrated, e.g., generating the human parsing at various levels of granularity by unifying annotations across different datasets.
|
['Xiaodan Liang', 'Meng Wang', 'Ke Gong', 'Yiming Gao', 'Liang Lin']
|
2021-01-26
| null | null | null | null |
['human-parsing']
|
['computer-vision']
|
[ 3.78938138e-01 3.89851213e-01 -3.90458345e-01 -6.24570310e-01
-5.93408048e-01 -9.38217998e-01 6.69746101e-01 3.29698354e-01
-1.56578273e-01 6.24277055e-01 6.86054602e-02 -1.53828725e-01
-2.93390274e-01 -1.18655562e+00 -7.74819613e-01 -5.82732677e-01
2.11107641e-01 6.86435282e-01 4.41517234e-01 4.43311296e-02
-1.09222896e-01 1.71928272e-01 -1.24632311e+00 5.10442138e-01
7.98342586e-01 1.03042841e+00 3.85213733e-01 2.49228105e-01
-6.25744343e-01 7.72791922e-01 -3.36555392e-01 -7.42137015e-01
1.18513994e-01 -5.09897411e-01 -1.17439616e+00 2.94896394e-01
5.36622941e-01 1.86068520e-01 7.14444891e-02 1.47636187e+00
4.11621965e-02 -1.44575462e-01 5.16171455e-01 -1.31687844e+00
-9.75126088e-01 6.31701112e-01 -6.03311658e-01 -1.20803766e-01
1.81524688e-03 -1.77508146e-02 1.28672647e+00 -1.40197575e-01
8.70292783e-01 1.35423660e+00 6.00311995e-01 5.40930867e-01
-1.26287174e+00 -8.39213014e-01 5.00473142e-01 7.30899423e-02
-8.80066574e-01 3.17006707e-01 8.70030224e-01 -7.09421337e-01
5.02807319e-01 -1.40529156e-01 3.92899841e-01 1.25925231e+00
-1.09174728e-01 4.65833008e-01 1.36335409e+00 -8.92718807e-02
1.23321697e-01 -3.39705050e-02 3.31833720e-01 9.75406766e-01
3.10985625e-01 -1.51276082e-01 -5.06411791e-01 1.35059133e-01
7.09609509e-01 -1.25334501e-01 -3.93711329e-02 -4.51132059e-01
-1.28270507e+00 9.94573474e-01 7.41172135e-01 3.29277873e-01
7.14842090e-03 6.35983944e-02 5.47428012e-01 3.34907264e-01
6.21822536e-01 4.45632845e-01 -8.83274376e-01 5.82029521e-01
-5.95061362e-01 -5.60195697e-03 7.51781166e-01 1.22370684e+00
1.31269503e+00 -5.00404239e-01 -4.92469706e-02 5.82271636e-01
4.12341475e-01 2.19258383e-01 1.60518318e-01 -7.49091387e-01
7.37255633e-01 1.01758635e+00 -2.55649388e-01 -9.58531559e-01
-7.40699768e-01 -4.42814618e-01 -8.44483733e-01 6.78683966e-02
6.35369360e-01 -1.63350627e-01 -9.55409706e-01 2.23955560e+00
5.25802135e-01 3.37798208e-01 6.34559840e-02 8.53954792e-01
1.35482526e+00 3.02962631e-01 6.70031786e-01 1.72583178e-01
1.87919712e+00 -1.12471855e+00 -4.11115348e-01 -5.00145376e-01
7.78084815e-01 -4.28505808e-01 1.01907051e+00 2.60662120e-02
-4.77356315e-01 -5.92081845e-01 -7.06847906e-01 -3.15867573e-01
-7.80131400e-01 2.79103927e-02 8.90431404e-01 1.99005425e-01
-1.02683926e+00 4.73086983e-01 -4.83054101e-01 -7.65174985e-01
6.03190720e-01 2.26277277e-01 -5.50198734e-01 -1.96792573e-01
-1.37091959e+00 5.62266231e-01 7.50329435e-01 -4.31076556e-01
-7.28957355e-01 -6.68417215e-01 -9.35454845e-01 7.73034245e-02
6.65519059e-01 -9.82493937e-01 8.10274899e-01 -1.13261950e+00
-1.24523067e+00 1.52221501e+00 2.44062081e-01 -2.22904295e-01
4.20636326e-01 -6.44804686e-02 -3.77660900e-01 1.02155045e-01
5.87118864e-01 9.49278116e-01 7.98780322e-01 -1.08477724e+00
-6.61670983e-01 -6.20021820e-01 4.02335107e-01 3.12770963e-01
-1.37880325e-01 -7.36112371e-02 -4.61943388e-01 -5.13244808e-01
1.10644147e-01 -8.64417732e-01 -9.32746287e-03 -1.36533290e-01
-3.84627342e-01 -4.19017702e-01 7.04639196e-01 -4.92692500e-01
6.61834598e-01 -1.99319196e+00 3.61697793e-01 -1.85953096e-01
3.53979856e-01 -1.27317635e-02 -2.41576642e-01 2.25946933e-01
-3.51014696e-02 2.35382497e-01 -4.70380127e-01 -5.63972704e-02
-2.59029176e-02 3.77196580e-01 -1.91351846e-01 2.19024435e-01
4.73312527e-01 1.13800716e+00 -1.35985351e+00 -5.95095754e-01
4.73907627e-02 1.71937689e-01 -3.63724828e-01 3.19165438e-01
-6.49504602e-01 1.11974120e+00 -8.14773858e-01 4.98316109e-01
6.58041894e-01 -8.41693997e-01 5.04820585e-01 -5.15532315e-01
2.97168314e-01 2.20371723e-01 -9.56212997e-01 2.10392118e+00
-4.89549667e-01 1.95319250e-01 9.79736820e-02 -1.43338335e+00
9.87200260e-01 1.25435680e-01 4.32800978e-01 -6.14707053e-01
4.22790758e-02 1.38675347e-01 -3.07785779e-01 -4.89364624e-01
-4.20777686e-02 -2.46917292e-01 -3.98147017e-01 4.58871782e-01
6.89289510e-01 -1.48069844e-01 1.86545983e-01 1.36987403e-01
9.80711043e-01 3.75959843e-01 4.60998982e-01 -3.83561850e-01
5.03390193e-01 1.71448782e-01 5.80132663e-01 5.61165988e-01
-2.14922890e-01 2.85416186e-01 8.03227067e-01 -4.84610677e-01
-6.21414304e-01 -1.08089733e+00 -4.34794091e-02 1.48032880e+00
5.45725822e-01 -2.60700792e-01 -7.69554675e-01 -1.36810279e+00
6.19572168e-03 3.27691138e-01 -8.55187654e-01 2.06509978e-02
-2.83768803e-01 -7.55466700e-01 4.46243525e-01 3.78400296e-01
1.03836966e+00 -1.34394217e+00 -3.03224504e-01 -5.70063293e-02
-8.36546421e-02 -1.61530495e+00 -9.16395560e-02 3.27092767e-01
-7.77507901e-01 -1.35098135e+00 -2.94266224e-01 -1.06691539e+00
5.70677519e-01 1.57944381e-01 1.69803560e+00 5.99262267e-02
3.69945131e-02 3.71704042e-01 -3.81055802e-01 -1.39735574e-02
-3.71464521e-01 3.85044664e-01 -6.34923756e-01 2.77547687e-02
4.72906351e-01 -5.25200248e-01 -4.40256745e-01 3.79809439e-01
-8.65963042e-01 4.22712177e-01 4.08284932e-01 8.99568558e-01
7.06205606e-01 6.12626150e-02 7.58223593e-01 -1.63475645e+00
4.22574848e-01 -8.58587086e-01 -8.48613322e-01 5.35429597e-01
-4.53210354e-01 1.80283874e-01 5.95049858e-01 -1.90993994e-01
-1.19899201e+00 7.57872639e-03 1.59803435e-01 2.32026223e-02
-5.63714981e-01 4.85102236e-01 -6.73495114e-01 7.78444037e-02
5.65881491e-01 -2.45592043e-01 -3.19225818e-01 -3.63742560e-01
9.39513445e-01 2.32300535e-01 5.64484119e-01 -8.71989965e-01
6.98415935e-01 5.90444446e-01 1.46508545e-01 -3.18275839e-01
-1.45445871e+00 -5.37517965e-01 -7.85200119e-01 1.78933460e-02
1.54418242e+00 -1.00581098e+00 -3.53009611e-01 4.62703377e-01
-1.24099147e+00 -5.95689178e-01 -2.42580280e-01 1.55612826e-01
-4.81536686e-01 2.02678338e-01 -7.22924888e-01 2.46160775e-02
-1.27484739e-01 -1.15054786e+00 1.52061450e+00 3.80289048e-01
-1.58941606e-03 -1.49764562e+00 7.07475021e-02 5.52075505e-01
-3.35669406e-02 4.03898954e-01 1.18647456e+00 -7.85082817e-01
-6.45511866e-01 3.46070111e-01 -9.63886678e-01 1.09989464e-01
3.80374968e-01 -4.69769210e-01 -1.06652892e+00 -8.11315924e-02
-3.05775255e-01 -5.66797376e-01 9.45782125e-01 1.32233992e-01
1.12861300e+00 1.03559308e-02 -4.53546464e-01 7.19206870e-01
1.40153086e+00 -1.43863022e-01 2.23756462e-01 3.50463271e-01
1.14260924e+00 1.09144485e+00 3.91831100e-01 -1.19299322e-01
5.28529167e-01 4.65172857e-01 5.12119532e-01 -3.30144286e-01
-4.24565285e-01 -3.11729461e-01 4.46290411e-02 6.11718118e-01
1.81732208e-01 -2.84116328e-01 -8.58825147e-01 4.08680409e-01
-2.03253698e+00 -5.40467680e-01 -1.71732590e-01 1.70180166e+00
8.03006649e-01 1.11478738e-01 -8.62230659e-02 -7.04295695e-01
9.96012866e-01 2.57376194e-01 -7.64951885e-01 -1.78698614e-01
-1.52442098e-01 3.24531436e-01 5.22902250e-01 3.51978481e-01
-1.47726715e+00 1.52610898e+00 4.59439754e+00 6.74120665e-01
-9.51671004e-01 4.43075240e-01 6.06815934e-01 6.63283348e-01
-3.87042761e-01 2.96940714e-01 -6.54320598e-01 2.73051947e-01
5.13751924e-01 1.19788446e-01 3.71567667e-01 8.99668336e-01
-5.84539413e-01 2.47207776e-01 -1.01356471e+00 8.08104813e-01
-1.80080652e-01 -1.37216330e+00 1.68802813e-01 -3.65106352e-02
8.69976163e-01 2.08402663e-01 -2.46065855e-01 4.56588477e-01
9.35152888e-01 -9.43187296e-01 4.62827295e-01 1.89294979e-01
7.93585360e-01 -2.60066867e-01 6.87634945e-01 1.71519503e-01
-1.48085785e+00 1.98420167e-01 -3.41770858e-01 8.88976157e-02
1.06569543e-01 7.43501484e-01 -4.67799395e-01 9.92888689e-01
7.87835479e-01 9.81314778e-01 -6.78388178e-01 3.31657648e-01
-8.93005252e-01 5.05598664e-01 3.90462950e-02 4.27526712e-01
3.85835201e-01 -4.43324536e-01 1.87923953e-01 1.29819942e+00
1.48749635e-01 -1.31110132e-01 5.57485104e-01 1.07075846e+00
-5.05128920e-01 7.78383687e-02 -7.46347070e-01 -1.11598879e-01
2.95852363e-01 1.63379860e+00 -1.08803046e+00 -4.05644238e-01
-8.09273243e-01 9.64908183e-01 7.47213960e-01 5.24285018e-01
-8.46065104e-01 -1.43438764e-02 6.72789216e-01 -6.95031956e-02
2.08103269e-01 4.63735033e-03 -3.55995685e-01 -1.31400681e+00
-3.93420011e-01 -5.17997563e-01 9.32432055e-01 -6.45042479e-01
-1.86327720e+00 5.83709121e-01 8.24283659e-02 -9.17620659e-01
8.71326178e-02 -8.49685967e-01 -4.55950826e-01 8.42066646e-01
-1.82904696e+00 -1.74683022e+00 -6.01815999e-01 7.55716622e-01
3.13156337e-01 -1.38930202e-01 7.73142636e-01 1.36085793e-01
-2.85483271e-01 2.36312270e-01 -4.78672981e-01 2.39149123e-01
8.90257955e-01 -1.48189843e+00 4.82283533e-01 6.29126966e-01
3.87239486e-01 2.01580882e-01 9.24746245e-02 -8.01457226e-01
-8.25909972e-01 -1.59255171e+00 7.48348951e-01 -4.34821516e-01
1.00059307e+00 -5.98734081e-01 -1.24890196e+00 8.78493130e-01
1.93885624e-01 2.99554884e-01 6.73784196e-01 4.04844761e-01
-8.52182865e-01 1.07727647e-01 -1.03865480e+00 4.36749071e-01
1.59813499e+00 -7.32534707e-01 -5.38254082e-01 5.98657489e-01
1.11906588e+00 -2.79353231e-01 -9.74174976e-01 5.09143710e-01
1.50254918e-02 -8.91193449e-01 8.27789485e-01 -8.75406265e-01
4.94516850e-01 -3.84419739e-01 -2.09118813e-01 -1.23599923e+00
-4.63700205e-01 -2.66392708e-01 3.63130867e-01 1.50750673e+00
3.63581181e-01 -8.06916773e-01 6.93553448e-01 2.61311024e-01
-7.88882524e-02 -3.23880047e-01 -7.19211698e-01 -6.32310331e-01
2.75674284e-01 -3.87894511e-01 7.31416523e-01 1.50310910e+00
-2.28811681e-01 8.62205803e-01 -1.23728225e-02 5.19053221e-01
7.08160222e-01 6.58397555e-01 7.45615780e-01 -1.64454424e+00
-3.51046294e-01 -4.41940069e-01 -4.71976787e-01 -7.18331873e-01
7.34065652e-01 -1.62251651e+00 -2.01985046e-01 -1.74465501e+00
3.03754240e-01 -5.13323128e-01 -4.33872819e-01 7.64829814e-01
-3.42086047e-01 8.17683935e-02 1.08528540e-01 1.39257431e-01
-8.53837907e-01 2.63471037e-01 1.55453861e+00 -3.11488062e-01
1.36326507e-01 -4.54529166e-01 -8.31911087e-01 1.03689015e+00
7.66042292e-01 -5.85834801e-01 -7.00536489e-01 -5.41418612e-01
4.02511537e-01 -1.97941512e-01 6.60366297e-01 -8.61070633e-01
4.41025496e-02 -1.92604110e-01 9.03444663e-02 -2.88816821e-02
-3.54417831e-01 -7.73367345e-01 1.37500539e-01 1.08702995e-01
-3.23390871e-01 -2.79431313e-01 1.28557906e-01 7.94801712e-01
-3.05085421e-01 -6.41705394e-02 8.04681778e-01 -4.53613728e-01
-1.15897703e+00 4.72909123e-01 3.88695180e-01 6.29999995e-01
9.86434460e-01 1.66329011e-01 -7.93325245e-01 3.91171165e-02
-9.91189718e-01 3.32030207e-01 4.85036641e-01 4.94572550e-01
1.81370676e-02 -1.14761007e+00 -5.83721578e-01 1.27404362e-01
4.73579586e-01 2.28586271e-01 2.70712763e-01 5.06347954e-01
-1.65265754e-01 5.77527061e-02 -3.50261927e-01 -8.42415154e-01
-9.78935599e-01 6.17156148e-01 2.01886848e-01 -9.14732218e-01
-6.45393193e-01 8.48997891e-01 1.06103218e+00 -1.03337920e+00
-1.71377212e-01 -3.02153617e-01 -3.82719904e-01 2.90409327e-01
8.23755413e-02 -1.71168223e-01 2.43975632e-02 -5.35630822e-01
-3.11811656e-01 9.96595263e-01 1.08453475e-01 3.11526388e-01
1.23102510e+00 -9.95034128e-02 -3.46955389e-01 2.67285496e-01
1.15574300e+00 -3.40911090e-01 -1.43044674e+00 -5.08379579e-01
3.99801135e-01 -3.63132581e-02 -3.32695782e-01 -8.95093501e-01
-1.36845636e+00 9.52137291e-01 3.75090271e-01 3.20340872e-01
1.10740805e+00 7.52237737e-01 6.83404028e-01 9.36300009e-02
5.71246386e-01 -8.51828516e-01 1.35920346e-01 5.43342948e-01
6.52066350e-01 -1.53239024e+00 -1.22291036e-01 -8.60636771e-01
-6.72831178e-01 1.05101585e+00 7.11446643e-01 -3.58224288e-02
6.30541503e-01 -6.06541755e-03 1.44156903e-01 -6.90901220e-01
-3.32195222e-01 -6.59304082e-01 4.35055554e-01 7.69540727e-01
4.39032048e-01 4.70351815e-01 -1.81729853e-01 7.37707198e-01
3.94841507e-02 -1.83587343e-01 -1.31547034e-01 4.31049496e-01
-1.48242414e-01 -1.32279646e+00 5.81860133e-02 2.38587007e-01
-2.80449420e-01 -1.30955517e-01 -5.83017647e-01 9.07758057e-01
6.29103661e-01 8.58790755e-01 -1.14600628e-03 -2.28799984e-01
1.56645998e-01 1.34373903e-01 4.04124796e-01 -9.76306617e-01
-5.73820651e-01 -9.95041505e-02 1.03571579e-01 -6.78885102e-01
-8.34057331e-01 -3.58357817e-01 -1.40920269e+00 1.50294930e-01
6.57786727e-02 -5.12210317e-02 3.58223826e-01 1.02186275e+00
4.85339701e-01 7.76567221e-01 -2.56940536e-02 -4.42781419e-01
-2.77873818e-02 -6.93710983e-01 -6.82960987e-01 1.01586771e+00
-1.71742544e-01 -9.27757800e-01 -1.85765445e-01 2.48332947e-01]
|
[9.855098724365234, 1.3299022912979126]
|
5404a666-8e9b-49c9-a56c-24be1af00650
|
towards-safe-continuing-task-reinforcement
|
2102.12585
| null |
https://arxiv.org/abs/2102.12585v1
|
https://arxiv.org/pdf/2102.12585v1.pdf
|
Towards Safe Continuing Task Reinforcement Learning
|
Safety is a critical feature of controller design for physical systems. When designing control policies, several approaches to guarantee this aspect of autonomy have been proposed, such as robust controllers or control barrier functions. However, these solutions strongly rely on the model of the system being available to the designer. As a parallel development, reinforcement learning provides model-agnostic control solutions but in general, it lacks the theoretical guarantees required for safety. Recent advances show that under mild conditions, control policies can be learned via reinforcement learning, which can be guaranteed to be safe by imposing these requirements as constraints of an optimization problem. However, to transfer from learning safety to learning safely, there are two hurdles that need to be overcome: (i) it has to be possible to learn the policy without having to re-initialize the system; and (ii) the rollouts of the system need to be in themselves safe. In this paper, we tackle the first issue, proposing an algorithm capable of operating in the continuing task setting without the need of restarts. We evaluate our approach in a numerical example, which shows the capabilities of the proposed approach in learning safe policies via safe exploration.
|
['Santiago Paternain', 'Luiz F. O. Chamon', 'Miguel Calvo-Fullana']
|
2021-02-24
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[ 2.42197827e-01 5.08664608e-01 -3.43554646e-01 1.65794522e-01
-3.90787929e-01 -6.56568170e-01 6.21104598e-01 2.74420977e-01
-4.97165143e-01 1.15437222e+00 -4.29777056e-01 -4.89237130e-01
-5.78115880e-01 -5.34268260e-01 -7.39589214e-01 -9.53298986e-01
-3.24061990e-01 1.31499857e-01 5.04585028e-01 -3.43125761e-01
3.30334455e-01 6.18899226e-01 -1.67391527e+00 -5.19234002e-01
9.22305822e-01 1.06869459e+00 1.22702591e-01 6.82863295e-01
2.54659235e-01 6.87438488e-01 -3.36586356e-01 4.27203536e-01
5.33226013e-01 -4.52740699e-01 -7.81364024e-01 1.62258863e-01
-8.64439905e-02 -2.86122710e-01 1.53497579e-02 1.07386303e+00
2.34070212e-01 2.82144397e-01 5.01286030e-01 -1.47118711e+00
1.95710629e-01 3.47321302e-01 -2.20257297e-01 -2.62212694e-01
1.62088796e-01 3.99581283e-01 6.79997742e-01 -1.53186664e-01
3.46834213e-01 8.46805394e-01 2.81611204e-01 7.49990940e-01
-1.22855055e+00 -4.08102363e-01 2.73361653e-01 -1.87123880e-01
-1.38572836e+00 -5.01441360e-01 5.72482467e-01 -5.01904130e-01
7.41534650e-01 3.96959215e-01 6.51828110e-01 8.54824364e-01
5.69390297e-01 2.97618777e-01 1.32555640e+00 -5.71520686e-01
8.88331056e-01 4.54314917e-01 -1.21690519e-01 4.61126417e-01
4.52011436e-01 5.05034626e-01 4.15691640e-03 -2.95270652e-01
7.86373377e-01 -3.59007865e-01 -4.27310467e-01 -8.79333258e-01
-8.84302735e-01 7.71978438e-01 1.68130204e-01 1.96663201e-01
-3.87340844e-01 2.40745991e-01 3.52902412e-01 5.72510600e-01
-1.10322356e-01 7.63383985e-01 -4.34634238e-01 -8.73426348e-02
-5.49791515e-01 4.32887733e-01 9.79251027e-01 7.73225248e-01
3.57572645e-01 2.95307040e-01 1.71177626e-01 1.30526707e-01
2.86776513e-01 3.05315197e-01 1.34233549e-01 -9.21966672e-01
2.88369268e-01 4.24576700e-01 5.71034610e-01 -6.98082566e-01
-2.86045492e-01 -2.31177986e-01 -5.09531856e-01 1.00153542e+00
5.51599562e-01 -4.18419093e-01 -6.07806683e-01 1.88744617e+00
5.73034942e-01 -7.71527737e-02 2.37537652e-01 8.97418678e-01
-3.27125818e-01 5.99615216e-01 -9.37158763e-02 -6.78340793e-01
9.38506007e-01 -4.13373798e-01 -8.41528237e-01 -9.97964740e-02
3.31820428e-01 -3.61628830e-01 1.05078363e+00 7.26530671e-01
-1.11720681e+00 -2.74844199e-01 -1.48049617e+00 7.73375809e-01
-1.90632924e-01 -2.48512894e-01 2.18620628e-01 7.23735332e-01
-9.41372693e-01 7.49929428e-01 -9.89198446e-01 -4.44386005e-01
-1.74724549e-01 7.51072049e-01 -2.12514117e-01 5.72241783e-01
-1.14046657e+00 1.21014845e+00 7.65316129e-01 1.64918974e-01
-1.12820613e+00 -3.24371219e-01 -6.60248280e-01 8.54773074e-02
1.01295626e+00 -4.53982025e-01 1.29895961e+00 -1.08574283e+00
-1.94931316e+00 9.01775360e-02 4.84544158e-01 -4.35816437e-01
8.84569466e-01 -2.03560457e-01 -1.38492182e-01 1.55964717e-01
-3.88405830e-01 1.39113307e-01 1.10285234e+00 -1.50173533e+00
-7.93822050e-01 6.86021000e-02 6.18911207e-01 1.14099436e-01
-5.07093191e-01 -2.42193818e-01 7.23090246e-02 -2.18273684e-01
-3.01578164e-01 -1.25026202e+00 -5.47612667e-01 7.94508681e-02
-2.50357032e-01 -6.31774366e-02 7.57016122e-01 -1.25032067e-01
1.24121749e+00 -2.04797530e+00 2.17740014e-01 4.93914545e-01
-1.17945172e-01 3.65314275e-01 1.76250070e-01 5.31774044e-01
-5.77927865e-02 -4.31346633e-02 -2.95098841e-01 1.09286308e-01
8.75386968e-02 3.51217121e-01 -5.68695843e-01 6.69375062e-01
3.17854136e-01 1.93023682e-01 -8.24345887e-01 -4.40095812e-01
2.23768532e-01 2.29896814e-01 -5.49368560e-01 3.86451006e-01
-5.02307117e-01 6.20687783e-01 -8.62805605e-01 -9.20926258e-02
2.35771254e-01 2.13794485e-01 4.38395768e-01 4.30875540e-01
-4.83295828e-01 9.76365581e-02 -1.63027823e+00 1.00532305e+00
-4.48798895e-01 -1.06766582e-01 5.96378922e-01 -1.27823400e+00
7.19048500e-01 5.68812132e-01 5.29445469e-01 -4.12405670e-01
3.68309826e-01 3.02880138e-01 5.94011098e-02 -3.86740863e-01
9.42748860e-02 -3.74558866e-01 -2.13209659e-01 3.13587219e-01
-3.18721473e-01 -3.71872216e-01 9.23834890e-02 -2.94647872e-01
1.02478576e+00 3.20238769e-01 6.36282086e-01 -7.59929717e-01
1.02421975e+00 1.43685758e-01 7.00794339e-01 5.42744756e-01
-7.17139319e-02 -1.17812060e-01 7.72259533e-01 -1.67059347e-01
-9.00372803e-01 -7.68022597e-01 -1.38680479e-02 5.33685148e-01
1.05344981e-01 -1.34988755e-01 -9.07150805e-01 -6.55909479e-01
-1.11933410e-01 6.95092022e-01 -5.66340446e-01 -3.75070661e-01
-7.30545163e-01 -3.02383035e-01 1.91030994e-01 2.45964408e-01
5.81987575e-02 -8.35704565e-01 -1.46256638e+00 3.60099137e-01
4.09101933e-01 -8.61322403e-01 -3.75435948e-01 6.81568027e-01
-8.97248507e-01 -1.19378769e+00 -2.25497618e-01 -4.63656574e-01
7.98450291e-01 -7.75436684e-02 4.87541229e-01 3.59333098e-01
-1.19262032e-01 3.92045707e-01 -2.80069076e-02 -1.84682548e-01
-7.01224327e-01 8.84214863e-02 4.41702604e-01 -1.61056384e-01
-5.64795732e-01 -4.67586279e-01 -2.85795748e-01 4.26774174e-01
-1.08706427e+00 -9.63708460e-02 5.26527226e-01 8.42832446e-01
4.81988072e-01 5.66633523e-01 8.35861325e-01 -5.25991857e-01
6.49889946e-01 -1.30576894e-01 -1.35322654e+00 2.53068268e-01
-8.19305956e-01 4.51492310e-01 1.19345772e+00 -5.45734644e-01
-8.22025359e-01 5.30880868e-01 6.92859367e-02 -2.80744672e-01
-1.91403121e-01 3.01706791e-01 -3.63054603e-01 -3.00063699e-01
4.42523003e-01 1.31472811e-01 4.52210903e-01 -9.77804363e-02
2.26126298e-01 3.99944067e-01 1.47445157e-01 -8.92791033e-01
1.00252223e+00 1.32078782e-01 3.47951800e-01 -9.05158460e-01
-4.44514334e-01 -1.35047972e-01 -6.45530224e-01 -3.41809303e-01
6.40072048e-01 -3.73614192e-01 -1.26968253e+00 -2.80232523e-02
-6.35443211e-01 -4.47935432e-01 -3.55680346e-01 2.21349210e-01
-1.11732924e+00 3.08934599e-01 -8.98154825e-02 -1.36380172e+00
-3.99165191e-02 -1.10466611e+00 4.72396135e-01 1.73543632e-01
-3.16022694e-01 -8.54234457e-01 1.51387498e-01 -3.15725207e-01
4.37112480e-01 5.98544478e-01 8.68050575e-01 -3.73102576e-01
-4.64506894e-01 -1.25734329e-01 5.35853922e-01 3.14977050e-01
2.34633654e-01 -4.80744755e-03 -6.94075942e-01 -7.63170362e-01
4.26826745e-01 -4.24391001e-01 2.21088275e-01 1.87385738e-01
8.69819582e-01 -7.18195915e-01 -3.40190887e-01 1.95569649e-01
1.49427629e+00 5.81774473e-01 3.96056443e-01 5.00926912e-01
2.38551691e-01 8.57788384e-01 8.62691164e-01 5.54475963e-01
1.36774378e-02 7.81703532e-01 7.13971674e-01 1.93497166e-01
5.11221349e-01 -6.09452166e-02 4.98458296e-01 3.75505626e-01
-6.61930442e-02 1.40956804e-01 -8.26113462e-01 2.93167442e-01
-2.13412881e+00 -7.64566004e-01 1.39619529e-01 2.81243396e+00
8.48163724e-01 5.00582814e-01 3.34901661e-01 4.61080939e-01
4.39914465e-01 -2.61015207e-01 -6.29339755e-01 -6.42161548e-01
4.93533671e-01 -8.20669383e-02 5.01759410e-01 6.90494120e-01
-1.14384329e+00 5.93942702e-01 6.06076288e+00 5.15121877e-01
-1.27137113e+00 -2.43096128e-01 1.55704394e-01 1.75298713e-02
-8.20719171e-03 2.92540163e-01 -5.94568729e-01 3.13707054e-01
1.01351845e+00 -3.19971353e-01 5.06244779e-01 9.92284775e-01
4.55717176e-01 -3.47430229e-01 -1.03659606e+00 3.15991402e-01
-3.96405756e-01 -5.25202453e-01 -4.43712950e-01 7.23349899e-02
4.08824503e-01 -7.38775551e-01 -7.40830675e-02 4.24473703e-01
7.92575926e-02 -9.38118398e-01 8.97143126e-01 3.12740415e-01
4.21496719e-01 -1.21950448e+00 4.89601463e-01 7.58450091e-01
-1.03804612e+00 -4.07047808e-01 -1.23540930e-01 -1.79309025e-01
8.78644958e-02 2.00714484e-01 -6.20178640e-01 5.78298151e-01
2.21503749e-01 7.01477975e-02 -1.65759251e-01 1.01538312e+00
-2.90658295e-01 3.18352729e-01 -3.92701238e-01 -2.61517316e-01
3.87630939e-01 -1.74242303e-01 5.10495424e-01 7.20525086e-01
2.27382928e-01 1.17455110e-01 6.41659439e-01 7.57744074e-01
8.29628587e-01 -1.28477588e-01 -7.85296738e-01 -1.34148911e-01
1.23358168e-01 1.08675146e+00 -8.36097777e-01 3.76316942e-02
-1.26480967e-01 5.32621145e-01 2.30404839e-01 1.31577194e-01
-1.06055486e+00 -3.98431957e-01 7.81289697e-01 2.25204661e-01
1.87819317e-01 -5.23343146e-01 -1.85808986e-01 -7.22444713e-01
-1.61717180e-03 -1.04206336e+00 2.69496530e-01 1.93986762e-02
-8.32451999e-01 4.03654516e-01 2.01549113e-01 -1.36943340e+00
-5.73609293e-01 -5.13440967e-01 -3.63963991e-01 6.39926732e-01
-1.56278300e+00 -6.11980796e-01 1.60960212e-01 7.45360076e-01
2.24876896e-01 1.04136325e-01 8.03081810e-01 -4.35049422e-02
-5.24237216e-01 2.34030604e-01 -1.00097053e-01 -5.60445130e-01
5.69947958e-01 -1.34713721e+00 -2.21310377e-01 8.41466784e-01
-5.31215370e-01 6.28073931e-01 1.12876570e+00 -5.74417114e-01
-1.73289216e+00 -8.05977345e-01 5.29551506e-01 -1.33676201e-01
7.38246381e-01 -3.79964769e-01 -9.70021248e-01 4.58733886e-01
8.01260397e-02 -8.61444101e-02 -3.80534381e-02 -1.68338805e-01
2.07161605e-01 -1.23982511e-01 -9.86918092e-01 6.66174352e-01
4.89843786e-01 -9.72915515e-02 -5.45952320e-01 1.66244119e-01
5.48953354e-01 -3.04740638e-01 -7.51141667e-01 4.46896583e-01
4.64760214e-01 -7.29035676e-01 6.38096452e-01 -5.84001184e-01
-1.24272309e-01 -6.57988787e-01 1.71605781e-01 -1.25118613e+00
-1.13614827e-01 -9.82819974e-01 -1.82128489e-01 9.18282926e-01
3.09647918e-01 -7.41123438e-01 5.95336020e-01 8.24240804e-01
-6.31071106e-02 -7.74109483e-01 -1.24764514e+00 -1.27720547e+00
2.16174185e-01 -2.51761287e-01 2.41820797e-01 7.71255195e-01
4.60772157e-01 1.85203910e-01 -5.98287106e-01 5.27875423e-01
3.49002928e-01 -8.25702697e-02 6.27463043e-01 -1.03460276e+00
-5.70718110e-01 -4.22617733e-01 -1.39055364e-02 -4.98901814e-01
3.13070118e-01 -2.83804566e-01 6.11747921e-01 -1.22968638e+00
-2.53253013e-01 -6.47751629e-01 -3.00645739e-01 6.09345376e-01
2.69565824e-03 -6.32757723e-01 2.08506659e-01 -2.18862798e-02
-3.97053897e-01 6.23745561e-01 1.07672393e+00 1.18748643e-01
-4.27377194e-01 3.42257768e-01 -3.62571329e-01 6.43419623e-01
9.95734870e-01 -3.32606077e-01 -8.06269825e-01 1.03930913e-01
1.94850460e-01 5.73047459e-01 1.36131048e-01 -1.10670614e+00
1.80647999e-01 -6.70233428e-01 -1.46417171e-01 -6.13805726e-02
2.62171596e-01 -1.43370593e+00 3.45152855e-01 1.17527342e+00
-2.87754983e-01 1.28711730e-01 2.59933919e-01 6.42017603e-01
6.30096719e-03 -4.32259262e-01 1.07455158e+00 2.39368081e-01
-5.13782024e-01 -5.09023257e-02 -7.48360336e-01 -2.22990572e-01
1.50979745e+00 -7.92650357e-02 1.90743245e-02 -2.64216781e-01
-5.29238045e-01 4.98516768e-01 5.60486138e-01 4.30029780e-01
2.88169414e-01 -9.06174600e-01 -8.95978808e-02 1.13512553e-01
1.54829975e-02 -2.78868318e-01 -1.01208650e-01 8.78392100e-01
-2.04818487e-01 5.26069105e-01 -3.77266198e-01 -3.78721267e-01
-9.92392898e-01 1.06588733e+00 4.02587563e-01 -2.11099833e-01
-6.10115707e-01 1.05257869e-01 9.71220285e-02 -1.88968211e-01
4.02279466e-01 -4.45387542e-01 -1.80712909e-01 -1.57098532e-01
4.86701012e-01 5.65329455e-02 2.01720726e-02 -3.46320868e-02
-5.18471837e-01 5.83623767e-01 1.32852837e-01 -2.55528092e-01
1.08262730e+00 1.56616420e-02 1.44464955e-01 4.28307503e-01
4.34666038e-01 -2.20362678e-01 -1.55221188e+00 2.55758822e-01
3.74393225e-01 -2.80997187e-01 1.19599119e-01 -5.36375105e-01
-6.89110696e-01 5.76758862e-01 4.75280285e-01 4.54477042e-01
1.23538768e+00 -6.16361976e-01 2.94656336e-01 5.32395720e-01
7.57348537e-01 -1.40753770e+00 4.24498506e-02 4.76173431e-01
7.94795096e-01 -9.27508891e-01 -7.09970444e-02 -3.69689912e-01
-5.95253944e-01 1.29854178e+00 7.26168036e-01 -3.09112638e-01
5.58013916e-01 5.59464037e-01 -1.09468460e-01 2.15728700e-01
-8.23115528e-01 -1.69503137e-01 1.68508291e-01 4.22542900e-01
1.26054779e-01 -1.67572647e-01 -6.30263388e-01 2.48813108e-01
2.26641268e-01 -7.12528899e-02 5.54640114e-01 1.31313789e+00
-7.94289470e-01 -1.18311965e+00 -5.22672415e-01 -1.02644637e-01
-2.69608587e-01 5.02164900e-01 -1.62364691e-01 1.28482807e+00
-7.44589642e-02 9.41550136e-01 -4.71229851e-01 -1.67216621e-02
5.62183619e-01 4.18229848e-02 4.67734754e-01 -5.69950163e-01
-5.15061378e-01 1.27279699e-01 1.62089571e-01 -7.13884652e-01
-1.71183392e-01 -5.71936429e-01 -1.30613244e+00 1.17289729e-01
-3.85309666e-01 3.32124949e-01 4.86623913e-01 1.00241208e+00
1.73660624e-03 6.38030887e-01 7.32576013e-01 -6.10315919e-01
-1.35418439e+00 -4.08715665e-01 -7.02209532e-01 -1.28309786e-01
6.01819992e-01 -8.62502515e-01 -3.73814791e-01 -1.19105317e-01]
|
[4.740832805633545, 2.093233346939087]
|
fbb2d805-e633-4136-991e-93c305bc3b19
|
natural-logic-guided-autoregressive-multi-hop
|
2212.05276
| null |
https://arxiv.org/abs/2212.05276v1
|
https://arxiv.org/pdf/2212.05276v1.pdf
|
Natural Logic-guided Autoregressive Multi-hop Document Retrieval for Fact Verification
|
A key component of fact verification is thevevidence retrieval, often from multiple documents. Recent approaches use dense representations and condition the retrieval of each document on the previously retrieved ones. The latter step is performed over all the documents in the collection, requiring storing their dense representations in an index, thus incurring a high memory footprint. An alternative paradigm is retrieve-and-rerank, where documents are retrieved using methods such as BM25, their sentences are reranked, and further documents are retrieved conditioned on these sentences, reducing the memory requirements. However, such approaches can be brittle as they rely on heuristics and assume hyperlinks between documents. We propose a novel retrieve-and-rerank method for multi-hop retrieval, that consists of a retriever that jointly scores documents in the knowledge source and sentences from previously retrieved documents using an autoregressive formulation and is guided by a proof system based on natural logic that dynamically terminates the retrieval process if the evidence is deemed sufficient. This method is competitive with current state-of-the-art methods on FEVER, HoVer and FEVEROUS-S, while using $5$ to $10$ times less memory than competing systems. Evaluation on an adversarial dataset indicates improved stability of our approach compared to commonly deployed threshold-based methods. Finally, the proof system helps humans predict model decisions correctly more often than using the evidence alone.
|
['Andreas Vlachos', 'Rami Aly']
|
2022-12-10
| null | null | null | null |
['fact-verification']
|
['natural-language-processing']
|
[ 2.88028777e-01 -5.92755601e-02 -3.78748536e-01 -8.21438897e-03
-1.29092276e+00 -8.23997974e-01 9.23955083e-01 6.29511833e-01
-4.31903571e-01 8.62907648e-01 -6.44963905e-02 -4.15465295e-01
-4.60275054e-01 -1.02992666e+00 -8.18951786e-01 -5.34986198e-01
-7.24350214e-02 6.70915902e-01 5.25237620e-01 -2.34858155e-01
7.04691947e-01 4.09285396e-01 -1.60841584e+00 2.72065639e-01
5.88140786e-01 1.06364489e+00 -1.50669292e-01 7.88228154e-01
2.67337896e-02 9.72796857e-01 -7.81639457e-01 -6.66412413e-01
1.85986042e-01 -6.02520593e-02 -8.87108207e-01 -4.50232804e-01
4.07546967e-01 -8.06381106e-01 -2.33394518e-01 9.84530985e-01
2.92868286e-01 6.41590878e-02 5.86534381e-01 -9.74053979e-01
-5.69683194e-01 7.86279559e-01 -3.25557232e-01 2.10627258e-01
9.24530268e-01 -3.40395302e-01 1.26609123e+00 -7.79544115e-01
6.44011259e-01 1.17626119e+00 2.27637008e-01 3.86182964e-01
-1.12713861e+00 -7.58359432e-01 1.34897634e-01 3.37282628e-01
-1.48485518e+00 -4.45585042e-01 5.31632781e-01 -2.07057133e-01
1.23987007e+00 4.90408093e-01 3.67402554e-01 9.65976834e-01
3.16396981e-01 8.26081276e-01 8.73485863e-01 -6.10008121e-01
4.00749862e-01 2.07884133e-01 5.23417413e-01 8.50447178e-01
5.36190271e-01 2.80321926e-01 -8.80387425e-01 -7.83340752e-01
1.28814295e-01 2.09078148e-01 -2.08300889e-01 -8.29265863e-02
-7.41532445e-01 8.17371428e-01 1.43850863e-01 1.98354170e-01
-6.04605436e-01 2.56787717e-01 3.97043020e-01 4.51626867e-01
3.91050071e-01 3.38889629e-01 -2.81351030e-01 9.28763002e-02
-1.47555554e+00 5.67034841e-01 8.81351233e-01 6.54864192e-01
6.39643848e-01 -5.26913226e-01 -3.44948322e-01 3.66247445e-01
5.79161346e-01 6.81304157e-01 4.28763479e-01 -6.11534774e-01
5.47403872e-01 6.41125023e-01 2.05288067e-01 -1.29547286e+00
2.00945035e-01 -1.25437081e-01 -6.10759139e-01 1.06889801e-02
-2.17189059e-01 4.57817882e-01 -9.22037184e-01 1.57886708e+00
8.97566229e-02 2.42143437e-01 1.37399018e-01 6.90905213e-01
3.86994809e-01 5.80286741e-01 1.32275326e-02 -1.83788493e-01
1.18886459e+00 -4.85548288e-01 -4.88689214e-01 -7.16727152e-02
4.74362940e-01 -6.43855095e-01 6.80871367e-01 8.08153987e-01
-1.30038762e+00 1.75510310e-02 -1.39683998e+00 1.73342526e-01
-6.72212660e-01 -8.60120803e-02 5.67770302e-01 7.54458308e-01
-1.17193794e+00 7.96696305e-01 -5.80794334e-01 6.00531437e-02
1.76388085e-01 3.34557116e-01 -1.47612646e-01 -1.61163956e-01
-1.71973467e+00 9.99838471e-01 4.59402025e-01 5.62906079e-02
-1.16769850e+00 -4.78206903e-01 -5.80326021e-01 2.44343027e-01
5.65928817e-01 -6.64393425e-01 1.08283007e+00 -1.00030176e-01
-1.29290700e+00 6.28300786e-01 -1.44378081e-01 -8.80100131e-01
4.48642015e-01 -5.42224467e-01 -4.73892361e-01 6.92910135e-01
-2.14641672e-02 1.82159945e-01 1.23065186e+00 -1.16185176e+00
-3.84877324e-01 -1.79108217e-01 3.48296613e-01 -1.91273227e-01
-4.06681538e-01 1.16725840e-01 -5.66539526e-01 -4.49106485e-01
6.50726706e-02 -8.99435163e-01 -9.53370854e-02 -3.04490715e-01
-5.80407381e-01 -2.68822223e-01 5.64875066e-01 -6.06024683e-01
1.65989625e+00 -1.83707190e+00 6.80293962e-02 7.59528160e-01
1.29003882e-01 3.69826078e-01 -7.41571188e-02 5.91425121e-01
2.72231400e-01 4.21298057e-01 2.41496097e-02 -2.47248217e-01
6.21186532e-02 -6.02524355e-02 -1.05835152e+00 3.19843352e-01
6.62293732e-02 6.36138260e-01 -8.76512229e-01 -7.09971249e-01
1.98040977e-02 3.01483423e-01 -5.20804584e-01 3.37267108e-02
-3.89489740e-01 -4.28536981e-01 -6.68049097e-01 6.29148364e-01
4.88586217e-01 -4.73062783e-01 1.43439889e-01 2.44842142e-01
4.44839597e-01 5.39426506e-01 -1.23618734e+00 1.53852201e+00
-3.19488794e-01 1.60453469e-01 -1.40348598e-01 -7.01741815e-01
7.99231529e-01 4.38256115e-01 9.95394662e-02 -5.75488269e-01
2.58989981e-03 3.32316488e-01 -6.57148123e-01 -3.58075976e-01
9.12507176e-01 4.28103050e-03 -2.71990269e-01 8.82495582e-01
-2.88199782e-01 -3.68889838e-01 5.64501762e-01 9.69143450e-01
1.42274547e+00 -1.52676553e-01 1.56826451e-01 1.93043187e-01
5.76532722e-01 -7.37301856e-02 8.57315361e-02 1.27442122e+00
2.39281386e-01 2.35871047e-01 2.11000353e-01 -6.58139214e-02
-6.31486237e-01 -1.10689080e+00 8.63495693e-02 9.32480097e-01
2.61854470e-01 -8.55131686e-01 -5.62134326e-01 -8.47661674e-01
3.19687039e-01 8.79499078e-01 -5.00449955e-01 -4.12417740e-01
-2.95076698e-01 -2.56358832e-01 7.83593833e-01 4.32847768e-01
1.61293849e-01 -9.46868896e-01 -8.74098122e-01 2.11164579e-01
-2.46068791e-01 -5.42839050e-01 -2.17111800e-02 5.12814373e-02
-7.48798907e-01 -1.15742421e+00 -4.88692790e-01 -1.35332555e-01
7.08868563e-01 1.40955821e-01 1.05718410e+00 5.57159543e-01
-1.93661287e-01 4.74891871e-01 -3.94067645e-01 -2.39994869e-01
-5.02546668e-01 -1.80400252e-01 1.00682043e-01 -3.06274295e-01
3.86956722e-01 -2.78630435e-01 -4.96297896e-01 6.40303865e-02
-1.25109076e+00 -5.15058458e-01 8.22033942e-01 9.77289677e-01
5.54044187e-01 2.93941051e-01 5.64957321e-01 -9.56686854e-01
8.96129549e-01 -5.15868247e-01 -7.78913856e-01 6.64276779e-01
-1.15716290e+00 3.01276177e-01 4.58557993e-01 -4.67782646e-01
-7.54368365e-01 -4.25597042e-01 4.41265792e-01 -6.30431294e-01
3.02910924e-01 7.77275741e-01 3.11046213e-01 1.45758003e-01
6.74962223e-01 3.54391158e-01 -3.69796157e-01 -2.42420301e-01
4.40885752e-01 5.57966411e-01 3.84651840e-01 -7.18222797e-01
9.34907973e-01 2.85233945e-01 -3.77284996e-02 -1.93501070e-01
-6.50530159e-01 -5.24025440e-01 -1.91380948e-01 -1.99135900e-01
2.21923172e-01 -6.55431271e-01 -8.68543565e-01 5.74838929e-02
-1.25934827e+00 2.56639451e-01 -1.31627947e-01 4.25074995e-01
-2.32127428e-01 4.42365021e-01 -7.40103245e-01 -1.20421433e+00
-6.67931259e-01 -8.49286020e-01 1.25184155e+00 -8.31579044e-02
-4.82878983e-01 -6.42266095e-01 7.18739405e-02 3.79847765e-01
3.96232635e-01 -9.56163332e-02 1.17240989e+00 -9.90993619e-01
-1.03254390e+00 -1.04986942e+00 -5.54802567e-02 2.15836748e-01
-3.27710629e-01 -6.63695671e-03 -8.40237975e-01 -3.88629109e-01
-2.16100603e-01 -6.13883853e-01 1.17427278e+00 -5.67713045e-02
8.76339614e-01 -5.17123103e-01 -5.62538564e-01 -2.21303001e-01
1.41010845e+00 9.00932178e-02 5.93110621e-01 3.94111693e-01
-1.01671815e-02 3.77135843e-01 7.77030408e-01 5.50175846e-01
-2.34362334e-02 2.42273003e-01 3.07852387e-01 4.45185363e-01
2.57827580e-01 -3.46707910e-01 5.14317036e-01 6.22031689e-01
1.10634975e-01 -4.63195592e-01 -8.25168431e-01 5.74877560e-01
-1.76879692e+00 -1.17207706e+00 4.55874622e-01 2.37997794e+00
1.02876496e+00 5.27913034e-01 -3.12312722e-01 5.92033505e-01
5.73843360e-01 2.25500658e-01 -3.31960946e-01 -2.87570029e-01
5.07922061e-02 3.68540287e-01 3.19947004e-01 5.50131321e-01
-7.16429353e-01 8.95088434e-01 7.17188072e+00 9.00069237e-01
-8.72959971e-01 -1.78869158e-01 3.36716026e-01 -4.59980667e-01
-5.93511343e-01 2.58222222e-01 -9.42502022e-01 2.27642953e-01
1.03764665e+00 -4.60817277e-01 4.70054418e-01 8.47526252e-01
-2.71015108e-01 -5.01047015e-01 -1.23845983e+00 6.30303562e-01
4.81291205e-01 -1.37640202e+00 4.00266916e-01 -1.09235011e-01
4.08368796e-01 -3.16140532e-01 1.83251426e-01 2.60414779e-01
4.23911899e-01 -7.57807374e-01 8.47077191e-01 8.28492165e-01
5.37685633e-01 -6.68416679e-01 7.10669339e-01 3.34729373e-01
-8.78247917e-01 -1.15286037e-01 -2.58631229e-01 2.93348461e-01
6.61237314e-02 8.56539607e-01 -1.12921667e+00 8.37628007e-01
6.24085546e-01 2.87170529e-01 -5.62974632e-01 7.27894068e-01
-4.43700761e-01 5.90093672e-01 -4.63923812e-01 -4.29396480e-01
-1.04079563e-02 2.05288395e-01 7.13756025e-01 1.14554358e+00
2.48155981e-01 -7.63194542e-03 1.91112190e-01 7.00691879e-01
-1.22879572e-01 6.41680211e-02 -6.34431899e-01 -2.45378301e-01
7.96831012e-01 8.91094208e-01 -6.98061645e-01 -8.49581122e-01
1.26935884e-01 7.19479144e-01 1.57364875e-01 3.16430330e-01
-6.61926210e-01 -4.69483674e-01 -9.05921832e-02 5.06991111e-02
3.31822485e-01 4.59939949e-02 9.53088254e-02 -8.61234367e-01
5.15597999e-01 -8.78580391e-01 6.84958816e-01 -7.22451389e-01
-1.20392883e+00 5.97002685e-01 1.84155688e-01 -1.18514705e+00
-7.24168718e-01 -2.50772953e-01 -1.99690223e-01 7.15446234e-01
-1.66710985e+00 -7.84459054e-01 2.53550887e-01 6.67260528e-01
1.96419343e-01 -1.85544312e-01 1.19104469e+00 1.68683708e-01
-2.51204014e-01 5.83122015e-01 -6.70554563e-02 -9.29365978e-02
7.13870645e-01 -1.04090357e+00 -6.16152696e-02 8.57840240e-01
2.60975003e-01 1.27632880e+00 5.89324176e-01 -1.07372594e+00
-1.55579031e+00 -7.28284597e-01 1.19704449e+00 -3.63330573e-01
8.78517091e-01 -1.94140270e-01 -8.69758248e-01 3.91933292e-01
-1.62187498e-02 -2.84329742e-01 6.01754427e-01 2.62739569e-01
-8.41158092e-01 -8.95729586e-02 -1.17831230e+00 5.29350758e-01
5.24540246e-01 -9.74363029e-01 -8.92140746e-01 3.95062685e-01
5.79227686e-01 -2.08132267e-01 -5.34752369e-01 2.77290493e-01
6.28334343e-01 -7.23695993e-01 9.79246736e-01 -6.24140739e-01
5.18752813e-01 -3.59400928e-01 -3.05326790e-01 -8.40842545e-01
-1.81302633e-02 -8.46269786e-01 -6.02696955e-01 1.12612939e+00
7.11583614e-01 -6.01357579e-01 5.91841996e-01 8.95233452e-01
4.42116529e-01 -7.06481993e-01 -9.03726399e-01 -8.92341912e-01
-2.91417867e-01 -4.95613635e-01 7.23208606e-01 6.83772147e-01
2.77389079e-01 1.47217885e-01 -2.72172809e-01 2.02853993e-01
5.50256193e-01 3.80955189e-01 5.09705782e-01 -1.28520024e+00
-4.32671547e-01 -3.29631269e-01 -4.13138479e-01 -8.00396621e-01
3.50679845e-01 -1.12396097e+00 1.45695001e-01 -1.45880544e+00
4.04319078e-01 -4.96693969e-01 -6.95505321e-01 7.00989366e-01
-2.17481807e-01 -1.57490402e-01 7.43781924e-02 4.02179182e-01
-8.92497241e-01 2.70112872e-01 5.42630315e-01 -5.19376993e-01
-2.00126823e-02 -1.59113288e-01 -5.37879348e-01 4.61954266e-01
5.24185538e-01 -7.45103002e-01 -4.41286981e-01 -1.06613174e-01
5.27242064e-01 2.74857551e-01 6.13351643e-01 -8.26809645e-01
8.03110719e-01 3.14107426e-02 3.24644148e-01 -1.01449633e+00
5.11445940e-01 -7.86059022e-01 -5.33447005e-02 4.97362286e-01
-7.49237061e-01 1.72478378e-01 3.72779630e-02 9.78146493e-01
-1.96485713e-01 -5.45674264e-01 2.31804192e-01 2.94894930e-02
-2.63865292e-01 1.12454943e-01 -4.53126609e-01 -2.51376808e-01
9.00374949e-01 1.35099471e-01 -5.78980863e-01 -3.90565753e-01
-4.30441082e-01 1.25940740e-01 1.28923446e-01 2.85451114e-01
1.02380025e+00 -1.12399578e+00 -6.34267986e-01 8.10871646e-02
2.23225318e-02 -1.82148859e-01 2.55736690e-02 5.03577411e-01
-1.35185838e-01 6.93948567e-01 4.09442425e-01 -3.93625319e-01
-1.28143489e+00 8.47622216e-01 -2.67078400e-01 -8.89728546e-01
-5.37475049e-01 8.35508287e-01 -3.97812217e-01 2.48325482e-01
3.31595898e-01 -2.61644572e-02 -1.08636543e-01 1.68186918e-01
7.68668234e-01 3.78236055e-01 2.93203950e-01 -1.16498649e-01
-6.70459449e-01 3.75111997e-01 -7.10498571e-01 -6.70785666e-01
1.17041600e+00 1.88773319e-01 -2.92719424e-01 2.96440929e-01
9.12684858e-01 3.82306665e-01 -3.86077374e-01 -3.93955439e-01
2.62655139e-01 -6.28887534e-01 1.03219658e-01 -1.01440489e+00
-7.31163859e-01 4.92880344e-01 2.67754257e-01 4.51318145e-01
1.11737347e+00 -3.52904461e-02 7.80979931e-01 9.52911794e-01
7.22562194e-01 -1.19532967e+00 2.20542356e-01 2.28875235e-01
9.96546507e-01 -8.14816475e-01 4.77967769e-01 -2.00170353e-01
-3.05511355e-01 9.26300049e-01 4.58487235e-02 -2.91303694e-01
6.15745485e-01 2.25739643e-01 -3.02181095e-01 -3.87152195e-01
-1.07268476e+00 3.06315511e-01 3.18367153e-01 5.99872656e-02
4.40401062e-02 -2.28220802e-02 -1.87391564e-01 6.26352787e-01
-1.31504461e-01 2.40573391e-01 2.72081643e-01 1.37339175e+00
-4.03671175e-01 -1.18133128e+00 -4.92653161e-01 5.36913335e-01
-6.58594370e-01 -4.27713037e-01 -5.97970784e-01 4.16632295e-01
-4.07839656e-01 1.33707142e+00 -3.81792843e-01 -5.30367255e-01
1.92767248e-01 2.76567638e-01 4.14699286e-01 -5.01335859e-01
-7.36020029e-01 -2.56371796e-02 2.91782886e-01 -6.03496313e-01
-2.91164935e-01 -5.71490347e-01 -1.26931596e+00 -3.40476811e-01
-7.91227102e-01 4.42182064e-01 7.16553748e-01 9.19331908e-01
4.82845306e-01 2.86506712e-01 7.11237788e-01 -4.81283993e-01
-9.66678739e-01 -6.26866162e-01 -4.75371003e-01 3.17167133e-01
3.48435640e-01 -7.32538283e-01 -6.20164037e-01 -3.37593704e-02]
|
[11.49457836151123, 7.623379707336426]
|
fa82566c-5a7d-4f73-b2c9-6dda99e087c1
|
conjectures-tests-and-proofs-an-overview-of
|
2109.03721
| null |
https://arxiv.org/abs/2109.03721v1
|
https://arxiv.org/pdf/2109.03721v1.pdf
|
Conjectures, Tests and Proofs: An Overview of Theory Exploration
|
A key component of mathematical reasoning is the ability to formulate interesting conjectures about a problem domain at hand. In this paper, we give a brief overview of a theory exploration system called QuickSpec, which is able to automatically discover interesting conjectures about a given set of functions. QuickSpec works by interleaving term generation with random testing to form candidate conjectures. This is made tractable by starting from small sizes and ensuring that only terms that are irreducible with respect to already discovered conjectures are considered. QuickSpec has been successfully applied to generate lemmas for automated inductive theorem proving as well as to generate specifications of functional programs. We give an overview of typical use-cases of QuickSpec, as well as demonstrating how to easily connect it to a theorem prover of the user's choice.
|
['Nicholas Smallbone', 'Moa Johansson']
|
2021-09-07
| null | null | null | null |
['automated-theorem-proving', 'mathematical-reasoning', 'automated-theorem-proving']
|
['miscellaneous', 'natural-language-processing', 'reasoning']
|
[ 2.70865589e-01 6.15960956e-01 -1.28696173e-01 9.81941074e-02
-6.56500638e-01 -8.60469699e-01 4.38069016e-01 2.86650509e-01
5.36923468e-01 9.55397606e-01 -3.37118477e-01 -1.40306020e+00
-2.17514619e-01 -1.07815790e+00 -8.92936587e-01 2.11380459e-02
-7.00245619e-01 6.04996085e-01 4.59974408e-01 -2.16274545e-01
3.87433022e-01 2.54607975e-01 -1.89352727e+00 4.01604325e-01
1.02255070e+00 6.03172064e-01 -1.44685417e-01 8.06668460e-01
-2.46350572e-01 8.68097067e-01 -2.90418237e-01 -4.83858794e-01
2.99334228e-01 -2.84496009e-01 -1.37350154e+00 -1.05055027e-01
1.19238332e-01 -7.98337460e-02 2.25437075e-01 1.33234859e+00
-4.08649266e-01 -4.02663797e-01 4.12384234e-02 -2.08027291e+00
3.67890634e-02 1.05907738e+00 -1.81841515e-02 -3.56433019e-02
8.70227695e-01 2.92221010e-01 1.35871828e+00 -4.07427162e-01
5.87263584e-01 1.42392850e+00 4.99893278e-01 4.24285710e-01
-1.73215032e+00 -7.36247361e-01 -1.26361683e-01 3.95174831e-01
-1.53475881e+00 -4.72310990e-01 7.63700783e-01 -3.85674596e-01
1.13598049e+00 8.71505916e-01 6.42796814e-01 2.93554872e-01
3.06663454e-01 6.70108020e-01 1.02010489e+00 -9.15819883e-01
4.25935775e-01 4.79312658e-01 3.80224407e-01 8.67514908e-01
6.11021221e-01 1.60142630e-01 -1.99353635e-01 -8.89558434e-01
5.00809848e-01 -6.92360342e-01 -7.58606121e-02 -3.88434559e-01
-1.26180696e+00 1.00705755e+00 -3.67943823e-01 3.73782158e-01
1.19483225e-01 3.18207085e-01 3.51258546e-01 7.20397949e-01
-6.85231760e-02 1.09415185e+00 -8.36755276e-01 5.00856079e-02
-8.04452956e-01 7.95595706e-01 1.41656256e+00 1.06665528e+00
8.49622130e-01 -1.25010505e-01 5.61773814e-02 -1.74962103e-01
3.70137334e-01 3.38855803e-01 -1.59194916e-01 -1.25680280e+00
1.70313656e-01 8.22043777e-01 5.78362197e-02 -8.36838543e-01
-4.64051776e-02 -4.04964685e-01 -7.77247846e-02 3.42814207e-01
3.29595834e-01 -1.83926269e-01 -2.75159717e-01 1.66125953e+00
2.93744206e-01 2.49209434e-01 9.72295627e-02 3.15894544e-01
3.22689831e-01 5.82858026e-01 -3.33719075e-01 -6.01700127e-01
1.15423441e+00 -2.95040190e-01 -2.38394633e-01 -1.16780223e-02
8.98390472e-01 -3.26154411e-01 7.35351682e-01 8.04834843e-01
-1.20528233e+00 -7.61670619e-02 -9.95063424e-01 1.33128956e-01
-1.33019820e-01 -3.99110466e-01 1.42240870e+00 7.93726861e-01
-1.23005807e+00 3.67534637e-01 -5.64925849e-01 3.36583108e-01
9.19800475e-02 7.75033414e-01 7.61765465e-02 -2.39875361e-01
-1.30718076e+00 5.77846169e-01 6.47501111e-01 -2.23227948e-01
-9.09236848e-01 -1.02149117e+00 -1.02100909e+00 1.20644003e-01
1.08724594e+00 -7.38129735e-01 1.67611039e+00 -6.53462172e-01
-1.13803136e+00 4.42524672e-01 -3.82581264e-01 -8.08800340e-01
1.82330340e-01 4.93566185e-01 -5.31962633e-01 1.64911170e-02
2.04731539e-01 2.87207395e-01 2.61900127e-01 -9.22205865e-01
-6.19340241e-01 -1.90562923e-02 1.02236331e+00 -6.44681931e-01
2.40795285e-01 3.40842366e-01 1.46554157e-01 6.36091456e-02
1.23718530e-01 -8.16834152e-01 -4.76927847e-01 -1.16131760e-01
-7.26329148e-01 -5.19797266e-01 5.80268085e-01 -6.39866740e-02
1.25063026e+00 -1.55647230e+00 -5.84641211e-02 7.92780042e-01
4.30566341e-01 -6.04423136e-02 1.64310366e-01 6.20433092e-01
-4.16939259e-01 5.63260019e-01 6.70373067e-02 6.48082733e-01
4.81678784e-01 5.08095957e-02 -5.85133135e-01 1.46127537e-01
5.17157733e-01 9.90929902e-01 -8.65867734e-01 -5.83451569e-01
1.08650930e-01 -4.50128675e-01 -9.03342724e-01 6.46021515e-02
-1.27789509e+00 -8.19984749e-02 -4.88747209e-01 6.28071368e-01
7.65553534e-01 -3.36766601e-01 3.62048417e-01 4.04137850e-01
-3.66335809e-01 9.95108128e-01 -1.49060535e+00 1.16231775e+00
-4.62761790e-01 4.21455294e-01 -7.92791694e-02 -8.48383665e-01
6.58050835e-01 3.04437608e-01 4.47123237e-02 -2.27879971e-01
6.80031627e-02 3.25516462e-01 2.30492577e-01 -4.92667049e-01
3.78528014e-02 -5.43065012e-01 -4.83547151e-02 7.81879127e-01
-1.81754231e-01 -3.37465525e-01 6.62368655e-01 4.58405614e-01
1.30223179e+00 -2.77045131e-01 4.20991093e-01 -5.96549690e-01
9.87353384e-01 4.56093609e-01 5.56836188e-01 6.68527007e-01
6.99283540e-01 -2.60028481e-01 1.19932485e+00 -5.44805884e-01
-8.86922538e-01 -6.80690765e-01 -3.08534168e-02 6.28904879e-01
-1.56048834e-01 -1.04183364e+00 -4.13733751e-01 -6.26702905e-01
-1.60513222e-02 1.24855733e+00 -4.97066170e-01 -1.20084666e-01
-2.34769821e-01 3.84068377e-02 4.60122496e-01 1.46938682e-01
2.00571150e-01 -9.01955545e-01 -7.72987962e-01 -3.76719646e-02
-1.43328205e-01 -8.47786248e-01 2.18388885e-01 2.91482568e-01
-8.07273269e-01 -1.64335573e+00 4.41966563e-01 -5.86221337e-01
9.13778961e-01 7.31399730e-02 1.13414562e+00 5.60517788e-01
-2.80917257e-01 1.65879339e-01 -2.18860507e-02 -5.09962440e-01
-8.86711419e-01 -9.27390456e-02 -2.04188764e-01 -6.59526408e-01
2.62639999e-01 -5.76741636e-01 4.64222103e-01 1.69718251e-01
-1.01527166e+00 2.48472095e-01 3.69016975e-01 6.66678250e-01
8.83969292e-02 8.31336796e-01 5.56280874e-02 -1.00421095e+00
6.77871883e-01 -4.40756053e-01 -1.34313631e+00 6.61282003e-01
-6.46974683e-01 7.68965423e-01 9.32333648e-01 -3.32157671e-01
-4.92945045e-01 -1.31323943e-02 1.83057785e-02 -6.32838160e-02
-1.31433263e-01 9.50717390e-01 -3.45337093e-01 -1.93467274e-01
6.04564488e-01 2.20647395e-01 -2.81813871e-02 -6.76190183e-02
1.95067599e-01 1.82147443e-01 3.14661682e-01 -1.32755888e+00
1.41447484e+00 -1.96554318e-01 3.72944444e-01 -4.11345273e-01
-4.79980230e-01 -1.72365054e-01 -1.91717342e-01 9.12578925e-02
-4.03283417e-01 -3.20727050e-01 -1.39836454e+00 -2.09639832e-01
-1.16240335e+00 -4.40311193e-01 -3.07020843e-01 1.17011905e-01
-5.66855848e-01 1.69024736e-01 9.48987156e-02 -1.20263231e+00
-1.77962631e-01 -1.14040565e+00 7.21289158e-01 -4.50414717e-02
-5.40504277e-01 -8.56592953e-01 1.08269922e-01 -7.38360882e-02
1.08488761e-01 2.71019042e-01 1.34369981e+00 -8.96045148e-01
-8.83529842e-01 -2.30201840e-01 -7.36335814e-02 -6.97328225e-02
-1.24907859e-01 1.28598943e-01 -6.67151809e-01 -1.68806925e-01
8.68093371e-02 -2.71379590e-01 7.04627335e-02 -4.11868505e-02
1.07305300e+00 -8.96060050e-01 -4.20918286e-01 2.52765477e-01
1.29617369e+00 -1.33360878e-01 7.66431034e-01 3.87813240e-01
-3.02412689e-01 3.09699774e-01 5.72495878e-01 2.24760130e-01
2.21125826e-01 4.35054839e-01 2.49871656e-01 3.97883654e-01
5.54443181e-01 -3.63067180e-01 1.00443237e-01 -1.43888354e-01
-6.18644170e-02 3.76252681e-01 -1.06284750e+00 5.46892107e-01
-1.49636412e+00 -1.11919928e+00 -3.99032623e-01 2.21893859e+00
1.23811603e+00 5.78671873e-01 2.50742525e-01 6.28019869e-01
4.58772004e-01 -6.09798729e-01 -2.60612309e-01 -7.15268016e-01
2.89892316e-01 6.98561549e-01 2.94807911e-01 7.79449344e-01
-5.16414523e-01 6.91369236e-01 7.13645220e+00 5.72038472e-01
-8.06082308e-01 -3.51628721e-01 7.83764943e-02 2.96544433e-01
-9.03456092e-01 9.73294795e-01 -6.47103786e-01 -1.81095093e-01
1.02399397e+00 -1.11556554e+00 6.73127592e-01 1.08177519e+00
-1.64457913e-02 -2.52723515e-01 -1.42665792e+00 3.52849476e-02
-2.88044363e-01 -1.51829088e+00 -7.91620463e-02 1.43810753e-02
5.59969783e-01 -6.02268517e-01 -2.87154615e-01 4.10925806e-01
5.42977035e-01 -1.18871951e+00 7.14266896e-01 9.22504887e-02
6.57737255e-01 -9.87766027e-01 6.08412325e-01 6.63644433e-01
-8.27976882e-01 -2.37873271e-01 -8.55368450e-02 -6.57266915e-01
-4.75803703e-01 5.73985398e-01 -1.59600616e+00 5.44742823e-01
3.04149743e-02 5.99003024e-02 -5.48057735e-01 1.39079857e+00
-5.14895022e-01 7.36590922e-01 -7.07022905e-01 -3.75829905e-01
-4.19142805e-02 2.44124979e-01 6.54184878e-01 9.50314283e-01
3.62174064e-02 2.92963207e-01 2.59160809e-02 1.69719112e+00
3.21888268e-01 -2.65172392e-01 -4.57693368e-01 -1.03809528e-01
3.38305593e-01 9.23540711e-01 -4.86363769e-01 -4.31681514e-01
-3.52227420e-01 -1.90786994e-03 -5.59468381e-02 1.40618205e-01
-7.75050759e-01 -5.59636950e-01 4.07020181e-01 3.64931256e-01
1.47054300e-01 -3.67651731e-01 -3.12317848e-01 -1.09586692e+00
1.73632964e-01 -1.47293961e+00 3.15090179e-01 -7.04247594e-01
-4.92090195e-01 3.19316477e-01 4.53268498e-01 -8.18930924e-01
-5.07441163e-01 -5.09610891e-01 -8.46790731e-01 9.34880137e-01
-1.18531144e+00 -9.13328826e-01 2.96844542e-01 5.51960647e-01
3.36383522e-01 2.19709560e-01 1.13971245e+00 -4.18165624e-01
-1.87539861e-01 5.14375746e-01 -6.99022293e-01 -3.03275079e-01
-1.24316953e-01 -1.22506154e+00 4.71345693e-01 1.08381510e+00
1.66798308e-01 1.33995485e+00 1.36558044e+00 -6.50798500e-01
-2.11650586e+00 -6.63362443e-01 1.18969548e+00 1.80746950e-02
1.24822760e+00 -3.46945256e-01 -8.00462186e-01 7.31147945e-01
-1.46042258e-01 -2.33155504e-01 3.54455501e-01 5.36548257e-01
-6.29458129e-01 -6.84154779e-02 -1.02205753e+00 6.59027696e-01
7.67039359e-01 -4.94626790e-01 -8.25090647e-01 6.41753197e-01
7.69390702e-01 -4.84790266e-01 -7.27061570e-01 2.26648599e-01
5.42128623e-01 -7.61862338e-01 6.62018418e-01 -9.23022568e-01
3.23246390e-01 -7.35513031e-01 -3.15387994e-02 -6.23201132e-01
-1.58404961e-01 -1.49957907e+00 -3.62908632e-01 9.51245666e-01
9.48969185e-01 -7.68595994e-01 4.22518045e-01 1.24717867e+00
1.02650739e-01 -6.76216185e-01 -7.62628317e-01 -8.13373983e-01
-3.58335003e-02 -1.02611971e+00 9.51929033e-01 7.45820165e-01
1.18861592e+00 3.83466154e-01 1.45275807e-02 3.85820597e-01
5.28701782e-01 7.03289390e-01 8.83511961e-01 -1.37559688e+00
-6.46161735e-01 -2.74195611e-01 -4.48869526e-01 -5.67975342e-01
3.21441114e-01 -9.67119813e-01 -1.16158366e-01 -1.15385318e+00
3.10899228e-01 -5.70550144e-01 2.04206809e-01 9.03775752e-01
3.06222409e-01 -1.75709978e-01 -2.48591438e-01 -2.16210529e-01
-6.44237399e-01 -1.88317254e-01 1.00545812e+00 -2.43977562e-01
-1.37516066e-01 5.60974360e-01 -1.25968063e+00 6.33592129e-01
7.64447153e-01 -3.79685163e-01 -4.20164764e-01 2.61649609e-01
1.06947565e+00 4.43532914e-01 6.86888099e-01 -9.00468588e-01
3.48741055e-01 -7.62556076e-01 -4.79235381e-01 -3.74359608e-01
-4.62465078e-01 -6.53263330e-01 5.99307001e-01 6.65338814e-01
-6.85744286e-01 -9.41995531e-02 6.08220577e-01 -4.96316440e-02
5.02053685e-02 -7.14515150e-01 2.16615409e-01 -3.44861858e-02
-6.38518274e-01 -1.70492172e-01 -2.96714306e-01 1.76370535e-02
1.16086411e+00 1.80598736e-01 -2.92558342e-01 -2.69582778e-01
-4.26909149e-01 4.99568492e-01 6.58483744e-01 1.41538745e-02
5.23305416e-01 -8.76866519e-01 -5.78048527e-01 3.58621389e-01
8.16126168e-02 -7.66941607e-02 -3.67571443e-01 8.80187809e-01
-3.72564495e-01 8.40555310e-01 2.62265623e-01 -1.95526674e-01
-1.47496140e+00 9.69641984e-01 2.32247442e-01 -3.67275029e-01
-4.90275383e-01 4.71678048e-01 -1.28441080e-01 -1.39148802e-01
3.14150304e-02 -8.70698571e-01 3.20572287e-01 -1.06306171e+00
8.89287651e-01 1.71280950e-02 2.33795606e-02 1.72505170e-01
-7.17362881e-01 1.54492915e-01 1.47361279e-01 -1.03116482e-01
1.36494625e+00 3.25436980e-01 -7.52363205e-01 2.25518525e-01
8.01822424e-01 3.01468045e-01 -1.78678289e-01 -7.23751262e-02
1.27265304e-01 -3.28341782e-01 -1.20499887e-01 -8.93636763e-01
-4.64889199e-01 4.58452433e-01 -4.38535869e-01 7.22325742e-01
1.14596140e+00 3.18478674e-01 3.94362390e-01 9.72477317e-01
8.31029356e-01 -4.50638384e-01 -5.75549483e-01 4.38272119e-01
7.85639584e-01 -6.54306293e-01 3.39242190e-01 -8.04263234e-01
-2.08720624e-01 1.52665305e+00 3.20227623e-01 -6.79518208e-02
1.96110815e-01 7.83669174e-01 -6.43082738e-01 -1.38109282e-01
-1.40603352e+00 -1.16922840e-01 1.97282478e-01 3.86111349e-01
1.93404719e-01 -1.33944169e-01 -3.02171111e-01 5.45943677e-01
-7.23276138e-01 4.73669082e-01 7.72628129e-01 9.37087893e-01
-5.48723936e-01 -1.50819135e+00 -8.43316793e-01 3.10646623e-01
-3.43174934e-01 -2.82923162e-01 -5.28516054e-01 8.85792971e-01
1.74617246e-01 1.12732005e+00 -5.98847270e-01 -4.47708130e-01
-1.73911810e-01 1.06401250e-01 7.67692447e-01 -8.34457755e-01
-7.99485371e-02 -2.05643043e-01 8.39630902e-01 -4.29921120e-01
-2.07221895e-01 -7.60607243e-01 -1.21835756e+00 -5.14675438e-01
-8.31795096e-01 9.57901180e-01 3.98026526e-01 1.30041480e+00
-8.29402730e-02 2.49178782e-01 7.19980478e-01 8.97797123e-02
-6.00256741e-01 -5.86032629e-01 -3.93271565e-01 -4.87400204e-01
2.85436869e-01 -4.27835435e-01 -4.04838085e-01 1.04858413e-01]
|
[8.817025184631348, 6.894043445587158]
|
c77d849a-c984-472c-9653-67f50c88f246
|
synthetic-demographic-data-generation-for
|
2306.17109
| null |
https://arxiv.org/abs/2306.17109v1
|
https://arxiv.org/pdf/2306.17109v1.pdf
|
Synthetic Demographic Data Generation for Card Fraud Detection Using GANs
|
Using machine learning models to generate synthetic data has become common in many fields. Technology to generate synthetic transactions that can be used to detect fraud is also growing fast. Generally, this synthetic data contains only information about the transaction, such as the time, place, and amount of money. It does not usually contain the individual user's characteristics (age and gender are occasionally included). Using relatively complex synthetic demographic data may improve the complexity of transaction data features, thus improving the fraud detection performance. Benefiting from developments of machine learning, some deep learning models have potential to perform better than other well-established synthetic data generation methods, such as microsimulation. In this study, we built a deep-learning Generative Adversarial Network (GAN), called DGGAN, which will be used for demographic data generation. Our model generates samples during model training, which we found important to overcame class imbalance issues. This study can help improve the cognition of synthetic data and further explore the application of synthetic data generation in card fraud detection.
|
['John Hawkin', 'Charles Robertson', 'Xianta Jiang', 'Terrence Tricco', 'Shuo Wang']
|
2023-06-29
| null | null | null | null |
['synthetic-data-generation', 'synthetic-data-generation', 'fraud-detection']
|
['medical', 'miscellaneous', 'miscellaneous']
|
[-2.39588302e-02 1.70332775e-01 -4.55354387e-03 -4.92112607e-01
-2.56638110e-01 -1.17539197e-01 5.41027784e-01 3.24782342e-01
-3.19818437e-01 1.03747702e+00 4.73950319e-02 -1.03410870e-01
7.57924497e-01 -1.61998057e+00 -7.75580287e-01 -5.28237164e-01
1.26238927e-01 7.80127525e-01 -2.74488717e-01 -5.42096198e-01
2.93560475e-01 1.62707075e-01 -1.43513548e+00 5.60185373e-01
1.13391399e+00 9.40141559e-01 -4.31254596e-01 2.52529740e-01
-2.47185096e-01 8.56490493e-01 -1.23289871e+00 -9.74696040e-01
2.34416351e-01 -7.98617482e-01 -2.89121568e-01 -3.93341422e-01
-2.72981018e-01 -7.11633682e-01 -1.59773752e-01 8.96318078e-01
5.07938504e-01 -4.16717976e-01 8.04002047e-01 -1.36485267e+00
-7.17259109e-01 9.56386685e-01 -5.48266649e-01 -1.73331261e-01
3.26494068e-01 1.55848756e-01 4.66397822e-01 -4.21559185e-01
5.88896036e-01 1.22363043e+00 9.07884896e-01 6.65024757e-01
-1.02390373e+00 -9.91450965e-01 -5.95551968e-01 -7.18253106e-02
-8.69589567e-01 7.13280812e-02 8.85431349e-01 -3.48127425e-01
5.30257344e-01 1.24601923e-01 1.09744716e+00 1.79260767e+00
1.28529608e-01 1.05381370e+00 9.29122567e-01 -5.15618265e-01
2.13916406e-01 2.85505891e-01 -1.90620288e-01 4.22212064e-01
9.82231140e-01 9.53444541e-02 -3.87088984e-01 -3.51058930e-01
8.94128263e-01 2.87750125e-01 1.90277889e-01 -1.91003188e-01
-1.12793434e+00 1.23061097e+00 2.42739469e-01 2.27168649e-01
-4.32140797e-01 3.05233300e-01 6.43534303e-01 4.08936858e-01
4.93326992e-01 6.76797748e-01 -2.06894502e-01 -2.78409541e-01
-7.99819946e-01 7.73503661e-01 8.67999077e-01 7.12965310e-01
6.58238649e-01 4.32054728e-01 -2.42462620e-01 8.76484215e-01
-3.46731991e-02 6.99384809e-01 8.25533211e-01 -6.50870025e-01
7.91013658e-01 1.08499360e+00 2.50703007e-01 -1.06014621e+00
-3.03693146e-01 -2.18669102e-01 -1.17330587e+00 4.42242809e-02
6.63521409e-01 -5.54991841e-01 -9.16846037e-01 1.40911281e+00
-1.28831482e-03 -2.41480559e-01 5.13578579e-02 6.06702328e-01
5.85937798e-01 4.01585519e-01 -9.56652164e-02 1.92333907e-01
1.18033969e+00 -3.43858272e-01 -6.18388116e-01 -9.41620301e-03
8.35245013e-01 -4.39623386e-01 8.42080653e-01 2.39710987e-01
-8.27372551e-01 -3.60816181e-01 -1.09601974e+00 2.55111516e-01
-5.99658906e-01 -9.09392722e-03 1.10786319e+00 1.31980324e+00
-4.25766051e-01 6.00795150e-01 -8.39113951e-01 -8.05822238e-02
8.53552222e-01 2.51744539e-01 -1.31008834e-01 -4.42841016e-02
-1.64208484e+00 6.88860655e-01 5.64333916e-01 -4.94392104e-02
-4.74868268e-01 -5.68761230e-01 -9.48144019e-01 7.98944235e-02
-1.02661319e-01 -9.09112811e-01 9.74367201e-01 -1.26406181e+00
-1.19627309e+00 8.05334985e-01 4.25345898e-01 -1.00964844e+00
9.47106004e-01 -1.98861258e-03 -3.04033637e-01 -2.05086321e-01
1.02908626e-01 6.09942377e-01 7.51922190e-01 -9.43419158e-01
-2.97675759e-01 -4.44584370e-01 -2.28803605e-01 -2.29627758e-01
-3.59547168e-01 -1.36722907e-01 -2.92909835e-02 -1.23263729e+00
-1.87083110e-01 -7.92263329e-01 -2.02964887e-01 -4.06845897e-01
-4.08757657e-01 1.52121395e-01 6.34245932e-01 -8.59100699e-01
1.12671733e+00 -1.79093909e+00 -4.43753719e-01 2.41188705e-01
-1.92394674e-01 5.35011172e-01 2.80428708e-01 6.18246496e-01
1.26360700e-01 4.08769608e-01 -4.10104811e-01 -2.98053563e-01
-5.95740899e-02 -1.10727623e-01 -3.58959675e-01 1.60870329e-01
4.58875477e-01 1.12114477e+00 -7.05945134e-01 -3.40657741e-01
1.89449973e-02 2.98728764e-01 -8.38819087e-01 2.36105576e-01
-3.04764092e-01 2.97111660e-01 -3.84743243e-01 8.71668160e-01
9.43752050e-01 2.47897416e-01 -4.59337011e-02 1.58925697e-01
5.56541383e-01 1.25435770e-01 -9.47567701e-01 1.23630202e+00
-2.61726677e-01 4.57654744e-01 -4.58165228e-01 -1.17973042e+00
1.20440972e+00 1.64491944e-02 6.08510673e-01 -9.76685226e-01
1.48674741e-01 2.09894031e-01 4.55864631e-02 -3.83517802e-01
6.27021492e-01 -1.67629823e-01 -5.78988671e-01 7.25777030e-01
-2.26550028e-01 -3.09896737e-01 1.69754475e-01 -2.84861699e-02
7.27181792e-01 3.99214216e-02 1.20233536e-01 1.96143255e-01
7.56361037e-02 6.66076243e-02 7.25797534e-01 6.98768914e-01
8.67291614e-02 7.97142923e-01 9.93566930e-01 -7.05951571e-01
-1.35245955e+00 -8.11736166e-01 1.25083447e-01 3.38796586e-01
-9.32739675e-02 -1.13420412e-01 -1.04193676e+00 -6.43582702e-01
3.80526602e-01 6.50205195e-01 -8.30342650e-01 -5.29575586e-01
-5.51225364e-01 -1.47536612e+00 1.00976682e+00 4.86039132e-01
8.26424539e-01 -1.52840996e+00 -5.42953372e-01 3.66882950e-01
-2.79091567e-01 -6.26964986e-01 -1.35433733e-01 -1.93002477e-01
-1.12343061e+00 -1.08903575e+00 -8.73021066e-01 -5.19090891e-01
7.28939950e-01 -4.31575477e-01 1.15497231e+00 1.42440289e-01
-2.64178216e-01 -4.08385336e-01 -5.85822046e-01 -1.12567532e+00
-9.26351666e-01 2.76516557e-01 -3.46172675e-02 1.32169873e-01
5.66334367e-01 -4.67984796e-01 -6.89409375e-01 2.13185683e-01
-1.06939137e+00 2.40995094e-01 4.45968121e-01 1.10759223e+00
1.15084887e-01 -3.09429150e-02 9.72179115e-01 -1.65971994e+00
1.03311038e+00 -5.60393929e-01 -2.12669194e-01 1.05891312e-02
-6.35676324e-01 1.37089863e-01 7.06197023e-01 -3.97080362e-01
-1.04488587e+00 -4.37398374e-01 -9.64216068e-02 -1.08252674e-01
5.34667633e-02 4.17971253e-01 -1.51918575e-01 4.13501382e-01
6.90232515e-01 2.18989879e-01 2.97482967e-01 -3.30575794e-01
-8.35686177e-02 8.35506856e-01 1.23288617e-01 -4.76198405e-01
5.12504816e-01 3.36013734e-01 -1.93151817e-01 -4.09737438e-01
-1.45639360e-01 1.68312237e-01 -3.00799251e-01 2.68543869e-01
2.89901674e-01 -8.74625564e-01 -6.92810178e-01 1.40738213e+00
-9.22759593e-01 -3.03186208e-01 -3.61101925e-01 3.56321365e-01
-4.18103009e-01 1.08411796e-01 -8.62907410e-01 -7.61640608e-01
-6.23873591e-01 -7.73396552e-01 7.21328378e-01 2.15727583e-01
-2.90851533e-01 -8.84385049e-01 4.76295315e-02 4.50146317e-01
5.92598677e-01 9.26513374e-01 1.01882720e+00 -7.23139167e-01
-3.61456871e-01 -6.15087450e-01 -7.67153651e-02 5.69281101e-01
1.35139525e-01 -2.91830413e-02 -6.89368904e-01 -2.00719580e-01
1.15213186e-01 -4.56243783e-01 9.58818376e-01 2.23670632e-01
1.37754381e+00 -4.56237644e-01 -2.60315716e-01 6.50510132e-01
1.19328070e+00 4.16668087e-01 1.10497296e+00 3.70258331e-01
6.93676293e-01 6.07150376e-01 6.93773329e-01 6.30419672e-01
3.89999181e-01 5.30934989e-01 4.06820506e-01 -3.56221080e-01
2.37766266e-01 -6.71428382e-01 3.08635265e-01 3.54027689e-01
-2.16232553e-01 -1.52338848e-01 -8.73197556e-01 4.22720373e-01
-1.60405159e+00 -1.23799443e+00 -1.92488953e-02 2.03874564e+00
8.47178042e-01 3.45873088e-01 3.09194118e-01 3.75501931e-01
7.30058551e-01 -1.44913658e-01 -7.09909320e-01 -5.55689216e-01
-2.72203267e-01 4.44383562e-01 5.15087068e-01 -2.56618679e-01
-9.01085317e-01 6.44116163e-01 5.91697788e+00 6.17821097e-01
-1.10915005e+00 -2.44159788e-01 1.17394972e+00 4.94884774e-02
-5.59929907e-01 -3.45850140e-01 -5.61628282e-01 1.14658689e+00
8.16978753e-01 -1.24909207e-01 2.99350411e-01 8.88339698e-01
1.81635126e-01 -2.11880431e-01 -1.12395740e+00 9.93704200e-01
9.46864784e-02 -1.36086833e+00 4.61272568e-01 1.65272087e-01
8.91589701e-01 -3.44357044e-01 -1.04188919e-03 6.16608679e-01
3.00288230e-01 -1.20858598e+00 4.93031681e-01 5.32286525e-01
7.84028292e-01 -1.16769230e+00 1.39501369e+00 4.25548553e-01
-3.74265194e-01 -2.43722834e-02 -1.80795997e-01 -3.08700521e-02
-7.69097283e-02 6.80865228e-01 -1.08566046e+00 4.48385984e-01
6.45621955e-01 5.71897566e-01 -8.08553696e-01 8.22056413e-01
2.82158535e-02 7.06682205e-01 -1.89683259e-01 -1.56538934e-01
-3.09407592e-01 -5.30032456e-01 -3.06776147e-02 9.90574598e-01
6.20405436e-01 -2.83515006e-01 -4.55903500e-01 1.10277152e+00
-2.97884077e-01 1.33289114e-01 -8.37494969e-01 -2.17474729e-01
3.14669102e-01 5.63221097e-01 -5.71753025e-01 -2.47785971e-01
-1.08123407e-01 9.26621079e-01 3.95857319e-02 1.07768290e-01
-9.03020561e-01 -6.97595477e-01 6.56709969e-01 4.88004625e-01
1.12185016e-01 2.07279354e-01 -7.47299194e-01 -1.48440540e+00
2.09647149e-01 -1.22058332e+00 3.04101795e-01 -4.38223809e-01
-1.23711538e+00 2.15731382e-01 -2.69849896e-01 -1.31559074e+00
-5.96340179e-01 -3.80769521e-01 -7.60775506e-01 8.83525312e-01
-8.39113593e-01 -1.14291656e+00 -1.66356489e-01 3.58824402e-01
2.05502436e-01 -6.25685453e-01 9.37488854e-01 2.62072057e-01
-5.57798982e-01 1.06014991e+00 2.24818453e-01 7.68301249e-01
6.64525688e-01 -1.03708065e+00 8.63846898e-01 4.96808171e-01
-8.81861299e-02 6.15198672e-01 5.45862079e-01 -7.81874955e-01
-1.01991582e+00 -1.20821452e+00 7.31751204e-01 -4.61206138e-01
1.43363386e-01 -6.49264395e-01 -7.91714966e-01 6.12632573e-01
-1.50844604e-01 -4.20705497e-01 6.36782348e-01 -1.11872524e-01
-7.81645104e-02 -4.04854476e-01 -1.70126224e+00 5.00574410e-01
6.64886177e-01 -2.80276597e-01 -4.19938594e-01 -1.11579904e-02
5.08339882e-01 -4.73304093e-01 -8.04968596e-01 5.31457067e-01
6.06914103e-01 -1.32246256e+00 8.79055738e-01 -5.19730151e-01
7.97445416e-01 3.04633200e-01 3.45390767e-01 -1.57328403e+00
3.17288429e-01 -3.88889849e-01 -3.87902677e-01 1.46813095e+00
5.42753100e-01 -9.78648067e-01 1.32192469e+00 7.01951981e-01
3.51070911e-01 -8.46632421e-01 -5.27650297e-01 -3.87411714e-01
2.65898287e-01 -2.48737276e-01 1.28443027e+00 1.10275662e+00
6.10920563e-02 -1.21997356e-01 -5.97239017e-01 -5.44885516e-01
4.98177499e-01 1.45476997e-01 1.09755409e+00 -1.20467579e+00
-1.95943296e-01 -4.07975346e-01 -5.01910150e-01 -2.16723666e-01
-7.79544562e-02 -5.43592572e-01 -7.46971130e-01 -1.30432045e+00
4.57441956e-02 -6.83200240e-01 -6.90365657e-02 2.36589998e-01
-3.33267003e-01 3.71622741e-01 6.41729757e-02 6.49729604e-03
1.59123242e-01 6.40323341e-01 1.35920441e+00 -3.05309802e-01
-1.74034998e-01 4.70694959e-01 -8.43131304e-01 3.64036739e-01
1.09205079e+00 -5.38761854e-01 -2.51843274e-01 -1.31235540e-01
3.39644402e-01 8.43237564e-02 7.57743567e-02 -8.64823222e-01
-3.27668339e-01 -1.51116639e-01 8.44761610e-01 -3.84472966e-01
1.43331975e-01 -4.83188003e-01 3.99009973e-01 7.95292497e-01
-6.90551177e-02 7.66555145e-02 -7.89243504e-02 2.79058337e-01
-5.17453969e-01 -1.89277545e-01 4.69201893e-01 -3.67725372e-01
-6.84361458e-02 1.89811081e-01 -1.97573885e-01 2.95277797e-02
1.10750246e+00 -4.37840581e-01 -2.60019839e-01 -6.52169466e-01
-2.30453268e-01 2.41098270e-01 8.49945366e-01 4.63586122e-01
4.87538338e-01 -1.44040322e+00 -8.68777990e-01 5.84717035e-01
1.66005239e-01 1.28293112e-01 2.04065815e-02 3.08578789e-01
-1.10880578e+00 2.43669540e-01 -7.12852061e-01 -2.05284759e-01
-6.91094160e-01 4.52477902e-01 3.18805218e-01 -5.16122758e-01
-3.01740527e-01 4.70616698e-01 -2.11149037e-01 -6.30360603e-01
-5.78534529e-02 -4.80073720e-01 -1.83840647e-01 2.50181228e-01
2.20076069e-01 4.55054134e-01 2.78783534e-02 -1.36518970e-01
1.28778245e-03 -3.83333396e-03 -8.75647143e-02 4.79892232e-02
1.51709390e+00 4.54543263e-01 -4.68921512e-02 2.89317936e-01
8.67071569e-01 -8.02587792e-02 -1.05643094e+00 1.99781492e-01
-4.07854915e-02 -4.63123173e-01 -6.65527105e-01 -7.10158646e-01
-1.28858650e+00 7.49478459e-01 3.13974440e-01 4.45751637e-01
9.56527054e-01 -6.41160369e-01 1.11742783e+00 1.45490691e-01
8.71872962e-01 -1.20193839e+00 2.98702944e-04 2.08853975e-01
7.66619921e-01 -1.53327560e+00 -2.17048228e-01 -6.53787106e-02
-8.35753202e-01 1.01548183e+00 7.07378507e-01 -2.68773109e-01
2.41988555e-01 2.82969087e-01 5.50367646e-02 4.12437916e-02
-3.52494240e-01 4.23384339e-01 -3.81274223e-01 7.85674632e-01
4.59843487e-01 3.36878419e-01 -5.53768992e-01 8.65875959e-01
-6.94085479e-01 1.61767662e-01 7.81396747e-01 6.42247140e-01
2.05016226e-01 -1.66566253e+00 -4.14748698e-01 1.00869191e+00
-7.44793236e-01 7.31202513e-02 -4.33972955e-01 8.65112305e-01
4.64953929e-01 5.15357256e-01 2.60536730e-01 -3.52750808e-01
3.17652076e-01 1.39824361e-01 3.61505181e-01 -6.09382868e-01
-8.87236297e-01 -7.53601015e-01 2.19284564e-01 -1.03646882e-01
-1.32729217e-01 -7.84581423e-01 -9.16338682e-01 -7.73919702e-01
-1.42719403e-01 1.92746669e-01 6.39038324e-01 4.57172185e-01
1.81763679e-01 2.24006951e-01 6.63368583e-01 -5.92226148e-01
-5.59940755e-01 -1.33429122e+00 -6.85104668e-01 1.01809943e+00
4.90231439e-02 -3.90340537e-01 6.11573132e-03 1.66836575e-01]
|
[8.852909088134766, 4.281287670135498]
|
efc1889e-d704-4173-b794-f04546378e35
|
defx-at-semeval-2020-task-6-joint-extraction
|
2103.1709
| null |
https://arxiv.org/abs/2103.17090v1
|
https://arxiv.org/pdf/2103.17090v1.pdf
|
Defx at SemEval-2020 Task 6: Joint Extraction of Concepts and Relations for Definition Extraction
|
Definition Extraction systems are a valuable knowledge source for both humans and algorithms. In this paper we describe our submissions to the DeftEval shared task (SemEval-2020 Task 6), which is evaluated on an English textbook corpus. We provide a detailed explanation of our system for the joint extraction of definition concepts and the relations among them. Furthermore we provide an ablation study of our model variations and describe the results of an error analysis.
|
['Leonhard Hennig', 'Robert Schwarzenberg', 'Christoph Alt', 'Marc Hübner']
|
2021-03-31
| null |
https://aclanthology.org/2020.semeval-1.92
|
https://aclanthology.org/2020.semeval-1.92.pdf
|
semeval-2020
|
['definition-extraction']
|
['natural-language-processing']
|
[ 2.87693232e-01 4.56296772e-01 -1.69834509e-01 -4.56604511e-01
-7.48502314e-01 -8.72305751e-01 1.10187149e+00 2.57925332e-01
-8.52322340e-01 1.18860447e+00 1.16886802e-01 -5.17062783e-01
-3.96937281e-01 -6.76156461e-01 -2.35304818e-01 1.44895211e-01
1.52431488e-01 8.55523944e-01 3.58637601e-01 -4.80431437e-01
2.14818090e-01 -2.53404229e-04 -1.24150133e+00 4.36196297e-01
6.40494168e-01 3.77295136e-01 8.12382326e-02 6.65172577e-01
-3.65540713e-01 8.41818869e-01 -9.85210538e-01 -9.78720009e-01
-8.18600059e-02 -1.52342111e-01 -1.41010225e+00 -5.29813349e-01
5.61059058e-01 1.00512095e-01 1.16524473e-01 9.91247714e-01
3.42574179e-01 3.36658806e-01 7.83857584e-01 -1.10908723e+00
-4.36531752e-01 1.49315906e+00 -5.24207391e-03 3.68497372e-01
6.70389175e-01 -3.55520904e-01 1.38813400e+00 -1.04755616e+00
1.21753025e+00 8.21093976e-01 7.82590151e-01 7.67837524e-01
-1.03993773e+00 -4.38583881e-01 -1.75299972e-01 6.65882051e-01
-1.69715929e+00 -5.17975509e-01 4.67124939e-01 -3.44458640e-01
1.94633925e+00 5.59777021e-01 4.36992675e-01 1.00955963e+00
-2.37123847e-01 8.20266068e-01 8.85075688e-01 -9.81962621e-01
-8.36167708e-02 3.11582178e-01 7.94258535e-01 6.58443987e-01
6.41998470e-01 -4.41599414e-02 -5.45407891e-01 -5.91128320e-02
6.61628783e-01 -1.02150297e+00 -4.20957536e-01 8.39005262e-02
-1.02469623e+00 5.56584477e-01 -2.38839388e-01 5.05992651e-01
-1.67200297e-01 1.59095705e-01 4.16011691e-01 1.55059755e-01
2.03385904e-01 1.12119687e+00 -1.12206483e+00 -3.01058620e-01
-1.02987599e+00 6.32196188e-01 1.25875115e+00 1.50200748e+00
4.30156499e-01 -4.95473415e-01 1.71215147e-01 7.39324570e-01
6.12071855e-03 2.45268852e-01 3.78332108e-01 -7.94888854e-01
3.53483260e-01 4.41148788e-01 1.80318609e-01 -6.58743918e-01
-4.80940849e-01 -3.08385313e-01 -3.05095881e-01 -4.66518819e-01
9.39300656e-02 -4.32168722e-01 -5.72117209e-01 1.49727869e+00
-3.52901332e-02 6.33063242e-02 1.23541504e-01 2.37778336e-01
1.44226003e+00 -8.94871727e-02 3.74047697e-01 -4.54308391e-01
1.34624469e+00 -8.56432259e-01 -1.24789476e+00 -1.56018257e-01
1.07061362e+00 -7.06995010e-01 5.64973414e-01 2.71271080e-01
-1.15620029e+00 -3.14982414e-01 -1.06746864e+00 -1.91827089e-01
-1.04496002e+00 -4.88730147e-02 8.84584010e-01 4.91858333e-01
-6.15005791e-01 4.84358907e-01 -6.75405145e-01 -3.24402362e-01
1.42403588e-01 1.47441790e-01 -3.15893859e-01 2.21716598e-01
-1.70816171e+00 1.60462093e+00 9.84515846e-01 -1.05294555e-01
-1.88950837e-01 -1.01109731e+00 -1.02136385e+00 -4.16008234e-01
9.22874689e-01 -9.41919863e-01 1.56917393e+00 -3.55630726e-01
-9.19377446e-01 1.20501995e+00 -9.22946930e-02 -8.59661758e-01
5.98306134e-02 -6.56542838e-01 -8.00019085e-01 -2.67885238e-01
3.42594743e-01 1.92677900e-01 5.23435250e-02 -1.08076179e+00
-8.57525170e-01 1.98164135e-01 3.97592932e-01 1.09283134e-01
2.62705594e-01 5.49806833e-01 -4.56876814e-01 -5.82955897e-01
-4.15610135e-01 -5.03888011e-01 -1.68380126e-01 -6.12247765e-01
-5.76949239e-01 -7.76266575e-01 1.59758568e-01 -5.14403105e-01
1.99832106e+00 -1.45694327e+00 3.44737321e-01 1.15013249e-01
3.35106879e-01 2.41660804e-01 3.94333839e-01 6.26117945e-01
-2.93649703e-01 6.58387661e-01 -4.63978499e-01 -3.38722467e-01
4.96374875e-01 5.03313601e-01 -2.03336805e-01 -2.79084295e-01
-3.41558189e-04 1.24616945e+00 -1.06202197e+00 -8.62270713e-01
3.76369923e-01 1.37898996e-01 -7.35758469e-02 5.82180955e-02
-5.07779896e-01 -4.70177114e-01 -3.04669291e-01 3.43664974e-01
1.41179264e-01 -1.00376561e-01 4.78431672e-01 -3.17316592e-01
-6.26403466e-02 8.11264038e-01 -1.18112636e+00 1.94518340e+00
-5.40887117e-01 6.58861995e-01 -3.87642652e-01 -4.12909836e-01
3.03420782e-01 5.08575022e-01 1.13269232e-01 -1.73036888e-01
9.36831720e-03 3.57218474e-01 9.23113674e-02 -5.05582035e-01
7.44654059e-01 -2.44796634e-01 -4.31427002e-01 3.87535125e-01
4.79149073e-01 -6.21608377e-01 8.42719615e-01 5.78441978e-01
1.10096383e+00 6.11536860e-01 1.14257908e+00 -5.03276408e-01
6.91300333e-01 2.59322107e-01 5.20353317e-01 7.55846560e-01
6.54242188e-02 3.54693741e-01 2.28793278e-01 -4.48975027e-01
-5.87670207e-01 -1.03561497e+00 -5.26544690e-01 7.16931581e-01
-7.18535809e-03 -1.57718098e+00 -5.31979978e-01 -1.14888752e+00
-2.32379153e-01 1.55258214e+00 -7.20316887e-01 3.17938805e-01
-9.23680007e-01 -5.45581877e-01 1.17004550e+00 6.15125000e-01
5.31813204e-01 -1.17287087e+00 -9.05765474e-01 2.48644695e-01
-3.32999378e-01 -1.43353343e+00 -3.30353305e-02 2.24639311e-01
-6.25758618e-02 -1.61706209e+00 2.00181141e-01 -7.65091598e-01
6.45773187e-02 -7.05179095e-01 1.84171307e+00 1.85729757e-01
-2.33557433e-01 4.19513792e-01 -4.12485659e-01 -1.00158679e+00
-1.88656420e-01 2.91501075e-01 -1.98294818e-01 -1.11334634e+00
1.21337831e+00 -2.40364119e-01 8.31512287e-02 -1.35266379e-01
-6.82012439e-01 3.62251878e-01 3.45919877e-02 6.01005793e-01
4.41626966e-01 1.45806058e-03 4.91755545e-01 -1.74730206e+00
9.13282812e-01 -2.87911624e-01 -3.61239940e-01 6.95506990e-01
-9.33846295e-01 5.42871095e-02 2.00577229e-01 2.79023379e-01
-1.17339587e+00 -6.36009052e-02 -1.98899135e-01 2.92016238e-01
-1.27856523e-01 9.93523717e-01 -1.66143671e-01 4.40683216e-01
7.79817700e-01 -1.52369931e-01 -7.55645096e-01 -5.71000755e-01
9.78256404e-01 4.87603694e-01 6.54954255e-01 -1.00705123e+00
6.63015187e-01 -2.21624717e-01 -6.34643435e-02 -8.98065150e-01
-1.05720162e+00 -5.55768013e-01 -9.83619630e-01 1.29874781e-01
6.45079672e-01 -5.28282106e-01 -5.39777726e-02 8.02720636e-02
-1.51203001e+00 -2.52806723e-01 -5.42084575e-01 2.74313778e-01
-4.58266377e-01 1.33313283e-01 -3.75968128e-01 -5.04396141e-01
-5.92390597e-01 -4.77073073e-01 5.98274410e-01 9.70966443e-02
-9.03245628e-01 -1.49435675e+00 3.40674192e-01 2.03224808e-01
3.22221130e-01 3.98610413e-01 9.70388889e-01 -1.08776319e+00
-3.57604064e-02 2.11188793e-02 -4.86731268e-02 2.09617764e-01
1.57170847e-01 9.36008468e-02 -5.33035994e-01 3.35189492e-01
-3.82958978e-01 -2.77803659e-01 6.90473735e-01 -4.69505936e-02
1.01725793e+00 -2.56677955e-01 -6.62768960e-01 3.24265212e-01
1.59458852e+00 -1.45072430e-01 4.90122706e-01 4.99195218e-01
4.41740155e-01 5.85357010e-01 6.32529140e-01 8.69109407e-02
5.41362047e-01 6.53548777e-01 -3.26333225e-01 1.71570793e-01
-6.37445033e-01 -7.05728829e-02 -3.45148593e-01 9.05673862e-01
-3.64332020e-01 -2.93782532e-01 -1.34487224e+00 1.06471419e+00
-1.84791887e+00 -8.23600173e-01 -3.80267769e-01 1.73226690e+00
1.62454319e+00 2.88851023e-01 -1.90288708e-01 -1.15479544e-01
4.58050191e-01 -1.40088484e-01 1.47670776e-01 -5.80792785e-01
-4.47389275e-01 1.21439493e+00 2.79784709e-01 8.43872666e-01
-1.12794042e+00 1.35459471e+00 7.53332710e+00 9.70425189e-01
-6.18905351e-02 1.62973404e-01 -5.50484598e-01 -7.26725385e-02
-3.96964103e-01 3.56731325e-01 -9.79878008e-01 -2.70543303e-02
1.06043541e+00 -7.95433402e-01 -8.31404775e-02 3.73440027e-01
-5.61740220e-01 -1.52834028e-01 -1.54339635e+00 7.74899840e-01
3.29198956e-01 -1.50379694e+00 2.58796275e-01 -3.70669872e-01
5.31806171e-01 2.42058318e-02 -6.25649214e-01 6.42313302e-01
3.86888325e-01 -1.24954045e+00 2.92724282e-01 6.13531649e-01
8.42034996e-01 -7.16038644e-01 1.06760764e+00 1.56562015e-01
-1.08545244e+00 6.76611483e-01 2.84342188e-02 -2.46340826e-01
3.97757232e-01 4.05896693e-01 -7.09187448e-01 9.63218272e-01
1.43256620e-01 7.98232198e-01 -6.97909117e-01 6.47989392e-01
-1.14510047e+00 5.95608234e-01 -3.49093169e-01 -3.12040865e-01
-2.54950702e-01 1.70192912e-01 7.89208710e-01 2.03580523e+00
-6.68258667e-01 5.43176770e-01 -1.55505016e-01 9.79151130e-01
-2.09814444e-01 6.18030690e-02 -4.27751660e-01 2.05213819e-02
7.20656872e-01 1.08503997e+00 -1.67088166e-01 -7.88909674e-01
-4.97349232e-01 5.83005071e-01 5.58346272e-01 2.31764346e-01
-7.94093549e-01 -1.02154756e+00 4.13631946e-01 -3.43408287e-01
2.78007209e-01 -2.61749566e-01 -4.34631169e-01 -1.43160760e+00
7.31600001e-02 -7.81739116e-01 5.67458749e-01 -7.04915285e-01
-1.43487799e+00 5.87280035e-01 7.72581995e-01 -5.62901139e-01
-6.91822708e-01 -6.44719541e-01 -1.76250115e-01 8.59012306e-01
-1.40901089e+00 -1.23666370e+00 2.55439460e-01 3.21466386e-01
4.64881152e-01 1.06437534e-01 1.52421081e+00 3.81113917e-01
-3.46947521e-01 7.29524195e-01 -5.97063243e-01 5.47086954e-01
7.94935226e-01 -1.72264683e+00 6.01594865e-01 1.00613570e+00
5.28250158e-01 1.39755690e+00 9.43398118e-01 -7.67502844e-01
-7.77622700e-01 -9.86360133e-01 1.65573788e+00 -1.12299836e+00
7.96000540e-01 -8.55103508e-02 -6.99375033e-01 1.16998744e+00
7.45342195e-01 -5.35008967e-01 1.07234740e+00 7.80609071e-01
-5.54685652e-01 5.28884470e-01 -8.68130863e-01 4.17735845e-01
1.33995867e+00 -4.71717805e-01 -1.85167754e+00 2.84477562e-01
6.26745999e-01 -6.18582070e-01 -1.17044592e+00 6.49161220e-01
7.09742725e-01 -3.62011403e-01 5.72956920e-01 -9.06132638e-01
5.07649779e-01 -3.98148179e-01 -2.80877113e-01 -1.19834530e+00
-8.94993618e-02 -6.41618788e-01 -5.76208532e-01 1.28026307e+00
1.13312149e+00 -2.33231306e-01 2.13070631e-01 8.85645866e-01
-7.22827911e-02 -6.46534085e-01 -9.16364551e-01 -7.47459352e-01
4.60351616e-01 -9.83525455e-01 3.23567569e-01 1.18271685e+00
5.96205413e-01 9.99101639e-01 2.24465609e-01 -2.12757066e-01
7.82092750e-01 1.44656554e-01 5.39265513e-01 -7.79917955e-01
-4.25168455e-01 -4.62557822e-01 -4.11803991e-01 -9.59514916e-01
3.08872640e-01 -1.01872051e+00 -4.41683270e-02 -1.89694166e+00
2.93644071e-01 -2.64111608e-02 -2.58827895e-01 7.97215462e-01
-3.30230027e-01 6.99618086e-02 -1.75351650e-02 -1.81562528e-01
-9.92320120e-01 1.34076238e-01 5.54344058e-01 7.03259706e-02
6.53645471e-02 -3.64601433e-01 -6.65075302e-01 7.66529799e-01
5.04302204e-01 -3.39834660e-01 -3.12232465e-01 -8.60860944e-01
5.16545236e-01 -5.06537199e-01 2.43829004e-02 -6.16196871e-01
3.70328218e-01 -2.44823679e-01 -1.24286592e-01 -6.18028820e-01
6.11099191e-02 -4.48831677e-01 -8.53016749e-02 1.89248994e-01
-4.49994057e-01 9.45987776e-02 3.57554018e-01 -5.37015572e-02
-1.31479457e-01 -4.45255160e-01 2.95462340e-01 -2.72287667e-01
-1.09285581e+00 -2.52576292e-01 -2.65265256e-01 5.26036203e-01
9.49160576e-01 1.84118077e-01 -3.69050413e-01 1.75370857e-01
-6.64619684e-01 6.17321491e-01 2.20744297e-01 4.57532018e-01
5.92438638e-01 -9.18652356e-01 -1.08384097e+00 -3.07749897e-01
2.68319637e-01 1.69629574e-01 -5.01073182e-01 3.54632914e-01
-4.30350751e-01 7.72123039e-01 1.14883058e-01 4.36239280e-02
-1.34575355e+00 4.18699741e-01 5.84276617e-01 -2.75086731e-01
-3.33865196e-01 1.12260795e+00 -3.95273805e-01 -4.56003338e-01
2.78802384e-02 -1.55336067e-01 -5.21442354e-01 -2.59793580e-01
7.93206453e-01 3.11131686e-01 1.73841417e-01 -3.31921756e-01
-9.56540763e-01 3.16050768e-01 -2.69417495e-01 -3.41798395e-01
1.33256567e+00 -5.96063547e-02 -4.85751361e-01 5.85416555e-01
7.23606765e-01 2.10608408e-01 9.69546065e-02 -4.92145658e-01
6.08715534e-01 -4.65243012e-02 5.82361706e-02 -1.48892498e+00
-2.53546864e-01 2.25642368e-01 -2.91814834e-01 7.56573081e-02
9.97611523e-01 2.79393047e-01 6.78724825e-01 5.34544230e-01
6.08218729e-01 -1.32199848e+00 -8.76140118e-01 8.89599621e-01
9.77490604e-01 -9.11045611e-01 5.70880175e-01 -8.11696470e-01
-2.20181540e-01 1.26120281e+00 6.76838815e-01 -8.75445083e-02
7.02414393e-01 4.82036620e-01 -5.15205450e-02 -7.11738646e-01
-1.17328095e+00 -5.77519417e-01 3.86275321e-01 7.99981773e-01
9.16821659e-01 2.05786735e-01 -1.12081206e+00 1.15091383e+00
-6.05046213e-01 5.69768958e-02 5.06902039e-01 1.18433046e+00
3.70106325e-02 -1.51566780e+00 5.08057833e-01 3.64316761e-01
-6.14994466e-01 -8.37226212e-01 -1.12022889e+00 1.30638015e+00
6.47285223e-01 1.03083193e+00 -3.37585360e-01 -1.83539882e-01
5.95647633e-01 3.17497373e-01 1.16064155e+00 -1.47894990e+00
-6.76134944e-01 -4.54818279e-01 1.09789824e+00 -3.89221877e-01
-8.81204307e-01 -7.56461024e-01 -1.41246068e+00 -1.56115936e-02
-6.11711204e-01 5.00299156e-01 4.56971794e-01 1.33643126e+00
-4.37263735e-02 5.88492334e-01 -2.71709859e-01 3.72786671e-01
-2.50764966e-01 -1.32360017e+00 -3.66702855e-01 3.64539176e-01
-1.29934713e-01 -4.99185711e-01 -3.85182947e-01 3.44233096e-01]
|
[9.786608695983887, 8.939358711242676]
|
0397bec6-7ae3-4e93-b3d5-ed80d22bb51e
|
breaking-the-limit-of-graph-neural-networks
|
2106.06586
| null |
https://arxiv.org/abs/2106.06586v1
|
https://arxiv.org/pdf/2106.06586v1.pdf
|
Breaking the Limit of Graph Neural Networks by Improving the Assortativity of Graphs with Local Mixing Patterns
|
Graph neural networks (GNNs) have achieved tremendous success on multiple graph-based learning tasks by fusing network structure and node features. Modern GNN models are built upon iterative aggregation of neighbor's/proximity features by message passing. Its prediction performance has been shown to be strongly bounded by assortative mixing in the graph, a key property wherein nodes with similar attributes mix/connect with each other. We observe that real world networks exhibit heterogeneous or diverse mixing patterns and the conventional global measurement of assortativity, such as global assortativity coefficient, may not be a representative statistic in quantifying this mixing. We adopt a generalized concept, node-level assortativity, one that is based at the node level to better represent the diverse patterns and accurately quantify the learnability of GNNs. We find that the prediction performance of a wide range of GNN models is highly correlated with the node level assortativity. To break this limit, in this work, we focus on transforming the input graph into a computation graph which contains both proximity and structural information as distinct type of edges. The resulted multi-relational graph has an enhanced level of assortativity and, more importantly, preserves rich information from the original graph. We then propose to run GNNs on this computation graph and show that adaptively choosing between structure and proximity leads to improved performance under diverse mixing. Empirically, we show the benefits of adopting our transformation framework for semi-supervised node classification task on a variety of real world graph learning benchmarks.
|
['Jianzhu Ma', 'Pan Li', 'Jennifer Neville', 'Vinith Budde', 'Susheel Suresh']
|
2021-06-11
| null | null | null | null |
['node-classification-on-non-homophilic']
|
['graphs']
|
[ 7.86781460e-02 -4.54562604e-02 -4.38799530e-01 -2.52774805e-01
-3.99498828e-02 -8.06183875e-01 7.27695107e-01 7.24496067e-01
-1.85148433e-01 5.19067466e-01 -5.70615791e-02 -3.22775304e-01
-6.56947911e-01 -1.37887478e+00 -7.11614668e-01 -9.12029564e-01
-5.47109842e-01 5.13726532e-01 1.69894025e-01 -2.72908986e-01
-9.77815911e-02 6.32399321e-01 -1.13038468e+00 -3.62640284e-02
6.76072299e-01 8.61234546e-01 -2.55216420e-01 6.09666884e-01
-4.07108739e-02 8.68811429e-01 -4.47882354e-01 -4.50614572e-01
5.27624726e-01 -4.05652136e-01 -6.56260133e-01 -1.52988553e-01
3.71110380e-01 3.30922008e-01 -6.56400084e-01 1.22479403e+00
1.56919107e-01 -6.82856739e-02 6.73375845e-01 -1.75511098e+00
-5.68812191e-01 1.29686677e+00 -6.46755695e-01 2.98896074e-01
4.21107933e-02 -3.40002542e-03 1.53304553e+00 -3.91614646e-01
6.31446660e-01 1.16493726e+00 8.24908853e-01 -5.87223321e-02
-1.54435074e+00 -6.75327659e-01 1.20488681e-01 -8.40755031e-02
-1.64500725e+00 6.37768507e-02 1.01933372e+00 -2.67015666e-01
6.88489914e-01 4.21061724e-01 7.94849098e-01 9.26651895e-01
3.42904180e-01 3.93965423e-01 1.04251277e+00 -9.45078507e-02
9.91757810e-02 -1.44431621e-01 3.25666755e-01 9.20508087e-01
6.89581573e-01 -1.47650883e-01 -4.13035512e-01 -1.04560688e-01
5.45785606e-01 3.83454829e-01 -2.13704094e-01 -6.92972124e-01
-1.02910900e+00 8.91164720e-01 1.14582932e+00 4.13185775e-01
-2.71884292e-01 3.28561604e-01 4.03678209e-01 7.61220336e-01
3.68899196e-01 3.11663806e-01 -2.24325269e-01 2.47725382e-01
-5.92342138e-01 -1.45469233e-01 9.86211002e-01 6.59537971e-01
1.12629867e+00 -1.13970876e-01 2.67825089e-02 8.23479414e-01
3.10852855e-01 1.82120517e-01 3.19106609e-01 -5.29424191e-01
4.22123075e-01 1.26180410e+00 -7.05285668e-01 -1.58132041e+00
-6.29133105e-01 -9.08476353e-01 -1.52199292e+00 -1.43935055e-01
2.78007954e-01 1.22716084e-01 -8.50264072e-01 2.34335279e+00
3.30860317e-01 1.63890585e-01 -1.21861614e-01 4.44094568e-01
8.79985690e-01 4.03810203e-01 -1.05613479e-02 -2.04293896e-02
9.37908173e-01 -6.42968774e-01 -3.83288860e-01 -9.81557220e-02
8.08756232e-01 -1.64402828e-01 7.95303643e-01 4.01622802e-02
-8.11158419e-01 -1.75546706e-01 -1.24173462e+00 2.17476442e-01
-5.53527296e-01 -5.46692014e-01 7.78707087e-01 7.69370675e-01
-1.32942796e+00 9.22269881e-01 -7.37290680e-01 -3.70982438e-01
5.26625752e-01 4.25202638e-01 -7.68577039e-01 -2.87257373e-01
-1.13305879e+00 4.22833234e-01 5.69060385e-01 -8.63985941e-02
-6.56286776e-01 -6.56774759e-01 -9.02423203e-01 2.35433966e-01
3.89475703e-01 -7.47939408e-01 4.46695298e-01 -7.70579457e-01
-9.73328531e-01 6.42509103e-01 3.95047605e-01 -4.78681058e-01
4.00166392e-01 4.76632118e-01 -3.78474116e-01 9.13930237e-02
-8.02286267e-02 2.75889605e-01 5.86993277e-01 -1.25658166e+00
-3.55327517e-01 -5.98749101e-01 2.47935489e-01 5.19515425e-02
-5.87177873e-01 -4.60501045e-01 -1.67082012e-01 -5.72614193e-01
6.01120174e-01 -1.02303112e+00 -1.80988207e-01 -1.17426641e-01
-7.16668069e-01 -2.88797051e-01 6.02814138e-01 1.50116473e-01
1.38695455e+00 -1.87231481e+00 1.07993171e-01 8.44578266e-01
1.18480754e+00 1.61893852e-02 -3.42372864e-01 7.63517559e-01
-1.66036487e-01 3.59016001e-01 -2.71648496e-01 -2.23689988e-01
1.06643904e-02 3.10662895e-01 1.01850942e-01 7.74829686e-01
1.70242131e-01 1.00211585e+00 -9.54372585e-01 -3.60953182e-01
-6.56987578e-02 3.81963104e-01 -4.78847325e-01 -9.18385194e-05
1.34089038e-01 -1.31872967e-01 -2.74858892e-01 5.46452880e-01
6.78802073e-01 -8.46128643e-01 6.56343222e-01 -2.84548789e-01
5.33527315e-01 1.44998088e-01 -1.19314837e+00 1.20675015e+00
-1.56046584e-01 4.82390940e-01 3.74086411e-03 -1.26566768e+00
1.03834975e+00 -1.37644798e-01 6.43538117e-01 -3.91831249e-01
1.57752782e-01 -3.77449463e-03 5.21353483e-01 2.61273794e-02
2.75993466e-01 9.93894935e-02 -4.01181467e-02 6.05924666e-01
1.67567998e-01 3.98033679e-01 4.16801155e-01 7.20530272e-01
1.87237346e+00 -6.89780295e-01 4.60000962e-01 -5.80641091e-01
2.61725157e-01 -3.08450818e-01 4.52653497e-01 9.26224470e-01
-1.32294759e-01 3.73842716e-01 1.02677476e+00 -3.93357366e-01
-8.46240997e-01 -1.34570885e+00 2.81906705e-02 1.15502930e+00
1.68154255e-01 -6.87813640e-01 -3.35343331e-01 -7.65895188e-01
2.55675018e-01 -1.18877299e-01 -7.93066144e-01 -5.31430721e-01
-4.92990971e-01 -1.17783058e+00 7.25091040e-01 3.38413417e-01
3.16976041e-01 -7.17557669e-01 1.48423493e-01 5.54386675e-02
1.95533767e-01 -1.22318649e+00 -2.93345511e-01 3.68528992e-01
-8.35810065e-01 -1.28656495e+00 -2.32009917e-01 -6.56057000e-01
7.27160513e-01 4.22831237e-01 1.37641561e+00 6.27110243e-01
-1.10464497e-02 2.19380960e-01 -3.58754456e-01 1.97580814e-01
-5.42949557e-01 4.12863910e-01 1.90434158e-01 3.33544463e-01
-1.62668303e-02 -1.33626091e+00 -5.70186079e-01 2.52938896e-01
-1.23684776e+00 2.59752255e-02 7.39742398e-01 6.60219908e-01
2.52068877e-01 2.50687569e-01 4.92227823e-01 -1.27481556e+00
7.50281930e-01 -8.04489315e-01 -2.34986871e-01 2.65058547e-01
-8.63727927e-01 1.83976084e-01 8.55239570e-01 -4.11308587e-01
-2.43240401e-01 -3.24868113e-01 2.69578755e-01 -2.78399020e-01
2.62358248e-01 9.44386959e-01 -2.56406248e-01 -3.86428237e-01
6.46566212e-01 -7.73395272e-03 1.58054471e-01 -1.28372744e-01
3.45460743e-01 2.27517083e-01 2.89083123e-01 -4.50723708e-01
1.02289617e+00 4.73830462e-01 6.78192377e-01 -6.93255782e-01
-4.15852934e-01 -2.60723919e-01 -4.48592275e-01 -8.25771913e-02
2.51969606e-01 -6.55932546e-01 -1.05265439e+00 4.19159681e-01
-6.17628098e-01 -3.47726971e-01 -6.37088045e-02 2.43252844e-01
-2.06580618e-03 4.56552058e-01 -9.15084541e-01 -4.74134445e-01
-2.86209553e-01 -8.85076761e-01 5.98217368e-01 8.41823593e-03
3.49785164e-02 -1.27881360e+00 1.34984180e-01 -1.32462397e-01
4.26052332e-01 5.63038826e-01 1.33764040e+00 -1.01661956e+00
-6.98903143e-01 -2.92538583e-01 -5.19207954e-01 8.59397575e-02
3.59772027e-01 1.39527515e-01 -4.72963244e-01 -5.28301299e-01
-5.08444369e-01 -1.31373256e-01 1.06998825e+00 2.21942350e-01
8.95262778e-01 -4.05569315e-01 -3.57994795e-01 7.57030606e-01
1.73556817e+00 -3.57191503e-01 2.87737936e-01 -5.63988537e-02
1.23890984e+00 4.74603653e-01 -2.56204605e-01 2.36989081e-01
5.03599226e-01 3.25777411e-01 6.08892679e-01 4.48302925e-02
-6.31503910e-02 -4.98450309e-01 1.44050643e-01 1.13219321e+00
-6.05701432e-02 -7.38943815e-01 -1.05502963e+00 2.70192951e-01
-1.81789696e+00 -7.25034773e-01 -1.77114069e-01 2.03051066e+00
6.09034777e-01 2.78251052e-01 2.80730456e-01 2.57743716e-01
9.13461864e-01 5.16537726e-01 -4.08582538e-01 7.78152281e-03
-4.60258275e-01 8.87453333e-02 7.09207118e-01 4.62073028e-01
-8.06012154e-01 6.94351017e-01 5.69211626e+00 7.13725090e-01
-1.00440025e+00 -1.25556245e-01 7.61081100e-01 1.99704483e-01
-5.72303951e-01 3.63848172e-02 -2.76995420e-01 3.76801282e-01
9.03614283e-01 -3.66442710e-01 5.10711968e-01 5.82356036e-01
-3.21383506e-01 2.88645774e-01 -1.27637470e+00 8.78488421e-01
-1.66399285e-01 -1.33475184e+00 3.31094712e-01 2.71307975e-01
7.33971715e-01 3.76412481e-01 3.43742110e-02 2.05282837e-01
9.33431268e-01 -1.17542338e+00 2.84770936e-01 2.30908930e-01
5.49911082e-01 -6.78346515e-01 5.21156788e-01 3.32627565e-01
-1.61192524e+00 -2.00181857e-01 -1.57050610e-01 5.35868704e-02
-2.51785636e-01 6.38175666e-01 -7.40212917e-01 8.07141960e-01
5.59097707e-01 8.99671137e-01 -8.90304267e-01 8.60369146e-01
2.87313700e-01 4.31783408e-01 -5.87726355e-01 -1.35259509e-01
3.23041767e-01 -4.90071267e-01 5.28118134e-01 1.05812693e+00
1.58781573e-01 -1.91026807e-01 2.56311566e-01 8.25836301e-01
-6.16826892e-01 9.72972736e-02 -8.09854150e-01 -3.29322577e-01
6.41202509e-01 1.46983278e+00 -1.17933023e+00 -1.82944819e-01
-2.97822297e-01 4.78005558e-01 8.09661865e-01 2.90243536e-01
-5.93219697e-01 -2.82704026e-01 7.49988735e-01 2.01882362e-01
5.27249537e-02 -2.40698308e-01 -1.45646244e-01 -9.98697937e-01
4.63489676e-04 -8.33302617e-01 7.13837981e-01 -3.21756035e-01
-1.69193459e+00 8.61496687e-01 -1.34432092e-01 -1.04013360e+00
1.56838790e-01 -5.28356910e-01 -6.67416096e-01 5.02794266e-01
-1.13039458e+00 -1.28249133e+00 -5.09402871e-01 6.52312875e-01
-2.15480119e-01 -1.72563672e-01 5.35721660e-01 3.62285644e-01
-7.05505550e-01 8.29435050e-01 -8.41276422e-02 4.08290863e-01
2.72550911e-01 -1.46157181e+00 4.03648883e-01 7.58431971e-01
5.23982942e-01 7.82089412e-01 6.06293559e-01 -7.67012715e-01
-1.64124918e+00 -1.19576526e+00 4.16525930e-01 -3.66411805e-01
9.88870382e-01 -4.83709782e-01 -9.68977392e-01 7.71238387e-01
-2.51228452e-01 6.39980495e-01 4.71624821e-01 3.44821095e-01
-8.24931741e-01 -5.62109649e-01 -1.04706526e+00 6.23263299e-01
1.52716959e+00 -7.50870764e-01 -2.20154412e-02 2.25937933e-01
8.34969401e-01 2.11009756e-02 -1.17517424e+00 7.48109400e-01
5.07272959e-01 -1.05611372e+00 9.21390653e-01 -7.74661660e-01
3.82091433e-01 -1.00517489e-01 -2.77207285e-01 -1.41567802e+00
-4.89550054e-01 -6.70605361e-01 -1.26684412e-01 1.12025964e+00
5.76128244e-01 -9.41981852e-01 1.10211372e+00 2.50317037e-01
1.65526003e-01 -8.69476914e-01 -7.99781084e-01 -7.35052347e-01
1.43533587e-01 -1.83781832e-01 8.33905160e-01 1.21932483e+00
-1.05522625e-01 6.69835806e-01 -1.72774866e-01 1.73879996e-01
8.01110625e-01 1.16873384e-02 7.54545212e-01 -1.65666795e+00
-1.52145997e-01 -9.04523015e-01 -9.36386704e-01 -6.51750207e-01
1.16198309e-01 -1.45060134e+00 -4.20441747e-01 -1.32720780e+00
3.76404107e-01 -7.43851423e-01 -6.68748856e-01 2.62067884e-01
-3.94228883e-02 4.19818431e-01 5.90611398e-02 2.40322918e-01
-7.19594598e-01 3.61982852e-01 1.09064329e+00 -2.65202045e-01
-2.43350655e-01 -1.90178342e-02 -7.73414433e-01 4.89341587e-01
7.34436452e-01 -5.50042033e-01 -5.74789047e-01 -5.58949783e-02
8.14605296e-01 -1.92529988e-02 4.00435001e-01 -9.70550716e-01
4.50890839e-01 3.08463629e-02 1.45268485e-01 -2.07784280e-01
1.89656690e-01 -9.23717976e-01 5.23144662e-01 4.91183966e-01
-5.33460081e-01 4.76135045e-01 -3.64036322e-01 9.25845027e-01
-1.00803114e-01 3.16979885e-01 6.71500623e-01 2.54519992e-02
-2.89306343e-01 8.31847906e-01 2.13908061e-01 8.77435431e-02
7.50762105e-01 -1.17320031e-01 -6.91644907e-01 -4.62446094e-01
-4.56940800e-01 5.34662530e-02 5.52588105e-01 3.43564332e-01
4.71127629e-01 -1.45818794e+00 -6.87869966e-01 1.77294880e-01
1.95927113e-01 -3.00202787e-01 1.43576607e-01 9.83710587e-01
-5.20445585e-01 -1.32296070e-01 -2.16229975e-01 -7.80848324e-01
-1.25496602e+00 5.04123628e-01 3.49466145e-01 -6.13536716e-01
-5.13091862e-01 7.24001348e-01 1.44346446e-01 -6.08446836e-01
-2.11825650e-02 -2.47335479e-01 -6.53741583e-02 5.78403212e-02
-1.39356861e-02 3.09165955e-01 1.19767047e-01 -7.53097951e-01
-2.87625611e-01 2.72255689e-01 -1.28567755e-01 2.76831627e-01
1.37266231e+00 -8.16170592e-03 -5.56132853e-01 5.71292758e-01
1.57704389e+00 -4.97510061e-02 -7.57766545e-01 -6.31184876e-01
1.59598276e-01 -3.47054094e-01 -1.81770101e-01 -3.00804377e-01
-1.40914261e+00 4.90923285e-01 1.83054745e-01 1.12308836e+00
9.61663425e-01 2.27463648e-01 5.06253183e-01 5.85175037e-01
2.37681210e-01 -5.94123185e-01 7.41537511e-02 3.77472997e-01
4.90740150e-01 -1.20080864e+00 1.65659845e-01 -5.53610206e-01
-1.08777240e-01 7.81469584e-01 5.78215420e-01 -3.92745763e-01
1.15356243e+00 2.28397995e-01 -3.20318848e-01 -5.33227503e-01
-7.53245473e-01 -1.18334129e-01 3.81774247e-01 4.16828901e-01
1.16087757e-01 4.24573839e-01 -1.80701818e-02 3.68306130e-01
-4.11298275e-01 -6.65956736e-01 4.07582074e-01 5.30902326e-01
-2.94341296e-01 -9.58343387e-01 2.41839021e-01 8.66679430e-01
-3.87386233e-01 -7.75753856e-02 -5.96456647e-01 9.74915743e-01
-2.73682386e-01 7.06981778e-01 1.49389401e-01 -8.13655913e-01
-9.56673175e-02 -3.69530439e-01 2.75434166e-01 -4.50807989e-01
-6.59132659e-01 -5.05383551e-01 3.97041775e-02 -5.47825634e-01
-3.78742784e-01 -1.21675707e-01 -9.59323883e-01 -8.86405289e-01
-1.40907913e-01 6.28112406e-02 3.40158045e-01 8.55309069e-01
3.40579420e-01 4.42022502e-01 8.48145664e-01 -4.42849368e-01
-5.12863874e-01 -8.40189636e-01 -9.17023301e-01 7.09551394e-01
4.61274147e-01 -5.10483325e-01 -7.22861409e-01 -7.65619457e-01]
|
[6.929956912994385, 6.098354816436768]
|
ff5ff21d-1719-4514-a5a5-ca26e90a81b5
|
optimal-transport-based-identity-matching-for
|
2209.12172
| null |
https://arxiv.org/abs/2209.12172v1
|
https://arxiv.org/pdf/2209.12172v1.pdf
|
Optimal Transport-based Identity Matching for Identity-invariant Facial Expression Recognition
|
Identity-invariant facial expression recognition (FER) has been one of the challenging computer vision tasks. Since conventional FER schemes do not explicitly address the inter-identity variation of facial expressions, their neural network models still operate depending on facial identity. This paper proposes to quantify the inter-identity variation by utilizing pairs of similar expressions explored through a specific matching process. We formulate the identity matching process as an Optimal Transport (OT) problem. Specifically, to find pairs of similar expressions from different identities, we define the inter-feature similarity as a transportation cost. Then, optimal identity matching to find the optimal flow with minimum transportation cost is performed by Sinkhorn-Knopp iteration. The proposed matching method is not only easy to plug in to other models, but also requires only acceptable computational overhead. Extensive simulations prove that the proposed FER method improves the PCC/CCC performance by up to 10\% or more compared to the runner-up on wild datasets. The source code and software demo are available at https://github.com/kdhht2334/ELIM_FER.
|
['Byung Cheol Song', 'Daeha Kim']
|
2022-09-25
| null | null | null | null |
['facial-expression-recognition']
|
['computer-vision']
|
[-5.02487533e-02 -2.80484110e-01 1.25818828e-03 -6.25554085e-01
-5.07622421e-01 -5.00298321e-01 2.49871537e-01 -4.59309578e-01
-2.66676396e-01 4.96980071e-01 -2.78705388e-01 1.58791646e-01
-2.88942695e-01 -6.02840960e-01 -4.97905314e-01 -7.20068097e-01
-3.75219993e-02 1.53613552e-01 -2.00505912e-01 -3.25459033e-01
1.52038082e-01 7.45216489e-01 -1.62230301e+00 -1.70367137e-01
8.91810834e-01 1.43859422e+00 -1.32796526e-01 3.08344752e-01
-2.11522520e-01 5.51788151e-01 -2.84943819e-01 -6.68035030e-01
5.16928375e-01 -6.34911180e-01 -7.60976195e-01 6.09334335e-02
4.59042668e-01 -1.87492877e-01 -1.41736045e-01 1.45886970e+00
6.52720630e-01 1.51675150e-01 4.78513628e-01 -1.96499181e+00
-7.05497026e-01 8.09260905e-02 -9.00660217e-01 -2.54154243e-02
2.00080320e-01 -2.20443159e-01 7.27139413e-01 -1.03896976e+00
6.10946178e-01 1.32557046e+00 7.15227544e-01 6.08374476e-01
-9.86061811e-01 -1.21534479e+00 -9.77103505e-03 5.62274396e-01
-1.95359516e+00 -6.53188109e-01 8.66123736e-01 -2.24233732e-01
1.48480609e-01 1.99657977e-01 5.92673063e-01 6.43439472e-01
-3.01861346e-01 7.44259477e-01 9.98270333e-01 -1.78925112e-01
-7.97495153e-03 5.43674119e-02 2.99103484e-02 9.69630659e-01
3.17274197e-03 -2.52900243e-01 -4.85648632e-01 -4.17055748e-02
5.80420017e-01 -8.49114284e-02 -2.80466288e-01 -5.17445654e-02
-5.52354038e-01 8.11314702e-01 4.68775451e-01 7.89112523e-02
-2.99434155e-01 1.18575349e-01 3.14596653e-01 3.04986238e-01
4.54338938e-01 -9.98663679e-02 -1.95510596e-01 -1.16907418e-01
-7.62818873e-01 2.18094900e-01 5.46959639e-01 1.02426755e+00
1.16693103e+00 4.26810123e-02 1.46596998e-01 1.07673764e+00
2.57586896e-01 6.02832854e-01 9.52129588e-02 -1.39134049e+00
2.00256214e-01 5.35315216e-01 5.43566979e-02 -1.56355917e+00
-2.84451276e-01 -1.08204573e-01 -8.61884832e-01 1.05031669e-01
4.42814827e-01 -4.54806685e-01 -4.59522963e-01 1.88608134e+00
5.19863248e-01 6.23212576e-01 -1.03550695e-01 9.80971694e-01
6.70038223e-01 5.20557404e-01 1.70980483e-01 -2.31227413e-01
1.28993356e+00 -8.32673371e-01 -7.81921804e-01 1.28285497e-01
5.03536224e-01 -8.66773129e-01 6.28494203e-01 1.44594219e-02
-9.15330410e-01 -2.81181753e-01 -7.87968516e-01 1.36626080e-01
-2.94976890e-01 3.13809216e-01 6.69935763e-01 6.32712722e-01
-1.19840431e+00 5.86172581e-01 -3.71477962e-01 -5.16419470e-01
7.25553095e-01 6.09104633e-01 -5.79706788e-01 5.39544076e-02
-1.28311908e+00 4.96604055e-01 4.49577486e-03 6.10437214e-01
-4.63281393e-01 -7.03119278e-01 -8.36271763e-01 -2.50157416e-01
1.92825288e-01 -4.58425283e-01 9.96912122e-01 -1.67468786e+00
-1.59329283e+00 1.00229585e+00 -5.69337368e-01 -1.27413794e-02
5.35816908e-01 -5.46357967e-03 -4.09321874e-01 1.69327050e-01
1.22844778e-01 6.75907969e-01 6.60726488e-01 -1.20236111e+00
-5.60870469e-01 -5.54996014e-01 2.68882364e-02 2.89245635e-01
-3.82076532e-01 5.17743945e-01 -6.63557887e-01 -3.32046568e-01
1.62920095e-02 -1.08349586e+00 1.60960704e-01 6.23166203e-01
-1.42092764e-01 -1.17812119e-01 8.84270310e-01 -6.49733722e-01
1.01856470e+00 -2.12673616e+00 -1.09905019e-01 3.78765255e-01
9.45736840e-02 1.66268528e-01 -3.00086200e-01 5.12713045e-02
3.61995623e-02 8.19396451e-02 -2.72343755e-01 -3.36826146e-01
1.16468417e-02 1.09139949e-01 1.81972206e-01 7.84903765e-01
2.56126106e-01 6.67758226e-01 -7.35739827e-01 -7.38050342e-01
-1.03223875e-01 5.21123171e-01 -3.48823249e-01 7.42495134e-02
3.36070716e-01 4.38785851e-01 -5.25040269e-01 9.70668614e-01
1.18295133e+00 -2.27414202e-02 1.44036319e-02 -4.88543272e-01
-5.07160574e-02 -7.80130684e-01 -1.34121501e+00 1.53271699e+00
-3.67356718e-01 6.87944353e-01 3.83824468e-01 -1.04142678e+00
1.15425777e+00 9.52466130e-02 7.92768955e-01 -6.00440919e-01
4.75023389e-01 2.81283915e-01 -1.69819847e-01 -6.24846339e-01
4.48476940e-01 -1.35674089e-01 1.42429963e-01 2.11549371e-01
-4.32820991e-02 4.62008834e-01 7.49100670e-02 -2.06278577e-01
7.26751089e-01 1.82739496e-01 1.14744127e-01 -2.36236766e-01
8.75630796e-01 -1.85374558e-01 9.46729541e-01 1.73118666e-01
-6.04543209e-01 4.03367251e-01 3.06304455e-01 -2.68145382e-01
-7.77631342e-01 -1.01219380e+00 -2.33926907e-01 8.54718387e-01
6.78012729e-01 7.53998682e-02 -9.95014489e-01 -4.25447613e-01
5.01716770e-02 9.88612697e-02 -6.31274283e-01 4.66121919e-02
-4.16220486e-01 -5.42914629e-01 8.32238555e-01 2.30892166e-01
1.05626881e+00 -7.47339606e-01 -1.97414204e-01 -1.77489579e-01
-2.71952212e-01 -1.04699862e+00 -5.73030353e-01 -4.94540602e-01
-4.06367183e-01 -9.76487517e-01 -8.94209266e-01 -8.41679871e-01
9.60034490e-01 1.62362218e-01 6.20861351e-01 2.14171678e-01
-3.70960057e-01 1.79558992e-01 -3.43164235e-01 -2.13674173e-01
1.07145429e-01 -1.22000143e-01 1.80358678e-01 7.94853151e-01
6.33258820e-01 -5.87519646e-01 -7.58669317e-01 6.16301417e-01
-4.89143848e-01 -1.77141741e-01 2.79173851e-01 7.67261803e-01
6.55618310e-01 -4.57452759e-02 6.69137299e-01 -3.70980293e-01
5.64788997e-01 -5.26768684e-01 -6.26983762e-01 3.44357193e-01
-4.76925969e-01 -8.01390409e-02 5.46673536e-01 -3.56375515e-01
-1.14591992e+00 2.82070100e-01 -9.01489407e-02 -8.17803681e-01
-1.03675343e-01 1.00505672e-01 -4.12092596e-01 -5.99303901e-01
1.55352265e-01 2.11753637e-01 3.17832708e-01 -2.48653874e-01
3.03643644e-01 7.99708068e-01 3.79819721e-01 -7.28912055e-01
8.22894812e-01 6.44288778e-01 2.11684570e-01 -8.21720541e-01
-5.33438027e-01 -3.95798564e-01 -6.20153606e-01 -6.80338085e-01
8.57838154e-01 -6.91055238e-01 -1.20493484e+00 7.17812717e-01
-1.16170180e+00 3.16817909e-02 2.10067511e-01 2.15233192e-01
-4.63336378e-01 3.99451017e-01 -3.92182559e-01 -9.58101809e-01
-4.68776137e-01 -9.93026793e-01 9.60872710e-01 5.89195490e-01
-3.49055696e-03 -8.82011354e-01 -1.65560976e-01 1.81908607e-01
3.45364064e-01 5.90562165e-01 4.01512861e-01 -2.54290551e-01
-3.93121868e-01 -3.98639292e-02 -5.99691212e-01 2.16581881e-01
2.91975290e-01 2.65157372e-01 -1.00842392e+00 -2.83720680e-02
-3.54219377e-01 -1.16019212e-01 4.13637966e-01 1.54682770e-01
1.17417026e+00 -4.21666473e-01 -1.90749660e-01 1.00186265e+00
1.53029251e+00 3.98706466e-01 5.81108868e-01 3.41068298e-01
5.84636033e-01 8.55139077e-01 7.90208459e-01 6.59856260e-01
4.04066861e-01 7.58960783e-01 4.03736204e-01 -2.10107237e-01
1.91544607e-01 -1.86211810e-01 1.94250524e-01 6.61165476e-01
-2.87227333e-01 2.05433648e-02 -7.59511232e-01 5.24161875e-01
-1.90473020e+00 -1.12783742e+00 -2.54671667e-02 1.94587505e+00
5.55048287e-01 -6.49334133e-01 5.20809107e-02 -1.08396880e-01
1.08076167e+00 -1.60661899e-02 -6.33158565e-01 -4.99203146e-01
-1.73599243e-01 8.12055618e-02 5.52719176e-01 4.73677218e-01
-1.02272248e+00 8.83882761e-01 5.23933077e+00 9.54093874e-01
-1.18969858e+00 1.04852974e-01 6.88799262e-01 2.98250858e-02
8.26287176e-03 -1.45781204e-01 -7.72810459e-01 5.47518790e-01
6.18927419e-01 -5.90915918e-01 5.94888449e-01 7.85291731e-01
5.63975155e-01 7.58521929e-02 -7.10543394e-01 1.35282815e+00
2.04389185e-01 -1.04052389e+00 -2.57649720e-02 -1.08717754e-01
5.17034709e-01 -3.69846493e-01 7.76424408e-02 -7.46984705e-02
8.10252875e-02 -8.80992651e-01 6.17690027e-01 7.75519133e-01
8.01617146e-01 -1.01075852e+00 6.75920606e-01 -1.24686785e-01
-1.67732298e+00 -3.61278802e-02 -4.06061888e-01 1.78823084e-01
1.82578117e-01 1.28428623e-01 -1.87389717e-01 6.46051168e-01
7.64447689e-01 8.63713205e-01 -3.57976675e-01 9.80713785e-01
3.23648900e-02 1.67502731e-01 -3.31722975e-01 1.55374501e-02
9.77061689e-02 -6.32210076e-01 3.09232503e-01 1.05041420e+00
6.42129004e-01 1.23454899e-01 -1.43089727e-01 8.87409985e-01
-3.41510445e-01 4.54009801e-01 -5.84651053e-01 2.15932250e-01
7.18949020e-01 1.42338443e+00 -3.21667790e-01 5.41303307e-02
-3.02937746e-01 1.32036078e+00 1.95213079e-01 3.74504268e-01
-1.16154993e+00 -6.05310977e-01 1.21270537e+00 -1.03389062e-01
1.01042970e-03 1.97566338e-02 -4.10330668e-03 -8.72158289e-01
1.00466773e-01 -6.25039041e-01 2.39542186e-01 -8.48084033e-01
-1.23349285e+00 6.88153028e-01 -8.37834924e-02 -1.39703262e+00
1.26853690e-01 -4.45907533e-01 -6.44694209e-01 9.67471480e-01
-1.70953119e+00 -1.30595112e+00 -7.82002687e-01 8.68663311e-01
2.88922220e-01 -1.72774181e-01 7.06399560e-01 7.94027448e-01
-9.50282335e-01 1.17790461e+00 9.45915282e-02 4.29909617e-01
5.45783341e-01 -6.17979884e-01 -1.45589754e-01 7.09269166e-01
-1.80089742e-01 4.86787051e-01 4.84625399e-01 -2.20174611e-01
-1.29106760e+00 -1.24610949e+00 7.58338690e-01 1.34173185e-01
7.38735974e-01 -9.66462791e-02 -7.91220427e-01 4.68289763e-01
1.80358633e-01 2.93317914e-01 6.33297265e-01 -3.60054851e-01
-2.74000287e-01 -5.50329208e-01 -1.42063761e+00 6.01589859e-01
1.38206506e+00 -4.66091454e-01 1.46081254e-01 2.70427633e-02
3.65323365e-01 -1.65566742e-01 -1.01151049e+00 2.44251519e-01
7.69751966e-01 -8.26014102e-01 7.27010250e-01 -4.09005135e-01
2.98721373e-01 -5.76373994e-01 -2.50610262e-01 -9.95591938e-01
-3.11719030e-02 -7.98146486e-01 3.47176582e-01 1.58374143e+00
2.51825243e-01 -7.51154602e-01 8.20469737e-01 7.89346874e-01
3.05895746e-01 -7.54017770e-01 -9.84463930e-01 -9.72082317e-01
-1.15005322e-01 -2.03220025e-01 7.81024218e-01 1.16833377e+00
-2.47253761e-01 -1.66597977e-01 -3.94360244e-01 2.52439559e-01
9.22410548e-01 1.91468894e-02 6.75097823e-01 -1.07992053e+00
1.53152481e-01 -4.84461755e-01 -7.63267398e-01 -7.69962490e-01
7.60350049e-01 -1.01399934e+00 -7.61102065e-02 -1.22725832e+00
1.27828807e-01 -6.43225908e-01 -1.63150012e-01 6.15066111e-01
1.01863444e-01 4.12636459e-01 2.59389281e-01 1.70106277e-01
-4.32508379e-01 7.11601555e-01 1.11013782e+00 -1.92802444e-01
5.47698662e-02 -2.36798927e-01 -5.54401517e-01 7.50637174e-01
1.03384304e+00 -3.54140013e-01 -4.24483240e-01 -4.23724949e-01
-1.20751850e-01 1.46059389e-03 4.33816105e-01 -9.16056693e-01
4.21925366e-01 -3.60286891e-01 1.28601030e-01 -1.14847124e-01
4.38973814e-01 -9.78549659e-01 4.74404603e-01 2.10202575e-01
-1.59618139e-01 2.84271032e-01 2.97854930e-01 3.76074195e-01
-3.60827297e-01 -1.51462942e-01 9.74267602e-01 1.02696836e-01
-1.06816423e+00 7.26800799e-01 -9.39803720e-02 -1.64441951e-02
1.30059350e+00 -5.48936903e-01 -2.82844067e-01 -4.26090688e-01
-6.79970801e-01 2.94453084e-01 3.98069918e-01 4.03374583e-01
5.80268562e-01 -1.66821182e+00 -6.60324991e-01 7.43388291e-03
1.66900188e-01 -4.59751844e-01 3.55267435e-01 8.71345401e-01
-6.14953399e-01 -3.28500308e-02 -4.97476488e-01 -4.52941477e-01
-1.65434110e+00 1.15491450e-01 8.79634678e-01 2.98440844e-01
-1.36091247e-01 8.84328783e-01 1.40494443e-02 -3.85158718e-01
-1.47190262e-02 4.37521696e-01 -1.90542996e-01 1.43556491e-01
4.07928884e-01 5.69708705e-01 -3.55547786e-01 -1.29027414e+00
-5.27461708e-01 1.24530840e+00 2.32370302e-01 1.16882533e-01
1.07628417e+00 -3.90068144e-01 -3.73364151e-01 -4.26436365e-02
1.72352171e+00 -3.18508416e-01 -1.09567523e+00 -1.53620675e-01
-1.31617501e-01 -7.85424173e-01 -1.85997277e-01 -2.98097759e-01
-1.58748281e+00 7.04284132e-01 9.03941274e-01 -3.43425125e-01
1.52664220e+00 -2.97036886e-01 7.97395468e-01 1.50658444e-01
4.40545887e-01 -1.10239816e+00 -2.40354702e-01 2.05748692e-01
8.79744053e-01 -1.16675436e+00 -1.54015318e-01 -5.65588772e-01
-6.18523955e-01 1.06191933e+00 8.05684447e-01 -8.60281140e-02
9.14068937e-01 1.94853693e-01 2.42389187e-01 -1.04705773e-01
-2.99021959e-01 -3.49707484e-01 -5.08736707e-02 5.36170900e-01
2.04577044e-01 -2.21663564e-02 -4.52687442e-01 3.82245988e-01
-1.08684577e-01 2.54312485e-01 2.06191286e-01 6.73081815e-01
-1.64286137e-01 -1.04272366e+00 -2.37950206e-01 3.23957086e-01
-4.45844769e-01 1.88637733e-01 -3.35626930e-01 7.21583605e-01
3.71114194e-01 1.00787568e+00 2.56509691e-01 -6.38211489e-01
3.60510468e-01 -6.63661882e-02 3.46171558e-01 1.13302410e-01
-3.32774401e-01 -2.40106896e-01 3.57295722e-02 -7.13420928e-01
-7.11676657e-01 -7.00786889e-01 -1.41335404e+00 -7.88020551e-01
-1.84590563e-01 1.25146285e-01 6.17333651e-01 7.35806942e-01
5.80267429e-01 -1.38833746e-02 1.04915512e+00 -6.32357359e-01
-2.23473623e-01 -5.34776688e-01 -7.59881556e-01 5.83529472e-01
8.45262855e-02 -8.50251019e-01 -4.32590276e-01 1.86646223e-01]
|
[13.615278244018555, 1.5245046615600586]
|
a5dde0da-bb15-4b62-bc1c-437647a975da
|
localizing-moments-in-long-video-via
|
2302.13372
| null |
https://arxiv.org/abs/2302.13372v1
|
https://arxiv.org/pdf/2302.13372v1.pdf
|
Localizing Moments in Long Video Via Multimodal Guidance
|
The recent introduction of the large-scale long-form MAD dataset for language grounding in videos has enabled researchers to investigate the performance of current state-of-the-art methods in the long-form setup, with unexpected findings. In fact, current grounding methods alone fail at tackling this challenging task and setup due to their inability to process long video sequences. In this work, we propose an effective way to circumvent the long-form burden by introducing a new component to grounding pipelines: a Guidance model. The purpose of the Guidance model is to efficiently remove irrelevant video segments from the search space of grounding methods by coarsely aligning the sentence to chunks of the movies and then applying legacy grounding methods where high correlation is found. We term these video segments as non-describable moments. This two-stage approach reveals to be effective in boosting the performance of several different grounding baselines on the challenging MAD dataset, achieving new state-of-the-art performance.
|
['Bernard Ghanem', 'Alberto Mario Ceballos-Arroyo', 'Fabian Caba Heilbron', 'Mattia Soldan', 'Wayner Barrios']
|
2023-02-26
| null | null | null | null |
['video-grounding', 'video-understanding', 'natural-language-moment-retrieval', 'natural-language-visual-grounding']
|
['computer-vision', 'computer-vision', 'computer-vision', 'reasoning']
|
[ 2.93767810e-01 1.64126232e-01 -8.55618268e-02 -1.37933373e-01
-1.13280177e+00 -6.35682821e-01 6.52496934e-01 6.71912357e-02
-4.68814433e-01 3.87474656e-01 4.67329949e-01 -2.39546597e-01
1.13272421e-01 -3.46935838e-01 -8.68100762e-01 -3.22302371e-01
-4.48762700e-02 4.74287570e-01 4.52849060e-01 -3.66355062e-01
4.44120288e-01 5.00158556e-02 -1.48489618e+00 7.94477403e-01
5.96136510e-01 6.33799314e-01 2.27264553e-01 6.26304984e-01
-2.47109815e-01 6.68876886e-01 -5.54322064e-01 -5.84536731e-01
4.76778835e-01 -7.27854192e-01 -9.93884265e-01 2.15448856e-01
1.11382735e+00 -4.83469099e-01 -3.43211710e-01 9.63102281e-01
4.08656687e-01 2.56916851e-01 2.31800988e-01 -1.10396278e+00
-3.60638261e-01 6.08295858e-01 -4.62056994e-01 3.52725327e-01
8.76155734e-01 7.20012933e-02 1.26599562e+00 -9.44153726e-01
1.11003280e+00 1.27802873e+00 8.69637012e-01 6.40673637e-01
-1.24322069e+00 -2.84777135e-01 2.64779985e-01 1.94070041e-01
-1.39053309e+00 -5.36910355e-01 4.07809466e-01 -7.20110476e-01
1.15919900e+00 1.22971050e-01 5.74943304e-01 1.17367375e+00
9.26807076e-02 8.61781180e-01 8.58811557e-01 -6.05866790e-01
1.23809231e-02 -3.09443444e-01 2.67956764e-01 9.16743457e-01
3.06878984e-01 -1.71479896e-01 -1.01225674e+00 -8.77726600e-02
4.10314232e-01 -2.51222491e-01 -1.47261366e-01 -2.56636351e-01
-1.56158483e+00 6.38185024e-01 1.25021920e-01 4.96718973e-01
-4.02029514e-01 -3.00987922e-02 1.90140828e-01 2.59994209e-01
7.48745918e-01 8.07751477e-01 -3.12792152e-01 -4.21501905e-01
-1.57552099e+00 8.06014121e-01 6.54187620e-01 9.15626168e-01
6.73929513e-01 -4.64578688e-01 -7.59274185e-01 3.11154991e-01
1.78017065e-01 4.79001626e-02 2.79773563e-01 -8.63521218e-01
9.13239479e-01 5.90703189e-01 1.16373584e-01 -1.07068264e+00
-4.64775056e-01 -3.14529121e-01 -2.20827729e-01 -1.77538574e-01
8.04051101e-01 -1.09404966e-01 -9.29775894e-01 1.46995163e+00
2.48406559e-01 3.43160570e-01 -1.55389413e-01 7.82638848e-01
7.37973273e-01 5.31059146e-01 7.25649763e-03 -3.09214517e-02
1.39426303e+00 -1.16882944e+00 -6.50048494e-01 -3.51650119e-01
8.62800539e-01 -9.23648953e-01 1.41449082e+00 4.86166745e-01
-1.18188286e+00 -6.35807931e-01 -1.16272914e+00 -1.98312178e-01
-1.74939215e-01 6.33279905e-02 4.60738301e-01 2.31720954e-01
-1.15354526e+00 9.62231159e-01 -8.62208426e-01 -6.15623355e-01
3.07133734e-01 1.66149482e-01 -4.46569234e-01 -1.85804918e-01
-8.02549005e-01 8.64177823e-01 4.50570673e-01 -1.78078227e-02
-6.18329704e-01 -6.80724919e-01 -7.28415787e-01 -2.93176174e-01
6.81616247e-01 -7.64229774e-01 1.27606392e+00 -7.95906126e-01
-1.16420662e+00 1.22938561e+00 -2.65057892e-01 -6.92061543e-01
6.16580486e-01 -8.50412190e-01 -1.44117087e-01 3.26853812e-01
3.11085582e-01 8.96911144e-01 1.03309810e+00 -8.55384886e-01
-6.69828653e-01 -5.18072098e-02 2.04607934e-01 1.52196944e-01
-1.88850001e-01 2.38546610e-01 -7.94208407e-01 -8.78057957e-01
4.21629488e-01 -1.13472307e+00 -9.06348452e-02 -5.96011937e-01
-4.10911649e-01 -5.95474355e-02 5.80898464e-01 -9.47068632e-01
1.53709805e+00 -2.17143059e+00 5.59384823e-01 -1.50472194e-01
2.57642508e-01 1.80260316e-01 -3.16436291e-01 5.83736360e-01
1.10830896e-01 2.82383829e-01 1.18595310e-01 -6.24786317e-01
9.74617600e-02 1.75560936e-02 -4.26682055e-01 3.01633209e-01
5.66806316e-01 1.01552987e+00 -1.04946971e+00 -5.34201503e-01
-8.93954188e-02 9.94016528e-02 -9.39550102e-01 4.19947177e-01
-5.63944817e-01 5.05470872e-01 -2.33182684e-01 4.87686872e-01
3.27226847e-01 -1.86670706e-01 2.47412939e-02 -1.99234992e-01
-1.10442549e-01 5.13545215e-01 -1.08121192e+00 2.32479763e+00
1.67242259e-01 7.36740410e-01 -2.32971057e-01 -6.47101223e-01
4.81898040e-01 1.36723414e-01 4.80062723e-01 -6.67627454e-01
-1.26042411e-01 6.33797944e-02 3.74141335e-02 -6.61059260e-01
9.82944310e-01 4.68046553e-02 -3.27118114e-02 3.38849723e-01
3.64163548e-01 -2.38590464e-01 6.02737904e-01 4.32405561e-01
1.11489618e+00 6.85684681e-01 1.00838058e-01 -3.28393430e-01
3.08980763e-01 2.86602080e-01 3.07398230e-01 8.45187366e-01
-1.31709930e-02 9.51236844e-01 5.05158126e-01 -5.35530984e-01
-8.92346203e-01 -7.59867132e-01 4.60563183e-01 1.24933290e+00
-3.84573415e-02 -1.12115693e+00 -1.12634265e+00 -8.38115573e-01
-2.63493806e-01 5.12929738e-01 -7.12197125e-01 -6.33401200e-02
-8.01197231e-01 -5.56735098e-01 6.46664679e-01 1.85876414e-01
3.58009577e-01 -8.07103515e-01 -5.93573689e-01 3.26591074e-01
-6.31045401e-01 -1.57942820e+00 -3.44357789e-01 -8.88245329e-02
-8.96286190e-01 -1.00252330e+00 -6.08978033e-01 -3.28862071e-01
3.45148712e-01 3.98880720e-01 1.69660091e+00 4.52251315e-01
7.07930177e-02 1.76068336e-01 -7.08583236e-01 -2.70874143e-01
-3.75784516e-01 4.86640245e-01 -1.01666555e-01 -6.86721653e-02
4.49002713e-01 -4.45606112e-02 -4.55534309e-01 1.98960066e-01
-9.30002272e-01 2.49865919e-01 4.38020825e-01 5.16395390e-01
6.22482836e-01 -3.76534730e-01 1.94741622e-01 -6.20042920e-01
5.17971456e-01 -3.36596578e-01 -5.72717249e-01 7.84358606e-02
-2.27421299e-01 2.18176588e-01 1.36793479e-01 -2.52426803e-01
-4.70001221e-01 -1.48897246e-01 -1.72921777e-01 -2.90024608e-01
-1.21564101e-02 5.35808861e-01 8.98899063e-02 2.58039963e-02
6.03373528e-01 -1.88666463e-01 -2.42802635e-01 -6.23439431e-01
3.45491320e-01 1.46307230e-01 6.89488471e-01 -6.84652567e-01
8.56459200e-01 3.76220167e-01 2.34574080e-01 -6.76785290e-01
-1.43186986e+00 -7.41818428e-01 -8.65128338e-01 -4.81099933e-02
1.32481670e+00 -1.00399780e+00 1.73180904e-02 3.30908746e-01
-1.18030512e+00 -4.71327990e-01 -2.71569639e-01 3.70143175e-01
-5.17064512e-01 5.74271560e-01 -4.80710864e-01 -3.60976636e-01
-1.10403158e-01 -1.09081686e+00 1.40649021e+00 -1.81191608e-01
-5.95185697e-01 -6.81647062e-01 1.82006717e-01 5.85984230e-01
1.21474527e-01 3.24645162e-01 5.85535228e-01 -7.54318476e-01
-6.72565579e-01 -2.92424541e-02 -1.19700823e-02 2.32736856e-01
-1.19496845e-01 9.25109386e-02 -9.41814780e-01 -2.55083531e-01
-1.90753713e-01 -1.72851220e-01 8.95804465e-01 2.75782883e-01
6.79479241e-01 -1.13444537e-01 -7.02228099e-02 7.48482823e-01
1.03567183e+00 -2.67136276e-01 7.44543672e-01 7.11287320e-01
7.32333481e-01 6.32270932e-01 9.61454272e-01 2.36166209e-01
3.36241305e-01 1.09021759e+00 4.67043430e-01 -5.61142638e-02
-2.69431531e-01 -4.03065324e-01 5.08229792e-01 7.82496750e-01
-1.55839458e-01 -2.66883820e-01 -1.12363255e+00 4.64189917e-01
-2.05626273e+00 -1.16098547e+00 -2.38286331e-01 2.06229329e+00
6.41251802e-01 4.09164846e-01 3.29155087e-01 2.75736041e-02
2.90051788e-01 1.97749123e-01 1.43073454e-01 -7.12113306e-02
-1.71486661e-01 -9.22467373e-03 1.40451878e-01 4.12845433e-01
-1.35573113e+00 1.26407123e+00 6.78147364e+00 6.29883170e-01
-1.02055097e+00 7.62476176e-02 4.13400859e-01 -2.54794717e-01
-8.00237730e-02 9.16562006e-02 -1.06941605e+00 3.70921671e-01
1.01925361e+00 1.92114726e-01 4.28419203e-01 5.03694296e-01
3.39581341e-01 -2.35023871e-01 -1.41488981e+00 1.05820370e+00
2.63843060e-01 -1.54960740e+00 1.94466889e-01 6.20474340e-03
8.59395862e-01 1.81498781e-01 -1.97250977e-01 2.10662350e-01
-1.23697244e-01 -9.29841995e-01 1.24330795e+00 4.38094050e-01
5.55491805e-01 -3.19620937e-01 7.55134940e-01 2.26434693e-01
-9.52635527e-01 2.32686296e-01 -1.32365525e-01 -4.22184020e-01
2.50214875e-01 3.68006617e-01 -8.45095515e-01 8.16855788e-01
6.70084536e-01 6.71630859e-01 -9.72399771e-01 9.13066924e-01
-1.20160125e-01 6.62324667e-01 -2.10425615e-01 4.55344439e-01
7.27075577e-01 -2.13684455e-01 6.41777217e-01 1.40026581e+00
4.43014830e-01 -1.35812670e-01 4.58367050e-01 6.38132691e-01
-5.86043857e-03 8.23381692e-02 -6.69532478e-01 -2.66916245e-01
6.84769973e-02 1.02700818e+00 -9.99539077e-01 -3.60005826e-01
-5.03342628e-01 1.03832853e+00 1.66176289e-01 2.34962791e-01
-8.33338737e-01 -1.85108706e-02 4.50801134e-01 4.06736165e-01
4.10286188e-01 -5.90164959e-01 -4.65557538e-02 -1.37799203e+00
1.77885920e-01 -1.33668923e+00 4.54234213e-01 -8.21975529e-01
-8.22620034e-01 6.04046643e-01 1.51971206e-01 -1.16674507e+00
-5.17891824e-01 -4.44999933e-01 -2.99771577e-01 5.15669644e-01
-1.28608990e+00 -1.30012977e+00 -4.68516737e-01 6.21074498e-01
7.65029609e-01 -1.12514876e-01 5.41535497e-01 3.07366222e-01
-3.23198527e-01 3.71445745e-01 -2.66299397e-01 -9.16163400e-02
9.29353595e-01 -1.23954892e+00 7.78000057e-01 1.41768289e+00
7.35966444e-01 7.93006718e-01 9.95037735e-01 -7.71376431e-01
-1.35147679e+00 -9.50443685e-01 1.04326749e+00 -8.89446378e-01
8.93553615e-01 -5.32134235e-01 -8.15057576e-01 6.85808599e-01
3.05194080e-01 -1.65699422e-01 4.13735598e-01 1.93298221e-01
-3.39422613e-01 3.12146813e-01 -6.45013332e-01 6.88500345e-01
1.29745901e+00 -6.74873054e-01 -9.25959945e-01 3.41588318e-01
8.84733438e-01 -6.34201705e-01 -3.72082889e-01 1.48812339e-01
4.07591343e-01 -1.08732927e+00 9.64828670e-01 -8.10548544e-01
6.62764907e-01 -3.93975526e-01 -3.91717434e-01 -1.06144762e+00
-2.52451390e-01 -1.34022737e+00 -2.01270983e-01 1.28446615e+00
2.68780559e-01 6.35305792e-02 8.21988702e-01 1.30374730e-01
-4.14558232e-01 -6.09213591e-01 -7.55182266e-01 -8.24045002e-01
-1.58006579e-01 -5.78760445e-01 4.91495758e-01 8.05610895e-01
-6.97492808e-02 4.42013115e-01 -4.22779769e-01 -2.72798538e-03
3.45199347e-01 2.12799534e-02 1.06199563e+00 -1.05270743e+00
-3.76897305e-01 -2.03144908e-01 -4.42184687e-01 -1.14029765e+00
1.54089913e-01 -7.23002434e-01 7.67903030e-02 -1.56908119e+00
2.09701195e-01 -2.99558556e-03 -1.42161310e-01 2.98632175e-01
-4.33234692e-01 5.10885119e-01 5.98619699e-01 3.06608826e-01
-1.03343105e+00 4.59550060e-02 9.82754409e-01 -2.54109874e-02
-2.21926838e-01 -3.09504211e-01 -5.50706446e-01 9.30102050e-01
5.11565626e-01 -5.17452598e-01 -4.22147155e-01 -7.80073881e-01
7.01204836e-01 -3.42577785e-01 4.62513834e-01 -1.24807394e+00
2.14324726e-04 1.27373924e-02 -1.38541535e-01 -8.32919717e-01
8.31580684e-02 -5.45275927e-01 7.03507289e-02 2.08550841e-01
-1.30348817e-01 6.17318034e-01 3.69259983e-01 2.84738213e-01
-3.11478347e-01 -2.94262290e-01 3.69117200e-01 -7.50659704e-02
-1.02275503e+00 7.92778805e-02 -1.68074310e-01 5.32092690e-01
7.56016731e-01 -3.10585976e-01 -2.69348383e-01 -2.53649890e-01
-6.60616636e-01 -1.65073618e-01 7.13946819e-01 5.89164972e-01
1.94002837e-01 -1.11343658e+00 -7.69136608e-01 5.77697530e-02
6.40804619e-02 -9.36183184e-02 -4.12555505e-03 1.02876437e+00
-7.20516622e-01 4.68965501e-01 -1.78328529e-01 -7.54122436e-01
-1.37084579e+00 5.22046149e-01 2.29963697e-02 -6.90422118e-01
-6.62648797e-01 9.69569087e-01 4.01035249e-02 1.75048485e-01
1.25069141e-01 -6.69362485e-01 5.59874624e-02 4.13584292e-01
5.31960130e-01 2.53942013e-01 5.25992036e-01 -7.45652318e-01
-4.16012615e-01 6.70553744e-01 -8.16176087e-02 -2.86471486e-01
1.27965522e+00 -1.92518607e-01 -9.65979043e-03 3.58184963e-01
9.23777878e-01 2.63855785e-01 -1.30932283e+00 1.93543243e-03
5.10002315e-01 -4.94968295e-01 -1.06169209e-01 -5.06449103e-01
-4.63329434e-01 7.52000690e-01 1.90431431e-01 2.26473346e-01
8.45563531e-01 6.37120232e-02 8.16154122e-01 4.02484238e-01
3.76251370e-01 -1.02913284e+00 3.78782749e-01 7.62697577e-01
7.89796412e-01 -1.19553578e+00 3.01522672e-01 -3.88120294e-01
-4.18317974e-01 9.71450984e-01 3.52245361e-01 -8.14452246e-02
-3.14635457e-03 3.22557986e-02 7.72560462e-02 -3.44517022e-01
-6.77671373e-01 -2.94999599e-01 7.47859061e-01 3.73356819e-01
5.17811000e-01 -3.04707259e-01 -3.10491949e-01 5.11164427e-01
-3.62318665e-01 -4.18614224e-02 4.16098863e-01 9.03852701e-01
-3.25481087e-01 -1.13479030e+00 -3.55250448e-01 1.84974864e-01
-9.10572290e-01 -3.94983113e-01 -5.55114448e-01 9.43376720e-01
2.85622120e-01 1.02844691e+00 2.05036253e-02 -4.63628322e-01
4.14056599e-01 1.60128072e-01 6.06840491e-01 -8.31377029e-01
-8.46458912e-01 1.97515130e-01 3.19460005e-01 -1.07851851e+00
-5.82843840e-01 -8.51382554e-01 -9.71153498e-01 -1.40019119e-01
-1.86091885e-01 5.48129790e-02 3.83493870e-01 1.24610865e+00
4.93984461e-01 3.98360789e-01 -1.06566593e-01 -1.21414363e+00
-5.33280492e-01 -8.85887682e-01 -7.53152475e-04 8.72391880e-01
3.53342175e-01 -6.23219967e-01 1.23775918e-02 4.97387350e-01]
|
[10.249101638793945, 0.7987715601921082]
|
76eaa639-44b6-4672-a9ed-a0b70d29ad1c
|
improve-document-embedding-for-text
|
2006.00572
| null |
https://arxiv.org/abs/2006.00572v1
|
https://arxiv.org/pdf/2006.00572v1.pdf
|
Improve Document Embedding for Text Categorization Through Deep Siamese Neural Network
|
Due to the increasing amount of data on the internet, finding a highly-informative, low-dimensional representation for text is one of the main challenges for efficient natural language processing tasks including text classification. This representation should capture the semantic information of the text while retaining their relevance level for document classification. This approach maps the documents with similar topics to a similar space in vector space representation. To obtain representation for large text, we propose the utilization of deep Siamese neural networks. To embed document relevance in topics in the distributed representation, we use a Siamese neural network to jointly learn document representations. Our Siamese network consists of two sub-network of multi-layer perceptron. We examine our representation for the text categorization task on BBC news dataset. The results show that the proposed representations outperform the conventional and state-of-the-art representations in the text classification task on this dataset.
|
['Hadi Veisi', 'Erfaneh Gharavi']
|
2020-05-31
| null | null | null | null |
['document-embedding', 'text-categorization']
|
['methodology', 'natural-language-processing']
|
[-0.06163628 -0.09410255 -0.26591522 -0.62888914 -0.8051893 -0.16521281
0.95697916 0.64822304 -0.4282201 0.39360455 0.88595223 -0.01655727
-0.35017645 -0.66912806 -0.35810277 -0.45762295 0.05968419 0.8041328
0.18420334 -0.15960604 0.7606781 0.36496252 -1.6340264 0.73077357
0.64264065 1.1831114 0.28683126 0.7455993 -0.96527517 0.6587279
-0.81235373 -0.18038924 -0.12177745 -0.30212757 -1.1675308 -0.10378952
0.3559804 -0.13305521 -0.45315272 1.0457269 0.27685618 0.680954
1.336025 -1.1487877 -1.1182656 0.77377325 -0.67921007 0.40951282
0.15046073 -0.9818531 1.3690072 -0.96628684 0.52029717 1.7549664
0.12468615 0.25008386 -0.7717459 -0.4084677 0.45463058 0.24588068
-1.0626055 -0.08276176 0.9235063 -0.34418046 0.99798936 0.11561121
0.20762996 1.1245229 0.5279403 1.0128663 0.34343037 -0.39761552
0.37984493 0.3746098 0.8908251 0.3363503 0.41319 -0.8012225
-0.48383242 -0.40048504 0.15118451 0.60830396 -0.02590777 -0.35144132
-1.0212293 1.2355376 0.5905888 0.54639816 -0.5312331 0.25160295
0.79240185 0.19661367 0.879043 0.3892454 -0.26850095 0.06009178
-0.88990057 0.15983662 0.7254767 0.7540902 0.49872652 -0.18899651
-0.33391395 1.1655401 0.7983194 0.13048342 1.4997115 -0.36417368
0.9783898 0.8844867 -0.20294471 -1.4482371 -0.46166494 -0.42294106
-1.1924459 -0.1586201 -0.24466977 0.07994013 -0.57469004 1.1149498
0.11010787 -0.30077782 0.5273343 0.7736685 0.7888913 1.4046401
0.18204314 0.17952882 1.5962785 -1.2035853 -0.6701921 -0.18880652
0.71277136 -0.67788225 0.9054873 0.14698142 -0.63221085 -0.5327299
-0.9679231 -0.46979535 -0.7860539 0.33478573 0.5084393 0.31658918
-0.6759606 0.4310475 -0.51294506 -0.77377176 0.43927783 0.09870026
-0.3683222 -0.09296387 -1.1739925 0.5882087 0.76191914 -0.15860045
-0.47840616 -0.35539085 -0.75890446 0.61406523 -0.27654037 -0.32547688
0.9462042 -0.9952512 -1.3676547 0.6270261 -0.21059622 -0.45500764
-0.02152467 -0.25901678 -0.40428665 0.38111633 0.14864266 0.68631655
1.0294595 -0.8649924 -0.70783854 -0.49604043 -0.4935292 0.53121275
-0.99420714 0.17399223 -0.26487002 -0.74785036 0.443398 -0.5535684
0.01335219 0.01254903 -0.2911377 -0.9214173 1.2787304 -0.5530502
1.0249052 -2.0900297 0.25322506 0.07319293 0.16557986 -0.09839015
-0.11286753 0.53334814 -0.1292186 -0.01899757 0.10875303 -0.6644311
0.28572184 0.0076639 -0.8697665 0.5598238 -0.10274776 0.64758545
-0.5070755 -0.6358658 -0.06386584 0.41904655 -0.4779456 0.14006782
-0.29367736 -0.31588802 -0.95383304 0.22693476 0.5818249 -0.49128732
-0.07228051 0.03495049 0.25803134 0.3210967 -1.1529897 1.8692933
-0.37824926 0.93525416 -0.16380017 -1.400719 1.230184 0.24140307
0.3085054 -0.44441548 0.12050368 -0.02457267 -0.3443726 -0.26296958
0.95831186 -0.02125349 -0.32797348 1.1301926 0.20837164 0.15329336
-0.0597738 0.6381524 0.636628 -0.41405892 0.31885427 -0.5687385
0.67385054 -0.14976484 -0.11897481 0.48034897 -0.09893347 0.4766716
0.56767124 -0.53116375 -0.9124897 -0.50283253 -0.24080777 1.7358652
-0.00691183 -0.3479738 -0.32149488 -0.7116598 0.28923512 0.7229144
-0.7425635 -0.36523807 -0.2222505 -0.73497385 0.37131444 0.5689792
0.35523722 -0.9996582 -0.25781104 0.1269846 -0.02908263 -0.5561704
-0.48634103 0.39497802 -0.9433515 -0.5925571 -1.1161023 -1.1343266
0.67506194 0.43981668 0.69892955 -0.3053235 -0.01534057 0.28384992
-0.5778211 -0.22964706 -0.42240223 0.574977 -0.09532077 0.05319652
0.6817972 -0.13183504 -0.5924798 0.02668369 -1.0287919 -0.17167732
0.40164092 0.8342422 0.28826752 0.20334233 0.8532503 -0.7385922
1.3030914 -0.95912504 -0.34711763 0.22488101 -0.6143905 0.5222477
0.9759262 -0.346231 -1.1052533 -0.23807956 0.14735258 -0.10003594
-0.10170337 0.68107176 0.25787044 0.6389967 0.67389005 0.5421602
-0.22844756 -0.72129035 0.5232152 1.3625442 0.1044201 -0.5140601
0.22914994 0.3094592 -0.27803108 -0.84739333 -0.96506315 -0.9520272
-0.44269294 0.38434148 0.7833361 -0.73387694 -0.38226965 0.08103812
-1.4624109 0.6388471 -0.23782425 0.76956433 -0.34540412 0.49640673
-0.59832 -0.6307724 -0.8190057 -0.95827174 1.2778585 0.05824336
-0.26021054 -1.275873 0.44385555 0.12519032 0.58916163 -0.57791877
1.2853854 -1.3838145 -0.03557199 -0.53109556 -0.54656345 0.26959437
0.18610185 -0.39183182 -1.0077119 -0.3873924 -0.03410842 -0.4892157
1.326582 0.37964532 1.418401 -0.44368237 -0.47292206 0.30299613
1.2469535 -0.04922493 0.26203462 0.47462186 0.5604081 0.8215653
0.25455275 0.54682183 0.3269432 0.3202834 0.19692013 0.13677533
0.15555815 -0.30050275 0.15228167 0.9857713 0.6327911 -0.699303
-0.79838765 0.40011784 -1.9809009 -0.8780414 0.2457805 1.6203501
0.5443875 -0.01338006 -0.1975939 0.13586357 0.9039312 0.3467879
-0.48411775 -0.7320313 0.11213119 -0.33509272 0.02824118 0.09079809
-1.2679107 0.87200516 6.1212 0.79984844 -0.9651641 -0.07128812
0.6690667 0.02400816 -0.13274816 -0.45153677 -1.0425416 0.4149025
1.1867518 -0.52361214 -0.18291779 1.2587538 -0.04142844 0.17672208
-1.1790811 1.007525 0.66371965 -1.4426361 0.82927614 -0.15771797
0.7533317 -0.08524141 0.22180775 0.5005466 0.25670436 -0.9955787
0.41481164 0.5174595 0.25423744 -0.77889544 0.97246003 0.3427228
-1.0736684 -0.16748106 -1.1571106 0.2007425 -0.16094622 0.4904375
-0.7212796 0.22820793 0.6113429 1.0201379 -0.5223406 0.73331046
0.35907486 0.3896188 -0.11453728 -0.6434043 0.63465995 -0.11463146
0.16469912 1.3710573 0.3657253 -0.30637604 0.13219817 0.66767627
-0.5686679 0.7114642 -0.57350725 -0.29941833 0.23206955 1.0815256
-0.77697396 -0.7159892 -0.3589457 1.2075894 0.59781355 0.20826133
-0.2390706 -0.9515698 0.43337157 -0.41386753 0.24528855 -0.04529316
-0.17308623 -1.0177784 -0.01285274 -0.5092672 0.67666966 -0.5834727
-1.5714259 0.87920356 0.03740744 -1.1078273 -0.44462883 -0.7204031
-0.78056866 1.0325073 -1.5170442 -0.9450687 -0.13259639 0.6351671
1.0758154 -0.830531 1.0046611 -0.09492294 -0.48489884 0.31817546
0.9674848 0.05939194 0.67992604 -1.1809034 0.41346222 0.4016911
0.13021874 0.80816644 0.32248 -0.53272074 -1.2085317 -1.200401
0.935946 -0.07849485 0.80187786 -0.3934794 -1.0903267 0.6907808
0.3338559 -0.19444992 0.9618054 0.18587281 -0.43720722 -0.09538966
-0.9690199 0.3550376 0.24420974 -0.5894676 -1.2039835 0.6535657
0.9489821 0.37455365 -0.6237821 -0.6831606 0.35578278 -0.2986248
0.8849676 -1.0234364 0.6425071 0.13473655 -0.51907873 -1.6234893
-0.41577396 -0.05616115 -0.17036502 1.2699394 0.29049402 -0.667041
0.8523667 0.3887207 -0.00993008 -0.37712127 -0.95926 -0.37914526
0.60054165 -0.10014167 0.4379619 0.9069588 0.4651758 0.74259776
-0.07394472 -0.16885084 0.48298278 0.30343115 0.2969099 -1.7329557
0.09968812 -0.62938845 -0.60705173 -1.3858988 0.7742081 -1.430332
-0.09908778 -1.8701643 0.4905677 -0.23598595 -0.42193446 0.06257316
0.08279387 -0.29083374 -0.09686372 0.56496626 -0.8260698 1.1214757
0.94949955 -0.540203 -0.07152399 -0.01940819 -0.9366376 0.56276006
0.8742458 -0.6988939 -0.491464 -0.61913514 0.14214483 -0.21413432
-0.21326506 -0.8380173 0.6510467 0.04445136 0.7167658 -1.0705036
0.4078069 -0.8841541 -0.7056775 0.29969084 -1.1916804 0.08650379
0.01145274 0.7878493 -0.5406355 -0.68300635 0.7451905 -0.04106103
-0.4751487 0.14326978 -0.64729816 -0.08094998 0.8014756 0.05201596
-0.6087958 -0.28955585 -0.18740727 0.18423522 -0.08198475 0.6862085
0.88029265 -1.2617835 -0.81384355 0.27556404 0.34461933 -0.1355212
0.17515537 -0.04362755 -0.44208118 1.0709755 -0.24378689 -0.40108562
-0.88285315 0.6388148 0.1171384 -0.32241166 -0.89457726 0.69627047
0.35029194 -0.4548513 0.5020036 -0.5590143 -0.80218184 0.3169961
0.89933497 0.37943485 0.1476866 -0.5886195 -0.24914822 0.56296635
-0.7073353 0.01280177 1.2696792 -0.26938492 -0.18973526 0.56131536
1.9133652 -0.502143 -0.5807334 -0.4531227 0.21192658 -0.10952131
0.54890776 -0.31091785 -0.81753844 1.1910868 0.5545914 0.2833665
0.5548093 0.25889045 0.5593746 1.1919512 -0.2679016 -1.2907991
0.41613743 0.7355249 1.1342065 -1.2274333 0.07873322 0.15788938
-0.8144944 1.5064625 0.3557379 -0.24600448 0.95123357 -0.29798314
-0.16215576 -0.34306157 -1.004972 0.3091354 0.5911361 0.41166517
0.6179084 -0.20594476 -0.10228936 0.68147993 -0.2922737 -0.5797012
0.3905735 0.90504056 -0.9447697 -0.5505392 -0.45870632 0.7145988
-0.39007288 -0.01463342 -0.34049392 0.23653714 -0.7807382 0.89129066
0.48350742 0.10953567 0.0374138 0.4325502 -0.36495873 -0.783085
-0.35496524 -0.20857814 -0.30272758 -0.12582761 -0.07166041 -0.3794118
-1.4591416 -0.09472027 -0.20236236 0.6404912 1.1095923 0.7696006
0.4912015 0.66404045 0.7134963 -0.75565904 -1.0136734 -1.2977844
-1.087839 0.4825841 0.2291545 -0.52951115 -0.44676068 -0.08058945]
|
[10.470613479614258, 6.925899982452393]
|
dd11fc49-51dc-458f-b594-0574c93fee4d
|
machine-learning-based-classification-of-3
|
2212.04684
| null |
https://arxiv.org/abs/2212.04684v1
|
https://arxiv.org/pdf/2212.04684v1.pdf
|
Machine Learning-based Classification of Birds through Birdsong
|
Audio sound recognition and classification is used for many tasks and applications including human voice recognition, music recognition and audio tagging. In this paper we apply Mel Frequency Cepstral Coefficients (MFCC) in combination with a range of machine learning models to identify (Australian) birds from publicly available audio files of their birdsong. We present approaches used for data processing and augmentation and compare the results of various state of the art machine learning models. We achieve an overall accuracy of 91% for the top-5 birds from the 30 selected as the case study. Applying the models to more challenging and diverse audio files comprising 152 bird species, we achieve an accuracy of 58%
|
['Richard O. Sinnott', 'Yueying Chang']
|
2022-12-09
| null | null | null | null |
['audio-tagging']
|
['audio']
|
[-3.73402763e-05 -6.63486123e-01 1.34238794e-01 -2.18101680e-01
-6.88220024e-01 -7.48643339e-01 1.98776871e-01 2.27058813e-01
-6.56025887e-01 3.56644124e-01 4.68277693e-01 1.40688438e-02
1.55867487e-02 -2.98031151e-01 6.72946274e-02 -3.64363194e-01
-5.69765329e-01 -7.41103012e-03 2.62879521e-01 -2.68084943e-01
3.41892809e-01 4.57094222e-01 -2.06409526e+00 8.08302164e-01
1.46200418e-01 9.13414955e-01 -4.09662724e-02 1.44774067e+00
2.99979031e-01 6.97900832e-01 -9.43805993e-01 -2.52570689e-01
8.17494839e-02 -1.19895108e-01 -9.34371769e-01 -2.96669960e-01
7.58899450e-01 -1.36489198e-01 -7.76099274e-03 5.79078913e-01
7.95679390e-01 3.69600058e-01 5.64251423e-01 -1.04313850e+00
-3.76229614e-01 9.63658333e-01 -1.89340249e-01 8.34693551e-01
5.60992360e-01 -8.34152102e-02 1.25769484e+00 -7.07307100e-01
-6.82841241e-02 9.07687187e-01 1.41057301e+00 4.86235768e-01
-1.07545865e+00 -9.10518169e-01 -3.92818630e-01 3.43386739e-01
-1.47235954e+00 -7.11584151e-01 5.50316095e-01 -6.89187944e-01
1.42653334e+00 3.63935679e-01 1.01140201e+00 2.37353742e-01
-2.43584022e-01 4.68218893e-01 9.19917941e-01 -6.49427354e-01
3.64981294e-02 9.30771381e-02 -3.80137637e-02 3.66222858e-01
-5.36147356e-01 3.52857530e-01 -8.49744380e-01 -6.28721178e-01
5.40857017e-01 -4.44656968e-01 4.25137319e-02 6.45335257e-01
-1.32826829e+00 8.50234032e-01 3.23806494e-01 2.88192272e-01
-4.71494496e-01 2.24819213e-01 6.75816953e-01 4.66520071e-01
4.38688785e-01 8.77305806e-01 -8.18878889e-01 -5.20265281e-01
-1.27279282e+00 3.95880669e-01 1.01905465e+00 5.39398253e-01
2.40903273e-01 6.56011522e-01 2.45632023e-01 1.45713234e+00
1.95924073e-01 1.89416140e-01 7.49079823e-01 -8.16579700e-01
-1.22689374e-01 -1.33719936e-01 -1.19610757e-01 -8.66114676e-01
-4.12442893e-01 -4.12227154e-01 -3.96938920e-01 -2.06821278e-01
7.28244036e-02 -2.77363032e-01 -7.08394349e-01 1.17030621e+00
3.43081921e-01 4.70374286e-01 -2.27889895e-01 7.33468711e-01
1.15194166e+00 9.09949303e-01 2.05969989e-01 -1.37901157e-01
1.30367482e+00 -7.80401230e-01 -4.80438381e-01 3.94248031e-02
2.46125072e-01 -1.33135080e+00 8.20727706e-01 9.04473066e-01
-7.43827224e-01 -9.18519199e-01 -9.65911984e-01 3.32733810e-01
-5.10151982e-01 5.09359062e-01 7.51034319e-01 9.81207848e-01
-1.05561781e+00 3.90847266e-01 -5.06071627e-01 -2.02869207e-01
6.38225004e-02 6.84383094e-01 -3.90861660e-01 6.26230299e-01
-8.90382469e-01 5.80067575e-01 4.21209693e-01 -2.04555720e-01
-9.77222085e-01 -1.01912916e+00 -5.90376973e-01 -3.16962928e-01
-3.26598138e-01 -1.18764024e-02 1.68618488e+00 -6.34837151e-01
-1.34209311e+00 8.13061476e-01 3.50581765e-01 -7.83062339e-01
-3.16685766e-01 -4.88863468e-01 -8.37288260e-01 1.33797497e-01
-2.42986798e-01 9.03020918e-01 1.09559226e+00 -5.50992668e-01
-1.25880015e+00 6.07662424e-02 -2.59911358e-01 -3.28024887e-02
-3.39896291e-01 7.72928774e-01 5.11157691e-01 -1.05459356e+00
-2.09597737e-01 -9.75126088e-01 1.30797580e-01 -6.11216784e-01
2.68040061e-01 -1.19608738e-01 6.21716678e-01 -9.83563304e-01
1.57932305e+00 -2.30690932e+00 -1.92236900e-01 -1.06538542e-01
-3.00274521e-01 3.77804130e-01 -1.22293778e-01 4.84356344e-01
-4.14643101e-02 -1.77882724e-02 -5.34840561e-02 -3.10938433e-02
-2.36801863e-01 1.83989003e-01 -4.79843706e-01 3.81887674e-01
2.61572272e-01 1.05021641e-01 -8.24858367e-01 -1.58529982e-01
2.91202545e-01 6.19764507e-01 -9.94733870e-01 2.83369958e-01
2.31118768e-01 7.28162378e-02 4.70344126e-01 7.88290977e-01
2.64351159e-01 1.05213761e+00 -3.27850580e-01 -1.19232990e-01
-3.87274593e-01 7.35377133e-01 -1.28907204e+00 1.17073607e+00
-6.11729860e-01 8.60389233e-01 3.59969944e-01 -7.01563835e-01
9.37720656e-01 7.24763989e-01 2.69818842e-01 4.00370449e-01
6.22744998e-03 3.59908015e-01 6.08883202e-01 -4.48923737e-01
6.77917480e-01 -5.31222105e-01 -5.80626912e-02 2.81960219e-01
8.12222421e-01 -6.59112573e-01 1.91913724e-01 -2.95948625e-01
6.10682189e-01 -3.21282238e-01 4.30561334e-01 -3.99559468e-01
5.89921713e-01 6.10534362e-02 3.01691175e-01 4.44548279e-01
-3.93736869e-01 5.99567890e-01 -2.83448130e-01 -5.79177320e-01
-8.60191703e-01 -5.57776690e-01 -2.53302366e-01 1.95871484e+00
-1.03827953e+00 -7.25146413e-01 -6.35486543e-01 -1.96860597e-01
1.75404668e-01 4.45162743e-01 -4.06927168e-01 -6.02082051e-02
-4.24056739e-01 -7.45793819e-01 1.24040914e+00 7.40674019e-01
3.54693532e-02 -1.44248486e+00 -6.05827689e-01 5.18996358e-01
-8.17538984e-03 -7.75035739e-01 -5.79300225e-01 3.88397604e-01
-7.07548857e-01 -7.89799869e-01 -3.49236757e-01 -1.03966820e+00
-2.66549587e-01 5.22954389e-02 1.12091029e+00 3.33501622e-02
-6.48041427e-01 5.03006935e-01 -7.31912434e-01 -9.18363154e-01
-5.77998519e-01 1.49694934e-01 7.12190390e-01 -1.33346260e-01
4.40854341e-01 -5.50221443e-01 -1.15705155e-01 1.52291775e-01
-6.43623710e-01 -7.04790533e-01 1.95005134e-01 9.43310082e-01
4.06554371e-01 1.95033401e-01 9.79508162e-01 -4.39713210e-01
6.61827862e-01 -2.60665298e-01 -3.42318833e-01 -3.31382811e-01
-1.70889199e-02 -7.29737699e-01 6.08593464e-01 -8.26227367e-01
-4.14518446e-01 4.92043287e-01 -7.43784606e-01 -1.17828466e-01
-8.13625693e-01 4.93487507e-01 5.14382899e-01 -3.85682046e-01
8.51010859e-01 1.37447044e-01 -2.52733439e-01 -1.00840986e+00
2.61205852e-01 1.36164260e+00 8.82335901e-01 -1.62453189e-01
3.12557042e-01 -8.97931121e-03 -3.80092233e-01 -1.67102778e+00
-7.14917898e-01 -1.01091814e+00 -1.00939870e+00 -3.91590416e-01
5.13484120e-01 -9.32453632e-01 -4.48805481e-01 6.17528319e-01
-7.38833845e-01 1.33093581e-01 -3.38342100e-01 9.98006880e-01
-5.32974184e-01 -1.51493862e-01 -7.03312099e-01 -1.45556092e+00
-4.65980619e-01 -5.94365537e-01 9.48544919e-01 9.01471451e-02
-5.98085940e-01 -7.68106759e-01 5.70698559e-01 2.98721105e-01
3.87842685e-01 -1.06890015e-01 6.88830614e-01 -1.04214549e+00
5.98817050e-01 -3.82565886e-01 5.52700162e-01 6.54853284e-01
3.22312802e-01 4.18817848e-01 -1.61295807e+00 -2.93426335e-01
-3.04125994e-01 -3.92912954e-01 9.19974029e-01 4.29526001e-01
8.61956060e-01 -3.83018374e-01 3.93611401e-01 3.80913883e-01
8.51837993e-01 4.29846376e-01 9.11853686e-02 -5.22604920e-02
2.79427081e-01 6.39908254e-01 5.88563383e-01 4.57433999e-01
-7.82018751e-02 5.15501440e-01 1.48931965e-01 3.52392316e-01
-1.02457121e-01 -4.70885605e-01 5.19543171e-01 1.28501749e+00
-2.22871035e-01 3.25428665e-01 -1.07600403e+00 1.07858419e+00
-1.23177934e+00 -1.15957677e+00 -6.08034432e-02 2.04240870e+00
9.69558477e-01 -2.57110089e-01 1.01419115e+00 1.02554405e+00
6.25530183e-01 -9.69074592e-02 2.91784555e-01 -8.71965468e-01
2.81247944e-01 6.52721524e-01 8.38417411e-02 4.10527796e-01
-1.77844858e+00 9.51146662e-01 8.44048786e+00 6.90012693e-01
-1.13363445e+00 1.05532752e-02 -6.73072785e-02 -2.99562365e-01
6.79677546e-01 -3.17683339e-01 -7.60434806e-01 1.13170549e-01
1.36296546e+00 1.43368334e-01 8.52801085e-01 7.94056237e-01
-3.76162166e-03 8.48298594e-02 -8.13734293e-01 1.14031231e+00
2.33045742e-01 -8.84427726e-01 -1.00041464e-01 -2.17217237e-01
6.42706275e-01 2.10950181e-01 1.21746726e-01 3.76285762e-01
-5.06522646e-03 -9.92290854e-01 8.51394117e-01 2.74364632e-02
7.41013527e-01 -1.15881753e+00 9.01592553e-01 2.02453032e-01
-1.56140316e+00 -3.81568342e-01 -4.84732389e-01 -6.06289208e-01
-1.14524707e-01 6.40046149e-02 -1.45046461e+00 1.02257483e-01
1.19129574e+00 7.55299866e-01 -6.66506588e-01 1.60611629e+00
2.18759909e-01 1.37899995e+00 -6.58636689e-01 -8.45045000e-02
1.80015147e-01 3.54736686e-01 6.16862893e-01 1.86627483e+00
2.48674244e-01 1.53257519e-01 -1.18367560e-02 6.73735514e-02
2.53288656e-01 4.71467495e-01 -3.80586058e-01 -3.60338688e-01
6.04580879e-01 1.28802395e+00 -5.51399469e-01 -5.74137792e-02
-1.55083805e-01 2.64240861e-01 -2.66443372e-01 -2.89025635e-01
-4.80916589e-01 -7.78384149e-01 9.79731143e-01 -6.98393434e-02
3.46542746e-01 -1.06135137e-01 2.27034166e-01 -6.25484228e-01
-5.74604630e-01 -9.63003576e-01 6.79988623e-01 -4.29929495e-01
-1.42986476e+00 7.43094385e-01 -2.19258666e-02 -1.32936954e+00
-4.28443849e-01 -5.37973821e-01 -6.20145380e-01 5.87772071e-01
-1.05067456e+00 -1.17660892e+00 1.41276330e-01 2.81269848e-01
7.13304937e-01 -7.19737053e-01 1.17011404e+00 5.95010340e-01
2.21497007e-02 4.58475649e-01 -6.70385212e-02 2.36182362e-01
7.72336543e-01 -1.47445035e+00 5.36785722e-01 4.06364977e-01
1.00420940e+00 3.77183169e-01 6.29870594e-01 -3.39287102e-01
-9.66090679e-01 -1.11373436e+00 8.94152284e-01 -3.68721604e-01
7.86602259e-01 -1.61790669e-01 -7.83364236e-01 2.71993518e-01
2.43064210e-01 -1.85757563e-01 1.46178186e+00 9.64916646e-02
-2.03113675e-01 -1.75009653e-01 -1.11365807e+00 -2.31738701e-01
4.73471969e-01 -9.87294614e-01 -9.95875299e-01 1.16199004e-02
3.64078343e-01 -1.00421987e-01 -1.07209408e+00 2.72241712e-01
1.05112433e+00 -6.24788761e-01 1.03350317e+00 -1.02503085e+00
-7.08066449e-02 -4.62332368e-01 -3.80400032e-01 -1.63102949e+00
-4.51646328e-01 -6.12530589e-01 -2.69704522e-03 1.51108694e+00
3.33030671e-01 -2.25083411e-01 1.88488662e-01 -2.90706575e-01
-1.05276048e-01 -2.00418502e-01 -1.03338039e+00 -5.45374215e-01
-2.20227450e-01 -8.82909536e-01 5.41411340e-01 8.27681601e-01
2.10876673e-01 3.25436562e-01 -5.08960247e-01 1.29436374e-01
1.75225809e-01 -1.07431024e-01 5.82451820e-01 -1.47974193e+00
-2.86524385e-01 -2.87103206e-01 -9.82998252e-01 -5.56100667e-01
3.34357545e-02 -7.41507113e-01 2.99516588e-01 -8.96679163e-01
-3.04907650e-01 -1.67844594e-01 -4.45187569e-01 6.17880642e-01
1.66041911e-01 9.85275984e-01 3.15215081e-01 -7.93710127e-02
-3.09817139e-02 9.01266262e-02 3.59630078e-01 2.49572080e-02
-4.09892231e-01 5.49068809e-01 -3.64283204e-01 9.71582353e-01
8.20771515e-01 -4.71322536e-01 4.45927233e-02 -1.17328167e-01
-1.23767490e-02 -3.30667406e-01 3.49015027e-01 -1.27292395e+00
1.70190004e-03 7.21519515e-02 3.82836610e-01 -6.65381312e-01
5.06943285e-01 -7.31301308e-01 1.33904636e-01 5.80661356e-01
-4.74851847e-01 1.21789150e-01 6.37738228e-01 2.87637293e-01
-5.52267730e-01 -4.28177118e-01 9.84264135e-01 -3.29186432e-02
-7.62395620e-01 -1.94017276e-01 -9.61460769e-01 -2.11074859e-01
4.08800304e-01 1.62991703e-01 3.02806813e-02 -4.85727072e-01
-1.09158194e+00 -3.61525804e-01 -1.76743735e-02 6.76170290e-01
6.08582795e-01 -1.23376024e+00 -1.00281417e+00 6.85379624e-01
1.87053308e-02 -8.13637614e-01 4.09976915e-02 6.82766438e-01
-8.54877472e-01 6.06517792e-01 -5.04463673e-01 -4.83065546e-01
-1.96917677e+00 1.56604514e-01 2.93062210e-01 5.96430361e-01
-2.58575869e-03 1.17847133e+00 -5.85417330e-01 -5.48897326e-01
3.44682753e-01 -4.67914969e-01 -9.17493761e-01 4.95228291e-01
8.94062042e-01 6.88205242e-01 3.03420126e-01 -1.22032225e+00
-5.34663916e-01 4.33566749e-01 1.91032901e-01 -2.82874644e-01
1.45741248e+00 3.94880772e-01 -5.82053959e-02 8.59500170e-01
8.05867076e-01 2.77229309e-01 -4.68940765e-01 2.44091064e-01
5.64287603e-02 -6.47635937e-01 4.11082536e-01 -7.03721762e-01
-7.82326043e-01 1.17957866e+00 8.28150809e-01 7.81853497e-01
1.35021603e+00 -2.03580707e-01 4.94958639e-01 3.64642113e-01
2.17581376e-01 -1.13467908e+00 -5.00948846e-01 5.95353603e-01
8.53216887e-01 -7.53492773e-01 -1.07659295e-01 -1.30428895e-01
-7.75425792e-01 1.21602035e+00 3.03427041e-01 -3.38280767e-01
9.62545991e-01 4.14338410e-01 2.58970529e-01 2.92165160e-01
-1.05034518e+00 -3.75501841e-01 5.58483124e-01 1.12122858e+00
1.00752211e+00 2.67280281e-01 -6.28523901e-02 9.62819993e-01
-1.04509389e+00 -1.97477952e-01 4.29299414e-01 8.80371332e-01
-6.79644406e-01 -8.32273662e-01 -5.79935014e-01 7.35486209e-01
-1.19751859e+00 -2.08744496e-01 -9.67027247e-01 3.54501784e-01
3.81677806e-01 1.28217757e+00 1.24454819e-01 -8.09484780e-01
3.02318275e-01 4.01744217e-01 3.83366853e-01 -9.90310192e-01
-1.57424605e+00 5.15818596e-01 2.47375742e-01 2.07860544e-01
-9.57786560e-01 -8.55543435e-01 -8.40739071e-01 -7.92723522e-03
-7.55418599e-01 5.69968700e-01 6.07108235e-01 4.72659379e-01
-8.43427554e-02 3.12415272e-01 7.59557545e-01 -1.10657299e+00
-4.96922940e-01 -1.35950077e+00 -8.51056397e-01 -1.16125494e-01
7.52588451e-01 -4.61696625e-01 -5.49409926e-01 5.89964688e-01]
|
[15.285284996032715, 5.309927463531494]
|
09205b87-4c49-4da9-aba9-446021dafa33
|
are-semantically-coherent-topic-models-useful
| null | null |
https://aclanthology.org/P13-2027
|
https://aclanthology.org/P13-2027.pdf
|
Are Semantically Coherent Topic Models Useful for Ad Hoc Information Retrieval?
| null |
['Eric SanJuan', 'Romain Deveaud', 'Patrice Bellot']
|
2013-08-01
| null | null | null |
acl-2013-8
|
['ad-hoc-information-retrieval']
|
['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]
|
[-7.294112205505371, 3.5921502113342285]
|
0796eef3-46d9-4463-a3be-21912abf5af6
|
borch-a-deep-universal-probabilistic
|
2209.06168
| null |
https://arxiv.org/abs/2209.06168v1
|
https://arxiv.org/pdf/2209.06168v1.pdf
|
Borch: A Deep Universal Probabilistic Programming Language
|
Ever since the Multilayered Perceptron was first introduced the connectionist community has struggled with the concept of uncertainty and how this could be represented in these types of models. This past decade has seen a lot of effort in trying to join the principled approach of probabilistic modeling with the scalable nature of deep neural networks. While the theoretical benefits of this consolidation are clear, there are also several important practical aspects of these endeavors; namely to force the models we create to represent, learn, and report uncertainty in every prediction that is made. Many of these efforts have been based on extending existing frameworks with additional structures. We present Borch, a scalable deep universal probabilistic programming language, built on top of PyTorch. The code is available for download and use in our repository https://gitlab.com/desupervised/borch.
|
['Michael Green', 'Johan Gudmundsson', 'Lewis Belcher']
|
2022-09-13
| null | null | null | null |
['probabilistic-programming']
|
['methodology']
|
[-3.33999723e-01 4.88747030e-01 2.40461789e-02 -7.66469896e-01
-6.07910216e-01 -3.68724525e-01 6.49926066e-01 1.81141242e-01
-4.46623504e-01 6.59815133e-01 3.25980693e-01 -6.12195075e-01
-3.13548207e-01 -7.20492125e-01 -6.45407557e-01 -5.76016963e-01
-1.49414688e-01 8.26107204e-01 1.56791151e-01 2.98514545e-01
2.99208105e-01 4.64936405e-01 -1.45034254e+00 1.70630574e-01
1.61278114e-01 9.31316435e-01 2.21431963e-02 6.40593588e-01
-2.04761922e-01 9.10863757e-01 -1.08970411e-01 -6.95552707e-01
-6.58485889e-02 2.26676106e-01 -8.23834956e-01 -5.04013956e-01
1.05645597e-01 -5.16844153e-01 -5.58194757e-01 9.02408421e-01
3.52457255e-01 1.27417237e-01 6.47405028e-01 -1.39887571e+00
-3.66483837e-01 1.16589022e+00 -1.83207721e-01 8.83671865e-02
-1.82764247e-01 3.64914630e-03 1.25961316e+00 -6.68855608e-01
5.96855879e-02 1.38532853e+00 8.19571793e-01 5.23530245e-01
-1.45850730e+00 -6.92012608e-01 1.23420879e-01 2.01781750e-01
-1.32292867e+00 -5.73173463e-01 7.08643079e-01 -6.19384706e-01
1.24822223e+00 2.64582932e-02 4.27383691e-01 1.12928796e+00
2.96422809e-01 1.01117313e+00 9.90725696e-01 -3.98370326e-01
4.24830526e-01 3.81549478e-01 4.04721558e-01 4.78307456e-01
2.80460060e-01 3.09135973e-01 -3.80794227e-01 -3.65375578e-01
6.60898328e-01 9.42177605e-03 1.04664527e-01 -3.50969076e-01
-7.09580183e-01 8.05910826e-01 1.76086828e-01 2.12838590e-01
-2.57903248e-01 6.21782124e-01 2.52352715e-01 -2.30012764e-03
3.95744771e-01 1.44633219e-01 -8.23638201e-01 -4.47836816e-01
-9.99395669e-01 5.55373490e-01 1.28911507e+00 6.30746722e-01
6.18590117e-01 -2.39927083e-01 2.95677871e-01 7.73621738e-01
9.16973352e-01 -1.41404256e-01 3.50942820e-01 -1.25564754e+00
6.02987893e-02 1.69015542e-01 1.76127423e-02 -7.96353579e-01
-5.00690579e-01 -2.86600590e-01 -5.25340915e-01 4.03162152e-01
3.80968094e-01 -4.73186910e-01 -8.50712538e-01 1.69056976e+00
-7.68932886e-03 2.91750510e-03 -7.94617236e-02 4.89855796e-01
5.56146860e-01 4.94459093e-01 2.46856391e-01 5.94839394e-01
9.43001866e-01 -6.42091691e-01 -2.01291770e-01 -4.63098794e-01
4.45162177e-01 -6.91952705e-01 4.77715880e-01 5.91710269e-01
-1.02764523e+00 -1.41360536e-01 -1.19008875e+00 -6.75584301e-02
-3.04994494e-01 -3.66628587e-01 9.27303016e-01 9.12037373e-01
-1.27400804e+00 1.08019638e+00 -1.38880491e+00 -1.36748046e-01
6.31386638e-01 3.30865204e-01 -2.12845981e-01 1.23917526e-02
-1.21903598e+00 1.27767241e+00 8.86423886e-01 2.03502461e-01
-7.11591423e-01 -4.62156147e-01 -6.95494533e-01 2.18213886e-01
1.94001988e-01 -8.17144752e-01 1.66655767e+00 -7.26169765e-01
-1.39525104e+00 4.88667160e-01 -7.25367069e-02 -5.55486977e-01
4.79721069e-01 -1.25754952e-01 -2.53884010e-02 -4.86114770e-01
-3.92313480e-01 8.09272826e-01 9.00539309e-02 -1.11057413e+00
-6.67587996e-01 -2.98175931e-01 7.78718386e-03 6.60399795e-02
-4.06485647e-02 2.67049134e-01 -1.94910437e-01 -1.89105660e-01
2.56271660e-01 -9.96715248e-01 -3.71341288e-01 1.15089722e-01
-4.72146958e-01 -4.40337956e-01 2.97075808e-01 -5.25968492e-01
7.09930003e-01 -1.92406201e+00 -7.10568652e-02 2.21597821e-01
2.44670078e-01 -6.44017383e-02 2.13450536e-01 7.59842157e-01
-1.57089278e-01 3.63966078e-01 -3.84665966e-01 -7.49799550e-01
5.07063448e-01 3.68098915e-01 -4.68742281e-01 3.70314419e-01
2.04124168e-01 5.73926985e-01 -7.80370593e-01 -3.54622573e-01
3.35751057e-01 8.62229168e-01 -5.54784358e-01 -1.32409036e-01
-3.40707451e-01 -4.53635119e-02 -4.04640496e-01 4.04408187e-01
5.60729921e-01 -2.15274528e-01 4.13503498e-01 2.62122333e-01
-7.28568286e-02 8.66613090e-01 -1.35814977e+00 1.51627612e+00
-1.63963169e-01 7.21513391e-01 -1.23019956e-01 -8.49705219e-01
5.59017241e-01 3.81708562e-01 4.18263823e-01 1.19878620e-01
3.85409385e-01 -1.14471163e-03 1.49061844e-01 -7.03515857e-02
4.74906385e-01 -3.63910913e-01 3.12370181e-01 6.82989120e-01
4.21477616e-01 -9.61677805e-02 -5.49556501e-02 2.09252253e-01
1.12373066e+00 5.41443050e-01 1.28401637e-01 -7.97035620e-02
-2.25084841e-01 5.33691831e-02 6.46613002e-01 8.21730614e-01
-2.90652514e-01 5.47332048e-01 6.42392337e-01 -4.49312925e-01
-1.19892097e+00 -1.22122896e+00 -6.48397386e-01 8.82353783e-01
-5.48963547e-01 -4.61301714e-01 -3.78006190e-01 -2.14365751e-01
1.36507258e-01 1.22520888e+00 -4.68789279e-01 1.08812280e-01
1.01611197e-01 -9.16469574e-01 7.36420214e-01 7.31731117e-01
6.91910759e-02 -9.91740823e-01 -4.93629724e-01 3.73153567e-01
2.32965007e-01 -5.97325385e-01 3.42059970e-01 6.67852461e-01
-1.00146973e+00 -7.00903594e-01 -2.35731721e-01 -2.75616258e-01
1.48428768e-01 -2.42626280e-01 1.25717402e+00 -2.57533044e-01
-1.34473637e-01 3.58529538e-01 6.88274130e-02 -8.87324750e-01
-4.37462777e-01 -8.60090256e-02 1.04933515e-01 -6.92168593e-01
8.49885583e-01 -1.22135937e+00 -5.37512839e-01 -1.53009430e-01
-9.64864850e-01 2.55777508e-01 5.81898153e-01 6.04969740e-01
3.06557655e-01 3.06227267e-01 2.75554210e-01 -8.21258545e-01
6.55129373e-01 -8.78808439e-01 -6.04812562e-01 -7.63819814e-02
-8.98284912e-01 1.27660960e-01 -1.16348594e-01 6.73339004e-03
-9.72683907e-01 9.60571542e-02 -6.40551507e-01 -1.63576454e-01
-5.52215636e-01 8.23008001e-01 8.95721316e-02 3.40407819e-01
5.53478718e-01 3.99831636e-03 9.59651265e-03 -5.98749757e-01
5.38003623e-01 6.83090150e-01 1.16707578e-01 -8.12315524e-01
3.02965790e-01 2.57343650e-01 -3.14532667e-01 -4.49723065e-01
-4.80890512e-01 -1.26686439e-01 -4.23689038e-01 -3.79350968e-02
3.23301226e-01 -8.77256393e-01 -6.17510140e-01 5.46605647e-01
-1.19388974e+00 -4.73438561e-01 -1.21028878e-01 5.06772101e-01
-5.46321034e-01 3.00281078e-01 -8.00609827e-01 -1.16767538e+00
-2.04842567e-01 -1.03759396e+00 3.38476896e-01 3.00293088e-01
-4.78741795e-01 -1.17528009e+00 2.76287496e-01 1.96871310e-01
5.36657035e-01 -6.68365285e-02 7.62903512e-01 -1.07489395e+00
-5.50469875e-01 -4.93556768e-01 -2.07036823e-01 8.04556072e-01
-2.94689029e-01 4.09420431e-01 -1.38141167e+00 1.41504899e-01
4.88287471e-02 -4.61179167e-01 9.03527856e-01 4.91384089e-01
1.06245673e+00 -2.54923493e-01 -2.85634249e-01 3.31275851e-01
1.42953753e+00 2.24227533e-02 4.99688357e-01 4.30888504e-01
2.95774460e-01 5.60860574e-01 -1.68464482e-01 3.88853759e-01
8.10718656e-01 2.53131658e-01 5.04548967e-01 5.09319782e-01
3.18694711e-01 -2.57300228e-01 1.66779697e-01 7.34053254e-01
-6.94417134e-02 -6.12056367e-02 -1.43751693e+00 6.53946996e-01
-2.14008904e+00 -1.09841776e+00 1.85662836e-01 1.85942769e+00
1.16088271e+00 5.41610479e-01 -2.46852323e-01 -6.69529801e-03
3.70483577e-01 5.86159453e-02 -5.25742471e-01 -6.69694841e-01
1.50859281e-01 5.53730763e-02 4.18201327e-01 6.46462560e-01
-1.04467964e+00 5.77900231e-01 6.82109642e+00 6.67574286e-01
-9.02338326e-01 1.11078592e-02 9.63630259e-01 -1.64261594e-01
-3.30757678e-01 2.03357548e-01 -9.99722719e-01 3.75529349e-01
1.53249443e+00 -1.79037154e-01 3.45660150e-01 1.23769581e+00
4.53435257e-02 -3.11581016e-01 -1.32040143e+00 4.51821119e-01
-3.90826851e-01 -1.42337382e+00 -4.16129678e-01 1.79229025e-02
4.72329617e-01 9.13003564e-01 5.55943288e-02 3.68211269e-01
1.15969920e+00 -1.17105436e+00 8.07627320e-01 6.66121900e-01
5.53997904e-02 -6.61193311e-01 7.08110332e-01 5.10440707e-01
-3.96890879e-01 4.67532575e-02 -5.40315747e-01 -4.28466350e-01
9.57849994e-02 9.74558949e-01 -1.02422333e+00 1.93933144e-01
8.24298084e-01 4.08568740e-01 -1.87018961e-01 1.41757476e+00
-4.04640317e-01 9.09898162e-01 -6.49979353e-01 -1.37404576e-02
2.03122467e-01 4.49779890e-02 5.75066566e-01 1.37167895e+00
1.62462339e-01 -2.97428995e-01 -1.61023378e-01 1.13544953e+00
9.89116803e-02 -4.06977624e-01 -4.12927598e-01 -2.91219860e-01
5.96579790e-01 1.31857491e+00 -4.86901492e-01 -1.67670324e-01
-4.30007398e-01 3.05767387e-01 5.18131733e-01 1.05808914e-01
-7.35512674e-01 -3.06585401e-01 8.27633917e-01 -1.27867013e-01
2.34793916e-01 -4.81934369e-01 -6.44974053e-01 -8.60687077e-01
-1.11829855e-01 -5.66958189e-01 2.00395688e-01 -9.18994486e-01
-1.42212498e+00 5.58139384e-01 1.22901663e-01 -3.33388597e-01
-4.78840590e-01 -8.43119740e-01 -5.84848523e-01 1.24579942e+00
-1.25005829e+00 -9.89378989e-01 3.36775362e-01 1.61501527e-01
9.25496593e-02 1.91446319e-01 1.08120632e+00 1.54973030e-01
-4.39747930e-01 2.46056408e-01 4.73735034e-01 1.77264601e-01
5.14751673e-01 -1.36670220e+00 5.82703590e-01 6.01216018e-01
2.55800426e-01 9.07963693e-01 1.15380740e+00 -3.17481101e-01
-9.99334693e-01 -8.35576236e-01 1.02443326e+00 -8.52232933e-01
1.11836910e+00 -3.95170063e-01 -8.46989930e-01 1.31598651e+00
2.21518159e-01 -2.83850640e-01 9.28634048e-01 7.25151479e-01
-5.85340381e-01 9.98413116e-02 -9.81994033e-01 5.00458360e-01
4.37622190e-01 -5.12345552e-01 -8.48997116e-01 1.95693746e-01
4.98758584e-01 -2.62152791e-01 -9.41075265e-01 2.57753134e-01
8.48280549e-01 -1.16942036e+00 7.88697660e-01 -5.28729618e-01
6.31582320e-01 -3.50498036e-02 -3.72291088e-01 -1.25415182e+00
-3.93079102e-01 -2.78395236e-01 -2.39354461e-01 1.34836614e+00
6.28021300e-01 -9.99152124e-01 9.63584423e-01 1.28886080e+00
-2.13767245e-01 -9.12101328e-01 -1.01431370e+00 -4.24244851e-01
6.14280224e-01 -1.11005604e+00 5.08176029e-01 6.87951148e-01
1.61076486e-01 -5.48015870e-02 -1.39666706e-01 2.91108042e-01
6.40466571e-01 -3.08249950e-01 3.86355937e-01 -1.42451072e+00
-6.67137563e-01 -7.02605784e-01 -5.99791527e-01 -7.13218749e-01
-1.55491918e-01 -8.29513192e-01 2.26542473e-01 -1.78669763e+00
3.36958021e-01 -6.04021251e-01 -8.36830854e-01 8.97178710e-01
2.03232482e-01 -9.00998339e-02 2.00761199e-01 2.95824915e-01
-3.79149646e-01 3.02549630e-01 4.06757712e-01 6.61654845e-02
1.30518317e-01 9.54426900e-02 -1.17476654e+00 1.04229534e+00
1.17791796e+00 -7.92394698e-01 -2.63634890e-01 -7.20063806e-01
4.74248558e-01 -1.47599891e-01 5.36985576e-01 -1.34201288e+00
4.28453118e-01 -1.47390112e-01 5.36426246e-01 -5.72821558e-01
7.30866849e-01 -6.79692626e-01 4.70652074e-01 2.76002795e-01
-4.22936648e-01 -1.27128094e-01 6.13406479e-01 5.57757258e-01
1.97770540e-02 -4.90426272e-01 6.19430363e-01 -1.92298412e-01
-5.60523927e-01 2.27806613e-01 -4.95933563e-01 -2.94240832e-01
6.78597748e-01 1.97952226e-01 -3.58692795e-01 -2.95843869e-01
-8.75252008e-01 1.21633500e-01 4.53664333e-01 2.22805634e-01
3.15729946e-01 -8.20252001e-01 -6.08201206e-01 -3.01970840e-01
-1.84006080e-01 9.44776908e-02 2.23189965e-01 6.04427814e-01
-3.38050246e-01 7.18216300e-01 -1.56409293e-01 -2.65417546e-01
-7.74693847e-01 2.69961566e-01 4.45185989e-01 -1.76354930e-01
-6.24090374e-01 1.17910039e+00 -1.48131162e-01 -5.42135954e-01
4.88105088e-01 -1.81828380e-01 5.44751473e-02 -2.74844676e-01
4.93881047e-01 1.17420390e-01 -2.92063467e-02 -1.22770377e-01
-3.63066316e-01 -3.11313599e-01 -2.79297531e-01 -4.29324687e-01
1.75259244e+00 5.57996929e-02 -1.76363513e-01 9.42639410e-01
7.51478970e-01 -2.82986194e-01 -1.60169506e+00 -2.54964888e-01
2.33531907e-01 7.92756081e-02 5.47345161e-01 -1.13048875e+00
-6.84506774e-01 1.08598447e+00 4.38848317e-01 2.32959747e-01
4.93755937e-01 6.71862438e-02 3.04760575e-01 5.57450891e-01
1.98513225e-01 -8.40977252e-01 -8.05723548e-01 6.54200792e-01
7.13977873e-01 -1.08961225e+00 1.09648416e-02 3.08496892e-01
-4.41349626e-01 1.21925437e+00 2.53371924e-01 -3.21775883e-01
1.12711215e+00 4.46492314e-01 -1.26786932e-01 -3.95393334e-02
-1.18966854e+00 1.31400719e-01 1.07066799e-02 5.11872768e-01
7.48797178e-01 2.00984776e-01 6.88929632e-02 9.50264573e-01
-2.37104446e-01 3.48187208e-01 4.52912062e-01 1.01333916e+00
-6.52238488e-01 -1.32080901e+00 -1.80697039e-01 5.56922615e-01
-5.88943541e-01 -4.36063498e-01 -2.93069407e-02 4.94441926e-01
-5.18422984e-02 8.40459168e-01 1.25555530e-01 -4.76093739e-01
-2.44521573e-01 4.55552727e-01 3.91000241e-01 -6.99188650e-01
-1.57761782e-01 -2.91778833e-01 3.28588992e-01 -4.04727072e-01
-9.42736268e-02 -8.43412995e-01 -1.13532507e+00 -6.73928261e-01
5.13584800e-02 1.19294062e-01 1.19036293e+00 7.96237648e-01
3.95003587e-01 3.58335167e-01 -8.68734121e-02 -1.07291126e+00
-7.98526525e-01 -1.04644322e+00 -6.91878319e-01 -4.02783304e-01
5.52996546e-02 -5.26875854e-01 -4.62136179e-01 -2.14039117e-01]
|
[7.267086029052734, 3.873002529144287]
|
abcb4735-b5a1-403b-91fe-bf289b2a1d0f
|
learning-to-distill-the-essence-vector
|
1611.07206
| null |
http://arxiv.org/abs/1611.07206v1
|
http://arxiv.org/pdf/1611.07206v1.pdf
|
Learning to Distill: The Essence Vector Modeling Framework
|
In the context of natural language processing, representation learning has
emerged as a newly active research subject because of its excellent performance
in many applications. Learning representations of words is a pioneering study
in this school of research. However, paragraph (or sentence and document)
embedding learning is more suitable/reasonable for some tasks, such as
sentiment classification and document summarization. Nevertheless, as far as we
are aware, there is relatively less work focusing on the development of
unsupervised paragraph embedding methods. Classic paragraph embedding methods
infer the representation of a given paragraph by considering all of the words
occurring in the paragraph. Consequently, those stop or function words that
occur frequently may mislead the embedding learning process to produce a misty
paragraph representation. Motivated by these observations, our major
contributions in this paper are twofold. First, we propose a novel unsupervised
paragraph embedding method, named the essence vector (EV) model, which aims at
not only distilling the most representative information from a paragraph but
also excluding the general background information to produce a more informative
low-dimensional vector representation for the paragraph. Second, in view of the
increasing importance of spoken content processing, an extension of the EV
model, named the denoising essence vector (D-EV) model, is proposed. The D-EV
model not only inherits the advantages of the EV model but also can infer a
more robust representation for a given spoken paragraph against imperfect
speech recognition.
|
['Hsin-Min Wang', 'Shih-Hung Liu', 'Kuan-Yu Chen', 'Berlin Chen']
|
2016-11-22
|
learning-to-distill-the-essence-vector-2
|
https://aclanthology.org/C16-1035
|
https://aclanthology.org/C16-1035.pdf
|
coling-2016-12
|
['document-embedding']
|
['methodology']
|
[ 3.67455840e-01 2.00507417e-01 -2.41693631e-01 -3.76818240e-01
-5.84234715e-01 -1.74254715e-01 7.68290699e-01 6.28600180e-01
-2.89076179e-01 5.74522376e-01 7.69967735e-01 -5.75364009e-02
7.56860599e-02 -8.05235445e-01 -4.09216553e-01 -7.58566499e-01
4.93698925e-01 -1.97378874e-01 -1.68626994e-01 -2.56261438e-01
4.52364564e-01 3.28462064e-01 -1.66079092e+00 3.14552754e-01
9.09695089e-01 6.95612073e-01 2.16996789e-01 3.20515871e-01
-6.58356369e-01 7.25309312e-01 -9.46896791e-01 -4.42642748e-01
-3.80169719e-01 -7.30164528e-01 -5.03388643e-01 2.63192296e-01
6.39447123e-02 -9.57219973e-02 -4.12998021e-01 1.08085454e+00
5.53228021e-01 3.32369417e-01 9.03128564e-01 -7.79702961e-01
-8.14252019e-01 8.01390409e-01 -4.67428684e-01 1.65644422e-01
2.99028963e-01 -4.06689525e-01 1.23373055e+00 -1.02742326e+00
2.87951380e-01 1.10913646e+00 2.68297583e-01 4.43459302e-01
-9.83354807e-01 -2.90419161e-01 4.24771339e-01 2.47040093e-01
-1.19953525e+00 -3.58812332e-01 1.33527911e+00 -2.35206470e-01
7.89088905e-01 4.57021356e-01 6.56554103e-01 1.27097738e+00
2.75293678e-01 1.13420284e+00 8.22290480e-01 -6.38684452e-01
3.18525165e-01 5.33751070e-01 5.61226189e-01 3.34468842e-01
4.38532472e-01 -3.74054670e-01 -4.60868001e-01 -1.47639543e-01
2.01084182e-01 1.45542830e-01 -5.51357806e-01 -1.82987362e-01
-9.84631062e-01 1.08342314e+00 1.14652738e-01 6.24694824e-01
-5.73616147e-01 -1.32213712e-01 6.71371937e-01 7.87397251e-02
7.43221760e-01 2.26414099e-01 -2.37670336e-02 -1.31787494e-01
-9.95013237e-01 7.77148455e-02 6.88237071e-01 6.27703905e-01
6.41681671e-01 3.91080797e-01 -2.28527293e-01 9.82311666e-01
3.19205314e-01 3.60729635e-01 8.24316800e-01 -2.87756056e-01
4.46044147e-01 7.40610778e-01 -1.66100889e-01 -1.48204195e+00
-1.61401138e-01 -4.62615490e-01 -1.11672938e+00 -2.24693105e-01
-3.10290128e-01 -5.34965284e-02 -4.54339057e-01 1.49471092e+00
1.29987687e-01 -2.41918974e-02 4.57819670e-01 5.57706475e-01
1.09469056e+00 1.19341254e+00 -1.46789804e-01 -5.94205260e-01
1.50247991e+00 -7.47112751e-01 -1.11674023e+00 -2.46685803e-01
3.24413747e-01 -6.74586594e-01 7.57328808e-01 2.89798737e-01
-6.23343825e-01 -6.16874814e-01 -1.34296620e+00 -9.37980413e-02
-6.05823040e-01 1.74557015e-01 4.66781288e-01 4.89858270e-01
-5.50958693e-01 1.97927073e-01 -4.43424344e-01 -4.26683754e-01
1.01044573e-01 -2.63718478e-02 -3.82818967e-01 -7.72845224e-02
-1.33665454e+00 8.28559160e-01 5.05226374e-01 2.66668826e-01
-1.99795738e-01 -2.87025452e-01 -1.16828978e+00 2.30391175e-01
3.21022540e-01 -5.43746471e-01 9.64272261e-01 -9.31640387e-01
-1.48133993e+00 4.16323304e-01 -4.45363402e-01 -3.89523536e-01
1.26324251e-01 -2.58779943e-01 -5.24867952e-01 2.57093996e-01
-8.22969601e-02 1.42565280e-01 1.31150818e+00 -1.30630970e+00
-2.23871961e-01 -3.45855266e-01 -1.13017455e-01 2.66332179e-01
-7.99602568e-01 -2.87781835e-01 -1.94714293e-01 -1.05213642e+00
9.28626955e-02 -4.59712982e-01 4.40490693e-02 -2.31928006e-01
-5.20452559e-01 -6.17504299e-01 8.78731847e-01 -6.18232906e-01
1.56157625e+00 -2.29852962e+00 2.95486450e-01 -1.31535307e-01
9.20676589e-02 5.00315726e-01 2.95753847e-03 9.94110644e-01
-2.13526681e-01 7.46915489e-02 -4.98599559e-01 -5.25483251e-01
5.29802889e-02 3.47045034e-01 -8.61523271e-01 4.22011644e-01
3.29398870e-01 6.35093451e-01 -8.33326340e-01 -5.19768834e-01
4.19229269e-01 7.12470412e-01 -3.51164311e-01 1.51550606e-01
-1.64759113e-03 1.35490417e-01 -5.65369070e-01 2.94275910e-01
4.87537861e-01 1.66599110e-01 -3.38991992e-02 -3.68829757e-01
-2.82530159e-01 1.90860108e-01 -1.12035215e+00 1.39369178e+00
-4.32900846e-01 7.86214173e-01 -2.49592796e-01 -1.46803033e+00
1.17447615e+00 6.38612747e-01 3.69980067e-01 -4.69667703e-01
1.04442276e-01 6.10597953e-02 -2.86714226e-01 -7.71421731e-01
1.00390077e+00 -2.44283274e-01 -2.44309127e-01 3.10595572e-01
-3.61574180e-02 -1.71594471e-01 1.58273607e-01 2.57928967e-01
6.87698066e-01 -2.74869919e-01 6.91654444e-01 1.41198307e-01
8.35440218e-01 -3.34539175e-01 3.49012911e-01 4.81896549e-01
-2.97977272e-02 6.49813890e-01 5.82200587e-01 -2.36291006e-01
-8.98662686e-01 -8.26900721e-01 -2.62211591e-01 6.30101025e-01
1.70324314e-02 -5.89506984e-01 -4.82444584e-01 -6.00037038e-01
1.87517814e-02 9.54444408e-01 -6.61404073e-01 -3.86417925e-01
-4.52970147e-01 -6.60868883e-01 4.57420439e-01 4.70222980e-01
4.69718516e-01 -1.23880434e+00 -2.43379205e-01 2.92582482e-01
-4.32579279e-01 -8.04633737e-01 -2.97083884e-01 2.39340179e-02
-8.20928156e-01 -7.78729558e-01 -1.04512906e+00 -8.20697725e-01
8.20330441e-01 4.60257202e-01 6.73834324e-01 -1.13496907e-01
-9.77244973e-02 3.45451504e-01 -8.46178353e-01 -5.64189196e-01
-4.99117166e-01 -4.50767651e-02 7.94559717e-02 3.77516657e-01
5.69058478e-01 -3.19444686e-01 -3.29994589e-01 -2.90614575e-01
-1.17219603e+00 -1.58407658e-01 7.63506711e-01 9.57990408e-01
4.80132520e-01 1.35134876e-01 9.74022269e-01 -9.20314789e-01
1.05169499e+00 -5.31425357e-01 -1.04983419e-01 9.90843847e-02
-5.61361969e-01 1.33248031e-01 8.62329423e-01 -3.54113728e-01
-9.07669663e-01 -3.70941490e-01 -3.86099428e-01 -2.37275884e-01
-1.77841842e-01 8.39153349e-01 -2.73345739e-01 5.74299574e-01
2.75489539e-01 8.39083850e-01 9.45524573e-02 -4.93611395e-01
4.53991383e-01 1.14948905e+00 2.69635677e-01 -1.45647660e-01
6.55097246e-01 3.11073452e-01 -3.83800089e-01 -1.54294527e+00
-8.45264852e-01 -7.33974397e-01 -4.03823107e-01 -6.62937611e-02
7.80018032e-01 -6.73512399e-01 -3.42261404e-01 1.15033403e-01
-1.41379166e+00 6.98413372e-01 -4.55753326e-01 6.04119658e-01
-3.17687213e-01 9.38941598e-01 -1.72888726e-01 -1.04034686e+00
-5.47148585e-01 -9.66128647e-01 1.07708585e+00 2.09386677e-01
-4.27788138e-01 -9.64769900e-01 1.47228777e-01 1.37946486e-01
2.75702924e-01 1.13467820e-01 1.27574992e+00 -8.56183708e-01
4.04455252e-02 -5.79609036e-01 3.65763879e-03 8.61446500e-01
4.02251035e-01 8.88393912e-03 -1.09991741e+00 -2.43701220e-01
4.42543924e-01 7.01796040e-02 1.06428051e+00 2.06607893e-01
1.04127717e+00 -5.26165962e-01 -1.11188635e-01 2.01545492e-01
1.37283039e+00 5.22325374e-02 6.29978836e-01 2.89844811e-01
5.01315236e-01 6.71752274e-01 4.01985407e-01 5.47258675e-01
1.26452029e-01 5.08608103e-01 3.56891841e-01 2.70378143e-02
4.47976664e-02 -3.74624670e-01 5.30290484e-01 1.53780878e+00
1.59834877e-01 -4.58170831e-01 -3.25955927e-01 7.08109617e-01
-1.67133808e+00 -1.13837934e+00 7.34341666e-02 2.12879109e+00
6.48897111e-01 1.22404240e-01 -1.22220278e-01 6.44706964e-01
7.49778867e-01 7.54242778e-01 -4.04629320e-01 -8.04999948e-01
-2.56547838e-01 1.88198574e-02 -1.68501034e-01 1.78232238e-01
-9.38213170e-01 4.92631465e-01 5.27930784e+00 7.64512241e-01
-1.09140074e+00 -2.87394077e-01 1.36967093e-01 4.40005064e-01
-5.49784720e-01 -6.93001226e-02 -6.60104573e-01 6.49441719e-01
7.08619177e-01 -4.59077567e-01 -1.00570939e-01 8.37857068e-01
3.51530045e-01 1.13381520e-01 -9.32525516e-01 1.04076087e+00
7.17782199e-01 -1.13331819e+00 6.79244816e-01 -2.76158508e-02
4.75167423e-01 -5.41889966e-01 1.22410893e-01 4.64107841e-01
-6.35616481e-01 -6.72814250e-01 4.61408198e-01 4.94959056e-01
3.58952135e-01 -1.07628357e+00 1.09827220e+00 6.10763252e-01
-1.07872498e+00 9.14217830e-02 -7.28565216e-01 4.11201157e-02
1.72561079e-01 8.27838957e-01 -5.76781332e-01 9.53508377e-01
2.49251634e-01 1.01488221e+00 -4.24737394e-01 8.89207304e-01
-4.42568243e-01 7.32108653e-01 1.63942933e-01 -3.41224581e-01
2.16170907e-01 -3.24720889e-01 8.40203285e-01 1.33708179e+00
1.95767447e-01 2.61356849e-02 -1.69329554e-01 7.12450087e-01
-1.69542119e-01 5.19920588e-01 -1.03902543e+00 -4.17629629e-01
3.12033236e-01 1.10831535e+00 -4.15906876e-01 -4.21589136e-01
-4.79658902e-01 1.03419566e+00 1.68607831e-01 2.92033225e-01
-4.77350086e-01 -7.70856380e-01 5.91418982e-01 -3.49012226e-01
4.92094547e-01 -2.13627338e-01 -1.03473224e-01 -1.27457070e+00
2.01859221e-01 -7.20328867e-01 -1.01082399e-02 -4.35847044e-01
-1.31183171e+00 6.98748827e-01 -4.99400124e-03 -1.58624125e+00
-3.55869591e-01 -2.84271777e-01 -8.66485596e-01 6.59305036e-01
-1.75292337e+00 -7.82823145e-01 -7.82592371e-02 1.01540402e-01
1.07014525e+00 -1.56899482e-01 1.07067561e+00 5.07617146e-02
-6.58733070e-01 5.24499178e-01 5.32832682e-01 1.36888713e-01
5.89199066e-01 -1.15667415e+00 4.73975018e-02 7.24997282e-01
3.37759316e-01 8.63437772e-01 9.00765300e-01 -3.44223022e-01
-1.44388795e+00 -9.51953053e-01 1.28674269e+00 -1.07167646e-01
4.97341156e-01 -2.26321667e-01 -1.18047619e+00 3.84144753e-01
4.28425968e-01 -4.25332725e-01 9.29711401e-01 -9.90482941e-02
-2.34770641e-01 -1.22708358e-01 -7.66386390e-01 6.77672446e-01
2.44589373e-01 -6.75559580e-01 -1.25903690e+00 1.98861554e-01
7.93533146e-01 1.19903997e-01 -6.70366287e-01 2.08552077e-01
3.88475567e-01 -7.30420947e-01 8.56770337e-01 -2.79277027e-01
5.45283198e-01 -1.46097407e-01 -2.46229827e-01 -1.47816789e+00
-7.78853819e-02 -3.04076225e-01 -4.03527647e-01 1.69228184e+00
9.69187617e-02 -6.48844779e-01 4.34059948e-01 7.50224963e-02
-2.34985456e-01 -8.51744711e-01 -9.08229589e-01 -6.36524975e-01
-2.82382295e-02 -3.56557399e-01 3.46706063e-01 8.44202816e-01
2.03957677e-01 6.44617438e-01 -5.48373520e-01 1.40075684e-01
5.65037966e-01 3.23456883e-01 5.45854986e-01 -1.18199992e+00
-4.25737798e-02 -4.20896798e-01 -5.35787761e-01 -1.21157038e+00
3.70367616e-01 -9.44703758e-01 1.20219544e-01 -1.87879586e+00
1.47058725e-01 3.44837338e-01 -3.12090784e-01 5.23887090e-02
-3.11855435e-01 -8.45491290e-02 1.95920795e-01 2.80745059e-01
-3.03726792e-01 1.19159853e+00 1.03492737e+00 -4.86440688e-01
-3.08545679e-02 2.09747091e-01 -9.46748376e-01 7.36554503e-01
6.69903219e-01 -3.58220100e-01 -4.07227337e-01 -2.99096376e-01
5.59021272e-02 -7.88922161e-02 2.26746872e-02 -6.85831785e-01
1.57207265e-01 2.18811348e-01 2.19614595e-01 -8.56896222e-01
4.49030161e-01 -7.70780504e-01 -5.16582191e-01 2.91162997e-01
-4.35952872e-01 9.41380560e-02 1.28346935e-01 8.91173542e-01
-6.82735562e-01 -6.92801893e-01 4.37878191e-01 3.05919982e-02
-7.29407132e-01 -1.22031376e-01 -7.17351317e-01 -2.10039213e-01
7.88217664e-01 -4.10301179e-01 -2.61833280e-01 -4.66800183e-01
-3.64781708e-01 -2.30148658e-02 1.45879820e-01 7.38071322e-01
1.04754090e+00 -1.12411261e+00 -8.27807486e-01 2.74465114e-01
3.40875149e-01 -1.28279522e-01 3.23678881e-01 4.78210568e-01
-1.70938209e-01 6.01261020e-01 2.83704281e-01 -3.75134885e-01
-1.14803255e+00 6.92670584e-01 -2.79926002e-01 -2.30907723e-02
-9.34261978e-01 3.99078220e-01 2.17699826e-01 -9.91262645e-02
3.35041910e-01 -2.71633327e-01 -9.27402914e-01 4.82603133e-01
8.30941439e-01 2.93219030e-01 5.96121252e-02 -8.96847129e-01
-2.72365272e-01 6.43095553e-01 -3.67871553e-01 5.29224239e-02
1.33466101e+00 -1.88411102e-01 -2.51338750e-01 9.98139203e-01
1.39255643e+00 2.11188942e-01 -5.85224211e-01 -2.50860333e-01
2.00195424e-02 -1.74585789e-01 1.22921139e-01 -2.54995465e-01
-6.07500851e-01 1.04640388e+00 2.42933229e-01 5.85064828e-01
9.43532467e-01 -1.30973220e-01 8.47775042e-01 4.08723295e-01
3.33651947e-03 -1.04974580e+00 1.57818779e-01 4.78792012e-01
1.08402061e+00 -1.14088404e+00 2.44683459e-01 -2.61521697e-01
-6.92598104e-01 1.33672774e+00 1.90722242e-01 -1.19580127e-01
4.33817446e-01 -2.72516340e-01 -1.29570821e-02 -1.57023802e-01
-5.51286399e-01 -1.14338942e-01 3.66647512e-01 4.55130965e-01
5.57129145e-01 -2.01414213e-01 -7.03595757e-01 6.95099115e-01
-1.06370904e-01 -3.60156059e-01 5.79612136e-01 9.19632137e-01
-7.31223702e-01 -1.01996875e+00 -3.09056878e-01 4.03202057e-01
-3.53890091e-01 -7.29864091e-02 -4.83381271e-01 5.68357050e-01
-2.10202351e-01 1.03089094e+00 -4.50165570e-02 -2.32386246e-01
4.82196093e-01 1.26307204e-01 3.25962417e-02 -8.69835317e-01
-3.27100664e-01 -1.18938647e-01 -2.20889241e-01 -2.76596062e-02
-5.54324865e-01 -4.23832417e-01 -1.09965885e+00 -4.85346019e-02
-5.56918204e-01 5.87744236e-01 9.51959014e-01 1.09453583e+00
1.86599955e-01 7.74885535e-01 9.08771694e-01 -7.84711182e-01
-7.34771430e-01 -1.09706330e+00 -7.37448215e-01 3.14416319e-01
5.29094458e-01 -4.31593180e-01 -6.92815602e-01 -4.84472886e-02]
|
[11.2532377243042, 8.84883975982666]
|
0d0bce0c-0722-43f6-97c5-4231d5a90cb0
|
carl-g-clustering-accelerated-representation
|
2306.06936
| null |
https://arxiv.org/abs/2306.06936v1
|
https://arxiv.org/pdf/2306.06936v1.pdf
|
CARL-G: Clustering-Accelerated Representation Learning on Graphs
|
Self-supervised learning on graphs has made large strides in achieving great performance in various downstream tasks. However, many state-of-the-art methods suffer from a number of impediments, which prevent them from realizing their full potential. For instance, contrastive methods typically require negative sampling, which is often computationally costly. While non-contrastive methods avoid this expensive step, most existing methods either rely on overly complex architectures or dataset-specific augmentations. In this paper, we ask: Can we borrow from classical unsupervised machine learning literature in order to overcome those obstacles? Guided by our key insight that the goal of distance-based clustering closely resembles that of contrastive learning: both attempt to pull representations of similar items together and dissimilar items apart. As a result, we propose CARL-G - a novel clustering-based framework for graph representation learning that uses a loss inspired by Cluster Validation Indices (CVIs), i.e., internal measures of cluster quality (no ground truth required). CARL-G is adaptable to different clustering methods and CVIs, and we show that with the right choice of clustering method and CVI, CARL-G outperforms node classification baselines on 4/5 datasets with up to a 79x training speedup compared to the best-performing baseline. CARL-G also performs at par or better than baselines in node clustering and similarity search tasks, training up to 1,500x faster than the best-performing baseline. Finally, we also provide theoretical foundations for the use of CVI-inspired losses in graph representation learning.
|
['Evangelos E. Papalexakis', 'Neil Shah', 'Tong Zhao', 'Yozen Liu', 'Uday Singh Saini', 'William Shiao']
|
2023-06-12
| null | null | null | null |
['clustering', 'graph-representation-learning']
|
['methodology', 'methodology']
|
[ 2.37419471e-01 1.27322808e-01 -6.27558112e-01 -3.52870107e-01
-8.34619224e-01 -5.81897259e-01 6.11305892e-01 8.44392002e-01
-2.69701153e-01 3.37513894e-01 -7.55298045e-03 -3.58885705e-01
-4.85520571e-01 -8.89684618e-01 -5.59911788e-01 -7.88566768e-01
-5.12338340e-01 7.01786339e-01 2.47086072e-03 -9.68230739e-02
2.65778631e-01 3.35716426e-01 -1.33419645e+00 2.96457820e-02
7.54348755e-01 8.26323211e-01 7.44368061e-02 4.04025108e-01
-1.68707937e-01 6.21004045e-01 -3.17498177e-01 -4.29670334e-01
1.75181702e-01 -4.99782830e-01 -1.08862209e+00 1.52131826e-01
3.18583101e-01 5.08849025e-01 -4.68358845e-01 1.05518758e+00
3.48223120e-01 2.09921166e-01 1.10470951e+00 -1.48327017e+00
-6.10065460e-01 7.80844271e-01 -7.19587862e-01 8.17906633e-02
7.40341470e-02 -1.58740729e-01 1.45824993e+00 -6.46973014e-01
4.45746124e-01 1.15510285e+00 9.76424694e-01 4.30083185e-01
-1.70285785e+00 -5.06153643e-01 2.59076864e-01 2.93198861e-02
-1.65967071e+00 -3.50481540e-01 7.25600541e-01 -3.46246660e-01
7.92087555e-01 2.34820679e-01 3.57693642e-01 8.25774193e-01
-2.61682510e-01 7.92294681e-01 8.34622145e-01 -4.89707023e-01
2.80315965e-01 9.99471918e-02 2.94014484e-01 8.37020993e-01
5.75523317e-01 -8.38008970e-02 -1.53179348e-01 -1.81753993e-01
5.03417850e-01 1.34492293e-01 -1.39750227e-01 -9.01336849e-01
-9.72200871e-01 1.13321924e+00 9.12735581e-01 4.41497028e-01
2.62798518e-02 2.35405236e-01 5.28692961e-01 5.22402942e-01
4.94838119e-01 6.14240825e-01 -1.84114173e-01 1.77919969e-01
-1.00001419e+00 3.43034766e-03 7.73235679e-01 9.75011468e-01
8.84462655e-01 -1.98317170e-01 1.88982226e-02 8.15728247e-01
2.82481313e-01 -2.96067595e-02 2.73548633e-01 -7.62908936e-01
3.52899700e-01 7.71775961e-01 -5.57706237e-01 -1.19302022e+00
-6.41891062e-01 -8.02822649e-01 -1.34619761e+00 8.66579860e-02
4.35021728e-01 1.85577109e-01 -7.90194750e-01 1.83202624e+00
1.41398115e-02 1.51655525e-01 -1.47281930e-01 4.79460269e-01
6.97567344e-01 3.38753432e-01 6.34946972e-02 -2.24452615e-01
9.22493041e-01 -8.44773769e-01 -3.21562916e-01 -1.63256630e-01
1.06890404e+00 -3.66541862e-01 1.08400702e+00 2.88314044e-01
-1.02038026e+00 -3.17975879e-01 -1.13852739e+00 1.54542357e-01
-5.33115625e-01 -2.81083107e-01 9.88735378e-01 8.80069017e-01
-1.35041165e+00 9.65532601e-01 -7.70161748e-01 -6.21801138e-01
6.41468465e-01 5.03718078e-01 -4.76855725e-01 -2.26358756e-01
-6.99155331e-01 5.77148616e-01 2.75913805e-01 -3.60450566e-01
-5.82223237e-01 -7.73331046e-01 -1.00178015e+00 2.96141297e-01
4.80938762e-01 -6.01049960e-01 6.53472245e-01 -8.19488347e-01
-9.66424704e-01 1.00726581e+00 7.79011706e-03 -6.09235287e-01
2.19025239e-01 1.94320828e-01 -2.04527900e-01 1.32793829e-01
1.88959524e-01 5.94567060e-01 6.14661574e-01 -1.34380043e+00
-2.97005028e-01 -4.72878426e-01 -1.97043240e-01 1.57296538e-01
-7.96121836e-01 -2.99764037e-01 -7.59339333e-01 -6.86660767e-01
1.75308645e-01 -8.97348464e-01 -4.82560456e-01 -6.08566625e-04
-5.28601348e-01 -3.75529319e-01 5.78574717e-01 -2.37085242e-02
1.31475782e+00 -2.15348744e+00 2.42390886e-01 7.20498502e-01
7.06640542e-01 2.84932226e-01 -3.57040703e-01 5.81597507e-01
-2.26448447e-01 5.01603603e-01 -4.46935505e-01 -5.21979451e-01
-2.14489661e-02 1.56188965e-01 1.41626179e-01 5.83547652e-01
2.09253132e-01 8.16150904e-01 -1.12942791e+00 -6.28684759e-01
2.53277272e-01 3.98685575e-01 -5.52703381e-01 -3.40087935e-02
4.56268601e-02 1.87949181e-01 -1.96945533e-01 5.05097985e-01
5.08236468e-01 -8.42299163e-01 6.28032684e-01 -8.86562020e-02
5.00932097e-01 2.70983130e-01 -1.11805534e+00 1.75284886e+00
-2.91867584e-01 5.50750852e-01 8.67016017e-02 -1.71863639e+00
8.99569988e-01 -7.65896887e-02 6.33892953e-01 -4.61399823e-01
-4.38274890e-02 1.03232332e-01 -2.39602290e-02 -1.13239335e-02
1.98206097e-01 -7.11011607e-03 -5.18640615e-02 5.07247746e-01
6.16725534e-02 1.27805024e-03 2.71909058e-01 7.53924072e-01
1.62766635e+00 -3.45910519e-01 4.06825900e-01 -4.29487407e-01
3.22448134e-01 -4.60981019e-02 4.41600204e-01 8.84886324e-01
-5.03737628e-02 5.81360579e-01 6.21587753e-01 -2.10091412e-01
-7.51166403e-01 -1.12273943e+00 -2.85722557e-02 1.16790235e+00
1.14541300e-01 -7.93892026e-01 -5.82088232e-01 -1.05396295e+00
2.54545182e-01 3.23266268e-01 -6.48108006e-01 -3.28051835e-01
-3.66833031e-01 -9.94729280e-01 6.78217530e-01 6.80517256e-01
1.04862839e-01 -7.30122507e-01 1.97093800e-01 4.22237627e-02
1.07664779e-01 -8.69444966e-01 -3.90933931e-01 4.34245378e-01
-1.13544369e+00 -1.30983245e+00 -5.22428095e-01 -1.13236237e+00
9.40737963e-01 7.37857759e-01 1.49824512e+00 5.91617167e-01
-4.07044232e-01 4.09646511e-01 -4.33498949e-01 5.32811955e-02
-3.25486988e-01 5.88313639e-01 -1.57026708e-01 -2.97227859e-01
3.51247787e-01 -8.63888502e-01 -4.50084180e-01 1.31317466e-01
-6.99369669e-01 -3.64367098e-01 8.26008916e-01 8.09699893e-01
5.20775437e-01 2.11392477e-01 7.32796133e-01 -1.32759738e+00
6.17245376e-01 -7.87069917e-01 -3.47373635e-01 2.67849565e-01
-1.11872268e+00 2.65488625e-01 7.65755832e-01 -2.34339714e-01
-3.29468697e-01 -6.57919198e-02 1.18600331e-01 -5.31490386e-01
1.54484928e-01 6.91097677e-01 -1.89669490e-01 -1.38570338e-01
8.61942291e-01 -4.80700135e-02 2.67798036e-01 -3.33885819e-01
5.59900522e-01 3.60327005e-01 4.31436896e-01 -5.71640134e-01
9.83382583e-01 4.39703703e-01 3.20495784e-01 -7.88094580e-01
-7.56519437e-01 -7.32085109e-01 -6.66966736e-01 1.05187841e-01
4.98548865e-01 -8.15092564e-01 -6.74556851e-01 -1.60668995e-02
-4.96396035e-01 -4.08351839e-01 -2.22985685e-01 1.59588933e-01
-4.08852398e-01 6.81216717e-01 -5.89561105e-01 -6.68486178e-01
-2.86430180e-01 -8.36779237e-01 7.27115631e-01 -2.25994706e-01
-1.14318557e-01 -1.41608763e+00 -1.23433992e-01 2.15706512e-01
3.78234774e-01 2.84266204e-01 1.18325043e+00 -8.74014020e-01
-3.89678776e-01 -2.34809592e-01 -3.58530104e-01 3.23347837e-01
1.56432837e-01 -1.39326379e-01 -6.28850758e-01 -7.61138082e-01
-4.94436681e-01 -5.21563470e-01 1.23859608e+00 2.91157901e-01
1.48299801e+00 -2.28997141e-01 -8.70216250e-01 7.32767522e-01
1.51620162e+00 -3.27142835e-01 4.89059985e-01 5.96865751e-02
8.07457864e-01 6.42155528e-01 4.08082783e-01 1.10293061e-01
5.28025389e-01 6.17362082e-01 5.31598508e-01 -3.38578433e-01
-1.67956159e-01 -3.19586217e-01 -3.21846046e-02 7.92300701e-01
-1.31286643e-02 -2.94580400e-01 -8.85208905e-01 6.71590269e-01
-1.92956865e+00 -9.50320721e-01 -2.48928308e-01 2.43215156e+00
7.22886503e-01 3.41775030e-01 3.79270196e-01 4.16707754e-01
7.61008084e-01 1.95049733e-01 -4.55392092e-01 7.89345149e-03
-1.18181802e-01 3.20333511e-01 5.21037459e-01 3.36349547e-01
-1.27648437e+00 8.94861937e-01 6.05785847e+00 1.03505862e+00
-7.56779015e-01 -7.96229839e-02 7.63541102e-01 1.43674210e-01
-3.13428044e-01 1.17382362e-01 -3.98850888e-01 2.98198432e-01
8.76000047e-01 -1.23917870e-01 5.40502608e-01 8.32206130e-01
-2.08960935e-01 3.34706217e-01 -1.46247625e+00 1.08930862e+00
1.09842747e-01 -1.50773704e+00 -1.42709687e-01 3.54935646e-01
7.79138327e-01 1.52642787e-01 1.13657549e-01 3.92160028e-01
7.44496465e-01 -1.35362947e+00 9.20431763e-02 1.61477670e-01
6.52253687e-01 -7.90419579e-01 5.94683588e-01 2.79789954e-01
-1.43471098e+00 -5.52357472e-02 -4.75805521e-01 6.36922494e-02
-2.96165287e-01 8.17393482e-01 -8.13898265e-01 6.99901640e-01
6.02340937e-01 9.89135265e-01 -7.86347210e-01 1.07410288e+00
1.23874784e-01 8.13928366e-01 -3.26415390e-01 -1.68774482e-02
3.30370098e-01 -1.54910147e-01 2.13031888e-01 1.46171105e+00
9.78800207e-02 -2.62920231e-01 3.80843848e-01 7.69913316e-01
-5.52134037e-01 1.89599276e-01 -8.29037249e-01 -1.12482674e-01
9.12100375e-01 1.39861488e+00 -9.15681303e-01 -2.64694333e-01
-3.16157490e-01 8.30290198e-01 9.10067558e-01 1.73894912e-01
-5.19581735e-01 -5.00323772e-01 5.26126325e-01 3.44905496e-01
3.38220119e-01 -2.65481286e-02 -3.23142141e-01 -9.26995516e-01
-1.90628126e-01 -8.53430331e-01 7.64557064e-01 -9.62033942e-02
-1.79453433e+00 4.32697505e-01 -1.54268414e-01 -1.07956624e+00
-2.28436351e-01 -4.98457700e-01 -7.68486440e-01 3.28140974e-01
-1.45966280e+00 -1.08004224e+00 -2.92062491e-01 6.01532817e-01
1.59900203e-01 -1.75917804e-01 8.83412957e-01 3.72162133e-01
-5.19032896e-01 9.81910944e-01 3.58826667e-01 3.53950649e-01
7.64947414e-01 -1.57168770e+00 4.57984388e-01 4.90375727e-01
4.74995464e-01 6.26516879e-01 3.23295623e-01 -4.06584918e-01
-1.28623652e+00 -1.28528094e+00 7.52541006e-01 -3.59696984e-01
6.04827464e-01 -4.73280132e-01 -9.51844633e-01 6.27202094e-01
-6.16894066e-02 3.32962245e-01 8.42632830e-01 6.09934866e-01
-6.68653667e-01 -1.86288252e-01 -1.09675276e+00 5.04409015e-01
1.51928306e+00 -4.62773502e-01 -2.08120719e-01 5.87797701e-01
4.82570529e-01 3.02040786e-01 -1.06192017e+00 4.29414332e-01
1.05092175e-01 -8.98235559e-01 1.25360036e+00 -7.93650746e-01
9.32190865e-02 -1.07707642e-01 -1.62621617e-01 -1.22626984e+00
-6.43364906e-01 -7.02560365e-01 -7.07122758e-02 1.33808494e+00
4.73054409e-01 -5.08773685e-01 1.05324972e+00 1.56804278e-01
-1.56871408e-01 -7.93520570e-01 -6.16157174e-01 -9.98717487e-01
3.91095012e-01 -2.49936223e-01 4.29122835e-01 1.39214027e+00
3.58033806e-01 6.62184298e-01 -1.08930804e-01 -2.18436737e-02
1.01492095e+00 3.73348743e-02 7.74192870e-01 -1.74150443e+00
-2.41547927e-01 -8.60040069e-01 -6.72973454e-01 -9.70939994e-01
4.02851135e-01 -1.57020605e+00 -2.10005775e-01 -1.68365610e+00
3.90698493e-01 -9.21904445e-01 -4.26109731e-01 5.24923027e-01
-2.70888120e-01 3.16807717e-01 1.38397500e-01 4.18710947e-01
-9.88972187e-01 3.30481231e-01 7.04543471e-01 -3.11435014e-01
-1.61361054e-01 3.50687802e-02 -1.00869524e+00 5.72935939e-01
8.30743194e-01 -4.32872981e-01 -5.38855970e-01 -1.63017243e-01
2.40419537e-01 -9.40129757e-02 3.03094715e-01 -1.09776878e+00
4.62974966e-01 1.87156886e-01 2.85919935e-01 -4.07237500e-01
1.44689530e-01 -6.20248318e-01 -1.53405845e-01 4.57580656e-01
-6.01378918e-01 5.52030727e-02 -2.64956236e-01 9.60864067e-01
-1.17958814e-01 -1.54882893e-01 6.95091128e-01 -1.79963097e-01
-4.22508955e-01 4.52329904e-01 -1.38235867e-01 4.19346273e-01
9.27122414e-01 -1.33416951e-01 -3.21951896e-01 -4.36740428e-01
-7.06717849e-01 3.66898775e-01 6.30086064e-01 1.39561683e-01
5.15779197e-01 -1.25115681e+00 -7.88398623e-01 1.10615954e-01
2.59768814e-01 -3.06784138e-02 -2.36308388e-02 8.37079823e-01
-2.53820449e-01 1.37532651e-01 2.28100464e-01 -6.38889432e-01
-1.14152133e+00 9.10927236e-01 1.50482252e-01 -6.38121605e-01
-6.32321060e-01 8.21824372e-01 1.02036126e-01 -6.60094380e-01
5.83313644e-01 1.12533234e-01 -2.46747117e-03 -1.34567795e-02
1.78516418e-01 4.01409090e-01 1.92846820e-01 -3.16509157e-01
-4.38511342e-01 4.43798333e-01 -2.19553575e-01 2.98985809e-01
1.39923823e+00 1.04686655e-02 -2.54204441e-02 1.98299915e-01
1.49096131e+00 -1.88102841e-01 -8.57466578e-01 -4.65595841e-01
4.91948485e-01 -3.59855473e-01 2.69886777e-02 -5.68046510e-01
-1.29252291e+00 8.85915518e-01 3.33308339e-01 5.42686343e-01
9.53990221e-01 4.18400466e-01 4.30268288e-01 6.15817070e-01
3.67586136e-01 -9.74146545e-01 3.16091180e-01 1.33138999e-01
3.81598383e-01 -1.52380490e+00 2.37663031e-01 -6.82655632e-01
-5.30816734e-01 8.98908436e-01 3.53867829e-01 -2.64154553e-01
7.93385088e-01 1.29497290e-01 -2.42188126e-01 -4.58065480e-01
-8.33305717e-01 -4.25993621e-01 2.83505321e-01 1.01721692e+00
5.89091778e-01 1.24779776e-01 -1.00057840e-01 2.74771482e-01
-3.41718830e-02 -5.73329389e-01 1.94208428e-01 7.42742360e-01
-3.63696873e-01 -1.28888226e+00 3.30822617e-02 8.27614069e-01
-2.30696574e-01 -1.17879570e-01 -5.80164790e-01 9.96620297e-01
-2.39931941e-01 1.24679339e+00 1.27436891e-01 -5.74448824e-01
6.21092552e-03 -2.09123358e-01 3.80966395e-01 -7.12893009e-01
-6.75781429e-01 -1.50931180e-01 2.33865790e-02 -5.66971183e-01
-5.12503743e-01 -6.04901016e-01 -1.27707374e+00 -5.22355080e-01
-4.56843048e-01 3.96102250e-01 3.79319638e-01 5.56519389e-01
4.57828462e-01 3.25950474e-01 7.96462059e-01 -6.16523206e-01
-6.68455482e-01 -8.56948674e-01 -6.03648424e-01 6.00819349e-01
6.21533319e-02 -6.03340268e-01 -6.79184258e-01 -3.67807031e-01]
|
[7.119868755340576, 6.088598728179932]
|
747739fc-27b3-49b2-838f-c7f2f9770065
|
antifragile-and-robust-heteroscedastic
| null | null |
https://openreview.net/forum?id=B1lTqgSFDH
|
https://openreview.net/pdf?id=B1lTqgSFDH
|
Antifragile and Robust Heteroscedastic Bayesian Optimisation
|
Bayesian Optimisation is an important decision-making tool for high-stakes applications in drug discovery and materials design. An oft-overlooked modelling consideration however is the representation of input-dependent or heteroscedastic aleatoric uncertainty. The cost of misrepresenting this uncertainty as being homoscedastic could be high in drug discovery applications where neglecting heteroscedasticity in high throughput virtual screening could lead to a failed drug discovery program. In this paper, we propose a heteroscedastic Bayesian Optimisation scheme which both represents and optimises aleatoric noise in the suggestions. We consider cases such as drug discovery where we would like to minimise or be robust to aleatoric uncertainty but also applications such as materials discovery where it may be beneficial to maximise or be antifragile to aleatoric uncertainty. Our scheme features a heteroscedastic Gaussian Process (GP) as the surrogate model in conjunction with two acquisition heuristics. First, we extend the augmented expected improvement (AEI) heuristic to the heteroscedastic setting and second, we introduce a new acquisition function, aleatoric-penalised expected improvement (ANPEI) based on a simple scalarisation of the performance and noise objective. Both methods are capable of penalising or promoting aleatoric noise in the suggestions and yield improved performance relative to a naive implementation of homoscedastic Bayesian Optimisation on toy problems as well as a real-world optimisation problem.
|
['Alpha A. Lee', 'Alexander A. Aldrick', 'Miguel Garcia-Ortegon', 'Ryan Rhys-Griffiths']
|
2019-09-25
| null | null | null | null |
['bayesian-optimisation']
|
['methodology']
|
[ 5.38674653e-01 2.39074066e-01 2.21094772e-01 -2.06809893e-01
-9.51388299e-01 -5.21138370e-01 8.39748025e-01 1.97464183e-01
-6.60385251e-01 1.03488934e+00 1.14366077e-01 -6.87392235e-01
-9.19951618e-01 -5.08785546e-01 -6.68729961e-01 -1.05576003e+00
2.64268190e-01 9.25441027e-01 -1.07517382e-02 3.98185879e-01
3.58426332e-01 5.91242254e-01 -1.17122936e+00 -8.91307071e-02
7.93101907e-01 7.38265216e-01 2.20159456e-01 6.83305800e-01
3.03002357e-01 2.26644173e-01 -6.73014164e-01 -5.96213877e-01
1.98931217e-01 -1.36244386e-01 -5.47588050e-01 -7.00192004e-02
-3.62593144e-01 2.68194020e-01 5.04426122e-01 9.94345427e-01
9.47302461e-01 9.20326039e-02 1.12603486e+00 -9.96257484e-01
-3.53690535e-02 2.90993154e-01 -5.33062100e-01 4.41245101e-02
9.45330560e-02 4.00345534e-01 6.18656635e-01 -7.52780735e-01
2.37854108e-01 1.41374016e+00 5.26159823e-01 6.94227666e-02
-1.66903770e+00 -3.64442587e-01 -2.24009559e-01 -3.93243879e-01
-1.56287944e+00 -6.12308025e-01 2.73795873e-01 -6.46038771e-01
7.67336488e-01 4.41181242e-01 2.79785573e-01 1.02884901e+00
6.66849911e-01 4.89973158e-01 1.28080153e+00 -3.05832505e-01
8.33270550e-01 2.54456013e-01 -4.57380712e-01 -3.27850953e-02
3.18106651e-01 4.96339768e-01 -3.11451759e-02 -7.10309505e-01
5.92540681e-01 -2.22325832e-01 3.49984989e-02 -4.75150257e-01
-1.01311600e+00 1.06344295e+00 -2.38896325e-01 -1.73122838e-01
-7.55196273e-01 1.81237429e-01 2.35894293e-01 3.93958464e-02
3.83152127e-01 7.24813700e-01 -6.21180475e-01 -3.87198180e-01
-8.78712475e-01 3.22802216e-01 8.36437166e-01 5.87515056e-01
2.02773198e-01 -3.64192836e-02 -3.12908292e-01 7.69788265e-01
7.69651234e-01 2.24264011e-01 2.21016839e-01 -8.95458400e-01
1.24590136e-01 -2.04286296e-02 3.73445541e-01 -5.74457645e-01
-3.35954636e-01 -4.94856656e-01 -7.66143382e-01 3.00913751e-01
4.11255151e-01 -2.25727022e-01 -1.06482816e+00 1.62894452e+00
4.53913242e-01 -3.17595415e-02 7.22278431e-02 6.02394760e-01
3.63514543e-01 6.09497607e-01 5.10954201e-01 -6.11290514e-01
1.42685568e+00 2.82557104e-02 -6.86311185e-01 6.47118092e-02
4.75210875e-01 -1.09499180e+00 5.60291946e-01 8.20686102e-01
-1.15174961e+00 1.45940054e-02 -7.14494109e-01 5.21008909e-01
-1.00251377e-01 -3.24271142e-01 5.09397030e-01 1.26480877e+00
-6.31112516e-01 7.01683879e-01 -7.99936891e-01 7.28571042e-02
5.15957057e-01 7.48282790e-01 -1.85477167e-01 9.88405421e-02
-1.06987536e+00 9.62989151e-01 6.54423594e-01 3.03372085e-01
-8.23439658e-01 -8.86508286e-01 -6.36879206e-01 4.35117558e-02
6.15132928e-01 -7.15904534e-01 1.12324286e+00 -5.77341914e-01
-1.65365100e+00 3.48686963e-01 1.38798565e-01 -3.40821654e-01
6.75171971e-01 2.64375597e-01 -1.14639655e-01 -4.01482165e-01
-1.13239288e-01 4.13161010e-01 7.07700670e-01 -9.32437301e-01
-2.50887156e-01 -3.67532641e-01 -3.90861303e-01 1.85501769e-01
4.71727878e-01 3.30400765e-01 -8.06545764e-02 -7.44688988e-01
6.76443651e-02 -1.02091157e+00 -5.93596220e-01 -4.25894260e-01
-4.90477860e-01 -1.02649108e-01 3.73975962e-01 -5.73074102e-01
1.08348250e+00 -1.75396752e+00 9.72714871e-02 6.71345592e-01
-1.53839290e-01 1.42823234e-01 1.40925795e-01 3.31204653e-01
-1.96778983e-01 2.70136029e-01 -6.01712584e-01 7.15342015e-02
7.92030022e-02 3.32096338e-01 2.97717780e-01 8.05536509e-01
5.92759073e-01 6.24432981e-01 -9.44907367e-01 -2.66390204e-01
1.36753187e-01 5.91666102e-01 -4.65246171e-01 -8.55068956e-03
-2.06641302e-01 6.44603312e-01 -4.39761400e-01 5.05767524e-01
6.22847855e-01 6.87284097e-02 3.55124027e-02 1.72890469e-01
-1.29094481e-01 7.17834383e-02 -1.72719264e+00 1.10258269e+00
-2.04359904e-01 -9.73796844e-02 -5.26512191e-02 -9.42963719e-01
9.30533588e-01 4.94521767e-01 4.86679405e-01 -3.36343795e-01
2.60932833e-01 3.67733359e-01 3.15531313e-01 -5.79045005e-02
3.09346974e-01 -8.60857785e-01 5.98953292e-02 2.44356692e-01
-4.97002751e-02 -4.68715608e-01 -3.05638045e-01 -1.69089273e-01
1.14886498e+00 2.39078119e-01 4.12896693e-01 -6.64437711e-01
2.90381312e-01 -5.11699557e-01 7.49248147e-01 7.71159828e-01
6.44338131e-02 7.64434159e-01 9.10189033e-01 3.11118644e-02
-1.20426083e+00 -9.87226009e-01 -7.30195522e-01 4.15305942e-01
-5.55550456e-01 -1.77262917e-01 -2.22088471e-01 -4.56345797e-01
-1.15819566e-01 9.94678795e-01 -4.55694288e-01 -2.52329946e-01
9.53730196e-02 -1.63263524e+00 1.89412192e-01 2.25497499e-01
-2.01884091e-01 -7.43979335e-01 -5.52720487e-01 5.55407405e-01
4.43208426e-01 -4.95081663e-01 -1.74202502e-01 7.74316072e-01
-6.90989971e-01 -8.08764517e-01 -1.05608511e+00 5.15695810e-02
4.08719569e-01 -5.94292939e-01 9.70856905e-01 -6.62493229e-01
-3.15863371e-01 2.80723870e-01 -5.24069294e-02 -9.52783823e-01
-5.99075198e-01 -4.72119063e-01 -1.20804518e-01 -3.80934477e-02
4.44064945e-01 -2.61577457e-01 -6.43681169e-01 4.92027611e-01
-1.08636248e+00 -6.11593008e-01 4.51482207e-01 1.17898071e+00
6.80292904e-01 4.68906552e-01 6.49359763e-01 -9.47248876e-01
8.62491548e-01 -6.24797642e-01 -8.42240870e-01 -8.02942738e-03
-6.95108950e-01 2.95149267e-01 1.60310362e-02 -5.16496480e-01
-1.11353719e+00 1.68362319e-01 -2.65380114e-01 -2.26154432e-01
-1.03610188e-01 8.23793352e-01 -4.71095890e-01 4.44213897e-02
6.03310645e-01 -3.45563740e-01 8.99798200e-02 -4.25156742e-01
-1.60608198e-02 7.70695210e-01 1.39825106e-01 -8.47796559e-01
2.30796978e-01 1.43708840e-01 6.15977705e-01 -6.48355722e-01
-1.21636845e-01 -5.71419656e-01 -2.12772071e-01 1.34898812e-01
6.30819857e-01 -6.86825514e-01 -1.04779720e+00 1.65589258e-01
-1.01180518e+00 -1.26659229e-01 -5.71265757e-01 6.73590600e-01
-9.61477995e-01 4.06458914e-01 8.39579776e-02 -1.35715783e+00
-2.27976479e-02 -1.73754930e+00 1.06827569e+00 -4.16106842e-02
-5.45716107e-01 -1.07016516e+00 1.70268983e-01 1.16873458e-01
2.90003419e-01 4.56580430e-01 8.45114946e-01 -1.10340786e+00
-1.03635885e-01 -4.10444736e-01 2.84479037e-02 3.05157512e-01
-1.55503199e-01 3.45948115e-02 -9.04470861e-01 -1.41728789e-01
1.27097979e-01 -5.00400662e-02 5.40289700e-01 8.98439229e-01
9.73086417e-01 -3.41527700e-01 -1.18271858e-01 1.75796092e-01
1.23603630e+00 8.02142918e-01 9.46114361e-01 3.92642587e-01
1.85120896e-01 9.55524743e-01 5.75679064e-01 8.27559054e-01
-7.30728582e-02 1.00006688e+00 4.48362082e-01 -1.38321994e-02
5.61570704e-01 3.95927504e-02 1.00212395e-02 5.01081683e-02
-4.89354469e-02 -4.01764661e-01 -9.97600913e-01 5.81975937e-01
-1.84185386e+00 -9.31716383e-01 -4.42384720e-01 2.78702593e+00
9.94113028e-01 2.05749169e-01 3.41216475e-01 8.55417550e-02
6.80825055e-01 -4.77224648e-01 -4.38345343e-01 -9.92756248e-01
-1.75473288e-01 4.12059903e-01 7.43859589e-01 4.81831521e-01
-9.90443707e-01 2.98842341e-01 5.89598370e+00 1.20836735e+00
-4.29086208e-01 7.85013139e-02 1.11342466e+00 4.86703105e-02
-2.78986126e-01 2.46307135e-01 -6.78435981e-01 7.10957587e-01
1.50367105e+00 -1.07386604e-01 -6.71412647e-02 4.39105868e-01
6.08360827e-01 -5.95535815e-01 -1.05243456e+00 5.77723086e-01
-4.24599290e-01 -8.54146481e-01 -4.03072417e-01 4.96403217e-01
7.11997151e-01 -4.80777591e-01 2.02009499e-01 2.53124069e-02
6.67590141e-01 -1.35182655e+00 4.36844885e-01 6.19633198e-01
4.87523258e-01 -1.18216109e+00 1.16642416e+00 4.24920142e-01
-3.97804976e-01 2.16998514e-02 -2.96833694e-01 2.88031459e-01
3.24384540e-01 7.69553483e-01 -1.00260293e+00 5.04494011e-01
4.88092780e-01 3.42319906e-02 -4.29428481e-02 1.67480528e+00
1.42151430e-01 3.37981820e-01 -7.35570669e-01 1.25516117e-01
4.75080043e-01 -5.03812432e-01 7.46645093e-01 1.07393694e+00
5.03529727e-01 5.52003905e-02 -3.37765068e-01 9.19247150e-01
4.59430039e-01 7.01818168e-02 -4.20409888e-01 -1.22174598e-01
3.03826481e-01 7.33625650e-01 -7.65513003e-01 -1.36739062e-02
-6.63691908e-02 5.01717746e-01 -5.93523383e-01 3.44487727e-01
-4.97225434e-01 -2.52807975e-01 2.55241662e-01 2.73781866e-02
3.64446759e-01 3.65110695e-01 -4.51561511e-01 -5.03626764e-01
-2.70004690e-01 -1.03356552e+00 3.86738747e-01 -5.64037919e-01
-1.27107346e+00 2.54927188e-01 4.95756119e-01 -7.47209251e-01
-3.70515049e-01 -5.68218648e-01 -5.35492241e-01 1.45322299e+00
-1.10881793e+00 -7.35515773e-01 6.72851682e-01 7.34722316e-02
2.81779975e-01 -1.26947820e-01 6.18968070e-01 7.65948370e-02
-4.23967272e-01 2.27025345e-01 4.56303984e-01 -8.25810313e-01
5.74810624e-01 -1.30928516e+00 -3.53253745e-02 4.33723569e-01
-2.58323193e-01 8.12586725e-01 1.34837973e+00 -8.99463296e-01
-9.67856526e-01 -9.27922308e-01 7.77038813e-01 -5.81862867e-01
7.31760800e-01 9.99787450e-03 -8.64460289e-01 2.42411762e-01
-2.59678364e-01 -2.98353165e-01 7.21481860e-01 1.02153912e-01
2.19620094e-01 3.55582982e-01 -1.47292399e+00 5.71835518e-01
2.39285856e-01 -1.10096328e-01 -4.37864393e-01 6.10544205e-01
1.44699305e-01 -2.08104968e-01 -1.34848440e+00 6.95610106e-01
5.39994061e-01 -5.82294762e-01 9.47819889e-01 -6.49996340e-01
7.92631134e-02 -3.28918934e-01 -1.48503214e-01 -1.25342619e+00
-1.34810373e-01 -1.07009923e+00 1.61936969e-01 1.19791579e+00
4.88176495e-01 -5.97904623e-01 7.94541895e-01 1.20889091e+00
2.45051589e-02 -9.51115489e-01 -1.50082242e+00 -1.03580606e+00
3.94065231e-01 -5.53960264e-01 4.84985679e-01 4.93109047e-01
-2.29836106e-01 1.91158667e-01 -4.30643588e-01 -4.45826724e-03
8.03097665e-01 -7.51512706e-01 4.77279544e-01 -1.36929512e+00
-6.41567290e-01 -5.06781697e-01 -6.26558959e-01 -4.60690677e-01
-2.32390046e-01 -3.80117565e-01 2.92667121e-01 -1.08208346e+00
3.79972100e-01 -4.59421128e-01 -1.70992970e-01 3.98806036e-02
-1.84178516e-01 -2.07761526e-01 -1.29493549e-01 -1.38691276e-01
-8.40768814e-02 5.54313123e-01 8.38787556e-01 2.27449220e-02
-4.06042546e-01 6.70182109e-01 -7.18951464e-01 6.20417893e-01
4.85540688e-01 -9.29831386e-01 -4.44946826e-01 3.92300665e-01
4.99747962e-01 2.21469045e-01 3.69652331e-01 -1.85313016e-01
7.06167296e-02 -2.81989038e-01 2.68512219e-01 -4.80417341e-01
3.66920590e-01 -9.79219437e-01 1.07604575e+00 3.15115958e-01
-1.40498042e-01 -1.54896662e-01 3.69220465e-01 8.98259103e-01
7.73461163e-02 -7.98616707e-01 8.33816826e-01 -2.96915948e-01
1.21593915e-01 9.30969510e-03 -5.24800956e-01 -2.97270507e-01
1.02648330e+00 -5.71610868e-01 3.34524363e-02 -2.44254038e-01
-9.06908512e-01 1.22957528e-01 4.17177498e-01 -8.48989785e-02
4.81823117e-01 -9.36865985e-01 -7.38804221e-01 -3.57689373e-02
5.97634120e-03 4.74773422e-02 1.71191469e-01 1.12172186e+00
-2.70094037e-01 6.03878319e-01 2.00760752e-01 -6.73890293e-01
-1.17573643e+00 5.05470455e-01 2.42076978e-01 -3.59134287e-01
5.82206212e-02 8.57588768e-01 2.84196198e-01 -3.88958931e-01
2.25676686e-01 1.00332730e-01 -7.96618536e-02 1.71797141e-01
1.18254162e-01 5.02030730e-01 3.47905159e-01 -6.63872838e-01
-3.09195250e-01 2.72334397e-01 -3.84400114e-02 -2.73051441e-01
1.50520968e+00 3.00164800e-04 6.83460981e-02 3.26681703e-01
8.12397003e-01 -3.40399057e-01 -1.39395630e+00 4.62268814e-02
4.57868397e-01 -3.47187132e-01 6.21561706e-01 -1.03443086e+00
-3.45938504e-01 6.74027264e-01 5.60734093e-01 -1.16377082e-02
8.65917265e-01 -1.86043531e-01 -1.34426862e-01 2.88256463e-02
1.98018521e-01 -1.11818981e+00 -4.00404751e-01 8.29214081e-02
8.92888665e-01 -1.25845695e+00 5.16909420e-01 -1.50699437e-01
-9.48120117e-01 7.32458830e-01 -9.40896720e-02 3.56111020e-01
8.15767169e-01 1.83024615e-01 -5.77948689e-01 -3.01097602e-01
-7.82843649e-01 -3.97450216e-02 5.34200609e-01 7.29230881e-01
4.07891572e-01 3.62732969e-02 -6.07827067e-01 6.03089452e-01
3.15581948e-01 -6.14848286e-02 4.41567093e-01 1.15581453e+00
-1.13684990e-01 -1.20253026e+00 -8.19864333e-01 6.04799271e-01
-8.77921104e-01 -2.26530910e-01 -1.44414157e-01 6.56075180e-01
6.32523298e-02 9.98557925e-01 -3.99440713e-02 2.52986223e-01
4.33249325e-01 -1.73420813e-02 2.36840203e-01 -6.64885342e-01
-6.32280409e-01 9.59254324e-01 2.32378751e-01 -1.46617666e-01
-3.81307721e-01 -9.21486437e-01 -4.66930121e-01 -1.48063451e-01
-7.92001009e-01 2.30031669e-01 1.02546620e+00 9.75063145e-01
2.50107586e-01 5.00274003e-01 3.42698932e-01 -6.74909055e-01
-8.49653423e-01 -7.55743027e-01 -7.84440458e-01 -3.23385671e-02
1.11194886e-02 -8.37778747e-01 -3.84143323e-01 -1.85425088e-01]
|
[6.232745170593262, 3.7715375423431396]
|
02d21bd2-cfaa-4b30-85e9-2d9a074b1e04
|
learn-what-not-to-learn-action-elimination
|
1809.02121
| null |
http://arxiv.org/abs/1809.02121v3
|
http://arxiv.org/pdf/1809.02121v3.pdf
|
Learn What Not to Learn: Action Elimination with Deep Reinforcement Learning
|
Learning how to act when there are many available actions in each state is a
challenging task for Reinforcement Learning (RL) agents, especially when many
of the actions are redundant or irrelevant. In such cases, it is sometimes
easier to learn which actions not to take. In this work, we propose the
Action-Elimination Deep Q-Network (AE-DQN) architecture that combines a Deep RL
algorithm with an Action Elimination Network (AEN) that eliminates sub-optimal
actions. The AEN is trained to predict invalid actions, supervised by an
external elimination signal provided by the environment. Simulations
demonstrate a considerable speedup and added robustness over vanilla DQN in
text-based games with over a thousand discrete actions.
|
['Matan Haroush', 'Daniel J. Mankowitz', 'Tom Zahavy', 'Shie Mannor', 'Nadav Merlis']
|
2018-09-06
|
learn-what-not-to-learn-action-elimination-1
|
http://papers.nips.cc/paper/7615-learn-what-not-to-learn-action-elimination-with-deep-reinforcement-learning
|
http://papers.nips.cc/paper/7615-learn-what-not-to-learn-action-elimination-with-deep-reinforcement-learning.pdf
|
neurips-2018-12
|
['text-based-games']
|
['playing-games']
|
[ 1.36475116e-01 9.98681486e-02 -1.48452669e-01 -4.48233262e-03
-3.87342453e-01 -4.50640261e-01 4.91309226e-01 -1.11062646e-01
-8.88995111e-01 1.33584392e+00 2.58745790e-01 -4.00429845e-01
-1.99014381e-01 -8.42678070e-01 -6.98832273e-01 -7.70512044e-01
-1.85955837e-01 5.61835229e-01 2.78156906e-01 -6.58368886e-01
2.63585120e-01 3.67804348e-01 -1.45356596e+00 9.44444463e-02
7.84973681e-01 5.56505024e-01 7.01985434e-02 8.17725360e-01
2.54614145e-01 1.49070466e+00 -8.58409762e-01 3.56896361e-03
5.50154865e-01 -6.74713850e-01 -7.09811270e-01 -1.58661440e-01
-1.99168220e-01 -8.86866987e-01 -5.19989371e-01 1.04509211e+00
6.77138329e-01 5.01205742e-01 2.94169277e-01 -1.19333518e+00
-1.97647139e-01 8.26046884e-01 -4.02739614e-01 2.60459870e-01
3.30379605e-01 9.87355471e-01 9.22069430e-01 -1.49214536e-01
5.56227803e-01 1.57732666e+00 7.85277113e-02 8.42362463e-01
-1.15199590e+00 -6.09843075e-01 3.37925345e-01 2.90809005e-01
-9.23260868e-01 -1.67781025e-01 4.37420875e-01 6.55453280e-02
1.25642562e+00 -1.26782432e-01 7.65850246e-01 1.20593071e+00
6.08793318e-01 1.01477516e+00 1.07442939e+00 -2.26590425e-01
6.81561351e-01 -4.14110214e-01 -5.30028880e-01 3.98308218e-01
9.95378047e-02 7.65996635e-01 -3.52808446e-01 8.44559632e-03
7.64760673e-01 -3.50055820e-03 1.17318630e-01 -4.05622214e-01
-7.04430163e-01 9.66400802e-01 2.18790233e-01 -1.38823334e-02
-8.60455394e-01 5.87420821e-01 5.00423074e-01 8.80127370e-01
8.36979151e-02 8.22802007e-01 -5.63266873e-01 -5.22015452e-01
-3.97918284e-01 7.19092727e-01 5.44993877e-01 3.81542325e-01
6.84937894e-01 5.98505020e-01 -4.13359165e-01 4.85330015e-01
-7.54959211e-02 4.74490613e-01 5.07516205e-01 -1.37794101e+00
3.46232980e-01 4.85217124e-01 4.70770657e-01 -5.10013580e-01
-4.66716319e-01 -3.28123719e-01 -6.00137115e-01 9.99189138e-01
2.28391200e-01 -7.21273541e-01 -8.23321998e-01 1.98831308e+00
2.78056204e-01 3.71851549e-02 3.04031849e-01 8.93349051e-01
5.10697328e-02 4.43314433e-01 2.27005988e-01 -3.31246555e-01
5.67331135e-01 -8.73557866e-01 -8.37806463e-01 -4.93381113e-01
6.49974346e-01 -2.17109144e-01 9.31276739e-01 6.18729115e-01
-1.15625858e+00 -4.21378464e-01 -9.48943436e-01 4.44513828e-01
-3.08181554e-01 -3.39452684e-01 5.99702597e-01 1.32405505e-01
-1.02118206e+00 1.01888442e+00 -6.92532480e-01 3.27395275e-02
4.47905064e-01 8.34929168e-01 -1.89498410e-01 6.71438202e-02
-1.66988778e+00 1.16793764e+00 6.80470407e-01 -2.26136893e-02
-1.56929421e+00 -1.31431103e-01 -7.77538300e-01 2.05427304e-01
1.12734306e+00 -3.46144497e-01 1.76067245e+00 -1.33737898e+00
-2.03880858e+00 -3.15913111e-02 4.86901224e-01 -8.19648147e-01
7.36876965e-01 -3.62515628e-01 -2.26935536e-01 8.33661631e-02
1.06848188e-01 6.42362118e-01 1.08919215e+00 -6.60782099e-01
-7.47910082e-01 -1.66719183e-01 6.15084112e-01 6.79633737e-01
1.41610101e-01 -1.55998766e-01 1.30062634e-02 -5.28737426e-01
-5.54688334e-01 -9.92082477e-01 -7.23186791e-01 -2.53904730e-01
-2.72965282e-01 -3.80145013e-01 6.19274974e-01 -4.16323215e-01
1.18297958e+00 -1.94012094e+00 4.10191298e-01 9.79640037e-02
1.71934694e-01 5.40153742e-01 -5.68586826e-01 6.20328546e-01
-1.05847940e-01 -2.06591517e-01 2.19982304e-02 2.81982690e-01
1.94274455e-01 5.29376686e-01 -1.71135217e-01 2.48239741e-01
2.85283715e-01 8.48104537e-01 -1.16018486e+00 -4.73650433e-02
2.94195622e-01 -1.64900925e-02 -6.57147646e-01 2.28420660e-01
-6.87315702e-01 6.10021830e-01 -6.55267775e-01 2.74117261e-01
4.14392054e-01 2.70085424e-01 4.52391088e-01 6.57050073e-01
-6.40014187e-02 3.46662074e-01 -1.40365529e+00 1.24251926e+00
-2.24882483e-01 2.29656667e-01 7.18551725e-02 -9.48255420e-01
4.93221670e-01 2.81498075e-01 4.74935263e-01 -1.35475636e+00
2.76024759e-01 -6.22615814e-02 6.24848187e-01 -2.90992528e-01
3.34773123e-01 -1.65340826e-01 -2.35467553e-01 8.42670798e-01
-3.81717905e-02 -2.46993437e-01 6.98051453e-01 2.57192135e-01
1.61051464e+00 4.54958946e-01 5.21159172e-01 1.73661456e-01
4.99300241e-01 -1.41348794e-01 9.57585394e-01 1.20085216e+00
-4.94242102e-01 -1.90096319e-01 9.80440021e-01 -7.03338385e-01
-8.64699781e-01 -7.40353227e-01 6.00155175e-01 1.37804842e+00
5.84837720e-02 -4.41114515e-01 -6.11967266e-01 -8.98287892e-01
1.44071221e-01 9.81606245e-01 -6.22776926e-01 -6.91864550e-01
-8.06332350e-01 -2.32621759e-01 2.99428940e-01 6.56934857e-01
5.73059082e-01 -1.76469147e+00 -1.02544701e+00 5.72134435e-01
2.34464526e-01 -4.94971156e-01 -3.56175125e-01 5.14456928e-01
-6.01072490e-01 -1.10719693e+00 -2.48113677e-01 -1.74686745e-01
3.44182968e-01 -1.43662598e-02 9.86378074e-01 3.97512317e-02
1.56743005e-01 3.37034054e-02 -2.46791765e-01 -4.64234680e-01
-6.27201438e-01 -2.41215423e-01 3.73095155e-01 -4.53999758e-01
3.26823294e-01 -4.12572563e-01 -5.45240998e-01 1.67990535e-01
-9.30724859e-01 -8.73669609e-02 7.65681148e-01 9.97494698e-01
3.34889591e-01 3.45620275e-01 6.89942777e-01 -8.06523681e-01
1.04254615e+00 -2.99407214e-01 -8.82140934e-01 -3.73841077e-02
-3.68958116e-01 5.27753532e-01 1.29250014e+00 -5.67737222e-01
-1.04559481e+00 1.47589982e-01 -1.87656000e-01 -3.60262990e-01
2.60209590e-02 8.28174353e-02 -6.83613643e-02 1.98931232e-01
5.94423056e-01 2.54931986e-01 -8.61639604e-02 -1.14412524e-01
4.16239321e-01 3.47785324e-01 5.97060134e-04 -4.69137609e-01
5.96842170e-01 -9.74540934e-02 -3.37456912e-02 -2.39437908e-01
-7.16301203e-01 1.53357208e-01 -3.69005173e-01 -3.40804487e-01
6.49623573e-01 -6.94302380e-01 -1.14190137e+00 4.33902979e-01
-7.45307267e-01 -9.92689013e-01 -6.67925715e-01 4.00397450e-01
-9.12847340e-01 1.09724678e-01 -5.62262118e-01 -8.83046985e-01
-1.67613834e-01 -1.29719782e+00 5.42268217e-01 5.83001435e-01
-1.29517093e-01 -5.47209203e-01 2.72266179e-01 -3.36997956e-02
3.83900315e-01 -6.42455965e-02 6.93861127e-01 -6.41807199e-01
-4.88313168e-01 8.94416422e-02 3.04640085e-01 3.86835903e-01
1.57449558e-01 -2.46005297e-01 -4.22898203e-01 -6.12623870e-01
-2.45233953e-01 -8.05112064e-01 7.58263528e-01 4.97376233e-01
1.16245699e+00 -5.56291938e-01 1.24334889e-02 1.58946425e-01
1.07401860e+00 6.35133624e-01 7.03386128e-01 4.70475942e-01
2.01186582e-01 2.15484664e-01 9.33415532e-01 7.50698864e-01
-2.87748482e-02 5.26463270e-01 8.42389643e-01 3.31427418e-02
1.74077943e-01 -3.32178742e-01 7.90146828e-01 4.27163579e-02
-8.68664086e-02 -3.31289321e-01 -4.67525780e-01 1.61055312e-01
-2.07935095e+00 -1.24448597e+00 4.98113841e-01 2.26111078e+00
1.06108809e+00 6.07621431e-01 2.52899051e-01 6.75460277e-03
4.83199000e-01 2.38218620e-01 -1.26590645e+00 -8.55597794e-01
2.48682164e-02 5.28891802e-01 4.96003538e-01 5.49853444e-01
-1.11957955e+00 1.31655300e+00 6.72019196e+00 7.37370133e-01
-8.80525529e-01 -1.57595247e-01 5.43944359e-01 -3.07184219e-01
1.27015248e-01 -5.28547764e-02 -4.72282887e-01 4.05953884e-01
9.37516034e-01 -8.02749544e-02 8.12586665e-01 7.58905828e-01
6.30620718e-01 -6.07155979e-01 -8.36847246e-01 3.83327633e-01
-3.73136640e-01 -9.79963422e-01 -9.24459323e-02 -1.54584169e-01
7.31127024e-01 -1.12962443e-02 -3.23835462e-02 9.21929717e-01
1.37299049e+00 -1.00431812e+00 4.90961671e-01 3.23241115e-01
6.02596402e-01 -1.22864449e+00 6.14915192e-01 7.11641848e-01
-7.57593453e-01 -6.59961164e-01 -4.09070462e-01 -6.46915436e-01
-1.61784694e-01 -2.87259609e-01 -6.27998054e-01 2.45700434e-01
5.20610273e-01 4.54108298e-01 -3.18676233e-01 7.13147461e-01
-9.82078075e-01 4.77619618e-01 -6.57218844e-02 -2.94038773e-01
6.74365938e-01 -3.89060199e-01 5.13438523e-01 5.71958125e-01
-1.75829321e-01 4.09025878e-01 3.17836046e-01 6.35651410e-01
-5.60154021e-02 -3.32916737e-01 -6.76362336e-01 -1.21262021e-01
2.31953070e-01 9.17439044e-01 -4.94822562e-01 -3.44730228e-01
-1.52136952e-01 1.10702193e+00 5.93716383e-01 4.42701966e-01
-8.36304784e-01 -3.29876959e-01 9.35758173e-01 -2.88363934e-01
3.59172702e-01 -5.72592653e-02 4.32464600e-01 -7.92691529e-01
-4.35497373e-01 -1.30846334e+00 4.92877871e-01 -7.38479853e-01
-9.82752562e-01 2.11901546e-01 -3.07664603e-01 -1.20599389e+00
-6.46550119e-01 -4.23737884e-01 -6.71251714e-01 5.88445306e-01
-1.38886631e+00 -3.49935353e-01 3.86479080e-01 5.78825057e-01
7.04091311e-01 -1.79569349e-01 6.01842403e-01 -7.56957233e-02
-7.19015539e-01 3.00842404e-01 2.82574683e-01 1.22140527e-01
5.65454006e-01 -1.50357223e+00 1.03870139e-01 7.74928272e-01
-2.37021551e-01 2.33821556e-01 8.60133708e-01 -7.78607428e-01
-1.28520095e+00 -1.04465532e+00 2.90231258e-01 -2.90259123e-02
5.62226057e-01 -1.26070485e-01 -6.08736813e-01 7.23186255e-01
4.02998984e-01 -1.75123438e-01 2.25390851e-01 -3.58349442e-01
7.58776069e-02 -1.09638318e-01 -1.03772736e+00 1.11112177e+00
9.41524029e-01 -1.74247175e-01 -6.50155425e-01 3.97876978e-01
7.96919823e-01 -3.80229294e-01 -1.30034581e-01 -4.21435088e-02
1.98103935e-01 -9.65941012e-01 6.60073638e-01 -1.14884794e+00
4.20279503e-01 -4.02466625e-01 2.26875886e-01 -1.89311767e+00
-3.72307211e-01 -9.80633318e-01 -1.34271979e-01 4.51894253e-01
4.97469120e-02 -4.35450673e-01 7.79831886e-01 5.41648686e-01
1.25084192e-01 -6.23330057e-01 -1.03012121e+00 -8.28882098e-01
5.89987598e-02 -2.51367390e-01 6.06048465e-01 5.18802583e-01
-2.39767898e-02 4.84603882e-01 -7.66170382e-01 -1.93636119e-01
2.53864586e-01 -1.25789478e-01 7.20787525e-01 -7.56878972e-01
-5.84943295e-01 -4.07618254e-01 -8.42938945e-02 -1.02442193e+00
3.41147929e-01 -4.32747424e-01 2.32105017e-01 -1.41558814e+00
-1.37819573e-02 4.10198839e-03 -5.66036224e-01 8.51025641e-01
-2.47357637e-01 -3.25237453e-01 3.68616164e-01 -3.17087442e-01
-1.26291466e+00 1.06815255e+00 1.53432250e+00 -2.46930897e-01
-5.65102339e-01 1.48752674e-01 -5.19070506e-01 7.49456525e-01
1.13849151e+00 -5.73264301e-01 -4.94179159e-01 -3.28774691e-01
2.75888503e-01 4.72667128e-01 -2.92554665e-02 -1.01915896e+00
1.07205592e-01 -6.72210276e-01 3.67828995e-01 -4.39590871e-01
1.80160925e-01 -6.95607960e-01 -1.54093787e-01 1.08292902e+00
-6.09867573e-01 2.90377676e-01 2.41315395e-01 6.21645868e-01
3.43909226e-02 -1.14723861e-01 8.21750581e-01 -3.85490209e-01
-9.05256927e-01 2.31252998e-01 -1.00766945e+00 1.19005933e-01
1.26824737e+00 -5.36552966e-02 -1.71535790e-01 -8.86589646e-01
-6.38589621e-01 6.92314982e-01 -2.87830979e-02 3.36443752e-01
6.54419363e-01 -1.19195902e+00 -6.17582202e-01 1.27320647e-01
-2.88198590e-01 -2.56461084e-01 2.84884721e-01 5.76875031e-01
-3.45558196e-01 2.53952026e-01 -6.42451048e-01 1.13411359e-01
-1.01060843e+00 7.12392330e-01 6.60691202e-01 -6.51088893e-01
-6.06851280e-01 5.60845554e-01 5.18823508e-03 -5.17111480e-01
2.60951191e-01 -9.13748890e-02 -4.22542870e-01 -1.64400667e-01
6.49012983e-01 4.13221359e-01 -2.40945637e-01 -2.21268058e-01
-1.25853512e-02 -1.54532239e-01 -2.09544584e-01 -2.56009877e-01
1.46059668e+00 2.08569735e-01 1.69881627e-01 -7.72613892e-03
3.43984097e-01 -3.62968028e-01 -1.81315243e+00 -2.06832662e-01
-3.86387520e-02 -3.48896146e-01 -1.30864859e-01 -1.12837815e+00
-9.66917872e-01 6.37911260e-01 5.06361783e-01 1.85810570e-02
1.21283305e+00 -5.28040349e-01 4.55703348e-01 8.05592954e-01
3.76774430e-01 -1.82012379e+00 4.27025467e-01 9.48426783e-01
8.43899488e-01 -1.07072186e+00 2.60081198e-02 5.61182380e-01
-8.82412553e-01 9.90978539e-01 1.08369696e+00 -2.70977646e-01
1.16271734e-01 1.16002835e-01 -6.29824772e-02 -2.64125760e-03
-1.24443090e+00 -4.35933441e-01 -5.65908253e-01 6.28052950e-01
-1.78733349e-01 1.71496809e-01 -4.49838817e-01 2.63364017e-01
1.54317126e-01 4.52325083e-02 7.76973844e-01 1.21999097e+00
-6.22324765e-01 -1.33442724e+00 -1.37315452e-01 6.26448452e-01
-3.24421674e-01 3.62768658e-02 -4.29114342e-01 8.42150390e-01
1.46839857e-01 9.81084228e-01 -7.25450814e-02 -3.22839975e-01
3.58727306e-01 -9.91049111e-02 3.71814311e-01 -5.74589372e-01
-8.62733066e-01 5.22414111e-02 6.44105300e-02 -1.06263089e+00
-1.35711893e-01 -4.99810964e-01 -1.63539672e+00 -3.72012705e-01
6.95705181e-03 -1.91809554e-02 5.78042679e-02 9.88693357e-01
2.56670505e-01 8.60648632e-01 7.54055560e-01 -4.62574661e-01
-1.16519928e+00 -7.55151272e-01 -7.95394897e-01 4.18262869e-01
4.03075129e-01 -8.13694954e-01 -1.53672472e-01 -6.21190727e-01]
|
[3.8863189220428467, 1.7092944383621216]
|
790a8e83-4c53-489c-bb36-84fe32a4ea36
|
tasnet-surpassing-ideal-time-frequency
|
1809.07454
| null |
https://arxiv.org/abs/1809.07454v3
|
https://arxiv.org/pdf/1809.07454v3.pdf
|
Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation
|
Single-channel, speaker-independent speech separation methods have recently seen great progress. However, the accuracy, latency, and computational cost of such methods remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of time-frequency representation for speech separation, and the long latency in calculating the spectrograms. To address these shortcomings, we propose a fully-convolutional time-domain audio separation network (Conv-TasNet), a deep learning framework for end-to-end time-domain speech separation. Conv-TasNet uses a linear encoder to generate a representation of the speech waveform optimized for separating individual speakers. Speaker separation is achieved by applying a set of weighting functions (masks) to the encoder output. The modified encoder representations are then inverted back to the waveforms using a linear decoder. The masks are found using a temporal convolutional network (TCN) consisting of stacked 1-D dilated convolutional blocks, which allows the network to model the long-term dependencies of the speech signal while maintaining a small model size. The proposed Conv-TasNet system significantly outperforms previous time-frequency masking methods in separating two- and three-speaker mixtures. Additionally, Conv-TasNet surpasses several ideal time-frequency magnitude masks in two-speaker speech separation as evaluated by both objective distortion measures and subjective quality assessment by human listeners. Finally, Conv-TasNet has a significantly smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications.
|
['Nima Mesgarani', 'Yi Luo']
|
2018-09-20
| null | null | null | null |
['music-source-separation', 'speaker-separation']
|
['music', 'speech']
|
[ 7.15674758e-02 -4.52225059e-01 1.85515121e-01 -3.11424106e-01
-1.01260078e+00 -5.15229344e-01 1.80383772e-01 -1.44874871e-01
-3.26228797e-01 3.35404724e-01 1.76615015e-01 -2.51869828e-01
-2.76951462e-01 -1.28627583e-01 -1.72172412e-01 -9.18846250e-01
-2.69174933e-01 -1.70318171e-01 1.61342680e-01 -1.99138656e-01
-1.41073406e-01 5.27413487e-01 -1.69504714e+00 1.92432687e-01
1.08850431e+00 1.20012546e+00 3.88779283e-01 9.69800413e-01
7.99020678e-02 3.65160555e-01 -9.72574890e-01 -1.14403467e-03
1.67760372e-01 -6.10136211e-01 -2.08251476e-01 -1.84832618e-01
1.50604695e-01 -2.96917349e-01 -6.20693445e-01 9.76810217e-01
1.03392303e+00 3.33773881e-01 3.96327764e-01 -1.07561839e+00
-3.87978256e-01 6.87052190e-01 -2.37889171e-01 4.84498024e-01
1.29651636e-01 -8.28159526e-02 8.25458229e-01 -9.51054215e-01
-2.98918903e-01 1.07243896e+00 8.44947398e-01 4.40671712e-01
-1.25803852e+00 -9.64919627e-01 3.02895755e-02 2.74293542e-01
-1.51173401e+00 -1.06086171e+00 9.18926239e-01 -1.95716038e-01
1.43948448e+00 5.43599963e-01 4.98691320e-01 7.04245508e-01
-6.81285933e-02 5.64848363e-01 7.35952497e-01 -4.41499174e-01
1.21077642e-01 -5.27991317e-02 4.20532562e-02 1.89132586e-01
-2.74348915e-01 2.87904650e-01 -7.34165490e-01 4.46055941e-02
5.72916329e-01 -2.86758929e-01 -6.32609665e-01 1.97084740e-01
-7.26199210e-01 4.21018958e-01 2.33688980e-01 5.99469423e-01
-2.35106692e-01 -4.01497222e-02 3.98941278e-01 3.00371647e-01
5.42390883e-01 1.60880417e-01 -3.55246842e-01 -1.14624768e-01
-1.45768702e+00 5.48539385e-02 5.97033620e-01 5.61332643e-01
1.79063201e-01 8.99594903e-01 -2.75406819e-02 1.08752704e+00
3.51611465e-01 5.95095992e-01 9.19138491e-01 -5.51590681e-01
4.24739152e-01 -1.19265907e-01 -1.45468861e-01 -7.02000916e-01
-4.14676070e-01 -7.21100509e-01 -7.79561162e-01 2.10015655e-01
2.38664687e-01 -2.90213048e-01 -1.10087883e+00 1.85310507e+00
5.95122315e-02 4.02546018e-01 4.08254594e-01 1.07872450e+00
9.29963171e-01 9.98677015e-01 -3.21813047e-01 -4.92209643e-01
1.23302329e+00 -1.05800259e+00 -1.14819121e+00 -4.70964134e-01
5.99313006e-02 -1.00987065e+00 5.00083566e-01 4.24476147e-01
-1.35829997e+00 -9.12292302e-01 -1.29949677e+00 -1.01472504e-01
-2.29953960e-01 4.53045905e-01 -6.97113350e-02 9.12807167e-01
-1.15089393e+00 5.91069162e-01 -8.98357213e-01 2.20702991e-01
-7.69260898e-02 7.16043830e-01 -1.15002982e-01 4.60707188e-01
-1.30767894e+00 6.53491020e-01 4.66669276e-02 4.91872609e-01
-9.51825559e-01 -7.60149300e-01 -1.03291285e+00 5.20270824e-01
-4.50478680e-02 -1.64422110e-01 1.40726829e+00 -1.08149552e+00
-1.93671405e+00 3.05434048e-01 -5.15142858e-01 -6.84883714e-01
2.93051917e-02 -2.57648975e-01 -1.25420392e+00 2.08182305e-01
-1.51023090e-01 3.09656203e-01 1.36285937e+00 -7.92655349e-01
-6.04251504e-01 -5.89669384e-02 -4.34819221e-01 2.05882251e-01
-4.57759351e-01 2.77061433e-01 -2.51027316e-01 -1.00208688e+00
2.43382677e-01 -5.68828404e-01 -1.39655560e-01 -4.17503744e-01
-2.25675315e-01 2.42423892e-01 9.47247684e-01 -1.08896768e+00
1.48523569e+00 -2.83336163e+00 2.17378870e-01 -9.83929783e-02
7.24429488e-02 7.95582712e-01 -2.21778750e-01 4.05506760e-01
-4.74023849e-01 -1.93578541e-01 -3.20242733e-01 -7.64025390e-01
-1.89238600e-02 -2.56377041e-01 -5.29850125e-01 5.13053298e-01
3.11914206e-01 4.42069024e-01 -5.23655176e-01 -9.22274664e-02
2.51998693e-01 9.36099231e-01 -3.85508627e-01 3.81014556e-01
3.27413231e-01 3.58640462e-01 3.14002812e-01 2.01033905e-01
8.84806216e-01 4.26350266e-01 1.12767614e-01 -2.33253732e-01
-4.52670246e-01 1.02332294e+00 -1.35647142e+00 1.51263869e+00
-5.98804951e-01 1.01617086e+00 7.58814633e-01 -8.45976949e-01
9.48394418e-01 9.58331704e-01 3.55874836e-01 -7.13984847e-01
2.58983165e-01 6.26212478e-01 3.69744509e-01 -3.67699832e-01
2.33083084e-01 -4.32782561e-01 2.02466920e-01 2.45573044e-01
1.97414339e-01 -3.06789786e-01 -1.29313143e-02 -2.68994480e-01
6.85250700e-01 -3.99609894e-01 -4.93184216e-02 -5.07899607e-03
7.87839711e-01 -6.30786419e-01 6.86373115e-01 7.76393563e-02
-1.41341642e-01 8.25126052e-01 3.85366678e-02 -9.61425423e-04
-6.16309941e-01 -1.17214072e+00 -7.05219433e-02 9.08734560e-01
4.66637732e-03 -4.96487170e-01 -8.36587727e-01 3.60935018e-03
-1.94189444e-01 6.79419875e-01 5.82643040e-02 -3.66340280e-01
-7.80095100e-01 -3.79081309e-01 9.59564328e-01 5.68113148e-01
3.21768790e-01 -6.75925672e-01 -2.76501119e-01 5.01281261e-01
-3.74327600e-01 -1.15466142e+00 -9.07051146e-01 7.09684849e-01
-7.55984485e-01 -3.61843318e-01 -7.54943848e-01 -1.12923145e+00
3.13697785e-01 5.29865324e-01 2.80452877e-01 -3.08680207e-01
-3.01381927e-02 -1.28520116e-01 -4.72170040e-02 -4.72537845e-01
-2.55050033e-01 -1.14801355e-01 4.57290292e-01 4.37039405e-01
3.35749000e-01 -9.35640514e-01 -4.88465279e-01 3.64441454e-01
-8.15760255e-01 -2.39331007e-01 2.98596501e-01 6.24277830e-01
2.64266819e-01 6.11959279e-01 9.04591680e-01 2.17312187e-01
8.01216185e-01 -1.33916992e-03 -6.74523234e-01 -1.88048288e-01
-2.29536399e-01 -2.35384837e-01 9.67463315e-01 -7.50729859e-01
-1.22697616e+00 -2.07585804e-02 -4.56470847e-01 -5.22322476e-01
8.69235471e-02 4.26427424e-01 -3.48415315e-01 1.74173951e-01
3.95001471e-01 3.57915193e-01 6.77993521e-02 -7.30347514e-01
3.19422185e-02 1.18198144e+00 6.99987650e-01 7.24382401e-02
8.04329813e-01 2.14653984e-01 -6.17722809e-01 -1.15016627e+00
-4.14660573e-01 -5.75438738e-01 -4.65366721e-01 5.95814474e-02
7.51994073e-01 -1.13146377e+00 -5.59704781e-01 7.82199740e-01
-1.27938414e+00 -1.55739769e-01 -1.73327610e-01 1.06761622e+00
-2.67074704e-01 3.10955197e-01 -7.07767785e-01 -1.12901378e+00
-4.91406590e-01 -1.29970717e+00 7.93395817e-01 2.73772478e-01
-1.71446025e-01 -7.30633974e-01 -9.61131081e-02 2.29652405e-01
7.05711067e-01 -4.14776415e-01 7.48930395e-01 -6.02719069e-01
-2.52148062e-01 -2.02295199e-01 2.22797573e-01 8.32385182e-01
3.54435682e-01 -3.05626709e-02 -1.64109206e+00 -4.32803512e-01
6.13183200e-01 2.05440551e-01 7.40005612e-01 8.70987833e-01
8.82156849e-01 -1.45060912e-01 -2.17968598e-01 8.14847469e-01
9.96444762e-01 7.70473838e-01 4.72407907e-01 -2.18659550e-01
4.11558956e-01 4.84631419e-01 2.45811716e-01 1.45564824e-01
-1.04973003e-01 6.98146224e-01 7.25281760e-02 -4.02880877e-01
-6.05029523e-01 1.22185141e-01 7.29736030e-01 1.53144205e+00
1.76947773e-01 -3.70377302e-01 -6.75675750e-01 7.67640531e-01
-1.38065350e+00 -1.01177669e+00 -9.98640060e-02 2.36199927e+00
7.92450607e-01 1.99694663e-01 1.17268734e-01 7.58595943e-01
9.05676961e-01 2.60686427e-01 -4.92043078e-01 -6.58796966e-01
-4.90197353e-02 6.28340840e-01 2.74631441e-01 7.14376152e-01
-9.54796612e-01 4.79435176e-01 6.56352568e+00 9.30246055e-01
-1.71374059e+00 1.70594186e-01 9.00765359e-02 -4.95321691e-01
-3.15056443e-02 -3.21763963e-01 -6.38475180e-01 2.76025325e-01
1.47722745e+00 -2.43195906e-01 6.68291807e-01 3.14073354e-01
4.77062225e-01 2.86938787e-01 -1.09491503e+00 1.18647957e+00
7.03778863e-02 -8.46400440e-01 -4.50901717e-01 -1.79036990e-01
2.33850569e-01 -3.62549238e-02 2.90088624e-01 1.86227396e-01
-3.91140878e-01 -1.18449259e+00 1.13937402e+00 -4.17959839e-02
9.99371111e-01 -8.97664130e-01 4.26066220e-01 3.01092625e-01
-1.72457087e+00 -3.93662363e-01 -9.68244951e-03 -1.11554533e-01
3.42735142e-01 6.48746967e-01 -7.78601110e-01 5.08108318e-01
6.13416314e-01 4.04645294e-01 9.03419554e-02 1.02110255e+00
-5.32096028e-02 8.65853131e-01 -3.19303006e-01 3.74099255e-01
1.32692948e-01 4.85034799e-03 7.03162014e-01 1.41819918e+00
4.29573953e-01 -1.18178725e-02 -1.91151291e-01 7.12765634e-01
1.34947568e-01 -2.52190679e-01 -1.52839825e-01 -3.26888591e-01
5.89272559e-01 1.08242583e+00 -4.77841675e-01 -4.23718542e-02
-4.05852169e-01 7.22184718e-01 -1.01204701e-01 6.48058593e-01
-1.08893299e+00 -1.07369995e+00 1.06339347e+00 -1.25594631e-01
5.88109672e-01 -4.70804304e-01 -2.44430706e-01 -7.46407092e-01
1.87702745e-01 -1.02051783e+00 -1.28032908e-01 -5.85467100e-01
-9.11659062e-01 1.16936314e+00 -2.77579784e-01 -1.53853238e+00
-1.85472891e-01 -5.31998694e-01 -5.62319756e-01 1.35211337e+00
-1.68175101e+00 -6.60083413e-01 9.49790403e-02 5.48201084e-01
7.95710564e-01 -2.39957243e-01 8.72444153e-01 7.80965149e-01
-6.98616028e-01 6.11258686e-01 7.94999972e-02 9.86230671e-02
8.14165890e-01 -9.78764057e-01 6.41507685e-01 1.05821109e+00
-5.49857467e-02 5.86646080e-01 6.22455418e-01 -3.27754945e-01
-1.08225429e+00 -9.32604551e-01 1.07352376e+00 2.79514492e-01
3.86499166e-01 -7.05506623e-01 -1.01746655e+00 2.78422564e-01
3.89886200e-01 -1.13713354e-01 9.11972344e-01 -3.90189081e-01
-3.27106416e-01 -5.82207143e-01 -8.48819792e-01 4.26260263e-01
6.85219467e-01 -9.60278690e-01 -7.39455760e-01 -9.05681551e-02
9.85420048e-01 -4.85026807e-01 -4.17360514e-01 1.26258940e-01
4.99016285e-01 -9.87890840e-01 1.09608054e+00 -5.97596802e-02
6.85043558e-02 -5.45426905e-01 -1.38090596e-01 -1.56727755e+00
-5.14278471e-01 -1.02578640e+00 5.95103651e-02 1.35631382e+00
4.86098111e-01 -8.22331131e-01 1.62304118e-01 2.64026582e-01
-5.82749665e-01 -3.68564099e-01 -1.16527021e+00 -1.08929884e+00
-2.96885729e-01 -5.41872144e-01 7.13282168e-01 5.76708078e-01
9.00923386e-02 3.43530089e-01 -2.19135687e-01 7.52773881e-01
2.76262939e-01 -2.31546834e-02 1.16558671e-01 -9.08056796e-01
-3.59180033e-01 -6.53775811e-01 -1.82470351e-01 -1.29292309e+00
6.53761476e-02 -7.16412425e-01 4.09541041e-01 -1.33986187e+00
-5.97170532e-01 -1.62286714e-01 -4.00619268e-01 9.82814804e-02
-1.37663722e-01 3.71758305e-02 6.18713759e-02 -6.39769062e-02
2.00998574e-01 6.67631865e-01 8.21184158e-01 -2.93974489e-01
-5.40512562e-01 1.75421953e-01 -3.15613329e-01 6.33996248e-01
6.61091030e-01 -5.08644760e-01 -5.84655464e-01 -6.88318729e-01
-4.58834082e-01 4.07579154e-01 1.25771627e-01 -1.36068153e+00
5.45768321e-01 1.66063949e-01 2.13933766e-01 -7.24066734e-01
9.71827209e-01 -1.00224948e+00 3.01865667e-01 4.14561689e-01
-2.82565057e-01 -1.71424940e-01 5.78011751e-01 2.05703273e-01
-7.76215136e-01 -2.65296429e-01 9.01547909e-01 3.46531004e-01
-1.64834812e-01 -5.52472472e-02 -7.87349820e-01 -1.92768916e-01
6.66920245e-01 -4.00811374e-01 3.67148556e-02 -3.68733168e-01
-5.60458183e-01 -1.52673170e-01 -2.90837377e-01 2.75637567e-01
7.06393838e-01 -1.30659318e+00 -6.31157696e-01 7.03698874e-01
-4.91158009e-01 -6.40864670e-02 6.29644275e-01 1.05364215e+00
-2.43131384e-01 6.43136442e-01 -8.21421370e-02 -4.78090614e-01
-1.44997358e+00 4.56400841e-01 7.79313087e-01 2.30716228e-01
-5.46250641e-01 1.16205013e+00 3.25659454e-01 -6.74976632e-02
6.38446331e-01 -5.71475029e-01 -1.97889492e-01 9.87916812e-02
8.42822194e-01 4.52455908e-01 3.97572875e-01 -7.67904282e-01
-5.27404070e-01 5.91067016e-01 1.19725130e-01 -4.62558955e-01
1.42337501e+00 -2.70037800e-01 1.87122077e-02 4.17697698e-01
1.30301344e+00 5.06184042e-01 -1.03462684e+00 -9.93453786e-02
-4.06222969e-01 -2.82650799e-01 5.39231956e-01 -6.16718113e-01
-1.16883147e+00 1.14923358e+00 5.55314004e-01 5.18420458e-01
1.61948538e+00 -5.25853634e-01 1.23600519e+00 -2.69729376e-01
-2.65786529e-01 -9.45552766e-01 -1.59717798e-02 5.45385897e-01
8.70709598e-01 -5.98262250e-01 -3.82118076e-01 -4.16918010e-01
-3.29230905e-01 1.24207294e+00 3.63744706e-01 1.48751974e-01
7.63663173e-01 7.00730264e-01 5.02965510e-01 2.12088645e-01
-5.67241967e-01 -2.04138085e-01 5.43267429e-01 6.91188753e-01
5.86226821e-01 -3.79296504e-02 1.22465812e-01 1.00582981e+00
-4.56875354e-01 -3.71186763e-01 1.43471912e-01 6.06192470e-01
-3.42789441e-01 -1.07704222e+00 -8.44833791e-01 -9.64420289e-02
-5.17831266e-01 -1.68105483e-01 -3.09361100e-01 4.89128157e-02
9.79869366e-02 1.54794466e+00 2.20104188e-01 -5.29464245e-01
4.90811914e-01 2.95239896e-01 1.91828907e-01 -4.36291337e-01
-8.65227759e-01 8.39230180e-01 -2.30172779e-02 -1.31265298e-01
-1.94060162e-01 -3.18839967e-01 -1.52320826e+00 -5.69108725e-02
-6.46247268e-01 3.12446296e-01 8.01776350e-01 7.86906779e-01
4.25657809e-01 1.09324121e+00 8.28221202e-01 -1.00251257e+00
-4.79337037e-01 -9.35269833e-01 -7.33098507e-01 -8.49738196e-02
1.11341262e+00 -2.97987163e-01 -5.96158981e-01 8.40346888e-02]
|
[14.92463493347168, 5.915769100189209]
|
6334ce61-26d6-44b3-8a1c-4214c1e3bb8f
|
personalized-federated-learning-with-2
|
2108.01903
| null |
https://arxiv.org/abs/2108.01903v3
|
https://arxiv.org/pdf/2108.01903v3.pdf
|
Personalized Federated Learning with Clustering: Non-IID Heart Rate Variability Data Application
|
While machine learning techniques are being applied to various fields for their exceptional ability to find complex relations in large datasets, the strengthening of regulations on data ownership and privacy is causing increasing difficulty in its application to medical data. In light of this, Federated Learning has recently been proposed as a solution to train on private data without breach of confidentiality. This conservation of privacy is particularly appealing in the field of healthcare, where patient data is highly confidential. However, many studies have shown that its assumption of Independent and Identically Distributed data is unrealistic for medical data. In this paper, we propose Personalized Federated Cluster Models, a hierarchical clustering-based FL process, to predict Major Depressive Disorder severity from Heart Rate Variability. By allowing clients to receive more personalized model, we address problems caused by non-IID data, showing an accuracy increase in severity prediction. This increase in performance may be sufficient to use Personalized Federated Cluster Models in many existing Federated Learning scenarios.
|
['Tai-Myoung Chung', 'Hong Jin Jeon', 'Han Young Yu', 'Ah Young Kim', 'Eun-Hye Jang', 'Hyejun Jeong', 'Ha Min Son', 'Joo Hun Yoo']
|
2021-08-04
| null | null | null | null |
['severity-prediction', 'heart-rate-variability']
|
['computer-vision', 'medical']
|
[-6.74133375e-02 5.49177349e-01 -3.66554469e-01 -7.90540457e-01
-4.10124242e-01 -3.77859235e-01 -8.08856916e-03 3.02529365e-01
-2.83413529e-01 9.02494967e-01 1.98841080e-01 -4.27389830e-01
-5.95446110e-01 -8.46033752e-01 -3.76843810e-01 -7.04105675e-01
-2.72526026e-01 4.19964254e-01 -5.30678570e-01 2.73089826e-01
-2.14819402e-01 4.56117898e-01 -1.37346554e+00 6.32249355e-01
9.06789184e-01 8.50004673e-01 -3.70450139e-01 2.75051743e-01
-1.12742290e-01 1.02673542e+00 -3.23666781e-01 -8.71589065e-01
4.70930189e-01 -4.04841304e-01 -8.02818656e-01 -4.92921025e-01
-9.30080488e-02 -3.21893215e-01 -5.27347811e-02 8.00105989e-01
7.28400171e-01 -3.18556935e-01 3.19642365e-01 -1.57685697e+00
-4.15149629e-01 8.12070012e-01 -2.58178353e-01 -3.16540420e-01
3.04615378e-01 -1.95208713e-01 8.49136651e-01 -2.39810999e-02
5.44185340e-01 7.23897755e-01 8.25805128e-01 6.63147867e-01
-1.11500347e+00 -7.80551791e-01 -2.51093745e-01 1.04213201e-01
-1.29496992e+00 -1.17164694e-01 5.58276176e-01 -2.65193790e-01
5.92192709e-01 8.23314667e-01 5.41028976e-01 9.66140330e-01
2.86708593e-01 6.58830106e-01 1.08638954e+00 -3.25058132e-01
4.58041877e-01 6.45639598e-01 5.06290756e-02 2.32981831e-01
5.89959681e-01 1.66340232e-01 -4.56181943e-01 -7.42292047e-01
1.67779550e-01 5.81988633e-01 -1.74649060e-01 -6.16229355e-01
-7.02856064e-01 9.45873916e-01 1.80858105e-01 2.79477119e-01
-4.32686716e-01 -3.50533158e-01 5.78993738e-01 4.01143253e-01
5.46199799e-01 1.67303517e-01 -8.46503079e-01 3.76149900e-02
-8.43415678e-01 1.86961770e-01 1.15407884e+00 7.68308043e-01
5.26146233e-01 -4.98388886e-01 -5.84786832e-02 4.42920417e-01
1.01488061e-01 1.16530433e-01 6.73631489e-01 -1.12803400e+00
2.49550968e-01 1.00021839e+00 -1.68113336e-01 -1.06102169e+00
-5.00474095e-01 -3.55877250e-01 -1.23129940e+00 1.55807482e-02
3.11272651e-01 -4.31903332e-01 -2.33046025e-01 1.71975958e+00
5.68051100e-01 -8.10029730e-02 3.22281122e-01 7.15846717e-01
2.74042308e-01 5.42998239e-02 2.55467951e-01 -4.01768982e-01
1.17218757e+00 -2.59418637e-01 -9.42764223e-01 4.29481000e-01
1.06182110e+00 -2.84947097e-01 4.31391120e-01 6.48884535e-01
-7.25074947e-01 4.42098379e-02 -4.53217745e-01 2.18079016e-01
-4.53252614e-01 -4.11954820e-01 1.17102194e+00 1.31980526e+00
-7.84659743e-01 8.76486599e-01 -7.97297597e-01 -4.44457382e-01
9.74344671e-01 6.43198252e-01 -6.81472063e-01 -2.70224869e-01
-1.10322094e+00 6.58121109e-01 1.83469698e-01 -2.46687949e-01
-8.23616087e-02 -9.83033478e-01 -5.58231950e-01 3.42989117e-01
8.94642156e-03 -8.64429057e-01 7.06065118e-01 -9.29138601e-01
-8.33074391e-01 9.69195247e-01 3.66971582e-01 -7.55085289e-01
7.35414803e-01 3.99312004e-02 -6.10473812e-01 4.99181589e-03
-1.09506480e-01 1.41885653e-01 5.73596239e-01 -9.53783393e-01
-7.28670180e-01 -9.57703114e-01 -4.66927648e-01 -1.28242925e-01
-7.80408323e-01 -1.33341938e-01 2.70155013e-01 -3.79033059e-01
-9.12067667e-02 -6.69522762e-01 -5.88546753e-01 1.34563789e-01
-2.45066404e-01 2.34089792e-02 9.52681303e-01 -6.25308514e-01
1.33626747e+00 -2.19300795e+00 -4.42606717e-01 4.27416325e-01
3.70341390e-01 2.39332706e-01 2.91428119e-01 4.53483939e-01
-1.12017788e-01 3.58967125e-01 -2.77469248e-01 -2.92155832e-01
-4.04583812e-02 3.66512865e-01 -4.06633690e-02 5.59157610e-01
-2.88238913e-01 7.69889414e-01 -4.15115207e-01 -6.88000321e-01
-3.22438069e-02 5.83103776e-01 -8.02604020e-01 3.12792569e-01
1.71956584e-01 5.28861880e-01 -6.17850482e-01 5.37306726e-01
9.03959215e-01 -3.74022722e-01 7.57668614e-01 1.38856843e-01
8.21993351e-02 -2.44370654e-01 -1.13253450e+00 1.55484188e+00
-5.56777939e-02 5.89365587e-02 3.17340195e-01 -1.17395461e+00
8.24595094e-01 8.65254462e-01 1.23484623e+00 -5.89993358e-01
1.50688857e-01 1.36542395e-02 -1.28687337e-01 -7.22865105e-01
1.58833712e-03 -3.61531734e-01 2.19360590e-02 6.34788573e-01
-2.90542006e-01 5.01108110e-01 -5.37488878e-01 3.21284443e-01
1.22424448e+00 -1.79320797e-01 3.23414356e-01 -1.69412166e-01
2.85491019e-01 -2.91365683e-02 8.66346598e-01 5.08819878e-01
-3.58033657e-01 5.66575944e-01 4.32129741e-01 -7.24523962e-01
-7.27296174e-01 -7.51521766e-01 -3.79388362e-01 5.76219738e-01
-3.51570815e-01 -4.37914371e-01 -6.69237673e-01 -7.43189037e-01
5.17844081e-01 5.15902877e-01 -5.74057996e-01 -3.01870883e-01
-2.34857127e-01 -8.88749182e-01 7.07173467e-01 3.08882743e-01
3.09467494e-01 -8.33522081e-01 -9.70558941e-01 1.89418137e-01
-1.67196870e-01 -6.66253746e-01 8.44212249e-02 3.40428442e-01
-9.97824788e-01 -1.38993847e+00 -6.33012891e-01 -4.51788187e-01
4.74317968e-01 -1.01727448e-01 9.18052733e-01 1.24340631e-01
-6.26118660e-01 5.80327809e-01 -3.79814148e-01 -6.85597181e-01
-4.78682905e-01 9.52997133e-02 5.38355187e-02 3.75367641e-01
7.43612647e-01 -6.75684094e-01 -9.00377631e-01 7.29701295e-02
-1.04544353e+00 -3.44050556e-01 2.51834244e-01 6.25424564e-01
1.74073175e-01 1.32112369e-01 8.91969025e-01 -1.54679906e+00
5.00886500e-01 -9.08752620e-01 -1.74370214e-01 3.73441875e-01
-1.29805899e+00 -2.05139294e-01 6.44067466e-01 -1.34216562e-01
-1.09868312e+00 3.70494872e-01 1.05783805e-01 -4.29257959e-01
-5.24785995e-01 4.42552388e-01 -1.87997594e-01 1.26336589e-01
6.02788508e-01 -1.71789140e-01 6.77466035e-01 -6.53113484e-01
1.71282828e-01 1.06594217e+00 1.56865686e-01 -3.22164327e-01
3.53868872e-01 6.38544738e-01 1.16788499e-01 -3.42053980e-01
-4.24553186e-01 -6.14166737e-01 -4.59021270e-01 1.84583500e-01
8.69129062e-01 -8.57739091e-01 -1.24249005e+00 6.29263446e-02
-6.75089002e-01 9.22249332e-02 -4.86756086e-01 6.62581861e-01
-5.12129486e-01 4.01613295e-01 -4.48845148e-01 -9.61277783e-01
-6.34110928e-01 -5.32229781e-01 3.50911498e-01 -6.94486499e-02
-3.90355825e-01 -1.08609915e+00 2.95950890e-01 5.80733836e-01
6.05383694e-01 6.31338477e-01 1.01890230e+00 -1.11464310e+00
-3.07051361e-01 -5.82267702e-01 1.54044420e-01 2.18102112e-01
5.46619415e-01 -3.16207677e-01 -1.15527987e+00 -4.21175867e-01
3.11129957e-01 -2.82728016e-01 2.75120914e-01 2.88302481e-01
1.26469541e+00 -7.03181267e-01 -6.22373998e-01 6.19413435e-01
1.40674937e+00 1.20807160e-02 5.60880899e-01 8.27030241e-02
3.02948117e-01 1.05909002e+00 4.25154924e-01 1.02900565e+00
3.98908436e-01 3.15488756e-01 5.64446628e-01 -2.12693527e-01
5.15164614e-01 -5.19732274e-02 -1.76265657e-01 2.79543817e-01
1.00878455e-01 5.49042746e-02 -8.49727333e-01 3.59806806e-01
-1.97707415e+00 -1.11080420e+00 -3.49945456e-01 2.37478375e+00
7.81712294e-01 -5.06409943e-01 2.53575653e-01 7.21806437e-02
5.51459908e-01 -4.38019007e-01 -5.52833200e-01 -5.74689209e-01
-1.87817991e-01 1.20466314e-01 5.25476813e-01 -2.56494731e-01
-8.10319483e-01 2.90387213e-01 6.08488989e+00 2.68067926e-01
-1.02991021e+00 1.94135129e-01 9.48976815e-01 -1.33732945e-01
-3.43983650e-01 -1.27283290e-01 -2.41776302e-01 5.02165318e-01
1.03605080e+00 -3.12711924e-01 1.85301468e-01 1.00652635e+00
1.68581337e-01 8.81211087e-02 -1.07234240e+00 1.04791307e+00
-2.11414084e-01 -1.29585707e+00 -2.47556657e-01 4.32923168e-01
6.39181197e-01 -2.60085106e-01 1.12594210e-01 8.91811624e-02
2.96799600e-01 -1.22077572e+00 -8.53062794e-02 6.26252413e-01
6.70686841e-01 -9.35917497e-01 8.79534960e-01 6.62633061e-01
-5.69382191e-01 -4.86592203e-01 -2.99361050e-01 -6.54089749e-02
-5.62773198e-02 7.21332133e-01 -8.10190022e-01 7.74194896e-01
1.06317270e+00 3.96638066e-01 -3.16479802e-01 9.61711347e-01
5.67980170e-01 6.39265776e-01 -2.75384784e-01 4.08379078e-01
-3.51556659e-01 -2.24869028e-01 -1.70949921e-01 9.03140545e-01
3.70104700e-01 3.47712398e-01 -1.95752233e-01 5.43886542e-01
1.06530689e-01 5.37171185e-01 -1.02965117e+00 2.02144206e-01
1.85547113e-01 1.27514946e+00 -1.86230123e-01 -4.20790203e-02
-5.98016739e-01 9.46883976e-01 1.39741495e-01 -2.39024628e-02
-5.87325096e-01 -3.29523832e-02 7.94750154e-01 2.82700390e-01
2.05680802e-02 4.27572817e-01 -5.04941463e-01 -1.00022364e+00
-2.02249244e-01 -1.06421101e+00 1.14655232e+00 -7.27673471e-02
-1.54590702e+00 3.93897742e-01 -2.44439870e-01 -1.25649333e+00
-3.71377259e-01 -2.49253124e-01 -3.80133837e-01 7.83103824e-01
-1.25110006e+00 -1.29969358e+00 -1.80390831e-02 1.18666863e+00
-4.26795036e-01 -2.39194766e-01 1.32280922e+00 5.98505497e-01
-4.15864021e-01 8.34284782e-01 2.94654101e-01 9.74436179e-02
1.01147985e+00 -8.33830595e-01 -3.31316829e-01 2.32923970e-01
-7.71214217e-02 7.48940408e-01 3.15940380e-01 -8.17211270e-01
-1.41054869e+00 -1.15927041e+00 1.08493781e+00 -5.35393536e-01
-2.08346508e-02 -2.53388226e-01 -1.09035230e+00 6.97990537e-01
3.81869823e-02 3.08266878e-01 1.51099551e+00 2.65098065e-01
-3.79954994e-01 -4.05647874e-01 -1.89402211e+00 9.12875906e-02
5.96130610e-01 -2.14704901e-01 -8.36454630e-02 3.89341474e-01
5.48776627e-01 1.15417264e-01 -1.36510181e+00 3.36832881e-01
5.42283237e-01 -1.22882164e+00 6.94844484e-01 -1.05315638e+00
8.51027369e-02 1.49315044e-01 -2.64897078e-01 -8.20381761e-01
-2.71755189e-01 -8.48014414e-01 -1.15129434e-01 1.32279980e+00
3.32414806e-01 -8.48183036e-01 1.41406822e+00 1.58697069e+00
4.32774723e-01 -6.02367580e-01 -1.10340607e+00 -5.25474370e-01
2.31750056e-01 -3.21804345e-01 9.03445125e-01 1.57594323e+00
5.88679790e-01 -2.38601193e-01 -6.37308836e-01 2.33634830e-01
7.97021568e-01 2.15942025e-01 6.74920619e-01 -1.54071593e+00
-2.66728610e-01 1.91473722e-01 -4.57345277e-01 8.39395449e-02
2.80908978e-04 -1.04763222e+00 -7.54166663e-01 -1.21070492e+00
3.73268634e-01 -9.59331930e-01 -4.70116615e-01 7.42872298e-01
8.41535162e-03 -3.39278243e-02 1.15940608e-01 1.53122410e-01
-3.72233808e-01 1.95836395e-01 7.87941992e-01 1.14283763e-01
-2.14139506e-01 4.72713888e-01 -9.15027976e-01 2.62164265e-01
1.10249686e+00 -7.56729782e-01 -3.83316189e-01 5.64231426e-02
1.97483316e-01 3.37899864e-01 1.73539981e-01 -8.52791011e-01
4.18160290e-01 -1.32684544e-01 4.77156341e-01 -3.11113924e-01
-4.57396265e-03 -1.55629766e+00 7.50137508e-01 7.14730084e-01
-3.56684834e-01 -1.07069775e-01 -1.13803148e-01 5.84947407e-01
-1.27397120e-01 7.87478909e-02 5.18959045e-01 -2.67759293e-01
-7.82445222e-02 4.62275773e-01 -1.61781847e-01 -2.33171999e-01
1.37699676e+00 -2.11022094e-01 6.43562153e-03 -3.97476941e-01
-8.87435853e-01 3.11406344e-01 5.44344068e-01 2.27665752e-01
4.71628606e-01 -9.47848797e-01 -5.55831015e-01 3.50901544e-01
-2.92545781e-02 -1.26347035e-01 6.77560210e-01 8.00418854e-01
-1.73094317e-01 5.11562049e-01 -1.88480198e-01 -3.48558128e-01
-1.49890244e+00 1.16626143e+00 1.08286589e-01 -2.34032899e-01
-6.92002952e-01 2.58831590e-01 -2.17125922e-01 -6.99582100e-01
4.49582279e-01 1.93186790e-01 3.56238373e-02 1.39425620e-01
4.77771908e-01 5.18536270e-01 3.17827940e-01 -2.17294827e-01
-4.19192642e-01 -4.95156236e-02 -3.65702994e-02 3.07963252e-01
1.58388805e+00 -1.31553113e-01 -4.11067992e-01 2.90049762e-02
1.34013665e+00 -2.47873738e-02 -8.34317565e-01 8.50611106e-02
1.08998947e-01 -3.93230230e-01 -3.27968180e-01 -8.62636983e-01
-1.27626467e+00 5.52972734e-01 9.67668056e-01 2.35544100e-01
1.27637982e+00 -1.77768305e-01 4.66480583e-01 1.27487347e-01
5.98526478e-01 -8.31645429e-01 -6.30761087e-01 -1.96026474e-01
1.84373349e-01 -1.36899614e+00 1.95152581e-01 -2.12664843e-01
-8.71409535e-01 7.80081153e-01 3.50496441e-01 3.45991135e-01
1.00182962e+00 4.48751897e-01 4.51665640e-01 -8.77497941e-02
-7.28482664e-01 3.75335246e-01 -2.93043226e-01 9.67308104e-01
4.22231257e-01 2.32105330e-01 -4.48907971e-01 9.02981758e-01
-9.21473503e-02 4.25306648e-01 2.63125986e-01 1.04271650e+00
1.92825034e-01 -1.59270608e+00 -4.22489256e-01 7.73611188e-01
-1.01562929e+00 1.44005448e-01 -3.06322068e-01 5.60051501e-01
2.19507009e-01 9.45264339e-01 -4.78718895e-03 -8.57335702e-02
8.65594968e-02 4.36612219e-01 5.27250543e-02 -3.96324843e-01
-1.10873210e+00 -2.69516915e-01 -2.64611304e-01 -6.80490911e-01
-3.96946162e-01 -6.89966917e-01 -1.25405908e+00 -4.96222496e-01
-8.64619315e-02 4.32257622e-01 7.36537933e-01 5.37292898e-01
9.11162198e-01 3.65018211e-02 7.86773860e-01 7.09162876e-02
-8.18386436e-01 -3.93590331e-01 -9.95911777e-01 8.33317757e-01
1.95710003e-01 8.06367695e-02 -2.73513526e-01 -7.96231814e-03]
|
[6.152388095855713, 6.4823713302612305]
|
545fa4ae-314c-489c-959d-eb5b341bec9a
|
bayesian-neural-network-inference-via
|
2209.02188
| null |
https://arxiv.org/abs/2209.02188v1
|
https://arxiv.org/pdf/2209.02188v1.pdf
|
Bayesian Neural Network Inference via Implicit Models and the Posterior Predictive Distribution
|
We propose a novel approach to perform approximate Bayesian inference in complex models such as Bayesian neural networks. The approach is more scalable to large data than Markov Chain Monte Carlo, it embraces more expressive models than Variational Inference, and it does not rely on adversarial training (or density ratio estimation). We adopt the recent approach of constructing two models: (1) a primary model, tasked with performing regression or classification; and (2) a secondary, expressive (e.g. implicit) model that defines an approximate posterior distribution over the parameters of the primary model. However, we optimise the parameters of the posterior model via gradient descent according to a Monte Carlo estimate of the posterior predictive distribution -- which is our only approximation (other than the posterior model). Only a likelihood needs to be specified, which can take various forms such as loss functions and synthetic likelihoods, thus providing a form of a likelihood-free approach. Furthermore, we formulate the approach such that the posterior samples can either be independent of, or conditionally dependent upon the inputs to the primary model. The latter approach is shown to be capable of increasing the apparent complexity of the primary model. We see this being useful in applications such as surrogate and physics-based models. To promote how the Bayesian paradigm offers more than just uncertainty quantification, we demonstrate: uncertainty quantification, multi-modality, as well as an application with a recent deep forecasting neural network architecture.
|
['Daniel Edward Pagendam', 'Joel Janek Dabrowski']
|
2022-09-06
| null | null | null | null |
['density-ratio-estimation']
|
['methodology']
|
[ 1.72958493e-01 5.73822141e-01 7.47594088e-02 -4.39615369e-01
-1.01627111e+00 -3.88877451e-01 1.04632676e+00 -2.55046725e-01
-4.87506270e-01 1.21740556e+00 -3.62861827e-02 -5.19064963e-01
-2.35318825e-01 -8.52500498e-01 -1.11708629e+00 -1.05488980e+00
8.06519166e-02 9.43566322e-01 2.10805818e-01 1.48033485e-01
1.75639585e-01 6.67990506e-01 -1.27200258e+00 -2.36692801e-01
6.76886618e-01 1.21230161e+00 -2.76741683e-01 5.21735609e-01
-1.32982656e-02 6.24595761e-01 -6.03708863e-01 -5.57589233e-01
-8.30337629e-02 -1.41564608e-01 -4.25275177e-01 -5.29584289e-01
1.19117722e-01 -6.60782278e-01 -1.93137050e-01 1.15797853e+00
2.80688643e-01 2.65652627e-01 1.29865336e+00 -1.15047956e+00
-1.32253587e-01 5.41378260e-01 -8.53311792e-02 -1.27153262e-01
8.39760825e-02 1.68267041e-01 6.42847836e-01 -6.20789766e-01
4.34349984e-01 1.33500040e+00 9.36564326e-01 5.30878663e-01
-1.66077542e+00 -4.70380425e-01 3.82054388e-03 -1.93017930e-01
-1.47509956e+00 -5.10982156e-01 5.96796691e-01 -5.85319519e-01
6.12163544e-01 9.01225433e-02 6.51421547e-01 1.48861516e+00
4.24421936e-01 4.99609292e-01 1.35065198e+00 -5.69524646e-01
8.72083187e-01 3.32914650e-01 4.18583006e-02 5.75150669e-01
8.84061381e-02 6.80679977e-01 -4.06851113e-01 -6.75804138e-01
7.13785350e-01 -3.09720248e-01 -3.14273596e-01 -3.46395731e-01
-7.29871094e-01 1.06094754e+00 8.04227516e-02 -1.17737450e-01
-4.19729471e-01 8.27460408e-01 2.12748066e-01 -1.34685501e-01
5.14677763e-01 1.26745969e-01 -4.44143534e-01 -4.26218193e-03
-1.38982463e+00 6.67409062e-01 1.16252232e+00 6.54472172e-01
7.10298717e-01 1.82690263e-01 -1.27769977e-01 4.25106227e-01
9.70871866e-01 6.61318898e-01 2.16730479e-02 -1.23449647e+00
8.45557377e-02 -5.28984249e-01 5.62086403e-01 -3.70356739e-01
-3.74819823e-02 -3.03301454e-01 -7.32697725e-01 7.45517790e-01
6.94241226e-01 -2.54293054e-01 -9.83836532e-01 2.06003904e+00
3.00757736e-01 3.63361150e-01 -5.15251607e-02 5.50700128e-01
4.14435536e-01 7.19098568e-01 2.48598725e-01 -3.63327086e-01
1.05509531e+00 -4.10629153e-01 -7.48455405e-01 3.11428197e-02
8.22565183e-02 -4.23688531e-01 6.02154911e-01 6.58732474e-01
-1.31372428e+00 -1.58606291e-01 -1.29764962e+00 2.93300420e-01
-3.69387031e-01 -2.74656773e-01 4.67235208e-01 9.54508007e-01
-1.09253263e+00 1.02966058e+00 -1.08831406e+00 2.62944549e-01
2.76382506e-01 2.83677727e-01 -5.78944869e-02 1.62411466e-01
-1.47411048e+00 1.24245679e+00 4.85144585e-01 7.86596984e-02
-1.19514728e+00 -8.17356527e-01 -7.17994988e-01 9.80073884e-02
-2.83477157e-02 -9.79404330e-01 1.32207096e+00 -7.29622602e-01
-1.80580807e+00 4.45364237e-01 -9.61482748e-02 -5.64906657e-01
9.07001853e-01 -6.08673878e-02 -1.25981495e-01 1.22241959e-01
-3.50585818e-01 7.65652955e-01 1.25764549e+00 -1.27857125e+00
-8.83660614e-02 -1.38754979e-01 2.13240888e-02 -2.37564728e-01
3.48187357e-01 -1.47436991e-01 -1.72539949e-01 -5.45520723e-01
1.02329150e-01 -9.02364016e-01 -3.63160402e-01 4.19648021e-01
-6.91537976e-01 -3.33114807e-03 4.14341450e-01 -7.54418492e-01
7.90585041e-01 -1.84704888e+00 3.22150998e-02 5.75145125e-01
3.10546923e-02 7.49167353e-02 3.62365991e-01 1.87427506e-01
4.22068276e-02 2.24829420e-01 -7.48614669e-01 -4.89008933e-01
4.29850340e-01 2.73565561e-01 -6.25455201e-01 5.68364322e-01
3.80021870e-01 7.41415679e-01 -6.43422067e-01 -4.63305742e-01
2.97584683e-01 9.34244335e-01 -4.38033342e-01 1.21004999e-01
-5.74125826e-01 4.77325022e-01 -5.97042918e-01 1.24196418e-01
8.42472613e-01 -2.47064605e-02 -2.77971905e-02 -3.13575774e-01
2.19703197e-01 1.76910758e-01 -1.43953633e+00 1.33274686e+00
-5.62461019e-01 4.00581241e-01 1.00673325e-01 -9.04964685e-01
7.08554566e-01 4.28433955e-01 7.20751211e-02 1.27688870e-01
1.26175135e-01 2.52239913e-01 -3.68147552e-01 5.61611401e-03
2.45992720e-01 -7.65098274e-01 1.04170717e-01 4.10130531e-01
3.35803688e-01 -4.61363584e-01 -6.23519607e-02 1.61915794e-01
7.35832393e-01 7.48923123e-01 2.71612197e-01 -3.35978180e-01
1.50719449e-01 -3.39816481e-01 1.89936161e-01 1.15526772e+00
-8.72849077e-02 6.44591868e-01 6.20416105e-01 -6.48995861e-02
-1.10486448e+00 -1.60674942e+00 -7.80813932e-01 3.37755114e-01
-3.06282759e-01 2.02400684e-02 -8.47494602e-01 -4.03787464e-01
6.76494539e-02 1.27804852e+00 -6.07127368e-01 -1.96171269e-01
-1.96970925e-01 -1.09049153e+00 6.83908761e-01 4.75529343e-01
1.97056860e-01 -6.21997952e-01 -3.18195283e-01 3.56040657e-01
6.27521649e-02 -8.97628605e-01 5.20295762e-02 4.24212575e-01
-1.04870474e+00 -7.16758728e-01 -6.78752005e-01 1.90833837e-01
3.62596989e-01 -1.07364547e+00 1.19744670e+00 -5.00541270e-01
2.55441725e-01 4.35552627e-01 3.05785805e-01 -3.88554305e-01
-9.07459617e-01 -4.51471657e-01 1.78553522e-01 -6.47705942e-02
1.27544254e-01 -9.67281044e-01 -4.34876800e-01 -4.05128375e-02
-9.07494962e-01 -2.88987190e-01 3.82895797e-01 7.76871026e-01
6.02054358e-01 -1.11391671e-01 4.68629301e-01 -9.67748523e-01
5.27114928e-01 -4.68002647e-01 -8.39981616e-01 1.74526736e-01
-7.64911175e-01 3.94536912e-01 5.56231618e-01 -5.93905151e-01
-1.36703777e+00 -7.53389224e-02 -5.36888003e-01 -5.13637125e-01
-2.79522926e-01 5.34546137e-01 -1.81937248e-01 -7.95651749e-02
8.05069149e-01 -4.90121022e-02 -8.04156624e-03 -5.02371490e-01
5.08024871e-01 4.23710078e-01 6.51689768e-01 -9.83237207e-01
7.57695973e-01 5.42072594e-01 4.89827186e-01 -5.29256344e-01
-6.91253185e-01 2.92732507e-01 -4.17451233e-01 -1.95961788e-01
7.43138492e-01 -7.80245483e-01 -6.76564753e-01 4.28546667e-01
-1.26664424e+00 -2.34328091e-01 -6.71380281e-01 7.66745567e-01
-8.59333396e-01 2.11041376e-01 -6.82534397e-01 -1.32002497e+00
-2.76222881e-02 -1.17952716e+00 9.67028737e-01 1.90144569e-01
-2.21312582e-01 -1.30444717e+00 9.85841975e-02 -1.36667982e-01
5.08677244e-01 4.75255102e-01 1.00904000e+00 -6.61722124e-01
-7.06353366e-01 -5.12648642e-01 -1.92623939e-02 4.87000257e-01
-6.34658217e-01 3.95136118e-01 -1.45007658e+00 -9.91468877e-02
2.66260028e-01 -3.91389638e-01 1.00437951e+00 8.13190162e-01
1.00393629e+00 -4.08280611e-01 -3.51415664e-01 5.23000002e-01
1.46186399e+00 -9.95107964e-02 8.34188998e-01 -1.33600682e-01
3.50721776e-01 4.55491811e-01 1.38121575e-01 3.55314165e-01
-1.08655185e-01 6.17301226e-01 4.83331174e-01 4.23452675e-01
8.44824538e-02 -4.68045145e-01 3.88720870e-01 4.65034246e-01
-1.59325525e-02 -1.46129936e-01 -6.93926632e-01 2.51577795e-02
-1.63766634e+00 -8.48085344e-01 -7.23011047e-02 2.25495839e+00
1.07371521e+00 4.76026744e-01 -1.90315530e-01 -5.07673919e-02
4.79350120e-01 9.00448188e-02 -6.61605775e-01 -4.65867311e-01
2.59935379e-01 4.83883560e-01 3.27872038e-01 7.98829079e-01
-1.11305833e+00 3.80037665e-01 7.50718212e+00 1.37238193e+00
-7.28984892e-01 2.87438005e-01 7.13443875e-01 3.48827243e-02
-6.70682430e-01 2.03978866e-01 -8.57350230e-01 7.33304322e-01
1.47010422e+00 2.92893350e-01 3.94014865e-01 7.05831110e-01
3.99795920e-02 -4.72272724e-01 -1.28478408e+00 4.12227154e-01
-3.97850513e-01 -1.43822503e+00 6.63843304e-02 1.42828465e-01
5.12409508e-01 -1.86416022e-02 -4.91293594e-02 3.39767575e-01
5.22330999e-01 -1.16608083e+00 1.13215411e+00 1.26607513e+00
9.68548179e-01 -7.29113400e-01 6.61461532e-01 6.84739232e-01
-5.48055351e-01 4.04423237e-01 -2.18311876e-01 1.87356561e-01
4.28918391e-01 1.02855337e+00 -2.21618414e-01 3.37051749e-01
4.56571907e-01 2.57055424e-02 3.06015220e-02 8.34927142e-01
-4.23177868e-01 8.14166963e-01 -8.46721590e-01 -6.35848492e-02
3.18763815e-02 -4.16141748e-01 9.06005859e-01 1.02061510e+00
1.96699470e-01 -2.66256720e-01 -1.86387405e-01 1.54382765e+00
2.73951083e-01 -4.74610597e-01 -4.74212766e-01 1.59588218e-01
6.61302388e-01 9.52173173e-01 -4.63110834e-01 -2.35728338e-01
-1.08243357e-02 3.44577342e-01 1.62211567e-01 7.22692549e-01
-9.31365013e-01 -2.01007828e-01 2.94952899e-01 -1.82751417e-01
3.34796041e-01 -8.43844190e-03 -1.62918746e-01 -8.78481865e-01
-6.38844669e-02 -5.37601888e-01 1.05908122e-02 -1.00740731e+00
-1.35111749e+00 3.37349236e-01 6.48434043e-01 -7.27324545e-01
-9.92177546e-01 -8.76532912e-01 -6.44131482e-01 1.47314417e+00
-1.31543112e+00 -1.04257476e+00 3.00679177e-01 2.45965138e-01
-2.81817347e-01 4.01256802e-05 9.87156093e-01 -2.83532720e-02
-3.13380003e-01 3.96800220e-01 5.07550538e-01 -2.87173539e-01
2.18110085e-01 -1.23352015e+00 1.57084435e-01 6.64600313e-01
-1.39076501e-01 6.64374650e-01 1.29838121e+00 -5.26770175e-01
-8.71143520e-01 -7.45686948e-01 5.56634068e-01 -6.13890171e-01
7.92918384e-01 -1.89239264e-01 -8.26009512e-01 6.51934862e-01
-2.46798635e-01 -1.73596013e-02 4.66329604e-01 9.45586562e-02
-2.34873518e-01 1.67438313e-02 -1.57693720e+00 4.49000597e-01
3.13111424e-01 -6.08953059e-01 -5.46089053e-01 2.34183133e-01
5.19491732e-01 -4.12331551e-01 -1.03679144e+00 4.72320557e-01
7.57193267e-01 -1.08673966e+00 1.04631352e+00 -5.80694199e-01
3.87857199e-01 -2.56330609e-01 -4.22170401e-01 -1.15121317e+00
1.67423025e-01 -7.90439844e-01 -7.59469271e-01 1.24246979e+00
4.28678095e-01 -8.68895769e-01 7.47750282e-01 9.63916600e-01
2.05269153e-03 -9.81065392e-01 -1.43111324e+00 -7.07877636e-01
5.29571295e-01 -7.94596195e-01 5.00724375e-01 4.48695898e-01
-5.81330836e-01 -5.20861000e-02 -1.17844842e-01 2.26649493e-01
1.04372311e+00 -3.06304187e-01 3.16981286e-01 -1.28192770e+00
-6.86827004e-01 -3.90844762e-01 -1.87385380e-01 -1.06339049e+00
3.07521105e-01 -5.17671168e-01 2.25303411e-01 -1.15680671e+00
2.13920623e-02 -5.82783699e-01 -1.58883393e-01 3.32442159e-03
1.60632089e-01 6.85475320e-02 -2.98044920e-01 2.08205506e-01
5.30333072e-02 7.66274035e-01 8.65130663e-01 8.63107387e-03
3.04552406e-01 4.01236892e-01 -2.92683274e-01 1.03362727e+00
5.24523675e-01 -7.06661463e-01 -3.53417724e-01 2.90201008e-01
3.56837511e-01 4.31905806e-01 1.02774274e+00 -8.93045306e-01
9.51703712e-02 -5.44935465e-02 5.35212159e-01 -5.23029268e-01
7.10832298e-01 -6.15517557e-01 5.12612224e-01 3.31813574e-01
-2.88044363e-01 -6.05378330e-01 1.55830488e-01 8.80342245e-01
2.50660833e-02 -1.05189466e+00 9.41152155e-01 -1.28054470e-01
-1.11327477e-01 1.20070316e-01 -4.54828590e-01 8.25400203e-02
4.66209650e-01 -3.16744260e-02 -2.10523903e-01 -4.91727471e-01
-1.10520387e+00 -1.37806430e-01 2.88069934e-01 -4.70138818e-01
3.51622254e-01 -1.26132286e+00 -5.55587530e-01 8.45759641e-03
-3.07448030e-01 2.81882775e-03 1.12813532e-01 7.42286861e-01
-4.50130910e-01 2.84485757e-01 2.48163670e-01 -6.81310713e-01
-1.99918970e-01 2.76771396e-01 8.15886199e-01 -3.71319622e-01
-3.46672952e-01 8.08648884e-01 1.39781367e-02 -5.36488891e-01
3.52669895e-01 -2.66974926e-01 4.83003929e-02 -1.97872236e-01
3.87661159e-01 4.32560414e-01 -5.58078736e-02 -3.29133928e-01
-2.53125936e-01 2.58536607e-01 2.47745335e-01 -7.39839196e-01
1.01694596e+00 -7.42443204e-02 -4.30922098e-02 7.62539983e-01
9.58206654e-01 -2.28884742e-01 -1.82267368e+00 -2.73061134e-02
-1.95438176e-01 -7.28427246e-02 4.37988400e-01 -1.00169063e+00
-6.24826610e-01 1.07926989e+00 5.61192870e-01 1.80872753e-01
7.12268054e-01 -2.01209374e-02 3.29910278e-01 3.53139073e-01
3.45154703e-01 -9.57954288e-01 -4.25531268e-01 2.29947597e-01
8.26025724e-01 -9.09894466e-01 1.90053910e-01 -2.06916798e-02
-9.48799625e-02 1.05667102e+00 -5.61702618e-05 -1.97524369e-01
1.21987033e+00 5.43213725e-01 -2.50083864e-01 -3.76423332e-03
-6.02878749e-01 2.83277959e-01 5.54529250e-01 8.49310935e-01
4.05421853e-02 1.48751447e-02 5.45815341e-02 6.76241755e-01
-1.44149214e-01 1.21178307e-01 3.22701931e-01 6.21917427e-01
-2.59300292e-01 -8.84487391e-01 -3.57229263e-01 5.73155463e-01
-4.98819470e-01 -2.76342005e-01 3.28146368e-01 8.86329114e-01
-1.32553112e-02 7.74214208e-01 -3.87916155e-02 1.01668917e-01
4.97293309e-04 5.06046712e-01 7.51260996e-01 -1.74640805e-01
-7.84932356e-03 2.02548504e-02 2.25212336e-01 -4.91781801e-01
-3.75057697e-01 -7.50800550e-01 -8.87835324e-01 -4.26845551e-01
-5.33549905e-01 1.51097968e-01 1.01442409e+00 1.29118991e+00
-2.58170605e-01 4.09038901e-01 1.47777900e-01 -1.27542436e+00
-1.26364696e+00 -1.03900445e+00 -7.86911488e-01 -1.36979252e-01
4.22423035e-01 -8.18307459e-01 -7.79449761e-01 -1.33878261e-01]
|
[6.981794834136963, 3.834427833557129]
|
a6d6342d-3a6a-470a-af9c-f5233343689c
|
ucl-dehaze-towards-real-world-image-dehazing
|
2205.01871
| null |
https://arxiv.org/abs/2205.01871v1
|
https://arxiv.org/pdf/2205.01871v1.pdf
|
UCL-Dehaze: Towards Real-world Image Dehazing via Unsupervised Contrastive Learning
|
While the wisdom of training an image dehazing model on synthetic hazy data can alleviate the difficulty of collecting real-world hazy/clean image pairs, it brings the well-known domain shift problem. From a different yet new perspective, this paper explores contrastive learning with an adversarial training effort to leverage unpaired real-world hazy and clean images, thus bridging the gap between synthetic and real-world haze is avoided. We propose an effective unsupervised contrastive learning paradigm for image dehazing, dubbed UCL-Dehaze. Unpaired real-world clean and hazy images are easily captured, and will serve as the important positive and negative samples respectively when training our UCL-Dehaze network. To train the network more effectively, we formulate a new self-contrastive perceptual loss function, which encourages the restored images to approach the positive samples and keep away from the negative samples in the embedding space. Besides the overall network architecture of UCL-Dehaze, adversarial training is utilized to align the distributions between the positive samples and the dehazed images. Compared with recent image dehazing works, UCL-Dehaze does not require paired data during training and utilizes unpaired positive/negative data to better enhance the dehazing performance. We conduct comprehensive experiments to evaluate our UCL-Dehaze and demonstrate its superiority over the state-of-the-arts, even only 1,800 unpaired real-world images are used to train our network. Source code has been available at https://github.com/yz-wang/UCL-Dehaze.
|
['Jing Qin', 'Mingqiang Wei', 'Wenhan Yang', 'Haoran Xie', 'Fu Lee Wang', 'Xuefeng Yan', 'Yongzhen Wang']
|
2022-05-04
| null | null | null | null |
['image-dehazing']
|
['computer-vision']
|
[ 3.16418409e-01 -2.33594939e-01 3.25821966e-01 -2.25932240e-01
-7.84924686e-01 -3.12640667e-01 4.59803760e-01 -3.12107354e-01
-3.41155946e-01 8.39309573e-01 9.47845429e-02 -2.13545725e-01
4.72442098e-02 -8.68320942e-01 -8.84929478e-01 -1.22029352e+00
2.34111547e-01 -2.22910702e-01 -2.27700118e-02 -4.29928035e-01
4.00124900e-02 1.30391464e-01 -1.42735887e+00 1.75758496e-01
1.38058591e+00 9.39000428e-01 2.19339460e-01 7.01268911e-01
3.60663027e-01 9.25276041e-01 -9.37851906e-01 -3.43453109e-01
7.72423983e-01 -7.48327971e-01 -2.76442140e-01 4.82107624e-02
9.43508029e-01 -5.65804482e-01 -8.07054281e-01 1.53273082e+00
7.36950934e-01 3.25690985e-01 5.49114227e-01 -1.59746873e+00
-1.53413928e+00 -7.45637575e-03 -6.76884711e-01 3.77073646e-01
-1.61844000e-01 6.27621651e-01 4.48872507e-01 -1.13083363e+00
2.10755900e-01 9.92908359e-01 4.76942927e-01 7.10326314e-01
-9.93461370e-01 -1.11949956e+00 -1.02766648e-01 3.40672255e-01
-1.41543841e+00 -4.38781291e-01 1.02959526e+00 -2.59952813e-01
2.36892521e-01 3.08878690e-01 5.97633362e-01 1.02411091e+00
2.44113907e-01 7.05693543e-01 1.46305609e+00 -4.04647470e-01
2.47874513e-01 1.33236066e-01 -2.09806040e-01 4.20643568e-01
3.95453185e-01 4.58884150e-01 -2.76966393e-01 2.65580684e-01
6.64057493e-01 3.54126662e-01 -6.65625036e-01 -1.65305763e-01
-1.03928947e+00 7.06285834e-01 8.28899741e-01 4.93368655e-02
-3.28373820e-01 -1.71055167e-03 -1.14801802e-01 5.98195791e-01
5.53606212e-01 5.94250739e-01 1.01765722e-01 4.36573237e-01
-8.88791561e-01 2.71294177e-01 2.37411156e-01 9.25736904e-01
1.04692924e+00 4.55460340e-01 -1.01180203e-01 8.81272852e-01
6.90576900e-03 8.18717182e-01 4.61554527e-01 -9.83834922e-01
3.58431578e-01 2.14821115e-01 1.29907504e-01 -1.06175256e+00
3.35658133e-01 -4.02232677e-01 -1.25412214e+00 7.47947514e-01
2.49310240e-01 -2.11099640e-01 -1.40523529e+00 1.50534999e+00
2.11009055e-01 4.02700424e-01 2.50457495e-01 1.04730248e+00
7.34766126e-01 1.23720872e+00 -2.83189416e-01 -1.00921251e-01
8.59718382e-01 -1.38330925e+00 -9.06649113e-01 -4.45164233e-01
-3.79648991e-02 -7.51650095e-01 1.28668964e+00 4.23036277e-01
-1.21019256e+00 -6.07766569e-01 -1.44328058e+00 -2.58286029e-01
-6.20286584e-01 -3.29804063e-01 1.47207558e-01 4.42555934e-01
-1.14933908e+00 2.23377720e-01 -5.01853585e-01 2.41065815e-01
6.79636657e-01 3.11998394e-03 -4.30708587e-01 -7.08453417e-01
-1.42409778e+00 8.74906719e-01 2.98031420e-01 3.54485393e-01
-1.31631219e+00 -9.35704529e-01 -1.00159812e+00 -7.90932477e-02
1.97164223e-01 -6.19102001e-01 7.95778871e-01 -1.17141175e+00
-1.18496764e+00 7.28364706e-01 2.52406538e-01 -3.45238119e-01
4.42384452e-01 -2.37382978e-01 -8.48760545e-01 3.23231846e-01
1.95654169e-01 7.53226459e-01 1.44169223e+00 -1.84839010e+00
-4.28100258e-01 -5.03448509e-02 5.01651950e-02 1.64808735e-01
-4.39096451e-01 -4.02827829e-01 -2.88812071e-01 -1.22307181e+00
-2.54773915e-01 -6.90103948e-01 -4.30871025e-02 3.17869842e-01
-2.95456767e-01 3.60014439e-01 1.14717007e+00 -8.69903088e-01
9.79792416e-01 -2.51920795e+00 -1.30499423e-01 -1.02589438e-02
5.33892870e-01 6.41275048e-01 -3.92508030e-01 2.66705662e-01
-3.50340784e-01 -6.91681495e-03 -7.14732707e-01 -2.95857340e-01
-1.67747274e-01 1.94727078e-01 -4.71088797e-01 6.28638446e-01
4.37184095e-01 8.74693990e-01 -1.14969337e+00 -4.09805804e-01
4.13780242e-01 7.54503489e-01 -4.36929047e-01 5.29767215e-01
3.33419330e-02 5.14766574e-01 2.40927693e-02 6.06866300e-01
1.25169849e+00 -3.15453485e-02 -5.33145845e-01 -1.28637448e-01
4.99525778e-02 -2.03293949e-01 -7.58875549e-01 1.22041297e+00
-4.35812473e-01 8.34898829e-01 2.19485417e-01 -1.02464521e+00
7.43287802e-01 8.99031609e-02 -2.27642767e-02 -1.17387247e+00
8.59236624e-03 1.37389615e-01 -1.40890822e-01 -4.88063544e-01
4.88918453e-01 -5.64346433e-01 2.04222992e-01 7.91785866e-02
8.99309516e-02 -4.93338317e-01 -9.89317894e-02 1.32523119e-01
6.85341954e-01 -2.90678442e-01 1.26647782e-02 -1.86665878e-01
4.64204907e-01 -1.98856249e-01 6.14639044e-01 7.26219714e-01
-5.09334147e-01 9.81670022e-01 -6.88475445e-02 -2.19215229e-01
-1.08204567e+00 -1.41136289e+00 -1.08633369e-01 6.72364295e-01
6.05004489e-01 1.52949050e-01 -7.72729754e-01 -5.21274805e-01
-2.38783509e-01 8.20475459e-01 -8.62219095e-01 -7.06184447e-01
-5.19404650e-01 -6.98291600e-01 3.58092040e-01 -6.86626807e-02
1.10082984e+00 -8.50348413e-01 6.28492832e-02 -1.74572408e-01
-2.65646189e-01 -8.55795383e-01 -7.39458323e-01 -4.85737994e-02
-4.15227115e-01 -8.14084828e-01 -1.06624746e+00 -1.00967515e+00
8.10421884e-01 8.16140592e-01 1.01236868e+00 2.20576659e-01
-9.31219086e-02 8.13800022e-02 -4.28301096e-01 -7.74286866e-01
-5.02998769e-01 -4.58652526e-01 -5.33389524e-02 2.36328855e-01
1.10619374e-01 -8.91269743e-01 -1.16478479e+00 3.45817953e-01
-1.54035449e+00 1.71822384e-02 5.56990743e-01 1.16028214e+00
4.09097433e-01 6.05408549e-01 4.88974541e-01 -6.27029538e-01
3.21905345e-01 -6.34574831e-01 -4.26151395e-01 -3.80742624e-02
-6.35253906e-01 -2.20057949e-01 8.06895971e-01 -5.56994855e-01
-1.08031774e+00 -4.96990949e-01 -1.21490195e-01 -8.83719027e-01
2.80449893e-02 2.11611450e-01 -2.88604617e-01 -3.66097927e-01
7.10462868e-01 5.52265108e-01 1.20104119e-01 -1.29288733e-01
5.40259004e-01 6.57161772e-01 1.12085855e+00 -2.49137968e-01
1.62449431e+00 7.02748239e-01 -3.99443299e-01 -8.07831645e-01
-8.60200286e-01 -2.84283012e-01 -7.27915689e-02 -1.08558565e-01
8.45077395e-01 -1.23796320e+00 -1.72978431e-01 8.92362416e-01
-7.45099664e-01 -6.04933619e-01 -3.11808109e-01 3.29226464e-01
-2.37925828e-01 4.18440819e-01 -5.60564816e-01 -7.21720517e-01
-3.63221467e-01 -9.28279221e-01 8.62983286e-01 4.01437104e-01
5.74465394e-01 -1.01222336e+00 -9.91556421e-02 4.89734739e-01
6.22262716e-01 4.83419776e-01 8.64941120e-01 -1.28781348e-01
-7.18915045e-01 -1.61839828e-01 -2.46965602e-01 1.15205193e+00
4.26347524e-01 -2.20121682e-01 -9.92625773e-01 -6.57954216e-01
4.32422817e-01 -2.96528071e-01 9.65158641e-01 2.13911504e-01
1.22299278e+00 -5.35044789e-01 3.03759396e-01 9.63032961e-01
1.54022014e+00 2.16347426e-01 1.13898039e+00 5.42759776e-01
8.38790417e-01 4.31677043e-01 6.91101909e-01 -1.75284334e-02
1.98269725e-01 3.10982704e-01 8.67813408e-01 -6.36168778e-01
-4.18006331e-01 -3.41965765e-01 3.92804712e-01 8.32755268e-01
1.99694797e-01 -5.80834508e-01 -5.28081775e-01 7.41493225e-01
-1.34004593e+00 -9.09880757e-01 2.56012499e-01 2.12292838e+00
1.00332904e+00 -3.00978068e-02 -2.49325022e-01 1.54076070e-01
8.36039782e-01 6.03675663e-01 -4.93662149e-01 8.10901970e-02
-4.86710459e-01 3.38574439e-01 4.72807288e-01 5.74800849e-01
-1.20797694e+00 7.87815928e-01 5.42396212e+00 9.60196674e-01
-1.51029074e+00 1.81996763e-01 7.37562001e-01 -2.63394833e-01
-3.11380178e-01 -2.80098915e-01 -5.08508421e-02 9.18045819e-01
6.83428586e-01 -1.79355696e-01 7.55966544e-01 5.66912711e-01
1.87496454e-01 7.90937170e-02 -7.42268026e-01 1.23478150e+00
2.63175577e-01 -1.39512217e+00 1.85445100e-01 1.49123298e-04
1.27367508e+00 6.61698058e-02 7.43459225e-01 2.89378136e-01
3.23329210e-01 -1.07884300e+00 7.91249335e-01 3.81185174e-01
9.01964605e-01 -6.14463627e-01 7.94308066e-01 1.85139298e-01
-6.70417964e-01 -9.37645510e-02 -4.40243095e-01 6.18631057e-02
-7.64797106e-02 7.36356735e-01 -3.69371355e-01 7.17617631e-01
9.30346966e-01 7.60874152e-01 -7.19764352e-01 8.74233663e-01
-4.09528017e-01 7.11931586e-01 1.53429732e-01 6.10461056e-01
3.63390654e-01 -3.09550285e-01 4.49040681e-01 9.57589209e-01
3.93578917e-01 2.01821744e-01 -1.12033591e-01 9.10754621e-01
-2.37372667e-01 -4.29166049e-01 -6.54158533e-01 5.29652089e-02
4.79681075e-01 9.94502604e-01 -1.23472743e-01 -4.67818260e-01
-3.22841674e-01 1.26575601e+00 -4.40710485e-02 6.97423518e-01
-9.58734989e-01 -6.87812805e-01 7.77613997e-01 -4.27347757e-02
3.05717885e-01 -9.81576145e-02 -1.18732966e-01 -1.21375227e+00
1.44554168e-01 -1.23471510e+00 9.19666588e-02 -1.26601624e+00
-1.82571256e+00 6.59658790e-01 -1.50551140e-01 -1.61484659e+00
2.35425994e-01 -6.06549740e-01 -8.79342496e-01 1.06864417e+00
-2.00365853e+00 -1.04568601e+00 -8.78175735e-01 7.28287399e-01
4.41504031e-01 -8.56396277e-03 2.79985398e-01 5.43880403e-01
-5.44119954e-01 7.80838251e-01 4.53632712e-01 2.73168892e-01
1.07271039e+00 -1.27077925e+00 3.37947816e-01 1.43216753e+00
-2.63488293e-01 6.51571274e-01 8.74945462e-01 -3.66541773e-01
-1.15583849e+00 -1.59532225e+00 1.44714445e-01 -5.28454959e-01
4.99210149e-01 -3.80382746e-01 -1.27073288e+00 5.48985362e-01
6.42982423e-01 4.94373173e-01 4.11866426e-01 -9.24645245e-01
-6.13793015e-01 -3.93290013e-01 -1.22063398e+00 7.49410391e-01
7.75955081e-01 -7.10128725e-01 -6.32376373e-01 2.68479139e-01
1.07943380e+00 -3.84870738e-01 -6.64302945e-01 3.30233008e-01
1.37669533e-01 -1.07720661e+00 1.18890679e+00 -1.64727941e-01
8.69484723e-01 -7.25629091e-01 -1.30214170e-01 -1.70781279e+00
-2.10846707e-01 -7.02947140e-01 -8.17456916e-02 1.08163786e+00
1.68516353e-01 -9.00934517e-01 6.36359870e-01 1.02520123e-01
-4.46931213e-01 -6.16198003e-01 -6.64167225e-01 -9.15699065e-01
5.26421607e-01 4.27322611e-02 7.33178198e-01 1.33298862e+00
-6.81972206e-01 -4.81082238e-02 -8.69188309e-01 5.97794592e-01
9.51732934e-01 -1.37842178e-01 8.02233100e-01 -4.97148484e-01
-1.43831104e-01 -1.78855419e-01 -3.30979496e-01 -9.09034789e-01
2.35761963e-02 -6.53131604e-01 4.03288960e-01 -1.30645120e+00
-3.11696995e-02 -3.01059991e-01 -5.23302674e-01 3.90799522e-01
-6.69381261e-01 6.74123883e-01 2.56464630e-01 2.15350494e-01
-2.41400734e-01 9.42537427e-01 1.65296948e+00 -6.53916657e-01
7.10316300e-02 -3.69042933e-01 -9.78638768e-01 3.53927463e-01
7.71393180e-01 -5.23511708e-01 -6.70187652e-01 -5.60315430e-01
-1.45834371e-01 -2.37700015e-01 5.85832119e-01 -1.08113098e+00
2.11111903e-01 -2.06339285e-01 5.58726370e-01 -2.83754140e-01
3.78814101e-01 -8.29798222e-01 9.25970674e-02 3.05835485e-01
-1.17034182e-01 4.91030589e-02 7.14258179e-02 6.96375012e-01
-6.44928455e-01 6.66593239e-02 1.14971721e+00 -1.21538579e-01
-7.69018054e-01 5.62992036e-01 -4.30350453e-02 1.98884249e-01
9.40594018e-01 -3.81992728e-01 -9.30583239e-01 -6.79826915e-01
-2.01672092e-01 2.72876948e-01 8.05136740e-01 3.75513762e-01
1.00825799e+00 -1.36548316e+00 -7.66571343e-01 4.02118951e-01
3.36146802e-01 3.10846508e-01 6.76955342e-01 6.57372534e-01
-8.23138535e-01 -3.02462250e-01 -4.74599063e-01 -2.22000599e-01
-9.42801833e-01 1.19573295e+00 4.09461141e-01 2.07678422e-01
-6.52426243e-01 8.37265134e-01 8.03112030e-01 -3.44993770e-01
-1.62825640e-03 6.62838668e-02 2.26747647e-01 -3.86456013e-01
8.03161263e-01 2.78330773e-01 -2.59581264e-02 -5.13979614e-01
-7.07693845e-02 2.66294748e-01 -6.16313703e-02 2.23821830e-02
1.25565708e+00 -3.15047443e-01 -3.56832296e-01 2.21768811e-01
1.50556350e+00 1.82541102e-01 -1.58273840e+00 -2.79212654e-01
-6.31822228e-01 -8.91725302e-01 2.52538383e-01 -7.47178853e-01
-1.33142114e+00 1.07696891e+00 9.53532100e-01 1.51749551e-01
1.62580276e+00 -3.23017925e-01 1.04778028e+00 1.42179236e-01
7.06481794e-03 -6.94592118e-01 5.27964711e-01 1.70906022e-01
9.71897542e-01 -1.41597509e+00 -6.98011443e-02 -3.82620782e-01
-5.99368155e-01 4.99617875e-01 8.58797371e-01 -4.39516962e-01
5.56561530e-01 -2.31258571e-02 4.78792399e-01 -1.38916776e-01
-3.60278636e-01 7.37800524e-02 5.48753925e-02 8.75114918e-01
-2.16979221e-01 -1.69689804e-01 2.66995370e-01 1.19228132e-01
-2.78140217e-01 -2.62707770e-01 7.08597183e-01 9.69782650e-01
-2.44269818e-01 -7.25931883e-01 -6.12529695e-01 1.93131939e-01
-3.17606568e-01 -4.14374262e-01 -1.68213844e-01 8.25950444e-01
3.31751794e-01 1.03684688e+00 -8.94967746e-03 -4.88810599e-01
2.13904142e-01 -4.26455110e-01 3.02815706e-01 -3.55465680e-01
-2.59508550e-01 -2.05767155e-01 -4.10189241e-01 -2.03750193e-01
-4.14369732e-01 -3.93511392e-02 -8.21086168e-01 -5.25834084e-01
-2.01905578e-01 2.50875980e-01 2.57591158e-01 6.24915481e-01
2.61853576e-01 6.40269756e-01 1.17914069e+00 -1.05853033e+00
-3.89784068e-01 -8.13469052e-01 -6.83560133e-01 6.14726186e-01
1.12569678e+00 -5.12667716e-01 -8.04298580e-01 1.80399269e-01]
|
[10.931320190429688, -3.1396548748016357]
|
89301d1b-ac34-4b51-82c7-15cfee40b0a1
|
visda-2021-competition-universal-domain
|
2107.11011
| null |
https://arxiv.org/abs/2107.11011v1
|
https://arxiv.org/pdf/2107.11011v1.pdf
|
VisDA-2021 Competition Universal Domain Adaptation to Improve Performance on Out-of-Distribution Data
|
Progress in machine learning is typically measured by training and testing a model on the same distribution of data, i.e., the same domain. This over-estimates future accuracy on out-of-distribution data. The Visual Domain Adaptation (VisDA) 2021 competition tests models' ability to adapt to novel test distributions and handle distributional shift. We set up unsupervised domain adaptation challenges for image classifiers and will evaluate adaptation to novel viewpoints, backgrounds, modalities and degradation in quality. Our challenge draws on large-scale publicly available datasets but constructs the evaluation across domains, rather that the traditional in-domain bench-marking. Furthermore, we focus on the difficult "universal" setting where, in addition to input distribution drift, methods may encounter missing and/or novel classes in the target dataset. Performance will be measured using a rigorous protocol, comparing to state-of-the-art domain adaptation methods with the help of established metrics. We believe that the competition will encourage further improvement in machine learning methods' ability to handle realistic data in many deployment scenarios.
|
['Ben Usman', 'Piotr Teterwak', 'Kuniaki Saito', 'Kate Saenko', 'Samarth Mishra', 'Donghyun Kim', 'Dan Hendrycks', 'Dina Bashkirova']
|
2021-07-23
| null | null | null | null |
['universal-domain-adaptation']
|
['computer-vision']
|
[ 2.30866641e-01 -3.25129062e-01 -1.81011111e-01 -6.71321929e-01
-9.48054314e-01 -9.72042143e-01 7.14338839e-01 -5.64249679e-02
-5.25469601e-01 9.75825608e-01 4.56429087e-02 2.76424754e-02
1.68650210e-01 -2.81658143e-01 -7.94734299e-01 -4.95221496e-01
-2.46971268e-02 8.16295445e-01 3.83583248e-01 1.05087884e-01
-7.54057169e-02 5.06166160e-01 -1.50958157e+00 5.29041350e-01
5.07659256e-01 8.19236875e-01 -4.77844514e-02 8.25200796e-01
1.93138160e-02 4.24455822e-01 -9.80302989e-01 -4.47639763e-01
3.53363484e-01 -4.06574279e-01 -7.34940231e-01 1.73737600e-01
1.06383300e+00 -3.30733865e-01 -1.96620762e-01 9.29233909e-01
7.78544784e-01 1.02047391e-01 9.16107774e-01 -1.47283328e+00
-9.74453449e-01 1.00403920e-01 -5.46057701e-01 5.06796896e-01
3.33693117e-01 4.60816234e-01 4.07079935e-01 -8.46402884e-01
1.03777158e+00 1.10522258e+00 9.04100597e-01 9.09826219e-01
-1.57711458e+00 -6.77265704e-01 4.46577907e-01 4.23078477e-01
-1.11382222e+00 -5.63514769e-01 5.29978573e-01 -6.42032564e-01
9.29958224e-01 -1.44868046e-01 1.10248007e-01 1.77666199e+00
-1.83216974e-01 6.60753489e-01 1.08622003e+00 -4.47813570e-01
4.95191604e-01 6.18441880e-01 -3.30392987e-01 -1.19491562e-01
2.06291929e-01 2.58557618e-01 -5.55221975e-01 -5.84814814e-04
2.78661579e-01 -3.93107235e-01 -1.43635675e-01 -1.04624844e+00
-1.16725993e+00 5.69670618e-01 1.69174373e-01 3.96465249e-02
-4.56516854e-02 -1.97004691e-01 6.85352027e-01 4.70175982e-01
5.86173773e-01 3.28073829e-01 -9.40940797e-01 -3.83476138e-01
-1.00303566e+00 3.20977449e-01 7.44196296e-01 1.21079397e+00
5.43602228e-01 -4.72055860e-02 -2.12973177e-01 1.04204524e+00
-6.07657246e-02 7.53356934e-01 4.24377292e-01 -1.06355453e+00
5.44583797e-01 1.55902699e-01 2.17058003e-01 -2.97710449e-01
-3.31421077e-01 -4.59050208e-01 -5.16496897e-01 7.05153286e-01
7.65725911e-01 -2.58085907e-01 -1.17264760e+00 1.98687077e+00
3.95240813e-01 5.37466109e-02 7.25090504e-02 6.42578065e-01
6.16453648e-01 1.83196038e-01 3.63709152e-01 8.90009291e-03
8.46044600e-01 -6.95325077e-01 -3.19459736e-01 -3.04141015e-01
3.64335090e-01 -7.42684722e-01 1.47861314e+00 6.16098642e-01
-9.63497519e-01 -8.27312887e-01 -1.07404661e+00 2.77333260e-01
-5.62360883e-01 -1.12162404e-01 1.52426943e-01 8.07829380e-01
-1.04172218e+00 4.40766007e-01 -6.39875054e-01 -9.05653536e-01
7.86365747e-01 1.14075951e-01 -4.73085910e-01 -3.81226987e-01
-9.10860479e-01 1.13640559e+00 2.84415811e-01 -6.79495990e-01
-1.27648115e+00 -1.09053528e+00 -6.32729173e-01 -4.51186270e-01
1.62555091e-02 -6.69487834e-01 1.52915394e+00 -1.57997847e+00
-1.13303435e+00 1.28592753e+00 3.65373753e-02 -4.49488252e-01
9.00782585e-01 -1.80919886e-01 -8.98361802e-01 -6.32082894e-02
1.39496416e-01 8.62652898e-01 9.68915343e-01 -1.49539876e+00
-8.42435122e-01 -3.48628879e-01 -1.11497588e-01 1.88887253e-01
-5.00924408e-01 -1.39601231e-01 -2.62742668e-01 -4.71251190e-01
-4.70881015e-01 -9.13085818e-01 1.50948897e-01 1.13344565e-01
2.10058931e-02 1.17747210e-01 9.51470971e-01 -5.58463633e-01
8.93951058e-01 -2.48593926e+00 -3.73148844e-02 1.85588729e-02
-7.75030330e-02 2.05185696e-01 -3.75399232e-01 1.67473927e-01
-2.94174880e-01 -4.24301438e-03 -2.13186979e-01 -3.56370538e-01
1.76208522e-02 1.69719100e-01 -1.85732633e-01 5.21947205e-01
4.10276353e-01 4.44942266e-01 -7.64420927e-01 -3.59563708e-01
7.11322501e-02 2.55588412e-01 -5.95597804e-01 1.36463284e-01
-3.67848486e-01 6.39776945e-01 1.02851428e-01 6.73278689e-01
1.07876408e+00 -1.51267126e-01 3.43654826e-02 6.86346972e-03
2.48432815e-01 -1.88273773e-01 -1.07462215e+00 2.05377412e+00
-3.88923109e-01 7.38986254e-01 -1.78055897e-01 -8.22952211e-01
7.71880150e-01 5.13643138e-02 3.18440884e-01 -9.62737024e-01
-2.11561486e-01 7.20016733e-02 -2.79327715e-03 -4.55016226e-01
2.78752595e-01 -1.16063520e-01 -7.67595172e-02 1.04786374e-01
6.28190875e-01 -1.65287614e-01 1.72008246e-01 1.44499868e-01
1.18391335e+00 5.17257631e-01 1.63234472e-01 -2.30533779e-01
7.88392946e-02 3.37071389e-01 3.31233054e-01 9.68708992e-01
-7.41597772e-01 9.17670906e-01 3.11734170e-01 -4.86300796e-01
-1.54515934e+00 -1.67112696e+00 -3.83428007e-01 1.46759915e+00
2.65186820e-02 2.75308967e-01 -7.37438381e-01 -1.14101279e+00
3.10845673e-01 9.56172705e-01 -8.80465984e-01 -3.30225736e-01
-1.23407699e-01 -6.01297081e-01 6.11953735e-01 7.25464106e-01
2.97327399e-01 -9.54163194e-01 -4.79258716e-01 9.81159508e-02
5.01847193e-02 -1.18199861e+00 -2.89511591e-01 2.45141119e-01
-7.62662292e-01 -9.99297082e-01 -9.77448940e-01 -6.06861055e-01
4.86499757e-01 -2.11951602e-02 1.75913823e+00 -4.17605221e-01
-2.81212717e-01 9.52153563e-01 -4.78418291e-01 -7.40633547e-01
-5.70405722e-01 2.43535295e-01 1.83529407e-01 -2.51772285e-01
8.05622637e-01 -4.70760912e-01 -5.79242945e-01 5.83393455e-01
-9.11220729e-01 -3.32512081e-01 4.15535510e-01 7.87038743e-01
5.44697940e-01 -1.90823704e-01 7.20386982e-01 -1.09462702e+00
5.50359547e-01 -7.49138176e-01 -5.22265255e-01 4.14303273e-01
-6.85977161e-01 -1.07899934e-01 4.36318189e-01 -9.60388601e-01
-1.39952660e+00 5.04126027e-02 2.39581615e-01 -6.02113426e-01
-6.77853823e-01 -5.64997783e-04 -3.13067764e-01 1.34982124e-01
1.51301062e+00 3.08796123e-04 -1.28598154e-01 -4.04525906e-01
4.10202265e-01 5.15872955e-01 8.13140094e-01 -8.24311733e-01
9.46830571e-01 5.62349617e-01 -3.57034117e-01 -5.22860050e-01
-6.08278275e-01 -3.71149004e-01 -8.94373000e-01 -1.59986734e-01
5.44455886e-01 -1.14632154e+00 4.22132239e-02 4.92727339e-01
-8.73866081e-01 -7.38475323e-01 -6.86484158e-01 2.91483790e-01
-6.76116645e-01 1.43352047e-01 4.24212627e-02 -5.74380934e-01
2.10625261e-01 -6.97405636e-01 9.12683010e-01 3.23886305e-01
-2.93513596e-01 -1.23772526e+00 5.08828402e-01 1.40614375e-01
5.84690392e-01 3.14134359e-01 5.89394987e-01 -9.89641070e-01
-1.05876587e-01 -2.85546407e-02 -3.00061345e-01 7.86438763e-01
5.10737067e-03 9.82293189e-02 -1.34255326e+00 -6.44286990e-01
-3.96357238e-01 -6.94281101e-01 7.31313825e-01 3.39307010e-01
1.27415299e+00 2.03904226e-01 -3.25816661e-01 6.33212149e-01
1.31584191e+00 1.28049955e-01 7.36329913e-01 6.68467522e-01
2.53412455e-01 4.51053202e-01 6.62442923e-01 6.18161976e-01
2.94911891e-01 6.35428309e-01 3.32097203e-01 -2.24071026e-01
-6.02060020e-01 -1.62357643e-01 3.07158262e-01 -4.63917702e-02
2.70895213e-01 -2.87481517e-01 -1.16151237e+00 8.56161833e-01
-1.52030528e+00 -9.18564796e-01 2.23575816e-01 2.46329665e+00
9.61144209e-01 3.24772239e-01 4.83808219e-01 -3.16238195e-01
5.66809833e-01 -3.86113375e-01 -1.17403877e+00 -4.14489031e-01
-3.43267798e-01 1.95696965e-01 5.79109371e-01 1.31838039e-01
-1.15615940e+00 7.75002062e-01 7.12763119e+00 6.76226556e-01
-1.02331865e+00 3.23501080e-01 7.28643179e-01 -4.17493343e-01
-1.62118718e-01 -2.63661355e-01 -7.65873492e-01 5.36829770e-01
1.00007308e+00 -7.74913803e-02 4.39216316e-01 1.15224195e+00
-2.64123142e-01 -5.07637709e-02 -1.44555318e+00 9.60121632e-01
3.06289017e-01 -8.95808220e-01 -2.03587905e-01 -1.52305424e-01
1.08451986e+00 3.82844329e-01 4.62095112e-01 7.19923794e-01
7.23712802e-01 -9.29347873e-01 7.28341997e-01 5.17629802e-01
1.37259388e+00 -5.19138038e-01 5.88224828e-01 5.92307262e-02
-5.77323258e-01 -1.05962515e-01 -4.20163482e-01 2.29577124e-01
-1.50165424e-01 3.73974741e-01 -1.10797250e+00 1.70783341e-01
1.21274626e+00 6.19080245e-01 -9.90416169e-01 1.16818702e+00
1.29846051e-01 6.86599731e-01 -3.13841611e-01 3.68133068e-01
-2.35810265e-01 5.69373190e-01 3.69924694e-01 1.55429292e+00
1.79341093e-01 -4.39449221e-01 5.21840565e-02 4.85444516e-01
-1.89316452e-01 -2.16813833e-01 -7.60904312e-01 2.72063106e-01
6.35537386e-01 8.28766644e-01 -1.39415935e-01 -4.18043315e-01
-5.95282376e-01 1.08159471e+00 3.20547551e-01 6.37058973e-01
-9.45410609e-01 -3.15955095e-02 9.17453945e-01 2.60914981e-01
3.66136491e-01 6.18518032e-02 -4.71917152e-01 -1.18414378e+00
2.65434980e-02 -1.16764605e+00 8.16131294e-01 -8.20600331e-01
-1.93739438e+00 5.66347182e-01 4.06278849e-01 -1.51319146e+00
-3.01018834e-01 -7.25114703e-01 -1.63846329e-01 8.04379582e-01
-1.65129685e+00 -1.09568083e+00 -6.18109941e-01 8.63384545e-01
6.15765095e-01 -5.97315609e-01 9.08420503e-01 3.76618654e-01
-1.90105245e-01 1.08748913e+00 5.21966517e-01 -7.39027187e-02
1.59249723e+00 -1.40964079e+00 5.64157724e-01 7.00364411e-01
3.43055353e-02 1.68433264e-02 8.46494555e-01 -4.53575075e-01
-6.71008527e-01 -1.11347342e+00 2.49807790e-01 -1.17158806e+00
5.46741962e-01 -3.84006739e-01 -1.04485750e+00 8.50179672e-01
2.66354591e-01 3.11397165e-01 8.69009316e-01 3.44167560e-01
-8.61751318e-01 -3.18177998e-01 -1.58932567e+00 2.94374317e-01
1.28523970e+00 -3.48977655e-01 -4.05724138e-01 3.08102131e-01
4.91105288e-01 -5.91874421e-01 -7.97681034e-01 4.84850407e-01
6.02293611e-01 -1.02493477e+00 1.11648369e+00 -1.09292746e+00
3.75045598e-01 -2.50199825e-01 -4.79804456e-01 -1.66379893e+00
-4.13488716e-01 -1.17709137e-01 4.37833779e-02 1.41795838e+00
5.85767865e-01 -3.99146080e-01 8.18641126e-01 5.82100868e-01
9.28438753e-02 2.83333287e-02 -1.01499343e+00 -1.11537266e+00
5.54040968e-01 -4.66558456e-01 5.82480431e-01 1.03806782e+00
-4.17494893e-01 1.01040028e-01 -2.13182911e-01 1.50801450e-01
7.53957391e-01 -4.24777836e-01 1.18600488e+00 -1.15765977e+00
-3.03868920e-01 -3.34177673e-01 -5.52172124e-01 -6.39383733e-01
9.05102957e-03 -5.34242153e-01 9.02095530e-03 -1.16830373e+00
3.11156929e-01 -4.57785040e-01 -5.56826651e-01 3.36319476e-01
3.86442654e-02 3.19411606e-01 2.09394902e-01 1.77891031e-01
-9.62219656e-01 4.45826948e-01 1.00359893e+00 -3.06358516e-01
-1.02951124e-01 -1.89816341e-01 -6.74075603e-01 5.08897305e-01
7.92400897e-01 -6.20501101e-01 -5.91437221e-01 -7.48108387e-01
-6.18067458e-02 -5.90153635e-01 4.42744553e-01 -1.51118982e+00
-3.21029201e-02 -3.17403495e-01 1.03478765e+00 -2.30332926e-01
2.00238273e-01 -1.06084037e+00 1.11760721e-01 -7.84875602e-02
-4.06422049e-01 -2.30142269e-02 7.61548102e-01 5.29653370e-01
-3.63238566e-02 -2.00499985e-02 1.17634249e+00 2.05808207e-02
-1.11030924e+00 1.14390887e-01 1.12785049e-01 6.76211536e-01
1.26082075e+00 -4.64049548e-01 -5.94564557e-01 -3.17301124e-01
-1.03406835e+00 1.35930225e-01 8.50428700e-01 6.57673597e-01
3.40184063e-01 -1.44046986e+00 -1.03440571e+00 1.44891277e-01
8.30610812e-01 -1.59233674e-01 4.07964528e-01 3.31780016e-01
-2.91261137e-01 -2.17697218e-01 -5.35978794e-01 -9.97538269e-01
-1.16565025e+00 6.60669267e-01 4.58332360e-01 -1.69055164e-01
-1.44573048e-01 8.80232692e-01 2.67875552e-01 -6.06294274e-01
3.50539923e-01 5.43825775e-02 1.34950861e-01 -6.60577118e-02
5.80860019e-01 1.18462130e-01 2.52393872e-01 -2.47449607e-01
-5.22705674e-01 1.84243917e-01 -2.07839474e-01 -2.18487516e-01
1.08972228e+00 -2.14686781e-01 6.68207586e-01 4.97159481e-01
9.99299824e-01 -2.61785060e-01 -1.99670219e+00 -4.65505302e-01
-3.17289643e-02 -6.18639469e-01 -3.97055894e-01 -1.63109112e+00
-6.38498187e-01 8.99998844e-01 1.31958389e+00 -6.87623024e-02
1.28672087e+00 8.55543688e-02 2.26743624e-01 1.02570310e-01
2.95581877e-01 -1.48280835e+00 3.04677367e-01 3.43555987e-01
9.73356128e-01 -1.55252814e+00 1.07544214e-01 1.91383183e-01
-1.05814767e+00 9.04211342e-01 9.85702455e-01 1.19534388e-01
5.94349921e-01 3.13738525e-01 3.87894899e-01 3.01164299e-01
-7.87314236e-01 -2.56967507e-02 2.25962713e-01 1.58917260e+00
2.14755356e-01 1.04199924e-01 4.33975428e-01 4.22494978e-01
-4.53289449e-02 1.47357747e-01 3.47447902e-01 7.31794298e-01
-1.24672614e-01 -1.15775943e+00 -4.80949938e-01 3.57813627e-01
-1.94744885e-01 1.70886621e-01 -4.53827918e-01 1.14665723e+00
4.02531773e-01 7.95040309e-01 2.80126125e-01 -2.91500777e-01
8.49178433e-01 1.70874804e-01 6.91259861e-01 -7.62678564e-01
-1.83905199e-01 -2.94486076e-01 7.00931624e-02 -3.72722596e-01
-3.63395929e-01 -1.17611670e+00 -7.39989161e-01 -2.81041592e-01
2.24549815e-01 -3.22897643e-01 7.48723567e-01 6.52242362e-01
5.46565711e-01 4.76891458e-01 3.74946296e-01 -7.01584637e-01
-8.58234286e-01 -1.04555345e+00 -3.86698514e-01 1.15268672e+00
3.24687868e-01 -8.72524261e-01 -2.61804253e-01 5.46840131e-01]
|
[10.12208080291748, 2.7065601348876953]
|
72033dd7-b88e-454e-b126-ef9fc5887a17
|
automated-linear-time-detection-and-quality
|
1701.0194
| null |
http://arxiv.org/abs/1701.01940v1
|
http://arxiv.org/pdf/1701.01940v1.pdf
|
Automated Linear-Time Detection and Quality Assessment of Superpixels in Uncalibrated True- or False-Color RGB Images
|
Capable of automated near real time superpixel detection and quality
assessment in an uncalibrated monitor typical red green blue (RGB) image,
depicted in either true or false colors, an original low level computer vision
(CV) lightweight computer program, called RGB Image Automatic Mapper (RGBIAM),
is designed and implemented. Constrained by the Calibration Validation (CalVal)
requirements of the Quality Assurance Framework for Earth Observation (QA4EO)
guidelines, RGBIAM requires as mandatory an uncalibrated RGB image pre
processing first stage, consisting of an automated statistical model based
color constancy algorithm. The RGBIAM hybrid inference pipeline comprises: (I)
a direct quantitative to nominal (QN) RGB variable transform, where RGB pixel
values are mapped onto a prior dictionary of color names, equivalent to a
static polyhedralization of the RGB cube. Prior color naming is the deductive
counterpart of inductive vector quantization (VQ), whose typical VQ error
function to minimize is a root mean square error (RMSE). In the output multi
level color map domain, superpixels are automatically detected in linear time
as connected sets of pixels featuring the same color label. (II) An inverse
nominal to quantitative (NQ) RGB variable transform, where a superpixelwise
constant RGB image approximation is generated in linear time to assess a VQ
error image. The hybrid direct and inverse RGBIAM QNQ transform is: (i) general
purpose, data and application independent. (ii) Automated, i.e., it requires no
user machine interaction. (iii) Near real time, with a computational complexity
increasing linearly with the image size. (iv) Implemented in tile streaming
mode, to cope with massive images. Collected outcome and process quality
indicators, including degree of automation, computational efficiency, VQ rate
and VQ error, are consistent with theoretical expectations.
|
['Stefan Lang', 'Dirk Tiede', 'Andrea Baraldi']
|
2017-01-08
| null | null | null | null |
['color-constancy']
|
['computer-vision']
|
[ 4.11310762e-01 -2.98575144e-02 4.79497433e-01 -2.18532875e-01
-9.09795105e-01 -9.75381911e-01 4.80074912e-01 1.17444836e-01
-4.44542140e-01 5.82757235e-01 -6.36400998e-01 -6.03163600e-01
-1.01273425e-01 -1.20172977e+00 -7.56926596e-01 -8.91995490e-01
1.71539441e-01 5.22012532e-01 2.43662894e-01 1.22881392e-02
2.03400865e-01 6.80700481e-01 -1.79986906e+00 -3.05319969e-02
8.49845648e-01 1.63734603e+00 1.04798369e-01 1.35289347e+00
-1.25145316e-01 6.77048683e-01 -3.65861803e-01 -2.68015396e-02
8.94724607e-01 -4.39315081e-01 -7.22700715e-01 1.16792113e-01
5.43392122e-01 -2.16190785e-01 3.44667464e-01 1.31767750e+00
2.51617968e-01 -1.05107091e-01 6.04711354e-01 -1.61721122e+00
-4.17661160e-01 -1.60056770e-01 -4.83266354e-01 -1.65503040e-01
2.02143341e-01 6.73004091e-01 7.39861548e-01 -8.97032082e-01
3.84443313e-01 9.08948660e-01 7.48225808e-01 -3.54696274e-01
-1.50481546e+00 -2.84772038e-01 -4.97251809e-01 -3.92661728e-02
-1.80536592e+00 1.43014833e-01 2.97089517e-01 -5.67917705e-01
7.37359941e-01 8.05126488e-01 1.17466235e+00 -4.93777320e-02
1.76264957e-01 -1.55735686e-01 1.92750120e+00 -5.35906971e-01
7.62256622e-01 1.46685109e-01 -3.32662731e-01 6.78975284e-01
2.22401366e-01 1.15945056e-01 -1.88682616e-01 9.51306522e-02
9.91998494e-01 -2.20324978e-01 -1.75461799e-01 -2.58355528e-01
-1.20944142e+00 7.49587417e-01 6.38360441e-01 -1.40875638e-01
-5.42724073e-01 3.91356289e-01 1.31352201e-01 3.01055968e-01
1.92924634e-01 3.66897494e-01 -5.20716012e-01 -1.22699447e-01
-1.05646574e+00 -7.65310675e-02 5.79053223e-01 1.08799887e+00
1.41387105e+00 1.56350195e-01 3.67702208e-02 2.73300171e-01
5.26364207e-01 1.33739209e+00 9.27853435e-02 -1.49225581e+00
1.10732667e-01 1.04999435e+00 2.66974062e-01 -9.65845287e-01
-5.09408295e-01 -2.35715285e-02 -9.82961118e-01 9.65389192e-01
3.82525086e-01 1.42282978e-01 -9.59236085e-01 1.00175977e+00
4.96828914e-01 -4.35072482e-01 7.80952796e-02 9.80003476e-01
5.83190620e-01 8.73975039e-01 2.14470718e-02 -2.34191373e-01
1.64951241e+00 -1.48267418e-01 -2.86412001e-01 1.25374019e-01
3.84092182e-01 -6.05909109e-01 1.36400783e+00 7.09593892e-01
-9.09352183e-01 -5.21958590e-01 -1.02204001e+00 -8.26063305e-02
-7.07372844e-01 1.25809938e-01 5.58462799e-01 1.12342441e+00
-1.44878662e+00 1.74382150e-01 -6.49519563e-01 -3.03515881e-01
2.19161704e-01 1.35746032e-01 -3.78865093e-01 3.44750211e-02
-9.78490591e-01 6.90318704e-01 5.68067014e-01 4.34042335e-01
-6.39079690e-01 -7.92959034e-01 -6.99689984e-01 -2.28302747e-01
-2.49676537e-02 -3.87192130e-01 5.61430335e-01 -1.35931385e+00
-1.65146530e+00 1.01282024e+00 2.08438963e-01 -2.21535116e-01
7.80988455e-01 2.60328740e-01 -1.99333772e-01 4.86118406e-01
1.27143428e-01 8.00445139e-01 7.47448266e-01 -1.43896866e+00
-8.37692916e-01 -4.24884915e-01 -2.36448020e-01 1.63738161e-01
1.20660797e-01 -3.96674931e-01 -3.94094348e-01 -1.67243451e-01
5.18555880e-01 -9.13711131e-01 -1.46898881e-01 6.50105596e-01
-2.21272394e-01 1.70281127e-01 4.97127563e-01 -7.84156978e-01
8.24047387e-01 -1.94539595e+00 -3.06743741e-01 8.29771876e-01
-2.00541578e-02 6.12104610e-02 3.62968177e-01 5.38997948e-02
1.80186201e-02 4.82406504e-02 -6.62613750e-01 2.09734797e-01
-1.56397298e-01 2.77144730e-01 3.91137786e-02 8.61246467e-01
6.62531108e-02 6.97298825e-01 -1.12833476e+00 -6.31508768e-01
8.41299713e-01 3.08511913e-01 -1.59983709e-01 1.03954524e-01
-9.71112922e-02 2.85053700e-01 3.88128050e-02 9.18336034e-01
1.17764163e+00 -4.51367870e-02 -1.12784943e-02 -7.26264358e-01
-6.79503679e-01 -5.10369778e-01 -1.70742965e+00 1.35137510e+00
-4.88260716e-01 9.42439020e-01 1.91933751e-01 -4.93476808e-01
1.18288624e+00 3.19084642e-03 3.65518212e-01 -9.67375815e-01
-3.79097871e-02 3.30299824e-01 -6.71512425e-01 -3.65138441e-01
6.31778955e-01 3.49686705e-02 2.12499619e-01 9.05815065e-02
-1.45997375e-01 -7.67437875e-01 -3.41752581e-02 2.83839386e-02
7.08486855e-01 4.92460907e-01 4.34561223e-01 -5.63383043e-01
5.38836241e-01 6.75871611e-01 3.39935541e-01 7.38457620e-01
-2.31675401e-01 9.27570105e-01 3.69719326e-01 -3.44721437e-01
-1.42181122e+00 -1.15103042e+00 -3.40875536e-01 7.07002401e-01
3.62432033e-01 5.29522523e-02 -6.73798084e-01 2.29650244e-01
-1.26411496e-02 6.47566378e-01 -6.26282454e-01 5.22175193e-01
-1.99230790e-01 -8.52998972e-01 4.65896875e-01 2.75825858e-01
9.18652833e-01 -7.05544770e-01 -1.41306818e+00 9.66456681e-02
4.33713831e-02 -8.48452151e-01 3.10053289e-01 4.14198339e-01
-8.06231499e-01 -1.17869270e+00 -3.58005375e-01 -1.58059314e-01
6.13004029e-01 1.30634606e-01 1.19432342e+00 -6.73755035e-02
-6.34859443e-01 6.90520108e-01 -4.91928995e-01 -2.12766126e-01
-3.18132162e-01 -7.98704922e-01 -2.92562485e-01 1.99722107e-02
4.36631814e-02 -1.53048337e-01 -1.09064496e+00 1.18740052e-01
-1.09735012e+00 2.75365144e-01 5.86860180e-01 4.87522244e-01
1.11356533e+00 2.63221115e-02 -3.87933344e-01 -4.66346562e-01
-2.48974666e-01 -1.22842975e-01 -1.39908719e+00 3.34955007e-01
-9.63606775e-01 -8.84886533e-02 5.78271329e-01 2.23076686e-01
-8.11709464e-01 5.44615865e-01 1.86467618e-01 -2.75876880e-01
-2.44785041e-01 3.18957537e-01 -7.87316561e-02 -3.91526848e-01
8.03393364e-01 4.65105116e-01 1.61965955e-02 1.99559517e-02
7.11364448e-01 6.99168324e-01 8.38471830e-01 -1.92889452e-01
1.02619970e+00 9.36270952e-01 5.04894018e-01 -9.68189478e-01
3.18519063e-02 -5.72607338e-01 -7.50571191e-01 -9.05837417e-01
1.25233674e+00 -1.00836420e+00 -8.51687610e-01 5.86347282e-01
-8.19710433e-01 -6.32970572e-01 -2.13075742e-01 1.92586124e-01
-4.85990107e-01 2.96573997e-01 -1.63574219e-01 -1.13214993e+00
-4.36863303e-01 -1.11386442e+00 1.17026055e+00 3.79457653e-01
1.57443240e-01 -7.33514965e-01 -1.46844372e-01 1.71029970e-01
3.12436998e-01 8.19860876e-01 5.96847951e-01 3.90155762e-01
-6.66173339e-01 -1.84555382e-01 -7.37092257e-01 5.51159382e-01
1.79564785e-02 5.07624388e-01 -9.69045043e-01 4.73489165e-02
-3.50935012e-01 -1.64914340e-01 2.64568955e-01 3.20254833e-01
8.61218274e-01 -2.50847459e-01 2.63368517e-01 9.95195687e-01
2.41619372e+00 -5.41491061e-02 8.93545091e-01 6.91729128e-01
7.25914776e-01 4.71949667e-01 8.17211568e-01 6.68101788e-01
5.08243799e-01 4.41124052e-01 1.00939298e+00 -4.78970885e-01
-2.09397286e-01 2.07494453e-01 2.39926100e-01 2.44822234e-01
-3.29080880e-01 1.16611101e-01 -1.11443782e+00 3.37466210e-01
-1.42408609e+00 -7.29271233e-01 -1.12804687e+00 2.45115924e+00
7.46642768e-01 -6.73179328e-02 -5.32973297e-02 3.07683080e-01
5.28888822e-01 -3.31082433e-01 -3.17092776e-01 -4.86532807e-01
-3.94805819e-01 2.19188705e-01 1.43043244e+00 4.72326428e-01
-9.77976859e-01 5.90966940e-01 5.11982155e+00 5.45450270e-01
-1.16863286e+00 1.10039540e-01 7.72465527e-01 3.93738806e-01
-3.58569592e-01 1.65954664e-01 -1.34500101e-01 4.81870621e-01
7.44371474e-01 2.87266225e-01 6.79810464e-01 8.80011082e-01
4.21444058e-01 -7.63237715e-01 -5.72691679e-01 1.35068607e+00
-3.37296307e-01 -1.12047970e+00 -8.79467279e-02 -1.63277102e-04
8.13424408e-01 4.63063531e-02 -1.24677032e-01 -1.72477484e-01
4.15877134e-01 -8.57377410e-01 1.11098433e+00 8.30068231e-01
1.42964697e+00 -6.44241452e-01 7.99802780e-01 -1.79842889e-01
-1.56088889e+00 4.28633159e-03 -3.42575043e-01 1.10493243e-01
-1.34228125e-01 5.33817649e-01 -6.25059962e-01 7.18818307e-01
1.15856433e+00 -2.30643544e-02 -8.70941401e-01 8.40488136e-01
-2.83228964e-01 5.40924311e-01 -6.46493018e-01 1.73164085e-01
3.57494831e-01 -8.85463178e-01 1.97479367e-01 1.23849678e+00
3.43897432e-01 3.90234470e-01 -2.25803301e-01 9.94330764e-01
5.17918169e-01 2.82956427e-03 -6.54710680e-02 4.33490604e-01
6.08081818e-01 1.59468925e+00 -1.27621329e+00 -4.40436900e-01
-3.22328120e-01 1.14076507e+00 -3.89559865e-01 4.03785050e-01
-6.72604024e-01 -3.98976505e-01 3.71537834e-01 -4.63832058e-02
1.53261438e-01 -8.92607048e-02 -8.25274646e-01 -4.44286942e-01
-8.97313952e-02 -4.64020193e-01 2.35874340e-01 -1.32438135e+00
-9.01843548e-01 3.56190950e-01 -1.80785060e-01 -1.64259231e+00
1.22538060e-01 -7.73418307e-01 -2.39260972e-01 1.04780591e+00
-1.38896143e+00 -1.30560076e+00 -1.18050814e+00 6.78676486e-01
5.42031862e-02 3.51590127e-01 8.81324708e-01 -3.79313454e-02
-2.15500236e-01 1.04542725e-01 4.16065991e-01 8.28546733e-02
2.70643800e-01 -1.81679022e+00 -1.31202057e-01 1.05719244e+00
-4.21597600e-01 -2.39641443e-02 8.25420797e-01 -4.11145985e-01
-1.86126208e+00 -1.51546133e+00 2.29809359e-01 -3.04547220e-01
4.62783933e-01 -1.05546199e-01 -7.23475218e-01 2.88916707e-01
-2.36817390e-01 3.78903389e-01 3.27585906e-01 -9.03937757e-01
-1.25163019e-01 -5.76267362e-01 -1.45494282e+00 2.04418108e-01
2.74048060e-01 -4.71884012e-01 -4.21367437e-02 4.10021454e-01
4.89023328e-01 -3.77973497e-01 -1.32280993e+00 1.41687989e-01
5.75286925e-01 -1.19393623e+00 9.53908324e-01 2.63184190e-01
3.16005528e-01 -1.19760501e+00 -5.53039610e-01 -7.08724797e-01
-1.55705169e-01 -2.74900019e-01 3.34791869e-01 9.60952878e-01
2.38382712e-01 -5.72191834e-01 4.45712537e-01 8.55763435e-01
-1.21228211e-01 -2.58164853e-01 -1.34455550e+00 -7.53123879e-01
-2.73444593e-01 -7.89893627e-01 4.20124412e-01 8.95851135e-01
-5.79496026e-01 -4.94569629e-01 3.06125637e-02 7.37020791e-01
9.50179398e-01 2.07923144e-01 8.75677288e-01 -1.16841853e+00
-7.45949671e-02 -5.24710476e-01 -8.31514478e-01 -4.52291071e-01
-8.95959377e-01 -5.12483478e-01 1.83022767e-01 -1.71059871e+00
-1.61276922e-01 -6.09832048e-01 7.66963640e-04 3.02256584e-01
5.49633913e-02 9.12125289e-01 1.95302311e-02 2.94600040e-01
-4.29575205e-01 1.89344481e-01 6.75004959e-01 -2.00028047e-01
-2.69049883e-01 -3.83636951e-01 -1.55191468e-02 4.75371659e-01
6.01497710e-01 -3.09016198e-01 -9.71155614e-03 -1.90170929e-01
8.70334685e-01 9.23425704e-02 8.52578282e-01 -1.20731676e+00
1.51844680e-01 -3.53807360e-01 6.02307320e-01 -6.81428373e-01
8.75128359e-02 -1.16685450e+00 5.80869913e-01 6.61777139e-01
3.31242800e-01 1.91645801e-01 -2.87358928e-02 2.68716037e-01
2.22562421e-02 4.08210270e-02 1.05921352e+00 -1.76286265e-01
-1.08864462e+00 9.46702510e-02 -3.21953982e-01 -3.23418975e-01
1.18362260e+00 -8.98867190e-01 -1.63046345e-01 -3.06860089e-01
-3.66426766e-01 -2.19449867e-02 8.83765221e-01 -4.05817747e-01
5.38720191e-01 -1.22736084e+00 -7.35803306e-01 8.87911245e-02
4.11556154e-01 2.78543025e-01 3.73611778e-01 8.47427726e-01
-1.56112731e+00 -2.85354406e-02 -1.00794651e-01 -1.12177169e+00
-8.63534331e-01 4.39353675e-01 5.36025465e-01 3.98364156e-01
-6.23850465e-01 6.36190116e-01 -3.57420415e-01 -3.65973681e-01
-1.87470675e-01 -6.26519918e-01 1.20976202e-01 1.02693811e-01
4.44363952e-01 7.87711680e-01 2.64104068e-01 -7.03060031e-01
-3.91104251e-01 6.29502952e-01 1.21792436e+00 -3.05453956e-01
1.24296641e+00 -3.92114162e-01 -5.47813058e-01 5.70019007e-01
1.11373734e+00 -3.29284877e-01 -1.50841451e+00 1.63890392e-01
-2.27936357e-01 -4.84594911e-01 4.22119558e-01 -8.80706847e-01
-9.97626543e-01 4.96332020e-01 1.31858385e+00 4.23006713e-01
1.50002670e+00 -2.98431069e-01 2.42395531e-02 1.92828402e-01
4.63411033e-01 -1.32690167e+00 -5.01353264e-01 1.43118501e-01
7.52135694e-01 -1.31121516e+00 4.15292948e-01 -2.33733445e-01
-5.81944287e-01 1.54553771e+00 2.81777948e-01 3.20037454e-02
5.24214387e-01 1.18381873e-01 2.97498852e-01 -3.31362039e-01
-1.38796726e-02 -2.51669258e-01 4.76886109e-02 6.38490796e-01
7.98522756e-02 4.19395387e-01 1.18344307e-01 -1.80257872e-01
-2.43840620e-01 -2.52590925e-01 5.89260161e-01 6.06886983e-01
-6.31369412e-01 -2.04299390e-01 -1.13609958e+00 3.98578942e-01
2.90944874e-01 -3.66661787e-01 -9.13499296e-03 7.84379840e-01
4.28910643e-01 1.03289604e+00 4.31422293e-01 -1.60198793e-01
3.21965814e-01 -2.76281267e-01 2.30390683e-01 6.59929067e-02
-4.68501925e-01 -1.81796953e-01 -3.81043941e-01 -7.63190866e-01
-5.52603424e-01 -5.53715467e-01 -1.25116062e+00 -2.21917555e-01
-1.43162429e-01 -2.13565499e-01 1.18873870e+00 6.33428693e-01
1.44435372e-02 2.31994063e-01 7.72072911e-01 -9.36302662e-01
-6.31209761e-02 -9.11630690e-01 -7.62592077e-01 3.88621777e-01
2.12590322e-01 -3.62911820e-01 -4.84409779e-01 3.96915793e-01]
|
[10.239108085632324, -2.4366538524627686]
|
33912360-1bad-4c4c-b884-a127c4b490dc
|
hifacegan-face-renovation-via-collaborative
|
2005.05005
| null |
https://arxiv.org/abs/2005.05005v2
|
https://arxiv.org/pdf/2005.05005v2.pdf
|
HiFaceGAN: Face Renovation via Collaborative Suppression and Replenishment
|
Existing face restoration researches typically relies on either the degradation prior or explicit guidance labels for training, which often results in limited generalization ability over real-world images with heterogeneous degradations and rich background contents. In this paper, we investigate the more challenging and practical "dual-blind" version of the problem by lifting the requirements on both types of prior, termed as "Face Renovation"(FR). Specifically, we formulated FR as a semantic-guided generation problem and tackle it with a collaborative suppression and replenishment (CSR) approach. This leads to HiFaceGAN, a multi-stage framework containing several nested CSR units that progressively replenish facial details based on the hierarchical semantic guidance extracted from the front-end content-adaptive suppression modules. Extensive experiments on both synthetic and real face images have verified the superior performance of HiFaceGAN over a wide range of challenging restoration subtasks, demonstrating its versatility, robustness and generalization ability towards real-world face processing applications.
|
['Pan Wang', 'Wen Gao', 'Chang Liu', 'Shanshe Wang', 'Lingbo Yang', 'Siwei Ma', 'Peiran Ren']
|
2020-05-11
| null | null | null | null |
['face-hallucination', 'blind-face-restoration']
|
['computer-vision', 'computer-vision']
|
[ 7.38362134e-01 -1.09634340e-01 2.32797831e-01 -2.93937653e-01
-5.38069606e-01 -2.03652844e-01 6.92373872e-01 -5.75666666e-01
-7.73946196e-02 7.52094209e-01 4.60871875e-01 9.11756456e-02
-1.00608282e-02 -4.67724711e-01 -4.72544312e-01 -9.92558956e-01
2.16462240e-01 -1.50819421e-01 -1.06316373e-01 -4.40012872e-01
1.73317522e-01 4.87056077e-01 -2.04416823e+00 3.64241391e-01
9.61057723e-01 9.31155980e-01 3.87968063e-01 4.54397470e-01
6.06087269e-03 7.28157520e-01 -5.80066860e-01 -5.63691258e-01
5.44178367e-01 -5.41818142e-01 -4.96685028e-01 5.98161995e-01
6.48654461e-01 -3.53997946e-01 -2.74095327e-01 1.23037517e+00
7.29025483e-01 2.05508798e-01 2.31218651e-01 -1.28208208e+00
-7.54995644e-01 1.66890547e-01 -8.79472375e-01 1.71363935e-01
4.01833087e-01 2.14019507e-01 6.30451798e-01 -1.03870738e+00
4.80907679e-01 1.55825233e+00 6.98265672e-01 9.54441071e-01
-1.38439167e+00 -7.09135890e-01 3.80243659e-01 9.20084417e-02
-1.11605728e+00 -1.03860259e+00 8.67412031e-01 -3.74810815e-01
5.33175349e-01 2.07842141e-01 3.32240522e-01 1.08044493e+00
-1.29256621e-01 7.36257434e-01 1.45740974e+00 -3.19837868e-01
2.42251962e-01 -2.43766364e-02 -3.64901841e-01 6.30492926e-01
2.14751735e-01 3.16057980e-01 -7.17168868e-01 8.67384672e-03
8.28271449e-01 -7.24921003e-02 -8.03044856e-01 -2.32032627e-01
-7.44923651e-01 5.72268963e-01 4.12427485e-01 1.68337807e-01
-4.10082906e-01 -1.33354515e-01 2.41278008e-01 3.54036033e-01
8.21991920e-01 -1.17284908e-04 -2.20014259e-01 6.94500983e-01
-1.16137910e+00 1.83461368e-01 4.60199386e-01 7.21677423e-01
7.40691423e-01 3.82304281e-01 -5.28249145e-01 1.12043023e+00
4.37025666e-01 4.27610755e-01 3.45823258e-01 -9.60854650e-01
2.74833024e-01 2.75890499e-01 1.17926873e-01 -1.06610644e+00
-5.58170266e-02 -8.43966603e-01 -1.03266501e+00 5.92903197e-01
1.41317859e-01 1.23766609e-01 -1.37840986e+00 2.17314839e+00
5.78880906e-01 4.71658468e-01 1.61917672e-01 1.13756835e+00
8.03624928e-01 5.52626073e-01 2.41343871e-01 -4.82289910e-01
1.34023595e+00 -1.00844395e+00 -9.38177168e-01 -3.88237238e-01
-7.61141852e-02 -8.74644399e-01 1.02435315e+00 4.51037258e-01
-1.04072773e+00 -7.05590606e-01 -1.06995404e+00 -2.38216184e-02
2.23627891e-02 3.29354078e-01 4.97743130e-01 8.02287638e-01
-1.55089176e+00 5.66355586e-01 -4.49545413e-01 -2.90846556e-01
5.97162604e-01 3.16356391e-01 -4.78082299e-01 -5.65847099e-01
-8.85275841e-01 4.61000949e-01 1.29239902e-01 5.36065996e-01
-1.27827549e+00 -6.45928383e-01 -7.63626039e-01 -1.43315105e-04
5.19670010e-01 -8.97244275e-01 8.74879897e-01 -1.39021647e+00
-1.63854098e+00 7.93393373e-01 -1.61369786e-01 2.23545451e-02
5.95486164e-01 -2.22736254e-01 -4.49646622e-01 1.94737494e-01
1.69302195e-01 5.98237872e-01 1.65966403e+00 -1.70962977e+00
-4.95882452e-01 -7.39931822e-01 9.78359878e-02 3.60723644e-01
-5.28212488e-01 1.42973408e-01 -5.81849992e-01 -1.02380133e+00
1.85926244e-01 -5.53289890e-01 6.62310049e-03 5.41112805e-03
-1.76798612e-01 6.88123703e-02 9.57782924e-01 -1.12165952e+00
1.03917527e+00 -2.15571570e+00 3.88165683e-01 -2.25777440e-02
1.69183403e-01 4.48721170e-01 -5.74435472e-01 1.09815918e-01
-2.33459860e-01 -2.09314585e-01 -6.50464892e-01 -6.75741255e-01
-1.89583778e-01 2.12294087e-01 -3.68906111e-01 5.89396477e-01
3.95815492e-01 5.15617073e-01 -1.01468050e+00 -3.02564919e-01
-4.47325110e-02 8.29994202e-01 -6.58016980e-01 4.26556259e-01
-2.11716086e-01 7.57052422e-01 -1.24767885e-01 8.47514272e-01
1.03715611e+00 7.59447962e-02 2.16961816e-01 -3.71087641e-01
1.46180570e-01 -2.39815027e-01 -1.19186413e+00 1.65924418e+00
-4.77931350e-01 3.34925294e-01 9.42985952e-01 -7.78550267e-01
7.95962572e-01 2.93456912e-01 2.38251895e-01 -7.40338385e-01
-8.71996358e-02 2.47873351e-01 -3.90111506e-01 -4.10589486e-01
3.55716705e-01 -4.53134805e-01 4.70159739e-01 2.50069857e-01
1.99186847e-01 1.45936729e-02 2.24960446e-01 1.04329735e-01
9.67844903e-01 4.05319989e-01 -3.62714641e-02 -3.51486504e-01
1.05350530e+00 -6.22330189e-01 9.10547197e-01 3.64954472e-01
-2.38623977e-01 9.07787263e-01 3.04951966e-01 -4.30481471e-02
-6.86610401e-01 -9.36003566e-01 1.62662819e-01 1.27847302e+00
1.54101387e-01 -3.06080252e-01 -9.84764516e-01 -6.45317435e-01
-2.67685473e-01 2.66015172e-01 -6.68963969e-01 -1.24630585e-01
-5.84495604e-01 -1.10349381e+00 3.38420957e-01 2.63853371e-01
6.63002610e-01 -1.04619956e+00 -1.59281865e-01 3.24470103e-02
-3.89467448e-01 -1.16234994e+00 -5.62597752e-01 -2.36823380e-01
-6.07274950e-01 -1.09512138e+00 -8.54561448e-01 -8.46825242e-01
8.09869885e-01 6.40988469e-01 1.04929197e+00 3.81054074e-01
-3.35656077e-01 3.42428744e-01 -3.00554186e-01 7.94544145e-02
-3.09614450e-01 -5.16891420e-01 5.85482493e-02 6.32875860e-01
-3.95941526e-01 -7.56542265e-01 -7.17985034e-01 2.01094806e-01
-1.26678610e+00 3.35654914e-02 7.56223559e-01 1.17716360e+00
2.56243914e-01 2.55569398e-01 7.26372242e-01 -7.21200883e-01
6.12542450e-01 -3.36246222e-01 -5.59027016e-01 3.18620801e-01
-6.28773987e-01 -1.58450633e-01 5.14644146e-01 -2.35501885e-01
-1.64011681e+00 8.18092655e-03 -1.07133038e-01 -5.11033058e-01
-1.25364223e-02 4.98180576e-02 -7.57043183e-01 -2.28202507e-01
4.52797830e-01 4.28554773e-01 2.05527514e-01 -6.01426125e-01
4.06257629e-01 5.23981094e-01 9.08809125e-01 -6.30005240e-01
9.40145195e-01 6.80429757e-01 -1.01036638e-01 -7.58137167e-01
-7.78182268e-01 -2.49333203e-01 -2.57170767e-01 -3.11845064e-01
6.62503004e-01 -1.22887564e+00 -5.25321007e-01 8.81764770e-01
-1.06146395e+00 -3.26498479e-01 -1.08524524e-01 1.05026057e-02
-5.12306809e-01 6.94886267e-01 -6.85736656e-01 -8.69375885e-01
-4.58204478e-01 -1.12593222e+00 1.18241203e+00 3.44486386e-01
4.64441687e-01 -7.05453932e-01 -2.79378027e-01 7.33925104e-01
5.78590810e-01 8.53686407e-02 7.53093183e-01 5.79061992e-02
-5.66537261e-01 2.29103714e-01 -4.72550541e-01 7.88454771e-01
1.73122764e-01 -3.32490802e-01 -1.25326967e+00 -7.20134914e-01
3.20937783e-01 -2.68269509e-01 1.10070813e+00 1.10847600e-01
1.08269310e+00 -4.38389599e-01 -7.53744021e-02 7.49040127e-01
1.51858020e+00 -1.99340999e-01 8.82084429e-01 5.46917543e-02
6.37687504e-01 9.28414464e-01 5.38770854e-01 4.03306931e-01
8.51980522e-02 7.24411309e-01 6.63747370e-01 -2.05368593e-01
-8.36060524e-01 -1.87023297e-01 6.77633345e-01 5.64108014e-01
4.00957316e-02 -4.10031527e-01 -5.01765549e-01 5.22844076e-01
-1.59469080e+00 -8.61451805e-01 1.72800571e-01 2.04484868e+00
9.51923847e-01 -4.69825894e-01 -1.85460731e-01 2.29995102e-01
1.07553172e+00 2.66592652e-01 -4.35774148e-01 1.11586772e-01
-4.76716220e-01 1.21459886e-01 1.04666211e-01 4.67467487e-01
-1.03165710e+00 1.04126132e+00 6.12272167e+00 1.13406730e+00
-9.88324046e-01 3.42432886e-01 8.73247266e-01 1.08416481e-02
-3.37957084e-01 -1.22389033e-01 -6.49136603e-01 4.49012995e-01
4.11637962e-01 3.70272607e-01 7.46262670e-01 3.03190082e-01
1.57721236e-01 -8.08698535e-02 -6.92965746e-01 1.00302327e+00
4.60939288e-01 -9.27533209e-01 1.86720520e-01 -1.01173870e-01
9.75232482e-01 -3.79990280e-01 2.50387728e-01 1.67435817e-02
3.06211412e-01 -9.76132751e-01 8.25181305e-01 4.25506860e-01
9.52321410e-01 -3.53424162e-01 4.77661729e-01 5.16132824e-02
-9.83834565e-01 -5.11837721e-01 -2.05625594e-01 2.50050068e-01
1.86259165e-01 6.92005217e-01 -3.28487754e-01 8.62625659e-01
7.45764613e-01 7.24265814e-01 -4.97340530e-01 8.53179693e-01
-4.68746901e-01 3.91197503e-01 -8.08104947e-02 9.06209826e-01
-1.58901677e-01 -2.76961744e-01 7.06817448e-01 9.89262342e-01
3.51778299e-01 1.88678890e-01 9.90261286e-02 6.63061261e-01
-1.29380107e-01 -1.29907459e-01 -9.43535268e-02 1.23425812e-01
1.49298996e-01 1.50976300e+00 -8.12527239e-01 -1.25849262e-01
-3.75328273e-01 1.11218274e+00 6.60335049e-02 5.23785591e-01
-8.10708821e-01 2.53159367e-02 7.71584809e-01 1.44918427e-01
4.29867774e-01 1.30488664e-01 -1.97026014e-01 -1.18908072e+00
1.04371578e-01 -1.14145148e+00 2.44486257e-01 -8.47647905e-01
-1.39792109e+00 9.27945971e-01 -2.37621993e-01 -9.86747265e-01
1.50845766e-01 -5.38244069e-01 -4.43742245e-01 8.84267092e-01
-2.00356579e+00 -1.45943356e+00 -5.24841726e-01 9.23774958e-01
6.02816463e-01 -2.45865554e-01 3.83115023e-01 5.68313301e-01
-8.37870598e-01 5.57407558e-01 -1.07660271e-01 -2.22666040e-01
7.97791958e-01 -9.38209534e-01 -1.13881685e-01 1.35552979e+00
-2.70968795e-01 3.64893734e-01 8.54188263e-01 -7.55347729e-01
-1.38450861e+00 -1.37344217e+00 4.94520098e-01 1.33131832e-01
2.66742736e-01 -3.05289388e-01 -8.98159087e-01 3.30708563e-01
2.83672303e-01 1.12574682e-01 3.16949934e-01 -3.16058874e-01
-5.82385957e-01 -3.56359512e-01 -1.30610669e+00 5.26206970e-01
1.53201663e+00 -5.06606638e-01 -2.50085831e-01 2.47292474e-01
5.10534167e-01 -1.08048908e-01 -4.36860800e-01 6.48452342e-01
2.49447644e-01 -1.07859755e+00 1.15000272e+00 -2.51950413e-01
4.11902845e-01 -7.21336544e-01 -4.16261286e-01 -1.17135584e+00
-3.04522485e-01 -7.45685637e-01 -1.40833914e-01 1.58638322e+00
-2.13724405e-01 -6.18742466e-01 5.78620791e-01 8.78861472e-02
-3.05398703e-01 -5.10992348e-01 -7.69735396e-01 -6.13006175e-01
-4.68479633e-01 -2.59503163e-03 7.19064593e-01 7.95716763e-01
-6.76036239e-01 1.86337262e-01 -6.86138391e-01 2.88625956e-01
9.06857073e-01 1.04429554e-02 6.79583013e-01 -1.18768919e+00
-3.07258993e-01 -2.71588147e-01 -3.08573753e-01 -8.57854247e-01
3.41354400e-01 -6.91459477e-01 3.89630377e-01 -1.18665779e+00
2.44299337e-01 -2.30378255e-01 -2.48344943e-01 5.68255007e-01
-4.17710543e-01 7.96541929e-01 1.99779630e-01 2.69380182e-01
-4.30848181e-01 1.00463700e+00 1.31993973e+00 -1.22467369e-01
-3.29431854e-02 -2.53333390e-01 -1.13126183e+00 4.77633417e-01
4.00394678e-01 -1.40523925e-01 -6.95227325e-01 -6.56576037e-01
-9.76148993e-02 4.94807996e-02 6.11463964e-01 -1.02732706e+00
1.40941203e-01 -6.73905760e-02 4.06541198e-01 -7.90689364e-02
5.89720726e-01 -5.01071692e-01 3.16892833e-01 2.42483601e-01
-6.11592904e-02 -2.95730472e-01 1.85786366e-01 8.67207170e-01
-3.64951193e-01 1.64129063e-01 1.07423353e+00 -6.93056434e-02
-8.01878035e-01 3.66242945e-01 -3.47531773e-02 -3.06059927e-01
9.04834509e-01 -3.61033380e-01 -3.40165287e-01 -3.47409546e-01
-9.09722328e-01 -1.07625648e-02 5.02930403e-01 4.16940451e-01
7.57481933e-01 -1.10894310e+00 -1.06556857e+00 6.68594658e-01
-1.69449091e-01 -2.78209805e-01 6.54085577e-01 8.32258701e-01
-1.78181976e-01 2.21639010e-03 -3.05065155e-01 -4.89825487e-01
-1.36102247e+00 5.86513340e-01 3.26583236e-01 -1.76121116e-01
-5.55665970e-01 1.03862643e+00 4.96708840e-01 -1.90416530e-01
1.53636977e-01 4.00788277e-01 -3.08764368e-01 1.61948115e-01
7.74597347e-01 3.57577026e-01 2.11309314e-01 -1.01732576e+00
-1.47306144e-01 4.45549935e-01 -2.97729783e-02 -8.19808096e-02
1.43904054e+00 -5.99818885e-01 -4.36950058e-01 -3.31665784e-01
8.42305183e-01 -8.24080035e-02 -1.67898214e+00 -4.64945048e-01
-1.07296981e-01 -8.67816985e-01 2.70495087e-01 -9.31217372e-01
-1.71178532e+00 6.03407145e-01 8.22776020e-01 -1.25847101e-01
1.87723708e+00 -3.65151703e-01 6.13085508e-01 -2.77683794e-01
2.63302833e-01 -8.49268854e-01 2.66501665e-01 1.06418535e-01
1.29320061e+00 -9.96013403e-01 1.96777172e-02 -8.14841807e-01
-4.05054718e-01 8.12970221e-01 7.14936376e-01 8.63387659e-02
4.82283771e-01 5.62673286e-02 1.29833490e-01 2.22914945e-02
-6.54949009e-01 -3.12181681e-01 1.76192582e-01 6.85828567e-01
2.66812116e-01 -1.96770966e-01 -2.36423805e-01 4.95786786e-01
1.19663082e-01 2.12643538e-02 2.98996866e-01 7.72840679e-01
-2.86064357e-01 -1.17115784e+00 -6.97069526e-01 1.58186182e-02
-4.95688736e-01 -4.59917895e-02 -5.91337979e-02 3.33907723e-01
4.11203116e-01 1.36687636e+00 -3.34883779e-01 -3.46670002e-01
4.49388362e-02 -9.62944031e-02 6.16880715e-01 -5.27266622e-01
-5.50028563e-01 3.68622452e-01 -2.39055589e-01 -7.29259014e-01
-6.84484065e-01 -6.18825078e-01 -6.43500686e-01 -2.23442286e-01
-2.99756080e-01 -3.54677322e-03 5.36088467e-01 8.37666094e-01
5.85433662e-01 5.85297704e-01 7.82466650e-01 -1.10181773e+00
-5.50268650e-01 -1.02245355e+00 -6.21306419e-01 5.59445381e-01
6.18771613e-01 -8.58828485e-01 -3.79360497e-01 2.71286994e-01]
|
[12.826501846313477, -0.1687726080417633]
|
2d23ce78-5575-49dc-a5d8-bec16abf575d
|
smoothed-q-learning
|
2303.08631
| null |
https://arxiv.org/abs/2303.08631v1
|
https://arxiv.org/pdf/2303.08631v1.pdf
|
Smoothed Q-learning
|
In Reinforcement Learning the Q-learning algorithm provably converges to the optimal solution. However, as others have demonstrated, Q-learning can also overestimate the values and thereby spend too long exploring unhelpful states. Double Q-learning is a provably convergent alternative that mitigates some of the overestimation issues, though sometimes at the expense of slower convergence. We introduce an alternative algorithm that replaces the max operation with an average, resulting also in a provably convergent off-policy algorithm which can mitigate overestimation yet retain similar convergence as standard Q-learning.
|
['David Barber']
|
2023-03-15
| null | null | null | null |
['q-learning']
|
['methodology']
|
[-4.90276247e-01 4.05977190e-01 -4.10738409e-01 1.01579584e-01
-1.20606971e+00 -7.87814975e-01 2.85622567e-01 2.11922228e-01
-7.38137245e-01 1.50037348e+00 1.48048550e-02 -7.49433577e-01
-2.46928275e-01 -6.73688650e-01 -5.32642424e-01 -7.45996535e-01
-5.33216558e-02 4.97559637e-01 1.47986248e-01 -8.34503919e-02
5.80228209e-01 4.72025163e-02 -1.37964213e+00 -3.27079237e-01
1.03736305e+00 6.34537935e-01 -1.29963875e-01 7.82472014e-01
-9.59670022e-02 8.33939672e-01 -8.19778681e-01 -8.53822455e-02
4.25027698e-01 -6.49603724e-01 -5.86054146e-01 -1.15669884e-01
3.51253152e-01 -5.65878808e-01 -6.94348291e-03 1.06377399e+00
4.94906873e-01 3.44264001e-01 1.63044170e-01 -1.26406145e+00
-1.78587869e-01 4.39891845e-01 -5.73954999e-01 1.91301525e-01
5.59050560e-01 5.11590660e-01 8.77819538e-01 -2.94809371e-01
3.96203876e-01 1.38711214e+00 5.48724949e-01 4.56761897e-01
-1.26958656e+00 -6.77110076e-01 7.52129704e-02 8.92875418e-02
-1.06586063e+00 -6.32031485e-02 5.76815665e-01 1.41846895e-01
9.65327382e-01 1.06476165e-01 1.22640657e+00 6.32319331e-01
6.24581218e-01 8.80224466e-01 1.62739480e+00 -3.17077368e-01
8.40212882e-01 -5.58139049e-02 -3.38472515e-01 4.74302262e-01
3.31146479e-01 8.55853260e-01 -2.96434730e-01 -4.07554418e-01
8.62867653e-01 -1.31021366e-01 1.00787394e-01 -8.14630151e-01
-7.46791899e-01 1.02225268e+00 9.12604630e-02 -1.34882241e-01
-6.43347442e-01 6.96465611e-01 4.27101076e-01 8.01315784e-01
1.06728874e-01 8.09830725e-01 -6.10780895e-01 -8.79300833e-01
-9.13671613e-01 1.02970254e+00 8.73149812e-01 2.87190020e-01
8.39292169e-01 4.55164194e-01 -1.64267942e-02 2.92521030e-01
1.37069345e-01 4.73934889e-01 3.96411180e-01 -1.57570910e+00
2.15216368e-01 1.96399420e-01 8.95340383e-01 -3.12261045e-01
-3.68545562e-01 -8.04229736e-01 -1.71025679e-01 9.94438112e-01
7.33928382e-01 -7.14319527e-01 -5.13755560e-01 1.79918337e+00
3.95232290e-01 -2.50485182e-01 1.03311487e-01 7.58447647e-01
-4.42187011e-01 7.31542945e-01 2.12471142e-01 -8.02063465e-01
6.06345415e-01 -8.47811997e-01 -1.00270116e+00 -1.96200132e-01
3.72173011e-01 -4.24533069e-01 1.17637706e+00 7.38171279e-01
-1.53694272e+00 -2.66014814e-01 -1.05231035e+00 7.20274508e-01
-6.12174533e-02 -7.45978951e-01 6.21842384e-01 9.65195358e-01
-1.11063349e+00 1.07040858e+00 -7.95255959e-01 2.04620659e-02
3.68363798e-01 4.60153848e-01 2.35054106e-01 2.49602646e-01
-1.16606927e+00 1.43019235e+00 7.10105717e-01 -3.19847912e-01
-8.34911644e-01 -8.19594264e-01 -5.42656839e-01 -2.88692787e-02
9.51136291e-01 -4.87475961e-01 1.76538396e+00 -1.28724074e+00
-1.88235986e+00 -9.79304239e-02 1.59097686e-01 -6.21021330e-01
9.28578496e-01 -1.84506744e-01 -1.19857006e-01 -8.86927396e-02
-4.39006127e-02 6.56793892e-01 8.74322653e-01 -1.11545169e+00
-6.20765328e-01 -3.95431250e-01 2.36826763e-01 7.48238623e-01
-9.80291814e-02 -2.40870535e-01 5.06866932e-01 -1.82132483e-01
-2.40714192e-01 -8.95604491e-01 -7.39878356e-01 -3.40113416e-02
1.84434712e-01 -3.61507326e-01 5.51272154e-01 -1.27632022e-01
1.13863742e+00 -1.62565231e+00 -2.14683041e-01 2.36778423e-01
-2.26876754e-02 1.17554121e-01 -1.95189789e-01 7.39995778e-01
1.22253656e-01 -3.77419442e-02 -1.34930192e-02 3.26650411e-01
-3.45221907e-03 6.12810671e-01 -1.42161876e-01 5.30300915e-01
-1.16791315e-01 9.56354141e-01 -1.52751482e+00 -3.36079001e-01
3.34005386e-01 -1.55321330e-01 -7.50443697e-01 7.89415240e-02
-3.59293073e-01 3.47460538e-01 -3.62695456e-01 4.31024164e-01
4.70904648e-01 8.22817609e-02 4.90441144e-01 7.30574608e-01
-3.31541479e-01 2.20028058e-01 -1.47551322e+00 1.46129048e+00
-3.46595496e-01 2.74672598e-01 2.05533832e-01 -1.09943342e+00
6.99115396e-01 3.90810043e-01 8.10786307e-01 -9.04375494e-01
5.43913879e-02 3.33942056e-01 4.10596430e-02 -2.74463713e-01
4.54206556e-01 -6.03908241e-01 6.75892681e-02 8.17167878e-01
-1.86021671e-01 -5.52186966e-01 1.48137599e-01 -6.99045733e-02
9.11014438e-01 4.94197875e-01 4.86639142e-01 -6.52847350e-01
1.62614062e-01 1.72415063e-01 6.55421674e-01 1.09244847e+00
-7.54597187e-01 -1.93182781e-01 7.89681017e-01 -6.56961918e-01
-1.02807927e+00 -1.23345041e+00 1.01728126e-01 1.12202334e+00
2.18110338e-01 -3.62811923e-01 -4.94774520e-01 -7.91746736e-01
4.73291874e-01 8.99300218e-01 -4.87014145e-01 -2.49889866e-01
-4.05523330e-01 -6.52407050e-01 1.57066137e-01 3.63963127e-01
2.91941226e-01 -1.12394834e+00 -1.17118835e+00 5.99922419e-01
1.86003193e-01 -2.66165823e-01 -2.49906927e-01 2.95586973e-01
-1.24760020e+00 -8.71525347e-01 -6.48885846e-01 -7.03104883e-02
4.54802543e-01 -8.73940289e-02 1.13608479e+00 -9.06832069e-02
2.35106528e-01 3.09509337e-01 3.24086770e-02 -5.61524212e-01
-6.40392661e-01 -2.14858994e-01 2.14941297e-02 -9.09673274e-01
1.81529865e-01 -3.73747975e-01 -6.77725494e-01 5.58780655e-02
-7.93190598e-01 -3.52932334e-01 2.81947166e-01 9.95414913e-01
4.14490938e-01 1.09772071e-01 1.24174941e+00 -6.68002307e-01
1.21057367e+00 -3.05392653e-01 -1.03558135e+00 -1.38628734e-02
-1.35952199e+00 5.04849255e-01 7.89727032e-01 -6.30115688e-01
-9.99911308e-01 -5.02360873e-02 -2.34544799e-02 -4.19357091e-01
2.42878422e-01 3.71401817e-01 4.60675836e-01 -2.25493625e-01
8.26405525e-01 2.58390218e-01 5.43227136e-01 -1.35047287e-01
3.98219287e-01 2.43980750e-01 6.61282390e-02 -7.77942896e-01
6.11251712e-01 2.53972840e-02 -2.67464053e-02 -2.13393554e-01
-6.86728179e-01 -1.20443940e-01 3.55839953e-02 -4.19400960e-01
2.49332208e-02 -6.85799241e-01 -9.08541858e-01 -6.47504535e-03
-5.67807674e-01 -5.85736811e-01 -8.03890049e-01 3.98716152e-01
-1.10857713e+00 3.58710259e-01 -4.00791585e-01 -1.25005174e+00
-1.64972946e-01 -8.18932414e-01 2.47222260e-01 5.67694902e-01
-2.66609222e-01 -8.05664241e-01 5.17668068e-01 -4.03322838e-02
5.58028758e-01 1.18082166e-01 5.77940583e-01 -3.27940047e-01
-2.30370104e-01 -1.14010304e-01 2.41930783e-01 7.87593648e-02
-1.35744065e-01 -1.67555302e-01 -4.50902671e-01 -7.69232869e-01
-1.90130070e-01 -7.70181358e-01 4.24798608e-01 5.78187346e-01
7.44661272e-01 -4.96188909e-01 2.17838258e-01 -5.54898307e-02
1.74394214e+00 6.27583027e-01 4.64317501e-01 7.55188107e-01
-2.15207234e-01 4.24441397e-01 1.06603086e+00 8.68395448e-01
1.82979196e-01 2.07945496e-01 6.16667509e-01 8.27731565e-02
5.24027407e-01 -5.34792721e-01 5.80656230e-01 2.34848619e-01
9.84319821e-02 2.45015845e-01 -6.96332514e-01 5.37264109e-01
-2.05695581e+00 -1.18448377e+00 3.49888772e-01 2.24624968e+00
1.09955394e+00 4.51720774e-01 6.90670192e-01 1.63794279e-01
2.54883677e-01 1.00420907e-01 -1.08078325e+00 -1.08030951e+00
3.82118136e-01 3.80736589e-01 7.93796182e-01 6.77446723e-01
-5.58164001e-01 8.47649753e-01 8.33063221e+00 7.23417044e-01
-8.92733574e-01 6.41122684e-02 4.02445763e-01 -4.09596294e-01
-3.52530897e-01 3.13618243e-01 -2.93216616e-01 4.90387022e-01
1.20026433e+00 -6.85476005e-01 8.20509434e-01 9.85848606e-01
5.86257875e-01 -5.93140185e-01 -5.74561596e-01 6.57672524e-01
-5.69646299e-01 -1.03098357e+00 -3.41096163e-01 3.57682481e-02
1.05552769e+00 -1.73750415e-01 -7.21897930e-02 7.65620530e-01
6.35551095e-01 -7.70771980e-01 8.24581087e-01 9.83617008e-02
5.13592899e-01 -1.49442399e+00 6.88475251e-01 7.01109767e-01
-5.64745307e-01 -6.61732137e-01 -4.63454932e-01 -6.48294449e-01
5.95723577e-02 3.86316746e-01 -5.84028840e-01 1.94118440e-01
2.53070563e-01 3.19312513e-02 -1.15065418e-01 1.19600773e+00
-4.53429013e-01 5.41650951e-01 -3.39517862e-01 -4.38513994e-01
7.37661302e-01 -4.66205716e-01 4.03376639e-01 7.12912917e-01
1.03640318e-01 -5.25961332e-02 4.26440924e-01 1.01673090e+00
4.41150159e-01 1.11037165e-01 -4.34686720e-01 -6.27585649e-02
6.14791632e-01 8.82050335e-01 -4.86836344e-01 -4.61370021e-01
-6.09949827e-02 4.28889841e-01 2.56485492e-01 4.21507537e-01
-7.47235358e-01 -2.92869061e-01 5.74160516e-01 -2.44502518e-02
1.17803261e-01 -3.38946313e-01 -1.35120571e-01 -7.53078282e-01
-3.13266039e-01 -1.03765643e+00 4.08375651e-01 -2.55508274e-01
-7.51819432e-01 -1.25733897e-01 8.74352232e-02 -1.01768529e+00
-8.80317867e-01 -9.66284573e-02 -4.99801219e-01 6.17620051e-01
-1.49819362e+00 -2.13512063e-01 3.65650862e-01 3.59859943e-01
5.32793939e-01 2.05981284e-01 6.37645841e-01 -2.31446728e-01
-3.13403964e-01 3.30895245e-01 4.92504537e-01 -6.44439161e-01
4.98203188e-01 -1.68759823e+00 -8.41992944e-02 3.31018269e-01
-2.63759106e-01 3.55394483e-01 1.10331190e+00 -5.66406488e-01
-1.42524493e+00 -6.52370393e-01 2.98583746e-01 -1.10091180e-01
7.60403395e-01 2.60077357e-01 -7.81081080e-01 3.83062601e-01
3.41728747e-01 -3.05824339e-01 3.01849961e-01 1.70254871e-01
8.98103118e-02 -2.00343639e-01 -1.20188808e+00 7.19351053e-01
4.14242208e-01 -3.21523458e-01 -6.71919942e-01 2.51291245e-01
2.61134923e-01 -3.82639647e-01 -8.54105473e-01 7.32363015e-02
7.40087748e-01 -1.19864750e+00 6.92005038e-01 -5.48738658e-01
2.57481001e-02 -7.99558237e-02 5.28830886e-01 -1.60499358e+00
-1.60098329e-01 -1.24051476e+00 -3.91820818e-01 3.80884498e-01
1.47731006e-01 -7.62248456e-01 1.07246232e+00 5.39258957e-01
3.55432659e-01 -9.06662524e-01 -1.22667706e+00 -1.32297730e+00
6.13295078e-01 -3.18345338e-01 3.98524195e-01 6.25066936e-01
6.04254901e-01 -1.34987473e-01 -3.37907344e-01 -4.17497009e-01
1.08826900e+00 1.15862191e-01 4.44603890e-01 -6.72916293e-01
-3.50093365e-01 -5.47089696e-01 6.90352321e-02 -9.58659947e-01
5.61462156e-02 -2.15367049e-01 2.54259497e-01 -1.13792992e+00
6.62119761e-02 -3.60786498e-01 -7.51776397e-01 3.51029366e-01
-2.45629296e-01 -1.96194366e-01 1.93073764e-01 -7.24240839e-02
-8.50971520e-01 6.33838713e-01 1.65841138e+00 3.01147550e-01
-5.44388831e-01 9.15859714e-02 -6.60015047e-01 4.75614876e-01
1.05564344e+00 -6.28030777e-01 -7.22906888e-01 1.81339756e-01
3.98195028e-01 7.26864100e-01 1.21260129e-01 -8.10461402e-01
-4.63708602e-02 -8.42254817e-01 2.50822097e-01 -4.58889782e-01
-1.08056843e-01 -5.14822721e-01 -2.63127033e-02 1.13215518e+00
-5.18748462e-01 3.09586287e-01 3.41316462e-01 5.64199209e-01
2.37930696e-02 -4.30394262e-01 9.80643749e-01 -4.95880663e-01
-3.23438764e-01 2.25185007e-02 -1.00565743e+00 1.70774281e-01
1.15104246e+00 -3.85723025e-01 -2.08327151e-03 -8.37202847e-01
-5.30282438e-01 5.57532668e-01 5.10832071e-01 1.75748974e-01
4.46833938e-01 -1.02885699e+00 -3.52869302e-01 -3.34308157e-03
-4.56576407e-01 -5.32740414e-01 -1.14823483e-01 5.12727022e-01
-3.64516556e-01 4.02648687e-01 -4.79674518e-01 -7.90307745e-02
-7.52720952e-01 5.35387218e-01 7.17261612e-01 -5.16853452e-01
-4.91790742e-01 2.78918028e-01 -6.91425741e-01 -3.86273980e-01
2.34773293e-01 3.76038998e-02 1.84815586e-01 -4.89844866e-02
3.74482334e-01 5.86305559e-01 -3.07068825e-01 3.68704021e-01
-3.51080112e-02 2.22880423e-01 5.55156507e-02 -5.94267428e-01
1.11945379e+00 -2.90349182e-02 2.88102925e-01 3.41339827e-01
6.02025330e-01 -3.06714058e-01 -1.88166547e+00 1.88849851e-01
2.61941493e-01 -6.13656461e-01 3.47210169e-01 -9.89221871e-01
-5.88527501e-01 5.39194703e-01 7.61006474e-01 2.78898448e-01
8.42491746e-01 -6.62174642e-01 6.86048567e-01 5.97877979e-01
6.72173321e-01 -1.80624354e+00 1.46370038e-01 2.67209411e-01
4.46695805e-01 -1.02883494e+00 3.26289505e-01 4.70261425e-01
-7.05911696e-01 1.14274871e+00 7.05566406e-01 -5.17131269e-01
2.33980700e-01 3.45392346e-01 1.49445772e-01 1.36715263e-01
-1.16918039e+00 -1.25553653e-01 -6.41879618e-01 5.72895586e-01
4.81084622e-02 8.39617848e-02 -8.67098987e-01 -1.16604142e-01
2.55341316e-03 2.81763166e-01 6.18700624e-01 1.28715396e+00
-8.16674590e-01 -1.35317755e+00 -5.22148907e-01 3.21220607e-01
-4.38111603e-01 3.22090805e-01 5.00896573e-02 9.01392043e-01
-2.17920721e-01 8.99027824e-01 1.42300189e-01 1.22697510e-01
1.73847988e-01 6.95788488e-02 8.16934764e-01 -7.90547673e-03
-8.25475931e-01 2.94956535e-01 3.00505850e-02 -8.87090802e-01
-1.93954825e-01 -6.36494756e-01 -1.40482175e+00 -4.45774585e-01
-2.97710836e-01 5.14056802e-01 5.33100963e-01 8.30806196e-01
-5.49939834e-02 2.89526492e-01 8.28539371e-01 -4.19402808e-01
-1.69887221e+00 -6.45591140e-01 -7.86372185e-01 3.05570990e-01
4.55701083e-01 -6.59224153e-01 -4.12986249e-01 -7.63224065e-01]
|
[4.153663635253906, 2.3353660106658936]
|
91288513-1538-4dac-a9dd-e7e087259bcf
|
crossnet-an-end-to-end-reference-based-super
|
1807.10547
| null |
http://arxiv.org/abs/1807.10547v1
|
http://arxiv.org/pdf/1807.10547v1.pdf
|
CrossNet: An End-to-end Reference-based Super Resolution Network using Cross-scale Warping
|
The Reference-based Super-resolution (RefSR) super-resolves a low-resolution
(LR) image given an external high-resolution (HR) reference image, where the
reference image and LR image share similar viewpoint but with significant
resolution gap x8. Existing RefSR methods work in a cascaded way such as patch
matching followed by synthesis pipeline with two independently defined
objective functions, leading to the inter-patch misalignment, grid effect and
inefficient optimization. To resolve these issues, we present CrossNet, an
end-to-end and fully-convolutional deep neural network using cross-scale
warping. Our network contains image encoders, cross-scale warping layers, and
fusion decoder: the encoder serves to extract multi-scale features from both
the LR and the reference images; the cross-scale warping layers spatially
aligns the reference feature map with the LR feature map; the decoder finally
aggregates feature maps from both domains to synthesize the HR output. Using
cross-scale warping, our network is able to perform spatial alignment at
pixel-level in an end-to-end fashion, which improves the existing schemes both
in precision (around 2dB-4dB) and efficiency (more than 100 times faster).
|
['Lu Fang', 'Mengqi Ji', 'Haitian Zheng', 'Yebin Liu', 'Haoqian Wang']
|
2018-07-27
|
crossnet-an-end-to-end-reference-based-super-1
|
http://openaccess.thecvf.com/content_ECCV_2018/html/Haitian_Zheng_CrossNet_An_End-to-end_ECCV_2018_paper.html
|
http://openaccess.thecvf.com/content_ECCV_2018/papers/Haitian_Zheng_CrossNet_An_End-to-end_ECCV_2018_paper.pdf
|
eccv-2018-9
|
['reference-based-super-resolution', 'patch-matching']
|
['computer-vision', 'computer-vision']
|
[ 5.71343482e-01 -1.11668937e-01 2.80490667e-02 -3.36563438e-01
-1.62311184e+00 -2.98124731e-01 4.39199477e-01 -6.06281221e-01
-2.62060672e-01 7.39233136e-01 4.07128632e-01 2.27774903e-01
-4.62668389e-02 -8.07049692e-01 -8.76659274e-01 -5.91177166e-01
2.44340077e-01 -1.67618081e-01 5.86698472e-01 -3.91086370e-01
2.26210698e-01 6.63863897e-01 -1.57120502e+00 6.04356587e-01
6.12189114e-01 1.19337094e+00 4.81168777e-01 8.67675781e-01
3.99855256e-01 6.14158452e-01 -3.17868888e-01 -6.31406307e-02
7.86244810e-01 -3.31621140e-01 -7.27889299e-01 -1.87836021e-01
1.12707067e+00 -4.78740722e-01 -6.73653483e-01 1.16043282e+00
7.79856324e-01 2.02378333e-01 -4.38318700e-02 -4.02801871e-01
-1.09622002e+00 2.17202157e-01 -1.17262471e+00 4.23772305e-01
3.28287989e-01 1.84103504e-01 8.22272837e-01 -1.11833739e+00
8.29536676e-01 1.55493999e+00 6.47315502e-01 3.61225843e-01
-1.54876614e+00 -7.23268330e-01 -3.68346691e-01 -2.28832345e-02
-1.65355372e+00 -5.44059694e-01 6.26772404e-01 -7.70919630e-03
9.98256087e-01 1.79246366e-01 5.26289046e-02 9.48175311e-01
3.20823640e-01 -1.04864307e-01 1.56753671e+00 -1.18783891e-01
-1.81589380e-01 -2.25309625e-01 -3.22917163e-01 2.91084021e-01
-2.51269430e-01 6.23968005e-01 -6.64574206e-01 2.93426901e-01
1.55619037e+00 -7.59928077e-02 -5.10291338e-01 1.07088141e-01
-1.37549758e+00 4.54728484e-01 9.55741107e-01 2.81035811e-01
-3.69110852e-01 -1.42988786e-02 -6.27399608e-02 5.11284769e-01
3.54945570e-01 4.26856518e-01 -4.07792240e-01 4.35093254e-01
-1.13332379e+00 8.23856071e-02 -1.95782874e-02 7.04600573e-01
9.34781432e-01 -1.13258744e-02 -3.37133527e-01 1.05419564e+00
-1.35187255e-02 4.77102607e-01 5.21051288e-01 -1.27762496e+00
5.98688841e-01 9.63158980e-02 2.35508502e-01 -1.05262053e+00
-2.61192918e-01 -5.93744099e-01 -1.16506791e+00 5.67187369e-01
1.74824577e-02 1.73984230e-01 -1.03497267e+00 1.79978466e+00
2.63137877e-01 2.55043298e-01 3.13188225e-01 1.49056125e+00
9.09378886e-01 6.54911757e-01 -3.04334044e-01 -2.94326484e-01
1.54230750e+00 -1.11526036e+00 -5.62415540e-01 -2.41414785e-01
-3.11880678e-01 -1.14471030e+00 9.19061899e-01 1.07291505e-01
-1.56729853e+00 -1.22692752e+00 -1.31400859e+00 -8.70602071e-01
-1.19421504e-01 2.23454490e-01 -1.87348932e-01 -9.48064625e-02
-1.45240402e+00 6.60597324e-01 -4.18471754e-01 -1.20849013e-01
2.97008604e-01 2.33271673e-01 -7.12542117e-01 -1.80298179e-01
-1.17320096e+00 9.64855731e-01 1.31696224e-01 -1.06883990e-02
-5.70983648e-01 -9.83154058e-01 -8.96031499e-01 3.56429704e-02
2.10225016e-01 -8.74544442e-01 8.02582204e-01 -8.92487705e-01
-1.56560230e+00 1.02956438e+00 -1.81912750e-01 -3.60805839e-01
4.53901201e-01 -1.99497744e-01 -7.62535989e-01 1.67256221e-01
2.81348795e-01 6.96637273e-01 1.00978184e+00 -1.26787031e+00
-9.71139550e-01 -4.81682628e-01 -2.17927337e-01 3.72787982e-01
5.05587101e-01 2.02044830e-01 -5.94336450e-01 -8.03080559e-01
3.42929900e-01 -4.85743672e-01 -1.77643761e-01 -6.32867366e-02
-2.15289026e-01 4.95657504e-01 9.42072570e-01 -8.62299263e-01
9.48109269e-01 -2.36008739e+00 1.18394807e-01 -2.03845501e-01
4.42072570e-01 1.44426480e-01 -4.38200951e-01 -5.56628108e-02
-5.46782374e-01 -2.90263265e-01 -1.70138672e-01 -2.75235653e-01
-4.82318819e-01 -2.42475286e-01 -5.81059337e-01 6.05853975e-01
5.92445195e-01 8.95801008e-01 -7.28769362e-01 -3.06620836e-01
4.27926928e-01 1.00738919e+00 -1.74809620e-01 4.08540040e-01
2.98964202e-01 7.52674162e-01 3.48422565e-02 5.12896180e-01
1.10172725e+00 -3.01226616e-01 -4.26294133e-02 -8.85391533e-01
-5.18008947e-01 2.43891180e-01 -1.38164151e+00 1.86176753e+00
-6.54854834e-01 7.60318696e-01 2.31763184e-01 -3.77828747e-01
1.36116254e+00 -1.98063403e-02 2.66140908e-01 -1.32066357e+00
-2.12560296e-01 3.03125799e-01 -3.22826982e-01 -4.00119647e-02
8.25605452e-01 -6.44728839e-02 -5.14974222e-02 9.23052207e-02
8.61344263e-02 4.89435196e-02 -2.79358894e-01 -2.45227277e-01
9.37176883e-01 1.91202492e-01 3.43307644e-01 9.81500465e-03
7.35591292e-01 -3.95167232e-01 4.80626583e-01 4.18270499e-01
1.33865848e-01 1.46608973e+00 1.29720956e-01 -5.88180959e-01
-1.55355763e+00 -1.38741791e+00 -2.73674756e-01 7.71084309e-01
3.90432715e-01 -1.53557226e-01 -4.93558556e-01 -1.26710206e-01
-3.03713799e-01 2.90286571e-01 -6.37655675e-01 1.86071657e-02
-9.67377663e-01 -4.73884284e-01 1.36303321e-01 5.33206820e-01
1.05168867e+00 -7.22275078e-01 -5.77624142e-01 1.76562831e-01
-2.70778835e-01 -1.40214109e+00 -7.77227104e-01 -3.94378090e-03
-5.80632925e-01 -8.66623461e-01 -7.28631258e-01 -6.16756201e-01
3.82471859e-01 6.34342670e-01 1.24496555e+00 -2.72787362e-01
-6.00667536e-01 -2.52224535e-01 2.12092698e-02 5.88888645e-01
-1.71808347e-01 -1.74824193e-01 -2.34251589e-01 1.73876807e-01
-7.00751320e-02 -8.78637373e-01 -1.03946948e+00 5.89473248e-01
-8.35143268e-01 3.07283044e-01 8.89660180e-01 9.18689311e-01
1.13081956e+00 4.56418768e-02 2.94834048e-01 -3.94269675e-01
1.94745123e-01 -5.25302440e-02 -9.85138416e-01 8.60554203e-02
-5.17670631e-01 5.48254438e-02 6.79379404e-01 -1.64257839e-01
-1.18091083e+00 1.22478746e-01 -1.13609992e-01 -7.30039895e-01
-1.47419944e-02 -2.67840207e-01 -1.82555810e-01 -2.99516588e-01
8.07451189e-01 2.81700253e-01 -1.29876003e-01 -6.09888077e-01
6.74928188e-01 6.08079851e-01 1.32165015e+00 -1.67726502e-01
1.07143700e+00 5.86218894e-01 1.19894221e-01 -5.23790419e-01
-9.31574821e-01 -2.03514248e-01 -8.64170015e-01 8.27292949e-02
1.12531197e+00 -1.49386907e+00 -3.75194877e-01 1.35129303e-01
-9.90262926e-01 -2.04059228e-01 -3.26614678e-01 4.00124878e-01
-6.78900301e-01 2.37587932e-02 -6.20809674e-01 -2.39125207e-01
-6.02772176e-01 -1.25440061e+00 1.59256542e+00 5.88629782e-01
2.05128118e-01 -3.85816813e-01 1.59749482e-02 3.48322630e-01
8.91169310e-01 3.56269240e-01 4.72259939e-01 1.15426004e-01
-1.04750347e+00 2.98542231e-01 -1.06674802e+00 2.83863723e-01
-1.02994284e-02 -1.68539524e-01 -1.03937912e+00 -5.24767697e-01
1.72929298e-02 -1.68929160e-01 8.74122202e-01 4.38710719e-01
9.98055339e-01 -1.09906368e-01 -1.69456787e-02 1.25336838e+00
1.99375057e+00 -1.32210240e-01 1.10170829e+00 4.65571910e-01
6.93138301e-01 3.84502381e-01 7.35970676e-01 4.50027129e-03
3.22908878e-01 1.37347758e+00 2.41504118e-01 -6.25098765e-01
-8.86552691e-01 4.44925837e-02 3.91992420e-01 1.25777423e-01
-9.51348394e-02 3.22429001e-01 -3.90552491e-01 3.70439589e-01
-1.58955741e+00 -1.17580080e+00 5.57155274e-02 2.49940181e+00
1.14191461e+00 -1.85787350e-01 -1.14475891e-01 -1.93186775e-01
7.72693872e-01 5.07884860e-01 -6.40081167e-01 -1.82633534e-01
-4.68744308e-01 5.53750396e-01 7.33358264e-01 9.27812040e-01
-9.83169496e-01 9.43999529e-01 5.85321426e+00 9.37791526e-01
-1.48081970e+00 2.59345204e-01 8.82940352e-01 -4.16985899e-01
5.36924899e-02 -2.81925172e-01 -8.57298851e-01 2.26902932e-01
8.70547891e-01 1.24733649e-01 7.23818243e-01 5.61128438e-01
1.13685869e-01 5.22847138e-02 -8.91379416e-01 1.30320752e+00
9.47017968e-02 -1.40980077e+00 -3.30847353e-01 -7.73277804e-02
8.79655302e-01 3.18107188e-01 3.68340939e-01 -2.55894456e-02
2.97168642e-01 -1.40044510e+00 4.46530700e-01 5.59830070e-01
1.67384541e+00 -9.51204479e-01 4.86139476e-01 -2.02164888e-01
-1.45000708e+00 -5.84547967e-02 -5.54768801e-01 5.17530382e-01
1.56819269e-01 4.07425225e-01 -3.12340021e-01 7.84748018e-01
1.07799411e+00 5.38909853e-01 -6.12793207e-01 3.18217933e-01
9.54944082e-03 -3.46538693e-01 1.47799943e-02 1.19302130e+00
-1.41373295e-02 -2.50048488e-01 6.04634464e-01 1.07688725e+00
4.40298140e-01 2.59832174e-01 -1.30960301e-01 1.15434909e+00
-3.80678065e-02 -5.19726515e-01 -2.81465858e-01 6.34183586e-01
5.33270240e-01 1.65190232e+00 -3.72192323e-01 -1.51161656e-01
-5.09478271e-01 1.34791231e+00 3.31743419e-01 3.68260443e-01
-8.97597849e-01 -5.57387471e-01 8.83620560e-01 3.35048527e-01
4.69571292e-01 -3.67444903e-02 -3.79908174e-01 -1.03791142e+00
8.27647895e-02 -9.93670404e-01 1.33903503e-01 -1.18843520e+00
-1.16874540e+00 1.03004956e+00 -4.47224975e-01 -1.34269381e+00
-2.17431217e-01 -2.02093512e-01 -1.77572504e-01 1.63043702e+00
-1.91450834e+00 -1.08151901e+00 -6.50441825e-01 7.48708367e-01
5.65322518e-01 -1.02772810e-01 5.70036829e-01 4.40048695e-01
-3.75266910e-01 6.55073702e-01 1.13918319e-01 1.51641443e-01
1.11136675e+00 -1.11324692e+00 7.20591903e-01 1.05983865e+00
-2.13700101e-01 5.57797968e-01 6.37150764e-01 -4.34923410e-01
-1.30563176e+00 -1.54143763e+00 7.29575574e-01 -4.25035387e-01
2.27711439e-01 -1.56405494e-01 -1.11457717e+00 3.72636169e-01
1.10752933e-01 6.92560315e-01 1.15893736e-01 -3.94159108e-01
-7.69860625e-01 -5.92930079e-01 -1.39325213e+00 5.87873518e-01
1.04893553e+00 -7.89277673e-01 -4.60541934e-01 -1.60847455e-01
1.14996552e+00 -9.56519961e-01 -1.23594666e+00 4.85562563e-01
5.44511199e-01 -1.38257372e+00 1.59750068e+00 7.75144994e-02
9.10936117e-01 -8.49218011e-01 -5.42460442e-01 -9.47390974e-01
-7.99596548e-01 -6.02908134e-01 5.67898750e-02 1.02399302e+00
1.67874083e-01 -6.19968712e-01 2.04885229e-01 2.56483823e-01
-5.43838106e-02 -6.63806498e-01 -1.05311155e+00 -5.22187471e-01
-3.39211524e-01 2.02491611e-01 7.06532478e-01 7.86583722e-01
-7.26224601e-01 5.68753362e-01 -5.14977038e-01 7.17206061e-01
8.84231269e-01 4.80054498e-01 6.29500270e-01 -7.86868811e-01
-3.16512525e-01 -4.18918461e-01 -1.71421543e-01 -9.71672356e-01
-3.25760305e-01 -4.11970049e-01 -5.33039868e-03 -1.18759465e+00
9.56214741e-02 -5.65839671e-02 -3.01266611e-01 1.25149131e-01
-1.65571079e-01 8.67250621e-01 1.90566644e-01 2.38925979e-01
-4.13999766e-01 1.94139257e-01 1.44203568e+00 3.98943245e-01
-2.72845328e-01 -5.18742681e-01 -8.25028002e-01 3.34462941e-01
3.50458980e-01 -8.02547187e-02 -7.88833499e-02 -5.95586419e-01
-2.91227009e-02 4.48074490e-01 5.98897815e-01 -1.12792027e+00
1.17306828e-01 3.84900160e-02 1.06196952e+00 -6.69966042e-01
5.38043678e-01 -7.23996222e-01 4.74702895e-01 -6.46698996e-02
-5.85242569e-01 1.04636513e-01 7.53995439e-04 3.82522434e-01
-4.12187815e-01 5.95526814e-01 1.58173573e+00 2.75018741e-03
-5.36716163e-01 5.54889917e-01 5.46454966e-01 -1.41032159e-01
7.92277813e-01 -2.60604978e-01 -6.17365479e-01 -5.29900528e-02
-4.90668923e-01 -1.21233985e-01 8.09361100e-01 5.67389905e-01
7.42692888e-01 -1.45986640e+00 -1.01378918e+00 5.39089382e-01
-1.97311118e-01 3.39687854e-01 7.03925908e-01 8.21193993e-01
-4.19444919e-01 3.37424874e-01 -5.61569571e-01 -5.78894675e-01
-1.24418259e+00 5.90994239e-01 5.67916393e-01 -3.22176695e-01
-9.26672876e-01 8.25820267e-01 4.45651650e-01 -7.33513162e-02
-8.97118375e-02 -1.48321707e-02 -2.27444828e-01 -2.34211445e-01
1.14818001e+00 3.01763833e-01 3.38622145e-02 -9.55494761e-01
-1.45173281e-01 1.19440138e+00 -2.88400143e-01 -2.46912286e-01
1.53022015e+00 -5.97019732e-01 -1.28525048e-01 6.92011341e-02
1.51797032e+00 1.11605376e-01 -1.65901649e+00 -7.67767191e-01
-5.05782783e-01 -9.31218505e-01 3.78554195e-01 -8.85915160e-01
-1.29658985e+00 6.20104969e-01 1.03847027e+00 -1.37072995e-01
1.77981806e+00 -3.38970721e-02 8.39330554e-01 -4.18656975e-01
2.22735927e-01 -8.87072265e-01 2.94400882e-02 2.67690897e-01
1.22149777e+00 -1.27532029e+00 1.31622642e-01 -3.04347426e-01
-5.21995425e-01 1.18549597e+00 6.58175290e-01 -3.94691139e-01
1.70054287e-01 4.97182935e-01 9.83187109e-02 2.91465241e-02
-9.29145694e-01 -2.71536529e-01 4.74339455e-01 7.13278651e-01
3.84904653e-01 -1.48511827e-01 2.92815641e-02 4.55568284e-01
-3.59668702e-01 -7.87133798e-02 3.03152561e-01 1.32664397e-01
-3.18390727e-01 -7.81810880e-01 -5.74117541e-01 -9.99096557e-02
-5.60374260e-01 -2.65028089e-01 1.19397260e-01 5.57858765e-01
1.21274047e-01 6.28522992e-01 3.38216543e-01 -4.78480279e-01
4.72985923e-01 -6.32936895e-01 4.30697560e-01 -3.07951003e-01
-3.73712331e-01 3.17824364e-01 -1.62145659e-01 -1.31403065e+00
-2.53922045e-01 -2.71780580e-01 -1.06477451e+00 -2.42813841e-01
1.33969083e-01 -5.29434323e-01 4.31502640e-01 3.98851484e-01
6.98027551e-01 5.95537126e-01 1.04426885e+00 -1.09300649e+00
-4.74432737e-01 -9.57664132e-01 -4.61396486e-01 2.14805782e-01
8.67515504e-01 -2.49424502e-01 -2.43180364e-01 3.91943268e-02]
|
[10.923954010009766, -2.078573703765869]
|
bacabdc5-5626-4f5e-800c-284553a4ad29
|
bootstrapped-q-learning-with-context-relevant
|
2009.11896
| null |
https://arxiv.org/abs/2009.11896v1
|
https://arxiv.org/pdf/2009.11896v1.pdf
|
Bootstrapped Q-learning with Context Relevant Observation Pruning to Generalize in Text-based Games
|
We show that Reinforcement Learning (RL) methods for solving Text-Based Games (TBGs) often fail to generalize on unseen games, especially in small data regimes. To address this issue, we propose Context Relevant Episodic State Truncation (CREST) for irrelevant token removal in observation text for improved generalization. Our method first trains a base model using Q-learning, which typically overfits the training games. The base model's action token distribution is used to perform observation pruning that removes irrelevant tokens. A second bootstrapped model is then retrained on the pruned observation text. Our bootstrapped agent shows improved generalization in solving unseen TextWorld games, using 10x-20x fewer training games compared to previous state-of-the-art methods despite requiring less number of training episodes.
|
['Ryuki Tachibana', 'Daiki Kimura', 'Asim Munawar', 'Subhajit Chaudhury', 'Michiaki Tatsubori', 'Kartik Talamadupula']
|
2020-09-24
| null |
https://aclanthology.org/2020.emnlp-main.241
|
https://aclanthology.org/2020.emnlp-main.241.pdf
|
emnlp-2020-11
|
['text-based-games']
|
['playing-games']
|
[ 2.25895092e-01 1.85423851e-01 -2.73304045e-01 1.18823268e-01
-1.02196395e+00 -5.97619116e-01 5.19450903e-01 1.93390310e-01
-9.66632128e-01 1.40253592e+00 -5.95415086e-02 -5.27646244e-01
-2.69599855e-01 -8.14305723e-01 -5.63154876e-01 -6.77684069e-01
7.79026002e-02 8.99355292e-01 4.88099664e-01 -7.09459960e-01
3.93009216e-01 4.38196920e-02 -1.57527184e+00 1.00461878e-01
1.10182786e+00 5.62342107e-01 4.28013206e-01 8.63059103e-01
-1.02729432e-01 9.88342285e-01 -9.53686059e-01 -1.64973080e-01
3.60005766e-01 -6.95708990e-01 -9.81125355e-01 -1.48168579e-01
-2.19325423e-02 -5.07679641e-01 -4.73862350e-01 9.40575838e-01
6.77118361e-01 9.45747912e-01 2.99919933e-01 -9.46321309e-01
-3.94852191e-01 6.56082809e-01 -1.79135159e-01 5.47110140e-01
2.43793368e-01 3.50590676e-01 1.07099891e+00 -3.11860383e-01
4.76949811e-01 1.08115482e+00 2.68133640e-01 1.27046752e+00
-1.20609522e+00 -7.34637260e-01 2.81870216e-01 1.57751873e-01
-8.92212391e-01 -1.11233942e-01 2.86399364e-01 2.62171924e-01
1.59069443e+00 2.89858758e-01 1.03417373e+00 1.47382402e+00
4.30223644e-01 9.38275516e-01 1.28538907e+00 -4.51433063e-01
9.20554578e-01 -2.94877321e-01 -1.55609563e-01 4.24917221e-01
3.05977613e-01 6.77492738e-01 -6.37017429e-01 -2.72111773e-01
6.73437238e-01 -5.80808744e-02 3.40227365e-01 -1.07125510e-02
-4.99896139e-01 1.17734599e+00 3.94266173e-02 6.91870134e-03
-4.51475710e-01 5.54227173e-01 4.84699845e-01 6.63214087e-01
4.89956915e-01 9.75285053e-01 -6.43557727e-01 -8.54738772e-01
-7.87791252e-01 9.60112751e-01 6.81458771e-01 7.67794609e-01
7.08036065e-01 5.31241298e-01 -2.12192446e-01 8.99209559e-01
-8.86222050e-02 5.50933480e-01 1.05806315e+00 -8.40859175e-01
5.35472512e-01 4.23065335e-01 2.67717481e-01 1.79476410e-01
-4.46992099e-01 -4.58625078e-01 -1.82781577e-01 3.12950134e-01
2.80636102e-01 -5.60458422e-01 -1.34189415e+00 1.70139122e+00
2.53487676e-01 2.88917124e-01 5.10210931e-01 6.70536578e-01
4.75611150e-01 6.62668347e-01 3.92384440e-01 -1.80621386e-01
1.05232501e+00 -7.33126819e-01 -3.98805410e-01 -7.49594271e-01
8.54518056e-01 -2.59475559e-02 1.19179904e+00 6.62657738e-01
-1.19788718e+00 -2.07231775e-01 -9.03979540e-01 2.19488606e-01
-4.38597739e-01 -6.75081551e-01 5.87261975e-01 7.82526076e-01
-1.00587952e+00 8.61506104e-01 -9.59270954e-01 -3.59141022e-01
3.83566707e-01 7.53596544e-01 1.33802637e-01 8.02837834e-02
-1.54604757e+00 1.01222694e+00 8.55819523e-01 -4.31047618e-01
-1.46760833e+00 -3.99281859e-01 -9.57892776e-01 1.68895349e-01
8.98335874e-01 -5.89196324e-01 1.73427963e+00 -5.96842706e-01
-2.07905722e+00 4.08491701e-01 8.68156180e-02 -9.77930307e-01
4.77954805e-01 -3.57800752e-01 -2.45482117e-01 8.58499482e-02
9.83311012e-02 7.44164705e-01 1.08373439e+00 -9.55765784e-01
-8.87713671e-01 -1.68368921e-01 2.08800197e-01 6.90374732e-01
-7.88444467e-03 -2.57243752e-01 -1.04395591e-01 -7.27311254e-01
-9.63716023e-03 -9.40449774e-01 -5.52280784e-01 -1.15330458e+00
-2.37283096e-01 -5.36714911e-01 6.33961439e-01 -1.04366824e-01
1.15580177e+00 -1.76994407e+00 1.73507199e-01 2.95976140e-02
1.16689958e-01 3.43873620e-01 -3.93178076e-01 6.46965086e-01
7.14101121e-02 2.01092362e-01 5.50972894e-02 -3.95811796e-01
2.19543293e-01 7.08581269e-01 -5.83463371e-01 -1.13893993e-01
3.50918807e-02 1.17786646e+00 -1.19281542e+00 -2.57165790e-01
3.61464828e-01 -5.24994910e-01 -8.82238746e-01 2.29202826e-02
-8.40684414e-01 3.37046176e-01 -7.18144596e-01 3.95569324e-01
2.62074202e-01 -1.38599426e-01 3.17708492e-01 8.80496562e-01
2.26618484e-01 4.92049217e-01 -9.01179731e-01 1.59329700e+00
-3.14718544e-01 3.07040632e-01 -6.35125279e-01 -1.16681194e+00
8.14979315e-01 2.00713679e-01 3.21881324e-01 -1.14672995e+00
1.60428897e-01 9.38839391e-02 2.00644210e-01 -5.68090200e-01
8.76382768e-01 -5.63548267e-01 -4.40649271e-01 5.51249743e-01
2.67670423e-01 -6.61153018e-01 4.23099488e-01 2.29416564e-01
1.24417281e+00 2.94326484e-01 2.87137508e-01 -2.93615535e-02
4.44577774e-03 1.72757089e-01 5.98942518e-01 1.65489590e+00
-3.72293591e-01 1.55290082e-01 4.31789130e-01 -5.33137679e-01
-1.11330640e+00 -1.11066878e+00 1.32749870e-01 1.68217516e+00
1.87792122e-01 -6.38506711e-01 -7.87756145e-01 -9.46204901e-01
2.75750104e-02 1.09836841e+00 -7.97579110e-01 -6.31824851e-01
-5.99673212e-01 -9.12175059e-01 6.67765081e-01 5.60134113e-01
5.46582043e-01 -1.79550755e+00 -8.12297821e-01 6.80980146e-01
-2.18834002e-02 -6.41769886e-01 1.47255342e-02 8.29193115e-01
-1.02199817e+00 -7.18305349e-01 -3.14452231e-01 -4.89118963e-01
1.69161946e-01 -2.31804535e-01 9.44422364e-01 2.44940236e-01
-1.65493097e-02 1.97660983e-01 -6.43973112e-01 -7.38455474e-01
-5.67054927e-01 4.01958406e-01 3.17008883e-01 -9.25633192e-01
5.65911889e-01 -4.80487049e-01 -4.65025991e-01 4.17851843e-02
-9.46677268e-01 -3.64646524e-01 3.83755803e-01 1.33146489e+00
2.81128258e-01 6.95855990e-02 9.60943401e-01 -8.01248968e-01
1.11877573e+00 -4.11046952e-01 -7.65303016e-01 1.21588379e-01
-5.83208859e-01 3.38571489e-01 9.73116875e-01 -8.48591506e-01
-1.11427057e+00 -3.36676925e-01 -1.17596397e-02 -2.63282418e-01
-1.97312891e-01 5.24329722e-01 5.74762881e-01 1.73513100e-01
1.11829507e+00 5.47448099e-01 -2.48659432e-01 -1.03875205e-01
1.97384596e-01 3.60239863e-01 2.70402342e-01 -1.11329103e+00
5.93795478e-01 1.17747895e-01 -1.75442412e-01 -7.62487292e-01
-8.93081367e-01 -1.60553068e-01 -7.31615424e-02 4.59944755e-02
5.88470459e-01 -5.71177602e-01 -1.01726723e+00 1.76638082e-01
-4.38216418e-01 -1.18077469e+00 -9.03565228e-01 5.63250244e-01
-1.02755606e+00 1.35527462e-01 -8.70852053e-01 -1.16423786e+00
-2.43733048e-01 -9.80136156e-01 1.00893044e+00 5.60356736e-01
-1.52550355e-01 -9.10263360e-01 5.15446663e-01 3.37284923e-01
1.33060873e-01 -1.19043507e-01 8.76054645e-01 -1.00629652e+00
-2.61194795e-01 1.25260651e-02 4.21675295e-01 6.94800168e-02
-7.17484728e-02 -5.77574670e-01 -8.03252339e-01 -6.66247785e-01
-8.51305574e-02 -9.94681954e-01 1.06209886e+00 6.10151589e-01
1.01987171e+00 -2.33331010e-01 -9.62897241e-02 3.16727906e-01
1.24159813e+00 5.63663900e-01 7.43592262e-01 7.61999547e-01
4.86567542e-02 3.80070210e-02 8.83628488e-01 7.64439344e-01
-3.64132635e-02 2.48604149e-01 4.49083030e-01 2.72796065e-01
4.48524415e-01 -6.09545410e-01 4.30167645e-01 1.77968040e-01
-1.46344438e-01 -5.08471489e-01 -7.72226930e-01 4.94214356e-01
-2.15300012e+00 -1.13811827e+00 6.25807762e-01 2.13457036e+00
9.80716288e-01 6.40999436e-01 4.08989936e-01 -7.50172138e-02
2.17055842e-01 9.69491005e-02 -1.10162449e+00 -6.00887239e-01
-3.48280482e-02 9.42335069e-01 5.33677638e-01 6.21049821e-01
-8.17657650e-01 1.87448859e+00 6.89105892e+00 1.27300096e+00
-8.64829302e-01 -4.30174507e-02 3.49813044e-01 -4.88261968e-01
7.53476322e-02 1.26120940e-01 -7.68927157e-01 3.16921979e-01
1.16198897e+00 -3.58452708e-01 8.64359319e-01 8.09357822e-01
2.71937221e-01 -5.12289822e-01 -7.43658245e-01 5.01215756e-01
-2.61373401e-01 -1.18510199e+00 1.52163789e-01 1.75916940e-01
9.38362479e-01 1.33155555e-01 3.37228149e-01 1.18188298e+00
1.23656070e+00 -1.09922910e+00 5.49741507e-01 4.26115617e-02
4.80674177e-01 -9.17431653e-01 5.36470771e-01 6.42429233e-01
-7.56217718e-01 -5.95463514e-01 -8.62119675e-01 -3.89625609e-01
-1.78164825e-01 -2.31322199e-01 -9.84201550e-01 4.29491073e-01
6.74709141e-01 2.99832046e-01 -4.61857289e-01 8.10455918e-01
-3.67207557e-01 9.28925335e-01 -3.89569521e-01 -5.92619240e-01
7.70904839e-01 -1.11714639e-01 6.28994882e-01 5.45503974e-01
1.62911937e-01 5.27826905e-01 3.35680902e-01 6.70518935e-01
3.88059616e-02 -1.37918368e-01 -6.25163853e-01 9.06114560e-03
5.28277636e-01 7.38220394e-01 -6.89086139e-01 -5.99465787e-01
8.75888094e-02 9.86403108e-01 6.37045920e-01 6.64918780e-01
-7.53532052e-01 -2.80417293e-01 3.20709765e-01 -3.62443984e-01
5.61164498e-01 2.39632204e-02 5.85681610e-02 -1.16166294e+00
-3.83208364e-01 -1.13233280e+00 6.64555073e-01 -6.20336771e-01
-1.15716159e+00 4.40271795e-01 2.53810257e-01 -9.96875763e-01
-8.30784798e-01 -4.55286831e-01 -6.50500000e-01 4.34681356e-01
-1.13018584e+00 -5.84682882e-01 2.66726464e-01 6.37388647e-01
9.42135930e-01 -3.25787693e-01 8.26623023e-01 -4.89682078e-01
-6.06657207e-01 5.10893524e-01 4.50676113e-01 -4.59717251e-02
2.55737454e-01 -1.78257453e+00 6.61987662e-01 4.99443293e-01
1.12697788e-01 4.91420150e-01 9.42245305e-01 -9.83511031e-01
-1.01486099e+00 -8.45644772e-01 2.44390354e-01 -4.99071956e-01
6.16061926e-01 -4.85282898e-01 -7.71713257e-01 8.71420920e-01
3.19637023e-02 -1.95279509e-01 4.81895775e-01 1.93714648e-01
2.08812393e-02 2.34874398e-01 -1.15049994e+00 1.08749652e+00
9.51653361e-01 -3.37263674e-01 -1.02725577e+00 3.48391801e-01
7.02793002e-01 -7.24680543e-01 -3.64812106e-01 2.97795166e-04
1.72821686e-01 -6.40447259e-01 8.73821020e-01 -1.43880880e+00
1.62016869e-01 1.53405607e-01 8.69214907e-02 -1.67559981e+00
-1.62103042e-01 -1.14465892e+00 -2.23265514e-01 3.03095639e-01
4.77543563e-01 -6.74547672e-01 1.40342295e+00 4.65182424e-01
-3.70382629e-02 -8.46187353e-01 -1.20497751e+00 -1.18004978e+00
6.22589290e-01 -5.03635108e-01 5.34725010e-01 5.60579836e-01
6.58475637e-01 3.45967442e-01 -4.74268049e-01 -2.48204246e-01
4.42630231e-01 -8.07670578e-02 6.56761289e-01 -9.79721665e-01
-4.03437555e-01 -1.46135807e-01 -6.59851059e-02 -1.12728429e+00
2.47097462e-01 -5.77749968e-01 2.53044069e-01 -1.10786736e+00
9.92401466e-02 -2.90022850e-01 -3.73526365e-01 6.42238855e-01
-4.67094988e-01 4.84700277e-02 2.43107811e-01 -2.04845145e-01
-1.08589244e+00 8.67464840e-01 1.27106953e+00 5.58260195e-02
-6.92513406e-01 7.51473904e-02 -5.47399938e-01 4.52560484e-01
1.19345355e+00 -8.31574619e-01 -6.60601377e-01 6.16113134e-02
3.08141291e-01 2.30427146e-01 1.08008713e-01 -1.06199789e+00
2.43906856e-01 -6.55589700e-01 3.82684141e-01 -4.61236596e-01
5.12417436e-01 -1.94248721e-01 -5.05539715e-01 7.64585435e-01
-5.64776003e-01 1.37526870e-01 6.31992340e-01 6.89310372e-01
2.44114593e-01 -6.46167755e-01 5.45831621e-01 -4.27025646e-01
-4.62779462e-01 2.39593178e-01 -1.03321457e+00 5.68938196e-01
9.40276980e-01 -4.34860289e-01 -3.50798845e-01 -7.18705118e-01
-1.08177423e+00 4.55810040e-01 1.89318180e-01 2.42455721e-01
7.22320855e-01 -9.30569112e-01 -6.09228194e-01 5.13990410e-02
-1.16006054e-01 6.70377463e-02 4.00207192e-01 1.69953227e-01
-2.44057834e-01 4.54828858e-01 -1.62125364e-01 -1.54230669e-01
-8.26484084e-01 6.59125686e-01 6.46563768e-01 -8.16639423e-01
-7.00937688e-01 4.93820578e-01 -3.67946806e-03 -3.89471561e-01
-2.30814959e-03 -4.71780896e-01 -1.62513778e-02 -4.32793647e-01
2.46834934e-01 4.03682828e-01 6.63166807e-04 2.26385504e-01
4.65638898e-02 1.62799224e-01 -6.71445668e-01 -6.85813785e-01
1.34701443e+00 1.49915233e-01 7.70305037e-01 2.04681844e-01
5.48865020e-01 -3.38695347e-01 -1.55179870e+00 -1.60897821e-01
1.33248297e-02 -2.16220140e-01 -8.22478458e-02 -9.40886497e-01
-3.99308860e-01 5.23203433e-01 3.90287369e-01 5.42580843e-01
7.68102586e-01 -7.66848102e-02 7.76377261e-01 1.12787044e+00
4.47846323e-01 -1.72033989e+00 4.16152656e-01 8.69657934e-01
5.11748791e-01 -1.17635953e+00 1.43993953e-02 3.01573515e-01
-8.35627973e-01 8.34474862e-01 9.56184626e-01 -4.34078425e-01
7.18632713e-02 8.90260786e-02 -1.83932304e-01 -3.56449544e-01
-1.16382754e+00 -5.25774777e-01 -2.57287294e-01 8.27501953e-01
-4.27094311e-01 -5.17689735e-02 -1.36469513e-01 4.58443969e-01
-6.48158789e-01 -5.77065349e-03 5.82331240e-01 1.27615893e+00
-7.30379879e-01 -1.18135512e+00 -2.42069796e-01 7.37382650e-01
-4.74254638e-01 -4.05538917e-01 -3.58326972e-01 1.04187405e+00
-1.78940028e-01 1.02204275e+00 1.12759158e-01 -3.13310504e-01
1.90879703e-01 1.61049604e-01 6.67065322e-01 -8.60250890e-01
-9.29867029e-01 1.30452931e-01 -5.76632954e-02 -4.71054822e-01
1.24535412e-01 -6.86541796e-01 -1.52266777e+00 -2.87569731e-01
-5.86186409e-01 5.08935392e-01 1.27407268e-01 1.14171374e+00
2.40608260e-01 4.23827171e-01 3.42167526e-01 -4.96925443e-01
-1.02583849e+00 -1.08158755e+00 -8.14415574e-01 4.43360031e-01
3.54723454e-01 -7.88913429e-01 -4.28677857e-01 -5.20113945e-01]
|
[3.933366537094116, 1.432818055152893]
|
3cdd0154-cba6-485b-afdd-3e79e6a498e8
|
ps4-a-next-generation-dataset-for-protein
| null | null |
https://www.biorxiv.org/content/10.1101/2023.02.28.530456v1
|
https://www.biorxiv.org/content/10.1101/2023.02.28.530456v1.full.pdf+html
|
PS4: a Next-Generation Dataset for Protein Single Sequence Secondary Structure Prediction
|
Protein secondary structure prediction is a subproblem of protein folding. A lightweight algorithm capable of accurately predicting secondary structure from only the protein residue sequence could therefore provide a useful input for tertiary structure prediction, alleviating the reliance on MSA typically seen in today's best-performing models. This in turn could see the development of protein folding algorithms which perform better on orphan proteins, and which are much more accessible for both research and industry adoption due to reducing the necessary computational resources to run. Unfortunately, existing datasets for secondary structure prediction are small, creating a bottleneck in the rate of progress of automatic secondary structure prediction. Furthermore, protein chains in these datasets are often not identified, hampering the ability of researchers to use external domain knowledge when developing new algorithms.
We present PS4, a dataset of 18,731 non-redundant protein chains and their respective Q8 secondary structure labels. Each chain is identified by its PDB code, and the dataset is also non-redundant against other secondary structure datasets commonly seen in the literature. We perform ablation studies by training secondary structure prediction algorithms on the PS4 training set, and obtain state-of-the-art Q8 and Q3 accuracy on the CB513 test set in zero shots, without further fine-tuning. Furthermore, we provide a software toolkit for the community to run our evaluation algorithms, train models from scratch and add new samples to the dataset.
All code and data required to reproduce our results and make new inferences is available at https://github.com/omarperacha/ps4-dataset
|
['Omar Peracha']
|
2023-03-01
| null | null | null |
biorxiv-preprint-2023-3
|
['protein-secondary-structure-prediction', 'protein-folding']
|
['medical', 'natural-language-processing']
|
[ 3.13087732e-01 -1.93059966e-02 -1.98931351e-01 -3.85650277e-01
-7.91310251e-01 -8.29122484e-01 9.12462398e-02 3.80175442e-01
-3.32070917e-01 1.19194996e+00 1.84669673e-01 -7.90451467e-01
2.16440618e-01 -4.07042742e-01 -6.67328835e-01 -8.61435056e-01
8.08662847e-02 5.98553717e-01 3.41537148e-01 -1.67804062e-01
3.26482207e-01 5.16241312e-01 -1.45866871e+00 4.70915109e-01
8.94777298e-01 4.08161938e-01 6.43769920e-01 5.07930279e-01
-1.00337729e-01 5.28273344e-01 -3.96554738e-01 -1.05837293e-01
2.22377405e-01 -8.54498684e-01 -9.94719505e-01 -1.29061595e-01
1.03952482e-01 6.85226396e-02 1.01357751e-01 6.09635472e-01
5.28760672e-01 -1.51065603e-01 4.06410038e-01 -6.29845321e-01
-3.11125100e-01 1.18734436e-02 -3.19355011e-01 2.05718726e-01
5.58904767e-01 5.59238374e-01 1.06942511e+00 -7.69583881e-01
1.17294216e+00 7.69951999e-01 5.76406360e-01 4.37584370e-01
-1.69834626e+00 -5.22397578e-01 -3.19723994e-01 3.35882872e-01
-8.29244077e-01 -4.06099230e-01 3.00208002e-01 -3.17177474e-01
1.58744800e+00 2.13843480e-01 6.81352198e-01 1.16213977e+00
1.73299849e-01 3.86797905e-01 1.34431744e+00 -3.39590460e-01
2.44593635e-01 -2.46538505e-01 4.00596678e-01 6.50243461e-01
1.21302061e-01 4.84032780e-02 -4.39550042e-01 -7.07637370e-01
2.14868337e-01 -1.21963955e-01 -3.19025367e-01 -5.76963723e-01
-1.05667567e+00 6.69335604e-01 3.40649784e-01 -5.46749122e-02
-3.31447631e-01 -3.85720402e-01 3.14520955e-01 6.49724960e-01
1.13211818e-01 6.68627024e-01 -9.55910683e-01 -4.61405307e-01
-5.74378312e-01 5.39423764e-01 8.89811099e-01 6.50057614e-01
7.90744245e-01 -6.23330832e-01 6.45096302e-01 6.43205047e-01
1.34574100e-01 1.94356680e-01 4.95087445e-01 -7.70395577e-01
3.70066129e-02 8.06855500e-01 1.61620826e-01 -2.37374857e-01
-6.59177005e-01 -5.56669086e-02 -2.28073001e-01 2.77721345e-01
8.04098606e-01 2.76425574e-02 -1.03823006e+00 1.53055394e+00
4.61973369e-01 -3.34688157e-01 1.20978750e-01 9.15181518e-01
3.89233917e-01 5.33245802e-01 1.68810859e-01 -4.83446896e-01
1.35928321e+00 -5.04172623e-01 -5.21283299e-02 1.38137743e-01
1.01603985e+00 -1.12594306e+00 1.10050952e+00 4.28530306e-01
-5.74164152e-01 -1.40218019e-01 -1.19303322e+00 -1.90264076e-01
-2.98522055e-01 -3.17873210e-01 6.55292273e-01 3.21794808e-01
-5.94547808e-01 1.04492438e+00 -1.00411046e+00 -6.44664764e-01
3.04473400e-01 4.89468992e-01 -6.65913284e-01 -1.40151426e-01
-9.56708848e-01 1.32848895e+00 4.65638936e-01 -3.76623392e-01
-5.13108313e-01 -7.52554238e-01 -4.02383208e-01 -1.97293326e-01
2.39414752e-01 -5.32840431e-01 1.19729662e+00 -1.02991450e+00
-9.96548712e-01 8.06143939e-01 -4.57307577e-01 -4.22256887e-01
2.04447806e-01 2.79090088e-03 -1.86089590e-01 1.63278952e-01
6.33859783e-02 4.65232104e-01 3.68598908e-01 -9.23717916e-01
-5.99914432e-01 -5.02509832e-01 -1.95003450e-01 1.10266492e-01
4.12441820e-01 1.96566224e-01 1.88110486e-01 -2.50087231e-01
2.48556301e-01 -1.13545418e+00 -4.52410668e-01 -1.63790718e-01
-5.95606007e-02 -1.32129401e-01 7.27982402e-01 -8.62590134e-01
8.54604125e-01 -1.74290884e+00 2.31219590e-01 1.86171532e-01
2.10359201e-01 3.73866975e-01 -2.09353462e-01 7.92831659e-01
-4.93323207e-01 -1.13074034e-01 -3.20627868e-01 3.66649985e-01
-4.36251134e-01 2.22548217e-01 -4.20845412e-02 5.34808517e-01
2.84758896e-01 6.29596055e-01 -7.85574973e-01 -2.37311885e-01
1.24327149e-02 3.25996697e-01 -5.01877904e-01 1.12138942e-01
-5.13493180e-01 4.63839889e-01 -3.13537210e-01 7.21649587e-01
4.56689209e-01 -5.34556389e-01 7.65234351e-01 -8.91098306e-02
-1.48361608e-01 7.38795817e-01 -6.78707778e-01 1.59705198e+00
-3.34988721e-02 1.63493350e-01 -2.11691663e-01 -6.34661913e-01
8.59904706e-01 2.07135960e-01 6.61610186e-01 -6.00719035e-01
-7.25414678e-02 2.83759892e-01 6.51556909e-01 -2.86214441e-01
4.77244928e-02 -2.70347238e-01 4.24237281e-01 6.56172335e-01
9.63410661e-02 3.85751128e-01 4.27249491e-01 1.77937761e-01
1.53692615e+00 6.30876362e-01 3.64088565e-01 -3.60899627e-01
2.33543232e-01 6.34038866e-01 8.37048769e-01 1.03004925e-01
-1.28772929e-01 4.85965252e-01 6.15715683e-01 -6.84547365e-01
-1.62325871e+00 -1.00971580e+00 -4.51546878e-01 1.12838602e+00
-1.73930943e-01 -7.29254484e-01 -6.71125770e-01 -7.33196557e-01
9.38848872e-03 4.95058268e-01 -2.19702795e-01 -6.11664243e-02
-6.32054508e-01 -1.05104983e+00 2.11309984e-01 2.29583219e-01
-2.03810111e-01 -1.27762139e+00 -7.38063633e-01 5.26198685e-01
-2.09366843e-01 -4.95967716e-01 -8.90219510e-02 8.90899777e-01
-1.02092266e+00 -1.59749746e+00 -4.85078156e-01 -7.20456779e-01
5.37880898e-01 2.08551437e-01 1.17758751e+00 2.75069177e-01
-5.25893211e-01 -3.52043152e-01 -5.29181838e-01 -3.81587267e-01
-7.18422413e-01 2.82289326e-01 1.92952618e-01 -6.24658823e-01
9.23701108e-01 -7.90670335e-01 -6.71800494e-01 4.12521034e-01
-5.80881655e-01 3.40552449e-01 6.41906679e-01 1.07881832e+00
6.38227701e-01 -5.56257844e-01 7.52262533e-01 -9.97923851e-01
3.23624253e-01 -5.13865709e-01 -7.00660288e-01 1.68985128e-01
-8.96076620e-01 4.56431955e-01 7.46520579e-01 -3.77048217e-02
-8.04393411e-01 5.53553760e-01 -3.31773221e-01 1.13593780e-01
-2.72291899e-01 5.13593495e-01 -1.36845335e-01 4.77561057e-02
9.57167089e-01 2.59811461e-01 5.11967182e-01 -8.87946129e-01
9.08765644e-02 6.38785362e-01 1.37991115e-01 -5.37544727e-01
4.26166207e-01 1.58736091e-02 7.84632415e-02 -8.14712763e-01
-5.75068474e-01 -6.08102560e-01 -7.09371805e-01 3.65574658e-01
4.23227102e-01 -6.79445684e-01 -8.93384218e-01 1.30745515e-01
-6.94741011e-01 -2.79676706e-01 1.39749497e-01 3.25917512e-01
-6.74460292e-01 7.42052495e-01 -6.08586729e-01 -2.10271835e-01
-3.75140667e-01 -1.36144960e+00 6.43798292e-01 4.54450166e-03
-6.69498682e-01 -3.53965223e-01 3.69072199e-01 6.53821409e-01
6.11999892e-02 9.04152691e-02 1.44326711e+00 -7.86879420e-01
-4.41953361e-01 1.45067319e-01 -8.28839540e-02 1.02500245e-01
1.95965648e-01 -4.07883003e-02 -6.64089203e-01 -4.17614400e-01
-1.22595809e-01 -6.10169351e-01 8.15752566e-01 5.98049350e-02
5.28818667e-01 9.24726669e-03 -4.12465841e-01 4.04093713e-01
1.31758583e+00 2.57090062e-01 4.86963481e-01 6.46981537e-01
5.44898748e-01 7.30353832e-01 7.85643935e-01 2.03463227e-01
-1.30455857e-02 6.27982378e-01 2.78304011e-01 -3.99069786e-02
1.77464038e-01 -1.62176132e-01 2.87889570e-01 4.83185381e-01
-2.29396239e-01 7.20082000e-02 -1.09551561e+00 3.51573646e-01
-1.80933225e+00 -9.23383176e-01 -2.79098690e-01 2.18158793e+00
1.24464703e+00 1.00777954e-01 3.42816561e-01 -7.95915071e-03
3.78834248e-01 -4.32134926e-01 -8.81675005e-01 -3.36029410e-01
-6.35594428e-02 4.58127171e-01 6.00423217e-01 2.06891105e-01
-8.45995307e-01 8.42788577e-01 6.45079327e+00 3.77406687e-01
-1.04209900e+00 -2.34528020e-01 4.64966983e-01 -3.57005149e-01
-8.24666843e-02 4.58932817e-01 -6.26253009e-01 6.27207577e-01
1.47623992e+00 -1.27542660e-01 4.43500817e-01 8.55109036e-01
4.34093803e-01 -2.40096524e-01 -1.08405828e+00 6.17418230e-01
-5.58419943e-01 -1.44859421e+00 -2.84512788e-01 2.84205794e-01
2.95627654e-01 4.76309419e-01 -4.29171950e-01 -8.15911591e-02
4.62139785e-01 -9.72831786e-01 1.07269764e-01 1.85683087e-01
5.58985114e-01 -8.58576000e-01 6.48622155e-01 3.82663369e-01
-6.43182278e-01 1.59056127e-01 -5.21125615e-01 -8.15639645e-02
-1.32727008e-02 3.91584545e-01 -1.03899014e+00 4.63644862e-01
6.54116213e-01 5.02140641e-01 -7.20821440e-01 8.69106591e-01
1.02784157e-01 8.04439247e-01 -4.64386046e-01 -8.10951814e-02
2.64944658e-02 -3.53737712e-01 2.94340551e-01 8.57699096e-01
-9.32881013e-02 2.97319740e-01 2.65653372e-01 4.54689056e-01
1.84864681e-02 2.89953858e-01 -2.98592806e-01 -2.43339509e-01
2.50041485e-01 1.10787475e+00 -7.57644534e-01 -1.62771910e-01
-6.23548865e-01 7.68205643e-01 4.88763213e-01 1.08995393e-01
-5.82732975e-01 -3.59314770e-01 1.18740749e+00 4.23723191e-01
1.70726985e-01 -1.64780170e-01 8.75405446e-02 -1.15379524e+00
-8.27180818e-02 -1.46585786e+00 2.70723194e-01 -5.35907507e-01
-1.30064750e+00 3.06599975e-01 -4.04989690e-01 -8.68270755e-01
-3.29311252e-01 -7.79438913e-01 -3.19226533e-02 1.07297814e+00
-1.10437882e+00 -8.51797283e-01 1.56354278e-01 -3.18034664e-02
3.85982364e-01 -8.33631977e-02 1.10441983e+00 -9.37378313e-03
-3.77986312e-01 2.97049433e-01 6.20259821e-01 -2.87801266e-01
1.05712819e+00 -1.20207822e+00 7.25907207e-01 5.54707110e-01
-3.82660851e-02 8.83505702e-01 1.04465461e+00 -8.80886376e-01
-1.49999535e+00 -7.71840036e-01 7.88246274e-01 -8.05912673e-01
6.55907393e-01 -3.58263105e-01 -1.34767985e+00 5.72956979e-01
-1.41151458e-01 -3.67266774e-01 1.04116988e+00 3.35961998e-01
-2.70911664e-01 4.45664346e-01 -1.02972579e+00 4.64907259e-01
1.16265821e+00 -2.50555247e-01 -6.38777018e-01 2.86807925e-01
5.28652370e-01 -1.69544071e-01 -9.97334421e-01 2.47163519e-01
7.22785413e-01 -1.02909946e+00 7.63149738e-01 -1.04818058e+00
4.41450208e-01 -4.81602579e-01 -3.57446745e-02 -9.65711653e-01
-5.05748987e-01 -4.64298636e-01 -2.81307995e-02 8.16559792e-01
8.18753600e-01 -5.51532984e-01 1.12805760e+00 4.96075034e-01
-1.46967679e-01 -1.02839553e+00 -6.80516243e-01 -6.02862895e-01
1.68368548e-01 1.19068466e-01 4.40969735e-01 8.93289208e-01
2.85085946e-01 5.99955380e-01 -1.45233423e-01 -2.81875193e-01
3.52285922e-01 5.34726620e-01 6.99546158e-01 -1.16221893e+00
-5.89838386e-01 2.93519553e-02 -3.43801796e-01 -6.83179498e-01
-4.22133878e-02 -1.03256857e+00 -1.82266720e-02 -1.21516407e+00
5.54583728e-01 -2.59509802e-01 -2.82595813e-01 7.85673618e-01
-1.54041812e-01 2.02181727e-01 -2.83619575e-03 4.89660561e-01
-3.00162226e-01 3.04788202e-01 1.04551649e+00 1.74720436e-01
-2.77818114e-01 -1.24024965e-01 -5.44408321e-01 4.84897375e-01
8.87362421e-01 -6.13862991e-01 -3.31955403e-01 3.27858388e-01
1.53814061e-02 -1.16578676e-01 8.72298703e-02 -8.62303734e-01
-3.20558310e-01 -2.67917514e-01 7.34621108e-01 -5.37030280e-01
1.72184020e-01 -7.42543578e-01 5.23328781e-01 7.56797373e-01
-1.30406871e-01 2.45203912e-01 8.85576475e-03 3.34938586e-01
1.92714989e-01 -2.49823853e-01 9.12526846e-01 -2.70953029e-01
-5.74329138e-01 6.46814182e-02 -6.58082843e-01 -1.65126413e-01
9.26394105e-01 -2.33464181e-01 -3.89892548e-01 1.62109047e-01
-9.04298782e-01 4.03664149e-02 1.32087922e+00 2.33432561e-01
1.96433187e-01 -6.41490400e-01 -4.66385514e-01 1.84740528e-01
3.25499445e-01 -4.75551128e-01 -8.22902396e-02 6.38924718e-01
-9.30608988e-01 4.75519449e-01 -5.84187150e-01 -3.81672829e-01
-1.66282558e+00 7.97381401e-01 9.31516290e-02 -1.13717191e-01
-7.75002003e-01 2.75540382e-01 7.51884771e-04 -4.12678361e-01
-2.76698560e-01 -5.91050647e-02 2.48265356e-01 -1.39175504e-01
4.34055954e-01 8.14731717e-02 2.55154759e-01 -6.14581823e-01
-4.50808644e-01 3.92270610e-02 -5.54880440e-01 3.06628704e-01
1.45985961e+00 2.11748853e-01 -1.85879961e-01 1.77221403e-01
1.15971565e+00 -3.76500636e-01 -1.14437139e+00 -6.27607387e-03
4.88853812e-01 -2.42337510e-01 -5.14047682e-01 -1.04644275e+00
-2.13474765e-01 4.53160793e-01 8.08217943e-01 -3.31371158e-01
8.54176283e-01 8.02759901e-02 7.93598115e-01 7.60083735e-01
4.15201217e-01 -8.28405201e-01 -4.50775623e-01 5.45375645e-01
3.96256357e-01 -1.33616710e+00 1.74921930e-01 -2.38082603e-01
-4.51416075e-01 1.01723397e+00 4.68641132e-01 8.95859525e-02
1.69612363e-01 2.60918200e-01 3.43075216e-01 -1.75810039e-01
-1.20972919e+00 -1.00059241e-01 -3.28690320e-01 6.13575637e-01
9.24838364e-01 -1.28832534e-01 -7.54058540e-01 3.61898631e-01
-2.39661530e-01 -6.59356713e-02 4.17459428e-01 1.29945052e+00
-7.17316449e-01 -1.85643482e+00 -1.69155896e-01 7.47174621e-01
-6.66805983e-01 -1.72515348e-01 -8.10570896e-01 4.75053042e-01
-2.10456446e-01 7.27531612e-01 -3.26828629e-01 -2.33272582e-01
9.77230966e-02 5.46721458e-01 6.34363830e-01 -5.84123075e-01
-6.16881967e-01 -7.59445364e-04 3.47641170e-01 -5.59264660e-01
-1.18576646e-01 -7.27460921e-01 -1.58393574e+00 -5.42151034e-01
-3.78423274e-01 3.84252936e-01 5.73213875e-01 7.59724081e-01
7.59851336e-01 -1.64351193e-03 2.68874049e-01 -6.79701209e-01
-6.60495877e-01 -9.38166797e-01 -3.36409509e-01 6.10633850e-01
3.42569686e-02 -5.08588612e-01 1.07766576e-01 2.32208297e-01]
|
[4.769915580749512, 5.519698143005371]
|
9d8d08b4-01d4-4161-a931-d916a0fe2356
|
learning-object-affordance-with-contact-and
|
2210.09245
| null |
https://arxiv.org/abs/2210.09245v3
|
https://arxiv.org/pdf/2210.09245v3.pdf
|
Contact2Grasp: 3D Grasp Synthesis via Hand-Object Contact Constraint
|
3D grasp synthesis generates grasping poses given an input object. Existing works tackle the problem by learning a direct mapping from objects to the distributions of grasping poses. However, because the physical contact is sensitive to small changes in pose, the high-nonlinear mapping between 3D object representation to valid poses is considerably non-smooth, leading to poor generation efficiency and restricted generality. To tackle the challenge, we introduce an intermediate variable for grasp contact areas to constrain the grasp generation; in other words, we factorize the mapping into two sequential stages by assuming that grasping poses are fully constrained given contact maps: 1) we first learn contact map distributions to generate the potential contact maps for grasps; 2) then learn a mapping from the contact maps to the grasping poses. Further, we propose a penetration-aware optimization with the generated contacts as a consistency constraint for grasp refinement. Extensive validations on two public datasets show that our method outperforms state-of-the-art methods regarding grasp generation on various metrics.
|
['Qi Ye', 'Jiming Chen', 'Yuchi Huo', 'Xiang Li', 'Yang Zhou', 'Xinzhuo Lin', 'Haoming Li']
|
2022-10-17
| null | null | null | null |
['grasp-generation', 'robotic-grasping']
|
['computer-vision', 'robots']
|
[ 2.31491670e-01 1.29209995e-01 -2.57001609e-01 -2.93094337e-01
-7.41168141e-01 -8.02298069e-01 3.67491007e-01 -1.49762603e-02
2.28510480e-02 4.51060385e-01 -1.49691924e-02 2.34841377e-01
-2.71475047e-01 -9.78263080e-01 -1.15014613e+00 -6.84204578e-01
-1.49231434e-01 9.63314831e-01 2.41468370e-01 -8.23236555e-02
3.68024677e-01 7.65682101e-01 -1.51354158e+00 1.51240095e-01
1.04648256e+00 1.08235657e+00 6.31757200e-01 2.88110554e-01
-1.48496717e-01 -1.24036342e-01 -4.24112111e-01 -2.07826793e-01
7.28918672e-01 -2.60597039e-02 -6.69120669e-01 8.66156220e-02
9.43956450e-02 -5.79655170e-01 -2.68593477e-03 1.11279905e+00
2.07601875e-01 4.63499278e-02 8.54461372e-01 -1.10282242e+00
-6.46224201e-01 3.98834914e-01 -5.97169816e-01 -6.74418211e-01
5.07361591e-01 1.93208799e-01 8.85081947e-01 -9.45723474e-01
8.11446011e-01 1.44440901e+00 2.31105059e-01 6.07150614e-01
-1.29572535e+00 -4.19434339e-01 5.18974185e-01 -2.37805188e-01
-1.12057865e+00 3.65640596e-02 8.26583266e-01 -5.64765573e-01
6.30985320e-01 1.46573022e-01 7.06463039e-01 1.15985334e+00
3.65656435e-01 8.30280423e-01 9.00842667e-01 -2.78315485e-01
3.99795532e-01 -3.55515808e-01 -2.16571733e-01 5.13656080e-01
2.43204623e-01 -7.06571043e-02 -4.01241630e-01 -3.31611246e-01
1.28219759e+00 1.66391134e-01 -1.92405432e-01 -8.36633325e-01
-1.20001471e+00 5.68451285e-01 6.83477163e-01 -1.85308546e-01
-7.57636368e-01 5.74774370e-02 5.42021133e-02 -6.79755211e-02
3.39210123e-01 6.99310541e-01 -4.72408235e-01 1.25817463e-01
-3.17393750e-01 9.83474731e-01 8.85785699e-01 1.48408711e+00
7.87564278e-01 -5.18311620e-01 -4.57514435e-01 6.16069555e-01
4.45071131e-01 6.03644013e-01 -3.32770050e-01 -1.07273185e+00
7.28020132e-01 7.28032649e-01 5.26951551e-01 -9.05773699e-01
-1.19233735e-01 4.35110405e-02 -4.80999917e-01 2.13394642e-01
5.49496174e-01 1.05596878e-01 -1.34153664e+00 1.54182565e+00
4.38046813e-01 -3.80701780e-01 -3.82720858e-01 1.07072997e+00
3.46278101e-01 4.64033097e-01 -9.99282375e-02 -2.51861252e-02
8.86322558e-01 -7.24701405e-01 -6.37629390e-01 -2.51112044e-01
-2.39246875e-01 -7.59016812e-01 9.96171057e-01 3.45288754e-01
-1.31113744e+00 -3.15039217e-01 -8.07853460e-01 1.68396458e-01
-7.49601945e-02 3.14830482e-01 8.25345933e-01 7.88010433e-02
-5.21903157e-01 8.41609895e-01 -1.18393099e+00 -1.61550879e-01
5.29422998e-01 4.83991504e-01 -1.22527279e-01 -2.79296786e-01
-6.35606527e-01 9.88191545e-01 4.21089858e-01 2.71526843e-01
-1.03240156e+00 -6.67031944e-01 -5.93131065e-01 -1.17768034e-01
6.75188363e-01 -4.34598476e-01 9.46422160e-01 -1.43188640e-01
-1.66857243e+00 6.28077745e-01 -1.58924639e-01 2.49645308e-01
8.23523819e-01 -6.14530861e-01 8.24636102e-01 -5.69734983e-02
1.59150124e-01 8.42388034e-01 9.81864870e-01 -1.97054899e+00
-2.13754714e-01 -5.15412152e-01 4.18753892e-01 3.32545877e-01
3.55427549e-03 -5.00124633e-01 -7.54284918e-01 -5.37675142e-01
7.37720251e-01 -9.60130632e-01 -2.74490058e-01 3.26095134e-01
-7.75873721e-01 -4.84484076e-01 6.83470905e-01 -7.29037464e-01
4.86060590e-01 -1.83191156e+00 1.01285636e+00 4.05129045e-01
-1.89920925e-02 -1.76321968e-01 -2.51568913e-01 4.62070584e-01
3.85705262e-01 -1.18338682e-01 -3.91937166e-01 -3.69017810e-01
2.71631330e-01 2.81769216e-01 -5.91664910e-01 2.52641857e-01
3.69549036e-01 1.00164926e+00 -1.11013579e+00 -2.10118100e-01
3.26895177e-01 2.86911368e-01 -7.64188766e-01 7.68357992e-01
-7.32760429e-01 5.73509872e-01 -8.72062564e-01 1.08986974e+00
1.06538880e+00 1.56989828e-01 1.53989762e-01 -6.29449844e-01
-1.94803812e-02 9.12194257e-04 -1.02806509e+00 2.01816487e+00
-2.79558510e-01 -4.39826965e-01 3.70256722e-01 -6.44373417e-01
1.28938305e+00 4.41077948e-02 7.44888961e-01 1.25271812e-01
1.80233359e-01 3.86901766e-01 -3.07103813e-01 -2.61204004e-01
1.41754612e-01 3.84945311e-02 4.65494860e-03 3.60553265e-01
1.31229326e-01 -9.62439418e-01 -9.20023918e-02 -2.57744908e-01
6.68001235e-01 8.87962282e-01 -2.57170916e-01 -3.12358767e-01
-1.66433930e-01 -1.38768196e-01 4.43563581e-01 5.77291667e-01
3.92106026e-01 8.40818465e-01 5.33600867e-01 -1.98782131e-01
-9.67670858e-01 -1.62546337e+00 -6.83269352e-02 5.23528814e-01
6.48354590e-01 -1.09851770e-01 -9.89829779e-01 -6.80490315e-01
4.94064718e-01 4.56557900e-01 -6.59864604e-01 -1.41366422e-01
-7.48239875e-01 -6.39499903e-01 -7.75563344e-02 6.52048230e-01
1.18195310e-01 -1.23785198e+00 -7.22246587e-01 2.10552648e-01
-1.13508634e-01 -9.95331883e-01 -5.00057518e-01 2.77099431e-01
-8.58691156e-01 -1.21670771e+00 -1.00084579e+00 -6.69475496e-01
1.07136607e+00 -4.23063198e-03 7.92939126e-01 5.70621677e-02
-5.75549424e-01 1.83712974e-01 -5.48094690e-01 -5.26319802e-01
-1.38253540e-01 2.11086780e-01 9.42339078e-02 -4.23129320e-01
7.18908310e-02 -5.54501653e-01 -5.64901173e-01 3.56750399e-01
-7.59642482e-01 1.80269197e-01 7.02755392e-01 7.19079077e-01
7.93370485e-01 -2.06688076e-01 5.27704775e-01 -1.99511096e-01
5.92503726e-01 -2.81090915e-01 -6.61524236e-01 4.02163237e-01
-1.00779355e-01 1.16395786e-01 2.29972884e-01 -5.47746778e-01
-9.59841073e-01 3.99958819e-01 8.03626701e-02 -7.77872741e-01
-6.70376094e-03 7.71474317e-02 -4.62967068e-01 -2.63890494e-02
2.49846876e-01 -2.22510397e-02 8.31108466e-02 -6.84823990e-01
3.84772241e-01 3.29570651e-01 3.30217510e-01 -1.36533856e+00
1.01164043e+00 3.16992998e-01 1.20836079e-01 -4.58408952e-01
-8.00848663e-01 7.29487091e-02 -1.11716688e+00 -1.93465918e-01
1.03008091e+00 -4.07294512e-01 -9.94863808e-01 5.96219063e-01
-1.53363776e+00 -5.30261755e-01 -2.69605219e-01 3.06297451e-01
-9.18178439e-01 2.42388636e-01 -4.72536623e-01 -8.72018456e-01
-3.20159495e-01 -1.48498261e+00 1.57492721e+00 7.29145631e-02
-1.21302843e-01 -3.19663048e-01 -3.45330775e-01 3.23704481e-02
1.69804469e-01 7.28556395e-01 1.09065461e+00 9.53167751e-02
-1.14894092e+00 -2.44268715e-01 -3.31495762e-01 2.69666940e-01
7.06456423e-01 -8.60446170e-02 -5.33139348e-01 -5.63408673e-01
-1.59080386e-01 -4.24713045e-01 4.94204253e-01 5.39725065e-01
1.31894386e+00 -1.22995250e-01 -6.03651643e-01 4.79728580e-01
1.15744340e+00 7.30298311e-02 3.54010999e-01 6.43459409e-02
7.96868622e-01 9.27112639e-01 1.05691135e+00 4.90818888e-01
1.12015285e-01 8.50945294e-01 8.67160320e-01 2.96662569e-01
2.08152756e-02 -5.87013245e-01 -1.05398595e-01 3.73787045e-01
-6.35750771e-01 -8.47434402e-02 -7.34019041e-01 3.53284478e-01
-1.92581213e+00 -3.21978778e-01 4.02210325e-01 2.34892917e+00
8.97871852e-01 1.43750846e-01 -6.33125007e-02 -3.25811446e-01
5.19778550e-01 2.29027867e-02 -9.02268767e-01 2.56428421e-01
4.36785072e-01 1.85739502e-01 4.29035991e-01 5.24219394e-01
-7.66386449e-01 1.07230401e+00 5.93415070e+00 4.85711247e-01
-9.50166762e-01 -2.24006712e-01 1.39009180e-02 -1.69964477e-01
-4.19275492e-01 -8.60280767e-02 -7.19816446e-01 3.55943441e-01
-3.89497250e-01 2.02968106e-01 6.70277894e-01 8.57213795e-01
-1.47425920e-01 1.16801664e-01 -1.44418848e+00 5.82903981e-01
-2.68570125e-01 -1.09717786e+00 4.17359531e-01 3.47751714e-02
6.80752456e-01 -4.12776142e-01 -1.01340905e-01 -2.94967815e-02
3.02719861e-01 -9.69403923e-01 1.12170911e+00 7.70194113e-01
8.93356919e-01 -6.50405824e-01 4.28561985e-01 3.74617428e-01
-1.17041290e+00 3.02169006e-02 -5.43978512e-01 2.19721958e-01
4.15869117e-01 3.34920406e-01 -6.18243396e-01 7.53258109e-01
8.73238921e-01 3.86912227e-01 7.25610256e-02 8.88911068e-01
-3.51589948e-01 -1.48199797e-01 -3.65409940e-01 2.25470141e-02
-1.83854103e-02 -2.31223494e-01 8.30772221e-01 6.94747567e-01
3.09993833e-01 1.67090610e-01 6.18479371e-01 1.69081926e+00
1.14234105e-01 -2.89463937e-01 -4.33672994e-01 -4.64031734e-02
8.33367705e-01 9.42450047e-01 -7.75132000e-01 3.04859608e-01
2.15965614e-01 1.07718933e+00 4.14395154e-01 3.86071205e-01
-4.73281711e-01 -2.25700915e-01 7.35705912e-01 1.82686687e-01
2.27838010e-01 -5.06133378e-01 -4.35275525e-01 -9.77485538e-01
5.53421617e-01 -6.35542214e-01 -3.89752507e-01 -5.41574836e-01
-1.45340097e+00 3.79488140e-01 5.37080705e-01 -9.24929321e-01
-1.07211292e-01 -7.99503922e-01 -4.66797709e-01 1.13245976e+00
-1.24931371e+00 -1.29972875e+00 -5.98279536e-01 1.58507049e-01
6.95878088e-01 1.74082235e-01 8.05837274e-01 -2.26507634e-01
-1.70407761e-02 4.47009951e-01 -4.43221062e-01 -1.99528232e-01
5.44109404e-01 -1.18351293e+00 5.10901749e-01 2.90648490e-01
-5.21792769e-01 7.81151712e-01 6.96214080e-01 -1.00953913e+00
-1.86374342e+00 -9.15746450e-01 1.97035134e-01 -7.51669943e-01
1.64242119e-01 -7.27226615e-01 -9.34448063e-01 6.31255090e-01
-4.18909550e-01 -1.79109484e-01 -2.75991440e-01 -4.90185767e-02
-1.81088995e-04 1.86768204e-01 -1.34462070e+00 5.49878120e-01
1.34436166e+00 -1.11469857e-01 -5.04764080e-01 4.05696273e-01
7.38225818e-01 -9.70068812e-01 -9.85270321e-01 8.95284295e-01
7.41596043e-01 -4.80522037e-01 1.26462555e+00 -4.34907377e-01
5.90917766e-01 -2.17146069e-01 -4.13868010e-01 -1.26996839e+00
-1.42166421e-01 -6.03262544e-01 -1.64477900e-01 1.01930785e+00
1.69205293e-02 -3.52914065e-01 9.73292470e-01 7.29566455e-01
-2.85669774e-01 -1.25027227e+00 -8.95921707e-01 -7.56419837e-01
2.78966725e-01 1.02802299e-01 9.08009827e-01 4.96749461e-01
-1.54294506e-01 -3.35644007e-01 -3.35891508e-02 3.80873531e-01
9.30354178e-01 5.13529360e-01 7.54996240e-01 -1.20823300e+00
-2.67056465e-01 -3.51683646e-01 2.02624202e-02 -1.47533274e+00
2.44813263e-01 -7.18270659e-01 7.53849149e-01 -1.67876959e+00
4.17775631e-01 -1.12613547e+00 2.55521953e-01 5.62369108e-01
-3.63492429e-01 -2.41734087e-01 4.43233728e-01 3.61143798e-01
1.12469763e-01 7.50887573e-01 1.92884290e+00 -8.65324438e-02
-4.18464243e-01 1.83858961e-01 -3.05000901e-01 6.24855638e-01
4.54018205e-01 -2.21523345e-01 -3.75624508e-01 -6.60909534e-01
-1.58985570e-01 1.42782226e-01 4.18699801e-01 -5.80394745e-01
-2.03463495e-01 -7.28689015e-01 2.35015109e-01 -7.42324471e-01
5.77867210e-01 -8.49564970e-01 -2.30848212e-02 4.35431868e-01
-2.70182252e-01 -3.08714986e-01 1.54632971e-01 6.39937341e-01
2.91495711e-01 -4.03439194e-01 4.47229028e-01 -2.09731370e-01
-1.41132511e-02 6.16053224e-01 1.91695496e-01 -3.46004516e-01
1.03617406e+00 -1.32254153e-01 6.22325614e-02 -8.82721171e-02
-7.18254745e-01 2.71718115e-01 7.82692969e-01 8.10691118e-01
7.10402608e-01 -1.32658577e+00 -6.23935282e-01 2.45995790e-01
-9.91111696e-02 9.78290617e-01 -2.02501893e-01 2.80508816e-01
-3.20411175e-01 1.13160372e-01 -2.76332945e-01 -8.68063390e-01
-7.83354044e-01 2.96405435e-01 1.50122538e-01 1.09029949e-01
-6.13030255e-01 9.34871674e-01 2.66058743e-01 -8.03470671e-01
5.48486114e-01 -6.04990482e-01 2.89939195e-01 -3.25023115e-01
-6.93113878e-02 3.74921203e-01 -1.15403585e-01 -1.54559076e-01
-3.52116674e-01 9.27418888e-01 -1.37320340e-01 9.77579579e-02
1.60221505e+00 3.70422333e-01 -2.89707839e-01 1.81937262e-01
9.06248629e-01 -2.26402700e-01 -1.95739746e+00 1.28897086e-01
-2.30500534e-01 -9.01323974e-01 -5.18389642e-01 -8.17240953e-01
-8.77095759e-01 6.11263514e-01 3.08912754e-01 -5.39359637e-02
6.49481714e-01 3.88834208e-01 7.08727539e-01 4.71031308e-01
1.09377706e+00 -9.08136725e-01 4.04439300e-01 5.40558159e-01
1.62524617e+00 -1.00990891e+00 -5.02715744e-02 -1.12903261e+00
-3.06274325e-01 1.14614642e+00 9.20757234e-01 -4.61842716e-01
5.19673944e-01 4.21052545e-01 -1.77264467e-01 -3.47412437e-01
-1.59074634e-01 3.44214588e-01 4.76101995e-01 6.95424259e-01
1.50429651e-01 1.46670923e-01 -1.56639323e-01 4.65816826e-01
-1.71107963e-01 -4.64537293e-02 -3.71630751e-02 1.21423161e+00
-3.12169194e-01 -1.27346194e+00 -3.00605536e-01 4.93723869e-01
1.19696729e-01 4.19731051e-01 -3.68195176e-01 5.57430983e-01
7.09244162e-02 5.59975564e-01 5.00874296e-02 -3.42415512e-01
6.89352691e-01 -1.31212115e-01 1.17060316e+00 -9.95717645e-01
-1.91127002e-01 -1.40053928e-02 -4.67893928e-01 -8.16716373e-01
-1.23795763e-01 -7.30535865e-01 -1.39799249e+00 9.99878496e-02
-5.67598522e-01 -2.40814816e-02 5.85234582e-01 8.85916233e-01
2.52975881e-01 4.75843221e-01 4.89149928e-01 -1.83451211e+00
-1.08767319e+00 -9.03543949e-01 -4.76188660e-01 4.92853373e-01
7.81350210e-02 -1.25145376e+00 -1.56421721e-01 -1.36018381e-01]
|
[5.802511692047119, -0.8546320796012878]
|
ec7f5223-b5fe-4bee-a5a2-a04b5f9234f8
|
malcom-generating-malicious-comments-to
|
2009.01048
| null |
https://arxiv.org/abs/2009.01048v2
|
https://arxiv.org/pdf/2009.01048v2.pdf
|
MALCOM: Generating Malicious Comments to Attack Neural Fake News Detection Models
|
In recent years, the proliferation of so-called "fake news" has caused much disruptions in society and weakened the news ecosystem. Therefore, to mitigate such problems, researchers have developed state-of-the-art models to auto-detect fake news on social media using sophisticated data science and machine learning techniques. In this work, then, we ask "what if adversaries attempt to attack such detection models?" and investigate related issues by (i) proposing a novel threat model against fake news detectors, in which adversaries can post malicious comments toward news articles to mislead fake news detectors, and (ii) developing MALCOM, an end-to-end adversarial comment generation framework to achieve such an attack. Through a comprehensive evaluation, we demonstrate that about 94% and 93.5% of the time on average MALCOM can successfully mislead five of the latest neural detection models to always output targeted real and fake news labels. Furthermore, MALCOM can also fool black box fake news detectors to always output real news labels 90% of the time on average. We also compare our attack model with four baselines across two real-world datasets, not only on attack performance but also on generated quality, coherency, transferability, and robustness.
|
['Thai Le', 'Suhang Wang', 'Dongwon Lee']
|
2020-09-01
| null | null | null | null |
['comment-generation']
|
['natural-language-processing']
|
[-2.27825537e-01 2.73016512e-01 7.21868593e-03 -2.05501974e-01
-5.80094516e-01 -9.66826558e-01 1.04878008e+00 -7.37146959e-02
-1.80304736e-01 6.67080462e-01 -4.71224077e-02 -5.80206275e-01
8.66366208e-01 -8.77849519e-01 -9.70278323e-01 -1.22421429e-01
4.19503272e-01 2.40241885e-01 4.21854109e-01 -7.32662320e-01
3.75219464e-01 1.15981050e-01 -9.34150755e-01 6.73093557e-01
1.02256644e+00 5.41108727e-01 -7.59354055e-01 5.86518645e-01
2.22311348e-01 9.90449309e-01 -1.36707258e+00 -1.22193909e+00
2.46279836e-01 -4.83831882e-01 -5.59382737e-01 -3.20351303e-01
5.52957773e-01 -6.22603834e-01 -7.80967653e-01 1.68921232e+00
3.77064824e-01 -5.72046220e-01 4.70861316e-01 -1.51687264e+00
-1.19631207e+00 1.08993888e+00 -3.83343071e-01 2.42194921e-01
1.75082847e-01 4.78699327e-01 7.33579457e-01 -7.76340842e-01
7.02510536e-01 1.40008926e+00 7.16095448e-01 9.15223479e-01
-8.81034851e-01 -1.06821072e+00 5.40748127e-02 -7.71866962e-02
-8.12989712e-01 -5.19596517e-01 6.69622779e-01 -4.58562046e-01
4.34191525e-01 4.44775760e-01 4.66780365e-01 2.07744217e+00
5.02767980e-01 9.05850649e-01 1.28187323e+00 6.02909476e-02
8.90097916e-02 6.11558020e-01 1.44694746e-01 6.35205507e-01
6.75551295e-01 5.09731472e-01 -4.01652366e-01 -5.90783715e-01
3.28739494e-01 -1.49038732e-01 -2.41619587e-01 7.21255124e-01
-1.26594412e+00 1.24552202e+00 6.89610600e-01 3.10373753e-02
-9.57164541e-02 2.97817141e-01 6.09781921e-01 7.71656036e-01
1.07079768e+00 1.02598763e+00 -2.65618950e-01 1.28621444e-01
-6.27849698e-01 4.40013945e-01 1.13858747e+00 7.45005190e-01
1.33081421e-01 3.57049048e-01 -1.48257315e-01 4.65733886e-01
2.50651926e-01 1.09643722e+00 5.05278647e-01 -5.01637638e-01
4.78270590e-01 3.67586851e-01 5.29960871e-01 -1.65584421e+00
-1.95931390e-01 -7.13999808e-01 -8.23961020e-01 -6.50636777e-02
4.14880991e-01 -4.03206497e-01 -6.62119389e-01 1.44729924e+00
4.73635010e-02 3.45622718e-01 1.97450519e-02 1.09360576e+00
7.14757681e-01 7.81296253e-01 -2.50931144e-01 2.00417444e-01
1.33732283e+00 -1.15430713e+00 -6.60001934e-01 -5.52453697e-01
7.60649323e-01 -9.12024379e-01 1.03452671e+00 3.77772659e-01
-7.95730233e-01 8.26370344e-02 -1.08517230e+00 3.19538146e-01
-7.78435469e-01 1.17419725e-02 4.48340267e-01 1.00236845e+00
-6.87796295e-01 3.65344167e-01 -5.87371945e-01 4.77889851e-02
5.22673368e-01 -2.52971500e-01 -7.54720494e-02 9.14541483e-02
-1.87648284e+00 1.08390450e+00 -5.08718751e-02 -2.00466722e-01
-1.16001010e+00 -4.98708665e-01 -2.52726316e-01 -2.11292297e-01
8.94480944e-02 -5.77392578e-01 1.29573822e+00 -1.37917590e+00
-1.36676574e+00 8.57623696e-01 3.66692036e-01 -7.65518129e-01
1.06256318e+00 -2.64533103e-01 -9.32254493e-01 4.57356758e-02
2.58208632e-01 1.96563751e-01 1.17930353e+00 -1.21588576e+00
-4.26974922e-01 -8.96310657e-02 7.54384547e-02 -3.12087089e-01
-4.24383521e-01 3.23241532e-01 2.44118422e-01 -1.05873811e+00
-7.71185756e-02 -1.08691084e+00 5.97792864e-02 -1.02098145e-01
-9.28838253e-01 5.10521829e-02 1.25329685e+00 -5.44568777e-01
8.19097757e-01 -1.93168819e+00 -4.58472192e-01 2.16562133e-02
6.85336709e-01 6.98507905e-01 -1.20215192e-01 2.67193317e-01
2.01190248e-01 5.17749786e-01 2.42487505e-01 -2.95898944e-01
-3.72015708e-03 -2.42752507e-01 -1.25745499e+00 8.14113200e-01
2.65036792e-01 1.13284922e+00 -1.27967203e+00 2.38951996e-01
-2.91599602e-01 2.68602878e-01 -5.88452458e-01 1.00021232e-02
-4.18992758e-01 3.01619947e-01 -3.33541393e-01 7.22984433e-01
6.04831517e-01 -4.73135769e-01 -2.45075241e-01 2.45565012e-01
2.92730510e-01 5.04163146e-01 -4.45865095e-01 6.18120432e-01
-5.94532676e-02 1.01355982e+00 -1.59825295e-01 -8.23940694e-01
8.35470378e-01 2.30661646e-01 -1.88351437e-01 -5.67159355e-01
6.07920825e-01 4.93277609e-01 -1.48419738e-01 -3.11634213e-01
5.93626261e-01 1.23187244e-01 -4.01487350e-01 8.29530537e-01
-2.55878091e-01 4.08888198e-02 -4.32096452e-01 5.89124799e-01
1.34143317e+00 -6.94791973e-01 -1.87974900e-01 1.17754862e-01
1.43957764e-01 2.76035517e-01 1.32696539e-01 1.30228949e+00
-4.87466127e-01 3.21139455e-01 6.94415331e-01 -5.26572764e-01
-1.11049676e+00 -9.45434093e-01 3.11798096e-01 1.06637859e+00
3.52723449e-01 1.75462589e-02 -8.41552973e-01 -1.21494830e+00
1.81910813e-01 9.42404985e-01 -6.95392132e-01 -7.06309140e-01
-3.17512065e-01 -1.01101863e+00 1.49814188e+00 -1.66792810e-01
7.81348288e-01 -8.49063516e-01 4.67482209e-02 2.46872216e-01
-4.80830610e-01 -1.22909296e+00 -5.70605338e-01 -5.97877383e-01
-2.55046189e-01 -9.36989129e-01 -4.53896075e-01 -3.58276010e-01
7.24309742e-01 7.58733273e-01 1.03240061e+00 4.62777555e-01
1.21789172e-01 -2.55039245e-01 -7.42218733e-01 -7.62809694e-01
-1.27006114e+00 8.55716784e-03 3.57452542e-01 1.03602983e-01
2.92043865e-01 -5.73519580e-02 -3.44973326e-01 7.46069670e-01
-9.51815665e-01 -1.30244374e-01 5.66316247e-01 8.14150095e-01
-3.18453223e-01 -2.75197595e-01 1.19284976e+00 -1.34003615e+00
9.82434988e-01 -9.69226837e-01 -7.58242309e-01 -8.08599114e-04
-8.28992903e-01 -2.86041439e-01 1.03559649e+00 -9.42860961e-01
-5.52295327e-01 -7.13907182e-01 -4.38576415e-02 -3.42613429e-01
1.85922742e-01 3.69203061e-01 3.27309787e-01 -2.74800748e-01
1.46580362e+00 3.16757083e-01 -2.33944133e-03 4.80229445e-02
4.99641746e-01 1.04701531e+00 4.84654397e-01 -7.14566633e-02
1.36153162e+00 7.91547060e-01 -6.97530329e-01 -5.16117930e-01
-1.24598324e+00 -1.35799691e-01 3.24595273e-01 -2.28331387e-01
3.68948758e-01 -9.93895769e-01 -5.41177511e-01 1.14172804e+00
-1.75298667e+00 4.93649786e-05 2.54689574e-01 2.45973259e-01
3.11161168e-02 3.14564377e-01 -1.10347545e+00 -5.60124099e-01
-5.24675488e-01 -9.66497660e-01 6.04704738e-01 -1.64356679e-01
2.33495235e-02 -9.07655358e-01 -3.82147059e-02 5.94995201e-01
7.97811925e-01 2.74061590e-01 2.37017855e-01 -1.23381293e+00
-3.81704539e-01 -8.00822079e-01 -6.11363709e-01 5.55220187e-01
-2.71177560e-01 1.08569950e-01 -1.09444499e+00 -2.58392274e-01
7.79373422e-02 -2.82702923e-01 8.11683655e-01 -2.74273038e-01
5.24024189e-01 -1.06390440e+00 -3.53157341e-01 3.73869061e-01
8.08548570e-01 -2.94294983e-01 6.05698526e-01 3.13086390e-01
6.47405505e-01 4.18923825e-01 4.47650880e-01 2.40699708e-01
3.24158579e-01 4.18636769e-01 7.38835454e-01 -2.67985221e-02
9.03454423e-02 -6.15199208e-01 7.51046836e-01 7.02002525e-01
5.17513275e-01 -7.17785478e-01 -6.35168552e-01 3.01524013e-01
-1.79208362e+00 -1.26680660e+00 -5.45067430e-01 1.67150140e+00
8.43101859e-01 4.95167226e-01 1.98411167e-01 -1.55052230e-01
1.02878869e+00 2.73060322e-01 -4.94408846e-01 -2.66521275e-01
-3.87968987e-01 -4.33624446e-01 9.31631804e-01 4.18590069e-01
-1.25179219e+00 1.48422003e+00 6.34586763e+00 6.57335162e-01
-1.50527084e+00 4.87877876e-01 7.16741860e-01 1.53323919e-01
-3.86285365e-01 -5.59310503e-02 -6.97214127e-01 1.00456309e+00
1.06319499e+00 -1.04906648e-01 5.72616398e-01 1.16197705e+00
3.93943667e-01 4.70347166e-01 -4.71221149e-01 6.01050735e-01
3.39740098e-01 -1.76854181e+00 1.87339455e-01 3.86597402e-02
1.09833086e+00 4.33135688e-01 3.14729780e-01 4.66097057e-01
8.80898356e-01 -9.96332407e-01 1.01628017e+00 2.90158689e-01
5.31769633e-01 -4.96505976e-01 1.00659335e+00 8.19121718e-01
-1.03896655e-01 2.75155399e-02 -4.77534384e-01 -1.39283389e-01
1.05669335e-01 8.98225188e-01 -1.11705542e+00 9.18395147e-02
3.52785707e-01 5.69677889e-01 -5.78784406e-01 7.82884896e-01
-6.43132210e-01 1.12359965e+00 -1.78983316e-01 -4.95989680e-01
4.92202699e-01 3.56927961e-01 8.32926095e-01 1.29209673e+00
3.47248875e-02 -1.81317449e-01 -1.01233587e-01 1.14662778e+00
-6.18675590e-01 -3.43409061e-01 -8.21404696e-01 -2.74341345e-01
7.49177754e-01 9.35416400e-01 -4.18654799e-01 -5.90799809e-01
-1.07328422e-01 1.16623664e+00 1.33563951e-01 1.71261117e-01
-1.28309143e+00 -4.83388573e-01 6.20835602e-01 2.24182889e-01
-3.46594781e-01 9.76253301e-02 -3.23764294e-01 -1.70375371e+00
-1.52472705e-01 -1.25742686e+00 -6.70822337e-02 -6.79264784e-01
-1.88500464e+00 7.28185713e-01 -5.97976327e-01 -1.24656355e+00
1.12398684e-01 -3.39218229e-01 -8.29006970e-01 4.92335528e-01
-1.40100789e+00 -1.05997717e+00 -1.29102170e-01 4.48444009e-01
2.98774928e-01 -5.90169072e-01 5.75391412e-01 2.92258233e-01
-5.06163120e-01 9.73562837e-01 1.92947596e-01 7.40890205e-01
9.83693898e-01 -8.79177034e-01 1.21869421e+00 1.04763460e+00
1.48424357e-01 3.73697519e-01 1.10299957e+00 -8.97780478e-01
-1.04916930e+00 -1.35219014e+00 8.85425270e-01 -9.10721362e-01
1.17102194e+00 -5.29474318e-01 -8.68916690e-01 6.52082682e-01
-1.63185135e-01 2.30749875e-01 2.39018902e-01 -5.00633597e-01
-1.09736562e+00 3.35314035e-01 -1.44862413e+00 7.33341992e-01
6.77013636e-01 -6.43800676e-01 -5.13098001e-01 1.01887417e+00
1.14037693e+00 -4.66883510e-01 -8.27081650e-02 -1.94593698e-01
4.90868449e-01 -1.07296634e+00 7.67444730e-01 -1.05414784e+00
6.85442984e-01 -9.19827744e-02 2.84192443e-01 -1.67711663e+00
-2.17209831e-02 -8.76215458e-01 -1.24220312e-01 8.05259943e-01
7.45189965e-01 -1.30494690e+00 7.67747879e-01 -6.50830194e-02
-7.23538250e-02 -3.02101701e-01 -7.09915519e-01 -8.45767617e-01
1.59605652e-01 -3.19820255e-01 3.58989209e-01 1.60646188e+00
-8.61770064e-02 2.33273596e-01 -9.25004303e-01 6.00453079e-01
5.77210486e-01 -4.02194679e-01 1.05920172e+00 -1.08132327e+00
-2.62086272e-01 -4.82929498e-01 -3.44032794e-01 -1.07369280e+00
1.68973789e-01 -7.97332406e-01 -1.57021061e-01 -8.51695418e-01
-1.76813900e-01 -2.25729555e-01 1.60778269e-01 3.21212828e-01
-2.05902487e-01 6.17392004e-01 4.74994853e-02 6.68320000e-01
-4.29049492e-01 2.45208040e-01 1.18440807e+00 -4.21541542e-01
1.37812391e-01 2.99407005e-01 -9.69121099e-01 8.39875937e-01
7.16683447e-01 -1.09580672e+00 3.89310233e-02 -5.09839892e-01
7.24710882e-01 -7.42223784e-02 8.72077823e-01 -6.15100026e-01
1.14984304e-01 -1.54247656e-01 -7.16479495e-02 -1.19850978e-01
5.65062985e-02 -3.46889675e-01 -4.54626918e-01 7.48147666e-01
-3.98280770e-01 9.93246064e-02 -3.09071749e-01 8.86028886e-01
6.23200322e-03 1.12633314e-02 1.09356129e+00 -1.78806543e-01
-1.59391891e-02 7.37864897e-02 -6.60343230e-01 3.11826557e-01
9.78607833e-01 3.92711550e-01 -1.56889117e+00 -8.88629556e-01
-4.16731685e-01 1.55081656e-02 4.49999690e-01 5.82601666e-01
6.15564525e-01 -1.03791118e+00 -1.18369293e+00 2.10055690e-02
2.54729748e-01 -5.95714271e-01 -1.78233758e-01 7.59442687e-01
-8.44858468e-01 1.24479517e-01 -1.19769527e-02 -1.61773324e-01
-8.70494843e-01 5.52028120e-01 4.47913170e-01 -1.32994235e-01
-3.05876553e-01 1.00969172e+00 -3.55104476e-01 -5.94102859e-01
-1.35180354e-01 7.76650608e-02 3.19910973e-01 -1.50523826e-01
6.62362158e-01 3.08235317e-01 -3.28456275e-02 -5.77307403e-01
-1.47911072e-01 -4.27540988e-01 -4.11183864e-01 4.32662144e-02
8.32266271e-01 1.87899098e-01 -1.00488096e-01 -7.41417473e-03
9.29985404e-01 3.57794970e-01 -6.91067636e-01 -3.61343920e-01
-2.14346856e-01 -6.07920885e-01 -3.57647352e-02 -1.25227153e+00
-9.57812786e-01 5.76640725e-01 6.81652129e-02 9.65816975e-01
2.13041484e-01 3.27235996e-03 1.20329618e+00 5.74456334e-01
3.70436877e-01 -8.66612017e-01 5.81682444e-01 7.15089083e-01
8.06297898e-01 -1.41847587e+00 -2.93913394e-01 -4.35534984e-01
-8.02285552e-01 8.24813426e-01 5.06677270e-01 -6.84039474e-01
5.30542612e-01 8.01606220e-04 5.17994285e-01 -3.03147048e-01
-8.43899071e-01 6.53060794e-01 2.17315834e-02 3.23047221e-01
-4.14212607e-02 3.07486355e-01 -2.44706810e-01 6.47859693e-01
-5.02170503e-01 -5.33728242e-01 1.13710284e+00 4.93963420e-01
-7.06437647e-01 -6.28684223e-01 -5.66045344e-01 5.50949275e-01
-6.94664299e-01 -2.27003247e-01 -9.47521329e-01 4.54087079e-01
-2.07290173e-01 1.37002814e+00 -3.48372698e-01 -8.89131486e-01
2.13284820e-01 -2.79273838e-01 -3.91733348e-01 -5.61953187e-01
-1.04120803e+00 -4.18103039e-01 2.49542266e-01 -3.81188065e-01
1.85012966e-01 -1.76062077e-01 -7.64859378e-01 -1.05468094e+00
-8.26202452e-01 1.18164778e-01 8.89065683e-01 1.01229739e+00
5.49920321e-01 8.26407671e-02 1.13204741e+00 -4.56828028e-01
-1.33859003e+00 -1.18969882e+00 -1.65346116e-01 5.82676232e-01
4.11956906e-01 -4.15914834e-01 -1.13842547e+00 -3.11621815e-01]
|
[8.135107040405273, 10.228588104248047]
|
cdd03311-5f57-44c2-a4f9-9bb7d4e4c801
|
landmark-aware-and-part-based-ensemble
|
2104.11274
| null |
https://arxiv.org/abs/2104.11274v2
|
https://arxiv.org/pdf/2104.11274v2.pdf
|
Landmark-Aware and Part-based Ensemble Transfer Learning Network for Facial Expression Recognition from Static images
|
Facial Expression Recognition from static images is a challenging problem in computer vision applications. Convolutional Neural Network (CNN), the state-of-the-art method for various computer vision tasks, has had limited success in predicting expressions from faces having extreme poses, illumination, and occlusion conditions. To mitigate this issue, CNNs are often accompanied by techniques like transfer, multi-task, or ensemble learning that often provide high accuracy at the cost of increased computational complexity. In this work, we propose a Part-based Ensemble Transfer Learning network that models how humans recognize facial expressions by correlating the spatial orientation pattern of the facial features with a specific expression. It consists of 5 sub-networks, and each sub-network performs transfer learning from one of the five subsets of facial landmarks: eyebrows, eyes, nose, mouth, or jaw to expression classification. We show that our proposed ensemble network uses visual patterns emanating from facial muscles' motor movements to predict expressions and demonstrate the usefulness of transfer learning from Facial Landmark Localization to Facial Expression Recognition. We test the proposed network on the CK+, JAFFE, and SFEW datasets, and it outperforms the benchmark for CK+ and JAFFE datasets by 0.51% and 5.34%, respectively. Additionally, the proposed ensemble network consists of only 1.65M model parameters, ensuring computational efficiency during training and real-time deployment. Grad-CAM visualizations of our proposed ensemble highlight the complementary nature of its sub-networks, a key design parameter of an effective ensemble network. Lastly, cross-dataset evaluation results reveal that our proposed ensemble has a high generalization capacity, making it suitable for real-world usage.
|
['Tapan K. Gandhi', 'Rohan Wadhawan']
|
2021-04-22
| null | null | null | null |
['face-alignment']
|
['computer-vision']
|
[ 1.63036641e-02 -2.17740908e-01 -1.36559337e-01 -6.69044256e-01
-3.05069208e-01 -3.07064444e-01 3.34740013e-01 -6.24235630e-01
-4.63054955e-01 6.29501343e-01 -2.10511118e-01 9.96295139e-02
-6.25767559e-02 -4.06211078e-01 -4.96825546e-01 -9.60565269e-01
-1.70189291e-01 -9.45321620e-02 -3.88538480e-01 -3.45820755e-01
-4.27610986e-02 8.79236579e-01 -1.70432305e+00 3.00476849e-01
5.44379532e-01 1.68623257e+00 -3.68500143e-01 2.10175157e-01
-4.73872349e-02 7.21213818e-01 -6.90725923e-01 -4.53639090e-01
2.24159956e-02 -2.83230275e-01 -4.29671228e-01 2.10010353e-02
3.92207891e-01 -2.65345454e-01 -4.18206565e-02 8.65714550e-01
7.38699377e-01 -3.69502828e-02 7.55032778e-01 -1.72176874e+00
-6.18414342e-01 -3.13878387e-01 -9.52490091e-01 -1.04152985e-01
2.19269022e-01 -4.44241613e-02 6.20794654e-01 -9.81113076e-01
4.60903227e-01 9.94442880e-01 7.38189936e-01 7.38969743e-01
-8.94417763e-01 -1.30941367e+00 -3.86542417e-02 2.08632007e-01
-1.44962084e+00 -6.96372867e-01 1.01955533e+00 -3.82920951e-01
7.33446300e-01 9.84303579e-02 4.65599060e-01 1.20457196e+00
1.06803618e-01 7.33233452e-01 1.45279968e+00 -2.19466105e-01
4.40117791e-02 2.90228397e-01 -2.52965033e-01 8.39696169e-01
-3.46436292e-01 -7.20775872e-02 -6.32621109e-01 -8.41166526e-02
6.75862074e-01 1.43963620e-02 -2.10896447e-01 -1.22441478e-01
-4.07447815e-01 7.14000523e-01 6.04425550e-01 1.86722100e-01
-4.79143351e-01 6.97633624e-02 5.31261444e-01 2.61578172e-01
6.89206302e-01 1.85228083e-02 -5.34896731e-01 -1.94090813e-01
-5.77696562e-01 -3.33530307e-02 5.31694233e-01 6.69285655e-01
7.05879569e-01 1.21368147e-01 8.85601565e-02 1.21410251e+00
2.85373539e-01 5.35273731e-01 3.63807738e-01 -7.36660182e-01
-2.96765892e-03 6.05812490e-01 -1.67781457e-01 -1.31476057e+00
-4.94016439e-01 -2.87013978e-01 -1.07093239e+00 4.52535033e-01
1.72946662e-01 -3.68469685e-01 -6.59435511e-01 2.00238347e+00
3.51614058e-01 1.56158850e-01 4.82645445e-02 8.53758872e-01
9.54627872e-01 4.24849778e-01 3.37209404e-01 -2.18569413e-01
1.16360140e+00 -8.15865517e-01 -6.79532647e-01 5.36956824e-04
5.68075597e-01 -5.54747581e-01 6.16474748e-01 4.43761230e-01
-7.73401499e-01 -6.27012610e-01 -7.70523727e-01 3.35002810e-01
-3.40676188e-01 5.48071146e-01 8.44229579e-01 4.87984836e-01
-9.36425745e-01 4.12102520e-01 -3.27069342e-01 -3.10692281e-01
6.92668974e-01 7.38355279e-01 -8.93095255e-01 2.36887485e-01
-9.86340523e-01 8.81831288e-01 -6.68764953e-03 5.94557524e-01
-4.50221807e-01 -3.79429281e-01 -7.69518912e-01 6.72174692e-02
-5.10335565e-02 -2.97861516e-01 9.97221410e-01 -1.78530681e+00
-1.74078226e+00 9.77643669e-01 -1.74155220e-01 1.65978283e-01
1.64592370e-01 -1.47907332e-01 -6.56563103e-01 1.30243838e-01
-2.66015917e-01 6.68509603e-01 9.11708415e-01 -1.09113300e+00
-2.11806595e-01 -7.66806602e-01 -2.94684440e-01 -1.25886721e-03
-7.16056526e-01 3.15482378e-01 -1.35164529e-01 -2.75269747e-01
-1.62758172e-01 -1.07188439e+00 1.84141561e-01 4.12614137e-01
2.32458077e-02 -3.72110188e-01 1.13846767e+00 -5.01947701e-01
8.64221275e-01 -2.30828762e+00 1.44073904e-01 3.61370236e-01
5.46020903e-02 3.75999421e-01 -2.60788143e-01 1.18478350e-01
-3.49590570e-01 -5.98620102e-02 5.18807955e-02 -4.43904042e-01
-2.27592722e-01 2.02791721e-01 7.17904121e-02 5.71540952e-01
4.08188969e-01 8.24082613e-01 -5.70097506e-01 -5.06729484e-01
1.57036647e-01 7.18699336e-01 -2.15350255e-01 2.99051106e-01
1.40374228e-01 7.00017989e-01 -4.32271183e-01 9.03571188e-01
7.79022574e-01 -1.10692874e-01 2.02670738e-01 -4.25195754e-01
1.45650193e-01 -6.03618622e-01 -8.55494320e-01 1.47071695e+00
-7.40144491e-01 6.46283448e-01 2.36358643e-01 -1.01992691e+00
1.42761314e+00 3.75129789e-01 6.18722796e-01 -8.00951719e-01
4.54910457e-01 2.57677674e-01 1.17828744e-03 -7.68405020e-01
-5.47518320e-02 -5.29269390e-02 1.56156734e-01 3.19760263e-01
4.62727815e-01 3.07507128e-01 -2.19143167e-01 -4.40542519e-01
6.66851401e-01 2.33236298e-01 3.40670913e-01 5.67746535e-02
7.96863675e-01 -6.99088752e-01 6.04287326e-01 -5.10702580e-02
-3.99616092e-01 1.91780046e-01 4.74752367e-01 -6.72058821e-01
-7.30480969e-01 -6.50180817e-01 -1.79975405e-01 1.33189261e+00
-1.95113912e-01 -1.14406772e-01 -6.65675640e-01 -5.96464396e-01
-9.75165889e-03 1.76295057e-01 -8.91825438e-01 -1.74029201e-01
-3.54690701e-01 -6.58805788e-01 7.31499672e-01 5.92526257e-01
7.68356025e-01 -1.31801021e+00 -5.18687963e-01 -1.76422641e-01
5.16383909e-02 -1.20421016e+00 -7.68448487e-02 -1.17732383e-01
-6.02100790e-01 -9.97163057e-01 -8.07759345e-01 -7.84126878e-01
8.61945212e-01 -1.08848855e-01 7.63726771e-01 1.29655391e-01
-2.11588055e-01 3.03891599e-01 -3.39290053e-01 -4.94258046e-01
-6.79077432e-02 -2.88498342e-01 1.21587738e-01 8.39639783e-01
4.98989910e-01 -8.64897668e-01 -6.55150712e-01 4.60756391e-01
-6.15700066e-01 -1.71485633e-01 7.03087747e-01 1.01486981e+00
2.68442512e-01 -3.53815764e-01 6.68873966e-01 -5.02954602e-01
6.52275860e-01 -5.53500533e-01 -3.51859182e-01 2.12737262e-01
-2.62941748e-01 -3.22794557e-01 6.15357935e-01 -5.15059412e-01
-1.01677394e+00 3.36813748e-01 -1.47974089e-01 -6.93091869e-01
-2.72683650e-01 4.27173853e-01 -1.38721853e-01 -6.32144332e-01
5.05495250e-01 2.14806765e-01 6.69266939e-01 -1.98742166e-01
1.41569883e-01 8.13162625e-01 4.31107849e-01 -5.54304063e-01
3.79144251e-01 4.25246835e-01 2.93895960e-01 -9.08129394e-01
-6.36757016e-01 -3.32835257e-01 -6.67310476e-01 -6.19389951e-01
8.50806892e-01 -8.52068901e-01 -1.22674668e+00 7.81810582e-01
-1.22044003e+00 7.82683268e-02 4.67814654e-01 1.44031689e-01
-4.72523659e-01 4.71696146e-02 -4.32295740e-01 -1.03124011e+00
-5.47824860e-01 -1.07398248e+00 1.21172440e+00 4.03631687e-01
-2.94400483e-01 -8.74262810e-01 -1.07901148e-01 2.88947374e-01
5.92406392e-01 7.48504639e-01 6.25546873e-01 -4.84462351e-01
4.99829911e-02 -3.06454033e-01 -2.88178951e-01 6.02272153e-01
2.68717170e-01 3.24377090e-01 -1.32176697e+00 -1.59168527e-01
-2.31853873e-01 -7.28286266e-01 4.69079077e-01 2.24658489e-01
1.59006703e+00 -2.05211803e-01 -2.09300518e-01 8.48994017e-01
1.13727081e+00 3.94463360e-01 6.27914310e-01 2.31818512e-01
3.74882370e-01 7.58591712e-01 4.25426334e-01 6.11916244e-01
1.19668603e-01 7.92759538e-01 6.09050930e-01 -4.92778242e-01
2.93459803e-01 4.44037318e-02 2.18268529e-01 4.91013795e-01
-5.12687385e-01 1.73234105e-01 -6.44654453e-01 1.38431057e-01
-1.74107718e+00 -9.83780384e-01 2.66033709e-01 1.78676105e+00
5.26755810e-01 -3.71290207e-01 1.74045101e-01 8.61884058e-02
3.47200423e-01 1.36225997e-02 -5.01278818e-01 -6.41826868e-01
-1.65015876e-01 6.04077578e-01 3.44899073e-02 -6.94048852e-02
-1.26047730e+00 8.23126674e-01 5.38745975e+00 5.72434008e-01
-1.69194055e+00 6.33049384e-02 8.97432625e-01 -6.11940958e-02
3.99890363e-01 -6.34832919e-01 -5.18395126e-01 4.53245252e-01
7.27413416e-01 4.87011150e-02 1.30429894e-01 1.06952858e+00
1.31686106e-01 1.29354805e-01 -9.06828463e-01 1.32624173e+00
3.13052565e-01 -9.73150373e-01 -2.68615037e-01 -5.93412369e-02
5.49399018e-01 -4.07841802e-01 4.75307792e-01 3.63791585e-01
-2.71484256e-01 -1.46468830e+00 2.24293709e-01 4.82700467e-01
9.61188018e-01 -9.11456823e-01 9.63537872e-01 9.98029858e-02
-9.97091591e-01 -2.28667691e-01 4.35646437e-03 -8.04887116e-02
-2.26636268e-02 5.40282428e-02 -6.34262443e-01 3.31961751e-01
9.14854586e-01 7.12028086e-01 -4.14950788e-01 4.97266859e-01
-1.31753042e-01 3.28529298e-01 -2.49206543e-01 -2.28454471e-01
1.46467134e-01 -2.81115919e-01 3.71168032e-02 1.19516742e+00
4.17943627e-01 1.61663011e-01 -1.60935789e-01 7.98948169e-01
-1.19259894e-01 3.42496574e-01 -7.97359645e-01 1.15135156e-01
1.47961780e-01 1.72988105e+00 -1.96144417e-01 4.31120507e-02
-3.79962176e-01 9.28091347e-01 4.90289837e-01 3.97678047e-01
-9.02123153e-01 -2.15186343e-01 1.03669643e+00 -3.36798102e-01
1.95702314e-01 1.10693648e-02 -2.32306924e-02 -7.06510425e-01
1.06997311e-01 -9.74727809e-01 4.90318350e-02 -8.67903829e-01
-1.12686026e+00 9.58862305e-01 -2.31820911e-01 -1.22604847e+00
-3.41960728e-01 -1.04130638e+00 -6.74365997e-01 9.06657219e-01
-1.37715137e+00 -1.43660474e+00 -8.02937567e-01 7.34165609e-01
1.44765735e-01 -4.50111896e-01 1.24822938e+00 4.13981140e-01
-7.88439155e-01 9.87136722e-01 -2.36181095e-01 4.78835315e-01
6.66965067e-01 -6.64060414e-01 -3.06806087e-01 2.25035846e-01
-5.90457395e-02 4.55910504e-01 3.56322795e-01 6.15932718e-02
-1.24594390e+00 -1.03374815e+00 5.61271787e-01 -9.29447338e-02
4.17828411e-01 -3.36633801e-01 -7.44596720e-01 5.56812346e-01
6.35148436e-02 5.24814367e-01 1.01162612e+00 2.47977391e-01
-5.03361523e-01 -5.39961219e-01 -1.20822489e+00 3.93400997e-01
8.10126781e-01 -4.54977244e-01 -1.55686187e-02 2.46282801e-01
4.73049283e-03 -3.41217220e-01 -1.06623793e+00 6.11787915e-01
1.11359513e+00 -1.23202264e+00 6.56236589e-01 -8.46529126e-01
6.62425518e-01 -8.52833409e-03 -1.32360533e-01 -1.39318609e+00
-1.10935688e-01 -4.33359802e-01 5.69604598e-02 1.21923244e+00
3.29606533e-01 -6.35736048e-01 8.24313581e-01 4.24834460e-01
1.29003525e-01 -1.22494173e+00 -9.83568192e-01 -5.25520921e-01
-1.15790658e-01 -1.93358138e-01 5.88857949e-01 9.98520255e-01
-1.29612371e-01 3.88043433e-01 -4.77904171e-01 -1.25715673e-01
3.69317293e-01 1.26537364e-02 1.02209795e+00 -1.12958479e+00
-6.49583340e-02 -5.76100111e-01 -7.54408956e-01 -6.02687418e-01
8.11258435e-01 -7.40460217e-01 -4.24428254e-01 -8.93300653e-01
1.39986023e-01 -4.23693448e-01 -4.52045381e-01 7.79420793e-01
2.12285578e-01 3.47847551e-01 2.64587462e-01 -4.58557978e-02
-2.99402952e-01 7.32266843e-01 1.24562335e+00 -8.85913670e-02
-6.57485798e-02 3.27580459e-02 -4.27670777e-01 8.10011089e-01
7.76716232e-01 -3.66905741e-02 -2.98306137e-01 -2.42909089e-01
-1.41848713e-01 8.54269788e-02 4.64669049e-01 -6.85422897e-01
1.84994370e-01 1.49476603e-02 8.13108146e-01 -6.14245906e-02
8.14059436e-01 -1.06569242e+00 2.62345076e-01 1.62598714e-01
-9.55078080e-02 1.57087937e-01 4.52630609e-01 6.86316341e-02
-5.44368327e-01 1.55054852e-01 9.09429669e-01 8.38473514e-02
-1.04557443e+00 5.30551910e-01 -2.29467228e-02 -4.56665963e-01
1.35452294e+00 -3.73690993e-01 -2.01784670e-02 -6.52935028e-01
-7.13743746e-01 -1.99662894e-03 -1.88336975e-03 5.05942822e-01
7.21979737e-01 -1.51037741e+00 -6.13766670e-01 3.94687265e-01
3.08299631e-01 -2.37894803e-01 3.63164902e-01 1.09433353e+00
-2.66633302e-01 1.08080171e-01 -7.05954015e-01 -7.43910849e-01
-1.70546770e+00 1.38654128e-01 6.22099102e-01 1.30262390e-01
-1.08899120e-02 9.46496844e-01 1.17987990e-01 -4.63184923e-01
9.89072770e-02 3.02427918e-01 -3.80172282e-01 4.37552668e-02
4.67804790e-01 2.49291152e-01 2.41665039e-02 -1.02567387e+00
-4.37111914e-01 9.24498439e-01 5.80033958e-02 3.29500407e-01
1.36324215e+00 3.31054747e-01 -3.88154864e-01 1.06313720e-01
1.64549792e+00 -3.94445956e-01 -9.29114223e-01 -1.12556607e-01
-2.29389757e-01 -3.00965309e-01 -5.40041663e-02 -8.12530041e-01
-1.38726330e+00 8.64130795e-01 7.72597373e-01 -3.95556271e-01
1.75200653e+00 -1.80185825e-01 4.70542461e-01 1.73556671e-01
4.29343134e-01 -8.94488573e-01 2.77758926e-01 2.49096215e-01
1.10515022e+00 -1.43085849e+00 -2.19189495e-01 -2.66826183e-01
-7.58370161e-01 1.37631035e+00 1.03177226e+00 -2.67410800e-02
7.51582384e-01 2.05002397e-01 3.24884385e-01 -2.03017384e-01
-7.38157094e-01 1.40946552e-01 2.90095955e-01 6.03203893e-01
5.96652031e-01 4.02539596e-02 7.83286691e-02 5.56076169e-01
-1.95765495e-01 2.56689161e-01 -1.13781430e-01 7.13221312e-01
-8.77777711e-02 -8.16690683e-01 -1.72520399e-01 2.76098639e-01
-4.67632830e-01 3.10757577e-01 -4.95798498e-01 1.07543147e+00
2.60811985e-01 7.47705281e-01 2.09021911e-01 -5.64629018e-01
3.57655466e-01 1.85759217e-01 6.31226242e-01 -4.27627861e-02
-5.39397717e-01 -1.22274593e-01 -5.41157797e-02 -6.85519695e-01
-7.38047540e-01 -4.15040970e-01 -9.18818474e-01 -2.50830680e-01
-7.15475455e-02 -4.66773361e-02 8.23788404e-01 8.30207407e-01
5.79267383e-01 1.55347660e-01 1.01308453e+00 -9.96983588e-01
-6.05813801e-01 -1.18573701e+00 -6.03112757e-01 5.75300217e-01
3.14156592e-01 -1.03837574e+00 -4.30210173e-01 -8.30738246e-02]
|
[13.582588195800781, 1.7392829656600952]
|
4acced89-f79c-468a-be42-f4b634794cab
|
uncertainty-driven-trajectory-truncation-for
|
2304.0466
| null |
https://arxiv.org/abs/2304.04660v1
|
https://arxiv.org/pdf/2304.04660v1.pdf
|
Uncertainty-driven Trajectory Truncation for Model-based Offline Reinforcement Learning
|
Equipped with the trained environmental dynamics, model-based offline reinforcement learning (RL) algorithms can often successfully learn good policies from fixed-sized datasets, even some datasets with poor quality. Unfortunately, however, it can not be guaranteed that the generated samples from the trained dynamics model are reliable (e.g., some synthetic samples may lie outside of the support region of the static dataset). To address this issue, we propose Trajectory Truncation with Uncertainty (TATU), which adaptively truncates the synthetic trajectory if the accumulated uncertainty along the trajectory is too large. We theoretically show the performance bound of TATU to justify its benefits. To empirically show the advantages of TATU, we first combine it with two classical model-based offline RL algorithms, MOPO and COMBO. Furthermore, we integrate TATU with several off-the-shelf model-free offline RL algorithms, e.g., BCQ. Experimental results on the D4RL benchmark show that TATU significantly improves their performance, often by a large margin.
|
['Xiu Li', 'Le Wan', 'Jun Yang', 'Jiangpeng Yan', 'Xiaoteng Ma', 'Jiafei Lyu', 'Junjie Zhang']
|
2023-04-10
| null | null | null | null |
['offline-rl', 'd4rl']
|
['playing-games', 'robots']
|
[-4.87652749e-01 2.96704695e-02 -4.35251743e-01 1.38889104e-01
-9.24247861e-01 -8.54647219e-01 4.53620017e-01 5.49213327e-02
-4.09588993e-01 1.32685018e+00 -2.59476304e-01 -5.67625880e-01
-2.18278587e-01 -7.31080592e-01 -1.23081255e+00 -7.24193633e-01
-2.80457050e-01 6.60287261e-01 2.68491328e-01 -5.12807779e-02
-2.25312207e-02 2.52768040e-01 -1.45456553e+00 -2.81857133e-01
1.28855693e+00 9.12868619e-01 1.11971729e-01 4.11852181e-01
2.52553225e-02 8.20537448e-01 -6.36729002e-01 5.81706954e-05
4.41150486e-01 -3.31109554e-01 -4.51906830e-01 -3.43759134e-02
-2.79223114e-01 -7.70213068e-01 -2.68378288e-01 9.31880534e-01
2.60155380e-01 4.18924868e-01 3.39195907e-01 -1.45213962e+00
3.74194197e-02 8.62752497e-01 -3.54684532e-01 -7.58644938e-02
9.42158177e-02 6.44351661e-01 5.73320150e-01 -6.08971238e-01
6.25669837e-01 1.19598949e+00 5.80182076e-01 4.45810348e-01
-1.22341537e+00 -7.39046514e-01 4.82796520e-01 -1.16961330e-01
-1.28347266e+00 -2.34457821e-01 5.45270205e-01 -2.43776947e-01
6.16424799e-01 5.16398773e-02 7.55265892e-01 1.24716687e+00
7.65147209e-02 1.14874446e+00 1.22719669e+00 2.00774055e-02
7.72019327e-01 2.86068954e-02 -2.19743505e-01 5.17058134e-01
3.07913631e-01 7.10402548e-01 -2.70468444e-01 -2.55827636e-01
7.15852320e-01 -1.97516203e-01 -1.11091211e-01 -5.93349457e-01
-1.13695216e+00 6.90051973e-01 2.61851758e-01 -1.44909054e-01
-2.32251748e-01 4.92664516e-01 4.47417259e-01 4.36909795e-01
3.03071052e-01 3.83435041e-01 -4.59923595e-01 -6.21699750e-01
-8.25450897e-01 7.05185473e-01 5.75073957e-01 1.26506603e+00
6.78205490e-01 2.28174612e-01 -1.00737847e-01 3.51599842e-01
1.37229994e-01 5.83925188e-01 3.96993965e-01 -1.12861955e+00
5.75656652e-01 2.69100994e-01 9.09243286e-01 -4.48401600e-01
-2.87005603e-01 -2.39761502e-01 -5.01586914e-01 1.60912767e-01
7.45826304e-01 -6.14922583e-01 -6.71483636e-01 1.85143745e+00
6.16938174e-01 5.29093623e-01 2.20818669e-01 9.70198929e-01
-3.26197706e-02 6.75850034e-01 -8.90135765e-02 -3.94940078e-01
4.03708905e-01 -9.81712997e-01 -6.44377112e-01 -7.17304051e-02
8.01600814e-01 -7.12380335e-02 1.36683416e+00 5.09474099e-01
-9.68910873e-01 -4.73835140e-01 -9.13072646e-01 6.74280822e-01
-7.75723979e-02 3.13435227e-01 5.88520110e-01 4.61445510e-01
-5.49340189e-01 1.00717509e+00 -1.30868554e+00 -1.17165856e-02
4.00395840e-01 2.36549303e-01 7.84503892e-02 5.49743995e-02
-1.11298680e+00 5.90950310e-01 7.32778668e-01 -1.06996521e-01
-1.45108092e+00 -8.16627383e-01 -7.40639627e-01 -2.02317819e-01
1.03036761e+00 -3.46366256e-01 1.55735803e+00 -7.58738458e-01
-1.92782187e+00 -9.36833248e-02 -4.32949662e-02 -8.67961645e-01
1.09964442e+00 -4.31567371e-01 -1.34871453e-01 -6.69580251e-02
-1.89641461e-01 4.69367504e-01 1.00122190e+00 -1.33853650e+00
-6.52225792e-01 -4.46630083e-02 1.37955353e-01 4.70972992e-02
-2.30257902e-02 -4.62120414e-01 -2.77994961e-01 -4.44703221e-01
-3.15506309e-01 -1.18398774e+00 -5.34764647e-01 -9.90216285e-02
-4.59905833e-01 -1.78715706e-01 6.90329015e-01 -3.79009068e-01
1.42646050e+00 -1.95957994e+00 -2.05885656e-02 2.63816118e-01
-1.45232707e-01 3.74940932e-01 -2.03020079e-03 4.73858505e-01
4.11226124e-01 1.46852314e-01 -3.68529320e-01 -2.81272799e-01
2.71035880e-01 5.26823759e-01 -8.93559217e-01 4.32216227e-01
-7.60269398e-03 9.62804496e-01 -1.38128459e+00 -1.62634283e-01
2.82233357e-01 -2.83545971e-01 -5.42474687e-01 1.55183494e-01
-9.98885512e-01 6.73949897e-01 -7.33546495e-01 4.20251578e-01
5.48139334e-01 -9.23570842e-02 3.86263072e-01 4.33739960e-01
-1.95186764e-01 1.78754315e-01 -1.33936763e+00 1.56959152e+00
-5.02355576e-01 2.79370755e-01 -1.66166648e-01 -7.28266358e-01
7.31680214e-01 4.02653217e-02 5.14622808e-01 -4.63738978e-01
7.92546123e-02 2.87096858e-01 -5.25917970e-02 -2.54625738e-01
5.21543205e-01 6.46433979e-02 5.27634285e-03 6.77465737e-01
-2.77956814e-01 -9.89691466e-02 2.26391807e-01 -3.09696328e-02
9.99417424e-01 7.60647714e-01 1.87264591e-01 -4.79059629e-02
1.83662176e-01 1.85494572e-01 7.50219226e-01 9.87698793e-01
-4.98069286e-01 2.18935266e-01 6.17745161e-01 -1.45259589e-01
-1.04194677e+00 -9.50129151e-01 9.88843963e-02 6.86444640e-01
2.78107822e-01 -4.92992222e-01 -8.98331404e-01 -9.98809397e-01
3.68226260e-01 9.86395597e-01 -5.03098607e-01 -1.06530026e-01
-4.80363071e-01 -6.80131316e-01 3.37392360e-01 6.45940185e-01
4.39115077e-01 -8.41121793e-01 -8.26026678e-01 5.39793313e-01
9.93864685e-02 -9.99579668e-01 -2.89324969e-01 -1.22745425e-01
-9.55943763e-01 -9.62486804e-01 -2.75080085e-01 3.39240991e-02
5.86928487e-01 3.14853817e-01 7.60950506e-01 -7.39506334e-02
3.61099273e-01 2.49628574e-01 -4.07228202e-01 -3.88780743e-01
-6.02779269e-01 -2.39551943e-02 4.10774529e-01 -2.33054176e-01
-3.79663371e-02 -4.27285731e-01 -3.51959527e-01 5.13725817e-01
-7.04859376e-01 -7.75424689e-02 2.39695713e-01 8.01207244e-01
9.14894998e-01 3.08028877e-01 8.75098467e-01 -6.29058957e-01
6.82692409e-01 -5.84476292e-01 -1.15327632e+00 2.07476437e-01
-6.85676336e-01 2.38164335e-01 1.01370001e+00 -7.56691039e-01
-6.77629352e-01 8.42944682e-02 2.76247077e-02 -1.04773796e+00
-4.94914083e-03 5.85575223e-01 -3.99483815e-02 2.04197481e-01
5.14300883e-01 4.39632773e-01 2.32639924e-01 -3.56623054e-01
5.03300130e-01 4.21537519e-01 3.15114558e-01 -8.95566106e-01
8.53229940e-01 3.93775672e-01 -1.57929048e-01 -4.65512156e-01
-8.22983205e-01 -4.22738418e-02 -3.26015055e-01 -3.94199967e-01
3.10306758e-01 -8.79362822e-01 -8.77768159e-01 3.61777514e-01
-6.68419778e-01 -1.15642595e+00 -5.38074076e-01 5.42142332e-01
-1.05285323e+00 1.75573826e-01 -3.18131506e-01 -1.24951530e+00
6.92474991e-02 -1.15681159e+00 7.99700499e-01 2.48247251e-01
-9.74329375e-03 -8.13790023e-01 2.08134666e-01 -1.58519730e-01
2.08146945e-01 5.53055406e-01 7.30048299e-01 -5.50485969e-01
-6.58662856e-01 -3.02804895e-02 3.09926033e-01 1.98274791e-01
-3.98092680e-02 2.07450837e-01 -7.05151498e-01 -6.05884671e-01
-2.26394832e-01 -7.15626359e-01 7.56003141e-01 4.33545619e-01
1.40961778e+00 -4.87351298e-01 -3.62423122e-01 3.61952335e-01
1.31300998e+00 2.94838727e-01 4.09217417e-01 3.94111186e-01
3.60839099e-01 3.12613130e-01 1.34675586e+00 8.79277587e-01
2.95922071e-01 6.29665613e-01 5.43777704e-01 5.27667701e-01
3.83039117e-01 -8.51905763e-01 8.12449753e-01 5.07213652e-01
2.65694737e-01 -2.95072258e-01 -7.44434893e-01 4.32062596e-01
-2.22899771e+00 -8.31332624e-01 1.22723237e-01 2.55927920e+00
9.03280854e-01 4.19277698e-01 5.40223241e-01 7.53552541e-02
2.98717082e-01 6.01171330e-02 -1.16533482e+00 -1.75141394e-01
1.20555274e-01 -1.37713015e-01 6.38118863e-01 4.53569621e-01
-1.05264521e+00 1.09678566e+00 6.09824562e+00 1.12418222e+00
-1.02819610e+00 -7.16412961e-02 5.32357752e-01 -3.59032184e-01
-2.45195314e-01 8.93681049e-02 -7.87926435e-01 7.02510536e-01
1.24292254e+00 -4.03824806e-01 8.63310754e-01 1.07153976e+00
6.30768180e-01 -3.38273078e-01 -1.07615459e+00 6.27375066e-01
-5.90816557e-01 -1.23129153e+00 -3.59491825e-01 1.21719643e-01
9.84763086e-01 8.24907273e-02 6.99272454e-02 9.30805624e-01
7.58706212e-01 -8.80414963e-01 9.66618478e-01 4.53882426e-01
6.05149806e-01 -1.14992762e+00 3.42784435e-01 9.16355252e-01
-9.00453150e-01 -3.07559788e-01 -5.32784522e-01 7.30958283e-02
1.25091255e-01 4.65138078e-01 -7.64605165e-01 7.15137541e-01
3.49149078e-01 8.05212736e-01 -2.82332689e-01 1.00178397e+00
-4.16234493e-01 1.02283621e+00 -6.72044456e-01 -6.64544329e-02
5.50358355e-01 -4.57516015e-01 8.27275991e-01 5.29518425e-01
3.58869851e-01 -5.98046370e-02 4.09012496e-01 1.16990364e+00
5.75222373e-02 -2.59297341e-01 -6.49243414e-01 -2.96013653e-01
8.08520913e-01 8.31739068e-01 -3.21745366e-01 -2.64776140e-01
3.93217802e-03 5.25165617e-01 4.04549539e-01 3.76882911e-01
-1.16282952e+00 -8.84595960e-02 6.75751925e-01 -2.02848941e-01
4.39764500e-01 -5.40167928e-01 -4.78186198e-02 -1.00554872e+00
3.09934979e-03 -1.09346616e+00 8.20372254e-02 -5.04521251e-01
-9.83169317e-01 2.42727622e-01 5.25684990e-02 -1.63344836e+00
-6.56154394e-01 -2.48997986e-01 -4.84643012e-01 4.57376450e-01
-1.31457496e+00 -6.31679952e-01 4.33621481e-02 2.45579019e-01
4.98164684e-01 4.61426228e-02 4.59161758e-01 -1.47092581e-01
-9.77530539e-01 5.65042853e-01 6.79642022e-01 -2.37295598e-01
4.50708598e-01 -1.34451795e+00 3.63402307e-01 7.38492548e-01
-9.74376276e-02 4.33266342e-01 7.35578954e-01 -8.25416803e-01
-1.65960193e+00 -1.58477151e+00 -5.37274256e-02 -5.48893869e-01
8.82344365e-01 -2.87985206e-01 -9.15930808e-01 7.51199186e-01
-2.76257396e-01 1.78548843e-01 1.32625610e-01 -2.18822479e-01
1.35553986e-01 -6.10751472e-02 -1.14818466e+00 8.59155059e-01
9.50189829e-01 -7.14754686e-02 -2.15323925e-01 3.78236115e-01
9.69480813e-01 -7.80085325e-01 -9.31491792e-01 3.87475818e-01
4.37101394e-01 -8.30793083e-01 7.67799318e-01 -9.43049967e-01
2.29232416e-01 -4.65876073e-01 -8.55295658e-02 -1.60650492e+00
3.84673804e-01 -1.13277805e+00 -7.91026056e-01 1.09610450e+00
3.08167160e-01 -5.96005678e-01 6.55842662e-01 5.38143635e-01
-1.24154920e-02 -1.10369587e+00 -9.95742679e-01 -1.50278854e+00
3.12004507e-01 -7.53022075e-01 1.06657672e+00 5.31154871e-01
4.33268622e-02 -2.58893847e-01 -6.14770055e-01 1.65989175e-01
7.24116862e-01 3.29947323e-01 1.19205725e+00 -6.66853547e-01
-5.43028831e-01 -1.84055552e-01 2.36348003e-01 -1.31166732e+00
3.89765084e-01 -4.80474263e-01 2.94392198e-01 -1.19660175e+00
-2.33744070e-01 -9.12057161e-01 -2.46089801e-01 3.48306149e-01
-3.12962174e-01 -5.57432055e-01 2.07453430e-01 2.58826077e-01
-7.89068222e-01 1.20461452e+00 1.42512882e+00 1.18018694e-01
-6.31764948e-01 2.83749431e-01 -1.27374202e-01 6.75019920e-01
1.08566642e+00 -5.47305167e-01 -7.54413903e-01 -6.64249361e-02
-1.26693053e-02 3.43910336e-01 1.52109727e-01 -9.64641690e-01
-1.57146111e-01 -7.05426872e-01 -5.38518280e-02 -8.16513658e-01
2.97070928e-02 -6.30476117e-01 1.01465926e-01 6.25348985e-01
-4.27024543e-01 -1.40437260e-01 4.26506251e-01 1.01401198e+00
5.60801625e-02 -6.07122108e-02 6.42542601e-01 1.48460880e-01
-5.24949610e-01 5.23798764e-01 -4.51998353e-01 3.09166074e-01
1.22233915e+00 2.05300391e-01 -2.63912916e-01 -4.33598459e-01
-4.78220999e-01 8.25441360e-01 5.66353858e-01 2.03485891e-01
4.85293508e-01 -1.28032744e+00 -2.45507702e-01 4.32678908e-02
-3.17746960e-02 1.07054338e-01 3.31035629e-02 9.26784813e-01
-9.82591435e-02 4.92517978e-01 2.28466287e-01 -4.95090276e-01
-5.62372804e-01 9.11372781e-01 4.81245548e-01 -3.18746924e-01
-7.61990428e-01 3.19925010e-01 -2.18099982e-01 -5.93676209e-01
2.89569348e-01 -8.25908124e-01 1.84258759e-01 -3.03279698e-01
5.45311213e-01 5.18047571e-01 -2.71180868e-01 -8.82731006e-02
-1.33196265e-01 1.97955668e-01 1.74493089e-01 -4.36348617e-01
1.08093822e+00 -1.78923588e-02 4.63996440e-01 5.74561715e-01
6.69858813e-01 -9.40901488e-02 -2.16750813e+00 -1.03265762e-01
5.56294955e-02 -5.43957710e-01 -7.17506707e-02 -8.08474958e-01
-8.13498557e-01 6.53120995e-01 3.86115372e-01 1.51440844e-01
8.68963361e-01 -5.38291335e-01 1.02937996e+00 5.49047887e-01
1.01671731e+00 -1.39875114e+00 -7.72483200e-02 4.76570100e-01
8.44253123e-01 -1.21384704e+00 -4.87504937e-02 4.73627783e-02
-8.92051101e-01 7.68873751e-01 7.60419428e-01 -3.25314939e-01
3.64447743e-01 1.83889896e-01 -3.06289792e-01 4.14283961e-01
-1.11188769e+00 -2.97314584e-01 -1.09834671e-01 4.62284625e-01
-4.93026227e-01 2.30562761e-01 -2.50009924e-01 8.08466554e-01
-9.86046419e-02 2.79157370e-01 5.85228503e-01 1.02360559e+00
-5.89329779e-01 -1.04185843e+00 -3.29056472e-01 2.71711051e-01
-9.34035033e-02 4.49995160e-01 -1.28925636e-01 1.01505673e+00
-1.53668091e-01 9.28326249e-01 -1.68466061e-01 -5.59986711e-01
2.47278765e-01 -1.40243694e-01 5.01717567e-01 -8.58430564e-02
-4.08612937e-01 1.09896496e-01 2.01028675e-01 -9.08553481e-01
-1.86864119e-02 -7.79925108e-01 -1.60150719e+00 -4.04062957e-01
-1.88444465e-01 1.05111286e-01 4.74973738e-01 1.04301751e+00
4.81344014e-01 3.52177918e-01 8.50073159e-01 -7.12298036e-01
-1.36698902e+00 -8.54399264e-01 -5.20579457e-01 3.26061994e-03
4.44756061e-01 -1.00462222e+00 -3.79678607e-01 -3.28742683e-01]
|
[4.136801719665527, 2.15238618850708]
|
8623a753-a2e4-456e-9eec-324e48b2233b
|
boundary-aware-transformers-for-skin-lesion
|
2110.03864
| null |
https://arxiv.org/abs/2110.03864v1
|
https://arxiv.org/pdf/2110.03864v1.pdf
|
Boundary-aware Transformers for Skin Lesion Segmentation
|
Skin lesion segmentation from dermoscopy images is of great importance for improving the quantitative analysis of skin cancer. However, the automatic segmentation of melanoma is a very challenging task owing to the large variation of melanoma and ambiguous boundaries of lesion areas. While convolutional neutral networks (CNNs) have achieved remarkable progress in this task, most of existing solutions are still incapable of effectively capturing global dependencies to counteract the inductive bias caused by limited receptive fields. Recently, transformers have been proposed as a promising tool for global context modeling by employing a powerful global attention mechanism, but one of their main shortcomings when applied to segmentation tasks is that they cannot effectively extract sufficient local details to tackle ambiguous boundaries. We propose a novel boundary-aware transformer (BAT) to comprehensively address the challenges of automatic skin lesion segmentation. Specifically, we integrate a new boundary-wise attention gate (BAG) into transformers to enable the whole network to not only effectively model global long-range dependencies via transformers but also, simultaneously, capture more local details by making full use of boundary-wise prior knowledge. Particularly, the auxiliary supervision of BAG is capable of assisting transformers to learn position embedding as it provides much spatial information. We conducted extensive experiments to evaluate the proposed BAT and experiments corroborate its effectiveness, consistently outperforming state-of-the-art methods in two famous datasets.
|
['Jing Qin', 'Lei Zhu', 'Qichao Zhou', 'Liansheng Wang', 'Lan Wei', 'Jiacheng Wang']
|
2021-10-08
| null | null | null | null |
['skin-lesion-segmentation']
|
['medical']
|
[ 5.10165513e-01 -1.49247870e-02 -4.28149194e-01 -2.88878381e-01
-7.07312942e-01 -3.53475690e-01 6.06598914e-01 1.93960845e-01
-4.76306558e-01 5.24438918e-01 1.85947478e-01 -2.30241537e-01
-1.35228068e-01 -7.62278855e-01 -4.59347069e-01 -1.05495322e+00
4.26208824e-01 -6.95637837e-02 4.33448166e-01 -3.35432768e-01
1.46886602e-01 4.82853502e-01 -9.19655621e-01 1.33395001e-01
1.45105684e+00 9.82121706e-01 2.82531410e-01 2.22626016e-01
-4.08072948e-01 5.32101572e-01 -2.44441032e-01 -4.74167436e-01
-5.34086488e-02 -4.22131777e-01 -6.68883264e-01 1.53348846e-02
3.50989997e-01 -8.93834755e-02 -2.02862769e-01 1.21527910e+00
4.49382514e-01 -2.06787303e-01 5.21669269e-01 -5.59188664e-01
-4.83934134e-01 1.65051684e-01 -9.37141776e-01 3.10306638e-01
-1.79718226e-01 2.09462285e-01 9.62839007e-01 -3.60660851e-01
4.57139909e-01 7.41259038e-01 6.53546870e-01 7.06308007e-01
-9.60773468e-01 -4.37919647e-01 5.23741126e-01 1.43915102e-01
-1.24998248e+00 -8.43772516e-02 1.06469703e+00 -6.29134104e-02
5.17783463e-01 6.28280044e-01 7.33888626e-01 1.23667777e+00
3.17471474e-01 8.44667375e-01 1.20378685e+00 -2.59765446e-01
-3.13454196e-02 1.84290662e-01 6.03800006e-02 6.19662285e-01
-1.73436075e-01 -3.58326048e-01 -1.44944996e-01 1.41090557e-01
1.11328101e+00 2.02783316e-01 -4.97092664e-01 -1.57703698e-01
-1.03075600e+00 7.36099064e-01 1.29895735e+00 5.87485135e-01
-4.03488815e-01 7.34180287e-02 3.55107814e-01 -2.63830900e-01
5.98242521e-01 2.54159093e-01 2.68551242e-02 1.87975153e-01
-5.88370383e-01 -8.57736170e-02 1.50191039e-01 3.51273149e-01
3.42005342e-01 -3.18173319e-01 -4.45049405e-01 7.79110491e-01
-2.57933140e-02 3.89331244e-02 5.98880291e-01 -3.11771333e-01
3.10061812e-01 1.13475788e+00 -1.91232249e-01 -9.12815154e-01
-4.46981490e-01 -9.39473391e-01 -1.27152479e+00 -9.80823934e-02
4.60128844e-01 2.13663161e-01 -1.15587699e+00 1.77290678e+00
4.19907808e-01 1.83492512e-01 -3.71358246e-01 9.76101458e-01
7.07555056e-01 2.41130725e-01 5.17978132e-01 -1.03406217e-02
1.39415693e+00 -1.15865803e+00 -7.27670670e-01 -4.29865450e-01
5.53354383e-01 -4.57381785e-01 1.12553680e+00 6.02402687e-02
-7.42670953e-01 -3.02794099e-01 -8.40482354e-01 -1.71325684e-01
-3.03243101e-01 1.79581359e-01 9.32524145e-01 4.90677446e-01
-9.49800789e-01 1.77851379e-01 -9.69612181e-01 -3.94730419e-01
9.05617952e-01 3.60248774e-01 -3.39488149e-01 -1.70929998e-01
-1.07137346e+00 6.59470916e-01 7.68366084e-02 5.44317544e-01
-5.29623151e-01 -6.45258009e-01 -7.48077154e-01 -5.96375316e-02
3.96522373e-01 -6.08862936e-01 7.24643409e-01 -1.13441563e+00
-1.34004354e+00 7.26398170e-01 -3.78015280e-01 -1.98094562e-01
5.38965344e-01 2.98017031e-03 -2.26897165e-01 4.49222714e-01
1.24895070e-02 7.90985644e-01 7.74095893e-01 -1.07217872e+00
-4.15204853e-01 -6.37130320e-01 1.30417719e-01 4.07236159e-01
-9.06114757e-01 -3.24151546e-01 -6.48128688e-01 -7.80867338e-01
1.04786970e-01 -7.91513205e-01 -6.17092133e-01 3.82836759e-01
-5.96828818e-01 -1.51521847e-01 6.81827247e-01 -8.18663180e-01
1.41476667e+00 -2.02689910e+00 2.35942289e-01 1.28564000e-01
3.14774156e-01 5.33557892e-01 -1.27824649e-01 1.11061208e-01
1.05451047e-01 2.64120162e-01 -5.20922303e-01 -1.61997631e-01
-3.89184028e-01 1.01168975e-01 -1.19771935e-01 4.16215122e-01
5.71618915e-01 1.34617114e+00 -8.29223156e-01 -6.44784093e-01
4.06365901e-01 7.72127271e-01 -3.49835634e-01 -3.63971181e-02
-3.27687681e-01 7.35940218e-01 -7.01121926e-01 7.43288159e-01
7.50702262e-01 -4.64898378e-01 1.90982550e-01 -4.23753381e-01
9.58367363e-02 1.45338438e-02 -5.66094279e-01 1.78265429e+00
-5.22793472e-01 2.78364897e-01 3.28159720e-01 -9.79007781e-01
5.51176965e-01 1.73336968e-01 4.47374701e-01 -8.42022836e-01
1.27740964e-01 1.25570104e-01 8.28031823e-02 -5.32796621e-01
3.63298915e-02 -3.06262672e-01 2.15489507e-01 -1.15889594e-01
-3.33390355e-01 1.65424600e-01 -2.17394561e-01 1.36068344e-01
1.04921639e+00 4.22962848e-03 2.19651267e-01 -1.44279838e-01
8.42340171e-01 -2.25240096e-01 6.13067269e-01 2.78563023e-01
-3.53458494e-01 6.68650568e-01 5.49458444e-01 -2.30241954e-01
-5.28542995e-01 -9.56407309e-01 -2.38942564e-01 7.33427167e-01
5.02645671e-01 -1.91746995e-01 -7.78864443e-01 -1.11729181e+00
-2.36666903e-01 2.39221051e-01 -1.10292673e+00 -5.16886078e-02
-5.72204411e-01 -1.07650042e+00 4.66847569e-01 8.53780866e-01
7.47648835e-01 -8.90562356e-01 -3.82676959e-01 7.58897737e-02
-2.78624505e-01 -1.01984394e+00 -5.33303082e-01 8.00915137e-02
-7.13953435e-01 -9.98783410e-01 -1.01078665e+00 -7.44661033e-01
1.04222322e+00 3.19437921e-01 5.89089692e-01 2.87440091e-01
-6.46280766e-01 -1.28304392e-01 -9.51957852e-02 -2.12260023e-01
-7.05174776e-03 3.76527369e-01 -6.87884927e-01 2.67384529e-01
3.16101253e-01 -4.57326770e-01 -9.07062590e-01 3.70651454e-01
-1.24095857e+00 1.22960970e-01 9.85636294e-01 1.20815146e+00
6.87535644e-01 8.73131007e-02 6.83513105e-01 -1.09030354e+00
4.30082470e-01 -2.96953887e-01 -1.71057552e-01 2.90409505e-01
-1.35693297e-01 -3.00002158e-01 7.41481066e-01 -2.06503928e-01
-1.22629631e+00 -1.23045512e-01 -5.66601396e-01 -1.70822024e-01
-1.61969498e-01 4.54709917e-01 -1.95674986e-01 -4.81187612e-01
3.53753805e-01 2.84048319e-01 2.44514063e-01 -2.90202528e-01
2.04782941e-06 5.12134194e-01 3.36359143e-01 -2.20298737e-01
5.78434765e-01 7.95918405e-01 1.56256035e-01 -7.73458540e-01
-1.00455844e+00 -5.25770545e-01 -5.98771155e-01 1.43282870e-02
9.44707155e-01 -6.90710604e-01 -5.11581004e-01 7.68708587e-01
-7.37284720e-01 -2.06649408e-01 -1.03446394e-01 2.11609136e-02
-1.46824177e-02 5.83428025e-01 -7.29175150e-01 -5.73497355e-01
-3.87769789e-01 -1.22534275e+00 1.14656198e+00 5.53343415e-01
1.33501679e-01 -1.40717614e+00 -2.24629611e-01 5.43021500e-01
7.52757490e-01 4.75727767e-01 1.13784301e+00 -2.96494603e-01
-5.03322780e-01 -3.20569664e-01 -4.97042865e-01 3.69302303e-01
4.86062199e-01 -6.03502281e-02 -1.09841955e+00 -2.85737216e-01
-1.07111752e-01 -2.05261305e-01 1.24156153e+00 4.64171171e-01
1.53019643e+00 7.96076085e-04 -6.96353853e-01 7.92292356e-01
1.50380301e+00 -2.27318838e-01 6.68750644e-01 1.44594327e-01
9.76742864e-01 7.15922475e-01 5.57413936e-01 -9.94765386e-02
3.25600237e-01 5.86086929e-01 7.66460657e-01 -7.09535122e-01
-3.06100488e-01 -3.09377044e-01 -1.74462214e-01 5.45874834e-01
-1.45114854e-01 -1.29009411e-01 -6.67800784e-01 7.67606735e-01
-1.71607471e+00 -4.45411831e-01 1.08637353e-02 2.14853525e+00
8.95760596e-01 2.88583398e-01 -1.13186620e-01 -3.11577730e-02
6.88983858e-01 4.03674096e-01 -6.94510877e-01 -1.62934549e-02
-1.06025651e-01 3.31075549e-01 3.93906444e-01 3.69835794e-01
-1.30198920e+00 9.13813531e-01 5.10238409e+00 1.23397839e+00
-1.43189764e+00 -2.16850061e-02 9.04278696e-01 1.26608491e-01
-5.48465073e-01 -2.07762793e-01 -6.66108191e-01 5.50729394e-01
2.03892648e-01 4.78191167e-01 -1.93832237e-02 3.91512096e-01
-9.15263221e-03 -1.15710691e-01 -6.50354564e-01 6.63270772e-01
-4.38130200e-02 -1.21273232e+00 2.74130136e-01 2.07972988e-01
7.64866710e-01 -2.95371175e-01 4.52177316e-01 -8.31858590e-02
-3.45511734e-01 -1.29612720e+00 -1.03455051e-04 4.87134129e-01
9.81227934e-01 -6.88194513e-01 1.00918806e+00 3.56488824e-01
-1.08682311e+00 -1.78867429e-01 -3.56355697e-01 9.21693966e-02
-8.49404335e-02 6.66513801e-01 -6.78893268e-01 6.84537888e-01
3.79092127e-01 7.72807658e-01 -8.63264561e-01 1.01205587e+00
-3.81913871e-01 6.46750271e-01 -2.56468683e-01 -4.24724706e-02
6.40249372e-01 -1.03281818e-01 4.34274614e-01 9.58512425e-01
-1.43797966e-02 -2.74653733e-02 -2.33833399e-02 8.69809866e-01
5.26395626e-03 2.61078775e-01 -3.89687926e-01 -2.16088165e-02
1.58896774e-01 1.62623000e+00 -8.26932073e-01 1.65681466e-01
-3.31202418e-01 9.75091219e-01 3.69174004e-01 4.11608964e-01
-7.92597950e-01 -4.59009767e-01 5.54430187e-01 3.41172010e-01
2.63160467e-01 5.89966476e-02 -3.78498107e-01 -1.04794014e+00
2.58015245e-01 -6.74713790e-01 3.18849862e-01 -2.74230212e-01
-1.35671246e+00 5.55496395e-01 -5.41352868e-01 -1.02453637e+00
2.31276587e-01 -4.63107020e-01 -9.33417559e-01 9.27771688e-01
-2.13067508e+00 -1.75382292e+00 -7.43233919e-01 6.41631782e-01
3.64500076e-01 3.52201313e-01 8.66483867e-01 1.87660113e-01
-8.29116225e-01 8.33346426e-01 -1.60674423e-01 1.23385109e-01
8.71745944e-01 -1.32874966e+00 7.12619051e-02 6.71673715e-01
-2.33257651e-01 7.58911490e-01 1.00386947e-01 -5.29021144e-01
-1.15879190e+00 -1.20309711e+00 3.50936711e-01 -3.70711118e-01
5.77089667e-01 -3.52999955e-01 -1.04507637e+00 4.06811506e-01
1.71428502e-01 3.34000230e-01 6.59616709e-01 9.75391865e-02
-1.54255584e-01 -3.85316402e-01 -1.07707560e+00 7.25236177e-01
9.23264861e-01 -3.90810281e-01 -1.70645386e-01 2.22135633e-01
4.60942477e-01 -4.83115733e-01 -7.76910245e-01 7.42775381e-01
2.76621997e-01 -1.03460145e+00 9.14784312e-01 -1.55646026e-01
6.05467677e-01 -7.01163039e-02 3.63252372e-01 -1.18843114e+00
-2.77238369e-01 -3.16410631e-01 2.13906765e-01 1.27034330e+00
8.00599083e-02 -9.08695579e-01 9.83811736e-01 2.32852429e-01
-1.76196501e-01 -1.45885432e+00 -9.75483716e-01 -2.02034712e-01
1.72328085e-01 3.06364782e-02 3.32473457e-01 8.76968205e-01
-1.81360379e-01 1.81882665e-01 -6.58026636e-02 1.71407009e-03
5.59988499e-01 7.38443360e-02 3.20081204e-01 -8.12149882e-01
-1.02658920e-01 -6.52687252e-01 -5.35614610e-01 -1.13386536e+00
-5.32901846e-02 -8.50273192e-01 -1.86431006e-01 -1.70906270e+00
3.61215144e-01 -6.26380682e-01 -6.87507331e-01 5.06779313e-01
-6.07308626e-01 5.81634521e-01 -1.45576149e-01 5.18200174e-02
-5.03468513e-01 5.82964122e-01 1.84023511e+00 -3.16377610e-01
-4.29507382e-02 9.03385580e-02 -1.12550914e+00 7.84690320e-01
5.89869022e-01 3.48785520e-02 -4.02381361e-01 -4.18938100e-01
5.14665432e-02 -1.81908384e-01 6.85114264e-01 -8.45456958e-01
3.75533342e-01 -9.38215256e-02 5.46340764e-01 -4.50913340e-01
2.84276843e-01 -6.73648775e-01 -3.65239024e-01 3.52435231e-01
-2.92123944e-01 -5.11317372e-01 2.84945935e-01 8.17153335e-01
-5.09035945e-01 1.99509393e-02 8.05372357e-01 -2.30552614e-01
-7.10324645e-01 4.61846858e-01 -2.62979139e-02 -4.82079573e-02
1.20232141e+00 -1.81709543e-01 -3.88996571e-01 3.20774925e-05
-6.17148161e-01 3.05715114e-01 6.57840192e-01 2.18358576e-01
4.68506843e-01 -1.05701959e+00 -5.60047388e-01 3.15639317e-01
1.61353022e-01 3.59075218e-01 7.57865965e-01 1.25727546e+00
-5.41329622e-01 5.69849968e-01 -1.41561851e-01 -7.49825358e-01
-1.02461410e+00 3.10575575e-01 4.94407445e-01 -6.61550224e-01
-6.18470848e-01 1.16603255e+00 8.07210863e-01 -2.82877356e-01
2.11273730e-01 -3.46937627e-01 -3.41651559e-01 -2.24913850e-01
4.28284913e-01 -5.88495284e-02 1.44976288e-01 -3.30477864e-01
-3.88949096e-01 9.33679461e-01 -4.68976885e-01 4.91998166e-01
1.20287037e+00 -2.51642019e-02 -2.40368217e-01 -2.65165027e-02
9.59121883e-01 1.36765316e-01 -1.39051199e+00 -2.80499756e-01
-4.51140523e-01 -4.78607476e-01 2.43530124e-01 -1.04146016e+00
-1.24526608e+00 1.40052021e+00 4.99598175e-01 2.12512594e-02
1.48955715e+00 -2.06889048e-01 1.00469756e+00 -1.11877963e-01
1.28860801e-01 -8.55160534e-01 2.20174506e-01 -1.01529518e-02
6.22750998e-01 -1.31373906e+00 -1.12147421e-01 -8.21723938e-01
-7.23454237e-01 9.28082108e-01 8.48821640e-01 -7.20608383e-02
4.18242455e-01 1.78122982e-01 1.72230765e-01 -2.09472731e-01
-3.80071610e-01 -2.12799296e-01 3.86180550e-01 5.35543561e-01
4.17943239e-01 -7.80145898e-02 -3.68354797e-01 6.10792816e-01
5.05955040e-01 -1.42291933e-01 9.92498323e-02 6.96797192e-01
-9.23266634e-02 -1.15476203e+00 3.08338646e-02 5.83664954e-01
-6.61161184e-01 -9.62089151e-02 -4.52326268e-01 9.09069061e-01
1.03516564e-01 4.69098002e-01 -1.30541131e-01 -1.90337390e-01
3.97081338e-02 -2.88252026e-01 5.31260788e-01 -5.60726106e-01
-6.95509553e-01 2.76809692e-01 -2.90970474e-01 -4.38641220e-01
-3.21334660e-01 -4.03199703e-01 -1.15738678e+00 1.89449415e-01
-4.30663764e-01 -1.35624856e-01 5.66962004e-01 8.16242397e-01
1.33273438e-01 8.59005809e-01 6.77031696e-01 -4.23758358e-01
-6.05162084e-01 -8.92234862e-01 -4.96426105e-01 3.77856165e-01
3.35939556e-01 -6.11742914e-01 -2.00676441e-01 -3.92764241e-01]
|
[15.608097076416016, -2.913339138031006]
|
f7b5331b-3532-4080-b7fa-a60dcb4996d9
|
best-answer-prediction-in-q-a-sites-using
|
2212.08475
| null |
https://arxiv.org/abs/2212.08475v1
|
https://arxiv.org/pdf/2212.08475v1.pdf
|
Best-Answer Prediction in Q&A Sites Using User Information
|
Community Question Answering (CQA) sites have spread and multiplied significantly in recent years. Sites like Reddit, Quora, and Stack Exchange are becoming popular amongst people interested in finding answers to diverse questions. One practical way of finding such answers is automatically predicting the best candidate given existing answers and comments. Many studies were conducted on answer prediction in CQA but with limited focus on using the background information of the questionnaires. We address this limitation using a novel method for predicting the best answers using the questioner's background information and other features, such as the textual content or the relationships with other participants. Our answer classification model was trained using the Stack Exchange dataset and validated using the Area Under the Curve (AUC) metric. The experimental results show that the proposed method complements previous methods by pointing out the importance of the relationships between users, particularly throughout the level of involvement in different communities on Stack Exchange. Furthermore, we point out that there is little overlap between user-relation information and the information represented by the shallow text features and the meta-features, such as time differences.
|
['Takayuki Ito', 'Kai Yoshino', 'Ahmed Moustafa', 'Rafik Hadfi']
|
2022-12-15
| null | null | null | null |
['community-question-answering', 'community-question-answering']
|
['miscellaneous', 'natural-language-processing']
|
[-4.65368390e-01 -1.25547096e-01 1.27261337e-02 -3.13822508e-01
-6.87248826e-01 -6.57846749e-01 4.89039540e-01 9.05267715e-01
-4.19866174e-01 4.28155899e-01 5.28176010e-01 -1.78539380e-01
-5.62296152e-01 -8.44891131e-01 5.68618141e-02 -1.45131171e-01
-1.97612830e-02 1.24379382e-01 8.80641878e-01 -6.31881058e-01
1.07352412e+00 -1.33027822e-01 -1.70662832e+00 6.48218870e-01
1.22110701e+00 9.44441319e-01 1.64466545e-01 6.43141568e-01
-9.45463538e-01 1.39366949e+00 -8.11482847e-01 -6.45835400e-01
-2.94452198e-02 -6.55812740e-01 -1.18105304e+00 -2.98558742e-01
2.47786880e-01 2.28596182e-05 -1.57231867e-01 5.99352419e-01
3.36151570e-01 -3.85167003e-02 1.51059940e-01 -1.23783064e+00
-5.34761012e-01 3.29072684e-01 -3.43288243e-01 5.61686397e-01
9.33012724e-01 -1.91378593e-01 1.46528351e+00 -5.33455133e-01
6.43744290e-01 9.34686184e-01 7.73118496e-01 1.64990142e-01
-6.69677436e-01 -3.90610993e-01 -6.21458739e-02 6.26181066e-01
-8.84317994e-01 1.54334843e-01 7.19744027e-01 -7.27761567e-01
6.87001944e-01 6.21201098e-01 4.11578983e-01 2.57212281e-01
2.65544455e-04 7.23232567e-01 1.16661608e+00 -5.30206800e-01
2.90579855e-01 3.91161323e-01 7.36183822e-01 5.19832790e-01
-2.37353697e-01 -8.36690247e-01 -6.90447628e-01 -8.72650981e-01
4.05809507e-02 1.77836552e-01 -2.88090795e-01 9.15582702e-02
-9.15109038e-01 1.15236855e+00 3.05487484e-01 7.99053133e-01
-3.53627801e-01 -3.97595942e-01 4.97491151e-01 6.28246546e-01
4.92963612e-01 7.08880603e-01 -5.26643753e-01 -3.95380974e-01
-7.68886507e-01 3.98654044e-01 1.38013148e+00 4.63397145e-01
1.01448429e+00 -9.24027503e-01 -4.04033422e-01 8.37087810e-01
3.85354280e-01 -1.02545619e-01 6.73396349e-01 -1.25045371e+00
5.33282995e-01 1.37426960e+00 2.75676608e-01 -1.43571603e+00
-1.96982175e-01 -2.18623847e-01 -2.61194669e-02 -5.13672769e-01
6.71573639e-01 -2.24687994e-01 1.07961856e-01 1.29062307e+00
4.09522682e-01 -2.00123832e-01 -5.11799276e-01 6.00499213e-01
1.04867077e+00 4.24644589e-01 -9.24307182e-02 -3.26532751e-01
1.41158998e+00 -8.49096477e-01 -1.01009011e+00 6.28151596e-02
7.75370359e-01 -1.05377579e+00 9.16457057e-01 -1.41141534e-01
-7.33335316e-01 -3.81875217e-01 -4.55429047e-01 9.19076279e-02
-4.10223395e-01 -1.56342164e-01 6.34237945e-01 7.80487955e-01
-1.04864895e+00 3.98857921e-01 -3.27851400e-02 -7.88920343e-01
-1.07649369e-02 1.61229610e-01 2.87860222e-02 1.23886555e-01
-1.23670161e+00 5.91159046e-01 -7.76465908e-02 -2.77459323e-01
8.91275331e-02 -5.73496997e-01 -1.37602195e-01 1.23762138e-01
2.28565469e-01 -5.78283370e-01 1.26129103e+00 -9.09008980e-01
-1.19855416e+00 6.34422481e-01 -2.56457746e-01 -1.76694989e-01
2.05656722e-01 -1.11154154e-01 -3.92969728e-01 2.08533362e-01
2.13405773e-01 9.04872492e-02 3.21840554e-01 -6.26299679e-01
-9.47730362e-01 -6.86029434e-01 4.94937390e-01 -5.99735044e-02
-7.63498187e-01 5.64358473e-01 -2.40601107e-01 -7.88370818e-02
2.06220105e-01 -5.28728068e-01 -1.34590253e-01 -1.22139134e-01
1.34773687e-01 -9.01590109e-01 9.41213429e-01 -1.06520987e+00
1.91450524e+00 -1.78465760e+00 -3.45538855e-01 2.32300743e-01
4.00412917e-01 2.53564894e-01 4.87096161e-02 1.15786970e+00
4.05270964e-01 2.89881110e-01 1.29406422e-01 2.21757054e-01
-4.82036322e-02 -9.89113748e-02 -1.04766473e-01 7.88931847e-02
-2.60485262e-01 6.03683531e-01 -1.03811884e+00 -5.79034209e-01
-5.20612419e-01 8.12039599e-02 -6.65491045e-01 3.34822774e-01
-4.35334355e-01 4.31567788e-01 -8.35371852e-01 2.91983455e-01
3.39843690e-01 -6.25581086e-01 2.85034806e-01 2.57809043e-01
-2.44714513e-01 5.78696072e-01 -9.37263727e-01 1.21726799e+00
-3.39073837e-01 7.13785529e-01 1.02682933e-01 -6.35225892e-01
1.20796216e+00 5.22320330e-01 5.05760550e-01 -9.13138866e-01
3.82258371e-02 5.38216770e-01 1.07047729e-01 -1.04294932e+00
4.90839392e-01 3.50256354e-01 1.99750766e-01 7.54212737e-01
-2.68754572e-01 1.57616615e-01 4.62261647e-01 4.21419203e-01
1.39943624e+00 -2.20850199e-01 3.43038529e-01 -2.33715802e-01
9.49821770e-01 -2.80887671e-02 2.61953354e-01 7.25720227e-01
-4.15030062e-01 3.92855346e-01 6.56716108e-01 -2.28601769e-01
-5.50913393e-01 -2.71031529e-01 1.08756892e-01 1.22581434e+00
3.29773799e-02 -7.55840540e-01 -8.19424450e-01 -7.14788616e-01
5.35054654e-02 3.43730092e-01 -4.68758225e-01 3.37860256e-01
-6.24134421e-01 -2.91264653e-01 1.20746799e-01 1.33619085e-01
6.50233805e-01 -8.78474832e-01 -4.31232184e-01 2.92955786e-01
-7.50838816e-01 -7.37880468e-01 -5.02987862e-01 -5.06406784e-01
-1.17786026e+00 -1.31432569e+00 -6.59484982e-01 -7.25554347e-01
1.31807402e-01 4.75596517e-01 1.27140951e+00 7.45624244e-01
7.06905052e-02 8.09762120e-01 -6.84197009e-01 -4.69158813e-02
2.55329069e-02 3.17950070e-01 -5.62719405e-01 2.67804742e-01
8.01814914e-01 -5.22198796e-01 -9.81176138e-01 6.03957772e-01
-6.87483490e-01 -5.19061983e-01 -4.66425940e-02 4.06506389e-01
-2.03508437e-01 -2.28393734e-01 1.04648280e+00 -1.09334004e+00
1.20466888e+00 -1.06546044e+00 4.31486480e-02 3.90183687e-01
-6.91245437e-01 -1.80934191e-01 -1.12916133e-03 -1.17228426e-01
-1.10570300e+00 -5.98240674e-01 -2.63346374e-01 6.20148122e-01
8.75793844e-02 7.48839796e-01 3.51147622e-01 -8.84272158e-02
7.96926379e-01 -2.34712571e-01 -1.85121316e-02 -7.27532089e-01
4.23721634e-02 1.06727171e+00 -2.34501302e-01 -3.62209141e-01
4.45508599e-01 3.12176555e-01 -5.76784968e-01 -8.48441422e-01
-9.10386562e-01 -1.33797693e+00 -3.45923573e-01 -5.90842128e-01
7.58197129e-01 -3.37623835e-01 -1.01641071e+00 3.13920796e-01
-1.29178178e+00 5.07538438e-01 3.21922377e-02 3.18491995e-01
-5.04440181e-02 6.08537674e-01 -7.96487153e-01 -1.11445296e+00
-3.22256684e-01 -6.76040590e-01 3.47763479e-01 6.04846656e-01
-6.05946779e-01 -1.16244638e+00 4.86818939e-01 1.25972486e+00
8.45800698e-01 -4.10820474e-04 1.19146228e+00 -1.05025828e+00
-7.24832714e-01 -3.42481256e-01 -1.54616460e-01 -3.97808850e-02
1.29615352e-01 -9.81742963e-02 -7.86554992e-01 -3.04148141e-02
4.32517141e-01 -2.03125030e-01 6.00312114e-01 -8.27048495e-02
7.47973382e-01 -3.33902806e-01 -1.37829021e-01 -6.39554262e-01
1.18184400e+00 1.08159520e-01 5.51399171e-01 5.71405947e-01
3.10133904e-01 1.18926477e+00 4.14882243e-01 5.55132806e-01
7.99635351e-01 5.42512119e-01 3.92415524e-01 5.83814502e-01
1.76776767e-01 -1.48741543e-01 2.37720966e-01 1.34657419e+00
8.75112321e-03 -1.14903077e-01 -1.04826236e+00 6.98597133e-01
-2.09321070e+00 -1.06159556e+00 -8.53645146e-01 2.07877970e+00
7.02445626e-01 -8.85217488e-02 3.68385106e-01 3.16333890e-01
8.27170074e-01 1.87738508e-01 -1.58287019e-01 -3.44724864e-01
1.72564313e-01 1.52140424e-01 -8.07407573e-02 4.23867375e-01
-4.44473237e-01 7.72507582e-03 5.84059286e+00 6.14491224e-01
-5.22021651e-01 3.21744531e-01 4.93933916e-01 3.65150481e-01
-6.53063118e-01 2.85363823e-01 -5.19826412e-01 5.49555779e-01
1.01738524e+00 -1.79635182e-01 1.06625423e-01 5.15028656e-01
2.22329199e-01 -3.84356201e-01 -7.50837326e-01 4.41127837e-01
2.13736296e-01 -1.35167217e+00 -4.39318389e-01 -1.86335519e-01
6.53574169e-01 -1.07424177e-01 -3.03025007e-01 4.93402660e-01
-1.04245164e-01 -3.84503156e-01 2.98181355e-01 6.73871458e-01
-3.34937453e-01 -2.92841494e-01 8.68776619e-01 7.61851490e-01
-9.56289709e-01 -5.34087002e-01 1.53636765e-02 -5.25243580e-01
-9.40117463e-02 5.24980903e-01 -6.90557241e-01 4.01522636e-01
9.71408427e-01 3.70863318e-01 -9.54667211e-01 1.45770335e+00
5.29377721e-02 9.70669925e-01 -5.55664897e-02 -6.39287174e-01
1.50607020e-01 -2.88991481e-01 1.90452293e-01 8.88313472e-01
3.31217617e-01 -3.99578921e-03 -6.62136897e-02 5.62800109e-01
6.55855387e-02 8.21870565e-01 -1.83453590e-01 -1.48261845e-01
4.37311083e-01 1.12429059e+00 -6.78451180e-01 1.74781382e-02
-1.00839853e+00 4.60558444e-01 2.37300500e-01 2.47394890e-01
-2.41477549e-01 -5.37575662e-01 1.07846558e-01 6.75188541e-01
3.34916890e-01 -2.75375992e-02 -2.29739249e-01 -8.67642164e-01
3.99594098e-01 -8.95058513e-01 7.46091664e-01 -5.00541449e-01
-1.60835552e+00 3.71247321e-01 -3.72464120e-01 -9.33879912e-01
-1.73288852e-01 -2.47974947e-01 -9.32017386e-01 9.81491327e-01
-1.42413354e+00 -7.80300736e-01 -4.51429456e-01 3.78938019e-01
4.59344745e-01 -3.57727706e-02 6.32738054e-01 6.34996712e-01
-1.71297446e-01 2.25962684e-01 1.07598707e-01 1.07393026e-01
6.78247213e-01 -1.16104293e+00 -1.31959498e-01 2.44172901e-01
1.95786413e-02 9.24534976e-01 5.87399781e-01 -5.34765065e-01
-1.00778949e+00 -2.35636845e-01 1.76224244e+00 -7.16412187e-01
9.15081978e-01 1.66552961e-02 -1.13071156e+00 9.59660485e-02
6.21075332e-01 -4.37583953e-01 1.12552965e+00 5.71074128e-01
-3.68073642e-01 -1.05817981e-01 -1.08218849e+00 2.87995696e-01
4.77953017e-01 -9.22923446e-01 -6.38904154e-01 3.22423726e-01
6.22841239e-01 2.02103183e-01 -9.57836509e-01 6.70956820e-02
3.74044150e-01 -1.35644579e+00 4.48956817e-01 -5.54427385e-01
4.09711838e-01 -2.67549813e-01 -9.85001922e-02 -6.57101274e-01
-1.11495771e-01 -5.75693727e-01 -1.52266845e-01 1.61163974e+00
6.59878373e-01 -5.78222930e-01 8.72488201e-01 8.75466883e-01
5.09462118e-01 -7.24457502e-01 -6.07649148e-01 -2.29440078e-01
-8.67135078e-03 1.25628114e-01 5.02838433e-01 1.09248674e+00
3.74595851e-01 4.86807048e-01 5.06766662e-02 -2.67056763e-01
5.25495186e-02 3.13874990e-01 5.42693675e-01 -1.67412460e+00
-3.00887913e-01 -1.93020225e-01 -2.90415496e-01 -9.32176888e-01
-1.88814402e-01 -8.31353843e-01 -4.10190493e-01 -1.69043612e+00
5.72544336e-01 -3.34403843e-01 -2.41294175e-01 -2.74548829e-01
-5.72068989e-01 -1.68101713e-01 1.42052561e-01 6.57778919e-01
-1.02413404e+00 2.59964615e-01 1.31188059e+00 1.73554733e-01
-1.47976041e-01 4.63457972e-01 -7.06158400e-01 6.22972608e-01
7.62395322e-01 -5.90735078e-01 -3.73299956e-01 -1.81459635e-01
1.06552589e+00 4.03576702e-01 -2.32162941e-02 -7.05852568e-01
6.57076716e-01 -4.11559790e-02 -3.64692628e-01 -5.54901183e-01
-1.31103247e-01 -6.61796212e-01 -2.09164321e-01 3.28774810e-01
-6.25750422e-01 2.36156210e-01 -4.33981985e-01 7.75474489e-01
-6.13756895e-01 -8.74961793e-01 1.41860828e-01 -2.86495954e-01
-4.56743479e-01 3.67076439e-03 -5.01678765e-01 4.82382357e-01
7.43765414e-01 -3.08325976e-01 -5.00896871e-01 -8.92196119e-01
-4.15938556e-01 4.51435447e-01 1.18324623e-01 6.72507524e-01
2.27224499e-01 -1.07091534e+00 -6.39416099e-01 -4.40022826e-01
2.95277596e-01 -7.55378664e-01 2.89737940e-01 8.53004336e-01
-4.91775870e-01 4.98272866e-01 6.64086640e-02 -5.05106568e-01
-1.32947481e+00 1.00695401e-01 2.62609988e-01 -4.34008241e-01
-2.96449102e-03 6.40929878e-01 -3.27060252e-01 -7.90087104e-01
1.88282728e-01 -3.23686488e-02 -9.08318639e-01 5.89329302e-01
6.97129965e-01 7.80110955e-01 1.26541376e-01 -2.88986772e-01
-1.43527478e-01 2.71139652e-01 -7.29091689e-02 3.40142730e-03
1.20682991e+00 -3.54487538e-01 -7.39721000e-01 5.57015479e-01
1.15194988e+00 3.34324211e-01 -5.77112377e-01 -6.26526117e-01
8.85175884e-01 -6.66990280e-01 -2.50099570e-01 -6.14648283e-01
-8.29173625e-01 7.49829948e-01 4.80850458e-01 1.00811553e+00
7.14119911e-01 9.83447731e-02 8.67455006e-01 3.98077190e-01
3.93222779e-01 -1.18262362e+00 5.89036763e-01 7.01381981e-01
6.57206535e-01 -1.29946434e+00 -3.28809112e-01 -3.63554090e-01
-3.36492062e-01 1.11452222e+00 6.82627201e-01 2.55834013e-01
9.61690605e-01 -3.86940062e-01 3.12014282e-01 -5.10740697e-01
-8.05219829e-01 -1.22380346e-01 1.76164553e-01 2.49773860e-01
9.79613483e-01 -1.91567466e-01 -1.08194780e+00 6.71502829e-01
-2.01158645e-03 -1.51123316e-03 4.98525798e-01 1.05918920e+00
-8.19447398e-01 -1.31693065e+00 -2.83239275e-01 6.87294841e-01
-9.12797511e-01 1.41487807e-01 -5.09464085e-01 2.98870802e-01
-3.40236095e-03 1.64176166e+00 -1.05265021e-01 -3.25787395e-01
1.20912910e-01 3.37705195e-01 -1.97620969e-02 -6.95368350e-01
-1.38043547e+00 -5.94498158e-01 2.99104989e-01 -3.74542952e-01
-7.54851937e-01 -7.74792790e-01 -9.40401018e-01 -3.69480491e-01
-7.01195240e-01 8.03499162e-01 6.52844012e-01 1.05872059e+00
4.43735182e-01 1.45136192e-01 7.80617833e-01 1.26151055e-01
-4.60036248e-01 -9.93381619e-01 -2.69087166e-01 5.20255864e-01
2.46402547e-02 -6.96964487e-02 -5.26768982e-01 -3.21734488e-01]
|
[11.456694602966309, 8.054780960083008]
|
0e311248-ad93-4ff7-9985-8c10b9eff50e
|
semeval-2020-task-6-definition-extraction
|
2008.13694
| null |
https://arxiv.org/abs/2008.13694v1
|
https://arxiv.org/pdf/2008.13694v1.pdf
|
SemEval-2020 Task 6: Definition extraction from free text with the DEFT corpus
|
Research on definition extraction has been conducted for well over a decade, largely with significant constraints on the type of definitions considered. In this work, we present DeftEval, a SemEval shared task in which participants must extract definitions from free text using a term-definition pair corpus that reflects the complex reality of definitions in natural language. Definitions and glosses in free text often appear without explicit indicators, across sentences boundaries, or in an otherwise complex linguistic manner. DeftEval involved 3 distinct subtasks: 1)Sentence classification, 2) sequence labeling, and 3) relation extraction.
|
['Nicholas A. Miller', 'Carl Dockhorn', 'Franck Dernoncourt', 'Sasha Spala']
|
2020-08-31
| null |
https://aclanthology.org/2020.semeval-1.41
|
https://aclanthology.org/2020.semeval-1.41.pdf
|
semeval-2020
|
['definition-extraction']
|
['natural-language-processing']
|
[ 8.37972045e-01 1.60698116e-01 -4.26450849e-01 -6.32863224e-01
-7.97090232e-01 -1.10827374e+00 9.11492884e-01 5.72118044e-01
-6.53013229e-01 1.28393114e+00 3.46195132e-01 -6.98042929e-01
-1.96315855e-01 -6.26078486e-01 -7.37745985e-02 -1.05852865e-01
1.97450295e-01 2.98104733e-01 -1.00577455e-02 -3.73337746e-01
3.18252563e-01 1.40930042e-01 -1.12604177e+00 3.63611609e-01
6.17343426e-01 6.00499272e-01 -1.35263230e-03 7.23196387e-01
-6.20277643e-01 1.04438424e+00 -1.13345945e+00 -6.94816530e-01
-1.96216166e-01 -4.80911374e-01 -1.25285053e+00 -2.23027281e-02
5.23671210e-01 3.17766696e-01 2.50132322e-01 9.99492109e-01
1.86409503e-01 2.41281182e-01 4.78533953e-01 -8.76268506e-01
-4.08824801e-01 1.11008334e+00 -7.10287094e-02 4.51650798e-01
7.84704924e-01 7.83072785e-02 1.33031809e+00 -7.38909423e-01
9.47149217e-01 9.90235031e-01 6.98264122e-01 4.38894004e-01
-1.47315300e+00 -4.69714284e-01 1.65850312e-01 -2.31386602e-01
-1.38227940e+00 -4.63246346e-01 3.31430793e-01 -6.43445313e-01
1.70694017e+00 6.95926726e-01 4.68097448e-01 1.03422785e+00
2.39276409e-01 1.85942382e-01 1.16940176e+00 -9.69733894e-01
-2.58338097e-02 1.24036767e-01 1.05632770e+00 3.82327557e-01
4.47978437e-01 -2.56376594e-01 -4.66700464e-01 -2.56008565e-01
3.35961163e-01 -5.83611667e-01 -4.07693505e-01 3.66536051e-01
-1.11655891e+00 6.65854454e-01 -4.66632009e-01 7.80162454e-01
-1.49472699e-01 -8.40094930e-04 7.32830346e-01 5.07572830e-01
5.73923290e-01 9.11539555e-01 -9.78287518e-01 -2.95708090e-01
-7.59907067e-01 4.64375794e-01 1.23629129e+00 1.26008737e+00
6.57490909e-01 -5.44743657e-01 -6.94269985e-02 6.88670039e-01
-1.41608855e-02 1.70794725e-01 4.98552918e-01 -5.29568851e-01
6.82191730e-01 5.97477734e-01 2.23515704e-01 -8.22848558e-01
-4.66293156e-01 -2.02058941e-01 -1.84983641e-01 -3.93846899e-01
4.66753542e-01 -5.09406090e-01 -8.15418780e-01 1.75702059e+00
1.53852344e-01 -3.86586159e-01 -8.06650743e-02 5.33164203e-01
1.17592919e+00 1.06676184e-01 4.12507653e-01 -5.58261812e-01
1.65299129e+00 -5.96123695e-01 -1.16123664e+00 -6.34292245e-01
1.07955921e+00 -8.13587487e-01 8.10452998e-01 8.76755938e-02
-9.51023400e-01 -1.25862256e-01 -1.24573064e+00 -3.38545352e-01
-7.20943034e-01 -3.76130879e-01 7.22866714e-01 7.33493209e-01
-6.76704347e-01 2.93248802e-01 -4.54770267e-01 -2.84300566e-01
-9.38929617e-02 2.09106833e-01 -3.39612931e-01 1.94383040e-01
-1.45025277e+00 1.37925196e+00 5.58533847e-01 -4.02646102e-02
-2.50629652e-02 -6.34332418e-01 -1.26433110e+00 -3.00216287e-01
8.86886120e-01 -7.78046012e-01 1.59946680e+00 -1.04264331e+00
-8.19757640e-01 1.28363502e+00 -3.10480773e-01 -3.59678119e-01
-1.86471567e-01 -4.25578654e-01 -6.02078617e-01 -4.82310951e-01
5.21826446e-01 -2.51935780e-01 2.31661126e-01 -9.04607952e-01
-4.22682375e-01 -4.10669088e-01 4.43246782e-01 1.87678561e-01
3.35311383e-01 6.53238237e-01 -8.26903731e-02 -7.69830763e-01
-1.75486133e-01 -4.76911247e-01 -2.35922545e-01 -4.99649435e-01
-5.19261777e-01 -8.17400277e-01 4.71539386e-02 -5.92511117e-01
1.95962703e+00 -1.76501560e+00 1.30912811e-01 2.91061178e-02
5.23631811e-01 1.81589365e-01 2.58137286e-01 7.98194647e-01
-3.79092813e-01 7.37171233e-01 -4.51043278e-01 -1.60323098e-01
3.53066742e-01 3.21118534e-01 -1.99243322e-01 2.80983597e-02
1.15987048e-01 9.30654585e-01 -1.10408640e+00 -6.72150552e-01
5.94366416e-02 -5.83334528e-02 2.45290577e-01 -1.85317900e-02
-6.12250566e-01 -2.87404865e-01 -3.64829928e-01 3.13957334e-01
2.41469413e-01 -2.42102161e-01 7.02028751e-01 5.51660806e-02
-5.63816488e-01 1.24017394e+00 -9.95182455e-01 1.66681731e+00
-4.25855100e-01 8.02009106e-01 7.29239807e-02 -6.85135007e-01
4.82361019e-01 6.15759730e-01 1.82366177e-01 -2.07736492e-01
1.94779351e-01 1.95721537e-01 -5.29057253e-03 -6.44949496e-01
6.41537964e-01 -4.15557563e-01 -5.15716076e-01 5.97614467e-01
-1.20862145e-02 -4.61178750e-01 7.79815555e-01 3.32698435e-01
1.26767433e+00 6.86731637e-02 1.05169892e+00 -2.21567973e-01
5.27717233e-01 4.29448038e-01 7.63204217e-01 7.50178277e-01
-9.52193290e-02 2.39425719e-01 5.90012372e-01 -3.11813027e-01
-4.59114850e-01 -9.81327116e-01 -5.11268854e-01 7.88695991e-01
9.10056569e-03 -1.16596162e+00 -5.07572591e-01 -8.18273902e-01
-2.41813958e-01 1.03221834e+00 -4.44971949e-01 3.88703734e-01
-7.23870397e-01 -3.82928282e-01 8.30211878e-01 1.49962455e-01
2.70129055e-01 -1.04646111e+00 -7.04687357e-01 3.83057863e-01
-2.05289170e-01 -1.34635174e+00 -7.41200805e-01 3.80097538e-01
-1.77013263e-01 -1.28290486e+00 1.04629129e-01 -8.66173148e-01
3.32824767e-01 -2.19911858e-01 1.82979119e+00 2.92664349e-01
-2.19012767e-01 1.90968484e-01 -3.94678235e-01 -5.78171790e-01
-2.55872875e-01 3.74051519e-02 -2.37885877e-01 -6.75864041e-01
9.93487835e-01 -2.28545219e-01 1.86467439e-01 -2.95157671e-01
-9.15252924e-01 3.00689433e-02 1.67392910e-01 7.08140254e-01
2.07552955e-01 -1.58547074e-01 6.81236088e-01 -1.67164040e+00
1.07197118e+00 -3.82042319e-01 -3.08813393e-01 6.52091920e-01
-9.19869244e-01 6.83319420e-02 3.20038855e-01 -2.20088810e-01
-9.36371326e-01 -2.36447886e-01 -2.45222449e-01 5.06283343e-01
-4.16534781e-01 1.04452646e+00 -1.47740886e-01 4.30691987e-01
8.53768945e-01 -1.76850930e-01 -2.59375304e-01 -1.15164094e-01
5.09282947e-01 7.62162387e-01 3.03760380e-01 -4.76824582e-01
6.21662438e-01 -3.69967937e-01 -3.50265950e-01 -9.50854838e-01
-1.36924577e+00 -5.47287345e-01 -8.00415039e-01 6.58142716e-02
8.75649929e-01 -6.36920333e-01 -4.38788608e-02 -1.02561023e-02
-1.47083831e+00 -3.82556140e-01 -3.60564828e-01 3.63600165e-01
3.98567133e-03 4.12571549e-01 -5.55367112e-01 -8.04283321e-01
-5.43771803e-01 -4.74770546e-01 5.76672137e-01 8.90344307e-02
-1.14512491e+00 -1.42257023e+00 2.22225338e-01 3.43127519e-01
2.31276765e-01 5.67435205e-01 9.47023749e-01 -7.08283305e-01
-1.46226883e-01 3.74467522e-02 -1.03972636e-01 4.17400450e-02
5.56237221e-01 3.60305235e-02 -3.58120918e-01 2.71305054e-01
4.59063752e-03 -3.33633542e-01 3.77643049e-01 -1.21424846e-01
3.22472870e-01 -6.25913560e-01 -3.49999398e-01 7.21485391e-02
1.38002324e+00 2.47233704e-01 2.98493534e-01 3.70405078e-01
3.48729640e-01 6.68928027e-01 5.47520101e-01 1.26727939e-01
5.07474601e-01 5.47255278e-01 -3.68348211e-01 -2.62833480e-02
-8.00186545e-02 1.52599946e-01 8.69038329e-02 7.03971744e-01
4.14839745e-01 -3.09017330e-01 -1.18801439e+00 7.80580282e-01
-1.67567480e+00 -9.43140864e-01 -4.15866166e-01 1.72885990e+00
1.61887300e+00 4.39156085e-01 -1.33494124e-01 2.61872150e-02
6.26368225e-01 3.87387633e-01 -1.34015560e-01 -7.93703198e-01
-2.59751946e-01 7.86777854e-01 2.62059301e-01 7.45055914e-01
-1.01822245e+00 8.96852612e-01 6.75229216e+00 5.85934877e-01
-8.16335738e-01 -5.00869751e-02 -3.77115677e-03 8.85132402e-02
-6.45467103e-01 3.16651940e-01 -6.53734326e-01 4.20223802e-01
7.19508469e-01 -5.20200014e-01 1.44607434e-02 2.38532871e-01
-5.26140295e-02 -2.42153615e-01 -1.30412805e+00 8.89328241e-01
1.82890818e-01 -1.21925092e+00 -1.15684398e-01 -2.72939324e-01
2.73820907e-01 -1.30255267e-01 -5.76286674e-01 2.83801377e-01
4.62820381e-01 -1.18019474e+00 5.78396976e-01 3.68293375e-01
8.62717032e-01 -4.29584801e-01 8.46649170e-01 3.76077533e-01
-1.15410805e+00 4.81738061e-01 3.26616138e-01 -7.00071454e-01
4.51330364e-01 8.53812695e-01 -6.85970902e-01 5.83079875e-01
1.28685189e-02 6.31506026e-01 -4.08713311e-01 5.49256384e-01
-7.93610275e-01 7.31669307e-01 -2.12013379e-01 -6.25589073e-01
6.58931062e-02 -2.34721676e-01 6.58351600e-01 1.93533170e+00
-4.92197514e-01 7.40608811e-01 1.91538736e-01 8.11561763e-01
-4.79492657e-02 2.09969565e-01 -7.83815920e-01 -4.48410034e-01
6.33671284e-01 1.21642005e+00 -6.10698283e-01 -3.75761479e-01
-6.58584237e-01 6.15227580e-01 4.03535783e-01 3.94730449e-01
-6.44911289e-01 -7.92842627e-01 7.07765579e-01 -1.52687371e-01
-7.35653639e-02 -6.01930320e-01 -6.35918558e-01 -1.33414078e+00
3.88531327e-01 -7.57799327e-01 4.19541866e-01 -4.49166685e-01
-1.44334865e+00 6.90828264e-01 1.06615789e-01 -5.40093958e-01
-4.76782590e-01 -5.09825408e-01 -4.34682190e-01 1.05427444e+00
-9.51580942e-01 -9.28500772e-01 7.82723576e-02 1.78021178e-01
6.14616215e-01 4.25478965e-01 1.18294001e+00 3.61360312e-01
-6.32536113e-01 5.25404990e-01 -7.42750645e-01 4.79006350e-01
6.37524307e-01 -1.42262948e+00 5.27304530e-01 1.07769442e+00
3.22683960e-01 1.29415917e+00 8.66214931e-01 -6.90785468e-01
-1.08918536e+00 -7.54574120e-01 1.95849276e+00 -8.59617829e-01
9.06085074e-01 -4.41689670e-01 -7.92503178e-01 8.81019592e-01
6.70355678e-01 -5.01556218e-01 1.18002152e+00 7.62421608e-01
-5.44506371e-01 2.79487371e-01 -8.29851925e-01 6.34139955e-01
1.21334815e+00 -8.66502821e-01 -1.40713656e+00 5.22718668e-01
9.54888284e-01 -5.57988405e-01 -8.81672859e-01 2.61968404e-01
6.36617064e-01 -2.77314574e-01 3.72820407e-01 -8.37849557e-01
1.96443781e-01 -2.24908069e-01 -8.71095899e-03 -9.56287146e-01
-1.38642415e-01 -9.24518347e-01 -7.28350738e-03 1.41442943e+00
1.08095932e+00 -4.88137215e-01 1.85471341e-01 1.34609365e+00
-4.12355885e-02 -8.13019037e-01 -7.96540082e-01 -5.76515734e-01
8.79603475e-02 -7.17226565e-01 4.13785160e-01 1.20868361e+00
6.13921404e-01 1.34055853e+00 3.29922855e-01 -2.93240696e-01
3.09489161e-01 2.46039480e-01 4.19886440e-01 -9.59442377e-01
-1.21865101e-01 -5.78240395e-01 -2.84186006e-01 -1.13884091e+00
3.43410492e-01 -8.93240929e-01 1.61609486e-01 -1.80629241e+00
6.00505546e-02 -3.54424685e-01 1.02054462e-01 5.94731152e-01
-2.94059902e-01 -2.58445084e-01 6.37883991e-02 -1.72078848e-01
-7.11450398e-01 4.92057167e-02 7.41473258e-01 -2.34341651e-01
-1.23525858e-01 -2.86301315e-01 -8.70736778e-01 5.93378127e-01
6.03562176e-01 -4.88789737e-01 -3.15720379e-01 -6.81463599e-01
6.20653570e-01 -1.52058126e-02 -9.84148011e-02 -3.21181685e-01
1.32223845e-01 -5.32364368e-01 -2.46702373e-01 -3.12579691e-01
-2.66477535e-03 -4.05120760e-01 6.79915845e-02 -5.81783876e-02
-5.53912640e-01 2.30181649e-01 1.56306192e-01 -2.34509464e-02
-2.77205497e-01 -4.46457773e-01 2.87203580e-01 -3.20778847e-01
-6.22408032e-01 -1.41122580e-01 -6.45098746e-01 5.38968682e-01
9.23448145e-01 -1.77817002e-01 -4.80595231e-01 -5.35696521e-02
-7.82593071e-01 2.81501055e-01 2.47538432e-01 3.81610930e-01
6.05604291e-01 -8.52165937e-01 -7.60441601e-01 -1.11464441e-01
6.97807744e-02 1.50618717e-01 -5.24723172e-01 5.50965786e-01
-3.47169906e-01 5.14188409e-01 4.84159917e-01 -5.53976111e-02
-1.47387421e+00 4.71042693e-01 3.48581642e-01 -5.39257288e-01
-5.41346908e-01 8.76045883e-01 -1.72875479e-01 -4.79814321e-01
5.07560074e-02 -4.27928060e-01 -3.36486757e-01 1.70345590e-01
6.03248537e-01 -8.58988687e-02 1.11524528e-02 -6.17473841e-01
-7.41275966e-01 3.20617110e-01 -1.11747645e-01 -3.76792073e-01
9.87838507e-01 -2.26728648e-01 -5.98723292e-01 9.61782098e-01
1.02171659e+00 2.75602698e-01 -2.04313040e-01 -3.57574940e-01
8.46381843e-01 -6.15244992e-02 -2.16070250e-01 -1.06015658e+00
-8.83252770e-02 4.81043272e-02 -3.04833204e-01 6.36544466e-01
8.94379437e-01 7.45189041e-02 9.41804886e-01 3.10416579e-01
5.43440878e-01 -1.05066586e+00 -4.32334512e-01 1.09612012e+00
9.15921390e-01 -9.97067451e-01 2.48884201e-01 -7.03455806e-01
-1.99789539e-01 1.12676120e+00 5.07243037e-01 3.19524080e-01
6.72531605e-01 5.51706910e-01 2.67350256e-01 -6.09343171e-01
-1.09679461e+00 -4.96074200e-01 2.35167459e-01 3.50395232e-01
1.13866520e+00 1.67522296e-01 -1.19620204e+00 8.66896152e-01
-2.97845602e-01 -1.98195487e-01 4.66130108e-01 1.24346483e+00
-1.65614352e-01 -1.50845277e+00 3.09717413e-02 6.07474089e-01
-8.09829891e-01 -7.28677213e-01 -1.05739379e+00 9.12499070e-01
4.00986075e-01 1.54430974e+00 -1.34650350e-01 -2.66319782e-01
3.90753806e-01 1.93724111e-01 6.03747427e-01 -1.42958724e+00
-1.04721212e+00 -4.72651958e-01 1.07212532e+00 -2.38087431e-01
-6.04667962e-01 -8.27977777e-01 -1.59164715e+00 -8.17422494e-02
-7.20562398e-01 5.97075880e-01 4.16573763e-01 1.55448055e+00
-9.15047973e-02 3.86181861e-01 1.33316040e-01 1.63290232e-01
-3.97295356e-01 -1.25869715e+00 -4.12321270e-01 4.86676306e-01
3.14215720e-01 -2.93406010e-01 -6.01884604e-01 1.28440499e-01]
|
[9.913338661193848, 8.938593864440918]
|
5c161b05-bc8c-456d-858d-74c446afd116
|
context-modeling-with-evidence-filter-for
|
2010.02649
| null |
https://arxiv.org/abs/2010.02649v1
|
https://arxiv.org/pdf/2010.02649v1.pdf
|
Context Modeling with Evidence Filter for Multiple Choice Question Answering
|
Multiple-Choice Question Answering (MCQA) is a challenging task in machine reading comprehension. The main challenge in MCQA is to extract "evidence" from the given context that supports the correct answer. In the OpenbookQA dataset, the requirement of extracting "evidence" is particularly important due to the mutual independence of sentences in the context. Existing work tackles this problem by annotated evidence or distant supervision with rules which overly rely on human efforts. To address the challenge, we propose a simple yet effective approach termed evidence filtering to model the relationships between the encoded contexts with respect to different options collectively and to potentially highlight the evidence sentences and filter out unrelated sentences. In addition to the effective reduction of human efforts of our approach compared, through extensive experiments on OpenbookQA, we show that the proposed approach outperforms the models that use the same backbone and more training data; and our parameter analysis also demonstrates the interpretability of our approach.
|
['Jing Jiang', 'Wei Jing', 'Hao Zhang', 'Sicheng Yu']
|
2020-10-06
| null | null | null | null |
['multiple-choice-qa']
|
['natural-language-processing']
|
[ 3.67829174e-01 3.04850280e-01 1.55282095e-01 -6.10327482e-01
-1.07468379e+00 -6.26314282e-01 3.38764697e-01 6.32774591e-01
-4.42539632e-01 1.06336832e+00 4.82491285e-01 -6.52705669e-01
-4.12317812e-01 -5.65104723e-01 -7.90202916e-01 -1.89551979e-01
2.05782056e-01 3.48840594e-01 6.86273038e-01 -5.76823950e-01
6.30467653e-01 -1.87378585e-01 -1.81045973e+00 6.01819813e-01
1.42630136e+00 8.39381993e-01 4.52866852e-01 9.10852134e-01
-5.20910978e-01 1.07835066e+00 -8.10600758e-01 -7.01940358e-01
8.93938262e-03 -4.69533682e-01 -1.34839451e+00 -1.62552938e-01
8.19681346e-01 -2.15373337e-01 8.37355033e-02 9.44992065e-01
4.55140948e-01 3.49590987e-01 3.86077553e-01 -8.72986138e-01
-4.93083954e-01 6.92171693e-01 -1.07453652e-01 7.15642631e-01
8.69616628e-01 -8.91530886e-02 1.52686894e+00 -8.64381373e-01
4.77991134e-01 1.13693643e+00 3.25793743e-01 5.66612184e-01
-8.61373067e-01 -8.37195963e-02 4.41532612e-01 8.80577147e-01
-9.69421625e-01 -3.02294016e-01 5.83748460e-01 -2.17444301e-01
9.73915279e-01 7.40841806e-01 1.95446253e-01 9.63362694e-01
-9.53294262e-02 9.58919466e-01 1.28015244e+00 -9.61390853e-01
3.19467336e-01 5.66722527e-02 9.86159205e-01 6.22813046e-01
6.56117424e-02 -4.72414345e-01 -7.19382286e-01 -3.11319292e-01
6.02135248e-02 -2.81387419e-01 -5.72820246e-01 2.72165816e-02
-8.92202675e-01 7.49223292e-01 7.81164914e-02 2.65833467e-01
-2.73225039e-01 -1.58751711e-01 1.66479319e-01 4.59459275e-01
1.14848115e-01 7.00748265e-01 -8.19081247e-01 -1.60485700e-01
-6.30162477e-01 4.35552031e-01 1.04760861e+00 1.00825179e+00
6.09559476e-01 -7.45403707e-01 -5.54098427e-01 8.42761934e-01
2.90246636e-01 2.84958094e-01 2.67397135e-01 -9.85301673e-01
9.67938662e-01 9.27488506e-01 2.89879233e-01 -8.22812676e-01
-1.49373785e-01 -4.01702732e-01 -1.76938727e-01 -1.27572715e-01
7.27364421e-01 2.94919517e-02 -6.26085222e-01 1.70275879e+00
4.03787494e-01 1.20546035e-02 2.48278514e-01 8.14382374e-01
8.96432579e-01 4.82867837e-01 2.03773007e-01 -1.24298155e-01
1.76327741e+00 -1.24940407e+00 -1.07826650e+00 -2.98679978e-01
5.38790047e-01 -7.19771504e-01 1.43959415e+00 3.59186471e-01
-1.01128447e+00 -2.99592912e-01 -1.14338374e+00 -4.70543891e-01
-2.85537988e-01 5.11401240e-03 3.56560081e-01 4.35498714e-01
-6.53834581e-01 3.49616677e-01 -2.07005158e-01 -1.08455539e-01
1.35586157e-01 1.58349633e-01 -1.51046410e-01 -3.50620747e-01
-1.49429548e+00 1.13744044e+00 4.03034866e-01 2.57599615e-02
-5.36924839e-01 -4.89977568e-01 -6.87443852e-01 3.16508114e-01
1.02597630e+00 -8.40529561e-01 1.53265214e+00 -1.08922136e+00
-1.35958612e+00 4.33376640e-01 -6.53556585e-01 -5.34523666e-01
3.06225061e-01 -7.87011564e-01 -2.97636032e-01 4.77195531e-01
1.94462553e-01 2.65187562e-01 6.50250137e-01 -1.00099814e+00
-7.75948226e-01 -3.38169903e-01 6.30072474e-01 3.62976432e-01
-1.67327508e-01 2.97643125e-01 -3.39574665e-01 -4.46834147e-01
-2.00760621e-03 -5.96788466e-01 -2.13701904e-01 -3.75825197e-01
-2.89944768e-01 -7.34684587e-01 4.11682487e-01 -8.48169386e-01
1.56358206e+00 -1.79625237e+00 2.37877727e-01 1.88754592e-02
2.25257084e-01 1.86639234e-01 -7.40210712e-02 4.60342735e-01
2.15425432e-01 1.91166982e-01 -4.64499623e-01 -1.36235729e-01
3.21464166e-02 3.75159651e-01 -5.26737928e-01 -3.32744300e-01
5.00613630e-01 6.70288265e-01 -1.13030612e+00 -8.24838042e-01
-1.96516484e-01 -9.74881127e-02 -4.73225743e-01 4.81512368e-01
-7.57410407e-01 3.00120741e-01 -5.91177583e-01 5.62072754e-01
4.45569664e-01 -5.83047152e-01 2.31779590e-01 1.79604739e-01
2.16936067e-01 7.43639708e-01 -1.21892464e+00 1.57480121e+00
-2.05446303e-01 3.43845963e-01 -4.73750085e-02 -7.21256971e-01
6.90516710e-01 4.03825164e-01 -4.65677589e-01 -7.08053589e-01
-1.14173084e-01 3.78948539e-01 2.43626580e-01 -1.11833942e+00
5.59225500e-01 1.34845152e-01 1.27095208e-01 3.26677233e-01
1.35752067e-01 1.47628501e-01 4.58934546e-01 4.62874562e-01
1.25697553e+00 1.33396700e-01 5.63739657e-01 -2.71252364e-01
9.55894768e-01 1.61343277e-01 5.01249254e-01 1.06792486e+00
-2.00371191e-01 3.73335421e-01 4.82428610e-01 -2.47968853e-01
-7.02984512e-01 -8.38416100e-01 5.10314181e-02 1.31817567e+00
8.08715224e-02 -5.67743063e-01 -6.77085042e-01 -1.08365357e+00
-2.96197236e-01 1.05869603e+00 -6.82680964e-01 3.45257789e-01
-8.37368250e-01 -3.81227612e-01 3.77120882e-01 4.97255623e-01
4.52786595e-01 -1.00522113e+00 -7.70181239e-01 1.90778881e-01
-7.00096428e-01 -9.12461936e-01 -9.62826163e-02 1.02233864e-01
-7.14983821e-01 -1.41494226e+00 -2.82282621e-01 -5.73206306e-01
5.57416856e-01 2.82230020e-01 1.61433220e+00 6.84149563e-01
9.13474113e-02 4.57888007e-01 -7.63546586e-01 -6.49522126e-01
-3.17812383e-01 1.92616209e-02 -4.61159080e-01 -1.25594154e-01
6.47337019e-01 -3.11760664e-01 -6.11869872e-01 2.09098861e-01
-8.82681668e-01 -7.37960413e-02 5.58353841e-01 1.01382232e+00
4.05219883e-01 -1.91157117e-01 9.32373345e-01 -9.94977415e-01
9.73282516e-01 -7.67123759e-01 -3.91463876e-01 9.97901082e-01
-4.94696200e-01 3.04866195e-01 5.94656110e-01 6.01805933e-02
-1.27710319e+00 -2.89501339e-01 -1.46048293e-01 1.37220278e-01
-1.71027899e-01 6.22596920e-01 -2.25319147e-01 3.74975085e-01
5.22219419e-01 8.54909644e-02 -2.11393923e-01 -5.24825454e-01
4.79574800e-01 4.50412512e-01 3.35103393e-01 -8.97433996e-01
2.87119240e-01 3.87380794e-02 -9.59884748e-02 -4.99863476e-01
-1.38561916e+00 -5.08951783e-01 -4.95540172e-01 -9.12376568e-02
8.62905443e-01 -5.56515634e-01 -6.91708326e-01 -1.58844560e-01
-1.30610752e+00 2.03794196e-01 -7.14949518e-02 2.02780679e-01
-1.94779783e-01 8.08996499e-01 -4.32934403e-01 -1.10348427e+00
-4.14732188e-01 -9.17858839e-01 7.56242156e-01 4.37013537e-01
-3.79123718e-01 -8.31527472e-01 -3.29636075e-02 9.89139974e-01
2.36035496e-01 -2.50351608e-01 1.46350324e+00 -1.21657383e+00
-8.12153459e-01 3.09485272e-02 3.28122564e-02 2.98421323e-01
-1.01159602e-01 -7.29937404e-02 -8.76678646e-01 1.72442883e-01
1.83249265e-01 -5.30179679e-01 7.52039969e-01 -2.50313822e-02
1.15601790e+00 -2.63662964e-01 8.74826983e-02 -3.63127619e-01
1.17392433e+00 -4.67871092e-02 4.94898051e-01 4.49363142e-01
2.96838552e-01 1.14126122e+00 1.03234470e+00 8.36729184e-02
5.90380013e-01 5.03972828e-01 4.58770335e-01 2.68965662e-01
5.53207919e-02 -3.23826313e-01 3.07986178e-02 9.58985865e-01
-1.41114686e-02 -5.16619742e-01 -7.76488960e-01 6.85884416e-01
-2.10888124e+00 -1.07446647e+00 -3.75851542e-01 2.02344871e+00
9.75586176e-01 1.33059800e-01 -3.15514356e-01 2.88983941e-01
4.94472653e-01 -7.18022063e-02 -2.78377295e-01 -5.81112981e-01
-3.61876190e-01 4.83193994e-01 -2.56438762e-01 7.86455214e-01
-8.28617036e-01 6.51538193e-01 6.87496996e+00 8.32523763e-01
-2.96345234e-01 3.57333198e-02 6.07936203e-01 1.18251354e-01
-5.82294285e-01 2.22895563e-01 -8.47473979e-01 3.50004405e-01
7.71019220e-01 9.29047018e-02 1.14750579e-01 6.26648903e-01
-9.76690929e-03 -5.33753514e-01 -1.25546527e+00 2.87280649e-01
2.93433934e-01 -1.18267322e+00 2.38877803e-01 -5.65646946e-01
4.53990281e-01 -5.26210666e-01 -2.62586683e-01 4.47242290e-01
1.77076384e-01 -1.00431287e+00 6.28040135e-01 6.24646187e-01
-7.96501264e-02 -5.03891528e-01 9.53517020e-01 7.69008338e-01
-7.44464576e-01 -4.23226207e-01 -3.22682351e-01 -4.63840485e-01
1.71491563e-01 3.98190767e-01 -9.73794997e-01 7.95563877e-01
7.92839527e-01 1.24338113e-01 -9.42195475e-01 8.51652741e-01
-9.29336071e-01 1.05869818e+00 -1.06276765e-01 -5.86099565e-01
1.04194999e-01 -6.38961121e-02 6.18053198e-01 1.01499450e+00
3.39510702e-02 4.50787961e-01 -3.86713706e-02 6.52151823e-01
-9.22475383e-03 4.05438572e-01 -2.16931775e-01 2.87320971e-01
7.03088641e-01 8.55959117e-01 -2.53606766e-01 -6.72870040e-01
-6.37567818e-01 8.57397437e-01 7.92105913e-01 1.90484062e-01
-5.29705882e-01 -2.96886146e-01 8.33726823e-02 -2.39940256e-01
4.04052585e-01 9.53386724e-02 -3.48623961e-01 -1.10258722e+00
6.55892551e-01 -1.22922480e+00 8.91610026e-01 -9.06179845e-01
-1.46065283e+00 5.80251813e-01 5.05900420e-02 -9.80927587e-01
-1.81467101e-01 -5.48766136e-01 -6.53020680e-01 1.03118825e+00
-1.82348335e+00 -7.36935556e-01 -1.22396857e-01 5.04087031e-01
8.66692901e-01 1.28056869e-01 8.49738121e-01 7.50044286e-02
-4.63961303e-01 5.32848418e-01 -2.34191164e-01 -2.54427701e-01
6.44469798e-01 -1.65142143e+00 9.39628948e-03 1.17945671e+00
1.94000244e-01 1.03366327e+00 8.80627275e-01 -6.22578800e-01
-1.23066175e+00 -4.72925842e-01 1.50329828e+00 -9.33194280e-01
5.91731787e-01 -1.99032560e-01 -1.18881238e+00 5.48657298e-01
6.72832966e-01 -3.21426511e-01 1.02852690e+00 3.82555217e-01
-4.61993009e-01 8.83433074e-02 -9.45968151e-01 6.69982195e-01
9.03812170e-01 -4.45825219e-01 -1.63750923e+00 1.78670302e-01
9.84522939e-01 -2.35729948e-01 -5.63296139e-01 3.74090642e-01
2.45012805e-01 -1.02902651e+00 6.85680091e-01 -1.01000881e+00
7.31015861e-01 -5.28784752e-01 -2.88438439e-01 -1.00993574e+00
-1.50042539e-02 -4.42733854e-01 -2.98410386e-01 1.20693707e+00
8.67431462e-01 -3.88717085e-01 4.36681300e-01 9.55742478e-01
-2.33331308e-01 -8.50017309e-01 -1.20456612e+00 -2.64306396e-01
-8.29344541e-02 -3.02378684e-01 5.87273002e-01 7.32011497e-01
1.39465153e-01 6.38973475e-01 -2.69512773e-01 4.25052315e-01
2.43237510e-01 2.65156478e-01 6.22727513e-01 -9.93674397e-01
-4.87549961e-01 -1.13253653e-01 5.03453910e-02 -1.45287085e+00
-6.01842664e-02 -5.87780654e-01 2.01132759e-01 -1.53948522e+00
3.86344820e-01 -2.95581967e-01 -5.18570662e-01 2.97260582e-01
-1.12136757e+00 -3.97052616e-01 1.76823065e-01 -2.80105993e-02
-1.06745005e+00 4.43421990e-01 1.17544520e+00 -9.61395502e-02
1.20404497e-01 -6.05991716e-03 -8.12356353e-01 7.36426473e-01
5.89869797e-01 -3.12338322e-01 -6.80835724e-01 -7.25577712e-01
6.76490843e-01 1.48051128e-01 2.98849523e-01 -6.71269834e-01
2.94839084e-01 -7.61419460e-02 6.12311959e-02 -7.42905974e-01
2.82111228e-01 -9.00455713e-01 -5.97798288e-01 1.54994503e-01
-7.49539137e-01 4.00136709e-01 3.09648663e-02 8.60296369e-01
-3.82218540e-01 -7.71429121e-01 2.22169343e-04 -1.90925643e-01
-8.32957625e-01 -3.95112902e-01 -3.14205974e-01 3.56663793e-01
8.25477302e-01 6.95626065e-02 -7.37011671e-01 -4.15909141e-01
-6.38246238e-01 6.04472101e-01 -3.88692394e-02 4.64410484e-01
6.81521654e-01 -8.46676052e-01 -8.87533486e-01 -2.43312046e-01
3.50550562e-01 3.73413563e-02 2.47017145e-01 6.26259565e-01
-2.26709485e-01 4.15336370e-01 8.42058137e-02 -3.68530065e-01
-1.60562038e+00 5.13657510e-01 1.33072168e-01 -5.60199499e-01
-2.90260643e-01 9.82298672e-01 -2.10367888e-01 -2.86981940e-01
3.25188249e-01 -3.53788167e-01 -8.50001097e-01 -1.66340590e-01
6.98080659e-01 3.90732378e-01 2.37263158e-01 -1.28048271e-01
-2.72695750e-01 1.41494304e-01 -2.85960823e-01 -5.34730963e-02
9.69003618e-01 -3.30519199e-01 -3.84070992e-01 4.85112935e-01
4.18302000e-01 4.15903687e-01 -8.75832498e-01 -3.72355431e-01
5.93848944e-01 -4.62452412e-01 -4.99232203e-01 -1.26743591e+00
-1.42345876e-01 9.25026834e-01 2.81280965e-01 3.66325140e-01
1.04544508e+00 4.54882719e-02 5.96483409e-01 8.13277781e-01
2.07572252e-01 -1.15009534e+00 1.35107875e-01 6.48986518e-01
9.95441496e-01 -1.42723525e+00 4.05613370e-02 -7.62142420e-01
-5.48330247e-01 1.12565243e+00 9.65740800e-01 1.89244032e-01
1.72884360e-01 -2.27594033e-01 2.22913504e-01 -3.11280310e-01
-1.15145814e+00 -1.38729721e-01 4.25275326e-01 6.42710030e-02
5.63518703e-01 -5.88053428e-02 -9.77760792e-01 8.90521348e-01
-1.10520229e-01 -3.42686027e-01 4.66730446e-01 1.27800286e+00
-6.82105839e-01 -1.18974364e+00 -4.95325685e-01 4.02121931e-01
-6.85524285e-01 -4.14061964e-01 -6.61998928e-01 5.47072887e-01
1.05917528e-01 1.57759094e+00 -2.58831441e-01 2.57055182e-02
3.52862090e-01 4.30098742e-01 4.11895841e-01 -5.76875508e-01
-9.43366706e-01 -4.57803875e-01 5.91046214e-01 -3.18650097e-01
-6.58749402e-01 -3.77038747e-01 -1.09213126e+00 2.73544639e-01
-5.91544509e-01 4.95443016e-01 3.99580032e-01 1.42609406e+00
4.79400963e-01 4.77138191e-01 3.98357481e-01 2.53982961e-01
-9.29602623e-01 -1.13857377e+00 3.11408136e-02 7.30916202e-01
2.88326621e-01 -5.13006926e-01 -3.70238096e-01 -1.61128417e-01]
|
[11.253783226013184, 8.068102836608887]
|
95a2d4a8-a9ff-48bb-acc6-b125f7e74feb
|
multi-view-integration-learning-for
|
2101.09986
| null |
https://arxiv.org/abs/2101.09986v2
|
https://arxiv.org/pdf/2101.09986v2.pdf
|
Multi-view Integration Learning for Irregularly-sampled Clinical Time Series
|
Electronic health record (EHR) data is sparse and irregular as it is recorded at irregular time intervals, and different clinical variables are measured at each observation point. In this work, we propose a multi-view features integration learning from irregular multivariate time series data by self-attention mechanism in an imputation-free manner. Specifically, we devise a novel multi-integration attention module (MIAM) to extract complex information inherent in irregular time series data. In particular, we explicitly learn the relationships among the observed values, missing indicators, and time interval between the consecutive observations, simultaneously. The rationale behind our approach is the use of human knowledge such as what to measure and when to measure in different situations, which are indirectly represented in the data. In addition, we build an attention-based decoder as a missing value imputer that helps empower the representation learning of the inter-relations among multi-view observations for the prediction task, which operates at the training phase only. We validated the effectiveness of our method over the public MIMIC-III and PhysioNet challenge 2012 datasets by comparing with and outperforming the state-of-the-art methods for in-hospital mortality prediction.
|
['Heung-Il Suk', 'Eunji Jun', 'Yurim Lee']
|
2021-01-25
| null | null | null | null |
['irregular-time-series']
|
['time-series']
|
[ 1.19937494e-01 5.99898305e-03 -3.13914031e-01 -5.24487853e-01
-9.69241679e-01 -1.81880444e-01 1.78047433e-01 6.93659365e-01
-2.98567295e-01 8.32303822e-01 7.69593954e-01 -1.19298562e-01
-2.85755932e-01 -6.30967796e-01 -9.16246176e-01 -6.17376566e-01
-2.61664003e-01 5.19771039e-01 -8.07776749e-01 -8.75227712e-03
-1.43541664e-01 4.70875800e-02 -1.11207223e+00 4.50553834e-01
6.79459274e-01 1.05659151e+00 -2.75211722e-01 4.45212811e-01
2.01972798e-01 1.16692650e+00 -3.25898588e-01 -3.18254709e-01
-1.46277115e-01 -3.45281452e-01 -4.20017451e-01 -2.61135101e-02
-9.92130712e-02 -3.11265081e-01 -4.71615881e-01 4.17239934e-01
7.03240037e-01 -2.72072226e-01 7.22485840e-01 -1.03259993e+00
-7.08131433e-01 6.07193530e-01 -5.06437242e-01 2.60771930e-01
3.17446202e-01 1.58167660e-01 9.14801776e-01 -7.22468495e-01
4.15030926e-01 7.25183666e-01 9.73107874e-01 4.10312772e-01
-1.24144244e+00 -3.78679514e-01 1.08068071e-01 3.90950412e-01
-1.30060387e+00 -4.34766471e-01 7.30596244e-01 -5.97894430e-01
1.00863576e+00 3.49106491e-01 5.92590868e-01 1.45445883e+00
5.67155480e-01 7.70236194e-01 6.53891146e-01 1.38973027e-01
7.12691694e-02 -2.60785758e-01 1.54124200e-01 4.04917449e-01
-3.69466990e-02 4.98155132e-02 -2.86830634e-01 -5.60012758e-01
5.39111674e-01 8.54769409e-01 -2.40724534e-01 2.79843411e-03
-1.62180626e+00 5.59212267e-01 2.15050116e-01 7.16044381e-02
-9.16989625e-01 -5.95199130e-02 7.38003194e-01 3.57121378e-01
4.53775793e-01 1.95134133e-01 -1.02346957e+00 -1.01938784e-01
-6.62997246e-01 -1.19267993e-01 7.46157229e-01 7.04970419e-01
2.45021641e-01 -1.94495931e-01 -5.60171604e-01 4.91796255e-01
4.30822857e-02 2.54806519e-01 7.56161273e-01 -6.69363856e-01
7.93162465e-01 7.35951006e-01 1.56920284e-01 -7.48723447e-01
-7.76979685e-01 -5.38331091e-01 -1.48062849e+00 -4.94644165e-01
1.17481768e-01 -4.50433642e-01 -8.05786908e-01 2.01117468e+00
2.45982707e-01 6.10217750e-01 1.42915756e-01 5.19095719e-01
8.89314353e-01 2.99872488e-01 2.53006846e-01 -5.27116537e-01
1.40736783e+00 -6.24097168e-01 -1.02864707e+00 2.83042908e-01
1.04314673e+00 -1.34677947e-01 5.40394127e-01 1.84049979e-01
-8.75509083e-01 -4.79229242e-01 -7.17451155e-01 -1.65281639e-01
-2.68969625e-01 1.32462740e-01 4.25923914e-01 -1.64769411e-01
-3.38749528e-01 6.58443630e-01 -1.15754962e+00 9.77135226e-02
6.03015423e-01 3.69506180e-01 -6.84531152e-01 2.44831275e-02
-1.24828792e+00 6.85412169e-01 1.22721508e-01 1.88302696e-01
-6.99747503e-01 -1.16473782e+00 -1.01271963e+00 2.81331867e-01
1.82487458e-01 -1.14164340e+00 7.42610276e-01 -6.22811913e-01
-8.72729719e-01 6.32195234e-01 -3.02220166e-01 -6.52105272e-01
4.84749705e-01 -2.71250069e-01 -6.46937490e-01 -2.16580912e-01
1.57060131e-01 1.99983902e-02 3.51067185e-01 -7.42767572e-01
-2.64524907e-01 -7.84138203e-01 -4.44730639e-01 7.31518492e-03
-7.58145819e-04 -3.17024171e-01 -2.86708236e-01 -8.35380793e-01
-1.72688633e-01 -6.50842488e-01 -3.72824520e-01 -2.17893109e-01
-5.62017024e-01 -2.28857756e-01 3.74500930e-01 -1.08498383e+00
1.47729242e+00 -2.35743928e+00 3.98090750e-01 -3.19212042e-02
5.52210152e-01 -1.62211090e-01 6.93506002e-02 6.43329561e-01
-3.22055876e-01 -9.52315703e-02 -3.76233906e-01 -5.74494064e-01
-2.71713287e-01 2.00886071e-01 -1.98682725e-01 5.48216343e-01
3.06205779e-01 1.21753848e+00 -7.59334564e-01 -4.51146036e-01
1.30624637e-01 7.79769778e-01 -6.39276683e-01 5.35739362e-01
2.03864742e-02 9.46975470e-01 -5.25597215e-01 7.04399228e-01
4.17706549e-01 -7.66590297e-01 1.25651389e-01 -3.19033414e-01
9.97587293e-02 2.14500666e-01 -7.13786006e-01 1.95290101e+00
-4.64455485e-01 6.89483434e-02 -4.73579407e-01 -1.20041442e+00
6.70968115e-01 8.69832635e-01 1.12461185e+00 -6.19713008e-01
1.26121178e-01 -6.98546022e-02 -4.20266129e-02 -9.53109860e-01
-3.44290644e-01 -9.98377725e-02 -1.89807624e-01 1.25108466e-01
5.80241494e-02 7.26240754e-01 -1.70320123e-01 -5.14218137e-02
1.38315439e+00 -1.82505548e-02 7.59395123e-01 2.76141673e-01
4.94914740e-01 -3.94997150e-01 9.65536475e-01 6.21501207e-01
8.66063535e-02 8.55017304e-01 6.67339385e-01 -9.37242866e-01
-7.82733917e-01 -9.16986525e-01 -3.58425140e-01 5.98690033e-01
-3.95506054e-01 -4.50773388e-01 -1.47635609e-01 -8.06590319e-01
2.37788126e-01 4.27399904e-01 -1.38381028e+00 -2.16274455e-01
-4.38772082e-01 -1.06635368e+00 3.49466801e-01 8.47448170e-01
-7.53917322e-02 -1.07451999e+00 -6.50851786e-01 7.03275919e-01
-4.89014328e-01 -1.02642500e+00 -5.39016485e-01 4.97334123e-01
-9.83149230e-01 -1.24767137e+00 -5.67115903e-01 -3.08462560e-01
4.88448888e-01 -3.47930253e-01 1.34707880e+00 1.29834265e-02
-2.26707906e-01 1.37891740e-01 -2.12194934e-01 -5.98487616e-01
1.10048071e-01 2.23751087e-02 -1.41676173e-01 4.57227796e-01
5.86531818e-01 -7.09588706e-01 -8.84561419e-01 -1.63216412e-01
-7.20202863e-01 1.82801977e-01 5.90227067e-01 1.18881023e+00
1.02308381e+00 -5.39550960e-01 1.00499046e+00 -1.17141545e+00
2.21135393e-01 -1.33708608e+00 -3.38923968e-02 2.60586321e-01
-5.10042131e-01 1.46320418e-01 9.93187070e-01 -2.38170519e-01
-4.53746885e-01 1.82927966e-01 -1.91511363e-01 -5.90263128e-01
-3.39743197e-01 8.08324516e-01 -2.22159147e-01 8.03567171e-01
1.09065287e-01 4.59600151e-01 -8.15367268e-04 -6.63056672e-01
-7.97648057e-02 6.69998229e-01 5.12912214e-01 -2.15909243e-01
9.60935354e-02 5.08249462e-01 2.23735124e-02 -2.57710308e-01
-1.13373983e+00 -5.52943945e-01 -5.25721669e-01 4.43515480e-01
9.37893093e-01 -1.48474765e+00 -8.99317920e-01 4.15827274e-01
-1.03713036e+00 -1.00942329e-02 -3.83334011e-01 6.22287273e-01
-7.33416557e-01 7.60061070e-02 -6.45924330e-01 -6.05634570e-01
-6.73636496e-01 -1.02386129e+00 1.37810969e+00 -2.13711202e-01
-2.27895677e-01 -1.05631781e+00 2.24101782e-01 3.00949156e-01
2.28364944e-01 8.75498474e-01 1.09714687e+00 -8.50530744e-01
-2.13564247e-01 -3.06892097e-01 -1.57483846e-01 -7.75354579e-02
4.31158662e-01 -4.88471776e-01 -6.68498755e-01 -2.66380906e-01
7.89890438e-02 7.19199050e-03 9.23862517e-01 6.05715156e-01
1.84427166e+00 -5.75399935e-01 -3.82314026e-01 1.19719660e+00
1.37670541e+00 4.45443466e-02 5.52125454e-01 7.26962611e-02
9.30837810e-01 2.27273419e-01 1.10782146e-01 1.01828647e+00
1.03108156e+00 7.24063993e-01 4.37356949e-01 -4.15710121e-01
3.37466568e-01 -1.97153777e-01 -1.12135693e-01 1.01053226e+00
-1.29481733e-01 -1.60302788e-01 -7.20967531e-01 7.11697698e-01
-2.11943913e+00 -1.00048923e+00 -1.44203618e-01 2.16939878e+00
9.92979944e-01 -2.50308394e-01 4.66302484e-02 -4.23012599e-02
2.62983769e-01 6.65002735e-03 -8.89688373e-01 -1.26032606e-01
-7.20753968e-02 1.46533385e-01 4.04418170e-01 6.83936924e-02
-1.22008669e+00 8.63378346e-02 5.38402367e+00 1.09032020e-01
-9.84379113e-01 1.14711255e-01 8.16014528e-01 -2.72785693e-01
-2.14385852e-01 -4.43818301e-01 -3.87489617e-01 9.32854354e-01
1.23559761e+00 -1.03354238e-01 1.66929498e-01 2.97565490e-01
3.96705776e-01 6.04455292e-01 -1.45516253e+00 1.07772565e+00
-8.01407360e-03 -1.31760597e+00 -2.39704549e-02 7.83571005e-02
4.39562261e-01 3.41080099e-01 -3.45291980e-02 3.35233897e-01
1.40241729e-02 -1.11550510e+00 1.09517358e-01 1.10307622e+00
7.63722420e-01 -5.80618799e-01 1.05001688e+00 2.85875529e-01
-1.11414015e+00 -2.76654452e-01 1.63870692e-01 1.12291120e-01
1.73644856e-01 6.91876173e-01 -4.83451843e-01 1.01200521e+00
5.03330886e-01 1.39542520e+00 -4.44093049e-01 9.73578215e-01
2.76291609e-01 6.11361206e-01 -1.03206597e-02 6.72349453e-01
-1.61761314e-01 -2.12333947e-02 1.25829518e-01 9.14487422e-01
4.86893684e-01 2.76959121e-01 5.44604473e-02 6.03771389e-01
-1.82233766e-01 6.32172152e-02 -7.34073520e-01 1.51898623e-01
2.31797755e-01 8.74040008e-01 1.43240139e-01 -4.02663916e-01
-9.31929708e-01 8.18078876e-01 5.04746199e-01 3.73380691e-01
-1.03057170e+00 -4.17944156e-02 6.36927128e-01 1.80925310e-01
2.86814153e-01 1.29786268e-01 -4.56757754e-01 -1.60682046e+00
5.61006367e-02 -1.02386117e+00 8.93626451e-01 -4.48034406e-01
-1.66232610e+00 5.77845037e-01 -4.08753872e-01 -1.57588112e+00
-4.26186800e-01 -2.61224676e-02 -4.69385415e-01 8.81387591e-01
-1.45442343e+00 -1.12660658e+00 -2.39445984e-01 8.12223136e-01
4.33865011e-01 -2.15076908e-01 1.22242141e+00 6.02042913e-01
-7.86991119e-01 7.04070807e-01 3.05373549e-01 4.71359402e-01
7.31191754e-01 -1.10288978e+00 3.07219714e-01 2.81025261e-01
2.22324720e-03 5.55443466e-01 3.30796450e-01 -4.66159314e-01
-1.58373523e+00 -1.42695439e+00 1.02699268e+00 -7.31960595e-01
4.34072196e-01 -1.53271437e-01 -1.18612182e+00 9.65194166e-01
-1.22195065e-01 3.71964604e-01 1.28389752e+00 2.69546300e-01
-3.09008062e-01 -7.17531145e-02 -9.69513714e-01 9.79632437e-02
7.50142872e-01 -4.56827909e-01 -7.06101358e-01 1.66588485e-01
7.31595755e-01 -4.25910890e-01 -1.37340844e+00 7.64984548e-01
6.46415651e-01 -4.90212023e-01 1.14448535e+00 -1.20524752e+00
7.93111980e-01 -7.88143352e-02 -2.42457762e-01 -1.36156750e+00
-4.08480287e-01 -3.67429376e-01 -7.85188079e-01 7.97810376e-01
4.47831362e-01 -6.86436534e-01 3.98543924e-01 6.41726732e-01
-8.47216770e-02 -1.09881830e+00 -9.87923145e-01 -8.42135251e-02
-1.54770836e-01 -1.42363623e-01 9.30327475e-01 1.22461236e+00
-2.83905249e-02 2.97674417e-01 -8.19152057e-01 3.00307542e-01
5.89827299e-01 3.63815457e-01 4.75398630e-01 -1.37536311e+00
-4.96399581e-01 1.05968669e-01 -4.60953027e-01 -4.39273447e-01
1.49097443e-01 -8.65392029e-01 -5.32759011e-01 -1.50925004e+00
4.48635638e-01 -1.78738192e-01 -1.19654870e+00 6.43531203e-01
-6.06418431e-01 -1.87142149e-01 -1.25118420e-01 1.80370212e-01
-7.07151473e-01 7.14551866e-01 1.04090834e+00 -2.07439870e-01
-3.29546839e-01 1.32418394e-01 -8.47716451e-01 4.78342742e-01
6.09950423e-01 -6.85166359e-01 -9.76275280e-02 -5.89384198e-01
8.38727430e-02 8.90620232e-01 4.05962199e-01 -8.22389543e-01
3.32191028e-02 1.69777442e-02 6.57209754e-01 -7.54060268e-01
2.21236467e-01 -1.09124649e+00 3.93463850e-01 3.79410267e-01
-5.10256052e-01 4.15020496e-01 6.09550662e-02 9.18612599e-01
-3.56448501e-01 4.97863829e-01 2.10396320e-01 -8.63655508e-02
-1.48810014e-01 7.83111513e-01 8.91502425e-02 2.00144723e-01
7.62747228e-01 3.02599102e-01 -1.62052020e-01 -2.12873623e-01
-1.07394695e+00 4.83490020e-01 3.10407169e-02 4.89287525e-01
5.95827579e-01 -1.46867228e+00 -1.16520512e+00 6.04332209e-01
5.77746749e-01 3.75798047e-02 6.69424593e-01 1.32686412e+00
1.10240966e-01 1.91493735e-01 -1.46676496e-01 -5.21611512e-01
-8.46565008e-01 9.68977869e-01 2.26691023e-01 -4.76762563e-01
-9.28154647e-01 1.63294062e-01 2.89425224e-01 -3.18927109e-01
1.89329073e-01 -3.75172436e-01 -4.36711073e-01 1.09231226e-01
6.25875950e-01 3.08806971e-02 2.11694464e-01 -4.22052354e-01
-3.27310652e-01 4.03406739e-01 -8.03217813e-02 5.07616699e-01
1.82982481e+00 -1.12399094e-01 -7.78297111e-02 7.94171274e-01
1.49106431e+00 -3.88960779e-01 -1.15581262e+00 -3.70762229e-01
-1.04695141e-01 -1.10544331e-01 -1.27841830e-01 -8.44127595e-01
-1.05850971e+00 7.90525377e-01 5.74144423e-01 1.87643208e-02
1.23938942e+00 -8.77999216e-02 8.81095648e-01 1.21706299e-01
-1.58091318e-02 -6.45189703e-01 -4.42197502e-01 3.46951932e-01
8.79378259e-01 -1.61036348e+00 -2.76332647e-01 2.46986896e-01
-7.41689503e-01 8.14837515e-01 3.82598400e-01 7.10565001e-02
9.22149241e-01 3.20115417e-01 -2.39146017e-02 -2.03694344e-01
-1.41669452e+00 2.20239926e-02 3.69314671e-01 3.87261331e-01
7.87000120e-01 2.72518486e-01 -2.42871404e-01 1.14961421e+00
2.06139833e-01 4.70661312e-01 2.27251247e-01 5.53849578e-01
3.25532675e-01 -8.41304183e-01 -2.55786702e-02 9.39232528e-01
-9.49149191e-01 -2.44271681e-01 1.90530524e-01 4.01400119e-01
1.67801499e-01 7.23886967e-01 6.97957873e-02 -2.75600761e-01
5.55750072e-01 3.90492886e-01 3.46249081e-02 -4.38365042e-01
-8.10134113e-01 -8.18038210e-02 -1.57659367e-01 -6.88500464e-01
-2.00405821e-01 -7.19612241e-01 -1.15663230e+00 1.44107133e-01
2.29825810e-01 1.21699236e-01 3.35535973e-01 1.06949532e+00
1.00558209e+00 9.74948883e-01 9.39105034e-01 -3.90332252e-01
-6.30141795e-01 -9.83119667e-01 -4.13693190e-01 8.87721062e-01
9.96657133e-01 -4.44582909e-01 -1.99690893e-01 2.09588781e-01]
|
[7.902946949005127, 6.232944965362549]
|
a9e7769b-1e9b-4da3-a865-2037e888b807
|
fully-convolutional-networks-for-chip-wise
|
1910.02451
| null |
https://arxiv.org/abs/1910.02451v1
|
https://arxiv.org/pdf/1910.02451v1.pdf
|
Fully Convolutional Networks for Chip-wise Defect Detection Employing Photoluminescence Images
|
Efficient quality control is inevitable in the manufacturing of light-emitting diodes (LEDs). Because defective LED chips may be traced back to different causes, a time and cost-intensive electrical and optical contact measurement is employed. Fast photoluminescence measurements, on the other hand, are commonly used to detect wafer separation damages but also hold the potential to enable an efficient detection of all kinds of defective LED chips. On a photoluminescence image, every pixel corresponds to an LED chip's brightness after photoexcitation, revealing performance information. But due to unevenly distributed brightness values and varying defect patterns, photoluminescence images are not yet employed for a comprehensive defect detection. In this work, we show that fully convolutional networks can be used for chip-wise defect detection, trained on a small data-set of photoluminescence images. Pixel-wise labels allow us to classify each and every chip as defective or not. Being measurement-based, labels are easy to procure and our experiments show that existing discrepancies between training images and labels do not hinder network training. Using weighted loss calculation, we were able to equalize our highly unbalanced class categories. Due to the consistent use of skip connections and residual shortcuts, our network is able to predict a variety of structures, from extensive defect clusters up to single defective LED chips.
|
['Maike Lorena Stern', 'Martin Schellenberger']
|
2019-10-06
| null | null | null | null |
['small-data']
|
['computer-vision']
|
[ 3.95975560e-01 -1.13109998e-01 2.36891404e-01 -3.76425922e-01
-6.21521473e-01 -4.52373207e-01 2.43420482e-01 9.45225433e-02
-1.29022583e-01 7.60796905e-01 -6.85882032e-01 -1.17535107e-01
8.87217745e-02 -9.52665031e-01 -5.54509103e-01 -9.76052999e-01
4.91618723e-01 6.57574177e-01 1.11326225e-01 7.30557218e-02
2.77877569e-01 9.32477951e-01 -1.80853868e+00 2.75002986e-01
7.82573342e-01 1.19999802e+00 4.53980654e-01 5.56046546e-01
-1.13059290e-01 2.87522763e-01 -7.49048769e-01 -1.01354390e-01
4.80387136e-02 -4.52869207e-01 -3.09612453e-01 1.45051122e-01
3.83005708e-01 -3.19128603e-01 -1.34037569e-01 1.03316009e+00
3.76546830e-01 -4.14365083e-01 9.92853820e-01 -1.01991034e+00
-3.80260646e-01 9.50311646e-02 -2.95531154e-01 -2.57235050e-01
2.97720343e-01 4.28817570e-01 8.09063256e-01 -7.36101329e-01
5.95638633e-01 8.19722354e-01 5.57776868e-01 6.43667459e-01
-1.22350848e+00 -5.23769200e-01 -6.42904937e-01 1.93046436e-01
-1.01456034e+00 -3.64387125e-01 1.09316635e+00 -3.64127010e-01
8.92040908e-01 2.15876773e-01 3.92858356e-01 9.37028766e-01
4.08794224e-01 3.66074860e-01 1.29619229e+00 -6.77059710e-01
3.00355166e-01 1.96746051e-01 2.25891452e-02 9.23349440e-01
4.87521142e-01 2.93936610e-01 -5.87342791e-02 1.68623552e-01
4.05332029e-01 1.69338107e-01 -3.09480190e-01 -3.41518998e-01
-7.71279931e-01 3.53609860e-01 4.80724722e-01 6.06081426e-01
-1.02843091e-01 1.19033799e-01 3.11891019e-01 5.31209826e-01
4.05614406e-01 6.27158284e-01 -2.42407143e-01 4.96416502e-02
-9.47664142e-01 -4.08237249e-01 6.12936795e-01 4.03202385e-01
1.23177290e+00 -1.49682865e-01 -1.38643086e-01 9.93084669e-01
5.28806709e-02 4.64380115e-01 1.24472827e-01 -7.56795585e-01
-9.59162414e-03 9.32629108e-01 1.19663946e-01 -7.40739286e-01
-5.01522005e-01 -5.96674643e-02 -9.66578364e-01 9.54937160e-01
4.01495785e-01 2.31981520e-02 -1.20198691e+00 9.91409302e-01
-2.10308254e-01 -3.66369814e-01 -1.76202908e-01 6.84622407e-01
4.92963821e-01 3.46811861e-01 -3.26909363e-01 -2.11787894e-01
9.76089239e-01 -3.95669729e-01 -4.99928355e-01 -1.27462715e-01
5.95012188e-01 -6.97927833e-01 1.15214610e+00 6.47456288e-01
-1.00721538e+00 -6.09845281e-01 -1.49358594e+00 1.10145770e-01
-5.57166576e-01 5.62690496e-01 4.49464560e-01 1.03044224e+00
-9.91381466e-01 1.08275485e+00 -8.68867576e-01 -1.51178136e-01
8.53230000e-01 5.66176832e-01 -4.70624626e-01 -6.32935286e-01
-4.23460871e-01 6.96308494e-01 9.21518914e-03 1.08023211e-01
-7.61566222e-01 -5.88902056e-01 -4.44512695e-01 6.49114549e-02
1.67647541e-01 -2.03884691e-01 5.45230329e-01 -8.24792504e-01
-1.54900312e+00 1.22848117e+00 -1.20084241e-01 -1.07959017e-01
4.99339163e-01 3.62903684e-01 -4.94794697e-01 1.83359951e-01
-5.16735055e-02 3.76660317e-01 1.02783537e+00 -1.59698284e+00
-3.40633512e-01 -4.29653406e-01 -2.51961708e-01 -5.55328250e-01
-2.67254084e-01 -1.44363865e-01 -2.90284097e-01 5.88243231e-02
3.05572867e-01 -6.24429345e-01 1.32968485e-01 -7.21514225e-02
-7.41030157e-01 -2.92232901e-01 9.64846194e-01 -3.33896577e-01
7.04086959e-01 -2.17452216e+00 -3.70802104e-01 3.21852177e-01
1.41921639e-01 2.88135707e-01 -8.52452144e-02 3.67369533e-01
-1.44103363e-01 1.71127152e-02 -4.87743884e-01 -4.40449655e-01
-3.31508555e-02 7.85345659e-02 2.23593503e-01 6.74034953e-01
5.24526536e-01 8.53368461e-01 -4.22923714e-01 -1.42296255e-01
4.49596822e-01 3.35620940e-01 -7.19700158e-02 1.57646194e-01
-1.97962061e-01 3.31262022e-01 -1.53866455e-01 1.12587512e+00
9.45453346e-01 1.57189518e-02 1.78707354e-02 -4.20272946e-01
-6.75765947e-02 -2.01413035e-02 -5.73226035e-01 1.47561252e+00
-6.32698596e-01 7.83102036e-01 1.51651084e-01 -1.05363309e+00
1.34634018e+00 1.23071507e-01 4.24511939e-01 -1.13568807e+00
2.30045423e-01 5.90599954e-01 -1.19274028e-01 -4.68677074e-01
2.53510028e-01 -4.27044630e-01 -6.56346744e-03 4.65233743e-01
-1.44896522e-01 -5.47039390e-01 2.37355769e-01 -2.71019638e-01
1.43015766e+00 -1.14865735e-01 -4.52729106e-01 -2.35166416e-01
4.25460309e-01 -1.21973395e-01 3.86979640e-01 6.24103665e-01
1.23498559e-01 9.29608643e-01 6.98357403e-01 -3.63232255e-01
-1.31017423e+00 -9.38577414e-01 -4.55138564e-01 4.39036846e-01
1.01872303e-01 2.73849934e-01 -6.94323540e-01 -7.15643406e-01
1.56160042e-01 5.03779292e-01 -4.31937039e-01 -1.95913345e-01
-2.42812619e-01 -8.10270071e-01 3.43974888e-01 2.11051553e-01
3.70160371e-01 -1.22177947e+00 -5.56239605e-01 2.25068584e-01
5.16556323e-01 -7.71690488e-01 3.06160301e-01 5.53956807e-01
-8.06730270e-01 -1.54648757e+00 -6.00076258e-01 -6.43559039e-01
1.06447673e+00 1.79552101e-02 1.35856974e+00 3.28428566e-01
-1.23225164e+00 3.32638830e-01 -1.23197034e-01 -2.08616152e-01
-8.48843515e-01 -2.07026765e-01 -1.43306509e-01 -2.09144399e-01
3.84508759e-01 -5.46134710e-01 -6.40853763e-01 3.57908398e-01
-8.67811024e-01 -3.20394695e-01 8.60386550e-01 8.20267618e-01
4.96937573e-01 5.97168863e-01 6.00925863e-01 -1.15793490e+00
5.20747542e-01 4.49130721e-02 -7.42192149e-01 1.29200533e-01
-6.94938898e-01 -1.27433717e-01 6.77793205e-01 1.54598415e-01
-1.02014673e+00 4.37523425e-02 -4.51923937e-01 -3.09577644e-01
-7.29781568e-01 7.59830773e-02 -2.54657716e-01 -2.71898866e-01
7.36235142e-01 -8.89091268e-02 1.02211997e-01 -3.37876022e-01
-1.72842160e-01 6.98852301e-01 3.04343909e-01 -3.04930687e-01
6.22719824e-01 7.60239184e-01 2.37304986e-01 -9.73553956e-01
-3.58712703e-01 -3.80649030e-01 -6.46245539e-01 -2.52968550e-01
6.39926016e-01 -6.52751446e-01 -9.64562476e-01 7.76825607e-01
-1.02588725e+00 -3.22573483e-01 -3.46864522e-01 2.98854746e-02
-2.68987477e-01 4.24719661e-01 -7.10564077e-01 -9.23314750e-01
-7.80179575e-02 -1.04513085e+00 1.11178792e+00 1.23736739e-01
1.26773164e-01 -9.30333257e-01 -3.07751507e-01 2.83363402e-01
2.69593239e-01 3.87926668e-01 1.44697845e+00 9.40660536e-02
-6.13398552e-01 -7.51312554e-01 -4.88933623e-01 8.89953971e-01
6.46239877e-01 1.57032892e-01 -1.30377150e+00 -3.17927808e-01
3.07101719e-02 -4.44064826e-01 1.37265944e+00 4.39027876e-01
1.27600455e+00 5.12436390e-01 -5.75677514e-01 2.11255446e-01
1.69719660e+00 2.23085999e-01 1.10570300e+00 -6.32501245e-02
6.89546108e-01 6.52912259e-01 4.47117895e-01 8.42738617e-03
-4.27409232e-01 4.99000162e-01 7.36019611e-01 -2.71125287e-01
-2.34545246e-01 2.28338093e-01 2.78689563e-01 2.09707871e-01
-5.74042983e-02 -3.68802220e-01 -6.85810268e-01 4.26122844e-01
-9.78930175e-01 -6.83705330e-01 -5.03237307e-01 2.35604405e+00
5.24661481e-01 3.70504469e-01 -2.55573541e-01 4.48066831e-01
8.65166545e-01 -1.21007048e-01 -6.65786207e-01 -2.89617717e-01
-1.84612602e-01 4.42509979e-01 5.51066279e-01 1.14760138e-01
-7.27230430e-01 5.41559398e-01 6.05584288e+00 8.02976251e-01
-1.20859468e+00 -1.04734182e-01 7.33990610e-01 9.79174376e-02
-4.48834896e-01 -2.88546592e-01 -6.15886867e-01 6.82583570e-01
5.45349717e-01 5.05279064e-01 1.99675083e-01 4.57221299e-01
-3.90980057e-02 -6.42311871e-01 -1.26789474e+00 1.23673153e+00
-2.98688002e-02 -1.28761983e+00 -4.47088301e-01 6.45186380e-02
8.42345297e-01 5.27596697e-02 -1.08127538e-02 -1.75720826e-01
-1.29072905e-01 -1.12554312e+00 2.83290863e-01 5.24615943e-01
1.19500053e+00 -7.31369436e-01 7.11737514e-01 4.91711963e-03
-6.44530535e-01 -3.08841646e-01 -8.28123808e-01 1.28254145e-01
-1.34326249e-01 1.39161301e+00 -7.78854370e-01 4.64804709e-01
6.28450394e-01 6.63299680e-01 -3.26985598e-01 8.58060181e-01
-1.73544541e-01 3.71068716e-01 -1.87067583e-01 7.58906547e-03
-3.05665843e-02 -3.92878532e-01 2.10367553e-02 9.66812134e-01
7.22084820e-01 -4.68139082e-01 -3.26445013e-01 1.34102130e+00
-3.13533217e-01 -4.28323984e-01 -7.29668021e-01 -1.85696706e-02
4.96617407e-02 1.42965043e+00 -1.11912775e+00 1.06957696e-01
-3.24141353e-01 1.17587984e+00 -1.02915332e-01 2.16600284e-01
-3.13994795e-01 -5.43448210e-01 7.15173364e-01 2.50218093e-01
2.21160084e-01 -1.59245118e-01 -3.53241295e-01 -6.29136324e-01
6.59634843e-02 -2.14427114e-01 -1.18441455e-01 -6.50691807e-01
-1.38927114e+00 4.49707627e-01 -7.00592875e-01 -9.73297536e-01
2.38897607e-01 -1.20231712e+00 -9.22396302e-01 8.22851777e-01
-1.92695105e+00 -7.65943170e-01 -4.40728396e-01 1.76160306e-01
3.02475572e-01 2.85968129e-02 7.61995316e-01 3.70226860e-01
-5.53131342e-01 1.95206761e-01 3.68910789e-01 -1.64220870e-01
6.75976217e-01 -1.30104828e+00 1.13412298e-01 5.17572165e-01
2.71403380e-02 5.76335788e-02 5.94105065e-01 -5.25904655e-01
-1.43475187e+00 -9.66302156e-01 3.97965819e-01 -2.93847978e-01
1.91809446e-01 -3.47796559e-01 -9.30806041e-01 6.80410266e-02
1.22913152e-01 9.70228985e-02 3.25510859e-01 -5.39544821e-02
-1.44586591e-02 -2.62076825e-01 -1.38387728e+00 7.62945786e-02
8.40802968e-01 -6.79939926e-01 -6.60484061e-02 5.82571447e-01
-9.40675009e-03 -1.60339832e-01 -5.34959018e-01 4.40476507e-01
3.69532257e-01 -1.60242569e+00 6.57750607e-01 1.54040247e-01
6.57412052e-01 -1.10836029e-01 2.48228014e-01 -1.28140903e+00
-5.57608269e-02 -5.86037077e-02 6.18531585e-01 1.37298083e+00
5.10867536e-01 -6.45436466e-01 1.26354849e+00 4.44204330e-01
-4.62830067e-01 -5.28448462e-01 -9.40463960e-01 -9.16151941e-01
-1.16184734e-01 -5.41650772e-01 4.20674950e-01 5.76750517e-01
-3.73949796e-01 -1.95224937e-02 -6.65361956e-02 7.89108425e-02
5.22266567e-01 2.43499979e-01 2.34468460e-01 -1.75198722e+00
-9.85592976e-02 -4.97703552e-01 -3.90290380e-01 -4.13035482e-01
3.85963917e-01 -8.80207241e-01 2.86551833e-01 -1.65708864e+00
-2.38357158e-03 -9.75504100e-01 -3.63564402e-01 3.97032320e-01
2.43263051e-01 8.12455416e-01 -3.21118712e-01 4.82469797e-02
-2.62365818e-01 2.61670947e-01 1.00687003e+00 -5.77149510e-01
-1.08955383e-01 2.05869734e-01 -1.98703185e-01 4.94507939e-01
7.79665828e-01 -4.13772613e-01 -6.99240565e-02 -2.14926809e-01
5.58280408e-01 -2.58068353e-01 5.23601174e-01 -1.48841429e+00
1.59100275e-02 4.67816383e-01 6.10915005e-01 -4.17651087e-01
4.65100944e-01 -1.03534245e+00 3.79534476e-02 5.49396038e-01
1.76980183e-01 -8.28647137e-01 1.16560370e-01 4.84180003e-01
-2.11584568e-01 -6.97208285e-01 8.82011294e-01 -2.78940886e-01
-6.44581318e-01 9.65254009e-02 -3.17226619e-01 -6.78360045e-01
9.36787605e-01 -5.34579873e-01 -4.30745572e-01 9.17424485e-02
-3.05174351e-01 -2.15991631e-01 9.76537287e-01 -1.95494052e-02
7.05286682e-01 -1.03842139e+00 -2.61086702e-01 5.34419358e-01
3.55212003e-01 -2.06295699e-01 6.34095371e-01 5.25735736e-01
-5.54547369e-01 2.54895061e-01 -3.97803009e-01 -7.75156796e-01
-1.29455721e+00 2.50129640e-01 3.66294980e-01 -1.43981531e-01
-4.94195640e-01 8.67289901e-01 -1.42043680e-01 -2.70224810e-01
-2.14849666e-01 -3.62754822e-01 -5.93112931e-02 4.24763421e-04
1.64977655e-01 4.03902233e-01 6.96814537e-01 -1.66893423e-01
-2.23054081e-01 7.47573912e-01 1.23377331e-01 6.19384408e-01
1.36369240e+00 3.98596972e-02 -3.80352914e-01 3.77951801e-01
1.42914772e+00 -1.33239299e-01 -1.49062407e+00 3.81959766e-01
-2.30098758e-02 -4.39795583e-01 1.23043254e-01 -8.50544035e-01
-1.28595793e+00 1.10101271e+00 8.99777234e-01 6.85756624e-01
1.33771122e+00 1.27311796e-01 5.31437695e-01 4.12269473e-01
5.34196913e-01 -1.41737783e+00 9.27507281e-02 -1.87941047e-03
5.51417232e-01 -1.39172804e+00 -1.42858878e-01 -6.24280214e-01
-1.77109957e-01 1.36315691e+00 3.07167619e-01 -4.91762608e-02
2.38978669e-01 5.64484477e-01 2.51332857e-02 -4.99850184e-01
-3.63803893e-01 -7.63671324e-02 -9.94984731e-02 8.40415597e-01
3.58486772e-01 3.40756625e-02 3.77880409e-02 4.57695127e-02
3.21961612e-01 -2.04561278e-02 5.07089257e-01 7.13811457e-01
-5.54798424e-01 -1.57157600e+00 -3.54637414e-01 1.06131518e+00
-2.87714094e-01 2.73030668e-01 -2.31637925e-01 6.06434286e-01
2.26970762e-01 1.04793513e+00 2.59390026e-01 -1.96321040e-01
3.79410297e-01 2.00641662e-01 7.49406338e-01 -6.52420402e-01
-4.84964922e-02 -3.38867337e-01 -1.47027895e-01 -6.21358633e-01
-4.37489152e-01 -2.84881115e-01 -1.09967875e+00 -2.73044594e-02
-3.92881483e-01 -1.23890072e-01 1.08823287e+00 9.72764850e-01
1.62662387e-01 8.14701259e-01 9.06435370e-01 -8.79239619e-01
-6.11912981e-02 -7.34137356e-01 -1.27902484e+00 3.79954129e-01
2.17424616e-01 -5.39381921e-01 -5.32558978e-01 -1.45837232e-01]
|
[7.295166969299316, 1.9130761623382568]
|
f4956b97-fa0a-49bc-9d8a-e224f24f1834
|
idea-net-dynamic-3d-point-cloud-interpolation
|
2203.1159
| null |
https://arxiv.org/abs/2203.11590v1
|
https://arxiv.org/pdf/2203.11590v1.pdf
|
IDEA-Net: Dynamic 3D Point Cloud Interpolation via Deep Embedding Alignment
|
This paper investigates the problem of temporally interpolating dynamic 3D point clouds with large non-rigid deformation. We formulate the problem as estimation of point-wise trajectories (i.e., smooth curves) and further reason that temporal irregularity and under-sampling are two major challenges. To tackle the challenges, we propose IDEA-Net, an end-to-end deep learning framework, which disentangles the problem under the assistance of the explicitly learned temporal consistency. Specifically, we propose a temporal consistency learning module to align two consecutive point cloud frames point-wisely, based on which we can employ linear interpolation to obtain coarse trajectories/in-between frames. To compensate the high-order nonlinear components of trajectories, we apply aligned feature embeddings that encode local geometry properties to regress point-wise increments, which are combined with the coarse estimations. We demonstrate the effectiveness of our method on various point cloud sequences and observe large improvement over state-of-the-art methods both quantitatively and visually. Our framework can bring benefits to 3D motion data acquisition. The source code is publicly available at https://github.com/ZENGYIMING-EAMON/IDEA-Net.git.
|
['Ying He', 'Yixuan Yuan', 'Junhui Hou', 'Qijian Zhang', 'Yue Qian', 'Yiming Zeng']
|
2022-03-22
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Zeng_IDEA-Net_Dynamic_3D_Point_Cloud_Interpolation_via_Deep_Embedding_Alignment_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Zeng_IDEA-Net_Dynamic_3D_Point_Cloud_Interpolation_via_Deep_Embedding_Alignment_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['3d-point-cloud-interpolation']
|
['computer-vision']
|
[-3.87829095e-01 -2.79612273e-01 -9.77202412e-03 -1.22956261e-01
-8.76348138e-01 -4.88873124e-01 5.40507555e-01 -1.21645540e-01
-1.11339398e-01 4.30137485e-01 1.58820972e-02 -2.16765907e-02
-8.64050463e-02 -5.68945169e-01 -1.11612141e+00 -6.05044305e-01
-3.07927728e-01 4.00667638e-01 2.01254562e-01 -1.60522357e-01
9.98811051e-02 7.21427023e-01 -1.25577784e+00 -7.61887133e-02
9.85030532e-01 8.58273208e-01 1.30023479e-01 6.03375614e-01
2.17497006e-01 3.85734051e-01 1.50895998e-01 -7.32258782e-02
4.73867863e-01 2.78338846e-02 -5.64679861e-01 2.15658575e-01
6.91600561e-01 -7.28132010e-01 -6.07739270e-01 8.77667964e-01
2.72267610e-01 9.65935066e-02 4.17209595e-01 -1.24978375e+00
-6.97659552e-01 -1.91720128e-01 -7.17253983e-01 1.22782290e-01
2.24535137e-01 4.57941055e-01 7.73893118e-01 -1.28592730e+00
7.50422776e-01 1.16163254e+00 9.30569172e-01 4.15109426e-01
-1.21016645e+00 -4.98765290e-01 1.30773142e-01 1.24443628e-01
-1.33270359e+00 -3.93958032e-01 1.12811947e+00 -8.51328135e-01
7.02392697e-01 4.24312949e-02 7.07160294e-01 8.00056875e-01
1.92603916e-01 5.99048376e-01 6.66761100e-01 9.38237086e-02
1.92638487e-02 -4.28272545e-01 -2.03734934e-01 6.22786462e-01
1.38211865e-02 5.82628071e-01 -2.09020004e-01 -1.48256585e-01
1.34085357e+00 4.52669978e-01 -3.29675764e-01 -6.17549241e-01
-1.58040869e+00 4.36221421e-01 6.32727444e-01 5.59791643e-03
-4.52750653e-01 4.61061567e-01 2.22293645e-01 8.75702724e-02
8.92870724e-01 -1.01516381e-01 -5.40592134e-01 -1.60735279e-01
-9.01536345e-01 5.95468700e-01 2.35738114e-01 1.15422916e+00
8.72303367e-01 6.15199730e-02 3.80944684e-02 5.81448436e-01
4.89992946e-01 5.02200961e-01 3.25405747e-02 -1.17845905e+00
5.63860595e-01 4.37103182e-01 4.07661945e-01 -9.42326844e-01
-2.61626929e-01 -5.70732728e-02 -8.62436712e-01 5.02402723e-01
4.33859229e-01 -2.36839447e-02 -8.38177502e-01 1.58346641e+00
6.72878087e-01 9.93226707e-01 -2.60785311e-01 1.20988369e+00
4.52063680e-01 6.68380201e-01 -1.09570891e-01 -1.79372832e-01
1.01672316e+00 -9.88463581e-01 -5.01146257e-01 1.75943807e-01
4.44236904e-01 -6.90274060e-01 9.83395696e-01 -1.35099307e-01
-1.28415084e+00 -7.25290239e-01 -8.20941448e-01 -3.69291693e-01
8.10176954e-02 6.40080497e-02 2.57071584e-01 -4.23942983e-01
-9.59348440e-01 1.19479251e+00 -1.51513267e+00 -2.30150726e-02
5.63339174e-01 1.60047039e-01 -3.63520235e-01 -1.52734471e-02
-8.16785872e-01 5.65242767e-01 -1.11538351e-01 3.12812150e-01
-7.44564116e-01 -1.19448090e+00 -8.30136657e-01 -2.30794385e-01
1.50838435e-01 -8.46963108e-01 1.09282899e+00 -4.05254304e-01
-1.39412117e+00 7.51899302e-01 -4.91669089e-01 -2.35986426e-01
9.58163977e-01 -5.07053614e-01 -6.84745014e-02 -2.40609571e-02
1.62305355e-01 4.92766649e-01 6.95369482e-01 -1.33532262e+00
-3.51356357e-01 -2.60571659e-01 -3.53093237e-01 1.09549299e-01
1.74905211e-01 -3.69793504e-01 -6.24274909e-01 -6.69275403e-01
2.88140088e-01 -1.24203777e+00 -3.58606189e-01 5.33737004e-01
-2.42087558e-01 -2.54499972e-01 9.70160007e-01 -6.71934366e-01
9.52327251e-01 -2.31095004e+00 3.08312982e-01 -2.08728835e-01
4.30084765e-01 4.48784977e-02 1.73091125e-02 2.89488256e-01
-4.17407043e-02 -1.27197308e-02 -5.55992544e-01 -5.99270642e-01
-9.63473469e-02 1.49594739e-01 -4.17675495e-01 8.35901558e-01
5.75897574e-01 1.05548692e+00 -1.10765028e+00 -1.97167844e-01
5.56227446e-01 6.56429887e-01 -4.01658803e-01 2.93301761e-01
-3.07819545e-01 9.39538240e-01 -6.31096065e-01 7.37908006e-01
1.10245359e+00 -3.62352759e-01 -3.68658453e-01 -3.81117284e-01
-4.53979552e-01 2.73063421e-01 -1.00346613e+00 2.12592769e+00
-2.83357769e-01 6.57474458e-01 -8.71143267e-02 -6.24329925e-01
7.32842267e-01 3.80510032e-01 8.53386879e-01 -2.39890516e-01
-3.22800763e-02 4.30115938e-01 -2.79935449e-01 -5.82223594e-01
4.53590691e-01 2.93559041e-02 2.03913167e-01 4.11102436e-02
-2.94241369e-01 -2.80225843e-01 -3.52688491e-01 -7.64354765e-02
8.35587561e-01 7.57925034e-01 2.25603333e-04 -1.72453851e-01
4.40563083e-01 1.18408486e-01 7.20630169e-01 4.46643084e-02
-3.84351581e-01 1.00348783e+00 2.27613777e-01 -7.29645014e-01
-1.45431542e+00 -1.11827040e+00 -1.97293118e-01 3.52494687e-01
5.41438520e-01 -2.09706306e-01 -3.48321289e-01 -3.72502029e-01
4.67559606e-01 2.13633597e-01 -5.43518066e-01 2.16540441e-01
-1.03204811e+00 -1.94685981e-01 1.20865047e-01 7.12874711e-01
2.11177692e-01 -6.74919069e-01 -5.39371490e-01 4.03537512e-01
1.31287007e-02 -1.21154666e+00 -8.13098788e-01 -3.50681543e-01
-1.32940102e+00 -9.49500918e-01 -7.57557154e-01 -6.39550149e-01
6.71664596e-01 4.25966591e-01 1.08974946e+00 1.88810796e-01
2.44379155e-02 2.48588145e-01 -3.81038338e-02 -5.85983805e-02
-1.24068297e-01 -1.53703377e-01 1.47799313e-01 -1.58003598e-01
7.40836933e-02 -1.03372562e+00 -9.44536567e-01 3.83797824e-01
-6.73414767e-01 2.94358313e-01 2.04960465e-01 6.19089901e-01
1.09987175e+00 -4.41491336e-01 1.56546742e-01 -3.81085813e-01
2.16960236e-01 -6.32023335e-01 -7.95073450e-01 -1.15464658e-01
-2.61455446e-01 2.16648225e-02 3.05096626e-01 -5.07492244e-01
-5.95173478e-01 2.87420601e-01 -2.32092842e-01 -1.34746993e+00
-1.15396574e-01 2.17825904e-01 1.04208931e-01 -1.57348067e-01
2.30501607e-01 1.25690088e-01 4.75401059e-02 -4.75822240e-01
4.50905949e-01 1.69485033e-01 7.00449705e-01 -8.33985388e-01
1.11607158e+00 8.20365071e-01 1.49203211e-01 -5.45029521e-01
-4.70119923e-01 -5.27588665e-01 -9.42018092e-01 -3.79756808e-01
9.32765007e-01 -1.24130023e+00 -7.04219282e-01 5.38904548e-01
-1.52946877e+00 -5.42552233e-01 -2.77991712e-01 4.79229867e-01
-8.03537786e-01 4.73972023e-01 -8.40759754e-01 -6.59512341e-01
-2.79936373e-01 -1.17712724e+00 1.49588251e+00 2.35959738e-02
-4.36885804e-02 -9.74827349e-01 2.86295801e-01 -1.40893891e-01
8.98441523e-02 9.65848148e-01 4.57499236e-01 3.68017145e-02
-1.15507042e+00 2.94740056e-03 -2.00594157e-01 9.97101367e-02
3.64260316e-01 3.13882262e-01 -8.60560834e-01 -3.13026011e-01
1.17266728e-02 3.44808772e-02 6.27435684e-01 4.83173102e-01
1.12508869e+00 -2.44071946e-01 -3.54771137e-01 1.10428131e+00
1.50840282e+00 -1.43192440e-01 5.94820619e-01 2.35418305e-01
1.24166405e+00 3.35023344e-01 7.17918038e-01 4.44698364e-01
7.00435579e-01 8.83249819e-01 5.59176683e-01 8.18334296e-02
-1.01093732e-01 -3.87430638e-01 1.91226378e-01 1.12701702e+00
-5.11327386e-01 1.28033966e-01 -9.55569744e-01 8.08665633e-01
-2.13327599e+00 -8.25985909e-01 -3.86071175e-01 2.07421327e+00
7.78086364e-01 -6.74584135e-02 4.16793860e-02 -1.91174179e-01
5.85266232e-01 3.56332511e-01 -7.71628857e-01 1.02611810e-01
1.94109872e-01 -9.60171372e-02 4.33985889e-01 7.00987875e-01
-1.07416916e+00 8.60885262e-01 4.54598618e+00 5.40297031e-01
-1.39136684e+00 1.78239375e-01 3.49340945e-01 -1.59966469e-01
-4.43909764e-01 5.50717935e-02 -5.22735298e-01 7.54124284e-01
7.31533647e-01 -2.15061635e-01 3.37748051e-01 7.62962759e-01
5.36441088e-01 4.75940347e-01 -1.32060528e+00 9.12328124e-01
-3.83772314e-01 -1.71779203e+00 -1.79345861e-01 1.64556921e-01
8.68785739e-01 4.87230450e-01 -2.10040081e-02 -1.34567628e-02
1.45803809e-01 -7.12150395e-01 1.05413818e+00 7.80733466e-01
8.98159087e-01 -4.65702653e-01 2.96729892e-01 3.76316249e-01
-1.64598143e+00 4.28188771e-01 -2.97141135e-01 -3.52199897e-02
6.23134315e-01 6.56033874e-01 -2.96601087e-01 8.76134515e-01
7.37271726e-01 1.36769688e+00 -8.10580626e-02 1.10625553e+00
-2.86377873e-02 3.06077540e-01 -4.61686760e-01 6.05505347e-01
7.65417889e-02 -5.90811610e-01 8.48147213e-01 9.23634827e-01
7.35225976e-01 2.75375634e-01 1.84795141e-01 1.18018293e+00
-9.16461572e-02 -4.14819032e-01 -6.28933549e-01 2.58423477e-01
6.62875712e-01 9.55422103e-01 -3.02869916e-01 -2.67629832e-01
-4.64476407e-01 1.06163037e+00 4.52816516e-01 4.17750239e-01
-8.26795161e-01 7.84755647e-02 1.18045843e+00 3.84248078e-01
3.13485950e-01 -8.52015793e-01 -4.46634859e-01 -1.31657767e+00
5.30159175e-01 -2.54601300e-01 -4.76262979e-02 -8.23968291e-01
-1.37496412e+00 4.59942579e-01 -4.20987718e-02 -1.95643413e+00
-3.95381078e-02 -2.44227603e-01 -8.52590501e-01 1.02020419e+00
-1.81498015e+00 -1.12874019e+00 -5.56485832e-01 4.81426746e-01
6.61313295e-01 3.84365439e-01 2.64492571e-01 4.51083809e-01
-3.56533945e-01 2.80461222e-01 1.74205616e-01 1.21390313e-01
6.16422236e-01 -1.00456572e+00 9.10057485e-01 8.28462958e-01
-3.94113809e-02 5.52786052e-01 5.10571837e-01 -7.40379393e-01
-1.52914512e+00 -1.37728822e+00 8.49928737e-01 -8.25296104e-01
6.70165777e-01 -1.95753887e-01 -1.42962241e+00 6.98722124e-01
-1.55679733e-01 7.21102774e-01 -2.93037035e-02 -5.45736909e-01
-1.35474026e-01 1.45521788e-02 -9.92244959e-01 4.67850447e-01
1.24586368e+00 -5.88098049e-01 -3.78208280e-01 1.80525333e-01
1.29348910e+00 -1.14533019e+00 -1.11779869e+00 5.90600073e-01
3.77674341e-01 -7.95091629e-01 1.20011580e+00 -3.47639561e-01
8.22501421e-01 -6.38691843e-01 -3.13252397e-02 -1.10890853e+00
-3.52198690e-01 -7.24605501e-01 -5.11460304e-01 9.49799180e-01
2.30421405e-02 -4.87864137e-01 7.82343984e-01 6.63757741e-01
-4.98338401e-01 -1.07905400e+00 -1.12710607e+00 -9.48291719e-01
3.46252739e-01 -4.63943630e-01 9.19328153e-01 1.16867638e+00
-3.56169879e-01 -2.98198223e-01 -4.15423959e-01 5.51568627e-01
7.88459420e-01 2.25672796e-01 8.56638253e-01 -1.10536766e+00
-4.23099920e-02 -3.22974056e-01 -3.99819255e-01 -1.41033268e+00
5.69343455e-02 -6.86430991e-01 1.42885134e-01 -1.33333635e+00
-4.04973328e-01 -7.09560573e-01 -1.47534326e-01 1.32385373e-01
-2.87770241e-01 6.99679134e-03 1.96843013e-01 6.84796154e-01
-1.63864836e-01 8.94689500e-01 1.54532778e+00 4.40223068e-02
-3.64067793e-01 -4.26151082e-02 -6.61621168e-02 6.63952529e-01
5.48311293e-01 -4.68907326e-01 -1.97475672e-01 -9.20372546e-01
-7.71014169e-02 3.76748621e-01 8.59794736e-01 -9.37793195e-01
2.04100445e-01 -4.88492489e-01 1.66421786e-01 -8.76471281e-01
3.83959919e-01 -9.92967725e-01 4.21242028e-01 1.66548133e-01
-2.03668177e-01 3.77650619e-01 4.13437873e-01 8.10322940e-01
-1.05119452e-01 3.32841396e-01 6.70578837e-01 1.01452529e-01
-6.00364208e-01 1.04755116e+00 3.42921972e-01 -6.03028573e-02
9.50395584e-01 -1.45213589e-01 -5.87242655e-02 4.49888268e-03
-5.42710900e-01 3.53013635e-01 9.12447751e-01 4.43829209e-01
7.96356320e-01 -1.76020563e+00 -7.46982574e-01 2.38374636e-01
1.69556513e-01 7.38461137e-01 4.26902562e-01 1.03867400e+00
-6.97797060e-01 4.82430421e-02 -1.70630664e-02 -1.11916304e+00
-8.19443047e-01 4.03700262e-01 3.50368738e-01 2.66133063e-02
-1.18391573e+00 7.06147492e-01 5.68476394e-02 -3.99123520e-01
-7.46719493e-03 -8.02994609e-01 3.44168007e-01 -3.92659903e-01
1.67050734e-01 3.94172788e-01 -7.94962645e-02 -6.98676169e-01
-4.48928773e-01 1.13089049e+00 1.84058562e-01 5.72067201e-02
1.49698114e+00 -1.64055705e-01 7.83554763e-02 4.24188346e-01
1.45428288e+00 -6.13765791e-02 -2.05030704e+00 -2.67139792e-01
-7.83362761e-02 -7.90959179e-01 -1.14785433e-01 -8.06040540e-02
-1.18417168e+00 9.44334209e-01 5.71468234e-01 -1.08745478e-01
7.83001602e-01 -3.99411935e-03 1.24005044e+00 -1.05221488e-01
2.79306144e-01 -5.09801209e-01 -1.89973205e-01 4.61962521e-01
1.02621710e+00 -1.43496323e+00 8.22339728e-02 -4.81118947e-01
-4.03521985e-01 1.17217779e+00 4.53736275e-01 -7.17887044e-01
7.98466682e-01 7.90689439e-02 -7.87678808e-02 -3.12303007e-01
-7.93654740e-01 3.12353987e-02 4.66894835e-01 4.42754060e-01
3.06606770e-01 1.29542172e-01 -4.80721928e-02 2.16749683e-01
3.52670252e-02 3.47126186e-01 2.96223104e-01 8.16076875e-01
-6.77949414e-02 -9.57108438e-01 -1.91821441e-01 1.45550705e-02
3.50435227e-02 1.58185825e-01 1.83307350e-01 7.22232759e-01
1.56656906e-01 3.36054713e-01 2.52052724e-01 -5.29650152e-01
5.38808525e-01 -2.69033998e-01 3.57477874e-01 -2.75744736e-01
-6.90607950e-02 1.38458222e-01 -1.89528927e-01 -8.21655512e-01
-5.42993605e-01 -9.18545306e-01 -1.36954045e+00 -5.03882110e-01
1.04939528e-01 -2.17411712e-01 4.38806087e-01 8.01384449e-01
7.13962555e-01 3.52048516e-01 7.73413837e-01 -1.53055942e+00
-4.58593577e-01 -6.10885382e-01 -1.79711312e-01 7.06810474e-01
9.56839204e-01 -7.24060178e-01 -5.25380254e-01 3.61277401e-01]
|
[8.536781311035156, -2.155642032623291]
|
0d89d680-e387-443d-bc7e-aebc6dd6d632
|
creating-pro-level-ai-for-real-time-fighting
|
1904.03821
| null |
https://arxiv.org/abs/1904.03821v3
|
https://arxiv.org/pdf/1904.03821v3.pdf
|
Creating Pro-Level AI for a Real-Time Fighting Game Using Deep Reinforcement Learning
|
Reinforcement learning combined with deep neural networks has performed remarkably well in many genres of games recently. It has surpassed human-level performance in fixed game environments and turn-based two player board games. However, to the best of our knowledge, current research has yet to produce a result that has surpassed human-level performance in modern complex fighting games. This is due to the inherent difficulties with real-time fighting games, including: vast action spaces, action dependencies, and imperfect information. We overcame these challenges and made 1v1 battle AI agents for the commercial game "Blade & Soul". The trained agents competed against five professional gamers and achieved a win rate of 62%. This paper presents a practical reinforcement learning method that includes a novel self-play curriculum and data skipping techniques. Through the curriculum, three different styles of agents were created by reward shaping and were trained against each other. Additionally, this paper suggests data skipping techniques that could increase data efficiency and facilitate explorations in vast spaces. Since our method can be generally applied to all two-player competitive games with vast action spaces, we anticipate its application to game development including level design and automated balancing.
|
['Hyoil Lee', 'Seongho Son', 'Inseok Oh', 'Seungeun Rho', 'Sangbin Moon', 'Jinyun Chung']
|
2019-04-08
| null | null | null | null |
['board-games']
|
['playing-games']
|
[-1.33496791e-01 -6.19763657e-02 -3.49945277e-02 3.74968420e-03
-4.18214738e-01 -4.05916274e-01 2.54176736e-01 -5.98208234e-02
-8.57705951e-01 1.08000076e+00 -1.21416308e-01 -4.76720065e-01
-4.77983057e-01 -9.80630517e-01 -3.15406978e-01 -3.94050717e-01
-5.08940160e-01 7.61616528e-01 5.50473750e-01 -1.23148429e+00
6.08612418e-01 2.28293747e-01 -1.76930249e+00 2.01260448e-01
6.17703378e-01 6.74385726e-01 2.49295920e-01 9.06365395e-01
2.52540112e-01 1.37872839e+00 -1.06134593e+00 -1.69593468e-01
6.63186073e-01 -4.88255620e-01 -8.33718300e-01 -4.44367439e-01
-2.35890210e-01 -4.52677637e-01 -3.51463199e-01 5.41565418e-01
7.95537949e-01 5.34667730e-01 9.85660702e-02 -1.32567298e+00
-8.15932900e-02 7.62483537e-01 -6.42219782e-01 4.48052377e-01
3.22581291e-01 3.93599421e-01 9.30152237e-01 -1.50376946e-01
3.46529961e-01 9.14676130e-01 5.81514835e-01 8.47533941e-01
-7.96023548e-01 -9.14334059e-01 -1.20578013e-01 3.68740797e-01
-1.12599194e+00 -3.40316743e-02 6.26234353e-01 -2.78227359e-01
1.35887659e+00 6.55647591e-02 1.34632337e+00 7.12222815e-01
3.08851272e-01 8.34523618e-01 1.15730715e+00 -3.70083153e-01
6.15072012e-01 -4.20941293e-01 -5.56241512e-01 5.12779653e-01
-1.15323842e-01 8.93763125e-01 -6.81996047e-01 1.79117590e-01
9.71090317e-01 -5.64229846e-01 2.62613624e-01 -5.04828095e-01
-7.61804461e-01 1.19228601e+00 4.66931194e-01 2.75649846e-01
-4.45064873e-01 4.11081940e-01 5.37289619e-01 5.03493845e-01
-3.49391513e-02 1.24149203e+00 -4.41369593e-01 -9.86327171e-01
-9.16112483e-01 9.50337827e-01 9.35943246e-01 4.37212914e-01
3.41213077e-01 5.73632956e-01 1.93940476e-01 8.60035479e-01
-9.03334022e-02 2.37706080e-02 5.47088444e-01 -1.00656033e+00
3.69932115e-01 5.75243413e-01 1.11281745e-01 -8.78320813e-01
-8.69980454e-01 -4.56772268e-01 -4.08662438e-01 1.16908872e+00
5.11711061e-01 -5.43815553e-01 -6.63811803e-01 1.74749458e+00
2.72279322e-01 -1.54630229e-01 1.12763301e-01 1.06947803e+00
5.75588286e-01 3.25199306e-01 -6.58191293e-02 2.17789382e-01
1.15304136e+00 -8.95865858e-01 -3.83012980e-01 -5.16715467e-01
4.41794455e-01 -4.64878827e-01 9.44924951e-01 8.09080064e-01
-1.40210402e+00 -5.82378805e-01 -1.35900950e+00 5.15983284e-01
-2.32047722e-01 -5.78741729e-01 1.06188619e+00 1.00381088e+00
-9.28144395e-01 9.00450766e-01 -9.04525399e-01 -3.15872915e-02
4.08838689e-01 8.91437829e-01 -5.84080368e-02 4.68287259e-01
-1.49893928e+00 1.30159640e+00 6.82033777e-01 -1.03860699e-01
-1.03964710e+00 -5.39708555e-01 -7.10617781e-01 -1.12696402e-01
8.09364259e-01 -2.85434037e-01 1.66848958e+00 -8.25068116e-01
-1.93699729e+00 5.31462371e-01 9.72155094e-01 -7.53991961e-01
4.72458094e-01 -1.65555730e-01 -1.72553107e-01 -3.24384034e-01
5.89024797e-02 6.76871777e-01 2.28928447e-01 -7.83647716e-01
-9.29750741e-01 -9.64126810e-02 6.81402266e-01 7.62221575e-01
-1.24571651e-01 5.02036279e-03 1.94821730e-02 -5.69369435e-01
-3.40417564e-01 -7.34914362e-01 -6.42575145e-01 -4.73179132e-01
1.67787820e-01 -5.76812327e-02 3.48669142e-01 -1.75447702e-01
1.24700034e+00 -1.76208544e+00 2.29200721e-01 1.80398837e-01
1.34650066e-01 4.07993048e-01 -1.92012757e-01 7.29834855e-01
-2.64908951e-02 -2.39078000e-01 1.74777746e-01 3.21493387e-01
1.42812595e-01 4.40283865e-01 1.44685909e-01 1.81772500e-01
-1.65933505e-01 9.03472722e-01 -1.06026328e+00 -1.82040781e-01
2.26156026e-01 -1.01885773e-01 -1.07111883e+00 1.64627880e-01
-2.40843490e-01 2.91052967e-01 -2.46528193e-01 4.35836673e-01
3.19244951e-01 4.01583076e-01 2.88792253e-01 4.23815042e-01
-3.19432318e-01 2.75634885e-01 -1.41603220e+00 1.66868901e+00
-3.37805390e-01 5.83555877e-01 1.28698051e-01 -1.16461802e+00
1.08620250e+00 -2.70304200e-03 7.46089339e-01 -1.31968403e+00
4.61610258e-01 2.47208700e-01 7.19203889e-01 -2.65914112e-01
1.05791259e+00 -6.12952113e-01 -3.86044085e-01 3.23032409e-01
-1.30184395e-02 -8.00306618e-01 6.80607736e-01 -1.58811718e-01
1.35662723e+00 4.26152706e-01 3.44853252e-01 -3.00494134e-01
1.43271223e-01 4.52169210e-01 6.18817329e-01 9.95119810e-01
-4.37824965e-01 1.57397225e-01 5.84848344e-01 -8.87264967e-01
-1.06097507e+00 -9.10786092e-01 3.73091221e-01 1.58055091e+00
3.06901127e-01 -5.49459994e-01 -6.00947499e-01 -1.18270934e-01
-6.18852228e-02 5.21915495e-01 -7.05326378e-01 -2.78272480e-01
-8.14038694e-01 -5.29980779e-01 7.47797489e-01 7.53463089e-01
5.01009047e-01 -1.62678194e+00 -1.20704460e+00 7.68297136e-01
1.36049032e-01 -6.93022549e-01 -3.95521447e-02 6.55800104e-01
-4.76036668e-01 -1.24966848e+00 -3.27956706e-01 -7.19120800e-01
-1.68644413e-01 1.78642813e-02 1.28362703e+00 2.25337729e-01
-5.29736578e-01 -5.46489060e-02 -5.05555212e-01 -6.51324689e-01
-3.12628508e-01 8.37635100e-02 3.04455876e-01 -9.64423478e-01
2.88129330e-01 -6.54263914e-01 -6.70373559e-01 4.26031262e-01
-6.97105706e-01 1.65262043e-01 4.33208227e-01 1.21847808e+00
9.17071179e-02 5.12932181e-01 4.74899232e-01 -4.38411355e-01
1.13636076e+00 -2.33091801e-01 -7.71535993e-01 -3.30561936e-01
-3.31332088e-01 -5.06790392e-02 7.09066212e-01 -5.39587736e-01
-8.26959908e-01 -2.45663866e-01 -4.22333241e-01 5.75347170e-02
2.45867312e-01 4.13765490e-01 3.83029968e-01 -1.99547678e-01
1.11275423e+00 2.21346449e-02 3.08453441e-01 3.24130617e-02
8.56167227e-02 3.87858003e-01 4.53702778e-01 -8.93098474e-01
6.68036580e-01 -4.62393165e-02 -7.07247034e-02 -5.09292483e-01
-2.88148284e-01 -1.39323577e-01 -2.97578841e-01 -4.95311469e-01
6.54747665e-01 -7.68504918e-01 -1.46300125e+00 6.69248879e-01
-5.18985033e-01 -9.72177744e-01 -7.40280986e-01 4.39904124e-01
-1.04703963e+00 -4.54880483e-02 -7.75690675e-01 -8.06207061e-01
9.81040485e-03 -1.16557562e+00 3.49240690e-01 6.76420569e-01
-3.14304620e-01 -7.08856165e-01 6.18123472e-01 4.34180170e-01
5.71429133e-01 1.88347042e-01 3.82764667e-01 -4.44900572e-01
-1.65195018e-01 1.89667437e-02 3.37639898e-01 1.30957097e-01
-6.24613166e-02 -2.72588342e-01 -3.69434655e-01 -4.69514191e-01
-3.86243939e-01 -8.97678316e-01 1.94096699e-01 3.50901097e-01
8.40589643e-01 2.00302958e-01 1.66573316e-01 3.41255695e-01
1.20275307e+00 8.67099404e-01 7.52704024e-01 1.07559109e+00
1.85888559e-01 4.32165951e-01 8.41892123e-01 8.65605235e-01
2.26776272e-01 9.18940425e-01 8.28665197e-01 -9.97753143e-02
1.43017367e-01 -3.18168521e-01 1.29226163e-01 3.95905137e-01
-5.16632795e-01 -1.81075424e-01 -9.92271841e-01 4.64690745e-01
-1.82814527e+00 -1.28111434e+00 2.27996796e-01 1.78726900e+00
7.77957678e-01 6.00000620e-01 6.57143950e-01 3.03486079e-01
3.65969092e-01 1.00822002e-01 -7.00822651e-01 -9.93218660e-01
3.70093100e-02 8.56009960e-01 5.34936965e-01 4.60134357e-01
-9.94267762e-01 1.24778831e+00 6.75704432e+00 1.07464242e+00
-7.45580554e-01 -2.53073841e-01 3.23347986e-01 -6.07415557e-01
2.48087674e-01 -4.23381068e-02 -3.84915590e-01 2.70534039e-01
6.79376841e-01 -2.97814846e-01 8.42630148e-01 1.03127539e+00
1.23024821e-01 -4.02856320e-01 -6.62595451e-01 8.57818604e-01
-2.83970088e-01 -1.43107283e+00 -5.95307827e-01 2.41709813e-01
7.59232700e-01 -5.32109924e-02 -3.05023836e-03 1.00694644e+00
1.17274737e+00 -1.39372981e+00 7.68531978e-01 -7.95365497e-02
5.22113383e-01 -1.24825478e+00 6.98309541e-01 4.92180675e-01
-1.15406275e+00 -6.37513936e-01 -2.41868585e-01 -1.07717800e+00
6.61981031e-02 -3.93829435e-01 -7.77473390e-01 4.64075536e-01
6.76333666e-01 1.40219569e-01 -1.44501761e-01 1.19140291e+00
-9.86549929e-02 3.48339319e-01 -3.61486405e-01 -6.37864769e-01
6.90683365e-01 1.88725721e-02 2.56194234e-01 5.99005342e-01
3.89549434e-02 3.76438707e-01 3.47747833e-01 2.24327981e-01
4.53532845e-01 6.07692003e-02 -5.27489543e-01 3.95087972e-02
3.34239125e-01 9.39404845e-01 -8.39567959e-01 2.20734440e-02
-5.51289134e-02 5.61590850e-01 4.63613719e-01 -2.39706650e-01
-9.37030315e-01 -4.72065419e-01 8.98878038e-01 1.61137864e-01
1.95384115e-01 -3.40895057e-01 -3.08560193e-01 -7.40168810e-01
-3.40827942e-01 -1.34451866e+00 4.71483499e-01 -5.29439151e-01
-1.10713184e+00 7.25922823e-01 1.31772133e-02 -1.22243869e+00
-5.45695007e-01 -7.20303059e-01 -6.86205924e-01 5.00226855e-01
-7.88811743e-01 -4.97197598e-01 -1.35781392e-01 5.87869406e-01
6.67139888e-01 -8.34665537e-01 7.88612902e-01 2.16772437e-01
-3.23378325e-01 5.60709000e-01 9.13734734e-02 1.19581871e-01
2.50142127e-01 -1.29603338e+00 4.80353802e-01 3.74192208e-01
-2.58539487e-02 7.93818608e-02 7.95445383e-01 -4.56176817e-01
-1.21615028e+00 -1.32830456e-01 -1.24649562e-01 -6.77110255e-02
8.38133454e-01 -2.89572686e-01 -3.53989571e-01 4.40979376e-02
2.69457430e-01 -4.79620188e-01 6.71662807e-01 4.43939745e-01
1.90264016e-01 -9.96717438e-02 -9.35162425e-01 9.30629671e-01
1.00210774e+00 -1.46323755e-01 -6.33514225e-01 2.13514064e-02
1.08053781e-01 -1.02505159e+00 -5.29885650e-01 4.13322389e-01
7.78076768e-01 -1.18783319e+00 9.57858086e-01 -9.84677434e-01
5.76709032e-01 -2.20324412e-01 -5.31740412e-02 -1.61951494e+00
-5.99925935e-01 -6.52418554e-01 1.40354902e-01 4.65996295e-01
1.18973076e-01 -1.78451598e-01 1.42127013e+00 4.91224736e-01
-1.53317586e-01 -1.00321043e+00 -1.13564372e+00 -8.29484880e-01
5.71112216e-01 -5.16932487e-01 6.37494087e-01 7.05041170e-01
5.15608668e-01 2.45506972e-01 -8.34415317e-01 -4.43392754e-01
3.67488027e-01 -1.06118314e-01 9.51928735e-01 -9.63430762e-01
-8.11711729e-01 -8.26854527e-01 -7.15234756e-01 -9.06018615e-01
-1.76492214e-01 -3.95547271e-01 2.70518214e-01 -1.31053996e+00
-1.19200528e-01 -6.68313324e-01 -1.53766036e-01 4.92489636e-01
2.47765090e-02 4.36385185e-01 4.94116068e-01 -2.23888054e-01
-6.16963148e-01 4.82463241e-01 1.53000796e+00 -1.00258075e-01
-4.31154996e-01 -5.36556542e-03 -9.29106414e-01 6.36936486e-01
1.37438881e+00 -2.25970253e-01 -5.96181870e-01 -1.93923727e-01
5.01061082e-01 4.89243209e-01 -2.70671379e-02 -1.51341248e+00
4.49932963e-01 -5.95467031e-01 2.74650007e-01 -3.30731511e-01
4.76145446e-01 -5.00983238e-01 9.47784036e-02 8.67035389e-01
-2.03263044e-01 2.41356269e-01 6.33038104e-01 2.29893133e-01
-1.97015315e-01 -2.79228330e-01 6.65352881e-01 -3.64794284e-01
-1.08721232e+00 -2.00461224e-02 -9.97971117e-01 3.72336745e-01
1.39895809e+00 -7.36267447e-01 -7.44214356e-02 -5.46036482e-01
-6.76408470e-01 5.38628042e-01 1.33836731e-01 5.14653325e-01
4.51234102e-01 -1.21777165e+00 -6.70087099e-01 1.70052007e-01
-2.45213062e-01 -1.72592431e-01 5.04146755e-01 2.50107676e-01
-1.04391015e+00 1.14111081e-01 -1.12979221e+00 -1.74296588e-01
-1.22082579e+00 3.96358311e-01 6.54044807e-01 -6.37121618e-01
-5.71819961e-01 9.41761196e-01 -1.26593217e-01 -6.07642114e-01
2.01035976e-01 1.65340453e-01 -3.81594777e-01 -1.70498490e-01
5.30168772e-01 3.57490838e-01 3.19178705e-03 -1.80671141e-01
-4.40546185e-01 1.79202631e-01 -1.82371438e-01 -3.66442055e-01
1.38672590e+00 5.15137792e-01 5.07930279e-01 4.70358972e-03
3.59901905e-01 -2.39561781e-01 -1.49940598e+00 2.97336310e-01
-2.34307379e-01 -5.92359245e-01 -1.08489759e-01 -9.86089945e-01
-1.08097780e+00 5.58387995e-01 5.09519339e-01 3.42680454e-01
1.16156399e+00 -4.59632009e-01 6.61643684e-01 5.03215492e-01
8.36777985e-01 -1.66824567e+00 6.36824369e-01 9.38741207e-01
6.44049406e-01 -9.75436628e-01 -2.52538975e-02 3.20378006e-01
-9.42352533e-01 1.08447254e+00 1.22856307e+00 -4.68605071e-01
2.62964606e-01 8.15314829e-01 2.30897427e-01 -2.99619168e-01
-8.94535422e-01 -3.84071290e-01 -2.69058853e-01 9.65454638e-01
2.57098705e-01 1.56000957e-01 -5.37702799e-01 4.41779375e-01
-8.59626710e-01 1.07371196e-01 6.86337352e-01 1.25038338e+00
-6.54870272e-01 -1.44518292e+00 -4.30379003e-01 3.84386271e-01
-3.57211530e-01 8.21071118e-02 -1.01400867e-01 1.19388819e+00
3.09718132e-01 1.01765168e+00 2.43874844e-02 -7.58005142e-01
6.35635972e-01 -5.29194415e-01 7.10578203e-01 -4.55722481e-01
-1.17554009e+00 -3.20950568e-01 3.29528749e-01 -7.16785669e-01
-1.28179505e-01 -3.73459488e-01 -1.26924622e+00 -9.12065089e-01
-2.25360528e-01 4.22992170e-01 4.27049458e-01 6.81593120e-01
-1.37825593e-01 9.14184868e-01 4.18613076e-01 -1.01868784e+00
-6.33805931e-01 -6.62179232e-01 -8.32783222e-01 7.63772354e-02
-2.27526188e-01 -1.15561950e+00 2.53767788e-01 -7.24333048e-01]
|
[3.4871182441711426, 1.4848140478134155]
|
a514d489-73c2-4ae7-8dd3-34058a22c8b9
|
modeling-the-trade-off-of-privacy
|
2303.10435
| null |
https://arxiv.org/abs/2303.10435v1
|
https://arxiv.org/pdf/2303.10435v1.pdf
|
Modeling the Trade-off of Privacy Preservation and Activity Recognition on Low-Resolution Images
|
A computer vision system using low-resolution image sensors can provide intelligent services (e.g., activity recognition) but preserve unnecessary visual privacy information from the hardware level. However, preserving visual privacy and enabling accurate machine recognition have adversarial needs on image resolution. Modeling the trade-off of privacy preservation and machine recognition performance can guide future privacy-preserving computer vision systems using low-resolution image sensors. In this paper, using the at-home activity of daily livings (ADLs) as the scenario, we first obtained the most important visual privacy features through a user survey. Then we quantified and analyzed the effects of image resolution on human and machine recognition performance in activity recognition and privacy awareness tasks. We also investigated how modern image super-resolution techniques influence these effects. Based on the results, we proposed a method for modeling the trade-off of privacy preservation and activity recognition on low-resolution images.
|
['Yuanchun Shi', 'Shwetak Patel', 'Chun Yu', 'Yukang Yan', 'Xuhai Xu', 'Xueyang Wang', 'Yan Kong', 'Xin Yi', 'Zirui Cheng', 'Yuntao Wang']
|
2023-03-18
| null | null | null | null |
['image-super-resolution']
|
['computer-vision']
|
[ 7.09825337e-01 -1.05274830e-03 -1.03129640e-01 -5.91156840e-01
-6.76796257e-01 -5.33631146e-01 5.72622359e-01 -1.69619724e-01
-7.52660394e-01 5.46635389e-01 2.39943355e-01 -8.38989317e-02
8.20780545e-02 -6.96830571e-01 -8.25218379e-01 -7.45703042e-01
1.11785484e-02 -5.84493041e-01 -1.27839759e-01 2.92996973e-01
1.33990765e-01 8.14878047e-01 -1.69306791e+00 4.69277382e-01
6.04145944e-01 1.17401552e+00 -1.64263308e-01 6.47329450e-01
3.98602098e-01 7.86712825e-01 -4.38651919e-01 -6.29419744e-01
6.86983109e-01 -1.27971694e-01 -3.10093105e-01 2.16975480e-01
4.61900294e-01 -8.36978436e-01 -5.04828513e-01 1.31041276e+00
2.19615579e-01 -1.55637041e-01 2.77417243e-01 -1.47192264e+00
-1.08408046e+00 7.25389272e-02 -7.21455216e-01 6.25629127e-02
4.46664661e-01 1.23199567e-01 2.90308416e-01 -3.23501468e-01
5.60665369e-01 1.03789616e+00 5.58029413e-01 7.42740154e-01
-1.36857283e+00 -8.75799477e-01 -1.11987464e-01 4.19115990e-01
-1.61967909e+00 -6.69297755e-01 6.23147190e-01 -2.03502670e-01
6.45015001e-01 6.65360689e-01 4.98170525e-01 1.29215169e+00
3.92394781e-01 3.05740863e-01 1.40105605e+00 -2.54853606e-01
3.79723340e-01 5.79371035e-01 2.27376401e-01 5.30184090e-01
8.34890902e-01 4.81832206e-01 -6.27698243e-01 -3.81846368e-01
1.00092864e+00 2.74160802e-01 -3.66127819e-01 -5.60247123e-01
-7.51732349e-01 6.03218973e-01 1.11070715e-01 1.38142720e-01
-1.95359096e-01 -1.02316931e-01 6.62719458e-02 3.42138469e-01
4.69964147e-02 2.68604547e-01 -3.63439471e-01 4.29673910e-01
-5.82762778e-01 -1.07744977e-01 5.65539598e-01 8.81400049e-01
7.43242443e-01 -3.45038980e-01 -3.41598898e-01 3.84255916e-01
1.37825087e-01 4.48019475e-01 2.79920846e-01 -1.18715918e+00
2.59457439e-01 3.96341801e-01 4.57337350e-01 -1.08618355e+00
-7.89670348e-02 1.71985745e-01 -9.06374931e-01 6.74873054e-01
5.11878669e-01 2.85905693e-03 -8.04626524e-01 1.55971205e+00
1.04629275e-04 -2.65877426e-01 3.69051337e-01 9.55377281e-01
1.52620286e-01 3.31595302e-01 2.94629425e-01 -6.56152129e-01
1.68840837e+00 -3.36808890e-01 -9.71059859e-01 -1.15647048e-01
1.70025393e-01 -2.20721722e-01 1.09870100e+00 2.62515008e-01
-8.19433570e-01 -3.13123554e-01 -1.25243509e+00 -2.60418713e-01
-5.45990288e-01 -1.73069105e-01 6.24101758e-01 1.39251208e+00
-9.73071992e-01 3.62496078e-01 -9.37370777e-01 -3.84256989e-01
7.40686655e-01 4.12162066e-01 -7.00407326e-01 -1.65523499e-01
-1.00752115e+00 7.66714633e-01 -4.69742604e-02 -4.11891907e-01
-5.56192517e-01 -6.65612817e-01 -7.90541351e-01 -4.90485951e-02
1.21407129e-01 -6.38060331e-01 7.90974796e-01 -1.19633889e+00
-1.25212741e+00 1.14716363e+00 -5.19821830e-02 -8.77731264e-01
7.90328562e-01 -1.58336908e-01 -6.32662594e-01 7.36379176e-02
-2.39486679e-01 4.43756342e-01 1.03197372e+00 -1.19866705e+00
-6.22254312e-01 -1.14095891e+00 -2.25094825e-01 9.71445292e-02
-4.87674743e-01 1.19204722e-01 -1.57808051e-01 -2.26354733e-01
-2.20607609e-01 -6.11449599e-01 -1.11874864e-01 5.09758651e-01
3.33708972e-02 6.75717235e-01 1.23469448e+00 -9.85412538e-01
7.13523507e-01 -2.46799564e+00 -5.70316374e-01 2.04094291e-01
1.71354622e-01 1.85459122e-01 9.23655257e-02 -3.93765301e-01
1.83916450e-01 3.03838849e-01 -5.07795922e-02 -8.02559927e-02
-2.77743429e-01 9.50100720e-02 -1.29802972e-01 7.48116076e-01
-1.53058037e-01 8.02820086e-01 -3.14196736e-01 -4.00217205e-01
4.65905607e-01 1.07127392e+00 -3.16858292e-01 2.72903025e-01
2.40496919e-01 5.18468976e-01 -3.04281265e-01 7.41192400e-01
9.68500793e-01 4.21912149e-02 1.53804377e-01 -3.12440425e-01
-1.28671406e-02 -5.94739377e-01 -8.71211290e-01 1.31312490e+00
-2.20111832e-01 4.74165946e-01 4.27025139e-01 -4.04071897e-01
9.42124069e-01 5.62958121e-02 4.65356469e-01 -1.24362195e+00
1.13986999e-01 -4.08302397e-01 -4.87611532e-01 -3.79912496e-01
4.37271625e-01 1.43602282e-01 -5.64763732e-02 1.48061335e-01
-5.34612715e-01 5.47468007e-01 -6.57420397e-01 -2.40596876e-01
1.18261766e+00 -2.45960176e-01 6.57104015e-01 -1.67160213e-01
4.54252720e-01 -2.72600174e-01 4.99597847e-01 7.38505781e-01
-7.54804075e-01 5.23133695e-01 1.08862832e-01 -6.79305017e-01
-1.00695193e+00 -1.13022816e+00 -1.72933146e-01 8.25230002e-01
3.07986140e-01 -8.58952403e-02 -1.03941584e+00 -4.70085442e-01
1.49016008e-01 8.42235982e-01 -6.78555787e-01 -3.50352645e-01
-6.99723661e-02 -6.53911173e-01 7.69568443e-01 2.50289530e-01
1.01779139e+00 -6.09465718e-01 -1.25349259e+00 -4.84480232e-01
1.77289937e-02 -1.37368762e+00 -4.42163408e-01 -1.52707174e-01
-6.23727918e-01 -9.56663370e-01 -5.37899017e-01 -2.43032530e-01
8.01751792e-01 4.08140421e-01 7.11271048e-01 -3.32264751e-01
-6.71341956e-01 6.74044847e-01 -1.11243308e-01 -5.72814167e-01
-2.39873424e-01 -3.64760101e-01 2.19430938e-01 4.76658046e-01
6.93912506e-01 -6.77533865e-01 -6.32451117e-01 3.37498218e-01
-8.82858634e-01 -1.55968815e-01 5.83789825e-01 3.30526292e-01
9.51985180e-01 2.75375396e-01 -1.53841659e-01 -7.86275148e-01
5.93481421e-01 7.16771325e-03 -9.62898552e-01 5.72646499e-01
-9.68389213e-01 2.55435377e-01 5.19014060e-01 -6.22561932e-01
-1.27535474e+00 5.80320776e-01 4.03423160e-01 -6.24884605e-01
-4.14323032e-01 -5.06981432e-01 -8.26098979e-01 -4.14580226e-01
6.23017788e-01 3.17250520e-01 9.58710760e-02 -3.20834428e-01
3.65968466e-01 7.68552601e-01 6.35383129e-01 -7.61497170e-02
5.66457748e-01 8.66610169e-01 -9.38458461e-03 -1.10027623e+00
-2.02652141e-01 -4.64423336e-02 -3.63140821e-01 9.08223353e-03
1.10296476e+00 -1.04242289e+00 -9.67107415e-01 5.50769866e-01
-7.94802964e-01 2.27003872e-01 -3.27308267e-01 2.07697257e-01
-6.99105501e-01 4.78640378e-01 -2.49583930e-01 -1.31138873e+00
-3.48975360e-01 -7.26117432e-01 9.17954564e-01 3.89121085e-01
-1.88823268e-01 -3.99366021e-01 -2.37077847e-01 6.91985071e-01
4.06206399e-01 6.23452663e-01 5.35296440e-01 -1.54465705e-01
-9.03355658e-01 -1.63521007e-01 -4.39315379e-01 2.32604474e-01
2.02565685e-01 -5.28105080e-01 -1.27363682e+00 -4.19378400e-01
4.63087231e-01 -8.71055126e-02 5.68528116e-01 3.63533229e-01
1.50969970e+00 -9.46417034e-01 -1.66439384e-01 1.13507974e+00
1.60964715e+00 3.25045377e-01 1.38630390e+00 4.63435024e-01
6.72399580e-01 6.87932491e-01 4.85839218e-01 6.00286126e-01
4.20446806e-02 5.92065096e-01 3.60968500e-01 -3.77556793e-02
3.15775603e-01 -3.59519511e-01 5.82893491e-01 -3.24658871e-01
-8.86014849e-02 2.49880642e-01 -4.92614985e-01 2.44723693e-01
-1.80893660e+00 -1.03708208e+00 3.70330989e-01 2.44654346e+00
3.90860796e-01 -3.25350821e-01 1.70267090e-01 5.67433599e-04
7.22219169e-01 1.13829337e-01 -8.70483875e-01 -4.45378453e-01
-2.71633953e-01 -2.09133044e-01 1.39804542e+00 4.22287732e-01
-9.99040127e-01 6.27990007e-01 6.61206341e+00 3.02462816e-01
-8.12806606e-01 1.60664454e-01 9.44512010e-01 -2.71365941e-01
-1.28369644e-01 -4.35378492e-01 -6.71772957e-01 3.91388535e-01
9.60125446e-01 -3.10283273e-01 8.78639579e-01 1.07974243e+00
1.74769238e-01 2.45024450e-02 -1.16984391e+00 1.40866911e+00
1.55672610e-01 -9.70472932e-01 1.97119668e-01 4.74220365e-01
2.23503947e-01 -4.39596295e-01 4.75715935e-01 -3.24638754e-01
2.93113768e-01 -1.14466667e+00 2.67962426e-01 5.18637598e-01
1.03880179e+00 -9.77965951e-01 3.74037266e-01 2.67456979e-01
-1.01248336e+00 -3.76342058e-01 -3.89961958e-01 2.98658721e-02
-2.58798033e-01 2.25788429e-01 -4.88765419e-01 1.41839892e-01
1.08525872e+00 1.47728592e-01 -5.88863075e-01 1.32219076e-01
7.90736154e-02 -2.79569626e-02 -2.41099626e-01 2.37778708e-01
-5.14604509e-01 -2.52540797e-01 2.71315843e-01 1.06792176e+00
3.27266365e-01 4.59980518e-01 -4.03702319e-01 1.12966418e+00
1.04143070e-02 -1.36358112e-01 -1.04946077e+00 6.24725712e-04
4.48091477e-01 1.04099023e+00 -3.77708614e-01 2.94063330e-01
-5.13127327e-01 1.51273286e+00 1.76025904e-03 4.28244621e-01
-7.05655634e-01 -6.01047128e-02 1.43484378e+00 5.09396970e-01
1.98314264e-01 -6.60054535e-02 -8.69070590e-01 -1.20910144e+00
2.02167913e-01 -8.87090743e-01 5.10481596e-01 -7.19950378e-01
-1.04989195e+00 1.15791500e-01 -5.96531890e-02 -9.22757506e-01
-1.94241777e-02 -4.20475513e-01 1.03783652e-01 6.73294544e-01
-1.24105287e+00 -1.23042274e+00 -3.24004412e-01 1.19692290e+00
7.61951134e-02 -2.38262609e-01 1.02786410e+00 -8.47511576e-04
-3.98462683e-01 1.20316148e+00 3.75536876e-03 2.22890466e-01
3.44722271e-01 -6.17470324e-01 4.02955264e-01 1.21352446e+00
1.29035205e-01 6.35078430e-01 6.01841509e-01 -5.58861732e-01
-1.67396021e+00 -1.19114447e+00 4.08541411e-01 -8.33561122e-01
5.75019866e-02 -4.36724901e-01 -8.76331270e-01 7.15794981e-01
-9.14793983e-02 8.57049823e-02 6.62154078e-01 -4.51277196e-01
-7.14035451e-01 -3.97064626e-01 -2.22189307e+00 6.73938751e-01
1.12306821e+00 -8.51803660e-01 -1.90258875e-01 -3.16298991e-01
7.98575580e-01 7.47151971e-02 -9.07358646e-01 9.87707227e-02
9.25194085e-01 -1.07707107e+00 1.32351232e+00 -1.53852165e-01
-7.36443624e-02 -1.73215777e-01 -6.98989689e-01 -5.06130517e-01
-4.06026363e-01 -5.98185599e-01 -3.29749763e-01 1.09034419e+00
9.49093997e-02 -7.13319540e-01 1.06351125e+00 1.68868506e+00
1.13221204e+00 1.01709582e-01 -9.20701385e-01 -8.01829875e-01
-4.14441049e-01 -1.72765270e-01 6.45861030e-01 9.15885210e-01
-3.12419742e-01 -1.60443619e-01 -7.39530206e-01 7.72761762e-01
1.35552478e+00 -4.20055836e-01 5.73796213e-01 -9.17431831e-01
-1.86858810e-02 1.70470580e-01 -7.66877234e-01 -2.59409428e-01
4.42482680e-02 -2.64400542e-01 -4.57092613e-01 -9.25415695e-01
4.87571955e-01 3.40643823e-01 -5.50735116e-01 3.50382060e-01
2.42177457e-01 2.72764981e-01 2.44639561e-01 1.73138440e-01
-4.15500790e-01 -5.39280511e-02 6.95343137e-01 -6.57783449e-02
-2.41346806e-01 -1.03070596e-02 -1.14102745e+00 5.09069145e-01
9.65371966e-01 -3.83096337e-01 -5.04417598e-01 -2.70946354e-01
-3.23659629e-01 2.59240251e-02 5.94864964e-01 -1.00228226e+00
2.84996331e-01 -3.89392942e-01 7.98995554e-01 -8.80901366e-02
4.77226883e-01 -1.64279556e+00 5.22479177e-01 6.18445814e-01
-4.84741628e-01 -2.65155733e-01 1.36226833e-01 8.80886793e-01
1.95443466e-01 5.29459476e-01 1.10482919e+00 -3.20532173e-01
-8.68446231e-01 3.08439046e-01 -4.63785470e-01 -4.82436717e-01
1.36555576e+00 -7.51545250e-01 -3.99653673e-01 -4.61166054e-01
-5.05759001e-01 -1.38381004e-01 1.12760854e+00 5.42226911e-01
8.98354590e-01 -1.11180568e+00 -4.60570544e-01 7.50162244e-01
3.59600961e-01 -6.22009039e-01 1.24536097e-01 4.14357008e-03
-2.45926365e-01 2.17130631e-01 -8.47250700e-01 -2.09598064e-01
-1.88219953e+00 9.83024657e-01 3.73382539e-01 -1.41779482e-01
-5.22746146e-01 3.50830257e-01 3.14442098e-01 6.87063336e-02
5.15798092e-01 -7.64945969e-02 -3.35566923e-02 -3.88615876e-01
1.16930819e+00 5.52499473e-01 -2.29144856e-01 -6.09253705e-01
-5.74432552e-01 5.26518703e-01 -1.53305948e-01 -2.95874000e-01
9.49276686e-01 -6.46834970e-01 1.69028759e-01 -3.42630781e-02
1.28752112e+00 -2.33124077e-01 -1.49978149e+00 1.14411399e-01
-5.96975461e-02 -1.02166903e+00 1.83380693e-01 -7.72647560e-01
-1.08925402e+00 4.29241270e-01 1.33988047e+00 5.65252043e-02
1.59762621e+00 -2.95611501e-01 5.39289951e-01 2.99182475e-01
8.05405736e-01 -1.12152278e+00 -4.33471590e-01 -2.92066783e-01
5.86775482e-01 -1.34533226e+00 1.37136400e-01 -5.79853296e-01
-8.25862348e-01 5.32929897e-01 4.90370274e-01 1.14405327e-01
5.95479667e-01 7.15079069e-01 2.19596803e-01 3.04937184e-01
-4.28988010e-01 2.96034127e-01 -5.55885956e-02 1.40323126e+00
-1.14452444e-01 3.61989349e-01 -1.66388974e-01 8.04512739e-01
-1.19669281e-01 2.74457961e-01 4.14340824e-01 8.53878975e-01
-1.81902006e-01 -8.40295017e-01 -6.87426209e-01 3.08918148e-01
-7.18286216e-01 3.63603592e-01 -9.95773137e-01 5.40963471e-01
1.42583549e-01 9.95365441e-01 6.24853522e-02 -7.44555891e-01
4.55907106e-01 -9.40078124e-02 4.72329646e-01 1.21274188e-01
-4.91512865e-01 -6.50367141e-01 -2.31033396e-02 -1.14657915e+00
-3.71832430e-01 -7.00090587e-01 -7.11652398e-01 -5.94401479e-01
3.25436592e-01 -3.44627768e-01 8.44340503e-01 4.49297577e-01
6.22453690e-01 9.37241688e-03 5.37202120e-01 -4.26059663e-01
-5.50505280e-01 -1.11687995e-01 -9.62976694e-01 6.27140999e-01
5.93770742e-01 -1.93376523e-02 -4.80887085e-01 2.87291765e-01]
|
[12.745426177978516, 0.815586268901825]
|
123a5577-f20f-4fff-82f6-1507207441f1
|
mass-masked-sequence-to-sequence-pre-training
|
1905.0245
| null |
https://arxiv.org/abs/1905.02450v5
|
https://arxiv.org/pdf/1905.02450v5.pdf
|
MASS: Masked Sequence to Sequence Pre-training for Language Generation
|
Pre-training and fine-tuning, e.g., BERT, have achieved great success in language understanding by transferring knowledge from rich-resource pre-training task to the low/zero-resource downstream tasks. Inspired by the success of BERT, we propose MAsked Sequence to Sequence pre-training (MASS) for the encoder-decoder based language generation tasks. MASS adopts the encoder-decoder framework to reconstruct a sentence fragment given the remaining part of the sentence: its encoder takes a sentence with randomly masked fragment (several consecutive tokens) as input, and its decoder tries to predict this masked fragment. In this way, MASS can jointly train the encoder and decoder to develop the capability of representation extraction and language modeling. By further fine-tuning on a variety of zero/low-resource language generation tasks, including neural machine translation, text summarization and conversational response generation (3 tasks and totally 8 datasets), MASS achieves significant improvements over the baselines without pre-training or with other pre-training methods. Specially, we achieve the state-of-the-art accuracy (37.5 in terms of BLEU score) on the unsupervised English-French translation, even beating the early attention-based supervised model.
|
['Tie-Yan Liu', 'Jianfeng Lu', 'Xu Tan', 'Tao Qin', 'Kaitao Song']
|
2019-05-07
| null | null | null | null |
['unsupervised-machine-translation', 'conversational-response-generation']
|
['natural-language-processing', 'natural-language-processing']
|
[ 5.84271014e-01 4.90352809e-01 -2.51241773e-01 -3.41882735e-01
-1.43310964e+00 -3.43093306e-01 7.64909863e-01 -2.15065718e-01
-3.14742297e-01 1.23951876e+00 7.90804744e-01 -5.08201659e-01
7.22582817e-01 -7.50221193e-01 -9.40225720e-01 -3.63782316e-01
3.79009426e-01 7.14346588e-01 -2.53741711e-01 -6.05371714e-01
1.56757414e-01 -3.16400558e-01 -8.93905938e-01 8.45358670e-01
1.26552594e+00 4.62591559e-01 5.22121489e-01 8.74407649e-01
-2.05424562e-01 8.97279203e-01 -7.07907319e-01 -6.81270063e-01
-2.52574444e-01 -9.47028697e-01 -1.12124300e+00 -1.12672500e-01
-2.04423070e-01 -4.89900738e-01 -3.49225879e-01 7.00329185e-01
7.35990584e-01 -2.91306507e-02 7.20346808e-01 -6.20081127e-01
-1.12701917e+00 1.19941890e+00 -3.32333267e-01 6.90290704e-02
6.08153582e-01 2.54583240e-01 9.94732261e-01 -1.12899113e+00
5.85658669e-01 1.38158154e+00 2.20024467e-01 1.02770638e+00
-1.07125568e+00 -4.61246192e-01 -3.23534384e-02 -6.13838583e-02
-1.04615474e+00 -9.30363357e-01 4.34513330e-01 -2.09609523e-01
1.52702796e+00 7.74713308e-02 9.71092582e-02 1.52400017e+00
2.48943835e-01 1.06413138e+00 6.48891330e-01 -6.34761631e-01
-1.81321800e-01 1.20878192e-02 -2.01448694e-01 4.71313506e-01
-9.86454934e-02 -9.77624729e-02 -5.38640618e-01 5.35839014e-02
3.85481358e-01 -4.46683615e-01 -3.56940210e-01 6.86502576e-01
-1.57683671e+00 8.52450192e-01 2.40784645e-01 2.45686188e-01
-5.03229201e-01 6.67768419e-02 5.37123561e-01 4.84649956e-01
8.15510571e-01 5.85129738e-01 -5.20215392e-01 -4.07433093e-01
-9.08500612e-01 -4.15420085e-02 8.49538982e-01 1.21899068e+00
6.34003103e-01 2.54643798e-01 -7.56342709e-01 9.85741019e-01
1.29358798e-01 4.60244060e-01 9.88257051e-01 -3.31634074e-01
1.23838234e+00 4.60996866e-01 -2.68261246e-02 -3.78568023e-01
-1.19596701e-02 -4.94600981e-01 -1.22640610e+00 -7.52780318e-01
1.18950261e-02 -6.76772058e-01 -7.95887291e-01 1.79500318e+00
-8.52017850e-02 1.01579633e-03 6.75423324e-01 7.42802024e-01
9.35701013e-01 1.19914603e+00 -1.14833668e-01 -4.87135917e-01
1.17127717e+00 -1.59337556e+00 -6.23424709e-01 -4.74959940e-01
9.11640882e-01 -8.30354214e-01 1.25879860e+00 8.48900303e-02
-1.32562053e+00 -7.44275630e-01 -8.08664680e-01 -3.60232770e-01
8.82451460e-02 5.47514379e-01 2.47101262e-01 1.13389283e-01
-1.04210758e+00 5.57285666e-01 -6.51803076e-01 -1.62670910e-01
1.35281920e-01 8.55347812e-02 -2.93839812e-01 -1.52966455e-01
-1.58512080e+00 8.55777264e-01 5.37860811e-01 1.53702050e-01
-1.04411101e+00 -4.45230216e-01 -8.80183160e-01 1.01108514e-01
9.12712887e-02 -1.20247495e+00 1.47432053e+00 -1.10311091e+00
-2.07079625e+00 7.16555715e-01 -4.93918419e-01 -7.36390173e-01
5.12886524e-01 -4.80016023e-01 -2.68805474e-01 -2.85338834e-02
2.45387271e-01 8.33789408e-01 7.23153591e-01 -6.16010547e-01
-2.95500576e-01 1.10009521e-01 -9.22503471e-02 3.70344996e-01
-1.59071952e-01 2.48026833e-01 -2.15199590e-01 -7.28643954e-01
-4.24161017e-01 -7.62039483e-01 -2.32455015e-01 -1.03102136e+00
-7.12797523e-01 -4.28502351e-01 1.36619166e-01 -1.05539072e+00
1.38838303e+00 -1.79663849e+00 3.74261916e-01 -6.02943718e-01
-2.30498031e-01 4.37711895e-01 -6.51790082e-01 9.42023695e-01
1.04917614e-02 2.21269980e-01 -3.50022107e-01 -5.67696095e-01
-7.75923356e-02 3.08113415e-02 -6.87384307e-01 -9.63985771e-02
6.51428819e-01 1.46524847e+00 -1.07886195e+00 -2.54758209e-01
-1.44586757e-01 1.99280351e-01 -4.95343596e-01 6.52334750e-01
-5.04849911e-01 6.31217062e-01 -4.14756387e-01 2.09584042e-01
2.28648305e-01 -4.07696545e-01 2.84252670e-02 2.36585036e-01
-1.80529431e-02 1.13298571e+00 -2.61280149e-01 2.01069140e+00
-8.58573139e-01 5.33318400e-01 -2.87232488e-01 -9.00133908e-01
1.01260006e+00 7.74960101e-01 -9.71648470e-02 -6.34819508e-01
-7.89996833e-02 3.55604917e-01 2.20988750e-01 -5.34123778e-01
7.50701189e-01 -1.90497965e-01 -3.25965345e-01 8.26944411e-01
3.40115458e-01 7.56038167e-03 3.12033147e-01 3.95922035e-01
1.03145921e+00 2.39424825e-01 3.96881104e-01 1.14616528e-01
8.43761504e-01 -1.16592854e-01 2.57132232e-01 5.39323449e-01
2.73816109e-01 7.71630287e-01 4.86382127e-01 5.02342209e-02
-1.13014102e+00 -6.60674632e-01 4.15663511e-01 1.34463644e+00
-3.29161793e-01 -6.66206121e-01 -1.05987239e+00 -7.43407488e-01
-3.52241904e-01 1.08998060e+00 -3.26207757e-01 -5.08342743e-01
-8.40008080e-01 -7.91380405e-01 7.81819105e-01 4.57838297e-01
5.62303603e-01 -1.55136299e+00 1.27442375e-01 6.17466629e-01
-8.10944855e-01 -1.11785150e+00 -7.42518008e-01 -2.04276785e-01
-8.29654098e-01 -4.45892781e-01 -9.01129723e-01 -8.51244986e-01
7.22887993e-01 1.24101222e-01 1.28054714e+00 -4.74362075e-02
3.27511460e-01 -2.71021813e-01 -6.53221250e-01 -8.47109258e-02
-1.07179308e+00 6.92311764e-01 -2.39289954e-01 1.79337012e-03
1.39020666e-01 -4.30934757e-01 -4.44591790e-01 6.12055734e-02
-5.82753479e-01 7.06017911e-01 1.11882615e+00 1.11232710e+00
3.18810403e-01 -7.70142674e-01 1.29035521e+00 -8.89555633e-01
1.05007398e+00 -6.40455663e-01 5.20337885e-03 4.31408107e-01
-3.19038987e-01 3.49748492e-01 1.05035853e+00 -3.02995265e-01
-1.19224286e+00 -3.40063006e-01 -4.37275827e-01 1.41786233e-01
-6.78538084e-02 7.32702434e-01 -3.13474149e-01 7.47320890e-01
7.06372321e-01 6.89456522e-01 -1.70142338e-01 -5.74149132e-01
6.25262082e-01 1.20670319e+00 5.74616253e-01 -5.27325392e-01
4.69400734e-01 -2.74101317e-01 -6.92432523e-01 -4.54958320e-01
-9.87132549e-01 -1.18854620e-01 -5.70662498e-01 2.65261620e-01
8.28000247e-01 -1.20878446e+00 -1.94407016e-01 3.13056767e-01
-1.75864768e+00 -4.84398723e-01 -1.10749930e-01 4.32311714e-01
-6.57731652e-01 3.11322659e-01 -9.11203325e-01 -6.42988086e-01
-1.04089928e+00 -1.03282273e+00 1.30644512e+00 -3.21939662e-02
-3.08919966e-01 -8.14524293e-01 4.56043035e-02 6.03587151e-01
5.06464541e-01 -2.37375200e-01 8.49524736e-01 -7.67861605e-01
-4.03599322e-01 5.49593233e-02 -1.68800816e-01 5.67184746e-01
1.50402367e-01 -3.81984740e-01 -7.65844166e-01 -3.34407210e-01
-5.23796044e-02 -6.07981324e-01 1.07079983e+00 -5.05458564e-03
8.58030021e-01 -7.24496305e-01 -8.04589763e-02 3.60170931e-01
9.26143885e-01 -6.43351600e-02 7.60226190e-01 -4.92624082e-02
6.46157742e-01 5.28788090e-01 4.29654092e-01 2.04274073e-01
6.17258608e-01 5.70800304e-01 -3.01136132e-02 6.59346730e-02
-2.83753753e-01 -7.10364580e-01 1.06431699e+00 1.47304654e+00
-2.09908299e-02 -6.04335964e-01 -6.11479580e-01 5.03698051e-01
-1.94645953e+00 -9.05242383e-01 -4.56718653e-02 2.09556532e+00
1.47816527e+00 4.86251935e-02 6.20363222e-04 -3.89735669e-01
8.48413765e-01 2.43231148e-01 -3.85433793e-01 -6.80782437e-01
-8.83549899e-02 1.78337246e-01 1.23435974e-01 7.21622944e-01
-7.59874105e-01 1.50529647e+00 5.56974697e+00 9.97677565e-01
-1.16950357e+00 3.12476039e-01 8.60876560e-01 5.18534929e-02
-5.10960460e-01 -3.71816866e-02 -1.04435408e+00 6.74584627e-01
1.59875464e+00 -5.61206639e-01 5.86844385e-01 6.04145646e-01
4.08317387e-01 2.64563173e-01 -1.29218793e+00 6.57641530e-01
3.14250916e-01 -1.31965816e+00 3.90771210e-01 -2.21366763e-01
9.78712797e-01 2.98814297e-01 -3.06115568e-01 7.77847707e-01
4.19106334e-01 -1.23945105e+00 5.52130938e-01 4.64432746e-01
1.23344707e+00 -6.03908598e-01 8.22130799e-01 9.95994806e-01
-8.27686489e-01 2.28587344e-01 -5.43204367e-01 -2.78596282e-01
6.09063506e-01 6.30751669e-01 -1.31712437e+00 8.17960680e-01
-2.47775152e-01 7.97230482e-01 -2.00651795e-01 3.83825332e-01
-7.74824381e-01 9.91127431e-01 2.11517423e-01 -2.21282721e-01
2.58807361e-01 -1.99953213e-01 4.83608782e-01 1.56790495e+00
4.04034376e-01 -8.03837925e-02 1.05483882e-01 9.12789047e-01
-6.80074155e-01 2.61498868e-01 -4.89654869e-01 -3.32894623e-01
3.57521653e-01 1.07957864e+00 1.36432536e-02 -7.59294748e-01
-2.86517859e-01 1.37259018e+00 6.07015610e-01 3.90994728e-01
-6.83569551e-01 -4.93686050e-01 3.58169764e-01 -8.45101252e-02
1.98335558e-01 -1.74971208e-01 -1.65718824e-01 -1.49442530e+00
1.58669874e-02 -1.06720114e+00 -6.25765920e-02 -8.24392200e-01
-1.15472662e+00 9.87712801e-01 -3.38462532e-01 -1.09342933e+00
-1.08148885e+00 -1.31490946e-01 -8.01581621e-01 1.26726913e+00
-1.55911613e+00 -1.34025359e+00 2.29190901e-01 3.01685274e-01
1.09601188e+00 -3.90121400e-01 1.01608098e+00 1.93515658e-01
-6.06861174e-01 7.41294980e-01 5.76993413e-02 2.66131669e-01
8.43363643e-01 -1.08973527e+00 1.08045125e+00 9.82218683e-01
1.61256626e-01 6.16775811e-01 1.93710789e-01 -7.02942610e-01
-1.29751921e+00 -1.43747580e+00 1.70432079e+00 -3.07220280e-01
5.33971250e-01 -6.96152031e-01 -7.33273983e-01 7.83199072e-01
6.69150054e-01 -6.10222399e-01 4.86910313e-01 -1.49068147e-01
-6.09293170e-02 1.95503131e-01 -5.61984658e-01 6.93170667e-01
1.11587167e+00 -6.47226632e-01 -8.22869599e-01 5.87978363e-01
1.31026292e+00 -4.45874512e-01 -5.12815654e-01 2.39129379e-01
1.40498519e-01 -5.59710562e-01 6.28131032e-01 -9.06264484e-01
1.10954523e+00 1.12223186e-01 9.43815336e-02 -1.73707139e+00
-3.15167487e-01 -1.25406957e+00 -2.80798197e-01 1.46662652e+00
8.91035557e-01 -5.05315304e-01 3.42545360e-01 4.00516391e-02
-7.14009464e-01 -8.73317778e-01 -6.61203921e-01 -5.40876985e-01
3.72688174e-01 -2.06802875e-01 6.69468641e-01 4.65263575e-01
2.61896253e-01 1.13247871e+00 -6.68947518e-01 -3.72293264e-01
-2.38427650e-02 1.56558797e-01 1.01167941e+00 -6.31089926e-01
-4.92174059e-01 -3.32619905e-01 4.23202515e-01 -1.74278533e+00
5.41796207e-01 -1.32240081e+00 3.56159478e-01 -1.76034832e+00
2.31772959e-01 4.01561819e-02 4.51396070e-02 4.98950899e-01
-6.39718592e-01 4.77627032e-02 8.05431902e-02 3.20836425e-01
-6.82051957e-01 9.54271436e-01 1.55838025e+00 -4.74311039e-02
-1.39089525e-01 1.13797128e-01 -1.13960421e+00 2.49414474e-01
7.62006879e-01 -4.06590253e-01 -2.74900109e-01 -8.26917171e-01
2.32585952e-01 5.10274589e-01 -9.67726409e-02 -5.23363471e-01
2.49332227e-02 -2.29404438e-02 2.36172471e-02 -5.13305902e-01
5.55919707e-02 7.09289834e-02 -3.16758811e-01 3.94260526e-01
-7.51431286e-01 1.98058605e-01 -1.65760387e-02 2.69248813e-01
-4.15770859e-01 -2.98853517e-01 4.29441094e-01 -2.78991789e-01
-2.54348963e-01 1.59879237e-01 -3.87997389e-01 4.27222282e-01
4.73450005e-01 2.17403740e-01 -5.04745066e-01 -6.48991823e-01
-4.64363247e-01 2.86171108e-01 1.18309684e-01 5.95541596e-01
5.46627164e-01 -1.24473274e+00 -1.46750188e+00 4.05791968e-01
-6.11709356e-02 1.45851359e-01 1.51695594e-01 9.04974997e-01
-2.36470982e-01 7.31914878e-01 9.04206038e-02 -3.20329666e-01
-8.77576351e-01 2.13694021e-01 1.97146386e-02 -8.01644027e-01
-4.05389696e-01 8.50748360e-01 3.14858258e-01 -4.66664940e-01
-2.54552543e-01 -2.05930814e-01 -1.12847667e-02 -1.82736129e-01
7.04987705e-01 1.79645553e-01 1.02326781e-01 -4.91899192e-01
-5.59111573e-02 9.15104151e-03 -4.10774976e-01 -3.45537812e-01
1.19885480e+00 -2.03568742e-01 -3.11136335e-01 3.05764407e-01
1.30273116e+00 2.78009232e-02 -1.00827551e+00 -2.87997633e-01
-7.06069320e-02 3.58338058e-02 -4.38423932e-01 -1.03020620e+00
-5.58050990e-01 1.18731964e+00 -4.77253705e-01 -1.09266445e-01
8.82181287e-01 3.83260846e-02 1.30304480e+00 4.26377445e-01
2.07096800e-01 -9.14041221e-01 1.62646279e-01 1.17361391e+00
1.31846070e+00 -1.15337515e+00 -4.66756672e-01 -1.58968419e-01
-9.38228667e-01 1.00368881e+00 5.65113902e-01 -1.80298150e-01
3.01610511e-02 5.81121482e-02 -6.25074059e-02 4.36651379e-01
-1.42454290e+00 7.59576680e-03 4.80291754e-01 4.05572504e-01
1.04051077e+00 1.85536787e-01 -5.91636240e-01 8.98714125e-01
-7.21256554e-01 9.13801640e-02 4.38621491e-01 2.79141009e-01
-4.70290035e-01 -1.28291821e+00 7.95442015e-02 3.29021305e-01
-5.19108653e-01 -6.52746141e-01 -6.00371301e-01 2.99762577e-01
-3.28644395e-01 1.20008457e+00 -1.48534076e-02 -4.17598635e-01
6.48534819e-02 2.87158370e-01 2.86063612e-01 -1.10298884e+00
-8.39166641e-01 1.97241172e-01 6.10317111e-01 -2.25554034e-01
-8.34540203e-02 -3.77116770e-01 -1.24470174e+00 -1.27639666e-01
-3.55045229e-01 4.62708086e-01 3.94765258e-01 1.07978725e+00
7.49542832e-01 6.33548617e-01 9.12771642e-01 -7.35519469e-01
-9.70367312e-01 -1.78645349e+00 8.28683376e-02 3.14774185e-01
2.63420910e-01 9.58304927e-02 -1.29902452e-01 2.07241818e-01]
|
[11.834897994995117, 9.170478820800781]
|
0fb46715-0d8c-4efe-8950-518787c3219c
|
on-the-benefits-of-3d-pose-and-tracking-for
|
2304.01199
| null |
https://arxiv.org/abs/2304.01199v1
|
https://arxiv.org/pdf/2304.01199v1.pdf
|
On the Benefits of 3D Pose and Tracking for Human Action Recognition
|
In this work we study the benefits of using tracking and 3D poses for action recognition. To achieve this, we take the Lagrangian view on analysing actions over a trajectory of human motion rather than at a fixed point in space. Taking this stand allows us to use the tracklets of people to predict their actions. In this spirit, first we show the benefits of using 3D pose to infer actions, and study person-person interactions. Subsequently, we propose a Lagrangian Action Recognition model by fusing 3D pose and contextualized appearance over tracklets. To this end, our method achieves state-of-the-art performance on the AVA v2.2 dataset on both pose only settings and on standard benchmark settings. When reasoning about the action using only pose cues, our pose model achieves +10.0 mAP gain over the corresponding state-of-the-art while our fused model has a gain of +2.8 mAP over the best state-of-the-art model. Code and results are available at: https://brjathu.github.io/LART
|
['Jitendra Malik', 'Christoph Feichtenhofer', 'Angjoo Kanazawa', 'Georgios Pavlakos', 'Jathushan Rajasegaran']
|
2023-04-03
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Rajasegaran_On_the_Benefits_of_3D_Pose_and_Tracking_for_Human_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Rajasegaran_On_the_Benefits_of_3D_Pose_and_Tracking_for_Human_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['action-recognition-in-videos']
|
['computer-vision']
|
[ 3.78910489e-02 6.47315830e-02 -7.82418028e-02 -1.45553559e-01
-6.82854593e-01 -5.39399028e-01 9.58049834e-01 -1.37867257e-01
-5.13036132e-01 4.60658669e-01 3.96195889e-01 2.62800783e-01
1.59870416e-01 -4.60309178e-01 -6.87483788e-01 -6.07720733e-01
-2.09195882e-01 5.83596706e-01 4.00281101e-01 -1.31980121e-01
-1.37039810e-01 5.00043154e-01 -1.35363066e+00 1.62815779e-01
2.91504174e-01 8.42551470e-01 -3.38616788e-01 7.55868435e-01
3.84343237e-01 7.18802452e-01 -2.33684272e-01 -6.26079619e-01
6.43758714e-01 -3.70143086e-01 -7.94874609e-01 3.97206455e-01
9.38055813e-01 -4.96867895e-01 -6.48895323e-01 7.53550887e-01
4.28950429e-01 2.28919998e-01 3.63984972e-01 -1.34319651e+00
-2.12166160e-02 2.01078936e-01 -5.38670421e-01 1.22653067e-01
8.59319031e-01 3.42324615e-01 9.17976022e-01 -6.32789433e-01
6.83145463e-01 1.45342934e+00 6.42831862e-01 6.85681641e-01
-1.21522510e+00 -2.73198128e-01 6.26521230e-01 2.12073356e-01
-1.37699199e+00 -5.50880134e-01 6.28191948e-01 -6.07583404e-01
8.51341128e-01 2.93160468e-01 9.15255368e-01 1.39923894e+00
6.81349486e-02 1.24815595e+00 9.17768717e-01 -3.19434851e-01
-4.50367033e-02 -1.63904414e-01 1.57329425e-01 7.85409451e-01
1.11257151e-01 6.78777546e-02 -5.84459424e-01 -9.68059245e-03
7.00509548e-01 2.05549568e-01 -2.56408930e-01 -6.16877377e-01
-1.23528099e+00 4.01505083e-01 4.29839224e-01 4.10689227e-02
-5.44362068e-01 5.58058798e-01 2.52611071e-01 7.56715164e-02
6.88239396e-01 5.07670715e-02 -4.42778319e-01 -5.90471566e-01
-6.71058118e-01 5.47964633e-01 5.22205532e-01 6.60814464e-01
1.88469321e-01 -2.74800867e-01 -5.42469382e-01 3.87565017e-01
3.91688704e-01 6.61664963e-01 -1.93426728e-01 -1.15176225e+00
4.89833742e-01 6.82857931e-01 3.97506177e-01 -8.10328543e-01
-4.65681434e-01 -3.01252663e-01 -4.38423157e-01 2.97589362e-01
8.88738096e-01 -1.23368055e-01 -8.73419166e-01 1.77951062e+00
5.74644089e-01 3.83754492e-01 -2.21185446e-01 9.52837467e-01
3.70765477e-01 2.60648042e-01 -6.76478539e-03 -1.16563410e-01
1.42003083e+00 -1.11770952e+00 -5.68634868e-01 -3.98952454e-01
7.20467627e-01 -3.52825820e-01 6.65554643e-01 4.70233649e-01
-1.03864753e+00 -7.09197342e-01 -7.01247811e-01 1.49696752e-01
-1.71396405e-01 3.33087564e-01 4.66802478e-01 6.57332540e-01
-9.97395992e-01 6.54303491e-01 -1.53151631e+00 -7.77200103e-01
4.14382219e-01 2.64701426e-01 -4.09759313e-01 -4.34505418e-02
-8.54916573e-01 9.62545931e-01 -6.34207129e-02 -8.73199701e-02
-8.15189183e-01 -6.17360771e-01 -7.06060290e-01 -3.86085838e-01
7.31639922e-01 -8.51146042e-01 1.34591866e+00 -5.07997751e-01
-1.29345727e+00 7.94201195e-01 -3.82417619e-01 -6.64855123e-01
1.36047423e+00 -9.72961009e-01 -1.66027099e-01 3.27096552e-01
3.64937335e-02 6.21415913e-01 5.48274219e-01 -9.90435421e-01
-6.69416964e-01 -7.12356150e-01 4.42043275e-01 2.97356904e-01
-6.17380477e-02 2.54384149e-02 -9.45179343e-01 -3.83285254e-01
-1.10355169e-01 -1.48896348e+00 -2.08485320e-01 4.88164991e-01
-2.25337952e-01 -2.30971858e-01 6.45094335e-01 -7.78348446e-01
9.81834590e-01 -1.96921098e+00 4.26469386e-01 -1.23894043e-01
2.54473031e-01 3.04241329e-01 5.06533533e-02 4.68677312e-01
2.47242838e-01 -1.49414316e-01 1.69157386e-02 -9.86517966e-01
1.69096008e-01 4.81954925e-02 7.09001422e-02 8.92713845e-01
-1.51202930e-02 1.04970980e+00 -8.31427574e-01 -3.41470957e-01
5.40925264e-01 7.14598656e-01 -5.39760709e-01 -2.04059724e-02
-1.87181249e-01 9.22490478e-01 -7.20198452e-01 5.64191163e-01
5.46515524e-01 -2.49841869e-01 1.27347961e-01 -4.61115018e-02
-4.99490127e-02 1.36734888e-01 -1.29404235e+00 1.90577304e+00
-1.57322332e-01 4.56646800e-01 -6.37788773e-02 -5.08164406e-01
4.28856105e-01 3.52642626e-01 8.55331838e-01 -3.74574184e-01
2.24892810e-01 -1.76600620e-01 -1.35529861e-01 -3.19716603e-01
9.27672088e-02 1.05407506e-01 -3.36831212e-02 3.02028835e-01
-1.47823364e-01 5.59260666e-01 1.42888501e-01 1.73836604e-01
1.30304170e+00 6.89182997e-01 2.00078532e-01 -3.30839381e-02
6.58240795e-01 -2.33055592e-01 3.75494868e-01 7.29862034e-01
-4.86612707e-01 4.71484363e-01 2.80881315e-01 -5.92391253e-01
-7.41322219e-01 -1.09871006e+00 1.39637692e-02 1.07906389e+00
9.94979069e-02 -6.81246340e-01 -6.44383669e-01 -8.92967165e-01
1.22628085e-01 5.89598000e-01 -8.90259862e-01 4.24735658e-02
-7.31964588e-01 -3.43000263e-01 4.73856658e-01 7.75032699e-01
7.11899340e-01 -7.27149248e-01 -8.57423484e-01 -9.49589908e-02
-3.06321800e-01 -1.41685116e+00 -4.43940908e-01 -5.21957278e-01
-5.97189426e-01 -1.07736063e+00 -7.76147068e-01 -1.45588238e-02
4.28980231e-01 1.60943016e-01 9.56524611e-01 1.43983932e-02
-1.26996920e-01 1.02309382e+00 -6.20023131e-01 -2.62099266e-01
3.76854800e-02 -1.41058326e-01 3.81837010e-01 3.69757533e-01
3.49041224e-01 -5.97356617e-01 -7.58848369e-01 3.11822742e-01
-3.02790880e-01 1.00575410e-01 3.88397515e-01 2.08227307e-01
4.23078895e-01 -2.80933589e-01 -1.31653070e-01 -4.44550633e-01
-1.51468933e-01 -2.21178308e-02 -4.61490989e-01 2.86532760e-01
-2.76242465e-01 2.65900288e-02 2.20008820e-01 -3.44608337e-01
-9.84959364e-01 5.47429800e-01 7.78740570e-02 -5.95798671e-01
-4.05179292e-01 -6.58733100e-02 -3.18847477e-01 9.14373025e-02
4.66778249e-01 1.40348315e-01 3.50960381e-02 -7.26750851e-01
3.47279310e-01 3.53118181e-01 3.93551707e-01 -5.91711521e-01
7.65153706e-01 9.79254067e-01 2.19637543e-01 -6.81381345e-01
-1.05720866e+00 -6.07068539e-01 -1.16799688e+00 -7.01139688e-01
1.21416771e+00 -1.03559983e+00 -1.00488603e+00 6.18782461e-01
-1.08086979e+00 -4.37455922e-01 -3.23269404e-02 5.23981094e-01
-7.45205641e-01 4.58190233e-01 -3.89536381e-01 -1.23312247e+00
-8.89350921e-02 -9.86564159e-01 1.54199159e+00 -1.37870491e-01
-2.93205202e-01 -1.02102470e+00 2.34103277e-01 6.84123099e-01
-8.24061036e-02 6.50210857e-01 -2.56978184e-01 -5.11114359e-01
-5.64270079e-01 -4.14624929e-01 1.91416979e-01 -1.19258631e-02
1.32323235e-01 -4.19962972e-01 -1.03707922e+00 -5.25749862e-01
-2.67298698e-01 -9.15578157e-02 1.01524496e+00 3.00569177e-01
7.54737973e-01 -1.29752770e-01 -7.03788400e-01 4.37064022e-01
1.00982618e+00 -6.79968670e-02 4.43147421e-01 3.09459239e-01
9.56223071e-01 6.35353208e-01 8.04894865e-01 6.76861525e-01
5.17130554e-01 1.35112810e+00 3.60654980e-01 2.09694490e-01
-2.66527414e-01 -3.34244341e-01 5.67338645e-01 -6.10615313e-03
-5.49724102e-01 -2.04570130e-01 -1.10379195e+00 4.72314745e-01
-2.32082939e+00 -1.13914609e+00 -3.04683506e-01 2.33528996e+00
3.87468040e-01 2.59582669e-01 7.22230494e-01 4.25976049e-03
5.80382645e-01 3.61655593e-01 -3.96009624e-01 3.04296196e-01
3.38641375e-01 -1.69162512e-01 6.20183766e-01 6.26971483e-01
-1.57170606e+00 8.84558558e-01 4.97375917e+00 4.37937856e-01
-8.13755155e-01 4.03544232e-02 1.80727467e-01 -5.50940096e-01
3.58654678e-01 -3.66720334e-02 -1.01639795e+00 3.03621709e-01
7.70349920e-01 1.86965406e-01 3.38763148e-01 5.85062563e-01
4.44432974e-01 -3.13667543e-02 -1.41760242e+00 9.90274251e-01
1.12050354e-01 -8.14260185e-01 -2.65375763e-01 4.19531107e-01
4.33419317e-01 -5.35364896e-02 -2.49515055e-03 2.59602249e-01
2.67549217e-01 -7.59844482e-01 8.87197554e-01 8.45223606e-01
4.62654531e-01 -4.85399216e-01 5.88550806e-01 4.91511583e-01
-1.42404413e+00 1.20822396e-02 1.49445549e-01 -3.67562205e-01
6.15157008e-01 1.61298901e-01 -5.50869882e-01 7.77030408e-01
7.30767787e-01 1.19948804e+00 -7.76570559e-01 9.55921888e-01
-2.52045929e-01 4.49229598e-01 -4.13739592e-01 2.23982632e-01
1.73369750e-01 -8.75979885e-02 7.59843290e-01 1.14249814e+00
1.38828471e-01 2.17009649e-01 7.04097867e-01 4.45741713e-01
3.15750271e-01 -2.15488091e-01 -5.76094925e-01 1.58744782e-01
1.04939654e-01 9.97153759e-01 -5.20584583e-01 -3.03989172e-01
-4.39177305e-01 1.21516490e+00 2.08337277e-01 1.46134689e-01
-1.07658064e+00 3.76440465e-01 9.86406267e-01 4.50423390e-01
5.39826930e-01 -3.89608681e-01 6.88283145e-02 -1.28113508e+00
3.67094517e-01 -7.09222972e-01 4.18274730e-01 -5.59942424e-01
-1.05491805e+00 3.13880235e-01 3.98524374e-01 -1.55107844e+00
-2.65486777e-01 -6.84182286e-01 -3.51664394e-01 6.60516918e-01
-8.20668876e-01 -1.54299176e+00 -3.43625486e-01 5.05318105e-01
5.82471550e-01 1.31885514e-01 5.72755516e-01 1.94295466e-01
-4.85615820e-01 4.97199595e-01 -3.19864303e-01 4.44033355e-01
6.23306751e-01 -1.14437473e+00 6.77916527e-01 1.00438595e+00
3.21937203e-01 5.08778393e-01 8.43067944e-01 -8.00467730e-01
-1.50022030e+00 -1.04855466e+00 7.67285287e-01 -1.36166453e+00
6.49053514e-01 -4.26577747e-01 -4.72217590e-01 1.03558028e+00
-1.62355021e-01 2.13099569e-01 3.11594546e-01 2.36257374e-01
-1.80911481e-01 5.55056483e-02 -1.00757694e+00 6.84859335e-01
1.65619302e+00 -2.99933702e-01 -4.66997057e-01 3.04983169e-01
4.58827674e-01 -5.32166839e-01 -8.92295241e-01 3.73258412e-01
9.15648699e-01 -9.12836015e-01 1.23047066e+00 -7.10321426e-01
8.20536315e-02 -4.85134542e-01 -3.25033814e-01 -9.60063040e-01
-4.02848661e-01 -5.78585088e-01 -4.12450403e-01 9.77776766e-01
1.25616446e-01 -4.86803114e-01 9.62239802e-01 7.35274255e-01
7.96704218e-02 -6.89092338e-01 -9.93525088e-01 -1.04013586e+00
-1.51370496e-01 -6.42772734e-01 1.05253302e-01 5.03538251e-01
-1.43488988e-01 1.49741843e-01 -7.11328626e-01 3.68274242e-01
7.71661818e-01 -1.08443066e-01 1.12329292e+00 -1.24425805e+00
-6.17911458e-01 -2.90932119e-01 -7.57569790e-01 -1.41439342e+00
1.84563652e-01 -6.12199903e-01 -1.63755447e-01 -1.67632353e+00
3.25291842e-01 1.28335297e-01 -1.45077407e-01 6.86016977e-01
-1.63162768e-01 2.80277282e-01 7.90068507e-01 2.37730950e-01
-1.02793801e+00 6.23756468e-01 1.22876084e+00 -1.18774749e-01
-1.14761479e-01 2.19398022e-01 -3.25970411e-01 8.17634761e-01
8.06078196e-01 -3.51536900e-01 -2.49414399e-01 -2.94397324e-01
-1.35833025e-01 7.99443945e-02 9.76792216e-01 -1.28949082e+00
1.09617211e-01 -2.90194172e-02 4.42915857e-01 -5.78986585e-01
1.03755164e+00 -8.18434060e-01 1.74957648e-01 7.68282413e-01
-3.25490266e-01 -1.41657755e-01 1.74835443e-01 8.63469303e-01
2.75695235e-01 3.28997701e-01 3.90677452e-01 -2.06698567e-01
-6.99039698e-01 3.10480714e-01 -1.20208740e-01 -4.46170680e-02
1.12627769e+00 -3.14523548e-01 -1.75505996e-01 -6.50434375e-01
-1.03523457e+00 3.95269066e-01 6.24832273e-01 5.83842337e-01
3.11202139e-01 -1.41478872e+00 -8.88276339e-01 -5.34322038e-02
1.79828972e-01 -4.62867677e-01 3.34174007e-01 1.40455127e+00
-2.17529565e-01 5.85260868e-01 -1.33149892e-01 -8.08860898e-01
-1.61746192e+00 4.50649709e-01 4.25666004e-01 -3.47201258e-01
-8.01379502e-01 6.88891172e-01 2.55022824e-01 -4.19293970e-01
2.94182241e-01 -7.18401596e-02 -1.46035284e-01 -1.40491843e-01
6.59298539e-01 7.05035448e-01 -2.34577149e-01 -1.00796461e+00
-8.74684870e-01 7.98437119e-01 6.85539842e-02 -4.28836912e-01
1.19500649e+00 -1.28601819e-01 4.14995641e-01 5.47895491e-01
1.09554446e+00 1.68759823e-01 -1.77363682e+00 -2.39335477e-01
-2.23158404e-01 -6.84842646e-01 -1.97425842e-01 -8.87053311e-01
-9.34176147e-01 6.59075201e-01 7.90931642e-01 6.62255138e-02
8.55045617e-01 4.17392522e-01 4.89487112e-01 3.58001411e-01
6.49484277e-01 -7.56104350e-01 3.55055630e-02 3.84153128e-01
9.46216524e-01 -1.30196965e+00 2.67697811e-01 -5.06029427e-01
-8.11807394e-01 7.42149174e-01 5.49035668e-01 -2.15493873e-01
6.02093458e-01 -1.95464045e-01 9.60051268e-03 -2.40970865e-01
-6.31961942e-01 -6.33287966e-01 5.98117769e-01 4.77851838e-01
3.37940782e-01 1.97262585e-01 -1.41791672e-01 2.47984692e-01
-1.66005030e-01 6.11024722e-02 1.88896611e-01 1.07694721e+00
-2.17483521e-01 -1.17775750e+00 -3.83356333e-01 7.01747462e-02
-4.90384728e-01 2.99542487e-01 -7.11243033e-01 9.68598902e-01
1.01164728e-01 1.03378177e+00 -1.98337678e-02 -5.57921529e-01
5.75447440e-01 1.57583013e-01 7.75318801e-01 -3.85288328e-01
-3.02023232e-01 8.10337961e-02 4.00930822e-01 -1.04019046e+00
-6.98949873e-01 -1.23169243e+00 -1.03235936e+00 -3.32718760e-01
2.25734234e-01 -2.95376182e-01 2.00002551e-01 9.89961088e-01
3.89089346e-01 3.76080751e-01 2.68582106e-01 -1.13859463e+00
-5.49965084e-01 -9.68069136e-01 -2.88599610e-01 5.91929972e-01
4.06683534e-01 -1.00650632e+00 -2.71139085e-01 8.52301270e-02]
|
[7.25062370300293, -0.5724424719810486]
|
ec1dec23-b448-48a0-bbf2-67c788686f8a
|
human-preferences-as-dueling-bandits
|
2204.10362
| null |
https://arxiv.org/abs/2204.10362v1
|
https://arxiv.org/pdf/2204.10362v1.pdf
|
Human Preferences as Dueling Bandits
|
The dramatic improvements in core information retrieval tasks engendered by neural rankers create a need for novel evaluation methods. If every ranker returns highly relevant items in the top ranks, it becomes difficult to recognize meaningful differences between them and to build reusable test collections. Several recent papers explore pairwise preference judgments as an alternative to traditional graded relevance assessments. Rather than viewing items one at a time, assessors view items side-by-side and indicate the one that provides the better response to a query, allowing fine-grained distinctions. If we employ preference judgments to identify the probably best items for each query, we can measure rankers by their ability to place these items as high as possible. We frame the problem of finding best items as a dueling bandits problem. While many papers explore dueling bandits for online ranker evaluation via interleaving, they have not been considered as a framework for offline evaluation via human preference judgments. We review the literature for possible solutions. For human preference judgments, any usable algorithm must tolerate ties, since two items may appear nearly equal to assessors, and it must minimize the number of judgments required for any specific pair, since each such comparison requires an independent assessor. Since the theoretical guarantees provided by most algorithms depend on assumptions that are not satisfied by human preference judgments, we simulate selected algorithms on representative test cases to provide insight into their practical utility. Based on these simulations, one algorithm stands out for its potential. Our simulations suggest modifications to further improve its performance. Using the modified algorithm, we collect over 10,000 preference judgments for submissions to the TREC 2021 Deep Learning Track, confirming its suitability.
|
['Pablo Castells', 'Ellen M. Voorhees', 'Nick Craswell', 'Charles L. A. Clarke', 'Chengxi Luo', 'Xinyi Yan']
|
2022-04-21
| null | null | null | null |
['online-ranker-evaluation']
|
['miscellaneous']
|
[ 3.21239941e-02 -2.78824985e-01 -5.48989952e-01 -7.19997823e-01
-1.47563374e+00 -1.15625656e+00 2.70771325e-01 5.01805007e-01
-8.80974114e-01 7.67492712e-01 2.59942651e-01 -6.49310112e-01
-8.80574703e-01 -6.27642274e-01 -5.37500143e-01 -5.19550383e-01
-3.35718691e-01 1.09829938e+00 2.98921108e-01 -1.84535626e-02
4.26096469e-01 2.11527601e-01 -1.75146997e+00 7.15575695e-01
8.11783135e-01 1.10687089e+00 -3.14578647e-03 6.92731261e-01
7.84242377e-02 1.95631355e-01 -7.06241310e-01 -6.78120732e-01
5.92766404e-01 -2.40985453e-01 -9.45336521e-01 -3.99796218e-01
6.98595583e-01 -6.85763240e-01 3.52868326e-02 1.05700588e+00
5.76507270e-01 3.94518971e-01 7.85863221e-01 -1.05319870e+00
-5.19143820e-01 8.98337781e-01 -3.75121415e-01 3.12737018e-01
4.64548558e-01 -2.95240253e-01 1.95524383e+00 -6.49904966e-01
2.00700849e-01 1.14262712e+00 3.46010864e-01 2.63799667e-01
-1.26847827e+00 -6.25715196e-01 3.32936049e-01 3.98262173e-01
-9.44130957e-01 -3.22137475e-01 3.80493999e-01 -1.05809003e-01
5.13856888e-01 7.67525733e-01 6.96341217e-01 6.83804691e-01
-3.01425934e-01 1.04006410e+00 1.18327177e+00 -4.06075954e-01
5.43640435e-01 -4.57809716e-02 5.37067294e-01 2.94280142e-01
5.64137459e-01 2.00527608e-01 -7.35130191e-01 -6.08426273e-01
3.73916388e-01 -1.26468852e-01 -1.26925975e-01 -5.49199224e-01
-1.05163252e+00 7.62523472e-01 4.54127222e-01 -1.25595570e-01
-4.12659973e-01 2.09024146e-01 1.70973420e-01 8.27142417e-01
3.36279780e-01 9.19775069e-01 -4.37988460e-01 -1.87989086e-01
-1.25641763e+00 6.43624544e-01 9.05824065e-01 4.48660076e-01
5.96322417e-01 -7.37558961e-01 -6.78365409e-01 1.09576416e+00
3.22504073e-01 3.98150653e-01 4.83502448e-01 -1.10918534e+00
4.00305748e-01 3.79437327e-01 5.19294679e-01 -8.39507401e-01
-1.02681480e-01 -5.64396918e-01 -2.60181993e-01 2.48564169e-01
5.41452706e-01 9.20799300e-02 -8.51189971e-01 1.65380037e+00
3.33349034e-02 -4.41674590e-01 -3.51029992e-01 1.16571009e+00
2.53216743e-01 3.61581326e-01 -9.06807482e-02 -1.45659193e-01
1.28729939e+00 -7.19145834e-01 -1.17733657e-01 -1.44196704e-01
5.12952209e-01 -7.93039024e-01 1.19340301e+00 7.74366379e-01
-1.36162341e+00 -1.07114948e-01 -1.10767293e+00 3.07827085e-01
-1.53344363e-01 -3.15918654e-01 7.63337553e-01 8.21604669e-01
-1.24786651e+00 6.91303611e-01 -2.85426706e-01 7.07114190e-02
1.78630259e-02 6.77972853e-01 -1.73148699e-02 -1.52844891e-01
-1.47884715e+00 7.95623302e-01 -1.69914868e-02 2.03490108e-01
-8.75757456e-01 -2.82903641e-01 -2.23527655e-01 4.06967223e-01
3.68574411e-01 -5.69339335e-01 1.73866904e+00 -9.20395851e-01
-1.00141001e+00 8.05577815e-01 -2.22993314e-01 -4.40953612e-01
7.62243986e-01 -2.42192522e-01 -1.32730544e-01 -4.23835553e-02
1.63945019e-01 4.58093733e-01 2.22879067e-01 -9.34744060e-01
-1.14250290e+00 -3.01413476e-01 5.98822713e-01 6.01184905e-01
-4.58304644e-01 2.00050980e-01 -6.38133287e-01 -3.73433948e-01
2.84455359e-01 -9.63148713e-01 -2.95484126e-01 -1.36304468e-01
-3.61146666e-02 -5.03425658e-01 -2.07191542e-01 -1.55949205e-01
1.29638731e+00 -1.75086975e+00 -3.61070156e-01 9.19911981e-01
2.43419111e-01 3.38219665e-02 -5.32378018e-01 3.74960303e-01
1.71097070e-01 3.18332195e-01 2.43705615e-01 -5.48197664e-02
5.21804750e-01 -5.80273755e-02 -2.62140542e-01 3.42096001e-01
-4.36142027e-01 7.97565162e-01 -9.71211553e-01 -3.11795235e-01
-2.77879298e-01 -2.01765940e-01 -7.32039452e-01 7.31855258e-02
-5.06124422e-02 -5.79407215e-01 -6.33078933e-01 6.31055057e-01
3.91654432e-01 -3.59189928e-01 2.36657247e-01 4.76633012e-02
2.02307075e-01 6.46452665e-01 -1.39996064e+00 8.81608844e-01
-2.33830810e-01 4.29338157e-01 5.32743335e-02 -1.05014801e+00
7.81555891e-01 2.62437731e-01 3.89561564e-01 -1.00954545e+00
-2.86778599e-01 6.42700315e-01 2.56176770e-01 -3.52469273e-02
6.25763953e-01 -2.46868934e-02 -1.24663942e-01 8.89163613e-01
-4.69478697e-01 1.36888057e-01 6.56510055e-01 2.45135427e-01
1.11562610e+00 -3.44729811e-01 -1.70212463e-02 -4.44707692e-01
1.15495205e-01 1.69752333e-02 4.24454749e-01 1.50759149e+00
-2.67212927e-01 5.20785570e-01 3.76518548e-01 -6.61296964e-01
-8.45070601e-01 -1.30078053e+00 -1.19382098e-01 1.88903594e+00
2.45239854e-01 -1.09320335e-01 -3.89209509e-01 -6.82933271e-01
2.90447026e-01 3.47126693e-01 -5.59568763e-01 1.17437348e-01
-2.02587187e-01 -7.11177826e-01 4.02599812e-01 4.88821030e-01
4.63292077e-02 -9.52628374e-01 -6.02569878e-01 1.74721107e-01
-2.30260581e-01 -2.09723160e-01 -6.27410412e-01 5.15815794e-01
-7.19419897e-01 -1.19763076e+00 -1.09523296e+00 -7.11263657e-01
5.28745353e-01 5.21349907e-01 1.46829724e+00 2.17349634e-01
1.03485569e-01 4.20982748e-01 -3.13282549e-01 -1.40655741e-01
1.41929403e-01 6.63346052e-02 2.79045343e-01 -2.77460009e-01
6.23273134e-01 -2.78548956e-01 -1.01741552e+00 6.97279632e-01
-6.85592473e-01 -4.91327822e-01 7.98075080e-01 9.54918504e-01
5.72882235e-01 -1.83952987e-01 5.20079911e-01 -8.89445364e-01
1.35257840e+00 -4.03657645e-01 -7.63020933e-01 6.62815273e-01
-1.01736307e+00 3.61280888e-01 2.17130870e-01 -4.51179862e-01
-6.51029348e-01 -3.19147736e-01 2.87975222e-01 -9.29858070e-03
3.07822585e-01 9.81068909e-01 2.34319538e-01 1.30707204e-01
8.66257131e-01 -5.33494689e-02 -3.46291482e-01 -3.83376062e-01
1.87393636e-01 7.41362631e-01 2.93701857e-01 -1.16722405e+00
5.21970153e-01 -1.00641981e-01 -3.22139651e-01 -1.05699077e-01
-1.06663632e+00 -7.77285099e-01 6.34216443e-02 -1.84448719e-01
3.53412986e-01 -7.48220980e-01 -1.07530963e+00 -9.97109488e-02
-8.16971779e-01 -2.58513927e-01 -5.66816069e-02 5.49947739e-01
-4.21141744e-01 1.75883472e-01 -5.20878375e-01 -8.88830960e-01
-3.20524454e-01 -1.32543766e+00 7.49355376e-01 9.16702002e-02
-6.63773894e-01 -7.31647313e-01 2.64391273e-01 5.10694265e-01
3.92183334e-01 -5.85532129e-01 9.47742343e-01 -1.25325143e+00
-6.33958161e-01 -4.61675018e-01 -1.96278974e-01 1.61764488e-01
-1.38235211e-01 -3.05233687e-01 -7.89552271e-01 -5.52647650e-01
-4.27600384e-01 -6.43171906e-01 9.17955995e-01 4.61873084e-01
1.25898778e+00 -2.96032161e-01 -1.37208685e-01 2.76971847e-01
1.02627242e+00 2.98543513e-01 3.62445533e-01 6.76183343e-01
-1.81752481e-02 6.62563920e-01 7.49692798e-01 3.27330619e-01
1.20719701e-01 6.91005945e-01 2.27860600e-01 1.30216867e-01
4.63207722e-01 -1.11534484e-01 1.74612492e-01 5.25809646e-01
-4.89881374e-02 -4.94128615e-01 -8.98487091e-01 7.57820010e-01
-1.82399368e+00 -8.06558907e-01 1.75685748e-01 2.88863826e+00
9.25367475e-01 3.94170165e-01 1.42311722e-01 1.91737618e-02
7.62303233e-01 -3.52316946e-02 -6.83029234e-01 -4.33507323e-01
-6.51609898e-02 1.83747604e-01 5.03251791e-01 6.74659789e-01
-8.64064336e-01 3.62249941e-01 7.15792465e+00 9.29863572e-01
-7.40831435e-01 -3.96939397e-01 1.03777421e+00 -7.35549629e-01
-8.39026749e-01 4.21778709e-02 -7.68789470e-01 3.90405416e-01
7.30642736e-01 -5.44913590e-01 5.98227441e-01 7.45205700e-01
4.16544676e-02 -2.07107812e-01 -1.48725498e+00 8.80450010e-01
-2.52631038e-01 -1.01742947e+00 5.23095056e-02 2.51480967e-01
7.99679697e-01 4.83014435e-02 3.13220143e-01 3.73251438e-01
1.03580201e+00 -1.04149032e+00 7.10735261e-01 1.96542189e-01
5.46586752e-01 -8.47828567e-01 8.83730531e-01 1.49196818e-01
-7.89252222e-01 -1.00853950e-01 -6.18276834e-01 -1.95730969e-01
-1.58059195e-01 6.40373588e-01 -8.18148017e-01 1.16778374e-01
6.48128748e-01 1.63998883e-02 -5.26675224e-01 1.61665797e+00
-2.20253225e-02 7.52567351e-01 -5.83470881e-01 -5.23032606e-01
4.61933494e-01 4.13453951e-02 2.38337383e-01 9.48401809e-01
5.05663693e-01 -1.67719815e-02 2.25852504e-01 3.90591651e-01
-2.35770553e-01 2.18748838e-01 -7.35524744e-02 5.23921177e-02
9.82564211e-01 9.64230835e-01 -5.59575558e-01 -4.08132643e-01
-1.59115329e-01 5.99904180e-01 3.61069173e-01 4.97606575e-01
-4.75873023e-01 -3.35415542e-01 5.50448418e-01 1.99223328e-02
-1.94334965e-02 3.40357989e-01 -2.61475265e-01 -7.06013262e-01
3.10792983e-01 -1.24921989e+00 9.13475454e-01 -7.07073033e-01
-1.53412485e+00 5.35484254e-01 1.39851660e-01 -1.11952496e+00
-5.46781898e-01 -7.52560437e-01 -4.61723179e-01 1.09643340e+00
-1.22168958e+00 -3.70223463e-01 5.96839525e-02 3.91882539e-01
3.43517393e-01 -1.81233056e-03 6.55649543e-01 1.75665393e-01
1.33237327e-02 9.97685492e-01 4.94214356e-01 -2.02052757e-01
1.10512483e+00 -1.38751137e+00 7.01898709e-02 4.67381567e-01
4.88568842e-01 1.02168286e+00 7.57328987e-01 -3.91604215e-01
-9.41994786e-01 -3.98572981e-01 1.08726776e+00 -4.51081604e-01
5.88624716e-01 2.15862896e-02 -7.23575830e-01 2.14673296e-01
-3.43302786e-02 -3.54919523e-01 8.51032436e-01 9.68801200e-01
-4.72007871e-01 -4.08431411e-01 -7.84090877e-01 7.15879142e-01
8.33625376e-01 -5.95430672e-01 -7.00716794e-01 1.50566667e-01
2.79193848e-01 9.72970854e-03 -5.17139316e-01 3.75064611e-01
1.15325570e+00 -1.10440695e+00 9.30929482e-01 -6.72872245e-01
4.07969892e-01 -1.72450140e-01 -2.87016839e-01 -1.43240809e+00
-5.76657176e-01 -4.81314182e-01 1.86553732e-01 8.38129461e-01
8.93855274e-01 -5.31026661e-01 1.19612908e+00 1.21198809e+00
6.38425052e-02 -9.04392838e-01 -7.43047953e-01 -7.36704469e-01
1.50552765e-01 -3.74153197e-01 8.10873330e-01 6.50289834e-01
2.30936989e-01 1.38682693e-01 -2.16624767e-01 -1.30826205e-01
5.66939056e-01 4.45702642e-01 4.19407576e-01 -1.52356160e+00
-7.14316249e-01 -9.89025652e-01 3.70434448e-02 -1.43891823e+00
-2.71638066e-01 -7.70260990e-01 3.48842859e-01 -1.50968850e+00
6.56916559e-01 -7.47802854e-01 -1.18977606e+00 4.58123386e-01
-4.46195543e-01 3.64214271e-01 1.01102673e-01 3.09982479e-01
-1.11066008e+00 -5.20996302e-02 9.11003411e-01 -2.18281060e-01
-4.83175628e-02 3.96092266e-01 -1.12733126e+00 5.29181123e-01
6.62102759e-01 -4.05160427e-01 -5.12798011e-01 -6.30778134e-01
9.05600727e-01 1.16290256e-01 9.01627317e-02 -6.81217492e-01
5.93559504e-01 -3.05248499e-01 4.30275500e-01 -6.61390722e-01
1.51951075e-01 -6.33499563e-01 -1.15302108e-01 2.60388315e-01
-1.11504829e+00 3.52472931e-01 -1.33590728e-01 4.27838206e-01
-3.54638457e-01 -5.33560157e-01 3.94177496e-01 -1.44295573e-01
-4.85096335e-01 2.21320897e-01 -3.25168759e-01 1.12911209e-01
4.36657608e-01 -2.64475137e-01 -2.96443820e-01 -7.54723728e-01
-5.65339625e-01 5.88491380e-01 4.35365498e-01 7.89175257e-02
4.48849410e-01 -1.26463675e+00 -7.44532585e-01 8.37593805e-03
2.92921096e-01 -3.15924168e-01 -8.43909383e-03 3.52317274e-01
-1.58919379e-01 6.45821452e-01 -5.33844084e-02 -2.68575072e-01
-1.47444415e+00 2.88793266e-01 1.29230157e-01 -6.48469925e-01
2.50936389e-01 1.10447097e+00 3.06277931e-01 -2.53339469e-01
6.23198807e-01 -7.20315874e-02 -8.55886042e-02 3.00539225e-01
7.42824256e-01 4.31719333e-01 3.57845128e-01 1.80238426e-01
-2.94808865e-01 2.00633854e-01 -3.52295965e-01 -6.35282040e-01
1.14647412e+00 -2.89708413e-02 3.81687097e-03 2.24276721e-01
8.49315584e-01 1.50606662e-01 -1.03295040e+00 -2.66531914e-01
2.47054681e-01 -6.13087535e-01 6.38724193e-02 -1.22248602e+00
-6.73158407e-01 6.68455124e-01 5.02522349e-01 5.01035392e-01
1.08110380e+00 -1.85264975e-01 2.98070580e-01 9.46040332e-01
5.50572276e-01 -1.20693970e+00 -1.53065929e-02 5.01351595e-01
7.05174148e-01 -1.17621708e+00 2.28430238e-02 2.05302328e-01
-4.69203919e-01 8.27718258e-01 4.08577800e-01 -7.34462067e-02
2.70450264e-01 -2.13602200e-01 1.40994087e-01 -1.11693203e-01
-1.15325677e+00 -1.17446303e-01 7.05934286e-01 8.21072906e-02
6.86263859e-01 3.16600680e-01 -5.86269140e-01 5.06026030e-01
-3.56090099e-01 -1.08872607e-01 1.93939686e-01 5.64993382e-01
-7.94396758e-01 -1.35081279e+00 -3.83976966e-01 1.01284635e+00
-5.94617128e-01 -4.16659623e-01 -4.48921174e-01 2.54137874e-01
-4.58130330e-01 1.04812479e+00 1.61181927e-01 -3.10902804e-01
2.21637905e-01 -3.32216620e-02 4.21404511e-01 -4.54763591e-01
-5.70919812e-01 2.65349057e-02 3.74715686e-01 -5.20213246e-01
-5.89014180e-02 -5.55351496e-01 -6.24462664e-01 -4.78955239e-01
-2.37782136e-01 7.40833700e-01 3.46974462e-01 8.18748295e-01
8.23304653e-02 3.94387692e-02 5.41250050e-01 -5.27981460e-01
-1.29144669e+00 -7.33826816e-01 -7.04010129e-01 5.46876669e-01
2.87700564e-01 -6.43134534e-01 -3.50656956e-01 -4.48739082e-01]
|
[11.456686973571777, 7.5400776863098145]
|
091b5ebd-1753-4766-aa42-7b6b2f26a7e9
|
analysis-of-oversampling-in-uplink-massive
|
2306.17697
| null |
https://arxiv.org/abs/2306.17697v1
|
https://arxiv.org/pdf/2306.17697v1.pdf
|
Analysis of Oversampling in Uplink Massive MIMO-OFDM with Low-Resolution ADCs
|
Low-resolution analog-to-digital converters (ADCs) have emerged as an efficient solution for massive multiple-input multiple-output (MIMO) systems to reap high data rates with reasonable power consumption and hardware complexity. In this paper, we analyze the performance of oversampling in uplink massive MIMO orthogonal frequency-division multiplexing (MIMO-OFDM) systems with low-resolution ADCs. Considering both the temporal and spatial correlation of the quantization distortion, we derive an approximate closed-form expression of an achievable sum rate, which reveals how the oversampling ratio (OSR), the ADC resolution, and the signal-to-noise ratio (SNR) jointly affect the system performance. In particular, we demonstrate that oversampling can effectively improve the sum rate by mitigating the impact of the quantization distortion, especially at high SNR and with very low ADC resolution. Furthermore, we show that the considered low-resolution massive MIMO-OFDM system can achieve the same performance as the unquantized one when both the SNR and the OSR are sufficiently high. Numerical simulations confirm our analysis.
|
['Markku Juntti', 'Italo Atzeni', 'Nhan Thanh Nguyen', 'Mengyuan Ma']
|
2023-06-30
| null | null | null | null |
['quantization']
|
['methodology']
|
[ 3.53292495e-01 -2.55780756e-01 -4.25427139e-01 3.26071292e-01
-7.74338305e-01 -5.58598578e-01 3.78957480e-01 2.26435825e-01
-3.97999734e-01 9.87074196e-01 7.04605430e-02 -5.80115736e-01
-4.73223627e-01 -6.56686783e-01 -3.05017799e-01 -7.24493623e-01
-2.29633793e-01 -5.30339122e-01 -2.83485055e-01 -1.75260648e-01
-6.27537891e-02 3.99379343e-01 -1.14110434e+00 -2.37418637e-01
8.64441216e-01 1.47377336e+00 1.14418000e-01 9.60235953e-01
6.01835668e-01 7.64525831e-01 -8.45988393e-01 -7.82044157e-02
2.89206386e-01 -3.88928115e-01 5.09504005e-02 2.41279621e-02
1.65527165e-01 -7.45660901e-01 -7.88784921e-01 1.19817555e+00
6.28061652e-01 -2.03137353e-01 7.60915399e-01 -7.42531598e-01
-2.93709725e-01 4.37119663e-01 -5.60118139e-01 4.80627775e-01
1.65148359e-02 -3.08880538e-01 5.18629253e-01 -9.33015049e-01
1.77039057e-01 9.73515093e-01 2.15415791e-01 -2.06552342e-01
-1.05409157e+00 -7.38122940e-01 -5.70685983e-01 3.26122463e-01
-1.98939180e+00 -5.18091977e-01 3.99762988e-01 -2.62586296e-01
2.04662293e-01 3.26847821e-01 1.90880910e-01 2.55525827e-01
3.09584886e-01 -7.30147287e-02 9.03905869e-01 -8.15989852e-01
3.50215256e-01 1.71733230e-01 -4.11466330e-01 5.07164776e-01
7.94057488e-01 5.91428876e-02 -2.35635638e-01 -3.19969118e-01
1.27915776e+00 -2.86902636e-01 -4.94455814e-01 -1.77936494e-01
-1.21759605e+00 4.54102784e-01 3.47120911e-01 5.41192114e-01
-5.63550830e-01 4.14368302e-01 -1.21168047e-01 4.44706351e-01
-2.08381966e-01 1.18154734e-01 1.89006180e-01 1.12963915e-01
-9.58585501e-01 -1.63495481e-01 4.26871836e-01 1.01040483e+00
3.82179230e-01 4.75538552e-01 -2.73328841e-01 2.74280101e-01
4.00935933e-02 7.27381647e-01 3.99531871e-01 -8.95450056e-01
8.07775319e-01 -3.36084396e-01 4.74281579e-01 -7.42109537e-01
-1.30808845e-01 -1.12857139e+00 -1.33681321e+00 1.11346636e-02
4.48239326e-01 -5.80522358e-01 -1.74898724e-03 1.38156021e+00
-1.92966089e-01 -3.61191034e-02 5.32549798e-01 9.65864837e-01
-1.00033648e-01 5.58336616e-01 -1.19437486e-01 -6.90927207e-01
1.44037449e+00 -2.98046947e-01 -1.27626419e+00 -1.11050799e-01
1.12498485e-01 -8.70122075e-01 4.25594091e-01 2.77727515e-01
-1.07120252e+00 -5.07188678e-01 -1.79745734e+00 3.34947079e-01
3.63334179e-01 7.59382784e-01 8.39753002e-02 9.55746055e-01
-6.03722990e-01 5.77015042e-01 -2.92496681e-01 3.22385043e-01
1.69274688e-01 2.98135906e-01 9.76919010e-02 6.37837276e-02
-1.30181003e+00 6.55723333e-01 2.59191155e-01 -1.98049210e-02
-1.13003060e-01 -6.11955583e-01 -2.43224397e-01 5.14627397e-01
2.32749403e-01 -7.71885037e-01 1.19952977e+00 -6.14663661e-01
-1.42122996e+00 -3.01568866e-01 8.04950297e-03 -6.97562397e-01
4.72792208e-01 -1.74426958e-02 -6.33532047e-01 7.22677529e-01
-3.64673376e-01 -2.31174439e-01 1.01073802e+00 -6.26075268e-01
-5.63234270e-01 -3.75404894e-01 6.92997500e-03 2.19129980e-01
-6.55773103e-01 -4.28315043e-01 1.31903231e-01 -6.58897102e-01
1.42179474e-01 -5.44681787e-01 -3.27362239e-01 1.20649755e-01
1.31107703e-01 6.17904544e-01 5.52584648e-01 -4.43301231e-01
1.66100860e+00 -2.42610860e+00 3.50794643e-02 5.39552942e-02
-7.72803873e-02 4.94212329e-01 4.07917827e-01 3.50763381e-01
3.99160147e-01 -8.29535425e-02 1.03109553e-01 3.16979855e-01
-3.92394096e-01 -2.62982279e-01 -1.72045559e-01 6.44361854e-01
3.03124279e-01 3.37416202e-01 -8.41321349e-01 -2.40176558e-01
3.90896589e-01 5.26445627e-01 -2.05133557e-01 -1.32852718e-01
3.17342997e-01 3.68127137e-01 -2.69082218e-01 3.99268121e-01
6.85830057e-01 -3.26644778e-01 7.08553076e-01 -5.66813290e-01
-4.59302366e-01 -7.45319650e-02 -1.52793074e+00 8.48798811e-01
-8.38139176e-01 9.02857423e-01 5.09232521e-01 -4.93910611e-01
7.47138202e-01 7.46932030e-01 -3.65253538e-02 -1.00113451e+00
2.48815030e-01 6.74497426e-01 3.73065919e-02 -5.79259917e-02
5.76902211e-01 -5.36019623e-01 -2.09421352e-01 -1.28148153e-01
-3.31218578e-02 1.01321638e-01 3.03813726e-01 8.35140571e-02
5.45624495e-01 -5.57597756e-01 1.21169937e+00 -5.23881972e-01
8.81714761e-01 -7.83182502e-01 2.75260955e-01 3.76723021e-01
-8.03733524e-03 3.59157808e-02 3.90736789e-01 3.67016375e-01
-1.34349644e+00 -1.02895534e+00 -4.88893479e-01 1.93013981e-01
5.73859334e-01 -1.78781316e-01 -2.81540424e-01 5.34022391e-01
1.53235374e-02 5.25989294e-01 2.14867368e-01 9.54214931e-02
-3.24740142e-01 -6.88705206e-01 5.30960858e-01 1.54513612e-01
8.60954702e-01 5.22996962e-01 -5.75021029e-01 5.11166453e-01
-2.66787107e-03 -1.62210834e+00 -1.16459958e-01 -9.70826671e-03
-7.40817904e-01 -6.04150414e-01 -1.00123215e+00 -3.84873271e-01
4.58919495e-01 4.45298254e-01 1.99580401e-01 -4.38488454e-01
9.10032094e-02 2.00654700e-01 4.51061949e-02 -1.32092506e-01
-3.82541627e-01 -1.77106559e-01 3.42424870e-01 2.63531595e-01
-3.42279613e-01 -6.55534208e-01 -6.16243541e-01 3.82848710e-01
-6.70041263e-01 -2.36882865e-01 8.24959457e-01 3.41226310e-01
2.70208210e-01 3.87624770e-01 9.18100297e-01 -1.79935675e-02
6.28669739e-01 -3.18304747e-01 -1.16240180e+00 3.62628475e-02
-3.33047152e-01 2.98309058e-01 1.27360964e+00 -3.39588434e-01
-6.91115022e-01 -5.65117933e-02 -8.20985511e-02 -1.54983908e-01
3.35896909e-01 1.75690427e-01 -1.90774187e-01 -4.52352107e-01
7.64145315e-01 1.75494805e-01 -2.39939302e-01 -2.95509756e-01
3.67826998e-01 1.19074535e+00 7.58210063e-01 -1.47593230e-01
1.10521591e+00 4.59275752e-01 8.38740647e-01 -1.37717271e+00
-3.87988597e-01 -1.71854034e-01 -4.05316472e-01 -3.47694904e-02
5.73894791e-02 -1.54151034e+00 -5.44273734e-01 2.57688854e-02
-8.39526653e-01 1.44631013e-01 6.47455305e-02 1.26476598e+00
-3.53718489e-01 5.72436512e-01 -4.41543579e-01 -1.15659142e+00
-3.55553240e-01 -7.46737838e-01 3.39895219e-01 3.54269594e-01
1.02925591e-01 -5.51919341e-01 -7.24734247e-01 -2.30999336e-01
8.14857185e-01 4.42178696e-01 8.41239870e-01 3.27930182e-01
-8.55344415e-01 -6.95831105e-02 -3.27566296e-01 1.41077727e-01
-7.48323053e-02 -5.43192565e-01 -6.43338978e-01 -4.11187887e-01
-3.11851059e-03 3.34004521e-01 2.29408950e-01 2.31953472e-01
5.91288924e-01 -5.54916739e-01 -1.92648441e-01 4.91013557e-01
1.90431547e+00 -9.02850404e-02 6.40547872e-01 -1.97771758e-01
2.08045654e-02 -2.64922172e-01 7.25528896e-01 1.04485035e+00
-1.45855471e-01 8.60517144e-01 2.62387961e-01 2.87141562e-01
2.84345504e-02 1.12938344e-01 2.28989840e-01 7.03987360e-01
-5.68113513e-02 -2.59291679e-01 -7.66420886e-02 1.65681228e-01
-1.25863266e+00 -7.86435306e-01 -3.74202788e-01 2.77618194e+00
5.35958290e-01 -1.04661593e-02 -6.70686141e-02 7.57679641e-01
8.24897110e-01 -2.79259030e-02 -1.73675746e-01 -1.85329258e-01
-2.47598961e-01 7.70403817e-02 1.34425676e+00 5.79075456e-01
-8.50731790e-01 -1.08073339e-01 5.28597498e+00 1.14230752e+00
-1.25034666e+00 1.77249566e-01 1.93077758e-01 -2.26226285e-01
-1.09690325e-02 -2.97745883e-01 -6.71065092e-01 6.08230472e-01
1.47161615e+00 -7.31688559e-01 3.21731001e-01 6.10725164e-01
3.69515806e-01 -1.69397518e-01 -6.73470318e-01 1.01967371e+00
-4.27671671e-01 -1.36385632e+00 -1.00891199e-02 4.14193600e-01
5.45123994e-01 -5.62832594e-01 2.04832464e-01 -3.04779666e-03
-7.22737491e-01 -5.42774498e-01 6.87556684e-01 3.60242367e-01
1.39731419e+00 -8.77731442e-01 7.30624139e-01 4.65283036e-01
-1.59340000e+00 -7.17202425e-01 -4.52278972e-01 -3.28842103e-01
2.61647314e-01 1.03048348e+00 -5.30855656e-01 7.99107194e-01
-3.31021160e-01 -2.75500983e-01 -1.70234870e-02 1.14828360e+00
-2.46312097e-01 2.29515478e-01 -4.14104491e-01 -1.86973229e-01
1.77547038e-01 -1.55829683e-01 5.99046111e-01 1.04404521e+00
9.98508215e-01 6.24564886e-01 -1.51816785e-01 3.61998558e-01
-2.67223626e-01 -9.93724912e-02 -1.75394699e-01 -4.59489040e-02
1.04429030e+00 1.00425315e+00 -1.84045076e-01 -4.33803439e-01
-4.32369351e-01 6.68243110e-01 -4.52779919e-01 3.10534805e-01
-5.99352241e-01 -8.70142221e-01 5.96233010e-01 4.02050495e-01
4.65964705e-01 -8.33465576e-01 -2.22787291e-01 -7.96629488e-01
1.08392701e-01 -4.66042221e-01 6.68230504e-02 -3.15762788e-01
-2.89944649e-01 6.21146709e-02 -2.14688748e-01 -1.89164686e+00
-5.24818778e-01 -2.04248101e-01 1.50641829e-01 1.06057107e+00
-1.85302484e+00 -3.40392381e-01 -8.41253698e-02 3.02964300e-01
6.36969507e-02 1.02614528e-02 8.09027076e-01 9.40545321e-01
-5.58911078e-02 6.81113839e-01 7.99280405e-01 -1.74303632e-02
1.50480390e-01 -7.84964442e-01 -2.74242938e-01 9.99374092e-01
-4.04450238e-01 6.04840755e-01 7.23636448e-01 -1.02002501e-01
-1.66067421e+00 -9.77944672e-01 7.57513762e-01 6.99258089e-01
5.59846282e-01 -9.80583504e-02 -4.07275498e-01 -1.61226168e-02
1.06343172e-01 2.48156682e-01 6.98289692e-01 -7.49090850e-01
4.66088317e-02 -1.80135265e-01 -1.19781423e+00 6.63171053e-01
3.95823896e-01 -5.65829635e-01 -5.62900491e-02 -1.72951579e-01
4.37671423e-01 -4.15359437e-01 -1.21442676e+00 2.33877122e-01
7.41013646e-01 -7.83537686e-01 1.06529486e+00 3.36800307e-01
6.90387264e-02 -6.06316924e-01 -7.53496110e-01 -1.17313254e+00
-4.56528962e-01 -9.08509970e-01 -4.86180335e-01 8.04389060e-01
1.20559461e-01 -4.02964115e-01 1.19543634e-01 -2.77615845e-01
5.44215024e-01 -4.54541445e-01 -1.25990760e+00 -1.23621356e+00
-1.89845726e-01 1.77037805e-01 3.36593270e-01 2.94690311e-01
4.86654788e-01 5.61253786e-01 -5.97848713e-01 9.19433713e-01
9.11322057e-01 -1.38910070e-01 3.35448176e-01 -1.19847655e+00
-5.13492346e-01 -1.18152305e-01 -6.27830565e-01 -1.46443725e+00
-6.71146750e-01 -3.98496211e-01 -6.17954791e-01 -1.08697391e+00
-3.91248524e-01 -1.46522328e-01 -2.12372467e-01 -6.06347680e-01
3.17325257e-02 4.34005499e-01 3.12837899e-01 1.48029149e-01
-2.75144517e-01 4.22334760e-01 1.06005466e+00 3.61204177e-01
-2.67319456e-02 3.02684247e-01 -5.43179750e-01 4.66361642e-01
6.06212080e-01 2.73137838e-02 -2.89736658e-01 -4.00754586e-02
9.18693617e-02 8.69515359e-01 1.70405716e-01 -1.69176209e+00
2.04806849e-01 3.47367823e-02 4.29816008e-01 -4.17750508e-01
5.72140813e-01 -1.02149224e+00 2.18898952e-01 8.64012957e-01
-8.14213380e-02 -5.53262174e-01 -4.92417738e-02 7.05584049e-01
-1.37406528e-01 -2.10588828e-01 1.10573816e+00 2.27410436e-01
-3.91637534e-01 -1.42202795e-01 -9.46678162e-01 -3.50506574e-01
9.19475019e-01 1.22305207e-01 -4.30544078e-01 -7.11527884e-01
-4.25625265e-01 7.06302747e-03 4.15513702e-02 -3.60080861e-02
3.17366898e-01 -1.44620645e+00 -5.68515658e-01 2.53043354e-01
-2.83511132e-01 -7.17554212e-01 2.57064074e-01 8.34924936e-01
-4.43853438e-01 1.02603126e+00 -3.10027152e-01 -4.51985240e-01
-1.28981793e+00 1.51632324e-01 3.17476332e-01 -1.15933426e-01
-1.74081892e-01 1.01672545e-01 -4.22058344e-01 7.49405682e-01
-6.54454634e-04 -2.25999221e-01 1.75353885e-01 5.15190419e-03
1.13215208e+00 6.76565886e-01 1.51027262e-01 -4.58587617e-01
-2.06840724e-01 6.20422363e-01 5.79248190e-01 -2.17834383e-01
6.30985379e-01 -5.56873083e-01 4.35550958e-01 1.63228795e-01
1.37242830e+00 3.35849285e-01 -1.03697193e+00 -4.86929089e-01
-2.96536714e-01 -7.46194541e-01 3.72002631e-01 -1.29712850e-01
-4.43998575e-01 9.54128504e-01 7.39324033e-01 3.72653216e-01
1.08093238e+00 -5.53951502e-01 5.91861844e-01 3.39657962e-01
8.87663841e-01 -9.67601418e-01 7.03071244e-03 1.41734689e-01
4.70590502e-01 -4.29955363e-01 6.72150195e-01 -5.68129599e-01
-1.41058579e-01 1.44199324e+00 -2.28118479e-01 1.44319698e-01
2.79992640e-01 4.78262931e-01 -4.00242716e-01 6.26961648e-01
-3.97061616e-01 -2.29158878e-01 -3.42565626e-02 5.02827406e-01
4.51469302e-01 3.86985093e-01 -7.91879594e-01 6.09632850e-01
-1.61507875e-01 2.18383968e-01 1.14636779e+00 5.24722993e-01
-1.01295018e+00 -7.48667836e-01 -9.70151544e-01 2.75275558e-01
-7.79482782e-01 -4.82423231e-02 3.86029124e-01 4.57742602e-01
-5.87037541e-02 1.12131047e+00 1.99704587e-01 -6.53541163e-02
3.70075196e-01 -3.45380425e-01 7.59477258e-01 -7.37560838e-02
9.10502821e-02 2.53717154e-01 2.93506354e-01 -1.37633160e-02
5.53183593e-02 -2.46175185e-01 -1.23411429e+00 -4.99697626e-01
-3.71450186e-01 1.64333910e-01 8.14624667e-01 7.58179247e-01
5.26436090e-01 6.96437478e-01 1.03810537e+00 -4.00111228e-01
-1.00926721e+00 -5.96653163e-01 -1.00410438e+00 -5.11740446e-01
9.25191641e-01 -2.89202303e-01 -4.56384808e-01 -2.26443201e-01]
|
[6.383587837219238, 1.3327473402023315]
|
cf95f8ef-8c74-4728-89a9-a8cc3c8e50ba
|
agi-agent-safety-by-iteratively-improving-the
|
2007.05411
| null |
https://arxiv.org/abs/2007.05411v1
|
https://arxiv.org/pdf/2007.05411v1.pdf
|
AGI Agent Safety by Iteratively Improving the Utility Function
|
While it is still unclear if agents with Artificial General Intelligence (AGI) could ever be built, we can already use mathematical models to investigate potential safety systems for these agents. We present an AGI safety layer that creates a special dedicated input terminal to support the iterative improvement of an AGI agent's utility function. The humans who switched on the agent can use this terminal to close any loopholes that are discovered in the utility function's encoding of agent goals and constraints, to direct the agent towards new goals, or to force the agent to switch itself off. An AGI agent may develop the emergent incentive to manipulate the above utility function improvement process, for example by deceiving, restraining, or even attacking the humans involved. The safety layer will partially, and sometimes fully, suppress this dangerous incentive. The first part of this paper generalizes earlier work on AGI emergency stop buttons. We aim to make the mathematical methods used to construct the layer more accessible, by applying them to an MDP model. We discuss two provable properties of the safety layer, and show ongoing work in mapping it to a Causal Influence Diagram (CID). In the second part, we develop full mathematical proofs, and show that the safety layer creates a type of bureaucratic blindness. We then present the design of a learning agent, a design that wraps the safety layer around either a known machine learning system, or a potential future AGI-level learning system. The resulting agent will satisfy the provable safety properties from the moment it is first switched on. Finally, we show how this agent can be mapped from its model to a real-life implementation. We review the methodological issues involved in this step, and discuss how these are typically resolved.
|
['Koen Holtman']
|
2020-07-10
| null | null | null | null |
['mathematical-proofs']
|
['miscellaneous']
|
[ 2.80575097e-01 1.17986274e+00 4.01169807e-02 -9.38840676e-04
3.45372744e-02 -8.97224545e-01 8.97444963e-01 3.16499956e-02
-2.90617079e-01 8.83528829e-01 -2.91822944e-02 -8.51508796e-01
-4.68536735e-01 -9.74868715e-01 -7.42291689e-01 -6.13040864e-01
-4.22530383e-01 5.83778203e-01 2.82227308e-01 -2.45619893e-01
2.61940122e-01 2.98860848e-01 -1.45982206e+00 -1.26521811e-01
6.04970157e-01 5.02736986e-01 -1.21558733e-01 7.28035688e-01
5.69094121e-01 1.24050522e+00 -8.10729980e-01 7.50489533e-02
2.53499776e-01 -3.92182559e-01 -9.80264783e-01 -1.87029958e-01
-3.38034183e-01 -5.99612713e-01 6.17879368e-02 8.48446965e-01
2.01609787e-02 -2.62998998e-01 6.70930743e-01 -2.00036645e+00
-2.98731685e-01 7.90710986e-01 1.97741598e-01 -5.83014488e-01
5.80712199e-01 6.90179944e-01 5.24739742e-01 2.20972925e-01
6.35828733e-01 1.33068323e+00 3.53152931e-01 1.00436819e+00
-1.03193998e+00 -4.62641120e-01 3.68617773e-01 -8.18350315e-02
-9.54348624e-01 -2.30764657e-01 4.28284049e-01 -5.20996630e-01
1.20280111e+00 4.06935692e-01 7.89043307e-01 1.03709936e+00
3.91929269e-01 6.46716356e-01 1.29984593e+00 -5.51933348e-01
5.64292073e-01 3.86779964e-01 3.31843197e-01 7.48858929e-01
5.38272738e-01 8.68100464e-01 -1.68876842e-01 -5.36419153e-01
7.14641452e-01 -5.53638637e-01 -2.42034048e-01 -4.30598974e-01
-9.40537989e-01 7.47410655e-01 1.04728021e-01 2.09883526e-01
-3.97214562e-01 2.88462847e-01 1.35717764e-01 6.61165237e-01
-2.10186213e-01 8.99332583e-01 -6.24025345e-01 -8.14313367e-02
6.10164925e-02 5.35130739e-01 1.23825610e+00 6.96654916e-01
4.51684147e-01 7.95207266e-03 1.31990224e-01 -1.49345294e-01
6.06684148e-01 3.59904170e-01 -1.30403996e-01 -1.44594884e+00
-1.64886773e-01 7.26074219e-01 5.72749794e-01 -5.87339401e-01
-8.43873441e-01 -3.31948064e-02 -3.12172294e-01 1.14651847e+00
4.84601349e-01 -7.31572032e-01 -4.71334547e-01 2.05079412e+00
2.59842843e-01 -8.69704857e-02 4.49952275e-01 6.42965913e-01
1.39156818e-01 6.70066833e-01 1.58730999e-01 -4.70326245e-01
1.28417242e+00 -3.81462067e-01 -4.54782307e-01 -7.80182183e-02
8.48108232e-01 1.24940522e-01 7.87408590e-01 5.41030109e-01
-1.35792422e+00 1.85124919e-01 -1.36172163e+00 5.80843627e-01
-3.30023974e-01 -6.83156371e-01 8.58173966e-01 6.13438487e-01
-1.22029948e+00 4.52903479e-01 -8.89115274e-01 -4.27768677e-01
-3.34161781e-02 6.04660928e-01 -5.17382883e-02 4.21688497e-01
-1.52549052e+00 1.22372115e+00 4.13659573e-01 -2.19972014e-01
-1.36347830e+00 -4.96721804e-01 -8.40174198e-01 1.75334718e-02
5.90543747e-01 -1.13844407e+00 1.37336564e+00 -1.05254376e+00
-1.67468059e+00 4.68617916e-01 4.99626130e-01 -6.46785676e-01
4.81001049e-01 3.06025624e-01 7.46745244e-03 -1.58746108e-01
-3.57091948e-02 5.83667815e-01 7.41392851e-01 -1.50486267e+00
-8.61099541e-01 -5.18976986e-01 9.02514994e-01 1.06525198e-01
3.36639360e-02 1.78060398e-01 4.19514567e-01 -1.35581836e-01
-6.18892193e-01 -1.25573158e+00 -2.98133045e-01 -3.77070218e-01
-2.63635993e-01 -4.10983920e-01 6.40392840e-01 6.83922786e-03
1.09489381e+00 -1.84265995e+00 9.25102234e-02 3.10525566e-01
1.11194707e-01 2.24132821e-01 -2.47681439e-01 4.91001815e-01
-7.23733678e-02 3.72518331e-01 -2.30396181e-01 3.98355573e-01
2.70006269e-01 3.48035127e-01 -1.24830410e-01 3.71309012e-01
1.28329620e-01 6.99392915e-01 -8.88513625e-01 -2.18927190e-02
1.45009026e-01 4.19063158e-02 -7.05917239e-01 4.15984929e-01
-4.18039292e-01 4.19895321e-01 -5.86923420e-01 9.45677236e-02
2.38616079e-01 2.75837362e-01 4.71346110e-01 4.42166209e-01
-4.14161712e-01 4.92224038e-01 -1.01399302e+00 8.43482375e-01
-2.31137887e-01 1.59302399e-01 6.38614118e-01 -9.29303944e-01
3.42560381e-01 5.23043871e-01 3.10545415e-01 -3.75031799e-01
2.57280022e-01 -2.04958841e-02 3.30354899e-01 -4.86870229e-01
-6.04437217e-02 -2.99248636e-01 -4.11861718e-01 9.39882874e-01
-4.06836957e-01 -2.81470507e-01 3.10046617e-02 2.87537187e-01
1.32505131e+00 7.99337476e-02 3.28981608e-01 -5.23291171e-01
6.15702868e-01 3.17056030e-01 4.95886028e-01 9.94381070e-01
-3.21213275e-01 -3.10248017e-01 9.36477542e-01 -6.55340612e-01
-8.68971348e-01 -7.98856556e-01 2.46921986e-01 9.68279243e-01
1.28245547e-01 -3.57862949e-01 -1.05402780e+00 -7.46765971e-01
-6.54141456e-02 1.42856622e+00 -6.79541290e-01 -4.51424092e-01
-5.27826309e-01 -6.26269519e-01 6.88064575e-01 1.97682604e-01
3.22835654e-01 -1.29023731e+00 -1.42675471e+00 1.65118560e-01
-2.59478986e-02 -2.70205677e-01 1.04669638e-01 3.79001379e-01
-5.45273125e-01 -1.20516288e+00 3.60127360e-01 -4.59435433e-01
6.75162196e-01 2.12657154e-02 6.26276910e-01 4.17202622e-01
2.16865689e-01 7.12838769e-01 -1.51909396e-01 -8.24176431e-01
-1.01041639e+00 -9.35319513e-02 3.29397202e-01 -5.86960196e-01
2.12481603e-01 -4.87066090e-01 -2.88022369e-01 3.18021357e-01
-1.02226901e+00 1.95820272e-01 3.08853183e-02 6.30872726e-01
-4.24988061e-01 4.06119257e-01 6.38626039e-01 -4.85915542e-01
9.55814481e-01 -3.53824288e-01 -9.75681722e-01 2.16323748e-01
-8.40437055e-01 3.70983720e-01 5.34978390e-01 -3.18152338e-01
-1.03634274e+00 8.34169239e-03 2.92424083e-01 3.81242096e-01
-2.43188396e-01 4.53412116e-01 -4.00195926e-01 1.19030029e-01
7.10856259e-01 -2.50526100e-01 5.05539477e-01 6.32274076e-02
3.50339979e-01 6.51348472e-01 2.89598554e-01 -9.45304930e-01
9.04068351e-01 2.72456944e-01 4.59793629e-03 -2.35410839e-01
-1.53731078e-01 4.71385032e-01 -4.73405886e-03 -2.47984201e-01
5.60599923e-01 -4.69316602e-01 -1.54798615e+00 4.92869467e-01
-1.18260467e+00 -9.73296940e-01 -3.25890660e-01 8.10078159e-02
-9.01237309e-01 -8.96933898e-02 -4.94524896e-01 -1.15379798e+00
2.86456328e-02 -1.09196091e+00 4.56113279e-01 1.66537657e-01
-5.68607450e-01 -7.04900146e-01 1.84837192e-01 9.75436270e-02
3.92887384e-01 7.70805329e-02 1.24510074e+00 -7.50396192e-01
-5.27870119e-01 -5.36564039e-03 2.81474829e-01 7.60964379e-02
3.40566747e-02 1.67239815e-01 -6.66401803e-01 -3.86237174e-01
3.08821172e-01 -1.70411840e-01 7.89394826e-02 3.41081560e-01
4.37115699e-01 -1.02202773e+00 -4.34302747e-01 1.01479359e-01
9.75903153e-01 8.64784241e-01 7.41700590e-01 9.39011335e-01
4.73997556e-02 1.06430137e+00 7.00136483e-01 4.50304538e-01
5.64531088e-01 4.96273339e-01 7.99522400e-01 2.83373564e-01
3.30521375e-01 -1.42990053e-01 8.58198643e-01 1.48956617e-02
-4.35500413e-01 3.02254576e-02 -9.10549283e-01 3.43240559e-01
-2.12506819e+00 -1.02941418e+00 9.72347111e-02 2.28442097e+00
9.31107998e-01 2.62003392e-01 3.15564722e-01 9.15638823e-03
4.90547508e-01 -4.71287519e-01 -4.47983772e-01 -8.79913330e-01
3.05931300e-01 -3.53581995e-01 5.46888590e-01 1.11585212e+00
-7.62303114e-01 8.62084866e-01 7.21424150e+00 1.20715804e-01
-6.06048644e-01 -6.90502971e-02 3.89103860e-01 2.86323354e-02
-4.10648137e-01 3.84055465e-01 -4.11919981e-01 1.20233022e-01
1.13499546e+00 -5.09732604e-01 9.16327477e-01 7.66140878e-01
6.41082823e-01 -2.46513039e-01 -1.41090798e+00 7.72756264e-02
-3.17484766e-01 -9.36500907e-01 -1.12880871e-01 1.93392918e-01
2.35838905e-01 -4.02950495e-01 -2.59728581e-01 4.56738889e-01
9.57139969e-01 -1.05698705e+00 9.50641990e-01 1.12209544e-01
4.05290514e-01 -9.57179427e-01 5.67491472e-01 8.18341315e-01
-5.60696125e-01 -6.22229815e-01 2.01759815e-01 -9.01221216e-01
-3.86951603e-02 -1.68121085e-01 -9.61282074e-01 3.12209487e-01
3.77345175e-01 -2.17406973e-01 -1.98539019e-01 4.43036348e-01
-5.25313973e-01 4.43078190e-01 -5.49727798e-01 -1.85598418e-01
2.93294042e-01 -2.59834528e-01 8.43642533e-01 6.40531719e-01
2.74937861e-02 6.06843650e-01 -1.11130729e-01 1.04667604e+00
6.64090753e-01 -7.03586102e-01 -9.73954558e-01 7.16232508e-02
2.90528834e-01 9.37103331e-01 -2.99913615e-01 -4.68313515e-01
-2.31035262e-01 4.85988230e-01 3.98851037e-02 4.61964011e-01
-8.73109877e-01 -2.88592458e-01 9.40138102e-01 1.72441125e-01
-4.16713923e-01 -5.30468896e-02 -2.86501914e-01 -8.04271758e-01
-3.31780493e-01 -1.29027975e+00 3.61392766e-01 -8.36456895e-01
-8.20609093e-01 3.01128149e-01 4.10725266e-01 -5.79923451e-01
-8.39562654e-01 -4.62221354e-01 -4.98064935e-01 5.22224009e-01
-1.00081122e+00 -9.91255879e-01 1.35566831e-01 6.16331100e-01
1.11920640e-01 -7.41732121e-02 8.97194803e-01 -4.06328440e-01
-4.59658951e-01 1.89151600e-01 -5.86198449e-01 -2.13748217e-01
2.20007062e-01 -1.02907813e+00 2.22097933e-01 8.45969677e-01
-5.77755868e-01 9.51886237e-01 9.48061168e-01 -7.15459764e-01
-1.60303795e+00 -8.75205874e-01 6.16405368e-01 -7.18461812e-01
7.74989665e-01 -2.46876463e-01 -5.84572434e-01 1.09808791e+00
3.54126960e-01 -8.12061310e-01 8.80117193e-02 -2.36482490e-02
-2.07898468e-01 1.60788000e-02 -1.36561441e+00 1.09679234e+00
1.03530431e+00 -2.07237840e-01 -9.15417910e-01 7.75328428e-02
8.49804878e-01 1.86011940e-01 -4.74735141e-01 5.33939123e-01
6.50947392e-01 -8.45822096e-01 6.97151780e-01 -8.55415881e-01
-5.98127022e-02 -7.63188004e-01 3.02656382e-01 -1.49249029e+00
-5.42306840e-01 -1.28869891e+00 3.54418457e-02 9.38600183e-01
3.98840725e-01 -1.21028543e+00 3.13497245e-01 1.37698722e+00
-2.33406313e-02 -3.32138926e-01 -8.48795831e-01 -8.61447811e-01
3.26334208e-01 -4.62901980e-01 6.15609825e-01 8.35946262e-01
1.08607507e+00 3.61280620e-01 -1.18386440e-01 3.84788543e-01
6.27405345e-01 -2.58861929e-01 7.16167927e-01 -1.14944327e+00
-2.83369809e-01 -6.07007504e-01 -1.41328335e-01 -5.30573964e-01
1.33240577e-02 -6.15059197e-01 2.88121194e-01 -1.41169763e+00
1.39051884e-01 -5.23522437e-01 -1.00933082e-01 1.10385847e+00
1.03984594e-01 -6.53803527e-01 3.51663172e-01 1.07621476e-01
-2.66193062e-01 -3.25661153e-02 8.95743251e-01 -1.68046758e-01
-4.72465754e-01 -4.80400026e-03 -1.02486038e+00 1.09887230e+00
1.12317097e+00 -3.62628430e-01 -7.30942607e-01 -5.00237271e-02
5.74133277e-01 1.63076788e-01 6.24561489e-01 -7.66483963e-01
4.20461297e-01 -7.55165577e-01 -3.67217839e-01 1.65043160e-01
-2.49432251e-01 -9.90197480e-01 5.17981231e-01 1.06506240e+00
-4.93553728e-01 -1.23551100e-01 1.49011001e-01 6.65265024e-02
3.60314637e-01 -4.25473064e-01 6.58513010e-01 -1.51336804e-01
-2.61818290e-01 -3.53557289e-01 -1.30142033e+00 -4.94300961e-01
1.34418154e+00 1.33932412e-01 -6.16812587e-01 -6.09585524e-01
-4.68688548e-01 6.21371388e-01 7.91161060e-01 1.97440565e-01
2.92002082e-01 -7.32720017e-01 -5.30804515e-01 1.70204893e-01
-1.37323320e-01 -2.14469165e-01 -3.20270777e-01 7.18226135e-01
-2.98162252e-01 4.62864786e-01 -4.09800440e-01 -1.39499474e-02
-1.15980566e+00 1.13673484e+00 5.73325515e-01 -1.14807367e-01
-4.31227833e-01 2.68016547e-01 5.88646412e-01 -4.60972071e-01
3.33279699e-01 -3.44461501e-01 -2.11329415e-01 -2.14908153e-01
6.95542991e-01 1.89208210e-01 -4.28292900e-01 -8.50171596e-02
-6.65401518e-01 2.40162253e-01 2.57315665e-01 -5.56474984e-01
1.17314410e+00 -2.09763288e-01 -4.14758384e-01 7.87033960e-02
3.22666705e-01 -2.94462383e-01 -1.24027026e+00 4.92412597e-01
-1.35486647e-01 1.18884385e-01 -2.77112037e-01 -1.12191629e+00
-4.06523615e-01 3.45118403e-01 1.93194285e-01 8.78854990e-01
1.14523256e+00 -2.38755327e-02 2.14358971e-01 5.15660048e-01
9.05809283e-01 -1.14296651e+00 -3.39503199e-01 4.89359081e-01
9.28121805e-01 -6.50673687e-01 -1.46109968e-01 -2.33986333e-01
-6.55292094e-01 9.03631747e-01 4.38700974e-01 2.61896625e-02
1.87920481e-01 7.64835596e-01 -5.57285473e-02 -3.70574385e-01
-1.16678476e+00 -3.45461406e-02 -6.30718589e-01 1.08585036e+00
-2.85512537e-01 2.83104122e-01 -5.52311182e-01 6.49597526e-01
-1.30125597e-01 2.73118526e-01 1.05557942e+00 1.06960285e+00
-7.23014832e-01 -1.10061634e+00 -5.81258476e-01 -1.39461607e-01
-1.84545293e-01 1.85669035e-01 -6.63114250e-01 9.06224310e-01
2.39131704e-01 1.39988124e+00 -1.76913217e-01 -2.45584041e-01
2.78164029e-01 1.71001256e-01 5.28532147e-01 -2.07652688e-01
-6.52052224e-01 -3.14089149e-01 5.62450171e-01 -6.55156612e-01
-4.00890887e-01 -7.43538499e-01 -1.48622310e+00 -4.16333020e-01
6.11359663e-02 2.08945483e-01 6.05300546e-01 7.40613282e-01
1.51191011e-01 3.32193434e-01 6.38413489e-01 -6.25345707e-01
-5.49150586e-01 -5.89404523e-01 -1.46212190e-01 2.12005572e-03
3.58233929e-01 -5.77190220e-01 -7.51552522e-01 -2.27193516e-02]
|
[4.396336078643799, 2.0723371505737305]
|
ce3cd1e9-1111-40b8-804c-0045edaa1fcb
|
color-image-edge-detection-using-multi-scale
|
2208.07503
| null |
https://arxiv.org/abs/2208.07503v1
|
https://arxiv.org/pdf/2208.07503v1.pdf
|
Color Image Edge Detection using Multi-scale and Multi-directional Gabor filter
|
In this paper, a color edge detection method is proposed where the multi-scale Gabor filter are used to obtain edges from input color images. The main advantage of the proposed method is that high edge detection accuracy is attained while maintaining good noise robustness. The proposed method consists of three aspects: First, the RGB color image is converted to CIE L*a*b* space because of its wide coloring area and uniform color distribution. Second, a set of Gabor filters are used to smooth the input images and the color edge strength maps are extracted, which are fused into a new ESM with the noise robustness and accurate edge extraction. Third, Embedding the fused ESM in the route of the Canny detector yields a noise-robust color edge detector. The results show that the proposed detector has the better experience in detection accuracy and noise-robustness.
|
['Jinni Chen', 'Jie Ren', 'Weichuan Zhang', 'Yuandong Bi', 'Yunhong Li']
|
2022-08-16
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[-4.21941392e-02 -8.38904202e-01 7.30071962e-02 2.31023759e-01
-2.29655579e-02 -4.44342285e-01 -3.00055370e-03 -2.20350295e-01
-7.16328025e-01 4.67726111e-01 -9.29621011e-02 -2.17233270e-01
1.05467765e-02 -9.96540248e-01 -8.31602141e-02 -7.88029552e-01
8.68134722e-02 -6.85622990e-01 7.45989919e-01 -1.08912118e-01
3.78171235e-01 7.41730154e-01 -1.36059833e+00 -3.07037324e-01
8.19933712e-01 1.21184695e+00 2.22687483e-01 8.08863759e-01
-1.87536985e-01 4.54437643e-01 -3.72129738e-01 -1.34352162e-01
2.76046783e-01 -5.76512814e-01 -3.01407158e-01 3.20964009e-01
-3.92885387e-01 -4.05173153e-01 -3.33737701e-01 1.40888882e+00
3.86971682e-01 2.82384902e-01 7.10359514e-01 -1.03019512e+00
-8.01016986e-01 -2.45311499e-01 -1.10029948e+00 2.75949478e-01
1.26851514e-01 -1.60976380e-01 4.60616231e-01 -9.98514295e-01
5.13747752e-01 1.06829500e+00 4.36955869e-01 1.52991610e-02
-8.99812639e-01 -5.48363745e-01 -2.55534202e-01 4.96198028e-01
-1.62949860e+00 7.05066398e-02 9.73085999e-01 -1.21283699e-02
4.74318743e-01 4.84178156e-01 8.42671514e-01 1.90322116e-01
4.69759047e-01 5.66050231e-01 1.58648360e+00 -6.82626605e-01
8.59139338e-02 1.66655704e-01 4.54508960e-02 1.15311003e+00
6.77669346e-01 1.83574855e-01 -2.65408698e-02 2.85565257e-01
9.55529988e-01 3.41035575e-01 -3.43903214e-01 6.19381517e-02
-9.15768743e-01 4.09478217e-01 7.95240283e-01 7.36794293e-01
-4.61795390e-01 -2.25464582e-01 1.87350497e-01 -6.30650446e-02
1.44631125e-03 -1.97590828e-01 1.29065335e-01 -3.63242663e-02
-7.85795808e-01 -3.95138562e-01 4.03155714e-01 6.76726222e-01
6.90903962e-01 3.15601110e-01 1.69284970e-01 5.87589443e-01
5.59868455e-01 8.27752173e-01 4.90648419e-01 -5.43272972e-01
2.63434398e-04 7.51855671e-01 1.98888615e-01 -1.65885198e+00
-4.13960189e-01 -2.10653380e-01 -8.65706801e-01 8.43607068e-01
2.86233723e-01 -1.50721073e-01 -8.83486211e-01 1.08327651e+00
3.73839140e-01 1.81056917e-01 3.36577892e-01 9.42085683e-01
6.94122493e-01 9.55258906e-01 7.40598962e-02 -1.38777867e-01
1.77163994e+00 -6.42932653e-01 -1.18065608e+00 -3.06576908e-01
-1.94344237e-01 -1.21311462e+00 8.10110688e-01 4.15223062e-01
-8.13333452e-01 -6.54846430e-01 -1.35952687e+00 4.49573435e-02
-7.56057501e-01 7.95873344e-01 5.83871067e-01 7.07162619e-01
-7.63866246e-01 -3.12109776e-02 -7.72866249e-01 -2.92584270e-01
6.80714175e-02 3.71620730e-02 -4.83013570e-01 -1.39239177e-01
-9.56857800e-01 8.86313975e-01 7.54272103e-01 6.23628676e-01
1.80304527e-01 4.13641959e-01 -8.90512407e-01 1.40143394e-01
1.44623026e-01 -1.04189679e-01 6.36979461e-01 -9.47265446e-01
-1.58406115e+00 6.32800102e-01 -1.55410752e-01 2.76110411e-01
4.43420142e-01 1.81961417e-01 -9.38482821e-01 5.67498386e-01
-1.49864197e-01 2.25357115e-02 6.96348250e-01 -1.08066046e+00
-1.00661659e+00 -2.14998230e-01 -4.94735003e-01 1.65522456e-01
-3.92729282e-01 2.03638732e-01 -7.44385719e-01 -7.53283918e-01
6.03883922e-01 -5.82667291e-01 1.55312726e-02 -1.15590148e-01
-2.03513190e-01 1.70905039e-01 1.25653970e+00 -9.85747874e-01
1.60395002e+00 -2.43352199e+00 -2.21368864e-01 6.43326342e-01
-3.37673388e-02 3.97666454e-01 9.80598927e-02 2.03189999e-01
3.86753753e-02 -9.51982141e-02 -6.55409619e-02 3.90275329e-01
-3.16846907e-01 -2.12995142e-01 3.36208194e-01 3.78502518e-01
2.56246567e-01 5.65784276e-01 -6.85316324e-01 -7.74772704e-01
4.48897570e-01 6.57058895e-01 -3.92149836e-02 -3.22768018e-02
6.22454882e-01 5.09636141e-02 -5.68321288e-01 7.72120059e-01
1.12674594e+00 1.91353247e-01 -1.38843596e-01 -5.18934608e-01
-4.43360716e-01 -6.57169640e-01 -2.03777814e+00 9.37857211e-01
-2.31337368e-01 4.96344507e-01 2.98642427e-01 -5.77823579e-01
1.26802182e+00 1.45160705e-01 2.77624518e-01 -7.57303178e-01
4.97450769e-01 3.59742403e-01 -1.56442925e-01 -7.09408522e-01
5.28793156e-01 -5.08866161e-02 1.39171749e-01 1.19875945e-01
-3.54756653e-01 7.49209896e-02 3.80741954e-01 -1.65541157e-01
4.88391638e-01 3.93726900e-02 3.30998570e-01 -1.70759976e-01
8.04241896e-01 -7.63527527e-02 8.20354164e-01 1.74380839e-01
-1.88373983e-01 1.67061582e-01 3.87635380e-02 -1.38623461e-01
-7.94671834e-01 -8.36864352e-01 -1.61877722e-01 4.17289346e-01
6.52183950e-01 1.34147316e-01 -7.81758428e-01 -3.34005713e-01
-1.41790643e-01 3.22784632e-01 -3.09651464e-01 -2.73835033e-01
-4.78328943e-01 -5.95848441e-01 9.75934491e-02 6.24914408e-01
1.27762973e+00 -1.07195210e+00 -7.40505695e-01 2.21659571e-01
9.29694995e-02 -7.13499904e-01 -6.12593234e-01 -3.32060963e-01
-9.07170355e-01 -1.17744267e+00 -8.17090750e-01 -1.32222843e+00
9.75673854e-01 6.13770664e-01 3.31575274e-01 3.20296764e-01
-6.56274736e-01 1.37721658e-01 -5.85053921e-01 -1.51409879e-01
-9.08258557e-02 -5.52803218e-01 -1.43547371e-01 4.06177223e-01
5.29932380e-01 -1.41380087e-01 -9.35199022e-01 2.85834908e-01
-1.14041662e+00 -7.74204209e-02 1.01520061e+00 7.89672613e-01
4.59932297e-01 7.69200981e-01 1.89650819e-01 -3.07877958e-01
6.80800080e-01 1.48763195e-01 -8.20056558e-01 4.12039071e-01
-6.19551539e-01 -1.78510338e-01 6.48997068e-01 -3.54431778e-01
-1.34719336e+00 2.46631168e-02 -7.24848956e-02 2.26072773e-01
-8.24388713e-02 3.81483793e-01 -2.38051787e-01 -4.44554299e-01
3.81885856e-01 5.51712573e-01 7.44213536e-02 -3.27390313e-01
3.63008678e-01 9.68737066e-01 7.29665577e-01 -7.80762658e-02
8.78531814e-01 3.78032744e-01 1.50372654e-01 -9.25316870e-01
1.82673290e-01 -5.35982430e-01 -2.61100680e-01 -4.58795339e-01
1.13105643e+00 -5.48578382e-01 -9.96491015e-01 1.01587415e+00
-9.44214582e-01 3.32670957e-01 3.92051816e-01 9.05291200e-01
-6.27077147e-02 6.33489668e-01 -9.28035736e-01 -1.12844348e+00
-4.42694694e-01 -1.14730179e+00 4.40823555e-01 1.15346658e+00
4.53510791e-01 -9.71541226e-01 -5.78963816e-01 -3.05701256e-01
5.40004194e-01 3.19901347e-01 7.44898498e-01 8.34829137e-02
-5.44273853e-01 -6.42152727e-01 -5.73450625e-01 3.40538055e-01
5.25603473e-01 3.92435968e-01 -4.26654756e-01 -1.27204806e-01
-3.97286899e-02 3.20599169e-01 8.14662755e-01 4.04066622e-01
6.91150308e-01 -1.66713983e-01 -3.40108722e-01 5.81650138e-01
2.06278539e+00 8.67281616e-01 9.51412022e-01 6.05003595e-01
2.75747746e-01 5.55442609e-02 7.82226026e-01 2.86608189e-02
-2.88403109e-02 1.82289794e-01 5.03504723e-02 -9.17153060e-01
-1.47406310e-01 -1.54944211e-01 2.14417547e-01 7.43974626e-01
-1.86524078e-01 7.48634897e-03 -4.28656906e-01 2.75527745e-01
-1.51590085e+00 -9.08042669e-01 -2.87450224e-01 1.99674404e+00
5.95068216e-01 1.58520252e-01 -3.70514132e-02 4.57127631e-01
1.00947905e+00 -2.57009774e-01 -1.80137709e-01 -3.88821959e-01
-1.86704844e-01 3.34256172e-01 7.63598144e-01 5.16087294e-01
-1.07584631e+00 7.36678720e-01 5.91062975e+00 7.19323158e-01
-1.49301803e+00 -3.12345415e-01 4.56700027e-01 7.54916370e-01
1.98572770e-01 -1.44791886e-01 -4.33876306e-01 6.53507233e-01
-1.53218247e-02 -2.34158915e-02 3.60644400e-01 5.81731796e-01
2.82400072e-01 -6.63040638e-01 6.84934556e-02 1.21599293e+00
-1.67225197e-01 -7.32186854e-01 -3.25593472e-01 -1.49907738e-01
4.67272758e-01 -8.16016853e-01 1.62670240e-01 -1.52631342e-01
1.77645549e-01 -4.79629248e-01 5.74469566e-01 8.60319555e-01
7.38100290e-01 -1.00781250e+00 8.54502201e-01 -4.29016463e-02
-1.63594639e+00 -1.49224535e-01 -3.82960796e-01 1.20090798e-01
7.53676966e-02 4.79476482e-01 -5.38359582e-01 7.78989911e-01
6.38347805e-01 2.85308301e-01 -4.79802102e-01 1.44191885e+00
-3.92126083e-01 2.81587332e-01 -4.17847335e-01 -3.16974640e-01
2.56292254e-01 -1.02356887e+00 3.78170878e-01 1.45469153e+00
6.38611555e-01 3.53757888e-01 1.28037781e-01 7.75399745e-01
2.51732409e-01 4.96296406e-01 -3.15385610e-01 -1.60227433e-01
4.75178242e-01 1.63713276e+00 -1.37756336e+00 -4.58892792e-01
-6.26261711e-01 1.34697700e+00 -3.30345690e-01 6.25338614e-01
-6.78069651e-01 -1.22213864e+00 1.17553242e-01 -3.47309142e-01
3.61324400e-01 -3.23681355e-01 -2.18802601e-01 -8.51080477e-01
-7.21813664e-02 -6.81383252e-01 4.33071703e-01 -9.17066276e-01
-8.26095998e-01 5.17519951e-01 -5.06928086e-01 -1.30658317e+00
4.02983427e-01 -1.08094907e+00 -9.13938820e-01 1.07698107e+00
-1.44384325e+00 -1.07341111e+00 -4.35151070e-01 4.09307748e-01
9.24381316e-02 -3.12332269e-02 5.82732320e-01 3.95851165e-01
-8.75165105e-01 3.99703711e-01 4.78340715e-01 4.71758783e-01
4.44006890e-01 -1.20733976e+00 1.00659672e-02 1.54246676e+00
-1.96788058e-01 6.22420311e-01 5.72980523e-01 -7.06473947e-01
-1.21402168e+00 -5.66727281e-01 5.12104511e-01 4.35473084e-01
2.26675004e-01 9.01096761e-02 -5.96401393e-01 3.08420211e-01
1.04248002e-01 1.69590935e-02 2.77149945e-01 -7.34092414e-01
2.89724115e-02 -2.55181193e-01 -1.29393017e+00 6.15367770e-01
2.30739176e-01 -2.36922175e-01 -5.39540291e-01 -3.30533266e-01
5.49242236e-02 -3.50445181e-01 -7.37836003e-01 1.59627870e-01
7.58469403e-01 -9.03496623e-01 9.11557555e-01 -1.88253120e-01
-7.57066235e-02 -9.89349306e-01 1.58705413e-01 -1.05777454e+00
-6.27963245e-01 -1.20418571e-01 5.49570262e-01 1.30100536e+00
2.66343951e-01 -9.33289289e-01 4.31457162e-01 3.10280293e-01
2.06744686e-01 -4.79779422e-01 -4.96760905e-01 -6.40511751e-01
-6.24524117e-01 2.71240532e-01 2.43871972e-01 7.49994814e-01
6.68732869e-03 -8.49221125e-02 -1.26413241e-01 2.50272065e-01
6.30616367e-01 8.65707919e-02 2.20213994e-01 -9.76294160e-01
-2.51831058e-02 -6.34642065e-01 -5.53176701e-01 -6.22773886e-01
-5.32587707e-01 -5.34327209e-01 -2.27550641e-02 -1.73301113e+00
3.21775023e-03 -1.61382377e-01 -5.43293655e-01 1.56385601e-01
-7.10769534e-01 6.03308618e-01 9.49562490e-02 -8.09480175e-02
-2.35454500e-01 1.76926211e-01 1.24235773e+00 -5.22033870e-03
-3.53978425e-01 -1.04605416e-02 -4.83691514e-01 7.23238051e-01
6.59674346e-01 -8.43372270e-02 6.09172173e-02 -5.78053445e-02
-1.23120472e-01 -3.96359190e-02 2.11344808e-01 -9.98451293e-01
2.23212823e-01 -2.22927295e-02 8.66149306e-01 -3.77689123e-01
6.27125949e-02 -1.22462332e+00 1.30879745e-01 8.63735795e-01
4.00814205e-01 3.26174319e-01 3.44845831e-01 4.63586539e-01
-3.64097893e-01 -3.07056457e-01 9.59662080e-01 -2.45919283e-02
-1.36235499e+00 -7.94600546e-02 -3.41413260e-01 -4.85321939e-01
1.29420316e+00 -6.40622973e-01 -1.94253147e-01 -1.51668906e-01
-5.18541574e-01 -1.99667186e-01 5.39998770e-01 8.64493698e-02
8.09536219e-01 -1.63979840e+00 -4.21085745e-01 5.12100518e-01
-9.75331068e-02 -5.90608120e-01 4.04064745e-01 8.41784716e-01
-1.22878373e+00 -1.23721518e-01 -6.07034028e-01 -1.69778690e-01
-1.38170564e+00 6.92557931e-01 2.27176428e-01 1.69614896e-01
-6.58452094e-01 5.70934355e-01 -4.57700580e-01 3.42676938e-01
-3.02627627e-02 -2.05174267e-01 -3.99210691e-01 -1.07007347e-01
7.77976155e-01 7.24365950e-01 -9.89018232e-02 -7.95897007e-01
-3.98024946e-01 1.11586475e+00 3.57045591e-01 -2.67379522e-01
1.07343328e+00 -1.32866502e-01 -3.93675238e-01 -2.81089898e-02
1.20741892e+00 6.61300242e-01 -9.55644548e-01 -1.10526504e-02
-1.32825121e-01 -7.88087249e-01 3.40802073e-01 -8.59359920e-01
-1.14137745e+00 7.39211857e-01 1.03567660e+00 3.51101428e-01
1.65236318e+00 -6.56343400e-01 8.80706608e-01 -1.45212546e-01
3.99589501e-02 -1.42080462e+00 -9.86783803e-02 -3.97881418e-02
4.51629817e-01 -9.28855062e-01 -1.09072011e-02 -4.10970330e-01
-4.96457070e-01 1.53239322e+00 3.63953173e-01 -2.31317714e-01
4.03389692e-01 3.93561929e-01 3.05809736e-01 1.14342473e-01
3.20489228e-01 -6.20083213e-01 4.37785417e-01 4.71143305e-01
1.42519474e-01 -3.70236151e-02 -8.31935167e-01 5.10986686e-01
2.31868327e-01 -4.41544224e-03 4.48484600e-01 9.04915631e-01
-1.01728237e+00 -8.92283142e-01 -9.01073813e-01 9.45401788e-02
-5.59166014e-01 2.06797108e-01 -1.03091612e-01 8.26688528e-01
4.86760736e-02 1.23786449e+00 -1.06811814e-01 -4.99534130e-01
5.66051483e-01 1.15673468e-01 3.59513760e-01 2.24252537e-01
8.46585911e-03 4.65566635e-01 -4.04999614e-01 -2.21297637e-01
-1.88356563e-01 -1.29163593e-01 -1.65648329e+00 -1.44237533e-01
-5.04915833e-01 4.11149323e-01 9.39350128e-01 5.47381043e-01
9.95970815e-02 6.12271607e-01 7.53368795e-01 -4.04280543e-01
-1.34315342e-01 -7.43852615e-01 -1.05340827e+00 7.20403552e-01
2.04378471e-01 -6.52844131e-01 -4.21506703e-01 4.39493544e-02]
|
[10.91262149810791, -2.426302909851074]
|
c7661d54-ebbc-4a14-b43d-d8f963d00334
|
hyperbolic-temporal-knowledge-graph
|
2106.04311
| null |
https://arxiv.org/abs/2106.04311v1
|
https://arxiv.org/pdf/2106.04311v1.pdf
|
Hyperbolic Temporal Knowledge Graph Embeddings with Relational and Time Curvatures
|
Knowledge Graph (KG) completion has been excessively studied with a massive number of models proposed for the Link Prediction (LP) task. The main limitation of such models is their insensitivity to time. Indeed, the temporal aspect of stored facts is often ignored. To this end, more and more works consider time as a parameter to complete KGs. In this paper, we first demonstrate that, by simply increasing the number of negative samples, the recent AttH model can achieve competitive or even better performance than the state-of-the-art on Temporal KGs (TKGs), albeit its nontemporality. We further propose Hercules, a time-aware extension of AttH model, which defines the curvature of a Riemannian manifold as the product of both relation and time. Our experiments show that both Hercules and AttH achieve competitive or new state-of-the-art performances on ICEWS04 and ICEWS05-15 datasets. Therefore, one should raise awareness when learning TKGs representations to identify whether time truly boosts performances.
|
['Johannes Heinecke', 'Lina Rojas-Barahona', 'Sebastien Montella']
|
2021-06-08
| null |
https://aclanthology.org/2021.findings-acl.292
|
https://aclanthology.org/2021.findings-acl.292.pdf
|
findings-acl-2021-8
|
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'methodology']
|
[-3.45695883e-01 2.72612810e-01 -4.86821979e-01 -3.04023802e-01
-2.00310096e-01 -4.53337044e-01 7.69435644e-01 3.18346083e-01
-1.99800193e-01 6.01496994e-01 -9.95260552e-02 -4.09155667e-01
-4.68201429e-01 -1.10284138e+00 -8.86902750e-01 -5.78163266e-01
-5.20342529e-01 4.34740901e-01 5.02771914e-01 -4.21308756e-01
-1.42003044e-01 3.32550257e-01 -1.39172733e+00 -8.65941718e-02
9.81536448e-01 1.02218294e+00 -1.64717883e-01 3.63445997e-01
4.45361100e-02 7.68995941e-01 -1.85101688e-01 -8.81851435e-01
1.42926127e-01 -8.79021361e-02 -1.00127316e+00 -3.49645257e-01
2.49539629e-01 -5.84812425e-02 -8.41950893e-01 8.51379812e-01
2.02913538e-01 2.28052258e-01 4.47231352e-01 -1.58268940e+00
-7.66714156e-01 9.23617184e-01 -4.92522925e-01 4.06741589e-01
4.14363891e-02 -1.58670530e-01 1.25799930e+00 -6.56449854e-01
7.67600894e-01 1.07461786e+00 5.93364656e-01 1.65908739e-01
-1.17295349e+00 -4.63647991e-01 4.75657433e-01 1.07180643e+00
-1.53281128e+00 -3.15637365e-02 9.22712386e-01 -2.88364351e-01
9.44144070e-01 2.91663319e-01 8.50314319e-01 9.65196729e-01
-2.53538508e-02 1.05792153e+00 9.86504376e-01 -4.07376260e-01
2.04095557e-01 1.66628752e-02 4.19999719e-01 9.33939338e-01
3.59993577e-01 2.11227521e-01 -8.80019367e-01 1.57611817e-01
3.92400712e-01 -2.55953044e-01 -4.17848021e-01 -6.28666162e-01
-1.11070752e+00 7.76947021e-01 5.46351016e-01 3.39894712e-01
-8.84176418e-02 3.11159879e-01 3.04730594e-01 4.71519798e-01
6.62283719e-01 2.05264822e-01 -6.25434339e-01 -2.32575893e-01
-7.56365299e-01 1.97276935e-01 8.73610914e-01 9.64743555e-01
5.36890388e-01 -2.05877662e-01 -1.11692466e-01 5.63225627e-01
-4.63587604e-03 7.57381395e-02 1.35843381e-01 -5.71679354e-01
5.54966331e-01 6.59315884e-01 -2.72613317e-02 -1.22307694e+00
-7.26992786e-01 -8.99509549e-01 -8.03893566e-01 -2.43663222e-01
5.89162111e-01 3.57514262e-01 -6.02049649e-01 2.03578258e+00
4.31938946e-01 4.37796801e-01 -4.13094275e-02 8.51926327e-01
5.40760815e-01 5.47492325e-01 1.70747623e-01 -3.49057823e-01
1.12312627e+00 -8.91754568e-01 -8.68088663e-01 4.54264469e-02
1.00637436e+00 -2.61177182e-01 1.01219857e+00 5.14023066e-01
-9.83500957e-01 -3.40174139e-01 -1.22517562e+00 -1.65488422e-01
-8.65386069e-01 1.63596824e-01 1.31918788e+00 5.60998857e-01
-9.29008484e-01 1.12474179e+00 -9.83539283e-01 -4.44137692e-01
2.48465255e-01 2.67518044e-01 -1.85992151e-01 -2.49996006e-01
-1.69625795e+00 1.09084499e+00 4.60134506e-01 2.22249448e-01
-5.14440894e-01 -9.56641734e-01 -6.91101491e-01 9.57711637e-02
7.81822085e-01 -7.27534473e-01 8.25605452e-01 -2.42970288e-01
-1.23370361e+00 7.75523722e-01 2.15245321e-01 -8.44009399e-01
7.63881981e-01 -2.60720462e-01 -8.51618230e-01 7.25919232e-02
-2.01995492e-01 2.63620645e-01 7.10311174e-01 -8.93749237e-01
-3.28002751e-01 -5.16045034e-01 2.49255791e-01 -3.15549448e-02
-4.92921710e-01 -5.71379960e-01 -5.87794423e-01 -7.17081547e-01
1.09835923e-01 -9.40876424e-01 1.13479577e-01 -1.33747324e-01
-3.82056117e-01 -5.67581713e-01 7.39228725e-01 -7.17936277e-01
1.64030433e+00 -2.06624269e+00 5.28959811e-01 1.01628676e-01
2.34893620e-01 2.16246039e-01 -1.42031386e-01 6.17313504e-01
-1.79396912e-01 2.44022936e-01 9.01961178e-02 -3.29646677e-01
2.17445835e-01 2.59911776e-01 -4.31359172e-01 6.36254072e-01
-3.01258042e-02 1.13819993e+00 -1.03585374e+00 -4.30941939e-01
2.71269046e-02 4.06915843e-01 -2.26090148e-01 -2.37592310e-01
-3.81576329e-01 2.05525961e-02 -3.31648111e-01 5.09244621e-01
4.95911419e-01 -4.63911891e-01 3.80199939e-01 -5.16978323e-01
1.04068667e-01 4.26001310e-01 -9.26739454e-01 1.77965653e+00
-1.42189994e-01 3.90563548e-01 -5.02984464e-01 -1.06168532e+00
5.99581301e-01 1.59666762e-01 5.26944041e-01 -7.15981781e-01
-3.93804125e-02 -2.69455276e-03 -9.93214082e-03 -3.33641887e-01
6.31123483e-01 1.42607376e-01 1.77997470e-01 2.02274784e-01
-2.33494546e-02 2.25168899e-01 5.53133547e-01 6.59695804e-01
1.21184397e+00 4.03411180e-01 1.63279161e-01 -2.58303434e-01
3.65961641e-01 -3.32697183e-01 5.36360204e-01 3.78077507e-01
-1.53762415e-01 2.77622981e-04 8.14668655e-01 -3.81744951e-01
-6.89544022e-01 -1.04812145e+00 5.54964393e-02 1.01249373e+00
1.48025215e-01 -7.88635910e-01 -4.44778889e-01 -9.81292665e-01
2.24761277e-01 8.75317693e-01 -9.73610461e-01 -5.38825631e-01
-4.40657169e-01 -8.50072622e-01 4.37535733e-01 6.38604462e-01
4.69066352e-01 -5.16279280e-01 -1.79821402e-01 1.98064029e-01
-2.60551810e-01 -1.21998036e+00 -2.64038563e-01 -1.21822782e-01
-1.01110220e+00 -1.28293586e+00 -4.48062271e-01 -3.05413067e-01
3.90038460e-01 2.69212931e-01 9.65877414e-01 1.23354517e-01
-2.24787772e-01 4.73736644e-01 -6.16564214e-01 -1.91944495e-01
1.30818769e-01 3.01462561e-01 6.49096444e-02 1.93942249e-01
1.93370357e-01 -8.21265161e-01 -5.52196026e-01 3.29982817e-01
-7.18067229e-01 1.69819996e-01 4.25049245e-01 5.37144482e-01
5.25168836e-01 4.60922241e-01 7.38809943e-01 -7.50248492e-01
4.04812038e-01 -5.22984922e-01 -6.75956309e-01 5.82340240e-01
-1.05206215e+00 2.14142904e-01 4.30531174e-01 -3.95400077e-01
-7.74005055e-01 -3.82325172e-01 3.61137271e-01 -4.90649760e-01
6.23778164e-01 9.12311673e-01 -5.10156080e-02 -1.03836022e-01
4.16066706e-01 1.33408397e-01 -2.94191211e-01 -4.27118510e-01
7.02587545e-01 -1.66149423e-01 3.89764071e-01 -7.21028268e-01
8.87543380e-01 5.13059795e-01 3.81014764e-01 -5.54401696e-01
-1.10395491e+00 -4.35305327e-01 -6.58312201e-01 -3.31520438e-01
4.80257779e-01 -6.59866035e-01 -9.67518389e-01 2.43487239e-01
-8.79394233e-01 -3.20663333e-01 -3.49464774e-01 4.62797374e-01
-5.13297498e-01 5.03682494e-01 -5.17298400e-01 -8.10277700e-01
-2.12456256e-01 -5.74761927e-01 6.11319661e-01 -1.21098667e-01
-9.45737772e-03 -1.11472702e+00 -8.47205967e-02 3.54600906e-01
2.57779837e-01 3.80780935e-01 1.19323802e+00 -4.90429699e-01
-8.58280897e-01 -2.81373650e-01 -2.61734933e-01 -1.64918303e-02
-1.52439699e-01 -9.10861790e-02 -7.92247295e-01 -2.83198893e-01
-1.77233621e-01 -1.38459325e-01 1.10443604e+00 5.30292019e-02
1.31923950e+00 -3.43592852e-01 -4.55877095e-01 5.61133802e-01
1.47809970e+00 1.59767817e-03 8.59576166e-01 2.93171346e-01
7.13847816e-01 6.43041193e-01 8.24490607e-01 3.70880395e-01
8.44443977e-01 8.27954352e-01 5.51560819e-01 3.00663710e-01
-1.21713020e-01 -3.88691306e-01 2.57283896e-01 1.03294861e+00
-5.21293700e-01 -3.25662613e-01 -7.61771321e-01 6.98662102e-01
-2.17715931e+00 -9.38180685e-01 -3.83217007e-01 2.25544906e+00
7.64973879e-01 3.04803580e-01 1.75529167e-01 5.56651115e-01
3.86819482e-01 2.88448662e-01 -5.20170748e-01 2.02103723e-02
-1.62731454e-01 1.06504738e-01 6.41670465e-01 3.20359975e-01
-9.69033718e-01 9.37872350e-01 5.50480604e+00 9.23581719e-01
-7.58185327e-01 9.09951180e-02 3.44444215e-01 -7.68386722e-02
-2.42460430e-01 5.94946705e-02 -7.04111457e-01 3.05785507e-01
9.49360967e-01 -4.35823172e-01 6.72462881e-01 6.75291598e-01
-8.24741423e-02 -1.84040272e-03 -1.34112465e+00 8.81260693e-01
-7.93473329e-03 -1.20966542e+00 -1.79509982e-01 1.34641454e-01
6.26828492e-01 -1.93816185e-01 8.33215192e-02 6.21091187e-01
1.88514426e-01 -8.23154032e-01 6.69643700e-01 7.21638918e-01
4.73839641e-01 -7.08098948e-01 5.66606641e-01 1.72561169e-01
-1.43080389e+00 1.75794318e-01 -2.34940425e-01 9.36458185e-02
6.77281916e-02 7.70147562e-01 -8.16753089e-01 1.24471176e+00
6.02594376e-01 9.43240225e-01 -9.80405807e-01 8.72863412e-01
-4.01128262e-01 7.01430619e-01 -4.04575080e-01 7.22149387e-02
9.46470574e-02 -8.93906280e-02 3.59119266e-01 9.26197529e-01
2.25280449e-01 1.17474839e-01 4.97925468e-02 6.93512261e-01
-2.07812771e-01 1.73956811e-01 -5.58032274e-01 -4.71112520e-01
3.85005385e-01 1.17539096e+00 -6.24793291e-01 -2.27589145e-01
-3.05182636e-01 7.59551644e-01 6.81992590e-01 4.72232878e-01
-9.95883107e-01 -1.58425838e-01 3.67982090e-01 2.50983596e-01
4.20138896e-01 -4.69468147e-01 -2.24530533e-01 -1.18306017e+00
1.67121470e-01 -2.60761291e-01 8.50396335e-01 -7.21106589e-01
-1.28986692e+00 1.30998045e-01 6.79781213e-02 -1.09021962e+00
1.01596080e-02 -5.89262843e-01 -1.82355598e-01 4.12294745e-01
-1.64847505e+00 -1.25388134e+00 -2.05697313e-01 6.70873880e-01
1.07573224e-02 2.20589846e-01 6.56780601e-01 4.33333367e-01
-5.94744205e-01 6.52154267e-01 -7.05371425e-02 -3.13947015e-02
4.74979430e-01 -1.31298876e+00 2.36711591e-01 7.25194275e-01
1.20301098e-01 4.93642062e-01 8.05956542e-01 -6.94723070e-01
-1.84409022e+00 -1.12804174e+00 9.11597669e-01 -4.20737028e-01
1.10160112e+00 -1.65283948e-01 -9.94858503e-01 9.09474432e-01
-6.28882274e-02 2.06565544e-01 4.10855323e-01 6.51123524e-01
-6.51029885e-01 -4.30446684e-01 -9.09927666e-01 7.42943108e-01
1.42871153e+00 -4.67871785e-01 -4.31501836e-01 5.05441010e-01
9.47958529e-01 -2.18057096e-01 -1.30486667e+00 6.10845149e-01
3.91702652e-01 -8.17324579e-01 9.45804715e-01 -6.04481757e-01
2.29508430e-01 -2.98124492e-01 -1.64003670e-01 -1.20012128e+00
-2.75704056e-01 -6.83895409e-01 -8.71187508e-01 1.23286211e+00
3.35237443e-01 -6.71967566e-01 9.54301715e-01 3.53286773e-01
-8.99042934e-03 -1.01564407e+00 -9.39782321e-01 -1.38051689e+00
-2.83024777e-02 -6.28739119e-01 6.25473678e-01 1.27643728e+00
3.20232600e-01 4.51651961e-01 -5.47087133e-01 1.68342561e-01
6.87002838e-01 1.45736516e-01 5.97261667e-01 -1.44251561e+00
-2.72672027e-01 -3.64650488e-01 -5.03657579e-01 -6.66524529e-01
1.13622703e-01 -1.17642081e+00 -5.24173319e-01 -1.64611566e+00
-8.13517272e-02 -4.06589240e-01 -3.92310858e-01 5.07198215e-01
8.32129940e-02 -2.58733422e-01 2.54455134e-02 5.97366318e-02
-8.37045193e-01 8.84442508e-01 1.29478705e+00 -1.64875284e-01
-1.13853350e-01 -2.27291286e-01 -4.81983244e-01 4.45506275e-01
7.53336966e-01 -1.76341936e-01 -7.45683432e-01 -2.35551909e-01
7.09555030e-01 4.41929363e-02 5.44156492e-01 -7.44322062e-01
4.20900792e-01 -6.59175217e-02 -1.21488627e-02 -8.40080142e-01
5.57464063e-01 -6.73423350e-01 3.43663573e-01 3.90619546e-01
-1.58016309e-01 -1.01376325e-01 1.32865459e-01 1.01528347e+00
-1.46766767e-01 1.10080928e-01 3.57744426e-01 1.07874162e-01
-8.18814754e-01 5.66645086e-01 3.05830836e-01 2.02279072e-02
1.04271924e+00 8.54563564e-02 -7.02993393e-01 -2.85183907e-01
-8.71870160e-01 4.16588813e-01 1.86205488e-02 4.99645740e-01
5.19897878e-01 -1.44068289e+00 -4.34895903e-01 -1.69074625e-01
3.15178335e-01 -2.38622651e-01 4.80162174e-01 1.28639448e+00
-1.09311059e-01 6.42308831e-01 2.92229652e-01 -2.61427611e-01
-9.55339670e-01 1.07964945e+00 8.22236612e-02 -6.92566752e-01
-8.23134959e-01 6.68887734e-01 -6.17362931e-02 -1.07377671e-01
4.08646315e-01 -3.75846088e-01 -9.76511836e-02 2.49643892e-01
2.48518392e-01 7.55894244e-01 2.68916339e-01 -6.73444718e-02
-3.96138012e-01 4.41870362e-01 -3.03659439e-01 8.15729499e-02
1.41140604e+00 3.16701829e-02 1.70297343e-02 5.38247585e-01
1.09626448e+00 -2.42505774e-01 -9.50979173e-01 -4.20091510e-01
2.16883197e-01 -2.75939405e-01 5.98204136e-02 -9.91434753e-01
-1.32771957e+00 7.43999362e-01 4.12298828e-01 5.88201821e-01
1.07317889e+00 1.76072285e-01 6.29446089e-01 4.83107388e-01
6.67819083e-01 -1.23402071e+00 -1.43352170e-02 3.11117738e-01
1.03002298e+00 -1.08260572e+00 2.46781826e-01 -7.85525858e-01
-4.56343889e-01 9.65404570e-01 4.93115634e-01 7.25348815e-02
8.51979673e-01 -3.50498259e-01 -6.91326559e-01 -4.17853773e-01
-1.00983608e+00 -3.62985075e-01 4.77542371e-01 4.35026705e-01
1.71717778e-01 2.67576009e-01 -6.30840003e-01 6.98412716e-01
-4.32276607e-01 1.06851675e-01 3.53343904e-01 7.16789126e-01
-4.03280891e-02 -1.11055124e+00 7.27156699e-02 3.34689736e-01
-2.88830608e-01 -4.09029573e-02 -2.59544641e-01 1.14454043e+00
-1.31122008e-01 8.33522379e-01 -4.34765935e-01 -6.41956449e-01
4.63901192e-01 7.82307982e-02 7.59603262e-01 -1.77113846e-01
-1.41814172e-01 -3.68535370e-01 4.07555014e-01 -7.38376081e-01
-3.96721661e-01 -6.05411887e-01 -1.18166459e+00 -6.53035641e-01
-3.65891337e-01 1.09243535e-01 5.96874297e-01 8.88191760e-01
3.49700540e-01 6.44686282e-01 3.02931339e-01 -2.45557368e-01
-4.09753650e-01 -8.57814133e-01 -8.56629312e-01 3.29328805e-01
-5.92417009e-02 -1.02946579e+00 -3.55855197e-01 -1.57037631e-01]
|
[8.593626976013184, 7.901104927062988]
|
46562339-06c5-46c3-a284-ca3cd769599a
|
efficient-unsupervised-video-object
|
2211.05364
| null |
https://arxiv.org/abs/2211.05364v2
|
https://arxiv.org/pdf/2211.05364v2.pdf
|
Efficient Unsupervised Video Object Segmentation Network Based on Motion Guidance
|
Due to the problem of performance constraints of unsupervised video object detection, its large-scale application is limited. In response to this pain point, we propose another excellent method to solve this problematic point. By incorporating motion characterization in unsupervised video object detection, detection accuracy is improved while reducing the computational amount of the network. The whole network structure consists of dual-stream network, motion guidance module, and multi-scale progressive fusion module. The appearance and motion representations of the detection target are obtained through a dual-stream network. Then, the semantic features of the motion representation are obtained through the local attention mechanism in the motion guidance module to obtain the high-level semantic features of the appearance representation. The multi-scale progressive fusion module then fuses the features of different deep semantic features in the dual-stream network further to improve the detection effect of the overall network. We have conducted numerous experiments on the three datasets of DAVIS 16, FBMS, and ViSal. The verification results show that the proposed method achieves superior accuracy and performance and proves the superiority and robustness of the algorithm.
|
['Liqiang Zhu', 'Chao Hu']
|
2022-11-10
| null | null | null | null |
['video-object-segmentation', 'unsupervised-video-object-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 1.18253238e-01 -5.55433273e-01 -1.82171538e-01 -1.12720281e-01
-3.08995485e-01 -1.77663323e-02 3.13291699e-01 -4.00513411e-01
-5.41919768e-01 1.58220068e-01 4.56808209e-02 2.91772485e-01
1.40482392e-02 -7.21323669e-01 -3.84409249e-01 -9.69590127e-01
1.45405710e-01 -2.80857086e-01 1.01456463e+00 2.52612419e-02
2.62087643e-01 3.88123482e-01 -1.51824188e+00 2.53704190e-01
6.65428042e-01 1.31900740e+00 5.46344519e-01 5.63640833e-01
-5.05104698e-02 9.81093049e-01 -3.68231922e-01 1.94921240e-01
2.00767949e-01 -3.28782707e-01 -3.75158340e-01 5.17711282e-01
1.82790324e-01 -6.93964958e-01 -7.24196494e-01 1.37137246e+00
5.42129934e-01 3.73652577e-01 4.01474833e-01 -1.28638291e+00
-1.31851614e-01 1.74220771e-01 -9.34806287e-01 6.10029995e-01
-2.60402150e-02 1.25404835e-01 4.78586376e-01 -9.72437561e-01
5.14938891e-01 1.44056356e+00 3.79661232e-01 3.12586427e-01
-6.17574155e-01 -6.47962987e-01 3.33383858e-01 5.45341969e-01
-1.46977663e+00 -4.80826527e-01 8.65064263e-01 -3.12677264e-01
5.21234691e-01 -5.57908043e-02 6.65429473e-01 7.76287258e-01
3.79331112e-02 1.20415211e+00 4.25074309e-01 -1.46901742e-01
-9.75499153e-02 3.44864503e-02 6.19579367e-02 1.06574535e+00
2.18070596e-01 1.31010013e-02 -3.38109374e-01 2.24175662e-01
1.00248659e+00 3.72029841e-01 -1.79259837e-01 -3.16851288e-01
-1.02420366e+00 6.21517479e-01 6.46761477e-01 2.91686088e-01
-3.95109475e-01 2.29669109e-01 3.99519145e-01 -6.85719959e-03
3.93951625e-01 -3.87776941e-01 -2.27210432e-01 1.38320237e-01
-1.07376266e+00 -1.51232913e-01 2.76284695e-01 9.44442809e-01
6.85339212e-01 3.00258100e-01 -2.34944701e-01 7.80339658e-01
7.42584944e-01 5.96663415e-01 4.80070412e-01 -9.71972346e-01
4.12367821e-01 7.89022267e-01 2.65506487e-02 -1.56826496e+00
-5.00873983e-01 -6.12999856e-01 -8.49871576e-01 9.49250162e-02
2.31736913e-01 -2.08702654e-01 -8.81743252e-01 1.55385745e+00
6.52967036e-01 2.95474499e-01 6.33329228e-02 1.22210133e+00
9.80420709e-01 9.69465256e-01 1.96527049e-01 -4.13645387e-01
1.34116316e+00 -1.31622839e+00 -6.75021112e-01 -1.76833823e-01
5.50449014e-01 -6.98254347e-01 4.35307175e-01 1.70785055e-01
-1.03517771e+00 -9.04985249e-01 -1.20513773e+00 6.80597946e-02
-1.08074853e-02 7.31304169e-01 6.51098311e-01 2.10271761e-01
-7.37911940e-01 2.37695441e-01 -7.42086649e-01 -5.83826602e-01
5.73811471e-01 3.08101118e-01 -1.70667797e-01 -2.89511412e-01
-9.93709981e-01 5.06938338e-01 7.21993029e-01 5.17825663e-01
-1.12555873e+00 -1.67448640e-01 -7.14167416e-01 1.39824137e-01
4.64255542e-01 -6.10474110e-01 9.56893682e-01 -1.36194968e+00
-1.07404566e+00 4.26255524e-01 -1.44772634e-01 -1.41765490e-01
4.81736571e-01 5.79991303e-02 -5.44942558e-01 5.96360326e-01
1.73291713e-01 7.88430750e-01 9.97667670e-01 -9.72008109e-01
-1.29296625e+00 -2.31749430e-01 2.32568327e-02 5.13633192e-01
-6.10625565e-01 2.35279381e-01 -1.04880798e+00 -6.60402596e-01
3.12315971e-01 -7.24442065e-01 -2.50255644e-01 5.18235676e-02
-1.33461013e-01 -2.26811707e-01 1.43124461e+00 -7.03323364e-01
1.24894595e+00 -2.63758326e+00 2.47380853e-01 6.88953698e-02
1.42801940e-01 4.48668569e-01 -3.57239604e-01 -1.18697368e-01
5.17196730e-02 -3.61820072e-01 -4.07718569e-02 -2.61279941e-02
-5.60892642e-01 -8.12895074e-02 1.22673824e-01 6.00281298e-01
2.24813536e-01 8.72981787e-01 -8.38351548e-01 -7.85034060e-01
4.14524138e-01 2.23699346e-01 -3.41469914e-01 2.40019783e-01
5.36982790e-02 2.13248327e-01 -8.72461259e-01 7.19439089e-01
6.97284162e-01 -2.00033262e-01 -1.66993454e-01 -4.91576731e-01
5.24290577e-02 -2.42401257e-01 -1.42663884e+00 1.70083153e+00
1.06473327e-01 5.81546783e-01 2.09862828e-01 -1.06491959e+00
8.11303079e-01 8.48770514e-02 6.08292460e-01 -5.98192811e-01
4.24228847e-01 -4.23939936e-02 1.05545811e-01 -9.23225760e-01
4.38998133e-01 2.49643609e-01 2.17613921e-01 7.97959492e-02
5.44361956e-02 4.18285310e-01 2.86299556e-01 4.44446057e-01
8.75802875e-01 1.48056641e-01 -2.64987107e-02 -1.84148133e-01
9.20774281e-01 1.07194846e-02 7.62403488e-01 3.81293714e-01
-3.59390974e-01 3.23337048e-01 7.93574080e-02 -3.33215594e-01
-9.14619803e-01 -8.42389107e-01 2.18286544e-01 9.95789707e-01
7.88991749e-01 -2.03874141e-01 -6.92873836e-01 -8.53687108e-01
-1.59562349e-01 1.04932852e-01 -5.01697838e-01 -4.88547742e-01
-4.90920991e-01 -8.97513330e-01 2.02782974e-01 7.96475232e-01
9.97069478e-01 -1.07438552e+00 -3.40029746e-01 2.14226320e-01
-3.84794205e-01 -1.31956375e+00 -4.59794700e-01 -3.76480490e-01
-9.60925817e-01 -1.10742581e+00 -7.30409145e-01 -1.20079327e+00
8.95173550e-01 8.47866654e-01 2.54533857e-01 4.11013752e-01
-3.51192981e-01 2.45836437e-01 -2.98365593e-01 -1.29618049e-01
-5.43001518e-02 -1.85905784e-01 2.31444798e-02 3.59489560e-01
3.04249823e-01 -2.68301934e-01 -8.65640223e-01 3.74836266e-01
-9.61682856e-01 4.89282943e-02 8.15257967e-01 5.52442253e-01
2.32726917e-01 4.50878471e-01 5.10061502e-01 -3.49637568e-01
6.52545020e-02 -4.07280236e-01 -5.67838669e-01 -3.83448005e-02
-1.93872958e-01 -3.36347610e-01 2.24796906e-01 -5.18748462e-01
-1.29823041e+00 2.05348670e-01 1.67411983e-01 -5.36325276e-01
5.14052138e-02 2.72541851e-01 -3.99749339e-01 -1.65734500e-01
2.54416645e-01 5.96709371e-01 1.77984804e-01 -2.38140538e-01
2.69879639e-01 7.62910068e-01 6.15216315e-01 -5.63513227e-02
9.00504947e-01 6.90391421e-01 -9.27141830e-02 -8.79998505e-01
-7.89980471e-01 -6.23758614e-01 -4.19127911e-01 -6.83277607e-01
1.06555033e+00 -1.21688271e+00 -5.70854247e-01 8.76278102e-01
-1.11853480e+00 1.61533549e-01 6.18787669e-02 7.66973317e-01
-3.21829319e-01 6.92676604e-01 -7.60517478e-01 -7.95738101e-01
-3.19854379e-01 -1.19179928e+00 9.79335368e-01 6.61939085e-01
3.91475499e-01 -6.79175496e-01 -6.01679623e-01 3.23121935e-01
2.84935743e-01 -8.24820623e-02 5.02464175e-01 -1.29362509e-01
-9.60415542e-01 -3.00660878e-01 -6.30582809e-01 5.17686188e-01
2.02519782e-02 2.25683048e-01 -7.59041071e-01 -3.90633017e-01
1.59984484e-01 -1.94494182e-03 1.13566971e+00 6.31365180e-01
1.07954109e+00 6.90709203e-02 -4.87163365e-01 7.22775519e-01
1.37866783e+00 4.49801356e-01 5.63619256e-01 4.26669806e-01
9.90045667e-01 5.10608137e-01 1.06622612e+00 2.23793879e-01
1.84246927e-01 4.96991694e-01 5.69383562e-01 -3.20210338e-01
-2.39971489e-01 -4.94671650e-02 7.07388461e-01 8.41459572e-01
-7.78201669e-02 -7.73685798e-02 -4.50698793e-01 4.80613083e-01
-2.14231658e+00 -1.08951294e+00 -1.64723113e-01 1.85707450e+00
2.47014180e-01 1.81219876e-01 1.26276731e-01 -1.13930218e-01
1.04301131e+00 2.60450453e-01 -5.33283651e-01 2.35459924e-01
-1.83356181e-01 -7.62354016e-01 4.17822093e-01 2.29945257e-02
-1.39718640e+00 9.29010332e-01 5.89411783e+00 1.19168878e+00
-1.01124668e+00 8.22564512e-02 5.85735381e-01 -1.06091273e-03
3.62219959e-01 -5.75464033e-02 -9.48355377e-01 5.44684649e-01
3.22903156e-01 2.50000805e-02 2.19994575e-01 8.65533054e-01
3.25848371e-01 -2.75564939e-01 -7.09181130e-01 1.10676682e+00
1.93239748e-01 -1.15860069e+00 8.43704417e-02 -2.33770490e-01
7.61183262e-01 8.54112133e-02 -4.55262139e-02 1.13801390e-01
-1.89955041e-01 -4.47185695e-01 6.65744364e-01 6.19300365e-01
5.15089512e-01 -9.29314673e-01 8.11385334e-01 4.66807127e-01
-1.67868984e+00 -4.92533535e-01 -5.31017423e-01 8.49934965e-02
1.57246411e-01 3.40504974e-01 -2.25069895e-01 6.60942554e-01
7.03157485e-01 1.08299911e+00 -5.23569643e-01 1.17101443e+00
-1.72259539e-01 3.91324311e-01 -2.97036320e-01 5.80317862e-02
4.52405840e-01 -2.02869356e-01 6.50777340e-01 1.11441779e+00
2.66211480e-01 1.38634443e-01 3.84408414e-01 5.47420502e-01
9.83537957e-02 1.42243603e-04 -4.42375064e-01 -3.00868079e-02
3.36513072e-01 1.68877494e+00 -9.17677522e-01 -6.55476809e-01
-5.55036962e-01 9.80358362e-01 1.13584459e-01 4.86200720e-01
-9.93616283e-01 -3.15812916e-01 2.58634448e-01 -4.27387021e-02
5.28516471e-01 -3.00528675e-01 1.94405407e-01 -1.09879589e+00
-3.22536454e-02 -6.68605030e-01 4.49921101e-01 -9.35852170e-01
-8.97617757e-01 3.06164652e-01 -2.26261020e-01 -1.33930898e+00
2.60537863e-01 -5.56820631e-01 -8.33180904e-01 5.85597575e-01
-1.28159475e+00 -9.98621225e-01 -6.57411933e-01 6.75084531e-01
8.00885737e-01 -4.03766900e-01 1.24442562e-01 5.86292684e-01
-9.49709237e-01 4.08656448e-01 1.63442373e-01 4.02700961e-01
4.66668606e-01 -5.77920318e-01 -1.40022442e-01 1.29247999e+00
-3.61047834e-01 3.75921935e-01 1.27935916e-01 -7.17995763e-01
-1.41103673e+00 -1.42396867e+00 1.84860408e-01 5.98991998e-02
4.69486773e-01 -1.24469839e-01 -7.94583142e-01 2.78986603e-01
-1.19118087e-01 9.18311179e-02 -1.70572568e-03 -5.12840271e-01
8.24532211e-02 -3.04730177e-01 -9.09557879e-01 4.58469212e-01
9.78893220e-01 -2.71617025e-01 -4.87703532e-01 3.22287560e-01
8.57493997e-01 -1.25606298e-01 -7.00757802e-01 5.55240095e-01
5.05102336e-01 -6.20282412e-01 9.90918040e-01 -4.24125910e-01
5.40088594e-01 -7.72433937e-01 -4.36425023e-02 -7.28321195e-01
-5.38021147e-01 -2.07826674e-01 -1.45085946e-01 1.36941504e+00
2.66079400e-02 -3.33588481e-01 7.50492215e-01 9.61707011e-02
-2.82081421e-02 -6.73708141e-01 -6.25060618e-01 -5.36162674e-01
-6.93147838e-01 -2.62648195e-01 1.32290214e-01 7.18393266e-01
-3.00699413e-01 5.62242508e-01 -4.38717246e-01 3.02469343e-01
7.39964485e-01 1.08936273e-01 7.45898485e-01 -8.30304801e-01
-1.53189525e-01 -3.55607420e-01 -7.10925281e-01 -1.48379076e+00
-1.89646289e-01 -7.05086708e-01 1.04475215e-01 -1.59231782e+00
6.15442932e-01 -1.15619816e-01 -4.85242397e-01 1.38608336e-01
-4.49845105e-01 3.18981081e-01 2.92375535e-01 4.14356977e-01
-9.83515978e-01 7.22491980e-01 1.46347678e+00 -2.08490118e-01
-5.38523756e-02 -2.26391360e-01 -3.90326440e-01 1.07855392e+00
5.57033300e-01 -4.09802228e-01 -5.33998668e-01 -4.61316168e-01
-3.50298911e-01 1.39188349e-01 3.87690395e-01 -1.19490016e+00
5.39698124e-01 -2.03749780e-02 8.89939368e-01 -8.75219762e-01
2.90435433e-01 -8.46594334e-01 -2.75656700e-01 7.07510412e-01
-6.45458624e-02 -2.22809747e-01 1.62143975e-01 9.51521516e-01
-3.01490009e-01 -3.79416347e-02 9.68795896e-01 2.01082453e-02
-1.32700276e+00 6.81994379e-01 -4.27554935e-01 -2.88651913e-01
1.25083792e+00 -4.49947298e-01 -3.06464761e-01 -2.02293858e-01
-6.79335654e-01 5.80175877e-01 2.26175368e-01 7.50204921e-01
8.60755324e-01 -1.48859763e+00 -6.40303075e-01 3.90318334e-01
-3.16420160e-02 1.81256920e-01 5.76090634e-01 1.22744906e+00
-5.40078223e-01 8.74321610e-02 -2.01829210e-01 -8.80847573e-01
-1.31577456e+00 6.61141396e-01 3.56356472e-01 2.69680738e-01
-5.99134326e-01 9.00239348e-01 7.19506919e-01 2.37904966e-01
2.32563987e-01 -1.49551556e-02 -5.27202785e-01 1.05048396e-01
7.64973581e-01 5.38814545e-01 -4.38949436e-01 -1.01786566e+00
-4.97423530e-01 7.44537890e-01 -1.58841178e-01 6.32968321e-02
1.12729967e+00 -5.13058722e-01 -2.58054554e-01 5.14314100e-02
1.24553668e+00 -2.85640836e-01 -1.37220621e+00 -3.43485862e-01
-2.98811823e-01 -4.60144818e-01 1.94439515e-01 -3.18503112e-01
-1.50199366e+00 8.22731078e-01 8.02402437e-01 -1.69504762e-01
1.45953619e+00 -2.35963970e-01 7.67124534e-01 2.72642940e-01
1.59741133e-01 -1.20236742e+00 3.68175626e-01 3.69443059e-01
3.82265866e-01 -1.09363711e+00 4.35124747e-02 -5.67601621e-01
-5.21014988e-01 1.02828836e+00 9.60561037e-01 -3.98754030e-01
4.72399205e-01 1.02423802e-01 -2.30356939e-02 -8.66801739e-02
-4.57498431e-01 -1.86859906e-01 3.54432076e-01 4.63832796e-01
-1.21941283e-01 -3.94305825e-01 -3.21129858e-01 5.12737870e-01
5.86338043e-01 -2.14154329e-02 2.62428671e-01 7.86923170e-01
-9.65987146e-01 -5.37082851e-01 -3.70240003e-01 2.71852911e-01
-4.02035445e-01 4.79207560e-02 -1.01404622e-01 6.50702715e-01
2.73842424e-01 1.10491955e+00 -5.49285449e-02 -6.62979186e-01
9.24333110e-02 -3.77461314e-01 2.97356248e-01 -3.53760093e-01
-1.75536439e-01 4.87090081e-01 1.87830988e-03 -8.40392351e-01
-6.86604500e-01 -4.87319678e-01 -1.45283532e+00 1.96677949e-02
-6.38935089e-01 9.34630707e-02 4.46647435e-01 9.30523932e-01
2.09414124e-01 7.84507573e-01 7.88643003e-01 -9.58894074e-01
-4.41018671e-01 -9.66902673e-01 -6.74469709e-01 4.87701476e-01
1.63938537e-01 -7.88122356e-01 -2.11592048e-01 9.79384854e-02]
|
[9.190481185913086, -0.3291400969028473]
|
62ccf789-891e-47af-bccc-3aabc1c1ba02
|
humans-as-light-bulbs-3d-human-reconstruction
|
2305.01652
| null |
https://arxiv.org/abs/2305.01652v1
|
https://arxiv.org/pdf/2305.01652v1.pdf
|
Humans as Light Bulbs: 3D Human Reconstruction from Thermal Reflection
|
The relatively hot temperature of the human body causes people to turn into long-wave infrared light sources. Since this emitted light has a larger wavelength than visible light, many surfaces in typical scenes act as infrared mirrors with strong specular reflections. We exploit the thermal reflections of a person onto objects in order to locate their position and reconstruct their pose, even if they are not visible to a normal camera. We propose an analysis-by-synthesis framework that jointly models the objects, people, and their thermal reflections, which allows us to combine generative models with differentiable rendering of reflections. Quantitative and qualitative experiments show our approach works in highly challenging cases, such as with curved mirrors or when the person is completely unseen by a normal camera.
|
['Carl Vondrick', 'Ruoshi Liu']
|
2023-05-02
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Liu_Humans_As_Light_Bulbs_3D_Human_Reconstruction_From_Thermal_Reflection_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Liu_Humans_As_Light_Bulbs_3D_Human_Reconstruction_From_Thermal_Reflection_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['3d-human-reconstruction']
|
['computer-vision']
|
[ 4.21695054e-01 -3.66846565e-03 6.01656854e-01 -1.56088427e-01
1.11624740e-01 -4.62784231e-01 7.68207967e-01 -1.17017210e+00
-1.28158748e-01 2.52999455e-01 2.23350868e-01 3.28387529e-01
6.01068616e-01 -7.99068034e-01 -4.16634202e-01 -7.44849920e-01
6.98271275e-01 6.62039757e-01 3.88134345e-02 -4.23852578e-02
-3.18304747e-02 2.52916545e-01 -1.86517215e+00 3.30853462e-01
5.23499250e-01 3.76127928e-01 1.37870863e-01 8.95865738e-01
1.10491432e-01 5.69908619e-01 -6.57922864e-01 -4.51654434e-01
4.91703004e-01 -4.44002062e-01 -3.39153051e-01 5.48703790e-01
8.13016832e-01 -6.97109282e-01 -2.73552030e-01 8.11967731e-01
9.45594460e-02 2.92350054e-01 8.29413354e-01 -7.07859397e-01
-6.52004302e-01 -1.43503547e-01 -7.02551723e-01 -5.10506153e-01
1.12148285e+00 2.12181941e-01 5.35294950e-01 -8.39109659e-01
5.20731390e-01 1.50687766e+00 4.75305349e-01 8.86671007e-01
-8.76265466e-01 -2.98374146e-01 3.94906364e-02 -1.91774532e-01
-1.16318142e+00 -5.72477043e-01 9.96536970e-01 -4.27806139e-01
2.41743311e-01 6.71099424e-01 9.14774060e-01 1.42003977e+00
2.11356282e-01 4.69789296e-01 1.11275041e+00 -4.20192987e-01
6.43915385e-02 4.27093327e-01 -4.71764728e-02 5.30286312e-01
2.18308225e-01 2.71755129e-01 -3.98327768e-01 -3.32487702e-01
9.23980832e-01 6.21536672e-01 -4.53422904e-01 -1.01259118e-02
-1.09640861e+00 1.31991088e-01 2.16136232e-01 -4.73093353e-02
-3.06285679e-01 1.61619738e-01 -6.34077251e-01 1.61519617e-01
5.63715577e-01 9.60620269e-02 1.90474346e-01 2.85897046e-01
-4.32467461e-01 1.57887876e-01 8.14607620e-01 8.31647992e-01
7.10949242e-01 -2.69903660e-01 1.97575986e-01 7.05302000e-01
7.35459924e-01 1.09187758e+00 -2.24417582e-01 -9.05715823e-01
2.15788856e-01 6.60309076e-01 5.44552863e-01 -6.63750768e-01
-1.84104785e-01 -1.61476687e-01 -6.83055222e-01 7.05125332e-01
5.78223407e-01 -2.50590473e-01 -8.99681032e-01 1.20183539e+00
6.80525184e-01 3.18914860e-01 -8.30589086e-02 1.53519857e+00
7.74694204e-01 7.19774425e-01 -6.19887769e-01 -1.28796667e-01
1.59052181e+00 -6.52894676e-01 -5.82183540e-01 -4.99033302e-01
7.56686479e-02 -1.03568065e+00 1.10676551e+00 6.95776403e-01
-1.16792953e+00 -5.05098641e-01 -5.66980183e-01 -1.01865858e-01
3.07677656e-01 4.25146930e-02 2.08265573e-01 8.15259457e-01
-7.03459740e-01 1.40760571e-01 -7.69619465e-01 -3.22423995e-01
1.00991735e-02 -3.80274132e-02 8.52787346e-02 -2.61996329e-01
-6.68099642e-01 5.82771361e-01 -5.58132470e-01 4.92114723e-01
-5.95344126e-01 -6.02033734e-01 -5.78466535e-01 -4.46761936e-01
4.21771944e-01 -1.06812167e+00 1.00606453e+00 -1.05772305e+00
-1.94858801e+00 9.14010584e-01 -3.69053006e-01 4.48846221e-01
8.66090059e-01 -7.98637629e-01 -4.94810700e-01 1.76610515e-01
-4.23719972e-01 -2.22667173e-01 1.09020877e+00 -1.74926651e+00
-1.19329631e-01 -7.15063572e-01 1.26724169e-01 4.77231979e-01
-9.00295079e-02 2.48578861e-01 -5.18514812e-01 1.10747507e-02
2.73871213e-01 -1.27784741e+00 -2.22082958e-01 2.01818481e-01
-8.43108356e-01 2.09157720e-01 1.01835763e+00 -3.49657118e-01
4.12554741e-01 -1.94586682e+00 -1.00017227e-01 3.99167001e-01
2.36163571e-01 1.36878431e-01 3.00800234e-01 4.43594545e-01
1.46059409e-01 -4.21265543e-01 6.92990497e-02 -6.73771620e-01
-2.25594133e-01 4.26711515e-02 -5.07066607e-01 7.04290509e-01
-5.94166458e-01 6.21975482e-01 -7.08157778e-01 -1.32861957e-01
2.82916427e-01 1.03420556e+00 -1.77358061e-01 5.31182170e-01
-1.87088370e-01 8.35219741e-01 -5.49986005e-01 4.84260201e-01
8.64627302e-01 -1.03125155e-01 4.91089933e-02 -1.28498152e-01
-1.90771714e-01 -8.67255330e-02 -1.13843381e+00 1.03376412e+00
-4.88368243e-01 4.65341777e-01 2.07099155e-01 6.52618855e-02
1.01067197e+00 3.67127448e-01 2.54203260e-01 -5.45794487e-01
4.11552563e-02 -1.37385026e-01 -6.50535226e-01 -7.56806672e-01
4.90687370e-01 -5.71653664e-01 2.19126299e-01 8.13973546e-01
-1.01273024e+00 -6.21120222e-02 -4.99227554e-01 1.92158502e-02
7.30575383e-01 4.01625723e-01 -1.82576701e-01 1.02162823e-01
3.19155782e-01 -5.66035211e-01 3.43944937e-01 3.52964789e-01
5.12835801e-01 1.24529779e+00 -2.13019267e-01 -7.39560485e-01
-8.72175753e-01 -1.33011830e+00 1.22465692e-01 9.70891714e-01
4.58211988e-01 -1.22742258e-01 -7.16429651e-01 -1.41666695e-01
-8.10747519e-02 6.91200674e-01 -5.24426401e-01 -7.88602233e-03
-7.24945307e-01 -4.04362321e-01 2.61990074e-02 2.92794943e-01
2.42136359e-01 -9.17799950e-01 -1.09598482e+00 -2.60695636e-01
-3.45164210e-01 -1.14532590e+00 -5.28801084e-01 -8.87058377e-01
-6.22105479e-01 -1.14767730e+00 -9.11989391e-01 -2.51162320e-01
1.13153923e+00 6.35621190e-01 1.11413217e+00 3.09714556e-01
-6.46372557e-01 8.07397485e-01 -8.50560740e-02 -5.86564541e-01
-5.40146470e-01 -8.31822693e-01 1.09225810e-01 5.29094756e-01
1.92939475e-01 -4.18008655e-01 -9.14093614e-01 6.05963349e-01
-5.59233487e-01 2.54804432e-01 1.23263434e-01 -7.65351802e-02
1.21504910e-01 -3.50084193e-02 -6.64215267e-01 -1.00809896e+00
1.17123351e-01 -2.46582385e-02 -7.51619816e-01 2.38824755e-01
2.11026818e-01 -3.02530736e-01 5.03531516e-01 -3.73556167e-01
-1.60398197e+00 -6.09251522e-02 4.41863298e-01 -5.45691133e-01
-5.16953230e-01 -3.14007670e-01 8.32972676e-03 1.47983328e-01
8.26231480e-01 7.55402669e-02 -6.19017817e-02 -6.71575367e-01
8.53391141e-02 4.67525661e-01 6.52647555e-01 -3.83242399e-01
1.43313611e+00 1.31093574e+00 1.79554686e-01 -1.51941025e+00
-8.54447365e-01 -4.76019830e-01 -7.10705698e-01 -6.25310600e-01
9.15700793e-01 -7.22321451e-01 -1.26548970e+00 6.95330024e-01
-1.26436293e+00 -2.45649815e-01 -8.43093544e-02 5.78948379e-01
-1.73856556e-01 3.81156236e-01 -5.26741087e-01 -1.35167873e+00
-7.27159083e-02 -5.84825277e-01 1.25414777e+00 4.28107858e-01
-1.27689078e-01 -1.04377341e+00 3.51580471e-01 8.68263423e-01
1.31355777e-01 2.49022916e-01 3.03169638e-01 4.39188808e-01
-9.50563908e-01 -7.05397222e-03 2.73652107e-01 1.12471782e-01
3.40193033e-01 3.65769178e-01 -1.39346313e+00 -1.41604200e-01
1.83357581e-01 5.86511195e-02 7.01095700e-01 3.49890411e-01
8.67463470e-01 -2.71157622e-01 -5.10738015e-01 8.22724164e-01
1.12935305e+00 -1.86818838e-01 8.74864876e-01 -2.26613849e-01
1.17395914e+00 1.02798760e+00 3.22171122e-01 6.36253476e-01
1.76861167e-01 7.07237482e-01 4.66709048e-01 -3.26710254e-01
-4.76574786e-02 1.93360541e-02 5.69924593e-01 4.88985747e-01
-9.54310417e-01 -4.46316868e-01 -7.46067822e-01 -1.72617927e-01
-1.34924734e+00 -1.10118651e+00 -7.75274992e-01 2.73788714e+00
2.96608567e-01 -3.57287765e-01 7.93285146e-02 -3.29114318e-01
6.55309975e-01 -4.16817665e-02 -1.71509385e-01 -6.41018376e-02
2.03520223e-01 -1.97773829e-01 2.20116854e-01 6.70221865e-01
-5.25681973e-01 3.94582063e-01 6.98941040e+00 -3.52712840e-01
-9.72659469e-01 -2.04149395e-01 1.28681719e-01 -4.85600263e-01
-8.85941684e-01 -2.45860573e-02 -6.32715523e-01 3.70233506e-01
4.74854827e-01 3.84220332e-01 5.28296888e-01 3.60906392e-01
2.96577156e-01 -1.40193745e-01 -1.24697351e+00 1.10126066e+00
2.42100671e-01 -5.48437953e-01 -1.46885857e-01 3.99796486e-01
6.21294618e-01 -1.86804473e-01 3.40080768e-01 -5.08342445e-01
3.59135360e-01 -9.18641925e-01 6.10046983e-01 1.09265423e+00
6.17116392e-01 -3.69545341e-01 3.69716436e-02 5.40932178e-01
-8.89338493e-01 2.30368450e-01 -5.64889073e-01 -3.02245826e-01
4.41360980e-01 8.22690785e-01 -8.42280686e-01 3.63959551e-01
4.33052152e-01 4.25409943e-01 -1.18345730e-02 6.40938818e-01
-4.84213471e-01 4.97451156e-01 -5.13854682e-01 1.63948208e-01
-3.41906428e-01 -8.71704936e-01 9.00078177e-01 8.34855258e-01
3.68324697e-01 4.91131157e-01 1.19434722e-01 1.40731037e+00
1.11953773e-01 -3.25894058e-01 -5.18069983e-01 5.12442768e-01
-1.23045757e-01 1.29688120e+00 -4.94052708e-01 -1.29285231e-01
-4.73570108e-01 1.11448741e+00 -2.48603582e-01 9.41080511e-01
-7.24116087e-01 3.27815801e-01 7.81810343e-01 6.38463140e-01
-3.67746115e-01 -2.52007008e-01 1.31787704e-02 -1.32725036e+00
3.10799837e-01 -3.81223679e-01 3.40954401e-02 -1.13940203e+00
-1.11661887e+00 3.64956766e-01 -8.82332101e-02 -1.39232624e+00
7.10959733e-02 -4.66791153e-01 -9.22141969e-01 8.75476658e-01
-1.24229050e+00 -1.18328190e+00 -9.08426106e-01 8.61618638e-01
2.47930437e-01 2.80254900e-01 5.82058251e-01 -9.66574252e-02
-3.04155767e-01 5.61307259e-02 5.12519702e-02 -1.80915520e-01
6.75249279e-01 -1.09440458e+00 2.97871858e-01 7.32019722e-01
1.21247113e-01 7.24999666e-01 9.25726831e-01 -6.65885150e-01
-1.80422866e+00 -7.05488980e-01 4.04228270e-01 -6.97570503e-01
-1.32115960e-01 -7.71222055e-01 -8.94283354e-01 7.17296779e-01
-1.17561249e-02 -6.83908537e-03 4.92292136e-01 2.61900872e-01
-4.32438672e-01 -1.43404722e-01 -9.79018629e-01 7.77436376e-01
1.04314458e+00 -4.18482244e-01 -6.18151009e-01 5.62800884e-01
1.18203945e-01 -4.53594029e-01 -1.13001533e-01 -1.10559709e-01
9.58894253e-01 -1.33968782e+00 1.28986585e+00 -7.30281929e-03
2.12347865e-01 -1.47738770e-01 2.85218537e-01 -1.05422175e+00
-3.17971379e-01 -8.89019191e-01 1.23042971e-01 7.61513114e-01
-9.32650343e-02 -1.04021287e+00 8.32208872e-01 1.06357682e+00
5.42107746e-02 -8.73170868e-02 -5.17697096e-01 -5.57257175e-01
-5.00948846e-01 -3.47165644e-01 3.69486004e-01 6.65831625e-01
-3.02293479e-01 9.20164287e-02 -8.93870533e-01 5.31974554e-01
1.13999116e+00 4.14986759e-01 1.12753665e+00 -1.62616336e+00
-4.87131923e-01 6.07971847e-02 2.93550920e-02 -1.20396459e+00
-6.15201741e-02 -3.43663752e-01 2.66280711e-01 -1.45642424e+00
4.37232196e-01 -2.61744261e-01 2.95705825e-01 3.91774215e-02
-1.27143502e-01 3.24824542e-01 4.04468402e-02 4.19429779e-01
-1.75896049e-01 4.44973379e-01 1.55824244e+00 3.11611891e-01
-3.90866756e-01 5.43766141e-01 -3.11703831e-01 1.34854412e+00
2.86381006e-01 -3.02455574e-01 -4.77508068e-01 -5.72619438e-01
6.11273646e-01 4.09105308e-02 8.89409900e-01 -9.45888817e-01
9.93647650e-02 -5.86677492e-01 5.55197179e-01 -4.80219513e-01
7.73286700e-01 -9.46634293e-01 7.46507704e-01 3.44397247e-01
-1.36611992e-02 -5.35669148e-01 -4.95446771e-01 5.28228700e-01
6.39050543e-01 -4.54845764e-02 7.10164487e-01 -3.77468437e-01
-9.57237463e-03 2.87108004e-01 -3.87469351e-01 -1.77338645e-01
8.66572917e-01 -6.15718067e-01 -5.51024556e-01 -6.81667984e-01
-4.95177656e-01 -1.22210756e-01 1.00032675e+00 3.28133404e-01
8.62502456e-01 -9.72849131e-01 -8.53447616e-01 4.19472724e-01
-4.92807254e-02 1.50342703e-01 3.49907130e-01 6.77902818e-01
-5.27361512e-01 -1.36345193e-01 2.25628570e-01 -7.72756040e-01
-1.86961806e+00 3.53831470e-01 5.15663326e-01 4.57855523e-01
-1.27765012e+00 8.10839534e-01 1.10350502e+00 -2.70610094e-01
-2.43389890e-01 -2.18609869e-01 3.27732079e-02 -4.06233400e-01
8.59865785e-01 7.69226074e-01 -3.65323812e-01 -8.97720754e-01
-3.44352067e-01 1.06465077e+00 2.60293871e-01 -1.71198234e-01
1.15399194e+00 -3.28525722e-01 -9.71935019e-02 6.59729600e-01
8.81080985e-01 5.50615072e-01 -1.25971663e+00 -5.05271554e-01
-9.97707665e-01 -1.01011622e+00 -3.78988475e-01 -4.27700877e-01
-1.06137276e+00 8.88731062e-01 4.34384383e-02 -1.51125900e-02
9.10805762e-01 3.03453147e-01 7.69482613e-01 2.91164935e-01
6.57225609e-01 -1.04284227e+00 4.94111091e-01 1.89275980e-01
8.26436043e-01 -7.11992383e-01 1.41589016e-01 -9.15219963e-01
-5.61293304e-01 1.17010307e+00 5.15264928e-01 -6.72783554e-02
3.27607125e-01 4.66576815e-01 4.66330796e-01 -3.27923805e-01
-6.70536041e-01 1.26491189e-01 6.24208331e-01 5.33260345e-01
2.20858902e-01 1.51180238e-01 6.15265429e-01 -2.44194493e-01
-3.34573805e-01 -2.47650176e-01 7.72545040e-01 7.91887045e-01
-4.69004929e-01 -7.25052476e-01 -1.10157132e+00 8.86613727e-02
-2.34674439e-01 4.28443044e-01 -5.96677661e-01 7.19415396e-02
-1.30730927e-01 9.92663145e-01 2.34015092e-01 -6.36350513e-02
3.72062773e-01 -3.51543501e-02 7.99967229e-01 -6.74816847e-01
-3.09454560e-01 3.86934489e-01 -1.28022790e-01 -6.58431470e-01
-5.34668148e-01 -5.98942101e-01 -1.21387374e+00 -2.72537738e-01
-2.54637897e-01 -3.57220352e-01 5.71626425e-01 9.74261820e-01
-1.16394967e-01 1.94238469e-01 8.45565081e-01 -9.19683039e-01
-2.36063257e-01 -5.62433898e-01 -8.81229937e-01 8.37159157e-01
6.70999944e-01 -5.09404898e-01 -7.37442911e-01 2.46341363e-01]
|
[9.812159538269043, -3.004401922225952]
|
9374a4c8-0ba7-4a4c-b75b-a8634222ddd3
|
one-for-all-one-stage-referring-expression
|
2208.00361
| null |
https://arxiv.org/abs/2208.00361v3
|
https://arxiv.org/pdf/2208.00361v3.pdf
|
One for All: One-stage Referring Expression Comprehension with Dynamic Reasoning
|
Referring Expression Comprehension (REC) is one of the most important tasks in visual reasoning that requires a model to detect the target object referred by a natural language expression. Among the proposed pipelines, the one-stage Referring Expression Comprehension (OSREC) has become the dominant trend since it merges the region proposal and selection stages. Many state-of-the-art OSREC models adopt a multi-hop reasoning strategy because a sequence of objects is frequently mentioned in a single expression which needs multi-hop reasoning to analyze the semantic relation. However, one unsolved issue of these models is that the number of reasoning steps needs to be pre-defined and fixed before inference, ignoring the varying complexity of expressions. In this paper, we propose a Dynamic Multi-step Reasoning Network, which allows the reasoning steps to be dynamically adjusted based on the reasoning state and expression complexity. Specifically, we adopt a Transformer module to memorize & process the reasoning state and a Reinforcement Learning strategy to dynamically infer the reasoning steps. The work achieves the state-of-the-art performance or significant improvements on several REC datasets, ranging from RefCOCO (+, g) with short expressions, to Ref-Reasoning, a dataset with long and complex compositional expressions.
|
['Peng Wang', 'Rui Niu', 'Zhongzhen Huang', 'Zhimin Wei', 'Zhipeng Zhang']
|
2022-07-31
| null | null | null | null |
['visual-reasoning', 'visual-reasoning']
|
['computer-vision', 'reasoning']
|
[ 2.57585704e-01 1.77921504e-01 -1.60550237e-01 -5.85697711e-01
-3.97999525e-01 -5.48877478e-01 6.94865227e-01 2.04526439e-01
-3.75600278e-01 3.11717302e-01 2.12900698e-01 -3.46315473e-01
-9.49540921e-03 -9.80453253e-01 -6.56070292e-01 -3.24573129e-01
3.42598557e-01 6.02328718e-01 7.43031025e-01 -4.38356608e-01
2.74254143e-01 4.97740358e-01 -1.51523256e+00 7.31313109e-01
7.68674850e-01 1.06815982e+00 1.68932483e-01 4.30089891e-01
-9.34837401e-01 1.31574643e+00 -5.97455800e-01 -5.95189154e-01
-2.53857553e-01 -6.36249125e-01 -1.06575298e+00 -3.42611343e-01
1.00578219e-01 -4.51310873e-02 1.49919555e-01 1.16094625e+00
3.87740552e-01 1.52205870e-01 4.56849188e-01 -1.22540045e+00
-4.20378894e-01 1.02740836e+00 -8.79852653e-01 2.87254214e-01
6.79801643e-01 7.40950033e-02 9.78575051e-01 -6.22885764e-01
8.47598374e-01 1.70604885e+00 1.46969274e-01 4.01044935e-01
-8.16671312e-01 -6.66767716e-01 7.99095392e-01 7.91403651e-01
-1.26246047e+00 -2.19369501e-01 9.30862069e-01 -3.99351031e-01
1.08606029e+00 1.38567150e-01 8.51616144e-01 7.34946489e-01
4.56750803e-02 8.20120335e-01 9.09880519e-01 -4.57968622e-01
4.76779282e-01 -2.54464984e-01 2.28201911e-01 8.34460974e-01
-4.65000749e-01 -4.32408392e-01 -5.12793660e-01 2.08716363e-01
4.54091400e-01 -2.18090087e-01 -9.48504284e-02 -2.46282592e-01
-1.10454333e+00 5.07646501e-01 7.22074807e-01 3.17909777e-01
-4.65383798e-01 3.19866389e-01 7.36043692e-01 2.80623548e-02
4.34180629e-03 1.84494689e-01 -2.48531342e-01 -2.66852885e-01
-7.01067507e-01 3.75430405e-01 5.80889165e-01 1.04932690e+00
7.52771378e-01 -3.26624155e-01 -7.24420428e-01 6.97613418e-01
3.13018918e-01 4.06908840e-02 1.69599533e-01 -8.42065036e-01
5.64015090e-01 1.28022277e+00 -1.82114672e-02 -9.40638304e-01
-4.88055170e-01 -2.94332385e-01 -6.18093908e-01 1.48386687e-01
2.61292964e-01 2.19145045e-02 -7.22291350e-01 1.81691432e+00
6.04533494e-01 1.54636458e-01 1.25647023e-01 1.06468022e+00
1.03707826e+00 6.64236307e-01 6.05502546e-01 -9.29237828e-02
1.98587644e+00 -1.31127989e+00 -9.00473535e-01 -2.75494874e-01
4.48731005e-01 -5.46808720e-01 1.18674541e+00 2.70012945e-01
-1.20780909e+00 -4.85061705e-01 -9.39446390e-01 -5.54541707e-01
-5.93976915e-01 -4.33332063e-02 8.04987669e-01 -1.47790811e-03
-5.79407930e-01 -5.57950363e-02 -5.41783988e-01 -4.23019469e-01
4.54014033e-01 -2.83905794e-03 -1.68129548e-01 -1.74505085e-01
-1.35755467e+00 9.11195219e-01 6.10871434e-01 3.68613183e-01
-6.64361179e-01 -5.92223465e-01 -7.87387371e-01 3.32866997e-01
8.33572030e-01 -8.27821255e-01 1.32528710e+00 -1.02258182e+00
-1.47864699e+00 9.01746690e-01 -4.07928735e-01 -3.59853685e-01
6.90871239e-01 -3.89445066e-01 -4.29694593e-01 4.28229868e-01
1.56475452e-03 7.36680210e-01 7.14433789e-01 -1.08595443e+00
-7.31505752e-01 -4.36608821e-01 7.18459308e-01 3.76965761e-01
3.88446957e-01 4.28072810e-01 -9.63090479e-01 -3.79405022e-01
2.01103881e-01 -4.87795770e-01 4.40140963e-02 3.22046012e-01
-2.30358064e-01 -7.55612075e-01 7.82466948e-01 -4.36992586e-01
1.41984153e+00 -2.21855068e+00 4.99069452e-01 -2.21692342e-02
1.53227612e-01 8.59799162e-02 2.30802640e-01 2.10720316e-01
-1.99823350e-01 -5.60432337e-02 -2.11003169e-01 -8.49123523e-02
1.40750974e-01 2.16343164e-01 -3.30141783e-01 7.91846812e-02
3.39780033e-01 1.06719184e+00 -1.06028819e+00 -9.20400918e-01
2.08164707e-01 2.00283945e-01 -3.66391987e-01 4.63210046e-01
-8.45400870e-01 3.79546404e-01 -5.25031924e-01 7.34550953e-01
8.29214036e-01 -3.13443720e-01 2.48619258e-01 -5.64834952e-01
-2.12921098e-01 1.00178942e-01 -1.20157933e+00 1.99038458e+00
-5.78185678e-01 3.43639672e-01 4.26823869e-02 -9.58616078e-01
9.61016357e-01 4.06459272e-02 4.91103381e-02 -8.28968942e-01
1.02495559e-01 -9.56877600e-03 -2.10623611e-02 -8.29403222e-01
3.40156227e-01 -6.89463690e-02 -2.00818732e-01 3.31220001e-01
-2.70295948e-01 1.05058197e-02 4.15326148e-01 2.82924622e-01
1.09518695e+00 7.61918306e-01 5.25242746e-01 -6.81806654e-02
9.57125247e-01 1.52596623e-01 6.70393229e-01 4.04596537e-01
-1.44082919e-01 2.01231748e-01 1.11572683e+00 -4.61880237e-01
-4.32154298e-01 -1.00142455e+00 1.97404936e-01 1.33182573e+00
5.63804746e-01 -4.64298397e-01 -7.41616189e-01 -6.32305682e-01
-3.74685436e-01 1.06705809e+00 -7.56502092e-01 -1.10736363e-01
-8.17800522e-01 -2.65686393e-01 5.32430291e-01 7.67164409e-01
9.88339424e-01 -1.53527832e+00 -1.24116457e+00 1.75804392e-01
-1.16799034e-01 -1.21747899e+00 -4.36972976e-02 1.83697939e-01
-3.41515005e-01 -1.11361861e+00 -2.83606142e-01 -6.06637716e-01
8.23353887e-01 -1.72888801e-01 1.25492871e+00 2.16123506e-01
-5.71442954e-02 9.67545956e-02 -4.93220776e-01 -3.85876417e-01
-1.34912536e-01 1.10724851e-01 -8.60192955e-01 -3.57143246e-02
3.48590195e-01 -3.17342997e-01 -5.70948482e-01 3.23320031e-01
-8.54943037e-01 6.46788061e-01 5.42261600e-01 6.20392680e-01
8.40024650e-01 -1.35400775e-03 3.14703614e-01 -9.84801650e-01
5.14002621e-01 -4.43811119e-01 -7.61468947e-01 8.86989892e-01
-1.83725372e-01 3.21907967e-01 8.13333452e-01 -2.58253783e-01
-1.61470604e+00 -4.15980443e-02 -1.19154818e-01 -4.07961071e-01
-7.82646835e-02 5.42797625e-01 -3.76346439e-01 5.28753996e-01
3.41783762e-01 6.08370639e-02 -4.32902187e-01 -2.29063511e-01
6.88859403e-01 2.02070519e-01 6.39762402e-01 -9.94677782e-01
3.85758311e-01 4.20289785e-01 7.10084140e-02 -1.30779386e-01
-9.53095198e-01 -3.53820533e-01 -6.71919048e-01 -3.32858145e-01
1.13177443e+00 -7.91104138e-01 -1.04438651e+00 3.49600554e-01
-1.56212676e+00 -3.73205781e-01 -1.76163658e-01 -1.54552191e-01
-4.62487787e-01 -3.64592969e-02 -3.57732594e-01 -8.23301017e-01
-4.51459318e-01 -1.36432087e+00 1.12579048e+00 5.44655323e-01
-2.82702118e-01 -6.67286158e-01 -4.49260503e-01 2.27919161e-01
4.16895509e-01 2.92633563e-01 1.39122415e+00 -2.01899469e-01
-7.16539860e-01 3.16928476e-01 -7.07722902e-01 -3.81590098e-01
-6.96493164e-02 1.71079561e-01 -7.38818586e-01 3.29510301e-01
-4.55481470e-01 -3.36318463e-01 6.23796284e-01 -4.49360013e-02
1.60956538e+00 1.37950137e-01 -3.90554994e-01 7.09766626e-01
1.17967308e+00 3.91833991e-01 7.81383455e-01 3.24156612e-01
5.43037832e-01 7.25871921e-01 9.99252975e-01 2.83458441e-01
7.03074813e-01 7.89768815e-01 5.70221066e-01 4.38555330e-02
-1.20505914e-01 -4.52157766e-01 8.65416899e-02 2.81029016e-01
3.99932303e-02 -1.48029968e-01 -1.11559868e+00 3.62611145e-01
-2.18356419e+00 -9.92692649e-01 -1.16664335e-01 1.59327352e+00
7.16452539e-01 2.84623712e-01 -1.73263475e-01 4.66603693e-03
5.54155052e-01 2.65850335e-01 -7.98932672e-01 -5.68989933e-01
-1.13120778e-02 2.53088742e-01 -1.40400201e-01 3.36703002e-01
-7.35889852e-01 1.28606498e+00 5.54196262e+00 6.63983405e-01
-1.17527783e+00 -7.34425038e-02 4.81167912e-01 1.71347201e-01
-3.54973257e-01 2.04017639e-01 -7.01165915e-01 1.16445370e-01
5.70084751e-01 -1.70391843e-01 4.03719187e-01 7.35991240e-01
1.35166138e-01 -6.54825747e-01 -1.13884771e+00 9.62884367e-01
5.04422151e-02 -1.19442892e+00 2.38772720e-01 -8.16573083e-01
9.61704329e-02 -4.08297598e-01 -3.70411068e-01 5.17490268e-01
1.51022866e-01 -8.69692743e-01 9.90820289e-01 7.83004105e-01
6.80918396e-01 -7.16376960e-01 6.10775113e-01 2.60533810e-01
-1.59298408e+00 -1.71084136e-01 -3.24522406e-02 8.10991675e-02
4.54117209e-01 3.06007981e-01 -4.83502924e-01 7.73313582e-01
1.00493145e+00 6.16389513e-01 -7.64546990e-01 6.28961325e-01
-8.36491048e-01 3.60133916e-01 -1.76497936e-01 -2.75085270e-01
1.43400580e-01 -9.53274295e-02 1.87513426e-01 1.24879801e+00
1.88981425e-02 3.22969168e-01 -4.98367175e-02 1.20387840e+00
-7.62372613e-02 9.56072584e-02 -4.04400490e-02 1.65798888e-01
4.35395420e-01 1.26015246e+00 -8.68420422e-01 -5.27453482e-01
-3.37334931e-01 9.09673989e-01 6.87403321e-01 3.86883914e-01
-1.26657259e+00 -3.93547237e-01 2.89258778e-01 1.26134411e-01
3.21024001e-01 -1.26780504e-02 -1.92228645e-01 -8.69079828e-01
2.52280653e-01 -6.95620060e-01 7.34076381e-01 -1.40047538e+00
-8.46143663e-01 6.18447304e-01 4.25025910e-01 -9.41865146e-01
-1.23289794e-01 -5.29041886e-01 -6.84893072e-01 9.14014637e-01
-1.74863243e+00 -1.25912213e+00 -7.79180825e-01 5.69273412e-01
6.78314924e-01 1.69612989e-01 7.51378775e-01 1.43441349e-01
-8.19151938e-01 3.70294064e-01 -9.97441411e-01 7.86776319e-02
4.40066069e-01 -1.21584380e+00 1.66467473e-01 7.77659178e-01
-1.53693810e-01 7.65380681e-01 6.00554943e-01 -3.57514948e-01
-1.31380105e+00 -7.27145612e-01 7.09371567e-01 2.28051376e-02
5.68941176e-01 -3.26521367e-01 -1.02952588e+00 8.24392915e-01
3.64737332e-01 3.34789604e-01 3.65687758e-01 1.00583127e-02
-4.94409084e-01 -2.56351024e-01 -1.01214468e+00 9.06638265e-01
1.26880693e+00 -3.90996277e-01 -8.50616395e-01 -1.34372741e-01
8.08175147e-01 -8.48798752e-01 -6.25877261e-01 3.21156919e-01
5.00413001e-01 -1.08766401e+00 6.81663096e-01 -5.12246132e-01
6.56973660e-01 -9.03430283e-01 -1.21108908e-02 -8.06423843e-01
-1.12676978e-01 -3.82125676e-01 -2.56709725e-01 1.37974119e+00
2.64019161e-01 -3.03150088e-01 4.25306261e-01 6.27499044e-01
1.75521523e-02 -1.06229901e+00 -7.50755191e-01 -3.82013358e-02
-5.18383443e-01 -5.45456529e-01 9.08772230e-01 7.03912258e-01
1.14431663e-03 5.69314599e-01 1.67364985e-01 9.18162838e-02
5.81438504e-02 4.92107958e-01 7.86010981e-01 -8.69822383e-01
-2.30306253e-01 -4.69980836e-01 -1.12036593e-01 -1.32444119e+00
4.65224296e-01 -7.87673473e-01 -4.71893549e-02 -1.95931709e+00
8.42724591e-02 -3.33196670e-01 -2.32452214e-01 6.70842409e-01
-2.82367140e-01 -6.24373436e-01 2.86503136e-01 -1.30913123e-01
-1.05440140e+00 5.42869270e-01 1.31311107e+00 -3.14580441e-01
-1.08525515e-01 -2.18317822e-01 -5.95634818e-01 8.36118042e-01
4.28178132e-01 -1.81468815e-01 -6.88124537e-01 -6.87131107e-01
6.77504003e-01 1.63089320e-01 4.52656269e-01 -8.54348779e-01
5.16155422e-01 -4.43196058e-01 2.01890618e-01 -1.00289547e+00
3.58943373e-01 -1.02216923e+00 1.75864816e-01 2.03773201e-01
-5.17864287e-01 5.87831140e-01 3.45992923e-01 2.68606126e-01
-3.57209384e-01 -3.18407565e-01 7.15826094e-01 -3.20904046e-01
-1.35556042e+00 -9.96456072e-02 -4.72080223e-02 1.17982969e-01
1.26820850e+00 6.47979602e-02 -4.77513433e-01 -8.92047733e-02
-4.91469920e-01 5.46119273e-01 7.83021078e-02 4.69633847e-01
7.20151484e-01 -1.02392769e+00 -4.18587267e-01 -1.94129378e-01
4.94485945e-01 7.19645500e-01 3.68275970e-01 6.67676866e-01
-7.30746090e-01 -1.21508837e-01 -2.49140799e-01 -6.20495856e-01
-1.13224673e+00 6.49065316e-01 5.03336430e-01 -4.93165076e-01
-6.98084593e-01 8.09706032e-01 1.23442737e-02 -1.62670836e-01
1.78959355e-01 -3.73868972e-01 -4.90469456e-01 1.39304966e-01
6.28540277e-01 1.70217797e-01 -1.62046880e-01 -6.56927705e-01
-4.34262156e-01 7.84902692e-01 -1.03965677e-01 1.31551027e-01
9.45381641e-01 -1.40619008e-02 -5.25200784e-01 6.22097313e-01
6.79958403e-01 -1.51112154e-01 -1.05964863e+00 -2.01002955e-01
1.31676450e-01 -1.19091839e-01 -1.74636438e-01 -9.89980102e-01
-9.91814792e-01 9.75486279e-01 2.89451480e-01 -1.97596699e-01
1.43900561e+00 2.56020397e-01 3.93396974e-01 3.39031518e-01
4.16367561e-01 -9.73789573e-01 5.86629622e-02 4.88677740e-01
1.11530471e+00 -8.48234713e-01 1.44349307e-01 -7.68827021e-01
-5.99604726e-01 1.20394337e+00 1.19032168e+00 3.28255177e-01
4.12949473e-01 4.33282048e-01 8.05670097e-02 -4.48488951e-01
-8.39907825e-01 -1.28420755e-01 -5.71546704e-02 2.26915956e-01
5.07973611e-01 -5.88750355e-02 -4.57332999e-01 5.73494256e-01
-2.95856856e-02 2.45618552e-01 -7.82747641e-02 8.13006461e-01
-1.15847364e-01 -1.00790894e+00 -2.38075599e-01 1.09977894e-01
-1.20518491e-01 3.68178487e-02 -2.50793248e-01 7.46335804e-01
3.23163748e-01 6.89328134e-01 1.75852850e-01 8.96931887e-02
7.41763055e-01 2.05863133e-01 3.81838709e-01 -4.22980011e-01
-8.91660571e-01 -3.22371036e-01 2.36668326e-02 -7.47611403e-01
-8.41725826e-01 -4.49168473e-01 -2.14585590e+00 4.56278548e-02
-3.55587639e-02 -1.04527973e-01 2.90301055e-01 1.04098034e+00
3.34207237e-01 1.06363213e+00 9.38263610e-02 -3.21150869e-01
-8.73537734e-02 -7.13580668e-01 -8.74211416e-02 6.07414067e-01
-1.51304394e-01 -8.73964489e-01 1.76441252e-01 5.37461182e-03]
|
[10.424715995788574, 1.4257980585098267]
|
7b5efd11-1e7a-4303-8f08-247cb021de49
|
no-reference-image-quality-assessment-by
|
2108.04165
| null |
https://arxiv.org/abs/2108.04165v3
|
https://arxiv.org/pdf/2108.04165v3.pdf
|
No-Reference Image Quality Assessment by Hallucinating Pristine Features
|
In this paper, we propose a no-reference (NR) image quality assessment (IQA) method via feature level pseudo-reference (PR) hallucination. The proposed quality assessment framework is grounded on the prior models of natural image statistical behaviors and rooted in the view that the perceptually meaningful features could be well exploited to characterize the visual quality. Herein, the PR features from the distorted images are learned by a mutual learning scheme with the pristine reference as the supervision, and the discriminative characteristics of PR features are further ensured with the triplet constraints. Given a distorted image for quality inference, the feature level disentanglement is performed with an invertible neural layer for final quality prediction, leading to the PR and the corresponding distortion features for comparison. The effectiveness of our proposed method is demonstrated on four popular IQA databases, and superior performance on cross-database evaluation also reveals the high generalization capability of our method. The implementation of our method is publicly available on https://github.com/Baoliang93/FPR.
|
['Zhu Li', 'Shiqi Wang', 'Hanwei Zhu', 'Chenqi Kong', 'Lingyu Zhu', 'Baoliang Chen']
|
2021-08-09
| null | null | null | null |
['no-reference-image-quality-assessment']
|
['computer-vision']
|
[ 2.80796498e-01 -2.16093376e-01 -1.65470004e-01 -5.34896314e-01
-1.05349314e+00 -6.77557588e-02 5.73794246e-01 -1.89769655e-01
-9.42399651e-02 5.48010707e-01 3.19310665e-01 2.09419101e-01
-4.29269612e-01 -7.01541603e-01 -5.60841501e-01 -9.89487350e-01
2.64662728e-02 -3.81749541e-01 -3.77648056e-01 -1.14178114e-01
4.34328318e-01 4.18842047e-01 -1.75356936e+00 2.35072240e-01
1.19668639e+00 1.40073061e+00 1.74173728e-01 4.02627587e-01
4.47804898e-01 7.02208698e-01 -4.26490813e-01 -5.62806606e-01
3.20832342e-01 -2.70986587e-01 -4.56313729e-01 3.65731359e-01
3.31784010e-01 -4.70309079e-01 -6.19166195e-01 1.25484002e+00
5.84675312e-01 1.89048842e-01 5.00283599e-01 -1.37965703e+00
-1.15017104e+00 1.47603035e-01 -3.63317758e-01 1.33443639e-01
5.79779506e-01 4.57832545e-01 1.32071424e+00 -1.26002896e+00
3.37057173e-01 1.16887462e+00 1.76055998e-01 2.04106033e-01
-1.27597678e+00 -7.83507884e-01 -1.50649205e-01 6.46952689e-01
-1.59249020e+00 -7.35493720e-01 1.04464328e+00 -4.77204382e-01
4.05325830e-01 2.14898273e-01 5.62782764e-01 1.04524255e+00
2.81298995e-01 7.90892899e-01 1.35978436e+00 -1.59451723e-01
1.30934045e-01 1.19538464e-01 -2.88536325e-02 7.12814569e-01
6.72544539e-02 5.09748340e-01 -7.28975713e-01 2.50821617e-02
7.26449132e-01 2.52676867e-02 -5.98606884e-01 -4.44247782e-01
-1.38809812e+00 5.71620524e-01 6.66358531e-01 1.82802022e-01
-3.68389726e-01 -2.53168523e-01 1.16667166e-01 3.18279117e-01
2.94305861e-01 3.12286079e-01 8.55342150e-02 1.16556175e-01
-7.85039127e-01 -4.56710868e-02 2.38270208e-01 7.62466073e-01
7.76107430e-01 2.11105168e-01 -5.16443610e-01 9.20534670e-01
3.83249521e-01 6.83524668e-01 4.14292544e-01 -1.21840084e+00
3.10778260e-01 4.13714945e-01 3.06069106e-01 -1.40316164e+00
1.94654278e-02 -4.92757857e-01 -1.06718838e+00 3.63160729e-01
1.71324775e-01 4.15203303e-01 -4.87196714e-01 1.68660903e+00
1.15549386e-01 1.87947959e-01 2.23937228e-01 9.98000383e-01
8.11951756e-01 6.36292160e-01 -1.40849531e-01 -3.99566412e-01
1.18625665e+00 -6.13922477e-01 -8.23286593e-01 3.20864499e-01
-1.12213567e-02 -6.93290830e-01 1.23879302e+00 6.86426342e-01
-1.15993857e+00 -1.01970708e+00 -1.25518048e+00 -1.14181176e-01
1.12266904e-02 4.10248995e-01 3.81133527e-01 5.20257115e-01
-1.10383797e+00 7.27641761e-01 -5.89260638e-01 4.66008894e-02
5.88201106e-01 2.11378962e-01 -5.40179670e-01 -2.83034295e-01
-1.16132939e+00 7.50705004e-01 3.43493521e-01 4.61665869e-01
-1.16433525e+00 -5.39290249e-01 -9.33644235e-01 -6.54096305e-02
1.66519061e-01 -7.15886831e-01 5.99708259e-01 -1.00446415e+00
-1.67525840e+00 8.14445734e-01 -7.84869045e-02 -1.72231290e-02
3.74268442e-01 -2.39884585e-01 -6.72214150e-01 2.79683799e-01
9.62078199e-02 6.38660908e-01 1.12510943e+00 -1.50096691e+00
-3.05571645e-01 -4.60000247e-01 1.80189162e-01 3.68304402e-01
-3.63410920e-01 -1.38288215e-01 -2.68824041e-01 -6.52120709e-01
2.67170846e-01 -4.02518243e-01 1.18566900e-01 2.73631930e-01
-5.13577640e-01 -5.27476780e-02 2.35764548e-01 -8.87021184e-01
1.04346466e+00 -2.31499243e+00 2.72706032e-01 2.17876807e-01
3.18860888e-01 -3.72804748e-03 -3.51131290e-01 3.05718303e-01
-1.45606384e-01 -1.99365154e-01 -3.38322252e-01 -3.94409113e-02
1.29678845e-01 -1.34592369e-01 -2.58601576e-01 6.72310829e-01
4.74194437e-01 8.98737609e-01 -9.24868405e-01 -4.70971137e-01
3.26900452e-01 6.75739467e-01 -4.03877288e-01 6.41577363e-01
2.96087176e-01 6.70699656e-01 -3.70544791e-01 7.64497459e-01
7.66263425e-01 -1.69563726e-01 -9.59208012e-02 -8.72927248e-01
4.72773500e-02 -5.20565500e-03 -9.61791635e-01 1.86205912e+00
-3.20608139e-01 2.13199273e-01 -2.50875354e-01 -7.84258723e-01
1.13204908e+00 2.06694186e-01 3.06307614e-01 -1.18660283e+00
3.66560556e-02 1.42536074e-01 8.24814364e-02 -6.45080686e-01
3.86891872e-01 4.32678685e-02 2.26817235e-01 1.77448556e-01
2.84135431e-01 -7.27087110e-02 -6.54937699e-02 8.81441906e-02
5.77777088e-01 2.79056430e-01 4.70005989e-01 -1.89836428e-01
7.21250057e-01 -5.65920532e-01 6.84288740e-01 4.52592909e-01
-4.33676183e-01 6.26524866e-01 1.78204015e-01 -1.02103524e-01
-1.13643229e+00 -1.45743072e+00 -4.46112931e-01 5.63979387e-01
4.28928018e-01 -9.35982987e-02 -4.70446736e-01 -2.90556818e-01
-1.89501941e-01 6.31775022e-01 -5.83388150e-01 -3.87715667e-01
6.34819968e-03 -5.25786459e-01 3.41660768e-01 1.78542688e-01
7.30803668e-01 -1.06811666e+00 -2.11856246e-01 -7.61923119e-02
-3.58808428e-01 -8.84641171e-01 -3.16531658e-01 -4.45360929e-01
-6.73660517e-01 -9.10137951e-01 -7.34163284e-01 -4.71975088e-01
5.72632968e-01 3.44583511e-01 8.82666409e-01 -3.32989404e-03
-2.36327842e-01 2.98179150e-01 -2.52225161e-01 2.96628654e-01
-2.70461410e-01 -7.13676631e-01 1.94683135e-01 6.35613918e-01
3.00018564e-02 -7.72544265e-01 -1.00397217e+00 2.82816589e-01
-8.94904912e-01 2.02038378e-01 7.64514923e-01 1.00054204e+00
8.18676889e-01 1.65466458e-01 7.29236722e-01 -3.25743765e-01
6.00920737e-01 -3.94478232e-01 -4.34879959e-01 1.98212147e-01
-7.32651830e-01 -1.66819483e-01 2.94914484e-01 -3.05591345e-01
-1.23341608e+00 -3.87432903e-01 -5.52158616e-02 -5.10843992e-01
-1.68266714e-01 3.19931328e-01 -8.85923505e-01 -1.37256756e-01
4.14662838e-01 5.50028443e-01 1.06420211e-01 -2.80290365e-01
5.27955294e-01 5.95906794e-01 7.21798122e-01 -6.00984693e-01
8.73168945e-01 5.05546391e-01 -1.09602667e-01 -5.56239307e-01
-8.99525404e-01 -2.59289086e-01 -6.67699099e-01 -4.08206552e-01
7.34888434e-01 -1.11421907e+00 -7.61855423e-01 5.49124479e-01
-8.60229969e-01 6.03221096e-02 -2.14900091e-01 5.34120023e-01
-8.21776032e-01 6.46747589e-01 -5.17518759e-01 -8.05382252e-01
-2.51291364e-01 -1.37921286e+00 9.10028517e-01 1.81217924e-01
2.21930474e-01 -7.55637944e-01 -1.19616278e-01 5.69381654e-01
3.03805918e-01 3.07409286e-01 9.12222743e-01 -2.39307761e-01
-7.49738514e-01 6.81593120e-02 -5.40133774e-01 6.53564334e-01
3.25462222e-01 -1.17399208e-01 -1.14136171e+00 -5.00785768e-01
2.33861849e-01 -4.81453687e-01 5.86638749e-01 2.84587324e-01
1.33313131e+00 -4.33526754e-01 2.96367764e-01 8.43801796e-01
1.55007863e+00 1.49550453e-01 9.27595258e-01 1.47327751e-01
7.07555771e-01 5.10601521e-01 6.93677843e-01 5.40831625e-01
4.89330739e-01 7.05652058e-01 5.42319536e-01 -1.67176366e-01
-2.25703090e-01 -3.10223162e-01 5.31037986e-01 1.06636536e+00
-1.02472782e-01 -1.74758658e-01 -5.73608339e-01 3.83260638e-01
-1.44007707e+00 -1.10322559e+00 1.48799866e-01 2.30460167e+00
7.70696104e-01 -1.71127573e-01 -2.41636544e-01 2.72783369e-01
6.66463435e-01 3.17678362e-01 -7.10464835e-01 5.05839176e-02
-3.49399477e-01 -1.20696403e-01 -1.19595535e-01 3.98074120e-01
-9.52338636e-01 4.24012095e-01 5.65272570e+00 9.44957256e-01
-8.61357868e-01 -2.00534109e-02 8.51747751e-01 2.81146988e-02
-4.30236667e-01 -1.24458767e-01 -1.59733489e-01 3.35981905e-01
7.00373352e-01 -3.35150778e-01 6.46917403e-01 4.06427801e-01
4.98559535e-01 1.74809080e-02 -1.08064961e+00 1.29003155e+00
2.42284372e-01 -9.57554221e-01 3.90980184e-01 4.13780622e-02
6.06456399e-01 -2.76127756e-01 6.26509964e-01 5.10507822e-02
-1.02973051e-01 -1.10153961e+00 6.65775418e-01 1.02794242e+00
1.06057394e+00 -8.22189987e-01 5.45094073e-01 8.07740018e-02
-8.38226080e-01 -1.10639408e-01 -5.73198497e-01 1.19158641e-01
-6.21232763e-02 6.57875776e-01 -2.04541638e-01 9.62325275e-01
6.22340918e-01 1.13686669e+00 -8.21565151e-01 9.83197272e-01
-2.75698125e-01 3.96705389e-01 3.69462013e-01 5.23827851e-01
-3.00443739e-01 -4.89703178e-01 5.82619309e-01 7.50574052e-01
4.22697961e-01 2.09754363e-01 -5.29404543e-02 1.19916308e+00
-1.36942267e-01 3.03706229e-01 -4.24423009e-01 1.56987578e-01
3.25470120e-01 1.27481413e+00 -2.56635487e-01 -2.07231462e-01
-3.63302052e-01 1.08697009e+00 3.08442980e-01 4.21647578e-01
-7.72536576e-01 -3.09389740e-01 5.59006035e-01 -3.08850646e-01
1.01579800e-01 1.12394027e-01 -1.18571475e-01 -1.55027807e+00
2.72779405e-01 -1.13908505e+00 1.56966031e-01 -1.11408818e+00
-1.62383604e+00 7.67106891e-01 -1.53277397e-01 -1.74236572e+00
1.57699779e-01 -4.47439224e-01 -5.43855131e-01 1.00539565e+00
-1.73206723e+00 -1.12751436e+00 -5.25895953e-01 8.49541128e-01
2.84759700e-01 -1.09818690e-01 7.51432896e-01 3.77377868e-01
-5.75308442e-01 7.55921304e-01 8.13626871e-02 9.55036432e-02
6.75122797e-01 -1.03852177e+00 -1.48990571e-01 8.91591609e-01
1.75889179e-01 5.22574902e-01 5.29906571e-01 -3.62408161e-01
-1.51484859e+00 -9.49752688e-01 4.64206606e-01 -2.11143494e-01
6.96258664e-01 -1.24109946e-01 -1.14737368e+00 2.13739216e-01
2.65678763e-01 2.00700104e-01 7.62078047e-01 -2.07091525e-01
-7.31332123e-01 -3.61976981e-01 -1.17078674e+00 4.45079505e-01
9.79037404e-01 -7.80059040e-01 -4.87205982e-01 1.02035142e-01
7.51694739e-01 -6.80372864e-02 -1.28910708e+00 4.31704760e-01
6.81195736e-01 -1.21066940e+00 1.09325981e+00 -9.77417305e-02
6.67818785e-01 -5.22698700e-01 -6.09707654e-01 -1.16804349e+00
-3.71488959e-01 -2.08182201e-01 -6.52899146e-02 1.36051786e+00
1.76244363e-01 -5.14441192e-01 2.06165314e-01 3.65298271e-01
-8.03453848e-03 -6.67787790e-01 -9.02590513e-01 -7.09763288e-01
-1.50196001e-01 -3.76532465e-01 6.25024498e-01 9.32954550e-01
-1.25848368e-01 1.51932910e-02 -6.58319056e-01 4.81167972e-01
1.00645363e+00 2.94725329e-01 4.76068079e-01 -8.84696364e-01
-4.85458493e-01 -3.13138872e-01 -6.59145653e-01 -8.08525503e-01
7.37925768e-02 -1.04210114e+00 1.65226497e-02 -1.14678860e+00
5.47427595e-01 -2.34792233e-01 -7.36743450e-01 1.04971230e-01
-2.61812657e-01 4.02126104e-01 2.67415553e-01 5.57675779e-01
-7.11155295e-01 1.11106360e+00 1.55309105e+00 -3.15247774e-01
8.57942328e-02 -6.32336214e-02 -7.53272593e-01 4.04505253e-01
6.47393227e-01 -1.25037193e-01 -5.08510709e-01 -4.06706810e-01
-7.06672296e-02 4.26794320e-01 6.89324558e-01 -9.71489072e-01
-1.80684440e-02 -1.23583913e-01 5.50241530e-01 -3.95676345e-01
4.86511916e-01 -6.41512752e-01 1.44216478e-01 2.21862257e-01
-5.42469501e-01 -1.13527432e-01 -1.74792647e-01 6.73177540e-01
-5.22601366e-01 1.50028080e-01 8.16942692e-01 1.73715860e-01
-6.33995891e-01 5.57104230e-01 3.27199787e-01 -1.38537750e-01
7.61786938e-01 -2.43776470e-01 -1.91882744e-01 -3.84065270e-01
-8.35402548e-01 -4.35851440e-02 5.50807416e-01 4.21658367e-01
1.18604922e+00 -1.73769546e+00 -9.53073442e-01 2.99824715e-01
5.78878284e-01 -4.54978198e-01 6.54377162e-01 8.13797712e-01
-2.46983487e-03 1.15578726e-01 -5.86461127e-01 -6.97407782e-01
-8.24153423e-01 7.80129015e-01 3.60796243e-01 1.07500078e-02
-4.79614556e-01 4.77370083e-01 3.57756853e-01 -2.20088437e-01
4.06720936e-02 9.39166024e-02 -4.07225877e-01 -1.47414207e-01
6.79813921e-01 4.23939556e-01 -2.29845773e-02 -1.06087673e+00
-1.29962727e-01 4.79888707e-01 1.45019591e-01 -1.33241147e-01
1.20577800e+00 -5.73766351e-01 -9.19073224e-02 5.78452349e-01
1.42134404e+00 -9.26887915e-02 -1.43947768e+00 -4.94100660e-01
-1.11221679e-01 -9.36270058e-01 1.90544888e-01 -8.93470883e-01
-1.22281766e+00 8.90988052e-01 1.01933825e+00 -2.25750998e-01
1.52989972e+00 -2.12279439e-01 3.56862366e-01 8.16686153e-02
5.18395960e-01 -4.94782388e-01 2.66978502e-01 -1.70098603e-01
1.26632094e+00 -1.45914030e+00 -5.28249377e-03 -2.11220264e-01
-6.71517611e-01 9.09460545e-01 5.41517317e-01 -2.37639859e-01
5.03762245e-01 -3.37976605e-01 -2.44539771e-02 -1.14001773e-01
-6.09728932e-01 -6.55225217e-02 6.64638042e-01 7.02297688e-01
2.39668816e-01 2.15854809e-01 -9.28922445e-02 5.20193577e-01
-1.60399452e-01 -6.41630962e-02 4.47869509e-01 2.98532248e-01
-1.55821338e-01 -8.72717261e-01 -2.20257536e-01 2.60944605e-01
-2.21191555e-01 -1.93201900e-01 2.09396899e-01 4.34557438e-01
1.69139415e-01 1.14757359e+00 -1.62180096e-01 -5.98445356e-01
3.61616015e-01 -2.62832701e-01 4.84158278e-01 -3.62922519e-01
-1.49676666e-01 -7.62308436e-03 -3.24622065e-01 -1.06620288e+00
-6.52442396e-01 -6.48818195e-01 -7.97888875e-01 -9.29563269e-02
-2.15154782e-01 -5.26489317e-02 4.55037117e-01 6.93801641e-01
4.03711557e-01 3.53314936e-01 1.23220146e+00 -8.79027009e-01
-5.56766152e-01 -8.44509900e-01 -6.71177745e-01 7.63356447e-01
4.34791863e-01 -7.80575454e-01 -4.79373902e-01 1.95761621e-01]
|
[11.902091979980469, -1.86527419090271]
|
b7d3025d-69ec-43e1-aeaa-0f93a761e2be
|
emotion-cause-pair-extraction-as-sequence
| null | null |
https://aclanthology.org/2020.emnlp-main.289
|
https://aclanthology.org/2020.emnlp-main.289.pdf
|
Emotion-Cause Pair Extraction as Sequence Labeling Based on A Novel Tagging Scheme
|
The task of emotion-cause pair extraction deals with finding all emotions and the corresponding causes in unannotated emotion texts. Most recent studies are based on the likelihood of Cartesian product among all clause candidates, resulting in a high computational cost. Targeting this issue, we regard the task as a sequence labeling problem and propose a novel tagging scheme with coding the distance between linked components into the tags, so that emotions and the corresponding causes can be extracted simultaneously. Accordingly, an end-to-end model is presented to process the input texts from left to right, always with linear time complexity, leading to a speed up. Experimental results show that our proposed model achieves the best performance, outperforming the state-of-the-art method by 2.26{\%} (p{\textless}0.001) in F1 measure.
|
['Ruifeng Xu', 'Jianzhu Bao', 'Chuang Fan', 'Chaofa Yuan']
| null | null | null | null |
emnlp-2020-11
|
['emotion-cause-pair-extraction']
|
['natural-language-processing']
|
[ 2.43499309e-01 3.76311600e-01 -3.43333125e-01 -3.38552654e-01
-7.63687849e-01 -6.57443464e-01 3.38378221e-01 4.78444010e-01
-5.06515741e-01 6.92562461e-01 6.00059815e-02 -2.71033365e-02
-1.45655215e-01 -5.89392781e-01 -3.98949683e-01 -5.52880168e-01
-2.04023749e-01 2.56912589e-01 4.27942835e-02 -1.03249680e-02
1.50692627e-01 -7.27239847e-02 -1.51630878e+00 1.53771818e-01
9.97642994e-01 1.03322172e+00 3.22352238e-02 2.93107003e-01
-4.41407055e-01 8.80397797e-01 -4.77616757e-01 -1.05969799e+00
-3.47281903e-01 -5.87465167e-01 -8.12503159e-01 -1.57243326e-01
-2.32397884e-01 2.12681368e-01 6.18066341e-02 1.20739317e+00
4.38164324e-01 3.16929221e-02 5.81529558e-01 -1.27528322e+00
-2.56047249e-01 9.71926451e-01 -8.08532178e-01 -1.62509009e-01
5.08349001e-01 -7.32466280e-01 1.54691172e+00 -8.71839821e-01
4.65747714e-01 9.79837596e-01 3.20368409e-01 3.86928678e-01
-9.60023761e-01 -6.94353521e-01 3.75244141e-01 2.76698142e-01
-1.43125582e+00 -1.49126917e-01 8.90822232e-01 -3.73997055e-02
9.85410333e-01 3.99582297e-01 5.71962833e-01 7.38916934e-01
-7.79686272e-02 8.79690707e-01 1.08376968e+00 -5.19808829e-01
3.00019503e-01 7.03449454e-03 2.05126807e-01 7.00021803e-01
1.75918579e-01 -4.90269929e-01 -6.71483397e-01 -1.87242433e-01
3.27152014e-01 -4.07140464e-01 -2.67302483e-01 -2.00277939e-03
-1.10448015e+00 7.83023298e-01 1.98993579e-01 5.43695509e-01
-3.93511653e-01 -1.73166313e-03 3.53202373e-01 3.56016345e-02
5.19241512e-01 3.12563092e-01 -7.58654475e-01 -3.88448387e-01
-6.10709667e-01 2.75076628e-01 8.00936103e-01 1.02776659e+00
3.76344681e-01 -4.88263965e-01 6.98676407e-02 9.20223594e-01
2.05105290e-01 3.08699310e-01 1.72124147e-01 -5.05060732e-01
4.13789243e-01 8.04225862e-01 2.77060959e-02 -1.29590392e+00
-7.62684226e-01 -4.98699725e-01 -7.71103323e-01 -5.37587702e-01
2.76549309e-01 -5.54728687e-01 -3.85394514e-01 1.95944059e+00
6.12950385e-01 1.81382731e-01 3.18976969e-01 8.52277458e-01
8.02769363e-01 8.05437326e-01 2.39345104e-01 -7.45578408e-01
1.85722005e+00 -6.83885515e-01 -1.15046930e+00 -4.31560904e-01
8.60950291e-01 -9.31330979e-01 8.00378978e-01 6.04158103e-01
-7.53817677e-01 -7.43283927e-02 -8.39581549e-01 4.12192605e-02
-1.22852206e-01 3.39324027e-01 9.14002240e-01 4.11730021e-01
-4.67309862e-01 3.42276841e-01 -5.03468871e-01 -1.58055991e-01
-2.55644117e-02 3.33925039e-01 -2.57938772e-01 8.13036561e-02
-1.41578281e+00 6.66154504e-01 6.36903226e-01 1.49229541e-01
1.71403326e-02 -4.90422636e-01 -7.55482733e-01 1.08345494e-01
7.10036695e-01 -3.52354556e-01 1.12416971e+00 -8.78870428e-01
-1.32873249e+00 6.74534440e-01 -2.83428311e-01 -9.28680524e-02
1.07850544e-01 -5.20599246e-01 -6.76511705e-01 9.70920697e-02
1.96968801e-02 4.97376204e-01 4.37468946e-01 -1.21497393e+00
-9.01754200e-01 -2.78973460e-01 -1.61874387e-02 3.29827666e-01
-6.17433786e-01 3.88829321e-01 -8.54379177e-01 -6.55983627e-01
2.16771811e-01 -9.69256818e-01 -1.86958581e-01 -4.86256719e-01
-4.95887786e-01 -7.55188882e-01 3.82901460e-01 -6.23057008e-01
1.63440168e+00 -2.14864230e+00 1.54611006e-01 8.46588090e-02
1.95692644e-01 -9.86460745e-02 1.11644693e-01 3.13305765e-01
-2.27650106e-01 1.58096597e-01 -1.80309013e-01 -1.16201356e-01
2.22262561e-01 5.37918881e-02 -1.20455772e-01 2.88015425e-01
2.63839543e-01 5.86931705e-01 -1.08223796e+00 -8.72849822e-01
-1.69450760e-01 3.25568557e-01 -3.98303002e-01 4.15766358e-01
-1.35244861e-01 -8.32416583e-03 -5.80521822e-01 2.85516411e-01
5.85573316e-01 -2.05743775e-01 6.63703442e-01 -1.65375590e-01
1.21503705e-02 4.38729882e-01 -1.22245157e+00 1.55103970e+00
-4.53897655e-01 1.18974984e-01 -1.09107532e-01 -6.86271846e-01
1.11194062e+00 5.36260247e-01 4.85249072e-01 -5.72149932e-01
4.21689510e-01 2.77648449e-01 4.35526483e-02 -6.23415291e-01
4.37746167e-01 -3.68179202e-01 -6.53588414e-01 9.89294425e-02
-5.21462485e-02 2.82919824e-01 2.80904084e-01 2.09061593e-01
1.04898620e+00 -1.59141831e-02 4.58966374e-01 -1.38938636e-01
3.90053064e-01 -6.52284026e-02 9.07649994e-01 1.09812587e-01
8.05023834e-02 8.81866738e-02 1.02484012e+00 -1.48540154e-01
-6.44302309e-01 -4.77771521e-01 1.68954149e-01 8.63419890e-01
2.72169530e-01 -7.33605385e-01 -8.05472314e-01 -9.41547096e-01
-4.23161626e-01 8.67705047e-01 -5.94539583e-01 1.40878826e-01
-5.03538489e-01 -8.10682356e-01 4.56486613e-01 3.09881121e-01
2.38417476e-01 -9.02264178e-01 -7.65795469e-01 2.67101675e-01
-6.09038949e-01 -1.32903421e+00 -1.31620303e-01 4.35166121e-01
-5.35226345e-01 -9.12525356e-01 -5.66620648e-01 -8.45515788e-01
5.96580565e-01 -2.37436965e-01 1.08240914e+00 8.92420337e-02
1.42534956e-01 -4.11564678e-01 -6.94957733e-01 -2.57177144e-01
5.51651157e-02 1.72432829e-02 -1.52240336e-01 1.05010062e-01
5.57970166e-01 -5.10783434e-01 -3.40531379e-01 8.25762972e-02
-6.83994114e-01 3.12343836e-01 7.23251760e-01 7.33383715e-01
6.27673328e-01 4.17923957e-01 5.95385432e-01 -9.75764215e-01
5.44008732e-01 -4.85734791e-01 -4.53798145e-01 2.91499108e-01
-6.90777481e-01 1.62966698e-01 8.75186384e-01 -3.44956666e-01
-1.16596889e+00 3.45560730e-01 -1.23232171e-01 -2.10345760e-01
-2.15235978e-01 7.73529291e-01 -4.37919497e-01 6.24512315e-01
2.64015198e-02 7.69568458e-02 -4.85247403e-01 -4.76755321e-01
6.44476831e-01 5.09077370e-01 5.44712245e-01 -4.70712572e-01
3.71795297e-01 -1.53972479e-02 1.74641505e-01 -4.00816739e-01
-1.14513314e+00 -5.82782269e-01 -4.06059057e-01 -2.02588916e-01
8.56443524e-01 -7.90038884e-01 -9.68635440e-01 4.72781658e-02
-1.35678089e+00 2.10259631e-01 1.64655373e-01 5.58919370e-01
-2.06664488e-01 3.38323504e-01 -6.71283901e-01 -9.65200245e-01
-3.75244230e-01 -6.20959520e-01 8.88066471e-01 3.04712892e-01
-5.99280775e-01 -6.67470038e-01 -7.89295286e-02 2.20357329e-01
-2.53799915e-01 3.93554777e-01 1.18178487e+00 -9.65555549e-01
1.57707129e-02 -2.24415973e-01 -2.63688445e-01 -2.26234257e-01
-1.88193575e-04 -1.64886966e-01 -7.72981584e-01 2.53257990e-01
-3.08671631e-02 -3.23140502e-01 4.83350009e-01 -1.30259857e-01
8.39644611e-01 -5.58738291e-01 -3.83032739e-01 3.73613648e-02
1.39192355e+00 3.46645236e-01 4.32019889e-01 8.95455331e-02
6.11742258e-01 1.03824162e+00 1.06691158e+00 7.21231341e-01
4.67945576e-01 6.45679653e-01 3.94102246e-01 -3.00192058e-01
3.47870857e-01 -3.53810877e-01 1.62877217e-01 1.11600578e+00
2.52734214e-01 -6.00583673e-01 -7.09673941e-01 6.91538095e-01
-2.07083607e+00 -7.26238072e-01 -5.71157694e-01 1.88366032e+00
1.00358498e+00 2.42391646e-01 8.57380684e-03 3.54969710e-01
7.94626117e-01 6.41252249e-02 -7.39072859e-02 -5.02707958e-01
5.22202626e-02 6.80232421e-02 1.18429214e-01 3.00819874e-01
-1.12288034e+00 1.01078928e+00 5.28190041e+00 1.17541587e+00
-7.71039069e-01 -4.41734158e-02 6.41746879e-01 4.30260375e-02
-4.11813557e-01 1.19478665e-01 -5.80738604e-01 5.24794698e-01
7.95381248e-01 -1.51966721e-01 1.84225470e-01 7.44316399e-01
1.36287287e-01 -1.96474776e-01 -7.87043214e-01 1.05093443e+00
1.02684177e-01 -5.67356110e-01 -1.80690020e-01 -1.21880546e-01
5.30982137e-01 -6.41574502e-01 -3.76119524e-01 2.17049062e-01
2.30741892e-02 -5.77204227e-01 7.37493634e-01 2.89223611e-01
6.38684928e-01 -1.41019154e+00 9.92283642e-01 2.78613180e-01
-1.39038444e+00 1.38403565e-01 -1.25649005e-01 -1.54361039e-01
3.32507461e-01 1.09995425e+00 -5.98565161e-01 8.24230552e-01
5.87769628e-01 5.08690655e-01 -1.98282391e-01 7.77252972e-01
-6.94760561e-01 7.39919484e-01 -2.77402699e-01 -6.65076971e-01
1.08249210e-01 -1.52887674e-02 3.84942025e-01 1.36279511e+00
4.93504643e-01 6.66077018e-01 2.35922709e-01 5.11936128e-01
-3.49327654e-01 7.14623511e-01 -2.36410409e-01 -7.35126808e-02
4.51768607e-01 1.72819400e+00 -1.33169615e+00 -3.08535814e-01
-2.34525368e-01 1.10052371e+00 4.37606692e-01 -1.59399778e-01
-1.03231692e+00 -8.53950799e-01 3.43767852e-01 -3.66623223e-01
3.49206090e-01 6.56224787e-02 -2.77386695e-01 -9.95281875e-01
1.11311190e-01 -6.72380567e-01 5.67538917e-01 -6.50972605e-01
-1.21319747e+00 8.11877310e-01 -9.64972004e-02 -9.86032724e-01
-2.66526192e-01 -4.11452889e-01 -2.34424815e-01 4.33472991e-01
-1.38758290e+00 -8.35210443e-01 -5.39660603e-02 3.28343511e-01
1.78721592e-01 4.78601485e-01 9.42348242e-01 4.38915640e-01
-6.51460171e-01 6.01548493e-01 -3.27887326e-01 1.82359889e-01
6.61915064e-01 -1.25682998e+00 -2.84089986e-02 9.98307705e-01
1.58998683e-01 4.95243937e-01 7.81868279e-01 -5.66198707e-01
-1.30352128e+00 -7.94941187e-01 1.94214928e+00 2.02258304e-02
6.55705214e-01 -4.97318238e-01 -7.45965302e-01 1.87686980e-01
4.22495723e-01 -2.53846139e-01 8.42853010e-01 4.72415626e-01
-3.21185261e-01 -2.34322902e-02 -9.06431615e-01 6.42108858e-01
9.98265266e-01 -1.22808531e-01 -5.76275170e-01 6.41608238e-02
9.34203863e-01 -3.18468302e-01 -9.15470183e-01 4.25631493e-01
3.58030766e-01 -8.03698480e-01 4.22821224e-01 -1.71847761e-01
6.73623502e-01 -3.68823290e-01 1.27556724e-02 -1.02286220e+00
-3.27596188e-01 -8.09528887e-01 2.18730122e-02 1.85990858e+00
7.15478241e-01 -2.23300248e-01 4.26902115e-01 6.87304020e-01
3.91119644e-02 -9.36788023e-01 -9.49129045e-01 -4.92871732e-01
-4.34961081e-01 -6.77397966e-01 6.27594948e-01 1.10400426e+00
5.90426505e-01 9.55265641e-01 -5.15141547e-01 1.34460628e-01
4.01787698e-01 6.20642185e-01 1.74476519e-01 -1.27722633e+00
-2.88267493e-01 -3.69254291e-01 -8.63073245e-02 -9.06670630e-01
2.98662901e-01 -7.73200929e-01 3.13493311e-01 -1.39704418e+00
4.37331885e-01 -4.88641113e-01 -4.11720484e-01 9.28065836e-01
-6.89005494e-01 7.19699189e-02 1.90759733e-01 -7.34693408e-02
-9.31412697e-01 5.71399808e-01 8.69221151e-01 2.70091712e-01
-7.20432028e-02 -3.46270889e-01 -9.10274088e-01 8.42500567e-01
7.26613700e-01 -8.45810175e-01 -3.26140642e-01 -1.68870226e-01
6.70792699e-01 2.26028487e-01 -5.34167662e-02 -6.72889948e-01
1.34598136e-01 -1.97930127e-01 7.88440779e-02 -5.24099469e-01
1.22036599e-01 -8.90442967e-01 3.35968174e-02 1.44250646e-01
-3.93023968e-01 3.48065138e-01 1.46956578e-01 3.52917165e-01
-2.73082852e-01 -3.81417781e-01 3.87760013e-01 3.41909081e-01
-5.52941740e-01 -1.27706453e-01 -2.99934208e-01 1.57153666e-01
1.01097226e+00 5.02893627e-01 -6.18784018e-02 -2.34746948e-01
-4.16692138e-01 1.35190696e-01 1.52662583e-02 3.37557048e-01
3.56127888e-01 -1.27801657e+00 -4.52327192e-01 -3.32644641e-01
1.02173246e-01 -1.24032676e-01 2.12033227e-01 8.03599536e-01
1.63018197e-01 4.31123137e-01 2.85782337e-01 -4.87160385e-02
-1.52925134e+00 7.07310557e-01 -1.80089757e-01 -6.07846320e-01
-2.94737637e-01 1.00327849e+00 -5.28894551e-02 -1.46686152e-01
1.91942155e-01 -1.96125492e-01 -5.59174359e-01 3.59790117e-01
3.27196658e-01 1.38833448e-01 2.01724544e-02 -6.06537879e-01
-6.57917321e-01 5.57780385e-01 1.73619315e-01 -1.29709989e-01
1.25186431e+00 -4.28951532e-02 -4.67161864e-01 3.63413543e-01
1.23452711e+00 3.15259427e-01 -6.87336564e-01 -1.37535498e-01
3.48103076e-01 -9.44128186e-02 8.76614545e-03 -9.55597997e-01
-9.99222159e-01 7.14442372e-01 2.63841480e-01 2.81940132e-01
1.39082205e+00 1.73801616e-01 8.88425946e-01 1.67503402e-01
2.19661906e-01 -9.67496336e-01 -1.97483763e-01 4.84537214e-01
5.00015497e-01 -8.35752189e-01 -2.64923368e-02 -9.37430859e-01
-8.31530809e-01 9.52141583e-01 5.27275920e-01 1.76212355e-01
4.19987142e-01 4.85304952e-01 -3.81819718e-02 -2.27157950e-01
-9.95208204e-01 -3.69525045e-01 3.70393664e-01 -3.53120007e-02
7.87588239e-01 2.76311129e-01 -1.01350653e+00 1.33335817e+00
-2.96222389e-01 -1.79939374e-01 8.66856575e-02 8.47820640e-01
-2.48997688e-01 -1.40599418e+00 -1.50275692e-01 2.27051795e-01
-9.42436635e-01 -2.58275986e-01 -6.00855350e-01 5.62049329e-01
3.77965271e-01 1.35921121e+00 -7.92336240e-02 -5.52488923e-01
3.67131174e-01 1.81492791e-01 2.19125956e-01 -2.25467503e-01
-6.80891156e-01 6.27982795e-01 5.61809123e-01 -3.41093093e-01
-6.52349770e-01 -5.98193049e-01 -1.81440198e+00 -7.82190543e-03
-6.17110312e-01 5.57291746e-01 4.11162972e-01 1.02454054e+00
4.61346209e-01 5.72188079e-01 7.86501944e-01 -1.98926106e-01
-7.65140578e-02 -8.19493294e-01 -5.85220873e-01 4.80708539e-01
-2.12934017e-01 -5.56609154e-01 -2.05837339e-01 -2.45764349e-02]
|
[12.611377716064453, 6.167050361633301]
|
43c518e0-64fa-48db-90eb-4b21d019ea19
|
revisiting-grammatical-error-correction
|
2211.01635
| null |
https://arxiv.org/abs/2211.01635v1
|
https://arxiv.org/pdf/2211.01635v1.pdf
|
Revisiting Grammatical Error Correction Evaluation and Beyond
|
Pretraining-based (PT-based) automatic evaluation metrics (e.g., BERTScore and BARTScore) have been widely used in several sentence generation tasks (e.g., machine translation and text summarization) due to their better correlation with human judgments over traditional overlap-based methods. Although PT-based methods have become the de facto standard for training grammatical error correction (GEC) systems, GEC evaluation still does not benefit from pretrained knowledge. This paper takes the first step towards understanding and improving GEC evaluation with pretraining. We first find that arbitrarily applying PT-based metrics to GEC evaluation brings unsatisfactory correlation results because of the excessive attention to inessential systems outputs (e.g., unchanged parts). To alleviate the limitation, we propose a novel GEC evaluation metric to achieve the best of both worlds, namely PT-M2 which only uses PT-based metrics to score those corrected parts. Experimental results on the CoNLL14 evaluation task show that PT-M2 significantly outperforms existing methods, achieving a new state-of-the-art result of 0.949 Pearson correlation. Further analysis reveals that PT-M2 is robust to evaluate competitive GEC systems. Source code and scripts are freely available at https://github.com/pygongnlp/PT-M2.
|
['Min Zhang', 'Heyan Huang', 'Xuebo Liu', 'Peiyuan Gong']
|
2022-11-03
| null | null | null | null |
['grammatical-error-correction']
|
['natural-language-processing']
|
[ 1.10821404e-01 1.51511803e-01 7.14310408e-02 -4.00272548e-01
-1.24840009e+00 -4.36316937e-01 5.38557351e-01 3.50016415e-01
-5.36380351e-01 9.90905762e-01 3.40249538e-01 -4.00068432e-01
1.04150817e-01 -5.22022605e-01 -6.22158825e-01 -4.45641667e-01
8.31362903e-02 5.58885992e-01 3.09779756e-02 -4.70183462e-01
5.26310086e-01 -9.63521600e-02 -1.24291086e+00 4.81853992e-01
1.60350406e+00 5.49097598e-01 2.80003250e-01 7.92307079e-01
3.13757127e-03 3.53510976e-01 -1.07925308e+00 -8.87985647e-01
-1.73763826e-01 -8.87348592e-01 -9.18511689e-01 -5.24933100e-01
2.92943358e-01 4.53385152e-02 2.86573112e-01 1.20893431e+00
6.52708530e-01 7.33291432e-02 5.95198572e-01 -9.30259228e-01
-6.09691620e-01 9.69015300e-01 -3.43413502e-01 7.66917542e-02
4.61185634e-01 7.60820359e-02 1.13238084e+00 -1.10522807e+00
5.86641490e-01 1.20341527e+00 7.08312392e-01 7.60795891e-01
-9.54826057e-01 -3.95895630e-01 -5.78753650e-02 3.24569911e-01
-1.35535157e+00 -2.80954808e-01 3.93181056e-01 -2.04859488e-02
1.38568175e+00 5.39258718e-01 1.81150645e-01 9.20947731e-01
2.11427465e-01 8.75603914e-01 1.02237618e+00 -6.73879206e-01
8.67712405e-03 9.40482244e-02 7.16361627e-02 6.29911721e-01
3.08048874e-01 -2.19993860e-01 -4.06267613e-01 2.83968057e-02
2.13638186e-01 -6.74860239e-01 -3.90135229e-01 2.91767836e-01
-1.31641114e+00 8.26425314e-01 2.91996896e-01 5.64628959e-01
-2.83496678e-01 1.55098751e-01 5.26675880e-01 3.63960207e-01
5.62796175e-01 9.26326156e-01 -5.79791486e-01 -7.27435231e-01
-9.99568582e-01 3.69347930e-01 7.59140491e-01 9.35384393e-01
3.56964737e-01 7.23292679e-02 -4.79960829e-01 1.12065613e+00
-4.22901697e-02 5.51951170e-01 7.38114476e-01 -8.12605023e-01
8.37166190e-01 5.14268935e-01 -1.26677221e-02 -8.85464787e-01
-3.59630555e-01 -6.82097375e-01 -9.52757061e-01 -2.01165706e-01
2.18015850e-01 -3.12300086e-01 -7.91815102e-01 1.81844020e+00
-1.01264067e-01 -2.97152400e-01 1.16415530e-01 8.34967315e-01
9.05984879e-01 7.06300676e-01 -1.15785725e-01 -2.28687420e-01
1.04704165e+00 -1.14466321e+00 -6.84466004e-01 -1.18418522e-01
1.16625214e+00 -9.38435853e-01 1.23717761e+00 3.24811041e-01
-1.11418760e+00 -4.15833771e-01 -1.09448707e+00 3.94161493e-02
-2.71952897e-01 4.01393473e-01 5.12054205e-01 7.13878572e-01
-1.12495720e+00 9.52755272e-01 -7.25536108e-01 -4.42897320e-01
1.85601115e-02 1.57553673e-01 -2.59329528e-01 2.72252150e-02
-1.30366933e+00 1.24480987e+00 7.19796956e-01 6.67013675e-02
-4.51051682e-01 -3.21515143e-01 -6.24341130e-01 1.53521448e-01
4.15892124e-01 -5.44286966e-01 1.44849706e+00 -6.67376101e-01
-1.45929611e+00 5.81723094e-01 -1.26574025e-01 -5.84984541e-01
4.99359518e-01 -4.27511424e-01 -3.96584779e-01 4.68155369e-02
1.33986771e-01 7.23941743e-01 2.80579776e-01 -1.19031942e+00
-5.87715864e-01 7.72371739e-02 -8.83147344e-02 4.01902467e-01
-1.76859155e-01 1.70022517e-01 -3.92751098e-01 -7.34511614e-01
-4.20690933e-03 -7.87502825e-01 3.02869845e-02 -8.52803290e-01
-6.01025522e-01 -5.79414248e-01 2.30395317e-01 -8.44036102e-01
1.79296923e+00 -1.55084455e+00 9.91274789e-02 -1.03295399e-02
-2.69419830e-02 8.50376427e-01 -3.56456459e-01 5.76334357e-01
-1.20290015e-02 4.88485217e-01 -4.50505197e-01 -6.10861897e-01
2.24383157e-02 2.24779807e-02 1.79060742e-01 -1.06320426e-01
3.86945963e-01 1.01580608e+00 -1.13525712e+00 -5.68725586e-01
-4.44134623e-02 9.94485915e-02 -5.97164571e-01 1.44910747e-02
-2.05026284e-01 2.90968716e-01 -1.76517218e-01 5.03697157e-01
4.13944900e-01 -1.72715798e-01 1.98243722e-01 1.98695004e-01
-1.17134698e-01 6.30838871e-01 -8.37691784e-01 1.65334117e+00
-3.52515280e-01 5.56009650e-01 -5.03215671e-01 -1.01162875e+00
9.64897990e-01 4.17324156e-01 5.32931015e-02 -6.02196336e-01
2.57134527e-01 6.69791162e-01 3.46746981e-01 -1.53531000e-01
9.83395338e-01 1.07703835e-01 -2.74134398e-01 3.35966438e-01
2.92784840e-01 -1.82404995e-01 7.70638704e-01 4.93738234e-01
1.09845960e+00 -8.45753253e-02 4.45247501e-01 -3.29568803e-01
6.42803490e-01 6.99593127e-02 7.56774247e-01 8.16591322e-01
-2.01332092e-01 8.30905616e-01 4.90649283e-01 1.61297336e-01
-1.07470238e+00 -6.97209656e-01 6.22562729e-02 6.70399308e-01
-1.44270971e-01 -8.96799982e-01 -1.24123323e+00 -9.21272993e-01
-3.55514735e-01 1.15463746e+00 -4.01593864e-01 -2.06534103e-01
-6.59927607e-01 -8.67588341e-01 8.62536669e-01 5.14253676e-01
6.00249410e-01 -1.10039985e+00 -2.33780727e-01 3.42568487e-01
-8.56317937e-01 -9.13192391e-01 -6.71241403e-01 -7.18080550e-02
-8.71770561e-01 -8.83354783e-01 -7.34908640e-01 -6.71223938e-01
5.29444277e-01 1.63829178e-01 1.31337225e+00 4.32478756e-01
2.61648208e-01 9.24268886e-02 -8.19909096e-01 -4.66159195e-01
-7.05615938e-01 4.37067479e-01 -2.46912055e-02 -4.89637643e-01
4.45662409e-01 -3.04075509e-01 -4.12675560e-01 2.50092089e-01
-5.82786977e-01 2.82964617e-01 7.49651253e-01 1.02228689e+00
4.83196676e-01 -1.69970855e-01 7.94399738e-01 -1.08845425e+00
1.06684840e+00 -1.44981116e-01 -1.28159001e-01 6.45772398e-01
-1.05729306e+00 1.16146691e-01 6.23709857e-01 -1.88326046e-01
-1.01665401e+00 -6.61834896e-01 -3.66266370e-01 -6.11764903e-04
1.65697932e-01 9.53254521e-01 -1.38734668e-01 2.91478306e-01
6.95946991e-01 1.06548145e-01 -2.23336220e-01 -4.61308599e-01
1.42909542e-01 7.03775227e-01 6.24787450e-01 -5.59126914e-01
4.96032923e-01 -4.67421889e-01 -3.72774959e-01 -4.41655666e-01
-9.21389222e-01 -3.53817940e-01 -3.77420992e-01 -2.38259077e-01
6.67635024e-01 -7.61238635e-01 -3.92150491e-01 4.49347019e-01
-1.46521342e+00 -1.99174449e-01 -7.66981095e-02 5.36434591e-01
-4.53130901e-01 5.04232943e-01 -6.37174189e-01 -6.30975544e-01
-8.10353696e-01 -9.86080050e-01 9.09853041e-01 2.56202877e-01
-4.28701967e-01 -9.65034962e-01 1.40650496e-01 4.47041631e-01
4.75886554e-01 2.44924635e-01 8.70616555e-01 -7.59005964e-01
-1.85416058e-01 -2.76185066e-01 -1.26267880e-01 7.26760328e-01
-9.63360071e-02 -9.54228118e-02 -5.67664802e-01 -3.07650268e-01
-1.65476948e-01 -2.15760753e-01 9.73435640e-01 1.89447194e-01
1.00297010e+00 -3.71628731e-01 9.64885484e-03 2.96702772e-01
1.34314895e+00 1.36680365e-01 9.60300148e-01 3.15573007e-01
6.31908298e-01 4.27086264e-01 8.30923438e-01 2.15847507e-01
5.18164992e-01 7.58569002e-01 1.61618456e-01 1.11100525e-01
-2.97539473e-01 -4.14232254e-01 5.19525230e-01 1.48459899e+00
-4.06505734e-01 -7.23557234e-01 -9.40613985e-01 4.13608044e-01
-1.86877108e+00 -8.93667281e-01 -4.79416430e-01 2.34798741e+00
1.08249879e+00 2.81083405e-01 -1.98157251e-01 2.22824335e-01
8.55842948e-01 -1.67326167e-01 -1.27659112e-01 -7.52786338e-01
-3.84211242e-01 3.70061427e-01 2.49190733e-01 6.08655632e-01
-8.82178605e-01 1.10731125e+00 5.42794037e+00 1.13860953e+00
-7.92630732e-01 2.09925100e-01 5.77765822e-01 5.74734695e-02
-3.00891668e-01 4.70430031e-02 -9.19691145e-01 6.05244458e-01
1.17068827e+00 -4.77953672e-01 2.00332060e-01 5.28264642e-01
2.09973156e-01 -1.65676698e-01 -9.57414806e-01 9.24354196e-01
3.32983822e-01 -1.06210792e+00 1.34018824e-01 -1.38288215e-01
1.04139400e+00 -3.19601633e-02 -2.33446926e-01 7.59887516e-01
2.50592113e-01 -8.93447518e-01 7.20744133e-01 3.50480765e-01
8.34593713e-01 -6.53615057e-01 1.27311361e+00 2.99136788e-01
-7.96690226e-01 3.73154074e-01 -4.91139919e-01 6.98160231e-02
3.45407337e-01 7.91153431e-01 -8.52248073e-01 9.42080796e-01
4.91764307e-01 5.41600168e-01 -7.89874077e-01 1.06625855e+00
-6.49793804e-01 8.94564867e-01 -6.56013638e-02 -3.45026612e-01
2.61886805e-01 -1.78929150e-01 7.16734946e-01 1.42260408e+00
6.21986210e-01 6.23588413e-02 -1.18711516e-01 6.37977123e-01
-3.66124392e-01 3.61878067e-01 -2.50975072e-01 -9.78830177e-03
5.50771832e-01 1.12205648e+00 -5.36517024e-01 -5.83212614e-01
-1.68180093e-01 1.17692864e+00 5.60876310e-01 7.83005878e-02
-9.94646907e-01 -7.21133649e-01 2.35034153e-01 -1.82681903e-01
1.24766700e-01 -1.10149175e-01 -4.80236024e-01 -1.19939756e+00
1.94855094e-01 -1.11749756e+00 1.83846533e-01 -6.13379300e-01
-1.00435925e+00 8.49772513e-01 -4.89222119e-03 -1.33051181e+00
-5.07446349e-01 -3.84419709e-01 -7.76567757e-01 9.38053727e-01
-1.23720849e+00 -7.87422180e-01 -1.63836867e-01 8.05486664e-02
5.24721980e-01 -7.39643425e-02 9.13880885e-01 2.88155705e-01
-6.29082561e-01 1.13182735e+00 3.70044857e-01 2.59581357e-02
9.00164247e-01 -1.52680719e+00 5.87258399e-01 1.20176566e+00
1.73653215e-01 6.24122500e-01 7.50181556e-01 -6.91085696e-01
-6.76425874e-01 -9.69766855e-01 1.59444439e+00 -5.09826541e-01
2.93965727e-01 -1.04176566e-01 -9.13780808e-01 1.99125603e-01
1.99058264e-01 -6.47996008e-01 5.41742206e-01 2.79384851e-01
-8.78626332e-02 3.58077660e-02 -8.29371929e-01 5.03604174e-01
1.01087487e+00 -1.89177170e-01 -5.60529351e-01 3.63703728e-01
7.17501104e-01 -4.90116805e-01 -8.31391692e-01 6.64927363e-01
2.80083269e-01 -9.24855173e-01 4.29027408e-01 -3.56555313e-01
6.67397380e-01 -2.03140810e-01 -3.40645276e-02 -1.77872872e+00
-2.27780923e-01 -5.25502801e-01 -1.58599634e-02 1.47539663e+00
9.04919267e-01 -7.00224817e-01 5.15811622e-01 3.22203159e-01
-7.66323328e-01 -9.77316201e-01 -7.23324656e-01 -1.11139858e+00
3.44046563e-01 -2.83158153e-01 6.08822584e-01 8.50997806e-01
2.04241276e-01 6.11392677e-01 -3.46007794e-01 -2.17379719e-01
3.00543994e-01 -1.23327985e-01 6.58788741e-01 -9.88476634e-01
-3.20943832e-01 -6.24676526e-01 -1.40148029e-01 -9.22947824e-01
-8.87104943e-02 -1.12313914e+00 3.98662984e-01 -1.76993585e+00
2.76968062e-01 -3.04883242e-01 -2.11889639e-01 5.04068911e-01
-7.44410098e-01 2.48771563e-01 3.86691570e-01 2.15294957e-01
-7.53729582e-01 6.74268425e-01 1.15615356e+00 1.13094540e-03
-1.89999998e-01 -2.02118352e-01 -8.75448227e-01 3.51185620e-01
1.29783046e+00 -5.17288268e-01 -1.51364893e-01 -7.19976306e-01
1.91207051e-01 -9.27264392e-02 4.30465583e-03 -1.15115249e+00
6.66098371e-02 1.27826437e-01 2.79544555e-02 -5.89410543e-01
-1.68720067e-01 -4.04663868e-02 -3.41767222e-02 4.82748568e-01
-3.36697578e-01 4.07912046e-01 1.10760145e-02 1.46380082e-01
-5.51040828e-01 -6.88267350e-01 5.92870057e-01 -9.82381329e-02
-4.27773774e-01 -1.48029938e-01 -1.40120819e-01 3.28316271e-01
4.58761841e-01 -5.58534861e-02 -6.87280893e-01 -4.88563329e-01
-3.04939151e-01 4.50301260e-01 3.17101181e-01 3.80349904e-01
4.87331033e-01 -1.17360508e+00 -1.25570333e+00 -2.58806169e-01
2.84834266e-01 -1.48896039e-01 1.59677416e-01 9.75486934e-01
-5.99639356e-01 8.40472817e-01 1.44328684e-01 -4.05972898e-01
-1.45037973e+00 -3.66828926e-02 1.47619694e-01 -9.29856896e-01
-3.32793027e-01 8.26432526e-01 -2.27452695e-01 -5.18351138e-01
1.12361044e-01 -2.01188505e-01 -1.86638653e-01 -3.04770678e-01
4.56240922e-01 5.05947530e-01 5.54660559e-01 -4.98896539e-01
-2.99133152e-01 1.93177640e-01 -3.46974760e-01 -1.18247673e-01
9.95981574e-01 -6.20112233e-02 -5.90213537e-02 1.76648900e-01
1.00922751e+00 -1.47879764e-01 -6.19731009e-01 -8.76421332e-02
2.59575784e-01 -2.77043819e-01 -1.36932209e-01 -1.29831481e+00
-6.91440821e-01 8.94956529e-01 2.84652650e-01 -5.55268191e-02
1.12784338e+00 -2.20179439e-01 8.54152322e-01 4.97863054e-01
4.37944382e-01 -1.34724355e+00 6.14043847e-02 9.89779711e-01
1.05464816e+00 -1.26833999e+00 -4.27510105e-02 -4.81118083e-01
-9.21551824e-01 8.30892563e-01 7.59799719e-01 1.63410515e-01
7.55397677e-02 -2.30680093e-01 -3.17846537e-02 1.48425410e-02
-9.53707457e-01 -1.85890138e-01 3.46034825e-01 3.68651688e-01
9.02981341e-01 4.20374691e-01 -1.28752363e+00 6.70027792e-01
-6.88166559e-01 -3.95738333e-01 6.15883946e-01 6.28570199e-01
-4.12274808e-01 -1.43905568e+00 -7.86530226e-02 6.55992866e-01
-6.52244925e-01 -3.90968353e-01 -4.18899357e-01 7.85237074e-01
-2.67014243e-02 1.21132123e+00 -2.28057310e-01 -5.80465078e-01
4.40381616e-01 1.94618940e-01 5.31668663e-01 -7.65290737e-01
-8.58435452e-01 -1.03063941e-01 5.02895594e-01 -3.33078891e-01
-3.95128250e-01 -6.88049853e-01 -1.26774478e+00 -4.53975916e-01
-7.22913265e-01 6.40449703e-01 6.91359580e-01 7.07410514e-01
3.62203687e-01 5.09159088e-01 5.60809970e-01 -5.09602368e-01
-7.22974360e-01 -1.67063415e+00 -1.98404387e-01 5.72072804e-01
-2.59341061e-01 -4.29786623e-01 -3.12377721e-01 -1.90954268e-01]
|
[11.180749893188477, 10.479741096496582]
|
622f58fe-c660-43ae-a699-683663cb4efa
|
pancreas-segmentation-in-ct-and-mri-images
|
1803.11303
| null |
http://arxiv.org/abs/1803.11303v1
|
http://arxiv.org/pdf/1803.11303v1.pdf
|
Pancreas Segmentation in CT and MRI Images via Domain Specific Network Designing and Recurrent Neural Contextual Learning
|
Automatic pancreas segmentation in radiology images, eg., computed tomography
(CT) and magnetic resonance imaging (MRI), is frequently required by
computer-aided screening, diagnosis, and quantitative assessment. Yet pancreas
is a challenging abdominal organ to segment due to the high inter-patient
anatomical variability in both shape and volume metrics. Recently,
convolutional neural networks (CNNs) have demonstrated promising performance on
accurate segmentation of pancreas. However, the CNN-based method often suffers
from segmentation discontinuity for reasons such as noisy image quality and
blurry pancreatic boundary. From this point, we propose to introduce recurrent
neural networks (RNNs) to address the problem of spatial non-smoothness of
inter-slice pancreas segmentation across adjacent image slices. To inference
initial segmentation, we first train a 2D CNN sub-network, where we modify its
network architecture with deep-supervision and multi-scale feature map
aggregation so that it can be trained from scratch with small-sized training
data and presents superior performance than transferred models. Thereafter, the
successive CNN outputs are processed by another RNN sub-network, which refines
the consistency of segmented shapes. More specifically, the RNN sub-network
consists convolutional long short-term memory (CLSTM) units in both top-down
and bottom-up directions, which regularizes the segmentation of an image by
integrating predictions of its neighboring slices. We train the stacked CNN-RNN
model end-to-end and perform quantitative evaluations on both CT and MRI
images.
|
['Lin Yang', 'Le Lu', 'Jinzheng Cai', 'Fuyong Xing']
|
2018-03-30
| null | null | null | null |
['pancreas-segmentation']
|
['medical']
|
[ 2.04574406e-01 3.16744596e-01 -1.10271700e-01 -5.59414208e-01
-8.64664733e-01 -2.94780284e-01 -9.27660242e-03 2.28010803e-01
-4.07240719e-01 3.14828485e-01 3.43933403e-01 -4.11745727e-01
9.29026380e-02 -6.70010269e-01 -9.36578810e-01 -7.31805027e-01
-3.75287235e-01 4.06768888e-01 2.55939722e-01 2.63509542e-01
-3.26837540e-01 3.33697438e-01 -5.80457985e-01 3.21619838e-01
1.08300388e+00 1.00822031e+00 3.09577644e-01 6.07369184e-01
-2.15422451e-01 8.54982078e-01 -1.06971815e-01 -1.27115190e-01
2.02598333e-01 -7.54578292e-01 -7.42872119e-01 2.87248105e-01
-2.16210745e-02 -3.22148621e-01 -3.74417931e-01 1.33766687e+00
4.20428306e-01 -1.40894800e-01 2.59621739e-01 -5.03840685e-01
-7.81804264e-01 9.69085813e-01 -3.85216892e-01 2.45150506e-01
-4.61069532e-02 2.81953394e-01 4.02377337e-01 -6.07932866e-01
4.73222315e-01 8.67482722e-01 9.32264566e-01 4.77908552e-01
-1.22412360e+00 -4.21110928e-01 2.82375723e-01 -3.94409001e-01
-8.47279549e-01 1.30472884e-01 5.68862259e-01 -3.10116589e-01
5.15156567e-01 -5.66177033e-02 9.71399784e-01 9.21846032e-01
3.51530224e-01 8.61033320e-01 8.03503871e-01 -9.13027003e-02
-1.08267248e-01 -3.41733724e-01 4.81271297e-02 9.18649375e-01
2.93003887e-01 1.30993828e-01 3.48542869e-01 1.93299696e-01
1.39817512e+00 4.08854336e-01 -5.55522263e-01 -5.17628193e-01
-1.58018994e+00 7.72945404e-01 1.20749164e+00 4.98095572e-01
-8.31836820e-01 3.80790383e-02 4.78083193e-01 6.84467033e-02
2.22467154e-01 3.17612708e-01 -4.15061355e-01 2.95194387e-01
-8.65183175e-01 -3.06729525e-01 7.92682767e-01 6.75433755e-01
2.76890900e-02 -2.78748572e-02 -1.75210714e-01 7.60965884e-01
5.28088272e-01 1.30170807e-01 9.82224405e-01 -5.09919167e-01
3.31154317e-01 5.77588141e-01 -3.87925118e-01 -4.86856639e-01
-8.43779087e-01 -5.61531365e-01 -1.39143777e+00 3.29432897e-02
4.95773226e-01 -2.78774738e-01 -1.54964280e+00 1.44434679e+00
3.20193708e-01 2.11278841e-01 8.90774950e-02 1.40947461e+00
1.10961330e+00 4.80821669e-01 4.97157872e-01 -1.80068612e-01
1.38960314e+00 -1.37123334e+00 -5.66194117e-01 -5.11955761e-04
6.82588577e-01 -4.86003727e-01 6.99450910e-01 1.54514611e-01
-1.34935498e+00 -2.83354998e-01 -8.41466606e-01 -7.70574734e-02
-6.12524077e-02 1.28554568e-01 3.44555169e-01 1.33880824e-01
-9.75610495e-01 6.81280792e-01 -1.52587032e+00 -2.63421774e-01
5.52309394e-01 3.36846560e-01 -2.55490839e-01 -2.90594492e-02
-1.03340423e+00 8.91315937e-01 3.03026050e-01 5.13487875e-01
-8.05036128e-01 -7.90375471e-01 -1.24439049e+00 1.43416643e-01
2.39540458e-01 -9.68098164e-01 1.42022002e+00 -1.09982896e+00
-1.79955113e+00 8.71729374e-01 2.96072334e-01 -6.66979373e-01
6.84827685e-01 1.13469452e-01 -2.07760185e-01 2.31309652e-01
-7.67152011e-02 7.38023579e-01 5.42863607e-01 -1.21051228e+00
-2.88894087e-01 -4.28568155e-01 -3.55478197e-01 1.23609409e-01
2.86497802e-01 -3.17078903e-02 -4.49861318e-01 -9.91180122e-01
7.95235038e-01 -9.65447009e-01 -8.09450686e-01 1.22066334e-01
-6.28222525e-01 1.99623615e-01 2.38828510e-01 -9.96544242e-01
7.41685271e-01 -2.09182668e+00 1.97299272e-01 2.12627500e-01
1.63835317e-01 1.17650710e-01 -1.57132953e-01 -5.15460670e-01
-2.38038778e-01 2.49847095e-03 -4.54646260e-01 -8.83885548e-02
-3.71193200e-01 3.16747814e-01 2.46465251e-01 6.19199753e-01
2.22814307e-01 1.33565414e+00 -9.03254628e-01 -5.36883593e-01
2.59078890e-01 5.14362037e-01 -3.94091874e-01 4.74289298e-01
-2.55342424e-01 1.00790191e+00 -3.58552247e-01 7.88935721e-01
5.68683982e-01 -7.14938819e-01 3.19784969e-01 -3.46274644e-01
-1.36661485e-01 2.25809172e-01 -6.47437334e-01 2.09494734e+00
-4.19752151e-01 3.39022721e-03 3.17534178e-01 -9.22494471e-01
5.98127306e-01 5.20045757e-01 7.50925899e-01 -8.01738977e-01
3.33661735e-01 4.47984576e-01 2.96412557e-01 -7.43516326e-01
-5.97213283e-02 -4.01639044e-01 2.32703355e-03 2.75649518e-01
2.21039206e-02 -9.10191890e-03 -8.37838836e-03 -4.45443928e-01
7.37518847e-01 2.04404164e-02 2.09823459e-01 -2.28102893e-01
4.31979209e-01 3.57219800e-02 5.73252499e-01 3.40247810e-01
-2.46901855e-01 9.46222663e-01 4.31523293e-01 -6.44900441e-01
-1.00504041e+00 -1.08520448e+00 -3.06135595e-01 8.93978834e-01
3.28934640e-01 4.76115793e-01 -6.38524294e-01 -7.77934372e-01
3.35470736e-02 2.76665211e-01 -7.39974201e-01 2.74575558e-02
-1.03532720e+00 -7.92933345e-01 3.92553717e-01 7.54953682e-01
3.80604506e-01 -1.32644093e+00 -7.32873321e-01 5.32692432e-01
-6.94982754e-03 -1.01480222e+00 -7.41179109e-01 4.90378082e-01
-1.47241628e+00 -1.16098654e+00 -1.36158895e+00 -1.37147903e+00
1.05375600e+00 -5.11551611e-02 1.15828764e+00 4.16495837e-02
-2.06223980e-01 -1.48059681e-01 -8.80851522e-02 4.74177562e-02
-5.27373910e-01 1.63246378e-01 -5.18305302e-01 -4.04806018e-01
-1.19110778e-01 -4.53599364e-01 -9.09760654e-01 2.34851703e-01
-9.20194387e-01 2.41473883e-01 9.95846510e-01 1.07401156e+00
1.01298702e+00 -5.75390220e-01 3.98652464e-01 -9.14488792e-01
5.17648757e-01 -5.31580448e-01 -6.53099120e-01 3.62529784e-01
-3.56514663e-01 1.14981167e-01 6.83169663e-01 -3.84756595e-01
-7.65674412e-01 1.36208043e-01 -3.97411674e-01 -6.29436255e-01
-1.25955805e-01 8.69299889e-01 2.76286930e-01 -1.25197396e-01
2.28738189e-01 2.76066720e-01 3.05063605e-01 -4.11752462e-01
4.33718562e-02 1.21064618e-01 8.26194167e-01 -1.12634137e-01
2.12349460e-01 1.79788277e-01 -2.06424102e-01 -1.15126662e-01
-8.26867938e-01 -2.30523258e-01 -6.42175436e-01 -2.65444554e-02
1.18121612e+00 -8.41368854e-01 -5.64703703e-01 3.34074587e-01
-1.05761468e+00 -6.46086991e-01 -3.37892622e-01 9.43378091e-01
-2.63750046e-01 1.49558038e-01 -1.26330006e+00 -1.05464458e-01
-7.61071384e-01 -1.54153931e+00 8.61989439e-01 4.84987199e-01
1.67438790e-01 -1.16054440e+00 -1.16902431e-02 -1.00122519e-01
4.60516840e-01 5.70345283e-01 1.01236928e+00 -8.45446885e-01
-5.05560040e-01 -1.40249252e-01 -5.63437223e-01 3.27755511e-01
1.41426161e-01 -3.45152557e-01 -4.21096891e-01 -2.50980049e-01
-1.22690620e-02 -1.07464530e-01 9.32663023e-01 1.18493402e+00
1.42474651e+00 -1.95069239e-01 -1.62489250e-01 1.11867225e+00
1.31127644e+00 1.58567742e-01 2.78799176e-01 2.68143624e-01
8.40832233e-01 2.34989151e-01 -4.99502989e-03 1.33107647e-01
6.03032768e-01 3.38386782e-02 6.05934083e-01 -6.12155795e-01
-2.15184018e-01 -1.15086742e-01 3.96226645e-02 1.14770448e+00
1.14632644e-01 1.59809738e-01 -8.67259443e-01 4.73318547e-01
-1.57608354e+00 -4.72883284e-01 -7.38542974e-02 2.15023232e+00
8.59192848e-01 8.44090879e-02 6.85369596e-02 -5.18307447e-01
7.47120261e-01 -3.59576903e-02 -7.83751190e-01 -2.62487948e-01
1.88076437e-01 4.34195176e-02 6.50507569e-01 3.75574380e-01
-1.29565239e+00 4.60917205e-01 5.51415586e+00 5.07078171e-02
-1.58400261e+00 5.15688770e-02 9.13808823e-01 1.26629084e-01
-1.86364144e-01 -4.63870734e-01 -9.32445303e-02 5.25715411e-01
6.04839265e-01 2.31523201e-01 3.02680284e-01 7.09379971e-01
9.90430340e-02 4.13845256e-02 -1.15907693e+00 7.27549732e-01
3.62432972e-02 -1.20476151e+00 -6.11885861e-02 -2.76283562e-01
6.89151645e-01 5.59334517e-01 2.40705218e-02 4.98959929e-01
3.44006211e-01 -1.24375844e+00 5.10554969e-01 5.24154842e-01
7.89687216e-01 -4.64280456e-01 1.10324740e+00 2.40406305e-01
-1.01028287e+00 5.65730594e-02 -1.04812957e-01 6.49615347e-01
3.11346859e-01 4.90854234e-01 -8.07751119e-01 4.60416377e-01
4.96430159e-01 6.41279101e-01 -2.83613205e-01 1.35800898e+00
-1.65973961e-01 2.63574272e-01 -2.28956953e-01 1.50401443e-01
6.43113375e-01 -4.58724827e-01 3.46204907e-01 1.29463863e+00
4.49287176e-01 2.13173732e-01 6.09591603e-01 1.02543426e+00
-4.64923114e-01 -2.50106752e-02 -1.21111296e-01 3.10952514e-01
-1.98422462e-01 1.37873018e+00 -9.92608905e-01 -5.47969162e-01
-5.62600076e-01 8.50911796e-01 1.59405887e-01 3.56400013e-01
-8.71764719e-01 2.23994663e-04 1.35947049e-01 -6.80294931e-02
3.41499150e-01 1.85029328e-01 -5.30922234e-01 -1.28324318e+00
3.36095504e-02 -5.87915063e-01 6.67784393e-01 -4.87120152e-01
-1.42261708e+00 9.08535302e-01 -5.66152871e-01 -1.16666019e+00
-2.16921851e-01 -4.77929980e-01 -6.73000038e-01 8.31471741e-01
-1.86514843e+00 -1.22240639e+00 -4.98757362e-01 3.96065772e-01
5.01856565e-01 4.42981958e-01 7.26394832e-01 3.63319159e-01
-5.91359556e-01 4.10183191e-01 -6.25102222e-03 7.26035714e-01
5.06630242e-01 -1.37213397e+00 2.10988924e-01 6.83328748e-01
-3.11245978e-01 3.14481884e-01 2.65762359e-01 -7.21204460e-01
-1.18038464e+00 -1.58883905e+00 6.82908595e-01 1.54743314e-01
5.79255581e-01 1.18213437e-01 -1.08001292e+00 9.16235924e-01
9.46073979e-02 8.96544158e-01 5.48994780e-01 -3.85126472e-01
1.47121623e-02 2.43994310e-01 -1.24533772e+00 3.85742933e-01
6.39075994e-01 -8.02815333e-02 -6.55351043e-01 2.51663387e-01
8.21076632e-01 -1.19185698e+00 -1.35864115e+00 7.58887053e-01
4.65552241e-01 -7.52733886e-01 1.00789988e+00 -4.89482522e-01
6.41460001e-01 -1.82042047e-02 3.31853569e-01 -1.45972419e+00
-3.51039797e-01 -3.09894472e-01 1.17243350e-01 4.52879727e-01
5.81460118e-01 -5.62914908e-01 7.52773523e-01 6.77382350e-01
-6.59286857e-01 -1.14047551e+00 -6.92960083e-01 -3.12875062e-01
4.21373039e-01 -6.24615848e-02 4.56224859e-01 9.50707197e-01
8.28829259e-02 -1.31091937e-01 1.93682835e-01 2.83723265e-01
5.47150671e-01 5.18364072e-01 1.24204308e-01 -9.79461372e-01
-4.76551317e-02 -7.81766593e-01 -1.21828198e-01 -1.15533674e+00
-2.45369777e-01 -1.20184946e+00 2.34290227e-01 -1.74124777e+00
3.61995935e-01 -3.14738452e-01 -6.29923582e-01 4.51085150e-01
-3.15408260e-01 1.72422931e-01 1.84363812e-01 1.68526873e-01
-7.33171999e-01 3.49937826e-01 2.09444451e+00 -1.47749722e-01
-4.97287750e-01 2.20861152e-01 -3.37392777e-01 8.44350994e-01
5.93076408e-01 -4.99479353e-01 2.41466582e-01 -5.47159851e-01
-2.57897824e-01 7.36864626e-01 5.79133332e-01 -7.33517289e-01
4.54906464e-01 3.21244299e-01 8.05750608e-01 -5.69250524e-01
-3.71739179e-01 -9.05152380e-01 -1.30436048e-01 9.54818666e-01
-6.87065244e-01 2.56674029e-02 1.69658158e-02 2.57920355e-01
-5.14746130e-01 -8.91793519e-02 9.92849410e-01 -6.37118697e-01
-1.55969620e-01 8.38362813e-01 -2.72518992e-01 4.08736058e-02
7.78393090e-01 9.53327771e-03 2.44145915e-01 6.96036443e-02
-1.07580054e+00 4.88861084e-01 2.31250599e-01 6.68887794e-02
6.72145724e-01 -1.31951654e+00 -7.65350759e-01 3.45789999e-01
-1.91893280e-01 8.25827301e-01 3.91374767e-01 1.54993463e+00
-8.35735619e-01 2.93687493e-01 -1.64196268e-01 -9.39930201e-01
-7.30607986e-01 5.83479464e-01 6.83849037e-01 -6.94833696e-01
-1.15278697e+00 9.04202342e-01 1.97392642e-01 -7.15574324e-01
4.85713929e-01 -1.32977402e+00 -1.84843302e-01 -3.42400640e-01
2.39376202e-01 -2.32541561e-01 7.02986494e-02 -3.92288119e-01
-1.67486221e-01 4.76567030e-01 -9.89788696e-02 3.70819747e-01
1.43681169e+00 -1.28203742e-02 -1.28064603e-01 1.34405836e-01
1.22644889e+00 -5.78276277e-01 -1.43910289e+00 -3.33160669e-01
1.60945535e-01 3.93854566e-02 1.75585523e-01 -9.77015436e-01
-1.53057468e+00 7.74793267e-01 5.56524217e-01 1.99601725e-01
1.08656943e+00 -5.16553521e-02 1.33553159e+00 -5.85959926e-02
-3.30691673e-02 -6.62265241e-01 -1.44812420e-01 4.19125408e-01
7.53303111e-01 -1.52030802e+00 -2.71932364e-01 -1.16652749e-01
-7.20371723e-01 1.51968884e+00 5.11276841e-01 -4.02552396e-01
6.90845132e-01 4.99446779e-01 4.42762941e-01 -2.19046876e-01
-3.64883780e-01 1.86462494e-04 3.39353919e-01 2.69181319e-02
6.26418114e-01 2.53833026e-01 7.34793022e-02 9.66276526e-01
1.03384912e-01 2.85870641e-01 1.35585219e-01 7.72627831e-01
-2.23557696e-01 -5.59416711e-01 -3.19998026e-01 6.14334881e-01
-6.38664603e-01 -6.81359768e-02 3.14135760e-01 8.27260256e-01
-4.62840833e-02 2.20264733e-01 1.96380079e-01 1.84776783e-01
4.00228113e-01 -9.85672176e-02 4.98770922e-01 -5.05030155e-01
-1.10915029e+00 5.04691839e-01 -4.05710131e-01 -4.93731469e-01
-3.12859535e-01 -6.42499506e-01 -1.78070498e+00 1.72242999e-01
-2.43972883e-01 -5.70167182e-03 6.75459683e-01 7.88801074e-01
5.48237041e-02 1.11536837e+00 5.58985114e-01 -9.76971149e-01
-7.88547277e-01 -1.05571032e+00 -3.65852654e-01 5.34500778e-01
5.40494502e-01 -1.01197533e-01 -4.29373495e-02 1.74204156e-01]
|
[14.535605430603027, -2.6708078384399414]
|
0a61ab97-868b-4cc2-909e-704a7418c1dc
|
estimating-network-edge-probabilities-by
|
1509.08588
| null |
http://arxiv.org/abs/1509.08588v3
|
http://arxiv.org/pdf/1509.08588v3.pdf
|
Estimating network edge probabilities by neighborhood smoothing
|
The estimation of probabilities of network edges from the observed adjacency
matrix has important applications to predicting missing links and network
denoising. It has usually been addressed by estimating the graphon, a function
that determines the matrix of edge probabilities, but this is ill-defined
without strong assumptions on the network structure. Here we propose a novel
computationally efficient method, based on neighborhood smoothing to estimate
the expectation of the adjacency matrix directly, without making the structural
assumptions that graphon estimation requires. The neighborhood smoothing method
requires little tuning, has a competitive mean-squared error rate, and
outperforms many benchmark methods on link prediction in simulated and real
networks.
|
['Yuan Zhang', 'Ji Zhu', 'Elizaveta Levina']
|
2015-09-29
| null | null | null | null |
['graphon-estimation']
|
['graphs']
|
[ 2.68470734e-01 4.85107958e-01 -1.05682157e-01 -1.95757776e-01
-1.79599762e-01 -3.89261663e-01 2.25987867e-01 1.95728213e-01
-2.20214520e-02 7.19282269e-01 -6.19677268e-02 -6.13321066e-01
-5.45654416e-01 -9.51970696e-01 -7.06618428e-01 -6.81786239e-01
-8.21824074e-01 5.04563451e-01 4.76942807e-01 -9.38084498e-02
3.76787083e-03 4.08047378e-01 -7.67459095e-01 -3.45640421e-01
9.18845475e-01 8.50223660e-01 6.83677718e-02 7.70902932e-01
-2.34059021e-02 6.95168197e-01 -2.15053216e-01 -6.42860115e-01
2.10879147e-01 -5.70746720e-01 -6.80360258e-01 2.31907040e-01
2.89054930e-01 7.09720254e-02 -8.62632811e-01 1.34916699e+00
2.00084656e-01 -8.11853781e-02 8.82120490e-01 -1.34160161e+00
-2.38734037e-01 9.64213490e-01 -5.76808214e-01 3.20440680e-01
4.01898921e-01 -3.33509535e-01 1.26921749e+00 -4.91763175e-01
6.68081045e-01 1.22060311e+00 1.24512863e+00 -8.83797333e-02
-1.89543962e+00 -5.47785640e-01 3.51917185e-02 2.01586276e-01
-1.80588353e+00 -2.13041380e-01 7.21097231e-01 -4.50753361e-01
5.40707469e-01 3.68359566e-01 5.06673813e-01 7.52223611e-01
2.03390628e-01 3.19487095e-01 7.88254738e-01 -3.18232417e-01
-6.05763718e-02 1.57458350e-01 7.79297054e-02 9.81819987e-01
3.37344140e-01 -2.66508032e-02 -3.58651243e-02 -5.66653430e-01
8.83781374e-01 -1.88672692e-01 -4.65876341e-01 -5.56120396e-01
-1.00028753e+00 8.02786231e-01 5.60259044e-01 -3.23753878e-02
-3.30868661e-01 2.92183608e-01 1.85730651e-01 6.46434605e-01
5.01545131e-01 -1.43796271e-02 -3.70147437e-01 2.54852325e-01
-7.88868427e-01 -3.57891858e-01 1.28150678e+00 9.51557994e-01
8.87970448e-01 -1.57694489e-01 1.79924384e-01 6.55983090e-01
4.04285699e-01 4.69337165e-01 -4.51655746e-01 -8.19483042e-01
4.71852481e-01 5.85836053e-01 -1.93196684e-01 -1.53727901e+00
-5.61106622e-01 -4.92497265e-01 -1.33218896e+00 -8.34910199e-02
6.30141497e-01 -4.81222421e-01 -6.06930435e-01 1.61397457e+00
3.29286218e-01 6.45300925e-01 -4.01801705e-01 4.92433101e-01
6.08729839e-01 6.10866964e-01 -2.15463832e-01 -4.18791145e-01
6.90398157e-01 -5.60537457e-01 -6.93688571e-01 -2.01377928e-01
5.37884653e-01 -6.94248319e-01 4.57721680e-01 6.99055195e-02
-8.86141062e-01 -3.73269431e-02 -6.92500472e-01 4.03905064e-01
-1.27923802e-01 -2.72055771e-02 8.28736067e-01 7.43262410e-01
-1.18783939e+00 1.10626101e+00 -7.45445967e-01 -5.58272898e-01
9.67173800e-02 6.06876493e-01 -5.35467565e-01 4.22107838e-02
-1.02683616e+00 6.40562952e-01 4.70437139e-01 4.08950090e-01
-3.75983834e-01 -5.68489373e-01 -8.52236927e-01 4.25976187e-01
6.84667468e-01 -5.23590446e-01 3.84874761e-01 -8.38473976e-01
-1.36792862e+00 3.41864944e-01 -1.86100513e-01 -4.76711363e-01
4.77533102e-01 2.78577566e-01 -5.29681385e-01 1.60051193e-02
-1.71024695e-01 2.76552260e-01 9.15612280e-01 -8.88034761e-01
-1.22509345e-01 -3.58708687e-02 -1.78407833e-01 -2.73752689e-01
-1.87481016e-01 -3.46165538e-01 -5.61991513e-01 -7.75009632e-01
5.46803653e-01 -1.03125262e+00 -5.27551472e-01 1.51703492e-01
-8.91286373e-01 -5.24099879e-02 4.87923741e-01 -7.39702344e-01
1.35510910e+00 -1.93727398e+00 9.39955190e-02 1.17320359e+00
4.15993422e-01 -6.61386130e-03 -3.58286619e-01 7.75193691e-01
-2.81552196e-01 4.42571379e-02 -3.68777931e-01 1.48613349e-01
-2.06761986e-01 1.41255006e-01 -1.10812917e-01 6.39703274e-01
-3.92156793e-03 5.75591326e-01 -9.09718096e-01 -5.88112056e-01
9.15922895e-02 6.01045787e-01 -6.76193357e-01 4.08236235e-02
2.75178730e-01 1.48586676e-01 -2.26995200e-01 1.24287382e-01
7.46188760e-01 -8.38643312e-01 8.13537359e-01 -3.63946617e-01
3.99783581e-01 1.18667617e-01 -1.73605466e+00 1.06646359e+00
-2.06303805e-01 7.60720193e-01 4.42050606e-01 -1.25781870e+00
8.18375111e-01 2.39776000e-01 5.37313521e-01 2.22227708e-01
2.33726248e-01 -1.16051100e-01 2.53080755e-01 -2.24528134e-01
-1.91332981e-01 1.37062013e-01 2.40135208e-01 4.21263486e-01
2.13033371e-02 2.31984407e-01 5.24035633e-01 8.12942266e-01
1.67601895e+00 -5.74920237e-01 2.63324887e-01 -4.50326592e-01
5.19223750e-01 -3.03668767e-01 6.15629256e-01 7.56994247e-01
4.35641035e-02 1.71241909e-01 1.00718975e+00 -3.70327532e-02
-8.82314682e-01 -1.38131070e+00 -4.66878973e-02 5.27786851e-01
1.24071442e-01 -6.56826138e-01 -5.76524377e-01 -7.44089842e-01
2.65192121e-01 1.75510287e-01 -5.50364256e-01 -1.60496667e-01
-2.25729764e-01 -9.01988447e-01 9.83585939e-02 2.56436288e-01
8.44644979e-02 -5.04486978e-01 8.63493264e-01 2.37277210e-01
-1.57184526e-01 -1.29757059e+00 -8.66151094e-01 -1.91323292e-02
-1.00403678e+00 -1.39213967e+00 -2.93251097e-01 -8.74040306e-01
1.02080250e+00 1.66510403e-01 1.20337141e+00 5.17844319e-01
-2.53026307e-01 2.53589094e-01 1.80570246e-03 3.34885627e-01
-3.69879454e-01 9.62212235e-02 -9.35203023e-03 1.98245898e-01
4.18885350e-02 -1.12620878e+00 -3.39578092e-01 5.00994563e-01
-3.41462225e-01 -1.65194571e-01 8.48785698e-01 8.67283821e-01
5.13950169e-01 4.24277514e-01 4.92367715e-01 -1.42837739e+00
6.45039856e-01 -6.10698223e-01 -8.23502541e-01 2.37005293e-01
-7.22093821e-01 1.98362932e-01 4.53937948e-01 -4.49952096e-01
-6.72020018e-01 3.51215452e-01 6.02479875e-02 -9.14260894e-02
2.18563855e-01 6.51503444e-01 -1.71206981e-01 -4.69496161e-01
2.78107077e-01 -9.47726145e-02 1.19749062e-01 -4.79016662e-01
4.25399810e-01 1.34588584e-01 4.01965380e-01 2.18182616e-02
1.07528758e+00 4.48237300e-01 5.96489549e-01 -1.09761369e+00
-6.86782658e-01 -5.42770267e-01 -7.86210895e-01 -9.41009372e-02
3.35936517e-01 -7.19164252e-01 -1.08195579e+00 -2.06469912e-02
-1.09942973e+00 -3.03203613e-01 8.13473836e-02 5.61734498e-01
-2.46480912e-01 7.86333740e-01 -9.14844990e-01 -6.42918110e-01
-1.34016722e-01 -4.70108002e-01 3.55871588e-01 -2.09083959e-01
-1.82465851e-01 -1.57138491e+00 7.30397999e-02 -2.54046470e-01
1.37391359e-01 1.69125825e-01 8.89780760e-01 -2.78600663e-01
-6.21669531e-01 -5.13029397e-01 -6.52883530e-01 2.73837209e-01
2.64015645e-01 1.49838313e-01 -2.93120831e-01 -3.25332195e-01
-5.48718154e-01 4.62268323e-01 8.48149180e-01 7.29805827e-01
1.13648093e+00 -4.52039927e-01 -7.75509298e-01 5.06842852e-01
1.36938035e+00 -4.68985528e-01 5.54446638e-01 -3.96903455e-01
7.56852925e-01 4.99568790e-01 1.48982346e-01 4.78312731e-01
1.41874045e-01 4.39746141e-01 4.10139471e-01 -1.04330920e-01
-1.90010801e-01 -2.80035406e-01 3.42619240e-01 1.14511037e+00
1.23287432e-01 -3.66272390e-01 -7.89471209e-01 3.91554862e-01
-1.94831955e+00 -1.01962972e+00 -7.21953809e-01 2.10004139e+00
6.68299258e-01 4.73109365e-01 1.01665378e-01 1.27809867e-01
1.27849412e+00 -1.18797943e-02 -3.40160012e-01 6.18492439e-02
-3.26840580e-02 -1.57102555e-01 9.20901179e-01 1.09818649e+00
-1.05817389e+00 6.36873603e-01 8.28647232e+00 8.77384603e-01
-3.73881102e-01 -1.20477438e-01 2.70076454e-01 3.67303818e-01
-2.42667675e-01 3.86283398e-01 -6.78770661e-01 4.81023133e-01
8.97156775e-01 -1.99816301e-01 6.27317548e-01 6.42911434e-01
3.14572930e-01 -8.18287805e-02 -1.04632711e+00 8.32653582e-01
-3.74761462e-01 -1.20520258e+00 -2.98517019e-01 3.30605417e-01
8.75467062e-01 7.20165372e-02 -2.56804377e-01 -2.05435798e-01
6.73597872e-01 -7.87631392e-01 -1.49960920e-01 8.46941411e-01
4.52012569e-01 -6.42697632e-01 8.15691888e-01 2.31116742e-01
-1.48491848e+00 8.51481035e-02 -5.54534912e-01 -2.12609321e-01
4.46004480e-01 1.52587140e+00 -1.05443490e+00 2.27915302e-01
4.11905348e-01 8.89600694e-01 -4.10927981e-01 1.33574045e+00
-2.60312349e-01 9.37122047e-01 -7.44254649e-01 1.68653589e-03
-1.60557032e-01 -6.59184873e-01 7.94542968e-01 1.00502551e+00
2.78730661e-01 -2.12732255e-02 3.84818941e-01 6.78103745e-01
-2.68701971e-01 7.08991364e-02 -5.93594372e-01 -5.99359274e-02
6.38828754e-01 1.40340698e+00 -1.01038527e+00 -1.80775821e-01
-4.64558423e-01 8.40376794e-01 4.23678279e-01 4.24782664e-01
-6.71774864e-01 -6.44727528e-01 6.01207793e-01 4.94371146e-01
4.95216727e-01 -3.30612779e-01 9.08384994e-02 -8.44245136e-01
-1.11432582e-01 -3.82232130e-01 4.27508116e-01 -2.89094478e-01
-1.70606792e+00 5.32533765e-01 -2.56324708e-01 -1.01530790e+00
-3.40633169e-02 -6.50606811e-01 -8.89756739e-01 6.72338545e-01
-1.01567650e+00 -6.04235113e-01 -1.65793613e-01 5.62318206e-01
-2.58074611e-01 4.38376553e-02 5.41630268e-01 5.19941747e-01
-9.18653905e-01 4.64979768e-01 5.47190785e-01 3.46635908e-01
5.61503410e-01 -1.40643477e+00 3.81043017e-01 7.67470539e-01
1.97940424e-01 3.49762380e-01 1.04833424e+00 -9.16231990e-01
-1.22139287e+00 -1.14392877e+00 9.74049032e-01 -2.81551778e-01
1.16703475e+00 -3.18950534e-01 -8.77194822e-01 8.93489122e-01
-5.38388118e-02 4.46016163e-01 5.91645181e-01 5.70830107e-01
-2.05336753e-02 -2.16748163e-01 -9.92061257e-01 4.41664547e-01
1.30443263e+00 -4.35398519e-01 1.36260495e-01 6.92465782e-01
3.73459548e-01 2.72475630e-02 -1.22200274e+00 3.34431946e-01
2.60460466e-01 -5.73781371e-01 1.11792374e+00 -5.20802379e-01
-2.77537584e-01 -2.38464877e-01 2.23538592e-01 -1.37007678e+00
-6.88459337e-01 -8.60452890e-01 -2.78123885e-01 1.31312418e+00
7.56883919e-01 -8.16275954e-01 1.21789742e+00 3.22519094e-01
4.67490435e-01 -5.67680538e-01 -7.90984631e-01 -1.05962944e+00
-6.24800265e-01 -2.90047228e-01 2.70104498e-01 9.38046873e-01
1.91120431e-01 6.91482008e-01 -4.80154365e-01 5.01430452e-01
1.04544389e+00 -1.45225868e-01 7.28940308e-01 -1.75857544e+00
-3.31252128e-01 -3.53295892e-01 -7.78624177e-01 -1.22127175e+00
4.40696925e-01 -1.04643226e+00 -1.11148067e-01 -1.59428763e+00
1.35383606e-01 -6.25496566e-01 -1.14937760e-01 1.16153397e-01
-1.75783724e-01 2.95307428e-01 -2.96670288e-01 -4.42869961e-02
-6.75117552e-01 5.02032280e-01 1.00940776e+00 -4.84851860e-02
-1.14381202e-01 5.02532780e-01 -2.95910895e-01 1.02750385e+00
6.63508773e-01 -8.02980781e-01 -4.06279653e-01 4.15767962e-03
5.67122936e-01 2.17008397e-01 3.10387075e-01 -8.81573021e-01
3.40851277e-01 6.49150982e-02 1.79027826e-01 -6.49789572e-01
3.02550584e-01 -1.03318810e+00 4.60839480e-01 6.01336122e-01
-1.45269170e-01 1.15296341e-01 -2.36277983e-01 1.18175220e+00
2.44779105e-04 -3.37131947e-01 5.87583959e-01 1.73796415e-01
-3.73163104e-01 4.63909537e-01 -3.23220015e-01 2.64946707e-02
8.48059297e-01 4.71658185e-02 2.68302914e-02 -8.24893415e-01
-1.30144382e+00 3.27463388e-01 1.87846631e-01 -1.76953703e-01
4.72960621e-01 -1.50143123e+00 -7.67727673e-01 -9.75845978e-02
-3.97386760e-01 -5.65277457e-01 1.05968425e-02 1.25281775e+00
-4.65658814e-01 1.67787507e-01 3.37437332e-01 -4.89159465e-01
-1.42395365e+00 6.85661972e-01 2.28453457e-01 -4.70250160e-01
-6.90284252e-01 9.31393087e-01 -8.18436518e-02 -4.74355787e-01
2.40550846e-01 3.08367331e-02 -6.45425841e-02 -1.68360144e-01
2.63571531e-01 7.34326601e-01 -1.69298157e-01 -5.24136007e-01
-2.98565865e-01 5.41692376e-01 9.03135836e-02 3.60192567e-01
1.26428580e+00 -5.37541747e-01 -5.44908106e-01 3.06933671e-01
1.21524060e+00 1.34846523e-01 -1.08461738e+00 -4.89564389e-01
2.89912909e-01 -5.26474357e-01 1.39936015e-01 -2.86398858e-01
-1.36996365e+00 5.87303102e-01 3.65203857e-01 8.61784816e-01
8.01254392e-01 1.57281592e-01 6.62938774e-01 5.45788825e-01
1.75667271e-01 -8.70325148e-01 -1.82479575e-01 5.24400651e-01
6.26443505e-01 -1.31105816e+00 3.12377751e-01 -1.43467414e+00
-1.25259429e-01 9.83935237e-01 2.67802596e-01 -4.70678687e-01
1.54623079e+00 1.49539396e-01 -3.67616236e-01 -1.79486841e-01
-7.43076146e-01 -3.55025500e-01 4.96542007e-01 6.48512602e-01
3.27104956e-01 1.17309712e-01 -2.55214036e-01 1.20804869e-01
-8.55540633e-02 -4.56304431e-01 5.36960423e-01 1.93749860e-01
-4.65294212e-01 -1.07879126e+00 -1.21260740e-01 9.46276367e-01
-3.16623569e-01 -4.17395622e-01 -4.27980930e-01 6.38338745e-01
-3.80350590e-01 9.69070196e-01 1.45371869e-01 -3.95265758e-01
1.27315506e-01 -4.06572849e-01 4.01825041e-01 -3.93607616e-01
8.03204998e-03 -6.56659976e-02 4.58381444e-01 -4.33757007e-01
-2.82936603e-01 -6.24044359e-01 -8.61276209e-01 -1.01148868e+00
-1.02240062e+00 5.26665390e-01 3.44738185e-01 7.55578279e-01
4.68848944e-01 4.05754507e-01 7.65074193e-01 -6.32218361e-01
-1.95163652e-01 -9.51739192e-01 -9.63134944e-01 2.62033761e-01
1.47202820e-01 -5.38934290e-01 -6.77528322e-01 -2.71827698e-01]
|
[6.959219932556152, 5.365571022033691]
|
c301b6ec-28ea-4db2-a80a-a0a4d5db94af
|
structformer-joint-unsupervised-induction-of-1
|
2012.00857
| null |
https://arxiv.org/abs/2012.00857v3
|
https://arxiv.org/pdf/2012.00857v3.pdf
|
StructFormer: Joint Unsupervised Induction of Dependency and Constituency Structure from Masked Language Modeling
|
There are two major classes of natural language grammar -- the dependency grammar that models one-to-one correspondences between words and the constituency grammar that models the assembly of one or several corresponded words. While previous unsupervised parsing methods mostly focus on only inducing one class of grammars, we introduce a novel model, StructFormer, that can simultaneously induce dependency and constituency structure. To achieve this, we propose a new parsing framework that can jointly generate a constituency tree and dependency graph. Then we integrate the induced dependency relations into the transformer, in a differentiable manner, through a novel dependency-constrained self-attention mechanism. Experimental results show that our model can achieve strong results on unsupervised constituency parsing, unsupervised dependency parsing, and masked language modeling at the same time.
|
['Aaron Courville', 'Donald Metzler', 'Dara Bahri', 'Che Zheng', 'Yi Tay', 'Yikang Shen']
|
2020-12-01
|
structformer-joint-unsupervised-induction-of
|
https://aclanthology.org/2021.acl-long.559
|
https://aclanthology.org/2021.acl-long.559.pdf
|
acl-2021-5
|
['constituency-parsing', 'unsupervised-dependency-parsing']
|
['natural-language-processing', 'natural-language-processing']
|
[ 1.28258929e-01 6.73073709e-01 -2.26691142e-01 -8.41084838e-01
-6.79115593e-01 -5.52953720e-01 4.80106056e-01 2.23805755e-01
-7.75607070e-03 3.08993280e-01 5.46734869e-01 -5.39368272e-01
3.23817611e-01 -1.06275713e+00 -6.20209932e-01 -3.88587505e-01
2.02636078e-01 3.91976178e-01 1.19525708e-01 -3.56955588e-01
-2.26907923e-05 4.04995568e-02 -1.03103662e+00 2.28335634e-01
1.00167620e+00 2.48963684e-01 4.32380289e-01 4.19060558e-01
-7.44220793e-01 8.56119752e-01 -3.80075574e-01 -4.01126415e-01
-1.52309179e-01 -6.95496261e-01 -7.74611354e-01 1.44047186e-01
-2.16685921e-01 -2.13120624e-01 -8.24730098e-02 1.09538484e+00
-2.32527871e-03 -2.80221611e-01 3.38778257e-01 -8.33311021e-01
-1.09194064e+00 1.43186629e+00 -3.20031315e-01 1.87732369e-01
2.60444582e-01 -2.39841461e-01 1.53831041e+00 -6.00850701e-01
4.08666968e-01 1.51617515e+00 1.02569692e-01 5.56241632e-01
-1.24455976e+00 -6.72859371e-01 8.69116068e-01 -4.38906580e-01
-8.28784525e-01 -2.36327648e-01 1.04108644e+00 -2.22863138e-01
1.13784349e+00 -1.65056229e-01 2.91186333e-01 7.70818949e-01
3.58451188e-01 5.99741876e-01 1.01524234e+00 -7.72318959e-01
-1.92031488e-01 -2.12791100e-01 9.42742288e-01 8.81620109e-01
5.64316511e-02 1.27589881e-01 -2.50088453e-01 1.01123363e-01
9.74204302e-01 -3.14930499e-01 9.79600996e-02 6.22985885e-02
-1.00837803e+00 1.06038404e+00 1.43608391e-01 6.58934057e-01
-8.06986168e-02 2.88828313e-02 1.17278337e-01 2.20287666e-01
5.61638653e-01 1.73916504e-01 -5.09292245e-01 4.84592974e-01
-2.00721040e-01 -6.74571842e-02 7.14821279e-01 1.28597093e+00
1.08719110e+00 -2.59084199e-02 -2.05216467e-01 7.43909836e-01
5.61563551e-01 3.54361415e-01 3.20879340e-01 -4.90425944e-01
6.33635759e-01 8.91012192e-01 -3.16388130e-01 -8.11862111e-01
-4.93606180e-01 -2.85538018e-01 -5.30976892e-01 -2.70916104e-01
7.05467388e-02 -5.33051975e-02 -1.04724419e+00 2.25433826e+00
2.52800643e-01 -6.77732602e-02 1.61383152e-01 5.40979147e-01
9.61270988e-01 8.35943043e-01 5.04189074e-01 -4.63417411e-01
1.54527903e+00 -9.86109376e-01 -9.73815739e-01 -5.87368429e-01
6.93857491e-01 -4.61803794e-01 1.25381315e+00 -2.16777459e-01
-1.25574577e+00 -6.48997605e-01 -8.50870728e-01 -4.17426556e-01
-2.48474717e-01 -5.28310128e-02 8.01668644e-01 6.44217372e-01
-9.19597685e-01 2.03747153e-01 -8.89385104e-01 -8.45351368e-02
-1.57497004e-01 4.46036011e-01 -5.17123163e-01 1.81389630e-01
-1.24318981e+00 7.04287469e-01 5.78382194e-01 1.78268217e-02
-5.01590312e-01 -3.02071273e-01 -1.35103738e+00 2.47576326e-01
1.78006679e-01 -5.64193308e-01 1.17373013e+00 -1.05361068e+00
-1.45969367e+00 1.09993422e+00 -4.85873073e-01 -1.12034857e-01
-3.87502521e-01 -1.59238860e-01 -3.62871975e-01 -2.91831136e-01
3.43607455e-01 6.05165303e-01 4.11339551e-01 -1.30303848e+00
-7.06195891e-01 -5.78745186e-01 3.41346234e-01 1.01330340e-01
-1.56296104e-01 6.57058120e-01 -7.10988104e-01 -8.23453426e-01
4.66740400e-01 -6.22660339e-01 -3.89159799e-01 -9.75338995e-01
-6.21109009e-01 -6.04053974e-01 4.30181175e-01 -7.00748622e-01
1.42135108e+00 -2.05402374e+00 3.78858894e-01 -2.27629710e-02
1.73195079e-01 -9.08359960e-02 -3.76681566e-01 3.54264498e-01
-4.19247866e-01 4.04526293e-01 -6.15741491e-01 -5.69268405e-01
-2.50929091e-02 7.69634366e-01 -4.11339372e-01 -3.28173935e-02
6.87237203e-01 1.11404240e+00 -8.80272031e-01 -5.13758659e-01
-7.55105168e-02 2.35912278e-01 -6.25012875e-01 6.41553819e-01
-4.29715753e-01 7.25552619e-01 -6.04163647e-01 4.75912213e-01
6.62137926e-01 -6.83507398e-02 7.72055328e-01 -1.20278805e-01
-3.77967894e-01 8.80249739e-01 -8.96429479e-01 1.64695847e+00
-6.49794936e-01 -3.38646211e-02 4.35401909e-02 -1.08865929e+00
1.22970545e+00 2.41544619e-01 -4.72153798e-02 -6.40572011e-01
2.76577443e-01 -7.89139643e-02 2.92077512e-01 -4.21510190e-01
3.09618562e-01 -2.09882960e-01 -6.71825647e-01 5.68760037e-01
3.16343099e-01 3.90454456e-02 3.47795993e-01 2.62565196e-01
8.58528018e-01 1.60663798e-01 3.16257328e-01 -5.42907536e-01
4.51668680e-01 -1.93385452e-01 9.66597199e-01 5.00973165e-01
1.99732617e-01 4.28479582e-01 8.74449492e-01 -4.53327686e-01
-7.93239832e-01 -1.10402036e+00 3.04051749e-02 1.61304390e+00
2.08585814e-01 -6.19665027e-01 -8.86132777e-01 -8.91314328e-01
-5.56278467e-01 8.84870470e-01 -6.12875521e-01 -2.94517353e-02
-1.25646949e+00 -6.80489063e-01 2.47409865e-01 7.68646300e-01
2.17991024e-01 -1.60309470e+00 -8.27150643e-02 4.85390842e-01
-2.59380579e-01 -1.36200881e+00 -6.42303824e-01 5.42921364e-01
-8.08073044e-01 -9.33676362e-01 9.20144022e-02 -1.52840233e+00
8.53529990e-01 -1.24393217e-01 1.46481597e+00 3.35096568e-01
2.86753565e-01 1.47645041e-01 -4.36403543e-01 -2.84503520e-01
-7.17073679e-01 1.89747632e-01 -2.44614020e-01 -9.12931636e-02
5.05166709e-01 -8.70473862e-01 2.38166209e-02 -4.54275347e-02
-9.48257267e-01 3.65297586e-01 5.33866167e-01 6.57536268e-01
5.99382699e-01 -2.94921935e-01 4.88211006e-01 -1.34202397e+00
6.18676186e-01 -3.39554280e-01 -6.77793980e-01 3.83127511e-01
-4.15714651e-01 4.52421218e-01 8.05623412e-01 -1.76092908e-01
-1.23556137e+00 3.56732160e-01 -4.97586876e-01 2.42044598e-01
-4.06148314e-01 6.21278822e-01 -9.61395085e-01 4.23317999e-01
3.74189019e-02 2.44491547e-01 -3.62277329e-01 -5.46096027e-01
7.42720783e-01 2.92856932e-01 8.02291751e-01 -9.98834789e-01
7.72047758e-01 1.04779921e-01 -2.69553453e-01 -4.50369537e-01
-9.71805334e-01 -4.78342101e-02 -1.03507853e+00 3.95725906e-01
1.13393247e+00 -8.41329217e-01 -2.01466635e-01 2.12666452e-01
-1.72286451e+00 -2.10032851e-01 -5.38646542e-02 2.54766673e-01
-3.51272643e-01 4.42710668e-01 -8.14898908e-01 -5.98279476e-01
-3.15108001e-01 -1.09849405e+00 1.04616964e+00 1.42672151e-01
-2.27239989e-02 -1.02086115e+00 2.94420362e-01 7.34612858e-03
-1.85994301e-02 7.60399923e-02 1.44098699e+00 -6.81679130e-01
-5.30405939e-01 8.22936818e-02 -2.68008977e-01 3.20857495e-01
5.15149355e-01 -5.20658679e-02 -4.93375421e-01 8.27850997e-02
2.56654825e-02 -3.21019180e-02 7.47645020e-01 1.56846836e-01
7.02042341e-01 -1.68702364e-01 -2.30841175e-01 4.30273980e-01
1.15324581e+00 3.85101944e-01 5.72392642e-01 -1.66555494e-01
8.65017951e-01 8.67530525e-01 2.66631514e-01 -6.52461722e-02
9.12927985e-01 2.72694021e-01 3.80625844e-01 -3.18863302e-01
-4.91368212e-02 -5.91870010e-01 4.49341029e-01 1.53030741e+00
6.18502647e-02 -1.00123480e-01 -8.21417451e-01 7.48540819e-01
-1.77953541e+00 -3.30496311e-01 -2.69214630e-01 1.73621500e+00
9.52431321e-01 3.26330304e-01 -1.04133621e-01 -1.49572551e-01
1.03637099e+00 2.48664901e-01 -1.01834469e-01 -5.56604445e-01
-1.24775209e-01 5.42968631e-01 5.11888117e-02 8.95783842e-01
-1.11055470e+00 1.66827476e+00 7.14164257e+00 1.72814682e-01
-8.45604420e-01 1.58761546e-01 5.36381900e-01 6.42985344e-01
-8.91552210e-01 4.82892632e-01 -1.11870861e+00 2.37642780e-01
7.63684750e-01 -2.84283757e-02 1.06146999e-01 7.33865321e-01
9.29723307e-02 3.98423493e-01 -1.11229753e+00 5.28512001e-01
-1.43547013e-01 -1.14192557e+00 3.90441328e-01 -1.23151802e-01
3.56884241e-01 -3.31081450e-01 -4.44412827e-01 3.60129863e-01
8.04270983e-01 -9.03169334e-01 5.89022577e-01 2.71612167e-04
5.30899048e-01 -7.69227326e-01 3.02262455e-01 4.42152888e-01
-1.63905442e+00 9.46829021e-02 -3.62546831e-01 -3.46214354e-01
5.53228974e-01 4.60541666e-01 -7.79483467e-02 6.76326871e-01
3.22301060e-01 4.39839810e-01 -2.65068591e-01 2.00025037e-01
-1.00523150e+00 7.55838692e-01 -4.88203093e-02 -1.08804908e-02
1.92594409e-01 -4.82268602e-01 4.32725579e-01 1.40776682e+00
8.10374785e-03 5.45513570e-01 5.23535073e-01 9.39563215e-01
-4.51387912e-02 4.18965518e-01 -4.71231133e-01 -2.06327741e-03
5.65581501e-01 1.25987542e+00 -8.16219330e-01 -3.58145386e-01
-7.59058714e-01 7.18336642e-01 8.50724578e-01 1.45472810e-01
-8.88295174e-01 -2.07684979e-01 5.13975680e-01 1.76142678e-02
3.48314494e-01 -5.82029939e-01 -3.16541404e-01 -1.23858047e+00
1.43943310e-01 -7.42752254e-01 4.79011774e-01 -4.92727399e-01
-1.20047867e+00 1.00699055e+00 2.97532361e-02 -4.17801797e-01
-1.40503302e-01 -5.57996452e-01 -9.96765018e-01 9.60626185e-01
-1.61324775e+00 -1.51824391e+00 6.48248196e-02 5.47327518e-01
6.39874816e-01 2.91201174e-02 1.14821935e+00 8.01163465e-02
-8.25711548e-01 4.65515226e-01 -7.26227462e-01 5.64028561e-01
2.26154879e-01 -1.43208873e+00 9.91580129e-01 1.19358790e+00
5.11041820e-01 9.95451570e-01 3.32114518e-01 -7.07000673e-01
-1.18477285e+00 -1.17088032e+00 1.38813245e+00 -4.65870410e-01
6.15807593e-01 -7.96573520e-01 -1.09412920e+00 1.12646544e+00
4.75755066e-01 -6.02660365e-02 6.58867776e-01 3.26950312e-01
-6.48650289e-01 1.32061779e-01 -5.49912095e-01 4.71617162e-01
1.32343268e+00 -5.04421115e-01 -1.03808439e+00 2.19878852e-01
1.39371681e+00 -2.86965221e-01 -7.25787759e-01 5.22719979e-01
1.80637762e-01 -8.57792020e-01 5.90531468e-01 -9.36718524e-01
6.21808708e-01 -2.58700728e-01 -2.10802332e-01 -1.18842292e+00
-8.11699867e-01 -6.38115525e-01 1.50862440e-01 1.74106526e+00
8.43879402e-01 -7.08560288e-01 5.19468904e-01 6.03084326e-01
-3.86224240e-01 -3.29906762e-01 -6.35140657e-01 -6.39001429e-01
4.34885502e-01 -4.47342008e-01 8.44750524e-01 9.96609390e-01
1.18761815e-01 9.67926443e-01 -2.10783124e-01 5.50378501e-01
4.69255924e-01 4.28725600e-01 5.39992154e-01 -1.12811136e+00
-6.27046764e-01 -2.39665285e-01 6.13535456e-02 -1.35928488e+00
6.69216573e-01 -1.00472701e+00 4.21833396e-01 -1.61725140e+00
1.52056351e-01 -4.96889979e-01 -4.10016358e-01 7.03308225e-01
-5.97649038e-01 -3.15652907e-01 -4.26894836e-02 -6.32582977e-02
-2.62972772e-01 4.48046386e-01 1.10858607e+00 4.15836386e-02
-3.64979446e-01 -2.61625350e-01 -1.18276298e+00 7.24534035e-01
8.22183609e-01 -6.91760600e-01 -4.03780162e-01 -9.09419060e-01
7.21391886e-02 1.79116830e-01 -1.49151087e-01 -4.44519997e-01
4.84950803e-02 -3.15816849e-01 -2.40577742e-01 -4.01453465e-01
-2.91181713e-01 -4.61838454e-01 -2.24356264e-01 7.78615251e-02
-3.02083850e-01 3.87499273e-01 2.24178005e-02 2.92030275e-01
-3.07658613e-01 -1.68311179e-01 5.62301517e-01 -2.92859316e-01
-5.97977936e-01 3.14557582e-01 -3.13623369e-01 1.48440257e-01
6.11831427e-01 2.85868555e-01 -3.05753559e-01 -1.84116268e-03
-9.17617619e-01 1.72975287e-01 1.18494499e-03 7.02926755e-01
2.22785577e-01 -1.21142399e+00 -8.89379144e-01 4.85217661e-01
5.94491325e-03 1.76746577e-01 -3.07741135e-01 2.09211528e-01
-2.21891403e-02 3.04747880e-01 1.34219071e-02 -3.57838362e-01
-1.05102444e+00 6.59599185e-01 1.28519505e-01 -6.99648738e-01
-6.64848447e-01 8.62852812e-01 9.01775122e-01 -8.35103095e-01
-2.87254751e-02 -6.24577582e-01 -3.06985885e-01 -4.47348744e-01
4.76699233e-01 -4.94273841e-01 -1.59411401e-01 -9.25102353e-01
-4.05260801e-01 8.12781811e-01 -1.65897295e-01 -2.72685979e-02
1.09178042e+00 -1.11996636e-01 -7.24513590e-01 4.18319196e-01
7.94414759e-01 3.77456307e-01 -1.07327509e+00 -1.14488624e-01
3.70007634e-01 4.38480563e-02 -2.66862810e-01 -3.40237916e-01
-9.08139110e-01 7.33473122e-01 -1.07205257e-01 5.31543076e-01
1.21126878e+00 5.73435128e-01 8.16538334e-01 1.63939536e-01
2.49138936e-01 -6.79705203e-01 -3.39528397e-02 9.59973633e-01
7.91099668e-01 -9.80117321e-01 -5.77736974e-01 -1.16177344e+00
-4.03634191e-01 9.61549044e-01 7.77845621e-01 -3.46039474e-01
6.91370666e-01 6.39720738e-01 2.05078527e-01 -1.28153399e-01
-7.69292533e-01 -5.26486218e-01 5.00169210e-03 6.15597486e-01
9.26212728e-01 1.81968719e-01 -6.38120949e-01 1.34072649e+00
-2.66353250e-01 -4.70500410e-01 1.52947322e-01 8.46038103e-01
-6.20832860e-01 -1.90915418e+00 -2.30839215e-02 -1.04675367e-01
-5.02489984e-01 -5.78601718e-01 -4.98906225e-01 7.60241508e-01
9.25687477e-02 1.18479502e+00 4.23203200e-01 -2.93712378e-01
5.15649855e-01 1.79796681e-01 5.51858664e-01 -1.22227871e+00
-6.91445529e-01 3.72786283e-01 2.92217195e-01 -3.88443023e-01
-3.74659687e-01 -2.04439431e-01 -1.59635603e+00 2.31149495e-01
-4.79266137e-01 3.04393262e-01 3.70140463e-01 1.28277385e+00
1.97122514e-01 6.35734081e-01 7.89325118e-01 -4.04128045e-01
2.53286622e-02 -8.88120890e-01 -4.48387086e-01 3.80534172e-01
-3.68618332e-02 -2.39095494e-01 -1.59167796e-02 3.78080249e-01]
|
[10.326236724853516, 9.545710563659668]
|
fe4c5b54-b533-49fe-9a2d-9afd237a0505
|
light-weighted-cnn-attention-based
|
2210.15119
| null |
https://arxiv.org/abs/2210.15119v1
|
https://arxiv.org/pdf/2210.15119v1.pdf
|
Light-weighted CNN-Attention based architecture for Hand Gesture Recognition via ElectroMyography
|
Advancements in Biological Signal Processing (BSP) and Machine-Learning (ML) models have paved the path for development of novel immersive Human-Machine Interfaces (HMI). In this context, there has been a surge of significant interest in Hand Gesture Recognition (HGR) utilizing Surface-Electromyogram (sEMG) signals. This is due to its unique potential for decoding wearable data to interpret human intent for immersion in Mixed Reality (MR) environments. To achieve the highest possible accuracy, complicated and heavy-weighted Deep Neural Networks (DNNs) are typically developed, which restricts their practical application in low-power and resource-constrained wearable systems. In this work, we propose a light-weighted hybrid architecture (HDCAM) based on Convolutional Neural Network (CNN) and attention mechanism to effectively extract local and global representations of the input. The proposed HDCAM model with 58,441 parameters reached a new state-of-the-art (SOTA) performance with 82.91% and 81.28% accuracy on window sizes of 300 ms and 200 ms for classifying 17 hand gestures. The number of parameters to train the proposed HDCAM architecture is 18.87 times less than its previous SOTA counterpart.
|
['Arash Mohammadi', 'Amir Asif', 'Elahe Rahimian', 'Soheil Zabihi']
|
2022-10-27
| null | null | null | null |
['hand-gesture-recognition', 'hand-gesture-recognition-1', 'gesture-recognition', 'mixed-reality']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 4.17927086e-01 -2.96926081e-01 -2.94149518e-02 -2.05527529e-01
-4.49678510e-01 -2.11868472e-02 2.04403862e-01 -4.21864212e-01
-9.95622039e-01 8.00481558e-01 1.20882966e-01 -8.99581835e-02
-1.48554131e-01 -4.68260318e-01 -4.58121002e-01 -6.98086381e-01
-2.29797512e-01 -2.24105775e-01 -2.84573250e-02 -1.74472705e-01
2.33940825e-01 5.20208001e-01 -1.44846499e+00 2.33431280e-01
8.05004239e-01 1.03097379e+00 5.25703073e-01 1.07370889e+00
2.57761270e-01 4.92225498e-01 -6.44666433e-01 -7.45937899e-02
-8.14460963e-02 -4.30787951e-01 -3.80957603e-01 -6.16252363e-01
1.71270892e-01 -4.42565680e-01 -7.29564726e-01 8.07833493e-01
1.28958416e+00 2.13245988e-01 3.83291990e-01 -7.57845044e-01
-5.97410500e-01 2.98175484e-01 -4.04954165e-01 6.51174784e-01
3.75617266e-01 2.36075401e-01 5.91944456e-01 -8.17933977e-01
5.37084162e-01 7.88071573e-01 5.73958337e-01 9.23127353e-01
-8.33151460e-01 -6.77427471e-01 1.43400237e-01 6.22501493e-01
-1.26674521e+00 -2.18384728e-01 7.24909425e-01 -2.00757772e-01
1.50764430e+00 5.48259854e-01 9.34567153e-01 1.82023478e+00
5.42547822e-01 7.81641841e-01 9.08922672e-01 -2.08348095e-01
3.02393943e-01 -3.14462751e-01 3.29762042e-01 2.96976060e-01
4.42166537e-01 -4.94214706e-02 -9.05087650e-01 1.99294209e-01
9.72551227e-01 2.25501880e-01 -4.58357424e-01 3.97413820e-01
-1.13592434e+00 2.41175115e-01 5.07783592e-01 6.01992249e-01
-8.74490976e-01 3.58508617e-01 4.63064879e-01 1.12966597e-01
1.74068779e-01 2.07557365e-01 -3.41373235e-01 -8.71456623e-01
-7.20840812e-01 2.26494476e-01 5.90253711e-01 4.81925815e-01
-3.10144931e-01 4.31363493e-01 -1.03194537e-02 6.88852727e-01
3.07037413e-01 5.31328857e-01 7.67756879e-01 -4.91154075e-01
7.98057616e-01 4.65131342e-01 -3.63963321e-02 -9.05468047e-01
-7.70148158e-01 -4.40227985e-01 -1.07969844e+00 1.01693980e-01
2.37062186e-01 -3.30648750e-01 -9.28533375e-01 1.52343941e+00
-3.75237949e-02 1.97939023e-01 -1.20549239e-01 1.51078606e+00
8.92698824e-01 5.17792344e-01 4.03298080e-01 9.14497077e-02
1.17551851e+00 -2.52363324e-01 -9.29091036e-01 -3.19495231e-01
2.45989755e-01 -1.58949003e-01 1.18665385e+00 6.75469220e-01
-8.44765544e-01 -6.72937930e-01 -1.36937475e+00 6.34708554e-02
-2.78473765e-01 2.95346320e-01 6.51361465e-01 8.17980409e-01
-6.44854009e-01 8.02914619e-01 -1.47962415e+00 -4.62277949e-01
4.67425317e-01 9.18925703e-01 -1.89502642e-01 2.26616800e-01
-1.16712785e+00 8.62202227e-01 1.59953937e-01 8.57192039e-01
-5.14831901e-01 -3.56667221e-01 -4.91295010e-01 -1.68028533e-01
-2.44598277e-02 -3.95424336e-01 5.84960401e-01 -6.44881785e-01
-1.85036266e+00 4.68670249e-01 -2.97501008e-03 -3.22152138e-01
3.41544658e-01 -7.34420002e-01 -6.20719075e-01 4.74206582e-02
-6.99144423e-01 4.53511477e-01 6.78510845e-01 -5.29031754e-01
-2.47845352e-01 -9.43497598e-01 -1.99337631e-01 2.09874123e-01
-6.28068328e-01 2.79106230e-01 -2.52772644e-02 -7.01818585e-01
2.90332645e-01 -8.29345345e-01 8.02468732e-02 -1.60829082e-01
-1.22931018e-01 -2.37770408e-01 6.17237210e-01 -1.19295037e+00
1.43010294e+00 -1.91000187e+00 5.61203837e-01 5.02175018e-02
2.15312421e-01 8.15743685e-01 -1.26603797e-01 2.15926051e-01
6.18224517e-02 -1.64440051e-01 -2.14069430e-03 4.87252437e-02
-1.78815633e-01 2.49766111e-01 -7.35488161e-02 5.03389537e-01
1.03666842e-01 1.00746250e+00 -6.45157397e-01 -6.00507297e-02
6.00192368e-01 8.78781199e-01 -3.93337280e-01 3.64334494e-01
3.04197818e-01 8.40819955e-01 -3.44991654e-01 8.30381095e-01
4.16193694e-01 2.16602489e-01 2.95369893e-01 -3.38534266e-01
-1.43973762e-02 2.09181815e-01 -1.19342506e+00 2.01484394e+00
-5.90698242e-01 1.00929570e+00 7.07883202e-03 -8.32844079e-01
8.36838305e-01 5.08452654e-01 5.51952064e-01 -9.09512162e-01
6.85924470e-01 3.46662939e-01 3.91529799e-01 -1.17180693e+00
1.25440449e-01 9.63408351e-02 2.65448928e-01 7.50801573e-03
2.08061591e-01 6.32323027e-01 -2.94495732e-01 -3.61753017e-01
1.28095925e+00 4.06045347e-01 1.12319745e-01 3.83909717e-02
3.72252464e-01 -5.35411358e-01 4.25394326e-01 5.68397343e-01
-3.42687100e-01 4.98274297e-01 -1.21414721e-01 -5.48143327e-01
-6.75335109e-01 -9.40885723e-01 6.40250882e-03 9.43964660e-01
1.09179281e-01 -1.47490337e-01 -8.12402308e-01 8.60620141e-02
-2.86382467e-01 3.90884161e-01 -3.35964471e-01 -2.55865604e-01
-9.93163466e-01 -8.04613769e-01 7.84220457e-01 1.19119644e+00
6.58657432e-01 -1.46594286e+00 -1.32519996e+00 4.86827612e-01
-3.22967991e-02 -1.06705463e+00 6.87560216e-02 2.92754412e-01
-1.20979095e+00 -9.08880055e-01 -1.04013216e+00 -6.26789689e-01
2.78147697e-01 -2.48402327e-01 3.01681757e-01 -5.56622982e-01
-5.62069058e-01 3.48861307e-01 -5.06516218e-01 -4.98969525e-01
4.31190521e-01 2.60330975e-01 4.05571014e-01 -1.44418463e-01
6.74416780e-01 -8.42728674e-01 -7.42424965e-01 -1.97760686e-02
-6.39224231e-01 1.28312167e-02 6.62522376e-01 9.38241363e-01
2.13762179e-01 -6.98777020e-01 6.91608369e-01 -1.48349658e-01
9.61387813e-01 -2.01204017e-01 -8.35926533e-02 2.18460023e-01
-2.23892912e-01 -2.81546026e-01 5.66977322e-01 -8.57757986e-01
-1.12474191e+00 -7.13885948e-02 -5.63489854e-01 -2.05525756e-01
-3.15007329e-01 4.50115681e-01 -1.73735991e-01 -1.38283297e-01
6.25428200e-01 2.64141053e-01 -2.92475581e-01 -6.04184449e-01
4.10827585e-02 1.28136849e+00 7.91741610e-01 -4.55367178e-01
1.09676048e-01 1.16940834e-01 -1.66982919e-01 -1.16446304e+00
-1.19729742e-01 -2.65765816e-01 -7.18931258e-01 -5.32549381e-01
9.94550765e-01 -7.93962359e-01 -1.16665685e+00 8.65038991e-01
-1.23673773e+00 -4.25272554e-01 1.43703654e-01 1.11644924e+00
-3.40696245e-01 1.87043935e-01 -6.35256231e-01 -1.25702167e+00
-7.78066874e-01 -8.92721653e-01 8.16867888e-01 4.17307526e-01
-6.66349709e-01 -2.74856210e-01 -8.61127451e-02 2.63063252e-01
6.55183315e-01 3.46474946e-01 3.87088686e-01 -4.61086243e-01
-7.82323629e-02 -6.01171434e-01 -1.05079740e-01 3.91603589e-01
4.01639305e-02 -5.25000036e-01 -1.12480223e+00 -3.29308212e-01
9.88067165e-02 -2.58366913e-01 5.40015519e-01 5.19961536e-01
1.32256937e+00 6.58180267e-02 -9.48049054e-02 5.81342697e-01
1.26014066e+00 7.19739079e-01 9.72769201e-01 1.59989029e-01
8.17669988e-01 2.75704801e-01 1.90477520e-01 5.92850208e-01
3.11046932e-02 6.78966820e-01 3.33534390e-01 2.80707218e-02
5.46632558e-02 3.42192911e-02 3.65079880e-01 9.69107926e-01
-9.44151103e-01 -2.02394232e-01 -7.88758695e-01 3.31708670e-01
-1.74903774e+00 -7.88666010e-01 -1.15709931e-01 2.10837054e+00
5.94540715e-01 9.39949474e-04 -4.27776687e-02 3.81969988e-01
5.97464025e-01 1.06829353e-01 -8.81641448e-01 -5.78263879e-01
8.58239606e-02 8.95905793e-01 4.03739750e-01 5.23728468e-02
-9.77118790e-01 5.12931049e-01 4.87875605e+00 5.41319847e-01
-1.33757877e+00 1.98701113e-01 6.90083131e-02 -4.34059054e-01
4.61725950e-01 -8.83193672e-01 -6.40865147e-01 5.38762987e-01
1.28694808e+00 4.86650765e-01 6.37942374e-01 6.78126395e-01
4.34692681e-01 -2.77805869e-02 -8.86008739e-01 1.57798719e+00
1.61325842e-01 -9.54729199e-01 -4.15597230e-01 -2.16522743e-03
3.02750796e-01 1.24870494e-01 2.16069724e-02 1.57986656e-01
-6.75869882e-01 -1.09706736e+00 4.28560495e-01 6.34241700e-01
1.04662907e+00 -5.89139044e-01 9.36971307e-01 2.94087291e-01
-1.20416522e+00 -3.98218244e-01 -2.01688707e-01 -4.67774481e-01
1.67269140e-01 -4.98860441e-02 -2.08862916e-01 1.96336329e-01
8.78477395e-01 5.11761427e-01 -2.15769615e-02 8.70807528e-01
-1.24750942e-01 7.06935644e-01 -4.82887268e-01 -6.10957623e-01
8.22597817e-02 1.41973630e-01 6.46852851e-01 1.19131672e+00
3.27587813e-01 6.02890313e-01 -4.79748189e-01 7.29096889e-01
-2.01669652e-02 -1.23194985e-01 -3.51669729e-01 -2.76516140e-01
1.28005788e-01 9.28413510e-01 -5.19606531e-01 -1.34204507e-01
-2.59710163e-01 1.29944158e+00 -4.06470709e-02 3.97777289e-01
-8.17585051e-01 -8.91794562e-01 7.65339494e-01 -1.49807349e-01
-8.50879401e-02 -7.35689104e-01 -6.56230867e-01 -1.17541492e+00
4.82526571e-01 -7.61770189e-01 9.84090343e-02 -4.78683710e-01
-1.05765331e+00 6.98803127e-01 -2.40073800e-01 -1.16192031e+00
-1.62068680e-01 -1.07928348e+00 -4.90559191e-01 9.79945302e-01
-1.05910766e+00 -7.80766249e-01 -7.63855100e-01 5.32175481e-01
6.43118858e-01 -2.41355668e-03 1.06649363e+00 6.64056599e-01
-7.26393938e-01 7.02845931e-01 -7.85247050e-03 1.81901187e-01
3.57837409e-01 -8.27920556e-01 3.67923766e-01 7.69795358e-01
6.04041619e-04 9.18521464e-01 3.84891003e-01 -5.64059913e-01
-1.89219618e+00 -6.67879581e-01 5.23391843e-01 -1.52702555e-01
3.35649401e-01 -4.05962706e-01 -7.64874816e-01 4.49810177e-01
1.20455585e-02 -1.51448995e-01 6.44131601e-01 -8.13922510e-02
8.96919146e-02 -1.52178600e-01 -1.13263321e+00 7.15851307e-01
1.35925972e+00 -4.83948559e-01 -5.91359437e-01 -1.59370735e-01
1.25171304e-01 -6.00262880e-01 -9.63592649e-01 4.73324448e-01
1.27448809e+00 -5.05346060e-01 8.70987475e-01 -6.84762836e-01
3.65693867e-02 -1.68988295e-02 -2.77073443e-01 -9.50776517e-01
-7.05120247e-03 -4.96358812e-01 -5.68277717e-01 5.38625836e-01
-9.24256369e-02 -6.87713683e-01 8.16236019e-01 8.98429692e-01
-2.12285638e-01 -1.07427549e+00 -1.24519956e+00 -5.76883554e-01
-4.54337180e-01 -8.17336857e-01 1.83068871e-01 3.23014289e-01
4.66250509e-01 9.29669440e-02 -6.08441889e-01 -1.60608254e-02
3.90735805e-01 -4.64718252e-01 2.35569552e-01 -1.08333158e+00
-1.29930899e-01 -2.50936657e-01 -8.87294054e-01 -9.90187883e-01
-2.22345442e-01 -4.84328300e-01 -1.19351707e-02 -1.54276466e+00
-1.35706797e-01 9.93902981e-02 -6.72988355e-01 3.37938368e-01
-1.11952908e-02 4.65555370e-01 1.51529804e-01 -1.10593222e-01
-2.79695421e-01 5.79089165e-01 9.59829152e-01 -8.48406851e-02
-6.24389887e-01 8.19446966e-02 -1.32137045e-01 5.01972914e-01
1.00862873e+00 -1.83400571e-01 -1.05758466e-01 -5.05970597e-01
-2.96747256e-02 9.36563909e-02 5.12150884e-01 -1.50690091e+00
3.51355851e-01 2.44202659e-01 8.35801363e-01 -5.75112522e-01
5.52758515e-01 -7.52047658e-01 9.19471011e-02 7.08031952e-01
-4.50197220e-01 -9.70090181e-02 2.00484797e-01 4.00144637e-01
2.32151859e-02 6.43369481e-02 3.58959347e-01 5.06829582e-02
-9.82130349e-01 9.24837664e-02 -6.15994990e-01 -4.60360497e-01
7.56175876e-01 -6.67627454e-01 3.20007242e-02 4.21244241e-02
-1.06049931e+00 -1.62240401e-01 -4.72342521e-01 7.05816925e-01
1.10828006e+00 -1.21296906e+00 -3.46514851e-01 3.85225892e-01
-1.24885857e-01 -3.14278156e-01 6.87463343e-01 9.33183849e-01
-4.03280228e-01 7.35153437e-01 -7.48468280e-01 -3.90304238e-01
-1.27077806e+00 -2.26611987e-01 2.92394698e-01 1.50830343e-01
-1.02343690e+00 8.88064265e-01 -6.66582227e-01 -8.49261731e-02
7.75172353e-01 -6.09907031e-01 -3.05086672e-01 -1.23441197e-01
6.79886341e-01 9.53848720e-01 1.79952666e-01 -3.64020824e-01
-5.59293807e-01 5.63898921e-01 3.16888809e-01 -1.15381688e-01
1.57117879e+00 3.03815484e-01 3.05959523e-01 5.16647518e-01
1.01351237e+00 -6.50758505e-01 -1.15889633e+00 1.86889410e-01
-9.03539956e-02 -3.39264274e-01 2.62223691e-01 -1.05584490e+00
-9.27815974e-01 1.25983298e+00 1.38429058e+00 -6.06925428e-01
1.12834954e+00 -5.19280195e-01 1.21929157e+00 6.68983042e-01
6.48429990e-01 -1.31491518e+00 -7.63203204e-02 1.20114200e-01
9.22976971e-01 -8.90958965e-01 -2.50935674e-01 1.28347248e-01
-5.28123379e-01 1.32751322e+00 7.73200214e-01 -4.18895304e-01
5.49473405e-01 2.73229092e-01 -1.41599759e-01 -1.43347219e-01
-1.54058650e-01 4.83554155e-02 3.86160493e-01 7.46363521e-01
6.23025656e-01 2.18334660e-01 -6.92621589e-01 1.24318063e+00
1.63713381e-01 5.88481486e-01 2.08311193e-02 1.12820256e+00
-2.66055077e-01 -6.12485707e-01 -2.01783136e-01 7.22193241e-01
-4.89920169e-01 1.41545296e-01 -6.70635849e-02 6.43765509e-01
8.80362540e-02 8.67604971e-01 -9.75773856e-02 -9.73452508e-01
4.72214073e-01 2.13568091e-01 8.61561120e-01 -2.89966822e-01
-7.77503312e-01 -1.07750319e-01 -1.06339686e-01 -7.57723987e-01
-4.13205475e-01 -3.45231771e-01 -1.36395407e+00 -6.79646432e-02
-2.63824314e-01 -4.46869820e-01 1.09881771e+00 8.75590682e-01
3.90215695e-01 8.99982870e-01 -9.38556250e-03 -1.33714235e+00
-5.81389785e-01 -1.46280324e+00 -6.63143635e-01 3.18108127e-02
-1.35349324e-02 -7.20255136e-01 3.48476879e-02 -2.31185317e-01]
|
[6.81045389175415, 0.13164184987545013]
|
a6d6fb2d-0286-42a0-a0f7-679dc2b797e8
|
neural-embedding-allocation-distributed
|
1909.04702
| null |
https://arxiv.org/abs/1909.04702v1
|
https://arxiv.org/pdf/1909.04702v1.pdf
|
Neural Embedding Allocation: Distributed Representations of Topic Models
|
Word embedding models such as the skip-gram learn vector representations of words' semantic relationships, and document embedding models learn similar representations for documents. On the other hand, topic models provide latent representations of the documents' topical themes. To get the benefits of these representations simultaneously, we propose a unifying algorithm, called neural embedding allocation (NEA), which deconstructs topic models into interpretable vector-space embeddings of words, topics, documents, authors, and so on, by learning neural embeddings to mimic the topic models. We showcase NEA's effectiveness and generality on LDA, author-topic models and the recently proposed mixed membership skip gram topic model and achieve better performance with the embeddings compared to several state-of-the-art models. Furthermore, we demonstrate that using NEA to smooth out the topics improves coherence scores over the original topic models when the number of topics is large.
|
['Kamrun Naher Keya', 'James R. Foulds', 'Yannis Papanikolaou']
|
2019-09-10
| null | null | null | null |
['document-embedding']
|
['methodology']
|
[-6.03464127e-01 3.17010224e-01 -6.49023771e-01 -1.96525887e-01
-5.40037930e-01 -1.96261987e-01 1.05849373e+00 1.34361580e-01
1.18215643e-01 2.30859667e-01 1.09073794e+00 -2.59415895e-01
-1.73423976e-01 -8.96409750e-01 -1.74140066e-01 -5.88935137e-01
-3.77751768e-01 7.21343637e-01 -6.47059307e-02 4.56770360e-02
4.23580736e-01 2.29024161e-02 -1.26033437e+00 1.87511998e-03
7.58160412e-01 4.66270238e-01 9.14032161e-02 3.51240873e-01
-8.44668269e-01 1.27334893e-01 -6.24618709e-01 -1.64721504e-01
-2.03839898e-01 4.59951311e-02 -5.82208037e-01 1.02219127e-01
3.08048993e-01 -4.63994324e-01 -7.83666551e-01 8.06796372e-01
9.85792652e-02 4.17323321e-01 1.32262599e+00 -1.46356332e+00
-1.47345364e+00 7.34928787e-01 -8.08754563e-01 -8.63164589e-02
-2.25653231e-01 -4.53705490e-01 1.65299916e+00 -1.29143429e+00
5.35561919e-01 1.73495412e+00 5.45786798e-01 3.54356349e-01
-1.26417673e+00 -6.66143000e-01 6.42655432e-01 1.23009868e-01
-1.28043497e+00 1.67791322e-01 9.56191182e-01 -7.33439565e-01
1.03604233e+00 -1.38138786e-01 5.45104206e-01 1.20094323e+00
3.19198608e-01 8.92738402e-01 3.92003924e-01 -5.01730084e-01
1.70280948e-01 3.47066760e-01 9.17633832e-01 4.40353274e-01
6.34020388e-01 -4.12348032e-01 -5.19297659e-01 -8.58240724e-01
6.87791646e-01 5.40353715e-01 -8.60946774e-02 -5.75268984e-01
-1.07888865e+00 1.60888970e+00 2.25673258e-01 3.39052230e-01
-5.58808565e-01 4.60837066e-01 4.88513440e-01 -8.46116096e-02
1.16682768e+00 6.11272156e-01 -3.20272148e-01 1.60262883e-01
-9.46611226e-01 4.98851687e-01 6.28759027e-01 8.77002120e-01
8.25676620e-01 7.74180293e-02 -4.56354171e-01 1.08148479e+00
9.15467441e-01 2.14054272e-01 9.18751776e-01 -7.24590242e-01
1.36390612e-01 6.48796976e-01 -2.43825074e-02 -1.23303521e+00
-1.95942186e-02 -2.99356848e-01 -6.39809787e-01 -3.62383544e-01
-2.93554872e-01 6.96839988e-02 -9.16400969e-01 1.61111665e+00
-3.89864901e-04 2.05402866e-01 6.50564656e-02 4.82037365e-01
8.60793412e-01 1.43618512e+00 2.13123813e-01 -7.23502412e-02
1.71417570e+00 -1.23220706e+00 -1.24982882e+00 -1.01421945e-01
5.67439198e-01 -5.38955629e-01 9.81451809e-01 6.21252283e-02
-7.58634746e-01 -3.51420522e-01 -7.12590098e-01 -3.77311617e-01
-7.80947208e-01 2.77055949e-01 1.01280594e+00 3.62897694e-01
-1.19136584e+00 1.57940716e-01 -8.66679132e-01 -5.86170256e-01
3.77939641e-01 1.49030194e-01 -9.23125446e-02 -8.75487924e-02
-1.32780385e+00 5.99068701e-01 4.04647082e-01 -5.60208917e-01
-8.81618857e-01 -8.61656964e-01 -8.83504689e-01 6.63659871e-01
-4.37173620e-02 -6.16365552e-01 9.81401801e-01 -5.89075089e-02
-1.28814471e+00 5.18745780e-01 -5.81453145e-01 -3.20896894e-01
-4.37505543e-01 -6.35539949e-01 -3.05062234e-01 -2.01055445e-02
1.71391025e-01 9.65676367e-01 8.42131734e-01 -1.16642749e+00
-4.27841365e-01 -2.28030160e-01 -1.13980301e-01 -4.34487872e-02
-1.42758560e+00 -4.12393846e-02 -4.75660235e-01 -9.72779632e-01
1.06373884e-01 -6.76313162e-01 -1.21216200e-01 1.43282086e-01
-3.75736743e-01 -1.21591735e+00 1.31450510e+00 -5.76373279e-01
1.57012939e+00 -2.18700671e+00 2.83817202e-01 1.43052628e-02
5.68906426e-01 -1.86383143e-01 -4.05665964e-01 7.86159992e-01
-1.52047470e-01 3.19440514e-01 1.97713584e-01 -5.84867835e-01
3.93952340e-01 3.80647242e-01 -9.85089779e-01 3.64146948e-01
6.64006034e-03 7.77421892e-01 -6.76240027e-01 -3.96282911e-01
-5.47052622e-02 6.72447681e-01 -7.92029679e-01 2.77215183e-01
-2.57742763e-01 -4.95368212e-01 -6.86064899e-01 2.24513814e-01
4.65530425e-01 -5.56530833e-01 2.64035374e-01 -5.24807498e-02
1.17924854e-01 5.13711154e-01 -8.08442473e-01 1.72374928e+00
-4.40679133e-01 1.19254458e+00 -3.46268564e-01 -8.28241706e-01
1.12648189e+00 7.16527462e-01 4.16607976e-01 1.10882491e-01
-5.40047996e-02 -1.38475806e-01 -2.35958189e-01 -1.63024038e-01
1.11502528e+00 -2.98454519e-03 -9.85231400e-02 9.86438394e-01
4.17021036e-01 2.20939755e-01 -1.27081290e-01 8.31730068e-01
5.70599198e-01 -2.58461505e-01 2.61211812e-01 -5.01124024e-01
-1.25593930e-01 -2.22401828e-01 7.84449428e-02 5.53803742e-01
1.79724425e-01 3.82943124e-01 6.83743119e-01 -4.50588405e-01
-1.22338104e+00 -1.07465780e+00 -2.37448812e-01 1.35127270e+00
-1.82983354e-01 -9.30609345e-01 -3.29478830e-01 -5.66824615e-01
3.45597893e-01 1.05777526e+00 -9.28792059e-01 -2.13988587e-01
-2.58384407e-01 -7.20336497e-01 9.75471810e-02 8.22514892e-01
-3.12947690e-01 -6.36352718e-01 6.19994625e-02 9.41293016e-02
-7.09895268e-02 -5.98097444e-01 -5.65075696e-01 4.34855521e-02
-1.05132651e+00 -5.18649995e-01 -1.17167020e+00 -6.80273354e-01
6.95340931e-01 5.33226252e-01 9.47357357e-01 -4.14176702e-01
-1.50429696e-01 5.29349506e-01 -3.34752142e-01 -3.40158314e-01
3.78069393e-02 2.23423943e-01 3.34460497e-01 -3.21106374e-01
8.31448674e-01 -7.18050897e-01 -5.67140222e-01 -7.80416420e-03
-9.40408647e-01 -3.37897748e-01 2.99758345e-01 9.49101150e-01
6.81170076e-02 8.14745203e-02 6.82505906e-01 -8.18548441e-01
1.21085763e+00 -1.00458264e+00 -1.60399333e-01 1.22750886e-01
-1.00045419e+00 1.75132081e-01 8.45415145e-02 -7.61592925e-01
-8.00630569e-01 -9.11727130e-01 4.33000803e-01 -6.06695533e-01
7.11287558e-02 5.63653529e-01 1.15634255e-01 6.50737882e-01
3.44978899e-01 2.53966868e-01 -1.85963642e-02 -7.55540192e-01
1.15423286e+00 9.53880966e-01 3.82995382e-02 -6.76693439e-01
6.98286414e-01 5.15871644e-01 -5.15624106e-01 -8.21598232e-01
-6.14808381e-01 -9.64441359e-01 -3.79518718e-01 4.27711666e-01
8.43319595e-01 -1.17966199e+00 -9.95379407e-03 -1.78744838e-01
-1.71733654e+00 3.64158511e-01 -3.78723979e-01 7.55362451e-01
-2.87022203e-01 4.14267331e-01 -7.00870991e-01 -6.87430203e-01
-4.37159330e-01 -8.44703436e-01 1.13051212e+00 1.50716484e-01
-7.38446712e-01 -1.48229945e+00 4.99433935e-01 -2.20045403e-01
6.15054905e-01 -4.07638788e-01 1.58238769e+00 -1.08441281e+00
-3.81226569e-01 -2.77945280e-01 -4.67033088e-01 1.88536599e-01
3.26672137e-01 -3.26914825e-02 -9.10745561e-01 -3.54246467e-01
-3.75309318e-01 1.05702272e-02 1.20614517e+00 7.79833257e-01
1.04117084e+00 -5.99539876e-01 -8.30227256e-01 2.59044468e-01
1.18747294e+00 4.13824432e-02 4.80339289e-01 3.85872722e-01
6.30617559e-01 6.55300617e-01 1.60333470e-01 6.00537241e-01
4.48899239e-01 6.03070259e-01 1.73743442e-01 -7.58860856e-02
4.11553867e-03 -3.18049818e-01 3.44091713e-01 1.26462066e+00
1.81327611e-01 -5.30633569e-01 -6.38580203e-01 1.06848836e+00
-1.76636851e+00 -6.71738982e-01 3.69604258e-03 1.58006847e+00
5.85014284e-01 -2.03609258e-01 4.13014404e-02 -3.08317095e-01
6.35724306e-01 7.81061769e-01 -2.11982399e-01 -5.84642649e-01
1.90851003e-01 -1.41351759e-01 3.44873965e-02 2.87392288e-01
-9.68355536e-01 1.07039940e+00 7.54016924e+00 9.72981334e-01
-7.47734725e-01 4.33698356e-01 2.52454340e-01 -2.47635841e-02
-9.08822417e-01 7.31198341e-02 -1.01066339e+00 3.44818890e-01
9.90481853e-01 -5.77376187e-01 -1.72529474e-01 1.29100084e+00
8.42971951e-02 4.85731035e-01 -1.10889101e+00 6.01351023e-01
3.61173719e-01 -1.44621861e+00 7.43144333e-01 5.71443498e-01
9.86425877e-01 -2.28659630e-01 4.41904813e-01 5.14922380e-01
6.76945508e-01 -8.65946651e-01 6.07095435e-02 3.60891670e-01
5.25283635e-01 -6.17627740e-01 6.13531590e-01 -4.49470691e-02
-8.81911576e-01 -2.80619524e-02 -9.23837602e-01 9.27236006e-02
3.22014362e-01 7.29965389e-01 -7.46117473e-01 2.88435578e-01
4.01761174e-01 9.50688004e-01 -1.63897604e-01 7.58632302e-01
-1.93538278e-01 9.52214122e-01 4.56828140e-02 -1.65405482e-01
6.60305798e-01 -1.96982801e-01 6.38435721e-01 1.25487590e+00
5.47672510e-01 -1.32604226e-01 1.14548020e-01 1.17457199e+00
-2.24833906e-01 1.76259875e-01 -7.84606576e-01 -5.54127574e-01
6.34820700e-01 1.18988359e+00 -4.83915955e-01 -6.54398382e-01
-4.29686934e-01 6.06129944e-01 2.18691945e-01 6.66627109e-01
-4.70944166e-01 -4.13873345e-01 1.15886605e+00 1.10298254e-01
4.36711341e-01 -4.27724332e-01 -8.17231536e-02 -1.15720809e+00
-3.98083806e-01 -3.80551726e-01 2.49419495e-01 -7.61303842e-01
-1.59165394e+00 6.08881652e-01 4.15756613e-01 -9.62261260e-01
-4.44525540e-01 -5.98958194e-01 -1.00547457e+00 9.34273779e-01
-1.48027980e+00 -1.20002627e+00 1.97366998e-01 2.24193688e-02
9.85400975e-01 -6.44980252e-01 1.16558826e+00 -1.18318848e-01
-3.26851547e-01 3.67763519e-01 7.68552542e-01 -7.99770430e-02
7.22083986e-01 -1.27071941e+00 5.33542573e-01 2.30021939e-01
3.01223963e-01 1.13762999e+00 5.69818318e-01 -5.80310762e-01
-9.38910127e-01 -1.03945351e+00 1.16871190e+00 -2.77801901e-01
1.04397619e+00 -4.92388815e-01 -1.20870304e+00 9.60979283e-01
6.82055175e-01 -4.28361803e-01 1.18924677e+00 9.77575719e-01
-6.97956264e-01 3.52794468e-01 -4.52067584e-01 5.71644902e-01
4.03273463e-01 -6.41066074e-01 -1.19962060e+00 6.17121935e-01
1.50438225e+00 4.74541873e-01 -8.09569895e-01 -3.50086421e-01
5.66872120e-01 -1.41051710e-01 1.22648501e+00 -1.15256345e+00
8.39744508e-01 8.64358023e-02 -4.11865890e-01 -1.53757942e+00
-8.66521597e-01 -2.90653795e-01 -6.79147363e-01 1.45846868e+00
1.19448416e-01 -6.31325364e-01 7.05704570e-01 3.67040753e-01
-4.13574018e-02 -7.95187831e-01 -7.83735335e-01 -7.38291681e-01
4.83842760e-01 -9.45466086e-02 6.23915374e-01 1.28702259e+00
2.62152404e-01 4.69046295e-01 -4.90434945e-01 1.06286712e-01
5.79680383e-01 9.86605734e-02 5.92293799e-01 -1.68404317e+00
-2.81735528e-02 -5.66592932e-01 -3.54220271e-01 -1.28185046e+00
6.40349567e-01 -8.64881277e-01 -3.31454396e-01 -1.88926351e+00
5.54166973e-01 -3.54103088e-01 -4.63924915e-01 4.16885287e-01
-3.07287425e-01 -3.07357430e-01 -4.46790084e-02 6.03957713e-01
-5.42070746e-01 1.13932371e+00 8.70844245e-01 -3.34737211e-01
-5.99729866e-02 -4.26500112e-01 -8.86169791e-01 6.97830558e-01
4.22102064e-01 -6.36478901e-01 -5.36975861e-01 -5.76134503e-01
2.33802274e-02 -3.74280155e-01 1.32398270e-02 -2.84986466e-01
1.82985887e-01 -1.57845132e-02 7.33589828e-02 -1.01291776e+00
5.09124935e-01 -6.00117624e-01 -4.46521848e-01 1.68095171e-01
-8.37009609e-01 -1.24458130e-02 1.34709075e-01 1.05235803e+00
-3.58753622e-01 -3.22553337e-01 1.02373682e-01 6.47335872e-02
-4.69490081e-01 3.40738922e-01 -6.88583136e-01 -2.43272528e-01
7.11753488e-01 -4.60077263e-02 -4.62032288e-01 -6.08795524e-01
-5.58803618e-01 3.02236706e-01 6.94795996e-02 7.23446548e-01
6.61384046e-01 -1.61900508e+00 -6.71194434e-01 4.77178507e-02
2.79586524e-01 -2.34680504e-01 1.89552888e-01 2.57439524e-01
8.85177702e-02 9.07147706e-01 8.05196017e-02 -4.12437201e-01
-9.91176963e-01 7.41406500e-01 -3.65998834e-01 -5.53422868e-01
-7.24544346e-01 7.14870274e-01 9.27422404e-01 -4.43594724e-01
3.55817556e-01 -2.65584737e-01 -5.82801580e-01 2.98336446e-01
6.16908193e-01 4.09112513e-01 -5.60662866e-01 -3.05808365e-01
-9.17101130e-02 5.81610084e-01 -6.66403055e-01 -1.07635930e-01
1.53063071e+00 -1.42013341e-01 -2.79671669e-01 7.10317969e-01
1.53234315e+00 -2.65866667e-01 -8.00533295e-01 -5.92522800e-01
-4.16818820e-02 -3.93781781e-01 4.08129454e-01 -1.18404992e-01
-6.95472717e-01 1.40700817e+00 4.61492777e-01 3.66081238e-01
4.23094004e-01 4.31715459e-01 7.90763021e-01 3.07561010e-01
-1.67196214e-01 -8.23324263e-01 4.79096860e-01 4.15390432e-01
8.28954101e-01 -8.10828686e-01 1.92593038e-01 -3.67655635e-01
-4.54155892e-01 1.24354208e+00 3.26677233e-01 -3.39655131e-01
9.53426182e-01 -1.70262307e-01 -2.58538932e-01 -5.72071671e-01
-1.19737577e+00 2.37854257e-01 5.53232133e-01 6.38404012e-01
5.55213690e-01 1.68725744e-01 -4.38183755e-01 8.17740560e-01
-4.51538339e-02 -3.84505153e-01 4.15189564e-01 4.98064429e-01
-7.59547055e-01 -9.65376735e-01 -4.49292749e-01 4.71662432e-01
-2.96180010e-01 -3.25700939e-01 -8.84756520e-02 8.30775857e-01
-3.36720496e-01 4.86331701e-01 6.76158547e-01 -1.56849489e-01
-2.98601002e-01 4.52070236e-01 -3.38022947e-01 -9.58457470e-01
3.74351279e-03 3.54618698e-01 -2.36910567e-01 -6.30735382e-02
-7.47488886e-02 -4.35455978e-01 -8.26205134e-01 -1.58852294e-01
-6.78772926e-01 6.46027565e-01 9.22436774e-01 8.40184808e-01
6.81165874e-01 8.33869278e-01 5.41642010e-01 -6.98827326e-01
-6.29443228e-01 -1.41844475e+00 -9.13028359e-01 1.43786892e-01
-3.76962200e-02 -8.89317870e-01 -4.10414815e-01 3.32731903e-02]
|
[10.455467224121094, 6.980878829956055]
|
9a77ef92-3c1f-4331-aeda-b1b73f6210be
|
the-2023-video-similarity-dataset-and
|
2306.09489
| null |
https://arxiv.org/abs/2306.09489v1
|
https://arxiv.org/pdf/2306.09489v1.pdf
|
The 2023 Video Similarity Dataset and Challenge
|
This work introduces a dataset, benchmark, and challenge for the problem of video copy detection and localization. The problem comprises two distinct but related tasks: determining whether a query video shares content with a reference video ("detection"), and additionally temporally localizing the shared content within each video ("localization"). The benchmark is designed to evaluate methods on these two tasks, and simulates a realistic needle-in-haystack setting, where the majority of both query and reference videos are "distractors" containing no copied content. We propose a metric that reflects both detection and localization accuracy. The associated challenge consists of two corresponding tracks, each with restrictions that reflect real-world settings. We provide implementation code for evaluation and baselines. We also analyze the results and methods of the top submissions to the challenge. The dataset, baseline methods and evaluation code is publicly available and will be discussed at a dedicated CVPR'23 workshop.
|
['Matthijs Douze', 'Giorgos Tolias', 'Symeon Papadopoulos', 'Akshay Gupta', 'Sugosh Nagavara Ravindra', 'Gheorghe Postelnicu', 'Hiral Patel', 'Giorgos Kordopatis-Zilos', 'Ed Pizzi']
|
2023-06-15
| null | null | null | null |
['video-similarity']
|
['computer-vision']
|
[ 3.67102474e-01 -6.44974113e-01 -4.56912458e-01 -2.89389081e-02
-1.28345168e+00 -9.62397099e-01 5.62102854e-01 8.89296383e-02
-3.77168775e-01 2.66429245e-01 1.41991496e-01 1.79809723e-02
1.56805247e-01 -1.57572329e-01 -1.04555500e+00 -6.21952534e-01
-6.17981195e-01 1.67397812e-01 6.20971680e-01 2.65940934e-01
5.50882578e-01 2.65343189e-01 -1.51831508e+00 6.63774669e-01
4.44809236e-02 1.03564358e+00 4.40573305e-01 1.11663365e+00
3.79677266e-01 1.01412785e+00 -7.02172399e-01 -3.82351100e-01
4.10720974e-01 -2.92116970e-01 -8.53965044e-01 2.46088058e-01
1.06879663e+00 -3.34933728e-01 -7.37107694e-01 1.19948769e+00
3.60204458e-01 2.99724042e-01 4.08344418e-01 -1.41229641e+00
-4.78765249e-01 3.14398646e-01 -8.37883592e-01 9.77157533e-01
8.25204492e-01 9.85998809e-02 9.99427080e-01 -8.77453446e-01
1.14177632e+00 9.72182512e-01 6.95245802e-01 1.96925595e-01
-1.03506160e+00 -3.45445991e-01 1.46426484e-01 3.94289374e-01
-1.63594222e+00 -6.54614329e-01 1.76080123e-01 -6.97873771e-01
7.24616826e-01 4.14444059e-01 1.97414339e-01 1.18144345e+00
1.26167223e-01 9.45545316e-01 6.25282049e-01 -1.71846539e-01
1.78231373e-02 1.27452284e-01 6.33066073e-02 7.63651550e-01
6.02792799e-02 4.73758951e-02 -7.10156500e-01 -3.47773165e-01
5.00168145e-01 -7.22411126e-02 -6.32787824e-01 -6.10138893e-01
-1.52137554e+00 4.34745997e-01 1.10520780e-01 1.68258309e-01
-1.62748024e-01 2.20502183e-01 4.78484958e-01 5.00316799e-01
2.74270505e-01 3.49035233e-01 -1.39682800e-01 -2.55652964e-01
-1.27326739e+00 6.60749972e-01 9.61498260e-01 1.48769724e+00
5.66479385e-01 -6.47276402e-01 -5.90682745e-01 6.24196768e-01
-1.02450632e-01 3.57571125e-01 2.80200779e-01 -1.46414816e+00
5.56165516e-01 -5.67431934e-02 3.39038879e-01 -1.19561410e+00
1.12343736e-01 -4.19423908e-01 -1.90708220e-01 -6.63202554e-02
4.25878048e-01 4.34326008e-02 -7.13602602e-01 1.58089066e+00
1.35097623e-01 7.23030210e-01 -3.35638702e-01 1.12153363e+00
7.03513145e-01 6.82386935e-01 -2.78976858e-01 -2.31591731e-01
1.58938229e+00 -1.38283932e+00 -4.86342341e-01 2.11356450e-02
7.27722406e-01 -1.12132204e+00 8.10908973e-01 3.78799111e-01
-1.28078222e+00 -3.45506281e-01 -8.95112753e-01 -1.23415954e-01
-2.79506624e-01 1.96021244e-01 3.03015625e-03 3.30592901e-01
-1.31497848e+00 5.44316947e-01 -6.06222332e-01 -6.59555852e-01
1.17063917e-01 -3.68983261e-02 -3.79667014e-01 -3.09317172e-01
-8.24868977e-01 6.00292563e-01 3.02944481e-02 -3.94265383e-01
-1.23152947e+00 -6.90914989e-01 -5.72059572e-01 -1.04282245e-01
8.25739801e-01 -4.38657582e-01 1.73097372e+00 -9.34363484e-01
-7.07303405e-01 1.41155410e+00 -3.14858794e-01 -4.41272259e-01
6.83131039e-01 -3.32144648e-01 -4.43347096e-01 5.60303271e-01
5.29922664e-01 5.17457366e-01 8.43701422e-01 -1.07040715e+00
-1.03271544e+00 -2.01653734e-01 1.01139233e-01 2.37672701e-01
1.87924057e-01 5.97969532e-01 -1.53134882e+00 -8.99100721e-01
-2.64697909e-01 -1.00820696e+00 8.67160335e-02 2.90275682e-02
-4.94785815e-01 2.99935639e-02 8.24787915e-01 -5.52362978e-01
1.28720927e+00 -2.48789191e+00 1.01192437e-01 8.80412534e-02
3.29961032e-01 -5.74862100e-02 -3.69804502e-01 5.78301430e-01
6.99337423e-02 1.41434237e-01 1.63176224e-01 -3.64703238e-01
-1.37261435e-01 -2.89583713e-01 -5.05120933e-01 7.98294485e-01
-9.96787176e-02 5.57984352e-01 -9.58497465e-01 -4.79550958e-01
-1.84423834e-01 1.13163643e-01 -4.48454022e-01 3.40383708e-01
-3.91635150e-01 1.39362469e-01 -1.11931525e-01 7.59049833e-01
6.16923153e-01 -5.01157641e-01 6.30395859e-02 -2.43182838e-01
-1.21555671e-01 1.60824433e-01 -1.17835391e+00 1.79111326e+00
1.66487783e-01 9.12964523e-01 4.20821339e-01 -5.83785951e-01
2.97961295e-01 1.91730514e-01 3.96429628e-01 -5.54935932e-01
-2.23315209e-01 2.88175754e-02 -4.29021269e-01 -6.83649719e-01
9.14176524e-01 6.88510954e-01 3.98109183e-02 4.49936628e-01
9.23221931e-02 3.38313073e-01 7.77123988e-01 6.94455087e-01
1.79253066e+00 3.03917199e-01 3.67314927e-02 -9.12362337e-02
2.79164255e-01 1.29422128e-01 3.45801324e-01 1.18661726e+00
-4.60126609e-01 8.49094152e-01 7.40470350e-01 -2.08165109e-01
-9.76850152e-01 -1.27875519e+00 1.16718799e-01 1.43829846e+00
7.34936595e-01 -8.93930674e-01 -7.22086966e-01 -6.59424186e-01
4.01798636e-02 2.26896048e-01 -6.40161932e-01 2.97254026e-01
-6.18306875e-01 -2.17745930e-01 4.76898611e-01 2.73658216e-01
1.96435347e-01 -8.08241010e-01 -4.31369990e-01 -2.03845844e-01
-5.87979019e-01 -1.35919249e+00 -1.13096356e+00 -1.54640034e-01
-4.20845479e-01 -1.52272391e+00 -8.04292381e-01 -8.64990771e-01
3.55353296e-01 9.57121968e-01 1.46311188e+00 3.62532049e-01
-5.58172584e-01 7.51898170e-01 -4.16207850e-01 6.78669587e-02
-2.38975361e-01 -1.18135689e-02 -2.11695418e-01 -7.14626238e-02
3.91396821e-01 -3.98795530e-02 -7.10732281e-01 6.16827130e-01
-9.14057672e-01 -3.70501608e-01 3.55483443e-01 3.83375943e-01
9.20039892e-01 -1.24385662e-01 -1.62620023e-01 -9.09031153e-01
4.05995727e-01 -8.18374813e-01 -5.77728927e-01 3.00017297e-01
-1.04705669e-01 -3.94606560e-01 1.41626343e-01 -2.17358187e-01
-5.56492865e-01 4.51919399e-02 2.49201983e-01 -9.80218768e-01
-1.42675638e-01 1.74243972e-01 2.40814731e-01 -1.15526281e-01
4.95469123e-01 3.20931137e-01 -1.82184458e-01 -5.89093328e-01
1.67141244e-01 2.98658520e-01 1.06077611e+00 -5.89393020e-01
6.69149935e-01 5.24091959e-01 -3.79584551e-01 -7.25641429e-01
-6.74706519e-01 -1.16559076e+00 -3.37292492e-01 -1.86970055e-01
6.86486602e-01 -1.13464212e+00 -5.61855257e-01 2.58429974e-01
-1.21700799e+00 -2.50083804e-01 -7.82813430e-02 4.09474850e-01
-7.85359025e-01 7.17449546e-01 -7.38537192e-01 -4.42916721e-01
8.57269615e-02 -1.26695430e+00 1.31921148e+00 -7.23296851e-02
-6.46342933e-02 -5.51004708e-01 3.54836911e-01 3.18341285e-01
3.00804943e-01 1.54189929e-01 3.16729218e-01 -6.82116330e-01
-1.18529713e+00 -3.41581643e-01 -4.22804177e-01 2.39361711e-02
-3.09647381e-01 1.16090141e-01 -7.04710841e-01 -7.25711584e-01
-9.55527350e-02 -4.58503872e-01 1.09489512e+00 2.03819066e-01
1.28344762e+00 -2.40541011e-01 -7.48060107e-01 6.44425809e-01
1.49743831e+00 -4.14516628e-02 5.76917589e-01 4.08504963e-01
2.08931580e-01 4.26738232e-01 7.50200152e-01 3.40023577e-01
2.38705263e-01 9.76848364e-01 3.95727098e-01 1.55053824e-01
-2.87129968e-01 -6.43271729e-02 5.86635351e-01 4.19697106e-01
9.36753601e-02 -7.42456675e-01 -7.03410089e-01 7.15596139e-01
-1.87687385e+00 -1.46065593e+00 -1.06337667e-01 2.38185406e+00
3.86407614e-01 -6.48942366e-02 5.31209171e-01 -4.74604577e-01
1.07617664e+00 3.48939508e-01 -3.12062293e-01 1.59033552e-01
-1.18943505e-01 -1.48482069e-01 5.87741256e-01 2.07979470e-01
-1.59729004e+00 5.35652697e-01 7.45907974e+00 6.73027575e-01
-8.31079721e-01 2.05595091e-01 3.80837202e-01 -3.92049074e-01
1.37964606e-01 -8.80998671e-02 -9.30148304e-01 8.83196771e-01
9.50319231e-01 -1.27501383e-01 4.57523078e-01 7.45196939e-01
8.70821550e-02 -3.68274629e-01 -1.45948589e+00 9.48577821e-01
4.90524054e-01 -1.46807146e+00 -1.51998594e-01 -3.52020301e-02
6.31994486e-01 6.04044914e-01 1.89317346e-01 2.24097222e-01
1.97810262e-01 -7.39207745e-01 8.37224245e-01 4.66159046e-01
7.70559788e-01 -2.16290981e-01 5.43842316e-01 2.01274864e-02
-1.25315416e+00 1.06439017e-01 -2.63497382e-01 3.73574644e-01
1.52864560e-01 1.08309790e-01 -3.32959294e-01 4.50327784e-01
1.17126656e+00 8.65609944e-01 -7.33857632e-01 1.59448421e+00
1.98233873e-01 4.10688341e-01 -1.82817087e-01 2.71730572e-01
9.02669802e-02 1.95849203e-02 8.17722082e-01 1.80463326e+00
3.21329564e-01 -2.30292365e-01 5.32076657e-01 7.06907868e-01
-5.11784315e-01 9.37113762e-02 -6.29083395e-01 -5.96330781e-03
6.90679073e-01 1.06771171e+00 -7.02120662e-01 -4.09928709e-01
-6.29232645e-01 1.11748290e+00 2.24151000e-01 4.18036729e-01
-9.96171594e-01 -6.15750968e-01 7.26489007e-01 3.39517504e-01
7.68122256e-01 7.45542394e-03 5.24458408e-01 -1.41942775e+00
3.12115937e-01 -9.55713987e-01 6.34424925e-01 -8.03662121e-01
-1.28169560e+00 3.75899225e-01 5.51142730e-02 -1.31104052e+00
-1.09216027e-01 -4.15889025e-01 -5.60017288e-01 5.12513518e-01
-1.29023182e+00 -6.05355620e-01 -6.10727549e-01 5.83237290e-01
5.40177286e-01 -2.05752820e-01 4.19376850e-01 5.79962373e-01
-5.83605409e-01 7.52616704e-01 4.12835330e-01 2.73056120e-01
1.13528621e+00 -1.16215241e+00 4.24086392e-01 1.10104811e+00
1.37353137e-01 4.53828126e-01 8.45230341e-01 -5.45754969e-01
-1.60446727e+00 -1.32421756e+00 6.12864673e-01 -6.39934540e-01
9.13604438e-01 -4.05062675e-01 -8.24519873e-01 9.42621291e-01
4.05349076e-01 2.13012457e-01 4.28464949e-01 -1.84375197e-01
-6.65799320e-01 6.33640215e-02 -8.90671432e-01 2.86324650e-01
1.18049192e+00 -6.13353491e-01 -3.63290101e-01 8.13265979e-01
7.04532981e-01 -7.41053998e-01 -6.96064949e-01 1.81370810e-01
4.60727841e-01 -1.27920020e+00 9.86203969e-01 -4.84963119e-01
4.67185825e-01 -5.15305400e-01 -3.71709704e-01 -7.08917379e-01
-3.46905261e-01 -7.53507435e-01 -3.39927316e-01 8.97812605e-01
2.07049578e-01 9.06005353e-02 8.86707425e-01 2.97914241e-02
6.33795634e-02 -5.32712102e-01 -7.06242859e-01 -1.04613340e+00
-4.74980682e-01 -2.10824311e-01 2.06008144e-02 8.16786528e-01
-2.75746673e-01 1.02768637e-01 -3.88850808e-01 2.57693768e-01
8.18670392e-01 2.08646253e-01 9.63311911e-01 -5.49035192e-01
-6.06646895e-01 -4.22047168e-01 -7.07025290e-01 -1.48383439e+00
-2.22796097e-01 -7.99414039e-01 1.89252466e-01 -1.08479285e+00
6.34931684e-01 -7.63129964e-02 -3.09047341e-01 4.37378287e-02
2.57997531e-02 5.40377975e-01 3.56620252e-01 7.28408754e-01
-1.52061474e+00 -2.04911798e-01 6.67937279e-01 -1.49883196e-01
1.54998571e-01 8.98454040e-02 -3.49223256e-01 3.49220634e-01
4.02231216e-01 -4.64921296e-01 -2.82694608e-01 -5.11884332e-01
-1.09542264e-02 1.53567120e-01 6.09446466e-01 -1.15379751e+00
4.57240194e-01 5.39974049e-02 2.10414588e-01 -8.36885333e-01
4.59016353e-01 -5.77164114e-01 1.10353783e-01 2.29934648e-01
-6.67666912e-01 5.16440153e-01 5.25816344e-02 1.03215289e+00
-3.15782338e-01 -3.05416942e-01 7.13716686e-01 -2.99534321e-01
-9.50027287e-01 3.55545700e-01 -3.34675819e-01 3.72435391e-01
1.44523191e+00 -1.54456183e-01 -6.77169859e-01 -5.39872885e-01
-7.52767563e-01 4.83702958e-01 9.55447793e-01 5.32185614e-01
5.47897637e-01 -1.16645610e+00 -7.87603974e-01 -6.33126944e-02
3.41241181e-01 -5.03115475e-01 1.61630958e-01 1.07871807e+00
-7.97664523e-01 4.26334471e-01 6.53874204e-02 -7.30406582e-01
-1.53430510e+00 1.12590897e+00 2.93197036e-01 -1.29315972e-01
-4.53209370e-01 9.35477674e-01 2.63985902e-01 4.55831513e-02
8.39875698e-01 2.87391879e-02 1.33851171e-01 -2.58143485e-01
9.60138738e-01 5.19861519e-01 4.07483848e-03 -5.91895223e-01
-4.74493474e-01 2.12806135e-01 -4.63673115e-01 6.84193447e-02
9.16473567e-01 -2.57097989e-01 -1.35051653e-01 3.07329476e-01
1.45728421e+00 2.28211135e-01 -8.62654209e-01 -6.18796706e-01
2.10881874e-01 -8.32367003e-01 -1.44379646e-01 -6.41618788e-01
-9.66736376e-01 5.57054639e-01 5.93422949e-01 3.04659069e-01
1.02715826e+00 2.22006992e-01 7.29062855e-01 3.63045752e-01
3.03918988e-01 -8.58430266e-01 1.37608618e-01 3.50024134e-01
7.99372613e-01 -9.70013320e-01 5.81429377e-02 -4.75803435e-01
-4.82536197e-01 8.88401806e-01 6.83623612e-01 -2.95489758e-01
4.92730409e-01 4.66491371e-01 -1.51480421e-01 -2.87539959e-01
-1.07739019e+00 -4.21003550e-02 2.22538948e-01 3.71369749e-01
4.85170156e-01 -3.17583323e-01 8.74306709e-02 2.76859459e-02
2.52171606e-01 3.99083123e-02 6.30763590e-01 1.07756710e+00
-4.64856982e-01 -8.13801587e-01 -4.51430917e-01 5.35498619e-01
-7.52803266e-01 -4.43738997e-02 -5.11325836e-01 6.84920192e-01
-4.60532568e-02 8.39076757e-01 3.98840934e-01 -5.64297378e-01
2.38726735e-01 -3.47148180e-01 5.29077947e-01 -6.08612001e-01
-4.75538462e-01 2.74738461e-01 1.49758726e-01 -1.17274821e+00
-5.55949569e-01 -7.64950871e-01 -6.95583045e-01 -6.63344145e-01
-1.13283224e-01 2.38329321e-01 4.01604295e-01 4.54603642e-01
6.49867058e-01 1.29368663e-01 5.57534456e-01 -7.66993701e-01
-6.62917256e-01 -6.16649866e-01 -6.21510446e-01 5.72194993e-01
5.46731591e-01 -4.31024939e-01 -3.19323868e-01 2.71748364e-01]
|
[10.149237632751465, 0.588818371295929]
|
0ef22a8d-449b-4cd4-84bc-1b4775310762
|
equivariant-muzero
|
2302.04798
| null |
https://arxiv.org/abs/2302.04798v1
|
https://arxiv.org/pdf/2302.04798v1.pdf
|
Equivariant MuZero
|
Deep reinforcement learning repeatedly succeeds in closed, well-defined domains such as games (Chess, Go, StarCraft). The next frontier is real-world scenarios, where setups are numerous and varied. For this, agents need to learn the underlying rules governing the environment, so as to robustly generalise to conditions that differ from those they were trained on. Model-based reinforcement learning algorithms, such as the highly successful MuZero, aim to accomplish this by learning a world model. However, leveraging a world model has not consistently shown greater generalisation capabilities compared to model-free alternatives. In this work, we propose improving the data efficiency and generalisation capabilities of MuZero by explicitly incorporating the symmetries of the environment in its world-model architecture. We prove that, so long as the neural networks used by MuZero are equivariant to a particular symmetry group acting on the environment, the entirety of MuZero's action-selection algorithm will also be equivariant to that group. We evaluate Equivariant MuZero on procedurally-generated MiniPacman and on Chaser from the ProcGen suite: training on a set of mazes, and then testing on unseen rotated versions, demonstrating the benefits of equivariance. Further, we verify that our performance improvements hold even when only some of the components of Equivariant MuZero obey strict equivariance, which highlights the robustness of our construction.
|
['George Papamakarios', 'Théophane Weber', 'Andreea Deac']
|
2023-02-09
| null | null | null | null |
['starcraft']
|
['playing-games']
|
[-6.14836998e-02 -1.05386212e-01 8.21942762e-02 1.37574494e-01
-1.98234230e-01 -1.03237975e+00 7.89005518e-01 -1.14408396e-02
-5.68969965e-01 8.07797551e-01 1.57552540e-01 -1.41324580e-01
-5.09678543e-01 -9.91268575e-01 -9.58979487e-01 -7.82876134e-01
-5.06423414e-01 5.63404739e-01 2.18095616e-01 -9.33013201e-01
3.85556757e-01 4.70634580e-01 -1.42307997e+00 -3.53398845e-02
4.57113475e-01 2.66363829e-01 1.18489660e-01 8.25733542e-01
4.88591611e-01 6.60816610e-01 -4.53720629e-01 -7.19581023e-02
7.10318863e-01 -4.96960282e-01 -7.96085596e-01 -7.74688423e-02
1.66369498e-01 -2.09655970e-01 -4.08129752e-01 8.72017205e-01
2.10159719e-01 5.18305779e-01 5.73238969e-01 -1.09422064e+00
-1.69082329e-01 4.68042493e-01 -7.97718465e-02 -4.36175726e-02
2.26505339e-01 5.90174198e-01 1.09037352e+00 -1.38219297e-01
8.93431365e-01 9.75719452e-01 5.74769139e-01 5.99592209e-01
-1.53400230e+00 -5.30831575e-01 1.11519657e-01 -4.52288277e-02
-1.20631623e+00 -3.80187720e-01 5.23907125e-01 -2.56661266e-01
1.04770041e+00 2.62986988e-01 8.62669408e-01 1.24603415e+00
6.27584159e-01 3.11193824e-01 1.13149583e+00 -3.30065578e-01
6.69056594e-01 -3.67530704e-01 -3.22939038e-01 8.36473167e-01
2.73807406e-01 5.70988894e-01 -4.31350499e-01 -1.08249329e-01
1.04372549e+00 -4.13888663e-01 -1.15832724e-01 -1.20363295e+00
-1.47544992e+00 8.63151073e-01 4.44655001e-01 3.80806476e-02
-2.11375728e-01 3.23142678e-01 3.04503322e-01 5.52388310e-01
-3.65762442e-01 1.35022736e+00 -4.59085256e-01 -2.70430684e-01
-3.97656620e-01 7.16102362e-01 9.09640729e-01 7.49936342e-01
6.87198520e-01 2.35510901e-01 2.42810249e-01 4.57506597e-01
-1.47632891e-02 3.02466273e-01 7.50746191e-01 -1.27443480e+00
4.14658278e-01 3.28459144e-01 3.70739251e-02 -9.46982265e-01
-7.06588626e-01 -7.18516827e-01 -5.47957718e-01 7.99649537e-01
5.17133832e-01 -2.72276670e-01 -7.76669741e-01 2.36320066e+00
2.18563527e-01 7.10316375e-02 4.24889296e-01 7.32919097e-01
8.94478559e-02 4.31191146e-01 -1.03294417e-01 2.51059175e-01
7.38465667e-01 -5.65688431e-01 3.05647645e-02 -4.40992355e-01
1.03853881e+00 -2.40767106e-01 1.14183748e+00 4.92027014e-01
-1.08400023e+00 -2.16970816e-01 -1.36704457e+00 3.10189605e-01
-2.08995327e-01 -6.10932231e-01 9.32952166e-01 4.93069440e-01
-1.01355183e+00 8.23186874e-01 -8.82000268e-01 -3.71535033e-01
3.47746350e-02 7.38163948e-01 -8.72581422e-01 3.00638616e-01
-1.22373521e+00 1.15810251e+00 8.75260532e-01 -1.07530698e-01
-1.41399956e+00 -3.20311189e-01 -9.22979116e-01 1.50722772e-01
6.24355614e-01 -7.62701035e-01 1.38513052e+00 -1.01908159e+00
-1.76303649e+00 5.20700991e-01 4.17026311e-01 -6.26068532e-01
5.41706204e-01 1.79435790e-01 -5.51560968e-02 -1.18784957e-01
-4.40820754e-02 5.45165718e-01 6.51548505e-01 -1.19814050e+00
-5.32797694e-01 -2.54995346e-01 8.28364670e-01 5.40646255e-01
8.30041170e-02 -3.67008656e-01 2.79959831e-02 -4.61359441e-01
1.61896989e-01 -1.29711783e+00 -5.79516530e-01 -3.45208436e-01
-2.00799435e-01 1.26710951e-01 3.41618747e-01 -2.14518398e-01
7.52220690e-01 -2.16888094e+00 5.55617869e-01 5.74836552e-01
1.94360688e-01 6.90399203e-03 -4.42904383e-01 6.26589715e-01
-1.99851394e-01 6.90736026e-02 -2.91897625e-01 2.18587920e-01
2.83614188e-01 6.39455020e-01 -3.00687462e-01 6.31494701e-01
-2.00324282e-02 8.33972096e-01 -8.64305794e-01 -9.82685611e-02
1.29295185e-01 -5.62784411e-02 -1.16204000e+00 -4.71054055e-02
-4.46505129e-01 4.69149172e-01 -3.25387806e-01 -1.10113412e-01
1.11491784e-01 6.95823431e-02 3.69610041e-01 4.64197963e-01
-1.15185924e-01 2.47146398e-01 -1.50750077e+00 1.80071175e+00
-6.39609993e-01 2.92654365e-01 -1.67177338e-02 -9.70910072e-01
5.02890289e-01 3.60940695e-02 3.23709488e-01 -8.57714057e-01
9.66279879e-02 2.11893499e-01 7.69953072e-01 -3.52350026e-01
4.86563146e-01 -3.08298767e-01 -2.68190563e-01 4.33603942e-01
3.17439675e-01 -4.25696492e-01 3.54041308e-01 1.74144059e-01
1.28974807e+00 4.17014778e-01 3.85641396e-01 -4.82608914e-01
7.02756941e-02 1.82306603e-01 5.80370486e-01 1.13040137e+00
-6.10049032e-02 3.96516979e-01 7.93923855e-01 -4.40188199e-01
-1.31325233e+00 -1.19953561e+00 -1.38155118e-01 9.54556763e-01
2.00328216e-01 -5.53694487e-01 -9.25950348e-01 -6.41553581e-01
-2.25823849e-01 7.47817993e-01 -9.50262308e-01 -4.94985580e-01
-7.45981216e-01 -6.30007029e-01 6.48172319e-01 1.83762163e-01
6.14741325e-01 -1.17893088e+00 -1.05716848e+00 3.53183806e-01
2.54809737e-01 -7.14215219e-01 -2.79987883e-02 6.40751302e-01
-7.25192249e-01 -1.25759125e+00 -1.11291319e-01 -5.56410074e-01
4.31720883e-01 -4.24866825e-02 9.77766693e-01 -1.64698362e-02
-3.60002965e-02 5.55065870e-01 -1.69119194e-01 -1.75321758e-01
-4.46264476e-01 3.47712457e-01 2.58296698e-01 -4.20732558e-01
-2.81703740e-01 -8.64235222e-01 -2.75858849e-01 3.26649696e-01
-1.06743789e+00 2.76411802e-01 3.93615365e-01 1.03610742e+00
2.50218749e-01 3.59525114e-01 3.31352115e-01 -7.54248321e-01
5.03306389e-01 -4.83933508e-01 -8.04625809e-01 7.48170987e-02
-1.95173800e-01 5.77820480e-01 8.09549391e-01 -3.31748009e-01
-8.39443505e-01 -8.98567140e-02 -7.23579228e-02 1.91456705e-01
-1.86309353e-01 5.83505392e-01 -3.39498878e-01 -1.78912163e-01
1.09617937e+00 1.38074026e-01 -4.29615453e-02 -8.24853852e-02
2.75299609e-01 -8.35782140e-02 5.39555073e-01 -1.07872808e+00
1.05202675e+00 3.74976635e-01 3.55038941e-01 -7.53353536e-01
-2.76734143e-01 2.27560550e-01 -4.79410499e-01 5.00029624e-02
7.58939743e-01 -5.13049126e-01 -8.19863379e-01 4.25175220e-01
-7.73045957e-01 -9.75167990e-01 -3.98748040e-01 4.59561199e-01
-1.07265854e+00 -2.62726322e-02 -2.02929467e-01 -4.18804765e-01
4.10742581e-01 -1.30441713e+00 4.59282756e-01 7.74032110e-03
-3.62092704e-01 -1.09750366e+00 5.57203174e-01 -1.10473923e-01
5.01945257e-01 3.56817722e-01 1.20244551e+00 -8.04265738e-01
-4.65672970e-01 1.58773690e-01 4.76411939e-01 -9.77142993e-03
-6.62088767e-02 -3.65094934e-03 -6.33634090e-01 -4.74225670e-01
-8.56610015e-02 -4.15307254e-01 6.23896480e-01 -2.58373246e-02
1.02250540e+00 -2.86088914e-01 6.58337846e-02 8.27830732e-01
1.39900839e+00 2.52342761e-01 6.19507372e-01 1.01830804e+00
4.27210629e-01 6.01832449e-01 4.79594171e-02 2.46426657e-01
2.23778114e-01 7.79059172e-01 8.84692073e-01 1.34637281e-01
3.46354604e-01 -3.63146633e-01 5.20475149e-01 4.16504472e-01
-2.17162743e-01 -4.50484678e-02 -8.66582513e-01 3.73144746e-01
-1.88049948e+00 -1.19949400e+00 3.00058782e-01 2.22128129e+00
8.09200823e-01 4.04957026e-01 8.79679397e-02 4.08656001e-02
2.40099028e-01 1.78196430e-01 -6.17543936e-01 -6.60792351e-01
-2.83776224e-01 5.84865153e-01 5.72493136e-01 5.96351922e-01
-1.04867113e+00 1.05474997e+00 6.53448820e+00 6.44181550e-01
-1.14550698e+00 -2.33494997e-01 1.79002136e-01 -1.75297633e-01
-2.06833497e-01 2.62008399e-01 -3.68108898e-01 5.33747040e-02
8.35637391e-01 -1.78521052e-01 1.11835921e+00 8.82059753e-01
1.00599155e-01 -1.11871451e-01 -1.30226684e+00 4.77825165e-01
-2.24700928e-01 -1.35864317e+00 -1.37795433e-01 2.74205089e-01
7.87805617e-01 5.99762462e-02 1.55564755e-01 7.29183316e-01
9.75031316e-01 -1.30892467e+00 9.57312703e-01 1.29854858e-01
2.55126268e-01 -8.92679930e-01 3.65516782e-01 5.19610643e-01
-6.23941123e-01 -9.71609727e-02 -2.76624054e-01 -3.39177787e-01
-4.30819243e-01 -3.32696855e-01 -7.92192161e-01 4.56365436e-01
4.01055217e-01 3.21988106e-01 -6.32152259e-01 9.25298989e-01
-3.81615222e-01 4.75417465e-01 -3.64240855e-01 1.66400433e-01
6.21059775e-01 -2.55705804e-01 6.00642502e-01 8.97071600e-01
1.37092307e-01 2.59746999e-01 5.51660582e-02 6.40435994e-01
3.14930417e-02 -4.71332781e-02 -9.74616349e-01 1.14425816e-01
6.94673955e-02 9.97951508e-01 -6.64527178e-01 4.64477316e-02
-8.93334821e-02 6.43630624e-01 4.66891170e-01 5.47406316e-01
-8.16039622e-01 -2.63665348e-01 9.20830071e-01 -7.34112114e-02
3.24087471e-01 -6.13203585e-01 -1.64642707e-01 -1.35577917e+00
-2.03480378e-01 -1.41537714e+00 2.04177886e-01 -6.40584886e-01
-7.95138597e-01 3.03899288e-01 1.92006767e-01 -9.50571716e-01
-5.72563410e-01 -8.68279696e-01 -6.31018162e-01 6.19017780e-01
-9.17623222e-01 -8.95758212e-01 1.24823160e-01 7.78653383e-01
2.48216897e-01 -4.34113979e-01 1.14097095e+00 -3.20601195e-01
-4.77256089e-01 4.10964578e-01 2.70014226e-01 1.19924083e-01
3.89023274e-01 -1.35605955e+00 3.14317584e-01 8.70990932e-01
4.12520885e-01 1.03099656e+00 9.29562271e-01 -3.79089028e-01
-1.56030583e+00 -8.60200226e-01 6.86080940e-03 -5.26377916e-01
8.27986419e-01 -5.48016548e-01 -6.47107422e-01 1.01620603e+00
1.61385104e-01 -1.86086595e-01 2.86388069e-01 3.49481642e-01
-5.18419027e-01 9.36432332e-02 -9.46209788e-01 1.28577459e+00
1.31856310e+00 -3.10988337e-01 -7.07902730e-01 -4.32708338e-02
3.92311841e-01 -7.84180403e-01 -5.92706919e-01 2.93815762e-01
6.55176401e-01 -1.10312378e+00 8.52605164e-01 -1.29582024e+00
5.97905159e-01 -3.98461670e-01 -4.06396717e-01 -1.84228849e+00
-4.76741433e-01 -6.40897214e-01 2.45078057e-01 6.72429860e-01
3.82289767e-01 -7.83849239e-01 7.13176966e-01 5.10895491e-01
-1.27011999e-01 -4.04589027e-01 -9.80383217e-01 -8.49414110e-01
5.65645456e-01 -4.55952764e-01 7.90467381e-01 9.99552310e-01
1.33809492e-01 2.51382917e-01 -2.53850222e-01 2.26425469e-01
3.66581380e-01 -1.51653469e-01 1.11590743e+00 -9.97330785e-01
-8.31170321e-01 -4.96797144e-01 -5.56525767e-01 -5.82859635e-01
3.48352164e-01 -1.03671014e+00 -1.36745442e-02 -1.14796972e+00
2.12714449e-02 -7.37838447e-01 -2.58344859e-01 6.05676174e-01
2.86282152e-01 2.28833482e-02 3.31538796e-01 2.55343392e-02
-5.82600296e-01 6.47757530e-01 1.31876671e+00 -5.43434806e-02
-3.36544484e-01 -3.04757476e-01 -7.59459913e-01 8.92771006e-01
1.16951990e+00 -3.75450373e-01 -6.85190022e-01 -5.38998067e-01
7.22531140e-01 -5.99084496e-02 5.51384687e-01 -1.19802749e+00
7.76262581e-02 -7.04767525e-01 3.24340761e-01 4.78170514e-01
2.32576266e-01 -6.90065086e-01 3.02551329e-01 5.72327495e-01
-6.42667651e-01 3.09211195e-01 3.15698087e-01 3.44560027e-01
1.13247208e-01 -3.80680710e-01 7.32903719e-01 -2.99281061e-01
-8.32624435e-01 1.97069988e-01 -6.87794924e-01 4.77112442e-01
1.05920386e+00 -1.74080163e-01 -3.33608210e-01 -3.61817837e-01
-6.72028065e-01 1.23065263e-01 8.93181086e-01 2.81409532e-01
1.82623297e-01 -1.08822465e+00 -5.26148438e-01 4.07139838e-01
1.43899053e-01 7.34300986e-02 2.15680674e-02 5.65367818e-01
-7.35903323e-01 1.43962592e-01 -6.97489321e-01 -2.60482788e-01
-8.01591933e-01 4.66532588e-01 8.39906633e-01 -3.37486595e-01
-6.48958206e-01 3.73563617e-01 5.51896811e-01 -7.95747578e-01
-1.45559043e-01 -3.85262549e-01 -6.55840784e-02 -5.43966115e-01
1.37662753e-01 5.33231348e-02 1.49546131e-01 -4.64781016e-01
-1.45864308e-01 3.19540113e-01 9.51980129e-02 -6.71965420e-01
1.51552165e+00 4.50252324e-01 6.74288720e-02 7.87328631e-02
7.28286445e-01 2.08996937e-01 -1.46461916e+00 1.35266066e-01
-3.49348895e-02 -2.14513823e-01 -2.33018979e-01 -6.80378020e-01
-7.68408298e-01 6.28883779e-01 1.87106237e-01 1.20230116e-01
7.40771711e-01 -4.10974294e-01 6.26793280e-02 1.08494234e+00
7.52114713e-01 -1.16981828e+00 7.94885978e-02 9.65297461e-01
9.76150930e-01 -8.53066802e-01 -2.38722488e-01 3.53675246e-01
-6.14731014e-01 1.07372475e+00 7.05204725e-01 -5.68451941e-01
1.69327155e-01 1.21791802e-01 -1.80623129e-01 -6.69796169e-02
-9.18201447e-01 -5.52661493e-02 -1.91094249e-01 8.43248308e-01
-1.31456450e-01 9.64733772e-03 1.00645293e-02 2.47884750e-01
-8.53406787e-01 -4.53604847e-01 7.85891414e-01 1.01931953e+00
-4.23968852e-01 -1.32140410e+00 -3.37406605e-01 1.78696573e-01
-1.42233387e-01 8.02362040e-02 -2.75588036e-01 1.36949074e+00
1.12276725e-01 6.09584928e-01 -1.38369143e-01 -3.15276861e-01
3.63433123e-01 -5.56099229e-02 8.16177547e-01 -7.17123866e-01
-6.12124681e-01 -3.67873371e-01 1.61003143e-01 -7.33096600e-01
-1.35345971e-02 -7.11431384e-01 -1.31718445e+00 -4.46206897e-01
1.58427179e-01 2.08769739e-01 5.68139195e-01 1.02401710e+00
6.92274049e-02 4.75777745e-01 3.74658018e-01 -7.73548782e-01
-8.22310746e-01 -4.77963328e-01 -6.14608109e-01 4.71680909e-01
4.43985701e-01 -8.02554429e-01 -2.52577424e-01 -2.04726964e-01]
|
[4.005280017852783, 1.7001630067825317]
|
c555c592-0094-48f5-8159-0df1278ea8bc
|
generative-models-for-pose-transfer
|
1806.0907
| null |
http://arxiv.org/abs/1806.09070v1
|
http://arxiv.org/pdf/1806.09070v1.pdf
|
Generative Models for Pose Transfer
|
We investigate nearest neighbor and generative models for transferring pose
between persons. We take in a video of one person performing a sequence of
actions and attempt to generate a video of another person performing the same
actions. Our generative model (pix2pix) outperforms k-NN at both generating
corresponding frames and generalizing outside the demonstrated action set. Our
most salient contribution is determining a pipeline (pose detection, face
detection, k-NN based pairing) that is effective at perform-ing the desired
task. We also detail several iterative improvements and failure modes.
|
['Alexander Li', 'Gokul Swamy', 'Patrick Chao']
|
2018-06-24
| null | null | null | null |
['pose-transfer']
|
['computer-vision']
|
[ 3.63163352e-01 -5.26069291e-03 4.83169973e-01 -4.98463094e-01
-1.01711702e+00 -8.10935080e-01 8.31306875e-01 -6.01605654e-01
-1.27356812e-01 4.50243145e-01 4.67301875e-01 3.04070413e-01
2.80127227e-01 -4.81395423e-01 -7.50825286e-01 -4.24217910e-01
-1.56035826e-01 7.36989915e-01 1.97710693e-01 2.09351957e-01
2.17273846e-01 8.84689629e-01 -1.49708819e+00 5.61448455e-01
-7.09814951e-02 6.18414462e-01 -2.91156083e-01 1.30206060e+00
4.73641217e-01 5.80888629e-01 -8.76791060e-01 -7.71724999e-01
7.17565596e-01 -8.16319704e-01 -6.86698794e-01 3.09133172e-01
1.33288145e+00 -6.86593235e-01 -5.36260366e-01 7.12348521e-01
1.05213845e+00 3.24481964e-01 7.18882918e-01 -1.39359188e+00
-3.30755442e-01 3.63079965e-01 -3.88676107e-01 8.37793201e-02
1.27269423e+00 2.32366905e-01 1.77494675e-01 -1.02717400e+00
8.56170654e-01 1.52219784e+00 9.71873283e-01 1.04326785e+00
-1.10635686e+00 -4.66142416e-01 -2.47266665e-01 -2.40941659e-01
-1.69897318e+00 -9.89019156e-01 4.93385136e-01 -3.26202661e-01
1.02813137e+00 2.88577229e-01 9.54019010e-01 1.49837399e+00
1.50302630e-02 5.98994076e-01 7.56059766e-01 -3.76240194e-01
6.46075681e-02 -1.23730399e-01 -5.99872112e-01 7.51279235e-01
-2.96516240e-01 1.76963702e-01 -9.32993531e-01 -3.16214591e-01
1.21078372e+00 -1.11407869e-01 -1.65490195e-01 -3.50876540e-01
-1.35801351e+00 2.94450462e-01 1.10985480e-01 -9.74035934e-02
-4.46110278e-01 7.21208274e-01 2.02512696e-01 4.47661281e-02
9.56682563e-02 1.99150220e-01 -6.19298667e-02 -3.35730314e-01
-1.23475850e+00 6.42644346e-01 7.97061145e-01 1.35706258e+00
4.33577240e-01 -5.38486335e-03 -5.49009502e-01 1.85320526e-01
6.94858655e-02 5.20386338e-01 -4.43608314e-02 -1.49633884e+00
2.14364693e-01 2.04864055e-01 3.17244440e-01 -9.50250328e-01
3.43149640e-02 2.97365300e-02 -1.64645627e-01 2.73346961e-01
5.35934389e-01 -5.49990058e-01 -9.07710016e-01 1.46382225e+00
6.88976884e-01 6.03445649e-01 2.99764462e-02 8.61735880e-01
8.70420158e-01 3.81718129e-01 -2.86997873e-02 -7.33170733e-02
1.26310456e+00 -5.62720954e-01 -3.65789175e-01 -2.10348010e-01
1.72858760e-01 -1.08416271e+00 4.90610570e-01 2.62971252e-01
-1.50275075e+00 -7.50105262e-01 -5.63066840e-01 -6.58986717e-02
2.52388995e-02 2.90012985e-01 4.51349318e-01 7.22632408e-01
-1.37656498e+00 7.89186954e-01 -7.20027030e-01 -8.27901244e-01
3.86902779e-01 5.03336251e-01 -7.61243105e-01 1.56737179e-01
-7.61986077e-01 5.65639853e-01 1.72263175e-01 1.29723847e-01
-1.25717127e+00 -5.83703279e-01 -8.16364646e-01 -3.12718540e-01
2.29709759e-01 -1.07702351e+00 1.15935481e+00 -7.41336226e-01
-1.41858578e+00 1.08552265e+00 -4.87256311e-02 -3.91701519e-01
9.71715748e-01 -6.58966482e-01 -3.04262996e-01 4.02900070e-01
6.18879423e-02 1.48168540e+00 1.22753966e+00 -1.04702377e+00
-5.54514229e-01 -6.55292571e-01 9.15689319e-02 6.49604023e-01
-1.42490059e-01 3.97993237e-01 -1.01717532e+00 -6.63345039e-01
-5.44020683e-02 -1.11654460e+00 9.20272470e-02 1.50682151e-01
-7.10309386e-01 -2.14981079e-01 8.26967120e-01 -7.71944106e-01
9.28410590e-01 -1.98081088e+00 1.67057917e-01 2.15323552e-01
6.70220405e-02 1.51378587e-01 -1.35222569e-01 5.20159662e-01
-1.49110347e-01 -3.57857049e-01 3.16300064e-01 -7.98453391e-01
-8.81812498e-02 -1.22366309e-01 -7.31260106e-02 6.21128857e-01
2.15509813e-02 9.61835384e-01 -6.43450379e-01 -7.62517333e-01
3.35368305e-01 1.02254295e+00 -5.56512892e-01 4.30745006e-01
8.50764960e-02 4.51694965e-01 -6.79868683e-02 6.92250550e-01
3.55031878e-01 2.58781075e-01 1.79380879e-01 -5.66413820e-01
5.13259768e-02 -9.58137363e-02 -1.56352091e+00 1.78857279e+00
1.53638020e-01 5.59601724e-01 -8.89713168e-02 -6.48123100e-02
7.16713548e-01 4.20246750e-01 3.00142258e-01 2.58051544e-01
1.69600502e-01 -3.10507864e-01 -3.44455540e-01 -5.55344760e-01
2.90783495e-01 1.52183533e-01 7.28739947e-02 2.84963399e-01
1.93365097e-01 -1.69867322e-01 3.33559327e-02 2.75838047e-01
1.14848351e+00 7.56628752e-01 2.03141570e-01 4.91935834e-02
1.36358887e-01 -1.07476689e-01 3.29707891e-01 7.15024233e-01
-5.62313080e-01 1.14331627e+00 2.84430325e-01 -6.17704988e-01
-1.11125410e+00 -1.13760865e+00 2.62969047e-01 1.18135357e+00
-1.26867518e-01 -4.71773803e-01 -1.19431651e+00 -7.44699240e-01
-9.49569866e-02 4.60034013e-01 -8.10709774e-01 4.97151390e-02
-7.75783479e-01 -8.39719400e-02 8.46323907e-01 8.62572968e-01
5.32838464e-01 -9.65132833e-01 -6.63370252e-01 -2.64446497e-01
-5.12159131e-02 -9.43237603e-01 -1.08917475e+00 -3.53032857e-01
-4.98597175e-01 -1.10813296e+00 -8.48415852e-01 -8.67625296e-01
8.79642189e-01 9.19434130e-02 1.13056934e+00 -3.61288786e-01
-5.48478961e-01 1.05895758e+00 1.02572062e-03 -1.15419634e-01
-3.39146048e-01 -4.19071555e-01 3.85551602e-01 -4.88297567e-02
6.29449785e-01 -2.91603267e-01 -7.09019303e-01 3.24731857e-01
-3.30695182e-01 -2.66011715e-01 3.33205044e-01 1.81231827e-01
4.85540390e-01 -1.27788931e-01 -2.33138278e-01 -5.47474504e-01
5.03817618e-01 9.65159014e-02 -2.09520966e-01 3.10789585e-01
1.64245993e-01 -4.25890893e-01 1.41757742e-01 -4.18023765e-01
-1.07668233e+00 7.98511863e-01 -2.25399919e-02 -8.37089956e-01
-5.26574492e-01 -6.48132741e-01 -3.34755301e-01 -1.50914624e-01
9.55888629e-01 8.80791545e-02 6.55642822e-02 -1.47763774e-01
4.92789805e-01 3.23617041e-01 1.18670595e+00 -4.00373816e-01
8.75818074e-01 6.22387409e-01 -4.02950644e-02 -7.13294148e-01
-3.73130381e-01 -3.73856157e-01 -1.03803325e+00 -6.61026418e-01
1.12986755e+00 -9.18362141e-01 -9.17932212e-01 4.60300684e-01
-1.34713984e+00 -2.63745952e-02 -2.66062617e-01 4.19337392e-01
-9.02245104e-01 2.45730147e-01 -5.82604229e-01 -8.25872123e-01
-1.61848828e-01 -8.19104850e-01 1.49538624e+00 2.85531819e-01
-6.48418963e-01 -8.29948604e-01 2.93020248e-01 3.00909311e-01
-5.39919436e-02 2.58245975e-01 -2.19089538e-01 -4.89128500e-01
-3.98125470e-01 -3.68214548e-01 2.12449431e-01 7.22719543e-03
6.37464747e-02 2.42634729e-01 -9.38686728e-01 -4.55178767e-01
-3.31590086e-01 -2.15139210e-01 4.16332036e-01 6.02060199e-01
7.58952320e-01 -2.06721187e-01 -5.58884621e-01 6.64385974e-01
9.41604316e-01 1.93123266e-01 9.78412151e-01 -1.17062025e-01
7.24554002e-01 8.67924213e-01 3.98594558e-01 2.74083674e-01
3.25473733e-02 9.17753458e-01 1.20317183e-01 -7.65580870e-03
-3.49106342e-01 -7.07477152e-01 5.02430141e-01 -2.86658645e-01
-5.60923338e-01 -8.44844356e-02 -7.40794241e-01 3.49476427e-01
-1.68775594e+00 -1.44848323e+00 7.53639871e-03 2.26579952e+00
4.35769618e-01 -9.28416923e-02 5.91391265e-01 -1.81912199e-01
1.07308269e+00 -2.94377983e-01 -3.67039442e-01 4.64126579e-02
2.23813877e-01 -9.12542194e-02 3.73859227e-01 5.31104565e-01
-1.39251506e+00 1.09904587e+00 7.84866333e+00 5.40913343e-01
-3.90840173e-01 -3.14018309e-01 6.66505039e-01 -3.82538438e-01
1.80562139e-01 -1.58118710e-01 -1.27267694e+00 2.24767268e-01
5.42158067e-01 1.87574089e-01 4.28448975e-01 8.96076083e-01
2.99856544e-01 -5.68752624e-02 -1.64733315e+00 1.37674093e+00
7.61633217e-01 -1.18754780e+00 7.64417276e-02 -7.79211670e-02
7.25386262e-01 -5.72391868e-01 1.96458906e-01 1.55349243e-02
3.00876617e-01 -1.03747058e+00 6.73396051e-01 7.98357606e-01
9.22405958e-01 -8.57775629e-01 4.77785081e-01 -5.30373678e-02
-1.21264577e+00 3.74934584e-01 -1.71541452e-01 8.50944072e-02
2.69171000e-01 8.81655514e-02 -1.24480855e+00 3.12262196e-02
8.43917251e-01 4.63313490e-01 -5.46236515e-01 8.17162275e-01
-2.23101050e-01 1.65790379e-01 -3.21406543e-01 3.41847986e-01
-1.75295368e-01 2.79034059e-02 7.89497077e-01 1.32791054e+00
4.39903229e-01 2.23063603e-01 2.96258628e-01 6.94894969e-01
-1.55356899e-02 -2.24912211e-01 -9.72084463e-01 2.02748835e-01
7.39706099e-01 1.31665647e+00 -7.85399020e-01 -4.77650583e-01
-1.39973834e-01 1.66878331e+00 -3.80561054e-02 1.85067028e-01
-9.24245954e-01 -2.05204010e-01 7.84983218e-01 3.77787501e-01
2.03456387e-01 -1.97210714e-01 2.51706690e-01 -7.50288904e-01
1.30598722e-02 -8.35453093e-01 3.16943765e-01 -1.11232400e+00
-1.07056558e+00 5.88743746e-01 2.79057175e-01 -1.27774620e+00
-8.23426247e-01 -3.58298838e-01 -5.28777778e-01 7.54400969e-01
-2.15843722e-01 -1.58349836e+00 -4.15853560e-01 7.95339525e-01
6.51867628e-01 -2.62210369e-01 6.77122235e-01 2.05232069e-01
-4.05962855e-01 6.15286946e-01 -6.62197292e-01 3.56390446e-01
1.05438936e+00 -1.22389591e+00 6.92051947e-01 9.05621052e-01
3.61142725e-01 9.65489805e-01 8.25809419e-01 -1.02062738e+00
-1.36443615e+00 -1.00415444e+00 9.07540679e-01 -1.03883636e+00
7.99411312e-02 -4.89642709e-01 -1.88341781e-01 1.09240973e+00
5.45332320e-02 -6.31749211e-03 5.11949241e-01 -2.57375717e-01
-7.41252378e-02 4.26090322e-02 -1.35821974e+00 7.76562929e-01
1.60059559e+00 -6.42860055e-01 -5.22869468e-01 5.15708327e-01
1.60545275e-01 -8.57797861e-01 -7.04651773e-01 -6.77299453e-03
8.26194644e-01 -1.33363509e+00 1.36988163e+00 -6.71341360e-01
2.47880101e-01 -5.61492324e-01 -7.35888407e-02 -9.58291233e-01
-2.99741238e-01 -1.01396239e+00 -9.46224108e-02 1.33065403e+00
-6.98976442e-02 2.89715594e-03 1.17831922e+00 9.81635630e-01
2.21848503e-01 -4.26187754e-01 -6.75486088e-01 -6.06435180e-01
-4.95404273e-01 -3.40212733e-01 4.06069219e-01 5.38442314e-01
-1.58072278e-01 3.31904665e-02 -7.66059875e-01 2.94100702e-01
6.88797057e-01 -4.24315840e-01 1.33411002e+00 -6.13612711e-01
-3.97663087e-01 2.03331560e-02 -9.86423969e-01 -1.14519906e+00
-1.80226654e-01 -3.04875612e-01 -2.34033400e-03 -1.27762330e+00
3.07684004e-01 2.49567971e-01 4.18444604e-01 4.34839725e-01
-1.04969494e-01 7.84945250e-01 1.97779447e-01 3.17258120e-01
-6.88443840e-01 -3.58951464e-02 9.25826848e-01 2.89447725e-01
-2.05122903e-01 7.09190592e-02 -3.64186794e-01 9.64666009e-01
4.15038705e-01 -1.68342143e-01 -3.45265627e-01 -3.33572656e-01
3.28323543e-02 1.69541165e-01 9.96112764e-01 -1.40908337e+00
3.54278207e-01 1.41863585e-01 1.25200999e+00 -7.75592923e-01
7.33028471e-01 -5.82613051e-01 7.39014983e-01 3.51903021e-01
-3.14696044e-01 3.77662599e-01 -1.72575880e-02 4.21185821e-01
1.79905400e-01 -3.34917232e-02 6.59638166e-01 -3.49250317e-01
-7.79076517e-01 1.77281424e-01 -7.97615014e-03 -2.03191414e-01
1.37878394e+00 -6.63665652e-01 1.22145958e-01 -7.02945709e-01
-1.33615685e+00 -2.43065089e-01 6.75738931e-01 2.74236858e-01
7.95601189e-01 -1.45883322e+00 -7.87362754e-01 2.51423985e-01
-2.04868972e-01 -4.35939103e-01 2.39747718e-01 5.34686685e-01
-7.60051787e-01 1.45762846e-01 -1.40202329e-01 -5.57510316e-01
-1.92534459e+00 3.30761403e-01 5.13279140e-01 3.64005148e-01
-5.77506006e-01 1.44988656e+00 7.26781562e-02 -2.52138853e-01
3.02804500e-01 3.67742151e-01 1.00834984e-02 -1.86965227e-01
7.93177962e-01 9.38779652e-01 -1.19681992e-01 -9.28755939e-01
-6.03001773e-01 6.00317359e-01 4.76746820e-02 -4.08627778e-01
9.78719056e-01 -7.49456137e-02 1.55939028e-01 4.89181466e-02
1.06215549e+00 9.88800228e-02 -1.73777664e+00 3.91599029e-01
-7.70675302e-01 -8.75412703e-01 -4.06980425e-01 -6.67209625e-01
-7.57876456e-01 4.18904603e-01 7.16276526e-01 -1.74740598e-01
8.17840278e-01 2.98933744e-01 4.81670320e-01 1.97092190e-01
5.32141864e-01 -1.23641574e+00 2.55802870e-01 1.33486971e-01
9.65904117e-01 -8.36184978e-01 2.19678476e-01 -5.87193429e-01
-6.59889758e-01 1.10461712e+00 8.49263191e-01 -2.74913430e-01
1.87652066e-01 4.10168380e-01 1.67247668e-01 -3.78840387e-01
-3.38310510e-01 -2.47624163e-02 4.61916149e-01 1.20092106e+00
5.39223850e-01 -1.09613590e-01 1.93009079e-01 -2.90921479e-01
-4.61975515e-01 1.42278941e-02 1.34318709e-01 8.62896681e-01
-1.43372387e-01 -8.79004300e-01 -9.11334217e-01 2.46762499e-01
-2.84240782e-01 3.96271758e-02 -8.67186904e-01 7.17504740e-01
3.17860425e-01 7.52727330e-01 3.13976884e-01 -3.38004231e-01
3.74619454e-01 1.17653467e-01 9.46940362e-01 -6.17037952e-01
-6.25555158e-01 3.56974930e-01 1.60878956e-01 -8.77103329e-01
-5.20506799e-01 -1.17814040e+00 -8.37411284e-01 -4.59900886e-01
5.08982651e-02 -3.02864134e-01 3.53256792e-01 6.35388017e-01
3.91504943e-01 1.81410164e-01 3.72940212e-01 -1.31087923e+00
-2.41688922e-01 -7.96375632e-01 -3.36097777e-01 4.75284845e-01
-1.21882327e-01 -4.35499698e-01 -1.42263755e-01 6.82462215e-01]
|
[7.10809850692749, -0.9027119874954224]
|
299ed45c-8940-423a-8218-acf86d72e51c
|
approximate-bayesian-optimisation-for-neural
|
2108.12461
| null |
https://arxiv.org/abs/2108.12461v2
|
https://arxiv.org/pdf/2108.12461v2.pdf
|
Approximate Bayesian Optimisation for Neural Networks
|
A body of work has been done to automate machine learning algorithm to highlight the importance of model choice. Automating the process of choosing the best forecasting model and its corresponding parameters can result to improve a wide range of real-world applications. Bayesian optimisation (BO) uses a blackbox optimisation methods to propose solutions according to an exploration-exploitation trade-off criterion through acquisition functions. BO framework imposes two key ingredients: a probabilistic surrogate model that consist of prior belief of the unknown objective function(data-dependant) and an objective function that describes how optimal is the model-fit. Choosing the best model and its associated hyperparameters can be very expensive, and is typically fit using Gaussian processes (GPs) and at some extends applying approximate inference due its intractability. However, since GPs scale cubically with the number of observations, it has been challenging to handle objectives whose optimization requires many evaluations. In addition, most real-dataset are non-stationary which make idealistic assumptions on surrogate models. The necessity to solve the analytical tractability and the computational feasibility in a stochastic fashion enables to ensure the efficiency and the applicability of Bayesian optimisation. In this paper we explore the use of neural networks as an alternative to GPs to model distributions over functions, we provide a link between density-ratio estimation and class probability estimation based on approximate inference, this reformulation provides algorithm efficiency and tractability.
|
['Irina Rish', 'Nadhir Hassen']
|
2021-08-27
| null | null | null | null |
['density-ratio-estimation', 'bayesian-optimisation']
|
['methodology', 'methodology']
|
[ 1.64514378e-01 -1.38307035e-01 -1.39701933e-01 -3.39996129e-01
-7.20618844e-01 -4.02194709e-01 7.88056970e-01 7.84704611e-02
-4.53450263e-01 1.03304875e+00 -2.66592324e-01 -3.72680604e-01
-1.01065683e+00 -7.34518647e-01 -6.57919466e-01 -1.05210888e+00
-1.48362562e-01 8.77931774e-01 -1.58164188e-01 1.26664788e-01
5.20968556e-01 6.68771148e-01 -1.58629501e+00 -4.16499466e-01
9.63125348e-01 1.37946689e+00 3.36450726e-01 6.77755535e-01
-1.57438874e-01 1.19498387e-01 -4.94182855e-01 -3.53483558e-01
2.85318792e-01 -1.26793116e-01 -4.13614064e-01 -2.29717463e-01
-2.81809360e-01 8.29240493e-03 4.16964740e-01 9.99089301e-01
5.73538244e-01 5.52284598e-01 1.22218215e+00 -1.38910127e+00
-2.37910390e-01 1.29227474e-01 -3.16073596e-01 1.62371010e-01
-3.09042633e-02 1.51806027e-01 6.36122227e-01 -4.15072322e-01
-7.94467330e-02 1.22535431e+00 7.64558733e-01 8.12813938e-02
-1.43877888e+00 -3.32219750e-01 -1.08699709e-01 7.56859481e-02
-1.58149612e+00 -2.85711676e-01 5.21042705e-01 -4.57990021e-01
6.79593742e-01 3.12181026e-01 6.13936245e-01 8.80851269e-01
3.16338003e-01 3.60476702e-01 1.15134251e+00 -5.03333986e-01
6.23548806e-01 4.29892749e-01 -1.74125075e-01 5.80718890e-02
2.59378701e-01 3.68975073e-01 -2.84194857e-01 -3.01316589e-01
6.72431648e-01 -7.46909752e-02 -2.21813858e-01 -4.09503818e-01
-6.09160721e-01 1.09914446e+00 5.65738343e-02 2.36394126e-02
-6.08724952e-01 3.37310314e-01 2.33987957e-01 7.81952310e-03
2.70387888e-01 4.78228122e-01 -5.46480119e-01 -4.33230817e-01
-1.15652573e+00 4.61979568e-01 9.82100725e-01 5.15063226e-01
4.95593876e-01 1.03868358e-01 -7.81333074e-03 7.92491853e-01
7.71520376e-01 5.53836465e-01 5.51489234e-01 -9.53076243e-01
3.25080276e-01 1.55232534e-01 3.82020295e-01 -9.23068345e-01
-2.04671815e-01 -8.10957730e-01 -8.62805128e-01 3.61231685e-01
5.04437983e-01 -6.79839179e-02 -6.54033542e-01 1.45359051e+00
6.54307485e-01 3.07997502e-02 -1.17551304e-01 7.27368295e-01
7.16664195e-02 8.90081048e-01 2.05700710e-01 -4.06499982e-01
1.12763238e+00 -3.09380889e-01 -6.23203397e-01 4.57187928e-02
2.72649139e-01 -7.93516934e-01 9.33852851e-01 7.12878287e-01
-1.01093388e+00 -3.91278476e-01 -9.74287570e-01 6.07095242e-01
-4.00501400e-01 5.30672185e-02 4.30319846e-01 9.91392076e-01
-8.44597161e-01 8.27923238e-01 -8.11005890e-01 3.01721394e-02
1.65412664e-01 7.34683812e-01 -3.40542234e-02 2.83589572e-01
-1.01015508e+00 1.22310543e+00 8.30639362e-01 5.07125139e-01
-7.31115878e-01 -6.10557258e-01 -5.95006227e-01 2.33998448e-01
2.06117958e-01 -7.55770862e-01 1.14706933e+00 -9.97428417e-01
-1.85543883e+00 1.15705796e-01 1.89007148e-02 -6.47989511e-01
7.03775465e-01 -1.69830456e-01 -2.69894123e-01 -1.73868135e-01
-2.52678782e-01 3.36858511e-01 1.00424671e+00 -1.01397157e+00
-6.71783805e-01 -1.84061661e-01 -2.80311733e-01 2.10617110e-01
-7.42521137e-02 -9.63812768e-02 -1.14911236e-01 -4.94630426e-01
7.48159736e-02 -6.94300532e-01 -2.63047218e-01 -1.83705315e-01
-3.25264931e-02 -2.45445058e-01 4.74011987e-01 -6.62021577e-01
1.36059737e+00 -1.83895648e+00 -2.71174740e-02 5.85466266e-01
-4.75643158e-01 4.51869443e-02 3.00034374e-01 4.70306188e-01
-1.43000707e-02 7.78971761e-02 -5.35723627e-01 -1.56347767e-01
3.33930314e-01 2.52139419e-01 -3.49331558e-01 7.60164261e-01
2.44492695e-01 4.46941525e-01 -5.86478233e-01 -4.74964440e-01
4.49825704e-01 6.63907707e-01 -3.04187894e-01 2.58171499e-01
-2.45376483e-01 2.99264938e-01 -5.08758724e-01 4.64021742e-01
6.97037339e-01 -8.99637043e-02 -1.71498492e-01 -6.82064667e-02
-1.38684869e-01 3.71821374e-02 -1.67552459e+00 9.26675677e-01
-6.98935330e-01 3.53311300e-01 1.97538003e-01 -1.44565165e+00
1.10146141e+00 2.64551073e-01 3.61484915e-01 -2.74990022e-01
2.72725135e-01 4.59244370e-01 -5.16951531e-02 -4.44939971e-01
2.17326656e-01 -5.11730015e-01 2.07082063e-01 2.25975335e-01
-5.96559383e-02 -2.46223912e-01 1.23723209e-01 -5.80111921e-01
5.30495882e-01 6.45789802e-01 4.93841588e-01 -4.64950293e-01
6.28455698e-01 -9.24006104e-02 3.40671450e-01 7.78383195e-01
3.61593775e-02 4.23638552e-01 3.84104460e-01 -3.10267746e-01
-8.80998611e-01 -8.49192619e-01 -6.66325510e-01 6.82921767e-01
-3.22396815e-01 3.46343458e-01 -5.17241359e-01 -2.15502799e-01
7.25159645e-02 1.11998940e+00 -5.30054092e-01 -5.34724258e-02
-3.81357342e-01 -1.24935472e+00 2.06079423e-01 1.98082879e-01
1.97862610e-01 -8.27174485e-01 -8.01540971e-01 4.55810875e-01
2.62168616e-01 -5.34388185e-01 3.58933471e-02 5.13538837e-01
-1.05113864e+00 -6.76259577e-01 -8.43291342e-01 -8.18296447e-02
5.12867332e-01 -5.17059982e-01 1.06345463e+00 -3.41879576e-01
8.73835459e-02 1.68630019e-01 -3.59263420e-02 -5.20607650e-01
-4.00329053e-01 4.24746387e-02 -2.06955522e-02 1.28156289e-01
2.46608391e-01 -6.97991729e-01 -4.45969820e-01 3.95979911e-01
-9.23448324e-01 -3.58967215e-01 6.74320936e-01 8.18475187e-01
5.69379807e-01 4.80055720e-01 5.97123444e-01 -4.02400076e-01
8.58236909e-01 -7.97246814e-01 -1.24708283e+00 3.88373911e-01
-8.88599873e-01 4.56584334e-01 5.42410791e-01 -4.01484281e-01
-1.06791902e+00 -1.14107784e-02 -2.61041969e-02 -3.09266448e-01
-1.76263630e-01 7.24617243e-01 -1.87675834e-01 6.92250803e-02
5.19414842e-01 8.77240002e-02 1.37056142e-01 -4.66508478e-01
9.22173038e-02 7.53966570e-01 4.83585685e-01 -9.46419001e-01
4.81359929e-01 1.10120490e-01 6.29672706e-01 -6.28887355e-01
-6.15377545e-01 -4.65940803e-01 -4.92730021e-01 -2.17444301e-01
5.75118661e-01 -4.40396816e-01 -1.10343802e+00 -3.11082415e-03
-1.01315641e+00 -1.69450432e-01 -1.59993768e-01 7.16473222e-01
-7.75977671e-01 1.94397569e-01 2.56162405e-01 -1.60171998e+00
-3.78710628e-01 -1.23593557e+00 8.30435812e-01 2.29528397e-01
-3.69033813e-01 -1.19856226e+00 -1.59484253e-03 5.00157177e-02
4.92289007e-01 3.70038211e-01 7.61870682e-01 -6.29196882e-01
-4.98865932e-01 -4.19110656e-01 -9.39221159e-02 3.75488937e-01
-1.82595566e-01 1.99289799e-01 -7.87726521e-01 -6.54537380e-02
3.28625321e-01 5.24866655e-02 4.33530629e-01 9.56340194e-01
1.08364594e+00 -4.08626258e-01 -1.63491592e-01 5.83497286e-01
1.63449061e+00 3.90035540e-01 4.15861338e-01 6.17757380e-01
1.80630073e-01 8.29575062e-01 6.80887103e-01 5.90306461e-01
-8.86285305e-02 7.01594830e-01 5.51829696e-01 4.05755818e-01
4.20412779e-01 6.20505847e-02 2.25858510e-01 3.37239146e-01
-2.90280897e-02 -3.18173140e-01 -1.09623122e+00 4.23995644e-01
-1.93107796e+00 -8.34617913e-01 -1.08044252e-01 2.80239344e+00
8.13681483e-01 2.97939926e-01 2.60526150e-01 2.88488120e-01
6.88119113e-01 -4.51691866e-01 -3.24432582e-01 -6.50774539e-01
1.82460085e-01 8.01360533e-02 8.12007606e-01 6.58365369e-01
-1.04102170e+00 1.53026626e-01 6.33786440e+00 1.30989194e+00
-9.81498778e-01 -3.77197303e-02 7.41473019e-01 -3.89170647e-02
-2.77066469e-01 1.11288108e-01 -1.04324913e+00 8.10254991e-01
1.36984003e+00 2.90609729e-02 6.05637193e-01 8.45108151e-01
5.96887648e-01 -4.13178742e-01 -8.14561188e-01 8.80286753e-01
-2.41455987e-01 -9.98808086e-01 -1.98768884e-01 2.26808339e-01
6.04957521e-01 -2.14425594e-01 6.48140684e-02 2.53842413e-01
1.99252352e-01 -1.42769253e+00 8.18419576e-01 1.01412582e+00
3.89748365e-01 -1.08442509e+00 1.02872860e+00 5.79642653e-01
-7.42544651e-01 -2.74064094e-01 -3.54787976e-01 -8.66093114e-02
2.69665599e-01 7.89173961e-01 -9.56667423e-01 4.96541023e-01
6.31154716e-01 -3.17904241e-02 -7.64718950e-02 1.55699980e+00
-1.80516839e-02 5.96503377e-01 -1.00230944e+00 -3.86888981e-01
3.79838467e-01 -7.06480205e-01 5.63701451e-01 1.04659092e+00
6.48115456e-01 -3.35798472e-01 -6.84234053e-02 1.03703618e+00
6.18918717e-01 2.01173261e-01 -2.69108653e-01 2.63008568e-02
5.60596526e-01 9.82984781e-01 -8.00873697e-01 3.45206745e-02
-3.83425467e-02 1.46708369e-01 -2.41620634e-02 4.37381089e-01
-6.86230898e-01 -1.62836269e-01 1.45227268e-01 9.11857560e-02
3.33713084e-01 -1.53141111e-01 -5.19743323e-01 -3.71434212e-01
6.38579950e-02 -8.22098136e-01 3.81391913e-01 -5.41108847e-01
-1.22985601e+00 3.44324678e-01 6.38058007e-01 -1.06399035e+00
-6.81346416e-01 -6.93925321e-01 -6.03283167e-01 1.38285184e+00
-1.37570918e+00 -8.58261645e-01 4.46411110e-02 7.49358982e-02
3.45124066e-01 -1.64349988e-01 6.13255084e-01 6.70993477e-02
-4.85622227e-01 1.06468469e-01 6.08770788e-01 -6.96760178e-01
1.45275950e-01 -1.28241265e+00 -2.19806060e-01 5.50668061e-01
-1.70021743e-01 4.95382547e-01 1.26733243e+00 -5.25106370e-01
-1.03712749e+00 -5.94650447e-01 6.84130073e-01 -3.52911353e-01
6.43004119e-01 4.06008698e-02 -8.15216362e-01 8.53281021e-02
-1.97926968e-01 -1.87743217e-01 4.11736161e-01 1.81920789e-02
4.40976918e-01 -2.62275130e-01 -1.17311895e+00 4.23861742e-01
1.09766625e-01 -2.60948092e-02 -4.78804290e-01 1.56101346e-01
4.53298464e-02 -1.91939287e-02 -1.01303506e+00 4.94230151e-01
6.61982656e-01 -8.56532335e-01 1.00940418e+00 -3.41512918e-01
7.74150938e-02 -3.89299721e-01 -1.67954713e-01 -1.08788407e+00
1.35591820e-01 -7.70518839e-01 -3.47514182e-01 1.33874178e+00
2.97306091e-01 -7.03711271e-01 5.96252203e-01 9.92411196e-01
2.48847470e-01 -1.11876810e+00 -1.25761151e+00 -1.03907657e+00
1.68069135e-02 -7.13455021e-01 7.26138651e-01 4.24575418e-01
-6.70405686e-01 -2.59976219e-02 -2.37344280e-01 2.51496226e-01
7.08231747e-01 -1.27796963e-01 7.41022825e-01 -1.46723330e+00
-5.50566435e-01 -8.39986920e-01 -2.49452755e-01 -8.05695355e-01
-4.00266051e-02 -3.89118314e-01 1.51427805e-01 -1.17739618e+00
-3.14866543e-01 -6.74398065e-01 -2.19099715e-01 3.63909826e-02
4.86827567e-02 -3.03758502e-01 -1.40430689e-01 3.16892028e-01
4.20095064e-02 7.16053009e-01 7.45922267e-01 2.51328945e-01
-3.18482697e-01 5.99081516e-01 -2.24086195e-01 6.83260798e-01
7.70721793e-01 -7.21991003e-01 -5.78706920e-01 1.48133099e-01
6.29057169e-01 1.85745701e-01 5.10827422e-01 -7.98490822e-01
1.36859477e-01 -3.78839940e-01 4.05204266e-01 -6.37046456e-01
3.78955245e-01 -1.13560855e+00 6.88091457e-01 2.10704580e-01
-1.27573609e-01 -2.80050701e-03 7.31986314e-02 7.30602503e-01
-2.36084089e-01 -1.15227377e+00 6.82851017e-01 8.59523192e-03
-2.51897156e-01 7.39195338e-03 -2.27670833e-01 -2.75241256e-01
9.03413594e-01 -5.45436978e-01 3.68732214e-01 -5.00782073e-01
-6.29498839e-01 1.27038494e-01 6.11012429e-02 -5.06719612e-02
9.16839391e-02 -1.04598653e+00 -5.42217553e-01 -4.34322702e-03
-4.62328076e-01 8.52603465e-02 5.21475403e-03 1.01356304e+00
-6.31366730e-01 4.68179286e-01 1.00207992e-01 -6.80844426e-01
-7.04603136e-01 3.54142666e-01 5.59911132e-01 -3.80495280e-01
-1.42913088e-02 6.33845389e-01 -2.28663296e-01 -2.47027844e-01
3.35433900e-01 -1.54105633e-01 -2.59237051e-01 9.55365077e-02
2.12954551e-01 7.62928963e-01 7.59587511e-02 -3.77496690e-01
-1.08571969e-01 3.99964213e-01 5.30584276e-01 -4.54008639e-01
1.31660652e+00 -2.39905789e-01 -4.40847799e-02 5.21783531e-01
1.05055130e+00 -4.52999294e-01 -1.46621943e+00 1.76582828e-01
2.76669145e-01 -4.28923309e-01 6.01838708e-01 -7.89013863e-01
-5.49296498e-01 8.13765824e-01 6.77232504e-01 4.29616511e-01
1.07252455e+00 -5.23926795e-01 1.64952159e-01 3.51030290e-01
5.51256239e-02 -1.28661454e+00 -5.70550561e-01 1.27078593e-01
9.78369474e-01 -1.20566869e+00 1.94960967e-01 -4.11453359e-02
-3.01667631e-01 1.31998146e+00 3.21872532e-02 3.71603481e-02
8.49643409e-01 5.45051321e-02 -3.11936080e-01 -5.78815378e-02
-4.28109229e-01 7.22186342e-02 7.12236941e-01 5.65914512e-01
2.60057032e-01 -5.31938951e-03 -3.99784803e-01 5.60845912e-01
-3.23958755e-01 1.62940845e-02 -9.35080349e-02 7.51723409e-01
-4.00057942e-01 -1.11112404e+00 -8.80710304e-01 4.83111471e-01
-5.57554305e-01 -4.29702140e-02 4.26636726e-01 9.39367056e-01
3.89618054e-02 8.88293624e-01 7.30819069e-03 4.17895615e-01
5.27297147e-02 1.25197053e-01 1.57647312e-01 -2.36195415e-01
-3.09140801e-01 2.92367160e-01 9.34901163e-02 -5.06299287e-02
-3.79348099e-01 -8.34042490e-01 -7.53194273e-01 -2.29869694e-01
-7.45418012e-01 3.32334161e-01 1.18999457e+00 1.36399639e+00
9.00122076e-02 2.82427996e-01 3.29792410e-01 -8.81670535e-01
-1.23601639e+00 -9.63487387e-01 -5.20270348e-01 -1.60865322e-01
1.06173784e-01 -9.62220967e-01 -5.20345807e-01 -1.61473066e-01]
|
[6.2590813636779785, 3.659233570098877]
|
a2631750-13b3-4546-9141-2587621b87c9
|
distribution-estimation-and-change-point
|
2211.14577
| null |
https://arxiv.org/abs/2211.14577v2
|
https://arxiv.org/pdf/2211.14577v2.pdf
|
Distribution estimation and change-point estimation for time series via DNN-based GANs
|
The generative adversarial networks (GANs) have recently been applied to estimating the distribution of independent and identically distributed data, and have attracted a lot of research attention. In this paper, we use the blocking technique to demonstrate the effectiveness of GANs for estimating the distribution of stationary time series. Theoretically, we derive a non-asymptotic error bound for the Deep Neural Network (DNN)-based GANs estimator for the stationary distribution of the time series. Based on our theoretical analysis, we propose an algorithm for estimating the change point in time series distribution. The two main results are verified by two Monte Carlo experiments respectively, one is to estimate the joint stationary distribution of $5$-tuple samples of a 20 dimensional AR(3) model, the other is about estimating the change point at the combination of two different stationary time series. A real world empirical application to the human activity recognition dataset highlights the potential of the proposed methods.
|
['Qiuran Yao', 'Lihu Xu', 'Zhijie Xiao', 'Yingjun Mo', 'Jianya Lu']
|
2022-11-26
| null | null | null | null |
['blocking']
|
['natural-language-processing']
|
[ 3.08650862e-02 -2.59621769e-01 1.62970945e-01 -2.55125642e-01
-9.26772237e-01 -5.21943927e-01 3.70997965e-01 -5.67636967e-01
-2.55130798e-01 1.00761306e+00 1.07695833e-01 -1.42157063e-01
-9.25484523e-02 -7.66692162e-01 -8.60484838e-01 -1.01343894e+00
-3.54701549e-01 3.95306528e-01 -4.22317922e-01 2.47022748e-01
-1.75596565e-01 3.67326319e-01 -1.20268250e+00 -5.18145800e-01
5.05496442e-01 1.44665837e+00 -4.14148211e-01 8.19875181e-01
1.24536492e-01 8.78091633e-01 -1.24083292e+00 -3.53179812e-01
4.69407111e-01 -7.29591846e-01 -1.52471319e-01 -2.72487968e-01
-1.51387021e-01 -4.78062898e-01 -3.29984009e-01 9.72231269e-01
5.77585638e-01 2.37584263e-01 8.81160080e-01 -1.62428248e+00
-2.99703270e-01 4.35658306e-01 -7.08008945e-01 5.06229401e-01
-7.06700459e-02 -8.93656537e-02 4.12855387e-01 -2.26745829e-01
1.84230655e-02 9.83178556e-01 9.30601716e-01 4.42777276e-01
-7.00104594e-01 -8.72671902e-01 -2.20031634e-01 2.63949931e-01
-1.51366079e+00 -1.51119918e-01 1.08354509e+00 -2.69795269e-01
5.37115812e-01 8.42460692e-02 7.40870237e-01 1.52930760e+00
4.63094831e-01 8.22916090e-01 9.97040033e-01 -1.89097002e-01
5.47615349e-01 -2.58387536e-01 -2.52709240e-01 3.67921144e-01
5.17980307e-02 2.36454353e-01 -2.41716877e-01 -3.41615349e-01
9.84334409e-01 2.54026502e-01 -4.27982658e-02 -1.50116961e-02
-8.97617817e-01 8.83697987e-01 -5.05590849e-02 2.93741137e-01
-7.49626160e-01 7.48642266e-01 4.26042199e-01 2.32706442e-01
7.42736340e-01 -2.21759215e-01 -3.71872604e-01 -6.39892638e-01
-7.72108614e-01 2.33253911e-01 5.89002311e-01 8.93176138e-01
5.22965640e-02 6.87882066e-01 -3.17651480e-01 4.11687881e-01
3.36652607e-01 1.04180014e+00 6.54710472e-01 -8.59432876e-01
3.73258591e-01 -8.36411566e-02 3.57460499e-01 -9.51656759e-01
-1.55190406e-02 -5.21324217e-01 -1.37072635e+00 -2.12941289e-01
5.40470421e-01 -7.36230493e-01 -7.28131711e-01 2.02432895e+00
2.46460766e-01 7.55856693e-01 9.50898230e-02 4.06031907e-01
1.46067992e-01 7.85863519e-01 -2.30437424e-02 -5.61171412e-01
9.39167857e-01 -2.40963757e-01 -1.00900733e+00 1.63095102e-01
3.07513904e-02 -5.60352623e-01 6.29448891e-01 3.13477427e-01
-1.03926349e+00 -5.49456358e-01 -8.31052899e-01 5.50935805e-01
-1.17245778e-01 3.66074927e-02 5.46999633e-01 7.92548895e-01
-8.40735555e-01 2.85012126e-01 -1.07659841e+00 1.28479272e-01
5.01042724e-01 6.54544905e-02 -2.83199083e-02 2.42443636e-01
-1.32760572e+00 3.37542832e-01 9.29391757e-02 3.19319904e-01
-1.03364706e+00 -6.07193768e-01 -4.93644953e-01 6.20919093e-02
-6.83014048e-03 -7.75087893e-01 1.17522955e+00 -1.17863703e+00
-1.53238451e+00 4.97845918e-01 -2.14479685e-01 -8.92018914e-01
6.92501128e-01 -1.28275290e-01 -7.00766563e-01 -7.57332891e-02
7.67427236e-02 8.48491117e-03 1.20873189e+00 -6.37700796e-01
-4.75273937e-01 -3.96928310e-01 -4.59326923e-01 -1.96930334e-01
-1.26666963e-01 -2.37071097e-01 7.23289326e-02 -1.22083926e+00
-3.45330179e-01 -8.41881454e-01 -5.73806912e-02 -3.95158887e-01
-3.67588282e-01 -2.08806470e-01 7.63209105e-01 -8.71929169e-01
1.16019964e+00 -2.38390374e+00 -3.47842239e-02 3.10124606e-01
-1.41019046e-01 6.19531842e-03 1.39962256e-01 4.15179044e-01
-2.26111189e-01 -8.01368505e-02 -3.11162621e-01 -4.21803683e-01
-3.22623290e-02 9.27608162e-02 -6.73928916e-01 7.06581295e-01
-4.64025475e-02 8.53380620e-01 -5.85449696e-01 -5.33931563e-03
2.98020355e-02 5.63434184e-01 4.74367514e-02 4.28887099e-01
-6.10781685e-02 5.71637034e-01 -2.64189810e-01 2.76553899e-01
6.15604401e-01 -3.76236066e-02 -1.74579740e-01 1.41228540e-02
3.87063056e-01 -3.23805481e-01 -1.15349805e+00 1.32914233e+00
-3.76862615e-01 6.43381417e-01 -4.24287796e-01 -1.23058116e+00
9.94812667e-01 5.82188487e-01 7.45079339e-01 -6.52919352e-01
4.76491719e-01 2.77866665e-02 -1.95240751e-02 -2.73055464e-01
-4.89683673e-02 -4.25718278e-01 -2.43840948e-01 6.45534456e-01
-6.34542033e-02 5.03246337e-02 -2.41643667e-01 -2.63895124e-01
1.17505074e+00 -6.98194206e-02 3.66863579e-01 8.78746957e-02
3.41617614e-01 -6.04646146e-01 4.98003781e-01 7.00284779e-01
-2.59666890e-01 3.92304629e-01 5.20806313e-01 -4.11834121e-01
-1.09160531e+00 -1.26681733e+00 2.61133015e-01 2.39741474e-01
-2.49837607e-01 1.59139380e-01 -9.15275335e-01 -6.34460747e-01
-1.32674351e-01 1.07633221e+00 -9.02576149e-01 -3.57670069e-01
-5.46573222e-01 -8.22904468e-01 7.07318842e-01 8.57036173e-01
7.33947754e-01 -1.02738607e+00 -6.38042271e-01 2.39943832e-01
-2.90937871e-01 -1.13553762e+00 -4.26541358e-01 3.07522789e-02
-7.87822545e-01 -8.64537358e-01 -1.02806556e+00 -2.79706717e-01
3.92565519e-01 -1.48617495e-02 8.87066841e-01 -5.95324993e-01
-7.40612298e-02 9.11898494e-01 -6.94469502e-03 -9.71000910e-01
-5.29719830e-01 -1.60901189e-01 7.61266053e-02 3.42011511e-01
3.34512562e-01 -8.89361262e-01 -7.75137484e-01 4.04358655e-01
-9.22417819e-01 -3.57087493e-01 3.18765044e-01 5.55185974e-01
7.41503477e-01 5.91073453e-01 9.06190097e-01 -4.58022267e-01
8.47711444e-01 -7.52813220e-01 -7.20353305e-01 1.22324072e-01
-3.68710548e-01 3.10390890e-02 7.49983013e-01 -7.64247417e-01
-9.92028654e-01 -2.02373788e-01 -9.09999609e-02 -8.06567550e-01
-1.46391630e-01 3.65615815e-01 -1.48764446e-01 4.80934799e-01
2.97944665e-01 4.70524728e-01 1.03520975e-01 -1.71837822e-01
2.01549321e-01 5.18190503e-01 7.30911493e-01 -4.71663535e-01
8.60558867e-01 6.82737291e-01 2.05286756e-01 -6.04139626e-01
-7.04621196e-01 -3.18834901e-01 -1.05915941e-01 -2.41398603e-01
7.17426896e-01 -8.91766906e-01 -8.78360033e-01 1.04283237e+00
-1.08379126e+00 -4.35695201e-01 -6.11549854e-01 6.18635297e-01
-1.10648799e+00 1.68078750e-01 -3.99184048e-01 -1.20357871e+00
-5.59429646e-01 -5.13043165e-01 9.81982291e-01 1.23308457e-01
-1.63130611e-01 -1.21970177e+00 4.11111623e-01 2.63671409e-02
3.78295928e-01 6.72456741e-01 8.16082597e-01 -6.66981161e-01
-3.07693332e-01 -7.15633333e-01 2.72035360e-01 6.41433716e-01
3.00600052e-01 -1.18073359e-01 -8.00850630e-01 -2.74138480e-01
7.45853364e-01 3.22260648e-01 6.36007562e-02 9.68041182e-01
1.38312161e+00 -3.13209921e-01 -2.65066046e-02 5.35625279e-01
1.26423776e+00 7.53820062e-01 1.00217485e+00 -8.61804932e-02
4.83773261e-01 -8.39343444e-02 6.64793015e-01 7.83000052e-01
4.56140153e-02 3.67226750e-01 4.96912688e-01 2.54773289e-01
2.62651354e-01 -3.37203652e-01 4.20224786e-01 8.30986619e-01
-2.00736955e-01 -6.84707344e-01 -7.07018018e-01 5.89552462e-01
-1.64003479e+00 -1.31421578e+00 1.07892849e-01 2.41528487e+00
4.64399695e-01 -4.90121730e-02 4.74362910e-01 2.54821777e-01
7.22525418e-01 1.91461608e-01 -8.85449529e-01 1.82718169e-02
-3.89567576e-02 5.04973710e-01 5.96061766e-01 -4.86473739e-02
-9.86935079e-01 3.74674290e-01 7.08754826e+00 1.08242536e+00
-1.14312398e+00 1.39705762e-01 7.91885793e-01 1.61526049e-03
-1.11409657e-01 -5.54824948e-01 -5.65470099e-01 1.00474465e+00
1.31943786e+00 -4.06493902e-01 4.51516449e-01 6.11559749e-01
2.71326065e-01 1.73150953e-02 -9.43211496e-01 1.36790407e+00
1.81034002e-02 -9.85971808e-01 -1.95936993e-01 1.14760287e-01
6.69428945e-01 -1.69148117e-01 3.69150370e-01 2.27179512e-01
3.89357030e-01 -1.05662346e+00 5.92134655e-01 9.91377175e-01
7.97759175e-01 -1.11665285e+00 9.00500655e-01 5.26058197e-01
-9.23808396e-01 -7.17124045e-02 2.15604100e-02 8.68492499e-02
3.24893951e-01 7.06470490e-01 -7.16887236e-01 3.45250487e-01
5.84190905e-01 5.03639042e-01 -1.41878380e-02 8.35061848e-01
-1.77390561e-01 1.08099580e+00 -4.33838874e-01 -1.61439016e-01
-6.57079089e-03 -3.46719056e-01 6.16498530e-01 5.37155271e-01
8.50365818e-01 7.19146850e-03 -6.88868284e-01 8.22539330e-01
4.89522330e-02 -2.87650824e-01 -6.20360672e-01 -4.04716969e-01
4.76955414e-01 7.72056103e-01 -5.67917824e-01 -1.68984488e-01
-1.07326835e-01 9.53372359e-01 -2.81404685e-02 5.83059013e-01
-1.45582080e+00 -2.68456846e-01 5.20352185e-01 -1.07692637e-01
6.72601640e-01 -3.84421766e-01 -2.76742969e-02 -9.95610476e-01
2.49296069e-01 -9.44993854e-01 3.91209245e-01 -7.28426337e-01
-1.47524178e+00 3.75239462e-01 1.64930165e-01 -1.38044727e+00
-8.10923934e-01 -1.42968386e-01 -7.15978503e-01 8.88583720e-01
-8.78423154e-01 -7.33714104e-01 -8.95898566e-02 8.51290762e-01
3.55811179e-01 -4.54634726e-01 8.48489046e-01 2.51936883e-01
-5.34224629e-01 6.30401373e-01 4.47705448e-01 3.74701679e-01
1.05269350e-01 -1.04356360e+00 6.51595950e-01 9.98341203e-01
2.14200795e-01 4.27686006e-01 8.28081310e-01 -5.59146464e-01
-1.20855796e+00 -9.66609359e-01 1.90226406e-01 -3.63773823e-01
5.71039617e-01 -3.04051489e-02 -5.94460905e-01 8.21448624e-01
1.17607452e-01 -1.44756705e-01 8.19902837e-01 -4.63511080e-01
-1.09115340e-01 -3.14609170e-01 -1.44107473e+00 3.09376150e-01
7.05672503e-01 -3.28890145e-01 -1.72950968e-01 1.12993792e-01
3.41447473e-01 -3.64284754e-01 -8.85986328e-01 3.93995732e-01
5.00529826e-01 -9.51125264e-01 9.78231788e-01 -5.98601460e-01
3.26759398e-01 -1.32926166e-01 -3.61595243e-01 -1.46399474e+00
7.80524835e-02 -7.81704187e-01 -6.05976820e-01 1.20150781e+00
-1.54257163e-01 -8.97241712e-01 6.51799083e-01 2.85220355e-01
3.60926032e-01 -5.19466758e-01 -1.34083831e+00 -8.87372434e-01
-1.28362224e-01 -5.54995179e-01 8.05370748e-01 5.18070281e-01
-7.98966110e-01 9.84023362e-02 -6.98922515e-01 1.64636850e-01
7.67449081e-01 -1.96865097e-01 9.48255002e-01 -8.81889164e-01
-5.48921585e-01 -1.38983428e-01 -5.68994105e-01 -9.86241221e-01
2.93353140e-01 -1.24180295e-01 9.60923880e-02 -9.46710646e-01
-1.65275127e-01 -1.57981709e-01 -4.00524348e-01 -8.54554698e-02
9.11082402e-02 -1.15106024e-01 -1.50976986e-01 -3.95143405e-02
-2.64946967e-01 7.36942112e-01 7.52980888e-01 -3.66333388e-02
1.84043229e-01 8.04876626e-01 -4.23755854e-01 6.42889082e-01
9.02688682e-01 -5.14201880e-01 -8.52620542e-01 -1.19339965e-01
2.04506189e-01 4.05942023e-01 4.46870029e-01 -1.15216434e+00
-2.69676119e-01 -3.01057678e-02 4.57347870e-01 -7.34458327e-01
3.65766078e-01 -1.19106102e+00 6.99247897e-01 4.52898234e-01
-1.54601723e-01 1.49603322e-01 1.67352661e-01 1.14558017e+00
-2.85057276e-01 -4.33607474e-02 6.87081993e-01 8.16197395e-02
-2.49159381e-01 5.70064187e-01 -3.40851307e-01 2.04102367e-01
1.19585586e+00 8.92725214e-02 8.80692340e-03 -1.12297964e+00
-4.63831246e-01 -2.55618453e-01 -2.37542599e-01 2.36542359e-01
4.80216503e-01 -1.63818002e+00 -4.09407794e-01 3.72184008e-01
-3.80391628e-01 6.51333705e-02 4.23672408e-01 6.54835045e-01
-2.77657419e-01 8.88002664e-02 -1.25068516e-01 -5.21812201e-01
-9.66752827e-01 6.60707355e-01 4.54538703e-01 -3.09237510e-01
-1.30650192e-01 5.82502425e-01 -1.69856593e-01 1.84958026e-01
2.31724158e-01 -3.74205559e-01 9.25359428e-02 -4.32015732e-02
4.81151432e-01 7.31955647e-01 -1.15700439e-01 -4.20328736e-01
-1.79014221e-01 3.78141999e-01 2.63936669e-01 -3.42651933e-01
1.33567452e+00 -8.04955512e-02 6.41693994e-02 9.58884716e-01
1.44526267e+00 -9.89430621e-02 -1.16130459e+00 -1.34978116e-01
-3.82816643e-01 -2.32144907e-01 -2.96884209e-01 -5.16464889e-01
-1.27531707e+00 7.71492958e-01 9.29651618e-01 6.30712569e-01
1.33006871e+00 -3.24577212e-01 1.06628990e+00 -2.05517992e-01
4.32832718e-01 -7.49543190e-01 -3.05917207e-02 1.77404448e-01
7.67496824e-01 -8.32864583e-01 -4.10279661e-01 5.48703223e-02
-4.60833579e-01 8.34076762e-01 1.46415338e-01 -3.53265762e-01
8.48503530e-01 3.61524373e-01 5.91013320e-02 1.08798973e-01
-4.60868984e-01 1.92866981e-01 1.17304690e-01 8.88251424e-01
1.65702462e-01 2.99670875e-01 -5.94790727e-02 5.99293888e-01
-3.54159206e-01 2.81857073e-01 2.82046527e-01 4.86266106e-01
2.65378803e-01 -7.21698046e-01 -3.28118742e-01 3.38856667e-01
-6.61185324e-01 1.30951703e-01 1.01082839e-01 6.93200469e-01
-2.40687370e-01 7.61081398e-01 4.20803547e-01 -7.76758566e-02
1.47804826e-01 1.59765393e-01 3.67381305e-01 1.66075766e-01
-6.98214397e-02 1.18590020e-01 -3.56907696e-01 -2.85052359e-01
-5.60632288e-01 -7.35713720e-01 -7.99505055e-01 -5.18744826e-01
7.15000108e-02 1.16778940e-01 7.53436625e-01 9.45276201e-01
4.03793991e-01 5.98470271e-01 9.39017951e-01 -3.08711588e-01
-9.29936051e-01 -9.99514341e-01 -7.65595138e-01 4.31574792e-01
2.54827857e-01 -4.61498171e-01 -7.59134829e-01 6.31005242e-02]
|
[6.991392135620117, 3.250979423522949]
|
d8ba2e9c-f5af-4759-8386-48226907d7ca
|
unsupervised-deep-language-and-dialect
| null | null |
https://aclanthology.org/2020.coling-main.141
|
https://aclanthology.org/2020.coling-main.141.pdf
|
Unsupervised Deep Language and Dialect Identification for Short Texts
|
Automatic Language Identification (LI) or Dialect Identification (DI) of short texts of closely related languages or dialects, is one of the primary steps in many natural language processing pipelines. Language identification is considered a solved task in many cases; however, in the case of very closely related languages, or in an unsupervised scenario (where the languages are not known in advance), performance is still poor. In this paper, we propose the Unsupervised Deep Language and Dialect Identification (UDLDI) method, which can simultaneously learn sentence embeddings and cluster assignments from short texts. The UDLDI model understands the sentence constructions of languages by applying attention to character relations which helps to optimize the clustering of languages. We have performed our experiments on three short-text datasets for different language families, each consisting of closely related languages or dialects, with very minimal training sets. Our experimental evaluations on these datasets have shown significant improvement over state-of-the-art unsupervised methods and our model has outperformed state-of-the-art LI and DI systems in supervised settings.
|
['John P. McCrae', 'Theodorus Fransen', 'Bharathi Raja Chakravarthi', 'Rajdeep Sarkar', 'Koustava Goswami']
|
2020-12-01
| null | null | null |
coling-2020-8
|
['dialect-identification']
|
['natural-language-processing']
|
[-1.80981293e-01 -2.75219113e-01 -3.74299251e-02 -6.28007650e-01
-5.63412070e-01 -8.86842668e-01 8.05902958e-01 5.40177703e-01
-7.00672328e-01 2.29479343e-01 5.47960401e-01 -4.46540177e-01
1.24784023e-01 -4.68379617e-01 -2.39631325e-01 -4.64962453e-01
9.20714289e-02 1.18419611e+00 -2.36107782e-02 -2.25170046e-01
1.31677791e-01 3.04329038e-01 -1.15046310e+00 2.84970075e-01
8.95029247e-01 2.99225807e-01 2.26592213e-01 6.81585014e-01
-7.46562541e-01 6.38302445e-01 -2.04016656e-01 -5.71946859e-01
1.09016672e-02 -1.70741171e-01 -1.12400365e+00 -1.38184354e-02
4.87453133e-01 -5.04448637e-02 -2.24231169e-01 1.07103646e+00
4.92779285e-01 -2.50039786e-01 7.60049760e-01 -6.98117733e-01
-6.18503809e-01 1.11953318e+00 -5.19928873e-01 1.57686651e-01
3.05757642e-01 -1.13602251e-01 1.29875982e+00 -8.90774369e-01
5.47248304e-01 1.64889371e+00 7.18013227e-01 4.92238700e-01
-1.60234165e+00 -4.19854581e-01 1.93162903e-01 -1.89989805e-02
-1.71982527e+00 -6.91068470e-01 7.13228047e-01 -7.92286634e-01
1.19894278e+00 -3.55796479e-02 2.27494035e-02 7.32989728e-01
-4.04903382e-01 1.00265372e+00 7.95815885e-01 -8.91492248e-01
-2.15345733e-02 6.45902991e-01 7.48838782e-01 5.83753526e-01
1.60704888e-02 -3.85259777e-01 -3.93531919e-01 -6.68345094e-02
1.22216679e-01 -2.17765689e-01 -4.63973843e-02 -1.16641380e-01
-1.35217261e+00 1.01162708e+00 5.88494129e-02 7.57104397e-01
1.01178037e-02 -5.08109033e-01 6.48169041e-01 3.57748061e-01
6.93175733e-01 4.34176594e-01 -4.25057471e-01 -3.44188488e-03
-8.13460946e-01 2.63506565e-02 1.02254713e+00 7.24743307e-01
7.24587977e-01 -3.74556303e-01 1.07758492e-03 1.38548481e+00
3.12967420e-01 9.50567350e-02 7.49720275e-01 -1.26339391e-01
6.95603609e-01 8.95376325e-01 -2.40722835e-01 -5.56255102e-01
-5.05654454e-01 -2.77711023e-02 -8.88283908e-01 -1.93311349e-01
6.85396314e-01 -2.11133227e-01 -5.17722845e-01 1.66113877e+00
1.24469578e-01 -2.27011487e-01 2.41269633e-01 5.78426778e-01
8.57946396e-01 7.83202291e-01 -1.15928695e-01 1.27301663e-01
1.46327853e+00 -7.97181010e-01 -4.66983467e-01 -3.98780376e-01
8.65524650e-01 -1.06973898e+00 1.31604540e+00 1.61525175e-01
-5.47380149e-01 -6.32729232e-01 -7.39114106e-01 -2.72357702e-01
-5.01471043e-01 3.70727181e-01 5.62129557e-01 7.60785103e-01
-1.00175488e+00 2.26916939e-01 -5.15824020e-01 -9.53419983e-01
1.43886864e-01 4.61759508e-01 -6.21358752e-01 -1.50185317e-01
-1.10700405e+00 6.36926889e-01 4.34894621e-01 -1.40897050e-01
-4.80281353e-01 -6.33710861e-01 -9.62114632e-01 3.32448967e-02
-6.93195537e-02 -1.57073662e-01 9.71094072e-01 -9.77821171e-01
-1.29938400e+00 1.70497894e+00 -2.91515917e-01 -4.70969617e-01
2.14201406e-01 -2.09959850e-01 -4.26522672e-01 -2.54102975e-01
1.44300342e-01 5.40570736e-01 3.07985425e-01 -8.98624241e-01
-6.04464531e-01 -4.72296566e-01 -1.05033047e-01 2.50507891e-01
-8.38090003e-01 7.24400938e-01 -5.63094616e-01 -5.29225111e-01
-8.99146646e-02 -9.21810031e-01 -6.28191978e-02 -3.16577435e-01
-5.74016392e-01 -7.32747257e-01 4.52318877e-01 -8.17107737e-01
1.34930623e+00 -2.35386610e+00 1.57507092e-01 6.31037727e-03
8.59666988e-02 3.46532464e-01 -1.22648023e-01 6.62886381e-01
-6.44008294e-02 1.41749442e-01 -3.80685419e-01 -8.70788276e-01
2.03626469e-01 2.58769542e-01 -2.27576494e-01 4.47022289e-01
2.67040223e-01 4.79382217e-01 -6.66327775e-01 -4.18218315e-01
1.12617917e-01 2.15973869e-01 -3.29364628e-01 4.43939626e-01
-1.31481633e-01 3.31450224e-01 1.33424969e-02 3.47859919e-01
7.09570169e-01 7.46827051e-02 5.07580936e-01 4.36510965e-02
-3.81827772e-01 6.62420511e-01 -1.12791991e+00 1.51036441e+00
-5.76613903e-01 9.40526664e-01 -1.73796639e-02 -1.11473405e+00
9.37677741e-01 2.10303694e-01 2.80511558e-01 -3.55633885e-01
-6.41204938e-02 2.17327833e-01 1.95593163e-01 -4.48155969e-01
4.36706930e-01 2.65209526e-02 -3.92528117e-01 8.20848465e-01
3.31816107e-01 1.64404139e-01 4.37148690e-01 1.61451906e-01
7.98904896e-01 -3.71393263e-01 3.46581250e-01 -6.63887262e-01
8.73295248e-01 -4.63818312e-02 3.69649380e-01 6.49200201e-01
-1.35244280e-01 6.86839461e-01 5.06786942e-01 -5.44305384e-01
-1.09248149e+00 -1.06742871e+00 -3.57983202e-01 1.53445685e+00
-2.36102208e-01 -5.83122730e-01 -8.08890462e-01 -6.33969426e-01
-1.20827639e-02 5.96342444e-01 -2.90717542e-01 2.31271416e-01
-7.02356279e-01 -1.05723870e+00 8.51479828e-01 2.76854068e-01
1.14179008e-01 -9.80762839e-01 3.41179848e-01 1.53803229e-01
-4.90372516e-02 -1.33324265e+00 -6.86330497e-01 1.95865825e-01
-2.81668127e-01 -8.74139667e-01 -6.34972513e-01 -1.55614448e+00
7.41009772e-01 4.27698866e-02 1.31892073e+00 -6.23022392e-02
-1.05940521e-01 1.54372677e-01 -2.50778347e-01 -3.07550937e-01
-7.28938699e-01 5.51162302e-01 4.30128664e-01 4.08999383e-01
9.10613775e-01 -3.24736923e-01 -1.09553479e-01 1.23338401e-01
-7.32082129e-01 -4.43187542e-02 9.92337093e-02 7.37446725e-01
2.94221759e-01 1.10930212e-01 1.20114185e-01 -1.28303552e+00
5.52198708e-01 -3.27902585e-01 -8.22426200e-01 4.45044369e-01
-1.69679984e-01 4.28685933e-01 9.21934187e-01 -4.78021890e-01
-9.50459003e-01 3.28799576e-01 -5.03484011e-01 2.74768889e-01
-5.98009109e-01 4.77671176e-01 -5.01997769e-01 2.29814589e-01
5.02783418e-01 2.29378492e-01 -2.31073320e-01 -1.04579532e+00
2.58023322e-01 1.33641028e+00 5.13415933e-01 -5.76694429e-01
5.73187053e-01 2.38981292e-01 -7.04181015e-01 -1.25841486e+00
-7.21758306e-01 -8.75319958e-01 -1.25263214e+00 2.59111255e-01
9.19739842e-01 -1.01405394e+00 -4.69871163e-01 6.45817220e-01
-1.27094901e+00 -2.32054725e-01 1.75206929e-01 3.86627555e-01
3.53146531e-03 5.40104806e-01 -5.99574089e-01 -7.64248073e-01
-3.76349628e-01 -1.04728210e+00 9.37127471e-01 1.14823125e-01
-5.19676387e-01 -1.38058317e+00 3.41041327e-01 1.71193525e-01
1.45016000e-01 -2.89374113e-01 1.30635059e+00 -1.20996964e+00
-1.73422083e-01 -8.84702280e-02 -1.08520962e-01 3.26860100e-01
3.45090300e-01 1.13124356e-01 -1.12480891e+00 -3.02394748e-01
-2.57620543e-01 -2.47502670e-01 9.49645460e-01 1.10559039e-01
7.58400500e-01 -2.15269849e-01 -2.04250216e-01 7.73557723e-01
1.21208441e+00 -2.04554304e-01 4.32562009e-02 3.50780040e-01
1.02026546e+00 1.03621042e+00 1.95616633e-01 2.85747945e-01
7.14495778e-01 6.08148098e-01 -2.38973767e-01 -1.99674621e-01
-6.03336003e-03 -1.86200529e-01 6.84818685e-01 1.27844441e+00
6.85772538e-01 -2.65288353e-01 -1.42732811e+00 9.69451189e-01
-1.63624001e+00 -6.85109496e-01 -2.64916062e-01 2.23618031e+00
1.05216908e+00 9.61779058e-02 2.78684556e-01 1.50249884e-01
8.61918151e-01 5.26939593e-02 -2.61456728e-01 -6.65954292e-01
-3.89227688e-01 5.50062433e-02 3.06951284e-01 7.74994910e-01
-1.37976730e+00 1.37158656e+00 5.87664318e+00 6.26383722e-01
-1.08449256e+00 -6.95930272e-02 6.87097073e-01 3.70223582e-01
-2.57241100e-01 1.07785113e-01 -1.33259916e+00 3.92479748e-01
8.45936179e-01 -2.06519380e-01 5.49006283e-01 5.81979990e-01
-3.79737429e-02 8.00339505e-03 -1.45239675e+00 1.08772981e+00
2.49814153e-01 -1.01252544e+00 -6.75397962e-02 -2.00992376e-01
5.55920660e-01 5.00393271e-01 -1.98708430e-01 1.95394456e-01
5.28118730e-01 -9.64236200e-01 4.96799201e-01 1.76539943e-01
8.65520895e-01 -8.15793633e-01 6.63636923e-01 5.94151735e-01
-1.12307549e+00 1.48434311e-01 -5.15038788e-01 2.60632727e-02
3.32055911e-02 6.46801353e-01 -1.00011396e+00 1.32740095e-01
6.78937793e-01 8.28457952e-01 -8.49530339e-01 7.28765786e-01
-2.19423890e-01 1.00970149e+00 -5.20852745e-01 -2.34660223e-01
1.75176635e-01 -3.13995153e-01 3.96289796e-01 1.69779754e+00
-7.96536207e-02 -4.05963451e-01 3.72870088e-01 7.00217128e-01
-1.30633697e-01 4.24876928e-01 -6.86068952e-01 -3.40275526e-01
2.35614792e-01 1.34336293e+00 -5.92379212e-01 -3.45871329e-01
-5.24143696e-01 1.06035864e+00 6.73304260e-01 8.22329000e-02
-2.24549472e-01 -5.27951777e-01 1.01411307e+00 -5.25074527e-02
2.28835464e-01 -4.19700742e-01 -3.00548494e-01 -1.31915343e+00
-7.06200376e-02 -8.22425961e-01 7.68237710e-01 -2.92261448e-02
-1.72918200e+00 7.02057898e-01 -3.31841350e-01 -9.59590852e-01
-4.21688139e-01 -8.91849101e-01 -6.43924415e-01 1.07649124e+00
-1.40653467e+00 -1.33267927e+00 2.38885865e-01 4.76135194e-01
6.93184257e-01 -7.11908698e-01 1.12981534e+00 3.84440958e-01
-7.78386593e-01 8.35596323e-01 7.83266783e-01 7.43001461e-01
8.85102451e-01 -1.49042010e+00 7.96763599e-01 8.20623577e-01
6.73850358e-01 7.47631311e-01 3.98289621e-01 -1.58490658e-01
-1.33404744e+00 -1.00945687e+00 1.72030175e+00 -4.42384571e-01
9.57575440e-01 -1.24614918e+00 -1.04764163e+00 5.16392350e-01
4.20991957e-01 -3.10422599e-01 1.12399995e+00 6.14289105e-01
-2.60580331e-01 -2.54915923e-01 -7.65991211e-01 5.81564367e-01
7.90721297e-01 -9.66437995e-01 -5.66305041e-01 6.28103673e-01
6.27401590e-01 1.15782827e-01 -7.66774118e-01 -2.18802050e-01
4.03016627e-01 -8.00107777e-01 8.44380260e-01 -5.32387674e-01
1.38769284e-01 -3.70222092e-01 -7.37337023e-02 -1.28243470e+00
-3.46702963e-01 -4.93475854e-01 4.20098871e-01 2.18106937e+00
4.86365259e-01 -4.95825082e-01 3.35462064e-01 3.11702460e-01
1.52254656e-01 -3.86506021e-02 -7.29154468e-01 -6.51041687e-01
2.75781035e-01 -4.06596512e-01 5.62013626e-01 1.20515716e+00
1.16917633e-01 9.14471447e-01 -7.18901083e-02 2.52290726e-01
4.13262516e-01 2.27558747e-01 8.47540855e-01 -1.24657214e+00
-3.56778800e-01 -6.39509618e-01 -3.70532691e-01 -9.91722286e-01
7.95604706e-01 -1.29355299e+00 1.17061228e-01 -1.31876695e+00
2.74390012e-01 -3.95733297e-01 -1.51575893e-01 3.42047215e-01
-2.80645013e-01 1.39684789e-02 3.74948569e-02 1.60423100e-01
-4.90568042e-01 3.58258724e-01 1.83614641e-01 -5.46115637e-01
-2.60926396e-01 6.60339147e-02 -7.14814305e-01 7.93767333e-01
6.91506922e-01 -4.69032615e-01 -8.47374648e-02 -9.58793104e-01
4.36264612e-02 -6.17529869e-01 -2.56670833e-01 -8.52919757e-01
3.17431062e-01 1.81295186e-01 6.51823431e-02 -6.20787740e-01
-1.77029908e-01 -5.03207982e-01 -4.69253689e-01 2.13168487e-01
-5.00471354e-01 2.22529694e-01 1.88493371e-01 -7.07642436e-02
-3.57254058e-01 -3.75594079e-01 8.39294851e-01 -3.67874205e-02
-8.74649823e-01 2.00727999e-01 -6.58141792e-01 2.71971971e-01
5.16822338e-01 7.84600899e-02 2.02337205e-01 -6.42587543e-02
-2.97343373e-01 4.36497241e-01 4.69456077e-01 6.69234574e-01
2.83245236e-01 -1.20885551e+00 -1.17096293e+00 6.18429184e-01
5.71091056e-01 -1.09121948e-01 -2.35474795e-01 4.75283653e-01
-6.73752844e-01 4.28816408e-01 2.15633214e-01 -7.18841553e-01
-1.55519581e+00 6.22654080e-01 3.06803435e-01 -4.06630695e-01
-3.09496611e-01 9.95676935e-01 5.53536892e-01 -1.24171865e+00
4.23911542e-01 -1.40592277e-01 -7.09859252e-01 2.13330269e-01
5.93243003e-01 -8.95838588e-02 5.29714823e-02 -1.17922723e+00
-6.03563607e-01 7.23438740e-01 -5.94079196e-01 -1.13910690e-01
1.34399593e+00 -2.70665824e-01 -5.64421117e-01 7.01572120e-01
1.37921143e+00 1.85200930e-01 -4.46619213e-01 -6.38284087e-01
6.74415648e-01 -1.90025613e-01 -1.32665843e-01 -3.67008537e-01
-6.00195467e-01 1.24301589e+00 5.63276589e-01 1.36230052e-01
8.11149895e-01 2.13730529e-01 7.04664230e-01 4.28117394e-01
3.05285174e-02 -1.23377275e+00 -4.11797792e-01 1.12804949e+00
4.26735461e-01 -1.60619545e+00 -3.95919204e-01 -1.75979286e-01
-6.17489576e-01 1.23613417e+00 4.79208291e-01 1.15910754e-01
7.20131516e-01 2.41415143e-01 3.46709073e-01 2.26923689e-01
-5.21551669e-01 -4.81553137e-01 3.96478683e-01 5.32721341e-01
1.04839182e+00 1.27824888e-01 -1.41576260e-01 6.19149089e-01
-3.54914308e-01 -6.93326771e-01 6.94561452e-02 4.43628401e-01
-2.40814015e-01 -1.41661024e+00 -4.80340213e-01 1.96642280e-01
-4.74645168e-01 -4.66899335e-01 -7.63099313e-01 4.33437526e-01
1.41795173e-01 9.94689941e-01 4.11604315e-01 -2.63931900e-01
1.65507764e-01 1.50177464e-01 1.84123665e-01 -8.47991168e-01
-8.02234828e-01 -2.49276813e-02 2.24054769e-01 1.08214580e-01
-7.71481916e-02 -8.68321598e-01 -1.26078260e+00 -3.97985369e-01
-1.43091595e-02 1.89872161e-02 4.13970232e-01 1.03830802e+00
-3.23880441e-03 3.44981514e-02 8.78576756e-01 -5.87383866e-01
-3.32845211e-01 -9.84817266e-01 -6.47430480e-01 7.86380529e-01
3.29060644e-01 -1.12383710e-02 -1.07038967e-01 1.36309341e-01]
|
[10.60677433013916, 10.151123046875]
|
5f8a753e-7502-4b9d-84b9-f82fba2def91
|
event-based-frame-interpolation-with-ad-hoc
|
2301.05191
| null |
https://arxiv.org/abs/2301.05191v1
|
https://arxiv.org/pdf/2301.05191v1.pdf
|
Event-Based Frame Interpolation with Ad-hoc Deblurring
|
The performance of video frame interpolation is inherently correlated with the ability to handle motion in the input scene. Even though previous works recognize the utility of asynchronous event information for this task, they ignore the fact that motion may or may not result in blur in the input video to be interpolated, depending on the length of the exposure time of the frames and the speed of the motion, and assume either that the input video is sharp, restricting themselves to frame interpolation, or that it is blurry, including an explicit, separate deblurring stage before interpolation in their pipeline. We instead propose a general method for event-based frame interpolation that performs deblurring ad-hoc and thus works both on sharp and blurry input videos. Our model consists in a bidirectional recurrent network that naturally incorporates the temporal dimension of interpolation and fuses information from the input frames and the events adaptively based on their temporal proximity. In addition, we introduce a novel real-world high-resolution dataset with events and color videos named HighREV, which provides a challenging evaluation setting for the examined task. Extensive experiments on the standard GoPro benchmark and on our dataset show that our network consistently outperforms previous state-of-the-art methods on frame interpolation, single image deblurring and the joint task of interpolation and deblurring. Our code and dataset will be made publicly available.
|
['Luc van Gool', 'Kaiwei Wang', 'Qi Jiang', 'Kai Zhang', 'JieZhang Cao', 'Peng Sun', 'Jingyun Liang', 'Christos Sakaridis', 'Lei Sun']
|
2023-01-12
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Sun_Event-Based_Frame_Interpolation_With_Ad-Hoc_Deblurring_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Sun_Event-Based_Frame_Interpolation_With_Ad-Hoc_Deblurring_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['deblurring', 'video-frame-interpolation']
|
['computer-vision', 'computer-vision']
|
[ 2.40593046e-01 -6.42311931e-01 -2.00620651e-01 -1.76556975e-01
-6.51541948e-01 -4.91638899e-01 5.60012758e-01 -4.44348633e-01
-3.83118510e-01 7.22822905e-01 5.73481977e-01 -2.88227767e-01
1.22445181e-01 -3.98073077e-01 -9.81072903e-01 -6.83162391e-01
-5.20633347e-02 -1.83389813e-01 3.89077246e-01 1.29089057e-01
1.55912086e-01 2.37671569e-01 -1.29448938e+00 4.68059421e-01
6.99171245e-01 9.10436928e-01 2.45893940e-01 1.01436651e+00
2.64269918e-01 1.23234773e+00 -4.50614631e-01 -9.58781242e-02
3.13100964e-01 -6.82237327e-01 -8.38443637e-01 2.06606612e-01
5.61951220e-01 -8.66420150e-01 -7.26078629e-01 8.33113790e-01
3.78667504e-01 3.66494179e-01 3.54913294e-01 -9.10230279e-01
-9.29362774e-01 5.06286681e-01 -7.16768265e-01 5.74725509e-01
4.62528557e-01 4.53008085e-01 5.76475143e-01 -9.44404721e-01
7.48908937e-01 1.09595835e+00 7.09288597e-01 4.69219595e-01
-1.27785850e+00 -5.11975348e-01 1.33619189e-01 5.62801182e-01
-1.23007286e+00 -7.52034903e-01 5.72418630e-01 -4.85164911e-01
6.81994677e-01 2.13912025e-01 3.93599242e-01 1.16357291e+00
1.16466455e-01 6.14219666e-01 8.25887203e-01 -2.30499044e-01
2.66732395e-01 -5.30879796e-01 7.28051439e-02 2.30825171e-01
-2.48090178e-01 3.44771326e-01 -6.11357570e-01 3.42020877e-02
1.04313016e+00 1.73550844e-02 -7.73412228e-01 1.28338844e-01
-1.55918574e+00 3.31877708e-01 3.99435937e-01 1.76972985e-01
-6.02711737e-01 3.29521537e-01 5.74794114e-01 2.93988317e-01
6.48658454e-01 -4.06537354e-02 -1.45515367e-01 -1.24408461e-01
-1.32774031e+00 2.94728905e-01 4.63090807e-01 8.30979884e-01
6.19524062e-01 9.30427313e-02 -5.48162699e-01 6.08972073e-01
-1.13260731e-01 8.69549215e-02 3.95080864e-01 -1.27529478e+00
4.15671766e-01 -1.41788498e-01 6.36687100e-01 -7.75329113e-01
9.76637080e-02 -1.83878379e-04 -1.09791732e+00 1.29795834e-01
6.11586750e-01 -2.36515149e-01 -9.26596642e-01 1.69648385e+00
1.66993767e-01 1.07747865e+00 -8.61080438e-02 1.35877275e+00
6.59497797e-01 9.31187212e-01 2.43104473e-02 -3.83907795e-01
1.25272393e+00 -1.15250945e+00 -9.80696917e-01 -9.77561846e-02
6.92306906e-02 -8.81426692e-01 8.55295181e-01 3.75810504e-01
-1.50103176e+00 -8.12263608e-01 -9.08619225e-01 -4.49003696e-01
1.65932938e-01 4.23232839e-02 3.28568667e-01 1.37621447e-01
-1.17735291e+00 6.11003518e-01 -9.21705127e-01 -1.99418738e-02
2.01775521e-01 3.54227312e-02 -1.55559838e-01 -2.97282368e-01
-1.39995742e+00 6.41741395e-01 2.58778363e-01 3.84468168e-01
-8.12171876e-01 -8.92541707e-01 -8.31069112e-01 5.49365655e-02
2.48726889e-01 -9.01766598e-01 1.19690406e+00 -1.33746743e+00
-1.38176692e+00 4.07370329e-01 -5.68831682e-01 -7.36011982e-01
8.65004659e-01 -4.32889968e-01 -3.54448885e-01 2.00141698e-01
-1.06636062e-01 6.65422440e-01 1.24810016e+00 -9.66952443e-01
-6.35660946e-01 1.45127147e-01 1.81466311e-01 1.96481928e-01
4.79590744e-02 2.08536297e-01 -7.25134730e-01 -1.05097675e+00
-2.33589694e-01 -7.73092151e-01 -2.22827613e-01 9.56186093e-03
-1.48454979e-01 1.61037847e-01 1.02533543e+00 -9.68537033e-01
1.49657393e+00 -2.51046610e+00 3.60924244e-01 -4.55760121e-01
7.15509728e-02 2.15910986e-01 -1.44803405e-01 1.98769495e-01
-2.82496780e-01 -3.76545563e-02 -4.06826228e-01 -3.67846251e-01
-3.85654241e-01 1.04282714e-01 -7.07066655e-01 6.19849324e-01
1.57802537e-01 8.32534611e-01 -1.09616745e+00 -2.75155365e-01
4.08296317e-01 8.39089453e-01 -5.87047994e-01 3.17568481e-01
-1.73767999e-01 8.71585369e-01 2.61281300e-02 2.45003104e-01
8.19145083e-01 -4.23364371e-01 -1.59204647e-01 -5.26405931e-01
-3.41610193e-01 2.95424640e-01 -1.22855687e+00 1.81108546e+00
-5.60660064e-01 9.62138653e-01 1.19799897e-02 -4.21960950e-01
1.62697971e-01 6.20122671e-01 4.53250855e-01 -3.02235872e-01
-2.06589684e-01 -1.44071905e-02 -2.79943317e-01 -5.92003405e-01
9.00995672e-01 2.79008985e-01 4.69202667e-01 4.07790273e-01
-2.16981485e-01 1.96904346e-01 3.51797998e-01 1.37578011e-01
9.74263251e-01 5.25597632e-01 -2.78384564e-03 2.38068979e-02
5.33303618e-01 -3.37826669e-01 5.18916130e-01 6.17429972e-01
-1.34514943e-01 1.18847024e+00 2.30027035e-01 -5.53284526e-01
-1.12517107e+00 -9.89103556e-01 -4.49988917e-02 9.79508400e-01
4.51882064e-01 -2.32995212e-01 -7.35020638e-01 -3.49747151e-01
-4.56535965e-01 6.36640429e-01 -4.67775762e-01 8.00893903e-02
-9.23005283e-01 -6.78995788e-01 2.77039886e-01 4.90507692e-01
6.47873282e-01 -8.66396844e-01 -4.49562371e-01 3.91606301e-01
-6.34476781e-01 -1.40477300e+00 -1.15702868e+00 -3.00553799e-01
-6.73442006e-01 -8.75620663e-01 -9.50595498e-01 -6.86108947e-01
5.24520934e-01 5.45607805e-01 1.13574135e+00 4.79107872e-02
1.92938168e-02 6.91113994e-02 -2.38635615e-01 4.39174801e-01
-3.67273390e-01 -2.30727449e-01 -2.88563013e-01 3.69202256e-01
-1.11252524e-01 -5.88084638e-01 -1.07186580e+00 3.73287499e-01
-1.29307687e+00 4.44202214e-01 9.85585153e-02 9.44528282e-01
3.21874589e-01 -4.54815887e-02 4.26369458e-01 -5.77202022e-01
4.37076509e-01 -6.22559011e-01 -4.69439924e-01 1.37592494e-01
-7.30582327e-02 -3.16499248e-02 6.29406214e-01 -7.14087129e-01
-1.29729211e+00 2.76882797e-02 -2.01723814e-01 -8.66376817e-01
-9.55113322e-02 3.00802737e-01 1.85603246e-01 7.15793148e-02
5.51995099e-01 1.89927116e-01 -3.83602917e-01 -3.01712692e-01
4.19724077e-01 4.36757714e-01 1.03285217e+00 -3.37095559e-01
5.93417048e-01 6.49756193e-01 -1.41554087e-01 -6.38445616e-01
-4.83711094e-01 -2.32934371e-01 -4.33869243e-01 -2.87879705e-01
8.58145237e-01 -1.28571677e+00 -5.64316988e-01 7.56459355e-01
-1.49784029e+00 -5.88838220e-01 -2.13111073e-01 4.79335815e-01
-4.30955410e-01 7.16068327e-01 -1.09347284e+00 -5.29696405e-01
-1.83533877e-01 -1.33038890e+00 8.64065647e-01 5.99014089e-02
-1.48564517e-01 -8.81976902e-01 -1.28060967e-01 5.76574169e-02
6.36488199e-01 1.26844600e-01 3.98638189e-01 1.72157586e-01
-1.00772238e+00 5.12208194e-02 -5.08767486e-01 2.69216746e-01
4.35620993e-01 1.20285533e-01 -9.60602582e-01 -1.64624140e-01
8.87618214e-02 -1.12166908e-02 1.23554695e+00 7.17300713e-01
1.25887382e+00 -5.23040295e-01 4.64932844e-02 8.25166464e-01
1.33006442e+00 -1.01264134e-01 9.87227201e-01 8.04801956e-02
9.14739430e-01 3.10526341e-01 3.62320930e-01 5.08786440e-01
5.09382188e-01 8.59497726e-01 3.29357713e-01 -1.19212866e-01
-5.08474469e-01 -2.11166907e-02 6.20013416e-01 6.39803290e-01
-3.27335894e-01 -3.42696935e-01 -4.50366944e-01 6.77775860e-01
-2.05406594e+00 -1.27245533e+00 -2.84346163e-01 2.27230692e+00
1.16816163e+00 -2.25722909e-01 3.07709817e-02 -5.26302643e-02
9.07337308e-01 4.38178033e-01 -4.68208790e-01 -1.14962064e-01
-2.84599364e-01 -5.19405492e-02 5.08726120e-01 8.88046741e-01
-1.14757478e+00 7.70962715e-01 6.05893087e+00 6.70877576e-01
-1.33416259e+00 2.23817557e-01 9.34861958e-01 -1.43558398e-01
-1.85788780e-01 2.13565864e-02 -4.72068638e-01 9.02480125e-01
9.86888409e-01 3.39327939e-02 8.67933512e-01 2.47635499e-01
6.80032194e-01 -2.27645099e-01 -1.32050467e+00 8.61677766e-01
-2.25795016e-01 -1.54588759e+00 -4.48018871e-02 -4.72525865e-01
8.80055606e-01 9.55515075e-03 1.58617556e-01 -8.61407593e-02
2.07871035e-01 -8.51271629e-01 9.51362967e-01 6.70858443e-01
8.95809352e-01 -3.66859287e-01 3.63779992e-01 8.86169225e-02
-1.27996898e+00 1.36215210e-01 3.99919190e-02 -1.78120241e-01
5.30987740e-01 6.18589878e-01 -2.93490708e-01 6.05200887e-01
7.63101399e-01 1.08521807e+00 -3.80748153e-01 1.05843174e+00
-1.66709125e-01 6.23882413e-01 -1.35397151e-01 8.24868500e-01
1.51485130e-02 -1.14172295e-01 5.04654765e-01 1.44656444e+00
3.82158905e-01 2.00909346e-01 -1.11186229e-01 8.29305291e-01
-1.04618795e-01 -4.62024808e-01 -8.91409218e-02 3.71734053e-01
4.00276095e-01 8.91510427e-01 -3.65790129e-01 -5.09753227e-01
-7.29351461e-01 1.34178758e+00 -1.05880067e-01 8.59669685e-01
-1.15668857e+00 -1.37367800e-01 8.27475727e-01 4.07403288e-03
5.59630334e-01 -3.47023964e-01 -2.77521372e-01 -1.63740039e+00
1.03825517e-01 -7.08455801e-01 1.89536378e-01 -9.98192489e-01
-1.17290986e+00 6.23207867e-01 -4.81459089e-02 -1.33043575e+00
-5.55163503e-01 2.13339720e-02 -5.61225057e-01 1.15379810e+00
-1.86890757e+00 -8.77072215e-01 -4.66415346e-01 6.88540161e-01
9.75758612e-01 5.77915549e-01 1.87028229e-01 5.52388012e-01
-4.47311193e-01 1.80038512e-01 2.81891767e-02 9.08862054e-02
1.12667584e+00 -8.48830104e-01 7.37755656e-01 1.35700309e+00
-2.05927312e-01 4.30347502e-01 9.35083807e-01 -5.95771253e-01
-1.46126425e+00 -1.35380292e+00 7.91871965e-01 -2.21772566e-01
6.00143015e-01 -4.45931777e-02 -1.27056468e+00 7.45889425e-01
6.39262438e-01 5.14395356e-01 -1.09225912e-02 -5.94635248e-01
-3.12000632e-01 -7.03636408e-02 -8.68291259e-01 7.35706151e-01
9.12077785e-01 -5.57411253e-01 -2.12936908e-01 1.83405742e-01
9.57988739e-01 -8.84203672e-01 -7.09023118e-01 2.92253584e-01
4.20864195e-01 -1.20977032e+00 1.26948154e+00 -1.48967177e-01
8.04626524e-01 -6.63891554e-01 1.01583555e-01 -9.97922719e-01
-5.06766260e-01 -8.09089243e-01 -4.18333918e-01 1.13553739e+00
-2.33055353e-02 -2.78491050e-01 3.60716432e-01 7.26628244e-01
-4.75491136e-02 -3.83717775e-01 -7.99343824e-01 -4.17924643e-01
-2.87529439e-01 -3.46918404e-01 4.74433035e-01 9.85871494e-01
-4.53772634e-01 2.23501641e-02 -9.46815968e-01 3.83035988e-01
6.24293208e-01 1.03919610e-01 4.47092384e-01 -5.10584772e-01
-4.14132029e-01 -2.64445066e-01 4.76248637e-02 -1.50156450e+00
1.23390339e-01 -2.87570447e-01 1.80193037e-01 -1.25394475e+00
1.39365885e-02 -8.22708011e-02 -9.21723694e-02 2.01774597e-01
-5.50999343e-01 4.77791131e-01 1.09245852e-01 5.04837871e-01
-4.32584465e-01 2.83137977e-01 1.25968325e+00 2.34070141e-02
-3.49995255e-01 -1.87410668e-01 -2.18001112e-01 5.48286676e-01
3.46415013e-01 -2.77410336e-02 -2.33514711e-01 -8.92911732e-01
7.14673027e-02 6.28000796e-01 7.56736219e-01 -7.76513159e-01
5.11180997e-01 -2.63497323e-01 4.41086292e-01 -3.87030452e-01
3.20754886e-01 -7.98250377e-01 5.90441346e-01 1.84812069e-01
-5.94813704e-01 8.46605673e-02 1.31107837e-01 6.25933826e-01
-2.76207715e-01 -3.17133702e-02 8.41073632e-01 2.41759531e-02
-8.19106519e-01 4.86912668e-01 -3.20801795e-01 -1.28414622e-02
6.88583851e-01 -1.70055449e-01 -3.13802958e-01 -5.66553950e-01
-6.70839429e-01 -4.46285047e-02 7.45743394e-01 3.93144131e-01
5.36915660e-01 -1.32479453e+00 -8.36648703e-01 9.63279754e-02
-4.50729966e-01 -3.64057645e-02 5.19704342e-01 9.92917717e-01
-4.73045975e-01 -3.00899372e-02 -1.46028936e-01 -5.81255317e-01
-9.44935739e-01 8.01715434e-01 2.27733269e-01 -5.88657036e-02
-7.68979490e-01 5.60239077e-01 4.61984634e-01 5.84442496e-01
2.88476288e-01 -5.81120968e-01 9.60528255e-02 -1.12253591e-01
8.55611026e-01 4.18765366e-01 -4.10688221e-02 -6.34790719e-01
1.93350296e-02 3.81104529e-01 -8.65108594e-02 -1.07458413e-01
1.07767415e+00 -7.00509727e-01 -1.58825740e-01 4.22784239e-01
1.04142392e+00 -7.78646693e-02 -1.90815079e+00 -4.10070419e-01
-2.93017536e-01 -6.97780669e-01 1.20872177e-01 -5.62793434e-01
-1.22944486e+00 6.29648328e-01 4.50848192e-01 9.50136259e-02
1.60052466e+00 -3.89835656e-01 1.07250178e+00 -2.22427875e-01
9.42210332e-02 -6.47487044e-01 -1.02054477e-01 5.13447404e-01
8.99619997e-01 -1.12615299e+00 -1.05348052e-02 -3.69926751e-01
-4.25779164e-01 1.15254331e+00 2.84788847e-01 -2.44912773e-01
3.94281238e-01 4.40936029e-01 -1.19335853e-01 5.95717669e-01
-9.36313689e-01 1.84505835e-01 2.89201826e-01 2.37426862e-01
5.58015883e-01 -4.23913836e-01 -2.75921911e-01 2.59894103e-01
3.04952264e-01 6.26576483e-01 6.75999224e-01 6.35351181e-01
-5.35844602e-02 -8.57534707e-01 -5.20378172e-01 -9.82517973e-02
-6.01164997e-01 -3.86308163e-01 2.15680271e-01 2.57474929e-01
8.99916217e-02 8.31771255e-01 2.02315047e-01 -3.51917334e-02
4.10332680e-02 -2.34965101e-01 4.15356249e-01 -1.47041045e-02
-7.33735442e-01 2.65627176e-01 -9.51368362e-02 -7.78769910e-01
-7.24347711e-01 -6.62368476e-01 -9.32613432e-01 -5.52447557e-01
-6.88236579e-02 -1.70426846e-01 2.65694201e-01 8.54207635e-01
5.70475757e-01 6.38285697e-01 5.72248101e-01 -1.36299229e+00
-8.76023918e-02 -7.52639830e-01 -1.02449074e-01 4.66859132e-01
1.08716106e+00 -1.99203029e-01 -4.78229046e-01 8.44460487e-01]
|
[11.190403938293457, -2.167649269104004]
|
94a4be13-c8b3-4a67-bfe1-5a652ae52da7
|
classification-and-self-supervised-regression
|
2210.14253
| null |
https://arxiv.org/abs/2210.14253v1
|
https://arxiv.org/pdf/2210.14253v1.pdf
|
Classification and Self-Supervised Regression of Arrhythmic ECG Signals Using Convolutional Neural Networks
|
Interpretation of electrocardiography (ECG) signals is required for diagnosing cardiac arrhythmia. Recently, machine learning techniques have been applied for automated computer-aided diagnosis. Machine learning tasks can be divided into regression and classification. Regression can be used for noise and artifacts removal as well as resolve issues of missing data from low sampling frequency. Classification task concerns the prediction of output diagnostic classes according to expert-labeled input classes. In this work, we propose a deep neural network model capable of solving regression and classification tasks. Moreover, we combined the two approaches, using unlabeled and labeled data, to train the model. We tested the model on the MIT-BIH Arrhythmia database. Our method showed high effectiveness in detecting cardiac arrhythmia based on modified Lead II ECG records, as well as achieved high quality of ECG signal approximation. For the former, our method attained overall accuracy of 87:33% and balanced accuracy of 80:54%, on par with reference approaches. For the latter, application of self-supervised learning allowed for training without the need for expert labels. The regression model yielded satisfactory performance with fairly accurate prediction of QRS complexes. Transferring knowledge from regression to the classification task, our method attained higher overall accuracy of 87:78%.
|
['Ru-San Tan', 'U Rajendra Acharya', 'Özal Yıldırım', 'Paweł Pławiak', 'Wojciech Masarczyk', 'Przemysław Głomb', 'Bartosz Grabowski']
|
2022-10-25
| null | null | null | null |
['electrocardiography-ecg']
|
['methodology']
|
[ 3.01575273e-01 4.03111018e-02 1.29498959e-01 -6.02624238e-01
-1.13496971e+00 -3.34862500e-01 -3.56711805e-01 1.75805658e-01
-2.63243794e-01 1.01100397e+00 -3.22307646e-01 -4.95694816e-01
-3.08659136e-01 -3.19191396e-01 -1.24639675e-01 -7.38354862e-01
-1.32104933e-01 4.46504444e-01 -5.22158682e-01 1.50140777e-01
-8.53558944e-04 5.57802677e-01 -1.09708595e+00 4.27152872e-01
9.76478279e-01 1.26905024e+00 -1.81081653e-01 8.27664733e-01
3.62712055e-01 9.77778316e-01 -9.07659411e-01 4.19081934e-02
-4.02325206e-02 -8.00620556e-01 -5.16020000e-01 -2.92216204e-02
-2.77211428e-01 -1.38574585e-01 1.83876902e-02 3.82582337e-01
1.09035611e+00 -3.55414212e-01 9.11612213e-01 -8.24443936e-01
-2.90180713e-01 4.73685890e-01 -5.78932166e-01 4.68692750e-01
1.14591174e-01 -2.28831321e-01 5.53732634e-01 -9.25673127e-01
2.31587186e-01 2.88807571e-01 1.09288681e+00 5.16147554e-01
-1.33076894e+00 -7.25887239e-01 -4.86043453e-01 3.23111087e-01
-1.60806215e+00 -4.11786199e-01 9.52237725e-01 -6.24181032e-01
7.95855224e-01 2.32738331e-01 7.59371161e-01 7.07897782e-01
1.51231050e-01 4.17424679e-01 9.98166561e-01 -7.08180845e-01
1.85794935e-01 3.74901295e-01 3.98364305e-01 4.69233215e-01
1.90390706e-01 4.54361923e-02 -2.37475112e-01 -3.71237040e-01
7.20061779e-01 -5.09133562e-02 -4.46331233e-01 1.34414285e-01
-1.17608678e+00 5.75691819e-01 1.09077156e-01 5.26571989e-01
-6.50068164e-01 -2.48331919e-01 5.88544965e-01 4.56567377e-01
5.54212451e-01 6.87307417e-01 -7.74601638e-01 -7.84124285e-02
-1.04300523e+00 -5.71765304e-02 6.81401372e-01 5.24043679e-01
2.07164466e-01 5.71293771e-01 -2.24057063e-01 1.03780007e+00
9.61541757e-02 5.06277561e-01 6.11720502e-01 -8.62719119e-01
2.09278539e-01 4.26397264e-01 8.24824274e-02 -9.49667096e-01
-7.77863681e-01 -9.30168927e-01 -1.25701880e+00 -2.83134058e-02
4.04815853e-01 -5.48793495e-01 -7.28696704e-01 1.44198024e+00
-4.21815142e-02 1.73183098e-01 3.22452813e-01 7.98823655e-01
8.79046977e-01 4.62635785e-01 2.70201080e-02 -8.76800239e-01
1.12177324e+00 -3.73930484e-01 -1.07513177e+00 2.02438354e-01
7.83445001e-01 -4.63851720e-01 5.27474463e-01 8.64050090e-01
-9.67662513e-01 -7.67375231e-01 -9.96491492e-01 4.65298235e-01
2.50500053e-01 6.38821483e-01 1.32531807e-01 7.93820620e-01
-9.31904435e-01 7.96395242e-01 -7.29826331e-01 -1.04010664e-02
6.20959461e-01 6.34240627e-01 -2.07884327e-01 3.49505663e-01
-1.18319321e+00 9.20293510e-01 1.99708864e-01 4.88983572e-01
-6.03450477e-01 -5.49558580e-01 -5.89765787e-01 -2.48263534e-02
-1.41360804e-01 -5.79906881e-01 1.07597697e+00 -1.03297687e+00
-1.30194402e+00 8.81301343e-01 -2.03367278e-01 -6.03839755e-01
3.72484744e-01 -1.96920127e-01 -7.06066787e-01 2.78170973e-01
-1.55795574e-01 1.23665206e-01 9.78726327e-01 -1.00053155e+00
-5.54847658e-01 -6.07311010e-01 -6.92253709e-01 -2.85503757e-03
-2.02417433e-01 -1.62813708e-01 1.65726766e-01 -6.22155249e-01
4.41416830e-01 -5.63035607e-01 -1.31810009e-01 -5.19622028e-01
-1.95791528e-01 -7.57089257e-02 6.06429935e-01 -1.04380441e+00
1.51755083e+00 -2.24524188e+00 -3.72677594e-02 4.21005160e-01
5.60833871e-01 5.57027280e-01 2.74813086e-01 2.20882148e-01
-5.17575085e-01 7.20920712e-02 -4.37210143e-01 9.01460201e-02
-4.81073499e-01 -1.74263082e-02 -3.28475475e-01 5.44786990e-01
3.90986890e-01 9.49860632e-01 -5.41121900e-01 -4.68791932e-01
3.68483931e-01 5.34459114e-01 -1.54622748e-01 4.08399940e-01
3.55317295e-01 9.27919447e-01 -2.90433139e-01 6.13707066e-01
3.01199198e-01 -2.69616693e-01 3.45955908e-01 -2.73611337e-01
3.07991743e-01 -4.98680733e-02 -9.57633436e-01 1.44865227e+00
-2.71191627e-01 6.80819869e-01 -3.10829252e-01 -1.48142421e+00
1.45680904e+00 8.88525784e-01 7.70934403e-01 -4.76480544e-01
3.48502517e-01 4.34082508e-01 3.48842204e-01 -9.25719023e-01
-3.03830862e-01 -2.51912713e-01 1.06575899e-01 3.87751579e-01
-8.38259608e-02 2.18761135e-02 -3.22738439e-01 -3.99627090e-01
9.38468635e-01 1.39787957e-01 5.63870847e-01 -2.46204391e-01
6.52148724e-01 -6.57769814e-02 6.48050606e-01 7.86302984e-01
-2.23862439e-01 7.26020098e-01 2.80344605e-01 -9.07270551e-01
-7.05265582e-01 -6.76902950e-01 -3.97963852e-01 3.76641303e-01
-1.83991581e-01 -1.91577241e-01 -7.36590326e-01 -5.32703459e-01
-2.71390200e-01 3.65890115e-01 -3.87486756e-01 -2.77633160e-01
-7.77783573e-01 -1.04504085e+00 8.58798265e-01 7.98805535e-01
1.63564026e-01 -1.29105568e+00 -8.11677337e-01 5.18405139e-01
-3.09451908e-01 -8.22071075e-01 2.70266056e-01 5.16003013e-01
-1.27323198e+00 -1.20686090e+00 -7.07468808e-01 -9.10523355e-01
5.75198174e-01 -3.49690586e-01 1.08584213e+00 2.68092722e-01
-6.44610465e-01 1.09934464e-01 -2.56418318e-01 -7.55155981e-01
-4.03231949e-01 -2.04102732e-02 1.17851183e-01 2.22940564e-01
3.34710956e-01 -6.82811260e-01 -7.15676188e-01 3.06963980e-01
-4.15795088e-01 -1.88170910e-01 4.26329195e-01 1.10629046e+00
5.09194195e-01 -1.08796872e-01 1.47123957e+00 -9.29834485e-01
7.94035316e-01 -2.43388221e-01 -1.24962337e-01 -5.58062531e-02
-8.89035225e-01 -4.07891542e-01 7.59534895e-01 -2.81352133e-01
-6.32140100e-01 2.00682327e-01 -4.25943941e-01 -3.15466642e-01
-2.34182388e-01 4.89387929e-01 8.00453126e-02 1.28875434e-01
9.47977662e-01 1.05340712e-01 -7.71373743e-03 -2.01013505e-01
-3.46053451e-01 1.14110231e+00 5.06829381e-01 -2.56853372e-01
3.26036543e-01 1.28363311e-01 -9.52995103e-03 -7.25139856e-01
-7.44672120e-01 -3.36767465e-01 -7.87417531e-01 -1.98007271e-01
7.75694430e-01 -8.84511173e-01 -6.00685596e-01 4.60303158e-01
-1.05869329e+00 -1.00400545e-01 -4.08796340e-01 7.10758209e-01
-4.66972262e-01 3.53222728e-01 -4.76092786e-01 -1.16196036e+00
-8.53933215e-01 -8.62350881e-01 7.27015018e-01 -2.26432502e-01
-7.05081999e-01 -9.07184482e-01 -5.38883023e-02 2.54499286e-01
2.96371400e-01 6.70994341e-01 1.08159828e+00 -1.06331825e+00
1.41667590e-01 -6.04826748e-01 7.16115581e-03 5.87482035e-01
3.74432504e-01 -2.58989692e-01 -1.39713681e+00 -2.88877308e-01
3.49657774e-01 -1.52480021e-01 4.88818288e-01 5.51574945e-01
1.35622287e+00 1.85734257e-01 -1.75786451e-01 5.36715269e-01
1.12184811e+00 7.75780141e-01 7.14428663e-01 -6.00471795e-02
5.11273503e-01 4.33338106e-01 5.32945454e-01 4.94112939e-01
9.60592180e-02 3.75712126e-01 1.27420370e-02 -5.85795522e-01
2.25578338e-01 2.67176688e-01 -2.44204655e-01 9.01483774e-01
-3.23843390e-01 6.52581900e-02 -1.16775382e+00 2.66583204e-01
-1.87551248e+00 -6.68698192e-01 -4.65630412e-01 2.24760509e+00
6.35597825e-01 1.17624089e-01 1.49737909e-01 1.04026031e+00
7.81273901e-01 -4.91980851e-01 -7.05073535e-01 -2.92032033e-01
3.65620642e-03 4.47569609e-01 -1.40618175e-01 1.82715908e-01
-1.04459012e+00 2.38843396e-01 6.68216658e+00 3.93622935e-01
-1.34629321e+00 6.06602021e-02 9.58631516e-01 2.33892038e-01
3.20558071e-01 -5.10169148e-01 -2.78530210e-01 2.63405502e-01
1.14792895e+00 1.79259375e-01 1.24539226e-01 4.19519901e-01
4.20387983e-01 2.19229251e-01 -1.16484153e+00 1.49307334e+00
1.21155597e-01 -1.09531474e+00 -3.80295485e-01 -2.50577301e-01
4.25472111e-01 -4.15775150e-01 -3.74663204e-01 2.75286555e-01
-6.92483246e-01 -1.14556885e+00 2.52679408e-01 7.12806523e-01
1.10572755e+00 -8.48970652e-01 1.18545341e+00 5.63658178e-01
-7.42103696e-01 -3.02806348e-01 2.87408121e-02 -2.96382338e-01
-1.83908448e-01 7.00551689e-01 -9.64144647e-01 6.00922763e-01
5.65210640e-01 1.02975404e+00 -2.37697154e-01 8.64406586e-01
-1.47458732e-01 1.02926397e+00 4.32065083e-03 3.27104926e-01
-4.68231916e-01 -1.92348063e-01 3.04303676e-01 1.06177127e+00
1.77361280e-01 3.31722409e-01 1.16914973e-01 5.64300895e-01
2.28711650e-01 3.51483107e-01 -5.42756319e-01 2.04710886e-01
3.22013438e-01 1.06589568e+00 -6.80008888e-01 -4.68403041e-01
2.65248623e-02 5.87071061e-01 1.67323023e-01 3.95677537e-01
-8.85598361e-01 -6.82732165e-01 -1.45372972e-01 1.92377731e-01
-5.26427738e-02 2.51531601e-01 -9.32840168e-01 -9.90193009e-01
1.31180093e-01 -9.48367894e-01 3.41914028e-01 -6.65936708e-01
-1.05335689e+00 1.14822519e+00 -2.92278737e-01 -1.45231664e+00
-5.50940514e-01 -3.58828366e-01 -4.10204768e-01 1.03210378e+00
-1.23301482e+00 -6.47482872e-01 -3.70572209e-01 4.63842452e-01
3.88295650e-01 -5.40656388e-01 1.36904299e+00 6.22131467e-01
-5.83008111e-01 5.50329864e-01 -8.32163915e-02 1.61369935e-01
7.42368877e-01 -1.27950251e+00 -1.10209778e-01 4.49300885e-01
1.62605181e-01 3.99947822e-01 4.20409292e-01 -3.00780505e-01
-9.56597388e-01 -1.07729232e+00 1.00666261e+00 -3.30820262e-01
1.90492868e-02 9.25562680e-02 -9.99312282e-01 2.33257160e-01
-7.87066743e-02 1.01669796e-01 1.09941816e+00 -7.84169436e-02
1.50078535e-01 -4.86116856e-01 -1.06740463e+00 1.33846611e-01
5.08613586e-01 -3.95411968e-01 -5.84000707e-01 2.02969417e-01
6.31996393e-02 -4.76261318e-01 -1.07751071e+00 8.87443304e-01
7.61570275e-01 -8.07208776e-01 6.70430779e-01 -5.98599792e-01
3.86065423e-01 -2.62736738e-01 1.71006933e-01 -1.14666843e+00
-4.10710186e-01 -5.40707052e-01 -2.49300078e-01 8.12762201e-01
6.40759408e-01 -6.38906837e-01 5.75699091e-01 3.38786960e-01
-1.62510931e-01 -1.13630760e+00 -6.01399183e-01 -1.48350775e-01
-3.55468571e-01 -5.05156517e-01 1.44887209e-01 9.56934452e-01
-9.37171727e-02 5.52806735e-01 -5.79209685e-01 8.23967084e-02
4.78340417e-01 -1.44051533e-04 3.85148048e-01 -1.50383067e+00
-3.34467500e-01 -4.24639434e-02 -4.54258084e-01 -2.45838940e-01
3.57849337e-02 -8.85268867e-01 -1.80771649e-01 -1.43566918e+00
-8.77602026e-02 -4.75327283e-01 -7.62623429e-01 4.56827044e-01
-2.46817172e-01 8.63354564e-01 -2.31570721e-01 4.05404717e-01
-2.09756091e-01 -2.59379670e-02 9.61910725e-01 -1.16676101e-02
-5.64731777e-01 5.51887631e-01 -8.30175638e-01 8.57446194e-01
1.13111770e+00 -5.10755897e-01 -3.79420489e-01 -1.99300870e-01
3.54451165e-02 6.31666064e-01 1.62041128e-01 -1.07660949e+00
-1.70811281e-01 5.92147887e-01 8.91917109e-01 -5.42978704e-01
2.50609159e-01 -9.77241874e-01 4.39776391e-01 6.97203100e-01
-4.18049097e-01 -4.26668227e-02 2.29298696e-01 4.27187324e-01
-3.61668617e-01 -1.17184855e-01 7.29134142e-01 1.16550334e-01
-1.52332038e-01 -4.12416197e-02 -6.26122952e-01 -7.91885406e-02
1.06899452e+00 -4.97921318e-01 3.01406473e-01 -2.95994669e-01
-1.40872431e+00 -5.21026254e-02 -4.32303369e-01 9.16855633e-02
9.36752856e-01 -1.08031213e+00 -9.25699294e-01 5.14309764e-01
8.77230242e-02 -9.07283872e-02 1.90501377e-01 1.13601553e+00
-6.80562615e-01 2.52338678e-01 -1.24840491e-01 -9.78951037e-01
-1.35612321e+00 2.38825932e-01 6.00087821e-01 -1.79230958e-01
-7.31946409e-01 3.82322460e-01 -2.10772440e-01 -1.32580504e-01
4.97775078e-01 -2.42093608e-01 -5.63665152e-01 8.10710117e-02
5.85162640e-01 4.52938050e-01 5.06836951e-01 -2.30891883e-01
-3.32291335e-01 6.76319242e-01 1.29263178e-01 3.37132335e-01
1.32611060e+00 -1.91617273e-02 -7.81934783e-02 7.41668463e-01
8.47304642e-01 -3.30399066e-01 -5.23284674e-01 -2.83693448e-02
1.25602439e-01 -6.51541352e-02 1.56463265e-01 -1.20965528e+00
-9.72993672e-01 1.10512471e+00 9.42771673e-01 1.92977101e-01
1.56108701e+00 -5.55027783e-01 4.91217464e-01 4.53514546e-01
2.86325723e-01 -7.98626840e-01 -3.06422710e-01 -3.88112031e-02
6.00913405e-01 -1.25319755e+00 -1.62449311e-02 -2.02064380e-01
-7.41360784e-01 1.36013186e+00 2.84255385e-01 -2.74833918e-01
6.35752916e-01 2.08063081e-01 5.76291680e-01 -1.12062298e-01
-5.85215151e-01 3.06955129e-01 4.43644106e-01 6.78483367e-01
7.73878098e-01 5.62710203e-02 -5.03504872e-01 1.01141143e+00
1.70944616e-01 4.86035407e-01 2.91023761e-01 9.42622364e-01
-1.90162256e-01 -8.29598606e-01 -3.60588104e-01 8.36355567e-01
-8.66548181e-01 -4.22936603e-02 -1.82288006e-01 5.45776844e-01
1.59627631e-01 1.19242942e+00 -1.59691706e-01 -3.88555408e-01
3.81841302e-01 3.42525184e-01 3.64807814e-01 -5.50418019e-01
-9.47425961e-01 4.13348287e-01 7.83312619e-02 -1.01251462e-02
-3.81604582e-01 -3.04759622e-01 -1.09424186e+00 4.85679775e-01
-5.21639585e-01 5.38517416e-01 5.47426760e-01 9.50345039e-01
3.75129551e-01 1.01397192e+00 8.86070311e-01 -5.03061414e-01
-6.51521266e-01 -1.17566133e+00 -9.37781692e-01 1.92794338e-01
6.10997438e-01 -1.49967030e-01 -1.50879472e-01 5.24859726e-01]
|
[14.311202049255371, 3.298381805419922]
|
3ba75bff-e4ee-4f52-822b-00e0dbb8a2f4
|
compilergym-robust-performant-compiler
|
2109.08267
| null |
https://arxiv.org/abs/2109.08267v2
|
https://arxiv.org/pdf/2109.08267v2.pdf
|
CompilerGym: Robust, Performant Compiler Optimization Environments for AI Research
|
Interest in applying Artificial Intelligence (AI) techniques to compiler optimizations is increasing rapidly, but compiler research has a high entry barrier. Unlike in other domains, compiler and AI researchers do not have access to the datasets and frameworks that enable fast iteration and development of ideas, and getting started requires a significant engineering investment. What is needed is an easy, reusable experimental infrastructure for real world compiler optimization tasks that can serve as a common benchmark for comparing techniques, and as a platform to accelerate progress in the field. We introduce CompilerGym, a set of environments for real world compiler optimization tasks, and a toolkit for exposing new optimization tasks to compiler researchers. CompilerGym enables anyone to experiment on production compiler optimization problems through an easy-to-use package, regardless of their experience with compilers. We build upon the popular OpenAI Gym interface enabling researchers to interact with compilers using Python and a familiar API. We describe the CompilerGym architecture and implementation, characterize the optimization spaces and computational efficiencies of three included compiler environments, and provide extensive empirical evaluations. Compared to prior works, CompilerGym offers larger datasets and optimization spaces, is 27x more computationally efficient, is fault-tolerant, and capable of detecting reproducibility bugs in the underlying compilers. In making it easy for anyone to experiment with compilers - irrespective of their background - we aim to accelerate progress in the AI and compiler research domains.
|
['Hugh Leather', 'Yuandong Tian', 'Benoit Steiner', 'Olivier Teytaud', 'Jia Liu', 'Somya Jain', 'Sahir Gomez', 'Jason Ansel', 'Brandon Cui', 'Jiadong Guo', 'Bram Wasti', 'Chris Cummins']
|
2021-09-17
| null | null | null | null |
['compiler-optimization']
|
['computer-code']
|
[-3.16821873e-01 -3.11086714e-01 -4.04706627e-01 -3.99561077e-01
-4.29163486e-01 -7.31592298e-01 2.70501167e-01 2.04725668e-01
-2.54783779e-01 3.19793880e-01 1.37360707e-01 -9.72681999e-01
2.65791178e-01 -8.19931567e-01 -6.87083423e-01 -1.88042089e-01
-1.56145230e-01 6.54994965e-01 -2.22104564e-01 -5.64793646e-01
5.51149428e-01 2.16602325e-01 -1.93383980e+00 4.03534323e-02
1.05665743e+00 4.16750044e-01 3.70164424e-01 8.85866761e-01
-3.03964823e-01 5.07806480e-01 -5.15136182e-01 -2.53705978e-01
5.19978344e-01 -1.47748262e-01 -1.20089495e+00 -1.94301844e-01
2.11448342e-01 1.24399237e-01 3.10961306e-01 1.00540709e+00
3.29160303e-01 -7.86933824e-02 4.87319147e-03 -1.38555753e+00
-5.89005291e-01 7.20447838e-01 -1.79315180e-01 4.78432998e-02
5.57511449e-01 7.03963339e-01 1.02465975e+00 -5.36170006e-01
5.68932950e-01 1.00167239e+00 6.80651426e-01 2.47230262e-01
-1.08453810e+00 -4.84191567e-01 -1.87745020e-01 3.15024168e-04
-1.13002229e+00 -6.55730546e-01 4.01964992e-01 -4.55032736e-01
1.44774306e+00 5.79501808e-01 7.50037193e-01 6.52409196e-01
1.96237743e-01 7.60550737e-01 7.06798255e-01 -6.27053916e-01
3.80040646e-01 2.75395155e-01 4.85547423e-01 1.18689001e+00
-1.58179961e-02 2.65945047e-01 -3.37090969e-01 -1.49028555e-01
2.98575222e-01 -1.10091574e-01 -1.01205021e-01 -2.87945896e-01
-1.58035779e+00 8.23247612e-01 4.07958597e-01 3.99046183e-01
-5.12622334e-02 1.55567572e-01 6.92330420e-01 6.44318521e-01
1.57274410e-01 1.30826259e+00 -7.35386729e-01 -8.77534509e-01
-6.61857188e-01 5.39522886e-01 1.34240663e+00 1.11407995e+00
8.34582984e-01 1.74155176e-01 3.79231751e-01 7.31958628e-01
1.53004706e-01 1.04425810e-01 7.58160830e-01 -1.03993022e+00
2.40148544e-01 1.04326975e+00 -3.42226177e-01 -8.79255474e-01
-4.91748273e-01 -3.43146890e-01 -4.97876406e-01 4.40408647e-01
2.55593777e-01 -1.80339575e-01 -5.31300068e-01 1.36075878e+00
3.06997776e-01 -3.14867914e-01 -1.42407818e-02 9.39408004e-01
6.43628359e-01 6.71780169e-01 -3.47023785e-01 3.23273195e-03
1.38583982e+00 -1.28231072e+00 -2.04832643e-01 -5.37653685e-01
1.26797092e+00 -1.04872572e+00 1.66310382e+00 3.22836936e-01
-1.12586629e+00 -3.84952813e-01 -1.20253623e+00 -2.95159519e-01
-7.13055611e-01 -2.31981993e-01 1.49341559e+00 9.16038811e-01
-1.22326541e+00 6.48027599e-01 -1.03422415e+00 -5.80698133e-01
1.94537029e-01 5.24218261e-01 -3.13639283e-01 1.45564880e-02
-5.17762065e-01 1.08728123e+00 4.96546179e-01 -2.10817799e-01
-5.13965607e-01 -1.22591209e+00 -1.07780826e+00 -1.79803625e-01
2.56507218e-01 -9.82272446e-01 1.46040785e+00 -1.03876472e+00
-1.58383334e+00 8.72360289e-01 -1.29733816e-01 -2.29720250e-01
1.32241681e-01 3.95052787e-03 -1.29339576e-01 -7.27250636e-01
9.24318880e-02 4.92762089e-01 4.42751378e-01 -6.05498314e-01
-4.77442920e-01 -1.79845527e-01 1.79333121e-01 2.65875906e-01
-2.14104503e-01 3.52127224e-01 -4.96376038e-01 -3.66108716e-01
-3.09993744e-01 -1.04548275e+00 -3.76416117e-01 -1.93094447e-01
-3.23046267e-01 5.41645810e-02 8.18648636e-01 -4.50025141e-01
1.10444272e+00 -2.17607474e+00 3.53730589e-01 -2.74172444e-02
2.66011000e-01 2.74996579e-01 -4.60457578e-02 3.68869990e-01
-1.28688022e-01 3.97746593e-01 -3.06739211e-01 3.54966260e-02
3.82001042e-01 3.11689526e-01 7.30661005e-02 3.52684349e-01
-1.75801553e-02 8.01051795e-01 -1.01813495e+00 -3.24602425e-01
2.98747867e-01 -1.12523876e-01 -1.17328334e+00 3.13573271e-01
-4.83621866e-01 1.43241763e-01 -3.63389760e-01 9.25988793e-01
3.31439584e-01 -1.42280772e-01 -7.26185441e-02 1.15865901e-01
-5.80853224e-01 5.68979323e-01 -1.20908296e+00 2.09361696e+00
-8.79959285e-01 7.06604898e-01 2.36241102e-01 -1.01177680e+00
7.02407420e-01 4.61771414e-02 7.19149709e-02 -6.05009377e-01
1.63229808e-01 1.98422790e-01 3.06714892e-01 -2.93459177e-01
8.06452751e-01 4.07570392e-01 -4.08354521e-01 9.42658424e-01
-8.40045065e-02 -6.86697006e-01 7.67517745e-01 3.04521974e-02
1.30573845e+00 1.33230900e-02 4.80863959e-01 -6.80205882e-01
3.60944748e-01 4.14220691e-01 3.87152642e-01 6.94054902e-01
1.50117904e-01 1.53060138e-01 4.23783869e-01 -9.27561402e-01
-1.28521621e+00 -8.15416515e-01 -1.34745672e-01 1.62775874e+00
-1.52884379e-01 -9.33601797e-01 -7.56046712e-01 -3.13461244e-01
-1.41464382e-01 8.57130229e-01 -1.94757536e-01 -3.82567719e-02
-5.48117757e-01 -1.02840507e+00 4.05417681e-01 3.12174648e-01
5.68072915e-01 -1.02081871e+00 -7.12681592e-01 1.63522646e-01
1.55455038e-01 -6.46765769e-01 -5.93773961e-01 3.47236425e-01
-8.01829517e-01 -8.22999120e-01 1.36025354e-01 -9.32983398e-01
5.40843904e-01 6.64147828e-03 1.81925809e+00 5.64804137e-01
-8.54380429e-01 2.42352381e-01 -1.41620576e-01 -5.82677901e-01
-9.18696165e-01 5.32524407e-01 -9.72876921e-02 -9.47136164e-01
1.37124315e-01 -7.05481946e-01 -1.11316487e-01 1.91490337e-01
-5.56183100e-01 5.48584342e-01 3.93614352e-01 9.76541221e-01
2.38944128e-01 4.06096503e-03 -8.06113407e-02 -1.06753969e+00
7.34862983e-01 -4.66466963e-01 -1.23424470e+00 8.46800134e-02
-1.09188902e+00 4.23833191e-01 8.13534439e-01 9.80444849e-02
-6.05391800e-01 1.36068612e-01 -2.73414582e-01 5.60355633e-02
-2.49800712e-01 5.99602401e-01 -1.96305275e-01 -3.09510529e-01
8.03367496e-01 -7.47767314e-02 2.37563103e-01 -4.58573967e-01
6.43799841e-01 4.42662984e-01 5.77464759e-01 -1.12026298e+00
8.00527871e-01 -2.09189907e-01 -1.83547974e-01 -7.64199257e-01
-1.85607344e-01 -2.35846758e-01 -3.02405506e-01 3.06335360e-01
5.09510934e-01 -5.08159637e-01 -8.86326492e-01 3.35755825e-01
-9.01518345e-01 -4.49309200e-01 -1.92862466e-01 8.54413956e-02
-4.58486706e-01 -3.01583968e-02 -4.33539033e-01 -1.65260866e-01
-4.90441680e-01 -2.05713081e+00 8.61394525e-01 3.33194017e-01
-5.89327037e-01 -1.19019938e+00 2.33851206e-02 4.05375123e-01
7.82355785e-01 -5.13817109e-02 1.15712523e+00 -5.03113091e-01
-8.77427876e-01 -9.10752863e-02 -9.30269659e-02 4.84714955e-02
-1.23712420e-01 4.08461720e-01 -6.94905519e-01 -8.41359422e-02
-2.53583819e-01 -1.37166306e-01 3.19516689e-01 1.11092716e-01
1.41377532e+00 -3.81869823e-01 -4.37595725e-01 1.19225895e+00
1.24700058e+00 -3.77578512e-02 3.27928662e-01 6.59088016e-01
5.98666966e-01 5.74095920e-02 3.58425051e-01 2.77292728e-01
5.52074432e-01 6.49678409e-01 3.39429319e-01 -3.39934886e-01
1.56297222e-01 2.28547618e-01 4.75111812e-01 1.19826567e+00
2.28563510e-02 3.53016518e-02 -1.31078815e+00 4.37169433e-01
-1.74482703e+00 -4.17294621e-01 -1.00776188e-01 2.17681980e+00
1.06042361e+00 2.42892597e-02 9.71855819e-02 6.86635599e-02
1.52335510e-01 -2.03443453e-01 -3.34200740e-01 -1.34227216e+00
2.95505494e-01 5.90288103e-01 4.73831385e-01 5.92603564e-01
-1.16503215e+00 1.07108462e+00 6.93473530e+00 4.50784743e-01
-1.12833166e+00 -1.11675113e-01 4.52343374e-01 -2.31311936e-02
-2.03451321e-01 3.52323890e-01 -6.04284704e-01 2.90175259e-01
1.19145250e+00 -7.20931649e-01 1.21333456e+00 1.39042640e+00
-1.91617936e-01 -1.20649375e-01 -1.53803885e+00 5.95548034e-01
-2.26792261e-01 -1.75490940e+00 -5.30316114e-01 -5.58250844e-02
5.01786292e-01 3.99223596e-01 7.11526871e-02 5.18079042e-01
8.04457426e-01 -1.16755581e+00 4.58776325e-01 4.31466550e-02
3.72486919e-01 -7.65425682e-01 7.31973290e-01 2.28000641e-01
-1.02650368e+00 -3.93300913e-02 -1.84271440e-01 -7.97220111e-01
-3.84466380e-01 4.42654192e-01 -1.20791519e+00 4.66855407e-01
8.98486435e-01 7.46344388e-01 -9.87356901e-01 8.85183275e-01
1.80991232e-01 4.59870845e-01 -3.37964684e-01 -2.67992288e-01
2.88390834e-02 -2.82474369e-01 4.49766874e-01 1.39245188e+00
6.30030334e-02 -3.08057129e-01 6.10544920e-01 1.07115364e+00
-9.40698981e-02 2.03506067e-01 -5.99096298e-01 -5.51559210e-01
4.95848179e-01 1.36351466e+00 -6.84013367e-01 -1.97437048e-01
-5.45006752e-01 7.48618901e-01 3.84382099e-01 -5.56850769e-02
-8.16313922e-01 -6.72745883e-01 1.17215443e+00 -3.51535201e-01
-1.87208652e-01 -2.66964704e-01 -7.53085673e-01 -1.11894274e+00
-6.74831197e-02 -1.68085158e+00 1.94052830e-01 -4.83457178e-01
-9.58087683e-01 3.31923753e-01 -1.16854668e-01 -5.71703434e-01
-5.49755394e-01 -6.17599845e-01 -6.71086788e-01 9.78284240e-01
-1.01596153e+00 -8.64573717e-01 -2.79410750e-01 1.20970368e-01
4.54719543e-01 -4.03962165e-01 1.19285035e+00 3.34196478e-01
-8.76228273e-01 6.33618653e-01 1.50920190e-02 -1.19132530e-02
3.88792694e-01 -1.45919859e+00 9.92678702e-01 8.64649177e-01
1.65693834e-01 1.03721011e+00 9.18612659e-01 -2.05841944e-01
-2.49775314e+00 -9.02487099e-01 4.90134984e-01 -5.89461565e-01
8.51195276e-01 -5.36168337e-01 -6.67785585e-01 8.53483260e-01
3.26676399e-01 -1.17129043e-01 5.05638421e-01 7.85917640e-01
-3.05424869e-01 -1.66839659e-01 -7.44879544e-01 7.07204580e-01
9.46003616e-01 -3.50333750e-01 -5.02144754e-01 7.88363755e-01
7.77034521e-01 -8.89354825e-01 -1.11680317e+00 7.83308037e-03
4.13243055e-01 -9.69746709e-01 9.87700343e-01 -3.62382084e-01
4.88926917e-01 -4.05929416e-01 -8.09854046e-02 -1.44356298e+00
-1.67528152e-01 -8.41359794e-01 3.48302037e-01 1.20497012e+00
7.01002300e-01 -8.37493777e-01 5.97922862e-01 7.41409123e-01
-6.64675713e-01 -5.62966943e-01 -3.09530079e-01 -6.18608832e-01
9.99691933e-02 -7.33294487e-01 1.06529582e+00 1.02347779e+00
4.70436424e-01 3.70887846e-01 2.01194465e-01 1.41646713e-01
2.43259236e-01 5.36480904e-01 1.29071236e+00 -9.19817924e-01
-7.11994052e-01 -1.00069845e+00 -4.40289915e-01 -4.92498517e-01
1.83015779e-01 -1.41709661e+00 2.82609370e-02 -1.05387008e+00
1.08753651e-01 -6.21240258e-01 2.44797945e-01 8.04731607e-01
-8.68172497e-02 7.63220936e-02 1.65829599e-01 1.36897704e-02
-3.71326745e-01 1.02064557e-01 8.76874864e-01 -1.24324359e-01
-2.45569289e-01 -3.15552920e-01 -1.03897953e+00 8.61759305e-01
6.21938407e-01 -3.53105426e-01 -2.06811517e-01 -6.62002265e-01
3.50684524e-01 -1.82521090e-01 1.01397717e-02 -1.18591547e+00
1.76153779e-01 -3.02613765e-01 -3.83372395e-03 -1.14560425e-01
-1.29023850e-01 -5.18412352e-01 2.87773609e-01 3.68255138e-01
-1.85521364e-01 6.42270863e-01 6.25980258e-01 -2.92149425e-01
-3.72424610e-02 -1.74613491e-01 6.76232338e-01 -4.34958041e-01
-9.28688288e-01 6.48387372e-02 -3.81546766e-01 2.65577525e-01
1.00331259e+00 1.09804943e-01 -4.54579651e-01 1.37548059e-01
-1.77389279e-01 3.97065848e-01 1.24640977e+00 5.34952760e-01
1.03297140e-02 -7.47026742e-01 -5.07748783e-01 6.50613308e-01
2.43609667e-01 -1.01021409e-01 -4.34695393e-01 6.32790804e-01
-1.06974232e+00 3.70447636e-01 -2.54615307e-01 -6.50324464e-01
-1.24478590e+00 8.41733992e-01 2.62987673e-01 -1.44070253e-01
-7.04894483e-01 9.56465065e-01 6.47183806e-02 -1.05024326e+00
2.19892766e-02 -6.24819636e-01 5.31621754e-01 -5.64958274e-01
5.93772113e-01 1.52317405e-01 3.87839228e-01 -2.56868731e-02
-5.23151517e-01 2.13473648e-01 -1.26323566e-01 2.38624081e-01
1.38025498e+00 3.60240787e-01 -6.35138810e-01 5.19270897e-01
1.13991916e+00 -4.91680652e-02 -4.21555936e-01 2.68059790e-01
1.12373404e-01 -4.65112686e-01 3.15741867e-01 -8.75060141e-01
-1.04262018e+00 6.12781525e-01 4.47973460e-01 7.73614645e-02
1.14115965e+00 -2.32731462e-01 5.40953815e-01 6.86160803e-01
3.75509739e-01 -8.39296758e-01 -4.48165059e-01 6.34368062e-01
6.88424706e-01 -1.26202595e+00 3.28609258e-01 -2.39659697e-01
-3.24276119e-01 1.16670036e+00 5.50624013e-01 3.88896316e-02
4.79244024e-01 9.17544246e-01 4.78952862e-02 -1.24603465e-01
-9.97011900e-01 1.97133012e-02 1.60812140e-01 6.06989980e-01
9.02065992e-01 1.74192593e-01 -1.59359500e-01 2.31912881e-01
-8.73424530e-01 4.24215198e-02 6.06345892e-01 1.14498556e+00
-8.85396823e-02 -1.57619238e+00 -4.93430227e-01 5.84590077e-01
-1.53612018e-01 -3.82787228e-01 -1.60006598e-01 8.97917926e-01
-6.24292046e-02 7.30679452e-01 2.59483028e-02 -3.86190146e-01
1.21041462e-01 8.26325417e-02 4.25965011e-01 -7.13133454e-01
-1.14919150e+00 -5.48051476e-01 4.61531073e-01 -9.32663083e-01
2.68136859e-01 -6.65749550e-01 -1.05333006e+00 -9.39565718e-01
-3.13374788e-01 2.53182977e-01 1.08903575e+00 7.65244007e-01
8.36780667e-01 6.08306408e-01 3.37669611e-01 -8.70453954e-01
-1.51061028e-01 -4.26660448e-01 -4.78573479e-02 1.03274658e-01
1.08411096e-01 -3.86836529e-01 8.94362479e-03 1.30580306e-01]
|
[7.792873382568359, 7.457464218139648]
|
aa88bbee-e338-4fac-b0ba-a06af2b3704c
|
what-can-human-sketches-do-for-object
|
2303.15149
| null |
https://arxiv.org/abs/2303.15149v1
|
https://arxiv.org/pdf/2303.15149v1.pdf
|
What Can Human Sketches Do for Object Detection?
|
Sketches are highly expressive, inherently capturing subjective and fine-grained visual cues. The exploration of such innate properties of human sketches has, however, been limited to that of image retrieval. In this paper, for the first time, we cultivate the expressiveness of sketches but for the fundamental vision task of object detection. The end result is a sketch-enabled object detection framework that detects based on what \textit{you} sketch -- \textit{that} ``zebra'' (e.g., one that is eating the grass) in a herd of zebras (instance-aware detection), and only the \textit{part} (e.g., ``head" of a ``zebra") that you desire (part-aware detection). We further dictate that our model works without (i) knowing which category to expect at testing (zero-shot) and (ii) not requiring additional bounding boxes (as per fully supervised) and class labels (as per weakly supervised). Instead of devising a model from the ground up, we show an intuitive synergy between foundation models (e.g., CLIP) and existing sketch models build for sketch-based image retrieval (SBIR), which can already elegantly solve the task -- CLIP to provide model generalisation, and SBIR to bridge the (sketch$\rightarrow$photo) gap. In particular, we first perform independent prompting on both sketch and photo branches of an SBIR model to build highly generalisable sketch and photo encoders on the back of the generalisation ability of CLIP. We then devise a training paradigm to adapt the learned encoders for object detection, such that the region embeddings of detected boxes are aligned with the sketch and photo embeddings from SBIR. Evaluating our framework on standard object detection datasets like PASCAL-VOC and MS-COCO outperforms both supervised (SOD) and weakly-supervised object detectors (WSOD) on zero-shot setups. Project Page: \url{https://pinakinathc.github.io/sketch-detect}
|
['Yi-Zhe Song', 'Tao Xiang', 'Subhadeep Koley', 'Aneeshan Sain', 'Ayan Kumar Bhunia', 'Pinaki Nath Chowdhury']
|
2023-03-27
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Chowdhury_What_Can_Human_Sketches_Do_for_Object_Detection_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Chowdhury_What_Can_Human_Sketches_Do_for_Object_Detection_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['sketch-based-image-retrieval']
|
['computer-vision']
|
[ 3.29348058e-01 4.78836633e-02 -6.73448816e-02 -2.78791487e-01
-5.74433267e-01 -1.01485550e+00 8.24799180e-01 -2.74216801e-01
-2.30154052e-01 1.38067126e-01 -3.52853648e-02 -2.94828653e-01
-3.59574258e-02 -7.85530210e-01 -1.00727487e+00 -4.68844265e-01
6.93975240e-02 5.20791471e-01 3.52025062e-01 -2.05837399e-01
2.57109761e-01 6.65809155e-01 -1.83028042e+00 4.89568800e-01
-8.17430019e-02 1.14251208e+00 4.12753016e-01 1.02526283e+00
-5.14208116e-02 6.28442883e-01 -4.56596851e-01 -6.69221580e-01
4.10780787e-01 -1.13258891e-01 -4.78077710e-01 1.94880396e-01
1.02631485e+00 -7.16362834e-01 -3.68695676e-01 8.19847882e-01
2.66598791e-01 -2.05117673e-01 9.26679552e-01 -1.55990231e+00
-8.96608233e-01 3.31639975e-01 -4.50409621e-01 -5.99848591e-02
1.80174172e-01 5.13252497e-01 1.30353904e+00 -1.37937939e+00
9.29893792e-01 1.30092585e+00 5.76201141e-01 7.88832366e-01
-1.42736816e+00 -8.15999746e-01 3.12692642e-01 -3.17809850e-01
-1.50748134e+00 -7.46380806e-01 7.06259251e-01 -5.65883458e-01
6.68567598e-01 5.19346952e-01 5.44526398e-01 1.33912349e+00
-4.69736814e-01 1.27203524e+00 8.64453912e-01 -4.24289137e-01
1.97762996e-01 3.83906364e-01 -8.16183463e-02 9.67371404e-01
7.73125142e-02 3.14209163e-01 -4.31284010e-01 -3.25224809e-02
1.19762266e+00 3.10400754e-01 -7.44949281e-02 -8.19120288e-01
-1.14695597e+00 5.89144528e-01 5.44131339e-01 1.86684400e-01
-2.13995248e-01 5.53537011e-01 2.73653418e-01 1.13935456e-01
1.36437893e-01 3.82731616e-01 -3.26864600e-01 2.53384739e-01
-1.09777129e+00 2.96075761e-01 7.76847661e-01 1.36368477e+00
8.13585043e-01 -1.77410513e-01 -1.94694430e-01 8.12232316e-01
1.67877540e-01 6.04608059e-01 -1.60076216e-01 -9.54848886e-01
2.03413069e-01 5.72423220e-01 2.46601686e-01 -6.64037943e-01
1.35143101e-01 -1.69526100e-01 -6.13166451e-01 6.96270108e-01
5.07832527e-01 1.50563121e-01 -1.08261645e+00 1.77197254e+00
1.38636962e-01 -8.53537917e-02 -3.43554020e-01 1.07479823e+00
7.82411337e-01 5.76374054e-01 2.82249093e-01 4.67165232e-01
1.72582459e+00 -7.98674107e-01 -1.03939377e-01 -3.25728238e-01
1.21836059e-01 -6.51948988e-01 1.41812897e+00 3.91883105e-01
-1.10405993e+00 -7.08798110e-01 -1.09752750e+00 -2.28196502e-01
-7.41971910e-01 5.80721319e-01 7.15854168e-01 6.02955580e-01
-1.01787198e+00 3.79807174e-01 -3.63049328e-01 -7.58552551e-01
5.88190734e-01 1.06171727e-01 -5.80672264e-01 -1.44562498e-01
-6.32910371e-01 6.96686983e-01 1.59246400e-01 -4.61788662e-02
-1.30176830e+00 -5.68056047e-01 -7.98489869e-01 1.54753745e-01
6.46984279e-01 -7.30206728e-01 1.10888600e+00 -1.35278869e+00
-1.10554898e+00 1.42301321e+00 6.34314865e-02 -1.47925779e-01
5.97410738e-01 -2.97002345e-02 -6.08012751e-02 3.00090790e-01
3.99223566e-02 1.33269656e+00 1.39888799e+00 -1.63237572e+00
-6.63542688e-01 -3.18464965e-01 4.43687677e-01 -5.36157824e-02
-1.41269594e-01 1.94271371e-01 -6.21988475e-01 -6.63626492e-01
-1.39844730e-01 -7.62477458e-01 2.18258709e-01 1.10062683e+00
-3.12820405e-01 -3.70337605e-01 8.33888113e-01 -3.24304670e-01
8.77317667e-01 -2.23038816e+00 -3.17815281e-02 2.90523637e-02
1.73122242e-01 4.01818544e-01 -4.45181102e-01 6.81324542e-01
-2.93256417e-02 1.74434394e-01 -2.30783924e-01 -4.49601144e-01
4.37331051e-01 3.72770876e-01 -7.98139453e-01 4.12026256e-01
5.49051940e-01 1.07272756e+00 -9.00502443e-01 -5.33395767e-01
4.39102948e-01 4.68746483e-01 -4.71624643e-01 4.66395080e-01
-5.28925240e-01 -1.49333298e-01 -1.78956389e-01 1.02206755e+00
5.87123692e-01 -2.46185705e-01 3.86418514e-02 -3.88834655e-01
-1.12063192e-01 -9.03408974e-02 -1.31744802e+00 1.70595789e+00
-3.94439071e-01 6.21957779e-01 5.38677990e-01 -7.94757843e-01
7.63526559e-01 8.68093073e-02 -5.77763617e-02 -4.26848382e-01
-1.79747641e-01 8.56001824e-02 -3.49617243e-01 -3.96762252e-01
3.79835546e-01 -2.59734750e-01 -1.22112274e-01 5.01156211e-01
3.85147899e-01 -2.53307641e-01 9.42107569e-03 4.35258865e-01
9.39870775e-01 5.46757460e-01 2.36709133e-01 -1.55135691e-01
2.82004356e-01 -5.98666705e-02 -1.13184594e-01 1.01931822e+00
-1.60606951e-01 8.25429201e-01 4.63120222e-01 -5.31178594e-01
-1.29508960e+00 -1.43229496e+00 -1.28699481e-01 1.76969683e+00
4.53010239e-02 -3.66821289e-01 -4.44788069e-01 -7.41402030e-01
3.66992235e-01 4.75273490e-01 -9.12688017e-01 2.60966331e-01
-3.34158927e-01 7.35789016e-02 8.76926959e-01 6.89807296e-01
6.05183467e-02 -1.36602581e+00 -9.20067072e-01 -1.11625075e-01
3.84792119e-01 -7.77551293e-01 -5.05087078e-01 3.70069295e-01
-2.54552782e-01 -9.81929243e-01 -7.90261745e-01 -7.47719288e-01
6.84709966e-01 3.36798161e-01 1.28448772e+00 4.41387296e-01
-7.82178819e-01 9.42381501e-01 -1.89294174e-01 -5.78958035e-01
1.05029577e-02 -4.37350720e-01 -1.71569571e-01 2.06810459e-02
3.26670617e-01 -4.21263218e-01 -7.98327863e-01 4.40453321e-01
-1.18254411e+00 9.74918976e-02 9.33698595e-01 8.03391933e-01
3.47306371e-01 -4.05643135e-01 2.60341823e-01 -5.84659457e-01
2.92397022e-01 -1.66621044e-01 -5.44264793e-01 5.67216456e-01
-1.69556588e-01 4.51899879e-02 3.69848341e-01 -6.59279227e-01
-7.95206904e-01 2.12565914e-01 3.92883550e-03 -7.52263725e-01
-6.05412424e-01 -2.17355728e-01 -1.62117824e-01 4.55157384e-02
7.26559997e-01 3.58087599e-01 -1.83084548e-01 -4.59214061e-01
9.74076509e-01 7.06656337e-01 7.66843915e-01 -1.02346778e+00
8.96971405e-01 8.25931430e-01 -2.12596551e-01 -8.83192122e-01
-5.90225160e-01 -7.15990543e-01 -6.67780757e-01 -2.81148404e-01
8.33180249e-01 -9.45788026e-01 -8.77184212e-01 1.26067325e-01
-1.29509068e+00 -4.08000678e-01 -3.61071944e-01 -1.83050469e-01
-6.40919387e-01 2.16012374e-01 -4.02659863e-01 -1.26718450e+00
-3.29103649e-01 -7.98270524e-01 1.68159473e+00 -5.54273799e-02
-2.56987602e-01 -4.74856228e-01 -1.48566470e-01 -1.03816073e-02
2.89118469e-01 3.13503556e-02 7.11974263e-01 -4.32363153e-01
-8.15733194e-01 -3.45301032e-01 -9.22067642e-01 4.39870745e-01
-3.60045999e-01 1.28078938e-01 -1.35461235e+00 -1.79134965e-01
-4.86768037e-01 -5.95451176e-01 8.76653492e-01 -8.41123909e-02
1.12171733e+00 -4.58174527e-01 -3.59627515e-01 5.85167706e-01
1.68368840e+00 -2.08411112e-01 5.02161145e-01 -3.26296762e-02
5.03938258e-01 6.87575698e-01 3.78073752e-01 3.02438736e-01
1.52335271e-01 7.35394895e-01 5.02657890e-01 -1.42117694e-01
-4.70802784e-01 -7.70515084e-01 4.47129428e-01 -1.57781854e-01
-4.51945737e-02 -1.49211198e-01 -7.27611244e-01 8.27672005e-01
-1.77730751e+00 -1.11570323e+00 1.69289991e-01 2.11623430e+00
7.17982352e-01 -2.79972907e-02 2.92494804e-01 -2.22928777e-01
5.25165915e-01 1.66165680e-01 -3.40081364e-01 -3.87863606e-01
4.95781153e-02 4.31320637e-01 3.93549919e-01 2.66632587e-01
-1.17493331e+00 1.04121757e+00 5.23114395e+00 8.11269045e-01
-9.47996497e-01 -3.69134098e-02 3.62352997e-01 -6.94355294e-02
-1.01290368e-01 1.63874626e-01 -8.47162247e-01 3.44860315e-01
3.05389911e-01 3.80181938e-01 7.70318091e-01 1.12976813e+00
-4.65487182e-01 3.83374356e-02 -1.68164253e+00 1.10024476e+00
2.88463533e-01 -1.23814774e+00 1.51033521e-01 -1.56038284e-01
2.62753576e-01 -2.56723195e-01 1.40949056e-01 5.63681543e-01
3.61906707e-01 -1.08814836e+00 1.19130218e+00 4.92816955e-01
1.27485847e+00 -9.51006860e-02 1.39894754e-01 4.10660595e-01
-1.37848842e+00 -1.48665279e-01 -5.60744703e-01 -7.67731443e-02
-2.08966136e-01 1.44206714e-02 -9.21184778e-01 1.77377760e-01
6.04616463e-01 4.45449352e-01 -6.14446402e-01 7.06127942e-01
-4.94899631e-01 1.55474469e-01 -3.56812894e-01 4.05828804e-02
2.42650941e-01 1.66434180e-02 5.44804573e-01 1.62932765e+00
5.96215092e-02 1.78882852e-01 3.66732255e-02 1.40880466e+00
-1.21588498e-01 -2.80180246e-01 -9.05842185e-01 -2.57050600e-02
4.70172703e-01 1.57371855e+00 -6.54402316e-01 -5.35178125e-01
-3.69784474e-01 1.18641877e+00 4.09125060e-01 4.20342624e-01
-6.07961297e-01 -3.04089427e-01 5.92668712e-01 3.48870456e-01
7.64213979e-01 -5.61834313e-02 -1.59076825e-01 -1.12438822e+00
8.43506232e-02 -7.18117654e-01 4.13889438e-01 -1.21069324e+00
-1.51773751e+00 1.66527152e-01 2.32351222e-03 -1.02860749e+00
8.18946864e-03 -1.04814970e+00 -7.00718343e-01 8.25464964e-01
-1.23596537e+00 -1.84214342e+00 -3.09466660e-01 5.85283995e-01
6.15061224e-01 1.94974422e-01 9.01533127e-01 1.41500115e-01
-2.01333523e-01 5.11978626e-01 -4.42164689e-01 4.43777442e-01
7.43958473e-01 -1.37818015e+00 4.10415262e-01 7.03369975e-01
5.64969659e-01 9.27630723e-01 5.23368776e-01 -3.97959411e-01
-1.62069178e+00 -9.50127363e-01 6.45029545e-01 -9.80657876e-01
7.23904192e-01 -1.06699562e+00 -6.24642432e-01 7.43488729e-01
-9.81585234e-02 2.44297549e-01 1.63699433e-01 1.70687065e-02
-9.41048503e-01 -4.27056588e-02 -1.05179167e+00 5.73638618e-01
1.30742753e+00 -1.00725234e+00 -7.39769876e-01 1.99002191e-01
3.47403765e-01 -5.52421622e-02 -5.43679714e-01 8.98493975e-02
1.24194491e+00 -1.00603330e+00 1.50779176e+00 -7.63037205e-01
5.13740897e-01 -4.41585571e-01 -3.77033144e-01 -6.70194507e-01
-1.91175178e-01 -2.96560377e-01 -1.21541478e-01 1.23846447e+00
4.10384685e-02 -1.39278010e-01 7.97959566e-01 5.90309680e-01
6.70995861e-02 -5.35810888e-01 -6.65340483e-01 -7.61940539e-01
-1.36025742e-01 -4.88645554e-01 4.58334953e-01 7.23959863e-01
-1.47238195e-01 2.26090670e-01 -3.48738521e-01 1.44391641e-01
6.48299634e-01 2.12448746e-01 1.29826951e+00 -1.10441375e+00
-5.41276634e-01 -6.52191222e-01 -3.08761805e-01 -1.22080302e+00
-2.23490253e-01 -8.59568477e-01 1.92633286e-01 -1.36141634e+00
3.94533455e-01 -5.93944430e-01 -2.43793711e-01 8.30725908e-01
-2.91900411e-02 5.32533884e-01 5.66537380e-01 2.37285525e-01
-8.28750908e-01 7.11733196e-03 8.64400685e-01 -3.00933182e-01
2.30252013e-01 -2.93556184e-01 -5.94646275e-01 6.33503795e-01
3.05284828e-01 -3.74206781e-01 -1.43736809e-01 -3.76721680e-01
3.69463027e-01 6.28583133e-02 1.25612736e+00 -7.85663307e-01
2.77415276e-01 1.69187668e-03 5.74618101e-01 -5.20431697e-01
5.18147767e-01 -9.61403787e-01 5.61627112e-02 1.90273643e-01
-4.10690486e-01 -3.29872251e-01 1.24662511e-01 6.05532408e-01
2.71851689e-01 -2.63921976e-01 7.09359050e-01 -4.68306780e-01
-9.02180135e-01 3.58149111e-01 8.79649222e-02 -3.60515833e-01
8.21884274e-01 -3.35206062e-01 -4.42073524e-01 -2.62704074e-01
-5.55250406e-01 -7.66579732e-02 8.51679265e-01 5.37904382e-01
6.68436050e-01 -1.08607185e+00 -5.15189230e-01 3.73495847e-01
6.85513496e-01 -6.14526533e-02 6.61841482e-02 4.75319535e-01
-1.94995910e-01 2.96298623e-01 -3.23012173e-01 -6.09216273e-01
-1.14930511e+00 9.84216690e-01 1.12985037e-01 1.49220467e-01
-6.19712472e-01 1.08983791e+00 8.58013749e-01 -4.62214231e-01
5.18909991e-01 -3.34747106e-01 4.15303200e-01 4.84063989e-03
5.33232033e-01 8.57472345e-02 -2.73270756e-01 -3.09254050e-01
-4.09243286e-01 4.81996417e-01 1.26695007e-01 -1.54824272e-01
1.25339973e+00 1.55978382e-01 -9.41819027e-02 3.01062763e-01
1.08713090e+00 -3.97110060e-02 -1.56330669e+00 -1.07944824e-01
-1.22292042e-01 -5.15109777e-01 -3.10820073e-01 -1.12111163e+00
-5.88271379e-01 1.17709386e+00 6.26116276e-01 1.50309727e-01
7.30724931e-01 4.17228848e-01 3.14162910e-01 4.33874130e-01
5.33403516e-01 -7.51968324e-01 3.13946337e-01 1.77054834e-02
1.15207136e+00 -1.19469368e+00 -2.11222712e-02 -5.31585305e-04
-5.47790408e-01 1.23047590e+00 4.37089980e-01 -4.62187260e-01
1.79457128e-01 2.80404776e-01 -2.09963292e-01 -5.01663685e-01
-7.06555068e-01 -5.82704842e-01 4.44815278e-01 7.73841083e-01
1.06965773e-01 2.37374410e-01 1.86024517e-01 6.13519549e-01
2.16771066e-01 -5.74398004e-02 -7.40707666e-02 8.48963857e-01
-5.70725679e-01 -9.68226433e-01 -4.47605550e-01 3.95289719e-01
-3.91336605e-02 -9.64965299e-02 -7.32763290e-01 9.88750219e-01
4.68767226e-01 5.48099220e-01 7.57676661e-02 -1.19761080e-01
2.72620469e-01 2.11489856e-01 7.50357747e-01 -7.19147742e-01
-4.74136353e-01 -1.41319782e-01 -7.85381272e-02 -4.93942171e-01
-2.22490996e-01 -4.47164834e-01 -6.73198283e-01 -1.57580303e-03
-2.39733532e-01 -3.34693164e-01 7.77638435e-01 4.93548810e-01
8.55240151e-02 -2.70523224e-03 2.78116554e-01 -1.19862056e+00
-5.46184540e-01 -8.11840653e-01 -6.20011449e-01 6.98275685e-01
4.17008579e-01 -6.90385699e-01 -3.31293911e-01 1.08530320e-01]
|
[11.654770851135254, 0.5219424366950989]
|
8b2d7b2e-d97e-4428-bde3-53499a658d08
|
throwing-away-data-improves-worst-class-error
|
2205.11672
| null |
https://arxiv.org/abs/2205.11672v2
|
https://arxiv.org/pdf/2205.11672v2.pdf
|
Why does Throwing Away Data Improve Worst-Group Error?
|
When facing data with imbalanced classes or groups, practitioners follow an intriguing strategy to achieve best results. They throw away examples until the classes or groups are balanced in size, and then perform empirical risk minimization on the reduced training set. This opposes common wisdom in learning theory, where the expected error is supposed to decrease as the dataset grows in size. In this work, we leverage extreme value theory to address this apparent contradiction. Our results show that the tails of the data distribution play an important role in determining the worst-group-accuracy of linear classifiers. When learning on data with heavy tails, throwing away data restores the geometric symmetry of the resulting classifier, and therefore improves its worst-group generalization.
|
['Martin Arjovsky', 'Kartik Ahuja', 'Kamalika Chaudhuri', 'David Lopez-Paz']
|
2022-05-23
| null | null | null | null |
['imbalanced-classification']
|
['miscellaneous']
|
[ 1.53431743e-01 3.33629966e-01 -5.27474940e-01 -5.47820032e-01
-7.63536215e-01 -5.45174778e-01 1.15375891e-01 4.03766960e-01
-4.04119432e-01 8.11682165e-01 -1.00568242e-01 -6.22290552e-01
-4.34916645e-01 -7.44879186e-01 -7.43809342e-01 -1.05846047e+00
-8.34185630e-02 3.95401686e-01 -9.65032354e-03 9.12878439e-02
4.79750901e-01 3.93820971e-01 -1.64142692e+00 2.41490290e-01
1.00523448e+00 9.26474452e-01 -7.76095271e-01 4.46462840e-01
1.11084901e-01 6.56783640e-01 -7.00632811e-01 -6.86957240e-01
5.43302596e-01 -4.54891413e-01 -5.12501061e-01 5.88399265e-03
6.02798581e-01 -1.49419188e-01 2.74167389e-01 1.16900444e+00
2.96238124e-01 -2.16569770e-02 1.20394242e+00 -1.58130348e+00
-4.33609605e-01 6.41317368e-01 -9.95342910e-01 9.99229997e-02
4.91224006e-02 -1.50269689e-03 1.30499923e+00 -5.63871741e-01
5.65116741e-02 9.19075012e-01 1.05954409e+00 5.41167974e-01
-1.25317085e+00 -8.37783158e-01 1.91457778e-01 -2.33253557e-02
-1.32120323e+00 -2.09955588e-01 5.00809371e-01 -5.65798104e-01
4.19534087e-01 4.03517097e-01 5.45434952e-01 5.93818784e-01
3.31114143e-01 4.46105808e-01 7.71929622e-01 -5.51000059e-01
5.65321207e-01 1.74987391e-01 3.10101509e-01 1.73156798e-01
9.76054311e-01 5.51228002e-02 -3.27479482e-01 -5.09627759e-01
1.45078599e-01 2.32115164e-01 -1.64492279e-01 -6.28297806e-01
-5.02901971e-01 9.60027933e-01 3.14693868e-01 3.48603576e-02
-1.92371711e-01 2.28564758e-02 4.53142911e-01 4.61600423e-01
7.85117686e-01 4.15911824e-01 -4.37057436e-01 1.63080722e-01
-8.23507726e-01 3.63759518e-01 8.79916847e-01 4.91748691e-01
3.28075349e-01 -4.94220108e-01 2.79163837e-01 7.31428504e-01
1.11286774e-01 1.84641287e-01 3.32720906e-01 -7.88083851e-01
6.51045740e-01 8.10234368e-01 -1.87647669e-03 -9.11173522e-01
-2.86733180e-01 -6.87734067e-01 -8.14621270e-01 5.23365378e-01
1.00266635e+00 -2.47156620e-01 -5.35553217e-01 1.80061674e+00
4.15294945e-01 -4.84265327e-01 7.18510523e-02 7.22638309e-01
-1.07025690e-01 3.16544056e-01 1.50600508e-01 -4.31221366e-01
8.61674309e-01 -3.42027813e-01 -4.68554705e-01 -2.93859214e-01
1.02932572e+00 -3.63251895e-01 1.22990274e+00 7.18368411e-01
-1.02582061e+00 -1.52869165e-01 -1.22907758e+00 1.83497846e-01
-7.47476332e-03 -5.13203382e-01 4.79084224e-01 1.12094355e+00
-4.14123267e-01 7.98313379e-01 -5.45623779e-01 2.66044736e-01
1.08777416e+00 3.72223973e-01 -2.95792222e-01 -1.44182414e-01
-7.81014204e-01 6.06984556e-01 3.74126375e-01 3.42776179e-02
-1.60807073e-01 -1.24850416e+00 -5.24824560e-01 1.71552002e-01
3.72470319e-01 -4.25621420e-01 9.42438841e-01 -1.09806681e+00
-7.93261647e-01 7.77329683e-01 3.54488432e-01 -5.91734827e-01
1.03240335e+00 -1.42176569e-01 1.75858274e-01 -3.59387070e-01
-1.99862137e-01 7.15407953e-02 8.29172850e-01 -1.18334007e+00
-5.87304354e-01 -8.61576915e-01 -2.62989402e-01 6.40105829e-02
-3.98189753e-01 -4.24562573e-01 5.77818871e-01 -4.31389660e-01
2.87172794e-01 -7.50610232e-01 -4.17642713e-01 -6.54497147e-02
-2.95900732e-01 -3.80219936e-01 4.64828372e-01 -2.06368446e-01
1.30325496e+00 -2.42632723e+00 -3.52444500e-01 4.47305948e-01
4.86743510e-01 5.64635843e-02 4.01964068e-01 1.51282236e-01
-3.45496535e-01 3.69631171e-01 -2.97502249e-01 -8.63128603e-02
5.31109497e-02 9.61000770e-02 -5.00806093e-01 8.65201533e-01
5.90871051e-02 4.49965596e-01 -6.42125309e-01 -2.75560349e-01
-2.19784811e-01 8.98314863e-02 -9.59995925e-01 -1.46254105e-02
-2.55560614e-02 -1.18608207e-01 -3.79031062e-01 2.62549609e-01
1.08200598e+00 -3.22268426e-01 3.30280155e-01 2.48323858e-01
2.88804531e-01 1.86997578e-01 -1.38947916e+00 6.11964583e-01
-1.74133196e-01 5.47385752e-01 -3.56028825e-01 -1.40392411e+00
8.07054222e-01 -8.94234106e-02 4.59675521e-01 -3.71047974e-01
2.38813341e-01 3.85730684e-01 3.15848470e-01 -3.33162785e-01
9.70955566e-02 -8.69436979e-01 -6.33380786e-02 4.72512662e-01
-3.84962201e-01 -7.44789466e-02 -6.71729743e-02 -4.49466221e-02
7.94643164e-01 -4.10604656e-01 4.61667269e-01 -4.04085517e-01
9.80126187e-02 -1.36207446e-01 5.34808397e-01 7.16349185e-01
-2.46917587e-02 6.55436277e-01 1.32341874e+00 -5.47949314e-01
-1.13050950e+00 -1.16991305e+00 -4.56840098e-01 9.33413744e-01
-1.74956143e-01 -6.52995780e-02 -9.09536660e-01 -1.05869210e+00
4.29641694e-01 8.71659279e-01 -9.09996271e-01 -5.33686817e-01
-2.18379200e-01 -1.23980343e+00 2.55339265e-01 5.22874236e-01
9.19373520e-03 -3.76811564e-01 -6.93090200e-01 -3.10764432e-01
1.88289776e-01 -3.82716447e-01 -3.00973475e-01 4.44802016e-01
-1.07564437e+00 -1.40749121e+00 -5.15053749e-01 -4.70288604e-01
9.18410838e-01 -1.00906871e-01 1.09657550e+00 2.93053210e-01
-1.66716918e-01 -1.35567695e-01 -2.24624604e-01 -8.50988746e-01
-3.00291151e-01 7.77054727e-02 -9.90651846e-02 -3.83796450e-03
4.91568625e-01 -5.14290094e-01 -4.55248952e-01 2.51796484e-01
-6.92544639e-01 -5.43830335e-01 2.58178473e-01 8.45643818e-01
4.37667251e-01 3.72039586e-01 9.42880511e-01 -1.08075464e+00
5.14740944e-01 -6.99408233e-01 -5.67404687e-01 2.66734391e-01
-1.00844586e+00 -8.12529400e-02 7.84582496e-01 -7.51785338e-01
-5.99271059e-01 -3.12581658e-01 1.81675121e-01 -2.01774180e-01
1.08106442e-01 1.65774122e-01 -2.66741812e-01 2.13485360e-01
1.02354479e+00 -2.99772739e-01 2.06500471e-01 -3.73448163e-01
-2.01188857e-04 8.68231118e-01 1.44597292e-01 -3.40945840e-01
4.95646477e-01 4.64254498e-01 2.87354469e-01 -4.08435345e-01
-1.33160245e+00 -2.12428346e-01 -5.44163704e-01 8.21646862e-03
3.84693265e-01 -4.69956189e-01 -8.48637819e-01 2.57211268e-01
-4.43844914e-01 -3.66632879e-01 -8.29446256e-01 5.43462694e-01
-5.61833084e-01 -5.79511793e-03 2.67483704e-02 -1.11583030e+00
-5.20234853e-02 -6.72957957e-01 6.14819944e-01 1.47300661e-01
-2.26594612e-01 -1.14041507e+00 -1.46771029e-01 2.80733913e-01
3.36945383e-03 4.59139764e-01 1.21349311e+00 -1.23184824e+00
6.13580532e-02 -6.73097610e-01 3.01941261e-02 7.23562896e-01
7.42109492e-02 -4.54445072e-02 -7.92332888e-01 -3.11141253e-01
2.66916484e-01 -2.62131095e-01 6.85344577e-01 5.94770968e-01
1.42908227e+00 -5.51859796e-01 -3.19510311e-01 5.05993068e-01
1.37278581e+00 4.24540602e-02 4.88100260e-01 2.25890577e-01
3.58383656e-01 9.89283442e-01 5.63023567e-01 6.48302317e-01
7.91156664e-02 8.62212703e-02 3.81016284e-01 8.33414793e-02
2.46985570e-01 -2.48620510e-01 8.99757594e-02 8.04255679e-02
3.53013337e-01 -2.37810507e-01 -1.07811439e+00 3.65142196e-01
-1.45106936e+00 -1.07450974e+00 -2.19522044e-01 2.70248675e+00
9.65568423e-01 5.91720164e-01 4.09263819e-01 6.11020267e-01
6.84914768e-01 -1.48878515e-01 -7.03833818e-01 -4.05970037e-01
1.94328744e-02 -2.21421719e-01 6.25753522e-01 4.98004168e-01
-6.68198705e-01 2.52915055e-01 7.06344271e+00 9.76120830e-01
-9.67241704e-01 -1.42887101e-01 1.18765318e+00 -6.13967121e-01
-2.62209892e-01 -5.12593309e-04 -8.67223501e-01 6.59304321e-01
9.80360866e-01 -5.53840935e-01 5.16297668e-02 9.05328512e-01
-1.31413937e-01 -2.80856013e-01 -1.48670542e+00 7.13533163e-01
6.43907338e-02 -1.01844680e+00 -1.36946619e-01 4.82813507e-01
7.59840369e-01 -2.53819257e-01 3.96637321e-01 1.54992208e-01
2.61531770e-01 -1.26142716e+00 5.95167696e-01 3.77108932e-01
5.77132821e-01 -1.24433219e+00 9.03183222e-01 6.75520599e-01
-3.47720355e-01 -5.78874946e-01 -5.32364190e-01 -3.58811378e-01
-2.42199257e-01 1.26840878e+00 -8.40757668e-01 -7.78713375e-02
5.42714894e-01 2.88881034e-01 -5.25862277e-01 1.07978201e+00
1.66783392e-01 8.05148780e-01 -4.74383295e-01 2.53359843e-02
-8.35385770e-02 -3.29561591e-01 2.58999974e-01 7.36048996e-01
4.91844267e-02 4.64960262e-02 -4.37444411e-02 5.75450480e-01
-1.64749309e-01 2.50803411e-01 -6.44800067e-01 1.50566816e-01
4.24519449e-01 7.10152149e-01 -8.57766032e-01 -3.56038421e-01
-1.95396796e-01 1.87145337e-01 2.89755464e-01 4.55623753e-02
-5.44386685e-01 -5.49617767e-01 5.72018206e-01 3.17326695e-01
7.57632330e-02 1.54002100e-01 -9.94492948e-01 -7.26944625e-01
3.74138325e-01 -9.74026561e-01 7.83059418e-01 -1.28484860e-01
-1.57369220e+00 6.50820062e-02 -9.01366845e-02 -1.26570368e+00
-1.12800427e-01 -5.54777861e-01 -5.15677691e-01 6.29736125e-01
-1.06338263e+00 -4.50337946e-01 7.38367140e-02 1.85648352e-01
1.94170922e-01 1.91766530e-01 2.33833179e-01 1.01067990e-01
-5.74609518e-01 1.12599993e+00 4.81183678e-01 -1.76190503e-03
5.60501873e-01 -1.25305665e+00 -4.77134436e-02 3.19372594e-01
-8.00615922e-02 6.01593137e-01 8.90055776e-01 -5.47577977e-01
-9.88657653e-01 -5.80778182e-01 6.93202436e-01 -7.00073183e-01
6.63554072e-01 -4.16409224e-01 -1.14564228e+00 4.74999845e-01
-5.31706393e-01 3.35489392e-01 1.20836055e+00 4.61350083e-01
-6.24914706e-01 -2.92108387e-01 -1.59511435e+00 3.96357715e-01
8.70854914e-01 -1.74463272e-01 -6.38815284e-01 4.54061925e-01
3.42674106e-01 -4.57052439e-02 -8.95699620e-01 4.09067750e-01
7.98339784e-01 -1.19362915e+00 5.91495633e-01 -1.21693540e+00
5.49184382e-01 1.86571926e-01 -2.45936930e-01 -1.47185373e+00
1.35370091e-01 -3.86908472e-01 1.63259298e-01 9.12567675e-01
5.82304835e-01 -8.88673663e-01 1.21626043e+00 7.84940422e-01
3.52116942e-01 -1.02389121e+00 -9.21470940e-01 -9.95204806e-01
8.71734500e-01 -3.82485032e-01 6.60413146e-01 9.33753252e-01
2.71283209e-01 6.76686764e-02 1.41358554e-01 -7.71500263e-03
8.32233310e-01 7.60471001e-02 6.33182526e-01 -1.58280659e+00
-7.82542303e-02 -5.57129741e-01 -5.67962050e-01 -5.03785908e-01
7.08186924e-02 -8.73752296e-01 -2.84938677e-03 -9.71341848e-01
4.03528005e-01 -7.64796138e-01 -2.25511611e-01 3.54942203e-01
-2.84887999e-01 2.83728600e-01 4.84398939e-02 1.88758656e-01
-4.90119249e-01 2.37108290e-01 9.31409180e-01 2.71556258e-01
-3.32564831e-01 4.72113609e-01 -1.28314686e+00 9.32024240e-01
8.87007892e-01 -8.63824069e-01 -4.64110732e-01 1.66836560e-01
7.34132230e-01 -3.06722760e-01 2.44951829e-01 -7.63491392e-01
-1.55292619e-02 -3.56169969e-01 6.54838324e-01 -2.65840083e-01
-2.46747285e-01 -9.04921412e-01 -9.44857150e-02 8.41685534e-01
-8.18859398e-01 -1.48029238e-01 -1.98341355e-01 4.76476252e-01
1.21536933e-01 -5.77665031e-01 1.13547826e+00 2.28730261e-01
5.39592206e-01 2.80427001e-02 -9.99828577e-02 5.45392692e-01
1.35135269e+00 -3.07076484e-01 -2.94722259e-01 -9.99390781e-02
-5.51059961e-01 1.74684823e-01 6.34972870e-01 9.25053060e-02
1.79830596e-01 -1.27808869e+00 -7.43835926e-01 3.00913781e-01
1.59636706e-01 1.86838254e-01 1.39472649e-01 9.65562403e-01
-2.72086889e-01 -2.20091641e-01 1.74976066e-02 -5.28338552e-01
-1.18800116e+00 7.02520013e-01 5.54234326e-01 -2.40460131e-02
-5.55300355e-01 9.84456778e-01 2.04105869e-01 -1.88004717e-01
3.04630965e-01 -9.69170332e-02 2.93823481e-02 5.36708653e-01
8.48886490e-01 7.00172722e-01 3.16950440e-01 1.24914711e-02
-2.26085529e-01 4.10773993e-01 -3.30942690e-01 1.20282650e-01
1.46367812e+00 1.29230082e-01 2.85972399e-03 5.99665403e-01
1.40039313e+00 2.46376455e-01 -1.40385520e+00 1.67872727e-01
2.73038507e-01 -8.93360138e-01 -1.81136027e-01 -4.81590807e-01
-9.83187795e-01 8.50864172e-01 4.07057911e-01 7.32914984e-01
1.07682681e+00 2.02454865e-01 2.55468190e-01 2.41095752e-01
1.11115061e-01 -1.22079766e+00 8.09915960e-02 2.14971110e-01
7.44338632e-01 -1.20869446e+00 4.11898345e-01 -3.10428619e-01
-5.73750079e-01 8.50378931e-01 4.08253223e-01 -3.38358283e-01
9.28229868e-01 4.07427251e-01 -2.35940665e-01 -8.34724605e-02
-8.97788763e-01 4.43244874e-01 1.74724758e-01 6.58796191e-01
2.89637417e-01 1.92594931e-01 -4.65533406e-01 8.04294348e-01
-6.66700363e-01 -1.38456836e-01 6.81953609e-01 7.81135499e-01
-6.63821399e-01 -7.81885266e-01 -5.88799238e-01 9.87866402e-01
-7.21152723e-01 6.92430586e-02 -3.99940699e-01 7.36516953e-01
1.54623538e-01 8.02960813e-01 3.77177000e-01 -1.11138612e-01
3.93743724e-01 2.69381106e-01 5.25672674e-01 -4.38205838e-01
-2.77226448e-01 -5.16662061e-01 -3.09316993e-01 -3.38355124e-01
1.46690205e-01 -9.89024699e-01 -1.12729704e+00 -5.66140115e-01
-5.87388754e-01 4.10462320e-01 5.45028389e-01 9.18976367e-01
9.70589295e-02 1.00862853e-01 9.53806996e-01 -3.21656585e-01
-1.44314837e+00 -6.23952031e-01 -6.30419850e-01 4.03818816e-01
7.58362055e-01 -5.42484999e-01 -1.15557837e+00 -3.05695891e-01]
|
[8.477668762207031, 4.493873596191406]
|
e4fb9399-7a51-4930-86d6-81f91ac81bb5
|
an-automated-end-to-end-deep-learning-based
|
2305.00046
| null |
https://arxiv.org/abs/2305.00046v1
|
https://arxiv.org/pdf/2305.00046v1.pdf
|
An automated end-to-end deep learning-based framework for lung cancer diagnosis by detecting and classifying the lung nodules
|
Lung cancer is a leading cause of cancer-related deaths worldwide, and early detection is crucial for improving patient outcomes. Nevertheless, early diagnosis of cancer is a major challenge, particularly in low-resource settings where access to medical resources and trained radiologists is limited. The objective of this study is to propose an automated end-to-end deep learning-based framework for the early detection and classification of lung nodules, specifically for low-resource settings. The proposed framework consists of three stages: lung segmentation using a modified 3D U-Net named 3D Res-U-Net, nodule detection using YOLO-v5, and classification with a Vision Transformer-based architecture. We evaluated the proposed framework on a publicly available dataset, LUNA16. The proposed framework's performance was measured using the respective domain's evaluation matrices. The proposed framework achieved a 98.82% lung segmentation dice score while detecting the lung nodule with 0.76 mAP@50 from the segmented lung, at a low false-positive rate. The performance of both networks of the proposed framework was compared with other studies and found to outperform them regarding segmentation and detection accuracy. Additionally, our proposed Vision transformer network obtained an accuracy of 93.57%, which is 1.21 higher than the state-of-the-art networks. Our proposed end-to-end deep learning-based framework can effectively segment lungs, and detect and classify lung nodules, specifically in low-resource settings with limited access to radiologists. The proposed framework outperforms existing studies regarding all the respective evaluation metrics. The proposed framework can potentially improve the accuracy and efficiency of lung cancer screening in low-resource settings, ultimately leading to better patient outcomes.
|
['Samiul Based Shuvo']
|
2023-04-28
| null | null | null | null |
['lung-cancer-diagnosis']
|
['medical']
|
[ 7.32841715e-02 4.54167984e-02 -4.27073896e-01 1.87653393e-01
-8.86334121e-01 -4.55439448e-01 2.59787560e-01 6.73278198e-02
-5.81181586e-01 3.40390682e-01 -1.38759062e-01 -6.17152572e-01
-1.17637902e-01 -9.03965414e-01 -3.50285858e-01 -7.68544197e-01
1.14214629e-01 5.63621998e-01 6.72724903e-01 5.38927972e-01
-3.28388184e-01 5.96943855e-01 -8.78163636e-01 2.32268631e-01
7.94650793e-01 1.08809292e+00 5.58253109e-01 1.02526438e+00
2.18582645e-01 9.93090153e-01 1.20822459e-01 -1.97611943e-01
4.28587884e-01 -4.28987920e-01 -8.09569955e-01 -6.92194253e-02
3.82106870e-01 -6.98147476e-01 -5.26045918e-01 7.78548360e-01
5.95464528e-01 -2.10403308e-01 7.40097463e-01 -5.35307646e-01
-3.93178910e-01 2.54990220e-01 -4.63951051e-01 4.71915454e-01
-4.14397210e-01 5.02230704e-01 6.44180536e-01 -7.84995317e-01
1.89826623e-01 6.04635000e-01 1.05940533e+00 6.68217123e-01
-5.25828362e-01 -4.94416803e-01 -7.55842030e-01 6.71704859e-03
-1.18214142e+00 6.01156731e-04 5.01288362e-02 -4.79156345e-01
5.63310385e-01 4.23026890e-01 6.24538898e-01 6.05308890e-01
4.98010904e-01 6.04800880e-01 8.28914702e-01 -2.32282892e-01
-9.38856304e-02 2.42923349e-01 -1.26760289e-01 1.40830708e+00
7.94408739e-01 7.18661696e-02 2.74933398e-01 -6.28848225e-02
1.13675499e+00 4.92338121e-01 -3.09749618e-02 -2.96279222e-01
-1.68059301e+00 6.38341725e-01 1.14470816e+00 5.97738743e-01
-6.39978647e-01 2.26745635e-01 4.18992847e-01 -4.51347649e-01
2.05419585e-01 1.33652434e-01 6.32265508e-02 2.13597745e-01
-1.02998269e+00 -3.56568992e-01 6.50586843e-01 4.56929058e-01
-8.12831447e-02 -1.43439874e-01 -9.74183202e-01 6.10895395e-01
3.64369720e-01 6.03858113e-01 8.78475606e-01 -6.53695941e-01
2.12382749e-01 5.27913213e-01 2.85512325e-03 -7.15699911e-01
-7.20916629e-01 -8.80081415e-01 -1.28377759e+00 -1.42137811e-01
5.13887584e-01 -2.49894753e-01 -1.18031526e+00 1.27089715e+00
3.66703123e-01 3.43613625e-01 4.99402545e-02 1.01320791e+00
1.29105103e+00 2.55161881e-01 2.71428168e-01 -1.08703986e-01
1.64229941e+00 -1.19290936e+00 -3.34071666e-01 1.64040290e-02
8.00104558e-01 -6.31494820e-01 8.48362982e-01 -3.37961227e-01
-1.09858012e+00 -5.92590034e-01 -7.02618182e-01 8.16938877e-02
2.38524556e-01 9.80399847e-01 3.70204598e-01 8.02165210e-01
-1.11044765e+00 3.53325635e-01 -1.22944796e+00 -7.86387146e-01
7.79022098e-01 5.01968086e-01 -1.21058777e-01 -1.20070793e-01
-8.47359121e-01 7.64289379e-01 3.76183659e-01 1.10658787e-01
-1.23232758e+00 -8.79998982e-01 -2.51664370e-01 1.73327252e-01
5.12408376e-01 -1.27982986e+00 1.65018642e+00 -3.20273638e-01
-9.33214366e-01 8.81723344e-01 3.73891219e-02 -7.48953521e-01
9.79290128e-01 1.08447053e-01 8.91523808e-02 4.81863111e-01
2.97848195e-01 6.09244764e-01 3.07478726e-01 -7.36630499e-01
-9.21869636e-01 -1.51997939e-01 -2.74369538e-01 4.58845526e-01
-4.60936785e-01 -4.01289195e-01 -7.34573305e-01 -3.56378675e-01
-1.68422207e-01 -1.15138936e+00 -4.13409024e-01 5.34162402e-01
-4.11716938e-01 -3.60844284e-02 9.12404358e-01 -7.66952217e-01
1.01410627e+00 -1.72495031e+00 -5.73039234e-01 7.48952925e-02
6.28431678e-01 7.61756003e-01 3.20377022e-01 -4.06894058e-01
1.73204646e-01 2.84432411e-01 -2.03003421e-01 2.48434454e-01
-5.40333509e-01 -1.99298307e-01 7.13313758e-01 4.12780046e-01
8.57541710e-02 1.23927629e+00 -9.30534124e-01 -9.63478267e-01
6.16920650e-01 5.39133847e-01 -3.09159726e-01 2.77923644e-01
1.77078724e-01 4.31610823e-01 -6.51036978e-01 8.72701049e-01
3.60425204e-01 -6.25611186e-01 1.19582482e-01 -3.53529215e-01
-1.17570115e-02 -3.10167909e-01 -5.47815323e-01 1.16152775e+00
-6.44335330e-01 5.23735762e-01 -9.69384611e-02 -6.19960904e-01
4.99939591e-01 7.08383560e-01 7.79786587e-01 -5.38373768e-01
2.69248128e-01 4.38683748e-01 4.95394170e-01 -8.20255220e-01
-3.33013505e-01 -4.07182127e-02 3.47795606e-01 1.41024992e-01
-3.06033820e-01 5.85500710e-02 2.14541197e-01 1.34530589e-01
1.61910844e+00 -3.87558609e-01 6.29909456e-01 -1.73770562e-01
7.97356665e-01 3.09653908e-01 2.04145610e-01 1.01002967e+00
-7.04277933e-01 4.96456385e-01 2.80055016e-01 -2.17288628e-01
-8.33295465e-01 -1.03579736e+00 -2.09158882e-01 5.64784050e-01
-1.13453284e-01 3.31848234e-01 -7.79178381e-01 -1.20004070e+00
-6.13306202e-02 4.45954233e-01 -6.73893631e-01 1.75930262e-02
-5.40987730e-01 -7.60981202e-01 7.52815127e-01 6.55156672e-01
1.02750778e+00 -1.07962263e+00 -9.84341800e-01 1.22517340e-01
-3.09845835e-01 -1.14423478e+00 -4.51242000e-01 5.64070269e-02
-1.10300851e+00 -1.41911566e+00 -1.47232223e+00 -1.01230478e+00
6.79911017e-01 5.74542165e-01 1.07781959e+00 1.73872039e-01
-7.65418470e-01 3.73203725e-01 1.33757055e-01 -3.18824023e-01
-6.41739726e-01 1.41578823e-01 -3.83413553e-01 -3.51285547e-01
2.04333797e-01 1.30767897e-01 -1.21174204e+00 3.84672731e-01
-8.23775053e-01 1.27462059e-01 1.37196326e+00 9.14897144e-01
8.70906174e-01 1.10526413e-01 3.77503961e-01 -9.56058264e-01
4.24020797e-01 -6.67145789e-01 -2.64363259e-01 2.89774686e-01
-3.40773821e-01 -2.91799873e-01 6.51370883e-01 -7.27058947e-02
-1.00824177e+00 3.51163983e-01 4.08925228e-02 -3.74504000e-01
-6.92049637e-02 2.82977730e-01 4.66099620e-01 -3.59337389e-01
7.38673568e-01 1.01543494e-01 2.90421396e-02 -1.44173846e-01
-1.40641183e-01 6.78417265e-01 6.61970675e-01 1.24344833e-01
7.71983743e-01 4.36969489e-01 5.78146279e-01 -6.84575617e-01
-1.05671620e+00 -8.31518531e-01 -7.12543368e-01 -3.73120576e-01
1.09941518e+00 -1.09903336e+00 -4.81865227e-01 2.75602102e-01
-6.75535083e-01 -1.45900428e-01 -2.22793102e-01 8.01588655e-01
-1.95113569e-01 2.97140777e-01 -5.73672473e-01 -3.86568129e-01
-1.02325678e+00 -1.27070153e+00 7.05696881e-01 4.79755133e-01
4.33610193e-02 -1.09636259e+00 -1.62807226e-01 5.67208648e-01
7.64783680e-01 2.99893141e-01 8.04212451e-01 -9.00789618e-01
-5.15925348e-01 -5.54161966e-01 -8.77821922e-01 3.27141136e-01
4.28970486e-01 -1.24419563e-01 -7.04236984e-01 -3.55895311e-01
-1.86648220e-02 -1.85287118e-01 8.77314925e-01 8.48667204e-01
1.45483387e+00 -3.13448310e-02 -6.92142665e-01 5.37696898e-01
1.50139523e+00 2.97243267e-01 5.46193495e-02 1.02492869e-01
9.03037965e-01 -8.56321007e-02 5.78635395e-01 2.14343905e-01
1.10156998e-01 1.73568621e-01 7.52801716e-01 -5.65717220e-01
-6.70297205e-01 7.02827796e-02 -2.50183046e-01 6.68624640e-01
-1.39103815e-01 -2.19819441e-01 -1.43112814e+00 7.48587310e-01
-1.47881186e+00 -7.32105970e-01 -4.87756997e-01 2.09969664e+00
5.31432450e-01 1.05853930e-01 6.80991113e-02 -1.88313499e-01
8.20078194e-01 -2.59653836e-01 -8.02426040e-01 2.16422960e-01
6.95725441e-01 2.59815753e-01 8.96328270e-01 -6.29167408e-02
-1.51479650e+00 3.64027232e-01 5.52729988e+00 7.82812774e-01
-1.27155888e+00 4.10030425e-01 9.85356331e-01 -1.10738926e-01
4.52311814e-01 -5.69031775e-01 -4.74600792e-01 2.62419045e-01
8.39504004e-01 -1.68919921e-01 -2.34311908e-01 8.74571860e-01
4.65823323e-01 -1.62946969e-01 -1.02519250e+00 6.94896936e-01
-1.30159840e-01 -1.34688234e+00 -1.49950981e-01 3.96190882e-02
7.60016680e-01 3.12033802e-01 1.22501723e-01 2.87324011e-01
-9.30327177e-02 -1.13006592e+00 -1.22915678e-01 5.19767880e-01
1.18752420e+00 -6.11727238e-01 1.30531645e+00 5.47299445e-01
-1.35703385e+00 4.90426309e-02 -2.02372670e-01 5.45491397e-01
-4.67133641e-01 5.97829401e-01 -1.74102962e+00 3.91859174e-01
7.17975557e-01 4.69684333e-01 -8.90805304e-01 1.50145960e+00
2.98562765e-01 9.37091172e-01 -3.63371342e-01 -3.79343867e-01
5.39339006e-01 2.52538174e-01 3.47251624e-01 1.42323589e+00
7.84885287e-01 6.24851659e-02 2.42771044e-01 6.03418767e-01
-5.00714719e-01 1.44608334e-01 -4.80062097e-01 8.91493335e-02
3.31375092e-01 1.66851377e+00 -1.17642975e+00 -5.54620504e-01
-4.36803460e-01 6.99545205e-01 -2.88751274e-01 -2.14530021e-01
-1.28041267e+00 -1.89923510e-01 -2.80117780e-01 4.67587411e-01
2.65834987e-01 4.37202930e-01 -2.22356841e-01 -6.81942940e-01
-1.21381275e-01 -5.77134669e-01 5.06502986e-01 -4.85630959e-01
-9.23138142e-01 4.46954995e-01 -1.85838521e-01 -1.40101576e+00
1.03454571e-03 -5.73831022e-01 -8.78560364e-01 6.47675216e-01
-1.42082930e+00 -1.15603220e+00 -9.37453270e-01 3.80469263e-01
6.02199852e-01 -2.55333006e-01 6.38925552e-01 2.55003750e-01
-7.76011348e-01 5.96737385e-01 3.68792325e-01 2.97325671e-01
6.19386137e-01 -1.40122795e+00 -7.44397345e-04 8.45150113e-01
-3.75708312e-01 1.46043643e-01 -4.20538671e-02 -6.71619773e-01
-1.22806168e+00 -1.81862700e+00 4.11325395e-01 -1.37588859e-01
3.51304919e-01 4.02274877e-01 -6.06901228e-01 3.28014612e-01
4.63340580e-02 6.74705148e-01 5.27489424e-01 -8.76786411e-01
4.37824428e-01 -1.32317273e-02 -1.56154525e+00 4.64061290e-01
6.23878002e-01 -4.22778316e-02 -1.04107752e-01 6.82540059e-01
3.38961005e-01 -5.48466682e-01 -9.29190814e-01 7.33782351e-01
6.12851918e-01 -1.02778113e+00 1.11870527e+00 1.75456926e-02
5.35423458e-01 -2.31294960e-01 2.86699235e-01 -8.07049870e-01
-7.22916722e-01 1.65426239e-01 -9.74491388e-02 7.35649228e-01
3.98783386e-01 -2.97372550e-01 1.19753623e+00 3.72435093e-01
4.62088361e-03 -1.15253580e+00 -7.58712709e-01 -3.80564481e-01
8.11077654e-02 -3.12405569e-03 -1.87250689e-01 5.01207292e-01
-9.31494772e-01 -8.69387761e-02 1.21247791e-01 2.70917676e-02
7.75179505e-01 -3.98394138e-01 2.75385201e-01 -1.05900609e+00
-1.76961139e-01 -5.41339695e-01 -2.53536671e-01 -6.20136440e-01
-5.46113074e-01 -1.21013379e+00 1.87172517e-02 -2.12565112e+00
8.51717710e-01 -1.96165696e-01 -5.76982796e-01 5.66021860e-01
-3.67571741e-01 5.30531287e-01 1.53709287e-02 5.04366100e-01
-4.62702066e-01 7.74566084e-02 1.72143269e+00 -2.26738006e-01
9.20370743e-02 5.51059246e-01 -5.94323039e-01 8.51210237e-01
9.29032505e-01 -5.34262180e-01 -2.55532771e-01 -1.57483950e-01
-6.34007514e-01 3.95000607e-01 7.19142497e-01 -1.62079704e+00
3.62304866e-01 -7.95901101e-03 9.09005284e-01 -8.28220487e-01
-2.57458235e-03 -1.05731642e+00 1.45837385e-02 1.44063389e+00
-2.38639429e-01 -4.09103543e-01 2.58353710e-01 6.16517127e-01
6.34756908e-02 -2.90685683e-01 1.20141697e+00 -4.52693015e-01
-4.36581433e-01 4.14244473e-01 -3.58022481e-01 -9.91856828e-02
1.66251349e+00 -2.25546330e-01 -1.14345223e-01 8.12443271e-02
-5.40341437e-01 1.05982035e-01 1.36789856e-02 -5.11252098e-02
5.06096959e-01 -1.31234646e+00 -9.04377520e-01 -1.40025601e-01
-7.61938989e-02 2.65954196e-01 3.15195024e-01 1.28641808e+00
-1.20376956e+00 8.89777124e-01 -6.84700012e-02 -9.61375058e-01
-1.56738031e+00 8.15406218e-02 8.91085386e-01 -8.61644566e-01
-5.79462111e-01 8.34250093e-01 2.65022576e-01 -3.34501594e-01
3.01077724e-01 -5.15423536e-01 -2.11982995e-01 -3.76888603e-01
4.19262387e-02 5.58968425e-01 1.42508924e-01 -2.96275705e-01
-4.96125698e-01 4.28673625e-01 -2.62236416e-01 3.69701713e-01
8.56196404e-01 2.25654989e-01 4.27143276e-02 3.66086774e-02
9.78889883e-01 -3.34092706e-01 -8.88862848e-01 -2.96320260e-01
-2.20253840e-01 -2.13298142e-01 5.26311338e-01 -1.13091552e+00
-1.25766325e+00 8.80872250e-01 1.38242733e+00 -6.53814971e-02
1.11418915e+00 -8.79337266e-02 9.56943989e-01 5.19708455e-01
-2.80288458e-01 -4.36249405e-01 2.72282869e-01 2.52274156e-01
4.01138723e-01 -1.77572751e+00 1.13282457e-01 -3.03853482e-01
-6.63311601e-01 1.19710672e+00 8.95612359e-01 -6.41597882e-02
5.71105480e-01 -5.00079244e-02 8.14040974e-02 -1.60087243e-01
-5.37570775e-01 -3.01021218e-01 6.71279013e-01 3.89688581e-01
8.14481735e-01 2.57940292e-01 -1.78408936e-01 4.23638433e-01
3.74674797e-01 1.62471995e-01 3.18251312e-01 8.09101164e-01
-6.11638486e-01 -2.89918929e-01 -5.56271434e-01 1.18553877e+00
-9.56978977e-01 3.56671065e-02 -3.19882661e-01 1.08446991e+00
1.87524363e-01 6.67870343e-01 -1.28122017e-01 2.96244081e-02
2.49503702e-01 -2.16549963e-01 5.09041138e-02 -6.49098098e-01
-7.64359176e-01 1.24225572e-01 -1.88334867e-01 -3.09024930e-01
-4.41545963e-01 -3.63387436e-01 -1.24026620e+00 -1.41392320e-01
-3.10203254e-01 -1.27092585e-01 5.08916676e-01 6.55414164e-01
1.59079939e-01 1.11954319e+00 7.02335954e-01 -5.24837554e-01
-6.55550003e-01 -8.71327877e-01 -1.09901562e-01 8.49590376e-02
1.21086501e-01 -2.39715874e-01 -1.08208440e-01 -7.20954314e-02]
|
[15.422245025634766, -2.1526052951812744]
|
f4dda5bd-f09d-4857-ad1d-2c30f94f019f
|
in-sample-policy-iteration-for-offline
|
2306.05726
| null |
https://arxiv.org/abs/2306.05726v1
|
https://arxiv.org/pdf/2306.05726v1.pdf
|
In-Sample Policy Iteration for Offline Reinforcement Learning
|
Offline reinforcement learning (RL) seeks to derive an effective control policy from previously collected data. To circumvent errors due to inadequate data coverage, behavior-regularized methods optimize the control policy while concurrently minimizing deviation from the data collection policy. Nevertheless, these methods often exhibit subpar practical performance, particularly when the offline dataset is collected by sub-optimal policies. In this paper, we propose a novel algorithm employing in-sample policy iteration that substantially enhances behavior-regularized methods in offline RL. The core insight is that by continuously refining the policy used for behavior regularization, in-sample policy iteration gradually improves itself while implicitly avoids querying out-of-sample actions to avert catastrophic learning failures. Our theoretical analysis verifies its ability to learn the in-sample optimal policy, exclusively utilizing actions well-covered by the dataset. Moreover, we propose competitive policy improvement, a technique applying two competitive policies, both of which are trained by iteratively improving over the best competitor. We show that this simple yet potent technique significantly enhances learning efficiency when function approximation is applied. Lastly, experimental results on the D4RL benchmark indicate that our algorithm outperforms previous state-of-the-art methods in most tasks.
|
['Zhaopeng Meng', 'Yan Zheng', 'Chenjun Xiao', 'Yi Ma', 'Xiaohan Hu']
|
2023-06-09
| null | null | null | null |
['offline-rl', 'd4rl']
|
['playing-games', 'robots']
|
[ 1.68385189e-02 1.41264305e-01 -8.89943600e-01 -4.49993946e-02
-1.27079475e+00 -6.21833026e-01 1.86415181e-01 2.30536625e-01
-8.80351603e-01 1.24788690e+00 5.26534393e-02 -3.74964982e-01
-3.23730737e-01 -4.56917346e-01 -1.09667170e+00 -7.65510678e-01
-1.80994183e-01 6.71519816e-01 -1.22170351e-01 7.97059163e-02
3.64507735e-01 3.05285662e-01 -1.44511282e+00 -2.24554278e-02
1.10962284e+00 1.03505981e+00 2.92923987e-01 4.16968852e-01
1.04126394e-01 1.13932812e+00 -6.41437948e-01 1.12767309e-01
5.58279991e-01 -3.02352577e-01 -5.77694833e-01 5.08953989e-01
4.12483573e-01 -8.23784232e-01 -1.81364641e-01 8.08243454e-01
3.23067397e-01 4.34015930e-01 3.95600572e-02 -9.80775476e-01
-4.49490905e-01 7.17815995e-01 -5.31114042e-01 1.03293210e-01
2.19507903e-01 4.89362687e-01 9.52239633e-01 -4.65508908e-01
4.76272464e-01 1.20137596e+00 3.33933651e-01 7.32527375e-01
-1.49572480e+00 -4.64549214e-01 8.02518070e-01 -2.53091548e-02
-8.71680856e-01 -3.54334176e-01 6.43245399e-01 5.75960614e-02
8.87331665e-01 5.57460375e-02 7.05285311e-01 1.01167154e+00
-2.28539094e-01 1.42699194e+00 1.38720894e+00 -2.79101610e-01
7.05091357e-01 1.74367011e-01 -3.79165933e-02 7.55159557e-01
2.42106587e-01 4.39736515e-01 -6.91257954e-01 -5.15489399e-01
7.38102734e-01 -1.05756354e-02 -1.94268495e-01 -6.75679326e-01
-8.24312747e-01 8.05766523e-01 8.31858441e-02 -4.54052500e-02
-6.73354864e-01 5.09744346e-01 3.37445527e-01 6.92193449e-01
5.20248592e-01 7.72943676e-01 -9.74484861e-01 -5.10197282e-01
-7.24142790e-01 6.42839074e-01 6.27787888e-01 9.16657984e-01
9.43598390e-01 2.95944691e-01 -4.89634931e-01 6.89368069e-01
-1.54156625e-01 4.84535933e-01 4.19871479e-01 -1.57605517e+00
6.02356911e-01 4.48291510e-01 7.32042074e-01 -1.13235399e-01
-5.33021651e-02 -5.37278593e-01 -3.53251666e-01 2.11850181e-01
6.02485955e-01 -3.22558820e-01 -6.09088659e-01 1.87551808e+00
5.87015510e-01 2.27436483e-01 8.73495936e-02 6.59876525e-01
-3.66173029e-01 2.83206940e-01 4.22644503e-02 -9.07658935e-01
6.78358197e-01 -9.54051435e-01 -6.82083130e-01 -1.27990216e-01
7.18563795e-01 -6.60907477e-02 1.47385597e+00 8.55693400e-01
-1.28577197e+00 -4.53874499e-01 -6.93480909e-01 4.38746542e-01
9.83078480e-02 1.36847720e-01 6.51612282e-01 4.36789185e-01
-8.79314542e-01 9.13403094e-01 -9.83142614e-01 1.53693408e-02
7.43985355e-01 4.25969422e-01 5.64771611e-03 1.17931351e-01
-8.17939579e-01 7.19576418e-01 3.87098551e-01 -3.99406701e-01
-1.61140251e+00 -1.07895052e+00 -4.78591770e-01 -1.18262865e-01
1.36660838e+00 -3.45685035e-01 1.87906861e+00 -9.20039594e-01
-1.78736770e+00 2.72479922e-01 -2.18315616e-01 -1.00591981e+00
7.88929522e-01 -6.50681555e-01 4.85330895e-02 2.46973470e-01
-2.09891982e-02 4.18898433e-01 1.14752650e+00 -1.40585506e+00
-1.09919298e+00 -3.66366923e-01 2.70630270e-01 4.02490050e-01
-5.00252426e-01 -5.87540805e-01 -3.66347253e-01 -4.39642876e-01
-4.60789710e-01 -7.38348603e-01 -6.70848131e-01 -1.67342722e-01
-7.32514858e-02 -5.45837045e-01 9.00742710e-01 -2.94185460e-01
1.51555371e+00 -1.83219302e+00 -3.31736170e-02 2.63652056e-01
1.85261920e-01 2.84713745e-01 -1.85748726e-01 4.29826826e-01
4.19013470e-01 8.40316061e-03 -1.92166775e-01 -4.75955933e-01
-3.91661786e-02 5.81427634e-01 -5.64798415e-01 6.57883584e-01
-1.09487131e-01 8.07051480e-01 -1.16931212e+00 -1.62043497e-01
2.17130139e-01 -2.78399378e-01 -9.71605599e-01 4.62335855e-01
-9.27780271e-01 7.56011605e-01 -8.50341439e-01 6.72538579e-01
3.36281717e-01 -2.16134205e-01 4.57604289e-01 2.55599767e-01
-1.18532851e-01 1.44521624e-01 -1.06415033e+00 1.68219852e+00
-5.25124490e-01 5.81802847e-03 1.56556174e-01 -1.40243089e+00
7.17248738e-01 1.07752465e-01 1.09276402e+00 -9.25235152e-01
-1.61875978e-01 1.36391506e-01 -4.90050375e-01 -4.78758216e-01
3.69708627e-01 2.60462821e-01 9.33775380e-02 6.05480552e-01
-1.85380757e-01 8.76006037e-02 4.34203118e-01 3.58854830e-02
1.23858595e+00 4.10808235e-01 4.67042834e-01 -1.83843955e-01
3.73670757e-01 2.79716421e-02 6.16722047e-01 1.42360735e+00
-5.16158402e-01 -2.23014206e-01 7.16322780e-01 -4.31412220e-01
-9.00038898e-01 -8.32177162e-01 1.45472139e-01 1.33419216e+00
-3.95642817e-02 -2.42964476e-01 -7.56977499e-01 -1.32096696e+00
6.36419237e-01 9.61505413e-01 -7.94633925e-01 -6.32609949e-02
-7.55435228e-01 -4.63293970e-01 2.34860390e-01 4.42473412e-01
4.79139298e-01 -1.05745304e+00 -8.69559288e-01 6.18187666e-01
5.34874760e-02 -9.07196999e-01 -5.84664404e-01 2.46816292e-01
-1.16759610e+00 -1.21479762e+00 -4.21602011e-01 -1.84440270e-01
8.04936469e-01 3.99927467e-01 1.07604814e+00 1.50234237e-01
1.66361004e-01 8.94639969e-01 -3.02935243e-01 -3.93698990e-01
-3.69089454e-01 1.20383829e-01 3.36789757e-01 -7.45077431e-02
2.99159616e-01 -4.37978715e-01 -6.27785802e-01 1.49354085e-01
-7.04073727e-01 -5.60793817e-01 5.37327766e-01 9.80616271e-01
1.02136064e+00 -8.34816992e-02 8.86014342e-01 -1.12842441e+00
8.69906366e-01 -5.34633160e-01 -9.81934190e-01 3.24269384e-01
-1.26098847e+00 5.11524439e-01 1.06561100e+00 -7.27346957e-01
-1.01403165e+00 5.62451854e-02 2.63370126e-01 -9.19660032e-01
-5.08041233e-02 2.12077066e-01 2.17983246e-01 2.39055440e-01
6.51443362e-01 5.53212702e-01 3.77703160e-01 -5.56551099e-01
3.87758225e-01 2.54509568e-01 2.89427906e-01 -1.20762467e+00
5.00408590e-01 4.87231195e-01 -2.60791034e-01 -5.08004844e-01
-1.24875140e+00 -4.73029256e-01 -4.55107801e-02 -2.79736429e-01
2.54027814e-01 -8.58544886e-01 -1.27866924e+00 1.29274011e-01
-4.37875897e-01 -9.42796111e-01 -8.43547523e-01 3.27264428e-01
-1.12563312e+00 1.96674600e-01 -2.90568292e-01 -1.28509533e+00
-1.80667862e-01 -1.03532600e+00 9.38303471e-01 1.88926727e-01
1.63470969e-01 -9.22203600e-01 2.30995759e-01 1.94567010e-01
1.45636261e-01 -1.43579114e-02 5.45526266e-01 -6.27504766e-01
-6.00342572e-01 2.08031580e-01 3.54173601e-01 4.21061665e-01
2.68040359e-01 -4.10970837e-01 -7.09637284e-01 -1.02202380e+00
-3.43800336e-02 -9.60662365e-01 8.56772780e-01 4.95656431e-01
1.94154167e+00 -7.10229099e-01 -1.19728781e-01 4.85671669e-01
1.45284367e+00 3.77006441e-01 1.37892485e-01 4.63435978e-01
3.97293478e-01 2.43810982e-01 1.23191524e+00 1.14187586e+00
1.66971728e-01 3.28560889e-01 5.07150292e-01 2.02530056e-01
2.52154201e-01 -7.11222529e-01 6.13504350e-01 1.79468557e-01
2.32738722e-02 2.76677795e-02 -3.34081978e-01 5.24469733e-01
-2.21646190e+00 -1.05767334e+00 4.91941959e-01 2.52604294e+00
1.33568001e+00 2.51334250e-01 5.23281336e-01 -9.99380425e-02
6.83806017e-02 3.43970060e-02 -1.46040046e+00 -1.56400830e-01
1.00575358e-01 3.57968479e-01 9.98780668e-01 5.90045214e-01
-9.74690974e-01 1.05905855e+00 6.44837618e+00 1.02185297e+00
-8.29883397e-01 6.57066032e-02 7.32164502e-01 -4.92798239e-01
-2.89978206e-01 -5.54011092e-02 -9.70835090e-01 4.20147985e-01
8.55260372e-01 -1.41433358e-01 1.18614340e+00 1.24434841e+00
6.85607374e-01 -4.02308911e-01 -1.28047299e+00 6.88976169e-01
-4.08704817e-01 -1.46684206e+00 -2.06966802e-01 1.42856628e-01
1.01381183e+00 -1.27976269e-01 1.53846040e-01 8.09661090e-01
8.11577499e-01 -6.11535609e-01 7.32254505e-01 4.87540752e-01
7.46800900e-01 -1.01926363e+00 2.26628613e-02 7.60964632e-01
-9.96328890e-01 -7.02866316e-01 -3.45116526e-01 7.21550779e-03
-2.38698676e-01 3.18255782e-01 -9.10420895e-01 3.00536096e-01
6.04878545e-01 7.20041990e-01 -2.78578997e-01 5.38658977e-01
-2.20895931e-01 9.00158823e-01 -2.85420299e-01 -2.07685843e-01
5.32164991e-01 -2.81098813e-01 5.31818449e-01 6.91954434e-01
-3.71933058e-02 -1.15568461e-02 8.10303628e-01 6.85517251e-01
-5.79761714e-02 6.21254370e-03 -6.88077927e-01 -1.36899710e-01
7.00542033e-01 8.15313399e-01 -3.34291570e-02 -3.79229695e-01
-2.36659750e-01 6.02117121e-01 1.08352590e+00 5.88252366e-01
-7.62038529e-01 2.35240802e-01 7.64053285e-01 -1.88675169e-02
4.93707418e-01 -3.03845644e-01 -1.23460978e-01 -9.15008783e-01
8.67680758e-02 -1.35568547e+00 5.95210433e-01 1.18412720e-02
-1.25983310e+00 -8.64920691e-02 1.15442783e-01 -1.23277986e+00
-3.74330193e-01 -2.58852869e-01 -1.61067680e-01 3.16408426e-01
-1.62311304e+00 -4.95406836e-01 2.01832116e-01 7.37105429e-01
9.12205517e-01 -2.92089075e-01 5.15207708e-01 -4.38696798e-03
-6.72375262e-01 7.73198187e-01 5.48214078e-01 -4.69059855e-01
6.55656934e-01 -1.52856314e+00 -1.86463326e-01 5.35323799e-01
-6.77505285e-02 4.80781883e-01 5.81649005e-01 -7.12413371e-01
-1.76782906e+00 -1.08907974e+00 -4.30719256e-02 -2.43045002e-01
7.48465896e-01 -1.11628421e-01 -7.91194141e-01 6.88421071e-01
-7.66959935e-02 -5.30417785e-02 3.65545332e-01 4.96549085e-02
-2.59307250e-02 -5.72661579e-01 -1.23929107e+00 6.35293782e-01
1.02601993e+00 -1.72049448e-01 -2.81794757e-01 4.89567161e-01
8.36159825e-01 -5.02206743e-01 -7.99224138e-01 2.30557755e-01
2.84268171e-01 -8.06789637e-01 6.94216669e-01 -1.09241307e+00
1.46013036e-01 -4.67834761e-03 -6.99355081e-02 -1.36113966e+00
3.29798013e-02 -1.27414680e+00 -1.00284362e+00 6.89784229e-01
1.72557101e-01 -7.47848630e-01 1.02549553e+00 4.94158268e-01
4.07360606e-02 -1.25509799e+00 -7.87850142e-01 -1.07587409e+00
1.04720831e-01 -4.27768707e-01 4.91261035e-01 6.34965539e-01
-2.01214597e-01 -1.84816957e-01 -6.99462533e-01 -8.22471902e-02
1.09102523e+00 1.15055464e-01 1.02296829e+00 -6.67859077e-01
-7.45958984e-01 -2.57984579e-01 7.91690707e-01 -1.76245272e+00
3.64948601e-01 -3.61613423e-01 9.08525288e-02 -1.01586986e+00
2.91686021e-02 -7.17866182e-01 -5.29126525e-01 6.45212710e-01
-3.57521117e-01 -6.13764942e-01 1.93614244e-01 2.37755179e-01
-1.00284898e+00 8.50133419e-01 1.72098112e+00 5.03234491e-02
-6.60489559e-01 2.89213479e-01 -7.79235840e-01 4.22428042e-01
1.00867319e+00 -5.86017847e-01 -8.76088500e-01 -1.28935084e-01
-6.59923702e-02 4.04392719e-01 6.27606362e-02 -6.19462788e-01
6.62344927e-03 -7.88160503e-01 1.46252081e-01 -6.27156377e-01
6.28355592e-02 -7.68513799e-01 -4.42717731e-01 7.27299273e-01
-8.78889501e-01 -9.51463506e-02 3.16806138e-02 1.03737605e+00
2.79404342e-01 -2.33607545e-01 8.56003821e-01 -4.03883964e-01
-5.04908025e-01 6.33022249e-01 -3.63236129e-01 4.06920105e-01
1.26757836e+00 4.74126153e-02 -2.13859737e-01 -4.40334857e-01
-6.10354900e-01 8.08940470e-01 1.43953398e-01 9.15813446e-02
5.59795320e-01 -1.14020729e+00 -5.64225793e-01 4.79333363e-02
-1.22078426e-01 3.25166285e-02 -2.67077778e-02 8.26777399e-01
1.59914643e-01 3.19917321e-01 2.28242621e-01 -5.15002131e-01
-7.99858928e-01 7.78806686e-01 3.91242743e-01 -9.01645422e-01
-5.15770853e-01 5.29788792e-01 -2.28840128e-01 -3.24071109e-01
7.45876849e-01 -4.99990195e-01 1.06389858e-01 -1.17658436e-01
4.99332607e-01 6.49770677e-01 -1.70490205e-01 5.28453171e-01
8.24242756e-02 3.84705551e-02 -3.81679207e-01 -2.67438859e-01
1.29515803e+00 -1.30728275e-01 4.52704012e-01 4.25465405e-01
9.31249082e-01 4.09390293e-02 -2.22485089e+00 -5.71332335e-01
9.01934877e-02 -6.29068553e-01 5.88136874e-02 -9.37929094e-01
-1.01255178e+00 1.87938601e-01 4.11031991e-01 3.48681360e-01
1.22628510e+00 -2.58780748e-01 7.76639998e-01 8.99443448e-01
7.61914551e-01 -1.77500045e+00 4.42136049e-01 3.56083632e-01
7.65352070e-01 -1.34984386e+00 1.58052295e-01 3.14410567e-01
-9.16017592e-01 9.23012257e-01 9.69096661e-01 -4.09869492e-01
3.63043457e-01 1.78906381e-01 -2.11348176e-01 1.29519314e-01
-1.30678785e+00 -4.16454732e-01 -2.36873165e-01 4.92185980e-01
-2.26165995e-01 6.24666326e-02 -3.39186281e-01 2.53287524e-01
2.09818155e-01 8.97955000e-02 3.20096791e-01 1.29303539e+00
-6.75225019e-01 -1.19484782e+00 -2.62763500e-01 6.92700326e-01
-4.24969405e-01 3.23108971e-01 -9.83655378e-02 9.38270509e-01
-2.67853439e-01 9.54032421e-01 8.40154216e-02 -2.89722774e-02
4.07902598e-01 -2.69619703e-01 7.27162778e-01 -2.68838763e-01
-6.93942249e-01 3.07016760e-01 4.98878770e-02 -1.16794848e+00
-4.26637352e-01 -7.63721406e-01 -1.41206324e+00 -1.78371340e-01
-4.03854921e-02 2.23400146e-01 2.53982604e-01 1.04212606e+00
3.53935957e-01 3.92542630e-01 1.23816824e+00 -4.33540016e-01
-1.68830502e+00 -7.35997558e-01 -5.11018693e-01 4.22431171e-01
7.04679370e-01 -7.95953810e-01 -2.70479292e-01 -2.09413871e-01]
|
[4.056641578674316, 2.283149480819702]
|
548d49d5-e5ad-4f0f-bcd6-bad8444c3d53
|
fast-and-efficient-calculations-of-structural
|
1711.05866
| null |
http://arxiv.org/abs/1711.05866v2
|
http://arxiv.org/pdf/1711.05866v2.pdf
|
Fast and Efficient Calculations of Structural Invariants of Chirality
|
Chirality plays an important role in physics, chemistry, biology, and other
fields. It describes an essential symmetry in structure. However, chirality
invariants are usually complicated in expression or difficult to evaluate. In
this paper, we present five general three-dimensional chirality invariants
based on the generating functions. And the five chiral invariants have four
characteristics:(1) They play an important role in the detection of symmetry,
especially in the treatment of 'false zero' problem. (2) Three of the five
chiral invariants decode an universal chirality index. (3) Three of them are
proposed for the first time. (4) The five chiral invariants have low order no
bigger than 4, brief expression, low time complexity O(n) and can act as
descriptors of three-dimensional objects in shape analysis. The five chiral
invariants give a geometric view to study the chiral invariants. And the
experiments show that the five chirality invariants are effective and
efficient, they can be used as a tool for symmetry detection or features in
shape analysis.
|
['Hanlin Mo', 'You Hao', 'He Zhang', 'Shirui Li', 'Hua Li']
|
2017-10-20
| null | null | null | null |
['symmetry-detection']
|
['computer-vision']
|
[-9.66678634e-02 -3.78554434e-01 -2.83148736e-01 4.42837253e-02
1.21305175e-01 -6.53322995e-01 8.46462607e-01 3.03897839e-02
-1.34890854e-01 6.15917802e-01 3.24310750e-01 -3.17039490e-01
-4.96476054e-01 -8.27322781e-01 -6.59692958e-02 -1.01650894e+00
-4.06494081e-01 4.92951721e-01 6.13922477e-01 -5.00782669e-01
7.44600892e-01 8.02800894e-01 -1.61873662e+00 -1.32797867e-01
8.30173016e-01 8.83712053e-01 -7.75153786e-02 4.35181975e-01
-1.34502754e-01 4.39987570e-01 2.33390182e-02 -1.58632249e-02
2.64760375e-01 -3.62454951e-01 -8.94219339e-01 -1.17899917e-01
-3.10787231e-01 2.02218611e-02 8.78617633e-03 1.15232849e+00
3.77821267e-01 -1.65125281e-01 1.19482863e+00 -1.19325161e+00
-5.60913026e-01 1.35125086e-01 -4.10234272e-01 -4.48169149e-02
4.99098897e-01 -3.50364953e-01 8.75359595e-01 -8.18291187e-01
6.00695550e-01 1.16601050e+00 4.12162632e-01 5.66663086e-01
-7.55845070e-01 -6.10880852e-01 -5.41950524e-01 3.90027851e-01
-1.18657172e+00 -2.33369127e-01 6.19983494e-01 -6.58905387e-01
3.47464681e-01 6.86030090e-01 7.25650489e-01 2.30607718e-01
5.08221447e-01 3.85257155e-01 1.35037518e+00 -5.29213011e-01
-2.70600140e-01 -1.05122022e-01 1.04852259e-01 9.56293523e-01
4.54141974e-01 9.05471519e-02 3.60423401e-02 -1.57260910e-01
8.55628490e-01 6.55591562e-02 -4.70953397e-02 -6.08392656e-01
-1.48942685e+00 6.35024369e-01 1.81010962e-01 6.62269354e-01
1.79951638e-01 -3.69419932e-01 2.92626858e-01 1.67175367e-01
-9.81511548e-02 5.60795367e-01 -6.67717606e-02 -6.68585449e-02
-2.38862872e-01 3.43220025e-01 6.60460234e-01 6.75243497e-01
6.10968411e-01 1.42950509e-02 2.95778625e-02 7.60520041e-01
1.75355792e-01 8.22156966e-01 4.22865242e-01 -4.18915749e-01
6.75115734e-02 7.55963683e-01 -9.19142216e-02 -9.45321202e-01
-8.26114476e-01 7.32002333e-02 -1.09689987e+00 1.72299400e-01
2.24560931e-01 5.25855601e-01 -5.58491051e-01 1.02116692e+00
3.43811810e-02 -2.72074282e-01 1.13837868e-01 8.01414907e-01
1.19281721e+00 6.68260217e-01 -2.41398856e-01 -2.20438540e-01
1.43656611e+00 -5.12997687e-01 -3.22760433e-01 6.67391419e-01
3.04351658e-01 -1.43270564e+00 5.60033679e-01 3.69628221e-01
-8.60019386e-01 -9.75195842e-04 -1.17343354e+00 3.36666286e-01
-5.21029294e-01 3.78820091e-01 1.06989133e+00 6.73163593e-01
-7.11718678e-01 5.82639635e-01 -1.65727973e-01 -3.00025761e-01
-3.65949631e-01 4.30756211e-01 -6.22666180e-01 3.89094442e-01
-9.86084521e-01 7.36186147e-01 4.80017275e-01 -6.00119419e-02
-2.80163884e-01 -7.56374821e-02 -4.89602774e-01 -1.86649859e-01
1.40911937e-01 -9.33874771e-02 8.74843180e-01 -5.65844536e-01
-1.29920936e+00 9.89789486e-01 -2.44007587e-01 3.48551691e-01
3.38270992e-01 6.46440148e-01 -6.85985804e-01 7.30068758e-02
3.79530251e-01 5.80495335e-02 2.41412938e-01 -9.30636048e-01
-3.26958239e-01 -3.20076764e-01 -1.16608362e-03 -7.06304982e-03
7.30352551e-02 1.60859302e-01 -5.44581339e-02 -3.83350194e-01
1.03780401e+00 -9.60251570e-01 8.90479386e-02 -3.64478588e-01
-4.36735988e-01 -2.49735728e-01 9.73679841e-01 -3.58528435e-01
1.08208394e+00 -1.89344895e+00 -1.50147581e-03 7.26667345e-01
1.32142812e-01 5.90577841e-01 9.38879699e-02 6.46951020e-01
-1.95273072e-01 3.10522407e-01 5.62834851e-02 1.07460940e+00
-9.33513343e-02 -5.14557287e-02 7.06557333e-02 4.85546380e-01
1.94346577e-01 4.76818532e-01 -6.90162241e-01 -5.05072534e-01
1.75704822e-01 2.24829599e-01 -3.79234344e-01 -2.37411022e-01
3.15791577e-01 4.69992667e-01 -7.30430782e-01 6.56673729e-01
1.03796744e+00 9.05978307e-02 2.82271326e-01 -5.31796336e-01
-5.66937745e-01 5.71043827e-02 -1.53586388e+00 4.80220169e-01
1.45225942e-01 4.17815179e-01 -2.04250976e-01 -8.80624592e-01
1.11404467e+00 1.58306718e-01 6.23645484e-01 -1.15206349e+00
3.62898558e-02 4.12838548e-01 3.51794243e-01 -4.57213700e-01
3.38485926e-01 -2.56280571e-01 9.93502885e-02 6.15554214e-01
-1.17085539e-02 -1.52865890e-02 7.23738849e-01 4.74591330e-02
3.62187326e-01 -2.48744771e-01 4.42462474e-01 -6.32843435e-01
1.28396940e+00 -6.04351878e-01 7.09968746e-01 7.26597086e-02
1.49239078e-01 2.41973177e-01 9.12259161e-01 -6.38162494e-01
-1.35259688e+00 -9.74987149e-01 -5.52867889e-01 4.69642580e-01
3.33457261e-01 -4.20019269e-01 -4.55086887e-01 -2.92820297e-03
-1.80923790e-01 -1.84625611e-01 -2.04075083e-01 -1.32917076e-01
-5.44304907e-01 -7.06827700e-01 3.22940737e-01 3.45186621e-01
9.06402171e-01 -7.59503543e-01 -2.40436018e-01 -1.04265280e-01
-6.43164366e-02 -8.25874269e-01 -2.47209072e-01 -1.52041093e-01
-1.02221286e+00 -1.66290975e+00 -6.74654424e-01 -6.88657701e-01
8.78256500e-01 4.45714265e-01 7.25500226e-01 3.00849676e-01
-3.81013215e-01 2.55081385e-01 -4.29079562e-01 -4.54258084e-01
-5.60091734e-01 -5.46210110e-02 2.84895778e-01 -1.65417209e-01
1.49723008e-01 -3.93559366e-01 -3.84496838e-01 6.72935903e-01
-6.85495019e-01 -1.02962695e-01 5.56002319e-01 7.95521438e-01
2.94370383e-01 -1.26078337e-01 2.16979280e-01 -6.35126591e-01
7.32384801e-01 1.71194598e-01 -9.95622993e-01 3.17098141e-01
-5.86260021e-01 8.59260142e-01 6.38721168e-01 1.37714937e-01
-7.03298271e-01 -2.23703131e-01 1.08784221e-01 3.44525605e-01
4.24185842e-01 1.89104334e-01 -1.15623638e-01 -6.53197229e-01
3.36711794e-01 4.80890810e-01 2.20527202e-02 -4.83850300e-01
-3.35794121e-01 5.96993625e-01 1.99932963e-01 -9.74280834e-01
7.99571931e-01 3.06696624e-01 1.27921045e+00 -1.17442286e+00
-3.61547530e-01 -7.17739582e-01 -9.00475621e-01 -2.74039298e-01
7.23196924e-01 -2.71576047e-01 -1.47326422e+00 3.84481341e-01
-1.05226576e+00 7.13958442e-01 4.45981115e-01 6.81792319e-01
-5.50372601e-01 8.28616619e-01 -1.73906669e-01 -1.02220535e+00
-3.87852579e-01 -1.18598986e+00 6.83981597e-01 3.77547085e-01
2.93843061e-01 -9.78204906e-01 5.71850315e-02 -1.74153998e-01
1.28944263e-01 3.23628306e-01 1.43532288e+00 -2.40555838e-01
-7.01772213e-01 -4.21362333e-02 -4.85568434e-01 7.64018595e-02
-5.60764819e-02 3.67762297e-01 -4.67265338e-01 -3.35988589e-02
-5.08646250e-01 1.47975430e-01 7.21864522e-01 -3.29040587e-02
8.91366601e-01 -1.98798880e-01 -2.14924350e-01 5.91310740e-01
1.23806000e+00 6.37579381e-01 7.67699182e-01 4.29551780e-01
3.34071994e-01 3.58745158e-01 7.26134717e-01 4.33011591e-01
-8.43677595e-02 6.25212312e-01 4.62402046e-01 -2.82121003e-02
2.49840587e-01 -1.27995685e-01 1.58576712e-01 1.16949403e+00
-1.11083281e+00 5.99754930e-01 -9.06851888e-01 6.09817654e-02
-1.69465387e+00 -1.43248796e+00 -6.19552493e-01 2.59677434e+00
4.62135941e-01 -1.56928554e-01 4.77526724e-01 4.13014203e-01
7.16361880e-01 5.67286313e-02 -4.42667753e-02 -7.79947221e-01
-3.38945091e-01 1.39619440e-01 6.64922237e-01 2.46617928e-01
-1.22900307e+00 6.77326560e-01 7.01315069e+00 4.98031408e-01
-1.23083615e+00 -3.26072723e-01 -1.19731329e-01 6.63856924e-01
-4.19963926e-01 3.78019989e-01 -6.97923124e-01 3.59370470e-01
1.09674737e-01 -5.57621360e-01 2.43359938e-01 6.04357183e-01
-3.80658768e-02 -2.92881787e-01 -6.68102801e-01 1.10042906e+00
1.51396478e-02 -1.20827138e+00 3.40453655e-01 2.60566056e-01
6.80833161e-01 -4.52908576e-01 1.76019073e-02 -2.63315380e-01
-4.65106517e-01 -9.64185655e-01 3.60145271e-01 6.03222549e-01
7.90019989e-01 -1.06576645e+00 7.20384896e-01 2.02076986e-01
-1.49201679e+00 4.26735803e-02 -8.02555203e-01 -1.69004668e-02
-2.45486394e-01 2.59435713e-01 -7.73630381e-01 8.15066516e-01
1.69143349e-01 5.25952637e-01 -4.63421077e-01 1.32796848e+00
-3.79122123e-02 2.12595854e-02 -1.60979524e-01 -7.12133169e-01
4.58474010e-01 -9.45639014e-01 5.61215222e-01 8.65265429e-01
4.77500856e-01 1.98363304e-01 -2.64634043e-02 6.48586571e-01
3.65979731e-01 5.76606512e-01 -7.91747153e-01 -3.58016282e-01
-3.21232416e-02 1.20271027e+00 -9.15854335e-01 9.95545015e-02
-2.44965106e-01 4.36450124e-01 -4.90629911e-01 4.03755121e-02
-4.69105631e-01 -7.29587257e-01 4.18308884e-01 2.87888616e-01
-7.12090731e-02 -4.59526449e-01 3.37349415e-01 -1.28823864e+00
-1.06744757e-02 -6.58547699e-01 1.23533569e-01 -4.89447773e-01
-6.51294768e-01 3.47326063e-02 5.22585064e-02 -1.53505552e+00
-2.48407140e-01 -1.55806518e+00 -7.24768519e-01 6.80297256e-01
-1.00217390e+00 -1.01596820e+00 -1.68327868e-01 6.40705645e-01
-2.09380686e-01 -7.25017726e-01 1.01425993e+00 4.15037304e-01
-1.53346881e-01 1.53153121e-01 5.74847579e-01 3.41021299e-01
4.76188779e-01 -1.17942965e+00 -1.76665962e-01 4.25268978e-01
-7.10818470e-02 7.54563928e-01 4.70197916e-01 -3.97332489e-01
-1.51102984e+00 -2.10700005e-01 9.72558081e-01 1.86181873e-01
3.53426486e-01 1.02414861e-02 -1.87428057e-01 1.46103501e-01
-3.02750021e-01 -4.05682147e-01 6.89932764e-01 2.38232300e-01
-3.97861242e-01 -2.85249442e-01 -9.73188877e-01 5.44399917e-01
7.86780238e-01 -4.71402138e-01 -3.21764648e-01 4.64599103e-01
-4.20101080e-03 -1.17391668e-01 -7.51392066e-01 4.77247894e-01
1.05459929e+00 -1.41176307e+00 9.44785833e-01 -7.02472389e-01
1.74726192e-02 -6.05070055e-01 9.68625918e-02 -7.34784484e-01
-6.17470205e-01 -5.16684175e-01 6.58230364e-01 8.78404737e-01
3.32474470e-01 -7.87908554e-01 4.35925961e-01 5.85510023e-02
1.33809030e-01 -5.17031312e-01 -8.99377167e-01 -1.08536220e+00
2.62967348e-02 2.57988453e-01 8.06900561e-01 1.10044122e+00
2.62622386e-01 4.38401729e-01 -3.31034571e-01 -4.03638363e-01
5.38366377e-01 1.96425349e-01 6.08325601e-01 -1.91893101e+00
1.80166870e-01 -5.07301211e-01 -1.28299057e+00 -4.18685615e-01
8.73974338e-02 -9.92669046e-01 -8.24139833e-01 -1.24059200e+00
7.24528909e-01 -3.21569622e-01 -2.22433075e-01 2.97949582e-01
6.45514071e-01 8.29687268e-02 -8.56413879e-03 1.40878215e-01
-3.74738455e-01 2.06091642e-01 1.58559144e+00 -1.32118568e-01
1.59883380e-01 2.52908647e-01 -2.61759877e-01 8.58328998e-01
8.61683369e-01 2.44021937e-02 -2.06658721e-01 4.26059812e-01
6.52023256e-01 -1.81614384e-01 2.94698536e-01 -1.00253272e+00
8.79465789e-02 -4.88455445e-01 4.22486067e-01 -7.02521920e-01
6.67849556e-02 -6.60770833e-01 6.36644214e-02 7.34628201e-01
4.61959571e-01 1.58450544e-01 -2.04390243e-01 -1.29530013e-01
-4.59867060e-01 -9.23218369e-01 9.38838243e-01 -2.84296304e-01
-1.06320322e+00 2.97050834e-01 -1.57769442e-01 -8.13966617e-03
1.08912683e+00 -3.99472535e-01 -3.94463211e-01 -3.64302456e-01
-5.16857743e-01 -2.19981894e-01 4.24993038e-01 2.39314407e-01
5.96088707e-01 -1.74045455e+00 -3.13982666e-01 5.02942026e-01
2.71639854e-01 -6.08344615e-01 -6.97661936e-02 1.01388264e+00
-1.22619104e+00 9.11082506e-01 -7.67724156e-01 -6.74145103e-01
-1.56627810e+00 3.22392225e-01 4.32437390e-01 -1.37231961e-01
-1.53713733e-01 -5.82474470e-02 2.24914372e-01 -5.62949300e-01
-1.74706712e-01 -2.32864469e-02 -4.94036436e-01 2.51057222e-02
4.05168712e-01 7.78925240e-01 -4.93748970e-02 -1.27358282e+00
-4.39912558e-01 1.46640015e+00 2.13650674e-01 -9.24716741e-02
1.17111802e+00 4.56525892e-01 -8.84946167e-01 3.21292192e-01
1.12452328e+00 2.31474057e-01 -7.67221302e-02 1.81567997e-01
7.75865540e-02 -4.74105686e-01 -5.69175899e-01 -3.65301192e-01
-5.01178026e-01 9.85443711e-01 5.60295224e-01 4.60302502e-01
8.91278982e-01 -6.46975487e-02 2.27210775e-01 5.78600466e-01
5.15283465e-01 -1.21472514e+00 -5.25363572e-02 1.03867877e+00
9.01362360e-01 -7.63029754e-01 2.38379374e-01 -7.02487946e-01
-4.20248091e-01 1.74222064e+00 4.16433662e-01 1.71209469e-01
6.29935980e-01 -7.93599337e-02 -2.89602250e-01 -4.08685692e-02
-2.69500464e-01 -4.78649557e-01 7.85780489e-01 6.34350002e-01
9.79372144e-01 2.12410003e-01 -1.11750734e+00 2.24858493e-01
-2.46507600e-01 -3.20614755e-01 4.42440897e-01 5.17773271e-01
-9.41428900e-01 -1.60033655e+00 -4.83869314e-01 3.04372460e-01
-2.90470809e-01 7.15242326e-02 -6.54381156e-01 7.84499049e-01
9.13845077e-02 5.42824090e-01 -2.40405142e-01 -3.51278305e-01
3.43841404e-01 6.80253878e-02 9.29761112e-01 5.02227247e-02
3.42100412e-02 -8.35225210e-02 2.20231689e-03 -4.93370682e-01
-5.10523200e-01 -5.05173147e-01 -1.24016595e+00 -6.48115516e-01
-1.81244776e-01 5.49295247e-01 6.49861395e-01 8.23964000e-01
6.25938997e-02 -9.07170475e-02 7.75784075e-01 -3.25453758e-01
-5.13485789e-01 -6.41780317e-01 -1.09140301e+00 3.90637994e-01
-6.92445859e-02 -9.19808865e-01 -9.53644738e-02 -3.68387312e-01]
|
[9.307497024536133, -1.7858189344406128]
|
3119cafa-3c14-4c82-8435-1d9ffba188f2
|
decentralized-structural-rnn-for-robot-crowd
|
2011.0482
| null |
https://arxiv.org/abs/2011.04820v3
|
https://arxiv.org/pdf/2011.04820v3.pdf
|
Decentralized Structural-RNN for Robot Crowd Navigation with Deep Reinforcement Learning
|
Safe and efficient navigation through human crowds is an essential capability for mobile robots. Previous work on robot crowd navigation assumes that the dynamics of all agents are known and well-defined. In addition, the performance of previous methods deteriorates in partially observable environments and environments with dense crowds. To tackle these problems, we propose decentralized structural-Recurrent Neural Network (DS-RNN), a novel network that reasons about spatial and temporal relationships for robot decision making in crowd navigation. We train our network with model-free deep reinforcement learning without any expert supervision. We demonstrate that our model outperforms previous methods in challenging crowd navigation scenarios. We successfully transfer the policy learned in the simulator to a real-world TurtleBot 2i.
|
['Katherine Driggs-Campbell', 'Neeloy Chakraborty', 'Weihang Liang', 'Peixin Chang', 'Shuijing Liu']
|
2020-11-09
| null | null | null | null |
['social-navigation']
|
['robots']
|
[-4.80217665e-01 2.74220526e-01 4.78871167e-01 8.55101389e-04
-1.33227244e-01 -2.79167801e-01 5.37245870e-01 -4.45452929e-01
-1.00757825e+00 1.26466489e+00 8.40758458e-02 -3.83413583e-01
2.92792231e-01 -6.68036342e-01 -7.97130406e-01 -6.87209427e-01
-3.91653329e-01 8.25866997e-01 7.32189178e-01 -1.06766629e+00
3.60418633e-02 2.42066592e-01 -1.25385511e+00 -4.84132767e-01
1.06084955e+00 2.41060823e-01 6.93856478e-01 9.72186804e-01
3.17551404e-01 1.62622571e+00 -5.85191309e-01 9.47541445e-02
3.31183612e-01 -2.85276890e-01 -7.69418120e-01 -1.21966787e-01
-3.95122498e-01 -5.22378564e-01 -8.81525934e-01 9.60719585e-01
7.62599230e-01 5.60497046e-01 6.74997807e-01 -1.42564368e+00
-3.96786034e-01 6.20455086e-01 -3.03685755e-01 -1.29126187e-03
2.34550372e-01 6.10769689e-01 3.21451098e-01 -6.06089771e-01
5.51702023e-01 1.47815287e+00 8.57573807e-01 9.33376968e-01
-4.10092801e-01 -4.40715283e-01 3.56975526e-01 2.51886100e-01
-9.51221108e-01 -2.10504442e-01 1.04519010e-01 -5.78036189e-01
1.11969268e+00 -7.30926514e-01 7.62638927e-01 1.37196124e+00
5.47534466e-01 7.99499094e-01 6.88805759e-01 1.59140408e-01
5.71555793e-01 -5.69184184e-01 -3.91136736e-01 7.96389580e-01
5.06607771e-01 1.70378238e-01 -3.24718654e-01 -1.88432783e-01
7.59119391e-01 -2.87251119e-02 -2.22801883e-02 -4.93239462e-01
-1.00182974e+00 8.05588663e-01 7.20767498e-01 -1.02586634e-01
-7.82526135e-01 7.38262415e-01 4.22546327e-01 2.63346165e-01
-2.80064195e-01 3.73223573e-01 -3.13220322e-01 -4.26429629e-01
1.02505654e-01 6.12082422e-01 1.17994940e+00 1.14900696e+00
6.35271788e-01 6.36501729e-01 2.15179056e-01 4.77890551e-01
3.46469790e-01 1.20597923e+00 6.84184968e-01 -1.58591735e+00
5.80443203e-01 2.31693894e-01 9.84151304e-01 -1.11304569e+00
-1.00914741e+00 2.20624986e-03 -6.58549190e-01 5.70283473e-01
5.04356861e-01 -9.56043303e-01 -7.49245346e-01 1.58980441e+00
2.76064903e-01 9.10440274e-03 5.89199007e-01 1.08615875e+00
6.64088309e-01 4.52335656e-01 -1.05644010e-01 3.17743242e-01
8.84564638e-01 -1.36197340e+00 -5.76860964e-01 -8.18597257e-01
6.70304596e-01 2.01187998e-01 6.46926641e-01 1.22254722e-01
-8.23269546e-01 -2.49573678e-01 -8.73367965e-01 1.34378716e-01
-3.36811304e-01 -1.35568336e-01 4.07296687e-01 4.84879203e-02
-1.43966532e+00 4.20061648e-01 -1.24958074e+00 -5.67815125e-01
1.51653871e-01 7.76858211e-01 -3.37497115e-01 1.76961735e-01
-1.38310623e+00 1.30930364e+00 5.35130091e-02 5.07158816e-01
-1.50986874e+00 4.80563730e-01 -1.13152063e+00 -3.68569195e-01
6.28728092e-01 -7.68611014e-01 1.97261226e+00 -3.53504330e-01
-2.09072804e+00 4.11308438e-01 -1.55982181e-01 -9.48325932e-01
5.93442798e-01 -3.85035485e-01 2.59073466e-01 2.83179078e-02
5.50653577e-01 6.68060243e-01 4.80040401e-01 -1.37856925e+00
-1.05403209e+00 -7.08252788e-02 2.65133232e-01 6.32537484e-01
3.51204008e-01 -4.27938700e-01 -1.88896582e-01 8.58855247e-02
-1.85352385e-01 -1.39291787e+00 -1.16510892e+00 -4.23935831e-01
-1.98918790e-01 -1.01532668e-01 5.76938689e-01 -4.25235182e-01
2.43643925e-01 -1.63722658e+00 1.32212415e-01 -9.96573865e-02
1.76658988e-01 3.22744250e-01 -2.53299028e-01 5.94191134e-01
8.17983508e-01 -4.60275590e-01 -2.06240833e-01 -4.42051440e-01
2.30035484e-01 7.38927186e-01 -4.43104088e-01 6.70852900e-01
-1.23764470e-01 1.01195705e+00 -1.51372051e+00 -1.58519879e-01
-6.32618889e-02 2.82007784e-01 -4.80820537e-01 3.10887456e-01
-2.01862186e-01 1.06938171e+00 -5.87389290e-01 2.83446074e-01
3.14790040e-01 -2.08612993e-01 2.71146923e-01 9.72974360e-01
-1.82509899e-01 1.40778035e-01 -9.07605588e-01 1.42160141e+00
-4.90765691e-01 7.45966077e-01 4.90198493e-01 -7.77451992e-01
8.10218573e-01 2.23789454e-01 2.99818933e-01 -8.73525977e-01
2.37053603e-01 4.53771979e-01 4.40090820e-02 -7.84552872e-01
8.43748152e-01 2.52308726e-01 -3.93872470e-01 3.06216359e-01
-2.67868459e-01 -3.11114460e-01 1.39028430e-02 1.35137245e-01
1.41467333e+00 2.31975004e-01 3.75177979e-01 -4.11069170e-02
2.64711738e-01 2.29100302e-01 7.67337859e-01 1.38005948e+00
-5.60614109e-01 2.43805096e-01 3.24259698e-01 -4.24925447e-01
-7.54261196e-01 -8.14623773e-01 8.30602050e-01 1.42828012e+00
7.92324960e-01 1.43835932e-01 -7.89600551e-01 -3.68045986e-01
9.99820622e-05 3.24640989e-01 -8.21629584e-01 1.41703963e-01
-1.22522748e+00 -4.58798051e-01 6.37818635e-01 5.47823191e-01
6.67743027e-01 -1.49909878e+00 -1.25447822e+00 5.65723419e-01
-2.32820794e-01 -1.28206170e+00 -1.71254128e-01 3.14384669e-01
-2.69977689e-01 -1.27198327e+00 -6.86959326e-01 -1.05197632e+00
6.03934586e-01 6.01782858e-01 9.77692723e-01 5.55751398e-02
2.59239286e-01 4.95327234e-01 -3.35458934e-01 -5.35519719e-01
-3.33716452e-01 1.87423363e-01 4.13948357e-01 -5.62710166e-01
1.80113669e-02 -4.33086276e-01 -6.24138653e-01 5.11024356e-01
-3.49414974e-01 -3.64873648e-01 2.87210196e-01 1.05241299e+00
-5.17264158e-02 -6.53287992e-02 6.25294328e-01 -4.37085420e-01
9.75484312e-01 -7.78772652e-01 -8.28906238e-01 -2.68835425e-01
6.52772933e-02 2.31145415e-02 7.39809632e-01 -3.56425792e-01
-1.03242028e+00 2.56267935e-01 -1.58307441e-02 2.33538136e-01
-1.17281102e-01 2.23380357e-01 1.18396841e-01 -1.02500021e-01
8.18590343e-01 1.64167419e-01 2.67149135e-02 1.55403554e-01
2.86390096e-01 5.08934498e-01 7.89769769e-01 -5.34205556e-01
5.89099526e-01 9.35514867e-01 2.04182398e-02 -8.83869231e-01
-3.79468245e-03 -2.84987271e-01 -4.94476110e-01 -1.29682392e-01
8.54593158e-01 -1.29304934e+00 -1.51477575e+00 8.06718469e-01
-1.38885903e+00 -1.21365106e+00 -7.32021332e-02 3.62049013e-01
-1.03821874e+00 3.44160050e-01 -8.88669431e-01 -1.38582945e+00
5.39129600e-02 -1.31202042e+00 9.61839437e-01 5.72853923e-01
-7.30195642e-03 -1.08967412e+00 3.04330438e-01 -2.77298838e-01
6.48154676e-01 1.37688503e-01 -1.97390601e-01 -5.55411935e-01
-5.88976026e-01 7.17288479e-02 1.57491818e-01 -3.03553969e-01
-2.58627255e-02 -4.82896686e-01 -6.35966837e-01 -2.87873179e-01
4.28464822e-03 -5.89591742e-01 9.06451941e-01 5.73448658e-01
7.39095509e-02 -3.56831729e-01 -5.49771130e-01 1.54240817e-01
8.28956842e-01 5.36583960e-01 2.73242980e-01 8.29780400e-01
7.49566972e-01 8.59448314e-01 7.15346515e-01 5.86637497e-01
1.33656251e+00 3.38146776e-01 9.86923456e-01 1.82665601e-01
5.62243998e-01 -4.31291342e-01 8.63655984e-01 5.41159689e-01
-1.78620040e-01 -5.39246202e-01 -1.26148260e+00 8.05264592e-01
-2.63873744e+00 -8.24582338e-01 -4.83433120e-02 1.48370671e+00
1.81032985e-01 -5.85928112e-02 3.66132677e-01 -4.01574641e-01
6.40726805e-01 -1.23416282e-01 -8.46623480e-01 -1.06840193e-01
-3.07994753e-01 -6.54542744e-01 8.96349549e-01 1.00873685e+00
-1.00449407e+00 1.45145357e+00 6.66287136e+00 1.11523464e-01
-8.80196154e-01 1.04161993e-01 -1.15147531e-01 2.19559222e-01
2.37122402e-01 -1.43498316e-01 -7.85205007e-01 2.52558500e-01
8.11407387e-01 1.12046048e-01 5.23456812e-01 1.04136288e+00
4.65085357e-01 -5.70204437e-01 -5.78848779e-01 7.81293988e-01
-1.70211375e-01 -9.21651065e-01 -5.95313013e-01 1.54963583e-01
8.46982718e-01 7.04448819e-01 7.42940307e-02 7.49055028e-01
1.68226981e+00 -1.23662663e+00 8.07204902e-01 6.10770643e-01
-1.14126466e-01 -6.35343909e-01 1.07272744e+00 8.62057567e-01
-1.08025682e+00 -5.80458283e-01 -6.57899022e-01 -6.54861450e-01
6.54309034e-01 -6.25863075e-02 -1.22584307e+00 1.88675448e-01
5.80276787e-01 6.28853798e-01 -1.21614665e-01 1.00990784e+00
-5.20947397e-01 1.91392437e-01 -3.51693243e-01 -7.16268420e-01
7.03704715e-01 -2.01301366e-01 5.89494884e-01 1.00068367e+00
3.02809626e-01 5.07046059e-02 4.95768160e-01 2.16857523e-01
2.97263771e-01 -4.60250527e-01 -1.22229052e+00 4.59682792e-01
4.38097507e-01 5.65933943e-01 -6.82855010e-01 -5.63210174e-02
-7.60170221e-02 1.06171060e+00 5.60197055e-01 8.32710028e-01
-6.58904791e-01 -4.23378825e-01 7.57059872e-01 -3.89907956e-01
4.74097311e-01 -7.78735697e-01 1.17234275e-01 -9.69631314e-01
-1.90436557e-01 -5.92209578e-01 -1.98219135e-01 -7.25094736e-01
-1.16389453e+00 8.74078453e-01 -4.32873130e-01 -1.11831975e+00
-9.27572310e-01 -8.06413651e-01 -4.75565732e-01 5.70393980e-01
-1.53387141e+00 -7.95889497e-01 -2.47886926e-01 5.14975369e-01
3.95274460e-01 -5.11680186e-01 5.46715200e-01 -3.49513412e-01
-2.87627906e-01 -5.47596402e-02 1.33639142e-01 4.50071782e-01
4.27812338e-01 -1.35422778e+00 1.09695101e+00 6.88594341e-01
-7.11379468e-01 4.79402006e-01 1.12436891e+00 -9.38753664e-01
-1.36648989e+00 -1.00853944e+00 3.92915219e-01 -6.60359025e-01
8.24840486e-01 -2.99671143e-01 -6.34657025e-01 7.82310426e-01
2.52132058e-01 -2.40952428e-02 3.66252176e-02 -4.63516027e-01
1.71516150e-01 4.22830582e-01 -9.13637877e-01 1.03801227e+00
1.14389157e+00 -2.33794376e-01 -4.29373115e-01 2.22049743e-01
8.48860204e-01 -9.61701095e-01 -2.56548487e-02 1.46595165e-01
4.00378376e-01 -8.58443797e-01 6.97449446e-01 -5.28156340e-01
-1.59570083e-01 -5.65045476e-01 -2.24777699e-01 -1.72349346e+00
6.52797818e-02 -9.54814196e-01 1.14087397e-02 2.66415626e-01
3.81594390e-01 -9.86222267e-01 8.80861878e-01 4.05204892e-01
-1.40207902e-01 -3.44113916e-01 -1.12632489e+00 -8.94757926e-01
2.59186119e-01 -2.05343693e-01 4.95816737e-01 3.59308869e-01
1.48836061e-01 4.15635377e-01 -9.35185730e-01 4.86838192e-01
4.63663131e-01 -6.50186002e-01 1.21417022e+00 -9.01203036e-01
3.71414633e-03 -2.30311319e-01 -3.50926995e-01 -1.53013408e+00
6.75527215e-01 -1.58243820e-01 1.09336269e+00 -2.06663299e+00
-3.35560381e-01 -3.01110238e-01 5.73170543e-01 2.12232888e-01
-7.88925365e-02 -3.75739902e-01 2.26100057e-01 1.13720544e-01
-1.34296751e+00 1.06619728e+00 1.29803753e+00 -2.31585838e-02
-2.93600321e-01 2.48142168e-01 -3.84169579e-01 9.55884635e-01
9.27512109e-01 -4.27725643e-01 -4.72147137e-01 -7.41714299e-01
4.63596940e-01 3.25988978e-01 2.35969022e-01 -1.17059481e+00
1.08035254e+00 -3.22807103e-01 -1.76897347e-01 -4.48288858e-01
4.79108393e-01 -7.08266258e-01 -4.51960862e-01 1.08512962e+00
-7.86071792e-02 4.71774757e-01 -3.22546661e-02 1.07357299e+00
-2.82304473e-02 -2.13343397e-01 3.44029814e-01 -4.49111164e-01
-1.06632566e+00 2.18624607e-01 -1.30266666e+00 3.41452211e-01
8.66564214e-01 1.90839414e-02 -5.18317401e-01 -1.12375867e+00
-6.30850434e-01 1.09577501e+00 6.09487474e-01 2.75951087e-01
7.07199872e-01 -9.17611182e-01 -6.25408113e-01 -1.82558671e-01
-2.06259489e-01 3.96198034e-01 1.68617010e-01 5.32126248e-01
-8.42467129e-01 4.28014815e-01 -2.79754668e-01 -4.51322347e-01
-4.55364466e-01 2.45060936e-01 8.53057742e-01 -2.10267305e-01
-5.45721412e-01 7.33052671e-01 2.17535242e-01 -1.20232379e+00
3.73047620e-01 -3.63681167e-01 -5.53384244e-01 -4.45108443e-01
4.49249119e-01 5.41631579e-01 -5.32355726e-01 -9.05990720e-01
-3.57548237e-01 2.88027704e-01 4.34572101e-01 -6.03355348e-01
1.24718308e+00 -4.96735930e-01 5.95254451e-02 4.90896702e-01
4.72032309e-01 -1.54978335e-01 -1.82865155e+00 -2.78350711e-01
2.10835651e-01 2.18041420e-01 -8.47036600e-01 -1.06984198e-01
-5.02727866e-01 6.92853689e-01 2.46391371e-01 5.88578684e-03
2.81937420e-01 -3.17987025e-01 7.39918470e-01 1.43696165e+00
1.06928337e+00 -1.24155867e+00 3.54853660e-01 1.76030004e+00
7.61021554e-01 -1.69744718e+00 -6.68073535e-01 2.27903381e-01
-1.26364434e+00 8.86182785e-01 1.08560312e+00 -6.10017300e-01
5.54914355e-01 3.94377053e-01 4.73450541e-01 -6.81312010e-02
-8.04909110e-01 -6.45473659e-01 -6.22247338e-01 1.31541276e+00
-3.44136834e-01 3.16433348e-02 4.34187144e-01 6.48991287e-01
-4.64928001e-01 -5.43878078e-02 1.02609587e+00 1.30690348e+00
-1.01024902e+00 -5.08780420e-01 -4.97141600e-01 -2.54990131e-01
-4.73634787e-02 2.97109246e-01 -1.63568795e-01 8.65189254e-01
-8.91066343e-02 1.36983061e+00 -4.74526361e-02 -3.89394075e-01
5.06056428e-01 -6.59387708e-01 6.54083267e-02 -5.36651075e-01
-3.48854393e-01 -4.12757158e-01 1.51344731e-01 -5.29543042e-01
-4.20705110e-01 -5.99545836e-01 -1.96212626e+00 -1.60758629e-01
1.52004823e-01 3.88646722e-01 5.06111741e-01 1.19545650e+00
2.60455608e-01 4.48036402e-01 3.82065475e-01 -1.41945112e+00
-5.70426226e-01 -9.35695887e-01 -4.84932005e-01 -1.46530256e-01
9.12919879e-01 -9.14259076e-01 -2.57203460e-01 -3.64927769e-01]
|
[4.748933792114258, 0.9750598669052124]
|
42e8d627-0fcf-4a7f-a22e-34d8cfe15da2
|
misnn-multiple-imputation-via-semi-parametric
|
2305.01794
| null |
https://arxiv.org/abs/2305.01794v1
|
https://arxiv.org/pdf/2305.01794v1.pdf
|
MISNN: Multiple Imputation via Semi-parametric Neural Networks
|
Multiple imputation (MI) has been widely applied to missing value problems in biomedical, social and econometric research, in order to avoid improper inference in the downstream data analysis. In the presence of high-dimensional data, imputation models that include feature selection, especially $\ell_1$ regularized regression (such as Lasso, adaptive Lasso, and Elastic Net), are common choices to prevent the model from underdetermination. However, conducting MI with feature selection is difficult: existing methods are often computationally inefficient and poor in performance. We propose MISNN, a novel and efficient algorithm that incorporates feature selection for MI. Leveraging the approximation power of neural networks, MISNN is a general and flexible framework, compatible with any feature selection method, any neural network architecture, high/low-dimensional data and general missing patterns. Through empirical experiments, MISNN has demonstrated great advantages over state-of-the-art imputation methods (e.g. Bayesian Lasso and matrix completion), in terms of imputation accuracy, statistical consistency and computation speed.
|
['Qi Long', 'Yiliang Zhang', 'Zongyu Dai', 'Zhiqi Bu']
|
2023-05-02
| null | null | null | null |
['matrix-completion']
|
['methodology']
|
[ 2.58560598e-01 -3.58156085e-01 -5.53040028e-01 -6.03217840e-01
-6.56531632e-01 -1.39250634e-02 -6.20879866e-02 -7.29683861e-02
-2.82593608e-01 1.28582132e+00 3.04015458e-01 -2.09561691e-01
-5.82715511e-01 -7.67555654e-01 -6.48369193e-01 -8.93904984e-01
4.01037969e-02 4.36081350e-01 -7.63014317e-01 9.56085138e-03
9.58617404e-02 3.06591451e-01 -1.17057335e+00 2.66973246e-02
1.20860779e+00 7.39814162e-01 -2.60501474e-01 -1.21878520e-01
-3.40282954e-02 4.26446497e-01 4.31734808e-02 -4.61209416e-01
6.30314276e-02 -2.97432691e-01 -2.99719691e-01 -3.25316548e-01
-4.14323024e-02 -1.39082968e-01 -1.66940942e-01 7.84398139e-01
6.63145900e-01 -5.35693578e-02 6.55832827e-01 -1.21797621e+00
-5.89300156e-01 6.33015990e-01 -7.39601612e-01 -2.72275448e-01
3.04918945e-01 -3.72531898e-02 5.00378728e-01 -1.38032162e+00
5.44763207e-01 1.06720638e+00 1.23719907e+00 4.19829249e-01
-1.71427751e+00 -8.45709145e-01 -1.49975911e-01 -1.13141999e-01
-1.52666795e+00 -6.81751549e-01 5.63386619e-01 -4.92903292e-01
4.80218738e-01 3.91454935e-01 3.45374733e-01 1.42556226e+00
4.91807431e-01 6.17054462e-01 1.14926970e+00 -2.55103499e-01
2.61454701e-01 -2.32373476e-01 1.97914749e-01 2.79670417e-01
4.66373742e-01 1.45480782e-01 -7.92145908e-01 -6.84416294e-01
7.26178527e-01 4.91039634e-01 -2.12823227e-02 -2.75130212e-01
-1.56057000e+00 7.37620234e-01 2.75331121e-02 -7.32242689e-02
-7.59469926e-01 5.09403320e-03 5.21347880e-01 3.05879354e-01
4.78003591e-01 3.50359082e-02 -4.44049686e-01 7.31269121e-02
-1.07680631e+00 3.03882122e-01 5.62273979e-01 7.68126726e-01
5.58695436e-01 2.83795118e-01 -4.11413431e-01 1.04493928e+00
2.74682701e-01 5.86171687e-01 4.43724960e-01 -1.05571389e+00
5.77149510e-01 6.85041308e-01 1.67543128e-01 -1.24464822e+00
-7.67199814e-01 -5.08718073e-01 -2.06695151e+00 -7.44076446e-02
4.06792700e-01 -3.35755080e-01 -6.93364978e-01 1.91750491e+00
4.76060122e-01 3.96537423e-01 9.17957947e-02 7.79438317e-01
7.72399604e-01 1.17264852e-01 3.18813115e-01 -6.38885558e-01
1.09068227e+00 -2.76234895e-01 -9.55398262e-01 -1.50530301e-02
5.66538870e-01 -5.32444119e-01 6.93250835e-01 5.65670490e-01
-1.03538978e+00 -3.41707557e-01 -4.91263092e-01 2.58841477e-02
-6.32270649e-02 2.60977685e-01 9.97558594e-01 5.52621424e-01
-4.57668006e-01 5.93030930e-01 -9.74420965e-01 -5.00355847e-02
6.10650718e-01 7.14604259e-01 -6.72919095e-01 -2.04415500e-01
-1.07323265e+00 5.83031416e-01 1.15516700e-01 7.74986982e-01
-4.63810831e-01 -8.13443184e-01 -7.45033562e-01 3.45487078e-03
5.20123355e-02 -1.10357273e+00 4.89750415e-01 -8.27173889e-01
-1.20431423e+00 4.60105002e-01 -4.93224591e-01 -1.41765013e-01
6.00859582e-01 -1.98527068e-01 -2.38967553e-01 -5.61740577e-01
1.84964940e-01 1.49570063e-01 7.71277368e-01 -6.77748621e-01
-1.05176046e-01 -8.67707670e-01 -8.45535398e-01 3.10627297e-02
-1.31269768e-01 -4.21666577e-02 1.66636586e-01 -9.23995614e-01
7.65524387e-01 -7.55241156e-01 -5.79706192e-01 -7.21197426e-02
-7.14028358e-01 -1.77459791e-02 1.28150880e-01 -7.82977819e-01
1.26574850e+00 -2.22712803e+00 2.72432983e-01 4.77343023e-01
1.35147735e-01 -5.65970168e-02 7.66388550e-02 4.14884567e-01
-4.56088968e-02 -4.63187788e-03 -5.53039312e-01 -4.26407516e-01
-9.42458212e-02 9.07084867e-02 -1.93651021e-02 8.19079578e-01
-1.29015952e-01 8.33243072e-01 -5.64644098e-01 -4.69749779e-01
5.77318221e-02 6.09467685e-01 -4.29482430e-01 1.30640015e-01
2.68332005e-01 1.06545532e+00 -5.68649113e-01 1.06694388e+00
9.55846965e-01 -2.07010642e-01 2.78622746e-01 -1.17441073e-01
-1.62194565e-01 -2.94615984e-01 -1.56196797e+00 1.79473579e+00
-2.55679846e-01 2.51840129e-02 2.99047679e-01 -1.11603522e+00
1.11560798e+00 4.80476320e-01 6.20722890e-01 -3.97429615e-01
-6.89097419e-02 4.52749670e-01 -2.88164765e-01 -7.98376381e-01
1.59510560e-02 -7.74235278e-02 -1.25322834e-01 1.91472366e-01
-2.78176785e-01 6.91445470e-01 -8.71069729e-02 -2.71384478e-01
9.08461690e-01 3.14592063e-01 3.86751503e-01 -2.62547046e-01
4.72699791e-01 -2.04236373e-01 1.40850675e+00 9.63873684e-01
1.85465455e-01 8.23761463e-01 3.74170512e-01 -6.27780080e-01
-8.30449641e-01 -1.05710733e+00 -6.88879073e-01 9.48061943e-01
-1.38971329e-01 6.69046268e-02 -3.39436948e-01 -1.16293743e-01
3.16523582e-01 3.65254879e-01 -5.19583166e-01 -2.57886667e-02
-5.94303489e-01 -1.59625363e+00 6.09340668e-01 4.18450654e-01
3.02547872e-01 -1.00677860e+00 -5.20989597e-02 6.15960896e-01
-3.89197946e-01 -6.67626560e-01 -1.04611702e-01 2.38285348e-01
-1.27206373e+00 -1.01129842e+00 -5.01219273e-01 -4.61089790e-01
9.17551994e-01 -3.62373404e-02 9.44978893e-01 5.34737110e-02
-1.73298717e-01 -2.88121194e-01 -1.71021208e-01 -1.59662098e-01
-6.07867911e-02 1.66395620e-01 4.08071220e-01 3.21671098e-01
5.54202616e-01 -9.11287069e-01 -8.15452278e-01 3.54771942e-01
-9.06367600e-01 6.17768466e-02 7.70846188e-01 1.41296124e+00
8.51815403e-01 -3.75004232e-01 1.10006702e+00 -1.25440979e+00
4.55335736e-01 -8.98386598e-01 -4.52998996e-01 3.23618919e-01
-8.04554820e-01 -1.46746844e-01 6.46043599e-01 -2.55751550e-01
-1.10400486e+00 2.78177977e-01 -3.59754801e-01 -9.26833451e-02
-2.24371105e-01 1.04716051e+00 -2.61956245e-01 -2.72057187e-02
6.51257277e-01 3.93678337e-01 3.43383491e-01 -7.28507698e-01
-1.12513185e-01 7.08277583e-01 3.31601918e-01 -4.73549426e-01
4.03756708e-01 4.28077847e-01 4.45061594e-01 -5.62433422e-01
-6.49535716e-01 -3.28042865e-01 -7.90633559e-01 2.78848231e-01
5.91304362e-01 -1.17535734e+00 -9.44018066e-01 6.24699593e-01
-8.59533250e-01 1.57957405e-01 6.44428432e-02 9.77231920e-01
-5.91229439e-01 2.54391372e-01 -5.48619807e-01 -7.90717304e-01
-5.78925848e-01 -1.15967834e+00 6.34022117e-01 -2.25971844e-02
-2.49487668e-01 -7.16883540e-01 -2.72338744e-02 4.00319338e-01
5.33837318e-01 6.66846097e-01 1.01523423e+00 -4.90031004e-01
-2.93029934e-01 -1.90870747e-01 -1.55913442e-01 2.90806964e-02
1.06468000e-01 -3.56833264e-02 -7.66943753e-01 -2.33251393e-01
-1.33959010e-01 -3.17314833e-01 6.92169607e-01 8.94800723e-01
1.49158800e+00 -4.52933192e-01 -3.85836750e-01 1.07643795e+00
1.35126221e+00 -2.04863146e-01 7.34291494e-01 2.30073348e-01
5.43806016e-01 4.14995879e-01 3.58134598e-01 9.06243682e-01
3.73185128e-01 5.13107061e-01 3.70699495e-01 -2.86696196e-01
4.72475410e-01 -3.93873602e-02 1.24084996e-02 7.61738539e-01
-7.70301744e-02 1.21744201e-01 -6.77491963e-01 3.51541489e-01
-2.31700659e+00 -1.06426811e+00 -5.16126871e-01 2.50711870e+00
9.53004241e-01 -2.90532142e-01 1.09655596e-01 7.87057802e-02
6.94220066e-01 -3.60809028e-01 -6.72338247e-01 -2.84493625e-01
-5.06306827e-01 -9.08952728e-02 3.51345122e-01 -1.41252559e-02
-1.04616177e+00 2.73679763e-01 6.51702309e+00 6.24753475e-01
-7.55113900e-01 1.14165366e-01 9.21251953e-01 -5.69057465e-03
-3.42678577e-01 -1.63205385e-01 -8.70408416e-01 6.75194919e-01
7.72964120e-01 2.78741658e-01 5.47180474e-01 5.86195529e-01
6.81101978e-01 -6.78619416e-03 -1.00520587e+00 9.26346540e-01
-2.24906638e-01 -1.34991896e+00 -3.80035728e-01 2.86624208e-02
6.71409845e-01 3.29544768e-02 -7.52307475e-02 3.02081168e-01
5.72868846e-02 -1.31888795e+00 1.32815078e-01 8.30159068e-01
8.10376167e-01 -7.71523952e-01 1.04908693e+00 5.26908398e-01
-5.45226038e-01 -2.02688545e-01 -5.11765480e-01 -1.58565551e-01
2.78840482e-01 1.32084298e+00 -1.95762351e-01 7.29376912e-01
7.13055611e-01 7.18901575e-01 -1.77784368e-01 1.20337009e+00
1.77060142e-01 6.98886514e-01 -3.99255633e-01 4.28191572e-01
-2.21384823e-01 -6.26050353e-01 3.51408839e-01 9.09055173e-01
6.44391477e-01 1.31062716e-01 1.83372170e-01 8.29791188e-01
-9.18561444e-02 3.09351087e-01 -6.59256220e-01 4.77057248e-01
6.04251742e-01 1.01075304e+00 -2.50372976e-01 1.38613805e-01
-5.23287535e-01 4.50700462e-01 3.27114433e-01 4.46488589e-01
-6.22450292e-01 -2.03741565e-01 5.10006130e-01 -2.00663190e-02
-8.84060785e-02 -1.51284233e-01 -8.65470052e-01 -1.32864916e+00
2.38513067e-01 -9.89615083e-01 5.33678949e-01 -3.67559254e-01
-1.74106455e+00 2.82650292e-01 -1.84704661e-01 -1.27486348e+00
-1.17000960e-01 -2.04218313e-01 -4.11350399e-01 9.63824868e-01
-1.32818615e+00 -1.25156605e+00 -1.78257927e-01 6.08509123e-01
6.54342994e-02 -2.78289169e-01 9.94874716e-01 5.25614619e-01
-1.04460323e+00 6.92488968e-01 9.12280917e-01 -2.27244236e-02
7.52123296e-01 -7.27744460e-01 -2.69255936e-01 4.93692636e-01
-3.87989610e-01 9.72601831e-01 4.56336856e-01 -8.15455616e-01
-1.81875384e+00 -1.07135141e+00 1.13750446e+00 -1.76341370e-01
2.45609626e-01 -3.85377914e-01 -1.01304996e+00 6.62075520e-01
-3.43825251e-01 2.11808920e-01 1.19664443e+00 6.63804412e-01
-8.57767090e-02 -4.19481426e-01 -1.41549170e+00 4.16404217e-01
8.19376528e-01 3.68620865e-02 -2.32381210e-01 4.88029420e-01
1.95360333e-01 -3.50922644e-01 -1.31802046e+00 7.40461886e-01
9.63889658e-01 -8.71693015e-01 1.29583716e+00 -7.47639656e-01
2.93485314e-01 -2.92604744e-01 -2.17877194e-01 -8.46986294e-01
-6.51205122e-01 -4.99600112e-01 -1.39433846e-01 1.43906498e+00
4.88720506e-01 -6.09231770e-01 9.13920045e-01 9.01028037e-01
4.20850366e-02 -6.59764171e-01 -1.48047507e+00 -6.62732244e-01
2.95758583e-02 -2.95451224e-01 8.20647418e-01 9.88772094e-01
-9.76102725e-02 -9.39029306e-02 -1.06130171e+00 6.38298243e-02
9.48055029e-01 2.60812193e-01 7.42857456e-01 -1.50266635e+00
2.59818672e-03 9.65006202e-02 -8.95807371e-02 -3.73398155e-01
3.36750180e-01 -8.85710359e-01 -1.12006158e-01 -1.23505676e+00
3.36666822e-01 -9.22414362e-01 -5.21745443e-01 6.99478447e-01
-2.78008312e-01 3.79850805e-01 -3.15509498e-01 5.20089030e-01
-2.99836010e-01 6.48304760e-01 9.94976819e-01 -1.19432554e-01
-5.86110771e-01 3.83786678e-01 -6.80797875e-01 6.23350620e-01
7.59371042e-01 -8.72989595e-01 1.81812569e-02 -5.23867249e-01
4.60419744e-01 4.95599687e-01 4.09633845e-01 -6.13324463e-01
1.92877218e-01 -5.28556705e-01 8.21244359e-01 -5.22565424e-01
2.19127893e-01 -9.35470462e-01 8.78483474e-01 3.74489129e-01
-3.09217334e-01 1.36130840e-01 -2.46401042e-01 6.23350203e-01
-1.85296074e-01 -1.60196140e-01 5.35241306e-01 -2.46644653e-02
-2.18349785e-01 4.25695211e-01 -3.78501892e-01 -1.95193902e-01
7.06307471e-01 -2.52567858e-01 -1.60737839e-02 2.90189367e-02
-1.12638366e+00 3.40124875e-01 1.18518054e-01 -7.28490204e-02
7.83691406e-01 -1.46002114e+00 -1.17109942e+00 4.71177936e-01
2.26046145e-02 8.07746127e-02 5.22416949e-01 1.48335171e+00
-1.10056221e-01 1.01385131e-01 -1.70100704e-01 -5.98055661e-01
-8.51240933e-01 3.79471332e-01 -1.15368828e-01 -1.05980858e-01
-6.55783117e-01 2.34997660e-01 -1.94124192e-01 -8.07938695e-01
2.46390000e-01 1.72394499e-01 -3.60904634e-02 -2.65903659e-02
3.61960113e-01 7.12575376e-01 1.35749578e-01 -3.08977455e-01
-4.38375771e-01 4.45535451e-01 2.29685679e-01 5.10155857e-01
1.65477252e+00 -3.58119428e-01 -6.21352434e-01 3.57338309e-01
8.36060166e-01 -2.60122299e-01 -8.76733899e-01 -1.01569340e-01
-9.47409123e-03 -4.67809469e-01 -3.62027474e-02 -7.26183414e-01
-9.01500940e-01 5.07578492e-01 5.10007858e-01 -2.22417846e-01
1.10914183e+00 -6.52090728e-01 4.68476892e-01 4.57324833e-01
4.84094113e-01 -9.39051628e-01 -7.41538167e-01 2.63930798e-01
7.76951075e-01 -1.71534324e+00 2.76463449e-01 -3.17769676e-01
-3.38239640e-01 1.05165839e+00 5.05285978e-01 -5.61071523e-02
7.55093575e-01 1.21857814e-01 -7.70036355e-02 8.70825425e-02
-5.55715322e-01 4.55488682e-01 1.45365391e-02 5.84532142e-01
7.33094335e-01 -2.92937532e-02 -7.46426702e-01 8.59367967e-01
2.37618685e-01 6.38304889e-01 3.15430611e-01 7.01340556e-01
1.81526735e-01 -1.23609912e+00 -6.10178113e-01 1.00525951e+00
-8.11416507e-01 -1.53660893e-01 2.95941740e-01 5.42980075e-01
1.56990811e-01 8.61364901e-01 -2.56271243e-01 8.18253979e-02
1.25445858e-01 7.96240121e-02 8.99919868e-03 -1.89289376e-01
-7.55693853e-01 8.05845708e-02 1.33789238e-02 -5.54138839e-01
-5.08546054e-01 -9.10449862e-01 -9.19826746e-01 -4.21860576e-01
-3.19325298e-01 1.48506865e-01 6.49329722e-01 9.57424641e-01
7.39556193e-01 9.10191834e-02 6.78039253e-01 -5.04392624e-01
-7.35411227e-01 -9.07275498e-01 -7.66714752e-01 4.08755094e-01
2.70688087e-01 -6.11078203e-01 -1.81652248e-01 -1.33394701e-02]
|
[7.614885330200195, 4.852677822113037]
|
0b1a06d3-5820-4488-beb1-319fdd2f6988
|
dont-let-notes-be-misunderstood-a-negation
| null | null |
https://aclanthology.org/W16-0310
|
https://aclanthology.org/W16-0310.pdf
|
Don't Let Notes Be Misunderstood: A Negation Detection Method for Assessing Risk of Suicide in Mental Health Records
| null |
['Harry Dean', 'Rina Dutta', 'Anika Oellrich', 'Maria Liakata', 'George Gkotsis', 'Sumithra Velupillai']
|
2016-06-01
| null | null | null |
ws-2016-6
|
['negation-detection']
|
['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]
|
[-7.282026767730713, 3.754664897918701]
|
42d43e20-9e13-49ca-b742-c0027458c5e6
|
memc-net-motion-estimation-and-motion
|
1810.08768
| null |
https://arxiv.org/abs/1810.08768v2
|
https://arxiv.org/pdf/1810.08768v2.pdf
|
MEMC-Net: Motion Estimation and Motion Compensation Driven Neural Network for Video Interpolation and Enhancement
|
Motion estimation (ME) and motion compensation (MC) have been widely used for classical video frame interpolation systems over the past decades. Recently, a number of data-driven frame interpolation methods based on convolutional neural networks have been proposed. However, existing learning based methods typically estimate either flow or compensation kernels, thereby limiting performance on both computational efficiency and interpolation accuracy. In this work, we propose a motion estimation and compensation driven neural network for video frame interpolation. A novel adaptive warping layer is developed to integrate both optical flow and interpolation kernels to synthesize target frame pixels. This layer is fully differentiable such that both the flow and kernel estimation networks can be optimized jointly. The proposed model benefits from the advantages of motion estimation and compensation methods without using hand-crafted features. Compared to existing methods, our approach is computationally efficient and able to generate more visually appealing results. Furthermore, the proposed MEMC-Net can be seamlessly adapted to several video enhancement tasks, e.g., super-resolution, denoising, and deblocking. Extensive quantitative and qualitative evaluations demonstrate that the proposed method performs favorably against the state-of-the-art video frame interpolation and enhancement algorithms on a wide range of datasets.
|
['Ming-Hsuan Yang', 'Wei-Sheng Lai', 'Zhiyong Gao', 'Wenbo Bao', 'Xiaoyun Zhang']
|
2018-10-20
| null | null | null | null |
['video-enhancement']
|
['computer-vision']
|
[ 2.61399269e-01 -5.78139007e-01 -1.11225285e-01 -3.43839288e-01
-6.18725538e-01 -3.02171931e-02 6.61637723e-01 -1.91668585e-01
-6.56682670e-01 8.94751489e-01 1.60908684e-01 -2.88558640e-02
1.91376582e-01 -6.41902566e-01 -7.91540384e-01 -7.24129617e-01
1.46294922e-01 -2.89704055e-01 3.59025478e-01 -1.38737604e-01
3.05516481e-01 4.16546136e-01 -1.54388690e+00 7.22222477e-02
1.18784475e+00 1.11188328e+00 3.60023946e-01 6.28830552e-01
4.53189723e-02 1.03166974e+00 -2.25391194e-01 -3.14562231e-01
3.17283154e-01 -4.62639213e-01 -4.94954079e-01 -1.08830463e-02
7.03261077e-01 -7.57804811e-01 -4.41453844e-01 9.80261266e-01
5.80451190e-01 3.81675512e-01 3.64722759e-01 -9.67423975e-01
-7.50267208e-01 4.00410265e-01 -7.26548254e-01 1.95604950e-01
2.08638489e-01 3.42417657e-01 5.77793181e-01 -1.27821720e+00
6.00853145e-01 1.20827925e+00 7.67676950e-01 4.93473679e-01
-1.36235034e+00 -6.72490895e-01 -1.41545698e-01 5.09484410e-01
-1.25000596e+00 -6.54140890e-01 1.03668451e+00 -5.53385913e-01
4.86352295e-01 2.36513768e-03 5.83077252e-01 8.60409617e-01
8.30897838e-02 7.21877933e-01 9.66671109e-01 -3.14200789e-01
2.01181769e-01 -3.06250770e-02 -4.22379732e-01 5.72337866e-01
4.63861860e-02 3.13237220e-01 -3.92280817e-01 1.87096223e-01
1.32502103e+00 -8.03399086e-02 -5.61219931e-01 -3.86171132e-01
-1.25135827e+00 5.48967302e-01 5.19747019e-01 1.45182401e-01
-5.68001926e-01 2.53806502e-01 5.06939232e-01 -1.26716629e-01
6.24352515e-01 1.06320165e-01 2.80591799e-03 -2.24237759e-02
-1.57394922e+00 4.29746270e-01 3.60414594e-01 7.77917325e-01
9.21933889e-01 5.00864267e-01 -4.47448522e-01 8.77104640e-01
2.31801644e-01 1.49122581e-01 3.48411560e-01 -1.30112660e+00
4.34544325e-01 1.82676330e-01 5.65954387e-01 -1.18310761e+00
-2.28373945e-01 -2.66658783e-01 -1.23522079e+00 5.23368537e-01
5.88490486e-01 -1.15214974e-01 -7.42727458e-01 1.60764670e+00
3.91183943e-01 7.02031374e-01 -2.14929774e-01 1.32156026e+00
6.67205691e-01 5.81341743e-01 2.59019881e-01 -1.18106417e-01
1.11856902e+00 -1.19215155e+00 -9.69194889e-01 -5.89619689e-02
2.69431621e-01 -9.12902772e-01 7.92304397e-01 2.29841858e-01
-1.66151929e+00 -9.96810853e-01 -1.13932836e+00 -4.53587651e-01
1.53501391e-01 2.97880650e-01 4.25997496e-01 4.88917500e-01
-1.14128613e+00 8.23064983e-01 -1.02932787e+00 4.47120070e-02
6.66310310e-01 2.84930617e-01 -1.77921757e-01 -1.05074361e-01
-1.23271930e+00 8.60102177e-01 5.58015645e-01 4.37076569e-01
-8.59838128e-01 -8.07526171e-01 -9.77427363e-01 1.55213609e-01
1.04148008e-01 -1.01889086e+00 9.63380516e-01 -1.25479949e+00
-1.72758114e+00 3.23131353e-01 -3.01209658e-01 -6.20705605e-01
8.63171697e-01 -3.77349973e-01 -3.43997449e-01 2.51370966e-01
-2.19870999e-01 9.09312010e-01 1.19938123e+00 -1.13188899e+00
-8.19840312e-01 1.18665077e-01 -6.22421131e-03 9.82185900e-02
-5.14111042e-01 8.51157010e-02 -5.03746271e-01 -1.14109838e+00
-3.16496611e-01 -5.40906250e-01 -1.95952997e-01 5.18202782e-01
-6.00760505e-02 2.23072350e-01 9.05110240e-01 -1.05949879e+00
1.38072300e+00 -1.96494055e+00 2.01607108e-01 -2.78462261e-01
1.31025121e-01 6.84640527e-01 -1.87229469e-01 1.64575558e-02
-6.01182282e-02 -1.23870581e-01 -4.68027860e-01 -4.80615199e-01
-2.57007122e-01 -8.32741708e-02 -2.79660169e-02 5.73548913e-01
4.92462814e-01 8.23436201e-01 -9.62951779e-01 -5.37934542e-01
7.25272655e-01 1.26531184e+00 -8.19900990e-01 2.94747263e-01
-1.17360026e-01 9.69847858e-01 -4.87502180e-02 3.24428141e-01
1.05054319e+00 -1.46367013e-01 -7.80287236e-02 -5.26910603e-01
-2.80623883e-01 6.59475997e-02 -1.25042689e+00 1.76380646e+00
-6.78681433e-01 9.13904667e-01 1.93307847e-01 -8.42508495e-01
7.40144312e-01 4.38534677e-01 5.80309987e-01 -3.99876267e-01
2.82551855e-01 2.92848825e-01 -1.60251394e-01 -4.73004580e-01
7.96040177e-01 6.92029074e-02 6.74089491e-01 8.90747756e-02
-1.20267272e-02 3.47406924e-01 3.49460900e-01 -2.24831685e-01
6.21339083e-01 6.12191141e-01 1.55802861e-01 -1.34463012e-01
1.06740737e+00 -2.99188256e-01 7.37823367e-01 3.30041856e-01
-2.13112876e-01 8.31343830e-01 -3.25229019e-02 -6.14933848e-01
-1.31846571e+00 -8.23938191e-01 -9.50338617e-02 9.62995052e-01
2.89135993e-01 -1.29741356e-01 -1.04269505e+00 -1.32810503e-01
-2.88601041e-01 4.25643116e-01 -4.03624117e-01 2.32884418e-02
-9.49157417e-01 -3.36841226e-01 3.01645011e-01 6.76456511e-01
1.07714832e+00 -9.58339512e-01 -4.83472943e-01 5.44121385e-01
-4.72975016e-01 -1.36856008e+00 -8.29227507e-01 -5.40442109e-01
-9.62578833e-01 -6.55631661e-01 -1.22955751e+00 -9.02151704e-01
6.43970370e-01 3.09590161e-01 8.77688110e-01 8.70823935e-02
-1.94479246e-02 -1.86735876e-02 -9.28618461e-02 1.29446730e-01
-4.18771505e-01 7.58664161e-02 -2.13849723e-01 5.65143704e-01
1.97218265e-02 -6.21830940e-01 -1.21540034e+00 4.17477161e-01
-1.15641308e+00 3.45339805e-01 4.18777585e-01 1.05515635e+00
4.75802511e-01 -3.03990543e-01 3.80281180e-01 -6.37586832e-01
4.73344088e-01 -2.95912892e-01 -7.76178598e-01 8.42496231e-02
-5.47864974e-01 -2.29527429e-03 8.82478416e-01 -5.15620828e-01
-1.41210532e+00 1.82543993e-01 -2.07830831e-01 -7.61981905e-01
-1.19626895e-01 2.77904004e-01 -7.66855776e-02 -3.97499859e-01
5.55313349e-01 2.46467561e-01 2.82917893e-03 -3.55618149e-01
3.38757306e-01 4.68613476e-01 8.36495280e-01 -3.04897696e-01
8.93264532e-01 6.47877157e-01 -1.04610724e-02 -8.20926428e-01
-4.28843796e-01 -2.26367801e-01 -6.23522580e-01 -3.89199197e-01
9.34899747e-01 -1.11750627e+00 -5.96297741e-01 7.52973974e-01
-1.34138584e+00 -2.41965815e-01 -9.15563330e-02 5.81760406e-01
-5.76279998e-01 6.15793467e-01 -7.83309519e-01 -7.41165280e-01
-4.16286170e-01 -1.42181945e+00 7.98155606e-01 4.05222505e-01
1.30628943e-02 -1.26204944e+00 -1.57868966e-01 2.40909815e-01
9.03035522e-01 4.10354763e-01 3.86788696e-01 1.88031927e-01
-8.59813333e-01 9.29522961e-02 -4.66724336e-01 5.14037609e-01
2.75550544e-01 6.61996901e-02 -9.65832293e-01 -4.93860930e-01
-1.00920349e-01 -1.14996945e-02 1.02223229e+00 6.99276984e-01
1.25837684e+00 -3.48685086e-01 4.72074635e-02 1.05575418e+00
1.45728517e+00 -1.44881131e-02 8.97951603e-01 4.53195184e-01
9.74621952e-01 4.21449512e-01 4.91444021e-01 6.21560574e-01
3.98277998e-01 8.80762339e-01 4.38314915e-01 -3.38850677e-01
-3.26733559e-01 -1.15459181e-01 3.52750510e-01 5.08922040e-01
-5.05144596e-01 -7.82598369e-03 -6.34815276e-01 5.78705311e-01
-1.89983702e+00 -1.08379579e+00 1.33379409e-02 2.12151408e+00
9.49428380e-01 -8.40870142e-02 8.46166462e-02 8.77005011e-02
9.62595999e-01 2.87115306e-01 -5.82193315e-01 -1.99377969e-01
-1.13135859e-01 2.09579781e-01 4.50164318e-01 6.36677623e-01
-1.31419301e+00 7.82233596e-01 5.29558706e+00 9.28943098e-01
-1.29030359e+00 9.74698216e-02 7.56647825e-01 1.87520981e-02
-3.03571731e-01 -9.10391882e-02 -6.48364961e-01 6.51461303e-01
7.82811046e-01 4.75031957e-02 5.54053009e-01 5.57467878e-01
8.07671309e-01 7.89601207e-02 -9.50918734e-01 1.10791171e+00
-1.27559423e-01 -1.77642047e+00 -2.42467821e-02 -2.86633909e-01
9.71694171e-01 -3.40072453e-01 1.37432322e-01 5.93984919e-03
-1.03582740e-01 -9.52133775e-01 8.47343862e-01 6.02542043e-01
9.14569974e-01 -6.94064558e-01 6.39025211e-01 1.57259330e-01
-1.37794173e+00 -7.21594645e-03 -2.34920412e-01 -4.06350289e-03
4.43567425e-01 5.80693364e-01 -2.41924092e-01 6.27223551e-01
6.70551002e-01 1.01532829e+00 -3.80847424e-01 1.29774654e+00
-1.15767427e-01 5.00062048e-01 1.93862524e-02 4.75549608e-01
1.45341933e-01 -3.68434906e-01 4.13787127e-01 1.24823225e+00
3.90641391e-01 -2.03360736e-01 -7.53949359e-02 1.09634101e+00
-1.21093653e-01 1.18670151e-01 -1.00101985e-01 2.82921344e-01
3.11882198e-01 1.36994791e+00 -4.82727408e-01 -3.77941579e-01
-5.89201748e-01 9.37608659e-01 1.99332759e-01 5.70316613e-01
-1.00435686e+00 -4.28152800e-01 7.46410668e-01 2.40005553e-01
5.53735077e-01 -2.77558923e-01 -2.71886200e-01 -1.31176603e+00
-8.28883890e-03 -6.15945756e-01 1.00302501e-02 -7.03204155e-01
-1.07354891e+00 6.54911995e-01 -1.92692965e-01 -1.60767353e+00
-3.68214905e-01 -4.47077841e-01 -6.16087496e-01 1.10619032e+00
-2.08231091e+00 -1.04001236e+00 -6.16708159e-01 6.39577270e-01
5.79638481e-01 -5.09588839e-03 2.79746234e-01 7.91122556e-01
-6.22051835e-01 5.30507505e-01 1.96605861e-01 2.41239473e-01
9.28869724e-01 -8.74176800e-01 4.28483695e-01 1.19527316e+00
-3.93415302e-01 4.77885455e-01 6.43591404e-01 -3.84022683e-01
-1.13264167e+00 -1.43536997e+00 5.68690896e-01 2.22932070e-01
3.60137522e-01 -6.32444993e-02 -1.20948696e+00 3.56473386e-01
4.28935766e-01 4.90528673e-01 1.26810580e-01 -7.31630445e-01
-4.99650314e-02 -3.48302126e-01 -1.13125300e+00 6.97466969e-01
6.58366144e-01 -2.57491410e-01 8.98199230e-02 -1.09701082e-01
3.57480824e-01 -6.14692569e-01 -9.87307012e-01 3.85375798e-01
6.10177636e-01 -1.16285336e+00 1.22567034e+00 -1.74479693e-01
8.98480296e-01 -5.58519185e-01 2.64210403e-01 -1.13962960e+00
-4.89041507e-01 -8.13256741e-01 -4.22400236e-01 1.32791924e+00
-6.93907496e-03 -4.63536799e-01 6.70853496e-01 7.35105753e-01
5.44244517e-03 -6.73925042e-01 -7.68546045e-01 -5.45271277e-01
-5.86055443e-02 -2.56004035e-01 4.44527507e-01 8.92799497e-01
-5.39656103e-01 -6.05198406e-02 -8.22359264e-01 7.88030326e-02
8.04440856e-01 -7.15256706e-02 6.86865032e-01 -8.32280993e-01
-2.20139369e-01 -5.70257843e-01 -2.73795217e-01 -1.33339083e+00
9.02798623e-02 -6.75604880e-01 -2.68325526e-02 -1.44002736e+00
-1.65558577e-01 -2.88942963e-01 -4.06475723e-01 1.57000795e-01
-4.88489628e-01 4.13628280e-01 2.20442176e-01 2.63367087e-01
-1.68775588e-01 6.30274475e-01 1.57206905e+00 -2.42528543e-02
-2.65039921e-01 -1.52040675e-01 -3.70011091e-01 8.03483903e-01
8.10571313e-01 5.11979684e-02 -2.65130907e-01 -6.35642946e-01
-9.53816995e-02 2.52068549e-01 5.97521186e-01 -1.10587919e+00
3.46525848e-01 -1.06088139e-01 6.31356597e-01 -3.47236574e-01
3.04036617e-01 -8.25130999e-01 2.21898109e-01 3.42820257e-01
-4.18995976e-01 -5.58011234e-02 1.53415650e-01 4.92492795e-01
-4.71374959e-01 -2.96049230e-02 1.20108151e+00 7.95858875e-02
-8.95207942e-01 4.94983792e-01 -1.82965308e-01 -2.15422325e-02
8.81590307e-01 -4.91390705e-01 -7.29037449e-02 -5.09946048e-01
-3.21233273e-01 -9.50812548e-03 5.06274402e-01 3.37331682e-01
8.87722850e-01 -1.50722599e+00 -8.92988503e-01 2.22457811e-01
-3.09730649e-01 -3.20983957e-03 4.83948350e-01 1.17141104e+00
-8.53387177e-01 1.10891849e-01 -5.49739540e-01 -6.01616681e-01
-8.63580406e-01 5.65423429e-01 3.59193206e-01 -1.44563928e-01
-5.90497375e-01 5.35763502e-01 2.21398696e-01 1.03064075e-01
2.07042485e-01 -3.97670209e-01 -3.22683811e-01 -1.60741135e-01
7.86407709e-01 7.15054631e-01 -2.08504826e-01 -7.95247018e-01
-1.16465099e-01 7.14163840e-01 3.74277644e-02 1.24437563e-01
1.25671387e+00 -3.51751834e-01 1.25400387e-02 6.15519984e-03
1.12117624e+00 -2.83975780e-01 -1.94350147e+00 -2.42151409e-01
-1.30319387e-01 -7.83600986e-01 3.21776658e-01 -3.28418136e-01
-1.54753423e+00 9.54720974e-01 6.70204759e-01 -1.01085164e-01
1.47573435e+00 -7.85197675e-01 1.14566660e+00 -3.82955432e-01
1.36439189e-01 -8.13799441e-01 -2.34612703e-01 2.05216199e-01
6.50303066e-01 -1.39957106e+00 -2.72722635e-02 -5.11455774e-01
-3.55869412e-01 1.35658038e+00 5.05682290e-01 -3.53384227e-01
3.88386577e-01 2.62609124e-01 -7.25112855e-02 6.05747044e-01
-4.98435050e-01 5.25699034e-02 5.15739083e-01 5.11650264e-01
7.14141250e-01 -4.14016098e-01 -4.87620234e-01 5.21053746e-02
2.51851887e-01 5.49264312e-01 5.24916887e-01 6.07870340e-01
-2.72378504e-01 -1.03943157e+00 -4.75791752e-01 7.91324675e-02
-5.09337962e-01 -2.08682612e-01 4.47934061e-01 6.27813518e-01
1.05296962e-01 7.69552648e-01 2.90511232e-02 -1.46920815e-01
1.09359868e-01 -2.97698170e-01 4.47997481e-01 6.56813830e-02
-6.04269505e-01 2.23947302e-01 -1.93205237e-01 -5.84383726e-01
-9.10596192e-01 -4.34562534e-01 -9.95200336e-01 -4.68099445e-01
-2.71229148e-01 -2.73443967e-01 4.55282748e-01 7.26307988e-01
4.19487864e-01 6.67425454e-01 5.26678681e-01 -1.44597507e+00
-1.63952589e-01 -7.29005039e-01 -3.08352888e-01 5.55617809e-01
5.90313196e-01 -6.84109271e-01 -1.97448358e-01 5.50934076e-01]
|
[10.821403503417969, -1.503053903579712]
|
bf92804d-9e61-4eaf-b70e-1c6bb7a02e3c
|
epik-eliminating-multi-model-pipelines-with
|
2211.1492
| null |
https://arxiv.org/abs/2211.14920v1
|
https://arxiv.org/pdf/2211.14920v1.pdf
|
EPIK: Eliminating multi-model Pipelines with Knowledge-distillation
|
Real-world tasks are largely composed of multiple models, each performing a sub-task in a larger chain of tasks, i.e., using the output from a model as input for another model in a multi-model pipeline. A model like MATRa performs the task of Crosslingual Transliteration in two stages, using English as an intermediate transliteration target when transliterating between two indic languages. We propose a novel distillation technique, EPIK, that condenses two-stage pipelines for hierarchical tasks into a single end-to-end model without compromising performance. This method can create end-to-end models for tasks without needing a dedicated end-to-end dataset, solving the data scarcity problem. The EPIK model has been distilled from the MATra model using this technique of knowledge distillation. The MATra model can perform crosslingual transliteration between 5 languages - English, Hindi, Tamil, Kannada and Bengali. The EPIK model executes the task of transliteration without any intermediate English output while retaining the performance and accuracy of the MATra model. The EPIK model can perform transliteration with an average CER score of 0.015 and average phonetic accuracy of 92.1%. In addition, the average time for execution has reduced by 54.3% as compared to the teacher model and has a similarity score of 97.5% with the teacher encoder. In a few cases, the EPIK model (student model) can outperform the MATra model (teacher model) even though it has been distilled from the MATra model.
|
['Anshuman Dash', 'Yash Raj', 'Bhavesh Laddagiri']
|
2022-11-27
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[ 9.57157165e-02 3.10418576e-01 9.88193080e-02 -4.13683832e-01
-1.36687684e+00 -7.42472291e-01 5.56416154e-01 -3.27696294e-01
-7.47669041e-01 5.74047327e-01 8.24106485e-02 -7.54035115e-01
3.57295245e-01 -2.82403111e-01 -8.26505244e-01 -3.50961208e-01
6.90423906e-01 1.00223291e+00 1.60960928e-01 -6.15271777e-02
-2.06225857e-01 -8.66887793e-02 -1.06391060e+00 6.77065730e-01
1.33480847e+00 5.63551784e-01 7.17311919e-01 8.21894228e-01
-2.87927557e-02 7.97833502e-01 -7.14117944e-01 -8.12942922e-01
2.27038696e-01 -4.47057664e-01 -1.15180016e+00 -4.24839735e-01
6.31857336e-01 -2.61572570e-01 -7.47520551e-02 6.11687541e-01
6.93756044e-01 5.37672937e-02 4.73318696e-01 -1.01652229e+00
-7.57784128e-01 1.17317498e+00 -1.89661741e-01 -2.26392210e-01
1.58418015e-01 4.47858423e-02 9.99981582e-01 -1.10532010e+00
3.38186622e-01 1.27529526e+00 7.25964606e-01 6.10205948e-01
-1.38341630e+00 -7.50497878e-01 -8.94004852e-02 1.17123447e-01
-1.22655904e+00 -6.61796272e-01 1.45267084e-01 -4.21520501e-01
1.50040567e+00 2.99820065e-01 2.32531086e-01 8.51801753e-01
1.82538688e-01 1.08659911e+00 1.31428182e+00 -4.11641568e-01
-2.39886150e-01 2.92757243e-01 -8.22197571e-02 5.40940404e-01
-1.86559498e-01 -7.44950539e-03 -7.84948826e-01 2.66483724e-01
3.37698847e-01 -4.20613229e-01 -1.17832504e-01 3.65605712e-01
-1.39238989e+00 4.14872497e-01 2.02118427e-01 2.17944056e-01
-2.99519300e-01 1.46200687e-01 6.29905939e-01 5.82403779e-01
3.16658169e-01 6.20915830e-01 -9.57273662e-01 -4.64984864e-01
-1.35537469e+00 1.51237160e-01 7.55618453e-01 1.10856998e+00
9.80465353e-01 2.25145236e-01 -3.20785940e-01 9.66293752e-01
2.27467522e-01 6.54987156e-01 8.47706378e-01 -6.53223395e-01
9.44147050e-01 4.83276099e-01 -1.68309957e-01 4.20927741e-02
-5.30423783e-02 -4.02435392e-01 -4.19192076e-01 -7.09352046e-02
5.92113376e-01 -2.62904555e-01 -1.27567887e+00 1.43666601e+00
1.60642013e-01 -1.49515226e-01 3.06425780e-01 5.12124300e-01
8.60241771e-01 9.43499863e-01 8.64517391e-02 1.54261678e-01
1.40362036e+00 -1.68209660e+00 -6.94360018e-01 -4.77318436e-01
1.06045449e+00 -1.21477079e+00 1.53278422e+00 5.10976434e-01
-1.40028787e+00 -1.06032217e+00 -7.08993137e-01 -6.49394035e-01
-3.77183080e-01 5.28127015e-01 1.75612405e-01 4.95335609e-01
-1.34598327e+00 3.83477837e-01 -9.75096822e-01 -1.80090383e-01
-1.20936774e-01 3.19010973e-01 -4.40006971e-01 -1.05852902e-01
-1.22395563e+00 1.40744531e+00 8.34523916e-01 3.78207676e-02
-1.20061946e+00 -1.01763368e+00 -9.46165502e-01 -4.59560333e-03
1.83380663e-01 -5.68560302e-01 1.80113745e+00 -9.70824301e-01
-2.01501250e+00 8.49480510e-01 -4.22396421e-01 -3.33022982e-01
7.64722228e-01 -5.06938994e-01 -3.18788409e-01 -2.78394759e-01
1.95330203e-01 7.30825365e-01 5.57033241e-01 -1.10078633e+00
-1.02842665e+00 -1.80942949e-03 -1.39026985e-01 5.10611057e-01
3.05137411e-02 2.25779504e-01 -5.69927454e-01 -5.46899199e-01
-2.14717850e-01 -1.03775477e+00 -1.89727091e-03 -6.42935514e-01
-3.88067603e-01 -3.37701768e-01 7.22693801e-01 -1.29439795e+00
1.33749843e+00 -2.00556278e+00 2.71471471e-01 -1.77963093e-01
-9.95164514e-02 5.72082698e-01 -2.34446079e-01 4.13361013e-01
-7.92763382e-02 1.34774640e-01 -3.10786486e-01 -7.93688178e-01
7.04257339e-02 2.51256794e-01 -1.24253221e-01 -4.26571779e-02
2.61610061e-01 1.29089868e+00 -9.57030118e-01 -1.89176574e-01
1.69898853e-01 4.04230505e-01 -4.70254898e-01 4.91140097e-01
-9.99670774e-02 4.04202104e-01 1.75380677e-01 3.92203242e-01
4.38290238e-01 1.51917696e-01 7.29789510e-02 -8.31337050e-02
-3.18201542e-01 9.29919124e-01 -1.06916893e+00 1.98258579e+00
-1.14867246e+00 6.23553991e-01 -2.91936342e-02 -6.04739845e-01
1.04379964e+00 6.09525084e-01 1.04195513e-02 -8.41123283e-01
-1.88048705e-01 8.06023479e-01 3.80418301e-01 -3.41828048e-01
7.69365788e-01 -1.23546040e-02 -3.84044349e-01 5.75787067e-01
3.89279693e-01 -5.58402240e-01 2.52565056e-01 -1.03741027e-01
8.84312749e-01 5.35239518e-01 1.19245015e-01 -2.92049408e-01
3.94382954e-01 2.26008758e-01 4.56839561e-01 6.82078004e-01
8.39855075e-02 5.02964556e-01 -6.90170517e-03 -3.69870782e-01
-1.21104264e+00 -1.31677675e+00 1.51810765e-01 1.45006776e+00
-6.05211377e-01 -6.90868080e-01 -1.02366495e+00 -7.38107204e-01
-1.15357906e-01 9.36150014e-01 -2.21168220e-01 -7.97797516e-02
-9.09607232e-01 -4.10252690e-01 1.00571096e+00 3.55967194e-01
8.11263621e-01 -1.08146405e+00 -4.56403941e-01 3.19869220e-01
-6.13266349e-01 -1.30006063e+00 -6.63568318e-01 5.43110967e-01
-5.02593100e-01 -5.68670571e-01 -7.57449985e-01 -8.78825843e-01
3.61584753e-01 -1.69702023e-01 1.33658934e+00 -3.77146453e-01
3.09566200e-01 -4.22385857e-02 -4.48851347e-01 -3.75952303e-01
-9.27347720e-01 6.21147394e-01 3.80554572e-02 -3.03498983e-01
3.47162992e-01 -1.40625656e-01 -5.08579127e-02 3.23948652e-01
-6.87726855e-01 5.93838215e-01 8.05244565e-01 7.94857502e-01
6.64405584e-01 -1.81531087e-01 6.37062669e-01 -9.16877151e-01
4.28078830e-01 -3.16637516e-01 -3.83267671e-01 3.41296136e-01
-5.10283053e-01 9.22686905e-02 1.11652851e+00 -5.64239621e-01
-1.18131816e+00 3.00049990e-01 -2.56483793e-01 -2.31636226e-01
-4.75592092e-02 5.69070339e-01 -3.59107167e-01 4.39304411e-01
5.50214708e-01 4.36457783e-01 -2.49602720e-01 -7.34369159e-01
6.33821249e-01 1.11436176e+00 7.49010980e-01 -7.12968767e-01
6.23265088e-01 -1.25808597e-01 -6.25345469e-01 -5.27689934e-01
-8.46701384e-01 -2.29518309e-01 -8.53569448e-01 2.37914640e-02
8.14480484e-01 -1.23367715e+00 -3.81498218e-01 7.28799701e-01
-1.20573127e+00 -1.00362122e+00 -3.18047285e-01 6.27200842e-01
-5.96110642e-01 -3.78798140e-04 -7.17336774e-01 -4.12806302e-01
-6.08135521e-01 -1.52697933e+00 1.05467474e+00 4.68490422e-02
-4.56165403e-01 -1.08831549e+00 -1.04633264e-01 6.61731064e-01
5.09468079e-01 -3.99838805e-01 8.76015484e-01 -9.13071394e-01
-5.93586490e-02 -1.07717945e-03 1.28354495e-02 8.14323783e-01
2.34432295e-01 -1.51857227e-01 -1.22447443e+00 -3.78502369e-01
-1.69644251e-01 -5.26058614e-01 7.53812134e-01 1.49606049e-01
7.99599290e-01 -1.80351436e-01 1.41552970e-01 6.80180311e-01
1.08801889e+00 2.91889995e-01 4.23583388e-01 2.45458782e-01
9.87670541e-01 5.60065687e-01 5.58518231e-01 -3.13945472e-01
7.81509995e-01 7.08017051e-01 -1.03056833e-01 -3.06940079e-01
-5.65135598e-01 -5.17660856e-01 1.05367863e+00 1.45753276e+00
2.43371099e-01 -9.92604047e-02 -1.32799542e+00 7.56269038e-01
-1.85587335e+00 -3.97725552e-01 -3.55875045e-01 2.23789477e+00
1.38318050e+00 7.15489835e-02 -4.80372496e-02 -1.81607883e-02
4.05544132e-01 -2.08005235e-01 -3.99465561e-01 -1.16477764e+00
1.38668776e-01 4.95276779e-01 4.14736390e-01 9.21652257e-01
-7.16484070e-01 1.65171409e+00 6.26854658e+00 1.05971634e+00
-1.13950145e+00 4.23377842e-01 4.32268947e-01 1.27540499e-01
-2.10079938e-01 2.66283214e-01 -1.17495334e+00 5.70216775e-01
1.52889383e+00 -9.73566771e-02 5.52007735e-01 7.54625559e-01
3.39570910e-01 -3.10731173e-01 -1.35497820e+00 8.42080832e-01
5.48865646e-03 -8.85125458e-01 1.20673582e-01 -1.61901578e-01
8.44325125e-01 3.21087092e-01 -3.11595220e-02 7.26504982e-01
7.47725129e-01 -1.22819698e+00 9.87926126e-01 2.28843525e-01
1.17819810e+00 -7.78999627e-01 6.53267384e-01 6.22716665e-01
-1.21825945e+00 2.30601534e-01 -1.45446807e-01 -2.21057698e-01
3.28741252e-01 4.90436882e-01 -1.25824487e+00 8.62118006e-01
6.06573164e-01 3.66331130e-01 -5.25274873e-01 6.75939083e-01
-5.95931411e-01 9.79996443e-01 -2.73996413e-01 4.19992834e-01
4.01543140e-01 -6.08493201e-02 1.32660761e-01 1.64083946e+00
2.82402307e-01 -6.30219400e-01 3.83886248e-01 6.06550276e-01
-1.89770788e-01 9.28369164e-02 -3.10843349e-01 1.41331449e-01
5.75123489e-01 1.23236728e+00 -1.94201261e-01 -7.15927243e-01
-3.64524186e-01 1.45290887e+00 4.85222340e-01 2.87049264e-01
-7.25108981e-01 -6.01353467e-01 4.87364560e-01 -1.04371041e-01
1.02329887e-01 -2.69858927e-01 -3.01855683e-01 -9.44539607e-01
4.76115085e-02 -1.37916851e+00 1.28928870e-01 -7.22903252e-01
-8.52147937e-01 8.72469544e-01 -2.80082338e-02 -7.62914121e-01
-5.30469716e-01 -5.78845024e-01 -4.25080597e-01 1.62719083e+00
-1.55229068e+00 -1.57799709e+00 2.75645014e-02 4.74847436e-01
8.53643417e-01 -1.54738456e-01 8.92944872e-01 4.91667122e-01
-5.43869793e-01 9.26501334e-01 1.00316986e-01 2.85075545e-01
1.02307630e+00 -1.44847143e+00 8.62922788e-01 1.00728786e+00
1.42955378e-01 3.47201675e-01 3.17605793e-01 -6.56124890e-01
-1.04171634e+00 -1.34184444e+00 1.71136498e+00 -7.32370555e-01
5.39136887e-01 -5.90770066e-01 -8.85319948e-01 1.05647850e+00
4.23503041e-01 -4.07703221e-01 5.81767142e-01 1.35215297e-01
-4.27363962e-01 -1.27058744e-01 -9.07882988e-01 5.26579916e-01
7.36699224e-01 -7.28670895e-01 -7.15295911e-01 8.48303959e-02
9.67274249e-01 -7.62456715e-01 -9.89625752e-01 1.04499191e-01
6.67451799e-01 -5.34172893e-01 5.53380787e-01 -4.91391838e-01
4.93719935e-01 -3.28885555e-01 -4.55575697e-02 -1.93703675e+00
-2.13070497e-01 -7.11918414e-01 -5.96481971e-02 1.39961934e+00
8.59903276e-01 -5.00568390e-01 1.80435061e-01 3.41859013e-01
-6.05260730e-01 -5.51406145e-01 -8.93479109e-01 -1.03200340e+00
3.68480623e-01 -6.45844340e-01 4.71637547e-01 9.64073002e-01
-2.67991185e-01 7.07694709e-01 -4.16155815e-01 -7.16462061e-02
1.47951409e-01 -2.23485902e-01 9.17302907e-01 -6.73120499e-01
-5.04433870e-01 -2.64555305e-01 2.53115475e-01 -1.14540124e+00
2.46955454e-01 -1.50363266e+00 2.53513634e-01 -1.72182024e+00
-2.11663954e-02 -5.60956359e-01 -8.67567863e-03 1.00254226e+00
-4.27754283e-01 2.16632307e-01 5.04761875e-01 5.21537699e-02
-1.50917366e-01 2.99819380e-01 1.29097581e+00 6.79210648e-02
-4.24798340e-01 -2.19746144e-03 -5.83468974e-01 5.35386205e-01
8.66580069e-01 -6.20834470e-01 -3.75633299e-01 -1.21167719e+00
2.28675559e-01 -2.32040823e-01 -1.21281579e-01 -9.41197693e-01
2.06912428e-01 1.39777049e-01 1.54788017e-01 -5.02365232e-01
2.11870223e-01 -6.17588460e-01 1.52141571e-01 4.80720222e-01
-2.47671887e-01 6.33880854e-01 3.61656308e-01 -3.15716416e-01
-3.37847173e-01 -1.91364422e-01 7.91742504e-01 -1.89904764e-01
-6.60221994e-01 -8.42300281e-02 -4.70641255e-01 1.51198879e-01
7.55149245e-01 -1.47033274e-01 -1.42623261e-01 5.52006625e-02
-7.46120214e-01 4.59538907e-01 1.56533375e-01 5.73681056e-01
2.87976474e-01 -1.19596374e+00 -1.15056586e+00 2.86168754e-01
-6.44751266e-02 3.81355315e-01 -1.36143520e-01 8.80887210e-01
-4.67830598e-01 4.02696878e-01 -1.11708812e-01 -5.14691412e-01
-1.17556095e+00 6.74726814e-02 4.44933385e-01 -7.49859035e-01
-2.71138221e-01 8.48651350e-01 -1.01064122e-03 -1.16535425e+00
-1.89409852e-02 -5.45673907e-01 3.32701772e-01 1.30672798e-01
3.48462164e-01 5.27785540e-01 4.94534850e-01 -7.01522410e-01
-2.61754066e-01 3.43289733e-01 -2.82623351e-01 -4.17436153e-01
9.44239497e-01 -6.82573244e-02 -8.20405111e-02 7.36223817e-01
1.11252034e+00 1.66090041e-01 -9.81548369e-01 -3.30992907e-01
1.52281716e-01 -1.93711538e-02 -2.13495102e-02 -1.53988731e+00
-6.39409840e-01 1.15562308e+00 1.17845148e-01 -3.08549434e-01
1.01400816e+00 -7.34636709e-02 8.93707812e-01 3.42983544e-01
5.34705296e-02 -1.35171282e+00 -1.72939137e-01 1.14919174e+00
7.02962101e-01 -1.07559586e+00 -5.53414643e-01 -8.71099308e-02
-9.65747297e-01 8.90545905e-01 8.11894357e-01 1.74727291e-01
2.22101867e-01 4.80609626e-01 3.60511631e-01 3.15843195e-01
-8.32178056e-01 -3.15767258e-01 3.52507621e-01 5.18570244e-01
7.65993536e-01 4.47401017e-01 -1.87607408e-01 6.56661153e-01
-8.20642769e-01 -1.87647313e-01 2.20293671e-01 5.54370821e-01
-2.48261392e-01 -1.27426040e+00 -2.25778908e-01 2.55506039e-01
-5.32814741e-01 -6.61209047e-01 -3.22999060e-01 6.66111946e-01
3.12512964e-01 1.05478287e+00 1.97919995e-01 -4.49900001e-01
4.54388946e-01 4.43076521e-01 3.13229889e-01 -9.04354513e-01
-1.30412412e+00 7.26938620e-03 2.55041897e-01 -3.07990193e-01
1.91822931e-01 -3.09529573e-01 -1.21584749e+00 -2.67412454e-01
-1.85473800e-01 3.11921507e-01 7.59886384e-01 1.02113748e+00
2.70640045e-01 6.90719843e-01 4.08931762e-01 -6.43431425e-01
-5.41671216e-01 -1.31188095e+00 4.79684630e-03 1.18677840e-01
2.65646368e-01 -3.71181965e-02 7.56789520e-02 2.80793101e-01]
|
[11.636188507080078, 10.300738334655762]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.