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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
363d1c91-958a-42c8-adfc-1f7b75010216 | attentive-neural-controlled-differential | 2109.01876 | null | https://arxiv.org/abs/2109.01876v3 | https://arxiv.org/pdf/2109.01876v3.pdf | Attentive Neural Controlled Differential Equations for Time-series Classification and Forecasting | Neural networks inspired by differential equations have proliferated for the past several years. Neural ordinary differential equations (NODEs) and neural controlled differential equations (NCDEs) are two representative examples of them. In theory, NCDEs provide better representation learning capability for time-series data than NODEs. In particular, it is known that NCDEs are suitable for processing irregular time-series data. Whereas NODEs have been successfully extended after adopting attention, however, it had not been studied yet how to integrate attention into NCDEs. To this end, we present the method of Attentive Neural Controlled Differential Equations (ANCDEs) for time-series classification and forecasting, where dual NCDEs are used: one for generating attention values, and the other for evolving hidden vectors for a downstream machine learning task. We conduct experiments with three real-world time-series datasets and 10 baselines. After dropping some values, we also conduct irregular time-series experiments. Our method consistently shows the best accuracy in all cases by non-trivial margins. Our visualizations also show that the presented attention mechanism works as intended by focusing on crucial information. | ['Noseong Park', 'Solhee Park', 'Seoyoung Hong', 'Heejoo Shin', 'Sheo Yon Jhin'] | 2021-09-04 | null | null | null | null | ['irregular-time-series'] | ['time-series'] | [ 1.35922268e-01 -3.73275229e-03 3.76962066e-01 -1.72055587e-01
-8.75292197e-02 -3.04361939e-01 6.77861333e-01 -3.68422531e-02
-1.32842243e-01 6.16484106e-01 1.12558894e-01 -4.56330657e-01
-1.83684140e-01 -5.83626628e-01 -4.44113851e-01 -9.03905749e-01
-5.78434765e-01 -1.39350910e-02 6.57151937e-02 -5.86432755e-01
2.29804680e-01 7.24930644e-01 -1.59640837e+00 -1.55471593e-01
9.90654707e-01 1.17535961e+00 -1.52703628e-01 6.05998755e-01
-1.05997063e-01 9.20610666e-01 -7.28422761e-01 -1.54289037e-01
1.45787194e-01 -4.14887398e-01 -3.73099178e-01 -3.13584894e-01
-5.91446869e-02 1.29031003e-01 -4.88413453e-01 7.72678435e-01
7.38846302e-01 4.01822358e-01 8.75866950e-01 -1.51538670e+00
-1.34131706e+00 5.47399879e-01 -3.23593199e-01 6.81165636e-01
-1.54998451e-01 3.26205462e-01 5.90529501e-01 -8.04964602e-01
4.24026161e-01 1.13076723e+00 9.97436643e-01 7.31744528e-01
-1.23166096e+00 -4.37364042e-01 2.53445566e-01 2.02903777e-01
-1.09776890e+00 -8.74229670e-02 1.14453781e+00 -4.87392485e-01
1.28625953e+00 2.82176942e-01 6.82744265e-01 1.16015637e+00
5.99205256e-01 7.39769876e-01 7.72767067e-01 -1.87204912e-01
5.22815108e-01 -1.46865752e-02 3.60700250e-01 1.22103713e-01
-6.77332506e-02 1.34800151e-01 8.65586028e-02 9.15337056e-02
7.80413568e-01 1.62093699e-01 -4.52558368e-01 -5.50993085e-02
-9.83687937e-01 8.36960196e-01 6.05483532e-01 7.26646602e-01
-7.83294499e-01 2.11537287e-01 5.47585368e-01 6.17980301e-01
8.77430022e-01 6.83827877e-01 -4.68192458e-01 -1.23704687e-01
-6.89722538e-01 3.06637168e-01 5.97338080e-01 6.75107896e-01
3.25964272e-01 7.80820251e-01 -3.91015857e-01 6.80859625e-01
-3.36762369e-01 3.54242861e-01 8.62362564e-01 -7.55086660e-01
-5.86081073e-02 4.92864221e-01 -1.83825508e-01 -1.20023525e+00
-4.92189258e-01 -6.32703006e-01 -1.52563095e+00 2.22964019e-01
2.26424709e-01 -6.06512785e-01 -9.78050470e-01 1.98935616e+00
-5.69805596e-03 4.38683987e-01 1.27694726e-01 7.54632592e-01
5.42672276e-01 1.22071660e+00 4.51480635e-02 -4.09708470e-01
9.64718044e-01 -6.84756756e-01 -1.08065498e+00 2.50280619e-01
5.66938579e-01 -2.15873674e-01 7.72483528e-01 1.85016304e-01
-1.09647143e+00 -6.87154055e-01 -8.52569282e-01 8.85139965e-03
-8.29575896e-01 -8.29889700e-02 5.91660023e-01 1.89842731e-01
-1.30620492e+00 9.44394648e-01 -9.83723462e-01 -2.57501602e-01
2.63456911e-01 2.49940649e-01 2.89844559e-03 6.67021930e-01
-1.69488311e+00 9.55448389e-01 5.99348135e-02 3.36240619e-01
-5.21004975e-01 -1.05264175e+00 -7.05542743e-01 2.63090104e-01
-8.21846500e-02 -4.24620003e-01 1.13539469e+00 -1.05703020e+00
-1.58632064e+00 5.21141708e-01 7.44768977e-02 -8.89098108e-01
3.97334963e-01 -1.83256656e-01 -7.18303144e-01 -8.71219411e-02
-2.12327793e-01 6.18058980e-01 8.19100201e-01 -7.18968749e-01
-1.74004942e-01 -4.35569584e-02 -1.21063769e-01 -2.71893263e-01
-5.87123454e-01 8.65295157e-03 7.28605613e-02 -1.10980237e+00
-1.09872997e-01 -8.57334852e-01 -3.73604089e-01 1.37171850e-01
-2.65348196e-01 -7.27151275e-01 1.07535660e+00 -6.72720730e-01
1.65301847e+00 -2.13044190e+00 2.86351830e-01 -4.65991572e-02
1.68408275e-01 4.03603852e-01 8.35491694e-04 4.94177908e-01
-6.74124897e-01 2.12076217e-01 -5.71515501e-01 -2.60495752e-01
1.90944523e-02 1.76531583e-01 -7.48349547e-01 2.88964778e-01
7.35843122e-01 1.03161752e+00 -7.07097709e-01 -1.24969810e-01
2.74834931e-02 9.60705996e-01 -4.47811544e-01 8.61844942e-02
-3.18725735e-01 3.48216027e-01 -2.16121301e-01 3.29143941e-01
5.10045886e-01 -1.75221965e-01 -3.56361002e-01 -1.54051306e-02
-4.44302142e-01 -4.97851111e-02 -8.91917527e-01 1.26385105e+00
-2.90402889e-01 1.15683901e+00 -3.47623676e-01 -1.28551066e+00
1.03788900e+00 4.19822603e-01 5.89413285e-01 -7.10913420e-01
2.75900126e-01 1.57027930e-01 1.87526003e-01 -6.58132017e-01
4.10243481e-01 -8.58322009e-02 2.31067166e-02 5.33691108e-01
-5.95784634e-02 -9.57368091e-02 2.56563246e-01 3.84964496e-02
9.10074294e-01 1.18399858e-01 8.80517662e-02 -3.67604017e-01
5.44110417e-01 -1.57508869e-02 3.97215843e-01 4.68870074e-01
-1.06580250e-01 6.09834611e-01 7.13398039e-01 -7.04796910e-01
-1.21254110e+00 -6.05229139e-01 -8.93215239e-02 9.15701509e-01
-3.38724583e-01 -6.49397597e-02 -8.28280091e-01 -2.56106239e-02
-1.28220499e-01 9.68124509e-01 -9.93154883e-01 -4.22209799e-01
-7.48989165e-01 -6.52689993e-01 6.80747628e-01 7.94618964e-01
3.47714424e-01 -1.53919661e+00 -8.34522009e-01 3.49515975e-01
2.87759632e-01 -5.63666046e-01 -3.57228518e-01 3.79197329e-01
-8.67446423e-01 -5.44642150e-01 -1.40569890e+00 -7.42012739e-01
4.90927547e-01 -5.55011183e-02 1.03796577e+00 -2.55259015e-02
-3.27529162e-01 2.76709646e-01 -1.84179053e-01 -6.95949972e-01
-3.60232651e-01 2.39317685e-01 2.82900900e-01 2.72341698e-01
2.57568270e-01 -9.80156064e-01 -3.99064004e-01 3.25701945e-02
-1.07275760e+00 -2.20293049e-02 3.00784320e-01 5.98940849e-01
3.73383254e-01 -1.31436303e-01 7.34776258e-01 -4.47831750e-01
1.12249196e+00 -8.02574694e-01 -5.69366515e-01 6.58016950e-02
-5.74630558e-01 2.47538567e-01 1.01986539e+00 -9.01954472e-01
-9.13036644e-01 -3.50304902e-01 -2.08294377e-01 -7.86463737e-01
-1.67228624e-01 6.07914746e-01 4.51383859e-01 3.54066938e-01
7.97205627e-01 3.41620684e-01 9.94516239e-02 -4.25498098e-01
1.63795605e-01 6.10338509e-01 5.05050838e-01 -3.50639373e-01
5.70418239e-01 1.91421956e-01 -1.44760199e-02 -9.29225266e-01
-3.75368655e-01 1.96231842e-01 -5.04112184e-01 -2.59390652e-01
8.64541829e-01 -4.11584556e-01 -6.53500795e-01 7.87398934e-01
-1.26524758e+00 -5.98746538e-01 -6.20696604e-01 3.50324035e-01
-3.70091468e-01 -1.35145903e-01 -6.70884371e-01 -1.23037708e+00
-3.84207040e-01 -6.87547207e-01 7.46204078e-01 3.39646041e-01
-3.25129032e-01 -1.36447179e+00 2.80561835e-01 -9.25209939e-01
1.07516956e+00 7.43391395e-01 9.72983897e-01 -6.00770950e-01
-6.68979660e-02 -2.06994250e-01 1.25718033e-02 3.59384567e-01
-3.68304141e-02 3.66169006e-01 -1.04983926e+00 1.46360658e-02
3.31846088e-01 2.02973578e-02 9.45143998e-01 6.41235948e-01
1.36713767e+00 -1.09688319e-01 -1.83558851e-01 6.56745791e-01
1.13502562e+00 6.78145766e-01 8.17874789e-01 -2.87012812e-02
4.55855370e-01 6.20382130e-01 -7.30415657e-02 2.85704792e-01
1.65780544e-01 3.81996453e-01 4.08578604e-01 -2.66839266e-01
2.74085283e-01 1.31500021e-01 4.79051590e-01 9.55732584e-01
-4.55220938e-01 -4.64834988e-01 -1.00241268e+00 5.77651620e-01
-1.72083437e+00 -1.28566372e+00 -3.67059261e-01 1.70280433e+00
5.58181107e-01 7.91861415e-02 9.56056416e-02 4.75793600e-01
8.00225556e-01 2.74570882e-01 -1.10924387e+00 -5.46322286e-01
-4.85824525e-01 2.64174670e-01 -7.90151730e-02 1.49160609e-01
-1.08609402e+00 5.19386172e-01 6.24204636e+00 3.58307570e-01
-1.65218806e+00 -7.34696314e-02 7.52641320e-01 -4.67176847e-02
-1.73743159e-01 -4.61704880e-01 -2.74578273e-01 4.72293466e-01
1.15034175e+00 -6.04458928e-01 2.14599848e-01 7.68323541e-01
1.65760130e-01 4.92695302e-01 -1.05935013e+00 1.03302479e+00
-2.11770266e-01 -1.32918453e+00 -7.47206509e-02 -1.68839410e-01
7.48661041e-01 -3.40125412e-02 2.53920555e-01 6.97614849e-01
6.67139292e-02 -1.00497925e+00 5.46863556e-01 8.87718320e-01
6.11823916e-01 -5.83434045e-01 6.47441089e-01 3.06321144e-01
-1.14986837e+00 -1.78341717e-01 -3.22918475e-01 -3.16093415e-01
2.68901229e-01 4.74393934e-01 -4.83054444e-02 3.26410741e-01
6.50347650e-01 1.03783703e+00 -4.47896302e-01 9.30442214e-01
1.62986130e-01 7.72966325e-01 -3.03447634e-01 -2.98653901e-01
5.15652597e-01 -1.83450162e-01 6.46276891e-01 1.19467449e+00
6.76510274e-01 3.48862648e-01 -4.63017702e-01 1.14239049e+00
2.34470993e-01 -4.25952822e-02 -6.95779860e-01 -2.65150279e-01
1.36287257e-01 1.14464986e+00 -7.49203384e-01 -4.07176942e-01
-2.16794327e-01 1.06811321e+00 2.15721041e-01 6.61763668e-01
-1.08438456e+00 -7.21624792e-01 8.39545667e-01 -5.35445102e-02
3.18798929e-01 -1.84275553e-01 -1.01779319e-01 -1.17246950e+00
-1.35379702e-01 -4.76567566e-01 3.84237856e-01 -1.19840777e+00
-1.42202222e+00 1.11955488e+00 -8.19258243e-02 -1.30081999e+00
-5.55170298e-01 -7.74053335e-01 -9.09777880e-01 1.03636360e+00
-1.22206008e+00 -2.99880058e-01 -2.68701553e-01 4.95416760e-01
5.89649498e-01 -6.24482483e-02 8.36419582e-01 3.04767281e-01
-8.82543743e-01 2.84132361e-01 6.86776191e-02 1.03369623e-01
1.49059653e-01 -1.19006634e+00 8.40501606e-01 8.80793631e-01
-2.23960534e-01 5.86684585e-01 8.30496550e-01 -3.77710313e-01
-1.25950181e+00 -1.14917183e+00 8.23400736e-01 -1.71064690e-01
7.25556374e-01 -3.33817124e-01 -1.58149958e+00 5.67570806e-01
5.16242087e-01 5.61903454e-02 2.75108725e-01 -4.93312389e-01
-2.46736631e-02 -7.59499893e-02 -8.10973167e-01 8.03502500e-01
1.06913626e+00 -3.45261961e-01 -5.05540371e-01 1.89061463e-01
8.47284496e-01 -3.92047226e-01 -8.06149721e-01 2.64301658e-01
2.93378741e-01 -1.00252831e+00 7.80452311e-01 -9.07692969e-01
7.43195474e-01 -8.93486068e-02 1.46451816e-01 -1.68767548e+00
-5.41270435e-01 -9.75027561e-01 -3.77637893e-01 1.14068568e+00
3.52355421e-01 -9.12458777e-01 2.04180077e-01 7.19249964e-01
-4.22668248e-01 -9.93471026e-01 -6.15871012e-01 -8.83450985e-01
4.49483961e-01 -5.46547472e-01 6.82911992e-01 1.26430631e+00
-1.85482070e-01 1.98210031e-01 -4.13648844e-01 -5.49396090e-02
9.97183099e-02 -1.81726776e-02 3.82864803e-01 -1.53330910e+00
5.88769745e-03 -9.12501335e-01 -4.15301740e-01 -9.03748155e-01
2.60844737e-01 -8.44423056e-01 -1.20756693e-01 -1.41536057e+00
-4.98854905e-01 -2.05421075e-01 -4.64241862e-01 3.64602000e-01
-1.44539699e-01 6.31669164e-02 1.95008218e-01 1.17211699e-01
-3.28340009e-03 7.73269594e-01 1.03578627e+00 5.06596453e-02
-4.84831631e-01 2.58662011e-02 -5.60026526e-01 4.45620745e-01
1.02867329e+00 -3.37929070e-01 -5.47966540e-01 -3.65931839e-01
5.88977300e-02 1.50453433e-01 4.61954653e-01 -1.07852435e+00
2.66332388e-01 -7.48646408e-02 2.27652311e-01 -4.84629482e-01
1.21367365e-01 -6.65324509e-01 3.21403205e-01 6.42853141e-01
-5.90390146e-01 6.75757289e-01 3.83377552e-01 2.99095184e-01
-2.31661499e-01 1.15473486e-01 6.43856287e-01 1.18375018e-01
-6.63667440e-01 4.70486701e-01 -7.37970412e-01 8.35578814e-02
9.59169626e-01 -1.18454628e-01 -8.24870467e-02 -7.09341466e-01
-8.23561192e-01 2.29113474e-01 -8.84832740e-02 6.11740053e-01
3.81340802e-01 -1.50590897e+00 -7.86593199e-01 4.71115589e-01
-2.34971255e-01 -1.54748484e-01 2.33569205e-01 9.01003718e-01
-3.77584606e-01 5.58501244e-01 -3.22236627e-01 -6.09675884e-01
-6.53836846e-01 8.94860566e-01 7.33098984e-01 -6.41442016e-02
-7.31416821e-01 7.02876329e-01 5.73756211e-02 -1.25596344e-01
4.53230500e-01 -9.01434302e-01 -4.20137614e-01 3.27701390e-01
5.81837416e-01 4.26425487e-01 -7.30311200e-02 -5.14253974e-01
-2.02156633e-01 6.14429176e-01 2.98459351e-01 -5.72154298e-02
1.85395980e+00 1.36391506e-01 -1.75267532e-02 9.16782439e-01
1.23820758e+00 -6.00620627e-01 -1.39524531e+00 -2.30871767e-01
4.22495082e-02 3.99733663e-01 5.59705216e-03 -4.37446862e-01
-1.36590874e+00 1.29724145e+00 5.77645361e-01 1.24305689e+00
1.45220923e+00 -4.84943658e-01 7.50700533e-01 2.48198375e-01
-1.40201613e-01 -8.47975016e-01 -1.58475220e-01 8.15883934e-01
1.30952597e+00 -8.62220943e-01 -5.33929825e-01 3.30905244e-02
-6.02240205e-01 1.33657134e+00 5.49551487e-01 -5.00595510e-01
8.88239741e-01 3.98814052e-01 3.21394429e-02 -7.19176233e-02
-1.15043390e+00 -1.29839361e-01 2.49969110e-01 6.73729479e-01
5.93452752e-01 -3.98785919e-01 -8.56228620e-02 6.00880802e-01
-1.98407263e-01 7.05895424e-02 3.65106553e-01 7.16985703e-01
-2.54928023e-02 -4.72671062e-01 -3.14375132e-01 2.64966756e-01
-4.62271184e-01 -3.71785313e-02 -3.54913443e-01 9.17545378e-01
-1.56606898e-01 4.97411072e-01 5.89276671e-01 -2.88145989e-01
4.59221125e-01 4.11938280e-01 -2.93239076e-02 -5.15832007e-02
-8.73911262e-01 -2.42395595e-01 -3.61088574e-01 -2.81619281e-01
-3.99723679e-01 -5.57121933e-01 -1.26015425e+00 -3.87893826e-01
3.88265564e-03 1.32537961e-01 4.70000058e-01 6.83528602e-01
6.58794940e-01 1.00185513e+00 6.20910227e-01 -1.28779936e+00
-5.32360554e-01 -1.14012206e+00 -3.69289726e-01 3.27566683e-01
6.67044520e-01 -4.92440701e-01 -6.01530731e-01 1.31765068e-01] | [6.988212585449219, 3.1679301261901855] |
6e444c13-3a5e-4977-b01c-5dd183e02f39 | dynca-real-time-dynamic-texture-synthesis | 2211.11417 | null | https://arxiv.org/abs/2211.11417v2 | https://arxiv.org/pdf/2211.11417v2.pdf | DyNCA: Real-time Dynamic Texture Synthesis Using Neural Cellular Automata | Current Dynamic Texture Synthesis (DyTS) models can synthesize realistic videos. However, they require a slow iterative optimization process to synthesize a single fixed-size short video, and they do not offer any post-training control over the synthesis process. We propose Dynamic Neural Cellular Automata (DyNCA), a framework for real-time and controllable dynamic texture synthesis. Our method is built upon the recently introduced NCA models and can synthesize infinitely long and arbitrary-sized realistic video textures in real time. We quantitatively and qualitatively evaluate our model and show that our synthesized videos appear more realistic than the existing results. We improve the SOTA DyTS performance by $2\sim 4$ orders of magnitude. Moreover, our model offers several real-time video controls including motion speed, motion direction, and an editing brush tool. We exhibit our trained models in an online interactive demo that runs on local hardware and is accessible on personal computers and smartphones. | ['Sabine Süsstrunk', 'Tong Zhang', 'Yitao Xu', 'Ehsan Pajouheshgar'] | 2022-11-21 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Pajouheshgar_DyNCA_Real-Time_Dynamic_Texture_Synthesis_Using_Neural_Cellular_Automata_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Pajouheshgar_DyNCA_Real-Time_Dynamic_Texture_Synthesis_Using_Neural_Cellular_Automata_CVPR_2023_paper.pdf | cvpr-2023-1 | ['texture-synthesis'] | ['computer-vision'] | [ 3.84316981e-01 1.25071347e-01 2.05430202e-02 2.68296570e-01
-3.07937026e-01 -5.80664873e-01 7.29038596e-01 -5.96435130e-01
-3.00744940e-02 7.89830446e-01 1.54742468e-02 -3.40345800e-01
2.69525975e-01 -1.02683640e+00 -9.52921093e-01 -7.53096581e-01
-2.57149667e-01 3.79060358e-01 4.96072233e-01 -3.70723248e-01
-3.49469180e-03 4.71804559e-01 -1.68575346e+00 3.32923412e-01
6.20308638e-01 9.92703080e-01 5.15755117e-01 1.29182625e+00
3.29115987e-01 1.09844053e+00 -3.06449682e-01 -4.51766625e-02
4.63730991e-01 -5.86073399e-01 -3.76150280e-01 3.65662545e-01
4.07442600e-01 -4.45233047e-01 -6.37014687e-01 6.76166594e-01
4.67683733e-01 7.07330033e-02 3.67557555e-01 -1.01633537e+00
-6.25483751e-01 5.03317893e-01 -2.19634816e-01 -8.65618885e-02
2.97248155e-01 5.24326384e-01 3.73298556e-01 -6.71289206e-01
1.01883304e+00 1.20544338e+00 5.96583366e-01 8.98430586e-01
-1.15347397e+00 -5.16262114e-01 3.50023545e-02 -1.94227919e-01
-1.39657748e+00 -8.69589627e-01 3.34422916e-01 -1.37811795e-01
9.94794667e-01 5.20166337e-01 1.09383297e+00 1.26195264e+00
4.32356089e-01 6.70721412e-01 1.05905259e+00 -5.61458409e-01
2.58245319e-01 -5.56279302e-01 -8.43585134e-01 9.59438741e-01
-1.12546831e-01 1.76838905e-01 -3.93464595e-01 3.20304260e-02
1.67333007e+00 -3.93261641e-01 -3.66946161e-01 -1.79959461e-01
-1.61725354e+00 3.19670260e-01 -2.63263375e-01 -4.38357890e-02
-1.47825927e-01 8.55264902e-01 2.04223558e-01 3.85674000e-01
3.11623842e-01 3.44490051e-01 -1.44904345e-01 -4.95050520e-01
-8.53511751e-01 3.84441704e-01 7.63995528e-01 1.28373814e+00
4.65539277e-01 5.50216258e-01 -2.06085846e-01 6.02288961e-01
-2.44836465e-01 9.40143764e-01 2.35401601e-01 -1.70732224e+00
1.41529247e-01 -3.35343368e-02 4.22075063e-01 -1.22906089e+00
-3.45566706e-03 1.27221256e-01 -9.79832888e-01 3.10755849e-01
4.27326590e-01 -3.63184243e-01 -8.17350745e-01 1.47702253e+00
2.73957670e-01 5.75110078e-01 -1.32778421e-01 7.11634338e-01
1.49360836e-01 9.90740836e-01 -2.16978550e-01 -4.27467674e-01
8.33819270e-01 -1.31815493e+00 -8.07192683e-01 1.78693637e-01
4.94643271e-01 -7.92116761e-01 1.38399661e+00 4.72711176e-01
-1.40328336e+00 -2.21189365e-01 -1.08176482e+00 7.87956268e-02
2.61432026e-02 1.06257141e-01 7.33401358e-01 5.59134305e-01
-1.44551671e+00 7.87776113e-01 -1.30611050e+00 -3.07931006e-01
1.53157830e-01 4.54704523e-01 3.83887291e-02 2.04876915e-01
-7.92238414e-01 4.63767111e-01 -1.30219400e-01 1.25645012e-01
-1.08889878e+00 -3.83869380e-01 -6.04447186e-01 8.58443882e-03
6.46204472e-01 -8.97491932e-01 1.42476690e+00 -1.19143343e+00
-2.48713613e+00 6.43100321e-01 -6.07443973e-02 -4.19621915e-01
8.54411066e-01 8.59699547e-02 -3.52933288e-01 3.58629525e-01
-2.30771273e-01 8.37182939e-01 7.22076416e-01 -1.24932349e+00
-4.58462596e-01 4.28260893e-01 3.91356945e-01 1.23987004e-01
-1.94019362e-01 -4.26879749e-02 -7.96626866e-01 -1.06199300e+00
-3.70706260e-01 -1.33517408e+00 -4.52239275e-01 4.70533758e-01
-2.30379924e-01 5.63485920e-01 9.72475886e-01 -2.99109757e-01
1.37089169e+00 -1.81792831e+00 1.99817881e-01 4.86997366e-02
8.66246372e-02 7.16120144e-03 -1.44537792e-01 3.33762586e-01
5.24326265e-01 1.91615298e-01 -5.21368124e-02 -1.55847102e-01
-1.10235326e-02 3.02596480e-01 -3.26479584e-01 2.23884657e-01
-2.89245218e-01 7.07092762e-01 -8.10524523e-01 -4.68737185e-01
2.09753945e-01 4.04607266e-01 -8.39383423e-01 -1.33566670e-02
-6.89049423e-01 5.01611412e-01 -3.12502980e-01 6.68242276e-01
3.98741096e-01 -3.09391737e-01 5.91139853e-01 1.88610762e-01
-3.34129572e-01 -1.91417173e-01 -1.08535981e+00 1.60354996e+00
-6.17986262e-01 8.82313311e-01 1.65428847e-01 -3.45149279e-01
6.11391068e-01 4.18726861e-01 3.29224586e-01 -6.29418492e-01
2.78283060e-01 2.60739386e-01 -2.51073748e-01 -3.12096983e-01
7.78705776e-01 3.19229960e-01 5.74874170e-02 6.74506903e-01
-4.13130790e-01 -4.09307927e-01 4.25035745e-01 -1.64723545e-02
1.10568213e+00 4.05509621e-01 -5.38534969e-02 -5.90184808e-01
1.72258973e-01 -9.62150097e-02 4.58059520e-01 8.63295078e-01
1.04271233e-01 8.01273763e-01 6.25101745e-01 -6.55761778e-01
-1.83893943e+00 -8.00582588e-01 3.20976734e-01 9.38688993e-01
3.24424654e-01 -4.31519926e-01 -1.22864056e+00 6.12507202e-02
-3.99922013e-01 3.50945562e-01 -6.86128557e-01 2.61045128e-01
-9.91013885e-01 -3.74654353e-01 6.40321255e-01 3.65697712e-01
5.60325265e-01 -1.14025354e+00 -8.06054711e-01 3.58460367e-01
-7.32318684e-02 -1.10710168e+00 -7.36263216e-01 -4.72202420e-01
-8.44029546e-01 -7.16617465e-01 -7.36614525e-01 -8.95200253e-01
6.05965436e-01 3.07453632e-01 9.20514882e-01 4.35539156e-01
9.93990619e-03 1.95669055e-01 -1.37698740e-01 6.02932163e-02
-7.03119397e-01 1.72423664e-02 3.78337651e-01 -6.81059361e-02
-5.85855901e-01 -8.63824844e-01 -7.32069314e-01 6.18467867e-01
-1.03314936e+00 1.02641952e+00 2.36007739e-02 7.84666181e-01
8.00362051e-01 -7.97645524e-02 2.44173318e-01 -7.32694387e-01
4.02778953e-01 3.85570824e-02 -9.22065020e-01 3.22462499e-01
-3.60183388e-01 -4.04721536e-02 8.68036509e-01 -7.98016906e-01
-1.09712231e+00 5.98212797e-03 -4.52360436e-02 -5.89154661e-01
2.07673237e-01 1.57404363e-01 -1.31418081e-02 -4.21522051e-01
4.28777725e-01 2.57126093e-01 -2.32297499e-02 -2.26556510e-02
1.77024677e-01 3.66310269e-01 5.93705833e-01 -8.90803099e-01
5.59622705e-01 7.63222933e-01 -1.40505344e-01 -7.54574120e-01
-9.62847322e-02 6.84690595e-01 -3.18911940e-01 -5.85214138e-01
5.20357847e-01 -8.41223478e-01 -9.88811314e-01 9.37057555e-01
-8.85211647e-01 -1.45790708e+00 -1.59563363e-01 8.22664425e-02
-1.22883868e+00 8.28800127e-02 -1.01374173e+00 -4.33056355e-01
-2.18047589e-01 -1.38326871e+00 1.04867315e+00 1.40085399e-01
-1.42489269e-01 -8.51958454e-01 -7.28266463e-02 -1.52406618e-01
7.40761220e-01 4.87489283e-01 7.03278244e-01 6.34798408e-01
-1.17508948e+00 1.53109297e-01 -2.12466088e-03 -1.90031439e-01
-1.01099126e-01 7.74854004e-01 -4.45418656e-01 -3.95007968e-01
-4.54929322e-01 -1.03934281e-01 4.11013663e-01 4.58301753e-01
1.40208483e+00 -6.76706195e-01 -3.58604670e-01 8.98549557e-01
1.40631711e+00 4.41289693e-01 9.23904896e-01 3.63348186e-01
5.96673548e-01 1.82090458e-02 3.81049573e-01 5.06449342e-01
2.43223935e-01 8.65189075e-01 8.34287480e-02 1.42030707e-02
-3.30523372e-01 -4.13795739e-01 4.81168807e-01 1.25431252e+00
-6.65346563e-01 -7.08885729e-01 -8.00062537e-01 3.89739662e-01
-1.95672131e+00 -1.04710186e+00 1.07588716e-01 1.97608864e+00
8.12634170e-01 1.12539776e-01 -3.12844142e-02 -1.13278612e-01
8.23389530e-01 7.30587542e-02 -3.73313665e-01 -4.26537573e-01
-3.76712143e-01 2.27543354e-01 6.46357656e-01 5.79229712e-01
-9.40207660e-01 1.39899313e+00 7.59451199e+00 1.28475928e+00
-1.41992056e+00 -7.56824315e-02 8.49820733e-01 -5.01342297e-01
-4.78586346e-01 -1.83217246e-02 -2.86985546e-01 4.93193626e-01
1.03980196e+00 -2.70164698e-01 8.90218258e-01 6.61077738e-01
6.91156805e-01 -2.10809365e-01 -6.58116698e-01 8.59278321e-01
-3.04998666e-01 -2.12093520e+00 2.97593921e-01 2.34790474e-01
1.17690313e+00 -2.96115637e-01 1.11572243e-01 -2.41186693e-02
5.41926205e-01 -9.97379482e-01 1.24985683e+00 6.68638051e-01
1.42191803e+00 -7.64165998e-01 1.92580484e-02 1.29889965e-01
-1.25856817e+00 1.73813999e-02 -2.59096980e-01 -2.31273457e-01
5.56614459e-01 5.94186969e-02 -4.69534956e-02 -1.22780889e-01
5.00986457e-01 5.43476820e-01 -3.18489224e-01 6.72415257e-01
1.81164648e-02 5.81827998e-01 -4.04523015e-01 -1.35440931e-01
1.25859216e-01 -1.58163816e-01 4.74267989e-01 1.08831656e+00
8.18553507e-01 4.84395325e-01 7.92559460e-02 5.49834371e-01
-2.59927977e-02 -1.66435599e-01 -5.44022799e-01 -1.36640340e-01
4.65215296e-01 8.24203312e-01 -1.17066061e+00 -5.97051382e-01
-1.22775823e-01 1.09177601e+00 -8.34461488e-03 4.26037401e-01
-1.26159191e+00 -3.95518661e-01 6.57791972e-01 2.98986197e-01
3.03054899e-01 -7.65668750e-01 -3.31810325e-01 -1.54435468e+00
-1.98448256e-01 -1.15094745e+00 -4.21765774e-01 -1.07433438e+00
-5.33464551e-01 6.40460551e-01 -2.21373782e-01 -1.19192469e+00
-3.28284979e-01 -4.61762846e-01 -2.64767647e-01 2.06229642e-01
-6.62620962e-01 -1.19839156e+00 -3.17476064e-01 6.02415800e-01
7.18318760e-01 -1.28801689e-01 8.88914227e-01 3.55552346e-01
-6.04147077e-01 5.12156665e-01 3.29337388e-01 -9.14245471e-02
4.10141468e-01 -7.49634743e-01 8.95784557e-01 8.29618931e-01
-3.29057842e-01 2.77888566e-01 1.00337732e+00 -6.85747027e-01
-1.70141912e+00 -8.59956861e-01 4.35551286e-01 -2.28782699e-01
7.75768220e-01 -6.21936440e-01 -5.44023633e-01 6.46171391e-01
3.49723041e-01 1.54089378e-02 1.02060862e-01 -6.67676687e-01
1.27954856e-01 9.63886157e-02 -7.95147836e-01 1.33607805e+00
1.63578427e+00 -4.13530529e-01 3.66305530e-01 1.88905776e-01
8.24973345e-01 -9.26953256e-01 -7.39597857e-01 2.47100770e-01
1.03760779e+00 -9.90994513e-01 6.84127510e-01 -2.97661815e-02
7.67911732e-01 -3.37659746e-01 -8.30271021e-02 -9.20821369e-01
-1.42868206e-01 -1.33838129e+00 -3.03981781e-01 6.38364196e-01
3.08550298e-01 -3.26492459e-01 8.61888945e-01 6.15461349e-01
-1.61597028e-01 -8.88752997e-01 -5.94926119e-01 -8.02706897e-01
-1.23700120e-01 -4.14580345e-01 7.22243130e-01 8.82853270e-01
4.45150174e-02 -3.19151461e-01 -7.33629584e-01 -6.24982119e-02
4.12228078e-01 6.14828095e-02 8.83532584e-01 -4.13443625e-01
-4.34848547e-01 -4.03906643e-01 -1.49590492e-01 -1.44530869e+00
-1.25915334e-01 -1.78651974e-01 3.55702788e-02 -1.06220686e+00
7.02007413e-02 -6.05143666e-01 4.53550965e-01 2.76721627e-01
3.90831560e-01 4.68983829e-01 2.69220829e-01 3.69493872e-01
-5.88216305e-01 5.18790424e-01 1.66475153e+00 9.22278985e-02
-2.16431931e-01 -4.52308208e-01 -1.90679327e-01 7.54492819e-01
7.32996464e-01 -1.93098083e-01 -4.31712627e-01 -7.78548539e-01
3.65671486e-01 5.58163047e-01 2.04136744e-01 -1.21551335e+00
1.97652861e-01 -7.75529504e-01 -3.18285674e-02 5.19471020e-02
3.17810535e-01 -4.28726315e-01 7.34903455e-01 3.55097473e-01
-3.40299755e-01 3.35945189e-01 1.55905768e-01 4.98167306e-01
1.35537073e-01 4.19600159e-01 8.50293994e-01 -3.00561607e-01
-2.95854509e-01 5.49188852e-01 -1.11982679e+00 -2.29098544e-01
1.05009031e+00 -4.89281088e-01 -4.72299874e-01 -7.16961801e-01
-8.65526378e-01 1.18765971e-02 1.28292513e+00 2.23378599e-01
4.23884690e-01 -1.44318748e+00 -3.34049881e-01 -1.71203949e-02
-3.39573681e-01 -1.21364102e-01 3.59408736e-01 5.22841990e-01
-1.60179865e+00 2.52890378e-01 -4.31945890e-01 -4.65615362e-01
-9.92289007e-01 5.30305803e-01 4.82066542e-01 -1.30732059e-01
-7.40554512e-01 4.79956299e-01 1.60730645e-01 -8.61811414e-02
-1.02383286e-01 -4.68095332e-01 2.77442962e-01 -5.66610694e-01
5.17046332e-01 4.34656739e-01 -3.16316813e-01 -3.96578014e-01
2.30837852e-01 5.94432414e-01 3.99519920e-01 -5.41232944e-01
1.20176685e+00 -4.18066412e-01 -1.28485844e-01 2.19303533e-01
6.37746513e-01 1.83597162e-01 -1.89926755e+00 2.28626475e-01
-5.60306728e-01 -4.90732640e-01 -2.31523186e-01 -4.24688011e-01
-1.14796257e+00 5.00265539e-01 2.58491784e-01 1.07433587e-01
1.11521447e+00 -5.52367687e-01 1.30642235e+00 4.47291285e-01
8.85604560e-01 -1.34625256e+00 1.12049863e-01 6.03206575e-01
8.30727100e-01 -5.31945169e-01 -2.36171186e-01 -6.22833669e-01
-5.60532272e-01 1.28175271e+00 5.71931005e-01 -3.58844042e-01
4.25819188e-01 8.89407039e-01 4.26941030e-02 2.32034519e-01
-1.29010999e+00 3.44282448e-01 -4.22752470e-01 4.43111300e-01
2.07317516e-01 1.41116872e-01 -2.37595126e-01 1.46527356e-02
-3.34077477e-01 2.83821374e-01 1.09400380e+00 8.92098367e-01
-3.63011539e-01 -9.78685498e-01 -1.78359136e-01 -1.04336930e-03
-4.55606520e-01 -6.65806141e-03 3.30974124e-02 7.36406803e-01
-1.53076038e-01 5.81754506e-01 2.67722309e-01 -4.99621689e-01
-1.44558772e-01 -3.56115878e-01 7.42874622e-01 -1.36881843e-01
-3.08309257e-01 3.25930834e-01 3.38374048e-01 -6.92604959e-01
-3.61016452e-01 -5.11773407e-01 -1.01148999e+00 -1.11270666e+00
-2.03125048e-02 -2.35942557e-01 3.30399454e-01 6.18616641e-01
6.57702506e-01 3.77615035e-01 4.94074941e-01 -1.32153201e+00
2.44876340e-01 -4.69989896e-01 -4.09117967e-01 -5.64261898e-02
1.31851077e-01 -5.08637369e-01 -4.01124284e-02 6.13341212e-01] | [10.880929946899414, -0.6492212414741516] |
5961df6b-aff0-411d-8e9e-58edd43ada57 | aida-upm-at-semeval-2022-task-5-exploring | null | null | https://aclanthology.org/2022.semeval-1.107 | https://aclanthology.org/2022.semeval-1.107.pdf | AIDA-UPM at SemEval-2022 Task 5: Exploring Multimodal Late Information Fusion for Multimedia Automatic Misogyny Identification | This paper describes the multimodal late fusion model proposed in the SemEval-2022 Multimedia Automatic Misogyny Identification (MAMI) task. The main contribution of this paper is the exploration of different late fusion methods to boost the performance of the combination based on the Transformer-based model and Convolutional Neural Networks (CNN) for text and image, respectively. Additionally, our findings contribute to a better understanding of the effects of different image preprocessing methods for meme classification. We achieve 0.636 F1-macro average score for the binary subtask A, and 0.632 F1-macro average score for the multi-label subtask B. The present findings might help solve the inequality and discrimination women suffer on social media platforms. | ['David Camacho', 'Javier Huertas-Tato', 'Alejandro Martín', 'Guillermo Villar-Rodríguez', 'Helena Liz', 'Álvaro Huertas-García'] | null | null | null | null | semeval-naacl-2022-7 | ['meme-classification'] | ['natural-language-processing'] | [ 2.30012491e-01 6.77162632e-02 -1.54326051e-01 -3.37542802e-01
-8.61879766e-01 -8.66231918e-02 8.51881921e-01 2.76915610e-01
-7.79539287e-01 7.17938483e-01 2.06917167e-01 1.11727091e-02
-1.47493869e-01 -6.32946432e-01 -3.69665593e-01 -5.34539282e-01
3.24324280e-01 4.07158256e-01 -1.60455272e-01 -2.31320381e-01
2.64598131e-01 2.90909648e-01 -1.71034122e+00 8.12131703e-01
7.26587653e-01 1.31497037e+00 -1.57878861e-01 7.89069533e-01
1.36246020e-02 8.23378384e-01 -5.27601361e-01 -1.12734687e+00
-3.41054022e-01 -1.80188134e-01 -6.78077877e-01 -2.31885910e-01
9.45687950e-01 -2.73586720e-01 -3.63103718e-01 1.34861612e+00
7.11099803e-01 -8.88540372e-02 1.20245898e+00 -1.14294755e+00
-5.87711036e-01 9.28109646e-01 -7.67095089e-01 3.22694212e-01
2.98708677e-01 -4.33669239e-01 6.96725965e-01 -1.24866509e+00
5.89058340e-01 1.56678998e+00 9.46648419e-01 5.93482971e-01
-7.75991738e-01 -1.06336033e+00 -4.39580411e-01 6.81692302e-01
-1.37966633e+00 -6.53029501e-01 3.82317752e-01 -4.06546354e-01
6.47080898e-01 3.84520113e-01 2.91225135e-01 1.33758700e+00
4.57629353e-01 7.70055830e-01 1.44021833e+00 -3.39321941e-01
-4.12502915e-01 3.96279633e-01 3.90471071e-01 8.57998013e-01
1.71402499e-01 1.36089986e-02 -8.81916225e-01 1.55610561e-01
2.07397997e-01 -3.07650357e-01 4.58327860e-01 6.39630258e-01
-1.00715637e+00 8.67535055e-01 5.46623543e-02 6.44018531e-01
-9.79443640e-02 1.05748028e-02 5.79922974e-01 2.12095156e-01
7.67097592e-01 1.07639112e-01 -6.46508709e-02 8.44413787e-02
-1.53679740e+00 4.09707487e-01 3.75901788e-01 5.45720279e-01
4.36543852e-01 -9.21387747e-02 -6.61403835e-01 1.09017658e+00
4.85237360e-01 7.70371974e-01 5.79481125e-01 -1.04655278e+00
5.87087631e-01 4.15337414e-01 -3.20188433e-01 -1.39172208e+00
-9.26499128e-01 -4.26296949e-01 -1.03474283e+00 9.77006182e-02
5.07527113e-01 -1.56835228e-01 -8.89602482e-01 1.44577873e+00
-3.06250483e-01 -1.42959401e-01 -1.19550854e-01 4.45447832e-01
1.50084352e+00 3.67558986e-01 3.43936175e-01 -1.19302962e-02
1.46222508e+00 -1.01907754e+00 -9.58536625e-01 -2.14965437e-02
5.13016641e-01 -8.80692720e-01 5.47014415e-01 2.65289813e-01
-1.39025927e+00 -6.86330497e-01 -1.06099439e+00 -8.09054747e-02
-7.23797679e-01 7.22898304e-01 3.81693482e-01 1.17227840e+00
-1.29560840e+00 4.02169645e-01 -3.36335748e-01 -4.86391664e-01
6.77563429e-01 5.83813131e-01 -3.79498273e-01 1.40143648e-01
-1.16157937e+00 9.80437517e-01 2.96576887e-01 2.10018739e-01
-7.02234745e-01 -4.63607579e-01 -7.87336051e-01 8.65478739e-02
-2.31896099e-02 -2.99781710e-01 9.16038811e-01 -1.12936378e+00
-1.04061365e+00 1.37551904e+00 -2.11649407e-02 -4.90685910e-01
5.02357960e-01 1.02683417e-01 -6.28588200e-01 3.17081034e-01
2.08715454e-01 1.06352019e+00 8.61474395e-01 -1.10491443e+00
-8.43915939e-01 -5.20680487e-01 -2.79755205e-01 1.95379034e-01
-6.44822121e-01 4.95170802e-01 -1.63658738e-01 -6.19754434e-01
-1.69715807e-01 -7.94285595e-01 3.29389602e-01 -5.01108766e-01
-3.47920030e-01 -3.80404815e-02 8.03985000e-01 -1.34286857e+00
1.38115060e+00 -1.92815316e+00 4.54927199e-02 3.28542799e-01
6.34097874e-01 3.38076651e-01 -9.60585773e-02 -4.03866954e-02
-2.18013719e-01 3.52735758e-01 1.78368568e-01 -7.05156803e-01
-8.39303136e-02 -3.64875704e-01 4.31632757e-01 3.44606400e-01
-1.00462861e-01 1.12561071e+00 -4.65002269e-01 -6.19275808e-01
7.18218237e-02 4.09572095e-01 -2.71162421e-01 -1.92925796e-01
4.72814620e-01 2.47278258e-01 -1.47260064e-02 9.95496452e-01
8.78194511e-01 -9.96628478e-02 7.56463483e-02 -4.84174430e-01
-1.20509841e-01 -4.11801338e-01 -7.25857913e-01 1.33801270e+00
-1.16621591e-01 9.87141669e-01 3.20406944e-01 -9.41518903e-01
6.65729344e-01 3.15738916e-01 4.35645342e-01 -1.10461509e+00
7.50897586e-01 3.36810112e-01 3.18417419e-03 -5.99251986e-01
1.01325512e+00 -2.47642145e-01 -1.59444451e-01 -5.78031428e-02
3.23483378e-01 6.23448789e-01 3.52957904e-01 1.90736800e-01
5.75098455e-01 -2.32706830e-01 -6.96198791e-02 -3.86668950e-01
9.69869018e-01 -4.25425380e-01 9.07569677e-02 8.71427357e-01
-4.29166228e-01 7.90950954e-01 5.76935709e-01 -7.48237446e-02
-1.06691813e+00 -6.44545317e-01 -2.54375756e-01 1.30056560e+00
-5.93986139e-02 -2.18047425e-01 -1.01974511e+00 -7.12167561e-01
-2.39324540e-01 4.20257419e-01 -8.74975920e-01 -3.94103974e-02
-3.82131815e-01 -1.26211512e+00 1.15304637e+00 3.88625413e-01
8.94946873e-01 -8.98693085e-01 -2.23671585e-01 -1.61328495e-01
-7.70102739e-01 -1.19938755e+00 -2.69812882e-01 -1.70555368e-01
-4.74626541e-01 -8.09400141e-01 -1.14456999e+00 -6.14079058e-01
3.75670582e-01 -1.93808675e-01 9.56825256e-01 2.67212745e-02
-4.43577878e-02 3.74392331e-01 -3.86085123e-01 -2.84147590e-01
-5.84604084e-01 4.23396498e-01 -1.16669379e-01 6.17240965e-02
3.89212251e-01 1.28333755e-02 -2.94943959e-01 1.46895960e-01
-7.16210485e-01 3.00992876e-01 3.76941323e-01 7.14786053e-01
7.04237819e-02 -1.21025912e-01 5.88563800e-01 -9.32325542e-01
5.65910637e-01 -7.48808682e-01 -1.24823667e-01 5.03089726e-01
-6.48448527e-01 -4.70763296e-01 -1.35743499e-01 -4.97921497e-01
-1.23171651e+00 -3.80536765e-01 -4.35994387e-01 -1.10756144e-01
-7.71170631e-02 6.14641726e-01 -5.97315952e-02 -2.67620564e-01
5.12374938e-01 -2.20315054e-01 -1.00657061e-01 -3.40602875e-01
-6.71580359e-02 9.97885048e-01 5.26183248e-01 -2.56348610e-01
1.78112283e-01 1.94278404e-01 1.88959077e-01 -8.66015673e-01
-8.07179987e-01 -1.87917769e-01 -2.93433696e-01 -8.20354700e-01
1.27361655e+00 -9.72588003e-01 -9.52559292e-01 1.11297405e+00
-8.67280960e-01 -8.09963718e-02 3.86892736e-01 2.36498877e-01
-1.72829807e-01 3.49733323e-01 -7.37186909e-01 -8.50015938e-01
-4.37278748e-01 -1.18752933e+00 9.42121327e-01 4.65726256e-01
-2.84928322e-01 -8.71508658e-01 -3.18877935e-01 1.00984740e+00
5.03253162e-01 1.57113656e-01 8.68572652e-01 -7.53784001e-01
9.31510329e-02 -1.01842217e-01 -8.41888189e-01 3.31878215e-01
-6.04244590e-01 -1.87190980e-01 -1.29875219e+00 -1.51742503e-01
-2.41474301e-01 -5.08592963e-01 1.16925466e+00 5.77177823e-01
1.12026250e+00 5.56925088e-02 -1.52821571e-01 3.85543585e-01
1.11290514e+00 1.26984671e-01 8.08657765e-01 1.78154707e-01
8.17236960e-01 9.19022620e-01 4.16954756e-01 2.44707778e-01
6.69201791e-01 7.03178048e-01 2.43489668e-01 -9.77860093e-02
-6.58589303e-01 1.17119402e-01 4.09201533e-01 7.48085022e-01
-1.56593040e-01 -3.45555693e-01 -1.03670049e+00 3.70999366e-01
-1.81950140e+00 -9.29479480e-01 -5.84656537e-01 1.64335406e+00
3.06995779e-01 -2.64465064e-01 2.47492433e-01 3.25165361e-01
9.08735156e-01 -4.00734432e-02 1.90868661e-01 -4.07700062e-01
-6.30699158e-01 8.82586464e-02 7.88111091e-01 5.34400403e-01
-1.34583461e+00 8.12315941e-01 6.68253946e+00 1.45630598e+00
-1.03018796e+00 5.50082207e-01 1.13196957e+00 5.96392527e-02
-6.07067812e-03 -6.78480148e-01 -9.66605723e-01 6.80339932e-01
1.44700038e+00 3.02772433e-01 2.28712678e-01 3.42541002e-02
-4.68762293e-02 -5.72535515e-01 -6.33399129e-01 1.19927979e+00
6.91393375e-01 -1.32274067e+00 -9.48753804e-02 2.13391975e-01
8.43943059e-01 -9.88567844e-02 6.02367342e-01 4.73902136e-01
-6.12630069e-01 -1.32689762e+00 9.65274394e-01 9.28177893e-01
1.16998291e+00 -8.03589463e-01 1.10276699e+00 -2.25742557e-03
-8.82357359e-01 -2.03593805e-01 2.25452380e-03 -3.95891182e-02
-2.15510517e-01 3.71286035e-01 -4.60577607e-01 5.23964822e-01
7.60865092e-01 6.60882175e-01 -1.20558667e+00 9.13914859e-01
2.88573027e-01 5.88852584e-01 -3.10307387e-02 -5.27921915e-02
8.87955353e-02 3.67236100e-02 5.19373655e-01 1.64814532e+00
5.87320447e-01 -3.59017044e-01 -3.54982615e-01 5.04782796e-01
-3.30865055e-01 3.83936286e-01 -3.47080350e-01 -2.61127263e-01
3.22412630e-03 1.39487100e+00 -1.04089773e+00 -3.83254409e-01
-3.75074446e-01 9.57341373e-01 -4.83335927e-02 3.21097076e-01
-7.96412170e-01 -1.47750184e-01 -1.16854042e-01 -8.97726417e-02
-5.25977351e-02 3.38116676e-01 -9.69209552e-01 -1.19812512e+00
-4.36023474e-01 -8.46743822e-01 5.07801950e-01 -6.88670814e-01
-1.16970384e+00 3.98383588e-01 1.95806861e-01 -7.45798290e-01
-1.34988585e-02 -8.65518272e-01 -2.40525052e-01 7.66747415e-01
-1.36620176e+00 -1.67760539e+00 -1.38505384e-01 4.05562520e-01
3.16494972e-01 -8.29556406e-01 7.42908776e-01 1.06497085e+00
-6.18640900e-01 1.02604854e+00 1.22166686e-01 -3.32290083e-02
7.98246801e-01 -8.98577034e-01 -4.38207030e-01 7.09290862e-01
-3.80879015e-01 4.91554337e-03 4.82107371e-01 -7.43434131e-01
-4.01166469e-01 -9.20750439e-01 1.44276738e+00 -4.91441727e-01
3.91123831e-01 -4.06626388e-02 -4.21292365e-01 3.01431090e-01
3.92937094e-01 -7.86390364e-01 7.27842510e-01 2.75659319e-02
-6.43080920e-02 5.45380227e-02 -1.64702725e+00 4.48347181e-01
5.44417560e-01 -5.13316810e-01 -5.38573898e-02 -8.86988044e-02
6.71085492e-02 -4.12832499e-01 -1.05971563e+00 5.33403993e-01
8.99003863e-01 -1.03204393e+00 1.04226947e+00 -2.42755845e-01
1.10660756e+00 2.52678424e-01 -3.09296250e-01 -6.20203257e-01
-2.91854709e-01 1.71096429e-01 -1.69176966e-01 1.41154599e+00
6.89314663e-01 -1.88598529e-01 6.90095186e-01 4.11529332e-01
1.89430192e-01 -4.40896779e-01 -1.06139195e+00 -1.22961141e-01
2.21758991e-01 -6.78389609e-01 2.16539353e-01 9.08124447e-01
-3.35189015e-01 3.96423250e-01 -6.92026794e-01 -1.32631302e-01
5.61681569e-01 -6.21372342e-01 1.44179642e-01 -1.23900640e+00
2.77999610e-01 -8.63216281e-01 -5.09744585e-01 5.36775496e-03
4.59731340e-01 -1.09786487e+00 -2.92500585e-01 -1.16831398e+00
8.04172635e-01 9.35134739e-02 -4.21465963e-01 3.74763131e-01
-4.85335104e-02 1.10052431e+00 5.21723866e-01 -3.89834531e-02
-6.83470368e-01 2.86675543e-01 9.34462786e-01 -4.20896649e-01
3.60935688e-01 -1.01664007e-01 -8.00648749e-01 8.55998218e-01
8.14763248e-01 -4.72908020e-01 -3.91493775e-02 -1.91627517e-01
6.94344401e-01 4.94773090e-02 4.18084860e-01 -1.05385733e+00
5.16435364e-03 2.69713223e-01 6.93112552e-01 -8.74170482e-01
5.03592551e-01 -5.76577663e-01 2.02669122e-04 3.41047943e-01
-4.46979582e-01 -5.31944111e-02 3.28696966e-01 1.24808028e-02
-1.06774829e-01 -7.45079458e-01 9.45990384e-01 8.68642479e-02
-4.79908049e-01 -9.85428914e-02 -8.11120868e-01 -8.00821036e-02
6.43586993e-01 -1.53828487e-01 -4.14411813e-01 -4.19864953e-01
-1.09595072e+00 1.58147693e-01 7.02463910e-02 3.45241994e-01
6.07779324e-01 -1.46719527e+00 -9.41074967e-01 3.53592820e-02
-4.61982228e-02 -1.13738227e+00 5.70865273e-01 1.29236424e+00
-3.81615639e-01 3.16747904e-01 -6.75909758e-01 -2.11374566e-01
-1.77870536e+00 -1.11387603e-01 4.24624890e-01 -4.41250712e-01
1.93940356e-01 9.92629111e-01 3.99959460e-02 -5.26606083e-01
4.19510037e-01 3.50968152e-01 -9.47504938e-01 6.89175248e-01
7.54434288e-01 1.14784753e+00 8.67539570e-02 -1.19706953e+00
-3.48462790e-01 3.47558737e-01 -1.51549444e-01 -4.68105078e-01
1.09446549e+00 -6.38909101e-01 -5.99110782e-01 5.01962781e-01
1.26447999e+00 -1.14310905e-01 -2.66585410e-01 1.15612648e-01
-1.44185334e-01 -1.89296752e-02 2.98033804e-01 -1.22583568e+00
-1.11275840e+00 9.84088480e-01 1.16639423e+00 2.04650275e-02
9.96643245e-01 -2.22300082e-01 5.48216045e-01 1.02103420e-01
-2.55325615e-01 -1.52686894e+00 -1.92868691e-02 7.58284330e-01
6.24130726e-01 -1.36231196e+00 -1.40719160e-01 -2.83140510e-01
-7.22460806e-01 1.33156145e+00 4.29514021e-01 3.14449042e-01
5.90967357e-01 4.92991321e-02 -5.27886711e-02 -3.17369521e-01
-1.16682142e-01 -1.72816753e-01 7.81316519e-01 6.37528896e-01
6.11825347e-01 1.68639541e-01 -8.66061032e-01 9.81994629e-01
-1.27122119e-01 -2.91949324e-02 5.32510936e-01 2.12773874e-01
-1.40286237e-01 -1.00460660e+00 -6.01826489e-01 8.75912428e-01
-1.03741336e+00 -1.27268031e-01 -4.24241811e-01 4.58878309e-01
7.51497746e-01 1.15001559e+00 -7.35468604e-03 -6.79077506e-01
-1.57441646e-01 2.97819525e-01 4.98400033e-01 -1.92441583e-01
-9.30454850e-01 1.14967093e-01 5.01484156e-01 -3.08809161e-01
-8.13446879e-01 -9.05225635e-01 -8.37674618e-01 -6.30130470e-01
-1.37236550e-01 -3.27868789e-01 8.28760207e-01 1.12545097e+00
-1.63848922e-01 7.24614859e-01 1.39705136e-01 -8.77233088e-01
1.18792810e-01 -1.20789444e+00 -5.57226300e-01 3.07849288e-01
1.44382119e-01 -5.66693902e-01 7.72562027e-02 -9.46424995e-03] | [8.565940856933594, 10.606555938720703] |
ed3316a1-dd63-4f05-afcd-64ff687b4948 | neural-network-for-heterogeneous-annotations | null | null | https://aclanthology.org/D16-1070 | https://aclanthology.org/D16-1070.pdf | Neural Network for Heterogeneous Annotations | null | ['Yue Zhang', 'Qun Liu', 'Hongshen Chen'] | 2016-11-01 | null | null | null | emnlp-2016-11 | ['multiview-learning'] | ['computer-vision'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.33184814453125, 3.6887059211730957] |
c0d5cc7f-7411-45d2-97ab-fb3a62ab8a45 | adaptive-remote-sensing-image-attribute | 2101.06438 | null | https://arxiv.org/abs/2101.06438v1 | https://arxiv.org/pdf/2101.06438v1.pdf | Adaptive Remote Sensing Image Attribute Learning for Active Object Detection | In recent years, deep learning methods bring incredible progress to the field of object detection. However, in the field of remote sensing image processing, existing methods neglect the relationship between imaging configuration and detection performance, and do not take into account the importance of detection performance feedback for improving image quality. Therefore, detection performance is limited by the passive nature of the conventional object detection framework. In order to solve the above limitations, this paper takes adaptive brightness adjustment and scale adjustment as examples, and proposes an active object detection method based on deep reinforcement learning. The goal of adaptive image attribute learning is to maximize the detection performance. With the help of active object detection and image attribute adjustment strategies, low-quality images can be converted into high-quality images, and the overall performance is improved without retraining the detector. | ['Chunhong Pan', 'Jian Wang', 'Yiwei Liu', 'Jiacheng Guo', 'Chunlei Huo', 'Nuo Xu'] | 2021-01-16 | null | null | null | null | ['active-object-detection'] | ['computer-vision'] | [ 4.70330745e-01 -2.28355035e-01 1.23821404e-02 -3.52589458e-01
-1.89519465e-01 -2.40900189e-01 1.42015561e-01 1.57803863e-01
-7.36503482e-01 2.58324981e-01 -3.77752095e-01 -1.60604790e-01
-4.09361959e-01 -1.27066839e+00 -1.41198114e-01 -1.10159242e+00
1.23292513e-01 -1.02572754e-01 2.49762952e-01 -7.96937868e-02
3.09829533e-01 7.32756436e-01 -1.29267514e+00 -4.89119262e-01
1.00612819e+00 9.94126558e-01 5.96309662e-01 4.90545630e-01
-7.94844255e-02 5.45639634e-01 -6.14903152e-01 2.84162074e-01
4.34914768e-01 -2.57077366e-01 -1.85092628e-01 5.42887568e-01
8.40264559e-03 -9.95716333e-01 -2.35714644e-01 1.21550560e+00
8.29249740e-01 1.88013539e-01 4.81440783e-01 -9.82650459e-01
-8.23956430e-01 3.10479522e-01 -7.02874124e-01 4.43657994e-01
-2.98271894e-01 4.28359061e-01 8.94133210e-01 -9.28823113e-01
-9.45590511e-02 9.21114683e-01 2.13747919e-01 3.50841731e-01
-1.19954038e+00 -6.79358661e-01 4.89624947e-01 2.57631302e-01
-1.39401138e+00 -3.78190964e-01 7.51720250e-01 -3.84792835e-01
8.00404072e-01 5.76844588e-02 8.85810792e-01 3.73539448e-01
-7.84334540e-02 9.23111260e-01 7.94476867e-01 -5.13883531e-01
3.33056331e-01 -3.88718545e-02 -9.86537933e-02 4.25105661e-01
5.30682266e-01 5.55861473e-01 6.80903122e-02 1.94036886e-01
1.06221640e+00 3.45882177e-01 -2.00299382e-01 -3.11149120e-01
-9.58059967e-01 8.92117739e-01 8.77444685e-01 1.16237596e-01
-4.69903618e-01 -5.39674945e-02 -1.66412629e-02 1.73638836e-02
3.97120416e-01 7.07519889e-01 -2.44090483e-02 3.96374792e-01
-6.48844421e-01 -1.27479389e-01 2.01628566e-01 3.93592089e-01
9.72763062e-01 5.00469446e-01 -1.36102065e-01 7.41178274e-01
5.12548387e-01 9.77272630e-01 1.25417151e-02 -1.00250232e+00
-1.45372106e-02 9.90911961e-01 3.00984472e-01 -9.19852972e-01
-3.65344733e-01 -6.55726612e-01 -7.41928697e-01 8.56362343e-01
-8.18772241e-03 -2.51302272e-01 -8.44881594e-01 1.07194972e+00
3.24890137e-01 -2.49439865e-01 1.00837477e-01 1.07618237e+00
8.50176871e-01 8.69234145e-01 2.62777597e-01 -5.64402878e-01
7.56971896e-01 -7.11180449e-01 -6.78836286e-01 -5.45728505e-01
1.59634247e-01 -6.19895160e-01 9.96144235e-01 2.60968387e-01
-8.00118029e-01 -6.98084235e-01 -1.24013484e+00 3.91699612e-01
-1.22843809e-01 4.05542105e-01 7.38025904e-01 5.04082680e-01
-5.61875582e-01 1.12417653e-01 -8.66731822e-01 -9.36542898e-02
3.21722716e-01 4.20320004e-01 2.15579852e-01 1.76932603e-01
-9.90973890e-01 7.28275239e-01 5.24677098e-01 5.29710293e-01
-8.77110422e-01 -3.08814108e-01 -5.18071830e-01 2.10606813e-01
6.56224549e-01 -3.66919756e-01 1.33727562e+00 -1.45380187e+00
-1.56585693e+00 5.49940467e-01 4.13291365e-01 -1.69414341e-01
3.89311939e-01 -2.42726058e-01 -3.59291345e-01 1.62191153e-01
-1.11788683e-01 5.88143468e-01 9.88453507e-01 -1.16817033e+00
-9.72694695e-01 -2.95972347e-01 2.83507049e-01 5.22471488e-01
-5.83927512e-01 4.52989750e-02 -1.19558401e-01 -2.81808585e-01
2.79182523e-01 -3.83777231e-01 -4.52007174e-01 5.33170819e-01
1.45393744e-01 -9.22494102e-03 1.09570932e+00 -2.45235562e-01
8.75402629e-01 -2.25660348e+00 -1.88952282e-01 -1.05201215e-01
1.25471160e-01 7.34880149e-01 -3.10687751e-01 6.18539564e-03
3.25235188e-01 -2.15062827e-01 -2.87616402e-01 3.21725905e-01
-3.97597343e-01 1.95995867e-01 -2.91661590e-01 4.05837506e-01
5.21469176e-01 7.76141465e-01 -1.03170228e+00 -4.99342889e-01
6.73331618e-01 2.82594234e-01 -1.86382666e-01 5.03228307e-01
-5.21563739e-02 3.32702905e-01 -7.13271499e-01 7.58771658e-01
7.74462163e-01 -1.74414366e-01 -1.01889871e-01 -9.83725712e-02
-5.05392671e-01 -2.08765551e-01 -1.24074125e+00 8.82901788e-01
-2.66805917e-01 5.94864786e-01 2.94533610e-01 -1.08914745e+00
1.23343492e+00 5.71082234e-02 7.16200769e-01 -1.09806979e+00
4.67047468e-02 6.46657050e-02 2.48024061e-01 -7.12617278e-01
5.51052153e-01 1.11582011e-01 4.66933459e-01 1.88162208e-01
-6.21528327e-01 -1.87535286e-01 -6.82019591e-02 -1.97502553e-01
4.60516512e-01 9.68461186e-02 5.90711057e-01 1.57569513e-01
3.23476464e-01 1.04844004e-01 7.05156207e-01 8.62215579e-01
-4.18278694e-01 2.81844079e-01 -2.26707593e-01 -4.61638004e-01
-9.88784611e-01 -1.24286258e+00 -2.32778862e-01 1.06547105e+00
6.48029387e-01 4.09647614e-01 -4.69422042e-01 -2.64004081e-01
-3.70534207e-03 3.39675188e-01 -4.28860635e-01 -3.83396953e-01
-5.91473639e-01 -1.39898002e+00 1.58556119e-01 6.77571654e-01
9.42084014e-01 -1.19928455e+00 -9.93589818e-01 5.15768647e-01
7.03004748e-02 -6.35746837e-01 7.48856589e-02 -1.29828572e-01
-1.04433179e+00 -8.53800952e-01 -5.46924889e-01 -5.61893761e-01
6.47418320e-01 9.06062186e-01 6.87754810e-01 4.38086569e-01
-6.29448235e-01 2.76836663e-01 -4.65409905e-01 -7.02708125e-01
-3.23242128e-01 -9.18052569e-02 -2.53964722e-01 4.98617217e-02
2.33689666e-01 -1.15011990e-01 -1.07536435e+00 3.57473820e-01
-8.83706033e-01 -1.06878970e-02 1.03902268e+00 7.48506248e-01
5.20097017e-01 2.68675655e-01 5.32757342e-01 -4.24319595e-01
1.63011387e-01 8.45817998e-02 -1.07499325e+00 2.51451105e-01
-8.92942190e-01 -4.03069437e-01 2.46905044e-01 -5.85120440e-01
-1.38846040e+00 3.41889918e-01 2.79262643e-02 5.69301322e-02
2.89957244e-02 5.57481945e-01 -3.50660980e-01 -3.40144008e-01
8.55672181e-01 4.22311783e-01 1.54708415e-01 -1.16211452e-01
1.87388230e-02 8.52103651e-01 2.69244522e-01 8.79308116e-03
7.07499921e-01 5.66539526e-01 -1.19800240e-01 -9.74639714e-01
-8.83434653e-01 -5.02966106e-01 -5.29962897e-01 -4.80279893e-01
6.11023486e-01 -9.06009793e-01 -7.08804667e-01 7.15292037e-01
-7.86120296e-01 -4.29784566e-01 -3.71906281e-01 7.26614475e-01
-2.05907896e-01 4.15362835e-01 -3.30922514e-01 -1.31780708e+00
-5.92479944e-01 -9.32933390e-01 7.73159027e-01 5.73649585e-01
4.73896325e-01 -5.87205768e-01 -3.17788780e-01 6.08757921e-02
7.35992014e-01 -1.10800669e-01 6.64500594e-01 2.01361224e-01
-1.04805315e+00 -2.95476288e-01 -4.73025352e-01 4.89619732e-01
3.28459352e-01 3.57141048e-01 -9.11313474e-01 -3.07813734e-01
7.62412623e-02 4.54211384e-02 8.73853445e-01 6.80109799e-01
1.02058685e+00 -8.93499702e-02 -1.47679269e-01 5.97371399e-01
1.70862401e+00 5.38585007e-01 7.92044997e-01 6.61153853e-01
4.43448484e-01 4.96256649e-01 9.19980645e-01 4.47582096e-01
-9.38683599e-02 5.33204019e-01 9.23828483e-01 -7.89068401e-01
-7.59786181e-03 1.15310937e-01 2.03856185e-01 3.25221688e-01
-2.50128746e-01 -2.34881863e-01 -8.35657775e-01 2.46523201e-01
-1.74056184e+00 -1.13819432e+00 -6.15555160e-02 2.20009947e+00
3.43873650e-01 1.23042829e-01 -9.75117385e-02 1.20929040e-01
8.16982508e-01 5.09072170e-02 -1.00446844e+00 2.62544543e-01
-2.12658092e-01 -4.90337253e-01 6.88423753e-01 4.06743705e-01
-1.29101157e+00 6.76422000e-01 7.01510954e+00 5.12748957e-01
-1.53910422e+00 -2.32952327e-01 3.96812856e-01 9.29156542e-02
9.52234119e-02 4.53454480e-02 -7.19241261e-01 8.52659345e-02
4.92410846e-02 -4.82663512e-02 2.92050153e-01 9.30425823e-01
6.03669167e-01 -3.27861905e-01 -5.42892992e-01 7.36822844e-01
-1.89087033e-01 -7.89837301e-01 4.52716611e-02 5.16533256e-02
4.53250825e-01 -2.73068044e-02 9.81347039e-02 1.22879893e-01
2.51824614e-02 -6.62500381e-01 6.30023718e-01 5.73331773e-01
4.24580216e-01 -5.56633770e-01 5.93081295e-01 5.61521530e-01
-1.05933857e+00 -6.48844540e-01 -7.51614511e-01 -4.05792505e-01
1.54111847e-01 5.83084106e-01 -7.38482535e-01 1.20041832e-01
5.93339801e-01 5.21898925e-01 -5.26206672e-01 1.42216611e+00
-3.97462100e-01 6.85726881e-01 -2.21586302e-01 -2.73301244e-01
5.39806113e-02 -4.03725922e-01 6.11210883e-01 7.87938416e-01
2.78617769e-01 3.37543011e-01 4.27973092e-01 1.03439462e+00
3.22985828e-01 1.32903278e-01 -3.03977996e-01 -1.10319465e-01
5.72203815e-01 1.38907576e+00 -7.78959036e-01 -2.94760525e-01
-4.14779812e-01 5.96838057e-01 4.77736481e-02 4.75262403e-01
-5.96348882e-01 -3.58266264e-01 2.23376155e-01 1.18843146e-01
2.34071389e-01 -3.96768957e-01 -2.55215883e-01 -8.80672693e-01
-1.92440763e-01 -4.92007524e-01 2.61066407e-01 -8.33932817e-01
-1.07893395e+00 1.79096997e-01 -9.37513560e-02 -1.31906712e+00
1.81079894e-01 -6.56249702e-01 -5.66565752e-01 6.82680488e-01
-1.67081475e+00 -8.99781406e-01 -6.51822686e-01 3.93366218e-02
4.99007195e-01 -2.06866026e-01 6.57291412e-01 3.04449946e-01
-6.58708274e-01 2.13325828e-01 4.66397852e-01 2.67450064e-01
3.58283639e-01 -1.02263343e+00 9.88268480e-02 1.08910286e+00
-7.00235963e-02 2.24931285e-01 5.91369629e-01 -2.60165393e-01
-1.17235827e+00 -1.07057869e+00 1.13363057e-01 6.14077859e-02
3.97850096e-01 -1.09891847e-01 -9.65802252e-01 1.20740741e-01
-1.45400211e-01 6.30173162e-02 3.80914390e-01 -2.28864849e-01
3.93171199e-02 -6.24908507e-01 -9.98051584e-01 3.33908170e-01
6.11733258e-01 -2.17775494e-01 -4.20249343e-01 1.38021082e-01
4.90928322e-01 -1.84837416e-01 -4.86706227e-01 6.95741534e-01
4.44016427e-01 -6.16738617e-01 1.16905785e+00 -1.08328968e-01
2.92203408e-02 -5.73851526e-01 8.93057883e-02 -8.59115779e-01
-7.71943271e-01 1.36947274e-01 1.84519500e-01 9.32247221e-01
3.16249520e-01 -5.03096700e-01 7.15867698e-01 2.26246908e-01
-1.05382390e-01 -4.07880068e-01 -5.88826954e-01 -6.83511913e-01
-2.71898985e-01 6.51340038e-02 4.32834595e-01 7.28010118e-01
-6.12907708e-01 5.43797076e-01 -2.86781132e-01 4.90945071e-01
8.76901627e-01 5.46435595e-01 6.31156147e-01 -1.34133089e+00
-3.99838686e-01 -5.07078648e-01 -3.00150603e-01 -1.06610692e+00
-4.80217874e-01 -2.74860650e-01 4.27730620e-01 -1.75062716e+00
3.69285166e-01 -4.42397565e-01 -3.52033198e-01 5.35113394e-01
-5.41399896e-01 3.15154523e-01 1.10298239e-01 5.61468601e-01
-5.25482059e-01 6.37349784e-01 1.46886134e+00 -4.80352581e-01
-2.77015418e-01 2.64719427e-01 -5.57287514e-01 5.79312146e-01
8.75212193e-01 -2.08463654e-01 -3.81524593e-01 -7.73240089e-01
3.10894549e-01 -2.81919353e-02 5.21650016e-01 -8.34645629e-01
2.09959134e-01 -5.45867145e-01 6.29990458e-01 -4.32210475e-01
1.74953043e-01 -8.68856430e-01 -3.18015784e-01 7.83997416e-01
-1.87259674e-01 -3.61973345e-01 3.61146443e-02 6.47908568e-01
-1.78658411e-01 -4.09821153e-01 1.04223645e+00 -2.14972585e-01
-1.26756310e+00 4.37354326e-01 -6.82710230e-01 -4.50333565e-01
9.09263015e-01 -5.98438203e-01 -3.10826153e-01 -5.81933521e-02
-5.92939019e-01 1.98823243e-01 2.08118603e-01 3.40822875e-01
7.72792578e-01 -1.04730642e+00 -8.31890523e-01 1.94898307e-01
2.31605396e-01 1.42655298e-01 3.63225281e-01 5.86851060e-01
-5.59756815e-01 -1.79661036e-01 -3.82079154e-01 -8.41769934e-01
-1.03667688e+00 4.84948605e-01 7.58286536e-01 2.63850033e-01
-4.96459484e-01 5.23567259e-01 3.83672386e-01 -1.62271619e-01
2.07180887e-01 1.32260144e-01 -2.89598018e-01 -4.88352440e-02
7.00586915e-01 3.72564405e-01 -9.67059880e-02 -2.62652189e-01
-4.68291230e-02 4.93790418e-01 9.53085870e-02 1.63713023e-01
1.37364173e+00 -4.17341322e-01 7.74224997e-02 2.71854550e-01
5.87226927e-01 -2.83249974e-01 -1.65676093e+00 -5.13526440e-01
-3.00357312e-01 -6.92014873e-01 6.29215658e-01 -6.97540760e-01
-1.27455819e+00 1.04289186e+00 1.19024158e+00 2.97957987e-01
1.47446024e+00 -3.60184222e-01 1.90794125e-01 7.30487287e-01
1.38666511e-01 -1.33700871e+00 4.72922862e-01 3.65322888e-01
6.31632209e-01 -1.62600648e+00 3.75298411e-01 -4.31984872e-01
-2.75939763e-01 1.12148619e+00 9.48400676e-01 5.81161045e-02
2.57056594e-01 2.90749907e-01 5.07112205e-01 -1.95063353e-01
-1.02125928e-01 -4.17161673e-01 1.12860864e-02 6.14594042e-01
1.90917552e-01 1.70279071e-01 -2.96511143e-01 -9.54027921e-02
5.67741930e-01 -1.50619045e-01 3.74120563e-01 8.30888033e-01
-1.29563355e+00 -8.18392336e-01 -4.72889245e-01 4.40217227e-01
-1.53310955e-01 7.59812817e-02 -1.88594326e-01 6.38002753e-01
1.53561667e-01 9.04979467e-01 2.99865007e-01 8.56260117e-03
4.08329725e-01 -7.36270487e-01 3.51626992e-01 -6.83882475e-01
-1.37073606e-01 2.16947749e-01 -4.81155634e-01 6.18626401e-02
-6.80581748e-01 -4.69224721e-01 -1.37215054e+00 3.50026265e-02
-1.03373945e+00 -2.34115832e-02 6.07255161e-01 8.09492052e-01
4.80623245e-02 4.34750974e-01 8.81388724e-01 -5.48810720e-01
-8.87396336e-01 -9.57290471e-01 -8.03015590e-01 8.98479521e-02
3.38451356e-01 -5.82810938e-01 -3.09472501e-01 -1.66788831e-01] | [8.99502182006836, -0.9670644402503967] |
50ba233c-ecce-4377-832d-9e845f86877b | meta-two-sample-testing-learning-kernels-for | 2106.07636 | null | https://arxiv.org/abs/2106.07636v2 | https://arxiv.org/pdf/2106.07636v2.pdf | Meta Two-Sample Testing: Learning Kernels for Testing with Limited Data | Modern kernel-based two-sample tests have shown great success in distinguishing complex, high-dimensional distributions with appropriate learned kernels. Previous work has demonstrated that this kernel learning procedure succeeds, assuming a considerable number of observed samples from each distribution. In realistic scenarios with very limited numbers of data samples, however, it can be challenging to identify a kernel powerful enough to distinguish complex distributions. We address this issue by introducing the problem of meta two-sample testing (M2ST), which aims to exploit (abundant) auxiliary data on related tasks to find an algorithm that can quickly identify a powerful test on new target tasks. We propose two specific algorithms for this task: a generic scheme which improves over baselines and a more tailored approach which performs even better. We provide both theoretical justification and empirical evidence that our proposed meta-testing schemes out-perform learning kernel-based tests directly from scarce observations, and identify when such schemes will be successful. | ['Danica J. Sutherland', 'Jie Lu', 'Wenkai Xu', 'Feng Liu'] | 2021-06-14 | null | http://proceedings.neurips.cc/paper/2021/hash/2e6d9c6052e99fcdfa61d9b9da273ca2-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/2e6d9c6052e99fcdfa61d9b9da273ca2-Paper.pdf | neurips-2021-12 | ['hypothesis-testing', 'hypothesis-testing'] | ['methodology', 'miscellaneous'] | [ 4.12427709e-02 -4.00260210e-01 -4.29425597e-01 -2.81225473e-01
-1.32355464e+00 -7.62745738e-01 7.77781427e-01 1.31202698e-01
-5.30307233e-01 9.60159481e-01 -4.80190724e-01 -6.81651175e-01
-4.26485687e-01 -4.36076581e-01 -4.96797264e-01 -8.59255672e-01
-3.48378718e-01 7.93013394e-01 5.72615504e-01 3.76328200e-01
4.06519920e-01 4.72041398e-01 -1.54408157e+00 4.04142141e-02
1.27404571e+00 7.55414128e-01 7.46413767e-02 1.05116236e+00
1.91618770e-01 5.85851908e-01 -5.26572943e-01 -3.24846864e-01
2.01456726e-01 -4.89722550e-01 -7.43248880e-01 -9.13024172e-02
5.37797630e-01 -2.12934479e-01 1.50477365e-01 1.11034155e+00
5.62219977e-01 1.50853917e-01 1.21329796e+00 -1.48122907e+00
-3.51061463e-01 3.19881052e-01 -5.67185581e-01 4.49942082e-01
-5.30041158e-02 7.60622025e-02 8.21260870e-01 -8.76602113e-01
1.40680417e-01 1.01780546e+00 7.47588873e-01 4.91105407e-01
-1.67885387e+00 -7.11650312e-01 -4.17405307e-01 1.43304944e-01
-1.17051458e+00 -4.06446904e-01 5.21631479e-01 -6.35955572e-01
4.94240463e-01 2.22645670e-01 1.82590604e-01 1.37838185e+00
-1.01558506e-01 1.14612794e+00 1.81117654e+00 -3.71694058e-01
6.99238658e-01 6.16533041e-01 4.93610770e-01 4.96233493e-01
4.44149464e-01 2.53776461e-01 -2.53460556e-01 -8.70399654e-01
6.36516511e-01 -2.28498220e-01 -2.71556973e-01 -7.82261074e-01
-1.32732785e+00 1.00398529e+00 -2.11361974e-01 1.29958540e-01
-2.24383086e-01 -2.55659997e-01 5.28898954e-01 7.02286839e-01
9.05599952e-01 5.90350449e-01 -9.19143140e-01 -3.95953059e-01
-1.21438897e+00 3.01218897e-01 1.09037209e+00 4.65668678e-01
7.70454347e-01 -1.39304698e-01 -2.63393670e-01 8.19473207e-01
-3.14471871e-01 5.04362822e-01 4.52250987e-01 -4.05850857e-01
3.55913490e-01 2.04690978e-01 1.70739993e-01 -2.85724759e-01
-1.24020971e-01 -3.68189633e-01 -6.84777677e-01 4.34679449e-01
7.87750900e-01 -2.92389601e-01 -8.17167163e-01 1.54224694e+00
4.81953710e-01 7.79796660e-01 1.97390020e-01 5.64948499e-01
6.65395260e-02 2.12280214e-01 -3.20033245e-02 -7.19380900e-02
1.02458739e+00 -8.64368856e-01 -9.86285135e-02 -8.39910135e-02
8.75129402e-01 -5.41724205e-01 1.27455211e+00 5.12905657e-01
-7.61332810e-01 -5.57239115e-01 -1.01992798e+00 6.59409881e-01
-4.36606079e-01 2.96985477e-01 8.57970238e-01 8.40523660e-01
-9.52551007e-01 6.42651439e-01 -5.75690031e-01 -2.69008756e-01
5.48000455e-01 3.77181351e-01 -6.53860211e-01 -1.09146029e-01
-7.90845394e-01 8.62978995e-01 5.02660811e-01 -2.71751612e-01
-1.11905539e+00 -9.56943095e-01 -6.06097519e-01 -1.56343400e-01
5.36604822e-01 -2.67741889e-01 1.27332389e+00 -8.15400779e-01
-1.32630169e+00 7.59118497e-01 8.83935206e-03 -4.75152433e-01
6.20968640e-01 -3.73298883e-01 -1.63781181e-01 1.82456240e-01
1.39916256e-01 1.35571152e-01 1.14411676e+00 -8.74839127e-01
-8.32243741e-01 -3.33353162e-01 -2.70183951e-01 -1.22743346e-01
-3.93823326e-01 -2.40125880e-02 -1.73178315e-01 -6.10212862e-01
-3.87623847e-01 -9.83769059e-01 -1.99579582e-01 -3.13058406e-01
-4.23686534e-01 -5.60334146e-01 7.95822322e-01 -2.79218465e-01
9.61007357e-01 -2.15998650e+00 -3.46016586e-02 3.47172111e-01
2.89943814e-01 5.58442950e-01 -3.88268083e-01 2.09159985e-01
-1.62579834e-01 -8.66150931e-02 -1.24184035e-01 -1.04423285e-01
2.63085868e-02 -1.81131139e-02 -4.53146368e-01 5.42355776e-01
4.26841021e-01 1.02378857e+00 -8.39137673e-01 -2.87773192e-01
2.94190366e-02 2.20021933e-01 -2.19949335e-01 4.26048279e-01
4.20802422e-02 2.76857525e-01 -4.87005949e-01 7.69747913e-01
7.03635395e-01 -3.74940246e-01 -1.13597892e-01 1.50964931e-01
1.90395296e-01 -2.65686270e-02 -1.36075759e+00 9.83240366e-01
-8.84720013e-02 5.82825482e-01 -8.30793604e-02 -1.47794259e+00
7.50530362e-01 -2.86441389e-03 7.76649341e-02 -3.67211759e-01
-1.77747801e-01 4.36184973e-01 9.44035202e-02 -3.97197604e-01
1.00990618e-02 -1.51348278e-01 -7.60043785e-03 5.65184236e-01
1.21401556e-01 -1.17874937e-02 5.91201894e-02 1.79994870e-02
1.59368420e+00 -1.26638949e-01 4.75883096e-01 -5.34235835e-01
4.93766338e-01 -1.09056152e-01 1.20740108e-01 1.27736533e+00
-3.15520465e-01 1.93837851e-01 8.20969522e-01 -1.76635280e-01
-9.87247169e-01 -1.42397845e+00 -2.84641653e-01 1.22978532e+00
-2.78176755e-01 1.70259960e-02 -5.37147284e-01 -1.07811606e+00
4.19499844e-01 5.05841553e-01 -9.06134188e-01 -2.38492578e-01
-1.34873897e-01 -1.01635563e+00 6.86447680e-01 6.56090260e-01
4.68688488e-01 -7.71121860e-01 -4.44863647e-01 -1.20774038e-01
3.38691860e-01 -9.71458435e-01 -1.36492610e-01 7.71573365e-01
-9.48354363e-01 -1.53325129e+00 -9.55741405e-01 -6.47462785e-01
5.61159253e-01 4.90473777e-01 1.05687869e+00 -3.66698891e-01
-4.19156373e-01 4.77806658e-01 -3.78431231e-01 -2.39841536e-01
-3.88822138e-01 1.92316651e-01 2.19814822e-01 -5.14021106e-02
6.56471372e-01 -4.71886218e-01 -1.90304250e-01 4.79187578e-01
-9.55504656e-01 -4.06528592e-01 1.06582093e+00 1.07060897e+00
2.33757094e-01 3.36110622e-01 7.93477654e-01 -1.21807110e+00
1.07249272e+00 -6.50788307e-01 -7.10689664e-01 4.55584645e-01
-7.85928488e-01 3.92572582e-01 5.85283697e-01 -1.15793788e+00
-5.52769721e-01 -2.38112763e-01 3.79253179e-01 -5.44402778e-01
-3.96607578e-01 5.31858683e-01 1.33014321e-01 -2.45190188e-01
9.29232955e-01 4.72679973e-01 1.01352006e-01 -4.03846979e-01
9.94233564e-02 5.40578246e-01 4.77641523e-01 -1.03075171e+00
1.08103764e+00 2.63331473e-01 2.36811507e-02 -8.32703412e-01
-9.31403458e-01 -9.67218816e-01 -8.39163363e-01 2.44901761e-01
4.36728507e-01 -8.64780486e-01 -5.80320597e-01 4.75669056e-01
-4.17482227e-01 -9.23780322e-01 -2.22599879e-01 7.93265760e-01
-6.95123553e-01 3.89537781e-01 -3.43179405e-01 -1.09347212e+00
-4.21278588e-02 -9.16499913e-01 1.03117335e+00 5.39566055e-02
-2.83953771e-02 -1.29674768e+00 5.83561301e-01 9.53964237e-03
3.98699194e-01 1.79477096e-01 1.02286875e+00 -1.51804137e+00
-2.26214036e-01 -4.77938294e-01 -2.70754725e-01 7.31256604e-01
1.17647208e-01 -1.70447081e-01 -1.03726578e+00 -5.58009744e-01
-1.80601746e-01 -9.58770752e-01 1.01034606e+00 3.21741968e-01
1.13774860e+00 4.82939072e-02 -3.25481206e-01 4.69577640e-01
1.25508451e+00 -1.79674789e-01 5.77706099e-01 3.69569808e-01
3.73899728e-01 5.24789929e-01 6.37489080e-01 2.37710938e-01
4.44176830e-02 4.87741143e-01 -1.32276699e-01 -9.12982374e-02
2.10864499e-01 3.76453213e-02 5.92238903e-01 4.51509148e-01
1.10047795e-01 1.30639061e-01 -8.71227622e-01 6.69709861e-01
-1.96109593e+00 -9.49168503e-01 3.59686557e-03 2.68773007e+00
9.80262935e-01 3.19774479e-01 5.55877566e-01 2.16414556e-01
5.58250666e-01 -2.48786882e-01 -7.75619686e-01 -9.72715914e-02
-1.76080436e-01 7.93664217e-01 3.45223010e-01 1.83036268e-01
-1.44557655e+00 6.51686847e-01 7.17076492e+00 1.38568032e+00
-8.73343289e-01 1.15278080e-01 6.19561315e-01 2.63388544e-01
5.03851846e-02 1.30989224e-01 -9.55072582e-01 4.35301661e-01
1.20013583e+00 -1.01320580e-01 2.08057046e-01 1.25031316e+00
-2.93457925e-01 -3.85978132e-01 -1.30325699e+00 9.66808259e-01
1.39078125e-01 -1.06789029e+00 -1.72579959e-01 2.60948569e-01
7.68107831e-01 3.42052355e-02 3.30802768e-01 7.87188053e-01
5.04171848e-01 -9.70798910e-01 1.72502711e-01 2.97369212e-01
6.60014331e-01 -8.82680953e-01 7.67624140e-01 6.97935104e-01
-8.43698025e-01 -1.00004897e-02 -6.18300617e-01 9.58637372e-02
-6.82665527e-01 9.08838272e-01 -1.05501914e+00 3.01164955e-01
4.04022187e-01 4.34956610e-01 -1.01170504e+00 1.27737737e+00
-2.16510773e-01 1.06261921e+00 -7.20751956e-02 -9.63256434e-02
2.21945137e-01 -6.72448948e-02 2.83321917e-01 1.28649604e+00
2.71956742e-01 -3.95069271e-01 3.54224592e-01 6.76760972e-01
2.21363589e-01 8.93263146e-02 -7.80946434e-01 -1.92659885e-01
2.31021926e-01 1.36893940e+00 -7.51241803e-01 -4.73385423e-01
-5.01473188e-01 1.39914465e+00 8.26466799e-01 4.07334447e-01
-7.20539629e-01 -4.92630094e-01 7.12313712e-01 -1.22280806e-01
4.79934305e-01 -2.75076330e-01 2.22664043e-01 -1.34693515e+00
5.19120023e-02 -1.00722849e+00 6.38751507e-01 -3.53148282e-01
-1.90322185e+00 1.80901825e-01 2.30004326e-01 -1.09870231e+00
-4.57546681e-01 -1.08399177e+00 -7.44243741e-01 7.87308693e-01
-1.61049223e+00 -9.19051826e-01 1.17259689e-01 8.03264678e-01
2.68328696e-01 -3.47483218e-01 8.74198496e-01 1.47957429e-02
-3.27168703e-01 9.04403746e-01 5.72192907e-01 1.29962787e-01
1.05077481e+00 -1.73409605e+00 9.96776018e-03 6.27909780e-01
2.52760082e-01 4.67851937e-01 4.24632668e-01 -6.77006960e-01
-1.45686889e+00 -8.39106858e-01 4.26837683e-01 -8.30091476e-01
1.04716873e+00 -5.22360981e-01 -1.09128869e+00 2.84721881e-01
-3.67363065e-01 2.46424749e-01 1.11429679e+00 5.11738241e-01
-7.54465997e-01 9.91481729e-03 -1.05405223e+00 4.46730912e-01
5.16469121e-01 -7.32217312e-01 -5.10586083e-01 4.06885296e-01
1.18391849e-01 1.49667814e-01 -6.42135084e-01 5.21976769e-01
4.80919600e-01 -1.11390495e+00 9.52934325e-01 -9.44113731e-01
8.54924470e-02 -6.62341937e-02 -1.95424678e-03 -1.40674436e+00
-1.83671549e-01 -6.63169920e-01 -3.79239619e-01 8.47089112e-01
3.43633384e-01 -9.17631865e-01 7.60716319e-01 3.89615029e-01
4.03701335e-01 -7.33823776e-01 -8.63436460e-01 -1.38907635e+00
2.06184790e-01 -5.19718349e-01 -2.94242660e-03 1.17568135e+00
-1.96674000e-02 1.87136233e-01 -3.16986740e-01 2.27565721e-01
9.06200826e-01 1.15685254e-01 1.33549726e+00 -1.68237209e+00
-5.76023996e-01 -5.60466528e-01 -6.37753069e-01 -8.51985157e-01
5.55122852e-01 -6.83101296e-01 3.98723595e-02 -6.75062537e-01
6.85528100e-01 -5.76504827e-01 -5.60025692e-01 4.29129183e-01
-7.05684781e-01 2.81953931e-01 -4.06249881e-01 2.11309731e-01
-7.19498217e-01 1.77979156e-01 8.98657560e-01 -4.40829694e-02
-1.78143829e-01 3.24706942e-01 -6.63699806e-01 4.62831438e-01
7.76263297e-01 -3.29095393e-01 -6.96478248e-01 3.90189320e-01
8.09281766e-02 -2.07890078e-01 7.64815807e-01 -1.03839993e+00
2.76637584e-01 -2.09645435e-01 6.84505582e-01 -4.99946326e-01
-1.97673682e-02 -4.34556097e-01 -3.42660993e-01 4.42151636e-01
-4.36667025e-01 -7.44899288e-02 2.72475928e-01 8.57404768e-01
-1.09372260e-02 -4.10701632e-01 8.09582591e-01 1.22349449e-01
-7.66297340e-01 3.82601678e-01 -1.73173770e-01 3.21044654e-01
1.28299141e+00 -3.00597459e-01 -2.72461504e-01 -2.12708741e-01
-6.00578785e-01 -4.48739342e-02 3.15660983e-01 1.26178235e-01
5.75834394e-01 -1.27327514e+00 -8.68276834e-01 3.34257782e-01
3.59362781e-01 -3.55882108e-01 -1.11029565e-01 1.09915209e+00
-1.20658375e-01 2.27273315e-01 -4.95392866e-02 -6.37924016e-01
-1.20544660e+00 6.90243840e-01 -5.09981774e-02 -5.69796383e-01
-5.50130725e-01 8.67446423e-01 3.62016469e-01 -4.91784304e-01
1.23475187e-01 -2.07491457e-01 1.30588636e-01 2.25625783e-02
8.47091138e-01 3.59880596e-01 -2.57337019e-02 -1.15371253e-02
-2.38440424e-01 2.69311696e-01 -2.90314466e-01 -1.05370991e-01
1.31729078e+00 4.06892002e-01 8.84868801e-02 8.13075185e-01
1.28903437e+00 1.60056800e-01 -1.38772392e+00 -3.36038798e-01
3.37227196e-01 -6.49443865e-01 -2.80411422e-01 -7.51680851e-01
-2.97144592e-01 7.78477490e-01 4.35930580e-01 4.26878929e-01
9.96492088e-01 2.94124305e-01 2.93480843e-01 6.99480891e-01
3.31813216e-01 -8.98457408e-01 5.46913803e-01 4.29926753e-01
2.39322305e-01 -1.57426190e+00 -2.24053226e-02 -1.45472229e-01
-5.22949040e-01 1.20726740e+00 6.07906163e-01 -2.18949109e-01
7.53034234e-01 2.65823364e-01 -1.91887140e-01 -1.22077256e-01
-7.86468208e-01 -5.48995554e-01 6.71410561e-01 9.10802186e-01
2.23173499e-02 1.59332082e-01 2.20886350e-01 3.80466104e-01
1.51204586e-01 -2.52393991e-01 1.59914613e-01 8.08392107e-01
-5.56643128e-01 -1.08163917e+00 -2.34026954e-01 7.97144830e-01
-3.30749631e-01 -2.05693290e-01 -4.04032350e-01 9.68676746e-01
-3.06950182e-01 5.53506851e-01 -1.79657221e-01 -4.24066842e-01
3.32499556e-02 4.37651783e-01 5.97494721e-01 -6.21161938e-01
-8.39221328e-02 -2.42702235e-02 -2.27152437e-01 -4.16291922e-01
-4.65252161e-01 -5.84701300e-01 -4.66915280e-01 -1.85423359e-01
-3.98683101e-01 4.04974431e-01 3.07368219e-01 1.02895987e+00
1.45537600e-01 -2.99982671e-02 7.39213884e-01 -7.06709206e-01
-1.16579473e+00 -9.24497008e-01 -1.05117095e+00 3.42467308e-01
4.09295321e-01 -8.62494051e-01 -8.35814118e-01 -2.05659851e-01] | [7.8446946144104, 4.004261016845703] |
7d07a5ab-a0c6-4c9b-91df-0d2539cb2398 | reinforcement-learning-for-low-thrust | 2008.08501 | null | https://arxiv.org/abs/2008.08501v1 | https://arxiv.org/pdf/2008.08501v1.pdf | Reinforcement Learning for Low-Thrust Trajectory Design of Interplanetary Missions | This paper investigates the use of Reinforcement Learning for the robust design of low-thrust interplanetary trajectories in presence of severe disturbances, modeled alternatively as Gaussian additive process noise, observation noise, control actuation errors on thrust magnitude and direction, and possibly multiple missed thrust events. The optimal control problem is recast as a time-discrete Markov Decision Process to comply with the standard formulation of reinforcement learning. An open-source implementation of the state-of-the-art algorithm Proximal Policy Optimization is adopted to carry out the training process of a deep neural network, used to map the spacecraft (observed) states to the optimal control policy. The resulting Guidance and Control Network provides both a robust nominal trajectory and the associated closed-loop guidance law. Numerical results are presented for a typical Earth-Mars mission. First, in order to validate the proposed approach, the solution found in a (deterministic) unperturbed scenario is compared with the optimal one provided by an indirect technique. Then, the robustness and optimality of the obtained closed-loop guidance laws is assessed by means of Monte Carlo campaigns performed in the considered uncertain scenarios. These preliminary results open up new horizons for the use of reinforcement learning in the robust design of interplanetary missions. | ['Alessandro Zavoli', 'Lorenzo Federici'] | 2020-08-19 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [-1.64470933e-02 2.58792490e-01 2.63068266e-03 -3.99903096e-02
-4.12944615e-01 -4.52355921e-01 9.57821369e-01 3.84165704e-01
-8.59383464e-01 1.21169710e+00 -2.97210008e-01 -6.90535665e-01
-9.02183294e-01 -6.27727747e-01 -8.11287701e-01 -1.03254676e+00
-5.46217144e-01 8.90168250e-01 -3.69991124e-01 -3.13873589e-01
1.81667224e-01 7.85256326e-01 -1.31383479e+00 -6.58431232e-01
7.39845276e-01 9.12337124e-01 6.23837225e-02 5.80249131e-01
4.64110583e-01 2.32568756e-01 -4.94764417e-01 2.72148937e-01
3.11263531e-01 -2.71837823e-02 -2.65042901e-01 1.58246353e-01
-3.43812704e-01 -8.61582160e-02 -1.41332686e-01 1.10012257e+00
3.94791842e-01 8.71868670e-01 6.94205523e-01 -8.04202020e-01
3.45587790e-01 1.19049616e-01 1.78133607e-01 1.00190401e-01
-3.89405177e-03 6.06114149e-01 3.98338437e-01 -4.15493935e-01
5.08133829e-01 9.98200178e-01 2.07960740e-01 -4.35309708e-02
-1.10279822e+00 -6.98898779e-03 -1.24757476e-01 1.89175949e-01
-1.04643202e+00 -2.35917211e-01 3.08155984e-01 -6.38680339e-01
7.76472390e-01 1.04506217e-01 6.71447337e-01 1.06429708e+00
7.81272471e-01 1.25553966e-01 1.10655653e+00 -2.83152908e-01
8.27166140e-01 -9.91108194e-02 -4.25482094e-01 3.95447999e-01
4.64146405e-01 1.07351708e+00 2.72123784e-01 -1.52736440e-01
3.56540233e-01 -4.26125288e-01 -2.21279725e-01 -5.95544040e-01
-1.07813704e+00 8.99863839e-01 3.98918957e-01 2.62408525e-01
-9.55954373e-01 -7.73099810e-02 4.49100852e-01 4.16241199e-01
6.99026557e-03 7.26325631e-01 -1.61008134e-01 -1.61809117e-01
-7.11803555e-01 7.71686018e-01 1.04420233e+00 5.94556749e-01
2.82040685e-01 1.06143141e+00 -4.70589511e-02 -2.19570622e-02
4.08664137e-01 8.11999142e-01 3.05832475e-01 -7.08177507e-01
4.34380621e-01 5.33198181e-04 9.60694075e-01 -7.40890980e-01
-6.98352218e-01 -8.35478067e-01 -5.97719252e-01 7.66877592e-01
2.93111235e-01 -9.18693423e-01 -5.83903909e-01 1.28120470e+00
5.66314578e-01 2.99247578e-02 6.51701093e-01 1.02503920e+00
-3.03214669e-01 8.71992171e-01 -2.55273849e-01 -5.22563696e-01
9.37808573e-01 -2.69564778e-01 -8.75075698e-01 -2.56261021e-01
1.96595341e-01 -4.81365979e-01 3.10798466e-01 6.55240953e-01
-8.19822252e-01 -3.62735152e-01 -1.41278541e+00 1.10479295e+00
-9.06586349e-02 3.36719930e-01 -4.28222306e-02 3.59069735e-01
-4.40152764e-01 9.26179588e-01 -9.13899004e-01 1.65452380e-02
-3.32543463e-01 2.42189810e-01 -6.89052641e-02 3.13769490e-01
-1.36742938e+00 1.43121541e+00 8.40393782e-01 3.93444240e-01
-1.62640941e+00 -3.37667525e-01 -7.22222269e-01 -8.14882591e-02
7.12379813e-01 -4.90615785e-01 1.31124914e+00 -6.84947848e-01
-1.92787445e+00 1.66305844e-02 4.02636617e-01 -1.05034781e+00
8.15558791e-01 -3.03211391e-01 -2.11952791e-01 1.73234463e-01
-2.46233538e-01 -1.07511416e-01 1.21314192e+00 -1.07598841e+00
-6.47372901e-01 -1.19525753e-01 -1.67810693e-01 4.10825372e-01
3.25460672e-01 -4.57969069e-01 2.57441401e-01 -5.30266404e-01
-3.86678010e-01 -1.27090311e+00 -4.60806906e-01 -6.31998956e-01
-1.16303936e-01 1.32089153e-01 3.12623769e-01 -6.14231527e-01
6.56453013e-01 -1.84721768e+00 8.14642549e-01 4.78736192e-01
-7.17066228e-01 5.63994162e-02 3.38915288e-01 6.43016219e-01
-3.01098637e-02 -5.75674832e-01 -3.03451419e-01 -9.18666273e-02
1.54449403e-01 3.50802362e-01 -5.02768040e-01 9.53957856e-01
8.81437659e-02 1.85588479e-01 -9.74013090e-01 1.52782068e-01
4.53177869e-01 1.61307976e-01 -3.38822715e-02 2.79869527e-01
-5.75131953e-01 9.26868916e-01 -6.79355860e-01 3.23630780e-01
2.53376514e-01 6.74950361e-01 1.30219027e-01 9.66285095e-02
-8.01101208e-01 -3.05419862e-01 -1.32407868e+00 1.32320726e+00
-6.70529008e-01 3.66560161e-01 6.59506202e-01 -1.28945029e+00
1.28135729e+00 2.96931595e-01 4.20162559e-01 -6.40466213e-01
7.33885229e-01 4.37787890e-01 1.48851082e-01 -5.51336348e-01
5.39120376e-01 -5.89881539e-01 -8.79506767e-02 3.28927953e-03
9.96650755e-02 -5.50667703e-01 1.46444991e-01 -4.48652625e-01
7.42579937e-01 1.31833777e-01 2.44205296e-01 -5.59291959e-01
7.53821850e-01 1.13617644e-01 6.22997344e-01 5.97246528e-01
1.54317766e-01 -2.52344877e-01 4.66597021e-01 -3.58496398e-01
-1.13209736e+00 -6.16852045e-01 -1.60788313e-01 2.80805230e-01
2.16810375e-01 5.35956204e-01 -3.07144314e-01 -1.26760036e-01
2.77261913e-01 1.27676570e+00 -4.10514027e-01 -3.20961684e-01
-5.01098692e-01 -7.87504852e-01 1.99580535e-01 -5.35455607e-02
3.65030691e-02 -8.67839456e-01 -1.07391059e+00 6.08693957e-01
2.91319281e-01 -8.23357463e-01 5.08073926e-01 4.00655180e-01
-8.79088819e-01 -1.12940562e+00 -4.56669092e-01 -7.58650675e-02
3.43853623e-01 -6.38482869e-01 3.91881585e-01 -3.82078797e-01
-3.25131677e-02 4.48639363e-01 -2.77345091e-01 -5.13748705e-01
-8.45247149e-01 -3.77379775e-01 7.06034362e-01 2.47613132e-01
-5.25927484e-01 -1.13655061e-01 -2.09025413e-01 4.27765995e-01
-8.58725786e-01 -5.13972104e-01 5.69853127e-01 9.93936479e-01
5.01958311e-01 5.93717277e-01 5.17471194e-01 -8.70119198e-04
8.84726882e-01 -4.51168865e-01 -1.57894719e+00 -7.14833215e-02
-5.42945981e-01 3.32050264e-01 8.36507797e-01 -3.31005514e-01
-1.30141032e+00 -1.30044753e-02 -6.68005049e-02 -4.85487998e-01
-2.62041003e-01 7.30493486e-01 1.50500715e-01 -3.34713846e-01
5.74931741e-01 3.09755504e-01 3.47264707e-01 -3.58610421e-01
1.62905350e-01 2.15290472e-01 5.66924572e-01 -8.71635437e-01
8.13805103e-01 2.88982183e-01 5.74284852e-01 -9.51445401e-01
-1.79557100e-01 -7.49707073e-02 -2.70894319e-01 -4.99267280e-01
6.77777171e-01 -7.22575426e-01 -7.77639925e-01 1.62854373e-01
-8.02502096e-01 -4.24683481e-01 -3.76762807e-01 8.64373982e-01
-1.12622273e+00 1.76315561e-01 -1.81271955e-01 -1.26936996e+00
-3.56621236e-01 -1.36410618e+00 4.46111441e-01 2.85972297e-01
1.30302638e-01 -1.07308161e+00 3.09283704e-01 -2.74987310e-01
3.57878178e-01 7.97556639e-01 7.10889578e-01 -4.52202648e-01
-2.55880922e-01 -6.01899922e-01 6.48643553e-01 3.96088928e-01
-4.46620017e-01 -2.17924669e-01 -2.87039369e-01 -8.04771483e-01
5.90202451e-01 -2.31243566e-01 3.92792284e-01 3.69169354e-01
2.27241263e-01 -3.36368442e-01 -1.47478864e-01 4.23372030e-01
1.66420162e+00 5.38010776e-01 5.81161752e-02 8.93927455e-01
-7.91778564e-02 5.51023543e-01 1.47550702e+00 8.29728782e-01
-3.13321382e-01 7.42069006e-01 1.07058716e+00 4.91246790e-01
7.79974401e-01 1.65269017e-01 3.95060867e-01 6.12818114e-02
-4.43457849e-02 -2.31627941e-01 -8.75003040e-01 6.84654832e-01
-1.81860209e+00 -8.39470148e-01 3.11354119e-02 2.70768952e+00
1.10323980e-01 4.79706049e-01 -2.07393736e-01 8.03399011e-02
6.16940379e-01 -2.53477823e-02 -3.65631223e-01 -7.86217630e-01
-3.86145315e-03 -1.42312229e-01 1.04015231e+00 7.73014367e-01
-1.07213616e+00 1.78869426e-01 4.77702141e+00 6.75255358e-01
-1.18385684e+00 -2.83262998e-01 -3.30348266e-03 -6.43916875e-02
8.81539062e-02 1.30234540e-01 -6.90527141e-01 4.11387771e-01
1.29526639e+00 -4.61392045e-01 4.65386420e-01 7.86691546e-01
8.08305860e-01 -3.91735762e-01 -6.15647376e-01 5.18637598e-01
-3.76887441e-01 -1.23980689e+00 -2.62528807e-01 -4.10808325e-02
7.60921597e-01 3.87335829e-02 -1.30790872e-02 4.06642914e-01
-1.90910045e-02 -4.30321664e-01 9.98574197e-01 9.43966269e-01
3.13653886e-01 -1.14160466e+00 1.05769873e+00 6.79602087e-01
-5.54380476e-01 -7.01815963e-01 -2.51401782e-01 -8.34311545e-02
6.85898840e-01 3.82761717e-01 -1.23808134e+00 1.17548347e+00
1.02535315e-01 2.26506993e-01 2.20375031e-01 1.37131500e+00
-3.81476194e-01 4.99377608e-01 -5.97750545e-01 -5.48667312e-01
8.65323126e-01 -5.05281508e-01 1.46437633e+00 8.35597634e-01
7.01856196e-01 -2.63950914e-01 2.49692932e-01 6.25854433e-01
6.50966346e-01 -1.74334973e-01 -7.21688569e-01 1.15730595e-02
2.16989532e-01 1.01018751e+00 -4.47116107e-01 -3.48711550e-01
1.58630952e-01 2.22810492e-01 -1.91190794e-01 3.96243006e-01
-7.05134630e-01 -5.78396857e-01 6.05111718e-01 -1.13116652e-01
4.35793728e-01 -4.58523721e-01 1.12020858e-02 -6.08940780e-01
-1.98535830e-01 -6.56647980e-01 6.79612756e-02 -2.26512998e-01
-7.52444625e-01 4.96585488e-01 3.16525251e-01 -1.58044946e+00
-8.70639384e-01 -6.62044525e-01 -8.12384725e-01 9.90796566e-01
-1.25444448e+00 -3.01718563e-01 2.69389600e-01 1.47926942e-01
4.51882392e-01 -6.04116797e-01 5.54691434e-01 -1.61974952e-01
-3.43220979e-01 -3.24338496e-01 9.07773316e-01 -5.79067349e-01
2.78486878e-01 -1.12121606e+00 -1.28894806e-01 7.91147709e-01
-6.96742654e-01 2.66379744e-01 1.63275564e+00 -9.14004028e-01
-1.75165093e+00 -1.02985072e+00 1.11803576e-01 3.53644222e-01
1.12592828e+00 1.62267745e-01 -7.59583592e-01 4.25166398e-01
2.09850937e-01 -2.50277489e-01 -2.32106000e-01 -4.14164215e-01
9.18895245e-01 1.42655363e-02 -1.05170202e+00 4.10439342e-01
-2.32178539e-01 1.35329768e-01 -8.85865867e-01 2.92180866e-01
4.58895564e-01 -6.78927183e-01 -1.06121409e+00 8.10504913e-01
2.12888315e-01 -4.89821702e-01 6.69680536e-01 -6.70441508e-01
-2.70817608e-01 -4.38660234e-01 2.21868336e-01 -1.86644781e+00
1.74471140e-02 -9.68295813e-01 -1.13658801e-01 5.74018657e-01
1.40905067e-01 -4.98210758e-01 5.55604219e-01 -3.02959885e-02
-3.97777498e-01 -7.65234172e-01 -1.65786600e+00 -9.68802154e-01
1.68504398e-02 -2.40675136e-01 1.04154900e-01 4.02234226e-01
-1.53103709e-01 -1.25735387e-01 -4.48039830e-01 7.25074768e-01
9.44181144e-01 7.07359314e-02 5.98237693e-01 -8.12798619e-01
-4.66807544e-01 -1.72047421e-01 -2.93232620e-01 -3.23412001e-01
3.18025768e-01 -4.65151995e-01 4.62009013e-01 -1.13589251e+00
-1.16655052e+00 -3.23830158e-01 8.57867524e-02 -2.67784983e-01
1.62449107e-01 -6.49743974e-01 1.66631013e-01 -2.40414321e-01
-1.33430228e-01 1.07172787e+00 8.58619213e-01 -1.83092237e-01
-2.81110585e-01 7.76844621e-01 2.54315108e-01 6.27697170e-01
8.21807325e-01 -4.39142287e-01 -3.15199584e-01 -1.34441927e-01
3.84411722e-01 8.56993318e-01 3.48371148e-01 -1.38429463e+00
1.96344092e-01 -1.72729224e-01 3.92115787e-02 -6.04404688e-01
2.95517683e-01 -9.55978990e-01 2.16057107e-01 9.18569446e-01
2.46074088e-02 5.91304079e-02 5.19567788e-01 1.01099801e+00
-4.25844491e-01 -7.44400024e-01 9.36030686e-01 4.76919524e-02
-6.99075580e-01 -4.37974483e-02 -8.35063100e-01 -3.74048024e-01
1.24528480e+00 1.97862327e-01 2.47215360e-01 -3.32240134e-01
-1.15850389e+00 5.51791310e-01 1.95466638e-01 1.61345050e-01
5.80028594e-01 -8.02656710e-01 -6.28379107e-01 9.48203430e-02
-2.40689859e-01 -3.04479688e-01 4.39262956e-01 9.21123743e-01
-4.42565352e-01 5.96258104e-01 -5.44134915e-01 -4.89254266e-01
-6.47261739e-01 7.06983149e-01 7.28454709e-01 -1.16443433e-01
-3.14044267e-01 1.98171750e-01 -5.51663220e-01 -3.02792847e-01
1.20838411e-01 -2.79518068e-01 -1.85677737e-01 1.89922675e-01
2.99874723e-01 4.72898364e-01 4.37923342e-01 -3.00264120e-01
1.95977017e-02 2.11830392e-01 5.03413439e-01 -5.37579060e-01
1.29828715e+00 -2.32372954e-02 2.72329122e-01 3.96021754e-01
6.22332633e-01 -2.75146395e-01 -1.54599643e+00 2.04539269e-01
1.16210297e-01 -1.13451801e-01 2.78361589e-01 -6.20649993e-01
-2.99771577e-01 5.46582818e-01 6.46973073e-01 1.65186822e-01
5.72375417e-01 -7.92556167e-01 1.07906833e-02 7.22693682e-01
5.48631966e-01 -1.16188538e+00 -3.75882000e-01 8.30504119e-01
1.18652594e+00 -7.99135208e-01 8.35869834e-02 3.87277812e-01
-7.16770291e-01 1.47753417e+00 2.80010164e-01 -6.06188178e-01
3.18211585e-01 1.44113630e-01 -1.76350906e-01 3.84565108e-02
-7.25325108e-01 -1.25354946e-01 4.69974428e-02 2.95475334e-01
-2.86956966e-01 1.55062452e-01 -6.42919958e-01 2.03136653e-02
1.09224655e-01 -1.39923885e-01 6.03053033e-01 1.05119634e+00
-7.35870004e-01 -7.15219319e-01 -1.04217279e+00 1.49634168e-01
-1.66925684e-01 4.51209098e-01 4.52806771e-01 1.16732883e+00
-2.67354269e-02 6.13695800e-01 -9.50474292e-02 1.34015098e-01
7.09127247e-01 -2.05062740e-02 3.25962126e-01 -2.73270518e-01
-4.21131372e-01 4.57303002e-02 3.25852662e-01 -3.76367211e-01
2.13797942e-01 -6.76920474e-01 -1.25156164e+00 3.04786861e-01
-1.27882659e-01 7.08348453e-01 1.12959933e+00 1.09033406e+00
-2.58835237e-02 7.42758691e-01 6.11110747e-01 -1.20133603e+00
-1.37924361e+00 -9.96914029e-01 -4.80843902e-01 -9.24354196e-02
4.32450026e-01 -1.01729155e+00 -4.80263561e-01 -5.55831671e-01] | [5.127792835235596, 2.2192676067352295] |
7b49a783-591f-4375-9c3f-a64cd94daa0e | cloud-transformers | 2007.11679 | null | https://arxiv.org/abs/2007.11679v4 | https://arxiv.org/pdf/2007.11679v4.pdf | Cloud Transformers: A Universal Approach To Point Cloud Processing Tasks | We present a new versatile building block for deep point cloud processing architectures that is equally suited for diverse tasks. This building block combines the ideas of spatial transformers and multi-view convolutional networks with the efficiency of standard convolutional layers in two and three-dimensional dense grids. The new block operates via multiple parallel heads, whereas each head differentiably rasterizes feature representations of individual points into a low-dimensional space, and then uses dense convolution to propagate information across points. The results of the processing of individual heads are then combined together resulting in the update of point features. Using the new block, we build architectures for both discriminative (point cloud segmentation, point cloud classification) and generative (point cloud inpainting and image-based point cloud reconstruction) tasks. The resulting architectures achieve state-of-the-art performance for these tasks, demonstrating the versatility of the new block for point cloud processing. | ['Kirill Mazur', 'Victor Lempitsky'] | 2020-07-22 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Mazur_Cloud_Transformers_A_Universal_Approach_to_Point_Cloud_Processing_Tasks_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Mazur_Cloud_Transformers_A_Universal_Approach_to_Point_Cloud_Processing_Tasks_ICCV_2021_paper.pdf | iccv-2021-1 | ['point-cloud-reconstruction'] | ['computer-vision'] | [-7.58156478e-02 -2.51642615e-01 3.32212806e-01 -2.70474255e-01
-7.47293234e-01 -5.42189896e-01 7.78927863e-01 1.74335297e-02
-2.35947505e-01 4.01632190e-01 -1.89773008e-01 3.55250686e-02
1.87290028e-01 -1.33743286e+00 -1.12422442e+00 -6.03052378e-01
-9.55402479e-02 7.62186706e-01 3.64674121e-01 -2.56456882e-01
2.97632486e-01 1.54143238e+00 -1.82742822e+00 5.36172032e-01
3.79037857e-01 1.19649136e+00 1.73214659e-01 8.21340680e-01
-4.87230599e-01 2.27650598e-01 -2.34248266e-01 -3.26021910e-01
4.21139270e-01 3.31317395e-01 -7.80676126e-01 1.94838136e-01
7.51143336e-01 -6.40501320e-01 -3.68368357e-01 4.96525884e-01
2.80865610e-01 4.33279127e-02 4.35176581e-01 -9.03735101e-01
-5.57336867e-01 -1.05403945e-01 -7.84482718e-01 1.65973827e-01
1.32933825e-01 2.01980948e-01 8.65553796e-01 -1.42402530e+00
7.31579900e-01 1.27609527e+00 1.01954532e+00 1.62973776e-01
-1.21193731e+00 -5.36709309e-01 3.26157659e-02 -2.77724683e-01
-1.23859239e+00 -2.48383418e-01 7.63407052e-01 -5.53979218e-01
1.33552408e+00 9.49884206e-02 1.04956543e+00 4.44564313e-01
2.58083344e-01 6.36924803e-01 5.89371145e-01 -2.25081015e-02
1.61496311e-01 -3.94124448e-01 -1.44562081e-01 7.25482166e-01
-3.13760079e-02 2.45801777e-01 -5.07950366e-01 -2.16704726e-01
1.42607725e+00 5.95623493e-01 1.24119826e-01 -5.56814134e-01
-1.22064340e+00 8.00731421e-01 9.50489640e-01 2.49247044e-01
-6.14571273e-01 7.13624060e-01 1.77938625e-01 9.61174965e-02
8.31519723e-01 2.52330154e-01 -4.25806671e-01 9.79532450e-02
-1.36079168e+00 6.75397873e-01 3.71691585e-01 1.16827834e+00
1.35830843e+00 6.87494799e-02 7.34174103e-02 5.41842937e-01
3.14124942e-01 5.12663782e-01 -1.32398516e-01 -1.14450240e+00
3.05136025e-01 7.44548202e-01 4.31105815e-04 -7.68969476e-01
-3.82340699e-01 -5.31444132e-01 -7.81644881e-01 7.23783731e-01
-9.21734720e-02 7.79927671e-02 -1.19133544e+00 1.00144637e+00
4.87967938e-01 4.41639453e-01 -3.34015936e-01 5.85117280e-01
8.46886873e-01 8.61884415e-01 -1.09579436e-01 3.73898238e-01
1.27548635e+00 -7.00410783e-01 -5.88258728e-02 -1.17599458e-01
2.28108093e-01 -5.71052313e-01 6.35712862e-01 1.49135634e-01
-1.72760618e+00 -6.67306006e-01 -9.32540834e-01 -6.16118610e-01
-3.82885903e-01 -9.81055126e-02 8.42626810e-01 -3.14047150e-02
-1.53439438e+00 8.84190857e-01 -1.16797733e+00 -3.44986171e-02
1.00917065e+00 5.91551900e-01 -3.76217216e-01 -1.47308305e-01
-5.73561192e-01 5.09315431e-01 -8.70677829e-02 1.76882080e-03
-7.38177001e-01 -1.32035780e+00 -9.02269721e-01 3.26589733e-01
-4.64647412e-01 -1.39007461e+00 1.25792038e+00 -6.17878258e-01
-1.24292123e+00 1.07427466e+00 -4.06405210e-01 -3.55624735e-01
4.10923481e-01 -2.34470114e-01 1.72962353e-01 2.76819408e-01
2.09658027e-01 1.02979672e+00 1.10063303e+00 -1.53305924e+00
-9.61408973e-01 -6.91911757e-01 -1.76164150e-01 2.17892706e-01
1.87372223e-01 -1.69438019e-01 -6.90614045e-01 -5.14567256e-01
2.71039039e-01 -6.54452264e-01 -2.56208390e-01 4.45504606e-01
-1.04008980e-01 -2.09086925e-01 1.05059040e+00 -4.24596310e-01
5.65542758e-01 -2.32834506e+00 2.39606440e-01 2.90896773e-01
6.25239193e-01 -7.26518333e-02 2.08255164e-02 4.40425277e-01
-2.51337379e-01 -1.07271615e-02 -4.81560171e-01 -9.80885804e-01
-2.15516150e-01 2.85811782e-01 -4.72759277e-01 4.81375456e-01
5.91371894e-01 1.02393961e+00 -7.52599418e-01 -7.67517164e-02
7.02034533e-01 7.36943543e-01 -6.56887472e-01 1.05646089e-01
-3.00698698e-01 4.37763691e-01 -4.34863448e-01 5.91034055e-01
1.27386546e+00 -4.34148997e-01 -6.10423148e-01 -1.64889708e-01
-4.69204426e-01 6.84399381e-02 -8.66559148e-01 2.16033912e+00
-7.04040051e-01 5.56603551e-01 2.25119099e-01 -4.06613380e-01
7.35091507e-01 1.15433522e-01 8.23354244e-01 -2.19379619e-01
-1.74368009e-01 1.65842324e-01 -5.74007988e-01 -2.14974917e-02
6.93239450e-01 -2.21182287e-01 2.59207368e-01 3.15322489e-01
3.51030260e-01 -7.56542981e-01 -2.57284284e-01 5.07134125e-02
9.86404777e-01 2.81485111e-01 -1.50912106e-01 4.65476187e-03
2.63890415e-01 1.75814331e-01 -3.53928469e-02 3.91577989e-01
3.87879789e-01 1.06426203e+00 3.24410409e-01 -9.12713528e-01
-1.48248684e+00 -1.35376489e+00 -2.29835317e-01 7.85412371e-01
-8.81321877e-02 -2.77863830e-01 -4.72139388e-01 -3.67386490e-01
4.27331418e-01 4.43774402e-01 -6.77645206e-01 2.44538113e-01
-8.57385755e-01 -3.03644538e-01 2.87765414e-01 7.31407404e-01
6.03254676e-01 -9.51147437e-01 -6.38373256e-01 1.55307934e-01
2.96219796e-01 -9.84056592e-01 5.06024696e-02 2.70933241e-01
-1.22227609e+00 -7.98222601e-01 -7.86795378e-01 -7.61612356e-01
7.02856719e-01 4.15078491e-01 1.55460119e+00 3.11320156e-01
-9.52224061e-02 3.80766124e-01 2.85805631e-02 -2.96756059e-01
-2.17120022e-01 9.54183191e-02 -4.09152299e-01 -1.12475500e-01
1.49340764e-01 -1.10850191e+00 -8.10474455e-01 -2.38000065e-01
-1.10675025e+00 5.58780506e-02 5.60335994e-01 6.33063674e-01
1.03734660e+00 -3.59844625e-01 -1.47566885e-01 -5.70044935e-01
1.65395126e-01 -4.81435567e-01 -7.53927469e-01 -3.14145774e-01
1.01005897e-01 -8.19060877e-02 4.72024620e-01 2.57954508e-01
-7.23400891e-01 4.79041278e-01 -6.77707911e-01 -9.79926467e-01
-4.36907351e-01 1.48848876e-01 2.09067613e-01 -4.06829000e-01
3.44191372e-01 2.37940431e-01 1.25308512e-02 -6.27025664e-01
6.91524684e-01 1.42402455e-01 6.57526255e-01 -4.20247644e-01
8.71123075e-01 1.23465025e+00 2.50628650e-01 -8.12544584e-01
-3.28674346e-01 -4.74076152e-01 -1.15913832e+00 -1.03621490e-01
1.22714686e+00 -1.07146180e+00 -5.43938518e-01 5.87857723e-01
-1.82997060e+00 -4.45589684e-02 -7.47785211e-01 -9.83572844e-03
-7.77454615e-01 1.37761518e-01 -7.03300297e-01 -3.83229733e-01
-4.05122280e-01 -1.26999593e+00 1.94701266e+00 -4.49556895e-02
9.07465816e-02 -9.12597239e-01 1.19157866e-01 -1.59982949e-01
2.22528860e-01 3.97463650e-01 9.58725512e-01 -1.21429972e-01
-1.20519269e+00 -1.81607053e-01 -2.60799378e-01 2.23012522e-01
-2.63775915e-01 2.63596863e-01 -1.23266006e+00 -1.73737079e-01
4.70193895e-03 -6.53181896e-02 9.99284804e-01 5.82072854e-01
1.47997558e+00 2.36841366e-01 -5.30657351e-01 1.23583710e+00
1.64106274e+00 -2.34258264e-01 8.56684089e-01 1.86177000e-01
9.18105245e-01 2.24732861e-01 -1.50355279e-01 4.65567857e-01
3.24784964e-01 5.35761774e-01 8.66546273e-01 -5.23351729e-01
-6.42792881e-02 -9.20067653e-02 -1.67748123e-01 5.97391605e-01
-3.29370826e-01 7.80177787e-02 -1.07527375e+00 6.51202500e-01
-1.60527754e+00 -1.02245116e+00 -3.92652124e-01 1.93480086e+00
2.49488741e-01 -2.29349449e-01 -5.87421618e-02 -1.63749039e-01
3.99862915e-01 2.71205783e-01 -5.25997758e-01 -3.94406468e-01
1.04466394e-01 9.44430232e-01 5.38521647e-01 3.80505800e-01
-1.09126794e+00 9.27877784e-01 6.98446465e+00 7.45982111e-01
-1.04396653e+00 8.01421106e-02 5.89267731e-01 -5.83365224e-02
-4.81099367e-01 -1.55325130e-01 -6.08154893e-01 1.66655660e-01
5.92013299e-01 3.14566255e-01 3.14506471e-01 8.64699781e-01
-9.01997909e-02 2.92558353e-02 -1.16916609e+00 1.33336079e+00
-2.89875090e-01 -2.15404177e+00 2.71653742e-01 1.91233486e-01
7.61471570e-01 8.10191691e-01 1.69698447e-01 -7.07253367e-02
4.56008643e-01 -9.77352083e-01 8.34213257e-01 5.19317746e-01
8.90965700e-01 -1.04489386e+00 3.34555447e-01 2.45012283e-01
-1.14961183e+00 1.39527977e-01 -4.53573912e-01 -1.68240443e-02
3.00376773e-01 7.43813932e-01 -5.56816041e-01 7.84415603e-01
8.09745967e-01 9.69222844e-01 -3.82902533e-01 9.65396106e-01
1.91064596e-01 -9.41530839e-02 -5.81802785e-01 5.71505189e-01
5.07514298e-01 -2.59823173e-01 5.28232515e-01 1.12314391e+00
4.78652030e-01 -8.95505920e-02 -2.49005511e-01 1.39833724e+00
-1.84886545e-01 -4.31868136e-01 -9.47443545e-01 4.22662735e-01
4.21912968e-01 1.30746675e+00 -7.05197930e-01 -4.88811672e-01
-5.56191146e-01 1.12501252e+00 6.26000643e-01 1.82382673e-01
-5.51731169e-01 -2.11549670e-01 1.17346632e+00 3.53247970e-01
8.06565106e-01 -6.56981468e-01 -8.82455289e-01 -8.76698315e-01
-1.83684379e-02 -1.62765190e-01 -4.02651429e-02 -9.73319829e-01
-1.32585788e+00 6.93094909e-01 -2.87400465e-02 -1.31723952e+00
-1.92534596e-01 -5.60768902e-01 -7.44643211e-01 1.30829573e+00
-1.61527038e+00 -1.36870027e+00 -4.65351999e-01 7.36408234e-01
4.96957660e-01 8.43660682e-02 7.85818994e-01 2.48394653e-01
-1.71378776e-02 7.90432561e-03 2.42535278e-01 -6.27514124e-02
6.17861468e-03 -1.23352695e+00 1.41840565e+00 6.42397821e-01
8.23184922e-02 4.93522316e-01 1.09417979e-02 -6.50822461e-01
-1.42238748e+00 -1.35315537e+00 5.24497211e-01 -6.36249602e-01
3.05882215e-01 -4.42486346e-01 -8.44953477e-01 7.26844609e-01
5.43512926e-02 3.71107072e-01 3.34628940e-01 -2.29112074e-01
-2.65833199e-01 1.06779732e-01 -1.22082329e+00 2.65372813e-01
1.01679623e+00 -5.05139112e-01 -3.58205378e-01 4.69551563e-01
7.71281600e-01 -8.16483736e-01 -8.68395686e-01 2.00446531e-01
3.06905240e-01 -1.29874778e+00 1.37543404e+00 -5.20850301e-01
8.23463619e-01 -2.44180262e-01 -8.34457427e-02 -1.30596602e+00
-7.71980941e-01 -3.44269067e-01 -3.89895448e-03 6.43588126e-01
1.73955128e-01 -2.94482201e-01 9.61180627e-01 2.99722582e-01
-7.71334529e-01 -9.79845881e-01 -9.96247411e-01 -2.48985633e-01
3.01137358e-01 -5.88218153e-01 1.04517496e+00 7.53313899e-01
-6.40215456e-01 -3.12557607e-03 2.43963748e-01 2.73126483e-01
5.05422473e-01 3.28905374e-01 7.94351518e-01 -1.34140444e+00
-6.20252118e-02 -5.21626055e-01 -6.15358889e-01 -1.32749021e+00
4.42288071e-03 -1.02028668e+00 -2.90253907e-01 -1.92302656e+00
-2.16846734e-01 -6.02082133e-01 3.47764194e-01 2.49464586e-01
1.90864746e-02 4.63074565e-01 2.92523146e-01 4.05049980e-01
-1.05843149e-01 4.20794427e-01 1.33227646e+00 1.58765256e-01
-1.71220005e-01 6.57129288e-02 -3.09590071e-01 7.17698693e-01
4.56768960e-01 -1.59673572e-01 -1.48427365e-02 -1.10144329e+00
1.63114205e-01 -3.46718058e-02 8.19697261e-01 -1.20202637e+00
1.57619938e-01 1.40926704e-01 9.37638521e-01 -1.15633726e+00
8.28574538e-01 -1.01021385e+00 4.13779229e-01 2.19153315e-01
2.23580599e-01 5.15840173e-01 4.00175899e-01 5.44459343e-01
-1.50915831e-01 2.61535227e-01 9.51737046e-01 -5.87343276e-01
-6.25330508e-01 8.99344444e-01 2.56682131e-02 -4.54936951e-01
1.00227606e+00 -4.65611398e-01 7.16695562e-02 -1.17732115e-01
-8.38569820e-01 6.98200464e-02 7.42423594e-01 1.63365990e-01
9.26816821e-01 -1.46129930e+00 -6.84691548e-01 7.03884065e-01
-1.91107184e-01 9.50194120e-01 3.61286342e-01 4.77921218e-01
-1.15333772e+00 2.67129719e-01 -3.29397917e-01 -1.10151184e+00
-7.89593518e-01 4.07580853e-01 4.91653949e-01 -1.11030094e-01
-9.85611796e-01 1.09073758e+00 3.93564403e-01 -4.54234779e-01
-1.69135258e-01 -6.43692017e-01 2.46266350e-01 -2.32295126e-01
3.03385139e-01 2.69178271e-01 6.63245499e-01 -6.48607910e-01
-1.51504129e-01 7.53875196e-01 -7.79938838e-03 -1.13373153e-01
1.82305706e+00 2.42995426e-01 -4.59574878e-01 2.80579627e-01
1.34605336e+00 -2.10488573e-01 -1.39260173e+00 -1.22343749e-02
-3.67013544e-01 -6.12962008e-01 2.29280680e-01 -3.02054763e-01
-1.20862997e+00 1.13925743e+00 3.60982001e-01 1.00734062e-01
8.62964571e-01 1.71754241e-01 9.54268038e-01 -2.53847577e-02
4.36298877e-01 -4.78462040e-01 -3.34120512e-01 7.62490153e-01
9.85056818e-01 -7.03588605e-01 6.66652918e-02 -5.56714356e-01
-1.64562687e-01 1.38242912e+00 4.17378277e-01 -7.03882515e-01
8.40923965e-01 4.87300843e-01 -4.15321440e-01 -8.50275993e-01
-5.24087667e-01 -5.63596524e-02 1.53523281e-01 7.20107019e-01
2.21076727e-01 -1.27225995e-01 3.88083726e-01 -5.62163368e-02
-2.57279158e-01 -2.60756519e-02 7.10528493e-02 9.54223990e-01
-2.95874536e-01 -9.66039002e-01 -4.62144732e-01 6.71039701e-01
1.88066829e-02 -7.81009719e-02 -2.47054785e-01 7.81878650e-01
3.48491132e-01 1.51555449e-01 8.87278259e-01 -2.85005271e-01
4.56661016e-01 7.65697956e-02 5.42888105e-01 -7.07996964e-01
-9.50802028e-01 2.63640564e-02 -4.81913805e-01 -9.03672457e-01
-3.57171685e-01 -6.68148100e-01 -1.21260941e+00 -5.91208220e-01
5.19112386e-02 -2.95663625e-01 7.90617228e-01 6.68412745e-01
8.26772928e-01 5.73790133e-01 6.95579171e-01 -1.89184129e+00
-8.28419030e-02 -5.94939947e-01 -4.66088742e-01 2.83359110e-01
6.80622876e-01 -5.04551589e-01 -2.98494864e-02 6.85800379e-03] | [8.048955917358398, -3.688333034515381] |
24fcdd6f-7c96-41b2-8ea9-e5202dc842ba | physics-informed-invertible-neural-network | 2306.17396 | null | https://arxiv.org/abs/2306.17396v1 | https://arxiv.org/pdf/2306.17396v1.pdf | Physics-informed invertible neural network for the Koopman operator learning | In Koopman operator theory, a finite-dimensional nonlinear system is transformed into an infinite but linear system using a set of observable functions. However, manually selecting observable functions that span the invariant subspace of the Koopman operator based on prior knowledge is inefficient and challenging, particularly when little or no information is available about the underlying systems. Furthermore, current methodologies tend to disregard the importance of the invertibility of observable functions, which leads to inaccurate results. To address these challenges, we propose the so-called FlowDMD, a Flow-based Dynamic Mode Decomposition that utilizes the Coupling Flow Invertible Neural Network (CF-INN) framework. FlowDMD leverages the intrinsically invertible characteristics of the CF-INN to learn the invariant subspaces of the Koopman operator and accurately reconstruct state variables. Numerical experiments demonstrate the superior performance of our algorithm compared to state-of-the-art methodologies. | ['Yue Qiu', 'Jianguo Huang', 'Yuhuang Meng'] | 2023-06-30 | null | null | null | null | ['operator-learning'] | ['miscellaneous'] | [ 2.28707373e-01 3.98852192e-02 -3.04355383e-01 3.80313933e-01
-2.07657784e-01 -7.76693285e-01 4.28187668e-01 -6.60866559e-01
1.14719108e-01 7.30439186e-01 1.57989934e-01 -7.75779188e-01
-4.64000642e-01 -2.73111641e-01 -5.10271847e-01 -7.87197292e-01
-2.81305313e-01 -3.53157409e-02 -5.60464501e-01 -1.89582229e-01
-1.76114678e-01 4.49258417e-01 -9.61205602e-01 -2.65084296e-01
8.26005220e-01 8.88979673e-01 -2.65373826e-01 6.92116976e-01
4.80600774e-01 8.96653593e-01 1.01486426e-02 3.77440721e-01
1.03067553e+00 -6.23510540e-01 -7.26963878e-01 4.71559614e-02
3.58627558e-01 -5.81883252e-01 -8.94030988e-01 1.34648311e+00
4.74307872e-02 3.59363198e-01 4.95187402e-01 -1.31165719e+00
-6.63208306e-01 2.53253043e-01 -1.16615012e-01 1.96900159e-01
5.91962412e-02 3.90657783e-01 1.15513992e+00 -1.07142031e+00
6.13848209e-01 1.13347518e+00 6.35703683e-01 3.52536857e-01
-1.79713833e+00 -5.37657559e-01 1.09327853e-01 -1.27749965e-01
-1.39058495e+00 -4.69785988e-01 7.66347587e-01 -9.20314312e-01
7.90409982e-01 2.22911268e-01 6.44957840e-01 7.84840226e-01
5.22206366e-01 5.04730046e-01 8.59116673e-01 -2.04953983e-01
1.02076598e-01 -2.20525473e-01 1.82161614e-01 9.35037851e-01
3.02968115e-01 6.88345790e-01 -4.32735831e-01 -3.57709587e-01
1.19626987e+00 2.90111870e-01 -8.45279694e-01 -9.03657019e-01
-1.39922893e+00 8.26611936e-01 2.64609039e-01 -9.95506346e-02
-3.24981958e-01 1.14881925e-01 2.82568246e-01 6.81381226e-01
1.93328336e-02 5.97536922e-01 -1.83177441e-01 -1.37080833e-01
-5.14881968e-01 2.84211963e-01 1.00302362e+00 6.88208163e-01
8.50570858e-01 5.00766158e-01 1.45315260e-01 -5.89444898e-02
3.13290626e-01 5.85191488e-01 1.14337437e-01 -1.47099173e+00
6.31953955e-01 7.55166352e-01 3.31575096e-01 -1.01445997e+00
-3.41257006e-01 -5.70954025e-01 -1.12235153e+00 4.19951379e-01
5.53321600e-01 -4.86951202e-01 -8.19365084e-01 1.99689257e+00
2.33957708e-01 1.45342425e-01 1.26161963e-01 1.24895763e+00
-7.60264620e-02 7.90375829e-01 -6.85190976e-01 -3.09106112e-01
7.05343902e-01 -6.88865721e-01 -8.21666420e-01 -1.60611078e-01
5.08896470e-01 -2.29564682e-01 9.91626203e-01 1.55901730e-01
-9.29617703e-01 -3.39344203e-01 -1.25474679e+00 2.34846711e-01
1.80701330e-01 2.00057286e-03 5.88858783e-01 2.69200772e-01
-9.24519598e-01 7.37031639e-01 -1.29363823e+00 9.35128704e-02
-2.06343323e-01 5.36725700e-01 -4.43090200e-01 2.72026300e-01
-1.29122996e+00 7.38744497e-01 3.82312089e-01 6.53574228e-01
-1.22417533e+00 -1.01397789e+00 -9.82924819e-01 1.29154965e-01
6.55145705e-01 -8.13995361e-01 1.07117689e+00 -7.78945148e-01
-1.62815797e+00 7.33530475e-03 -1.04504071e-01 -4.56244588e-01
5.08319199e-01 -2.68116653e-01 -1.61962166e-01 3.19100618e-01
-1.01739369e-01 -1.31165888e-02 1.11716855e+00 -9.70732450e-01
-2.60165513e-01 -4.70195152e-02 3.78260493e-01 1.03564098e-01
-4.02630597e-01 -6.51129961e-01 1.23432644e-01 -4.79039907e-01
2.61341393e-01 -1.19062138e+00 -3.13588887e-01 2.58014426e-02
-5.37758350e-01 3.30175102e-01 1.20175016e+00 -7.63660669e-01
1.64628088e+00 -2.12764311e+00 6.83212757e-01 4.15077925e-01
6.21058285e-01 3.52415800e-01 6.01494201e-02 7.67134845e-01
-2.90490746e-01 -2.28285175e-02 -3.29756200e-01 7.23703206e-02
8.38968828e-02 2.64318347e-01 -9.26986933e-01 8.29369545e-01
2.70486414e-01 6.47480130e-01 -8.73825848e-01 7.32465163e-02
4.28669304e-01 4.60546345e-01 -6.05548263e-01 1.23724394e-01
2.16048628e-01 8.99929583e-01 -5.86474776e-01 2.17095986e-01
3.05131048e-01 -3.61170322e-01 1.58764005e-01 -2.33063862e-01
-3.48131657e-01 2.97018111e-01 -1.46447968e+00 1.46086824e+00
-4.83717024e-01 6.56793773e-01 4.67253953e-01 -1.18631279e+00
3.92356396e-01 5.95844209e-01 7.97708809e-01 -2.54079372e-01
1.97087392e-01 2.28917807e-01 3.63420546e-01 -3.81622076e-01
1.23287372e-01 -3.01882207e-01 6.10233285e-03 4.49704409e-01
1.72049031e-02 1.99340060e-01 1.89097330e-01 2.33903438e-01
1.36789429e+00 1.00699849e-01 4.49115813e-01 -7.03918397e-01
8.36872339e-01 -1.44026607e-01 9.11015630e-01 6.21782899e-01
-5.99564373e-01 2.52485335e-01 8.12464833e-01 -3.76062214e-01
-1.03099906e+00 -1.30842721e+00 -2.10414082e-01 3.54856461e-01
7.62650520e-02 -1.68871596e-01 -6.17719769e-01 -2.97756165e-01
3.90587933e-02 2.74608940e-01 -6.32113636e-01 -6.02669716e-01
-8.34447384e-01 -2.59975046e-01 2.01455846e-01 4.20542687e-01
4.58939940e-01 -5.17740428e-01 -5.14271855e-01 2.01313257e-01
9.68411472e-03 -9.85601425e-01 -8.40329409e-01 9.67590287e-02
-8.36791337e-01 -9.85193551e-01 -4.03220385e-01 -6.25328302e-01
9.23692644e-01 7.46364072e-02 6.65897191e-01 -2.65859783e-01
-1.67085841e-01 4.09043401e-01 2.44732276e-01 1.76926330e-01
-4.95378286e-01 -1.03985411e-04 5.08491158e-01 5.24808109e-01
-2.29699194e-01 -7.09426820e-01 -3.87861818e-01 3.31542224e-01
-8.83790433e-01 1.41556740e-01 3.73427838e-01 1.26919258e+00
3.11182320e-01 1.29418686e-01 4.09195960e-01 -5.58323503e-01
7.23645389e-01 -2.56442189e-01 -1.14767754e+00 -2.50453334e-02
-6.06437325e-01 5.00544310e-01 1.16297877e+00 -6.16816044e-01
-7.83862770e-01 4.08237517e-01 5.84481418e-01 -9.87788916e-01
5.22460222e-01 6.23815477e-01 -1.85123444e-01 -3.01640689e-01
4.93715733e-01 2.24260956e-01 2.94806331e-01 -1.63017035e-01
2.43365079e-01 3.83178592e-01 8.13843250e-01 -5.27118325e-01
1.19426465e+00 6.09987557e-01 4.52909470e-01 -7.52099454e-01
-8.67959559e-01 -5.41797876e-01 -6.16941988e-01 -2.01558277e-01
6.02400064e-01 -6.81961477e-01 -1.07284951e+00 2.37645581e-01
-7.67959535e-01 -2.79104024e-01 -4.86517221e-01 6.84258223e-01
-6.72429442e-01 3.62865716e-01 -8.04949343e-01 -1.06462145e+00
-1.17740452e-01 -1.16953039e+00 6.97379291e-01 -4.45949584e-02
-2.00106665e-01 -1.14738822e+00 2.85327226e-01 -5.68121336e-02
3.33507985e-01 3.63835424e-01 7.86426544e-01 -3.13512273e-02
-8.91627192e-01 -3.85361552e-01 1.25629663e-01 4.75826830e-01
6.93563148e-02 -2.00923800e-01 -5.39799869e-01 -6.59422874e-01
3.30361187e-01 -9.74013507e-02 6.53161585e-01 3.46421123e-01
5.22286236e-01 -6.75294995e-01 -1.35304704e-01 6.99171305e-01
1.41141570e+00 1.25258178e-01 6.08698390e-02 -2.74887029e-02
8.11510623e-01 4.25540894e-01 6.86119720e-02 3.89300615e-01
1.13057815e-01 3.49476844e-01 2.93020546e-01 7.94231296e-02
4.47168380e-01 -3.39182258e-01 7.24020779e-01 1.00141954e+00
-1.40661001e-01 1.95412353e-01 -9.36109006e-01 5.06401896e-01
-1.94595551e+00 -8.66445124e-01 2.58495271e-01 2.23492956e+00
5.83498001e-01 7.37711787e-02 -1.09249055e-01 2.85241157e-01
6.32069767e-01 1.58290118e-01 -9.75649953e-01 -2.46778831e-01
9.18587297e-02 -2.33287543e-01 6.18364215e-01 7.83318937e-01
-1.15200257e+00 3.66757482e-01 6.42355108e+00 1.30437568e-01
-1.27397025e+00 -2.06242725e-01 -6.60114214e-02 1.59586787e-01
-9.14891884e-02 4.15217519e-01 -6.09371364e-01 1.78152785e-01
9.58597779e-01 -5.81888556e-01 1.04924643e+00 6.41050994e-01
5.47346830e-01 2.36735597e-01 -1.22861552e+00 7.81361699e-01
-2.68452704e-01 -1.07760441e+00 -1.91010624e-01 3.45637619e-01
9.50436592e-01 -2.57663339e-01 1.94262400e-01 2.20521510e-01
2.90647089e-01 -7.82881737e-01 6.57930911e-01 6.15508020e-01
5.34820735e-01 -5.17232537e-01 3.57069075e-01 4.38005656e-01
-1.35561967e+00 -4.13449436e-01 -7.18797520e-02 -5.07460296e-01
4.26454157e-01 4.81826305e-01 -5.01504481e-01 3.14709902e-01
2.04641625e-01 1.12068188e+00 2.11649626e-01 6.37451231e-01
-1.95369452e-01 7.99764872e-01 -3.33086163e-01 4.51921523e-01
3.77148360e-01 -6.34181619e-01 1.01924849e+00 5.32265663e-01
7.93459266e-02 3.62212360e-02 6.06827676e-01 1.34371173e+00
9.78488177e-02 -4.91705328e-01 -7.94831991e-01 -5.89879811e-01
1.36615768e-01 1.11381173e+00 -1.41617954e-01 -1.56454980e-01
-6.02035582e-01 5.36515951e-01 1.11866333e-01 8.44166815e-01
-5.93975484e-01 -3.52274448e-01 1.03868294e+00 -2.82586545e-01
2.39350274e-01 -7.76044965e-01 -2.04773068e-01 -1.55622387e+00
3.88525158e-01 -1.01425898e+00 2.95418769e-01 -2.10054845e-01
-1.09707546e+00 1.82858258e-01 -3.93721312e-02 -1.70856392e+00
-6.05164826e-01 -6.49816692e-01 -4.27268088e-01 9.15538192e-01
-1.00584364e+00 -7.59933949e-01 4.22716551e-02 5.14985800e-01
9.86814573e-02 -1.00251459e-01 6.34448290e-01 2.63621062e-01
-9.23898101e-01 5.96394725e-02 4.42274392e-01 3.19480926e-01
1.45645082e-01 -1.22196150e+00 2.10620075e-01 1.28057218e+00
-1.32130489e-01 1.06305611e+00 7.46493101e-01 -4.49262470e-01
-1.93981659e+00 -1.05271435e+00 2.71776885e-01 -3.94151509e-01
1.23701763e+00 -4.56097692e-01 -8.79355669e-01 1.08880460e+00
-1.09499261e-01 3.51903051e-01 1.22890860e-01 -2.11598977e-01
-1.30358487e-01 -1.14841990e-01 -5.45610011e-01 8.36011589e-01
1.02965331e+00 -9.47220862e-01 -4.16443408e-01 -3.98708619e-02
6.82311535e-01 -4.48223233e-01 -1.01477265e+00 3.84696275e-01
6.57905161e-01 -5.68855941e-01 1.12858999e+00 -9.88290846e-01
1.87492162e-01 -6.81816459e-01 -1.14869326e-01 -1.41488349e+00
-3.45049351e-01 -1.26385570e+00 -8.12223315e-01 5.06012976e-01
4.10936296e-01 -9.70710695e-01 4.87177968e-01 7.69879580e-01
-2.02835843e-01 -5.77144742e-01 -9.64660406e-01 -1.22287261e+00
1.42301247e-01 -1.46237984e-01 2.14990646e-01 1.10832775e+00
4.24922764e-01 3.59961569e-01 -5.62257111e-01 7.09680557e-01
5.20953834e-01 3.72408032e-01 6.29421294e-01 -8.09767604e-01
-5.52189171e-01 -3.56450468e-01 -4.66800004e-01 -1.00659597e+00
4.48163867e-01 -7.99609125e-01 1.62353143e-01 -1.05130994e+00
-1.10574521e-01 -7.57861212e-02 -4.54081327e-01 3.69054854e-01
-7.33500645e-02 -8.06414783e-02 2.78861731e-01 4.93537366e-01
-2.06910998e-01 6.35939598e-01 1.37887585e+00 -9.31331590e-02
-4.51998413e-01 -9.42606106e-02 -2.95389503e-01 6.73740387e-01
5.31627953e-01 -1.26504198e-01 -6.43289626e-01 -3.06714803e-01
9.35207084e-02 4.92151916e-01 5.33357799e-01 -1.11176836e+00
1.42134339e-01 -4.32999164e-01 -1.08019985e-01 -2.23942876e-01
2.64076829e-01 -8.86649430e-01 2.37963974e-01 8.11793447e-01
-3.87613505e-01 1.37407795e-01 -9.54436734e-02 8.06532204e-01
-3.50760162e-01 1.88080281e-01 6.07548594e-01 2.02539518e-01
-4.23580617e-01 3.64256501e-01 -5.62653840e-01 3.87524605e-01
7.21905947e-01 7.42085055e-02 -5.95425703e-02 -6.42036080e-01
-6.13795459e-01 2.49341935e-01 2.63814300e-01 2.39879206e-01
1.34624854e-01 -1.34267485e+00 -2.00932324e-01 6.45201564e-01
-1.76663190e-01 -5.78334332e-02 1.60933405e-01 1.19126427e+00
-3.27265918e-01 9.02162135e-01 -4.86522764e-02 -7.15635836e-01
-5.65800428e-01 5.38271546e-01 7.47960865e-01 -3.58536363e-01
-9.44806695e-01 3.53766978e-02 5.29446483e-01 -6.43593371e-01
-4.18269001e-02 -6.09528482e-01 3.38174254e-01 -5.51863790e-01
3.47595245e-01 6.61811113e-01 -1.95484683e-01 -5.99006414e-01
-1.14379227e-01 3.57219458e-01 3.75078022e-01 -3.08047265e-01
9.38892066e-01 -1.79051623e-01 -9.69792381e-02 5.68051994e-01
1.38236153e+00 -1.12228438e-01 -1.76263392e+00 -2.24542245e-01
-2.32572392e-01 -3.94687384e-01 2.91829139e-01 -1.98150635e-01
-1.12024748e+00 8.32822025e-01 2.87453681e-01 2.99620330e-01
1.10606599e+00 -5.90501964e-01 8.10090601e-01 8.57638597e-01
3.35656822e-01 -8.80313337e-01 -2.52985209e-01 7.11788535e-01
8.05189252e-01 -8.90284896e-01 -4.81595770e-02 -3.99829239e-01
-2.34209076e-01 1.14745045e+00 5.05609989e-01 -4.07864273e-01
7.39626467e-01 3.62410963e-01 5.82693331e-02 1.42585590e-01
-7.35102534e-01 1.90383539e-01 4.76641059e-01 1.91326559e-01
1.45468264e-04 -9.13794264e-02 -2.55123153e-02 1.31534711e-01
-3.49550247e-02 -6.63558301e-03 7.01981187e-01 1.18032432e+00
-2.83386782e-02 -7.83475101e-01 -4.83577073e-01 3.51921380e-01
-1.14750065e-01 2.40533054e-01 -2.60555714e-01 9.03864682e-01
-3.89101774e-01 5.70677638e-01 -1.80791110e-01 -2.63241559e-01
2.88261980e-01 1.19643129e-01 1.08280979e-01 -3.70149255e-01
-1.58270210e-01 -7.52869248e-02 -2.12808058e-01 -7.96714842e-01
-2.18258813e-01 -6.57819986e-01 -1.31481600e+00 -3.05511504e-01
-4.05770838e-01 -4.03291322e-02 9.98574123e-02 1.12740588e+00
4.80808437e-01 4.14304376e-01 7.10112453e-01 -8.82960379e-01
-1.19930327e+00 -5.78854799e-01 -6.46138310e-01 4.70348626e-01
1.15704250e+00 -9.08068299e-01 -7.99614310e-01 1.70828536e-01] | [6.493125915527344, 3.452052116394043] |
635cb650-952e-46d3-a07e-437a3651297b | composition-and-deformance-measuring | 2306.03168 | null | https://arxiv.org/abs/2306.03168v1 | https://arxiv.org/pdf/2306.03168v1.pdf | Composition and Deformance: Measuring Imageability with a Text-to-Image Model | Although psycholinguists and psychologists have long studied the tendency of linguistic strings to evoke mental images in hearers or readers, most computational studies have applied this concept of imageability only to isolated words. Using recent developments in text-to-image generation models, such as DALLE mini, we propose computational methods that use generated images to measure the imageability of both single English words and connected text. We sample text prompts for image generation from three corpora: human-generated image captions, news article sentences, and poem lines. We subject these prompts to different deformances to examine the model's ability to detect changes in imageability caused by compositional change. We find high correlation between the proposed computational measures of imageability and human judgments of individual words. We also find the proposed measures more consistently respond to changes in compositionality than baseline approaches. We discuss possible effects of model training and implications for the study of compositionality in text-to-image models. | ['David A. Smith', 'Si Wu'] | 2023-06-05 | null | null | null | null | ['image-captioning'] | ['computer-vision'] | [ 7.07154274e-01 4.70763981e-01 3.79762322e-01 -4.83994395e-01
-5.79854548e-01 -7.73909926e-01 1.22018027e+00 1.68596953e-01
-5.88088036e-01 4.40427870e-01 6.93650603e-01 -3.01582158e-01
4.63268042e-01 -9.01502609e-01 -8.05383027e-01 -2.96451539e-01
4.51572210e-01 4.16157514e-01 1.10024989e-01 -2.38176629e-01
8.41164827e-01 2.71474749e-01 -1.60127664e+00 6.93147004e-01
8.39540839e-01 4.52467561e-01 5.12701809e-01 9.86133099e-01
-2.75506467e-01 1.00392485e+00 -8.71594131e-01 -5.28244674e-01
1.48629576e-01 -8.53647768e-01 -7.30504394e-01 4.06384140e-01
9.93407130e-01 -1.85674042e-01 -2.79709995e-01 1.21179056e+00
7.13121116e-01 -7.02375825e-03 1.08696818e+00 -9.18776631e-01
-1.65951180e+00 1.12577140e+00 -6.11064494e-01 6.03571057e-01
8.42908978e-01 3.17339510e-01 8.25697005e-01 -1.00755215e+00
8.84737909e-01 1.79331899e+00 2.56327063e-01 6.54565156e-01
-1.37988472e+00 -3.64881963e-01 -1.80926293e-01 9.48306620e-02
-1.08811140e+00 -7.03803957e-01 5.97414732e-01 -6.92524076e-01
1.03789139e+00 3.50878716e-01 6.24369025e-01 1.11836088e+00
2.29323879e-01 4.57871675e-01 1.65586531e+00 -1.05155563e+00
-1.26334146e-01 3.68639320e-01 -4.99431603e-02 7.54333675e-01
8.34869370e-02 7.38628432e-02 -8.69136810e-01 2.92019039e-01
8.81506920e-01 -9.01949048e-01 -2.55768865e-01 4.59690541e-01
-1.59523106e+00 8.73772800e-01 1.01234317e-01 5.04845500e-01
-2.75140762e-01 1.39600962e-01 2.88766026e-01 3.02686393e-01
7.06189394e-01 7.19330788e-01 1.38014510e-01 1.17403224e-01
-6.96574867e-01 4.82062668e-01 5.51961958e-01 7.14163542e-01
4.23183292e-01 1.34597063e-01 -4.93264288e-01 1.01446390e+00
1.86323654e-02 5.75887978e-01 9.43972349e-01 -7.26781249e-01
4.88468856e-01 2.31305227e-01 5.42130917e-02 -1.44142270e+00
-2.41950080e-01 6.57527447e-02 -2.25192934e-01 1.22948803e-01
4.31055397e-01 8.90258551e-02 -8.45215440e-01 1.89017487e+00
-1.36834249e-01 -4.25866365e-01 7.33093098e-02 8.45219314e-01
8.60135913e-01 8.80435228e-01 4.05189931e-01 -3.23752075e-01
1.79734099e+00 -7.50449419e-01 -7.80580103e-01 -3.95255923e-01
4.48908359e-01 -1.02921879e+00 1.76515007e+00 -8.38632509e-03
-1.65922868e+00 -9.40472782e-01 -1.08987856e+00 -2.33582973e-01
-3.76275152e-01 -2.81259371e-03 2.39872351e-01 7.99970627e-01
-1.32870078e+00 2.54068375e-01 -1.45104572e-01 -6.57863259e-01
7.10021779e-02 -1.10381477e-01 -6.53159097e-02 2.69266456e-01
-1.18941319e+00 1.41697943e+00 4.49657261e-01 -2.43658930e-01
-6.62846923e-01 -2.34197930e-01 -9.63775277e-01 -1.53393283e-01
-1.44744158e-01 -7.83602774e-01 1.43223786e+00 -1.60402930e+00
-1.16266274e+00 1.67963529e+00 -2.22836450e-01 -4.53461766e-01
4.32459801e-01 1.29561722e-01 -4.55462456e-01 4.73354280e-01
3.02703738e-01 1.28109157e+00 1.00117469e+00 -1.40279460e+00
-1.81056306e-01 -1.69905499e-01 -5.87695278e-03 6.20692730e-01
-3.95178825e-01 3.56866181e-01 1.93625003e-01 -9.98735547e-01
-1.05010755e-01 -6.23205245e-01 2.03409687e-01 -3.00202426e-02
-3.80179077e-01 -3.06376934e-01 2.96349138e-01 -6.81893289e-01
8.78217220e-01 -1.94847953e+00 -1.80807769e-01 -9.29463468e-03
1.84509292e-01 -1.57838225e-01 -5.14696002e-01 4.02106673e-01
-9.73663777e-02 5.32688141e-01 -6.97400272e-02 -8.31688270e-02
1.98600009e-01 3.23609635e-02 -2.21733034e-01 2.29895264e-01
1.11199498e-01 1.25851345e+00 -7.14088857e-01 -9.93104637e-01
2.06816584e-01 2.00653702e-01 -4.40915793e-01 7.62250647e-02
-2.95523822e-01 7.61896297e-02 -3.51154692e-02 1.53129473e-01
4.55979079e-01 -2.70660762e-02 -8.56987312e-02 -1.38677731e-01
-2.06171632e-01 3.57053101e-01 -2.59012222e-01 1.39888835e+00
-3.59170914e-01 1.18931770e+00 -4.85700428e-01 -5.52521586e-01
8.49454939e-01 3.63326758e-01 -4.87428784e-01 -1.15528166e+00
2.03328326e-01 -1.13595963e-01 4.20655310e-01 -9.28941846e-01
8.47424805e-01 -7.38009930e-01 -9.81872901e-02 7.11582243e-01
-2.18610808e-01 -7.26000965e-01 4.90909874e-01 2.63215572e-01
5.74481487e-01 -1.06801409e-02 3.62673014e-01 -5.58018506e-01
4.11973447e-01 5.49210794e-02 -2.64378816e-01 9.82060134e-01
-6.64244071e-02 6.25284731e-01 4.19814974e-01 -3.41564626e-01
-1.52417707e+00 -1.11092067e+00 -1.20121673e-01 1.45476568e+00
1.45855933e-01 6.00044169e-02 -1.39444542e+00 -6.14290237e-02
-5.60454309e-01 1.40144575e+00 -8.87998998e-01 -1.56034455e-01
-3.68523777e-01 -7.86406338e-01 6.53096259e-01 2.53522754e-01
3.08741897e-01 -1.88190997e+00 -9.73309457e-01 1.18588425e-01
-3.84258807e-01 -1.36584163e+00 -6.55512929e-01 -5.78807294e-01
-5.98153114e-01 -4.26024526e-01 -7.66135752e-01 -1.27782309e+00
9.73948240e-01 3.77932489e-02 1.36439788e+00 2.72703499e-01
-4.56028908e-01 5.75734615e-01 -4.73039120e-01 -5.98632991e-01
-1.38300991e+00 -3.13199461e-01 -1.03679657e-01 -2.20125362e-01
2.22037837e-01 -2.97027588e-01 -4.79336143e-01 1.42494515e-01
-1.34389067e+00 6.32353723e-01 6.03901207e-01 4.88493323e-01
-1.35351047e-02 -3.98031086e-01 4.78096008e-01 -8.00757766e-01
1.43399620e+00 7.28311890e-04 -1.53477266e-01 3.44016165e-01
-4.56185549e-01 2.04325065e-01 2.28759095e-01 -6.49693370e-01
-1.40661418e+00 -2.68437952e-01 1.45024627e-01 2.33894005e-01
-4.27045107e-01 4.33427542e-01 2.62152284e-01 6.33363277e-02
1.22241509e+00 3.85288537e-01 1.03831686e-01 2.08875477e-01
7.29366839e-01 5.46016634e-01 9.26243067e-01 -6.61083579e-01
7.03512251e-01 3.43024254e-01 -5.57538092e-01 -1.07467544e+00
-6.20982349e-01 1.40712932e-01 -4.10151690e-01 -6.58113599e-01
1.20246673e+00 -7.80248165e-01 -3.87716234e-01 5.16630054e-01
-1.55638754e+00 -3.12077045e-01 -2.59100497e-01 1.67668089e-01
-7.84048319e-01 6.14460528e-01 -8.34935904e-01 -7.30414987e-01
-4.38404560e-01 -9.78426933e-01 1.01278222e+00 1.56879693e-01
-6.68906987e-01 -1.03469002e+00 1.21484526e-01 3.86092752e-01
1.79846823e-01 1.99984446e-01 1.31023943e+00 -5.53624690e-01
1.49952038e-03 5.12013026e-02 -4.50033754e-01 1.74662694e-01
-1.19168509e-03 9.77900997e-02 -1.03568113e+00 2.01104030e-01
1.94287390e-01 -7.41798699e-01 7.46493936e-01 3.55951369e-01
6.19243622e-01 -5.34820259e-01 9.12474692e-02 -1.32684395e-01
1.30348349e+00 1.89083040e-01 1.00149107e+00 7.34869838e-02
3.87378842e-01 1.05133128e+00 1.94238961e-01 1.73418835e-01
3.30972731e-01 3.73722672e-01 -1.36491776e-01 -2.43274704e-01
-4.08330083e-01 -5.09197712e-01 6.31046116e-01 1.06003845e+00
-9.87703577e-02 -7.19235480e-01 -7.62195349e-01 6.66785240e-01
-1.25491107e+00 -1.31921816e+00 -3.02466571e-01 1.71806359e+00
9.51978683e-01 2.21610323e-01 -3.84639017e-02 1.75045859e-02
1.03100491e+00 2.55758554e-01 -8.16436261e-02 -8.27602506e-01
-5.01327455e-01 -2.07536112e-04 1.23214625e-01 6.67820752e-01
-5.56910872e-01 1.22040331e+00 7.55473757e+00 6.81827009e-01
-8.34762096e-01 1.55230328e-01 1.03921044e+00 1.99186027e-01
-6.03505850e-01 -1.84173241e-01 -1.32133618e-01 3.06380391e-01
8.59728336e-01 -3.88536841e-01 4.31233376e-01 4.25118923e-01
4.97106016e-01 -4.78740841e-01 -1.03464568e+00 8.74787927e-01
7.86040902e-01 -9.82895970e-01 5.18000484e-01 -3.50473017e-01
8.45752060e-01 -4.38049465e-01 3.13650906e-01 -1.83903769e-01
-4.69924824e-04 -1.15817571e+00 1.27396762e+00 4.78706598e-01
9.83309209e-01 -2.79868037e-01 2.55818337e-01 1.58837184e-01
-6.04567826e-01 2.58499324e-01 -6.36525750e-01 -4.25380796e-01
2.92955339e-01 1.50265321e-01 -1.22593343e+00 -3.84878099e-01
8.44046324e-02 -1.12574756e-01 -9.68610108e-01 4.28379506e-01
-2.66698092e-01 5.04547358e-01 8.52221176e-02 -4.85969156e-01
8.30640867e-02 2.59760264e-02 4.11585778e-01 1.43694496e+00
3.54132921e-01 6.13477081e-02 -3.93323779e-01 1.20576060e+00
1.13360845e-02 6.35494471e-01 -7.44740069e-01 -2.53727883e-01
3.85892481e-01 1.03594613e+00 -1.19040036e+00 -7.57196724e-01
-5.09272397e-01 1.41716075e+00 1.88519999e-01 3.04148406e-01
-7.11354315e-01 -1.66777506e-01 -1.94957495e-01 3.50256026e-01
-1.68517247e-01 -9.77839455e-02 -4.98816401e-01 -8.49768758e-01
-1.12673201e-01 -9.80902970e-01 -2.41562009e-01 -1.68510377e+00
-1.30646646e+00 8.14864576e-01 3.37202132e-01 -5.38988113e-01
-3.72171998e-01 -7.01926529e-01 -7.33340502e-01 8.42538416e-01
-8.79515648e-01 -1.06656098e+00 -3.05132061e-01 3.59652728e-01
1.08567417e+00 -1.48839146e-01 6.44451618e-01 -3.20335031e-01
7.22392425e-02 5.48195243e-01 -6.17027164e-01 7.05176294e-02
7.14100301e-01 -1.21308827e+00 9.31735575e-01 8.90490711e-01
5.03015339e-01 4.39479858e-01 1.14137876e+00 -6.51910305e-01
-7.13248849e-01 -3.85065496e-01 1.25935018e+00 -5.50826967e-01
6.13230228e-01 -4.79369640e-01 -5.62947631e-01 4.30662364e-01
1.05678678e+00 -5.87353528e-01 3.91845673e-01 -5.88489175e-01
-3.79175186e-01 5.51581264e-01 -9.47590649e-01 9.87088680e-01
1.17008948e+00 -8.08598101e-01 -1.08102274e+00 4.97702271e-01
7.12069750e-01 6.40726357e-04 -3.95208985e-01 -1.99911129e-02
6.03125572e-01 -1.10386801e+00 8.39149058e-01 -4.07009453e-01
9.05367494e-01 5.77185676e-02 7.41557777e-02 -1.41001177e+00
-4.33062792e-01 -4.71511334e-01 7.91746974e-01 9.96733069e-01
8.44369471e-01 -4.19117421e-01 2.02388361e-01 6.74020529e-01
-8.02537277e-02 1.70811951e-01 -5.94105422e-01 -3.13605338e-01
2.16338322e-01 -2.91546941e-01 1.77865744e-01 9.27239537e-01
2.66033620e-01 8.47002804e-01 -2.39968896e-01 -2.73217589e-01
2.85913736e-01 -2.80994147e-01 4.76336062e-01 -9.57589805e-01
-3.13224226e-01 -7.24393249e-01 -3.16513032e-01 -7.53623247e-01
3.40991110e-01 -9.60814953e-01 3.19288433e-01 -1.53097725e+00
6.34133875e-01 3.17964762e-01 3.48420918e-01 7.62879103e-02
-4.50941443e-01 6.20354891e-01 5.45950770e-01 2.14621097e-01
-2.39837378e-01 9.99703780e-02 1.71162391e+00 8.18659216e-02
2.40108427e-02 -7.62667954e-01 -8.53154421e-01 7.62659848e-01
8.56161833e-01 -2.50620097e-01 -4.71213132e-01 -5.62456369e-01
6.07298434e-01 -7.60007948e-02 5.31847239e-01 -9.15972114e-01
-1.36972725e-01 -1.25259027e-01 4.67386395e-01 -9.71741825e-02
8.80309497e-04 -2.47072399e-01 -1.71271160e-01 3.37603211e-01
-9.72504616e-01 7.77153075e-01 2.84760207e-01 2.80279696e-01
-6.20400570e-02 -6.40662551e-01 8.20822179e-01 -5.24950683e-01
-5.69599032e-01 -3.94892693e-01 -1.03265440e+00 2.86443383e-01
7.86146045e-01 -5.02430975e-01 -4.71269637e-01 -8.37430596e-01
-6.70948029e-01 -4.01312977e-01 5.61425149e-01 4.88949418e-01
7.18278527e-01 -1.18081987e+00 -1.19603539e+00 -1.20081536e-01
1.29223078e-01 -7.43084073e-01 1.05379201e-01 5.16868889e-01
-9.10073042e-01 1.53462082e-01 -3.95621687e-01 -2.03468665e-01
-1.23676717e+00 8.29980373e-01 1.03407010e-01 2.79535621e-01
-2.86809206e-01 8.89265954e-01 7.71937966e-01 2.82607060e-02
-2.31726050e-01 -3.88535082e-01 -9.82269794e-02 1.92454964e-01
6.07291102e-01 1.00303747e-01 -5.35841286e-01 -9.28928494e-01
2.71134734e-01 6.36660576e-01 -2.49285139e-02 -9.89408851e-01
6.46022677e-01 -3.83022100e-01 -2.16938078e-01 7.33416378e-01
9.46190357e-01 7.15230256e-02 -7.16657758e-01 1.06985047e-01
-1.19060487e-01 -2.36055642e-01 -3.42331558e-01 -8.99168849e-01
-4.48747963e-01 8.63806546e-01 5.25405943e-01 5.07672966e-01
9.21674490e-01 3.34063858e-01 4.61318493e-01 2.25482225e-01
-1.61849454e-01 -1.26831925e+00 4.76530433e-01 1.43412367e-01
1.35344529e+00 -1.00900543e+00 -1.44080520e-01 -3.09062809e-01
-9.97981191e-01 9.58509207e-01 6.19776130e-01 1.02541819e-02
-1.27619743e-01 1.33617610e-01 4.36752439e-01 -2.98262537e-01
-7.80965805e-01 -9.48474705e-02 3.23135644e-01 7.62803495e-01
8.54780734e-01 8.09421688e-02 -7.51144767e-01 2.44326666e-02
-8.65394652e-01 -4.20397401e-01 8.39022934e-01 5.01803219e-01
-7.08267927e-01 -6.08332157e-01 -7.11403370e-01 4.86145973e-01
-4.88597721e-01 -6.55324936e-01 -8.42308223e-01 6.70060039e-01
1.24612898e-01 9.15226400e-01 5.01329601e-01 -7.95848295e-02
8.80916491e-02 3.39062184e-01 7.92590439e-01 -6.54866576e-01
-5.61655223e-01 1.80182174e-01 3.90054494e-01 2.18369871e-01
-7.60780156e-01 -6.41196966e-01 -1.08276868e+00 -1.77496195e-01
-2.43199706e-01 -1.31150633e-01 6.38091683e-01 8.74450088e-01
-6.73247650e-02 3.29216033e-01 7.16670379e-02 -7.97738194e-01
-8.67624804e-02 -1.57057369e+00 -4.64373201e-01 9.76425111e-01
-1.81654334e-01 -2.28472605e-01 -4.56296802e-01 6.76786184e-01] | [11.195024490356445, 1.1612523794174194] |
ea85cf75-f667-40e6-a74d-7119e79da2c6 | multi-camera-calibration-free-bev | 2210.17252 | null | https://arxiv.org/abs/2210.17252v1 | https://arxiv.org/pdf/2210.17252v1.pdf | Multi-Camera Calibration Free BEV Representation for 3D Object Detection | In advanced paradigms of autonomous driving, learning Bird's Eye View (BEV) representation from surrounding views is crucial for multi-task framework. However, existing methods based on depth estimation or camera-driven attention are not stable to obtain transformation under noisy camera parameters, mainly with two challenges, accurate depth prediction and calibration. In this work, we present a completely Multi-Camera Calibration Free Transformer (CFT) for robust BEV representation, which focuses on exploring implicit mapping, not relied on camera intrinsics and extrinsics. To guide better feature learning from image views to BEV, CFT mines potential 3D information in BEV via our designed position-aware enhancement (PA). Instead of camera-driven point-wise or global transformation, for interaction within more effective region and lower computation cost, we propose a view-aware attention which also reduces redundant computation and promotes converge. CFT achieves 49.7% NDS on the nuScenes detection task leaderboard, which is the first work removing camera parameters, comparable to other geometry-guided methods. Without temporal input and other modal information, CFT achieves second highest performance with a smaller image input 1600 * 640. Thanks to view-attention variant, CFT reduces memory and transformer FLOPs for vanilla attention by about 12% and 60%, respectively, with improved NDS by 1.0%. Moreover, its natural robustness to noisy camera parameters makes CFT more competitive. | ['Jihao Yin', 'Qian Zhang', 'Hongmei Zhu', 'Wenming Meng', 'Hongxiang Jiang'] | 2022-10-31 | null | null | null | null | ['camera-calibration'] | ['computer-vision'] | [ 9.51612815e-02 -9.15054753e-02 9.11221188e-03 -4.02491540e-01
-7.72189736e-01 -6.79849029e-01 5.53514063e-01 -3.88838232e-01
-6.49831951e-01 4.45522428e-01 4.43859957e-02 -1.71014041e-01
1.84622258e-02 -7.48968363e-01 -1.12893331e+00 -6.00665510e-01
7.53614187e-01 8.69607553e-02 5.54051459e-01 -4.06771034e-01
5.17157316e-01 5.65960228e-01 -1.84113002e+00 4.27590385e-02
7.97163963e-01 9.96516109e-01 6.48387730e-01 7.29833901e-01
4.96746935e-02 5.87670088e-01 -2.04027459e-01 -3.80009264e-01
4.02792215e-01 4.28058244e-02 -3.51213902e-01 7.86417425e-02
8.47630262e-01 -6.46751404e-01 -4.35813457e-01 9.15753126e-01
5.82122982e-01 2.49554947e-01 5.81160605e-01 -1.34371567e+00
-4.50575382e-01 -6.50599599e-02 -9.57465470e-01 8.56487900e-02
2.30267808e-01 2.94003665e-01 6.39424086e-01 -1.21212006e+00
5.34743428e-01 1.22503221e+00 4.69219416e-01 5.92034996e-01
-8.85605752e-01 -7.10773587e-01 4.47579801e-01 3.75143528e-01
-1.34707797e+00 -4.61531907e-01 8.95971298e-01 -2.35583052e-01
1.07444727e+00 2.34159946e-01 6.54553115e-01 1.00698364e+00
3.05379868e-01 5.11352360e-01 1.08779705e+00 -1.78311273e-01
-1.05619930e-01 4.55504447e-01 -1.22548034e-02 6.47302747e-01
1.50327891e-01 3.81944776e-01 -5.88609278e-01 5.28314173e-01
8.82813096e-01 1.79032594e-01 -4.70516413e-01 -6.74590349e-01
-1.39751172e+00 6.82532012e-01 4.87591743e-01 -2.53624678e-01
-9.00379717e-02 1.84902281e-01 1.43515199e-01 8.17895681e-02
2.71188736e-01 2.40565047e-01 -5.34342051e-01 -1.59574404e-01
-5.08030891e-01 5.83865047e-02 2.39084318e-01 1.41458738e+00
1.08809936e+00 1.42428443e-01 -7.15130866e-02 5.60311913e-01
2.81226158e-01 1.08647192e+00 3.95501465e-01 -1.20185578e+00
7.13161111e-01 6.19350851e-01 1.56331182e-01 -1.11852002e+00
-5.47049940e-01 -3.79347563e-01 -8.47983956e-01 4.86561805e-01
2.49248877e-01 -7.71926939e-02 -8.85150611e-01 1.58664227e+00
4.41543847e-01 -7.62930140e-02 1.30773053e-01 1.12047231e+00
9.53380167e-01 5.29948056e-01 -5.84114194e-01 -8.34234804e-02
1.37308395e+00 -1.26141798e+00 -6.66643620e-01 -6.70226276e-01
5.91642141e-01 -7.26007223e-01 1.12230575e+00 5.91216564e-01
-9.05338407e-01 -7.82171667e-01 -1.14742827e+00 -6.30845249e-01
-3.66768539e-01 2.42098600e-01 5.08745551e-01 6.06053650e-01
-1.18080091e+00 1.93121582e-02 -6.65437877e-01 -2.91803390e-01
2.26732582e-01 3.51266652e-01 -6.18507385e-01 -2.58318275e-01
-8.66683662e-01 9.29561138e-01 2.24257976e-01 1.12450957e-01
-1.06846333e+00 -7.95251548e-01 -1.03466654e+00 -4.26355004e-02
6.50610268e-01 -8.92271280e-01 7.92798758e-01 -5.85561872e-01
-1.71260107e+00 6.60119891e-01 -5.31417191e-01 -3.24799567e-01
7.22584307e-01 -5.59290111e-01 -1.84175596e-01 8.88447240e-02
5.69679635e-03 9.15133536e-01 1.09783316e+00 -1.39283466e+00
-7.59467423e-01 -4.82548624e-01 2.31142163e-01 5.52012384e-01
-3.95307481e-01 -4.70368832e-01 -8.12772393e-01 -2.32343242e-01
4.09380645e-01 -1.02219343e+00 -1.76734880e-01 3.02703053e-01
1.42465374e-02 1.05036676e-01 1.01810932e+00 -4.47331667e-01
9.28778410e-01 -2.20727539e+00 2.12046877e-01 -2.02963740e-01
4.20983940e-01 1.20101772e-01 -6.89299777e-02 2.63535623e-02
7.78032094e-02 -2.39550024e-01 -1.75252762e-02 -4.80940521e-01
-1.73157886e-01 1.36588290e-01 -2.21433774e-01 5.34152925e-01
2.22028613e-01 9.96039391e-01 -5.97747207e-01 -3.57650280e-01
8.08531880e-01 5.75996399e-01 -8.46387506e-01 1.10745482e-01
1.81894928e-01 4.90744412e-01 -2.68563896e-01 3.98031026e-01
1.09546411e+00 6.84514735e-03 -4.02008235e-01 -5.85524678e-01
-3.54353487e-01 -3.50850493e-01 -1.20638478e+00 1.97822499e+00
-8.61661494e-01 6.48194730e-01 2.30594411e-01 -7.40161359e-01
1.16884422e+00 -2.17385590e-01 1.43901870e-01 -8.80977094e-01
2.91518897e-01 4.41507325e-02 -3.86148959e-01 -4.22781199e-01
7.46157467e-01 3.23750168e-01 -5.42076072e-03 1.33896887e-01
-7.68622160e-02 -5.67927778e-01 -2.06482142e-01 2.22013667e-01
6.99757755e-01 3.72208625e-01 2.13155225e-01 -1.50746241e-01
8.11907053e-01 -1.97882667e-01 5.07053494e-01 4.45249647e-01
-2.35645652e-01 8.91580105e-01 6.42094687e-02 -4.20294613e-01
-1.00132000e+00 -9.09132421e-01 -1.10858932e-01 8.17686141e-01
7.44752407e-01 -1.82756186e-01 -6.10279381e-01 -4.93122965e-01
-1.25850812e-01 5.40708065e-01 -4.82765734e-01 -1.74283460e-01
-6.86485171e-01 -4.17819142e-01 4.10858579e-02 5.55089772e-01
8.12580228e-01 -5.49685359e-01 -8.71965766e-01 -5.34810387e-02
-2.81664968e-01 -1.45714974e+00 -7.48297811e-01 1.84609950e-01
-8.73646915e-01 -1.00056934e+00 -7.82168508e-01 -4.58312154e-01
6.00211501e-01 9.33023214e-01 8.48007619e-01 -3.92399818e-01
4.73475307e-02 3.39976251e-01 -2.34443218e-01 -2.98711449e-01
1.41524985e-01 1.41572118e-01 5.62191904e-02 2.11743549e-01
2.40484640e-01 -5.51105440e-01 -8.16448331e-01 5.38235009e-01
-5.81072390e-01 3.91357005e-01 8.41695249e-01 7.60222852e-01
6.07535005e-01 -3.66118193e-01 9.91629362e-02 -6.61215603e-01
-1.42603040e-01 -1.08966917e-01 -8.12099755e-01 -2.44054236e-02
-6.95864499e-01 1.44888639e-01 5.14990270e-01 -2.92651027e-01
-1.26794040e+00 3.89619738e-01 -1.86175015e-02 -9.17409241e-01
-9.89533514e-02 -3.24992120e-01 -3.69756460e-01 -3.75033498e-01
6.12787545e-01 4.95923668e-01 1.08395629e-01 -1.80514395e-01
5.08104146e-01 5.65190554e-01 6.15341485e-01 -1.84667379e-01
8.67579520e-01 8.05703282e-01 1.09817818e-01 -7.18119264e-01
-7.84067690e-01 -5.05768239e-01 -8.57248127e-01 -2.63589293e-01
1.19701624e+00 -1.29573417e+00 -9.60274577e-01 4.46552306e-01
-1.16676092e+00 -1.56768009e-01 9.79585499e-02 6.99907541e-01
-6.01549506e-01 4.66654122e-01 -1.26476929e-01 -6.82837605e-01
-3.08571786e-01 -1.39915991e+00 1.44406939e+00 7.41052702e-02
3.47504407e-01 -6.73405886e-01 -3.98146331e-01 7.38462508e-01
4.17097956e-01 1.27244085e-01 3.35060358e-01 1.49091989e-01
-1.00371587e+00 1.10997275e-01 -5.80074191e-01 4.14884537e-01
-5.28409146e-02 -2.01079458e-01 -1.34374797e+00 -4.59260166e-01
2.26408243e-01 -8.59573409e-02 9.05789971e-01 5.55951297e-01
1.00483668e+00 7.08101466e-02 -3.55552137e-01 1.37596381e+00
1.63359225e+00 2.48550698e-01 5.18617570e-01 4.51157510e-01
1.12714505e+00 5.98485649e-01 8.98502886e-01 4.20181483e-01
8.37464988e-01 7.63980269e-01 9.04344559e-01 -3.05287838e-01
-2.70752370e-01 -2.00422898e-01 5.92332721e-01 7.07583666e-01
-1.12673111e-01 -2.54350275e-01 -6.95643187e-01 3.20950091e-01
-1.67089307e+00 -6.13950551e-01 -3.36860955e-01 2.11969233e+00
2.53413230e-01 2.64898688e-01 -1.67308837e-01 -6.63309619e-02
5.93939900e-01 1.63624629e-01 -8.25313926e-01 -1.70184582e-01
-2.32748300e-01 -1.65575415e-01 1.01365829e+00 6.40451610e-01
-9.32873070e-01 9.95474458e-01 5.21628094e+00 8.02862585e-01
-1.08571315e+00 1.75403729e-01 4.45557982e-01 -3.61300498e-01
-3.16890538e-01 -8.64308849e-02 -1.22147644e+00 2.87416428e-01
5.64025044e-01 1.23946756e-01 4.09963459e-01 9.15489852e-01
2.85841078e-01 -1.61774695e-01 -1.05008948e+00 1.55533409e+00
4.50648904e-01 -1.06259739e+00 1.82326734e-02 2.30289102e-01
7.57047892e-01 2.27414072e-01 1.49552912e-01 1.75781444e-01
2.30230158e-03 -7.36000955e-01 9.51278031e-01 5.06707728e-01
9.85247612e-01 -8.57460916e-01 8.05923879e-01 3.79691273e-01
-1.35410988e+00 -3.32469910e-01 -5.86430132e-01 -5.47565296e-02
3.27253997e-01 3.79437536e-01 -2.85694838e-01 7.54459023e-01
9.06802833e-01 1.05320740e+00 -7.88506329e-01 4.90489095e-01
2.01594643e-02 -1.06199287e-01 -4.07291323e-01 1.47838950e-01
2.47473702e-01 -3.04663748e-01 5.32059968e-01 7.58642018e-01
4.49537158e-01 1.11096382e-01 -1.01076849e-02 7.80516326e-01
2.74460971e-01 -1.17176279e-01 -7.76462018e-01 7.10280299e-01
3.89292538e-01 1.41228628e+00 -5.42199612e-01 -1.99669585e-01
-5.93082905e-01 1.17496216e+00 2.87941903e-01 3.89452666e-01
-1.18692112e+00 -3.15413952e-01 6.47598445e-01 2.73726761e-01
4.56531197e-01 -3.33975166e-01 -1.67809814e-01 -1.26714003e+00
2.62507111e-01 -6.43089771e-01 -6.90080673e-02 -1.03084493e+00
-6.51035488e-01 8.38683903e-01 -5.96196763e-02 -1.65463400e+00
3.57273482e-02 -6.73816144e-01 -2.69245863e-01 7.78297782e-01
-1.86866069e+00 -1.39283586e+00 -9.76412952e-01 8.10906112e-01
7.71480739e-01 -4.45590131e-02 3.16618234e-01 3.75474155e-01
-4.76509541e-01 7.24219561e-01 -7.91123807e-02 -3.68989438e-01
1.02394581e+00 -1.04949427e+00 4.28879410e-01 8.99689972e-01
-1.16558686e-01 2.52907276e-01 3.56495142e-01 -1.99986830e-01
-1.74254882e+00 -1.18851459e+00 3.49700630e-01 -9.86485481e-01
1.51150838e-01 -5.68967998e-01 -7.18666911e-01 5.35883486e-01
7.81134516e-02 2.04358533e-01 -7.57789090e-02 -3.72584879e-01
-2.86943465e-01 -5.20785213e-01 -8.13301921e-01 4.57632124e-01
1.36402178e+00 -4.41096187e-01 -2.18177766e-01 9.94057655e-02
9.49035466e-01 -6.90042019e-01 -7.40380526e-01 3.11328292e-01
5.67481875e-01 -1.34296262e+00 1.11379707e+00 2.23380446e-01
2.00380757e-01 -6.18940055e-01 -2.84056336e-01 -9.69632208e-01
-2.75177300e-01 -5.16523659e-01 9.84989330e-02 1.16816890e+00
1.66420326e-01 -8.00429940e-01 8.06557119e-01 2.93357939e-01
-5.54307044e-01 -5.60476899e-01 -8.30160201e-01 -5.68482041e-01
-1.06949493e-01 -5.76282680e-01 5.26151597e-01 7.20658898e-01
-6.66556239e-01 6.72657967e-01 -6.03724122e-01 3.97414953e-01
5.53595841e-01 -9.00426228e-03 1.31472361e+00 -1.00083601e+00
-2.41251096e-01 -2.82506168e-01 -4.10873920e-01 -1.52736628e+00
-9.13080350e-02 -5.76991022e-01 -3.93918343e-02 -1.32390654e+00
8.60531107e-02 -1.28964961e-01 1.80066288e-01 1.07700191e-01
-1.58425435e-01 3.38616759e-01 3.16988945e-01 8.24278127e-03
-4.14211750e-01 8.13351452e-01 1.62181771e+00 -5.51652089e-02
-2.11675778e-01 -2.15325177e-01 -5.28555572e-01 7.47655213e-01
5.23854196e-01 -2.02697024e-01 -7.62148201e-01 -8.39068770e-01
1.96396992e-01 1.91384450e-01 4.20744181e-01 -1.14211619e+00
4.32261884e-01 -8.25761631e-02 3.62610072e-01 -9.79411781e-01
7.76110053e-01 -9.51591194e-01 1.81436479e-01 2.68927157e-01
2.38478526e-01 2.75552183e-01 2.79687524e-01 8.18212390e-01
-3.06182355e-01 -4.53851977e-03 8.27600837e-01 -1.00762315e-01
-1.11533797e+00 4.53644365e-01 -4.20676619e-02 7.53760571e-03
1.21767223e+00 -6.83207512e-01 -3.33014965e-01 -4.49174494e-01
-4.47706610e-01 3.17439258e-01 6.33320391e-01 5.16886890e-01
8.10777903e-01 -1.19565666e+00 -5.70437074e-01 7.03140497e-01
2.73159117e-01 5.30342996e-01 7.66096532e-01 9.39102530e-01
-6.53918266e-01 5.63064754e-01 -2.04015091e-01 -9.64843929e-01
-1.34223449e+00 7.81914294e-01 3.10188413e-01 1.62966266e-01
-7.75788128e-01 9.11468804e-01 9.24211442e-01 -4.64989156e-01
-2.29386501e-02 -5.68873346e-01 -2.39270568e-01 -5.94709925e-02
4.67570335e-01 4.43326890e-01 1.43367618e-01 -8.20886016e-01
-3.87386441e-01 1.43881190e+00 -7.61052668e-02 -4.04524505e-02
1.03706169e+00 -6.66817844e-01 2.91571945e-01 1.82032540e-01
1.25828385e+00 -3.45601700e-02 -1.80807865e+00 -8.25743973e-02
-6.56741738e-01 -6.77913189e-01 4.79804665e-01 -3.96999240e-01
-1.31022775e+00 1.12919712e+00 8.27891707e-01 -3.71286064e-01
1.17832911e+00 -2.45718002e-01 7.88660467e-01 3.39116961e-01
4.74019706e-01 -8.45109224e-01 1.69439524e-01 4.94634122e-01
9.61640537e-01 -1.67795765e+00 -3.69245633e-02 -6.13450408e-01
-8.75408828e-01 9.88469958e-01 1.13486993e+00 -3.40078250e-02
6.17847145e-01 1.63549423e-01 7.27278441e-02 -1.16942711e-01
-8.03125918e-01 -1.66022375e-01 3.15752536e-01 8.33573401e-01
-1.32298604e-01 -3.89640927e-01 3.17560285e-01 4.03558552e-01
-2.78104246e-01 -3.65534782e-01 5.79954147e-01 4.74055678e-01
-3.87176603e-01 -6.65768564e-01 -3.05890083e-01 1.87802002e-01
2.98773441e-02 -1.66711092e-01 7.91667253e-02 9.07913804e-01
2.06081882e-01 9.32978809e-01 2.26643875e-01 -5.55395007e-01
6.45334303e-01 -3.74246955e-01 4.38128531e-01 -4.14825439e-01
-2.96221137e-01 1.36979192e-01 -2.82859981e-01 -8.96710992e-01
-4.17247474e-01 -4.12301958e-01 -9.00347710e-01 -3.54080886e-01
-5.50893843e-01 -3.25226754e-01 5.60257137e-01 6.94841325e-01
4.88581210e-01 5.90438843e-01 7.63476610e-01 -1.13564730e+00
-2.78058708e-01 -8.27238142e-01 -3.82312030e-01 2.04327062e-01
5.91971457e-01 -9.20497656e-01 -5.84582448e-01 -5.14006764e-02] | [8.36109447479248, -2.3078646659851074] |
850f3ea2-fb61-4a00-bc03-596fdb1ebfb1 | probabilistic-surface-friction-estimation | 2010.08277 | null | https://arxiv.org/abs/2010.08277v3 | https://arxiv.org/pdf/2010.08277v3.pdf | Probabilistic Surface Friction Estimation Based on Visual and Haptic Measurements | Accurately modeling local surface properties of objects is crucial to many robotic applications, from grasping to material recognition. Surface properties like friction are however difficult to estimate, as visual observation of the object does not convey enough information over these properties. In contrast, haptic exploration is time consuming as it only provides information relevant to the explored parts of the object. In this work, we propose a joint visuo-haptic object model that enables the estimation of surface friction coefficient over an entire object by exploiting the correlation of visual and haptic information, together with a limited haptic exploration by a robotic arm. We demonstrate the validity of the proposed method by showing its ability to estimate varying friction coefficients on a range of real multi-material objects. Furthermore, we illustrate how the estimated friction coefficients can improve grasping success rate by guiding a grasp planner toward high friction areas. | ['Ville Kyrki', 'Fares J. Abu-Dakka', 'Francesco Verdoja', 'Tran Nguyen Le'] | 2020-10-16 | null | null | null | null | ['material-recognition'] | ['computer-vision'] | [ 8.54072198e-02 3.88312489e-02 -2.14587003e-01 1.30736992e-01
-2.29407489e-01 -6.23009980e-01 2.47803286e-01 4.91690576e-01
-1.80606335e-01 6.94795728e-01 -2.77655810e-01 -2.03033700e-03
-5.16714156e-01 -7.72898674e-01 -7.76545107e-01 -5.64503491e-01
-4.25779432e-01 4.75329459e-01 4.33314800e-01 -2.51473427e-01
5.29125988e-01 9.45298731e-01 -1.94774950e+00 3.67935807e-01
8.61318886e-01 1.22568512e+00 1.16972268e+00 4.49154288e-01
-1.24311976e-01 -5.56724332e-02 -3.25414181e-01 1.48812369e-01
2.52197653e-01 3.57805461e-01 -6.30621612e-01 2.76961364e-02
-1.01098001e-01 -7.39856541e-01 2.20348239e-01 8.46787035e-01
5.30417077e-02 1.39927775e-01 8.89838696e-01 -9.24731314e-01
-3.15167010e-01 2.48373702e-01 -4.43928063e-01 -5.87513745e-01
6.58246636e-01 -5.23084328e-02 9.05167997e-01 -7.49811769e-01
1.04527891e+00 1.21148360e+00 1.97597116e-01 5.49458027e-01
-1.00310004e+00 -5.68638407e-02 2.00020254e-01 5.13268188e-02
-9.75860476e-01 4.80549335e-02 9.90088642e-01 -5.21339297e-01
6.39482617e-01 4.35922533e-01 7.66053498e-01 6.58461928e-01
3.17297459e-01 7.41378486e-01 1.17621768e+00 -5.01803279e-01
4.15672034e-01 3.37709218e-01 -2.67726243e-01 4.48231876e-01
3.31754416e-01 1.32317960e-01 -5.73229611e-01 -2.33407408e-01
1.19642556e+00 -6.04970902e-02 -5.28774559e-01 -1.26438141e+00
-1.21012270e+00 3.27278137e-01 6.25330091e-01 1.13234773e-01
-5.91232955e-01 2.25492105e-01 2.83509940e-01 1.91380218e-01
2.21745700e-01 7.78557420e-01 -3.86430949e-01 -4.53224242e-01
-1.54757768e-01 3.77499282e-01 9.56874073e-01 1.10155535e+00
6.99905396e-01 -5.62995017e-01 9.25843641e-02 6.57820344e-01
4.78757650e-01 6.80313766e-01 -3.78270775e-01 -1.06283975e+00
3.70810717e-01 5.30014396e-01 1.02954757e+00 -8.90468836e-01
-4.23617482e-01 4.00587529e-01 -2.42256775e-01 9.63927150e-01
5.24026990e-01 1.41575247e-01 -6.51827216e-01 1.24669862e+00
5.06400824e-01 -8.93910289e-01 -3.18038017e-01 1.34123623e+00
4.03352022e-01 1.43833712e-01 1.70673341e-01 -1.51183251e-02
1.18978369e+00 -5.14274597e-01 -8.11556041e-01 -1.56565428e-01
4.21693265e-01 -6.80733085e-01 1.15148723e+00 6.16330147e-01
-1.17412031e+00 -4.06118333e-01 -9.04529691e-01 1.18869670e-01
-3.59297097e-01 3.52309495e-01 1.01480651e+00 2.88067430e-01
-6.93680882e-01 1.06208992e+00 -9.61365283e-01 -2.82171071e-01
2.84988552e-01 5.50205648e-01 -3.34181905e-01 -3.78048681e-02
-7.55555272e-01 1.15600681e+00 4.35840994e-01 3.29847693e-01
-3.15496355e-01 -4.75972116e-01 -5.67212760e-01 2.21508592e-02
4.36586469e-01 -4.00181383e-01 9.77719426e-01 -4.58916545e-01
-1.83220232e+00 4.52334911e-01 4.32552472e-02 2.34720170e-01
1.06507587e+00 -6.74047232e-01 4.63245600e-01 4.67380226e-01
-3.06247324e-01 7.77339160e-01 8.66440117e-01 -1.92021370e+00
-9.11562592e-02 -3.07264924e-01 2.82591701e-01 1.96761563e-01
-1.86303511e-01 -5.33741117e-01 -1.47154436e-01 -2.92052895e-01
3.90258461e-01 -7.14509904e-01 -1.55620709e-01 1.03177953e+00
-3.55905503e-01 -1.79598987e-01 9.82741654e-01 -5.02743840e-01
4.79900807e-01 -2.06217313e+00 3.77764165e-01 5.41422427e-01
-6.66734055e-02 -3.17692198e-02 2.90415995e-02 9.19494152e-01
5.44821620e-01 -9.24406201e-02 9.03590620e-02 4.71010283e-02
8.61582011e-02 7.09616521e-04 -2.86540866e-01 2.32461691e-01
1.40446171e-01 9.12226498e-01 -1.05204642e+00 -3.06866676e-01
3.95041198e-01 5.47686040e-01 -3.88967633e-01 3.45307022e-01
-4.27690566e-01 4.75763470e-01 -1.02199125e+00 8.79732072e-01
9.19627309e-01 2.50243768e-02 3.97102803e-01 -4.81392592e-01
-1.91019550e-01 7.84414820e-03 -1.10566974e+00 1.68231273e+00
-7.34554291e-01 2.47878760e-01 6.93305671e-01 -6.00052416e-01
1.19869697e+00 1.06143311e-01 3.93968225e-01 -4.91519541e-01
1.29551381e-01 4.91974235e-01 -1.96871459e-01 -7.30953813e-01
4.33975041e-01 2.17223447e-03 3.82897705e-01 3.89938414e-01
-4.64020610e-01 -7.58202195e-01 -1.65977851e-01 -2.08057880e-01
6.49748504e-01 6.65231824e-01 1.02859795e-01 -5.08795261e-01
-1.00443169e-01 -3.23083997e-02 -3.02173942e-01 7.65108168e-01
1.92283586e-01 2.57979393e-01 2.20983133e-01 -2.01900408e-01
-1.18223202e+00 -1.35514498e+00 -3.03548634e-01 6.31416678e-01
9.50769007e-01 -4.71874811e-02 -4.48221207e-01 -2.82103211e-01
7.49228001e-01 3.46587718e-01 -7.47406721e-01 3.71633917e-02
-4.44189876e-01 1.04794063e-01 -6.38784051e-01 6.81604505e-01
1.32474869e-01 -1.38956857e+00 -1.53814149e+00 2.21313477e-01
2.34978506e-03 -9.80625749e-01 1.21633202e-01 6.64422065e-02
-1.17349041e+00 -1.02530026e+00 -7.65517890e-01 -4.79709297e-01
8.81806970e-01 4.22926754e-01 7.41179943e-01 3.25509638e-01
-6.87171638e-01 5.66063702e-01 -4.86055374e-01 -2.96728224e-01
-4.34688449e-01 -2.15091094e-01 -1.00153461e-01 -4.60495144e-01
-2.86847830e-01 -4.04065520e-01 -7.24229276e-01 5.42904675e-01
-6.35643899e-01 4.61771607e-01 5.63474596e-01 5.57069957e-01
4.38951105e-01 -4.25530344e-01 5.30959487e-01 -4.50158864e-01
6.47306323e-01 -1.77661777e-01 -5.42657077e-01 2.97235787e-01
-1.54565852e-02 -6.86560795e-02 2.22352058e-01 -1.01816201e+00
-1.08480513e+00 -4.47796285e-03 3.10200304e-01 -3.81905586e-01
-5.25735766e-02 4.56042320e-01 -5.54347737e-03 -2.69206256e-01
3.35825950e-01 -3.62201184e-01 1.83436006e-01 -6.30636692e-01
2.67879069e-01 5.49634099e-01 1.25524536e-01 -8.55216026e-01
4.97074425e-01 7.00231850e-01 3.29574347e-01 -9.08565402e-01
-1.47779956e-01 -3.76877099e-01 -8.56726348e-01 -5.24687290e-01
3.59030932e-01 -2.30570555e-01 -1.55407226e+00 1.86006710e-01
-1.17856681e+00 -6.51886404e-01 -2.94077247e-01 6.97324395e-01
-1.00344002e+00 4.07190472e-01 -5.51339328e-01 -1.36337209e+00
-1.09677598e-01 -1.28112376e+00 1.44461787e+00 -4.05978598e-02
-3.68848234e-01 -7.73836136e-01 -7.17765331e-01 -1.03565836e-02
5.64760804e-01 3.30044717e-01 8.37480187e-01 3.10955048e-01
-1.04685998e+00 -1.23860866e-01 -4.21330988e-01 -1.79895282e-01
5.53056359e-01 -1.25838503e-01 -8.39094341e-01 -4.39518273e-01
-2.83569753e-01 -5.52161515e-01 6.72744751e-01 2.56776422e-01
1.21877205e+00 1.76108032e-01 -7.44069934e-01 2.24273894e-02
1.36689305e+00 1.77488655e-01 4.36999351e-01 3.32535565e-01
5.82029760e-01 1.00477886e+00 1.35577524e+00 6.87457860e-01
-2.52348930e-01 9.76140559e-01 9.12111282e-01 2.25001611e-02
3.85228135e-02 5.13866059e-02 -6.68316782e-02 1.67351708e-01
-6.17811382e-01 -5.11955209e-02 -5.39639950e-01 3.39842886e-01
-1.51232219e+00 -2.19502106e-01 9.25964396e-03 2.41275954e+00
6.53731883e-01 1.42921343e-01 -1.54105261e-01 3.26012932e-02
3.59080553e-01 -4.02671754e-01 -6.87067330e-01 -3.79310668e-01
3.64626288e-01 -9.91349965e-02 3.90070200e-01 5.09882152e-01
-5.84992647e-01 6.99450254e-01 6.05083084e+00 6.07005298e-01
-1.19897425e+00 -4.35839444e-01 -2.04037875e-01 2.58821249e-01
-6.68790340e-01 -1.08705737e-01 -2.71924585e-01 2.06984743e-01
-3.82035598e-02 3.07029992e-01 5.20406604e-01 9.23749208e-01
8.03237483e-02 -8.96058202e-01 -1.24878061e+00 4.89647508e-01
-4.77386624e-01 -9.77310061e-01 4.31835651e-02 -4.90806140e-02
2.23242611e-01 -4.22835022e-01 -4.85008359e-02 -4.08819377e-01
-2.31763050e-01 -6.70175135e-01 8.25178683e-01 7.69627035e-01
8.72859359e-01 -3.59080046e-01 4.56352651e-01 4.73372400e-01
-1.01544023e+00 -6.29126206e-02 -3.13892901e-01 -1.10923685e-01
2.87131846e-01 3.98640990e-01 -9.69570398e-01 4.76845741e-01
8.16905499e-01 3.95885929e-02 7.09556788e-02 1.20597744e+00
1.89325530e-02 -2.12046519e-01 -5.04756987e-01 -4.66451645e-01
-1.45309404e-01 -1.33270890e-01 6.88294113e-01 7.77329862e-01
3.83009225e-01 1.07749358e-01 1.54889107e-01 1.22836280e+00
4.57712561e-01 1.79025933e-01 -6.65430188e-01 -3.81651595e-02
4.30765718e-01 1.09800899e+00 -9.82328653e-01 5.11143059e-02
-3.65565233e-02 8.94706964e-01 4.53806669e-01 4.86552536e-01
-1.82393029e-01 -3.45637739e-01 3.88371080e-01 2.50861526e-01
4.07784104e-01 -7.22340047e-01 -3.27176034e-01 -8.06028426e-01
4.30863351e-01 -1.05955012e-01 -5.23434818e-01 -1.13371933e+00
-9.86421585e-01 2.09676549e-01 1.56634986e-01 -1.35865378e+00
3.88268866e-02 -1.08824730e+00 -1.47472590e-01 1.02649724e+00
-1.49569988e+00 -1.15254557e+00 -5.98698080e-01 -8.10160637e-02
3.62132788e-01 5.07838011e-01 8.98639619e-01 -4.18305278e-01
5.46944141e-01 1.16488179e-02 1.93711463e-02 -5.75513363e-01
5.30398607e-01 -1.13043249e+00 6.29775003e-02 -1.17339440e-01
-6.30703270e-01 7.86398053e-01 9.59917009e-01 -9.41401601e-01
-2.03062797e+00 -2.81451255e-01 1.12957820e-01 -1.21260181e-01
5.13447583e-01 -6.04366958e-01 -1.28577030e+00 7.43530840e-02
-2.92711496e-01 -2.35147141e-02 -2.18246832e-01 1.36911459e-02
8.02128762e-02 3.75685185e-01 -1.33134270e+00 6.81019485e-01
1.02877569e+00 -4.48646456e-01 -1.68766692e-01 1.99637994e-01
2.28897184e-01 -7.71218121e-01 -1.13492191e+00 6.81028724e-01
1.16754174e+00 -7.82753050e-01 1.15540564e+00 -2.46748462e-01
2.76089281e-01 -1.05525076e-01 -1.05855197e-01 -1.20898771e+00
-4.63194437e-02 -3.74449432e-01 -2.25475147e-01 6.97846711e-01
7.49251395e-02 -6.08620822e-01 8.61498594e-01 6.25131845e-01
-3.44748702e-03 -1.00200701e+00 -8.75679612e-01 -8.14043343e-01
-2.68287927e-01 -3.31499875e-01 2.50845224e-01 4.93222266e-01
4.40410942e-01 -6.32948101e-01 -1.76106066e-01 3.08132023e-01
5.32831967e-01 7.42932916e-01 5.81856847e-01 -1.44173980e+00
-3.05570960e-01 -2.81377554e-01 -1.80741116e-01 -1.20984101e+00
-3.45707312e-02 -4.54657406e-01 3.57992649e-01 -1.69732523e+00
1.77123368e-01 -1.07147360e+00 1.46500632e-01 2.05651402e-01
-9.96489301e-02 -1.00527920e-01 2.87804544e-01 2.47274563e-01
-5.23658507e-02 6.01460457e-01 2.15980363e+00 1.46267675e-02
-4.04510498e-01 1.14403844e-01 1.02188073e-01 5.74669242e-01
6.13158405e-01 1.08202472e-01 -2.98042536e-01 -3.40491027e-01
2.19129413e-01 4.79479820e-01 6.69268966e-01 -5.67014754e-01
-1.18918799e-01 -4.64179903e-01 3.31051499e-01 -4.12508667e-01
8.04273248e-01 -1.17713332e+00 8.59514475e-02 5.90152323e-01
-2.26637885e-01 -5.05777597e-01 5.59367359e-01 6.71184778e-01
1.60198078e-01 -3.89505625e-01 3.09300780e-01 -4.24039140e-02
-6.01216137e-01 -5.02396598e-02 -1.36913553e-01 -6.14280283e-01
8.68166149e-01 -4.33312893e-01 -1.00950107e-01 -2.44259655e-01
-9.47066247e-01 8.78687203e-02 8.80077243e-01 4.17998254e-01
8.95107090e-01 -1.10169554e+00 -1.07408993e-01 2.40756720e-01
1.18355550e-01 -8.67606029e-02 2.34198630e-01 5.94425678e-01
-5.69969773e-01 1.22210614e-01 -4.33229834e-01 -8.60587597e-01
-1.20104218e+00 5.79455912e-01 5.30756973e-02 2.93467402e-01
-6.72992051e-01 5.10157824e-01 3.49275410e-01 -1.54622674e-01
2.13866457e-01 -6.95904553e-01 -5.91217652e-02 -3.12217385e-01
1.70758694e-01 3.72736126e-01 -3.47583443e-02 6.99874684e-02
-2.38404229e-01 1.02631021e+00 1.04094565e-01 -4.83713374e-02
1.28351581e+00 -2.20230326e-01 2.85977237e-02 6.29738450e-01
7.91403770e-01 5.66321611e-02 -1.71434593e+00 1.10077314e-01
-6.65836707e-02 -8.48027945e-01 -2.54135817e-01 -8.63495946e-01
-5.30124605e-01 1.08040786e+00 4.22445148e-01 3.75518024e-01
9.50935304e-01 2.48203799e-01 4.17124391e-01 7.34991312e-01
1.06155336e+00 -1.21679080e+00 2.29749441e-01 2.26872712e-02
1.48965442e+00 -1.20666194e+00 9.47669223e-02 -1.23235333e+00
-4.56770301e-01 1.47992027e+00 5.34495413e-01 6.30965978e-02
5.54752111e-01 5.18039882e-01 3.01861018e-02 -2.10679010e-01
-4.93579209e-01 1.51764629e-02 3.96843910e-01 6.19783223e-01
9.35422555e-02 1.79616764e-01 -1.69200182e-01 -1.24537773e-01
1.93999559e-01 1.36775658e-01 3.10952932e-01 1.45199168e+00
-6.76899552e-01 -1.02894258e+00 -3.21498930e-01 3.72365266e-01
-1.09480530e-01 4.65019554e-01 -2.17103064e-02 1.04113710e+00
-3.77173185e-01 5.53658307e-01 4.45251772e-03 8.35761800e-03
4.02621299e-01 -4.79956716e-01 1.05716586e+00 -4.31755126e-01
-5.27465381e-02 1.21912211e-01 7.91602880e-02 -7.21748829e-01
-4.23567265e-01 -6.54003561e-01 -1.34408712e+00 2.59667277e-01
-6.48300648e-01 3.89403142e-02 1.27146542e+00 7.68911242e-01
1.79181948e-01 3.20829421e-01 3.71570021e-01 -1.59407818e+00
-5.88650823e-01 -9.07873809e-01 -8.11403394e-01 6.13409758e-01
3.88979644e-01 -1.47902334e+00 -2.08499253e-01 -2.40442738e-01] | [5.880251407623291, -0.8866768479347229] |
4bea048c-0565-418e-9439-c9b0cbbf5758 | momentum-contrast-for-unsupervised-visual | 1911.05722 | null | https://arxiv.org/abs/1911.05722v3 | https://arxiv.org/pdf/1911.05722v3.pdf | Momentum Contrast for Unsupervised Visual Representation Learning | We present Momentum Contrast (MoCo) for unsupervised visual representation learning. From a perspective on contrastive learning as dictionary look-up, we build a dynamic dictionary with a queue and a moving-averaged encoder. This enables building a large and consistent dictionary on-the-fly that facilitates contrastive unsupervised learning. MoCo provides competitive results under the common linear protocol on ImageNet classification. More importantly, the representations learned by MoCo transfer well to downstream tasks. MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins. This suggests that the gap between unsupervised and supervised representation learning has been largely closed in many vision tasks. | ['Yuxin Wu', 'Haoqi Fan', 'Saining Xie', 'Ross Girshick', 'Kaiming He'] | 2019-11-13 | momentum-contrast-for-unsupervised-visual-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/He_Momentum_Contrast_for_Unsupervised_Visual_Representation_Learning_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/He_Momentum_Contrast_for_Unsupervised_Visual_Representation_Learning_CVPR_2020_paper.pdf | cvpr-2020-6 | ['self-supervised-image-classification'] | ['computer-vision'] | [ 7.49597102e-02 2.17218939e-02 -6.03529871e-01 -1.17587194e-01
-5.20536959e-01 -4.90256011e-01 8.46209764e-01 1.73381656e-01
-8.13499510e-01 5.04466653e-01 9.18125734e-02 -1.61439851e-01
1.78044811e-01 -4.57249612e-01 -7.65146255e-01 -8.71049821e-01
-5.80982938e-02 5.38742185e-01 3.55749540e-02 -2.25061804e-01
1.34683266e-01 3.48535776e-01 -1.48848748e+00 3.67025167e-01
3.70436668e-01 9.29790974e-01 4.81592685e-01 7.51006246e-01
-1.03111668e-02 1.35109603e+00 -1.75716758e-01 -3.74297857e-01
4.01589394e-01 -3.15107852e-01 -8.09259593e-01 2.21317366e-01
9.07826006e-01 -1.02661133e-01 -5.05289555e-01 1.05126143e+00
4.33220357e-01 1.78326905e-01 8.53561878e-01 -1.03944063e+00
-7.94065952e-01 7.30325162e-01 -5.90572655e-01 8.02564144e-01
-1.93874583e-01 3.82964969e-01 1.40610611e+00 -1.19575000e+00
1.00623810e+00 9.80633438e-01 6.45767391e-01 5.74863672e-01
-1.49610257e+00 -2.71955431e-01 4.07319784e-01 3.16636801e-01
-1.08614922e+00 -4.02196825e-01 5.74005067e-01 -6.96030736e-01
1.34002018e+00 4.95950505e-02 8.52207541e-01 1.20147765e+00
2.24428341e-01 1.01073158e+00 1.19087589e+00 -4.20192093e-01
1.25571713e-01 1.19860567e-01 1.93053991e-01 8.86012793e-01
1.85665861e-02 5.85780084e-01 -7.93924689e-01 1.59520969e-01
7.52622604e-01 4.89400290e-02 -1.15282265e-02 -6.06882572e-01
-1.11906850e+00 9.32469845e-01 5.92355192e-01 1.16859272e-01
-3.69593412e-01 4.80542630e-01 7.70773768e-01 8.61815572e-01
3.92329335e-01 6.39064848e-01 -3.53909969e-01 -9.79977623e-02
-1.09154665e+00 -7.28409737e-02 7.19778061e-01 9.04503226e-01
9.59126770e-01 4.79588330e-01 -2.90400326e-01 6.75960481e-01
6.59962837e-03 3.99793684e-01 4.64188844e-01 -9.59133744e-01
2.24360138e-01 8.60046893e-02 -3.68156791e-01 -7.13753879e-01
-2.93067366e-01 -8.91283274e-01 -9.46276724e-01 4.86830145e-01
4.16088730e-01 8.31585228e-02 -1.13248646e+00 1.53930390e+00
-1.98073864e-01 2.45245114e-01 -1.10081099e-02 9.93750870e-01
8.10185313e-01 5.53476572e-01 2.35454187e-01 -2.84250915e-01
9.17465508e-01 -1.44424260e+00 -4.61731881e-01 -3.41320515e-01
4.93679821e-01 -4.65225130e-01 1.15266716e+00 7.07043767e-01
-1.05541730e+00 -8.07371140e-01 -1.21533203e+00 -3.53937864e-01
-2.44494528e-01 1.16820276e-01 9.28526402e-01 2.33915314e-01
-1.23694420e+00 6.52529359e-01 -8.82676303e-01 -3.28238577e-01
7.07531989e-01 1.91380024e-01 -3.42425227e-01 -4.18885387e-02
-5.39068580e-01 9.45355952e-01 5.61835587e-01 -2.88107060e-02
-1.75167942e+00 -6.58524394e-01 -9.80663180e-01 6.45222962e-02
2.74038762e-01 -7.05345333e-01 1.13524330e+00 -1.35186160e+00
-1.51255143e+00 1.23341584e+00 -1.86586066e-03 -1.08787835e+00
5.67062497e-01 -3.82198751e-01 -3.71438786e-02 2.31072590e-01
4.91567329e-02 1.16064370e+00 1.22417402e+00 -1.08925450e+00
-6.33708298e-01 5.36888279e-02 1.14657216e-01 3.60886782e-01
-2.12521181e-02 -1.85683861e-01 -4.72463489e-01 -7.23183513e-01
1.44913257e-03 -8.43206823e-01 -5.09034395e-01 1.52045161e-01
-1.47688299e-01 -6.81397617e-02 6.00521564e-01 -3.87255400e-01
8.39524150e-01 -2.38712406e+00 4.82630759e-01 6.87863976e-02
5.55605173e-01 5.51826283e-02 -3.81744057e-01 1.44548804e-01
-1.20408155e-01 -3.11123013e-01 -2.62562305e-01 -6.44135654e-01
-1.27941683e-01 5.27298689e-01 -5.25442302e-01 7.33771443e-01
3.38221431e-01 1.26306415e+00 -1.09343278e+00 -4.32288915e-01
2.56114781e-01 1.46364421e-01 -8.58547807e-01 1.14568442e-01
-3.25290680e-01 6.03329182e-01 2.46748209e-01 7.62086987e-01
4.36033487e-01 -5.04452765e-01 2.44984478e-01 -7.23291794e-03
-1.54374272e-01 2.52845019e-01 -7.76777625e-01 2.05331230e+00
-3.27854723e-01 1.20912695e+00 -3.91459875e-02 -1.43828642e+00
6.32537723e-01 -2.99455039e-02 3.32857996e-01 -9.77607191e-01
5.84827438e-02 9.99866799e-02 4.89289127e-02 -1.91415086e-01
5.83469093e-01 -7.37468749e-02 2.11307988e-01 1.38085395e-01
8.93583179e-01 8.32775549e-04 3.79221022e-01 4.12291408e-01
8.60275984e-01 1.06472142e-01 2.57610202e-01 -5.91825366e-01
2.68185735e-01 1.98430359e-01 4.26545560e-01 1.14843249e+00
-3.16460907e-01 7.31296957e-01 4.96611953e-01 -4.61669505e-01
-1.01350307e+00 -1.27915359e+00 -1.54675037e-01 1.52837539e+00
-2.38305312e-02 -6.36204958e-01 -2.54587293e-01 -8.89971375e-01
7.69172981e-02 2.81716704e-01 -8.31819832e-01 -8.00113678e-02
-5.19213617e-01 -4.48496342e-01 3.49164426e-01 6.87320054e-01
3.56376082e-01 -1.21898317e+00 -4.51352090e-01 2.32123673e-01
3.72219950e-01 -1.03722644e+00 -3.70095521e-01 9.39525485e-01
-1.00514853e+00 -9.81966734e-01 -6.99289560e-01 -1.16271186e+00
6.93041682e-01 3.17185670e-01 1.42656970e+00 -2.15192083e-02
-3.99834454e-01 6.62718415e-01 -1.67558894e-01 -2.61851966e-01
-1.84658110e-01 2.01338872e-01 -1.75393149e-02 -3.10262680e-01
1.28009871e-01 -6.16801202e-01 -4.98605341e-01 -4.12362702e-02
-4.98153389e-01 -1.79127678e-01 4.82336491e-01 1.29217231e+00
1.13932526e+00 -4.79127586e-01 3.55025381e-01 -9.99475062e-01
2.89838821e-01 -5.09147406e-01 -7.64472783e-01 -9.21372175e-02
-7.63810039e-01 1.42141268e-01 6.18533492e-01 -4.41803068e-01
-6.70714915e-01 1.99637964e-01 -2.61126310e-02 -9.16632473e-01
1.44332126e-01 5.19382417e-01 3.95867527e-01 1.89787522e-02
9.55535591e-01 4.61836755e-01 2.03040782e-02 -4.69230890e-01
9.45368588e-01 9.40085649e-02 8.31375360e-01 -4.26968694e-01
7.27093220e-01 6.40725374e-01 -2.57215083e-01 -8.04455459e-01
-1.02356184e+00 -6.09912395e-01 -7.45331407e-01 -2.93654203e-02
8.64452958e-01 -1.32859242e+00 -2.70716369e-01 3.49742472e-01
-8.22477400e-01 -8.09306145e-01 -9.87688303e-01 4.88121033e-01
-6.04457736e-01 2.01681525e-01 -7.68295705e-01 -2.77265221e-01
-2.88931578e-01 -1.00655460e+00 6.54749215e-01 8.19356963e-02
1.25561565e-01 -1.12969756e+00 2.00044319e-01 -9.07274708e-02
4.82358456e-01 -1.09222412e-01 4.90776658e-01 -3.49984676e-01
-5.22812665e-01 4.11182165e-01 -2.38255426e-01 7.17107534e-01
-2.01697215e-01 -1.35301203e-01 -1.14953899e+00 -6.95749938e-01
-1.43122941e-01 -7.47745335e-01 1.67089248e+00 5.42325974e-01
1.09658885e+00 -1.40035227e-01 -7.41288960e-02 1.10747480e+00
1.43403900e+00 -2.17974573e-01 4.25611585e-01 3.51179421e-01
8.63731146e-01 1.62804306e-01 4.41529632e-01 2.18340591e-01
2.33065829e-01 3.30122381e-01 4.67636317e-01 -2.63206393e-01
-5.39704561e-01 -2.99852669e-01 5.72032511e-01 1.20135999e+00
-1.24222435e-01 1.77766711e-01 -7.83367515e-01 6.24470472e-01
-1.70792413e+00 -1.21210229e+00 2.37831041e-01 1.82543015e+00
8.02035630e-01 3.99914503e-01 1.78929567e-01 -1.76098883e-01
4.08639759e-01 5.20771146e-01 -6.32494271e-01 -5.87419450e-01
-4.33003873e-01 5.75715303e-01 7.61301279e-01 4.55864966e-01
-1.33195066e+00 1.36359811e+00 7.49332714e+00 6.93435371e-01
-1.71451616e+00 4.12447125e-01 5.64926624e-01 -1.06416740e-01
-8.12267661e-02 8.22576210e-02 -5.98393798e-01 1.36063576e-01
6.83078408e-01 -4.28362563e-02 5.74586034e-01 1.24587178e+00
-2.78196424e-01 5.74985892e-03 -1.32326770e+00 1.41628623e+00
3.94454114e-02 -1.76166153e+00 1.29258975e-01 -9.15675145e-03
1.13824427e+00 8.75899374e-01 4.02931273e-01 7.00248241e-01
4.68964845e-01 -1.20060074e+00 9.23880875e-01 4.11969632e-01
9.40447688e-01 -4.76677597e-01 4.31439787e-01 2.73534566e-01
-1.10452986e+00 -1.01274908e-01 -7.82966614e-01 -1.82682008e-01
-2.04770982e-01 4.04564559e-01 -5.30282199e-01 1.68661073e-01
6.62152946e-01 1.39822972e+00 -7.07574069e-01 1.08285511e+00
-3.91966313e-01 9.50378597e-01 -8.32506493e-02 3.23368788e-01
5.40180683e-01 -1.62819609e-01 5.93625724e-01 1.64074802e+00
-3.70112121e-01 -3.28320444e-01 4.18524772e-01 7.61231720e-01
-4.17851210e-01 9.08640772e-02 -5.56707561e-01 -6.69236630e-02
2.60211259e-01 1.20553684e+00 -5.97770095e-01 -6.49326444e-01
-4.34610844e-01 1.11372328e+00 8.31901312e-01 3.86234224e-01
-8.32715094e-01 2.62812842e-02 5.22699773e-01 -1.35321915e-01
8.21473360e-01 -3.95396948e-01 -2.08614394e-01 -1.42350304e+00
-3.00993919e-01 -1.05242395e+00 5.48028827e-01 -2.98649520e-01
-1.36866283e+00 5.42452335e-01 -2.22538650e-01 -1.32222760e+00
-2.09160224e-01 -8.87089431e-01 -5.89105248e-01 2.64766812e-01
-1.93448484e+00 -8.85458410e-01 -2.25065514e-01 7.05030501e-01
7.29391634e-01 -4.49107796e-01 6.91741049e-01 2.86796212e-01
-4.74741250e-01 6.49959207e-01 1.53019786e-01 2.61331469e-01
5.53404927e-01 -1.60883069e+00 3.30183834e-01 9.09039199e-01
9.41128790e-01 3.80239993e-01 3.57223749e-01 -1.61903515e-01
-1.25117040e+00 -1.02390623e+00 3.85782599e-01 -3.78172100e-01
6.81625843e-01 -4.12712812e-01 -7.98752725e-01 8.83590400e-01
5.40087342e-01 6.92661405e-01 3.16058248e-01 1.97100207e-01
-7.35116839e-01 -3.06134224e-02 -6.65959656e-01 2.60950059e-01
1.21134377e+00 -1.04374850e+00 -7.28973866e-01 3.42819363e-01
4.79339093e-01 -5.83537042e-01 -5.76102078e-01 1.46721199e-01
4.07818109e-01 -8.95905733e-01 7.99837112e-01 -7.97800541e-01
4.83690470e-01 -1.23504601e-01 -1.59882903e-01 -1.20368540e+00
-4.92309481e-01 -7.53340840e-01 -3.47764134e-01 8.30196917e-01
4.30401236e-01 -4.44945574e-01 7.61495531e-01 -3.47538382e-01
-3.00065547e-01 -7.12606788e-01 -8.16295624e-01 -9.28591907e-01
2.70290345e-01 -6.36515558e-01 -1.36898071e-01 1.03708076e+00
-2.18733579e-01 5.41159630e-01 -3.96840572e-01 -1.50118560e-01
6.63381577e-01 8.17019865e-03 6.72093809e-01 -9.83954608e-01
-6.52448118e-01 -6.59678698e-01 -5.34906089e-01 -1.48764181e+00
1.86991364e-01 -1.65427065e+00 1.93272866e-02 -1.25288200e+00
3.38176787e-01 -4.16149557e-01 -9.01670754e-01 4.92277950e-01
-5.34425266e-02 4.02244151e-01 4.58491027e-01 5.23491561e-01
-8.68154764e-01 4.32838798e-01 1.24951601e+00 -6.15393519e-01
-1.30045265e-01 -5.48859954e-01 -6.58928096e-01 7.46422946e-01
6.78636074e-01 -7.36673534e-01 -3.25903594e-01 -6.96681559e-01
9.82864723e-02 -3.18626046e-01 4.52219188e-01 -1.06866205e+00
2.72321999e-01 7.64994025e-02 2.15278983e-01 -3.63118857e-01
7.92983845e-02 -4.41837639e-01 -3.85324061e-01 5.30605376e-01
-5.07163465e-01 8.52354094e-02 2.02382460e-01 6.96004093e-01
-4.37034369e-01 -5.66782914e-02 9.87229109e-01 -2.24773586e-01
-1.28709042e+00 3.69055122e-01 -3.44457626e-01 5.79476655e-01
6.63707376e-01 -2.00639576e-01 -3.32675517e-01 -1.17639758e-01
-1.25259900e+00 1.14133671e-01 3.32540244e-01 3.09284776e-01
5.49237669e-01 -1.14617586e+00 -7.81648219e-01 2.67436236e-01
2.17935279e-01 -9.91342962e-02 -1.41783804e-01 9.85139787e-01
-6.15706325e-01 2.62647122e-01 -5.05008042e-01 -1.06805646e+00
-8.98005426e-01 5.42473495e-01 4.70118791e-01 -4.22134042e-01
-8.33300292e-01 1.30135357e+00 2.39671290e-01 -1.58127248e-01
5.37738442e-01 -4.26119119e-01 -1.50767744e-01 2.72204876e-01
4.19196635e-01 1.15553118e-01 -1.37671545e-01 -4.16546464e-01
-2.09361836e-01 3.33391219e-01 -2.50573039e-01 -4.01248224e-02
1.25303614e+00 1.12752449e-02 -1.16350643e-01 5.50884843e-01
1.24468184e+00 1.84966505e-01 -1.68970013e+00 -3.03174168e-01
1.09488368e-01 -1.08374059e-01 2.67055780e-01 -5.18437684e-01
-1.44172359e+00 9.61840391e-01 8.30604076e-01 -2.19872579e-01
8.24056029e-01 1.49577484e-01 3.19569319e-01 7.81803489e-01
3.62930715e-01 -1.13612187e+00 4.23961371e-01 8.21574330e-01
8.41898143e-01 -1.51420426e+00 9.49753150e-02 -1.21180518e-02
-8.86852622e-01 1.09228110e+00 5.18829525e-01 -7.22769201e-01
7.64571488e-01 2.35580742e-01 1.30782470e-01 -3.11288804e-01
-9.69574869e-01 -6.91793203e-01 3.43419194e-01 6.34870708e-01
1.71212494e-01 -3.41530964e-02 -1.24336623e-01 3.14954698e-01
-1.15886211e-01 -1.80263370e-01 3.02272230e-01 8.10055137e-01
-4.39803869e-01 -9.04918969e-01 2.73897439e-01 4.32612598e-01
-3.30442101e-01 -4.86237943e-01 6.32646866e-03 8.86531115e-01
2.24510476e-01 5.25254786e-01 4.21421438e-01 -1.98734164e-01
6.62045693e-03 -1.13333531e-01 7.31923461e-01 -8.91642570e-01
-7.46950746e-01 3.46554779e-02 -1.97960977e-02 -7.60254383e-01
-5.06107807e-01 -5.49905360e-01 -1.18620265e+00 -5.30118831e-02
-2.69041151e-01 1.48497950e-02 1.93080172e-01 8.60986352e-01
1.70157954e-01 5.24699330e-01 4.80024278e-01 -8.94586205e-01
-5.94446838e-01 -8.23593736e-01 -6.06460154e-01 4.37766224e-01
5.12949109e-01 -6.47386491e-01 -4.01727468e-01 2.06169426e-01] | [9.447884559631348, 2.401327610015869] |
66e5cfd9-76a9-4b50-a254-0b74e7ddfb04 | unsupervised-cross-modal-alignment-of-speech | 1805.07467 | null | http://arxiv.org/abs/1805.07467v2 | http://arxiv.org/pdf/1805.07467v2.pdf | Unsupervised Cross-Modal Alignment of Speech and Text Embedding Spaces | Recent research has shown that word embedding spaces learned from text
corpora of different languages can be aligned without any parallel data
supervision. Inspired by the success in unsupervised cross-lingual word
embeddings, in this paper we target learning a cross-modal alignment between
the embedding spaces of speech and text learned from corpora of their
respective modalities in an unsupervised fashion. The proposed framework learns
the individual speech and text embedding spaces, and attempts to align the two
spaces via adversarial training, followed by a refinement procedure. We show
how our framework could be used to perform spoken word classification and
translation, and the results on these two tasks demonstrate that the
performance of our unsupervised alignment approach is comparable to its
supervised counterpart. Our framework is especially useful for developing
automatic speech recognition (ASR) and speech-to-text translation systems for
low- or zero-resource languages, which have little parallel audio-text data for
training modern supervised ASR and speech-to-text translation models, but
account for the majority of the languages spoken across the world. | ['James Glass', 'Wei-Hung Weng', 'Yu-An Chung', 'Schrasing Tong'] | 2018-05-18 | unsupervised-cross-modal-alignment-of-speech-1 | http://papers.nips.cc/paper/7965-unsupervised-cross-modal-alignment-of-speech-and-text-embedding-spaces | http://papers.nips.cc/paper/7965-unsupervised-cross-modal-alignment-of-speech-and-text-embedding-spaces.pdf | neurips-2018-12 | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 3.84243935e-01 3.65615785e-01 -9.98143330e-02 -5.31053066e-01
-1.11174619e+00 -5.85202694e-01 1.02163279e+00 -7.74467587e-02
-6.26937747e-01 2.50874639e-01 4.94877130e-01 -5.71140409e-01
4.51384991e-01 -3.78448844e-01 -5.40505171e-01 -6.34216428e-01
2.12986618e-01 6.88784540e-01 -1.41066983e-01 -4.01576728e-01
-3.39756042e-01 2.42372766e-01 -1.13705492e+00 9.89541337e-02
6.08711362e-01 5.33271968e-01 1.68422222e-01 7.97074556e-01
-4.19462264e-01 3.59512627e-01 -2.55581111e-01 -5.97268283e-01
2.02269360e-01 -5.75818300e-01 -8.53750944e-01 1.11218110e-01
4.87760961e-01 4.81471755e-02 -7.64126778e-01 9.62752342e-01
6.82154655e-01 2.65878085e-02 7.59427547e-01 -1.08868086e+00
-9.13322508e-01 7.41387725e-01 -2.36520935e-02 -7.67017975e-02
3.33214432e-01 -2.34364942e-01 1.24928761e+00 -1.19894958e+00
4.13717687e-01 1.17080986e+00 3.83158118e-01 6.92555130e-01
-1.50942278e+00 -3.40294182e-01 -2.24912949e-02 1.02806576e-01
-1.52518618e+00 -1.09346616e+00 6.98552012e-01 -3.63483787e-01
1.12639248e+00 2.72082984e-01 3.66631150e-01 1.49344540e+00
-8.17034394e-02 7.65123188e-01 8.31296325e-01 -9.25887942e-01
7.36126974e-02 4.82570320e-01 -3.32277536e-01 5.61759531e-01
-2.90460557e-01 -5.29935642e-04 -9.19400215e-01 -5.01041450e-02
3.38543206e-01 -2.52256513e-01 -2.11193934e-01 -3.26387584e-01
-1.58685446e+00 9.25521970e-01 8.33840482e-03 6.25411332e-01
-1.33482695e-01 -5.94810732e-02 5.73830664e-01 6.95777833e-01
6.96904957e-01 2.95486629e-01 -3.50199103e-01 -1.17736295e-01
-8.42383683e-01 -1.69765621e-01 8.18202317e-01 1.01512432e+00
5.82453966e-01 4.72710460e-01 1.70069650e-01 1.09994698e+00
4.70048785e-01 9.83672202e-01 9.87564683e-01 -4.08367127e-01
6.23650193e-01 -1.20639289e-02 -2.36978143e-01 -8.50427926e-01
-2.11853292e-02 1.79273531e-01 -6.88579977e-01 -2.59482831e-01
1.48983896e-01 -9.48098376e-02 -7.16405630e-01 1.81958795e+00
9.90501866e-02 1.64747387e-01 7.24801958e-01 6.53816223e-01
4.62513089e-01 8.32904279e-01 -2.33809635e-01 -1.48787260e-01
1.28763592e+00 -9.94700909e-01 -9.75468755e-01 -5.81589162e-01
8.54401529e-01 -1.01928782e+00 1.28165293e+00 -1.68424919e-02
-1.06068361e+00 -5.43698072e-01 -1.07767999e+00 -1.36461347e-01
-5.58636487e-01 8.80083814e-02 5.13009652e-02 6.38244092e-01
-1.07038772e+00 1.48236558e-01 -9.13644135e-01 -7.53320873e-01
-1.40343294e-01 1.66635811e-01 -6.52644634e-01 3.31234699e-03
-1.34603858e+00 1.13406146e+00 3.29807073e-01 6.64476901e-02
-6.04960382e-01 -2.27934852e-01 -1.23485303e+00 -1.53322488e-01
4.61841039e-02 -3.16515207e-01 1.18425274e+00 -1.19310284e+00
-1.86728704e+00 1.12147284e+00 -2.54179478e-01 -5.14311790e-01
1.85864121e-01 -2.47527778e-01 -7.07724869e-01 -1.09726273e-01
-8.50055292e-02 5.11044860e-01 1.14550734e+00 -8.72195482e-01
-4.07986760e-01 -5.38407326e-01 -4.25383180e-01 3.32063705e-01
-7.60739982e-01 2.50581622e-01 -2.59956419e-01 -7.69021034e-01
6.79780096e-02 -9.89892781e-01 -1.14779556e-02 -2.26092234e-01
-2.87863553e-01 -1.81937158e-01 7.04271793e-01 -8.28413427e-01
8.91671360e-01 -2.31956458e+00 6.99211359e-01 6.38373345e-02
-3.27867180e-01 2.12800771e-01 -5.97131848e-01 8.38686407e-01
-1.16020277e-01 1.55452807e-02 -3.35464358e-01 -8.19318235e-01
2.77152896e-01 6.77032232e-01 -8.54321301e-01 5.56035578e-01
4.37973469e-01 7.60607243e-01 -7.86686182e-01 -3.55519623e-01
2.42071867e-01 6.53363585e-01 -2.76203483e-01 6.74661458e-01
4.44617420e-02 4.21635002e-01 -6.34307340e-02 2.87011117e-01
9.35382470e-02 4.60162371e-01 3.41069549e-01 -5.44657633e-02
3.35392654e-02 7.61966348e-01 -8.26099277e-01 2.02685452e+00
-8.29768777e-01 8.86113524e-01 4.24599200e-02 -1.39316416e+00
1.05817211e+00 8.18562806e-01 3.17735851e-01 -6.77517116e-01
1.21407218e-01 4.55235392e-01 -1.74776651e-02 -4.92363781e-01
4.95912611e-01 -3.09786379e-01 -1.64689109e-01 6.33720040e-01
5.47106385e-01 -5.62444925e-01 -2.26878479e-01 4.10083644e-02
7.95528531e-01 -2.38918081e-01 1.90897286e-01 -1.18644290e-01
7.11167216e-01 -2.10997716e-01 5.07576168e-02 3.45324755e-01
2.66613322e-04 6.73204005e-01 3.69269066e-02 -1.82996899e-01
-1.32618022e+00 -1.17948258e+00 -1.58362180e-01 1.23877263e+00
-2.41087645e-01 -4.17254120e-01 -6.56565011e-01 -4.93829161e-01
-2.93117821e-01 7.74981737e-01 -3.41858119e-01 -2.02017471e-01
-5.96255839e-01 -2.04428330e-01 1.00491273e+00 4.29473728e-01
-2.50374436e-01 -9.26522255e-01 2.30765596e-01 2.71542937e-01
-2.45328128e-01 -1.61669421e+00 -9.02819216e-01 3.26861709e-01
-4.92197126e-01 -6.65819824e-01 -7.97074258e-01 -1.16204393e+00
4.26054358e-01 1.14758778e-02 8.46013188e-01 -3.27047855e-01
-1.66031215e-02 8.17244291e-01 -5.44980109e-01 -4.35851723e-01
-1.11572838e+00 2.05010965e-01 8.09357584e-01 4.64826703e-01
4.58254546e-01 -6.01277649e-01 2.69550122e-02 2.99444050e-01
-1.05959404e+00 -7.57148266e-02 2.79543847e-01 9.80379283e-01
4.70978975e-01 -6.65349782e-01 6.10393286e-01 -5.19182980e-01
6.10981822e-01 -2.04397202e-01 -4.34984416e-01 2.95973182e-01
-5.49469292e-01 3.12621474e-01 6.90930486e-01 -6.25487387e-01
-5.50547898e-01 2.55145282e-02 -4.35462892e-01 -4.33909297e-01
-1.56517670e-01 4.70554024e-01 -2.95430571e-01 2.24856481e-01
5.04153967e-01 5.43605924e-01 1.63894907e-01 -3.36135060e-01
8.62828553e-01 1.32575440e+00 7.04092383e-01 -3.90476197e-01
1.05825877e+00 2.07133397e-01 -5.66998243e-01 -1.23315072e+00
-4.61407036e-01 -5.12049139e-01 -9.96680975e-01 2.78674960e-01
1.10998893e+00 -8.75673771e-01 -2.78891847e-02 2.36106887e-01
-1.46226704e+00 -1.96096256e-01 -4.28511620e-01 7.57636487e-01
-8.42327058e-01 3.16919714e-01 -3.37331533e-01 -7.51526475e-01
-2.60834754e-01 -1.11697781e+00 1.16982388e+00 -3.83937210e-01
-3.05517614e-01 -1.33547854e+00 5.56874812e-01 2.38626093e-01
5.54588675e-01 -5.64990759e-01 8.29797864e-01 -1.33543909e+00
-5.31029934e-03 -3.15348864e-01 3.43646854e-01 7.52716064e-01
3.56024057e-01 -1.18647605e-01 -1.20767713e+00 -4.89773065e-01
-7.79222101e-02 -4.36382711e-01 4.14094716e-01 -8.30981880e-02
4.63299125e-01 -5.15785158e-01 3.40060983e-03 5.16979873e-01
1.01022422e+00 -7.10389018e-02 4.16183770e-01 1.44351110e-01
8.08112979e-01 7.35551476e-01 2.33205557e-01 -4.96172495e-02
2.46214926e-01 1.07197511e+00 5.14220037e-02 -1.89468414e-01
-7.55923167e-02 -3.56086016e-01 1.02831304e+00 1.85819578e+00
4.60059643e-01 -1.51426703e-01 -1.03940320e+00 9.14207816e-01
-1.60926580e+00 -7.71417499e-01 3.93034726e-01 2.33566141e+00
1.09750295e+00 -2.16180339e-01 2.22324636e-02 1.05655715e-01
4.80096459e-01 4.47484285e-01 -1.04425937e-01 -7.17615664e-01
-1.01950295e-01 3.85581821e-01 3.14392686e-01 8.06524754e-01
-1.00848746e+00 1.20088434e+00 6.45164871e+00 6.73666596e-01
-1.32176638e+00 4.51040536e-01 6.52022809e-02 2.46595275e-02
-3.99091184e-01 -1.45002142e-01 -4.90222663e-01 8.30452591e-02
1.54189444e+00 -3.82987350e-01 6.04659319e-01 5.11650741e-01
1.03687592e-01 7.45459557e-01 -1.64734244e+00 1.05577743e+00
4.02062982e-01 -1.07678723e+00 1.17315471e-01 -1.09429978e-01
5.03890276e-01 4.31992471e-01 1.25124931e-01 3.27896625e-01
2.08653405e-01 -1.13605702e+00 6.99799240e-01 3.46103087e-02
9.88025427e-01 -5.11809528e-01 7.20350981e-01 3.63137096e-01
-1.05152488e+00 3.47656280e-01 -3.51854354e-01 2.57367909e-01
3.12222600e-01 7.29155391e-02 -1.03812420e+00 5.32634377e-01
3.23610574e-01 6.29414022e-01 -1.77432388e-01 8.89125019e-02
-1.19428746e-01 6.80737197e-01 -2.29442775e-01 1.00989006e-01
2.03477576e-01 -3.35778773e-01 7.38755226e-01 1.34865904e+00
4.41966444e-01 -4.46834564e-01 2.42578417e-01 5.74096143e-01
-1.55474424e-01 5.37417412e-01 -1.08134508e+00 -7.32386112e-01
4.63279128e-01 8.51228714e-01 -9.39134359e-02 -2.29944691e-01
-7.52752721e-01 1.34051275e+00 2.97347993e-01 4.49261159e-01
-5.23496151e-01 -3.68737221e-01 1.01777315e+00 -1.21099669e-02
1.98297739e-01 -6.11296773e-01 1.31625943e-02 -1.42480111e+00
-4.34681727e-03 -1.06339598e+00 -1.07563220e-01 -5.26403546e-01
-1.54163921e+00 1.10012734e+00 -3.01741898e-01 -1.23015559e+00
-6.57968998e-01 -7.22595930e-01 -4.50761884e-01 1.06567168e+00
-1.43771243e+00 -1.42159379e+00 4.29039419e-01 7.25952685e-01
8.16668928e-01 -8.24174404e-01 1.41833234e+00 1.68897420e-01
-2.89152563e-01 9.11706567e-01 4.17637348e-01 4.76390064e-01
8.92223001e-01 -1.22097945e+00 7.46496081e-01 8.96556973e-01
1.11011624e+00 3.86203200e-01 7.15007961e-01 -1.17001377e-01
-1.62177694e+00 -1.02372003e+00 1.22323895e+00 -5.22342801e-01
1.25793993e+00 -8.11787784e-01 -9.92580652e-01 9.60792124e-01
6.97594702e-01 5.14180921e-02 9.63430166e-01 1.91722766e-01
-6.05837286e-01 -1.94635317e-01 -5.62335193e-01 7.96017110e-01
7.12921560e-01 -1.38194048e+00 -8.92907917e-01 5.25468528e-01
1.02981448e+00 -2.64935553e-01 -7.87175238e-01 2.99397912e-02
4.66745317e-01 -4.14888591e-01 8.48801017e-01 -9.59717631e-01
1.29685968e-01 -1.05262129e-02 -8.34052682e-01 -1.60819626e+00
3.00201178e-01 -7.29669809e-01 5.82347699e-02 1.38055015e+00
5.46789229e-01 -8.47656906e-01 2.55037546e-01 2.22095415e-01
-2.44084254e-01 -2.80944794e-01 -1.35100830e+00 -8.85677159e-01
3.52822512e-01 -6.31329536e-01 3.06096047e-01 1.10461545e+00
1.69595078e-01 6.77671254e-01 -4.31322366e-01 4.97758538e-01
3.16111803e-01 -1.58908635e-01 8.09617877e-01 -8.53798926e-01
-3.91525567e-01 -2.86150545e-01 -6.26706839e-01 -1.04231095e+00
7.45355725e-01 -1.31052268e+00 3.72233599e-01 -9.98673379e-01
-3.30310404e-01 -2.70549655e-01 -1.78476110e-01 5.88549316e-01
2.09368512e-01 1.96589977e-01 1.57620773e-01 2.58209169e-01
-3.10696680e-02 1.02106452e+00 6.44300163e-01 -4.54546243e-01
-6.97482452e-02 -2.13146463e-01 -2.68757373e-01 5.31028092e-01
5.09204745e-01 -5.96563578e-01 -3.00041288e-01 -7.74941325e-01
-9.13522989e-02 -4.13714275e-02 -6.36277162e-03 -7.02483416e-01
1.83094278e-01 -1.42773399e-02 -3.37895066e-01 -1.30465284e-01
4.75128114e-01 -1.11507130e+00 -3.22883397e-01 1.12547828e-02
-6.06588066e-01 1.13338806e-01 1.61700457e-01 4.38754082e-01
-6.74037457e-01 -1.73412725e-01 6.61943614e-01 3.61372977e-01
-2.09835112e-01 2.42985293e-01 -5.87998092e-01 1.15060844e-01
7.52658308e-01 -9.52476461e-04 6.60160184e-02 -5.81387341e-01
-8.74430299e-01 -9.89677459e-02 2.99740255e-01 8.81748557e-01
6.17245615e-01 -1.64738977e+00 -1.03015804e+00 7.27208495e-01
4.37357783e-01 -3.64798993e-01 -3.11515301e-01 8.64319861e-01
-1.52494743e-01 5.51299810e-01 6.31571934e-02 -6.85228765e-01
-1.30959308e+00 2.93645918e-01 3.46555680e-01 3.55340838e-02
-4.24726397e-01 6.63331389e-01 4.93525714e-02 -9.47573602e-01
4.02222484e-01 -1.08106092e-01 1.98595986e-01 -2.59994473e-02
4.38731104e-01 -1.57476828e-01 2.11910352e-01 -1.16785228e+00
-3.91827703e-01 4.95682061e-01 -9.40766931e-02 -7.04844058e-01
1.37800848e+00 -4.18382704e-01 5.87831205e-03 1.02928555e+00
1.39080834e+00 2.72085756e-01 -5.89243650e-01 -6.43735886e-01
3.04955095e-02 -2.01317683e-01 -2.10266662e-04 -1.46104291e-01
-6.47432029e-01 1.33229601e+00 5.90092540e-01 3.84814203e-01
7.75515556e-01 2.34663323e-01 8.91691208e-01 5.98923743e-01
1.94404840e-01 -1.01790392e+00 1.30211245e-02 8.22398603e-01
1.00062299e+00 -1.29473674e+00 -5.98552465e-01 -1.81382953e-03
-6.86853349e-01 1.11324608e+00 7.61573315e-02 7.05175474e-02
5.05703986e-01 3.10725570e-01 5.41660547e-01 2.10603282e-01
-7.27372110e-01 -2.32324392e-01 5.26814044e-01 7.63449371e-01
6.49588823e-01 -3.86618525e-02 3.29891965e-02 3.72482240e-01
-3.35641682e-01 -6.44987404e-01 3.92158717e-01 6.59508705e-01
-1.46858305e-01 -1.59662330e+00 -2.89207578e-01 -9.89496708e-02
-3.49100530e-01 -4.59964752e-01 -5.12962699e-01 5.28715551e-01
-5.24414778e-01 8.65981340e-01 2.83325016e-01 -4.72980440e-01
3.95094275e-01 7.16737092e-01 3.99175704e-01 -8.61123860e-01
-1.82311848e-01 1.04480334e-01 5.25434352e-02 -4.24071312e-01
-4.00867462e-01 -4.97417569e-01 -1.11963737e+00 -5.69799244e-02
-2.19022244e-01 2.29343265e-01 8.49847376e-01 1.13636506e+00
1.51413336e-01 2.06970006e-01 8.76626611e-01 -1.00281811e+00
-8.45956445e-01 -1.17739677e+00 -6.20902121e-01 4.37108636e-01
6.11182570e-01 -1.98639542e-01 -4.55856472e-01 3.89024943e-01] | [14.40589427947998, 7.060170650482178] |
e5c583c4-f6bf-4ce4-a568-ac3cad092938 | further-improving-weakly-supervised-object | 2301.01060 | null | https://arxiv.org/abs/2301.01060v1 | https://arxiv.org/pdf/2301.01060v1.pdf | Further Improving Weakly-supervised Object Localization via Causal Knowledge Distillation | Weakly-supervised object localization aims to indicate the category as well as the scope of an object in an image given only the image-level labels. Most of the existing works are based on Class Activation Mapping (CAM) and endeavor to enlarge the discriminative area inside the activation map to perceive the whole object, yet ignore the co-occurrence confounder of the object and context (e.g., fish and water), which makes the model inspection hard to distinguish object boundaries. Besides, the use of CAM also brings a dilemma problem that the classification and localization always suffer from a performance gap and can not reach their highest accuracy simultaneously. In this paper, we propose a casual knowledge distillation method, dubbed KD-CI-CAM, to address these two under-explored issues in one go. More specifically, we tackle the co-occurrence context confounder problem via causal intervention (CI), which explores the causalities among image features, contexts, and categories to eliminate the biased object-context entanglement in the class activation maps. Based on the de-biased object feature, we additionally propose a multi-teacher causal distillation framework to balance the absorption of classification knowledge and localization knowledge during model training. Extensive experiments on several benchmarks demonstrate the effectiveness of KD-CI-CAM in learning clear object boundaries from confounding contexts and addressing the dilemma problem between classification and localization performance. | ['Jun Xiao', 'Yi Yang', 'Fei Gao', 'Qiyi Li', 'Shengjian Wu', 'Yawei Luo', 'Feifei Shao'] | 2023-01-03 | null | null | null | null | ['weakly-supervised-object-localization'] | ['computer-vision'] | [ 1.93236217e-01 5.04892282e-02 -2.86897361e-01 -3.01071793e-01
-3.24460834e-01 -4.85479861e-01 6.20467067e-01 2.34709322e-01
-5.37955642e-01 4.47545767e-01 5.49315996e-02 -1.68934688e-02
-4.18625534e-01 -5.19240618e-01 -9.07774806e-01 -1.08774555e+00
2.08271608e-01 1.22776158e-01 3.30540001e-01 1.00635402e-01
2.58197784e-01 1.37118995e-01 -1.50963318e+00 2.69576192e-01
1.23136616e+00 1.02291691e+00 4.01608706e-01 -8.18399712e-03
-1.43765152e-01 5.58317423e-01 -6.53132260e-01 -8.39752778e-02
-1.14580086e-02 -5.89598596e-01 -6.11752927e-01 -1.08867861e-01
6.19897127e-01 -1.62992224e-01 3.52021419e-02 1.29281747e+00
3.37208629e-01 -5.02012596e-02 7.13923931e-01 -1.38897705e+00
-9.67970550e-01 7.01894939e-01 -8.45524907e-01 5.00831604e-01
-1.87376775e-02 1.53261349e-01 9.57950056e-01 -9.30249274e-01
8.29691216e-02 1.30289972e+00 4.29179490e-01 3.84397328e-01
-1.44383740e+00 -9.62385297e-01 7.70866573e-01 4.83272761e-01
-1.68134308e+00 3.71581279e-02 9.04894233e-01 -6.46684766e-01
3.22761625e-01 1.80551454e-01 6.08762681e-01 1.04276943e+00
-6.20884895e-02 7.97665715e-01 1.50549495e+00 -5.02561569e-01
2.51616776e-01 3.54290217e-01 3.62868994e-01 5.22908866e-01
2.60965973e-01 2.51174867e-01 -6.18094802e-01 1.25027478e-01
7.68706977e-01 1.10989679e-02 -4.31805551e-01 -3.23671907e-01
-1.31477416e+00 6.30002797e-01 1.02109003e+00 4.80453819e-01
-1.25736117e-01 -3.17545645e-02 -6.48496076e-02 -1.58137530e-01
3.28711659e-01 5.33606887e-01 -4.21575308e-01 5.02860188e-01
-6.14403665e-01 1.07447363e-01 2.24260330e-01 7.51420021e-01
1.02546632e+00 -2.02929378e-01 -3.83248210e-01 5.21732688e-01
3.67129028e-01 4.16590631e-01 4.82100725e-01 -4.47086990e-01
3.03681642e-01 7.92461574e-01 -1.28631279e-01 -1.13636410e+00
-2.53202945e-01 -7.57152021e-01 -6.77275419e-01 9.89644043e-03
3.77452791e-01 7.33240321e-02 -9.74527776e-01 2.13974190e+00
5.43992043e-01 6.39834523e-01 -1.51156023e-01 1.12078524e+00
1.00115240e+00 3.58521014e-01 4.59560275e-01 -1.49722770e-01
1.52510071e+00 -1.02468050e+00 -7.21655309e-01 -4.35987413e-01
4.54936236e-01 -2.36514971e-01 1.20804477e+00 1.85982045e-02
-5.98818004e-01 -7.03429639e-01 -1.04503596e+00 3.84732969e-02
-4.45818871e-01 1.48432061e-01 6.43210292e-01 2.42978930e-01
-6.00956321e-01 2.50974983e-01 -5.87550640e-01 2.31189933e-02
7.33346581e-01 3.72577041e-01 -2.41898015e-01 -1.49801001e-01
-1.22119713e+00 9.38784361e-01 7.09322929e-01 2.60094047e-01
-1.15262532e+00 -9.41437125e-01 -5.93970835e-01 2.02305913e-01
6.76078320e-01 -3.65946233e-01 7.45901346e-01 -1.29859817e+00
-1.03335500e+00 5.46458125e-01 1.30131319e-02 8.26354474e-02
1.72685459e-01 -2.64748454e-01 -1.52226299e-01 -5.99691048e-02
2.34227881e-01 1.00976896e+00 8.45305324e-01 -1.56354880e+00
-8.25960755e-01 -5.68817317e-01 1.88907176e-01 4.57584441e-01
-5.05435824e-01 -3.34451914e-01 -3.30404878e-01 -5.95160186e-01
2.78888762e-01 -7.29097962e-01 -2.28255317e-02 -1.25533104e-01
-2.43577123e-01 -5.85104764e-01 8.38151038e-01 -3.11846554e-01
1.16534531e+00 -2.42043424e+00 2.46260852e-01 -2.98744012e-02
2.66775548e-01 1.34168923e-01 -2.33154073e-01 -1.32373974e-01
-1.47070244e-01 2.08758399e-01 -2.19054386e-01 7.86320642e-02
-3.54342192e-01 3.71165335e-01 -4.25468266e-01 4.69930738e-01
4.69153196e-01 1.02210057e+00 -1.17870498e+00 -7.20489264e-01
2.61127762e-02 4.05057073e-01 -5.21134138e-01 1.38959959e-01
-1.98000968e-01 8.03744316e-01 -3.33541244e-01 5.08998275e-01
6.11588478e-01 -2.22807661e-01 3.09342351e-02 -4.46530044e-01
-9.95506644e-02 1.22376502e-01 -1.09139740e+00 1.52869344e+00
-3.01889569e-01 3.62023026e-01 -1.39916584e-01 -1.14030004e+00
6.07085764e-01 1.62557438e-01 1.85317099e-01 -8.25161278e-01
2.41651423e-02 2.57603675e-01 5.58776855e-01 -4.83668625e-01
-2.20331885e-02 -2.68090516e-01 1.46807641e-01 1.58487797e-01
2.49172226e-01 1.85907602e-01 -1.95398688e-01 9.41301435e-02
5.96235812e-01 1.79707512e-01 1.81101784e-01 -5.89992523e-01
4.35425699e-01 -1.76980302e-01 7.54841447e-01 7.62697279e-01
-3.03310037e-01 2.92987883e-01 5.23436427e-01 -7.28891715e-02
-4.41640377e-01 -1.00823116e+00 -4.16892350e-01 1.23356247e+00
8.64958525e-01 -1.18627869e-01 -7.02318966e-01 -1.09001887e+00
1.44546293e-02 7.42574275e-01 -9.98338938e-01 -4.90188420e-01
-4.76868212e-01 -9.10085618e-01 4.05699164e-01 6.24394000e-01
5.82842112e-01 -9.16385055e-01 -6.16183579e-01 -1.48225456e-01
-2.64473647e-01 -9.45708811e-01 -3.39868844e-01 4.41431791e-01
-6.45792067e-01 -9.76165652e-01 -5.38733482e-01 -1.01000106e+00
1.03696191e+00 4.31805581e-01 7.92150438e-01 1.36311397e-01
-1.03061236e-01 2.02309698e-01 -1.46572486e-01 -4.93448466e-01
8.06350335e-02 -1.45678416e-01 -2.44180132e-02 2.08256662e-01
3.43604416e-01 -5.48124492e-01 -6.59069896e-01 6.17942452e-01
-8.37790012e-01 1.71185076e-01 7.67793477e-01 9.98298466e-01
6.18726969e-01 3.19467902e-01 7.75313973e-01 -5.34263790e-01
3.06791276e-01 -6.81801796e-01 -4.24733579e-01 4.69417185e-01
-7.66807556e-01 3.06829158e-02 2.35805631e-01 -1.00987685e+00
-1.16754889e+00 3.41154379e-03 3.33854914e-01 -4.82424825e-01
-1.93478540e-01 5.47764421e-01 -5.19491255e-01 -6.82555065e-02
6.17696643e-01 2.52998799e-01 -3.74510735e-01 -3.36082220e-01
5.38358212e-01 3.00641418e-01 6.00901842e-01 -6.45265162e-01
5.97725034e-01 4.57503855e-01 -1.18236020e-01 -4.07718331e-01
-1.21236587e+00 -4.38311279e-01 -7.74066389e-01 -2.01139465e-01
9.54707861e-01 -1.02844846e+00 -5.73465407e-01 2.86312997e-01
-1.10349047e+00 -1.60997778e-01 -1.82034284e-01 6.00467503e-01
6.02579035e-04 1.03511497e-01 -9.24904421e-02 -5.82926631e-01
2.37959340e-01 -1.18594348e+00 8.69834900e-01 4.92004275e-01
2.72724897e-01 -8.25702310e-01 -2.17211321e-01 1.35502100e-01
2.72236854e-01 2.83291582e-02 1.22562778e+00 -6.61662459e-01
-7.06056774e-01 5.65976985e-02 -6.18941784e-01 2.08367139e-01
1.51341975e-01 -3.92119497e-01 -1.13532925e+00 -6.48436844e-02
1.20234147e-01 -3.43689203e-01 1.09850657e+00 2.36164898e-01
1.32663834e+00 -5.44360578e-01 -5.28567314e-01 2.98673004e-01
1.33533347e+00 2.10142046e-01 3.63749534e-01 -2.15982925e-02
1.02491140e+00 8.90168905e-01 6.82573199e-01 -9.05434117e-02
3.40117365e-01 7.81488240e-01 6.06602073e-01 -1.69551566e-01
-2.89117157e-01 -4.47779328e-01 7.52622858e-02 5.87560058e-01
5.42829707e-02 -1.02489047e-01 -9.18661773e-01 6.94847345e-01
-1.87063301e+00 -6.04040623e-01 -2.30390921e-01 2.08344007e+00
1.10513663e+00 5.43656759e-03 -2.72161990e-01 -8.25836360e-02
7.84834802e-01 -1.19469918e-01 -5.14500976e-01 3.26583028e-01
-1.09359808e-01 -1.38715386e-01 3.20820689e-01 4.24066842e-01
-1.07408249e+00 9.16469097e-01 5.55278778e+00 1.14260674e+00
-1.21518481e+00 4.05767083e-01 6.11992538e-01 6.19585104e-02
-3.60375404e-01 2.86136597e-01 -9.55398858e-01 5.68463743e-01
3.99617940e-01 2.19171941e-01 2.64949381e-01 8.07993591e-01
-1.86810285e-01 -4.54955935e-01 -1.36688519e+00 8.39906335e-01
2.34202340e-01 -8.70661259e-01 7.83016682e-02 7.28136823e-02
7.94118106e-01 -2.31850788e-01 2.13183001e-01 4.79851723e-01
6.24413649e-03 -1.06740797e+00 9.79423523e-01 3.41213226e-01
5.53534746e-01 -3.47871959e-01 6.78223252e-01 6.21848762e-01
-1.06001091e+00 -3.54311585e-01 -1.25224829e-01 -1.06240936e-01
-1.71624318e-01 4.91595954e-01 -6.30916893e-01 3.92902941e-01
7.45334923e-01 4.89674985e-01 -8.90005410e-01 9.23621833e-01
-6.68411493e-01 6.90380156e-01 -1.43873155e-01 1.27675369e-01
2.71597564e-01 3.52269039e-02 4.02555168e-01 9.92434263e-01
-5.70090003e-02 3.72756064e-01 3.52116585e-01 1.35016918e+00
1.05720736e-01 -1.05048791e-01 -3.01302165e-01 2.09706664e-01
5.12805641e-01 1.08061004e+00 -9.03962493e-01 -3.54369938e-01
-1.98484272e-01 6.30223274e-01 5.52304387e-01 4.41308081e-01
-1.02005863e+00 -1.05090469e-01 2.71548927e-01 -5.92997819e-02
2.71911055e-01 -5.82508780e-02 -5.00100076e-01 -1.05409074e+00
7.06093535e-02 -6.91783845e-01 2.54720420e-01 -6.00609481e-01
-1.20792449e+00 2.15085626e-01 4.26117480e-01 -8.21361661e-01
5.15627205e-01 -4.06197160e-01 -5.92009664e-01 7.95144916e-01
-1.71568668e+00 -1.25241446e+00 -4.82213616e-01 5.06611466e-01
3.74258786e-01 2.94775128e-01 4.63844121e-01 4.21004385e-01
-7.83219397e-01 5.99839866e-01 -2.59842128e-01 -2.64751771e-03
6.52614057e-01 -1.18130529e+00 -5.24406910e-01 7.43263006e-01
3.03819060e-01 8.04988325e-01 4.49915916e-01 -7.52151847e-01
-9.87740695e-01 -1.06550848e+00 6.66476369e-01 -6.20221376e-01
4.21905130e-01 -5.53148746e-01 -1.29071558e+00 4.55026865e-01
4.54586484e-02 2.26512104e-01 5.07526040e-01 2.87858009e-01
-6.37118816e-01 -2.64563769e-01 -8.24310064e-01 4.84213829e-01
1.07051837e+00 -5.23262501e-01 -8.93887997e-01 2.04140916e-01
9.28136885e-01 -1.44172922e-01 -4.12976891e-01 6.74991131e-01
3.49641919e-01 -7.68589437e-01 9.27698255e-01 -4.50744450e-01
4.66941535e-01 -5.70597231e-01 7.12899445e-03 -1.20356512e+00
-5.02389073e-01 1.21856086e-01 -4.10837792e-02 1.53245246e+00
2.89065838e-01 -5.49764216e-01 2.55524516e-01 3.24250489e-01
-8.28738958e-02 -8.13932002e-01 -1.04804981e+00 -7.57921100e-01
9.97325927e-02 -1.32889867e-01 5.07069051e-01 1.22086835e+00
-9.86659676e-02 6.35236382e-01 -7.13893697e-02 5.77913523e-01
4.90450352e-01 2.07681030e-01 3.11615109e-01 -1.11275256e+00
-1.93859056e-01 -5.91783941e-01 -2.00026035e-01 -9.89167154e-01
9.93625298e-02 -9.14789200e-01 4.36296016e-01 -1.30323648e+00
5.54207981e-01 -8.83792520e-01 -6.86870873e-01 8.97737205e-01
-6.98540390e-01 2.01964408e-01 8.98955166e-02 4.42807943e-01
-6.86534882e-01 6.33671343e-01 1.33284640e+00 -2.90739298e-01
-1.97208196e-01 -3.93885851e-01 -9.03585315e-01 7.22802699e-01
4.22035694e-01 -7.32769251e-01 -6.49194241e-01 -5.75826228e-01
8.76602829e-02 -3.80414277e-01 9.01696682e-01 -7.94627190e-01
2.89721459e-01 -3.93163711e-01 3.99392247e-01 -3.75719666e-01
-2.09487099e-02 -9.31391895e-01 -2.35224187e-01 4.46300864e-01
-5.28668463e-01 -3.73991877e-01 1.38830289e-01 9.14438665e-01
-2.40524128e-01 -1.57211661e-01 7.91511655e-01 -6.33454695e-02
-8.70395243e-01 9.98226851e-02 1.14190519e-01 9.18011665e-02
1.16636682e+00 -4.74552251e-02 -5.38107455e-01 1.99291185e-01
-5.33799231e-01 4.43955094e-01 7.66597614e-02 5.92044652e-01
4.11749989e-01 -1.27696717e+00 -5.09661376e-01 1.96201861e-01
3.26886892e-01 3.10758263e-01 3.89694810e-01 1.14897132e+00
2.40609303e-01 3.81318957e-01 -1.12425856e-01 -9.46691811e-01
-8.97750139e-01 9.01064992e-01 3.64131808e-01 7.38605298e-03
-3.20016921e-01 1.13027966e+00 1.16240883e+00 -1.26160607e-01
4.47690636e-01 -3.88517469e-01 -4.35516119e-01 2.43196681e-01
4.16732550e-01 1.57896299e-02 -2.28910029e-01 -5.62687039e-01
-5.57859182e-01 5.93992829e-01 2.58972794e-02 1.08802296e-01
9.15943325e-01 -1.52047873e-01 -2.21957266e-01 5.06836951e-01
8.96225035e-01 -1.39119253e-01 -1.49442756e+00 -3.96190584e-01
3.47721092e-02 -4.73758340e-01 2.45090604e-01 -1.16190553e+00
-1.00469887e+00 1.03269601e+00 8.63398135e-01 -1.24549374e-01
1.05878472e+00 3.07958037e-01 1.50494903e-01 2.99247485e-02
1.25216648e-01 -8.30517292e-01 4.90589142e-01 1.89904675e-01
8.62908900e-01 -1.38376081e+00 -2.28061024e-02 -4.89791095e-01
-6.36554539e-01 6.44136131e-01 1.18786705e+00 5.86252511e-02
6.44485593e-01 -7.55705088e-02 -6.06717877e-02 -2.25017041e-01
-4.26572531e-01 -3.58302712e-01 5.62490106e-01 5.57569385e-01
1.09166466e-01 2.19244286e-02 -4.12615955e-01 9.21107352e-01
3.31287891e-01 -2.78987825e-01 4.67875414e-02 7.58702517e-01
-3.65679264e-01 -6.84255779e-01 -3.69182020e-01 1.09823450e-01
-2.15940148e-01 -2.99663723e-01 -3.48193139e-01 7.24134326e-01
9.68082011e-01 7.82447100e-01 1.60563782e-01 -3.59010637e-01
6.12996221e-02 -2.38338243e-02 4.38133091e-01 -7.10150301e-01
-4.82248068e-01 1.34554133e-01 -4.82536823e-01 -2.66061038e-01
-5.90387583e-01 -2.61138141e-01 -1.30718839e+00 4.15671557e-01
-9.24170911e-01 1.96642116e-01 6.37235582e-01 1.17805314e+00
3.37489218e-01 6.89002693e-01 5.17428041e-01 -6.37759268e-01
-5.36614120e-01 -9.34962690e-01 -4.40274775e-01 3.62070352e-01
4.39102322e-01 -1.20656741e+00 -4.79142338e-01 -6.83603734e-02] | [9.775738716125488, 1.5081939697265625] |
ba25ebf1-5883-4ff4-88ef-1ddb2277f2e2 | lexical-simplification-with-the-deep | null | null | https://aclanthology.org/I17-2073 | https://aclanthology.org/I17-2073.pdf | Lexical Simplification with the Deep Structured Similarity Model | We explore the application of a Deep Structured Similarity Model (DSSM) to ranking in lexical simplification. Our results show that the DSSM can effectively capture fine-grained features to perform semantic matching when ranking substitution candidates, outperforming the state-of-the-art on two standard datasets used for the task. | ['John Lee', 'Xiaodong Liu', 'Lis Pereira'] | 2017-11-01 | lexical-simplification-with-the-deep-1 | https://aclanthology.org/I17-2073 | https://aclanthology.org/I17-2073.pdf | ijcnlp-2017-11 | ['learning-word-embeddings'] | ['methodology'] | [ 2.52550751e-01 -3.12892497e-02 -6.57482028e-01 -5.91366112e-01
-8.19623709e-01 -1.64391577e-01 7.85394549e-01 6.56519651e-01
-9.21193600e-01 4.35821414e-01 7.66034067e-01 -6.04571290e-02
-3.00220251e-01 -5.56022644e-01 -5.74794948e-01 6.19508743e-01
2.53784090e-01 1.29020977e+00 4.93471295e-01 -1.10009348e+00
5.79336464e-01 7.76967555e-02 -1.89532804e+00 8.16588759e-01
1.05134964e+00 7.37555504e-01 2.89147466e-01 -1.31020039e-01
-5.11030257e-01 7.73351252e-01 -7.18575299e-01 -1.05745268e+00
1.90230682e-01 1.19451508e-01 -1.11167991e+00 -9.91388619e-01
7.31905043e-01 -1.70206621e-01 -4.01010841e-01 1.09526241e+00
6.03318095e-01 5.12480974e-01 7.90443897e-01 -7.34407067e-01
-1.00825846e+00 1.17293823e+00 9.63131115e-02 3.87156159e-01
8.67322743e-01 -5.53124547e-01 1.65626311e+00 -1.20455539e+00
1.14380801e+00 1.87599695e+00 1.03543639e+00 4.12671596e-01
-9.31876421e-01 -5.07093012e-01 7.48895630e-02 6.65469885e-01
-1.39358020e+00 -5.71062863e-01 4.24497128e-01 5.63230291e-02
1.84748793e+00 3.72757107e-01 4.32904154e-01 9.83277202e-01
-1.51571721e-01 9.26217794e-01 7.13381410e-01 -8.03161561e-01
-2.23070592e-01 -3.97397667e-01 7.32257307e-01 5.53011715e-01
4.21753436e-01 1.97573230e-01 -9.09793556e-01 -3.97079170e-01
1.74711779e-01 -1.45701453e-01 2.49044970e-01 -1.95418701e-01
-9.58683908e-01 1.09127736e+00 4.10119742e-01 1.40000224e-01
-4.33049947e-01 1.00599982e-01 8.20716202e-01 7.64999986e-01
4.84018058e-01 1.10172188e+00 -6.34791017e-01 -1.00269221e-01
-1.15589046e+00 7.58611917e-01 6.61274374e-01 9.36594129e-01
3.70638043e-01 -2.40152851e-01 -2.83355653e-01 1.36016595e+00
2.32885052e-02 2.70171165e-01 7.96247721e-01 -9.89144921e-01
5.56254625e-01 6.40248477e-01 -1.96831807e-01 -6.62914515e-01
-6.28970027e-01 -2.26306409e-01 -1.70303762e-01 -4.52846378e-01
-3.86598021e-01 8.49089265e-01 -8.17579806e-01 1.78712440e+00
-2.83111408e-02 -2.71975491e-02 -1.16087042e-01 6.95954680e-01
1.30704641e+00 2.24149108e-01 5.16277015e-01 1.71329752e-01
1.30138242e+00 -1.04325855e+00 -6.54530108e-01 -3.44582707e-01
1.00024998e+00 -1.05941963e+00 1.43219721e+00 -8.01912844e-02
-1.37163615e+00 -7.65708506e-01 -9.95859206e-01 -6.74574375e-01
-5.96266508e-01 -3.60920846e-01 1.00446117e+00 2.04094291e-01
-8.81222725e-01 1.23471189e+00 -4.58057582e-01 -6.50770128e-01
3.39115947e-01 3.43392521e-01 -4.18862343e-01 -3.15425366e-01
-2.00622749e+00 1.67673409e+00 4.12101746e-01 -5.48048139e-01
-4.16742444e-01 -1.26513135e+00 -9.32399929e-01 -1.70540772e-02
1.45980924e-01 -1.31532669e+00 1.56231654e+00 -3.93084317e-01
-1.32213187e+00 1.31089902e+00 -2.18261749e-01 -8.05954814e-01
1.35374039e-01 -5.60491145e-01 -5.36335945e-01 -1.40836239e-01
6.64863765e-01 7.49102354e-01 3.69717628e-01 -4.91235375e-01
-6.35389626e-01 -2.00938433e-01 1.70433700e-01 5.95437765e-01
-1.73152938e-01 6.82294846e-01 -4.04379636e-01 -1.17617536e+00
-1.72762394e-01 -7.66077220e-01 -9.32712331e-02 -6.34375334e-01
-3.32438529e-01 -6.01047993e-01 1.71142906e-01 -8.48279297e-01
1.81272888e+00 -1.68719590e+00 6.26984954e-01 -4.96356469e-03
-9.54413638e-02 7.69392490e-01 -4.63022262e-01 8.33362520e-01
1.11171275e-01 2.21522272e-01 1.06024668e-02 -6.39131069e-01
4.29073781e-01 8.74517709e-02 -5.50883412e-01 -6.48458824e-02
-2.07184449e-01 1.54781783e+00 -9.41046298e-01 -3.44492406e-01
1.12145185e-01 2.50286609e-01 -7.68337071e-01 1.05814196e-01
-1.25300303e-01 -2.68176526e-01 -1.92185804e-01 6.71560764e-01
2.41317138e-01 5.13111651e-02 3.83014798e-01 -3.02739233e-01
5.33605993e-01 1.47226346e+00 -7.15348005e-01 2.19671011e+00
-7.61494100e-01 3.42893451e-01 -5.49922049e-01 -6.67829871e-01
6.09652460e-01 -1.65816858e-01 3.83002870e-02 -1.34176219e+00
-1.30395010e-01 8.08753192e-01 -3.03880908e-02 2.82095909e-01
9.78222907e-01 -7.97493011e-02 -5.48685610e-01 1.82343438e-01
2.52555221e-01 -3.82969260e-01 1.49966836e-01 4.23871070e-01
1.20928466e+00 3.00328005e-02 6.14811122e-01 -5.99604249e-01
5.12148201e-01 4.03415672e-02 3.43099803e-01 1.02243650e+00
2.74270028e-01 2.41987020e-01 -1.68815553e-01 -5.97467422e-01
-8.68515730e-01 -9.27236617e-01 3.40612419e-02 1.84857142e+00
2.64505059e-01 -9.72586453e-01 -5.60087919e-01 -5.74319839e-01
6.53627813e-01 8.31995785e-01 -3.85206193e-01 -5.10179818e-01
-8.77095103e-01 -3.49538893e-01 8.28610182e-01 4.66131926e-01
-8.97128284e-02 -1.37641132e+00 -2.93503433e-01 4.31433290e-01
-2.23935530e-01 -1.05721509e+00 -5.12819588e-01 2.73632914e-01
-7.33108401e-01 -1.01865792e+00 -2.21927285e-01 -9.16616201e-01
-1.18847102e-01 4.11455631e-01 2.00519371e+00 4.25998867e-01
-2.74958432e-01 -1.02798596e-01 -4.21047032e-01 -3.69000822e-01
-5.22496998e-01 5.99551201e-01 2.38262862e-01 -1.12455165e+00
7.23843753e-01 -4.04225439e-01 -4.04170752e-01 5.23922034e-02
-6.43335700e-01 -2.42842332e-01 2.50323385e-01 9.23771858e-01
6.80674613e-01 -2.63442904e-01 5.84863663e-01 -1.52827644e+00
1.02307928e+00 -3.07268053e-01 -1.91058502e-01 2.35457197e-01
-8.47576499e-01 3.39235783e-01 6.94711864e-01 -1.47114888e-01
-7.36545682e-01 -5.64899504e-01 -3.21459234e-01 -1.86465129e-01
3.52599055e-01 7.16197431e-01 -7.05159381e-02 -6.13157824e-02
5.50600946e-01 -4.02214825e-01 -3.56323719e-01 -8.75045717e-01
9.01839435e-01 4.22059149e-01 6.54073954e-01 -6.79863095e-01
5.16770899e-01 2.42819101e-01 1.37416482e-01 -3.17761719e-01
-1.30069137e+00 -6.01999581e-01 -6.22150242e-01 7.52559423e-01
3.63844424e-01 -9.77184236e-01 -6.77857024e-04 2.70187020e-01
-1.32844448e+00 1.43555198e-02 -4.79437262e-01 9.26655605e-02
-3.45418841e-01 4.93701279e-01 -8.93023074e-01 2.04292521e-01
-6.04885340e-01 -9.89578247e-01 1.47974992e+00 -4.91782427e-01
-1.06850076e+00 -1.25887668e+00 3.42136890e-01 4.96274024e-01
7.66585708e-01 -2.73919404e-01 1.55522227e+00 -1.21164227e+00
-1.10967018e-01 -2.70180821e-01 1.37098236e-02 -5.52990241e-03
-3.20332468e-01 -4.22335356e-01 -5.34644723e-01 1.31192803e-01
-4.21365827e-01 -4.68830049e-01 1.35635519e+00 4.14464653e-01
9.90600467e-01 -2.21364312e-02 -4.02368814e-01 9.84790266e-01
9.30916905e-01 -3.78099531e-01 5.62922060e-01 9.38850582e-01
9.52214777e-01 5.44682682e-01 1.03860974e+00 5.35241663e-02
8.91221285e-01 1.29803598e+00 2.49510899e-01 -1.38327256e-01
-6.73130870e-01 -6.52261734e-01 6.92012906e-02 1.25380075e+00
3.54371607e-01 -5.69679216e-02 -5.77260315e-01 3.10049474e-01
-2.08517933e+00 -8.63994300e-01 -1.75676830e-02 1.99279201e+00
1.05763781e+00 2.40600735e-01 -3.48811150e-02 3.39149684e-02
6.59125328e-01 4.76034731e-01 -3.16964358e-01 -1.01810074e+00
-5.68715155e-01 1.15674913e+00 4.83758420e-01 9.06462312e-01
-1.14885175e+00 1.76365077e+00 7.25607157e+00 1.33116043e+00
-4.02629554e-01 2.77360946e-01 -5.50724454e-02 9.64970961e-02
-7.78998375e-01 1.06832929e-01 -9.10611689e-01 2.64758795e-01
1.08525574e+00 -2.24304721e-01 6.94654524e-01 6.70818686e-01
-5.41721106e-01 3.74700040e-01 -1.33151221e+00 8.43654931e-01
2.21734449e-01 -1.45627439e+00 8.79072189e-01 -3.31173986e-01
6.80719376e-01 2.62864739e-01 4.08454239e-02 7.39372015e-01
5.36742568e-01 -1.30281687e+00 8.78948927e-01 4.43339705e-01
8.50065172e-01 -7.48352230e-01 7.60694027e-01 -5.79782575e-02
-1.24249685e+00 3.62019353e-02 -5.68154454e-01 -2.67917275e-01
1.26291603e-01 3.73416603e-01 -5.04844427e-01 6.44692063e-01
4.43593055e-01 1.00666714e+00 -7.42522776e-01 5.94718933e-01
-2.84223497e-01 8.38665292e-02 -1.30341858e-01 8.78310502e-02
4.70622033e-02 2.71230400e-01 6.62235439e-01 1.45675051e+00
1.63548952e-03 -3.33153874e-01 1.49834543e-01 2.58506715e-01
-4.43014055e-01 3.98621470e-01 -3.62392068e-01 8.73382911e-02
1.31036103e+00 6.59596026e-01 -2.03164235e-01 -4.95104849e-01
-3.58840585e-01 1.34839940e+00 8.95997286e-01 -1.47733942e-01
-6.84474230e-01 -5.36083877e-01 1.03925443e+00 1.05856486e-01
3.30138862e-01 2.86608875e-01 -4.58708942e-01 -1.30144823e+00
-8.82501453e-02 -1.21478355e+00 8.05609703e-01 -3.95995498e-01
-1.78223038e+00 5.93428493e-01 2.21569717e-01 -7.84031034e-01
-4.61244941e-01 -5.56336641e-01 -3.75856012e-01 9.89516258e-01
-1.85810554e+00 -1.33736444e+00 8.56950879e-02 3.64097297e-01
6.32520020e-01 -6.21134162e-01 1.14020705e+00 4.89198327e-01
-2.51193121e-02 8.88165653e-01 1.40268385e-01 -4.05784488e-01
8.44888508e-01 -1.11496997e+00 1.39734364e+00 5.39578617e-01
5.53783357e-01 1.05854499e+00 4.47259456e-01 -6.99309349e-01
-1.10157335e+00 -9.41375256e-01 1.62899816e+00 -7.90465474e-01
5.78846693e-01 -2.74751127e-01 -7.93780982e-01 4.55199778e-01
-6.51630014e-02 -2.31332824e-01 7.37867594e-01 4.11512405e-01
-8.67442250e-01 1.45926818e-01 -8.61782014e-01 5.20004272e-01
1.84236181e+00 -9.86944914e-01 -1.45052135e+00 2.90980607e-01
1.22128654e+00 -5.05811751e-01 -6.77219212e-01 7.85231709e-01
6.12093151e-01 -7.45898366e-01 1.60195541e+00 -1.25762165e+00
4.58975792e-01 8.45867023e-02 -6.02799892e-01 -1.65092766e+00
-7.49717951e-01 -7.41228819e-01 -4.89000887e-01 9.29839611e-01
4.88814533e-01 -3.89051527e-01 3.72532487e-01 1.10763900e-01
-4.77665573e-01 -7.17746615e-01 -1.06526601e+00 -9.38363552e-01
2.80703455e-01 -3.15372139e-01 1.05638242e+00 1.01435542e+00
-8.34718049e-02 7.42985010e-01 5.77777848e-02 -4.01207328e-01
2.96994150e-01 6.44735992e-02 2.95992523e-01 -1.77754176e+00
-1.74926609e-01 -8.90526891e-01 -6.40210152e-01 -8.13570321e-01
1.12397182e+00 -1.53009152e+00 -7.64690280e-01 -1.74131048e+00
3.00750345e-01 -3.44198972e-01 -6.36092722e-01 2.28877231e-01
-6.37617350e-01 3.74175787e-01 3.13747793e-01 4.21964049e-01
-1.04574406e+00 5.25884211e-01 4.63102549e-01 -1.05726130e-01
6.14504106e-02 -3.83170664e-01 -1.14688861e+00 6.67900205e-01
3.17810297e-01 -6.59626961e-01 -3.88920158e-01 -9.50491011e-01
5.14751613e-01 -3.87431979e-01 -5.90927377e-02 -4.79504287e-01
-8.09162930e-02 1.45057589e-01 -7.69067556e-02 -4.58165973e-01
4.56322074e-01 -4.26240712e-01 -3.31249759e-02 4.40923125e-01
-8.20310771e-01 5.39177656e-01 1.64405465e-01 2.28409812e-01
-3.93785387e-01 -3.48787159e-01 4.36310589e-01 -2.44436592e-01
-1.06601000e+00 -4.02482413e-02 1.54891267e-01 7.50545800e-01
2.80991524e-01 1.67458490e-01 -6.30861700e-01 -1.66670114e-01
-2.24820137e-01 9.93178561e-02 7.55369127e-01 8.19576085e-01
5.84860206e-01 -1.61856568e+00 -6.07962847e-01 -1.45410880e-01
4.53719318e-01 -4.72570986e-01 -3.52686018e-01 2.34744519e-01
-5.20344257e-01 9.27215040e-01 -1.14490323e-01 -3.48920859e-02
-1.10854685e+00 6.27538085e-01 7.80456364e-02 -8.97965193e-01
-4.18803662e-01 1.15285051e+00 -1.54062390e-01 -8.82601440e-01
3.10039163e-01 -3.50379139e-01 -2.50304848e-01 9.66262817e-03
2.81564772e-01 6.25948966e-01 6.24395072e-01 -7.85161436e-01
-7.85461068e-01 5.36216736e-01 -5.16628027e-01 -1.31341606e-01
1.44894302e+00 -2.54817396e-01 -4.69756901e-01 1.02431424e-01
1.38531125e+00 1.25442654e-01 1.78231876e-02 -7.23553777e-01
6.22696996e-01 -3.92604291e-01 -9.07269120e-03 -1.04225075e+00
-4.24670070e-01 5.78642964e-01 2.50752717e-02 -2.73413211e-01
7.48339593e-01 -3.55174057e-02 1.38172448e+00 7.75318921e-01
4.72506076e-01 -1.27867222e+00 -2.27887988e-01 1.16383314e+00
7.39459455e-01 -1.02930057e+00 1.89358205e-01 -5.83709776e-01
-1.01898217e+00 8.93772721e-01 2.95166880e-01 -4.76903915e-01
5.95114231e-01 -2.59772599e-01 -4.25276421e-02 -4.75336581e-01
-8.59262884e-01 -4.98129517e-01 9.61747348e-01 6.37418747e-01
7.87806869e-01 1.83134347e-01 -7.22121000e-01 9.20895398e-01
-5.97828448e-01 -4.63581383e-01 -1.20670609e-01 8.46344233e-01
-3.18230629e-01 -1.57893491e+00 3.49944204e-01 5.79349935e-01
-5.04518986e-01 -1.16677868e+00 -9.32546794e-01 7.93858171e-01
-5.16680777e-02 6.71889424e-01 -8.55293649e-04 -3.50310445e-01
8.41295123e-01 2.20779210e-01 7.03505039e-01 -1.08121645e+00
-9.95835662e-01 -5.71372330e-01 7.09650755e-01 -1.06768334e+00
-2.73801029e-01 -5.66600800e-01 -1.18024600e+00 -3.66164178e-01
-1.12840481e-01 -1.31556839e-01 2.69837081e-01 8.46010923e-01
5.86077869e-01 3.47310811e-01 3.93586844e-01 -7.81810164e-01
-8.77194464e-01 -9.28875804e-01 -2.06471324e-01 1.14673913e+00
-3.45288366e-02 -1.00666475e+00 -3.90621349e-02 -7.44420350e-01] | [10.925140380859375, 10.33533763885498] |
b5e1676e-f1e9-4257-862f-5d91acc00489 | post-training-quantization-on-diffusion | 2211.15736 | null | https://arxiv.org/abs/2211.15736v3 | https://arxiv.org/pdf/2211.15736v3.pdf | Post-training Quantization on Diffusion Models | Denoising diffusion (score-based) generative models have recently achieved significant accomplishments in generating realistic and diverse data. These approaches define a forward diffusion process for transforming data into noise and a backward denoising process for sampling data from noise. Unfortunately, the generation process of current denoising diffusion models is notoriously slow due to the lengthy iterative noise estimations, which rely on cumbersome neural networks. It prevents the diffusion models from being widely deployed, especially on edge devices. Previous works accelerate the generation process of diffusion model (DM) via finding shorter yet effective sampling trajectories. However, they overlook the cost of noise estimation with a heavy network in every iteration. In this work, we accelerate generation from the perspective of compressing the noise estimation network. Due to the difficulty of retraining DMs, we exclude mainstream training-aware compression paradigms and introduce post-training quantization (PTQ) into DM acceleration. However, the output distributions of noise estimation networks change with time-step, making previous PTQ methods fail in DMs since they are designed for single-time step scenarios. To devise a DM-specific PTQ method, we explore PTQ on DM in three aspects: quantized operations, calibration dataset, and calibration metric. We summarize and use several observations derived from all-inclusive investigations to formulate our method, which especially targets the unique multi-time-step structure of DMs. Experimentally, our method can directly quantize full-precision DMs into 8-bit models while maintaining or even improving their performance in a training-free manner. Importantly, our method can serve as a plug-and-play module on other fast-sampling methods, e.g., DDIM. The code is available at https://github.com/42Shawn/PTQ4DM . | ['Yan Yan', 'Bingzhe Wu', 'Bin Xie', 'Zhihang Yuan', 'Yuzhang Shang'] | 2022-11-28 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Shang_Post-Training_Quantization_on_Diffusion_Models_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Shang_Post-Training_Quantization_on_Diffusion_Models_CVPR_2023_paper.pdf | cvpr-2023-1 | ['noise-estimation'] | ['medical'] | [ 2.55684555e-01 -5.35517693e-01 -1.61497459e-01 -1.09198533e-01
-1.04994392e+00 -4.38350320e-01 4.35452193e-01 -2.38094747e-01
-2.84407943e-01 4.93935317e-01 2.93903768e-01 -3.66380602e-01
-1.45466208e-01 -9.59455431e-01 -5.17689645e-01 -9.35892999e-01
2.54336923e-01 2.19213411e-01 2.68452764e-02 -1.98539227e-01
1.71792358e-01 3.38647813e-01 -1.16870224e+00 2.65831321e-01
9.41077828e-01 9.91053820e-01 4.60307628e-01 8.64292026e-01
2.79591009e-02 7.48006761e-01 -8.48209620e-01 -5.40678918e-01
3.93431604e-01 -8.32495570e-01 -2.60064602e-01 -1.03201933e-01
5.93191162e-02 -6.33331358e-01 -7.99997211e-01 1.23812819e+00
9.91145670e-01 -1.60110459e-01 4.51056123e-01 -9.88275290e-01
-7.64039278e-01 7.87283897e-01 -3.99226576e-01 3.15374315e-01
-5.99866100e-02 4.75273043e-01 6.80313587e-01 -7.85095572e-01
5.36051571e-01 9.57872272e-01 8.95541191e-01 7.21899390e-01
-1.29627943e+00 -7.86973298e-01 -2.93667555e-01 1.09683856e-01
-1.57292557e+00 -8.14621449e-01 9.70361054e-01 -1.01232074e-01
7.89716005e-01 2.95423895e-01 6.16696417e-01 1.38904142e+00
6.85173869e-02 8.41963530e-01 8.99841070e-01 -1.50118321e-01
5.89487970e-01 -3.61253351e-01 -2.99100518e-01 3.12329322e-01
2.83702701e-01 1.60455927e-01 -6.60364747e-01 -1.87852785e-01
1.01816583e+00 -5.63297905e-02 -3.87246192e-01 -6.04584217e-02
-8.99208188e-01 7.65864909e-01 1.05911590e-01 2.61493325e-01
-4.10720825e-01 5.47655463e-01 4.89499748e-01 5.14243424e-01
4.68773097e-01 9.29471850e-02 -1.51646763e-01 -7.01918721e-01
-1.54016924e+00 2.67119050e-01 6.92635655e-01 9.74709451e-01
4.71299559e-01 4.90134984e-01 -3.34354132e-01 7.52219856e-01
8.51427391e-02 6.03985667e-01 5.91214061e-01 -1.23191535e+00
5.53006768e-01 2.48316973e-02 -2.26605922e-01 -9.92418468e-01
-6.11950085e-02 -7.26920009e-01 -1.50444591e+00 -5.51461689e-02
1.92425236e-01 -2.92880893e-01 -7.98161268e-01 1.57801771e+00
1.26661316e-01 2.14774489e-01 3.00718900e-02 7.81151354e-01
3.58806789e-01 7.35931396e-01 -3.40597421e-01 -3.50932896e-01
8.64460647e-01 -8.68031681e-01 -8.46634328e-01 5.17938510e-02
5.77405334e-01 -8.07176769e-01 1.00541878e+00 9.37852740e-01
-1.51717854e+00 -6.59864843e-01 -1.19164205e+00 -1.44331619e-01
1.20990425e-01 1.48578227e-01 4.89195555e-01 8.50732625e-01
-1.21707964e+00 9.72474515e-01 -9.89859402e-01 1.26858100e-01
6.61942184e-01 1.97533682e-01 2.91432261e-01 -2.68459290e-01
-1.08631766e+00 4.00312990e-01 -3.67218032e-02 2.21113011e-01
-1.06883824e+00 -6.13226295e-01 -5.24148285e-01 7.72403628e-02
1.81773633e-01 -8.50519001e-01 1.34455276e+00 -8.61919105e-01
-1.86729038e+00 2.59775817e-01 -2.13683993e-01 -7.18986213e-01
8.05832148e-01 -7.51627162e-02 -6.41475379e-01 1.85110644e-01
-1.31108239e-01 2.93621540e-01 1.17137682e+00 -1.01704395e+00
-4.02430028e-01 -8.75760466e-02 -1.68209612e-01 -8.33545998e-02
-5.10209978e-01 -2.96345204e-01 -7.55416691e-01 -1.08951521e+00
1.44178793e-01 -5.87050676e-01 -5.05573273e-01 1.71952829e-01
-3.04772198e-01 1.82003722e-01 8.20174336e-01 -5.85664272e-01
1.81358755e+00 -2.38366747e+00 5.19553237e-02 2.10073918e-01
4.45354998e-01 5.63479125e-01 -1.78011253e-01 4.83262956e-01
2.15859309e-01 1.99925929e-01 -3.54107887e-01 -7.89951801e-01
9.83245596e-02 2.37082332e-01 -3.48219573e-01 3.68642479e-01
1.79664731e-01 8.96026790e-01 -9.43622589e-01 -2.62833267e-01
-9.19484347e-03 6.93910182e-01 -8.90005231e-01 2.95588449e-02
-9.19693485e-02 3.89619529e-01 -1.82001695e-01 6.11431897e-01
8.23734701e-01 -2.47829840e-01 1.65932760e-01 -4.41115648e-01
8.99837259e-03 3.91647846e-01 -1.28505361e+00 2.01507068e+00
-4.71986622e-01 6.77911937e-01 2.13593915e-01 -9.94462669e-01
9.24570680e-01 2.14428410e-01 3.70068997e-01 -8.73453736e-01
1.27421886e-01 5.21401584e-01 -1.38275445e-01 -3.81708711e-01
7.73413241e-01 -8.14928934e-02 3.78488451e-02 4.08011705e-01
-6.34617032e-03 -3.29922825e-01 3.76451612e-01 2.08268568e-01
1.22379911e+00 -1.11854255e-01 -1.47126466e-01 9.03397426e-02
6.99543208e-02 -3.85536224e-01 6.18709147e-01 9.05163527e-01
-1.82718813e-01 1.01449704e+00 4.22822267e-01 1.60517059e-02
-1.37050998e+00 -1.11298001e+00 -1.24376528e-01 5.04920483e-01
5.62407449e-02 -7.40015507e-01 -8.14814091e-01 -3.59764814e-01
-2.56958395e-01 6.94126725e-01 -1.63930565e-01 -3.96709204e-01
-5.98103940e-01 -9.46080387e-01 1.00770712e+00 5.15001595e-01
6.55170858e-01 -5.35408497e-01 -3.35630119e-01 4.37448919e-01
-8.07490721e-02 -7.39490807e-01 -5.96100330e-01 2.81333834e-01
-1.02883577e+00 -5.22657335e-01 -9.53331232e-01 -4.78686035e-01
4.05078620e-01 2.61085182e-01 9.44286883e-01 -3.63456602e-05
1.44614652e-01 6.43712059e-02 -2.32512221e-01 -1.23718090e-01
-6.86458468e-01 1.95816845e-01 -9.93790850e-03 -1.39401898e-01
3.80413890e-01 -1.03621268e+00 -9.36944842e-01 1.47043034e-01
-1.23169565e+00 3.37240100e-02 6.82085693e-01 8.13145399e-01
6.59493089e-01 3.43803614e-01 5.56611598e-01 -5.53614795e-01
1.03214526e+00 -5.43849826e-01 -4.81403172e-01 -1.13245398e-01
-8.10697615e-01 3.93620580e-02 7.26570666e-01 -5.85030079e-01
-7.18859792e-01 -3.28241706e-01 -5.98564863e-01 -7.58946359e-01
3.40179116e-01 4.67132658e-01 -9.35209692e-02 1.31478831e-01
8.11345220e-01 5.45589089e-01 5.90759777e-02 -4.86920148e-01
4.00988072e-01 8.64460111e-01 4.70829070e-01 -5.71872294e-01
8.23939443e-01 4.56864595e-01 -2.03169182e-01 -5.93662977e-01
-3.58849704e-01 -9.78037268e-02 -3.99921238e-02 2.54388731e-02
4.03674304e-01 -1.04477429e+00 -4.04681683e-01 5.80257356e-01
-1.00419712e+00 -5.16872883e-01 -6.16810322e-01 2.52748340e-01
-4.69627142e-01 5.83884835e-01 -1.04104269e+00 -6.92533791e-01
-5.40349305e-01 -1.23947155e+00 9.71451998e-01 -3.83407250e-02
-1.91373408e-01 -7.82144606e-01 -1.65855676e-01 1.19762390e-03
9.68929052e-01 -1.48262531e-01 5.92641950e-01 -1.15481637e-01
-7.39647567e-01 -1.32241935e-01 -7.54178390e-02 7.63343573e-01
-6.57090843e-02 6.71693459e-02 -8.61215234e-01 -4.30147111e-01
4.44336325e-01 -1.03493810e-01 9.90971088e-01 4.77256984e-01
1.43330002e+00 -3.33599061e-01 -4.77230437e-02 1.02388477e+00
1.45181668e+00 1.62004247e-01 9.22439396e-01 2.11688820e-02
4.11983848e-01 -1.37494266e-01 1.01694331e-01 6.82692170e-01
2.05986381e-01 4.82914299e-01 1.61750197e-01 -7.42482319e-02
-5.33968925e-01 -3.53987396e-01 5.11222363e-01 1.40024829e+00
-5.46890758e-02 -4.84819353e-01 -6.31468534e-01 4.18183059e-01
-1.43765581e+00 -9.34753597e-01 -4.34470810e-02 2.20730615e+00
1.10327792e+00 4.51382309e-01 -9.66874659e-02 4.75375473e-01
4.22556996e-01 2.31175929e-01 -6.95007503e-01 -1.96050614e-01
-1.89359933e-01 4.45176005e-01 4.80593860e-01 3.18845958e-01
-8.13857973e-01 4.36914176e-01 5.84994984e+00 1.38799167e+00
-1.38574409e+00 4.23966348e-01 8.55318010e-01 -4.40161645e-01
-5.05264342e-01 -1.96219295e-01 -8.28125417e-01 8.51074159e-01
1.20834196e+00 -2.49085680e-01 6.22219324e-01 8.08286965e-01
4.44082558e-01 1.18193969e-01 -9.56694186e-01 1.28487504e+00
-1.72070473e-01 -1.51321781e+00 1.07391253e-01 7.19673058e-04
8.71001363e-01 1.84284195e-01 3.14668298e-01 2.84110934e-01
1.58359811e-01 -7.92468846e-01 8.95648003e-01 5.23204923e-01
1.12923408e+00 -6.88387096e-01 4.88513082e-01 3.79819125e-01
-9.78389919e-01 -1.61399662e-01 -5.00928283e-01 -4.59947102e-02
4.58188146e-01 1.32146060e+00 -2.86494672e-01 4.87845987e-01
5.46650648e-01 6.28790975e-01 -2.03378439e-01 1.00142169e+00
-1.01378426e-01 9.17040765e-01 -4.25348431e-01 1.68255761e-01
1.49521679e-01 -2.73834914e-01 5.01067162e-01 1.17655349e+00
8.79174650e-01 -1.05405942e-01 -2.54450083e-01 1.05454993e+00
-2.07034409e-01 -3.19684893e-01 -3.48768741e-01 -1.10617779e-01
6.77207708e-01 8.83062363e-01 -4.40929085e-01 -4.19526249e-01
-2.16781065e-01 1.17484057e+00 -5.32122031e-02 4.78390783e-01
-1.02329135e+00 -3.15731913e-01 7.30986297e-01 1.07330829e-01
5.07445991e-01 -5.87394714e-01 -5.29560566e-01 -1.31156373e+00
1.59085140e-01 -1.09853554e+00 -1.72450975e-01 -5.05753040e-01
-1.36849236e+00 5.83931565e-01 -3.52368951e-01 -1.45893848e+00
-2.11636692e-01 -6.25270307e-02 -3.64659846e-01 8.27040792e-01
-1.45607734e+00 -6.56431139e-01 -3.06613684e-01 3.49322081e-01
6.96446598e-01 8.33632350e-02 5.78121364e-01 8.80304992e-01
-6.29064739e-01 8.39216113e-01 5.68691254e-01 -1.12748981e-01
6.19244337e-01 -8.44553590e-01 6.91266656e-01 1.00172150e+00
-2.63217334e-02 6.66037679e-01 5.52876532e-01 -6.38574183e-01
-1.54854524e+00 -1.08722889e+00 6.28285050e-01 -1.53689206e-01
5.51307499e-01 -4.53964978e-01 -8.24982524e-01 1.55120686e-01
9.25684571e-02 -5.05189300e-02 3.81499559e-01 -3.06282580e-01
-1.94536988e-02 -2.73666710e-01 -1.02530253e+00 8.36056590e-01
1.23135459e+00 -6.57589495e-01 1.01008177e-01 1.06250636e-01
7.38545299e-01 -6.67040288e-01 -9.13239539e-01 1.31554171e-01
2.98710614e-01 -1.14485264e+00 9.91001189e-01 5.47988415e-02
7.93052971e-01 -3.00862908e-01 -2.63694227e-01 -1.16190600e+00
-3.17725748e-01 -1.13754845e+00 -6.79513752e-01 1.34696412e+00
2.08316505e-01 -4.69249487e-01 8.60817432e-01 2.84331799e-01
-2.09939510e-01 -9.15735424e-01 -1.01230574e+00 -8.00207675e-01
7.78359100e-02 -8.21879327e-01 6.93303168e-01 7.69430101e-01
-4.83386815e-01 1.61188319e-01 -4.95774001e-01 1.91458706e-02
5.42707264e-01 -2.48264298e-01 6.63969278e-01 -6.00373566e-01
-7.38151431e-01 -6.13488078e-01 -9.56664085e-02 -1.70747387e+00
-6.66990995e-01 -7.41537333e-01 -5.21166958e-02 -1.17394209e+00
-1.45401403e-01 -7.11677492e-01 -5.66421151e-02 -8.84668902e-02
-9.08610690e-03 2.72765219e-01 2.06467122e-01 5.13676167e-01
-3.25092286e-01 7.83068895e-01 1.28417730e+00 -1.41498998e-01
-1.46564111e-01 2.48739379e-03 -7.16749847e-01 4.75848109e-01
9.09335792e-01 -5.65276265e-01 -6.57652974e-01 -8.08825254e-01
4.02261436e-01 3.94960791e-02 2.71538466e-01 -1.45918226e+00
4.96217072e-01 2.02475443e-01 2.30970919e-01 -4.39261794e-01
3.16903353e-01 -7.02176869e-01 4.67651606e-01 4.83468145e-01
-9.10891518e-02 1.73401877e-01 -6.77963793e-02 5.92806756e-01
-2.88262963e-01 -3.58274847e-01 6.23474777e-01 8.16859305e-03
-4.85817373e-01 3.98125827e-01 -2.92358071e-01 9.38775316e-02
5.36617279e-01 -2.98365951e-01 -2.40585074e-01 -5.46447396e-01
-5.05836606e-01 -2.13057548e-01 5.70325315e-01 -4.25473191e-02
5.78071654e-01 -1.34238684e+00 -6.08508348e-01 5.11528909e-01
-4.80961680e-01 4.26026493e-01 4.12512839e-01 9.22033846e-01
-5.16736865e-01 -1.42917475e-02 2.52878159e-01 -5.42137086e-01
-6.89785540e-01 4.66210455e-01 2.71863878e-01 -4.16120499e-01
-5.72368503e-01 8.26846659e-01 -2.74164557e-01 3.38275619e-02
2.71141678e-01 -4.95213032e-01 5.57635128e-01 -9.75534841e-02
6.13848925e-01 4.48702514e-01 2.04718664e-01 -1.31699850e-03
2.86798030e-01 3.33849758e-01 6.68799505e-02 -7.65021071e-02
1.32582116e+00 -2.01601312e-01 7.05955923e-02 1.75345346e-01
1.29200709e+00 1.25072628e-01 -1.45029962e+00 -2.13035986e-01
-3.90206903e-01 -3.21089715e-01 2.67599165e-01 -4.50413942e-01
-1.32761610e+00 9.55039024e-01 6.96847975e-01 2.09744602e-01
1.56760406e+00 -4.25299972e-01 1.33471811e+00 4.47010957e-02
3.60592067e-01 -1.22598195e+00 1.74015969e-01 3.56436044e-01
6.75209820e-01 -7.53877759e-01 -4.47093695e-02 -2.56279171e-01
-2.25678921e-01 9.65394557e-01 1.97181240e-01 3.16211279e-03
5.92077792e-01 7.32475340e-01 -2.55964547e-02 -3.43096703e-02
-7.45934606e-01 2.47201845e-01 -1.93786666e-01 5.57996809e-01
1.86346114e-01 -1.85390636e-01 -4.08628941e-01 7.15600014e-01
-2.25611567e-01 4.68386322e-01 4.60390806e-01 9.94317472e-01
-1.77978143e-01 -1.22432637e+00 -2.50402510e-01 5.28230190e-01
-3.53664279e-01 -4.51630235e-01 3.31645936e-01 3.12128723e-01
2.49406397e-01 7.83575594e-01 7.93309733e-02 -6.08894348e-01
2.05369502e-01 -2.39517674e-01 2.78044403e-01 -1.08693853e-01
-5.00070214e-01 2.11547658e-01 6.33965135e-02 -5.21329403e-01
-2.61516958e-01 -5.60427308e-01 -8.85502696e-01 -8.32351446e-01
-2.66021550e-01 -8.74804258e-02 6.36108637e-01 5.32207966e-01
7.56375492e-01 6.58616424e-01 5.97654045e-01 -8.66467297e-01
-1.06515455e+00 -8.59687746e-01 -4.75911856e-01 2.65192777e-01
2.95688599e-01 -2.20635474e-01 -3.37035000e-01 1.84532127e-03] | [11.248915672302246, -0.49389275908470154] |
ff29cb22-346b-417c-bd9e-70559cbbbd1e | geometric-signatures-of-switching-behavior-in | 2209.03324 | null | https://arxiv.org/abs/2209.03324v3 | https://arxiv.org/pdf/2209.03324v3.pdf | Geometric Signatures of Switching Behavior in Mechanobiology | The proteins involved in cells' mechanobiological processes have evolved specialized and surprising responses to applied forces. Biochemical transformations that show catch-to-slip switching and force-induced pathway switching serve important functions in cell adhesion, mechano-sensing and signaling, and protein folding. We show that these switching behaviors are generated by singularities in the flow field that describes force-induced deformation of bound and transition states. These singularities allow for a complete characterization of switching mechanisms in 2-dimensional (2D) free energy landscapes, and provide a path toward elucidating novel forms of switching in higher dimensional models. Remarkably, the singularity that generates a catch-slip switch occurs in almost every 2D free energy landscape, implying that almost any bond admitting a 2D model will exhibit catch-slip behavior under appropriate force. We apply our analysis to models of P-selectin and antigen extraction to illustrate how these singularities provide an intuitive framework for explaining known behaviors and predicting new behaviors. | ['Robijn F. Bruinsma', 'Casey O. Barkan'] | 2022-09-07 | null | null | null | null | ['protein-folding'] | ['natural-language-processing'] | [ 5.43486476e-01 -2.22878337e-01 -7.33140349e-01 -1.46640524e-01
-3.41824561e-01 -7.54585803e-01 4.23796475e-01 3.51428688e-01
-1.67313918e-01 1.04628980e+00 -3.04288417e-02 -5.70771873e-01
-2.54227221e-01 -8.41234744e-01 -8.43971193e-01 -1.03007543e+00
-6.38163924e-01 4.05663192e-01 4.87713695e-01 -6.60188019e-01
4.51052278e-01 7.23252833e-01 -9.58420038e-01 2.87741005e-01
7.77446091e-01 4.56179112e-01 -2.36931324e-01 1.00329721e+00
-2.89797395e-01 -2.10225150e-01 -1.99493468e-02 -2.84022450e-01
1.82771608e-01 -5.24708927e-01 -7.79155076e-01 -3.31233919e-01
4.45026338e-01 2.93070465e-01 -1.79176971e-01 5.87057829e-01
2.80838132e-01 -1.98316313e-02 1.33084774e+00 -6.77016795e-01
-5.46838760e-01 -1.24840625e-01 -4.01476890e-01 4.13187742e-01
5.98412693e-01 6.14329129e-02 9.48286951e-01 -6.94232166e-01
1.24020720e+00 1.02232373e+00 5.03569841e-01 8.99351597e-01
-1.71672678e+00 -1.45351112e-01 -8.68013203e-02 -3.14648718e-01
-7.80394435e-01 -3.74723315e-01 7.26462185e-01 -7.73794413e-01
1.01669669e+00 4.68971997e-01 9.15780783e-01 9.83887136e-01
1.20387256e+00 2.31113225e-01 9.24369872e-01 -4.30423796e-01
2.89028585e-01 -4.96897608e-01 5.07214129e-01 9.08510506e-01
7.91680574e-01 1.96455106e-01 -8.29892516e-01 -7.04767883e-01
9.12179470e-01 8.49911049e-02 -3.71368200e-01 -5.42374909e-01
-9.40406919e-01 3.58611822e-01 -1.70623381e-02 3.69984776e-01
-3.45940031e-02 3.23781759e-01 5.18708639e-02 3.60043466e-01
-2.31136948e-01 5.53306520e-01 -4.69195545e-01 -1.45275310e-01
-2.68456340e-01 3.42408180e-01 8.88746083e-01 2.46485710e-01
8.75995636e-01 -5.95349669e-01 4.14436013e-01 5.07713974e-01
-8.69925693e-03 5.63893139e-01 2.29998752e-01 -7.26887107e-01
2.92105526e-01 7.51107156e-01 3.46285790e-01 -9.11680222e-01
-7.46836841e-01 1.01774847e-02 -6.04151249e-01 1.27673864e-01
9.64439034e-01 -1.53852524e-02 -6.66597247e-01 1.91957450e+00
1.34883225e-01 -2.36005142e-01 1.31896347e-01 8.07108939e-01
2.47536212e-01 3.64643037e-01 2.96385318e-01 -5.58118165e-01
1.32967627e+00 6.55482486e-02 -4.57499355e-01 5.53580709e-02
6.77126110e-01 -4.80737865e-01 1.05126822e+00 -7.92129710e-03
-1.19730449e+00 5.42584062e-02 -1.11551869e+00 1.90784466e-02
-3.48818392e-01 -7.55599916e-01 6.78017080e-01 3.39957774e-01
-4.46700335e-01 1.07678509e+00 -1.14954329e+00 -7.64836371e-01
1.37396351e-01 4.07771796e-01 -3.88376474e-01 3.72447908e-01
-1.17883193e+00 6.24946535e-01 -3.63882601e-01 1.93743091e-02
-3.16917002e-01 -4.95600313e-01 -2.67022133e-01 -1.56145498e-01
-1.11737698e-01 -8.71513188e-01 8.27606857e-01 -2.56787449e-01
-1.37433946e+00 1.01226354e+00 -5.92379689e-01 -1.71443760e-01
2.52199024e-01 -4.91120890e-02 -3.78731877e-01 1.60814673e-01
-1.40924543e-01 3.78468744e-02 1.65186018e-01 -1.07004094e+00
3.64212389e-03 -4.75597113e-01 -4.21682507e-01 2.22596806e-02
9.24617052e-02 -4.62886363e-01 4.42909390e-01 -3.29353958e-01
4.89603281e-01 -1.04899168e+00 -4.98382032e-01 -4.09648716e-02
-5.52670419e-01 1.40383363e-01 4.79527175e-01 1.37502626e-01
1.25807095e+00 -1.70416844e+00 4.45590377e-01 5.80756068e-01
3.59901845e-01 -9.37362388e-02 1.20626830e-01 1.09770942e+00
1.71593964e-01 4.92813617e-01 -4.47686523e-01 7.70491481e-01
-2.05323398e-01 -4.19251667e-03 -3.60374451e-01 5.18717587e-01
2.38759145e-01 9.94532645e-01 -8.57028067e-01 -2.86755085e-01
-2.26432532e-01 4.08326834e-01 -6.57204330e-01 -2.34571785e-01
-1.58451647e-01 5.50819099e-01 -9.29408312e-01 7.94518113e-01
6.43479168e-01 -3.95283401e-01 8.34994137e-01 -1.09155566e-01
-2.76437372e-01 1.04049176e-01 -6.33530319e-01 1.02753937e+00
1.77665576e-01 4.33184355e-01 -7.78062716e-02 -7.97792852e-01
8.04018617e-01 -3.32197663e-03 4.87951279e-01 -5.01075029e-01
-5.45273311e-02 6.91822052e-01 2.11739600e-01 -5.56894839e-01
-1.81596979e-01 -7.18723476e-01 -3.89231026e-01 3.18135500e-01
-3.58438462e-01 -3.12798172e-02 2.28947639e-01 6.35597706e-02
1.23279977e+00 -1.12859033e-01 8.26565176e-02 -8.39783549e-01
5.66060007e-01 1.48387417e-01 7.01705396e-01 7.73683131e-01
-4.54506457e-01 1.24692045e-01 1.10613823e+00 -8.83317351e-01
-9.40682530e-01 -1.25646818e+00 -4.26827997e-01 7.06440449e-01
6.87502205e-01 -1.21125154e-01 -5.79110801e-01 7.47907087e-02
5.86900711e-01 -1.31019264e-01 -6.38438582e-01 -5.09502053e-01
-9.48223412e-01 -1.21983850e+00 4.75236237e-01 2.81356037e-01
1.42375454e-01 -4.63892937e-01 -6.85160697e-01 4.47650313e-01
6.80358782e-02 -7.08346486e-01 -5.19105196e-01 2.54613310e-01
-1.08148754e+00 -1.45917344e+00 -3.67387146e-01 -5.68700492e-01
8.35140586e-01 4.87864874e-02 6.01512432e-01 1.12802349e-01
-8.19875479e-01 -1.99537333e-02 3.21517110e-01 1.71257406e-01
-4.51226562e-01 8.52161124e-02 4.34797525e-01 -2.25845963e-01
3.64916921e-01 -7.59636283e-01 -1.12228310e+00 7.29342878e-01
-7.35222995e-01 -3.82444799e-01 1.83707237e-01 7.36777902e-01
8.65390003e-01 -6.53706610e-01 7.11643159e-01 -6.91094816e-01
7.54313052e-01 -2.77663916e-01 -4.69759613e-01 2.35442057e-01
-4.41325694e-01 4.82799381e-01 7.79225886e-01 -3.37518185e-01
-1.18268514e+00 6.03523217e-02 3.97899710e-02 6.09658659e-01
-4.06220444e-02 2.29836777e-01 7.56141590e-03 -4.47511733e-01
8.42032731e-01 1.89333752e-01 2.11152121e-01 -3.96956980e-01
-5.16130030e-03 2.22294360e-01 1.53015912e-01 -9.43086624e-01
5.92999756e-01 9.37714279e-01 1.00098240e+00 -9.95060980e-01
-4.11032736e-01 5.69813550e-02 -6.17467701e-01 -6.12195581e-02
8.15930307e-01 -4.64311987e-02 -1.36101806e+00 3.89814705e-01
-7.49452591e-01 -5.09518564e-01 -9.01088044e-02 3.62554252e-01
-8.20808947e-01 3.80789727e-01 -9.27435279e-01 -7.10577369e-01
1.80074558e-01 -9.77225125e-01 8.55577767e-01 4.90986049e-01
-3.89831960e-01 -1.23435402e+00 7.94977248e-01 3.74501199e-02
5.21464109e-01 6.32409036e-01 1.64776516e+00 -4.19926494e-01
-6.07337832e-01 -1.76778123e-01 4.29619253e-01 -5.46960354e-01
6.85773343e-02 4.40013826e-01 -3.29176515e-01 -1.31284535e-01
-5.23078501e-01 1.98082045e-01 1.16352355e+00 6.93544030e-01
4.53753620e-01 -3.20510566e-02 -1.06189597e+00 7.28543222e-01
1.45405364e+00 7.25208044e-01 3.35924596e-01 3.50563973e-02
1.65013403e-01 5.41689336e-01 1.58017874e-01 -1.48993433e-01
-2.94466317e-01 6.10878706e-01 -1.07719012e-01 -1.45759568e-01
1.15184464e-01 -3.30108643e-01 1.67689279e-01 4.28070843e-01
-4.04990107e-01 2.57937107e-02 -1.05038369e+00 1.43311694e-01
-1.66073847e+00 -9.65640783e-01 -3.45726967e-01 2.55655265e+00
9.57473040e-01 5.31060696e-01 1.58559874e-01 -3.43116730e-01
5.45806825e-01 -7.25236014e-02 -9.83959794e-01 -6.57186449e-01
-6.24249160e-01 2.56044149e-01 6.00814581e-01 1.09045899e+00
-6.16829991e-01 8.62752378e-01 8.41006279e+00 2.77697057e-01
-1.23890388e+00 -3.77132863e-01 4.99956578e-01 8.32478479e-02
-5.81536829e-01 4.46739078e-01 -6.66370094e-01 2.39021420e-01
6.46927953e-01 -2.69402713e-01 -2.19758786e-02 1.54798433e-01
1.94226608e-01 -2.68454701e-01 -9.64356720e-01 3.84328455e-01
-7.47962832e-01 -1.90596974e+00 2.66417801e-01 3.17067474e-01
6.24272645e-01 -4.81301069e-01 -9.90501791e-02 -4.44010496e-01
6.28796965e-02 -7.94186652e-01 -2.69122660e-01 7.63811707e-01
1.04719830e+00 -3.99587572e-01 1.87654272e-01 5.50234579e-02
-1.08480740e+00 3.18564773e-02 -2.30004102e-01 -2.39314362e-01
5.94704211e-01 7.77785540e-01 -3.87234926e-01 1.11638241e-01
1.32836372e-01 3.10975790e-01 1.66195244e-01 4.84191477e-01
5.30652285e-01 3.87647122e-01 -4.08387244e-01 -3.12420130e-01
-2.42404327e-01 -5.57694495e-01 8.48259389e-01 1.01410627e+00
-7.04560578e-02 3.52775365e-01 -4.23115492e-01 5.68892598e-01
6.07859306e-02 5.52850105e-02 -8.24708581e-01 -3.06080222e-01
3.87054116e-01 6.85812354e-01 -1.04783046e+00 -6.45812005e-02
-2.49404367e-02 6.68875039e-01 2.61972278e-01 6.55319512e-01
-3.92175227e-01 -8.36371541e-01 1.32714307e+00 7.50324130e-01
-5.08415140e-02 -5.59388936e-01 -1.47180006e-01 -1.34382367e+00
-1.43868506e-01 -5.57648502e-02 1.87736705e-01 3.02281901e-02
-1.14413989e+00 4.53595482e-02 -1.40977263e-01 -6.03402019e-01
1.48191512e-01 -9.08961654e-01 -5.24422646e-01 4.10244018e-01
-1.18397880e+00 -4.74335253e-01 2.58116983e-02 2.07673058e-01
-3.48057859e-02 2.34300688e-01 6.63689733e-01 6.04335889e-02
-6.63559020e-01 2.90798038e-01 3.77826542e-01 -3.07542354e-01
5.66685081e-01 -9.79776263e-01 4.31563795e-01 3.90734673e-01
-5.36125779e-01 1.03206980e+00 9.85767066e-01 -1.03932011e+00
-1.92965400e+00 -2.70657539e-01 8.56331706e-01 -6.85628593e-01
7.06815839e-01 -4.97040033e-01 -8.04363728e-01 4.55636740e-01
-5.05626678e-01 -1.05964445e-01 1.00023830e+00 -2.52262093e-02
-1.13638580e-01 -3.43111111e-03 -1.00283945e+00 9.39967930e-01
1.57793188e+00 -4.30389673e-01 -2.20170587e-01 4.08716768e-01
2.32283294e-01 -1.78038180e-01 -7.79638350e-01 4.49382842e-01
1.14680052e+00 -9.63537753e-01 7.85894334e-01 -1.44274962e+00
2.28593051e-01 -1.25218868e-01 -1.19947910e-01 -7.80978501e-01
-3.96669060e-01 -1.38738203e+00 -1.45257562e-02 3.03038955e-01
6.09040082e-01 -1.01036692e+00 8.30703795e-01 7.36727536e-01
1.38382122e-01 -1.17633820e+00 -1.32291734e+00 -6.06292665e-01
5.95712364e-01 3.67061198e-01 3.25214624e-01 6.50787652e-01
7.49690115e-01 2.24182129e-01 2.32675478e-01 -4.42063093e-01
7.47421503e-01 2.15555906e-01 5.33427417e-01 -1.19949770e+00
2.75398251e-02 -6.49772823e-01 -4.08788979e-01 -1.18422902e+00
-1.86128631e-01 -7.21791267e-01 -4.23092484e-01 -1.28633106e+00
3.91566008e-01 -3.95673543e-01 -2.33170941e-01 -4.24654316e-03
1.83670655e-01 -1.16083153e-01 -3.11092466e-01 6.70605958e-01
-9.90063995e-02 1.50647789e-01 1.38667822e+00 4.12793547e-01
-5.87015033e-01 -2.17650115e-01 -7.34020293e-01 6.78034365e-01
6.77894294e-01 -3.10074508e-01 9.38350558e-02 2.71577209e-01
8.77129972e-01 3.59245777e-01 1.44974053e-01 -4.56981987e-01
-1.17855228e-01 -7.69271195e-01 1.65924996e-01 -6.56145364e-02
2.46768624e-01 -4.28230703e-01 3.50334376e-01 1.14227641e+00
-4.56335217e-01 -1.37254298e-01 -2.23225847e-01 8.06544840e-01
1.67236447e-01 4.21154380e-01 9.50643659e-01 -2.87815859e-03
-4.31196839e-02 6.25546351e-02 -1.02793348e+00 1.12016469e-01
1.27868044e+00 -6.10134780e-01 -9.57171738e-01 3.44667671e-04
-1.21524131e+00 -5.32996655e-02 9.86279428e-01 -1.45922810e-01
3.78735691e-01 -1.14447534e+00 -1.79499522e-01 4.96625453e-01
-1.82918966e-01 -5.18635273e-01 3.79724056e-01 8.54338527e-01
-1.02121019e+00 5.12066245e-01 -4.52579290e-01 -5.68378389e-01
-7.85301030e-01 1.07611358e-01 9.76736844e-01 -2.87902296e-01
-1.98171452e-01 7.33702958e-01 2.89736092e-01 -9.16572213e-02
-6.48378730e-01 -3.87939692e-01 2.63558090e-01 -3.50978255e-01
6.13482343e-03 5.31035662e-01 -1.95820227e-01 -4.82625216e-01
-8.40660453e-01 1.45332944e+00 -5.24749085e-02 8.52846652e-02
9.29693580e-01 -5.36815915e-03 -3.48483890e-01 4.62981313e-01
9.51302409e-01 2.29734406e-01 -1.24912858e+00 2.74568826e-01
4.42894585e-02 -3.06504697e-01 -8.18001330e-01 -3.90814960e-01
-4.95732874e-01 7.22171426e-01 1.25270769e-01 4.68975991e-01
5.64889073e-01 3.94339770e-01 1.12381196e+00 5.81451178e-01
5.01515210e-01 -9.47319627e-01 6.09631762e-02 3.04129213e-01
5.55361152e-01 -5.82967997e-01 -1.88732952e-01 -6.17704809e-01
-1.54267311e-01 1.13333392e+00 5.57456374e-01 -4.45189685e-01
1.01131535e+00 3.73882383e-01 -4.12869789e-02 -3.24037254e-01
-1.14983487e+00 3.73200476e-01 -4.87729684e-02 4.61958438e-01
6.21307373e-01 1.78334221e-01 -1.15336347e+00 2.27613434e-01
1.37102038e-01 -1.74245983e-01 5.96161485e-01 1.21266413e+00
-9.63121533e-01 -1.50736558e+00 -4.25152443e-02 3.55802804e-01
-3.86132449e-01 4.06151682e-01 -7.88760126e-01 7.25922704e-01
-4.56265092e-01 3.08825940e-01 5.09887785e-02 -2.65134573e-01
5.51842690e-01 4.88776177e-01 7.09919333e-01 -1.01331085e-01
-5.80889881e-02 1.26329288e-01 1.64414681e-02 -6.63734019e-01
-1.13540158e-01 -6.34324789e-01 -1.81769502e+00 -5.32656372e-01
-1.67690009e-01 2.56114155e-01 2.37743512e-01 6.41823232e-01
9.55907822e-01 -4.29392383e-02 3.73929679e-01 -6.56490147e-01
-4.39155430e-01 -1.88581675e-01 -6.78145409e-01 7.30570793e-01
6.93050742e-01 -8.85785520e-01 -6.16448104e-01 8.89109671e-02] | [4.861719131469727, 5.186827659606934] |
5da06e85-ba96-4b34-9069-b4f1fc11ef40 | learning-multi-agent-intention-aware | 2307.03119 | null | https://arxiv.org/abs/2307.03119v1 | https://arxiv.org/pdf/2307.03119v1.pdf | Learning Multi-Agent Intention-Aware Communication for Optimal Multi-Order Execution in Finance | Order execution is a fundamental task in quantitative finance, aiming at finishing acquisition or liquidation for a number of trading orders of the specific assets. Recent advance in model-free reinforcement learning (RL) provides a data-driven solution to the order execution problem. However, the existing works always optimize execution for an individual order, overlooking the practice that multiple orders are specified to execute simultaneously, resulting in suboptimality and bias. In this paper, we first present a multi-agent RL (MARL) method for multi-order execution considering practical constraints. Specifically, we treat every agent as an individual operator to trade one specific order, while keeping communicating with each other and collaborating for maximizing the overall profits. Nevertheless, the existing MARL algorithms often incorporate communication among agents by exchanging only the information of their partial observations, which is inefficient in complicated financial market. To improve collaboration, we then propose a learnable multi-round communication protocol, for the agents communicating the intended actions with each other and refining accordingly. It is optimized through a novel action value attribution method which is provably consistent with the original learning objective yet more efficient. The experiments on the data from two real-world markets have illustrated superior performance with significantly better collaboration effectiveness achieved by our method. | ['Tie-Yan Liu', 'Yong Yu', 'Weinan Zhang', 'Dongsheng Li', 'Jiang Bian', 'Li Zhao', 'Weiqing Liu', 'Kan Ren', 'Zhenggang Tang', 'Yuchen Fang'] | 2023-07-06 | null | null | null | null | ['reinforcement-learning-1'] | ['methodology'] | [-2.61217296e-01 2.29762256e-01 -4.24486965e-01 -1.75470158e-01
-5.76488197e-01 -6.84190750e-01 4.88614768e-01 8.45579356e-02
-5.06205201e-01 1.04410434e+00 -9.58924443e-02 -1.67648554e-01
-5.25160491e-01 -8.97280335e-01 -6.09374583e-01 -9.09785807e-01
-4.55975652e-01 1.08034039e+00 -3.02470196e-02 -2.77527630e-01
4.63628888e-01 8.42155218e-02 -7.91556060e-01 -6.39951453e-02
8.21458399e-01 1.00574601e+00 2.66697615e-01 5.30205786e-01
8.01737010e-02 1.20926356e+00 -7.58325338e-01 -3.60591322e-01
7.32278705e-01 -4.86270308e-01 -4.87713426e-01 5.08662760e-01
-7.09895551e-01 -8.59069288e-01 -4.68100607e-02 1.07878458e+00
2.09950000e-01 9.22667608e-02 3.19116265e-01 -1.58110034e+00
-7.38779664e-01 1.18117964e+00 -6.56527638e-01 -5.36470069e-03
9.46143419e-02 2.19063058e-01 1.47309625e+00 1.39000854e-02
3.58666599e-01 9.82976317e-01 -1.54070646e-01 4.43790764e-01
-8.37215900e-01 -7.86635816e-01 5.27146518e-01 2.52652109e-01
-7.73067236e-01 7.54991323e-02 8.37005794e-01 -1.48822621e-01
6.51071668e-01 2.53649175e-01 9.20216620e-01 4.30676311e-01
4.31967378e-01 9.69596505e-01 1.33578885e+00 -1.31115645e-01
4.18701500e-01 2.26469800e-01 -1.47660688e-01 3.46643031e-01
2.82660633e-01 4.89096284e-01 -5.09824872e-01 -2.09897965e-01
1.01267624e+00 1.88029334e-01 -4.23891060e-02 -2.72584081e-01
-1.29450715e+00 9.33164239e-01 2.10325181e-01 1.45151854e-01
-8.31020534e-01 -5.55735640e-02 3.81942131e-02 8.96583676e-01
3.40818465e-01 5.85921109e-01 -6.30708992e-01 -1.59504473e-01
-5.69687188e-01 6.75509572e-01 1.10577345e+00 8.25820565e-01
6.93830192e-01 2.28737369e-01 -7.30719119e-02 4.17256057e-02
6.02112889e-01 1.49989918e-01 4.03495401e-01 -1.29972208e+00
7.11040437e-01 6.27885818e-01 6.73151731e-01 -8.98327589e-01
-3.45747560e-01 -5.11909783e-01 -6.71341658e-01 5.14550984e-01
2.79369295e-01 -4.93663371e-01 -7.35336542e-02 1.45855474e+00
3.22335273e-01 1.73107222e-01 4.00988728e-01 1.03342247e+00
-1.25598339e-02 7.10357070e-01 -1.61556497e-01 -9.19979632e-01
1.16753197e+00 -1.21318591e+00 -9.40655947e-01 1.63081750e-01
4.59226757e-01 -4.94804174e-01 5.51674008e-01 7.96024621e-01
-1.11405563e+00 -1.88678339e-01 -1.04559028e+00 7.35590041e-01
3.86220925e-02 -1.24516308e-01 8.56605589e-01 4.65087146e-01
-7.57876873e-01 6.38901949e-01 -7.66275167e-01 4.94447142e-01
2.19813377e-01 7.59239435e-01 1.58703372e-01 6.25557542e-01
-1.14004934e+00 6.04798138e-01 4.63650614e-01 2.75570601e-01
-9.99238014e-01 -5.69748938e-01 -2.97930807e-01 3.37539576e-02
1.00904250e+00 -2.89038092e-01 1.57770920e+00 -1.14066422e+00
-2.05285144e+00 1.80581257e-01 3.08001399e-01 -7.90422261e-01
8.65621924e-01 -3.11058313e-01 -3.59565407e-01 -4.02968787e-02
-4.80965935e-02 2.18289912e-01 6.10953689e-01 -1.41534257e+00
-1.08121610e+00 -1.70710057e-01 6.49545789e-01 6.22420132e-01
-2.67207831e-01 -2.20322341e-01 6.59116507e-02 -6.03432298e-01
-5.96914254e-02 -7.95230806e-01 -5.34903765e-01 -5.75202525e-01
-2.33475700e-01 -4.94230747e-01 5.96716523e-01 -5.52377582e-01
1.12150919e+00 -1.80199158e+00 2.04483598e-01 3.05128843e-01
3.80295277e-01 -2.23057225e-01 5.55564091e-02 6.34571910e-01
3.92049670e-01 5.94745576e-02 -1.19258769e-01 -2.41519958e-01
4.47438955e-01 4.10940528e-01 -4.38875347e-01 4.65984911e-01
-1.26744911e-01 8.56743157e-01 -8.98533523e-01 -4.47544128e-01
1.05240397e-01 -2.74179101e-01 -6.00382447e-01 7.65645146e-01
-5.27706981e-01 7.96170413e-01 -1.01254916e+00 6.98810220e-01
5.72647393e-01 -3.03137273e-01 5.04247129e-01 1.75654754e-01
-1.08041503e-01 -1.96153745e-01 -1.63984609e+00 1.04555798e+00
-2.49051228e-01 -1.93146735e-01 3.22554588e-01 -1.11016119e+00
8.88157368e-01 5.90381801e-01 9.38284814e-01 -7.51520574e-01
7.60093518e-03 3.75919908e-01 2.86816210e-01 -3.76677454e-01
3.03139001e-01 -1.82659104e-01 4.21259232e-04 1.18837285e+00
-3.94115061e-01 2.73251850e-02 4.40188855e-01 7.65094236e-02
7.59404123e-01 2.72789299e-01 4.41605330e-01 -2.51048021e-02
6.24924779e-01 -1.04062058e-01 8.79201353e-01 8.48546982e-01
-2.57572621e-01 -1.03528075e-01 7.21015513e-01 -5.82817435e-01
-7.86056936e-01 -8.13401461e-01 4.74727362e-01 9.47602630e-01
4.62578833e-01 -2.55597048e-02 -5.67900181e-01 -8.53524566e-01
9.92169604e-02 5.71698070e-01 -2.45934829e-01 2.29218885e-01
-7.26085663e-01 -8.80475163e-01 -1.18560441e-01 1.53710783e-01
8.19973290e-01 -1.41212022e+00 -8.69004071e-01 6.07084990e-01
1.31674722e-01 -8.27980220e-01 -7.33729303e-01 1.12403445e-01
-6.47455394e-01 -1.25282717e+00 -4.84410703e-01 -4.59630787e-01
7.79382527e-01 -1.99364945e-01 8.37334216e-01 2.43387535e-01
3.85167032e-01 2.77215421e-01 -2.88221389e-01 -3.93610477e-01
-3.65720183e-01 1.47851214e-01 1.47416636e-01 4.27097678e-01
1.31087571e-01 -3.53615791e-01 -6.37919664e-01 2.13512197e-01
-9.32087719e-01 -2.07721572e-02 8.11359048e-01 7.43200839e-01
4.62370127e-01 6.05394542e-01 8.23278129e-01 -1.03156233e+00
1.05888522e+00 -4.06590700e-01 -1.09961152e+00 3.79977793e-01
-1.04395127e+00 1.16774261e-01 8.11511874e-01 -3.86470407e-01
-1.16163516e+00 -2.30572239e-01 5.34024894e-01 -1.56212553e-01
1.93178564e-01 4.20267642e-01 -2.12477729e-01 -5.16351983e-02
-2.28280529e-01 4.10123020e-01 1.62840724e-01 -1.56242132e-01
-6.28703237e-02 5.53149104e-01 4.62895110e-02 -7.16138005e-01
9.42856371e-01 8.71226937e-02 1.60673797e-01 -1.28440261e-01
-6.40129268e-01 1.05519015e-02 -3.77772212e-01 -3.91961902e-01
4.81871605e-01 -5.91056228e-01 -1.80646920e+00 4.34520811e-01
-9.96962428e-01 -2.52809972e-01 -2.15042964e-01 7.62617409e-01
-8.24302077e-01 2.28649914e-01 -7.74003923e-01 -1.10768843e+00
-9.94788185e-02 -1.31756604e+00 5.48565865e-01 3.96601915e-01
1.53473839e-01 -1.10789812e+00 2.86080390e-01 4.32680935e-01
3.46059680e-01 9.13726836e-02 7.38893688e-01 -1.04546368e+00
-1.20606816e+00 4.35332917e-02 1.75800025e-01 1.46199822e-01
3.26178670e-01 -3.10483366e-01 -2.49682575e-01 -5.14970839e-01
3.25267404e-01 -3.63078415e-01 2.31409609e-01 2.41535455e-01
8.03932607e-01 -9.26639557e-01 5.33890165e-02 2.05743387e-01
1.46549749e+00 8.52149248e-01 1.12884073e-02 7.74639368e-01
2.76545465e-01 7.18578577e-01 1.12570786e+00 1.00702059e+00
5.77027380e-01 3.63362312e-01 7.03522086e-01 1.73109174e-01
7.36157954e-01 -1.25331342e-01 4.84767765e-01 8.72351885e-01
-3.49298686e-01 -3.74886274e-01 -3.53254408e-01 4.40733768e-02
-2.37543654e+00 -8.93165827e-01 4.06720161e-01 2.09753251e+00
9.19900954e-01 2.72779047e-01 1.81831628e-01 -1.12602472e-01
5.56596875e-01 2.39463300e-01 -8.65909278e-01 -2.27625787e-01
-4.93012853e-02 -1.95300102e-01 8.00186992e-01 7.35027492e-01
-8.80146801e-01 6.89909399e-01 5.99346399e+00 6.09761953e-01
-7.02793181e-01 -3.14802304e-03 7.45005667e-01 -1.44338652e-01
-4.56512958e-01 2.29553744e-01 -7.74188221e-01 4.64780778e-01
6.60576046e-01 -5.25125861e-01 1.11536551e+00 5.90454876e-01
5.77866077e-01 1.85876209e-02 -1.15256906e+00 7.46424139e-01
-3.90623629e-01 -1.33238447e+00 1.32618338e-01 2.85692424e-01
8.31439257e-01 -5.22832096e-01 -1.23187669e-01 2.88023680e-01
7.10354507e-01 -7.21733332e-01 9.18300033e-01 5.85818887e-01
-1.83609188e-01 -1.07777882e+00 8.90693426e-01 6.30815446e-01
-1.25774062e+00 -3.44142944e-01 -1.12261355e-01 -3.02674055e-01
2.17084795e-01 1.45436376e-01 -5.41826010e-01 9.75510657e-01
3.33399057e-01 7.20123827e-01 -2.79505234e-02 5.33548594e-01
-3.30770910e-01 3.04883242e-01 -9.03271139e-02 -3.86382580e-01
4.49469954e-01 -9.00201976e-01 5.04726946e-01 2.60926485e-01
1.19995050e-01 2.76719779e-01 8.69176328e-01 7.81880736e-01
-1.80392131e-01 1.04896456e-01 -1.92227751e-01 -2.90614218e-01
4.84445333e-01 1.25596571e+00 -6.61233425e-01 -3.51975322e-01
-2.96044618e-01 6.62662148e-01 2.92666107e-01 4.93902624e-01
-8.87171865e-01 -2.12951928e-01 2.82280296e-01 -2.36296013e-01
1.21909931e-01 -3.96042407e-01 -4.22967076e-02 -9.38917279e-01
3.68966013e-01 -1.11696625e+00 4.95125085e-01 -1.49118572e-01
-1.12824821e+00 3.02651793e-01 -1.71068624e-01 -1.41880238e+00
-4.55227375e-01 -3.06680351e-01 -5.00170767e-01 4.91514713e-01
-1.91067088e+00 -8.23546231e-01 3.78846407e-01 5.27668118e-01
6.52638316e-01 -6.10564232e-01 3.63719612e-01 3.86568978e-02
-4.73061472e-01 1.10177211e-01 1.49440989e-01 2.61302348e-02
3.77109110e-01 -1.45992601e+00 -4.42801267e-01 4.49676335e-01
1.02411970e-01 2.63880074e-01 6.47383511e-01 -8.44605625e-01
-1.62204087e+00 -8.17007184e-01 5.85835099e-01 1.17240660e-01
9.98095274e-01 4.32561003e-02 -5.89015603e-01 9.32293415e-01
6.34482086e-01 -4.69793737e-01 3.20943594e-01 -3.12307179e-01
3.62232924e-01 -5.59556901e-01 -1.05962086e+00 7.25638628e-01
6.04055107e-01 1.02565236e-01 -5.36144495e-01 5.38185060e-01
9.40059483e-01 -3.66267025e-01 -1.01072276e+00 1.54952601e-01
2.75019467e-01 -8.60008121e-01 4.42725033e-01 -5.78740001e-01
7.59353340e-02 -3.83603781e-01 1.12655498e-01 -1.27785802e+00
-1.09826989e-01 -1.16912711e+00 -3.58263165e-01 1.19271684e+00
3.64894062e-01 -1.15100074e+00 9.41070914e-01 7.12308109e-01
1.69532925e-01 -6.78512156e-01 -1.00569212e+00 -7.97415733e-01
-2.22344995e-01 1.66204706e-01 1.04933202e+00 7.47393668e-01
1.86631791e-02 -1.48427904e-01 -7.69814312e-01 3.86130065e-01
1.05920708e+00 5.50493717e-01 6.80278242e-01 -8.66010070e-01
-8.37376416e-01 -4.92801696e-01 2.80376554e-01 -1.02815866e+00
3.06621075e-01 -5.45146286e-01 -3.19591649e-02 -1.09161663e+00
1.37766987e-01 -5.38603902e-01 -5.37008524e-01 2.43856072e-01
1.90488458e-01 -4.90305841e-01 4.58300859e-01 5.59481561e-01
-9.34930682e-01 6.03770316e-01 1.73288405e+00 -2.57552207e-01
-4.54766333e-01 3.58447850e-01 -6.45316362e-01 4.90388662e-01
9.86431122e-01 -5.03809452e-01 -4.63244706e-01 -2.20326453e-01
2.67075032e-01 5.98762989e-01 1.01352058e-01 -5.13135850e-01
5.17142415e-01 -8.05439353e-01 2.32091732e-03 -5.40894389e-01
9.63659957e-03 -1.07461357e+00 3.33377481e-01 1.01820505e+00
-7.10951924e-01 3.06158066e-01 -6.04276359e-01 6.88536584e-01
-3.75276774e-01 -4.55897689e-01 3.40752065e-01 -5.12094557e-01
-5.84751070e-01 6.63866937e-01 -1.63593888e-01 -6.36005998e-02
1.56078684e+00 2.42578268e-01 -1.59322962e-01 -4.82617080e-01
-5.87407351e-01 9.10558045e-01 -5.77057339e-02 3.30664128e-01
4.41647798e-01 -1.18221951e+00 -8.76434326e-01 1.79583523e-02
-4.94454026e-01 -5.80839552e-02 3.90388668e-02 6.21754587e-01
-3.57378751e-01 5.36445022e-01 -2.34462604e-01 -2.69880354e-01
-8.33020449e-01 5.34235179e-01 4.67711926e-01 -9.14190352e-01
-3.51286113e-01 1.82562247e-01 6.95252344e-02 -4.37537640e-01
2.20251217e-01 -2.27538973e-01 -4.59653050e-01 6.47415817e-02
3.17883968e-01 4.67097998e-01 -3.12163740e-01 -8.24620947e-02
8.11666623e-02 4.82070684e-01 -3.31113309e-01 -3.68523180e-01
1.53989351e+00 -3.11350703e-01 -2.53262103e-01 5.17680764e-01
5.52915096e-01 -1.53710201e-01 -1.55766034e+00 -2.88507462e-01
3.45174491e-01 -5.65036595e-01 -1.53978512e-01 -6.50369167e-01
-1.42395914e+00 1.07261121e-01 1.29982799e-01 7.71265209e-01
1.00183046e+00 -2.25530207e-01 6.85130537e-01 5.46976328e-01
8.57273817e-01 -1.48211169e+00 2.35888824e-01 5.83524778e-02
7.63939738e-01 -1.23821926e+00 9.01990831e-02 5.56854792e-02
-9.29985225e-01 1.24176407e+00 7.27604449e-01 -4.04442340e-01
5.74688196e-01 3.09557736e-01 1.79058723e-02 -6.00202046e-02
-1.06186271e+00 1.75273821e-01 -3.90588462e-01 3.49387914e-01
-2.98459083e-02 3.85995507e-01 -5.68855941e-01 8.07282031e-01
-2.33919863e-02 -1.80314891e-02 7.81692922e-01 1.21229351e+00
-5.04048288e-01 -1.52119315e+00 -3.87627095e-01 4.39638257e-01
-4.88994539e-01 2.62565821e-01 -1.44941419e-01 8.88181686e-01
-1.69236258e-01 9.94688153e-01 5.81741072e-02 6.18454739e-02
3.03049207e-01 -3.51603270e-01 1.62561208e-01 -3.91427189e-01
-9.40150201e-01 3.60975534e-01 -2.31897950e-01 -4.94752228e-01
-7.37978280e-01 -7.39227295e-01 -1.53336012e+00 -3.68200421e-01
-1.63916469e-01 5.80374122e-01 1.64846078e-01 1.09327352e+00
2.57276949e-02 6.30637050e-01 1.45145845e+00 -5.43296158e-01
-1.40355492e+00 -5.49363494e-01 -1.20033979e+00 1.68160528e-01
5.27217448e-01 -5.78815937e-01 -5.91142237e-01 -1.07279204e-01] | [4.354043483734131, 3.6243700981140137] |
ecc8c2d1-c243-42c2-8b01-8e88727e6f01 | avsegformer-audio-visual-segmentation-with | 2307.01146 | null | https://arxiv.org/abs/2307.01146v2 | https://arxiv.org/pdf/2307.01146v2.pdf | AVSegFormer: Audio-Visual Segmentation with Transformer | The combination of audio and vision has long been a topic of interest in the multi-modal community. Recently, a new audio-visual segmentation (AVS) task has been introduced, aiming to locate and segment the sounding objects in a given video. This task demands audio-driven pixel-level scene understanding for the first time, posing significant challenges. In this paper, we propose AVSegFormer, a novel framework for AVS tasks that leverages the transformer architecture. Specifically, we introduce audio queries and learnable queries into the transformer decoder, enabling the network to selectively attend to interested visual features. Besides, we present an audio-visual mixer, which can dynamically adjust visual features by amplifying relevant and suppressing irrelevant spatial channels. Additionally, we devise an intermediate mask loss to enhance the supervision of the decoder, encouraging the network to produce more accurate intermediate predictions. Extensive experiments demonstrate that AVSegFormer achieves state-of-the-art results on the AVS benchmark. The code is available at https://github.com/vvvb-github/AVSegFormer. | ['Tong Lu', 'Wenhai Wang', 'Guo Chen', 'Zhe Chen', 'Shengyi Gao'] | 2023-07-03 | null | null | null | null | ['scene-understanding'] | ['computer-vision'] | [ 2.69522399e-01 -1.78004488e-01 -5.49213924e-02 -3.64061683e-01
-1.08984077e+00 -4.73885506e-01 2.88834870e-01 2.52577700e-02
-2.26048335e-01 1.51929989e-01 2.64612347e-01 9.67434868e-02
2.56432414e-01 -4.45964605e-01 -8.04050922e-01 -6.10533893e-01
1.69475913e-01 3.13360281e-02 6.63212478e-01 7.66492710e-02
1.79452598e-01 1.99743927e-01 -1.68353868e+00 7.44467080e-01
7.43892372e-01 1.34509552e+00 6.15181029e-01 6.74804866e-01
1.52880788e-01 7.30683386e-01 -2.35208228e-01 -1.82978481e-01
1.28900215e-01 -3.09687704e-01 -7.80626833e-01 6.04874752e-02
7.24200964e-01 -2.28553087e-01 -5.15920877e-01 9.73036766e-01
5.36463678e-01 2.66255260e-01 4.03535932e-01 -1.37802875e+00
-3.92330110e-01 6.49111629e-01 -6.86444759e-01 4.32183802e-01
1.90518737e-01 4.47508723e-01 1.45903683e+00 -1.13645589e+00
2.81906694e-01 1.28900850e+00 1.74030870e-01 3.78514022e-01
-1.22636735e+00 -6.61588967e-01 5.48276484e-01 6.59163058e-01
-1.46028614e+00 -6.61614597e-01 1.01336348e+00 -4.02969152e-01
5.04615724e-01 3.98968160e-01 7.77591825e-01 9.38470483e-01
-1.93389967e-01 1.41259360e+00 7.22349763e-01 -2.09772542e-01
1.47371277e-01 -7.20566809e-02 -1.32188439e-01 4.68316793e-01
-4.07111198e-01 -8.98061991e-02 -1.00040829e+00 2.12593764e-01
5.80623209e-01 -7.92962834e-02 -5.50093949e-01 -3.43669504e-01
-1.11626267e+00 6.60119057e-01 6.05121911e-01 8.09682980e-02
-3.56568158e-01 4.55221087e-01 3.68733376e-01 -4.52255532e-02
4.18813556e-01 1.23348169e-01 -3.50807667e-01 -7.86134675e-02
-1.14317572e+00 1.63157836e-01 2.56245911e-01 6.56598389e-01
5.07211030e-01 -3.61670256e-02 -6.32436275e-01 1.02887416e+00
5.99831462e-01 3.04858685e-01 4.41740230e-02 -1.32949698e+00
5.29818296e-01 3.61228257e-01 -1.93292931e-01 -7.73201168e-01
-1.85890898e-01 -5.72744906e-01 -6.76326692e-01 -1.38848368e-02
5.95152713e-02 2.15582848e-01 -9.20229435e-01 1.59594512e+00
5.02558947e-01 6.11121058e-01 -4.00811553e-01 1.28358507e+00
9.09921169e-01 9.61660266e-01 7.09930807e-02 -2.62036757e-03
1.38079822e+00 -1.33914387e+00 -4.07395482e-01 -3.27834547e-01
-7.85612389e-02 -7.05441594e-01 1.35996878e+00 4.09557223e-01
-1.10542142e+00 -6.89160347e-01 -7.67648816e-01 -2.44475067e-01
1.19770303e-01 4.58606929e-02 3.30094457e-01 1.06939748e-01
-1.09922349e+00 1.47775248e-01 -1.07638347e+00 8.10527653e-02
6.32538378e-01 1.49429724e-01 1.83480382e-01 -5.72176911e-02
-1.10862064e+00 1.06639601e-01 1.43357962e-01 2.71763057e-01
-1.36623991e+00 -8.06445360e-01 -8.99465442e-01 1.54210165e-01
6.79490983e-01 -6.59009516e-01 1.48940063e+00 -1.01817250e+00
-1.36913252e+00 6.23459458e-01 -4.02348697e-01 -4.21800196e-01
5.48400283e-01 -2.22331718e-01 -2.51803726e-01 5.68413675e-01
2.47876674e-01 1.14134657e+00 1.13554776e+00 -1.30774546e+00
-8.41659606e-01 -3.29592466e-01 3.65161784e-02 3.03546965e-01
-3.02411199e-01 2.95496266e-02 -1.10349131e+00 -8.79610121e-01
1.43299222e-01 -7.47107863e-01 -9.31227803e-02 2.46767312e-01
-5.71856260e-01 -2.24621072e-01 7.17604816e-01 -5.17690182e-01
1.12596321e+00 -2.43136692e+00 2.74521172e-01 6.66878605e-03
4.13444489e-01 1.60774261e-01 -3.72168720e-01 1.00711636e-01
1.48852706e-01 -1.24583781e-01 -3.16292584e-01 -5.56083202e-01
-9.62957740e-02 -1.79777801e-01 -4.52499002e-01 2.30198294e-01
2.78590798e-01 8.68775547e-01 -7.44489312e-01 -6.51570678e-01
2.34455228e-01 6.36183798e-01 -8.67838144e-01 2.30549887e-01
-4.30517823e-01 6.04854882e-01 -4.84734178e-01 8.39225590e-01
5.09957552e-01 -3.92539740e-01 -2.97332525e-01 -2.71608591e-01
-1.25282139e-01 2.92905957e-01 -1.01945126e+00 1.83319080e+00
-4.61581022e-01 8.32056642e-01 3.10833931e-01 -8.71383488e-01
5.75317502e-01 1.38051137e-01 5.00376165e-01 -7.67497003e-01
4.99790460e-02 5.72178848e-02 -2.86322355e-01 -5.31879127e-01
4.46823508e-01 3.21886122e-01 3.68057564e-02 2.26041544e-02
8.37762803e-02 -5.83142340e-02 1.63578317e-01 2.40843207e-01
8.04884732e-01 -2.18469277e-02 -2.50489801e-01 3.00401807e-01
6.78690434e-01 -2.79543877e-01 5.89912832e-01 5.04271388e-01
-3.03311199e-01 9.74890590e-01 4.06199008e-01 1.81498937e-02
-4.60037977e-01 -1.24512696e+00 -8.04920122e-02 1.46774960e+00
4.09589916e-01 -4.27202880e-01 -6.21656418e-01 -4.88665491e-01
-1.10111780e-01 5.85462034e-01 -4.30089623e-01 -2.17382424e-02
-3.10832858e-01 -3.05039823e-01 4.48494464e-01 4.98457611e-01
4.21836555e-01 -9.42562878e-01 -6.80282950e-01 1.24741793e-01
-6.66399181e-01 -1.15035450e+00 -8.01362753e-01 6.44136444e-02
-5.61938226e-01 -8.62212181e-01 -9.13240671e-01 -8.42182636e-01
3.30044895e-01 5.19002438e-01 9.30833220e-01 -1.81463018e-01
-2.75240153e-01 5.12592196e-01 -3.80009621e-01 -2.18921751e-01
-6.01306632e-02 3.40724647e-01 -4.74014342e-01 5.00567436e-01
-8.52680579e-03 -7.49152720e-01 -1.02723157e+00 3.18673849e-01
-8.47013474e-01 3.25033844e-01 4.94162023e-01 6.72598481e-01
1.05809307e+00 -3.92542750e-01 5.28315187e-01 -5.48066497e-01
2.56547660e-01 -3.28386575e-01 -7.19650447e-01 2.18960419e-02
-2.36704513e-01 -1.86261579e-01 4.76541668e-01 -3.23529989e-01
-8.24530721e-01 3.69506866e-01 -4.02958095e-01 -7.39640236e-01
-1.28215283e-01 3.26520443e-01 -4.05488074e-01 6.22657537e-02
2.65719563e-01 3.75016749e-01 -2.23898977e-01 -5.86605430e-01
5.13648450e-01 8.07972372e-01 7.76414931e-01 -2.95008421e-01
4.68933076e-01 6.16075277e-01 -3.40268970e-01 -8.60283792e-01
-9.40405965e-01 -7.06842780e-01 -4.11667407e-01 -6.01916373e-01
8.64565432e-01 -1.26427913e+00 -7.31246293e-01 4.84421521e-01
-1.14909101e+00 -5.49273968e-01 -2.60761052e-01 2.79339641e-01
-5.75982273e-01 1.38123676e-01 -5.04062891e-01 -6.15032017e-01
-2.59737968e-01 -1.51815140e+00 1.38696623e+00 2.74562389e-01
-2.81630773e-02 -4.42897618e-01 -1.09645739e-01 6.91018462e-01
1.68180376e-01 -9.67850834e-02 5.31092644e-01 -4.19878066e-01
-9.93159056e-01 1.85410455e-01 -3.78572375e-01 2.74264008e-01
-1.63419425e-01 -4.15255837e-02 -1.41784704e+00 -1.58388168e-01
-2.72587329e-01 -3.73022556e-01 1.46125340e+00 5.75095892e-01
1.65336430e+00 -7.24551976e-02 -2.37403050e-01 8.52572143e-01
1.04537439e+00 8.09213445e-02 3.40652466e-01 1.49576753e-01
9.26746845e-01 4.55124944e-01 7.64925003e-01 5.65315723e-01
7.23159254e-01 9.16775107e-01 7.43789077e-01 -2.45002329e-01
-3.72869730e-01 -3.55119973e-01 3.52440804e-01 6.76767707e-01
3.17933828e-01 -3.26471329e-01 -9.29450691e-01 6.57950401e-01
-1.83029115e+00 -6.46294773e-01 -7.33367056e-02 1.91140676e+00
9.67475116e-01 2.78433878e-02 1.93490714e-01 1.58130527e-01
7.25671589e-01 6.79732800e-01 -7.80201018e-01 1.33917823e-01
7.85348937e-02 1.22478433e-01 1.45246059e-01 5.90751827e-01
-1.38647246e+00 1.04519498e+00 5.00330496e+00 9.80906010e-01
-1.36558318e+00 2.83325821e-01 7.72587359e-01 -5.22352576e-01
-3.19192052e-01 -8.40650424e-02 -6.44074261e-01 5.26661992e-01
6.19177997e-01 1.74296666e-02 4.01867688e-01 6.93746746e-01
5.65440953e-01 -1.19987793e-01 -9.71633792e-01 1.11557806e+00
8.11801404e-02 -1.22796714e+00 2.21803010e-01 -2.07230359e-01
5.24880767e-01 2.18350172e-01 3.63661647e-01 -6.53057694e-02
-1.35437682e-01 -7.44829655e-01 1.14339161e+00 3.95320088e-01
6.51788473e-01 -7.57791996e-01 1.89174041e-01 -4.58310768e-02
-1.56977272e+00 -1.22204691e-01 -8.35758168e-03 3.74005139e-01
2.87945807e-01 5.44248343e-01 -6.31062746e-01 3.14056396e-01
9.81273472e-01 7.97131777e-01 -6.34855092e-01 1.48506784e+00
-3.74294966e-01 8.04798722e-01 -2.72613436e-01 2.27856100e-01
1.79945260e-01 1.39123380e-01 7.85822511e-01 1.25638080e+00
6.87563494e-02 -1.35813296e-01 2.65009582e-01 9.65059698e-01
-1.66397706e-01 1.58793911e-01 -7.03517497e-02 6.23536184e-02
5.05189300e-01 1.25454676e+00 -7.49855161e-01 -6.88956305e-02
-3.39700848e-01 1.11052442e+00 5.47635555e-02 4.49952960e-01
-1.03377736e+00 -3.64395052e-01 6.81973934e-01 2.20335662e-01
5.59087753e-01 1.34921474e-02 -1.98369458e-01 -1.03530264e+00
1.87641919e-01 -7.22289205e-01 4.16923493e-01 -8.38102460e-01
-7.59923637e-01 4.57836360e-01 -2.30739370e-01 -1.28463125e+00
-1.35934558e-02 -2.08340272e-01 -4.68224823e-01 5.34222066e-01
-1.69702744e+00 -1.17163777e+00 -4.88698810e-01 6.73983753e-01
9.63318765e-01 1.30990341e-01 1.57850549e-01 5.89308262e-01
-4.77772921e-01 5.72824180e-01 -1.28627390e-01 1.51545092e-01
7.17945755e-01 -9.59180117e-01 2.93798685e-01 8.89876127e-01
4.62841183e-01 -4.43354715e-03 6.52715385e-01 -2.68902242e-01
-1.27585781e+00 -1.42950010e+00 5.72046459e-01 -3.41332108e-02
5.39681613e-01 -4.98410702e-01 -9.86062944e-01 4.07530457e-01
1.20362841e-01 2.84862995e-01 4.40825045e-01 -2.13851482e-01
-4.41078693e-01 -4.59989578e-01 -5.07649183e-01 4.96292830e-01
9.87637341e-01 -7.84985781e-01 -1.26511440e-01 2.20176652e-01
1.02503943e+00 -4.75389808e-01 -3.49917471e-01 2.42000118e-01
4.66719359e-01 -9.37815368e-01 1.06402695e+00 -1.21375002e-01
4.64745522e-01 -5.88519573e-01 -1.26934126e-01 -1.12516022e+00
-1.08107388e-01 -4.93699759e-01 -1.47430137e-01 1.36554408e+00
4.68661696e-01 -2.76683837e-01 5.70883274e-01 6.23488203e-02
-3.40830296e-01 -8.09056342e-01 -1.01677120e+00 -4.91294622e-01
-4.97204661e-01 -8.01019430e-01 2.63728559e-01 5.54197609e-01
-3.31242472e-01 3.39526504e-01 -3.17298383e-01 3.75630766e-01
5.10499358e-01 2.77589202e-01 6.36567652e-01 -1.08467436e+00
-5.77415407e-01 -4.83857632e-01 -3.70258689e-01 -1.67407286e+00
9.92027819e-02 -9.22314167e-01 3.67906094e-01 -1.63401556e+00
1.33783415e-01 -1.15168586e-01 -4.80749160e-01 5.51250875e-01
-1.64743751e-01 5.19580305e-01 4.97276336e-01 1.52751088e-01
-1.08484709e+00 8.55007946e-01 1.34912169e+00 -4.19264793e-01
-4.08477843e-01 2.30798706e-01 -7.97809422e-01 6.06929243e-01
6.50624216e-01 -4.15795505e-01 -5.32421887e-01 -6.23638391e-01
6.60345405e-02 1.87200189e-01 7.51757801e-01 -9.92482901e-01
5.01185894e-01 -1.14753604e-01 1.55378848e-01 -8.41088593e-01
6.89548612e-01 -5.72741985e-01 -1.67200387e-01 1.75262854e-01
-6.06122911e-01 -2.11457804e-01 2.17045069e-01 6.92663848e-01
-6.38003469e-01 1.89431116e-01 9.04296458e-01 1.56861335e-01
-8.62034500e-01 5.59007823e-01 -2.44470894e-01 2.57652581e-01
8.70127738e-01 2.14637369e-02 -5.13504073e-02 -4.43845958e-01
-8.02112222e-01 6.57252073e-01 1.55769855e-01 6.00742579e-01
9.30064678e-01 -1.14535046e+00 -7.02021241e-01 5.25757633e-02
1.75913155e-01 3.43360066e-01 6.52489126e-01 1.08866024e+00
-1.87339753e-01 2.09267721e-01 1.15445167e-01 -9.94992971e-01
-1.38728988e+00 3.14273804e-01 3.78299832e-01 2.87415296e-01
-6.29268527e-01 1.14937091e+00 6.53410316e-01 1.60647482e-01
6.14938021e-01 -4.35668796e-01 -2.67920434e-01 3.17405045e-01
5.80080867e-01 2.35746533e-01 -3.19261476e-02 -6.66738689e-01
-3.20574820e-01 4.00739670e-01 -1.12598248e-01 -3.10046822e-01
1.34304285e+00 -4.51929003e-01 7.13530630e-02 6.70859754e-01
1.14979732e+00 -6.70970306e-02 -1.67291427e+00 -3.46062094e-01
-2.96505749e-01 -5.61538637e-01 3.39803129e-01 -6.71470046e-01
-1.52195978e+00 1.26239026e+00 6.97121561e-01 9.60653201e-02
1.38129413e+00 3.07677448e-01 8.18306744e-01 -5.02355359e-02
-6.37801141e-02 -9.49200332e-01 3.95704240e-01 4.04197961e-01
1.08199811e+00 -1.10314453e+00 -4.46075380e-01 -6.33530617e-01
-7.83552587e-01 8.94303679e-01 6.17888510e-01 1.47217259e-01
5.81761003e-01 2.38871649e-01 1.47878349e-01 -3.28601077e-02
-9.37123597e-01 -4.36974317e-01 5.11644840e-01 5.15388131e-01
2.59727299e-01 4.65320144e-03 2.83696771e-01 6.97087824e-01
1.27074309e-02 -6.37369901e-02 1.44864678e-01 5.78705907e-01
-5.84884703e-01 -7.41258025e-01 -3.46679091e-01 2.86938339e-01
-4.25454378e-01 -3.27043504e-01 -3.78104150e-01 2.56846875e-01
-2.21267100e-02 9.77829278e-01 1.58028468e-01 -4.07128662e-01
3.07998925e-01 -9.73176062e-02 2.17246130e-01 -5.04399776e-01
-3.28606814e-01 6.30560040e-01 -2.66837776e-01 -7.98436105e-01
-3.12153727e-01 -7.09870040e-01 -1.29217625e+00 2.73970425e-01
-8.97769555e-02 4.08999808e-02 3.13672811e-01 6.79441452e-01
5.46764433e-01 8.26535821e-01 7.16091692e-01 -1.02207124e+00
-2.27252185e-01 -6.52196586e-01 -4.11408544e-01 1.26185697e-02
6.55186236e-01 -5.61239779e-01 -2.48853356e-01 2.52626866e-01] | [14.802176475524902, 4.744424819946289] |
88450bc7-a741-4d30-95bb-06ba5c725c57 | cleme-debiasing-multi-reference-evaluation | 2305.10819 | null | https://arxiv.org/abs/2305.10819v1 | https://arxiv.org/pdf/2305.10819v1.pdf | CLEME: Debiasing Multi-reference Evaluation for Grammatical Error Correction | It is intractable to evaluate the performance of Grammatical Error Correction (GEC) systems since GEC is a highly subjective task. Designing an evaluation metric that is as objective as possible is crucial to the development of GEC task. Previous mainstream evaluation metrics, i.e., reference-based metrics, introduce bias into the multi-reference evaluation because they extract edits without considering the presence of multiple references. To overcome the problem, we propose Chunk-LEvel Multi-reference Evaluation (CLEME) designed to evaluate GEC systems in multi-reference settings. First, CLEME builds chunk sequences with consistent boundaries for the source, the hypothesis and all the references, thus eliminating the bias caused by inconsistent edit boundaries. Then, based on the discovery that there exist boundaries between different grammatical errors, we automatically determine the grammatical error boundaries and compute F$_{0.5}$ scores in a novel way. Our proposed CLEME approach consistently and substantially outperforms existing reference-based GEC metrics on multiple reference sets in both corpus-level and sentence-level settings. Extensive experiments and detailed analyses demonstrate the correctness of our discovery and the effectiveness of our designed evaluation metric. | ['Ying Shen', 'Hai-Tao Zheng', 'Shirong Ma', 'Yangning Li', 'Qingyu Zhou', 'Yinghui Li', 'Jingheng Ye'] | 2023-05-18 | null | null | null | null | ['grammatical-error-correction'] | ['natural-language-processing'] | [ 6.82787597e-02 -9.73638073e-02 3.63683164e-01 -5.70534945e-01
-1.15689063e+00 -5.31821489e-01 2.14389384e-01 6.76363349e-01
-4.89272147e-01 7.79939175e-01 1.88834593e-01 -2.19858959e-01
-1.79651678e-01 -5.84885180e-01 -6.72956944e-01 -2.66098559e-01
2.81088263e-01 3.11728120e-01 3.13653618e-01 -3.24665636e-01
9.12790656e-01 -2.37112388e-01 -1.62329125e+00 9.28126127e-02
1.77345026e+00 4.75700110e-01 4.59643036e-01 7.25674212e-01
-1.52913347e-01 4.86049742e-01 -1.03792965e+00 -7.40659833e-01
-2.47717664e-01 -8.11057508e-01 -9.16895390e-01 -5.18708646e-01
5.49788535e-01 1.64081678e-01 6.51899576e-01 1.34552085e+00
5.21924734e-01 1.02505319e-01 5.14565587e-01 -8.61043811e-01
-8.67488086e-01 7.55864739e-01 -2.30235368e-01 2.84235418e-01
4.63220000e-01 -1.58098280e-01 1.14886451e+00 -1.07383025e+00
7.77814209e-01 9.93397415e-01 8.39885294e-01 5.70952177e-01
-8.75178397e-01 -2.87387282e-01 2.54217777e-02 2.29490355e-01
-1.47487748e+00 -4.59711850e-01 4.38773394e-01 -3.34438562e-01
1.16264939e+00 3.45828265e-01 4.92154323e-02 7.60845602e-01
1.34867474e-01 3.23602259e-01 7.74717033e-01 -9.73461390e-01
3.99401225e-02 1.55760363e-01 4.88291353e-01 5.79172134e-01
3.56582910e-01 -1.28966898e-01 -4.93223995e-01 1.99519813e-01
-4.13020700e-03 -6.99491739e-01 -5.72074294e-01 2.59208053e-01
-9.75134492e-01 6.14866972e-01 -8.60525444e-02 4.62170899e-01
-2.09538966e-01 9.56032723e-02 6.54621065e-01 3.67431790e-01
4.45598334e-01 6.26865745e-01 -3.07869285e-01 -6.35670781e-01
-1.10486650e+00 3.01244259e-01 6.74486518e-01 1.24567246e+00
4.17370975e-01 -7.97107220e-02 -5.05337119e-01 1.05757928e+00
3.00098121e-01 4.09906626e-01 5.64989388e-01 -7.08042383e-01
8.82726550e-01 6.34317398e-01 1.79745346e-01 -1.09696007e+00
-2.01339155e-01 -6.32927418e-01 -4.97985005e-01 -1.34634718e-01
2.73534209e-01 3.22827995e-02 -2.89111197e-01 1.96945143e+00
2.19828442e-01 -2.33499125e-01 -2.39255186e-02 6.89601660e-01
9.21309173e-01 3.03636432e-01 1.99111432e-01 -2.79304385e-01
1.02817822e+00 -7.74068475e-01 -8.02878082e-01 1.56546474e-01
1.19217443e+00 -9.70008671e-01 1.38134718e+00 5.84206581e-02
-1.01965177e+00 -6.25476480e-01 -1.09388697e+00 -1.67338356e-01
-1.55950561e-01 3.97470444e-01 -1.73162092e-02 7.65206575e-01
-8.92059207e-01 7.49654651e-01 -3.86990607e-01 -3.23653132e-01
-2.14114293e-01 -8.34534541e-02 -1.62436038e-01 8.80510882e-02
-1.21745038e+00 1.07681262e+00 5.08970559e-01 1.25392213e-01
-3.19730133e-01 -6.86729133e-01 -7.41202474e-01 8.76501575e-02
2.01443896e-01 -3.91969472e-01 1.22587812e+00 -7.11862564e-01
-1.08866465e+00 7.73098171e-01 -3.56820524e-01 -2.74600297e-01
3.62188280e-01 -2.93828458e-01 -5.78593791e-01 -2.11637035e-01
2.73778945e-01 2.94898182e-01 3.89066309e-01 -1.02443218e+00
-9.10588086e-01 -1.17728367e-01 -4.57030237e-02 2.39478037e-01
8.01163018e-02 3.52052659e-01 -3.54069769e-01 -6.47996664e-01
-3.55473720e-02 -6.58445060e-01 3.45182121e-01 -7.34843194e-01
-4.14719075e-01 -5.88825583e-01 2.45037898e-01 -9.47919488e-01
2.00180221e+00 -2.11069989e+00 9.63526294e-02 7.54096508e-02
9.67412442e-02 4.69594002e-01 -2.40466133e-01 4.43798542e-01
6.85095713e-02 4.13184792e-01 -3.60522926e-01 -4.70063776e-01
8.83715302e-02 -1.36375368e-01 1.36334807e-01 4.86712083e-02
2.92512178e-01 6.90535069e-01 -1.06856561e+00 -7.02440977e-01
-1.35374829e-01 -2.02665161e-02 -5.75909972e-01 1.04954563e-01
-3.88807394e-02 2.94646263e-01 5.79426717e-03 5.76581180e-01
5.51341236e-01 4.25620861e-02 3.24066281e-02 4.26201373e-02
-3.34040761e-01 6.78730845e-01 -1.03263009e+00 1.57656384e+00
-5.39604902e-01 4.26278919e-01 -2.79109687e-01 -8.09955060e-01
1.10165739e+00 2.12234259e-01 -1.42790124e-01 -7.27479339e-01
-2.43258774e-02 6.78866446e-01 2.84391016e-01 -4.67493176e-01
1.13368094e+00 -5.81179513e-03 -2.30831295e-01 4.18377995e-01
1.77581519e-01 1.55938640e-01 6.36241555e-01 2.74382383e-01
1.18975675e+00 3.27942260e-02 4.29584652e-01 -4.32205975e-01
7.77476788e-01 -5.53954653e-02 9.24489856e-01 8.60595047e-01
-4.10911709e-01 7.68637002e-01 4.22687501e-01 1.75511628e-01
-9.71685112e-01 -7.97070622e-01 -1.84415013e-01 8.77066374e-01
1.43571705e-01 -8.52011681e-01 -1.15223753e+00 -9.14753139e-01
-1.36705026e-01 1.07545996e+00 -5.60828030e-01 -1.30402073e-01
-7.03854084e-01 -6.28022552e-01 7.91046381e-01 3.52221310e-01
3.16797316e-01 -1.06931269e+00 -5.06266236e-01 4.40017819e-01
-6.38317347e-01 -8.52085710e-01 -8.18497360e-01 -1.59924611e-01
-6.20673239e-01 -1.25772727e+00 -2.71944612e-01 -6.71868324e-01
5.37630320e-01 2.74873190e-02 1.43375945e+00 6.04477048e-01
-5.36871783e-04 1.73896998e-01 -9.21829760e-01 -3.14820528e-01
-7.70489872e-01 1.69342205e-01 -1.39466926e-01 -4.47156131e-01
5.76394260e-01 -2.20028654e-01 -3.39720905e-01 1.94752365e-01
-4.78078216e-01 -1.75717831e-01 3.59241873e-01 8.61202896e-01
5.87736189e-01 -2.00590149e-01 1.18909502e+00 -1.12695050e+00
1.05212843e+00 -4.09678996e-01 -4.16039824e-01 8.89316797e-01
-1.02665961e+00 1.48298085e-01 5.78845441e-01 1.09566912e-01
-1.01408696e+00 -6.24634206e-01 -4.04514015e-01 1.64360687e-01
1.23153348e-02 8.22934210e-01 -2.44620234e-01 5.75048588e-02
7.14338005e-01 2.06358820e-01 -3.21928829e-01 -5.42411268e-01
1.63123101e-01 8.02366614e-01 6.50859058e-01 -8.08120906e-01
3.65893245e-01 -5.61058879e-01 -3.87008756e-01 -5.20821393e-01
-6.72439873e-01 -4.24191415e-01 -6.25922740e-01 -3.05988908e-01
5.96039057e-01 -7.00573504e-01 -3.84305716e-01 3.17318320e-01
-1.56166041e+00 8.95761400e-02 -3.16989329e-03 3.96398991e-01
-3.20319742e-01 6.31062090e-01 -3.04128140e-01 -6.47062540e-01
-6.98783875e-01 -1.05756843e+00 9.63800907e-01 1.78495273e-01
-5.89854777e-01 -9.31577086e-01 2.46996284e-01 1.91150144e-01
3.63373935e-01 1.79665625e-01 1.12580490e+00 -5.64754367e-01
-2.78371364e-01 -1.85229510e-01 -2.23950967e-01 4.46782619e-01
3.12256403e-02 3.00055623e-01 -4.57406670e-01 -1.23890996e-01
-1.61156908e-01 8.61374959e-02 5.42061746e-01 1.24462031e-01
1.00303912e+00 -1.32922307e-01 2.25822497e-02 4.10701841e-01
1.56616628e+00 2.94028312e-01 7.32908607e-01 3.13555926e-01
5.36296546e-01 5.82624078e-01 8.51466358e-01 2.94857770e-01
5.82650840e-01 8.07430923e-01 8.66751894e-02 5.56020617e-01
-2.17812389e-01 -2.37787127e-01 2.69440651e-01 1.54625475e+00
-7.25565106e-02 -6.06961250e-01 -9.14892077e-01 8.03564548e-01
-1.63872075e+00 -7.97728598e-01 -5.85079849e-01 2.51950574e+00
1.09167206e+00 1.43053740e-01 -2.52572775e-01 2.65242040e-01
1.05250442e+00 -3.46880674e-01 -1.14360310e-01 -8.32428873e-01
-2.11790636e-01 3.81618619e-01 5.46020046e-02 5.39192677e-01
-8.12962770e-01 8.45104516e-01 5.86572886e+00 8.44177127e-01
-6.97438896e-01 2.38250360e-01 5.08883633e-02 -1.48964338e-02
-5.42815030e-01 5.06768189e-02 -1.09842193e+00 8.86405230e-01
1.17440844e+00 -4.07383323e-01 1.34546518e-01 5.46329021e-01
2.84959465e-01 -2.68505424e-01 -1.09459388e+00 8.51988971e-01
2.99452007e-01 -1.07590401e+00 -1.86309144e-01 -3.24404240e-01
7.07923770e-01 -2.57555962e-01 -3.22910339e-01 4.66864109e-01
2.06008062e-01 -7.16975510e-01 1.09784126e+00 5.63318908e-01
7.87883520e-01 -6.54730856e-01 9.52174246e-01 2.56344438e-01
-9.65738416e-01 3.22579563e-01 -4.40913260e-01 -3.50524038e-02
2.28117988e-01 8.38707447e-01 -4.98323143e-01 8.34307790e-01
5.33431530e-01 5.95458448e-01 -8.79585743e-01 1.12806952e+00
-5.42197585e-01 7.85718083e-01 3.76693811e-03 -2.13496312e-01
2.63613164e-02 -9.46257412e-02 5.92277288e-01 1.49957657e+00
6.44360662e-01 -8.98670126e-03 -3.66004825e-01 9.44797397e-01
-2.76573271e-01 5.42766809e-01 -2.87536502e-01 2.20819145e-01
1.15637755e+00 8.34505022e-01 -3.08121383e-01 -1.37024865e-01
-2.44781986e-01 8.88413310e-01 8.11920166e-01 4.43011150e-02
-8.64214659e-01 -1.08311903e+00 4.07611489e-01 -3.42901707e-01
1.37622938e-01 -5.87493367e-02 -5.96259475e-01 -1.10333538e+00
5.42875230e-01 -8.85393560e-01 1.89176902e-01 -3.84839147e-01
-1.09097934e+00 7.07156301e-01 -2.45469078e-01 -1.38278997e+00
-2.83005744e-01 -1.63970843e-01 -7.59720743e-01 1.18265629e+00
-1.37438917e+00 -6.67911828e-01 -3.43495876e-01 -3.95732112e-02
4.93796945e-01 9.18239057e-02 7.36881673e-01 5.62763572e-01
-8.94476056e-01 1.28688085e+00 3.37945044e-01 6.48389012e-02
9.70184982e-01 -1.48062241e+00 4.13227499e-01 1.35070050e+00
1.44005064e-02 9.06548917e-01 6.07202411e-01 -7.87140667e-01
-6.29074991e-01 -1.12440348e+00 1.71780837e+00 -6.48158252e-01
2.86674142e-01 -1.62605420e-01 -1.19145977e+00 1.45534188e-01
-5.08201160e-02 -3.38573217e-01 6.73809171e-01 3.81373703e-01
-3.30028176e-01 1.00788297e-02 -1.00814366e+00 2.14002326e-01
1.12251055e+00 -5.11680901e-01 -7.66034901e-01 -2.45641768e-02
8.30590010e-01 -4.85390365e-01 -1.01582980e+00 3.39537263e-01
3.94344807e-01 -1.04152226e+00 4.11156237e-01 -4.01558310e-01
7.45730400e-01 -4.13562149e-01 -1.96445078e-01 -1.59762311e+00
-2.34855786e-01 -2.36116454e-01 1.98259011e-01 1.82224321e+00
7.54623950e-01 -4.33202356e-01 8.42787623e-02 4.64798868e-01
-8.39171290e-01 -5.54028749e-01 -9.05408025e-01 -1.01427352e+00
3.02218497e-01 -3.11831266e-01 8.75313640e-01 8.69215608e-01
1.96531639e-01 1.19066514e-01 -1.39003813e-01 1.44732028e-01
6.24691844e-01 -1.04742229e-01 4.29520309e-01 -1.12124944e+00
-1.54651478e-01 -4.74036038e-01 -2.38711923e-01 -5.48421204e-01
7.18891770e-02 -9.80559111e-01 5.35866916e-01 -1.27777350e+00
1.25660881e-01 -7.28875458e-01 -1.84967697e-01 1.87848836e-01
-7.64709055e-01 -9.68328789e-02 1.71179265e-01 3.32677096e-01
-7.41636634e-01 6.28275692e-01 8.29335213e-01 2.38166407e-01
-6.63728118e-02 -4.71943527e-01 -8.46592724e-01 2.59381354e-01
7.92406023e-01 -6.25057161e-01 -1.60989314e-01 -8.11592042e-01
3.86237144e-01 -2.08188206e-01 2.56183576e-02 -1.06122541e+00
1.65558785e-01 -5.15762642e-02 -2.67842710e-01 -4.52468485e-01
-5.75001836e-01 -1.55391335e-01 4.47495468e-02 6.72965422e-02
-4.97812718e-01 3.12095851e-01 -9.05350298e-02 4.05059218e-01
-4.92773145e-01 -9.64210808e-01 7.42350936e-01 1.82032399e-02
-6.53656542e-01 -3.08727682e-01 2.71824971e-02 5.64423084e-01
6.28882647e-01 -1.72711283e-01 -6.92527354e-01 1.20131299e-01
-2.51726002e-01 2.70974547e-01 5.51818073e-01 4.06705111e-01
4.29241449e-01 -1.15162694e+00 -1.03549278e+00 -1.01234697e-01
5.96950591e-01 -1.28593430e-01 2.00028509e-01 8.88567924e-01
-5.87221861e-01 4.97577310e-01 3.96631956e-02 -3.78601402e-01
-1.41302598e+00 9.38480720e-02 2.58511305e-01 -3.70885551e-01
-3.17041576e-01 8.78525555e-01 -2.81211734e-01 -5.18087029e-01
1.19887494e-01 -3.90818149e-01 -2.93012470e-01 -1.83907375e-01
5.35480976e-01 5.25841951e-01 6.50154412e-01 -6.75015330e-01
-3.63238096e-01 3.39545131e-01 -1.90227821e-01 1.33059740e-01
8.98005128e-01 -5.18006682e-01 -2.27106154e-01 3.55316520e-01
9.89112318e-01 6.36163726e-02 -4.95185792e-01 -2.89162770e-02
4.66489226e-01 -5.43103158e-01 -1.53373763e-01 -1.11104250e+00
-5.00061750e-01 7.52691031e-01 3.53529185e-01 -6.19499385e-02
1.02632928e+00 -3.75593245e-01 6.96466029e-01 1.81791782e-01
4.30000693e-01 -1.43856478e+00 -2.90249884e-01 8.42043757e-01
8.25320005e-01 -1.25062478e+00 -3.46838593e-01 -5.21166921e-01
-4.20072764e-01 8.92829418e-01 8.59009504e-01 2.60709286e-01
2.49247819e-01 -1.76966995e-01 -6.99488819e-02 -1.44415364e-01
-7.55620360e-01 -8.39725286e-02 3.36883426e-01 5.03747284e-01
9.56520915e-01 1.00927532e-01 -1.30886984e+00 8.17282557e-01
-4.34002668e-01 -1.27750620e-01 7.08473206e-01 7.45657623e-01
-5.24338782e-01 -1.29095984e+00 -1.15337178e-01 4.38463509e-01
-3.93531919e-01 -3.17918360e-01 -1.76763073e-01 6.34247363e-01
2.46650428e-01 1.16680443e+00 -1.85196660e-02 -4.56437349e-01
4.71957356e-01 2.52706766e-01 4.84119982e-01 -7.86395848e-01
-8.45026076e-01 -5.12383163e-01 3.27394456e-01 -2.98371673e-01
-3.40133131e-01 -7.45181561e-01 -1.20290387e+00 -3.20069790e-01
-7.71349669e-01 5.98553658e-01 6.78412676e-01 1.04932988e+00
6.01570189e-01 4.81681168e-01 5.13893247e-01 4.22885874e-03
-6.75045967e-01 -1.43512130e+00 -4.21965420e-01 6.83929920e-01
-4.99606095e-02 -7.39665329e-01 -4.02531028e-01 4.22596782e-02] | [11.071444511413574, 10.72245979309082] |
ab99e938-6c2c-46da-b973-7d8b1a2de4e4 | measuring-and-mitigating-constraint | 2305.15338 | null | https://arxiv.org/abs/2305.15338v1 | https://arxiv.org/pdf/2305.15338v1.pdf | Measuring and Mitigating Constraint Violations of In-Context Learning for Utterance-to-API Semantic Parsing | In executable task-oriented semantic parsing, the system aims to translate users' utterances in natural language to machine-interpretable programs (API calls) that can be executed according to pre-defined API specifications. With the popularity of Large Language Models (LLMs), in-context learning offers a strong baseline for such scenarios, especially in data-limited regimes. However, LLMs are known to hallucinate and therefore pose a formidable challenge in constraining generated content. Thus, it remains uncertain if LLMs can effectively perform task-oriented utterance-to-API generation where respecting API's structural and task-specific constraints is crucial. In this work, we seek to measure, analyze and mitigate such constraints violations. First, we identify the categories of various constraints in obtaining API-semantics from task-oriented utterances, and define fine-grained metrics that complement traditional ones. Second, we leverage these metrics to conduct a detailed error analysis of constraints violations seen in state-of-the-art LLMs, which motivates us to investigate two mitigation strategies: Semantic-Retrieval of Demonstrations (SRD) and API-aware Constrained Decoding (API-CD). Our experiments show that these strategies are effective at reducing constraints violations and improving the quality of the generated API calls, but require careful consideration given their implementation complexity and latency. | ['Yi Zhang', 'Nikolaos Pappas', 'James Gung', 'Sailik Sengupta', 'Sebastien Jean', 'Shufan Wang'] | 2023-05-24 | null | null | null | null | ['semantic-retrieval', 'semantic-parsing'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.39601409e-01 2.31824726e-01 -2.47377679e-01 -6.01724803e-01
-1.20256782e+00 -8.86027038e-01 7.41576850e-01 -7.64172059e-03
-1.24494918e-01 2.82582879e-01 3.69304717e-01 -6.53436244e-01
6.74603134e-02 -4.23470438e-01 -9.45997000e-01 2.79595125e-02
9.03241038e-02 4.57794815e-01 2.01118633e-01 -6.53878897e-02
2.26131722e-01 1.72562242e-01 -1.74084091e+00 7.46879697e-01
1.04194999e+00 7.29135036e-01 4.79694486e-01 7.59139836e-01
-4.34547812e-01 1.03259993e+00 -7.40978658e-01 -3.60195309e-01
-1.29377484e-01 -1.88155904e-01 -1.06971061e+00 4.61859293e-02
2.42209211e-01 -2.95931876e-01 1.69181272e-01 1.02855563e+00
1.77449957e-01 3.63930427e-02 3.50867450e-01 -1.51685500e+00
-3.66153330e-01 7.49595284e-01 2.68761635e-01 -3.45318764e-02
7.70165920e-01 3.94666046e-01 1.21769941e+00 -7.29437649e-01
5.81476569e-01 1.38607156e+00 2.71288037e-01 8.70478809e-01
-1.11773717e+00 -2.44450882e-01 3.52181315e-01 1.00367665e-01
-1.10505760e+00 -8.41675222e-01 6.43111527e-01 -5.40235281e-01
1.54589069e+00 6.15563273e-01 4.74855825e-02 1.50480878e+00
-1.30539596e-01 9.98825431e-01 8.10916781e-01 -6.69129074e-01
3.99132013e-01 2.04371095e-01 3.17502916e-01 5.89533925e-01
-1.83233172e-02 6.68237209e-02 -7.42436230e-01 -3.11797202e-01
6.08958006e-02 -2.42176369e-01 -1.39318109e-01 -1.83161378e-01
-1.21236694e+00 7.85111427e-01 -1.89401150e-01 1.43946677e-01
-1.44764036e-01 1.44276112e-01 6.07406378e-01 2.60108829e-01
2.64867276e-01 6.99595034e-01 -6.80407047e-01 -6.87476277e-01
-7.17192829e-01 4.70989466e-01 1.03644669e+00 1.52652454e+00
6.00777924e-01 -6.77614585e-02 -2.31792510e-01 7.12103605e-01
2.62731045e-01 5.41395962e-01 4.20888811e-01 -9.55401480e-01
7.66360998e-01 6.92181349e-01 8.48219991e-02 -5.09039402e-01
-2.34711990e-01 9.92676467e-02 6.37906278e-03 -1.58815965e-01
2.90065259e-01 -1.07282355e-01 -6.23020530e-01 1.89805710e+00
1.46583155e-01 4.75802384e-02 2.25668237e-01 8.32673907e-01
4.34666187e-01 4.95970964e-01 3.49967331e-01 1.56555623e-02
1.52506149e+00 -9.73943830e-01 -5.06519318e-01 -7.42686868e-01
9.50469613e-01 -8.22252333e-01 1.86072671e+00 1.93928361e-01
-9.23061132e-01 -3.41006100e-01 -9.21924293e-01 6.36156183e-03
-2.05238566e-01 -5.20273745e-02 8.03171277e-01 6.12096310e-01
-7.88464785e-01 3.11123967e-01 -9.35728192e-01 -2.74311125e-01
1.44381613e-01 7.22413063e-02 5.00951745e-02 -1.27253562e-01
-1.00758076e+00 7.82686472e-01 2.03533739e-01 -1.86214313e-01
-1.25352859e+00 -6.54112339e-01 -1.12157619e+00 9.15792435e-02
8.08037400e-01 -5.13501823e-01 1.90454209e+00 -7.82935798e-01
-1.47483444e+00 7.78942645e-01 -5.30721962e-01 -3.39931369e-01
1.67016283e-01 -3.40170145e-01 -3.57063353e-01 -1.18142687e-01
1.39574558e-01 5.30483007e-01 8.52127731e-01 -9.53905761e-01
-7.78757155e-01 -1.96641982e-01 6.40594721e-01 2.01308668e-01
-2.62035996e-01 3.52081090e-01 -5.48265696e-01 -1.64297909e-01
-1.18498549e-01 -1.11923850e+00 -1.41164009e-02 -5.26612520e-01
-7.47765958e-01 -4.48633879e-01 4.96415138e-01 -5.42872488e-01
1.30386555e+00 -2.17583323e+00 2.48468906e-01 -5.69055900e-02
4.81568713e-04 2.06501409e-01 -3.41256797e-01 3.79711747e-01
2.36378327e-01 3.12349945e-01 -2.36954406e-01 -6.49356961e-01
4.81948406e-01 3.80888224e-01 -5.19276440e-01 -1.06022852e-02
4.13425684e-01 9.30238426e-01 -8.30406964e-01 -4.72957373e-01
3.29634666e-01 1.09397501e-01 -9.36625957e-01 7.62909830e-01
-9.64681268e-01 2.97682852e-01 -8.47867250e-01 9.22701001e-01
1.55137673e-01 -8.77482891e-02 1.94159567e-01 -4.53532748e-02
9.93749034e-03 9.63255405e-01 -7.42490590e-01 1.77863216e+00
-1.00120747e+00 5.37608743e-01 -2.27602050e-02 -1.02645946e+00
4.49344695e-01 3.41142207e-01 -1.54854497e-02 -7.91684210e-01
-1.96008354e-01 3.53614807e-01 -7.39087313e-02 -8.15944135e-01
3.79171610e-01 1.77578330e-01 -5.16103268e-01 4.20473397e-01
-1.20559521e-01 -2.63551265e-01 1.46657884e-01 1.99706435e-01
1.28053379e+00 2.74127394e-01 1.79049715e-01 -1.60719439e-01
5.44115543e-01 2.34288365e-01 2.95082897e-01 8.49285245e-01
-7.23052770e-02 4.28571254e-01 6.27099574e-01 -2.56181985e-01
-9.31232214e-01 -6.99684680e-01 2.02400938e-01 1.49518716e+00
-1.88979939e-01 -4.73204553e-01 -1.16290390e+00 -9.59633410e-01
-1.54819518e-01 1.25504518e+00 1.43562639e-02 -2.69667119e-01
-7.58242428e-01 -3.10115308e-01 7.07387149e-01 2.73753256e-01
5.19986898e-02 -1.39031291e+00 -9.01659966e-01 1.52375102e-01
-4.37387705e-01 -1.55203807e+00 -5.33079982e-01 2.63996810e-01
-6.01211429e-01 -1.32234800e+00 1.02286320e-02 -5.36555052e-01
6.05375469e-01 1.09590113e-01 1.48998737e+00 1.74366966e-01
-1.80327501e-02 5.75685441e-01 -5.97881734e-01 -4.21498358e-01
-9.81166184e-01 1.92711383e-01 -2.39033680e-02 -3.45453411e-01
7.60601521e-01 -4.53860760e-01 -1.55125871e-01 2.11850330e-01
-9.36635137e-01 2.08517328e-01 5.21565259e-01 4.61164653e-01
4.57632005e-01 -2.71330059e-01 3.83576453e-01 -1.12848365e+00
8.97996366e-01 -3.26454788e-01 -8.69985938e-01 3.94542456e-01
-6.56656444e-01 2.15309560e-01 9.82594013e-01 -4.01356846e-01
-9.99148667e-01 8.84659737e-02 -2.73914576e-01 -3.07558894e-01
-4.86992389e-01 6.47823632e-01 -4.86697465e-01 2.47890636e-01
7.70673931e-01 3.41808647e-01 -3.22808981e-01 -4.59295332e-01
5.33949554e-01 9.68815327e-01 6.32533371e-01 -1.22949469e+00
6.66558683e-01 3.19738798e-02 -5.34237564e-01 -6.29594922e-01
-1.12806606e+00 -4.69645619e-01 -1.20621413e-01 9.59524140e-02
8.41605008e-01 -7.88506031e-01 -6.87449932e-01 4.67863791e-02
-1.41208243e+00 -6.34219885e-01 -1.09560058e-01 1.20088883e-01
-8.47029209e-01 3.53089929e-01 -3.98544073e-01 -8.84872317e-01
-3.14970940e-01 -1.62657452e+00 1.34682572e+00 -7.78594315e-02
-6.65484846e-01 -7.16256380e-01 -2.39745751e-01 5.76691151e-01
4.67919856e-01 -1.57405555e-01 1.28793049e+00 -1.04366815e+00
-8.45527709e-01 -6.67413846e-02 -1.44814938e-01 5.40785253e-01
-8.08942616e-02 -2.81820923e-01 -1.16758060e+00 -1.05440304e-01
1.72647238e-01 -4.44187194e-01 1.15763761e-01 7.24779591e-02
1.25538349e+00 -5.15520573e-01 -2.27716357e-01 5.27243018e-01
9.79139447e-01 3.28264683e-01 3.82995188e-01 2.15792567e-01
6.38789833e-01 6.92553878e-01 6.98145807e-01 2.83855110e-01
4.68897253e-01 9.03823853e-01 3.97598118e-01 4.06216115e-01
-1.13699019e-01 -5.21158874e-01 6.44583642e-01 7.39129007e-01
4.65601355e-01 -2.31566250e-01 -9.49147403e-01 5.10088325e-01
-1.84747696e+00 -5.08956015e-01 2.94597633e-02 2.20971012e+00
1.02119339e+00 4.31708217e-01 -2.59618849e-01 -2.52253801e-01
3.46460074e-01 1.75922498e-01 -5.73250175e-01 -5.57235062e-01
3.03965330e-01 2.35952601e-01 1.26533151e-01 5.85630953e-01
-9.91940320e-01 1.10936558e+00 5.94725418e+00 6.45876467e-01
-8.68921399e-01 1.89227119e-01 2.84424275e-01 1.29312739e-01
-5.31908393e-01 1.76596537e-01 -9.75561678e-01 5.62275648e-01
1.38426852e+00 -1.43400848e-01 9.16009128e-01 1.37570298e+00
2.08844781e-01 7.33511942e-03 -1.94060409e+00 7.39250422e-01
-6.71525076e-02 -1.26665521e+00 5.53733623e-03 -2.38469481e-01
2.93552846e-01 2.14125454e-01 -1.39491022e-01 8.91791046e-01
1.78261220e-01 -8.76531780e-01 1.05368745e+00 8.09932649e-02
8.19027722e-01 -4.22106743e-01 5.22805989e-01 6.95695102e-01
-1.05434978e+00 2.74869241e-02 -1.60494432e-01 -1.54756337e-01
1.88190937e-01 4.38979030e-01 -9.18948770e-01 1.39706194e-01
3.90141308e-01 1.77184224e-01 -4.39827502e-01 3.62274408e-01
-4.20103639e-01 7.45589972e-01 -2.62386203e-01 -1.63123012e-01
1.75233006e-01 1.67881399e-01 6.10214770e-01 1.27886701e+00
1.35606453e-01 -4.57879156e-02 3.55707586e-01 1.31651306e+00
4.63704169e-02 -1.95863754e-01 -5.83952665e-01 -5.67202628e-01
7.81426370e-01 9.39733386e-01 -2.78622895e-01 -4.45071280e-01
-5.96837938e-01 9.95573580e-01 3.17664534e-01 4.19645548e-01
-9.23920393e-01 -1.28898814e-01 1.03120077e+00 -1.24601968e-01
-1.11058973e-01 -2.34333873e-01 -1.92936033e-01 -1.25088298e+00
5.39665103e-01 -1.47177529e+00 2.57237703e-01 -5.25673807e-01
-9.51521814e-01 6.02269709e-01 2.29365945e-01 -8.55080843e-01
-6.75147355e-01 -5.63895524e-01 -4.97100711e-01 7.30752230e-01
-1.71448302e+00 -1.07744122e+00 -1.17147289e-01 3.74649197e-01
1.13385367e+00 -1.52976528e-01 1.15859354e+00 3.39549184e-01
-5.39806843e-01 6.37679160e-01 -5.15011311e-01 -2.12283477e-01
2.74510264e-01 -1.11437404e+00 7.92045712e-01 1.04140580e+00
3.10706139e-01 8.15105915e-01 8.24557662e-01 -5.92768729e-01
-1.91619086e+00 -1.12763774e+00 1.17338717e+00 -8.50607753e-01
5.80743551e-01 -6.94172561e-01 -8.57604742e-01 9.22235012e-01
-3.14672470e-01 -2.05307141e-01 4.42692369e-01 2.26807788e-01
-5.37342250e-01 3.08705896e-01 -8.43138814e-01 6.91293955e-01
1.17549443e+00 -1.04934669e+00 -7.25819826e-01 7.43075371e-01
1.29399562e+00 -6.82934940e-01 -4.97736603e-01 1.92120522e-01
2.62653977e-01 -9.60190296e-01 1.00502706e+00 -8.20278406e-01
4.75393355e-01 -1.72005862e-01 -5.22730827e-01 -1.04456913e+00
3.16653252e-01 -8.57108712e-01 -1.55408159e-01 1.16261387e+00
6.46905303e-01 -4.92846489e-01 6.40434027e-01 1.21673489e+00
-7.26150513e-01 -5.84866703e-01 -7.40903258e-01 -9.44401681e-01
-2.79710203e-01 -1.04749966e+00 7.62651265e-01 7.46369600e-01
2.06205875e-01 4.06634212e-01 -6.53959736e-02 3.72141451e-01
3.53129059e-01 1.51504084e-01 8.60615909e-01 -8.17733765e-01
-4.98903275e-01 -3.35484833e-01 1.01099715e-01 -1.17427826e+00
7.41925120e-01 -9.37953353e-01 3.48657846e-01 -1.17804372e+00
1.22784548e-01 -7.56819189e-01 8.60953182e-02 5.42614102e-01
-1.42491341e-01 -4.18425143e-01 2.32285604e-01 1.87860414e-01
-8.16740334e-01 2.17488274e-01 6.20945156e-01 -1.20677441e-01
-2.65685350e-01 1.87581465e-01 -6.47194147e-01 7.43479908e-01
5.31594455e-01 -4.76920635e-01 -6.62153125e-01 -7.04624176e-01
1.76337942e-01 1.05584465e-01 2.04702094e-01 -9.15024519e-01
4.68945727e-02 -4.88046646e-01 -5.79352677e-01 -1.25528127e-01
1.69411778e-01 -7.63731718e-01 -1.52686452e-02 1.74896270e-01
-6.41312778e-01 -2.23111913e-01 1.72492176e-01 3.71387750e-01
-2.16015056e-01 -5.76247275e-01 3.26297492e-01 -9.50259343e-02
-9.48928833e-01 -7.13582635e-02 -4.57468539e-01 3.75116467e-01
6.96791828e-01 1.44128084e-01 -3.80822897e-01 -2.84552842e-01
-3.68484497e-01 1.81531847e-01 4.02204633e-01 7.26296008e-01
5.75896442e-01 -8.78313839e-01 -4.16625887e-01 2.58322656e-01
6.45904541e-01 2.11520195e-01 -6.07847944e-02 6.50283277e-01
-2.42493972e-01 8.70195866e-01 3.82250398e-01 -6.75702572e-01
-1.09887671e+00 5.47957063e-01 1.29289180e-01 -1.96781561e-01
-5.28726935e-01 9.11329627e-01 3.03820431e-01 -6.95043743e-01
5.31199038e-01 -7.08067834e-01 2.27362037e-01 -4.64712322e-01
4.78955507e-01 -1.42587543e-01 3.54758769e-01 -3.03015709e-01
-4.90391552e-01 1.36700775e-02 -3.18276510e-02 6.68181852e-02
1.06307745e+00 1.01037867e-01 5.09740524e-02 1.25269264e-01
1.15379000e+00 -2.39423811e-02 -1.14810562e+00 -2.78944165e-01
5.77474594e-01 -2.91334093e-01 -1.24530382e-01 -8.71601522e-01
-4.93772775e-01 8.73131275e-01 1.14153154e-01 2.51696289e-01
9.47952747e-01 1.95101917e-01 1.09824073e+00 6.92844033e-01
5.57472646e-01 -9.94573951e-01 1.42219260e-01 5.47018707e-01
6.78208232e-01 -1.21420121e+00 -4.28543597e-01 -6.04211032e-01
-5.89363337e-01 7.96866536e-01 6.24667764e-01 3.75416964e-01
-6.07707091e-02 4.86486197e-01 -4.87078801e-02 -2.33612061e-01
-9.68023837e-01 -1.37329102e-01 2.61098593e-01 7.40386605e-01
4.79397058e-01 3.56058359e-01 5.75765185e-02 7.90410221e-01
-2.73329616e-01 -1.30019024e-01 3.88674170e-01 1.04377103e+00
-3.87451500e-01 -1.28508222e+00 -1.46056458e-01 3.18340480e-01
-2.62571543e-01 -3.91452074e-01 -3.44468057e-01 6.16544008e-01
-2.51878768e-01 1.11007512e+00 -1.80544302e-01 -2.50367910e-01
6.34008050e-01 5.75609207e-01 2.47430965e-01 -1.04803061e+00
-4.33992207e-01 -1.59825817e-01 6.54507399e-01 -1.05656052e+00
-1.26970693e-01 -5.71262717e-01 -1.43906713e+00 2.44902968e-02
-2.19906494e-01 1.59018245e-02 8.97041440e-01 1.23906171e+00
5.10705411e-01 5.33799052e-01 5.13351440e-01 -6.50454998e-01
-1.00323820e+00 -6.88006938e-01 8.17719698e-02 6.53727531e-01
2.40381911e-01 -3.96646261e-01 -4.58295912e-01 2.97724962e-01] | [10.563557624816895, 8.14789867401123] |
ae01aef6-d2d8-45d9-acd5-a8c83cf1719c | representation-online-matters-practical-end | 2305.15534 | null | https://arxiv.org/abs/2305.15534v2 | https://arxiv.org/pdf/2305.15534v2.pdf | Representation Online Matters: Practical End-to-End Diversification in Search and Recommender Systems | As the use of online platforms continues to grow across all demographics, users often express a desire to feel represented in the content. To improve representation in search results and recommendations, we introduce end-to-end diversification, ensuring that diverse content flows throughout the various stages of these systems, from retrieval to ranking. We develop, experiment, and deploy scalable diversification mechanisms in multiple production surfaces on the Pinterest platform, including Search, Related Products, and New User Homefeed, to improve the representation of different skin tones in beauty and fashion content. Diversification in production systems includes three components: identifying requests that will trigger diversification, ensuring diverse content is retrieved from the large content corpus during the retrieval stage, and finally, balancing the diversity-utility trade-off in a self-adjusting manner in the ranking stage. Our approaches, which evolved from using Strong-OR logical operator to bucketized retrieval at the retrieval stage and from greedy re-rankers to multi-objective optimization using determinantal point processes for the ranking stage, balances diversity and utility while enabling fast iterations and scalable expansion to diversification over multiple dimensions. Our experiments indicate that these approaches significantly improve diversity metrics, with a neutral to a positive impact on utility metrics and improved user satisfaction, both qualitatively and quantitatively, in production. An accessible PDF of this article is available at https://drive.google.com/file/d/1p5PkqC-sdtX19Y_IAjZCtiSxSEX1IP3q/view | ['Nadia Fawaz', 'Ashudeep Singh', 'Shloka Desai', 'Bhawna Juneja', 'Pedro Silva'] | 2023-05-24 | null | null | null | null | ['point-processes'] | ['methodology'] | [-3.71264279e-01 -3.51573139e-01 -4.61061388e-01 -2.54843026e-01
-7.68172920e-01 -1.09572875e+00 2.58064151e-01 2.17201009e-01
1.34297967e-01 3.18194449e-01 6.53603911e-01 -2.57053673e-01
-5.99699914e-01 -7.12521374e-01 -2.17526332e-01 -1.55521750e-01
1.00792609e-01 5.98149955e-01 -4.31475453e-02 -6.62382245e-01
6.03722572e-01 3.56836259e-01 -1.59215486e+00 5.12643754e-01
1.06820774e+00 1.17006242e+00 3.05874646e-01 5.92608988e-01
-3.06658357e-01 3.95725578e-01 -5.21634281e-01 -7.00494051e-01
5.65386295e-01 -1.67612731e-01 -6.81819022e-01 5.19146770e-03
3.03911269e-01 -3.32845658e-01 1.38952330e-01 6.21012449e-01
7.88379610e-01 2.61793286e-01 3.94007653e-01 -9.44540620e-01
-8.76013637e-01 5.86017191e-01 -6.86946332e-01 1.81643918e-01
6.04012966e-01 1.15813665e-01 1.48570621e+00 -9.52586353e-01
7.38527000e-01 1.20382702e+00 2.89797395e-01 4.53890041e-02
-1.19504690e+00 -7.11639464e-01 2.26143003e-01 -1.86924279e-01
-1.31716490e+00 -4.29490298e-01 5.45084894e-01 -4.78062987e-01
6.33035481e-01 8.30637813e-01 7.96798348e-01 7.20630109e-01
-2.45607104e-02 7.75689721e-01 5.96129417e-01 -1.06719501e-01
2.18186006e-01 4.32088107e-01 6.70313090e-02 2.89341092e-01
2.52455175e-01 -4.71154749e-01 -7.66012132e-01 -4.84384328e-01
6.58335745e-01 2.36077085e-01 -1.08683392e-01 -2.87665337e-01
-6.18801892e-01 8.94354999e-01 1.68293029e-01 -1.70318913e-02
-6.89976156e-01 -2.53470689e-01 2.53531754e-01 5.16825676e-01
4.35909659e-01 1.04375088e+00 -5.06640315e-01 -4.62788701e-01
-9.43625093e-01 5.26802003e-01 1.02901125e+00 1.20646346e+00
4.56139594e-01 -5.25786042e-01 -5.36845386e-01 1.24022150e+00
1.78895310e-01 3.99939507e-01 4.63910639e-01 -1.22713459e+00
3.79213512e-01 9.17094290e-01 3.36638927e-01 -1.08172190e+00
-3.47551942e-01 -8.69398654e-01 -1.57291532e-01 -2.34123752e-01
8.10609460e-02 -3.51918668e-01 -5.40571809e-01 1.50010824e+00
3.04180205e-01 -1.00976467e+00 -4.57082808e-01 9.32610512e-01
6.04907393e-01 6.01762354e-01 6.08474165e-02 -1.37061521e-01
1.32628930e+00 -9.33287442e-01 -4.08648431e-01 2.06281453e-01
3.29948395e-01 -1.24029636e+00 1.37651348e+00 5.78110278e-01
-1.59520662e+00 -4.53077346e-01 -7.24467993e-01 -1.13353416e-01
-1.28446952e-01 -4.55970056e-02 4.64352041e-01 5.85378647e-01
-1.07355547e+00 4.10136312e-01 -9.81264487e-02 -4.76879448e-01
1.95958331e-01 4.05955881e-01 2.33077496e-01 -2.87833139e-02
-1.06382036e+00 5.15707254e-01 -6.48987591e-02 -3.94352823e-01
-4.00519907e-01 -8.32173824e-01 -9.07359198e-02 2.98594296e-01
4.49370533e-01 -7.97193170e-01 1.24498427e+00 -1.02019799e+00
-1.37695229e+00 3.92262965e-01 2.58794934e-01 1.78997904e-01
4.78527755e-01 -3.48522723e-01 -5.40938914e-01 -1.25636026e-01
9.97064412e-02 5.70337415e-01 5.95753372e-01 -1.28221846e+00
-9.24848974e-01 -2.51008838e-01 2.30680510e-01 7.05581009e-01
-6.86116219e-01 1.60027474e-01 -8.05661023e-01 -6.09878480e-01
-3.69187305e-03 -9.05736327e-01 -1.03300013e-01 -2.62529731e-01
-2.63027281e-01 -8.85161608e-02 3.38832498e-01 -8.27401996e-01
1.84551096e+00 -2.18889022e+00 2.12710515e-01 6.73488617e-01
1.41093269e-01 -1.31271511e-01 -2.65813500e-01 9.66488898e-01
3.00655901e-01 5.02140343e-01 6.39444053e-01 -1.81339514e-02
2.44854465e-01 -2.95412689e-01 -2.62036234e-01 -1.06405036e-03
-1.19448297e-01 7.80432761e-01 -8.40646267e-01 -2.78680503e-01
-3.24119836e-01 1.23743780e-01 -9.46693718e-01 8.14039167e-03
-3.84107232e-01 -1.64444335e-02 -5.23534417e-01 1.15401340e+00
4.93195832e-01 -5.52226543e-01 2.26726979e-01 -7.67556131e-02
-2.75685668e-01 2.67324239e-01 -1.17100143e+00 1.54482090e+00
-5.90605080e-01 7.71484748e-02 2.54688084e-01 -1.54688686e-01
8.37230444e-01 -7.58700743e-02 7.98271537e-01 -1.20730400e+00
1.27385989e-01 2.45361775e-01 -1.51671842e-01 -3.41696322e-01
1.22414804e+00 4.00829136e-01 -4.38868493e-01 4.37150300e-01
-3.21672410e-01 -6.66315332e-02 7.11391211e-01 3.84019315e-01
1.06507611e+00 -9.92146134e-02 -5.88777417e-04 -4.26187545e-01
4.76027206e-02 2.92918384e-01 2.16259941e-01 5.44051886e-01
2.59111851e-01 4.80084300e-01 4.49503392e-01 1.01678163e-01
-1.01367080e+00 -1.03749537e+00 -1.54853538e-01 1.71625900e+00
3.88299882e-01 -6.58587337e-01 -3.05950314e-01 -2.26479203e-01
4.91832584e-01 8.78429055e-01 -1.97775275e-01 -1.72593612e-02
-2.76223958e-01 -3.59237015e-01 -4.37606797e-02 9.37295705e-02
1.79942325e-01 -8.54112685e-01 -3.68629426e-01 2.04247862e-01
-1.98325425e-01 -2.54755259e-01 -8.78229856e-01 1.28722787e-01
-6.27298295e-01 -5.61238408e-01 -8.30847323e-01 -3.66190284e-01
4.25522089e-01 4.31189716e-01 1.28205562e+00 1.15484081e-01
-3.22472900e-01 4.73841488e-01 -6.05878890e-01 -1.46997973e-01
-1.84458680e-02 4.35233295e-01 -1.23143919e-01 -3.12182248e-01
5.11319563e-02 -4.80378509e-01 -1.08872545e+00 6.80064559e-01
-9.30103898e-01 -2.08098710e-01 4.84833121e-01 4.88152236e-01
3.90708208e-01 -6.60336092e-02 5.56217015e-01 -5.22303820e-01
1.22369826e+00 -1.05112898e+00 -3.81196111e-01 2.38203600e-01
-9.46904063e-01 -7.92357251e-02 5.11603773e-01 -3.87405157e-01
-8.11606348e-01 -3.16227376e-01 -8.60991143e-03 -1.81015238e-01
5.42593598e-01 6.03723228e-01 1.00661151e-01 2.38099650e-01
8.73378813e-01 -2.22453356e-01 7.89139345e-02 -5.90081751e-01
6.69845879e-01 8.12721848e-01 -4.21289802e-02 -5.91485262e-01
5.03864229e-01 2.38961026e-01 -6.47431433e-01 -4.07740295e-01
-4.46346730e-01 -8.00859272e-01 4.74404357e-02 -1.85777366e-01
2.60142058e-01 -9.36884940e-01 -7.18185484e-01 -1.33993745e-01
-4.06625032e-01 -2.52107173e-01 -6.33063316e-01 8.79199952e-02
-1.98484212e-01 7.91588649e-02 -5.71223795e-01 -6.13450646e-01
-7.04909444e-01 -8.61321092e-01 9.37634170e-01 5.57364166e-01
-6.79595351e-01 -4.57983941e-01 5.93516901e-02 5.08989871e-01
8.41350913e-01 -2.22561657e-01 1.00680232e+00 -7.96456039e-01
-6.17915750e-01 -1.50013581e-01 -1.92821220e-01 -1.14082828e-01
1.82653651e-01 2.31037423e-01 -1.49651796e-01 -3.32931936e-01
-5.01844764e-01 -1.49768844e-01 4.06868964e-01 2.82467425e-01
8.94083142e-01 -3.61425430e-01 -1.33533835e-01 3.42504174e-01
1.25498247e+00 4.55885589e-01 4.91583586e-01 6.08001113e-01
2.98206180e-01 6.54268861e-01 9.66262639e-01 1.12872148e+00
4.42712158e-01 8.63362849e-01 2.36948505e-01 -5.73741123e-02
-4.53663953e-02 -1.88308939e-01 3.33821118e-01 6.75940037e-01
3.11984837e-01 -6.05383873e-01 -5.15645921e-01 5.08481383e-01
-1.74364030e+00 -8.62784684e-01 3.97429168e-01 2.22709465e+00
8.60843182e-01 8.18632245e-02 6.01611435e-01 -3.44794005e-01
4.67735261e-01 -7.51137584e-02 -6.34728134e-01 -8.64488363e-01
9.52104628e-02 -3.29376310e-02 5.71810782e-01 3.10775846e-01
-4.34479654e-01 6.60102546e-01 6.14580679e+00 7.90233493e-01
-1.17616332e+00 -1.45983145e-01 8.57331812e-01 -9.39904749e-01
-9.73591626e-01 -1.77388653e-01 -8.59086156e-01 7.31336474e-01
6.75525665e-01 -4.09511209e-01 7.51561821e-01 9.28013563e-01
2.54141271e-01 -2.31568459e-02 -6.26557231e-01 8.60746205e-01
-1.13215975e-01 -1.33284378e+00 2.55656075e-02 2.51504064e-01
8.71982276e-01 -8.72997195e-02 4.92619067e-01 3.01841736e-01
6.81644678e-01 -5.69737136e-01 9.17742133e-01 5.53613961e-01
6.34136736e-01 -7.92033195e-01 1.42828286e-01 1.36366889e-01
-1.07840133e+00 -6.23125017e-01 2.58254353e-02 2.66402096e-01
3.94183189e-01 7.14497983e-01 -6.82946920e-01 3.66849303e-01
8.44294488e-01 1.55463919e-01 -4.07175958e-01 1.12558067e+00
3.37739706e-01 1.55634135e-01 -2.63058424e-01 -3.23270530e-01
-3.55233401e-02 -3.57208848e-01 5.39767623e-01 1.01283550e+00
6.21058345e-01 -4.41246405e-02 1.69497997e-01 5.79297245e-01
-1.88077345e-01 5.34476697e-01 8.32934380e-02 -2.25068301e-01
7.85391450e-01 1.44890010e+00 -6.59905553e-01 -9.41991135e-02
-5.08703068e-02 7.75900543e-01 -5.58280237e-02 2.33591318e-01
-8.54910076e-01 -3.37454289e-01 7.70117402e-01 7.28124082e-01
5.18352449e-01 -5.54970838e-02 -4.00325805e-01 -6.85165882e-01
9.22577754e-02 -1.24844372e+00 5.03589213e-01 -6.19773507e-01
-1.20630693e+00 4.92329001e-01 6.02031983e-02 -1.08199978e+00
-2.89459378e-01 1.43157330e-03 -2.38006696e-01 8.70109320e-01
-1.08102727e+00 -8.21784019e-01 -2.91782379e-01 3.72915089e-01
5.38432777e-01 -1.20555788e-01 3.60883862e-01 6.10324800e-01
-3.69507670e-01 7.58652985e-01 5.10271788e-01 -8.03822815e-01
8.77901256e-01 -8.81995678e-01 3.66794467e-01 3.23354274e-01
-3.26520771e-01 9.61477339e-01 5.54852903e-01 -8.53128374e-01
-1.57124984e+00 -5.05412698e-01 6.46858633e-01 -3.13481122e-01
6.52090192e-01 -3.35164666e-01 -3.63328189e-01 -1.64396409e-02
1.29794002e-01 -7.55521059e-01 8.90930653e-01 5.78445673e-01
-2.17977390e-01 -2.69769132e-01 -1.22280467e+00 9.23445821e-01
1.15863216e+00 -2.16426134e-01 1.36931017e-01 2.71712422e-01
5.97595036e-01 -4.09291118e-01 -1.14677441e+00 1.26434699e-01
1.06857955e+00 -6.92777812e-01 9.66104209e-01 -2.28837520e-01
5.46075523e-01 -5.50755486e-02 -3.40990484e-01 -1.07553291e+00
-8.77272010e-01 -8.77914190e-01 -7.38747120e-02 1.40136766e+00
7.60743678e-01 -4.19823170e-01 6.48309052e-01 8.30577374e-01
-7.62802586e-02 -1.14631581e+00 -3.83796781e-01 -4.78323013e-01
-2.29906589e-01 1.15382433e-01 7.78455496e-01 6.47841990e-01
1.18213966e-01 3.56193900e-01 -2.70738006e-01 -3.12007278e-01
9.52820703e-02 4.82046485e-01 7.33420491e-01 -1.05126667e+00
-5.25979340e-01 -9.66942608e-01 1.94926843e-01 -1.23138165e+00
-8.38682353e-01 -7.27973282e-01 -4.79960620e-01 -1.59919691e+00
2.48814747e-01 -8.68403375e-01 -3.32835585e-01 1.92464247e-01
3.62625942e-02 5.25938086e-02 4.12177920e-01 5.41418672e-01
-9.77071822e-01 7.13661760e-02 1.26438105e+00 1.75373778e-01
-6.83122098e-01 8.10970888e-02 -1.59314060e+00 1.64712206e-01
7.29281783e-01 -1.46243304e-01 -6.09172702e-01 -2.39174917e-01
8.64313900e-01 1.51041336e-02 -2.76970655e-01 -5.79829931e-01
6.10805154e-02 -3.04725677e-01 4.00254846e-01 -6.64416850e-01
4.47260082e-01 -6.12437665e-01 6.04183435e-01 1.99781850e-01
-4.47983742e-01 5.82885861e-01 1.55855134e-01 1.59137309e-01
1.46326393e-01 -9.43287387e-02 2.70111173e-01 -3.90412137e-02
-4.39226300e-01 2.32523769e-01 -2.69177824e-01 7.35259578e-02
7.82273173e-01 -3.85576665e-01 -4.57044959e-01 -7.12853789e-01
-5.12154877e-01 6.74229562e-01 7.78498113e-01 7.46831298e-01
1.20902032e-01 -1.22513521e+00 -5.89259684e-01 -8.95815715e-02
1.44786701e-01 -4.12171572e-01 4.21953976e-01 6.25135064e-01
-4.61041033e-01 3.11673224e-01 -3.15247416e-01 -2.75704563e-01
-1.20806849e+00 3.42240632e-01 -1.06898002e-01 -4.56814945e-01
9.47259441e-02 1.11933732e+00 -5.11966385e-02 -1.02689937e-01
3.58906597e-01 -1.80727497e-01 -2.44607702e-01 5.52345753e-01
2.34947324e-01 6.42396629e-01 6.33674711e-02 -3.14405173e-01
-2.53626168e-01 2.79959679e-01 -4.93928403e-01 -1.38228849e-01
1.11424291e+00 -3.81113976e-01 -3.09477579e-02 5.38032092e-02
1.02389109e+00 6.50931060e-01 -9.29737866e-01 -4.12127352e-04
-1.21991225e-01 -6.89910769e-01 -1.16271168e-01 -1.28890240e+00
-8.96397173e-01 -1.03319637e-01 5.77683091e-01 7.86261857e-01
1.33706272e+00 6.77160174e-02 9.42430377e-01 -3.58584374e-02
2.37105265e-01 -1.38809729e+00 2.68016040e-01 1.98348403e-01
1.05236912e+00 -8.50980580e-01 3.63171220e-01 -2.84390599e-01
-1.13297772e+00 6.92496121e-01 4.58033800e-01 1.44698948e-01
6.00265205e-01 4.12013941e-02 -2.90247537e-02 -2.66083360e-01
-9.97439265e-01 6.18853383e-02 3.11686516e-01 -1.66307211e-01
5.65402925e-01 1.40135884e-01 -6.61265850e-01 5.88050187e-01
-3.67298990e-01 -1.60824418e-01 -7.52633512e-02 9.29742992e-01
-4.67777103e-01 -1.41865683e+00 -2.30780318e-01 8.74265254e-01
-6.15832865e-01 -3.11485380e-01 -5.37599266e-01 5.12341678e-01
6.38715103e-02 9.89249170e-01 1.33274226e-02 -5.38516462e-01
6.02729678e-01 -3.00501257e-01 9.05877426e-02 -4.64107394e-01
-1.18091953e+00 4.95083392e-01 4.83400851e-01 -5.78459978e-01
2.57978797e-01 -8.49586606e-01 -9.47087228e-01 -6.43575966e-01
-4.74388301e-01 3.39315832e-01 9.22350883e-01 1.88349903e-01
1.11341846e+00 4.25687999e-01 9.53555286e-01 -5.01491785e-01
-7.83609927e-01 -6.74365282e-01 -6.31604493e-01 3.64246070e-01
-8.72826129e-02 -5.56780398e-01 -1.77611381e-01 -4.90493536e-01] | [9.948929786682129, 5.756075382232666] |
0b8c2e67-732b-4bb8-9bef-503740eb12b0 | mlengineer-at-semeval-2020-task-7-bert-flair | null | null | https://aclanthology.org/2020.semeval-1.136 | https://aclanthology.org/2020.semeval-1.136.pdf | MLEngineer at SemEval-2020 Task 7: BERT-Flair Based Humor Detection Model (BFHumor) | Task 7, Assessing the Funniness of Edited News Headlines, in the International Workshop SemEval2020 introduces two sub-tasks to predict the funniness values of edited news headlines from the Reddit website. This paper proposes the BFHumor model of the MLEngineer team that participates in both sub-tasks in this competition. The BFHumor{'}s model is defined as a BERT-Flair based humor detection model that is a combination of different pre-trained models with various Natural Language Processing (NLP) techniques. The Bidirectional Encoder Representations from Transformers (BERT) regressor is considered the primary pre-trained model in our approach, whereas Flair is the main NLP library. It is worth mentioning that the BFHumor model has been ranked 4th in sub-task1 with a root mean square error (RMSE) value of 0.51966, and it is 0.02 away from the first ranked model. Also, the team is ranked 12th in the sub-task2 with an accuracy of 0.62291, which is 0.05 away from the top-ranked model. Our results indicate that the BFHumor model is one of the top models for detecting humor in the text. | ['Mahmoud Hammad', 'Malak Abdullah', 'Fara Shatnawi'] | 2020-12-01 | null | null | null | semeval-2020 | ['humor-detection'] | ['natural-language-processing'] | [-4.04960603e-01 5.93744934e-01 4.03587632e-02 1.30079299e-01
-6.32885158e-01 -3.30285132e-01 1.15467978e+00 3.60833794e-01
-5.13747215e-01 6.69802487e-01 9.12207723e-01 -2.58713663e-01
1.17128007e-01 -6.12154245e-01 -6.57171369e-01 -1.49772286e-01
1.97582111e-01 2.01553851e-01 1.23881297e-02 -7.43475974e-01
7.47665405e-01 5.05934320e-02 -1.23743820e+00 7.86517203e-01
7.36800313e-01 9.15923297e-01 -9.85102579e-02 8.42703223e-01
5.11752851e-02 1.94345820e+00 -9.26428437e-01 -9.78741109e-01
-1.37860015e-01 -7.45462835e-01 -1.10808671e+00 -5.28349161e-01
4.17025357e-01 -1.74202383e-01 -4.76179361e-01 9.58305120e-01
4.92068976e-01 1.83234997e-02 5.21719396e-01 -1.05744517e+00
-5.01849234e-01 1.13965225e+00 -2.37540066e-01 2.70262867e-01
4.97383893e-01 -5.95248006e-02 1.27703083e+00 -1.17350769e+00
8.75989020e-01 1.12299478e+00 7.71523833e-01 3.01582336e-01
-7.29528666e-01 -4.65351284e-01 -5.09348452e-01 5.76693952e-01
-1.01066458e+00 -4.16884363e-01 7.64443278e-01 -8.35310936e-01
1.02398360e+00 3.94877434e-01 3.33515674e-01 1.25704277e+00
5.20838976e-01 1.03216290e+00 1.14758074e+00 -3.14338058e-01
4.11865190e-02 4.84248012e-01 2.78397292e-01 6.83539867e-01
-2.87355751e-01 5.89690208e-02 -1.09040463e+00 -1.72259547e-02
1.81880102e-01 -5.47937393e-01 -4.42879677e-01 5.49955487e-01
-1.17353940e+00 1.16091335e+00 5.69427073e-01 6.17287397e-01
-4.66700286e-01 -4.09504324e-02 8.09292555e-01 5.55119455e-01
6.39325202e-01 8.81619990e-01 -1.78903490e-01 -4.39292133e-01
-1.21722341e+00 4.82682049e-01 1.16856825e+00 5.72477579e-01
1.86789498e-01 -3.71326171e-02 -5.51240802e-01 8.04877281e-01
-2.76801977e-02 2.97907349e-02 6.57208681e-01 -7.16484845e-01
3.79430801e-01 5.89062810e-01 1.62697896e-01 -1.26233244e+00
-6.07862890e-01 -8.97168458e-01 -6.85614645e-01 1.50216296e-01
3.71522605e-01 -2.46367857e-01 -2.94240147e-01 1.26426947e+00
-3.87076914e-01 -2.24718302e-01 -1.67741910e-01 1.06626296e+00
1.45370376e+00 1.01622057e+00 -2.12542638e-01 -3.97105515e-01
1.18828523e+00 -1.30099714e+00 -7.99490392e-01 -1.06697291e-01
5.95678151e-01 -9.62658644e-01 1.16379333e+00 8.86907220e-01
-1.17293000e+00 -4.14332360e-01 -1.33942688e+00 -4.44996953e-01
-4.32891309e-01 1.99905764e-02 -1.10997863e-01 1.52845293e-01
-7.75739133e-01 8.00851107e-01 4.63518500e-02 -2.50440091e-01
-5.79244718e-02 -5.48272491e-01 -1.21640027e-01 4.49944079e-01
-1.74271405e+00 1.53735363e+00 2.65999943e-01 -2.59612054e-01
-1.03726101e+00 -8.24535429e-01 -3.23917687e-01 7.09063634e-02
1.77927300e-01 -9.89753976e-02 1.33303308e+00 -8.50569725e-01
-1.37484765e+00 1.12192547e+00 1.07307345e-01 -9.85533714e-01
1.05561006e+00 -6.36958897e-01 -5.13035953e-01 -3.78033184e-02
2.82737482e-02 7.59989116e-03 7.70373106e-01 -9.73561585e-01
-3.33680063e-01 5.12411147e-02 4.57996242e-02 -1.19271524e-01
-4.78554875e-01 5.42451859e-01 2.15554506e-01 -5.79585969e-01
-5.93433022e-01 -5.04972398e-01 2.55778611e-01 -6.19462371e-01
-6.53236508e-01 -3.56356621e-01 3.95266384e-01 -1.16410148e+00
1.83192909e+00 -1.92148805e+00 9.81530100e-02 -2.13603392e-01
4.92912054e-01 1.14738449e-01 2.62077153e-01 7.33897448e-01
-5.91828488e-02 2.39684686e-01 1.98072553e-01 -1.96435809e-01
1.77959919e-01 -4.79565322e-01 -5.49052596e-01 4.20695156e-01
-4.10686433e-02 8.21696579e-01 -1.05056119e+00 -2.86219954e-01
-9.56084672e-03 2.77208656e-01 -1.83681875e-01 2.29961470e-01
-1.38847172e-01 8.58706832e-02 5.13519347e-02 -1.47734368e-02
2.65558928e-01 -1.09841526e-01 -2.49115050e-01 -6.66655898e-02
-6.39674485e-01 1.00692284e+00 -5.27985752e-01 1.18284941e+00
-4.41932440e-01 1.16186523e+00 -3.30072522e-01 -3.07414174e-01
1.39398813e+00 4.79259312e-01 2.02352375e-01 -9.72058475e-01
3.07279795e-01 3.33121300e-01 2.67850608e-03 -5.25314569e-01
9.60417032e-01 -4.42792982e-01 -3.41716498e-01 3.22119564e-01
1.41616240e-01 -1.04316950e-01 2.51144975e-01 4.94893104e-01
1.32421362e+00 -1.59572005e-01 6.49266422e-01 -4.44329709e-01
7.36590326e-01 1.44630754e-02 4.80591923e-01 6.97987795e-01
-2.98773766e-01 5.89280188e-01 1.02517724e+00 -5.60053647e-01
-1.20420408e+00 -5.48977196e-01 -1.59780920e-01 1.12916577e+00
-3.36902559e-01 -9.31680799e-01 -6.30739391e-01 -5.96811771e-01
-2.25686014e-01 1.57800388e+00 -7.64237583e-01 -1.98004305e-01
-3.17159563e-01 -2.84081489e-01 9.03475821e-01 1.80623196e-02
4.30690080e-01 -1.35315430e+00 -8.90720785e-01 2.13555962e-01
-6.74017429e-01 -8.51494551e-01 -2.81598046e-02 2.76629657e-01
-2.69063681e-01 -9.84612107e-01 -5.17147481e-01 -2.41962329e-01
-1.64253578e-01 -2.69014508e-01 1.31221128e+00 1.42073065e-01
1.62972093e-01 -3.00248146e-01 -8.52031410e-01 -5.66531360e-01
-8.64814937e-01 -3.58505249e-02 -2.23271266e-01 4.69133742e-02
4.36849624e-01 -7.04941079e-02 -1.19579941e-01 7.40536302e-02
-4.51785505e-01 2.72897333e-01 2.26695426e-02 1.01073182e+00
-1.92323387e-01 1.68356486e-02 6.78342164e-01 -9.54569161e-01
9.82206404e-01 -6.91796660e-01 -1.35631561e-01 -9.01297703e-02
-6.82342827e-01 -2.33900413e-01 1.01466250e+00 4.60041836e-02
-1.00644863e+00 -5.13826668e-01 -1.86773047e-01 -6.57695085e-02
8.50624964e-02 8.51470113e-01 1.76843703e-01 4.72963899e-01
1.20831406e+00 -2.01440211e-02 -4.43446040e-01 -6.77935123e-01
2.65025169e-01 8.44743550e-01 7.00229347e-01 -4.94654849e-02
6.01300180e-01 -1.34083375e-01 -4.25099850e-01 -8.17653775e-01
-1.28239870e+00 -4.73574638e-01 -4.31334734e-01 -7.28818893e-01
7.10904062e-01 -8.73267353e-01 -6.35225654e-01 4.52508777e-01
-1.64664781e+00 -2.77012557e-01 -9.30736065e-02 2.68901616e-01
-5.42799950e-01 1.92762375e-01 -9.89470124e-01 -9.70604837e-01
-6.98396325e-01 -4.94189620e-01 4.30676639e-01 -7.96039924e-02
-8.61738861e-01 -7.49206662e-01 1.99903861e-01 7.70621777e-01
3.09621900e-01 4.97957081e-01 9.40305650e-01 -1.00372112e+00
3.55845630e-01 -3.51645589e-01 -4.07175794e-02 4.98020113e-01
-5.25557935e-01 -1.57501046e-02 -1.08103585e+00 5.86876050e-02
3.82646471e-01 -6.80685282e-01 1.05851460e+00 -5.47488891e-02
6.44973457e-01 -7.82522798e-01 4.03698683e-01 3.17581929e-02
1.13160682e+00 -1.58322185e-01 1.03582704e+00 8.54325771e-01
3.99561226e-01 6.98496342e-01 6.60683393e-01 9.18208182e-01
2.87803710e-01 5.27563453e-01 5.87684631e-01 2.52427101e-01
-2.25389600e-01 -6.87566221e-01 9.88108337e-01 9.49600577e-01
-7.60972425e-02 -1.82960302e-01 -1.03184724e+00 4.16153699e-01
-1.85789585e+00 -1.36073494e+00 -9.79849637e-01 2.05625463e+00
8.53598893e-01 5.61955273e-01 3.84330153e-01 4.05680716e-01
4.92556751e-01 3.92200172e-01 9.85753313e-02 -1.08570528e+00
-2.72066504e-01 -3.25209200e-02 8.23649168e-02 6.69097304e-01
-8.44497919e-01 9.14487958e-01 5.74166107e+00 6.23318970e-01
-8.04954410e-01 2.98386604e-01 3.55843097e-01 -1.84986055e-01
-1.86819986e-01 1.30939977e-02 -6.06582880e-01 7.80835569e-01
1.46120203e+00 -6.42315626e-01 2.73318827e-01 7.29576647e-01
6.07983530e-01 -1.20561637e-01 -9.33281422e-01 7.57253349e-01
6.57856286e-01 -1.22698736e+00 -1.50496840e-01 -1.43897489e-01
6.56527340e-01 1.21540613e-01 -1.61310896e-01 6.50880933e-01
2.15654433e-01 -1.18751919e+00 1.26420975e+00 7.14575768e-01
2.06144005e-01 -9.11997259e-01 1.12758386e+00 7.90525913e-01
-3.71217042e-01 -2.68797398e-01 -3.69455606e-01 -4.89932418e-01
3.54411125e-01 1.15421820e+00 -9.38353777e-01 2.34702334e-01
6.51928008e-01 7.91967630e-01 -6.83817685e-01 1.02298689e+00
-5.42324960e-01 8.95687640e-01 2.51036316e-01 -4.16264802e-01
2.23493174e-01 2.09463477e-01 9.81111884e-01 1.69151163e+00
1.80542663e-01 -2.01498821e-01 -2.22201616e-01 9.84569192e-01
-3.71922821e-01 2.81281561e-01 -3.33493590e-01 -1.97713941e-01
2.78723329e-01 1.33133316e+00 -3.84435058e-02 -3.77020150e-01
-1.97081571e-03 8.99588227e-01 3.35866004e-01 -2.01348484e-01
-9.98138905e-01 -5.18988013e-01 -1.20654650e-01 4.20860320e-01
-1.30456805e-01 1.36912897e-01 -5.44067264e-01 -9.05559659e-01
-3.56221229e-01 -1.04905117e+00 3.11475575e-01 -1.07013357e+00
-1.24269152e+00 9.27579939e-01 -4.41370875e-01 -1.07353604e+00
-1.41588792e-01 -4.68445212e-01 -7.42530823e-01 7.54473627e-01
-1.16737568e+00 -9.62846696e-01 -3.84856194e-01 3.27212423e-01
5.95132351e-01 -3.77024442e-01 6.38696313e-01 1.39590219e-01
-4.17985231e-01 4.20805991e-01 -8.50146934e-02 2.51611978e-01
8.07168007e-01 -1.31873977e+00 1.81958631e-01 7.34706104e-01
-8.54542777e-02 2.63074309e-01 1.29108047e+00 -6.12404943e-01
-8.61476302e-01 -8.60326767e-01 2.10063291e+00 -5.66474438e-01
1.15325069e+00 -1.15481161e-01 -9.36054289e-01 4.14783210e-01
6.27516627e-01 -8.45564425e-01 5.60472071e-01 2.24414453e-01
-6.53845072e-01 1.97751299e-01 -8.36912036e-01 2.90674239e-01
4.19761658e-01 -6.42857730e-01 -1.01550078e+00 6.02158189e-01
6.22800171e-01 -2.50794888e-01 -8.15120876e-01 5.40106185e-02
3.26975137e-01 -1.33468699e+00 2.91791052e-01 -6.70317769e-01
1.65236437e+00 -1.33757085e-01 -3.25689986e-02 -1.38347626e+00
-5.31211436e-01 -7.30497479e-01 -2.22022399e-01 1.36867046e+00
4.95795220e-01 -2.12872066e-02 2.90455341e-01 1.97242871e-01
-1.20852135e-01 -4.04367566e-01 -7.54767418e-01 -6.79576218e-01
3.61965120e-01 -5.08122802e-01 -6.96197897e-02 1.11229384e+00
6.78603888e-01 9.71725345e-01 -9.46829915e-01 -6.89131021e-01
3.57227921e-01 -1.45335093e-01 7.99085915e-01 -1.17705011e+00
-3.52372497e-01 -7.54485667e-01 -2.55912006e-01 -7.06119239e-01
7.00038150e-02 -1.12604141e+00 1.92175254e-01 -1.45657134e+00
4.70299721e-01 5.35574317e-01 -3.28870803e-01 3.99641365e-01
1.46431297e-01 1.33307651e-02 6.39012992e-01 1.23277299e-01
-3.99592131e-01 5.08228004e-01 9.93739307e-01 -1.52745783e-01
-7.09510595e-02 -9.96192992e-02 -7.48911560e-01 7.48287618e-01
7.95198739e-01 -5.63332915e-01 1.49814397e-01 4.03739810e-02
7.95087814e-01 5.42359203e-02 6.51654959e-01 -8.08778048e-01
2.02906892e-01 1.20214736e-02 1.09129108e-01 -6.81715846e-01
5.42761646e-02 -2.80177027e-01 -1.10432468e-01 5.68243861e-01
-8.51000726e-01 -5.90647757e-02 -1.17209695e-01 3.38953994e-02
-3.25759470e-01 -6.29362941e-01 8.98393810e-01 -2.84245640e-01
-3.62861395e-01 -5.31287372e-01 -6.84563577e-01 1.89202875e-01
5.12777269e-01 1.70158863e-01 -6.65300071e-01 -8.08973908e-01
-4.71432030e-01 -6.62634000e-02 1.23733141e-01 6.36440933e-01
5.30774176e-01 -1.10362303e+00 -1.34265113e+00 -4.96355861e-01
4.53866310e-02 -8.17102194e-01 -2.14590319e-02 1.09634435e+00
-5.11724114e-01 3.38037521e-01 -3.14160973e-01 1.47625357e-01
-1.10223246e+00 3.94195169e-01 1.97188824e-01 -5.75619280e-01
-6.87874675e-01 7.92280674e-01 -3.17097872e-01 -3.87304872e-01
4.58121076e-02 4.08468157e-01 -5.61011553e-01 4.52685982e-01
9.16785836e-01 1.16748607e+00 3.09649706e-01 -9.09527957e-01
-2.26021022e-01 -1.78594485e-01 -1.40082583e-01 -3.08944881e-01
1.36986947e+00 1.89575642e-01 -4.46301371e-01 9.24297690e-01
1.39786053e+00 7.45563954e-02 -5.45932472e-01 -1.09134279e-01
4.03006405e-01 -2.69441158e-01 3.82713050e-01 -1.38668358e+00
-5.38465679e-01 1.04308391e+00 -3.29886898e-02 3.90084535e-01
7.68665373e-01 -2.95718849e-01 9.02594626e-01 2.58343846e-01
1.95204034e-01 -1.50417244e+00 2.55058855e-01 1.21516919e+00
1.81407535e+00 -7.94066191e-01 -5.28437970e-03 1.37655854e-01
-1.06052625e+00 1.43135786e+00 5.46893418e-01 -1.79578841e-01
3.00303221e-01 4.09261174e-02 -7.09649175e-02 -3.95717442e-01
-1.05913627e+00 2.94787288e-01 4.26514268e-01 -1.32380918e-01
1.01352990e+00 1.43342301e-01 -9.41028774e-01 1.06037259e+00
-9.35506880e-01 1.32160246e-01 9.34665501e-01 3.12224925e-01
-7.81785250e-01 -3.21125627e-01 -3.67418081e-01 3.38723034e-01
-5.34310520e-01 -2.70002753e-01 -1.06769824e+00 4.22773570e-01
-7.76508301e-02 1.27066207e+00 -3.66401255e-01 -9.51374233e-01
4.08976227e-01 2.27035493e-01 -4.64429101e-03 -2.61716157e-01
-1.37998378e+00 -1.98187917e-01 6.52376235e-01 -5.06086946e-01
-1.32715264e-02 -2.88160414e-01 -1.12243211e+00 -8.87035072e-01
-1.47111952e-01 3.69408548e-01 3.88277441e-01 9.37469959e-01
-1.82233956e-02 4.91788268e-01 7.82416999e-01 -4.92310971e-01
-6.69182837e-01 -1.34171462e+00 -5.61478794e-01 5.74554443e-01
1.89622030e-01 -3.26381415e-01 -7.36918807e-01 8.82495940e-03] | [8.848548889160156, 11.035077095031738] |
13dc19d5-692f-4d62-83be-c850060871de | leveraging-just-a-few-keywords-for-fine | 1909.00415 | null | https://arxiv.org/abs/1909.00415v1 | https://arxiv.org/pdf/1909.00415v1.pdf | Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training | User-generated reviews can be decomposed into fine-grained segments (e.g., sentences, clauses), each evaluating a different aspect of the principal entity (e.g., price, quality, appearance). Automatically detecting these aspects can be useful for both users and downstream opinion mining applications. Current supervised approaches for learning aspect classifiers require many fine-grained aspect labels, which are labor-intensive to obtain. And, unfortunately, unsupervised topic models often fail to capture the aspects of interest. In this work, we consider weakly supervised approaches for training aspect classifiers that only require the user to provide a small set of seed words (i.e., weakly positive indicators) for the aspects of interest. First, we show that current weakly supervised approaches do not effectively leverage the predictive power of seed words for aspect detection. Next, we propose a student-teacher approach that effectively leverages seed words in a bag-of-words classifier (teacher); in turn, we use the teacher to train a second model (student) that is potentially more powerful (e.g., a neural network that uses pre-trained word embeddings). Finally, we show that iterative co-training can be used to cope with noisy seed words, leading to both improved teacher and student models. Our proposed approach consistently outperforms previous weakly supervised approaches (by 14.1 absolute F1 points on average) in six different domains of product reviews and six multilingual datasets of restaurant reviews. | ['Luis Gravano', 'Giannis Karamanolakis', 'Daniel Hsu'] | 2019-09-01 | leveraging-just-a-few-keywords-for-fine-1 | https://aclanthology.org/D19-1468 | https://aclanthology.org/D19-1468.pdf | ijcnlp-2019-11 | ['aspect-category-detection'] | ['natural-language-processing'] | [ 1.54931620e-01 2.66806841e-01 -7.49859035e-01 -6.14142716e-01
-1.15474606e+00 -9.60055232e-01 6.83164775e-01 7.41826296e-01
-3.11743587e-01 5.35063386e-01 1.36663184e-01 -4.81402695e-01
3.08021575e-01 -9.69744265e-01 -6.25868738e-01 -5.78928590e-01
2.15154946e-01 4.09645081e-01 1.10708542e-01 -2.16560751e-01
3.43062639e-01 8.74875393e-03 -1.56960642e+00 1.44321650e-01
9.10773516e-01 8.52992654e-01 6.09982945e-03 4.59913522e-01
-4.59561616e-01 2.61449218e-01 -7.65863001e-01 -6.31488383e-01
-5.96646108e-02 -2.60240585e-01 -5.52214563e-01 4.17824805e-01
2.42859855e-01 -1.23156868e-01 6.13163292e-01 1.01852369e+00
1.52355537e-01 1.84582938e-02 8.91042054e-01 -1.00392568e+00
-7.20908940e-01 8.09239626e-01 -6.62204504e-01 1.50064910e-02
2.18384326e-01 -1.31783426e-01 1.69881630e+00 -1.22460997e+00
3.43124002e-01 9.75584090e-01 4.87161964e-01 2.48514861e-01
-1.18552411e+00 -4.63677913e-01 6.20801032e-01 -2.51513988e-01
-9.01824653e-01 -1.10471494e-01 7.81912863e-01 -2.70660132e-01
1.03036189e+00 1.52565077e-01 8.13225269e-01 9.19714332e-01
-9.67110917e-02 1.08686030e+00 1.16134810e+00 -4.15185452e-01
4.70032871e-01 8.15461993e-01 7.13700652e-01 3.86337489e-01
6.68305039e-01 -3.45441043e-01 -4.74665582e-01 -5.02307594e-01
2.60835499e-01 1.21874211e-04 1.04970457e-02 -3.74725938e-01
-8.11310112e-01 1.29916418e+00 -7.52845928e-02 2.19884485e-01
-4.81968880e-01 -1.74455851e-01 3.04350853e-01 2.15193942e-01
1.00780475e+00 6.71748579e-01 -1.03717589e+00 -2.10887149e-01
-7.92731941e-01 2.35347468e-02 1.03106773e+00 1.10693908e+00
1.06961608e+00 -1.12763621e-01 -7.53532127e-02 1.03313041e+00
5.01248479e-01 5.63761473e-01 4.16643173e-01 -2.79056340e-01
2.80675739e-01 7.80066252e-01 2.12948099e-02 -6.00300014e-01
-1.07210480e-01 -5.78626990e-01 -2.84723163e-01 -1.65031612e-01
6.52077422e-02 -2.51265556e-01 -1.08152139e+00 1.41382253e+00
5.73562026e-01 -1.65200606e-02 2.32389584e-01 4.36657667e-01
8.38203311e-01 6.53005898e-01 3.04568827e-01 -1.85160398e-01
1.77355754e+00 -1.23468995e+00 -5.76311588e-01 -5.13861299e-01
6.52682900e-01 -9.78033960e-01 1.21190464e+00 4.33239937e-01
-8.32488477e-01 -3.61498505e-01 -8.86490047e-01 2.65869290e-01
-7.21044838e-01 3.03404123e-01 8.21789265e-01 7.56156027e-01
-7.26268947e-01 2.56569684e-01 -5.94837666e-01 -3.03577960e-01
3.22300225e-01 3.43956143e-01 -1.98001251e-01 6.48970380e-02
-1.01970279e+00 5.98982453e-01 1.31058276e-01 -3.23255807e-01
-7.84425795e-01 -6.15407050e-01 -1.18953145e+00 5.90565540e-02
5.83936810e-01 -3.79098147e-01 1.36098516e+00 -1.27845788e+00
-1.22677374e+00 7.85442650e-01 -3.60953122e-01 -2.52492905e-01
-2.96140969e-01 -6.11646295e-01 -3.48082691e-01 1.73060834e-01
4.19150174e-01 5.16946793e-01 1.17333984e+00 -1.36353445e+00
-9.42251384e-01 -1.64065629e-01 4.91974443e-01 2.15253264e-01
-6.97663248e-01 2.36261189e-01 -5.00444114e-01 -7.12695479e-01
-2.46014297e-01 -9.51492727e-01 -4.21852767e-01 -3.28300774e-01
-5.13985872e-01 -5.17511487e-01 6.92875385e-01 -2.16150329e-01
1.14637721e+00 -2.05722165e+00 -3.13932449e-01 3.13849688e-01
1.32184476e-01 3.73547018e-01 -3.13320160e-01 1.87276378e-01
-1.46858975e-01 3.77313823e-01 -2.20415652e-01 -4.19345856e-01
-1.03416517e-01 2.22914472e-01 -3.52848887e-01 8.27435330e-02
6.95486069e-01 9.01605308e-01 -1.06522930e+00 -3.19938809e-01
-9.97905433e-02 4.95287776e-01 -5.27093768e-01 1.78021207e-01
-3.97487938e-01 -1.50526643e-01 -6.70325041e-01 7.83586979e-01
4.13913131e-01 -3.91581684e-01 -1.77245066e-02 -1.71597615e-01
2.70021912e-02 7.89778590e-01 -1.02607238e+00 1.13201272e+00
-7.73422062e-01 3.60805631e-01 -1.23663686e-01 -9.96209621e-01
1.01637244e+00 3.32944602e-01 2.93157518e-01 -7.91349038e-02
1.17545664e-01 2.62001693e-01 -2.31318131e-01 -2.24428758e-01
6.75857365e-01 -3.56974691e-01 -2.67107129e-01 8.79735291e-01
2.56023556e-01 -4.36355233e-01 3.57549489e-01 1.53051943e-01
8.84307861e-01 -7.90529102e-02 7.19078779e-01 -2.31504947e-01
4.86695617e-01 1.81423321e-01 3.87399107e-01 5.33659518e-01
-9.82233584e-02 6.36923492e-01 6.96217120e-01 -1.85341194e-01
-9.19919014e-01 -8.75812590e-01 -6.14679195e-02 1.38228047e+00
-1.09216347e-01 -7.19231606e-01 -5.43380082e-01 -1.24897051e+00
-7.92259723e-02 7.91059077e-01 -6.85926020e-01 -1.58700332e-01
-2.90430903e-01 -7.91720986e-01 4.86951284e-02 6.66108370e-01
-1.79396883e-01 -9.73232806e-01 -2.75922060e-01 2.75529593e-01
1.35228947e-01 -1.04172146e+00 -4.02702779e-01 7.46239960e-01
-9.66660738e-01 -9.67739642e-01 -5.67220330e-01 -8.62669170e-01
9.73522723e-01 4.81409520e-01 1.55588078e+00 8.68308768e-02
3.16840082e-01 4.49586958e-01 -7.19558716e-01 -5.90682983e-01
-1.96867526e-01 3.73003453e-01 -6.42899191e-03 8.84532090e-03
1.00807166e+00 -4.11192924e-01 -4.85585570e-01 2.52288669e-01
-9.83319283e-01 -3.42655271e-01 8.30534339e-01 9.72897768e-01
7.41081357e-01 -4.32311604e-03 7.37667322e-01 -1.49258280e+00
9.12234306e-01 -6.05338216e-01 -4.20430720e-01 1.62133723e-01
-9.20112312e-01 5.37346452e-02 6.00224793e-01 -7.65343130e-01
-9.44973469e-01 1.73881650e-02 -2.85003513e-01 1.23301990e-01
-3.27223271e-01 8.12841713e-01 -2.15093102e-02 3.96961987e-01
5.14315724e-01 1.78639770e-01 -2.81051069e-01 -2.60944217e-01
5.42456031e-01 8.51509094e-01 -1.62136719e-01 -4.75087345e-01
8.96811306e-01 2.85235614e-01 -5.89338362e-01 -6.84873998e-01
-1.46835291e+00 -9.01328325e-01 -5.54122865e-01 1.37015790e-01
5.91180205e-01 -1.15708399e+00 9.92979333e-02 1.37619689e-01
-1.02161539e+00 3.78819406e-02 -5.04567683e-01 4.50690806e-01
-2.01328367e-01 8.11896250e-02 -5.01541376e-01 -7.95885205e-01
-5.00559688e-01 -9.20353711e-01 1.37212074e+00 3.86335552e-01
-4.89102483e-01 -1.17949367e+00 7.39576817e-02 3.44820678e-01
3.66106778e-01 -2.40883484e-01 9.33634222e-01 -1.27194679e+00
-3.26360077e-01 -4.50915426e-01 2.68078297e-02 6.16504014e-01
5.88316560e-01 5.07830568e-02 -1.10996068e+00 -2.66493894e-02
1.22618496e-01 -4.17179555e-01 8.68533015e-01 2.73774058e-01
6.24120891e-01 -2.60343075e-01 -2.78168440e-01 1.58508420e-02
1.12996054e+00 3.43993381e-02 2.70987272e-01 3.50796521e-01
5.94017327e-01 7.18840778e-01 1.20020747e+00 3.29867154e-01
4.22095597e-01 3.23172837e-01 -7.95910433e-02 -2.94555515e-01
2.28418931e-01 -2.84388632e-01 6.67665422e-01 1.03034687e+00
1.50697783e-01 -1.27387792e-01 -5.54948747e-01 1.00201654e+00
-1.47289932e+00 -4.81686413e-01 -1.39478385e-01 2.02260661e+00
1.17982841e+00 5.19514620e-01 1.68108463e-01 4.95035127e-02
4.61879790e-01 2.99728930e-01 -4.92556661e-01 -6.34612679e-01
7.18705356e-02 4.78515923e-01 1.86844632e-01 4.43308711e-01
-1.29891396e+00 1.06737471e+00 5.67056179e+00 7.83855319e-01
-9.27381098e-01 2.78252184e-01 7.82266080e-01 1.36057228e-01
-9.00844216e-01 1.83619887e-01 -1.20225632e+00 1.19875185e-01
9.65790987e-01 -2.02734128e-01 -2.36807093e-01 1.37286270e+00
8.85585546e-02 -2.45301813e-01 -1.05026281e+00 4.10178125e-01
4.11564022e-01 -8.26003611e-01 1.16973594e-01 1.79554056e-02
1.01981342e+00 -9.85004157e-02 1.04935408e-01 5.85206687e-01
4.89139140e-01 -7.94703662e-01 3.72861534e-01 -1.55186608e-01
5.89767575e-01 -8.15483630e-01 9.22794282e-01 3.17809284e-01
-1.25636125e+00 2.13595733e-01 -4.68896180e-01 1.38029784e-01
2.08943054e-01 1.14720523e+00 -8.19480598e-01 3.33057582e-01
5.12451172e-01 8.04154456e-01 -4.89416808e-01 6.18195176e-01
-7.45337665e-01 1.03625870e+00 -2.38891453e-01 -3.88185173e-01
3.92622799e-01 -1.95902392e-01 3.61292511e-01 1.13218510e+00
2.25962073e-01 -1.07947059e-01 2.03228280e-01 6.60300791e-01
-8.84581134e-02 4.50153083e-01 -6.69370651e-01 -2.60138184e-01
1.59685299e-01 1.70383060e+00 -8.72215450e-01 -6.32602334e-01
-8.57776463e-01 6.92696512e-01 3.31052810e-01 3.60619068e-01
-4.55317944e-01 -3.27786446e-01 8.20768356e-01 -5.94470976e-03
9.39994693e-01 -5.87382633e-03 -3.07239622e-01 -1.25855494e+00
7.84374774e-02 -1.13609970e+00 2.40215987e-01 -4.22789305e-01
-1.45188391e+00 7.80626833e-01 -2.26243690e-01 -1.25561810e+00
-4.71233219e-01 -6.51881278e-01 -7.54281700e-01 6.90913796e-01
-2.01070166e+00 -1.10324967e+00 8.37032571e-02 2.43502840e-01
7.69383132e-01 -4.26333286e-02 9.24591243e-01 4.33319584e-02
-1.99403286e-01 5.08371830e-01 -1.97091326e-01 9.55956280e-02
8.71758640e-01 -1.61349285e+00 5.64114571e-01 5.94449162e-01
5.58599353e-01 1.00775719e+00 6.86734021e-01 -6.90691710e-01
-1.09491682e+00 -1.22913039e+00 1.27365434e+00 -6.39755189e-01
9.12027597e-01 -4.65153754e-01 -7.87577033e-01 5.96732914e-01
2.13193074e-01 -1.05414301e-01 1.26030290e+00 6.78009152e-01
-6.37315512e-01 -4.20698822e-02 -9.24946785e-01 4.38046455e-01
3.44083220e-01 -6.21064186e-01 -7.83285201e-01 2.95071453e-01
9.60503519e-01 1.78857322e-03 -6.56958401e-01 2.62923747e-01
4.11773384e-01 -6.30333126e-01 6.90369546e-01 -7.60505736e-01
5.37734628e-01 -2.73600817e-01 -2.51158811e-02 -1.60767472e+00
6.56202585e-02 -2.70630747e-01 -2.23444924e-01 1.59437180e+00
1.11649048e+00 -5.74279904e-01 7.94442654e-01 4.56103474e-01
7.72194341e-02 -1.22615993e+00 -4.35149580e-01 -6.00463033e-01
-4.11364762e-03 -6.47570312e-01 5.06077051e-01 7.48280466e-01
8.42589065e-02 9.26953912e-01 -7.40823373e-02 5.64073771e-02
2.59970397e-01 5.13209760e-01 6.33899450e-01 -1.17523074e+00
-4.04053360e-01 -2.48865202e-01 -1.95121035e-01 -1.28385544e+00
2.45056450e-01 -5.53549826e-01 3.13335687e-01 -1.48947275e+00
4.43119824e-01 -5.98561525e-01 -3.80105078e-01 5.00957191e-01
-7.76765585e-01 2.68829644e-01 -2.59680063e-01 -1.10563427e-01
-7.45120585e-01 5.14195025e-01 8.57166827e-01 -2.69300014e-01
-2.49896958e-01 5.82036853e-01 -1.29850197e+00 8.92994583e-01
8.85899484e-01 -7.11087167e-01 -5.39142787e-01 -9.31823701e-02
5.26773334e-01 -5.62028766e-01 -1.74453169e-01 -2.81283647e-01
-4.76832651e-02 -3.23315337e-02 -5.63791441e-03 -6.08090699e-01
2.07000464e-01 -7.02458262e-01 -7.15504289e-01 3.04629579e-02
-3.86136502e-01 6.89416751e-02 -1.53080618e-05 6.36635303e-01
-6.04516745e-01 -6.20314240e-01 3.44671518e-01 -1.43687092e-02
-2.47972921e-01 2.09932446e-01 -4.90876973e-01 2.10014239e-01
8.98031950e-01 1.32077157e-01 -1.27926707e-01 -5.24323225e-01
-3.82159382e-01 1.39206827e-01 3.40155959e-01 4.66093659e-01
6.03739023e-01 -1.16987860e+00 -5.30690789e-01 1.41532019e-01
5.69765747e-01 3.82263735e-02 -3.68504167e-01 6.29056275e-01
2.48988882e-01 3.17119002e-01 4.87941593e-01 -5.11048734e-01
-1.38563776e+00 5.47992229e-01 -2.46515051e-01 -5.69284260e-01
-1.32121041e-01 8.77392173e-01 2.78384864e-01 -6.64966583e-01
3.98160219e-02 -5.73374033e-01 -5.80845952e-01 5.90344131e-01
5.37351131e-01 -2.97928452e-01 8.86186063e-02 -5.53397417e-01
-2.76164174e-01 6.20340109e-01 -3.87226462e-01 -1.42300323e-01
1.57671010e+00 -1.15302368e-03 -1.15300052e-01 6.45495415e-01
1.15112364e+00 4.36612189e-01 -8.75499070e-01 -4.30311084e-01
2.43251234e-01 -1.02954581e-01 2.70280465e-02 -6.28024518e-01
-1.06165016e+00 8.31972539e-01 1.40424907e-01 5.37234962e-01
1.07435358e+00 4.60627913e-01 7.76819944e-01 3.99361104e-01
2.04438150e-01 -1.02785766e+00 1.04044862e-01 5.08298278e-01
3.15921277e-01 -1.48624575e+00 3.07909325e-02 -7.42502987e-01
-8.38197410e-01 9.64337945e-01 4.75172818e-01 3.03268693e-02
8.89042377e-01 3.76761943e-01 3.56258571e-01 -4.45585221e-01
-1.06698585e+00 -5.02763033e-01 5.52184165e-01 5.01915872e-01
7.74464309e-01 1.89121813e-01 -4.46956277e-01 9.67391789e-01
-2.00638816e-01 -5.22683203e-01 5.11594951e-01 9.65385973e-01
-6.87268853e-01 -1.40903628e+00 -1.41392902e-01 9.06370044e-01
-7.43191779e-01 -6.05074286e-01 -4.12356198e-01 4.27920133e-01
1.12478482e-02 1.12793553e+00 -1.53750509e-01 -1.20251589e-01
3.06213409e-01 -2.10080426e-02 -1.79305717e-01 -1.32149816e+00
-7.21382201e-01 3.23353678e-01 4.05155897e-01 -2.01022744e-01
-6.29893184e-01 -7.59176552e-01 -9.16452825e-01 3.31245571e-01
-7.73768902e-01 4.99936312e-01 7.08949566e-01 1.15416491e+00
1.83418840e-01 2.15894908e-01 8.31736863e-01 -2.69510090e-01
-4.47595268e-01 -1.07414377e+00 -5.25725901e-01 3.22984368e-01
3.38729531e-01 -4.94284064e-01 -6.53561354e-01 1.34725183e-01] | [11.36135196685791, 6.697432994842529] |
1fe2cb3a-b1b7-4a70-9d91-763de8f2add8 | speaker-and-language-change-detection-using | 2302.09381 | null | https://arxiv.org/abs/2302.09381v1 | https://arxiv.org/pdf/2302.09381v1.pdf | Speaker and Language Change Detection using Wav2vec2 and Whisper | We investigate recent transformer networks pre-trained for automatic speech recognition for their ability to detect speaker and language changes in speech. We do this by simply adding speaker (change) or language targets to the labels. For Wav2vec2 pre-trained networks, we also investigate if the representation for the speaker change symbol can be conditioned to capture speaker identity characteristics. Using a number of constructed data sets we show that these capabilities are definitely there, with speaker recognition equal error rates of the order of 10% and language detection error rates of a few percent. We will publish the code for reproducibility. | ['David A. van Leeuwen', 'Nik Vaessen', 'Tijn Berns'] | 2023-02-18 | null | null | null | null | ['change-detection', 'speaker-recognition'] | ['computer-vision', 'speech'] | [ 3.81544620e-01 4.22824085e-01 1.71439111e-01 -8.02530050e-01
-7.55001664e-01 -7.41739094e-01 1.02204382e+00 -1.15635373e-01
-4.40964907e-01 5.15148520e-01 3.83096188e-01 -7.38349378e-01
4.62087184e-01 -2.49420226e-01 -4.82201427e-01 -5.09016037e-01
-3.01010877e-01 5.26175261e-01 3.37413520e-01 -3.28168273e-01
-1.74914271e-01 6.28020823e-01 -1.53608680e+00 2.33203158e-01
1.36450872e-01 8.15273821e-01 -1.89642683e-01 9.67784047e-01
-2.76973337e-01 6.33286834e-01 -7.32129097e-01 -3.13421696e-01
1.12750605e-01 -3.85334820e-01 -8.89225543e-01 1.91836938e-01
4.56986487e-01 -1.50629684e-01 -3.30974132e-01 9.81528699e-01
5.74217975e-01 -1.65434077e-03 5.91722250e-01 -1.09721220e+00
-2.61718363e-01 1.05684149e+00 6.61423653e-02 6.95825458e-01
5.16616046e-01 -5.30010983e-02 9.66780484e-01 -9.25810754e-01
4.29154932e-01 1.39868724e+00 5.30781448e-01 7.16568172e-01
-1.22729027e+00 -7.24403918e-01 1.10039115e-01 2.61079192e-01
-1.32238543e+00 -1.40353167e+00 7.23361313e-01 -4.07438487e-01
1.31153917e+00 5.06078780e-01 1.31252319e-01 1.42160904e+00
-1.76257581e-01 5.21212697e-01 8.99440765e-01 -6.75718665e-01
1.56048452e-02 4.91623491e-01 3.47471565e-01 6.54761255e-01
-2.74955153e-01 2.91755766e-01 -4.60001051e-01 4.46452424e-02
4.84779030e-01 -5.88231504e-01 -3.89210075e-01 -3.02973185e-02
-1.30410111e+00 8.77245605e-01 1.15820020e-01 7.70042121e-01
-7.16185421e-02 1.30882084e-01 6.91579044e-01 7.61703849e-01
3.98396432e-01 3.50596875e-01 -5.54790616e-01 -1.97698042e-01
-8.94370914e-01 -3.25791538e-01 7.61465430e-01 9.59118903e-01
5.19429445e-01 7.25329638e-01 -1.98290408e-01 8.88936341e-01
3.85873355e-02 4.40631688e-01 9.21817243e-01 -6.36952519e-01
1.51089236e-01 4.44917148e-03 1.24244150e-02 -4.61306930e-01
-4.27495897e-01 -4.33701962e-01 -4.69295323e-01 -1.02894129e-02
4.06022429e-01 -1.07978538e-01 -1.09609866e+00 1.90221167e+00
8.88903961e-02 6.23497106e-02 2.23048180e-01 2.64795750e-01
9.34802651e-01 5.99050879e-01 -9.15315077e-02 -3.35001916e-01
1.35017240e+00 -5.31597078e-01 -7.41760492e-01 -1.69638872e-01
6.74327254e-01 -7.60942340e-01 1.09973800e+00 2.44133025e-01
-9.34829354e-01 -5.79948783e-01 -9.17315423e-01 1.67768762e-01
-3.33394706e-01 -6.57703076e-03 2.96540737e-01 1.11170042e+00
-1.39225793e+00 3.26348692e-01 -7.78403223e-01 -6.13941431e-01
-7.62139186e-02 2.30851248e-01 -3.73725563e-01 4.46911007e-01
-1.36022627e+00 1.12232924e+00 7.93609917e-02 -1.13745600e-01
-1.11691880e+00 -3.64429921e-01 -9.70941663e-01 1.69835508e-01
-1.31630823e-01 -5.84369376e-02 1.50668120e+00 -1.01892173e+00
-1.77006567e+00 1.05845571e+00 -5.38504779e-01 -6.46827519e-01
3.90210539e-01 3.80477518e-01 -1.06604898e+00 -9.23046395e-02
-3.05636346e-01 4.32650834e-01 9.82187629e-01 -1.04298472e+00
-5.63195288e-01 -2.07113311e-01 -2.04814240e-01 -1.38237536e-01
-3.26432407e-01 6.24270022e-01 1.05302319e-01 -5.42013288e-01
1.89892858e-01 -7.48863339e-01 2.60674864e-01 -4.57358629e-01
-4.41916138e-01 -3.73671144e-01 8.27171624e-01 -9.75354254e-01
9.18181479e-01 -2.16183591e+00 -2.96150953e-01 2.75994241e-01
-2.35814098e-02 2.23010093e-01 -4.19588655e-01 2.78289676e-01
-3.12551498e-01 1.54311687e-01 -1.20908052e-01 -4.66708928e-01
1.82084098e-01 2.20931694e-01 -6.32717252e-01 6.17459536e-01
2.41225258e-01 6.30078554e-01 -4.45957005e-01 1.67808437e-03
1.11535534e-01 5.51394701e-01 -3.20831329e-01 8.31345096e-02
1.88822284e-01 2.67230362e-01 2.16078445e-01 3.32435220e-01
2.65435815e-01 3.38054597e-01 1.51510552e-01 1.56202512e-02
-7.46544674e-02 1.23943591e+00 -1.18254781e+00 8.63027573e-01
-6.55673683e-01 1.06484056e+00 4.13024127e-01 -8.74006510e-01
1.03744292e+00 7.24347532e-01 3.84500995e-02 -4.88718301e-01
2.57673383e-01 6.36914894e-02 5.18343329e-01 -2.09398612e-01
2.19284758e-01 -5.49226999e-01 3.97771448e-02 3.25967431e-01
3.28913122e-01 -1.53805763e-01 -1.64696544e-01 -1.31071642e-01
1.06701362e+00 -7.06047416e-01 1.51476309e-01 -3.73652011e-01
7.79224515e-01 -4.07538861e-01 4.06813174e-01 7.42444992e-01
-4.46540534e-01 3.44987720e-01 2.71079242e-01 -1.54877409e-01
-8.90681863e-01 -1.11413646e+00 -4.00630981e-01 1.48427641e+00
-4.89294171e-01 -3.25438827e-01 -5.47749579e-01 -4.32929188e-01
-2.79227823e-01 1.16446757e+00 -4.73643452e-01 -1.67763442e-01
-5.82066536e-01 -2.87051558e-01 1.02757215e+00 5.86077690e-01
-1.61700847e-03 -1.13156617e+00 -5.59937432e-02 2.35543236e-01
9.59652439e-02 -1.11166811e+00 -4.85559911e-01 6.69708788e-01
-4.57047522e-01 -6.43979013e-01 -4.45037156e-01 -1.07505357e+00
2.45044842e-01 -2.86290437e-01 8.56637299e-01 -2.57159144e-01
2.22455919e-01 4.76221353e-01 -3.08288962e-01 -2.83360749e-01
-1.12840509e+00 -2.35212576e-02 5.30985773e-01 -6.14313446e-02
3.44958991e-01 -5.66624224e-01 2.16123670e-01 3.45037043e-01
-5.75556517e-01 -4.55202878e-01 3.84206069e-03 8.49689126e-01
3.00854235e-03 4.57551656e-03 5.79846084e-01 -6.67845368e-01
5.94709039e-01 -8.65900666e-02 -5.21193564e-01 1.62327990e-01
-5.11853397e-01 1.74600631e-01 5.14652193e-01 -6.45758390e-01
-8.16486299e-01 4.91857938e-02 -6.97279871e-01 -3.34700316e-01
-4.42743242e-01 2.42058858e-01 -2.86280066e-01 3.54738496e-02
5.83500743e-01 6.03741825e-01 2.89373398e-02 -4.45187956e-01
4.15613562e-01 8.21796060e-01 5.71020901e-01 -1.86117917e-01
7.20827043e-01 7.20027164e-02 -6.66713834e-01 -1.18871248e+00
-1.90366551e-01 -4.02775198e-01 -4.57122594e-01 2.40603741e-02
5.18195093e-01 -8.72811258e-01 -5.02059042e-01 3.87488753e-01
-1.29975665e+00 -3.22410315e-01 -3.65458220e-01 5.65047324e-01
-3.17057550e-01 3.08741271e-01 -7.96507597e-01 -8.74807715e-01
-2.45401293e-01 -1.11457062e+00 7.40505993e-01 -3.31003755e-01
-3.27139407e-01 -1.09396434e+00 9.18102562e-02 2.77067143e-02
7.62307703e-01 -4.29894716e-01 8.69131207e-01 -1.37662566e+00
1.03445485e-01 -1.87622577e-01 2.38729417e-01 5.55102110e-01
4.71089303e-01 -8.21866915e-02 -1.50369573e+00 -4.99118686e-01
1.40503615e-01 5.77675849e-02 8.59393537e-01 2.42366314e-01
8.97358596e-01 -5.70781708e-01 -3.00699115e-01 3.81724745e-01
7.23571420e-01 4.20824379e-01 5.36196530e-01 -8.74815136e-02
5.08062959e-01 6.64817274e-01 -4.02083337e-01 1.79202244e-01
1.46192968e-01 7.86199987e-01 -5.09363003e-02 1.39347181e-01
-3.16899776e-01 -1.56633660e-01 7.35900879e-01 1.19116461e+00
4.03016359e-01 -3.61932814e-01 -1.01702058e+00 7.80982018e-01
-9.45699334e-01 -1.16770399e+00 -2.02728301e-01 2.23468471e+00
9.51380789e-01 4.56616700e-01 2.44273186e-01 4.01149839e-01
8.30781817e-01 2.70552367e-01 -2.24465996e-01 -7.12841630e-01
-1.43439040e-01 3.27801198e-01 5.15264332e-01 1.10220850e+00
-9.94787753e-01 1.08283341e+00 8.07202435e+00 3.60788554e-01
-1.42748129e+00 3.84926677e-01 2.70612746e-01 1.31772652e-01
-5.36462247e-01 -1.97301105e-01 -8.59024167e-01 2.34765515e-01
1.47703350e+00 -1.28939420e-01 4.49224412e-01 5.13711154e-01
1.54013649e-01 5.58176696e-01 -1.31253862e+00 9.95417655e-01
2.73156047e-01 -7.97193766e-01 6.21175319e-02 -7.67242238e-02
3.21993560e-01 3.33747059e-01 -5.89811392e-02 5.95273077e-01
5.52280128e-01 -1.05592811e+00 7.73343742e-01 5.86833917e-02
8.48501563e-01 -4.46414083e-01 4.83933568e-01 1.93325147e-01
-1.04614460e+00 1.28619254e-01 -2.49902353e-01 -9.22405049e-02
1.84851736e-01 4.59493130e-01 -1.58209300e+00 -9.85942110e-02
3.82152021e-01 4.59461153e-01 -7.04584479e-01 6.37836516e-01
-1.86749846e-01 1.26689303e+00 -5.39342761e-01 1.02835400e-02
1.06022485e-01 3.50769907e-01 7.02107489e-01 1.34496939e+00
4.12824273e-01 -3.38200063e-01 -7.21763596e-02 5.60456932e-01
-1.03488490e-01 8.50060210e-02 -6.42427385e-01 -9.15530547e-02
6.45134389e-01 7.42635310e-01 -4.30037141e-01 -4.45868045e-01
-2.52971798e-01 1.01434958e+00 1.62327752e-01 3.18104148e-01
-5.55861294e-01 -5.04630268e-01 1.04179955e+00 -1.04461862e-02
7.26068735e-01 -2.73971915e-01 2.70266403e-02 -1.09370410e+00
-1.35755867e-01 -9.52208221e-01 1.37446582e-01 -5.60971677e-01
-1.12613988e+00 9.52152014e-01 -3.61458331e-01 -7.28085160e-01
-7.81854987e-01 -8.02549362e-01 -8.43728721e-01 9.39342260e-01
-1.19526350e+00 -8.40019107e-01 2.68576294e-01 7.02887237e-01
5.20109355e-01 -4.94263202e-01 1.07741559e+00 1.30371556e-01
-6.10378563e-01 1.03916574e+00 9.06384289e-02 4.57155079e-01
4.86362785e-01 -1.20729804e+00 8.05447698e-01 9.75961745e-01
6.50355458e-01 5.37304342e-01 1.08780801e+00 -2.41366476e-01
-9.24592853e-01 -9.16395307e-01 1.49569750e+00 -6.18668139e-01
8.15021634e-01 -8.04971457e-01 -1.06094623e+00 1.13439739e+00
2.50078619e-01 -1.66194826e-01 6.08909726e-01 5.32733262e-01
-7.78600693e-01 -1.37928262e-01 -9.88524318e-01 3.16283107e-01
1.01975310e+00 -9.84551847e-01 -9.84439850e-01 3.74095529e-01
8.81556869e-01 -1.30624637e-01 -5.07475317e-01 1.22691713e-01
2.37923563e-01 -7.24437296e-01 7.99916267e-01 -7.59007692e-01
-2.85811096e-01 -2.30234206e-01 -4.62182730e-01 -1.51567125e+00
-4.56419170e-01 -3.14312339e-01 2.98376769e-01 1.59440279e+00
6.99604392e-01 -1.02876532e+00 3.70065004e-01 1.94437414e-01
-1.49282590e-01 1.56122684e-01 -1.49071169e+00 -1.06292164e+00
1.82296664e-01 -8.77049267e-01 5.88295519e-01 1.03529739e+00
2.03817979e-01 7.17846811e-01 -2.86596656e-01 2.82279909e-01
1.02471657e-01 -3.82813364e-01 4.63159949e-01 -1.05160511e+00
-3.46800804e-01 -6.65080011e-01 -7.25173950e-01 -8.60320389e-01
5.52538574e-01 -1.09463882e+00 2.82669663e-01 -1.05621052e+00
-3.59756768e-01 -5.61027899e-02 -4.14947361e-01 6.94922566e-01
2.26072788e-01 1.22032233e-01 -9.32969376e-02 -5.98603562e-02
-8.42798278e-02 5.64418793e-01 4.44161475e-01 -4.43049550e-01
-2.39127204e-01 9.13277790e-02 -4.36772048e-01 4.73999143e-01
8.08833301e-01 -4.03965563e-01 -2.97748744e-01 -2.90875018e-01
-4.08934653e-01 -6.74193799e-02 2.89852083e-01 -1.12486911e+00
-4.36819755e-02 2.31554568e-01 1.24306083e-01 -1.95038185e-01
5.30639589e-01 -5.80806613e-01 -1.75061338e-02 5.28352618e-01
-8.06805551e-01 3.29794921e-02 3.29018772e-01 2.43693188e-01
-2.97429711e-01 -2.18156636e-01 7.73868561e-01 1.61391497e-01
-5.83250940e-01 -4.50932831e-02 -1.02927721e+00 -8.96771923e-02
4.87810314e-01 6.77389950e-02 -1.07295237e-01 -8.39939833e-01
-8.69414449e-01 -1.28605455e-01 1.84824720e-01 5.16532719e-01
3.37261677e-01 -1.26437175e+00 -9.02242362e-01 6.31545067e-01
2.40206629e-01 -9.26618218e-01 -3.32429469e-01 7.52935648e-01
1.03461901e-02 5.22364795e-01 7.49852955e-02 -5.63191831e-01
-1.53640640e+00 6.00733161e-01 7.07789600e-01 2.44832709e-01
-3.98039132e-01 1.12730062e+00 2.34431066e-02 -6.98888481e-01
3.89482319e-01 -7.03460395e-01 -1.24722674e-01 4.81546074e-02
6.71341658e-01 5.28508425e-03 3.34060937e-01 -9.93797302e-01
-8.20318103e-01 2.69338526e-02 -1.00343019e-01 -4.48449939e-01
1.14936757e+00 -2.28468865e-01 2.11446151e-01 8.76026988e-01
1.44493890e+00 3.43043625e-01 -6.72250509e-01 -2.98370093e-01
9.77676641e-03 3.50425616e-02 2.41851807e-01 -7.31158078e-01
-8.11830401e-01 1.00221968e+00 8.78784478e-01 4.90088135e-01
6.54574752e-01 3.63569081e-01 4.37747777e-01 4.79085296e-01
2.18851656e-01 -8.12713742e-01 -5.88038683e-01 9.47257698e-01
1.02193153e+00 -1.16614616e+00 -4.98531997e-01 -1.67986110e-01
-3.68024051e-01 9.55404162e-01 1.53712496e-01 5.76693937e-02
8.51865828e-01 3.00161213e-01 4.49971974e-01 -3.87088656e-02
-7.63556421e-01 -4.49845254e-01 2.65948504e-01 7.84688056e-01
6.57749295e-01 3.54625612e-01 5.02768084e-02 4.02532458e-01
-8.01957071e-01 -6.38010681e-01 5.55534124e-01 3.87141287e-01
-6.36105299e-01 -1.06235850e+00 -5.37605822e-01 3.22433561e-01
-2.41571486e-01 -2.95564920e-01 -5.97784698e-01 5.08682311e-01
-3.50944787e-01 1.17547429e+00 3.04227561e-01 -6.21979415e-01
4.07202691e-01 8.04935217e-01 2.94368863e-01 -6.43057704e-01
-4.54368025e-01 -6.12205602e-02 5.38278639e-01 -2.38152966e-01
-2.00006381e-01 -8.36252511e-01 -1.07285738e+00 -2.42598951e-01
-3.16539317e-01 9.97154713e-02 7.04377115e-01 9.89615738e-01
-4.58498150e-02 4.42984194e-01 6.59502447e-01 -5.95300972e-01
-6.50211990e-01 -1.43003643e+00 -5.94855905e-01 3.25237840e-01
6.72177494e-01 -4.99050319e-01 -7.93088853e-01 1.59257174e-01] | [14.381579399108887, 6.394554138183594] |
371a431c-7ee2-4ebc-94a1-ed0e4dae491f | predicting-privacy-preferences-for-smart | 2302.10650 | null | https://arxiv.org/abs/2302.10650v1 | https://arxiv.org/pdf/2302.10650v1.pdf | Predicting Privacy Preferences for Smart Devices as Norms | Smart devices, such as smart speakers, are becoming ubiquitous, and users expect these devices to act in accordance with their preferences. In particular, since these devices gather and manage personal data, users expect them to adhere to their privacy preferences. However, the current approach of gathering these preferences consists in asking the users directly, which usually triggers automatic responses failing to capture their true preferences. In response, in this paper we present a collaborative filtering approach to predict user preferences as norms. These preference predictions can be readily adopted or can serve to assist users in determining their own preferences. Using a dataset of privacy preferences of smart assistant users, we test the accuracy of our predictions. | ['Michael Luck', 'Natalia Criado', 'William Seymour', 'Marc Serramia'] | 2023-02-21 | null | null | null | null | ['collaborative-filtering'] | ['miscellaneous'] | [ 3.20088089e-01 4.69074965e-01 -3.01740915e-01 -1.32009745e+00
-6.02030993e-01 -8.19794357e-01 2.28092417e-01 1.26857176e-01
-3.76687020e-01 6.94515526e-01 8.64942491e-01 -3.45367461e-01
-1.44387797e-01 -6.92725360e-01 7.69927502e-02 -2.36181989e-01
2.72657365e-01 3.54190230e-01 1.07480409e-02 -1.13768548e-01
2.31150016e-01 1.88033804e-01 -1.69064760e+00 8.43796372e-01
1.01014972e+00 1.19298005e+00 -3.22972327e-01 5.10296166e-01
-1.17970079e-01 5.13140082e-01 -1.83913708e-01 -7.54019797e-01
2.16638058e-01 1.17229506e-01 -9.11095858e-01 -2.63926327e-01
3.20292801e-01 -7.28159308e-01 4.25440341e-01 8.52849603e-01
4.42532778e-01 3.34826469e-01 2.76349872e-01 -1.17519689e+00
-4.00072306e-01 6.41172647e-01 3.45283121e-01 -5.02428532e-01
1.39376473e+00 -1.99353009e-01 1.28701150e+00 -3.08450848e-01
4.35594827e-01 9.21667755e-01 6.84835613e-01 9.47298050e-01
-1.39842963e+00 -5.37501991e-01 4.17775214e-01 -5.28657846e-02
-1.18724155e+00 -8.97510946e-01 8.43592763e-01 -2.73910612e-01
5.98821223e-01 1.06256557e+00 4.56336170e-01 1.45519125e+00
-3.01207691e-01 1.02686119e+00 9.25815344e-01 -1.87760428e-01
7.93807209e-01 1.06667340e+00 2.35163033e-01 -2.30302885e-02
5.41885383e-02 -2.51429141e-01 -5.49139440e-01 -9.60858226e-01
5.51198907e-02 3.14479589e-01 -2.15587646e-01 -5.15745580e-01
-6.77115619e-01 5.99167287e-01 -1.61596507e-01 4.55106586e-01
-5.37168562e-01 -8.08997214e-01 1.25658244e-01 4.79808986e-01
4.38952833e-01 3.47305506e-01 -9.91164029e-01 -3.36215794e-01
-4.28166121e-01 1.68389097e-01 1.48780143e+00 1.16286981e+00
6.33002043e-01 -8.67933273e-01 -4.48233634e-02 8.12036872e-01
6.31864071e-01 -1.64227281e-02 7.07312167e-01 -1.06582248e+00
1.83891922e-01 9.20481622e-01 9.66321111e-01 -1.07382107e+00
-3.03279132e-01 3.67187887e-01 -6.61736846e-01 -1.01113142e-02
6.09945118e-01 -5.18331528e-01 -1.14564173e-01 1.62788475e+00
4.57176775e-01 -3.30625743e-01 7.55979940e-02 6.80570424e-01
2.04507321e-01 3.16318214e-01 3.71295601e-01 -5.34120739e-01
1.12344265e+00 3.18096913e-02 -7.65984356e-01 -8.59971568e-02
8.40092778e-01 -5.87977171e-01 1.31147969e+00 7.35308826e-01
-7.81031549e-01 -2.75465459e-01 -2.91111290e-01 1.54794410e-01
-3.42647672e-01 -3.57414126e-01 8.85334671e-01 1.46844578e+00
-1.05346107e+00 7.49532580e-01 -5.49797833e-01 -6.94140494e-01
1.03575498e-01 7.62049317e-01 -1.96478978e-01 2.49237061e-01
-1.21907151e+00 5.94985962e-01 -1.16581127e-01 -2.57357001e-01
2.77892500e-01 -6.57019079e-01 -5.57060003e-01 7.02724159e-02
9.22704488e-02 -4.91282970e-01 1.54761243e+00 -9.95198250e-01
-1.83655930e+00 7.12113261e-01 -3.76292884e-01 -1.19042844e-01
8.02836418e-01 -4.02465880e-01 -7.11614251e-01 -5.26524425e-01
-1.74675673e-01 2.11652473e-01 4.07443315e-01 -1.07600474e+00
-9.29517388e-01 -6.94595098e-01 1.13251597e-01 2.03981087e-01
-8.65496933e-01 1.70758694e-01 1.06019586e-01 1.52592301e-01
1.49076208e-01 -7.80788124e-01 -4.28669006e-01 -1.91041544e-01
-2.72951275e-01 -4.15511519e-01 7.90043831e-01 -4.32550758e-01
1.28026772e+00 -2.17850995e+00 -5.57513416e-01 7.17977822e-01
-8.74142051e-02 -5.99685349e-02 3.50019097e-01 3.14087778e-01
3.36590588e-01 4.76905823e-01 4.70812172e-02 -4.35824692e-01
5.90753138e-01 2.26085484e-01 -2.65536606e-01 9.92361307e-02
-6.39624655e-01 3.90910923e-01 -7.20431507e-01 -4.78680044e-01
3.29781204e-01 4.30277139e-01 -9.46684361e-01 6.19438827e-01
-1.27271086e-01 5.30416548e-01 -9.64466393e-01 4.31485295e-01
5.48105061e-01 1.38012469e-01 8.98614943e-01 4.81773578e-02
-1.70771897e-01 8.85495186e-01 -1.28442848e+00 1.24977219e+00
-6.21562541e-01 -5.63725047e-02 4.95006114e-01 -5.66264689e-01
9.09874439e-01 6.60535812e-01 7.44436681e-01 -4.01575863e-01
2.11136550e-01 7.02192932e-02 -5.09242475e-01 -6.11774147e-01
2.85691798e-01 1.26781434e-01 -1.93382666e-01 6.23367548e-01
-5.97455502e-01 1.70014411e-01 -3.71809870e-01 -9.19445045e-03
9.10210133e-01 -2.33035728e-01 7.26654828e-01 -2.96994954e-01
8.12314749e-01 -5.48149467e-01 5.49417734e-01 8.14009249e-01
-3.77842575e-01 2.17601076e-01 5.38873821e-02 -7.56830871e-01
-3.26723963e-01 -6.64489865e-01 2.84193233e-02 1.56616664e+00
-1.82580695e-01 -4.93007094e-01 -6.10900819e-01 -8.27914536e-01
-5.09024151e-02 1.03133738e+00 -3.86884212e-01 1.71757638e-02
-1.53312251e-01 -3.73446137e-01 2.43725069e-02 2.91011810e-01
3.81421000e-01 -8.94165099e-01 -6.23698056e-01 3.56352627e-01
-4.24138814e-01 -8.07455778e-01 -5.64048886e-01 -2.12733999e-01
-9.66945827e-01 -8.71034563e-01 -1.05231717e-01 -3.01248819e-01
5.84604204e-01 -1.51180224e-02 1.02819169e+00 1.72502548e-01
5.11243999e-01 6.34822726e-01 -4.33617502e-01 -2.58931071e-01
-1.36588842e-01 1.88569605e-01 4.85267311e-01 6.13561034e-01
1.05536795e+00 -1.10549772e+00 -6.83017373e-01 8.37144136e-01
-7.53778160e-01 -4.24577802e-01 7.25162774e-02 8.50556791e-02
6.74217418e-02 -6.77366555e-02 2.56373614e-01 -1.45284712e+00
1.13386726e+00 -3.97216469e-01 -1.97381198e-01 2.28292525e-01
-1.05768716e+00 -1.64349809e-01 8.32403958e-01 -4.15014446e-01
-1.37970567e+00 6.94395840e-01 -4.21913236e-01 5.83125591e-01
-9.63294864e-01 7.41151273e-02 -9.94210780e-01 2.28463244e-02
6.28453434e-01 -1.71830997e-01 -5.51032901e-01 -7.18579888e-01
2.59033114e-01 1.35409999e+00 1.10508099e-01 -5.48067093e-01
3.96466225e-01 2.06011280e-01 -8.41968477e-01 -6.76094711e-01
-6.96843624e-01 -7.07326889e-01 -6.50639057e-01 -1.65180653e-01
5.67011952e-01 -3.86811614e-01 -1.39068627e+00 -8.59164726e-03
-8.28483343e-01 -6.63916618e-02 -2.78401189e-03 3.89746279e-01
-6.09836042e-01 3.60321432e-01 -2.97388077e-01 -1.58651340e+00
-1.69764876e-01 -6.32011652e-01 6.60459161e-01 2.75894761e-01
-1.44538116e+00 -1.15019190e+00 -6.83590323e-02 5.13541162e-01
7.63361514e-01 -1.33390084e-01 5.68586826e-01 -1.19590211e+00
1.02581635e-01 -7.64205635e-01 3.23333681e-01 -8.86500552e-02
7.31550455e-01 -3.25023592e-01 -1.19038105e+00 -1.97028846e-01
3.11926812e-01 -1.47792697e-01 -1.50908083e-01 1.58009544e-01
1.30014288e+00 -1.03604126e+00 -5.32559335e-01 1.28755227e-01
9.69192147e-01 4.46542919e-01 6.90833271e-01 -1.06477626e-01
9.98131186e-02 9.52703297e-01 2.38441691e-01 8.57392550e-01
4.97189194e-01 6.93346500e-01 4.85123917e-02 4.31964248e-01
1.15297294e+00 -5.17471313e-01 3.68593067e-01 2.52248168e-01
-1.76757872e-01 -1.59751862e-01 -4.17041838e-01 6.64161295e-02
-1.94059730e+00 -9.08431768e-01 2.95226630e-02 2.56678391e+00
8.75227988e-01 -5.96402278e-05 4.32711273e-01 3.06978285e-01
5.05399704e-01 -8.07109699e-02 -6.80246949e-01 -5.87656975e-01
4.62756127e-01 -1.29926831e-01 5.01115322e-01 8.00299585e-01
-8.21157634e-01 5.98581016e-01 7.07022905e+00 -2.29351982e-01
-8.80962551e-01 -1.04372852e-01 8.26091349e-01 1.94243286e-02
-7.50188887e-01 -1.28740266e-01 -5.92082202e-01 4.89369452e-01
1.01984918e+00 -1.84335113e-01 6.50822937e-01 1.21596634e+00
4.14064199e-01 8.30238014e-02 -1.47972608e+00 9.21198606e-01
-4.84291524e-01 -6.13573790e-01 -3.27333122e-01 1.62454024e-01
3.64070266e-01 -5.32177508e-01 -6.11714832e-02 9.50999185e-02
9.53184664e-01 -6.68681681e-01 1.67598560e-01 6.33428633e-01
3.32781106e-01 -4.45130855e-01 6.11664414e-01 7.70565689e-01
-7.26234436e-01 -2.76737243e-01 -2.33395860e-01 -6.61199033e-01
2.46495381e-01 6.17299080e-01 -9.65261996e-01 5.28016165e-02
9.03081954e-01 4.24064636e-01 -3.69633920e-02 6.98877692e-01
1.09350949e-03 5.99562347e-01 -6.07045174e-01 -3.96208555e-01
-4.55316812e-01 -3.75016212e-01 -3.43166897e-03 9.31814075e-01
4.79613572e-01 6.74663603e-01 1.92653447e-01 5.49935400e-01
2.06614174e-02 4.88862008e-01 -6.98407292e-01 8.67848024e-02
6.24073267e-01 1.16906524e+00 -8.14948678e-02 -3.23922783e-01
-5.36755145e-01 1.09157109e+00 -7.77366012e-02 2.59605736e-01
-9.33813676e-02 -7.55885988e-02 1.31647217e+00 3.51541013e-01
-3.71127337e-01 1.83684781e-01 -4.49197441e-01 -1.23543465e+00
1.38392463e-01 -8.55253637e-01 6.22679770e-01 -4.17867392e-01
-1.45482242e+00 3.06644827e-01 -4.22303557e-01 -1.14650941e+00
-5.47631502e-01 -4.57768917e-01 -6.83606625e-01 6.36092842e-01
-5.95130801e-01 -4.49676812e-01 -2.46011987e-01 9.44316447e-01
1.40153661e-01 2.69195616e-01 1.41332269e+00 1.61901593e-01
-1.63467955e-02 5.96213400e-01 -5.27535565e-02 -1.35053262e-01
8.70411217e-01 -1.09838045e+00 1.42485544e-01 5.23934513e-02
-1.25313520e-01 1.05253577e+00 1.04067731e+00 -6.07319057e-01
-1.39001596e+00 -5.77599466e-01 1.44264960e+00 -8.67469549e-01
3.39196444e-01 -5.49360931e-01 -8.21565688e-01 8.55001628e-01
-3.28996368e-02 -3.42912138e-01 1.52160347e+00 6.22891963e-01
-2.12944999e-01 -3.15885812e-01 -1.82124603e+00 4.45279986e-01
1.17723536e+00 -5.45751154e-01 -5.56230783e-01 3.24079424e-01
9.41848308e-02 -5.70710329e-03 -9.92352605e-01 7.50071853e-02
1.05434585e+00 -1.28280854e+00 6.87761307e-01 -7.72919238e-01
-3.55021507e-01 -2.19334029e-02 -5.65815806e-01 -1.22883713e+00
-3.26080382e-01 -1.13238180e+00 1.50627196e-01 1.41065097e+00
7.41352916e-01 -9.42341268e-01 1.32306445e+00 2.46828938e+00
5.66991627e-01 2.13971594e-03 -7.05939651e-01 -9.11642686e-02
-4.46063608e-01 -7.66359568e-01 1.17078269e+00 1.21990013e+00
9.77284312e-01 1.70755193e-01 -7.53537536e-01 2.49685436e-01
4.90605831e-01 -1.31874636e-01 6.38803840e-01 -1.75418866e+00
-5.01465909e-02 -1.21741481e-01 3.79263498e-02 -1.13962984e+00
-5.41822501e-02 -3.51527095e-01 -2.69025508e-02 -1.35786867e+00
-3.79657090e-01 -6.51549637e-01 -3.89607459e-01 4.38994795e-01
1.96626097e-01 -2.73490667e-01 -3.28150779e-01 -1.03185512e-01
-8.26990843e-01 5.07859178e-02 3.39563698e-01 7.33047025e-03
-8.04269433e-01 9.51688170e-01 -1.41315830e+00 8.72761607e-01
1.10834801e+00 -8.91188011e-02 -5.30779541e-01 -1.55949490e-02
6.36170506e-01 1.00245133e-01 -3.32315177e-01 -6.39885187e-01
3.68623823e-01 -4.77117211e-01 6.10101521e-01 -9.80749950e-02
1.53396830e-01 -1.52218604e+00 2.84229338e-01 1.72542036e-02
-8.10486436e-01 -2.62207031e-01 -3.45668912e-01 3.41886193e-01
1.42499106e-02 2.76528485e-02 1.84959143e-01 -3.81291658e-01
-3.25393170e-01 2.02774331e-01 -8.35680962e-01 -4.91265535e-01
6.42803133e-01 -4.07848805e-01 3.52435559e-01 -1.00151932e+00
-1.13707495e+00 3.31016093e-01 6.13354027e-01 6.87518179e-01
4.76681381e-01 -1.24576819e+00 -5.42988293e-02 3.92914891e-01
1.60365954e-01 -3.92975658e-01 6.12424724e-02 4.46384877e-01
4.87465590e-01 3.44814003e-01 6.96586492e-03 -1.54986084e-01
-1.44982660e+00 5.18539131e-01 2.12403178e-01 2.28383511e-01
-1.83815062e-01 6.12755597e-01 -1.38059705e-01 -7.82469332e-01
5.63497126e-01 -3.23628843e-01 -5.26504934e-01 2.27813758e-02
1.05141377e+00 2.23764107e-01 6.21522069e-02 -2.55618989e-01
-5.45668602e-01 3.30223627e-02 -2.56435052e-02 -6.73788562e-02
1.30006766e+00 -5.11089742e-01 -1.39968535e-02 2.53991306e-01
1.04276097e+00 5.47720730e-01 -9.90688264e-01 -1.11832052e-01
5.66225410e-01 -8.76176238e-01 -4.50725943e-01 -8.89180422e-01
-6.95987701e-01 1.78766161e-01 6.41597331e-01 9.29996371e-01
1.14164627e+00 -2.35587105e-01 7.56775260e-01 5.61870039e-01
7.34506488e-01 -1.35892248e+00 -7.97382891e-01 9.54024345e-02
2.94745147e-01 -1.47698927e+00 -4.03245986e-01 -6.12353444e-01
-7.24756777e-01 8.51035535e-01 4.04990405e-01 4.04269427e-01
8.97862554e-01 -1.63045283e-02 3.81920189e-01 4.19535041e-01
-8.96404326e-01 3.33440423e-01 1.59641489e-01 8.03874254e-01
9.37980831e-01 5.10247886e-01 -3.78120482e-01 1.06930947e+00
-5.20315945e-01 2.87021399e-01 3.29931349e-01 8.62924814e-01
-5.31337500e-01 -1.61177969e+00 -2.88618267e-01 7.07325637e-01
-3.31350863e-01 1.56098336e-01 -1.11706400e+00 -1.23321503e-01
-1.49396867e-01 1.61770427e+00 -3.01368773e-01 -8.95002127e-01
7.96537995e-01 5.11520863e-01 -2.97587842e-01 -2.65760481e-01
-5.85320175e-01 -4.81256396e-01 4.63223934e-01 -9.51294780e-01
-4.21166033e-01 -9.38655317e-01 -7.10542262e-01 -3.45837027e-01
-8.05293676e-03 6.12833977e-01 3.64613593e-01 8.73969913e-01
4.78968382e-01 -5.94829977e-01 9.41971838e-01 -5.52226961e-01
-6.77170277e-01 -7.71004736e-01 -4.96169120e-01 9.35009360e-01
1.55125991e-01 1.62662414e-03 -3.92357409e-01 -5.41613773e-02] | [12.304174423217773, 7.625743865966797] |
0de2815f-eab8-43f2-a98e-e9df489febb4 | legal-and-political-stance-detection-of | 2211.11724 | null | https://arxiv.org/abs/2211.11724v1 | https://arxiv.org/pdf/2211.11724v1.pdf | Legal and Political Stance Detection of SCOTUS Language | We analyze publicly available US Supreme Court documents using automated stance detection. In the first phase of our work, we investigate the extent to which the Court's public-facing language is political. We propose and calculate two distinct ideology metrics of SCOTUS justices using oral argument transcripts. We then compare these language-based metrics to existing social scientific measures of the ideology of the Supreme Court and the public. Through this cross-disciplinary analysis, we find that justices who are more responsive to public opinion tend to express their ideology during oral arguments. This observation provides a new kind of evidence in favor of the attitudinal change hypothesis of Supreme Court justice behavior. As a natural extension of this political stance detection, we propose the more specialized task of legal stance detection with our new dataset SC-stance, which matches written opinions to legal questions. We find competitive performance on this dataset using language adapters trained on legal documents. | ['Kathleen McKeown', 'Emily Allaway', 'Noah Bergam'] | 2022-11-21 | null | null | null | null | ['stance-detection'] | ['natural-language-processing'] | [ 9.67664346e-02 4.40460503e-01 -6.07575297e-01 -6.54975712e-01
-1.09501791e+00 -1.28603220e+00 1.27727878e+00 6.36644959e-01
-7.35060096e-01 8.93989682e-01 1.32613671e+00 -1.12510133e+00
1.61960442e-02 -8.17209780e-01 -4.29816157e-01 -2.68754840e-01
8.87923777e-01 7.07245767e-01 1.25672981e-01 -6.84095979e-01
9.69618201e-01 2.92402685e-01 -1.00491858e+00 8.03161800e-01
8.87225509e-01 3.16240937e-01 -9.03826594e-01 3.58419806e-01
-4.73209202e-01 1.39394999e+00 -5.23825347e-01 -9.75041091e-01
3.04496408e-01 -3.10100704e-01 -1.13954175e+00 -3.86078060e-01
1.17374134e+00 9.94172618e-02 1.38403606e-02 1.07509375e+00
5.61010897e-01 -4.02004927e-01 7.39005268e-01 -4.54954028e-01
-8.50960791e-01 1.18428433e+00 -5.42020142e-01 8.67556691e-01
6.43514276e-01 -7.30157867e-02 1.52422404e+00 -6.05900347e-01
1.54936028e+00 1.70371258e+00 6.01043463e-01 1.78421974e-01
-1.39225578e+00 -4.50875074e-01 3.97547871e-01 -6.39136955e-02
-3.27488124e-01 -7.82649040e-01 9.07974184e-01 -1.14425242e+00
4.35742646e-01 1.35646075e-01 4.86090988e-01 1.37390018e+00
3.07217866e-01 2.98789620e-01 1.87755752e+00 -5.37760377e-01
5.42753227e-02 1.88283205e-01 9.30110633e-01 4.97161686e-01
5.50458431e-01 -3.23447019e-01 -2.89640933e-01 -9.34746385e-01
-2.64551267e-02 -5.92413962e-01 -1.11857438e-02 2.98701972e-01
-1.12831414e+00 1.15784585e+00 -3.88440443e-03 7.86207139e-01
-2.32634485e-01 -1.40157804e-01 6.44696891e-01 7.26466179e-01
7.86805511e-01 5.32349765e-01 -3.61858755e-01 -4.45474148e-01
-7.54716873e-01 5.60385346e-01 9.56766307e-01 1.67869896e-01
1.62693635e-01 -6.37150049e-01 -3.70135903e-01 8.35932910e-01
3.52006316e-01 5.67956388e-01 2.59978265e-01 -1.28912580e+00
8.89866650e-01 7.18670964e-01 5.60580194e-02 -1.07931709e+00
-2.98364103e-01 -4.44123715e-01 -6.09343755e-04 2.14312539e-01
6.37940288e-01 -9.09755006e-02 -4.40322846e-01 1.58492398e+00
3.75698715e-01 -8.56831968e-01 1.70334980e-01 4.36118722e-01
8.04506183e-01 2.39407644e-01 1.81762278e-01 -5.30241132e-01
1.56844842e+00 -1.44746900e-01 -5.84325731e-01 7.52084106e-02
4.64032650e-01 -1.25956547e+00 9.87967670e-01 -1.70937702e-01
-1.19018352e+00 -1.11745596e-01 -5.34901977e-01 -4.15231317e-01
-1.83167025e-01 -2.70721316e-01 5.73011577e-01 4.55084652e-01
-8.69612813e-01 4.39570457e-01 -2.01570123e-01 -3.44904155e-01
4.87664640e-01 -4.31677163e-01 1.45296991e-01 4.46386993e-01
-1.00531447e+00 1.00179029e+00 -3.28178614e-01 -5.01079679e-01
1.79872826e-01 -5.90700984e-01 -5.00209987e-01 -3.91027987e-01
3.22686762e-01 -5.02599478e-01 1.33814025e+00 -7.90568769e-01
-1.24671745e+00 1.76844299e+00 -2.34055921e-01 -1.56922475e-01
9.59744990e-01 1.29994541e-01 -6.56431854e-01 8.12229961e-02
9.04888272e-01 -1.28258303e-01 3.73071194e-01 -9.82026935e-01
-5.96856356e-01 -5.09608924e-01 4.39022332e-01 -4.10157084e-01
-1.16641164e-01 5.54249465e-01 5.21194935e-01 -5.77569723e-01
3.01186591e-01 -1.03178740e+00 1.62143424e-01 -5.61014116e-01
-4.83280331e-01 -7.10100055e-01 3.92729729e-01 -5.99046826e-01
1.29429519e+00 -1.80859172e+00 -2.90989101e-01 4.76480514e-01
5.95127761e-01 -2.24852175e-01 2.44470775e-01 7.17200100e-01
-1.97799295e-01 4.15903032e-01 4.62122355e-03 1.78660259e-01
3.61588269e-01 -2.19727550e-02 -9.34862971e-01 5.86049855e-01
-4.03052330e-01 1.01826704e+00 -7.59752750e-01 -7.54598200e-01
-5.23715734e-01 -2.07825154e-01 -8.59959543e-01 -8.40134144e-01
-4.45060283e-02 4.63555664e-01 -6.40871465e-01 7.42870271e-01
3.96799892e-01 -2.85243988e-01 6.66551888e-01 -2.72481889e-01
-8.42551708e-01 1.08032429e+00 -2.41648138e-01 1.05655456e+00
-3.38742658e-02 1.16054952e+00 1.04645044e-01 -8.52079988e-01
7.05145538e-01 1.89085841e-01 3.10139954e-01 -9.44871068e-01
5.79669178e-01 6.26455009e-01 6.62478983e-01 -6.11217737e-01
4.97825384e-01 -4.68108356e-01 -4.12585944e-01 8.08389187e-01
-6.79797351e-01 1.40627280e-01 5.50333083e-01 3.92558604e-01
1.16544259e+00 -2.16002092e-01 4.62538779e-01 -9.74940777e-01
6.02736592e-01 2.54927635e-01 6.26961768e-01 7.89298475e-01
-2.93236971e-01 -1.98429488e-02 9.06917930e-01 -7.19075859e-01
-9.85380113e-01 -7.47855663e-01 -5.78335643e-01 9.88577366e-01
-3.65374476e-01 -1.10085547e-01 -5.29520452e-01 -8.01697135e-01
2.97361076e-01 7.77945876e-01 -7.96348870e-01 5.44661403e-01
-9.48591650e-01 -5.45123398e-01 4.42708969e-01 -1.53289307e-02
4.29547668e-01 -8.14889967e-01 -5.87903321e-01 2.51138568e-01
-5.23161113e-01 -1.17659068e+00 -1.77990705e-01 -4.33351666e-01
-7.36906171e-01 -1.41038716e+00 -3.14819515e-01 -4.56197739e-01
4.07032996e-01 -3.04081529e-01 1.08176315e+00 1.04422495e-01
1.21247038e-01 2.81985402e-01 -1.23063415e-01 -4.30985153e-01
-1.25579739e+00 1.80686131e-01 1.82081237e-01 -6.61508366e-02
5.55218399e-01 -7.00968504e-01 -3.01793218e-01 -2.81585574e-01
-5.62625349e-01 -2.16965243e-01 8.47783759e-02 2.25188792e-01
-6.94032460e-02 -8.26286674e-01 4.73299801e-01 -1.41057193e+00
1.26818502e+00 -4.67288196e-01 -5.93001723e-01 2.49938592e-01
-6.93023324e-01 1.43829241e-01 3.73955131e-01 -8.17169473e-02
-1.01596916e+00 -1.02755702e+00 -3.12856644e-01 7.20341027e-01
1.97663143e-01 5.91819108e-01 6.05852842e-01 3.08292150e-01
9.68252122e-01 -4.88695145e-01 -2.16044694e-01 -3.97866070e-01
2.46478274e-01 8.44625711e-01 5.04886389e-01 -9.73571837e-01
9.38827455e-01 8.63275826e-01 -2.74568945e-01 -6.20314717e-01
-1.57887042e+00 -2.09548786e-01 -3.45657259e-01 -1.77924529e-01
9.25222337e-01 -6.46214783e-01 -1.12500548e+00 -1.73772573e-01
-1.57232797e+00 4.88958322e-02 -1.81335285e-01 4.43183362e-01
-2.71516860e-01 7.04598188e-01 -8.71840537e-01 -7.11605728e-01
-3.38921309e-01 -8.10815096e-01 8.28383088e-01 -2.77146906e-01
-8.74770880e-01 -9.89746511e-01 8.86977434e-01 1.01498532e+00
1.50016665e-01 6.44223332e-01 1.28739905e+00 -5.53550780e-01
1.45875111e-01 3.53926383e-02 -1.90399989e-01 1.04995027e-01
1.01142935e-01 3.42825681e-01 -6.35677338e-01 -5.58802150e-02
1.42235070e-01 -2.20470577e-01 9.16921556e-01 4.13821250e-01
1.76286086e-01 -6.13658607e-01 -1.88325703e-01 -1.12222187e-01
1.16313624e+00 1.23544477e-01 3.47057939e-01 8.01374435e-01
6.95315674e-02 6.33104384e-01 2.53512979e-01 2.96971142e-01
5.38253605e-01 5.25450230e-01 -5.39729953e-01 6.74283952e-02
-1.18484743e-01 -1.36433780e-01 4.65444028e-01 1.10145366e+00
-6.10713542e-01 -3.12538110e-02 -1.26363516e+00 4.75754768e-01
-1.81186938e+00 -1.60100889e+00 -4.94382292e-01 1.57151401e+00
9.57167149e-01 5.28963506e-01 2.95742005e-01 5.24246022e-02
5.47055483e-01 2.59776682e-01 -7.30331242e-02 -7.22850084e-01
-3.88560772e-01 2.51750082e-01 3.25906754e-01 1.07587087e+00
-8.02805126e-01 9.11496043e-01 6.86762285e+00 2.30219826e-01
-9.06247020e-01 2.59153426e-01 6.02735341e-01 -2.43004397e-01
-7.96880126e-01 2.68652231e-01 -7.57158279e-01 5.98864973e-01
7.37188995e-01 -4.48708683e-01 -2.40439828e-02 5.18933952e-01
3.88093442e-01 -7.19160065e-02 -1.11844420e+00 5.08174837e-01
1.12748653e-01 -1.83495867e+00 -2.12062359e-01 5.90281487e-01
9.38478768e-01 2.55304217e-01 6.33347780e-02 1.44380584e-01
8.29594731e-01 -3.36761832e-01 1.23760653e+00 3.88935000e-01
4.67080265e-01 -1.13231279e-01 5.16869605e-01 2.04239339e-01
-2.84791321e-01 -3.88940603e-01 -1.55212894e-01 -4.18133229e-01
4.86717224e-01 8.50165427e-01 -5.77808261e-01 2.65736245e-02
4.33402807e-01 5.16086400e-01 -4.66541916e-01 3.80990244e-02
-3.37281585e-01 9.28950965e-01 -6.06073216e-02 4.22965223e-03
4.81866837e-01 -2.73701221e-01 9.29886281e-01 1.02853119e+00
-3.81896757e-02 4.61884737e-01 1.17463455e-01 7.45467424e-01
-3.71669531e-01 4.52450573e-01 -7.51483977e-01 -1.57805443e-01
3.28663319e-01 1.00911391e+00 -7.72770584e-01 -7.76066959e-01
-2.81372964e-01 2.05148548e-01 5.11635542e-01 1.52647197e-01
-6.51176393e-01 3.68957430e-01 5.98292232e-01 4.25293207e-01
-2.74264038e-01 -2.06932463e-02 -3.68141055e-01 -1.50967360e+00
1.91944972e-01 -1.19424331e+00 4.68388289e-01 -2.50420123e-01
-1.41460097e+00 4.02785897e-01 -1.19758390e-01 -7.80919075e-01
-9.98956431e-03 -5.33007383e-01 -5.34943819e-01 4.87503767e-01
-1.41502154e+00 -8.05988550e-01 2.96686620e-01 1.04403436e-01
2.95103550e-01 -3.45090181e-01 4.01707292e-01 2.54377246e-01
-1.74763799e-01 1.06597155e-01 -6.34658113e-02 2.80793697e-01
8.56586874e-01 -9.22043860e-01 3.98971558e-01 6.13358974e-01
9.30754915e-02 1.13593423e+00 1.02155757e+00 -1.00032616e+00
-7.16657162e-01 -3.05948734e-01 1.51466000e+00 -1.05897725e+00
1.28613687e+00 -1.13818638e-01 -4.97608244e-01 7.20319569e-01
5.77186704e-01 -8.75321507e-01 9.97657478e-01 6.95244610e-01
-7.92338252e-01 4.10910040e-01 -9.80384767e-01 7.51866698e-01
1.39399648e+00 -8.11298609e-01 -1.43575370e+00 8.21928799e-01
4.44932282e-01 6.43358305e-02 -8.02997649e-01 1.58346161e-01
7.28037059e-01 -9.18294966e-01 7.04120398e-01 -9.76961553e-01
1.11510396e+00 -8.25226679e-02 -3.00715506e-01 -8.81336987e-01
-5.22385716e-01 -4.23714757e-01 7.43791103e-01 1.16226673e+00
7.24816978e-01 -1.18343759e+00 3.45872343e-01 5.24778068e-01
-4.64887992e-02 -4.06224936e-01 -1.20990837e+00 -2.93626368e-01
7.21446753e-01 -1.76883548e-01 2.01567620e-01 1.64719999e+00
3.55856806e-01 8.04793477e-01 3.42280149e-01 -2.30010897e-01
4.81435746e-01 9.78056073e-01 8.02656353e-01 -1.77215350e+00
-2.05457866e-01 -8.04798782e-01 -1.59749746e-01 -4.83778000e-01
4.77598667e-01 -1.27996349e+00 -6.82913184e-01 -1.65269661e+00
4.77351308e-01 -5.33686355e-02 9.30638313e-02 2.92412102e-01
1.97314247e-01 5.33725433e-02 1.24259204e-01 5.31827748e-01
-3.90918702e-01 -5.88375106e-02 1.12776208e+00 -3.11470896e-01
9.02411761e-04 -3.92169356e-01 -1.33665860e+00 1.22517908e+00
6.02045298e-01 -4.96477515e-01 2.38957167e-01 -4.14666325e-01
9.74745810e-01 -2.43618116e-01 2.46879205e-01 -3.97866368e-01
6.24809042e-02 -3.52790862e-01 -1.16785660e-01 -5.40395856e-01
-1.75066128e-01 -2.97691375e-01 -2.80815572e-01 5.81130803e-01
-8.14111471e-01 4.17232215e-01 -1.35374740e-01 2.29047701e-01
-2.20681563e-01 7.70539837e-03 8.07562768e-01 -1.44033119e-01
1.60435401e-02 -3.34390879e-01 -6.53950393e-01 5.68186224e-01
5.11768818e-01 1.51412329e-02 -1.15225446e+00 -3.36077541e-01
-6.12134695e-01 -1.39653355e-01 6.32303715e-01 2.86009967e-01
-1.93515018e-01 -9.82413650e-01 -1.43976891e+00 -6.87638402e-01
4.53350618e-02 -1.26408184e+00 -2.15108782e-01 1.28299761e+00
-5.00720739e-01 4.52863753e-01 2.98560202e-01 -1.51795015e-01
-1.55985010e+00 2.18138620e-01 1.17139537e-02 -3.34003359e-01
-6.00594997e-01 4.19158161e-01 -1.25852436e-01 -5.52406371e-01
-6.54218793e-01 -3.70884836e-01 -2.94580162e-01 6.56307459e-01
3.83498520e-01 3.11379522e-01 -2.02749133e-01 -8.60546350e-01
-4.49772060e-01 6.55360579e-01 -2.46189341e-01 -6.08561933e-01
1.41521025e+00 1.62722229e-03 -7.31847346e-01 7.10735559e-01
1.06798160e+00 1.35071635e+00 -1.29263192e-01 -2.84634739e-01
3.65528613e-01 -5.64624369e-01 -2.18890250e-01 -8.90839756e-01
-4.42387253e-01 2.70090789e-01 2.29226109e-02 7.13755667e-01
1.74682617e-01 5.50958872e-01 3.98379266e-01 5.54551601e-01
2.12921396e-01 -1.63801396e+00 -3.00976098e-01 6.91396832e-01
1.13460004e+00 -1.16605484e+00 3.57004702e-01 -3.92183691e-01
-3.90027553e-01 1.00438285e+00 -1.23463027e-01 -5.41935675e-02
3.94854337e-01 2.78887898e-02 7.66034782e-01 -9.57161963e-01
-7.30969906e-01 2.96677407e-02 1.59412444e-01 -3.60999197e-01
1.00292051e+00 6.81430027e-02 -1.53732944e+00 2.18037758e-02
-1.03473771e+00 -3.29675734e-01 6.67719007e-01 7.96895683e-01
-5.65935612e-01 -1.35274529e+00 -5.13708472e-01 4.21893060e-01
-9.17560339e-01 -2.88819999e-01 -1.21945620e+00 6.90185189e-01
1.37190580e-01 1.06985676e+00 -1.21033065e-01 1.56407699e-01
2.75277108e-01 2.29523063e-01 3.78344327e-01 -6.62204325e-01
-8.23563516e-01 -5.59972972e-02 8.87938440e-01 -3.16397995e-01
-9.63069677e-01 -1.17529762e+00 -9.38877583e-01 -7.19473422e-01
2.09665477e-01 3.38574767e-01 5.41935384e-01 1.08386767e+00
2.48793334e-01 3.81368279e-01 3.92140746e-01 1.19089797e-01
-5.85436761e-01 -9.37757492e-01 -9.43847224e-02 6.15990222e-01
3.12224507e-01 -4.22845632e-01 -4.06248569e-01 -1.37229741e-01] | [9.032991409301758, 9.93613052368164] |
53ede84e-9a9d-4110-86bf-13bfd4c4d71e | realgait-gait-recognition-for-person-re | 2201.04806 | null | https://arxiv.org/abs/2201.04806v2 | https://arxiv.org/pdf/2201.04806v2.pdf | RealGait: Gait Recognition for Person Re-Identification | Human gait is considered a unique biometric identifier which can be acquired in a covert manner at a distance. However, models trained on existing public domain gait datasets which are captured in controlled scenarios lead to drastic performance decline when applied to real-world unconstrained gait data. On the other hand, video person re-identification techniques have achieved promising performance on large-scale publicly available datasets. Given the diversity of clothing characteristics, clothing cue is not reliable for person recognition in general. So, it is actually not clear why the state-of-the-art person re-identification methods work as well as they do. In this paper, we construct a new gait dataset by extracting silhouettes from an existing video person re-identification challenge which consists of 1,404 persons walking in an unconstrained manner. Based on this dataset, a consistent and comparative study between gait recognition and person re-identification can be carried out. Given that our experimental results show that current gait recognition approaches designed under data collected in controlled scenarios are inappropriate for real surveillance scenarios, we propose a novel gait recognition method, called RealGait. Our results suggest that recognizing people by their gait in real surveillance scenarios is feasible and the underlying gait pattern is probably the true reason why video person re-idenfification works in practice. | ['Anil K. Jain', 'Annan Li', 'Tianrui Chai', 'Yunhong Wang', 'Shaoxiong Zhang'] | 2022-01-13 | null | null | null | null | ['person-recognition'] | ['computer-vision'] | [ 2.37717763e-01 -7.17873573e-01 -1.61229432e-01 -2.71378756e-01
-2.33667940e-01 -4.59365308e-01 4.70432699e-01 -2.59362042e-01
-5.96781671e-01 7.91716456e-01 2.09574044e-01 2.24291176e-01
1.60146996e-01 -7.13887811e-01 -4.06514287e-01 -7.14238524e-01
1.05608232e-01 5.43734908e-01 8.18980560e-02 -1.21621132e-01
-1.26701850e-03 2.34097496e-01 -1.96948969e+00 -2.63747908e-02
5.55540442e-01 5.53667724e-01 -2.12622136e-01 7.38492429e-01
3.31023127e-01 1.80601463e-01 -7.26980746e-01 -8.85897517e-01
4.33978617e-01 -2.99405307e-01 -5.62739253e-01 2.48383448e-01
8.17257106e-01 -6.86905742e-01 -6.02800548e-01 9.18816805e-01
7.30197847e-01 -5.04516251e-02 5.15585661e-01 -1.24610162e+00
-8.62706304e-01 -8.13024640e-02 -6.11869216e-01 3.29255402e-01
9.70939755e-01 2.82417983e-01 4.98719841e-01 -5.47969937e-01
5.19375980e-01 1.02476311e+00 1.29867804e+00 8.56865764e-01
-1.21423209e+00 -3.67314607e-01 -1.14570230e-01 3.54900509e-01
-1.55853760e+00 -5.69413245e-01 6.06174052e-01 -3.66808802e-01
8.49619687e-01 4.97948259e-01 1.00659597e+00 1.65949345e+00
-3.03905252e-02 6.02385700e-01 9.35782671e-01 -3.56662542e-01
-1.51118681e-01 -9.30496119e-03 2.42429584e-01 5.42234540e-01
1.05187047e+00 1.53457552e-01 -5.05773127e-01 -3.36407840e-01
6.55151010e-01 7.13810176e-02 -3.49357456e-01 -2.30624989e-01
-1.29741347e+00 2.88183689e-01 -2.01575786e-01 4.15580004e-01
-1.75629750e-01 -9.08575486e-04 6.12025380e-01 5.39795399e-01
2.30370626e-01 -1.55094981e-01 5.78385293e-02 -4.99014378e-01
-1.12446594e+00 5.40091515e-01 7.59262264e-01 7.17418730e-01
3.42319369e-01 1.15863696e-01 -9.05865282e-02 7.81118512e-01
1.12340607e-01 1.00230956e+00 7.72819281e-01 -4.20949996e-01
4.27599639e-01 4.81443077e-01 4.20683801e-01 -1.51280463e+00
-2.11476609e-01 -1.42448237e-02 -7.86212206e-01 -3.68123770e-01
1.02768838e+00 9.52071920e-02 -7.98863947e-01 1.53872621e+00
4.55480292e-02 4.27985221e-01 -2.25772485e-02 1.02857029e+00
5.86606801e-01 4.96321768e-02 1.63363799e-01 -8.10520723e-02
1.56943560e+00 -4.98663872e-01 -5.22545874e-01 -4.06310022e-01
5.18417895e-01 -6.23279631e-01 9.24050093e-01 1.12221085e-01
-4.26326752e-01 -6.89053714e-01 -9.45627093e-01 5.41958690e-01
-5.76841414e-01 2.35601142e-01 4.70024586e-01 1.53242409e+00
-1.00944769e+00 5.43487251e-01 -7.19396949e-01 -1.38830554e+00
1.31310999e-01 5.14746487e-01 -6.92667723e-01 -2.87673533e-01
-1.09188783e+00 8.10115516e-01 1.56723022e-01 2.19217375e-01
-3.53757828e-01 -1.49562150e-01 -8.15406322e-01 -5.32671332e-01
-1.41417440e-02 -8.94372642e-01 7.49823034e-01 -7.19907641e-01
-1.01535034e+00 1.38250709e+00 -3.26080203e-01 -5.38267791e-01
8.18880081e-01 -3.68361250e-02 -9.96432364e-01 8.35334733e-02
2.60420769e-01 1.01455696e-01 1.03067124e+00 -1.05235410e+00
-4.97145832e-01 -6.22832358e-01 -3.69621664e-01 7.67563656e-02
-5.10941684e-01 1.31365642e-01 -4.43631500e-01 -8.33463192e-01
-2.31450781e-01 -1.21349716e+00 2.45666668e-01 -4.02422994e-01
-7.75519684e-02 9.96680856e-02 9.69157100e-01 -1.07858288e+00
1.38761818e+00 -1.80489159e+00 -9.72352028e-02 1.63940921e-01
3.55382822e-03 4.29897934e-01 1.18438452e-01 3.34380180e-01
5.83071038e-02 9.96391326e-02 -1.82236537e-01 -4.48926300e-01
8.08329955e-02 1.45584777e-01 3.51168960e-02 7.77962029e-01
-2.77521998e-01 9.02266324e-01 -8.35575163e-01 -4.36879843e-01
4.90835696e-01 3.17051739e-01 -2.15857044e-01 3.43873091e-02
6.92877471e-01 4.52108264e-01 -2.51448303e-01 1.19365585e+00
8.23341072e-01 9.06791091e-02 2.51839370e-01 -3.62617821e-01
1.18109241e-01 -5.28131902e-01 -1.04994869e+00 1.32957530e+00
3.49771827e-01 7.03742683e-01 -2.60534137e-01 -1.11365187e+00
8.88687193e-01 3.48009735e-01 8.13813925e-01 -4.67833579e-01
8.19491819e-02 1.87608644e-01 -1.89623326e-01 -7.07421482e-01
6.81945503e-01 4.52810712e-02 -3.09562862e-01 2.77390569e-01
-1.50838435e-01 6.08777583e-01 4.50994849e-01 -3.69830608e-01
1.03335989e+00 1.73951328e-01 2.85832554e-01 -1.31349087e-01
6.95592165e-01 1.63901374e-01 4.73746508e-01 9.01717663e-01
-1.07947338e+00 7.96462834e-01 -3.90892774e-01 -7.90182590e-01
-1.26095653e+00 -1.08280849e+00 6.11822233e-02 7.87339151e-01
1.95559233e-01 -3.98956686e-01 -9.97558117e-01 -5.58961093e-01
2.26419643e-01 -1.13582872e-01 -6.36292100e-01 -1.31685674e-01
-7.40317345e-01 -1.33187699e+00 1.17592192e+00 5.56399405e-01
1.14501107e+00 -5.85774124e-01 -6.24427319e-01 3.69048744e-01
-5.74398398e-01 -1.49578559e+00 -7.52156913e-01 -7.00028300e-01
-4.55712765e-01 -1.35140073e+00 -1.40551758e+00 -7.79459774e-01
6.32623732e-01 7.69705713e-01 9.60664272e-01 3.24567825e-01
-2.62714297e-01 9.85300303e-01 -4.22125012e-01 -9.01260972e-02
-1.54767841e-01 -5.62121458e-02 8.90989900e-01 4.21131194e-01
1.18865216e+00 -2.52350628e-01 -6.25007629e-01 7.60512769e-01
-3.08615834e-01 -3.19550604e-01 -2.24732682e-02 8.90109658e-01
-2.41905749e-02 2.07827449e-01 5.19007146e-01 -2.83864379e-01
6.34247959e-01 -1.48065686e-01 -6.18191436e-02 6.51015460e-01
-4.12651926e-01 -4.20429707e-01 3.61546665e-01 -6.03993893e-01
-8.78443718e-01 -5.24366321e-03 5.07264882e-02 -2.53995925e-01
-5.44805110e-01 2.92187959e-01 -2.38225069e-02 -4.91451949e-01
4.39317226e-01 6.47500634e-01 8.16165209e-02 -4.45096225e-01
-2.94620037e-01 8.75622869e-01 7.86988378e-01 -6.47651136e-01
1.12364888e+00 6.72755778e-01 -3.46319020e-01 -1.15838099e+00
-2.17238460e-02 -7.31769025e-01 -8.59782338e-01 -6.49010837e-01
9.30651546e-01 -9.85723555e-01 -8.33066225e-01 1.23394251e+00
-6.91743135e-01 1.13927729e-01 -4.59973291e-02 4.54709738e-01
-4.90558207e-01 1.02399695e+00 -4.17186022e-01 -1.02503324e+00
-3.24868530e-01 -8.00363183e-01 1.09411156e+00 2.95592129e-01
-5.59357524e-01 -9.47682261e-01 5.46164624e-02 7.74240434e-01
4.12606150e-01 4.47950214e-01 1.06524430e-01 -3.68811399e-01
-6.58581555e-02 -6.76060259e-01 -1.33996591e-01 -8.27157721e-02
6.39760852e-01 -2.92114522e-02 -8.85590672e-01 -7.38975406e-01
-2.82879442e-01 9.32988003e-02 6.29852116e-01 2.56755769e-01
4.62435097e-01 -1.37789562e-01 -4.61195678e-01 3.71871591e-01
1.24527669e+00 1.38244823e-01 8.13026786e-01 6.55748188e-01
8.25467408e-01 6.45611167e-01 4.20852661e-01 4.12931353e-01
7.33084202e-01 9.91652668e-01 -2.08577171e-01 2.58329272e-01
-1.55245781e-01 -3.71669352e-01 5.73508501e-01 6.12765789e-01
-8.98758709e-01 -1.95181072e-01 -1.12471163e+00 5.49001932e-01
-1.99536467e+00 -1.61121404e+00 3.32625918e-02 2.61213493e+00
1.46075264e-01 -6.36504740e-02 9.16308165e-01 5.13507128e-01
1.16533792e+00 -5.95623180e-02 -1.71992615e-01 7.22043812e-02
-4.90254968e-01 -3.69747907e-01 1.03231525e+00 -9.86863300e-02
-1.37542605e+00 5.67249477e-01 6.64882326e+00 4.34795946e-01
-9.93570805e-01 -1.18493661e-01 2.84595460e-01 2.37542883e-01
4.92403716e-01 -4.98105675e-01 -7.63650179e-01 8.70803773e-01
1.03191471e+00 -3.70981991e-01 4.10121500e-01 5.27027786e-01
3.14921081e-01 -7.74621032e-03 -7.97194302e-01 1.73858202e+00
4.68292087e-01 -8.73563766e-01 2.08125487e-01 2.50958204e-01
2.15239972e-01 -4.98602211e-01 1.29000414e-02 1.39429450e-01
-1.48868829e-01 -8.41507733e-01 5.62720239e-01 4.32578087e-01
9.04615343e-01 -4.94793087e-01 9.15148795e-01 1.08171567e-01
-1.74048889e+00 -5.40937260e-02 -3.75448823e-01 -1.16737016e-01
2.95994729e-01 5.87345734e-02 -6.60170436e-01 9.06477034e-01
1.00756001e+00 1.00133526e+00 -9.71643865e-01 1.16877449e+00
5.45190394e-01 6.17688417e-01 -3.50320190e-01 2.22978219e-01
-3.00777733e-01 -1.27084419e-01 6.07573926e-01 1.39578021e+00
6.99750006e-01 -1.79089278e-01 1.21966869e-01 2.27701083e-01
3.50067645e-01 -1.43282264e-01 -9.54152524e-01 -1.78130031e-01
1.35548249e-01 7.48499036e-01 -7.46693611e-01 -2.37229124e-01
-6.50364459e-01 1.28623080e+00 -1.85829774e-01 4.38061386e-01
-8.87753010e-01 2.03092426e-01 8.87426913e-01 3.05149049e-01
-2.08499860e-02 -4.61991072e-01 5.51643297e-02 -1.74981356e+00
1.81077078e-01 -8.22960377e-01 6.78599477e-01 -3.04295152e-01
-1.69121754e+00 2.61208832e-01 9.93987843e-02 -1.60916853e+00
-3.00887614e-01 -7.96296239e-01 -4.12866175e-01 5.85081577e-01
-9.22492146e-01 -1.55961275e+00 -6.37561917e-01 9.00203824e-01
3.30783188e-01 -5.16090631e-01 9.44195628e-01 7.25627065e-01
-8.19902003e-01 9.45365667e-01 8.82027820e-02 5.22164464e-01
8.59935760e-01 -8.44263554e-01 7.93106079e-01 1.10543787e+00
-5.09375073e-02 7.55623102e-01 8.74624729e-01 -1.02515328e+00
-1.67255378e+00 -8.25549006e-01 7.33187854e-01 -8.28011513e-01
3.94833833e-01 -2.50827253e-01 -8.03703845e-01 5.21411359e-01
-1.32756531e-01 -1.65411040e-01 8.16209674e-01 -3.06734055e-01
-2.00067326e-01 -5.05678095e-02 -1.49975944e+00 5.50260544e-01
1.48071802e+00 -5.21079540e-01 -5.95530212e-01 4.47804760e-03
-1.32463515e-01 -1.24772549e-01 -9.58660543e-01 4.58616078e-01
1.08457017e+00 -6.79448187e-01 1.40092504e+00 -5.46976089e-01
-3.50434184e-01 -4.02497977e-01 -3.67675394e-01 -9.57039356e-01
-3.37554187e-01 -4.81771380e-01 -1.94030061e-01 1.35409796e+00
-3.27045232e-01 -7.91917562e-01 1.26860094e+00 1.02546060e+00
6.16281867e-01 9.18940976e-02 -1.05384481e+00 -1.28729320e+00
-3.56139541e-01 -2.38127723e-01 7.63731241e-01 1.09023011e+00
-2.78453846e-02 -3.55182886e-01 -1.04833519e+00 4.75898385e-02
1.24404478e+00 -2.35474452e-01 1.07202208e+00 -1.34123373e+00
-8.34529251e-02 -9.24602970e-02 -1.16103089e+00 -5.58118403e-01
1.21977799e-01 -5.04664242e-01 -3.77656847e-01 -1.14692998e+00
3.76892060e-01 -1.08546533e-01 -9.56880003e-02 2.36271277e-01
-1.67744368e-01 5.51628709e-01 2.72379220e-01 6.29131019e-01
-2.79120773e-01 1.34895444e-01 8.62185657e-01 -4.96617854e-01
5.35639338e-02 -1.83532131e-04 -4.07944381e-01 5.79571247e-01
7.57403314e-01 1.64508987e-02 -3.96882147e-02 -2.65071213e-01
-5.30820906e-01 -3.73679906e-01 8.62896085e-01 -1.45980585e+00
3.32000434e-01 -1.99475974e-01 6.59658372e-01 -3.07820141e-01
4.29066956e-01 -9.13476050e-01 6.04323626e-01 5.99227428e-01
4.72851336e-01 3.30582678e-01 1.30055234e-01 6.16884708e-01
-2.16814298e-02 8.21767971e-02 4.73366857e-01 -2.58812755e-01
-1.13629425e+00 3.43181074e-01 -5.66716254e-01 -2.37468228e-01
9.22296107e-01 -1.30457997e+00 -3.63411635e-01 -3.14986289e-01
-6.66282833e-01 -2.87324280e-01 9.86270368e-01 6.17728591e-01
6.44905210e-01 -1.53919196e+00 -7.99815357e-01 3.71618569e-01
4.11643803e-01 -1.07437408e+00 3.28856349e-01 5.19690573e-01
-4.32531834e-01 5.02374709e-01 -6.69894636e-01 -6.38993323e-01
-1.74860084e+00 4.16154444e-01 3.59498799e-01 -8.16997513e-02
-5.77205062e-01 1.25886098e-01 -4.33658034e-01 -2.21498981e-01
-3.76647487e-02 1.89796969e-01 -2.42804185e-01 1.18156850e-01
6.31450295e-01 5.11268497e-01 8.91997814e-02 -1.50006115e+00
-7.34723568e-01 9.04838562e-01 6.14542924e-02 9.39084962e-02
9.12431538e-01 -5.22557318e-01 3.36970150e-01 5.76470941e-02
9.71879423e-01 -2.59582162e-01 -9.52296972e-01 1.20721228e-01
1.06931396e-01 -9.73286271e-01 -8.16141903e-01 -2.43437201e-01
-9.55421448e-01 2.71215767e-01 1.11296093e+00 2.58749813e-01
1.07669854e+00 -6.66099787e-01 9.48524475e-01 3.25002372e-01
1.09459972e+00 -1.24409699e+00 -1.18082538e-01 2.92050153e-01
4.23696250e-01 -1.47973049e+00 -1.21323518e-01 -4.37584430e-01
-5.16035318e-01 9.87735093e-01 5.45141399e-01 5.62494844e-02
2.09476635e-01 5.66751696e-03 -1.48423955e-01 2.51874417e-01
1.83547214e-02 -1.81586608e-01 1.25632972e-01 1.23065197e+00
2.70557195e-01 1.89077348e-01 -2.33910754e-01 5.28727829e-01
-2.99484432e-01 3.59883189e-01 4.74818528e-01 1.06821978e+00
-1.00381278e-01 -1.19284511e+00 -9.57393229e-01 4.80079681e-01
-4.20111746e-01 4.07676548e-01 -4.23988253e-01 7.20774055e-01
9.37727019e-02 1.12654662e+00 -2.85376370e-01 -7.91364253e-01
4.47612345e-01 1.56255037e-01 7.35755146e-01 -1.22038007e-01
-4.54731166e-01 -6.17830396e-01 4.38035131e-01 -2.59030312e-01
-9.56973493e-01 -8.92669082e-01 -4.76781547e-01 -1.02434802e+00
1.38378337e-01 -1.32870063e-01 1.15430661e-01 8.37471902e-01
8.80136788e-02 -1.45793065e-01 1.97272256e-01 -9.94159460e-01
-3.21407586e-01 -5.12602448e-01 -6.02060378e-01 9.90959585e-01
1.30515397e-01 -8.99854720e-01 -2.42356762e-01 5.57973504e-01] | [14.27123737335205, 1.3794318437576294] |
3fd8b58c-f9fa-4d06-b2a7-574282442ef7 | consistent-multi-granular-rationale | 2305.09400 | null | https://arxiv.org/abs/2305.09400v1 | https://arxiv.org/pdf/2305.09400v1.pdf | Consistent Multi-Granular Rationale Extraction for Explainable Multi-hop Fact Verification | The success of deep learning models on multi-hop fact verification has prompted researchers to understand the behavior behind their veracity. One possible way is erasure search: obtaining the rationale by entirely removing a subset of input without compromising the veracity prediction. Although extensively explored, existing approaches fall within the scope of the single-granular (tokens or sentences) explanation, which inevitably leads to explanation redundancy and inconsistency. To address such issues, this paper explores the viability of multi-granular rationale extraction with consistency and faithfulness for explainable multi-hop fact verification. In particular, given a pretrained veracity prediction model, both the token-level explainer and sentence-level explainer are trained simultaneously to obtain multi-granular rationales via differentiable masking. Meanwhile, three diagnostic properties (fidelity, consistency, salience) are introduced and applied to the training process, to ensure that the extracted rationales satisfy faithfulness and consistency. Experimental results on three multi-hop fact verification datasets show that the proposed approach outperforms some state-of-the-art baselines. | ['Deyu Zhou', 'Yingjie Zhu', 'Jiasheng Si'] | 2023-05-16 | null | null | null | null | ['fact-verification'] | ['natural-language-processing'] | [ 2.62314022e-01 4.82910812e-01 -6.59861147e-01 -4.39062923e-01
-9.48577762e-01 -2.49475956e-01 5.16878843e-01 3.18749934e-01
3.23314905e-01 9.31928754e-01 6.41892254e-01 -3.55456144e-01
-1.70594931e-01 -5.17544806e-01 -6.84311330e-01 -4.93765473e-01
2.00794697e-01 -1.10884428e-01 -5.28534770e-01 1.77208949e-02
7.24509656e-01 6.55198470e-02 -1.42142165e+00 6.08422458e-01
1.21443164e+00 9.13407564e-01 -4.90429580e-01 5.44471681e-01
-3.46895084e-02 1.14657140e+00 -5.40483654e-01 -8.66915822e-01
1.77361280e-01 -7.48002887e-01 -6.63677394e-01 1.38929531e-01
5.00351369e-01 -5.48470914e-01 -3.55744809e-01 1.13683796e+00
1.28815174e-01 -1.66415274e-01 6.11431420e-01 -1.55097735e+00
-1.47926342e+00 1.03510594e+00 -6.52518034e-01 2.24717513e-01
2.65290976e-01 3.50720912e-01 1.63590908e+00 -1.01957989e+00
4.18165624e-01 7.77348638e-01 6.74625218e-01 6.42774105e-01
-9.17607248e-01 -7.04595566e-01 6.45396858e-02 3.92721474e-01
-1.28843808e+00 -6.46598637e-01 9.24654663e-01 -1.69930398e-01
1.03720164e+00 4.02769864e-01 3.52613598e-01 1.04028368e+00
2.70828962e-01 9.36604679e-01 1.03915381e+00 -2.94643939e-01
1.75129279e-01 3.42693567e-01 2.29598612e-01 8.83412480e-01
5.88285327e-01 6.81435838e-02 -1.00485826e+00 -2.82413244e-01
2.61661917e-01 -1.03093140e-01 -5.71771979e-01 7.52738267e-02
-1.11984444e+00 9.33367074e-01 4.58158046e-01 2.91393012e-01
-5.11368454e-01 3.24297547e-01 3.25038642e-01 3.48123729e-01
2.81794876e-01 5.41263759e-01 -4.41012919e-01 -1.03811271e-01
-1.27612948e+00 1.88303515e-01 5.29367864e-01 8.21497381e-01
5.29939294e-01 3.40704858e-01 -5.38213313e-01 1.56261161e-01
4.28699255e-01 2.38101512e-01 2.81628519e-01 -7.09751666e-01
9.05908287e-01 6.09321833e-01 9.12168473e-02 -1.29659557e+00
2.28500441e-02 -7.16320515e-01 -1.11916089e+00 -1.90079629e-01
-1.20126270e-01 2.82081276e-01 -8.58316898e-01 1.67954707e+00
5.32941893e-02 3.11133116e-01 2.46404901e-01 1.07166433e+00
7.78442442e-01 4.22720820e-01 1.42172083e-01 -3.00880998e-01
1.18518627e+00 -9.85627770e-01 -8.72337043e-01 -2.18956828e-01
6.63904309e-01 -5.69766045e-01 1.18277037e+00 3.64498943e-01
-1.07960737e+00 -3.50591511e-01 -1.27624249e+00 -2.37458780e-01
-5.98008819e-02 2.93672323e-01 7.95491397e-01 9.57052886e-01
-8.22485566e-01 7.11433291e-01 -2.60205984e-01 3.18193704e-01
6.22409999e-01 1.15743272e-01 -3.80992919e-01 -1.99058339e-01
-1.58344948e+00 7.22720623e-01 1.71651527e-01 1.68764219e-01
-7.90689230e-01 -6.41184866e-01 -8.60347211e-01 6.21959269e-01
3.77082586e-01 -9.39520061e-01 8.56231391e-01 -7.41075993e-01
-1.01841152e+00 5.07594645e-01 -2.76555151e-01 -5.30965269e-01
5.23424625e-01 -1.99633479e-01 -4.41845745e-01 3.46955284e-02
2.12155834e-01 3.18030745e-01 9.75251257e-01 -1.38469112e+00
-2.23265544e-01 2.08161827e-02 -1.11730797e-02 -6.30372092e-02
-4.76768315e-01 -2.74873286e-01 4.00162617e-06 -7.49425590e-01
4.13045883e-01 -5.71689129e-01 2.47001678e-01 -1.45997599e-01
-1.09055078e+00 -6.19504042e-02 5.20849407e-01 -1.01595140e+00
1.44731402e+00 -2.07106495e+00 -2.13863906e-02 1.09778427e-01
5.54351926e-01 -1.29865423e-01 -1.63188592e-01 3.90739799e-01
-1.38511732e-01 9.15769339e-01 -3.51828486e-01 -7.68988371e-01
3.22714865e-01 -9.93662626e-02 -7.47668624e-01 5.42673528e-01
4.31401998e-01 9.89941359e-01 -6.16821706e-01 -5.79910934e-01
-9.02329683e-02 3.11374158e-01 -7.27280974e-01 -5.80033436e-02
-2.26495102e-01 3.46032321e-01 -5.54759979e-01 1.02807987e+00
7.58100629e-01 -7.50888109e-01 -1.07271843e-01 -2.00813673e-02
2.01112881e-01 8.04751754e-01 -8.73222291e-01 1.31627440e+00
-2.48044193e-01 6.11581743e-01 -4.77864087e-01 -7.10750580e-01
8.69743109e-01 5.72192609e-01 3.56390506e-01 -5.08838892e-01
7.98461065e-02 3.98700953e-01 -2.76281107e-02 -6.48810327e-01
9.33483064e-01 -4.25836563e-01 -1.40019894e-01 7.02550769e-01
-5.40243685e-01 2.04287976e-01 -2.00908318e-01 5.41358113e-01
1.01075232e+00 -1.64887398e-01 4.91179556e-01 3.03904206e-01
5.13362527e-01 5.17402254e-02 7.67837763e-01 7.98478186e-01
-4.84850734e-01 7.67722607e-01 6.68253422e-01 -3.22407752e-01
-1.09563768e+00 -8.03933144e-01 1.37629598e-01 2.22076043e-01
2.84528375e-01 -4.64637011e-01 -3.06007028e-01 -9.03721333e-01
1.25962287e-01 1.09139526e+00 -8.08590233e-01 -2.51726598e-01
-2.46510774e-01 -4.37797219e-01 6.99562609e-01 4.14965242e-01
8.27082217e-01 -9.23958659e-01 -2.98661649e-01 3.66476700e-02
-6.71075821e-01 -1.10319519e+00 -4.48767722e-01 -2.49761775e-01
-6.93045914e-01 -9.89244640e-01 -7.57690370e-02 -8.50956589e-02
6.73783183e-01 4.70042706e-01 9.28360045e-01 9.92311299e-01
1.14551723e-01 -1.24658719e-01 -2.87984669e-01 2.52027392e-01
-4.28704709e-01 -5.70778474e-02 -2.46390980e-02 -7.70166423e-03
2.91996598e-01 -4.97560829e-01 -4.92704809e-01 -1.63917691e-01
-8.15022230e-01 3.08153033e-01 8.93645108e-01 9.43692625e-01
4.85628724e-01 1.39288917e-01 7.92931080e-01 -7.68054247e-01
6.56181931e-01 -6.37438416e-01 -4.04537618e-02 4.90594715e-01
-9.62428570e-01 1.94528967e-01 7.43389785e-01 -1.99335739e-01
-8.27158689e-01 -3.25214177e-01 7.69631863e-02 -5.12852550e-01
1.03520997e-01 8.65257680e-01 -2.32360691e-01 2.10960031e-01
4.48213488e-01 5.10591090e-01 -5.25039196e-01 -2.09130496e-02
4.68592823e-01 5.85414767e-01 3.83454949e-01 -1.41816959e-01
1.11681616e+00 3.44538987e-01 -2.04985529e-01 -1.19021982e-01
-9.79472816e-01 -1.72384810e-02 -2.90478766e-01 -1.37031719e-01
6.56402707e-01 -8.03843319e-01 -7.80817032e-01 -9.64600891e-02
-1.41173208e+00 3.10208708e-01 -2.00028345e-01 3.90269756e-01
-2.38560557e-01 4.28915888e-01 -5.54671526e-01 -1.07357156e+00
-4.16446745e-01 -9.30575371e-01 1.03507686e+00 1.90446779e-01
-3.56882662e-01 -6.97080314e-01 -2.92234510e-01 9.21581745e-01
2.49082938e-01 2.35745177e-01 1.18861544e+00 -7.06520915e-01
-1.19596887e+00 -4.37906533e-01 -4.33613509e-01 2.47254670e-02
8.89594629e-02 1.31938076e-02 -1.02484500e+00 8.66315588e-02
1.48008749e-01 -4.00140494e-01 1.12412345e+00 2.63614565e-01
1.20913208e+00 -9.56111431e-01 -1.09394901e-01 4.23079312e-01
1.43005371e+00 -4.58929241e-01 6.99974120e-01 3.50503474e-01
6.48926198e-01 4.05543357e-01 5.75110257e-01 7.27115631e-01
6.59795046e-01 2.69570053e-01 6.85444772e-01 1.26601353e-01
-1.68010771e-01 -5.00599980e-01 2.15041131e-01 6.05325162e-01
2.28810921e-01 -6.61720932e-01 -5.95142663e-01 6.30750000e-01
-1.76703537e+00 -1.44623351e+00 -4.21996921e-01 1.88144016e+00
6.41740441e-01 2.04089835e-01 -3.91157269e-01 6.03303492e-01
6.41401887e-01 5.18285990e-01 -5.47040105e-01 -4.86712009e-01
-3.52741212e-01 -6.45094752e-01 2.76276678e-01 5.83309114e-01
-6.54094577e-01 6.86130881e-01 5.96848631e+00 6.53475165e-01
-1.11658251e+00 1.83648452e-01 8.54503512e-01 -1.45738974e-01
-1.25404239e+00 2.56910503e-01 -5.63681722e-01 4.51936066e-01
4.08483148e-01 -1.19627401e-01 4.02037024e-01 7.31656909e-01
5.03725588e-01 -1.92184409e-03 -9.81905460e-01 6.79973781e-01
2.83078581e-01 -1.74986851e+00 3.40037532e-02 9.61958691e-02
1.09966350e+00 -5.35482049e-01 4.12270963e-01 3.37046534e-01
-3.88955884e-02 -1.24973083e+00 9.54896867e-01 4.55641747e-01
6.78440988e-01 -4.19449776e-01 8.22149992e-01 3.63194197e-01
-9.96009946e-01 -2.08738089e-01 -1.98119432e-01 -1.07122965e-01
3.01744998e-01 9.72528577e-01 -9.08759236e-01 8.14101100e-01
1.78828135e-01 6.85084939e-01 -4.59525108e-01 7.02384830e-01
-7.26280093e-01 6.52371883e-01 1.36852652e-01 -9.04431418e-02
1.72293216e-01 4.16984797e-01 4.63620454e-01 7.22397327e-01
4.54913437e-01 2.07062334e-01 -2.31856257e-01 1.37634110e+00
-6.12936974e-01 -1.91674531e-01 -3.41509253e-01 -3.29727948e-01
6.48384631e-01 8.86320293e-01 -4.19609606e-01 -5.36481917e-01
-3.78530800e-01 1.04553699e+00 3.18594575e-01 2.93817252e-01
-1.23177505e+00 1.10562079e-01 5.29418707e-01 -1.18719209e-02
3.15689057e-01 -1.04836322e-01 -1.37575698e+00 -1.41356313e+00
4.27554071e-01 -1.00227046e+00 4.36908394e-01 -8.83524656e-01
-1.24445617e+00 5.73583126e-01 -4.26521480e-01 -1.38070059e+00
-2.63682753e-02 1.12401597e-01 -7.81763077e-01 9.87096071e-01
-2.21343279e+00 -1.33004320e+00 -2.76118547e-01 4.18481141e-01
5.52178204e-01 -1.52321324e-01 5.31899452e-01 1.33873239e-01
-5.06449521e-01 8.39358032e-01 -3.64376754e-01 -1.46903336e-01
3.65299791e-01 -1.04338133e+00 3.93770635e-01 1.27190650e+00
2.60758519e-01 9.20158327e-01 1.07342827e+00 -1.06706583e+00
-1.38815749e+00 -8.58069777e-01 1.75265718e+00 -3.52898598e-01
6.09466195e-01 1.85529701e-02 -1.15776825e+00 5.45780122e-01
1.11107081e-01 1.47197014e-02 6.46567643e-01 2.09701091e-01
-6.72190607e-01 2.25149482e-01 -1.32600820e+00 6.41692102e-01
8.68710518e-01 -7.96564400e-01 -6.44801259e-01 3.22259307e-01
7.35644817e-01 -2.49183148e-01 -4.08970296e-01 2.39498079e-01
4.42825735e-01 -1.20310748e+00 6.09061360e-01 -6.43410027e-01
1.23194921e+00 -3.68276000e-01 -7.98971206e-02 -8.25821757e-01
-3.27284932e-01 -6.90309346e-01 -4.42825526e-01 1.27471185e+00
5.78472674e-01 -4.58949894e-01 9.42313492e-01 9.74294722e-01
-2.38518834e-01 -7.57953286e-01 -1.11639285e+00 -6.05723321e-01
-1.75935343e-01 -4.40317571e-01 1.01222718e+00 1.17537463e+00
3.47486287e-01 1.40245095e-01 -1.05513799e+00 6.96847677e-01
5.94628096e-01 6.04319870e-01 5.35314620e-01 -7.62569785e-01
-4.40864533e-01 -4.40914303e-01 -1.09076023e-01 -9.22898412e-01
1.88954264e-01 -9.35997248e-01 7.51410797e-02 -1.72479963e+00
4.37682360e-01 -3.93424273e-01 -2.36394927e-01 7.60538280e-01
-6.35417819e-01 3.63954306e-02 1.44193187e-01 5.48576295e-01
-3.34755629e-01 8.55411172e-01 1.13158762e+00 -3.50805730e-01
2.43956409e-02 -2.15402633e-01 -1.21889472e+00 2.34459579e-01
8.79313171e-01 -8.76012504e-01 -4.30807054e-01 -4.40458894e-01
5.31794846e-01 3.79454285e-01 6.42846346e-01 -6.55481875e-01
3.43584150e-01 -4.07561123e-01 1.30014420e-01 -5.27409077e-01
3.08820546e-01 -6.33417547e-01 1.11850813e-01 5.52012801e-01
-7.29671299e-01 1.06524276e-02 -3.31906974e-02 7.55035996e-01
-2.55297035e-01 -2.20469654e-01 4.11674142e-01 1.33031055e-01
-4.03426558e-01 3.13403934e-01 1.11830242e-01 -1.87308446e-01
6.96169138e-01 -3.15644175e-01 -8.11083198e-01 -5.48027158e-01
-3.40565234e-01 -2.22492237e-02 4.01965588e-01 3.43670100e-01
9.39535797e-01 -1.40622175e+00 -9.44481075e-01 1.63506553e-01
1.95363745e-01 -6.38026178e-01 5.76150715e-01 1.03666139e+00
1.20201737e-01 5.88498712e-01 5.96379600e-02 -1.34433448e-01
-1.10344136e+00 5.93308687e-01 1.62304163e-01 -5.03083408e-01
-5.65554559e-01 8.97872984e-01 -1.85532004e-01 5.75743541e-02
-8.38276893e-02 -3.19761574e-01 5.68662100e-02 -4.04085666e-01
3.75161886e-01 1.15211578e-02 -8.62814859e-02 -4.89051312e-01
-4.69105750e-01 2.27776930e-01 1.14008389e-01 -5.11010885e-02
1.14061749e+00 -2.24592134e-01 1.09574869e-01 1.26967356e-01
9.46584404e-01 3.32623690e-01 -1.13604474e+00 -4.71018031e-02
-3.67155224e-02 -1.05492580e+00 2.55192164e-02 -9.36369300e-01
-1.11890125e+00 9.32963192e-01 -2.38867849e-01 4.80344743e-01
9.74758983e-01 -1.72004700e-01 9.35608923e-01 6.84509575e-02
-2.60517281e-02 -7.40322471e-01 2.61994332e-01 1.19590491e-01
1.05415952e+00 -1.54286289e+00 2.53932714e-01 -5.73581696e-01
-9.95876253e-01 9.76086259e-01 5.35366893e-01 2.33308315e-01
1.27093613e-01 -3.76909636e-02 -1.91348106e-01 -3.91646475e-01
-9.75806415e-01 2.30902880e-01 3.80207986e-01 2.02938631e-01
4.27190274e-01 -1.22272924e-01 -3.83466899e-01 9.88267660e-01
-2.61395872e-01 4.05936353e-02 8.84819925e-01 5.12209773e-01
-4.31489646e-01 -7.02076077e-01 -2.74696499e-01 4.98069525e-01
-4.22712475e-01 -6.78467095e-01 -5.26553035e-01 6.68714225e-01
1.28298357e-01 1.16043234e+00 -5.69445193e-01 -6.41261578e-01
-1.20660119e-01 6.11930937e-02 -1.01213306e-01 -4.21133459e-01
-7.48301566e-01 -3.88354957e-01 2.94735432e-01 -3.41079772e-01
-2.24524915e-01 -6.94143653e-01 -1.43941820e+00 -8.52890372e-01
-6.30697191e-01 1.34124130e-01 6.47911310e-01 1.35026813e+00
4.05130237e-01 4.67141867e-01 7.80997813e-01 -7.08189383e-02
-6.80419683e-01 -8.18456650e-01 -4.40107793e-01 3.46730679e-01
7.97399461e-01 -3.82244855e-01 -7.86173284e-01 7.69646540e-02] | [9.162102699279785, 7.83399772644043] |
e667fc62-9171-4085-b363-e189e617e505 | adaptive-hybrid-activation-function-for-deep | null | null | https://doi.org/10.20535/SRIT.2308-8893.2022.1.07 | http://journal.iasa.kpi.ua/article/download/259203/255848/596453 | Adaptive hybrid activation function for deep neural networks | The adaptive hybrid activation function (AHAF) is proposed that combines the properties of the rectifier units and the squashing functions. The proposed function can be used as a drop-in replacement for ReLU, SiL and Swish activations for deep neural networks and can evolve to one of such functions during the training. The effectiveness of the function was evaluated on the image classification task using the Fashion-MNIST and CIFAR-10 datasets. The evaluation shows that the neural networks with AHAF activations achieve better classification accuracy comparing to their base implementations that use ReLU and SiL. A double-stage parameter tuning process for training the neural networks with AHAF is proposed. The proposed approach is sufficiently simple from the implementation standpoint and provides high performance for the neural network training process. | ['Serhii Kostiuk', 'Yevgeniy Bodyanskiy'] | 2022-04-25 | null | null | null | system-research-and-information-technologies | ['activation-function-synthesis', 'architecture-search'] | ['methodology', 'methodology'] | [-1.03910547e-02 1.54131085e-01 -6.64058402e-02 -5.12076735e-01
2.18546063e-01 -2.69959778e-01 6.43287241e-01 -2.13217899e-01
-9.56390381e-01 8.56731236e-01 -1.97064295e-01 -4.23918605e-01
-5.39095819e-01 -5.95036685e-01 -4.34279621e-01 -9.62438047e-01
8.37448090e-02 3.00020874e-01 3.06705952e-01 -3.97525162e-01
1.50998635e-02 8.19948673e-01 -1.51771450e+00 1.59201220e-01
6.84198558e-01 1.09380138e+00 1.33132041e-01 5.23609161e-01
2.98961997e-03 8.13555002e-01 -9.88050044e-01 -9.83498693e-02
3.23072225e-01 -1.67922556e-01 -5.91707289e-01 -4.06338215e-01
1.34809524e-01 -2.00474467e-02 -3.74357015e-01 5.13734698e-01
7.63833702e-01 2.62165338e-01 5.81189990e-01 -1.14523709e+00
-6.53859675e-01 9.78690684e-01 -9.52830762e-02 6.24391198e-01
-3.07458133e-01 -5.40622398e-02 6.22567415e-01 -7.22838402e-01
2.56955147e-01 9.63726521e-01 7.60797143e-01 7.14325607e-01
-1.22686183e+00 -8.66593957e-01 -1.06850034e-02 1.61527529e-01
-1.56555414e+00 -1.69960529e-01 6.34193003e-01 -9.92434099e-02
1.20343709e+00 1.24334872e-01 6.98227346e-01 9.49968576e-01
8.97751376e-02 7.83429742e-01 1.01365018e+00 -5.28313518e-01
3.66808951e-01 5.68773329e-01 5.71824491e-01 3.29747498e-01
5.13347566e-01 1.73440099e-01 -2.51725078e-01 6.18382841e-02
8.02983582e-01 -2.30103046e-01 -2.89412737e-01 4.92279157e-02
-5.53208590e-01 8.18319201e-01 8.64766657e-01 9.84276712e-01
-6.57105446e-01 4.50725555e-01 3.85242552e-01 6.53656960e-01
3.16902846e-01 8.03338289e-01 -4.37428057e-01 1.16517641e-01
-9.16901708e-01 2.50389218e-01 6.08693421e-01 5.04888773e-01
3.61660868e-01 8.36471736e-01 -3.04958642e-01 8.75854969e-01
4.19669412e-02 5.20748235e-02 8.27541649e-01 -5.57061315e-01
-1.78918079e-01 7.67315626e-01 -9.32785124e-02 -2.67046601e-01
-6.46984518e-01 -7.89121389e-01 -8.08421135e-01 5.63755333e-01
2.46930286e-01 -3.17951202e-01 -1.23033214e+00 1.58756340e+00
-6.46477565e-02 1.89719745e-03 4.68235403e-01 4.38720852e-01
1.17976654e+00 8.99237514e-01 1.57806426e-01 5.96568882e-02
1.14403868e+00 -8.26353908e-01 -8.83558035e-01 7.59330392e-02
5.49810290e-01 -2.20978320e-01 9.10950184e-01 6.39714777e-01
-1.17855012e+00 -9.81809199e-01 -1.28247440e+00 3.54775846e-01
-8.29646587e-01 3.65732878e-01 6.42053604e-01 1.00869846e+00
-1.31867886e+00 6.50660515e-01 -5.08778930e-01 -2.22668573e-01
3.30466896e-01 1.02876151e+00 -1.78645238e-01 3.90476495e-01
-1.30555153e+00 1.14982820e+00 1.08988559e+00 3.98033708e-01
-4.59240049e-01 -4.83160645e-01 -5.04461110e-01 3.51674587e-01
-1.96628988e-01 -4.11654651e-01 1.00220239e+00 -1.47631276e+00
-1.67120659e+00 3.70460898e-01 4.09844905e-01 -9.52761948e-01
2.43801311e-01 -9.93670747e-02 -4.86182868e-01 3.53144482e-02
-8.21884036e-01 9.42725539e-01 9.35488403e-01 -1.16960549e+00
-5.34255445e-01 5.63920960e-02 -4.92082201e-02 -4.51526977e-02
-6.89527631e-01 -1.29707694e-01 3.16868424e-01 -8.02989662e-01
-3.03019613e-01 -6.33054197e-01 6.47392496e-02 -4.14727420e-01
9.66356546e-02 -5.96454918e-01 1.10549414e+00 -1.82100937e-01
1.51401424e+00 -2.30279160e+00 -1.61876865e-02 6.28166854e-01
-2.41779223e-01 9.27235126e-01 -1.46545127e-01 1.07410453e-01
-4.43939298e-01 -7.20518529e-02 -4.27576080e-02 2.47739512e-03
-1.97910503e-01 6.12105846e-01 2.15833504e-02 2.67297685e-01
1.67300671e-01 1.00831842e+00 -4.67680901e-01 1.33353502e-01
2.33629063e-01 7.62533844e-01 -4.60151374e-01 1.20109238e-01
1.04032539e-01 3.05180848e-02 -1.61887169e-01 1.99616015e-01
5.10107517e-01 1.65692329e-01 -4.67721708e-02 -3.62843573e-01
-1.87690049e-01 -9.57192630e-02 -9.33811605e-01 1.15321541e+00
-4.13962245e-01 5.94084620e-01 -6.11355975e-02 -1.08496499e+00
1.47439981e+00 6.58526123e-01 1.12920247e-01 -7.75061309e-01
6.84037745e-01 4.38577116e-01 5.02373815e-01 -3.03065777e-01
3.92814368e-01 -8.70806724e-02 5.99367209e-02 1.06072880e-01
5.49724877e-01 2.17886358e-01 3.16394091e-01 -2.38861501e-01
8.60983431e-01 1.32712603e-01 1.41742066e-01 -7.02998817e-01
9.38859403e-01 -3.16257149e-01 4.02724385e-01 6.84638083e-01
2.14584887e-01 -9.74315871e-03 2.02992260e-01 -7.87256777e-01
-1.08459496e+00 -7.63120234e-01 -2.74983913e-01 1.03222311e+00
-1.87990263e-01 2.85383891e-02 -7.78982580e-01 -5.18542945e-01
-6.06179908e-02 1.01316476e+00 -7.13630497e-01 -5.33059478e-01
-8.45530987e-01 -1.01985717e+00 8.63166630e-01 7.30658412e-01
8.21829259e-01 -1.37835622e+00 -8.46739352e-01 2.54093915e-01
6.56291127e-01 -7.01286435e-01 2.58267432e-01 8.82479966e-01
-9.46050406e-01 -5.95204175e-01 -7.84930706e-01 -1.00551605e+00
6.81826174e-01 -3.94590735e-01 7.22399771e-01 9.27813798e-02
-9.61808022e-03 8.07651654e-02 -4.90779251e-01 -5.18001735e-01
-5.38645864e-01 3.72416466e-01 2.47348454e-02 1.74450409e-02
2.72304326e-01 -7.43608356e-01 -4.65130240e-01 2.05748975e-01
-1.03407371e+00 -2.99667984e-01 6.39245331e-01 1.18177640e+00
2.64424048e-02 4.15429831e-01 1.10032272e+00 -8.68217468e-01
1.03720117e+00 -1.91091657e-01 -5.41627884e-01 7.60128051e-02
-8.93566787e-01 3.53968292e-01 8.84150207e-01 -7.78051436e-01
-1.09358633e+00 6.19644523e-02 -5.16766250e-01 -4.82898712e-01
-7.19899535e-02 2.02908427e-01 1.00026630e-01 -3.56000006e-01
8.69507611e-01 -3.91168892e-02 -8.04793015e-02 -6.69971943e-01
3.24060135e-02 7.73024023e-01 6.33229136e-01 -3.12945306e-01
5.05781889e-01 -6.70156255e-02 -3.29197764e-01 -7.70424783e-01
-2.52765924e-01 -1.12674467e-01 -4.22100455e-01 -1.81629956e-01
7.13752031e-01 -6.22394860e-01 -7.67330825e-01 7.05257833e-01
-9.33249772e-01 -4.37434077e-01 -7.32255578e-01 4.62037116e-01
2.80133206e-02 -3.57856512e-01 -6.18191242e-01 -9.55855727e-01
-6.31161809e-01 -1.05414069e+00 8.53615329e-02 7.34204829e-01
-1.76339984e-01 -1.07524145e+00 -2.92601794e-01 -1.30605489e-01
8.25978875e-01 5.54809347e-03 1.16944480e+00 -1.11701667e+00
1.25342369e-01 -3.63066733e-01 9.38489586e-02 8.83364677e-01
-6.70370534e-02 1.21501714e-01 -1.23274350e+00 -4.86761242e-01
-1.18651785e-01 -2.78670937e-01 1.05201423e+00 4.73792344e-01
9.79909658e-01 -2.79771417e-01 3.55833173e-02 5.19224823e-01
1.52289951e+00 8.70201230e-01 7.67415047e-01 6.69467747e-01
1.73472792e-01 2.08178833e-01 1.64020266e-02 1.88585848e-01
-2.27412462e-01 4.81037050e-01 4.58620310e-01 -4.01070535e-01
-3.61308083e-02 2.22611994e-01 3.00889492e-01 6.18381739e-01
-3.47746283e-01 -5.09025395e-01 -6.66971028e-01 4.48351651e-01
-1.65181243e+00 -8.02913189e-01 3.44868675e-02 2.11405683e+00
4.65138495e-01 5.93650579e-01 7.40076229e-02 8.20961416e-01
5.31576753e-01 -1.46299705e-01 -2.95908064e-01 -1.19559085e+00
-3.44881862e-01 9.00330484e-01 6.13571882e-01 3.04736018e-01
-9.04722929e-01 6.14537716e-01 6.93351078e+00 7.70283222e-01
-1.46223629e+00 1.96993172e-01 4.67164159e-01 1.35518285e-02
4.87919524e-03 -4.13713783e-01 -7.69952357e-01 2.89415687e-01
1.30770135e+00 2.24230990e-01 2.74197787e-01 7.13724554e-01
8.27512369e-02 -4.96014506e-02 -8.28685343e-01 7.12877274e-01
-1.53987691e-01 -1.25404418e+00 1.99653387e-01 -3.26817930e-01
5.99343121e-01 -6.97165504e-02 8.13465938e-02 6.16115570e-01
1.06955372e-01 -1.07579708e+00 7.73648143e-01 3.09668481e-01
4.35462654e-01 -1.04876614e+00 1.15847290e+00 1.21045344e-01
-1.01010048e+00 -8.35578799e-01 -2.25187257e-01 -3.10368072e-02
-1.47675753e-01 1.56270966e-01 -8.36186230e-01 4.96801823e-01
6.52258039e-01 1.84442356e-01 -6.74641609e-01 1.03783846e+00
-7.47344345e-02 7.69739091e-01 -4.41049635e-01 -2.45034531e-01
6.20909691e-01 -1.59930494e-02 2.27153137e-01 1.27734137e+00
3.07320386e-01 -1.93164974e-01 -4.49071795e-01 7.30788589e-01
-3.62009704e-02 2.43029371e-03 -3.82929087e-01 -1.47132158e-01
4.47159350e-01 1.23715186e+00 -7.58368850e-01 -3.31308186e-01
-1.44126629e-02 6.81739986e-01 -2.90669990e-03 5.12617469e-01
-7.90205956e-01 -6.87440813e-01 1.29802674e-01 2.01926634e-01
4.49424148e-01 1.85406521e-01 -1.96541145e-01 -3.64203125e-01
-3.27023029e-01 -6.49014294e-01 3.67807955e-01 -7.91049302e-01
-8.12144399e-01 1.18495464e+00 2.46632159e-01 -8.34614694e-01
-2.81443119e-01 -9.78096664e-01 -8.83874774e-01 7.26998925e-01
-1.37943077e+00 -1.07951927e+00 -3.44663143e-01 5.78865945e-01
2.46110529e-01 -6.29523277e-01 7.27299690e-01 5.34211457e-01
-7.75711358e-01 7.94346929e-01 3.58799100e-02 6.39388636e-02
-2.86935046e-02 -1.10734093e+00 9.49172005e-02 5.83565652e-01
-1.34595022e-01 5.75892210e-01 7.15278983e-01 2.64892448e-02
-8.81788611e-01 -8.97915304e-01 7.09600091e-01 4.88107532e-01
2.07656175e-01 -3.08843017e-01 -1.01515150e+00 4.56777126e-01
6.49186730e-01 1.24539826e-02 4.44354236e-01 -3.07025969e-01
-6.19578362e-02 -5.45765340e-01 -1.31051826e+00 2.59277582e-01
3.43854368e-01 -5.56609072e-02 -8.33215952e-01 -2.56012022e-01
2.75301516e-01 -1.85857341e-02 -9.79347527e-01 6.48334622e-01
6.38008416e-01 -8.61118734e-01 8.37700665e-01 -2.87040442e-01
-6.00961410e-02 -1.14036538e-01 1.86298326e-01 -1.39137185e+00
-5.30309796e-01 -6.34936035e-01 -5.56451976e-02 1.19386971e+00
6.44855797e-01 -8.48200858e-01 7.12275326e-01 2.73433715e-01
-6.91030100e-02 -9.72910166e-01 -9.99785960e-01 -7.85540342e-01
3.91157940e-02 2.35216931e-01 4.14634407e-01 5.39053917e-01
-3.48785728e-01 7.03586459e-01 -2.31499836e-01 -4.31636512e-01
-2.88857315e-02 -4.75896001e-01 2.66144037e-01 -1.50766635e+00
-2.12652743e-01 -5.98943055e-01 -4.92536098e-01 -5.51471055e-01
2.78319921e-02 -8.71797323e-01 -2.45696530e-01 -1.34463143e+00
-4.93875444e-01 -7.56233871e-01 -8.54644299e-01 8.08791339e-01
2.55056590e-01 3.68593365e-01 1.82759523e-01 -8.88309628e-02
2.56356865e-01 3.85959089e-01 7.83973515e-01 5.77641390e-02
-4.36852932e-01 1.93362013e-01 -4.43254322e-01 4.82787490e-01
1.10105228e+00 -4.54414129e-01 -7.32954979e-01 -1.61498442e-01
1.63711160e-01 -4.11184460e-01 1.02544539e-01 -1.25803006e+00
6.78534880e-02 5.42955875e-01 5.55746496e-01 -3.76577705e-01
4.47442383e-01 -1.15098214e+00 4.05419320e-01 9.37434673e-01
-4.99545753e-01 4.66468126e-01 4.75786000e-01 -1.20295091e-02
-2.15885207e-01 -6.30540848e-01 1.14047110e+00 -8.55021328e-02
-7.48617291e-01 -2.09235147e-01 -5.83681822e-01 -5.73875725e-01
1.04189670e+00 -7.63495028e-01 -2.15849116e-01 -1.05722010e-01
-7.13877261e-01 1.33398265e-01 -1.48494869e-01 4.61581886e-01
5.89512050e-01 -1.28990710e+00 -6.01572156e-01 5.08999288e-01
-3.22684884e-01 -1.69339180e-01 -1.12291470e-01 7.65733361e-01
-6.41649961e-01 3.08827102e-01 -7.70477176e-01 -2.79551476e-01
-1.19984901e+00 4.73587722e-01 7.20681727e-01 -3.24669033e-01
-4.77352619e-01 8.54249418e-01 -1.77672684e-01 -3.16881478e-01
5.29800475e-01 -3.84544998e-01 -6.83122218e-01 1.02982253e-01
3.19715828e-01 3.84218425e-01 2.61051804e-01 -3.55183929e-01
-1.64964870e-01 4.23403531e-02 -1.86192051e-01 1.14228263e-01
1.62529099e+00 3.53081286e-01 4.88512926e-02 4.18296129e-01
9.62691307e-01 -5.02203047e-01 -7.02687979e-01 -6.57705665e-02
7.87343159e-02 2.50430256e-01 4.67020959e-01 -9.88488019e-01
-1.33845770e+00 8.00422490e-01 9.28601384e-01 2.11970404e-01
1.49477124e+00 -5.63480794e-01 4.75440979e-01 5.63138783e-01
-9.77331698e-02 -1.16173685e+00 -1.91517904e-01 3.55932564e-01
7.00712979e-01 -3.20954114e-01 -2.72294376e-02 1.95708677e-01
-4.14702296e-01 1.33926225e+00 7.64358997e-01 -5.31149447e-01
5.48181355e-01 7.03904331e-01 -3.00446935e-02 -1.02600446e-02
-9.03709233e-01 4.54155132e-02 2.42411032e-01 3.14812869e-01
4.55703676e-01 -2.90751487e-01 -7.60423362e-01 5.19286335e-01
1.06643945e-01 2.50550747e-01 3.36232275e-01 1.12308335e+00
-4.98469144e-01 -1.14284015e+00 -1.63856626e-01 4.99184549e-01
-5.28841078e-01 -2.15448011e-02 -3.21055651e-01 1.14168799e+00
5.06391108e-01 4.81738418e-01 2.66525507e-01 -4.73497391e-01
4.97009188e-01 3.51087511e-01 5.34132361e-01 -3.90079245e-02
-1.61236668e+00 -1.52578568e-02 1.56030759e-01 -9.99163389e-02
-5.73479176e-01 6.65248334e-02 -1.26345468e+00 -1.61115170e-01
-5.77578843e-01 5.06492078e-01 6.86408699e-01 7.49474823e-01
1.43247526e-02 9.24915791e-01 5.52796185e-01 -7.58527100e-01
-5.47343194e-01 -1.19083500e+00 -5.70672035e-01 1.23132937e-01
2.13853151e-01 -7.33481765e-01 -1.71400204e-01 -2.33462185e-01] | [8.453715324401855, 3.0694262981414795] |
f15c2cf6-5455-4270-8f4f-b544f34c15b5 | a-stochastic-lqr-model-for-child-order | 2004.13797 | null | https://arxiv.org/abs/2004.13797v1 | https://arxiv.org/pdf/2004.13797v1.pdf | A Stochastic LQR Model for Child Order Placement in Algorithmic Trading | Modern Algorithmic Trading ("Algo") allows institutional investors and traders to liquidate or establish big security positions in a fully automated or low-touch manner. Most existing academic or industrial Algos focus on how to "slice" a big parent order into smaller child orders over a given time horizon. Few models rigorously tackle the actual placement of these child orders. Instead, placement is mostly done with a combination of empirical signals and heuristic decision processes. A self-contained, realistic, and fully functional Child Order Placement (COP) model may never exist due to all the inherent complexities, e.g., fragmentation due to multiple venues, dynamics of limit order books, lit vs. dark liquidity, different trading sessions and rules. In this paper, we propose a reductionism COP model that focuses exclusively on the interplay between placing passive limit orders and sniping using aggressive takeout orders. The dynamic programming model assumes the form of a stochastic linear-quadratic regulator (LQR) and allows closed-form solutions under the backward Bellman equations. Explored in detail are model assumptions and general settings, the choice of state and control variables and the cost functions, and the derivation of the closed-form solutions. | ['Jackie Jianhong Shen'] | 2020-04-28 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-4.93507534e-01 1.09375611e-01 -3.52602482e-01 7.34572336e-02
-3.08012187e-01 -1.20350289e+00 3.36731374e-01 1.47691742e-01
-1.40190095e-01 6.96912289e-01 -2.65886843e-01 -4.19441581e-01
-5.69570184e-01 -6.22620463e-01 -7.17595100e-01 -5.88629603e-01
-9.62484628e-02 1.11654413e+00 3.65073271e-02 -2.25798309e-01
3.86144429e-01 3.97557408e-01 -8.63644838e-01 -2.24671796e-01
7.41149783e-01 1.24751627e+00 -1.24291867e-01 3.84227306e-01
-2.28826135e-01 8.14600527e-01 -4.48137641e-01 -4.66198146e-01
1.08941150e+00 -2.67249882e-01 -1.14793815e-01 2.72045255e-01
-3.47630441e-01 -8.47374380e-01 1.21770643e-01 8.53339255e-01
2.60981381e-01 -3.06320935e-02 4.80628788e-01 -1.20781767e+00
-3.97474766e-01 1.07957554e+00 -6.34832203e-01 2.57305712e-01
-1.83029398e-01 2.08043203e-01 1.15325844e+00 -3.58042657e-01
4.30359036e-01 9.74212587e-01 2.78162032e-01 3.43578547e-01
-1.62830019e+00 -5.67720056e-01 4.72131968e-01 -5.04366696e-01
-8.42606127e-01 5.66075556e-02 4.68715042e-01 -6.20501041e-01
4.52061206e-01 4.40500259e-01 1.15435088e+00 6.71613514e-01
5.18682837e-01 5.83175302e-01 1.11127448e+00 -6.18320182e-02
4.83820736e-01 4.07215565e-01 1.45789191e-01 -1.27122074e-01
7.91529655e-01 2.56077170e-01 -1.89203799e-01 -3.46671343e-01
1.05978584e+00 2.47984380e-02 -4.01428826e-02 -6.47123337e-01
-5.66098332e-01 1.01457369e+00 -1.41588300e-02 7.79805109e-02
-5.15174627e-01 -1.41563386e-01 8.08760375e-02 6.01687133e-01
3.79312277e-01 8.02830875e-01 -5.09307146e-01 4.54631224e-02
-9.87723708e-01 6.92243159e-01 1.22185302e+00 8.07702005e-01
3.58897805e-01 -5.67724369e-02 -1.74167514e-01 1.36060268e-01
2.37745449e-01 3.90259504e-01 1.62120640e-01 -1.23130870e+00
8.09784949e-01 4.81173545e-01 8.53649378e-01 -6.96023703e-01
-2.10132375e-01 -5.91223955e-01 -3.63168448e-01 1.57937735e-01
6.66083038e-01 -6.10955715e-01 -3.17749023e-01 1.50601423e+00
2.10368291e-01 -3.27556551e-01 -2.91257083e-01 1.00006843e+00
-4.21028346e-01 7.89000690e-01 -5.62134922e-01 -6.78895175e-01
1.12893641e+00 -7.43203700e-01 -7.63914704e-01 -3.99183519e-02
2.38786321e-02 -4.95066077e-01 5.81204236e-01 6.80218101e-01
-1.50989389e+00 3.20325643e-01 -7.94741571e-01 3.94063532e-01
-1.38667062e-01 -2.32659668e-01 4.38273430e-01 4.66150254e-01
-1.02235317e+00 5.34204662e-01 -6.58046663e-01 2.06196383e-01
-6.90601468e-02 5.10609925e-01 4.63146001e-01 7.23335207e-01
-1.06942260e+00 5.77553570e-01 -1.09387394e-02 7.10378408e-01
-7.57395446e-01 -1.05157101e+00 -1.84129924e-01 3.71395826e-01
1.10423028e+00 -3.86670977e-01 1.65537369e+00 -1.00037861e+00
-1.62490952e+00 3.08329523e-01 5.32550573e-01 -7.55555093e-01
1.16180408e+00 -3.97958457e-01 3.73303920e-01 -5.56696914e-02
9.23881233e-02 2.13313401e-02 3.86365533e-01 -9.86851454e-01
-4.34106857e-01 -4.29026037e-01 2.02999786e-01 3.87541771e-01
9.17694345e-02 -4.78046713e-03 7.21880645e-02 -6.71513081e-01
1.62908379e-02 -9.59583938e-01 -4.56593126e-01 -4.95613694e-01
-4.16007787e-01 2.24491447e-01 2.82381862e-01 -4.17602599e-01
1.34731150e+00 -1.82537210e+00 5.40503673e-02 2.38347352e-01
-8.67360830e-03 -2.22776383e-01 5.41201591e-01 9.33068156e-01
2.72172898e-01 3.32994074e-01 7.08776563e-02 -7.04533309e-02
4.74381864e-01 4.79117483e-02 -7.55356729e-01 3.33597004e-01
-2.73740143e-01 6.83120191e-01 -5.30027568e-01 -3.05499993e-02
-1.03518546e-01 -2.43991554e-01 -5.24284601e-01 1.34446472e-01
-4.20301944e-01 3.21002066e-01 -8.33453774e-01 6.60414457e-01
6.94685578e-01 -5.64017594e-02 5.56786954e-01 5.45944870e-01
-8.52610886e-01 1.17404342e-01 -1.49237454e+00 6.39539421e-01
-2.58777082e-01 2.46274471e-01 8.84620309e-01 -6.78594887e-01
5.22239387e-01 1.46114975e-01 4.46745515e-01 -5.73634088e-01
2.46833459e-01 5.60770750e-01 -1.55419791e-02 -6.29825071e-02
3.91106367e-01 -3.96291077e-01 -1.18636705e-01 6.60665810e-01
-5.00230730e-01 -2.20175594e-01 2.97410250e-01 9.45680737e-02
8.70187759e-01 -1.57391742e-01 9.90210921e-02 -7.16733336e-01
2.71516889e-02 8.35146233e-02 1.00980234e+00 7.09780991e-01
-3.40122953e-02 2.05134556e-01 1.33807278e+00 -2.44508177e-01
-9.24084246e-01 -8.15809071e-01 -8.35004598e-02 8.52310479e-01
2.92584449e-01 2.66925395e-01 -7.07147181e-01 -1.05433017e-01
6.78911150e-01 6.17258668e-01 -6.60390139e-01 2.00394958e-01
-6.83009326e-01 -7.32581437e-01 -1.35554686e-01 1.05398387e-01
3.11375320e-01 -8.31147015e-01 -1.08722353e+00 4.85420704e-01
3.54095280e-01 -7.30811119e-01 -8.21879208e-01 4.53332335e-01
-8.08679044e-01 -9.73071575e-01 -6.68823898e-01 -1.23928383e-01
5.78017354e-01 -1.36390328e-01 6.87192559e-01 -4.97424901e-01
8.38970318e-02 2.13126063e-01 -1.52805885e-02 -5.58224976e-01
7.75946723e-03 -1.41563164e-02 9.72378924e-02 4.90606010e-01
-1.90029725e-01 -2.36707851e-01 -7.81297445e-01 3.39355350e-01
-7.98242033e-01 1.22940363e-02 6.19679511e-01 6.84173703e-01
5.82500756e-01 1.20662823e-01 4.80822712e-01 -1.00776136e+00
7.85774767e-01 -4.11424428e-01 -1.63707733e+00 4.80498314e-01
-1.01214039e+00 2.12389439e-01 5.59861779e-01 -6.21221840e-01
-1.09617412e+00 -2.02708632e-01 7.08289266e-01 -5.66177666e-01
5.16952097e-01 8.52865204e-02 -2.39873514e-01 5.59080057e-02
-2.87826389e-01 3.94938849e-02 1.32471040e-01 -7.36639082e-01
7.80395418e-02 3.33637118e-01 7.32339621e-02 -9.01559532e-01
7.79902816e-01 2.98858881e-01 -3.21762590e-03 -2.50676811e-01
-4.48231995e-01 5.67983165e-02 -3.13084990e-01 -1.98952645e-01
5.74869215e-01 -4.08170164e-01 -1.26314187e+00 5.88403285e-01
-8.42299223e-01 -7.13570297e-01 -5.53567052e-01 2.82903671e-01
-6.66588426e-01 -2.01168567e-01 -1.00908101e+00 -1.40203178e+00
-1.52330905e-01 -1.04721689e+00 4.97924596e-01 6.20360851e-01
2.79717207e-01 -8.24642420e-01 3.57220232e-01 3.86482358e-01
6.42114282e-01 4.54238713e-01 7.37063408e-01 -7.55435348e-01
-1.10242975e+00 2.28380617e-02 1.54533818e-01 6.13272935e-02
-1.55165911e-01 -1.66963208e-02 -1.68636397e-01 -3.56578022e-01
5.68242073e-01 3.87208574e-02 2.58670866e-01 7.06014276e-01
3.11791390e-01 -1.10132241e+00 -1.50105700e-01 4.54099685e-01
1.61663020e+00 8.07902694e-01 -2.56750733e-01 5.85987210e-01
1.96382683e-02 1.12687373e+00 7.64798701e-01 8.16145420e-01
2.84532726e-01 6.20855272e-01 5.11849701e-01 4.59000647e-01
1.04624009e+00 -3.50141436e-01 2.68582761e-01 2.45312855e-01
2.16524392e-01 -2.32892245e-01 -8.59042645e-01 3.73454720e-01
-1.88628757e+00 -7.76949763e-01 2.50553638e-01 2.41540432e+00
6.71296179e-01 3.65637898e-01 4.02371645e-01 -2.99542934e-01
8.75047326e-01 -2.28094876e-01 -8.13728213e-01 -7.33163834e-01
-1.52863309e-01 -2.80742705e-01 1.19820487e+00 4.49435979e-01
-5.64085960e-01 5.01392841e-01 6.24254274e+00 3.60730350e-01
-1.10192978e+00 -4.89464290e-02 1.15853357e+00 -6.21691585e-01
-2.83047169e-01 3.24391037e-01 -9.77297664e-01 8.48301411e-01
7.69710720e-01 -3.66474479e-01 7.91339993e-01 4.72815365e-01
6.39636517e-01 -2.60181993e-01 -9.90571797e-01 5.22308588e-01
-5.65707564e-01 -1.23144186e+00 -4.50504214e-01 6.90904558e-01
7.13055491e-01 -4.25402522e-01 3.34295183e-01 1.46503940e-01
3.97504836e-01 -5.92543483e-01 1.42185342e+00 4.70081300e-01
2.74627268e-01 -7.36917138e-01 5.88148057e-01 4.79238689e-01
-9.86325979e-01 -7.17344284e-01 1.01634264e-01 -2.57714808e-01
4.26449895e-01 3.51574630e-01 -2.03533471e-01 4.20991331e-01
4.71068084e-01 -1.54320106e-01 8.31290521e-03 8.95470917e-01
5.45131639e-02 4.81077909e-01 -6.44275725e-01 -2.32141227e-01
4.87849265e-01 -1.06647646e+00 6.21127903e-01 3.91948998e-01
1.18504450e-01 3.90640795e-01 1.79996878e-01 1.14997411e+00
2.01303676e-01 5.59674762e-03 -1.79292172e-01 -2.93830395e-01
3.70922565e-01 1.04830003e+00 -1.07230520e+00 1.56827390e-01
-2.54412413e-01 3.04365128e-01 -1.50130972e-01 3.62639546e-01
-7.01962113e-01 -2.16256320e-01 5.39104939e-01 8.40233624e-01
2.50990450e-01 -1.62710235e-01 -5.26429057e-01 -1.06151533e+00
4.31915760e-01 -7.62512922e-01 3.43125522e-01 -3.04029416e-02
-9.48863983e-01 -3.99419144e-02 1.91949189e-01 -8.40363681e-01
-4.13769454e-01 -2.36688823e-01 -7.13073254e-01 5.13853908e-01
-1.15703642e+00 -3.67726803e-01 5.98791659e-01 -2.24005263e-02
5.19095063e-01 8.95278826e-02 -2.64173388e-01 8.63969773e-02
-8.67450953e-01 1.46802083e-01 7.33154535e-01 3.70631590e-02
8.94858465e-02 -1.54794705e+00 -2.83501986e-02 7.11713731e-01
-2.97770441e-01 6.44600689e-01 9.48059797e-01 -8.74576032e-01
-1.66388285e+00 -4.47197616e-01 4.61686879e-01 -1.79909140e-01
1.14810157e+00 -7.42344141e-01 -6.00836217e-01 7.34126866e-01
2.32742354e-01 -3.95744711e-01 -2.24807523e-02 -2.72491097e-01
2.09630489e-01 -5.87761045e-01 -1.06905389e+00 3.16109926e-01
4.60527509e-01 1.24853216e-01 -2.66876847e-01 -1.20948479e-02
6.64079428e-01 -4.12762552e-01 -5.46937168e-01 1.06437393e-01
6.19002998e-01 -1.01709282e+00 5.22088826e-01 -2.45472908e-01
-2.10258827e-01 1.67293370e-01 2.41053298e-01 -8.08177412e-01
4.87814508e-02 -1.56963730e+00 -9.50582251e-02 1.38072526e+00
5.54997385e-01 -1.04398465e+00 6.80423498e-01 1.10812569e+00
3.45668018e-01 -1.04744887e+00 -1.23099470e+00 -9.38803971e-01
3.70357931e-01 2.55184323e-01 7.05001116e-01 4.79831308e-01
3.99315590e-03 -1.18150406e-01 -3.14743787e-01 8.52116570e-02
8.19377661e-01 6.86059654e-01 2.43698984e-01 -1.00127912e+00
-7.09361017e-01 -6.46327376e-01 3.45984817e-01 -8.47448885e-01
-3.01604837e-01 -1.50812075e-01 1.25049027e-02 -9.59979713e-01
1.13505185e-01 -6.96058035e-01 -2.73540795e-01 -8.28142241e-02
5.12867212e-01 -4.79636759e-01 5.89706361e-01 3.67690086e-01
-4.61713135e-01 2.13017404e-01 1.02391613e+00 1.74490362e-02
-7.89087832e-01 5.46829939e-01 -7.51595974e-01 3.37967336e-01
7.30733216e-01 -2.96415150e-01 -4.28912252e-01 -2.33189061e-01
6.44710779e-01 9.50289011e-01 1.23924978e-01 -2.60646433e-01
2.91247904e-01 -7.80113161e-01 -2.51393050e-01 -6.18357182e-01
4.05366831e-02 -7.42253542e-01 6.47165060e-01 7.19873428e-01
-4.37159985e-01 3.89446437e-01 -2.69591063e-01 6.04348958e-01
-9.67151970e-02 -3.92096221e-01 6.40379190e-01 -2.69624561e-01
4.12139684e-01 2.00419679e-01 -3.05594623e-01 2.47903228e-01
1.41089952e+00 -7.48993680e-02 -7.51838237e-02 -4.76318181e-01
-7.90050864e-01 9.30647254e-01 4.73954380e-01 2.35725135e-01
-1.19094737e-01 -6.92024887e-01 -3.63742352e-01 1.92370921e-04
-8.02854538e-01 1.50209814e-01 4.32234220e-02 8.95592511e-01
-7.64988065e-01 7.61611521e-01 1.20193595e-02 -6.64802641e-02
-5.74342847e-01 5.53375483e-01 7.79637814e-01 -6.64685786e-01
-4.16082680e-01 4.45599288e-01 4.53191429e-01 -3.94601077e-02
1.18045248e-01 -6.57253444e-01 1.22373052e-01 5.64450502e-01
1.92983896e-01 6.64068282e-01 -2.92344779e-01 -1.29969925e-01
-2.13991608e-02 4.40544099e-01 -7.87658542e-02 -5.51276267e-01
1.33492398e+00 -2.81517565e-01 -2.00432792e-01 6.10824525e-01
2.47463748e-01 3.94716300e-02 -1.66633463e+00 1.97373018e-01
3.39710146e-01 -2.78854370e-01 -4.03426915e-01 -7.39366353e-01
-1.18388557e+00 4.31882054e-01 -1.01843521e-01 8.84275198e-01
8.31251562e-01 -1.82903230e-01 5.80181360e-01 7.23676011e-02
6.29254520e-01 -1.43088162e+00 -2.25341931e-01 2.08038524e-01
7.65425205e-01 -5.09599626e-01 -1.98665515e-01 -8.75489190e-02
-6.39380336e-01 8.30162644e-01 5.02597928e-01 -4.12869543e-01
7.54815102e-01 5.87999642e-01 -9.51974615e-02 2.41961643e-01
-1.00906038e+00 3.48497182e-01 -1.78860202e-01 -2.61914015e-01
-1.96015984e-01 2.90860295e-01 -8.34527791e-01 1.13562930e+00
-2.11487010e-01 -3.60896409e-01 9.11837339e-01 9.16022301e-01
-3.53266358e-01 -1.07630527e+00 -5.94758987e-01 5.57196617e-01
-9.28382576e-01 2.99110264e-01 -4.74731386e-01 9.52288806e-01
1.46811092e-02 6.48747802e-01 3.05122614e-01 3.73326391e-01
4.20710772e-01 -8.44900608e-02 6.48134500e-02 -5.17513037e-01
-1.08259213e+00 6.62355185e-01 -3.14982653e-01 -3.70210141e-01
2.35128060e-01 -9.91193056e-01 -9.93975818e-01 -4.13837463e-01
-6.04208827e-01 5.11478603e-01 5.10127902e-01 7.26862431e-01
1.69185430e-01 -2.90334243e-02 9.54729021e-01 -7.14793503e-01
-1.36770344e+00 -4.80179787e-01 -1.17029142e+00 -2.79289156e-01
4.13138837e-01 -6.52446806e-01 -8.03836823e-01 -6.55747533e-01] | [4.8195271492004395, 3.9224586486816406] |
0121220e-2e5a-49be-b62d-fc29ecd90001 | stock-price-prediction-using-convolutional | 2001.09769 | null | https://arxiv.org/abs/2001.09769v1 | https://arxiv.org/pdf/2001.09769v1.pdf | Stock Price Prediction Using Convolutional Neural Networks on a Multivariate Timeseries | Prediction of future movement of stock prices has been a subject matter of many research work. In this work, we propose a hybrid approach for stock price prediction using machine learning and deep learning-based methods. We select the NIFTY 50 index values of the National Stock Exchange of India, over a period of four years, from January 2015 till December 2019. Based on the NIFTY data during the said period, we build various predictive models using machine learning approaches, and then use those models to predict the Close value of NIFTY 50 for the year 2019, with a forecast horizon of one week. For predicting the NIFTY index movement patterns, we use a number of classification methods, while for forecasting the actual Close values of NIFTY index, various regression models are built. We, then, augment our predictive power of the models by building a deep learning-based regression model using Convolutional Neural Network with a walk-forward validation. The CNN model is fine-tuned for its parameters so that the validation loss stabilizes with increasing number of iterations, and the training and validation accuracies converge. We exploit the power of CNN in forecasting the future NIFTY index values using three approaches which differ in number of variables used in forecasting, number of sub-models used in the overall models and, size of the input data for training the models. Extensive results are presented on various metrics for all classification and regression models. The results clearly indicate that CNN-based multivariate forecasting model is the most effective and accurate in predicting the movement of NIFTY index values with a weekly forecast horizon. | ['Sidra Mehtab', 'Jaydip Sen'] | 2020-01-10 | null | null | null | null | ['stock-price-prediction'] | ['time-series'] | [-6.22005105e-01 -3.54024470e-01 -2.91737944e-01 -3.81106198e-01
-2.01582983e-01 -4.67482001e-01 6.99919045e-01 -1.26810193e-01
-3.13797683e-01 9.52381849e-01 1.18547700e-01 -8.73278916e-01
-2.57410258e-01 -1.37705028e+00 -4.87728000e-01 -6.03562713e-01
-5.26740015e-01 4.16627973e-01 1.04565285e-01 -7.14692414e-01
6.35270476e-01 5.11093199e-01 -1.32368970e+00 5.68735925e-03
4.39263970e-01 1.77721775e+00 -2.94421166e-01 5.75059772e-01
-7.42004514e-02 1.07327044e+00 -5.45473218e-01 -3.96696061e-01
9.68860447e-01 -2.16791421e-01 -4.08584803e-01 -5.10863662e-01
-1.52279139e-01 -5.69195271e-01 -3.11652035e-01 7.02263057e-01
1.74808070e-01 2.09075175e-02 5.16989410e-01 -1.13664305e+00
-6.58057153e-01 9.71546292e-01 -4.98776406e-01 8.20099950e-01
-3.41799080e-01 4.73865792e-02 1.16524279e+00 -7.46042728e-01
1.80347607e-01 8.08286011e-01 1.12526405e+00 1.77412678e-03
-9.14571285e-01 -1.11197531e+00 4.52744775e-02 1.28376022e-01
-1.03534806e+00 5.68301156e-02 7.78031111e-01 -9.14580882e-01
1.01933527e+00 1.99003384e-01 1.01240015e+00 4.03729767e-01
7.40376115e-01 2.82619536e-01 1.11584198e+00 -2.42231637e-01
2.04567075e-01 6.53479397e-02 1.09459609e-01 1.55183509e-01
1.56135768e-01 5.80427587e-01 -1.58449262e-01 -2.98786253e-01
7.39269435e-01 3.23146403e-01 1.16091058e-01 5.30151188e-01
-1.01488686e+00 1.26856625e+00 4.26591486e-01 5.52976668e-01
-7.23702013e-01 9.96480044e-03 3.71055603e-01 8.73876333e-01
1.10502577e+00 6.12113535e-01 -1.18763709e+00 -5.49304821e-02
-1.54587173e+00 7.63929605e-01 9.32611167e-01 1.63590744e-01
5.49294829e-01 5.74287236e-01 -3.23435552e-02 4.71335232e-01
1.97088242e-01 3.24032873e-01 1.01565313e+00 -3.90894473e-01
3.80331308e-01 6.11216903e-01 2.51498520e-01 -1.12252831e+00
-6.34482503e-01 -7.64213502e-01 -7.81681597e-01 5.57630539e-01
3.73377264e-01 -5.86857975e-01 -7.71156788e-01 1.24381185e+00
-1.45972282e-01 2.83758104e-01 2.11686596e-01 4.27053094e-01
2.42887139e-01 8.67616296e-01 -1.16805591e-01 -2.58193910e-01
1.00376570e+00 -7.98875391e-01 -4.07721490e-01 2.72620887e-01
5.84149301e-01 -5.15688002e-01 2.87846327e-01 3.40828747e-01
-9.39032495e-01 -7.05520511e-01 -1.02465940e+00 5.65571427e-01
-8.79203796e-01 -9.40292981e-03 5.74382484e-01 2.74494410e-01
-9.45799112e-01 1.30337572e+00 -5.26453674e-01 4.18426186e-01
2.83257306e-01 3.52865636e-01 2.40707546e-01 7.67740488e-01
-1.76458836e+00 1.07833540e+00 6.19698226e-01 3.27905983e-01
-3.53308082e-01 -1.06694889e+00 -4.21524346e-01 1.98476449e-01
-5.35791636e-01 -3.02182704e-01 1.18262398e+00 -1.15617526e+00
-1.29716122e+00 3.37858975e-01 5.81481159e-01 -1.24552214e+00
8.87593031e-01 7.14198127e-02 -7.55227745e-01 -5.44494867e-01
-1.30421251e-01 2.14170784e-01 7.71319509e-01 -6.08103395e-01
-1.08019710e+00 -1.57295465e-01 -1.43323958e-01 -2.19740644e-01
-2.12082818e-01 1.96127091e-02 3.92456651e-01 -1.08127022e+00
-4.82162423e-02 -7.41598010e-01 -2.72300214e-01 -5.10493159e-01
-4.16789502e-02 -3.00897241e-01 8.25018048e-01 -9.70854521e-01
1.48485577e+00 -1.82767808e+00 -5.05287707e-01 5.41744649e-01
8.64869729e-02 1.67653546e-01 2.51474410e-01 4.32103544e-01
-4.34117436e-01 1.92810595e-01 -1.37326643e-01 -2.27458235e-02
-7.73690566e-02 5.95383085e-02 -8.96849215e-01 2.18802944e-01
1.89835563e-01 1.02387762e+00 -3.92551661e-01 2.64785975e-01
1.65639266e-01 3.00149679e-01 -1.12933896e-01 1.86596811e-01
-2.70344883e-01 3.49855989e-01 -3.97517383e-01 4.70432222e-01
8.99163485e-01 -2.05152497e-01 -3.01444262e-01 9.63035449e-02
-6.45633638e-01 2.30263487e-01 -8.60455871e-01 5.82731664e-01
-3.29396218e-01 6.55649602e-01 -7.02671111e-01 -1.19556487e+00
1.45542192e+00 3.03601235e-01 6.11161411e-01 -8.08521390e-01
1.26282319e-01 6.72710538e-01 2.47051448e-01 -9.42265168e-02
3.43287081e-01 -5.06120026e-01 1.65369660e-01 5.68155587e-01
-4.35502619e-01 2.08764657e-01 1.53702751e-01 -6.15535378e-01
6.77280605e-01 -3.74460816e-02 1.12824313e-01 -4.02968079e-01
5.18321753e-01 1.40551746e-01 4.52835828e-01 4.54522371e-01
-1.51703060e-02 2.70515859e-01 4.96251613e-01 -1.44363213e+00
-1.32765627e+00 -5.01711249e-01 -5.98721445e-01 9.34287190e-01
-6.36679173e-01 4.08749163e-01 -1.51670024e-01 -3.71779919e-01
5.65285087e-01 8.19980621e-01 -1.09576213e+00 1.22407809e-01
-5.54504514e-01 -1.24462712e+00 3.74489099e-01 5.00550747e-01
7.69860923e-01 -1.43652129e+00 -7.46037841e-01 5.15699983e-01
4.89686728e-01 -5.10771215e-01 2.48926133e-01 4.11567271e-01
-1.13597071e+00 -9.56271529e-01 -9.84846294e-01 -5.29616475e-01
1.03622742e-01 -3.65181744e-01 1.22575939e+00 3.03438663e-01
2.36088306e-01 -3.81141812e-01 -1.67774022e-01 -9.65301812e-01
-1.49545640e-01 1.13307796e-01 -6.75626993e-02 1.04078420e-01
4.06020969e-01 -6.12111032e-01 -7.65281975e-01 -3.85420019e-04
-6.82454050e-01 -2.11614400e-01 3.35685402e-01 7.20092237e-01
3.02507550e-01 3.94694567e-01 1.01378524e+00 -6.61321521e-01
7.68664837e-01 -9.64118242e-01 -1.09636831e+00 -4.06774208e-02
-1.20846939e+00 -1.30013436e-01 7.88879573e-01 -3.05677116e-01
-6.12742066e-01 -4.13839430e-01 -1.48493946e-01 -5.05826533e-01
4.85941440e-01 1.01752007e+00 9.50994015e-01 6.64576441e-02
2.57705897e-01 4.23656732e-01 -9.79854912e-03 -6.07663691e-01
-1.51799574e-01 3.66009802e-01 3.02185386e-01 6.64393529e-02
8.87725532e-01 2.11038262e-01 -4.11773287e-02 -2.31172696e-01
-8.77578199e-01 -9.52624828e-02 -6.47723198e-01 -2.33161017e-01
4.67168152e-01 -1.03211188e+00 -5.80134869e-01 9.14165676e-01
-7.02217758e-01 -2.96411276e-01 -2.97085613e-01 5.67688644e-01
-1.82859525e-01 -4.15077388e-01 -6.60253763e-01 -9.62622881e-01
-8.08980823e-01 -8.26971769e-01 3.90017867e-01 1.88237086e-01
-1.62325412e-01 -1.51106572e+00 5.15653789e-01 -2.34109506e-01
1.01487792e+00 6.86851084e-01 9.18314278e-01 -1.11116707e+00
-2.19879463e-01 -6.51212573e-01 -1.88189879e-01 5.01137614e-01
4.30196971e-02 7.13435039e-02 -8.06010485e-01 -2.33169064e-01
9.31397974e-02 -1.17380768e-01 9.91179168e-01 6.91184819e-01
1.06124687e+00 -6.45640016e-01 1.99724525e-01 7.42064238e-01
1.67999136e+00 7.25865066e-01 6.27685845e-01 1.27608800e+00
1.97974592e-01 3.24910820e-01 4.04326349e-01 7.58713424e-01
3.29644054e-01 1.90900937e-01 3.64593983e-01 -1.65821984e-01
5.85855186e-01 -2.61633210e-02 2.06221700e-01 7.32896209e-01
-4.22972947e-01 2.21359745e-01 -1.06317878e+00 3.91247809e-01
-1.69006741e+00 -1.19740212e+00 5.02628647e-02 2.00422096e+00
5.61853588e-01 4.06010389e-01 3.63763154e-01 4.16652828e-01
3.14764827e-01 3.48845035e-01 -7.38604784e-01 -5.16797423e-01
-1.70742407e-01 2.09508359e-01 9.71291900e-01 2.35963225e-01
-1.26084876e+00 6.24447048e-01 6.82956028e+00 2.40335733e-01
-1.70759070e+00 -2.87205845e-01 1.30397892e+00 1.02767624e-01
-1.65058568e-01 -1.83766082e-01 -9.09515083e-01 7.82101035e-01
1.44976211e+00 -4.83131886e-01 2.68725157e-01 1.00505340e+00
4.74553347e-01 1.86627477e-01 -6.35641456e-01 5.81085205e-01
-3.62897366e-01 -1.84893799e+00 5.55955060e-02 1.65800244e-01
9.27960157e-01 3.39299500e-01 3.05717736e-01 5.89561522e-01
3.55958045e-01 -1.05660820e+00 9.45682824e-01 1.01067960e+00
3.34466845e-01 -1.02274168e+00 1.31094301e+00 3.60326469e-01
-1.15257263e+00 -5.72830141e-01 -4.09077108e-01 -6.72569036e-01
-1.00130096e-01 7.30315804e-01 -4.54917163e-01 4.10080582e-01
1.03428817e+00 8.11776519e-01 -3.81101996e-01 8.71567130e-01
3.68756860e-01 7.47726560e-01 -2.33561635e-01 -1.49226293e-01
5.87948680e-01 -3.11309904e-01 -8.27797037e-03 9.23710763e-01
6.95749581e-01 1.63186099e-02 2.07268400e-03 7.70731091e-01
3.83040868e-02 1.58163682e-01 -5.10341465e-01 1.58653229e-01
3.09101492e-01 8.60833228e-01 -4.75541562e-01 -4.86914396e-01
-5.30638695e-01 2.02655345e-01 1.16999345e-02 1.78750649e-01
-6.72516286e-01 -3.61340225e-01 5.84433258e-01 2.33200222e-01
6.26277924e-01 -5.78059107e-02 -7.73564756e-01 -9.40311491e-01
-8.05686340e-02 -5.89124143e-01 4.06336486e-01 -6.35550797e-01
-1.31761074e+00 8.10271204e-01 -4.58718203e-02 -1.49622679e+00
-6.54325128e-01 -6.99074388e-01 -1.24062145e+00 1.21737719e+00
-2.07850790e+00 -6.89876556e-01 2.43698657e-01 4.46372658e-01
4.78187144e-01 -8.40038180e-01 6.17609560e-01 -4.61068377e-02
-4.94850338e-01 1.69790342e-01 6.61368251e-01 5.83735526e-01
-9.92967039e-02 -1.18904257e+00 8.18714082e-01 4.99287456e-01
-2.67958075e-01 3.55958104e-01 6.10009015e-01 -7.41994560e-01
-7.26562679e-01 -1.20520329e+00 1.18580377e+00 1.49750235e-02
1.34119534e+00 1.61647767e-01 -8.96624565e-01 8.30521941e-01
1.32486239e-01 6.82163686e-02 5.40662348e-01 -1.84517875e-01
-2.06809625e-01 -2.74529368e-01 -1.08365166e+00 8.61519128e-02
-1.25915170e-01 -3.71530838e-02 -6.50754154e-01 2.62104362e-01
5.23224354e-01 -7.70121589e-02 -1.20158005e+00 3.90567303e-01
8.82330656e-01 -1.04660928e+00 8.56968701e-01 -6.63340509e-01
5.43904543e-01 3.83329876e-02 -3.94682027e-02 -1.47846377e+00
-5.25992095e-01 -3.15600127e-01 -1.39696941e-01 8.02625954e-01
8.84921312e-01 -1.12324846e+00 8.06268871e-01 6.64253771e-01
2.77146518e-01 -1.15241396e+00 -9.91669178e-01 -7.28630185e-01
7.14668810e-01 -4.41254526e-01 1.18581319e+00 1.06280422e+00
-2.96122044e-01 -1.74442649e-01 -5.03004491e-01 -2.59921886e-02
3.21349144e-01 4.75780547e-01 4.71210718e-01 -1.63922143e+00
-1.09130688e-01 -7.23487258e-01 -4.19373482e-01 -3.66535097e-01
3.11042257e-02 -5.53650677e-01 -6.36289954e-01 -1.11132944e+00
-3.28507513e-01 -6.06379211e-01 -8.07836175e-01 6.09413803e-01
2.57591844e-01 1.17355213e-01 2.42560878e-01 7.97842324e-01
4.05221969e-01 3.96123439e-01 9.14480507e-01 -9.65001658e-02
-2.94665992e-01 5.52025259e-01 -3.62294734e-01 7.08945811e-01
1.22482324e+00 -1.94269449e-01 -6.45984039e-02 -1.54939750e-02
4.27627325e-01 1.64260760e-01 1.82801932e-01 -1.01509893e+00
-1.44176990e-01 -2.72635311e-01 1.00806069e+00 -8.90345097e-01
-1.19626932e-01 -7.40567267e-01 2.68372029e-01 8.59313548e-01
-4.02233362e-01 7.68891335e-01 2.99701184e-01 1.70134500e-01
-5.32498956e-01 -1.57015711e-01 7.31079519e-01 -3.27684641e-01
-5.34975588e-01 6.12099588e-01 -2.43140176e-01 -2.74189144e-01
1.07756376e+00 -2.06592888e-01 9.75099131e-02 -4.48653966e-01
-8.01205635e-01 2.56864667e-01 -1.81921542e-01 4.72207993e-01
3.97820234e-01 -1.47162914e+00 -1.00895488e+00 3.98601115e-01
-3.30472678e-01 -5.27956307e-01 -2.04942241e-01 6.70500755e-01
-8.70551050e-01 5.81540942e-01 -3.61789584e-01 -5.26353084e-02
-4.91039783e-01 3.99432749e-01 7.92481005e-01 -7.00656056e-01
-6.91249073e-01 5.79183102e-01 -2.20032513e-01 -2.89788276e-01
1.70556754e-02 -6.52426898e-01 -7.51506388e-01 5.57265699e-01
8.84301305e-01 3.75080615e-01 2.49535561e-01 -5.15727878e-01
1.29866362e-01 6.32904053e-01 -2.04308368e-02 5.05960695e-02
2.02727985e+00 1.61349162e-01 -1.07818618e-01 7.91329384e-01
1.21735418e+00 -6.32979393e-01 -1.21950471e+00 -4.92903292e-02
3.18056107e-01 -1.74605981e-01 2.90856063e-01 -7.99083352e-01
-1.64380550e+00 6.09876812e-01 6.78625524e-01 6.69012010e-01
1.06152248e+00 -6.56175792e-01 9.17682588e-01 1.73462525e-01
7.95749128e-02 -1.06805623e+00 -5.72562695e-01 7.55388081e-01
1.03711665e+00 -1.22103095e+00 -4.45729569e-02 4.60336596e-01
-4.44540828e-01 1.67126918e+00 1.65676415e-01 -6.92696691e-01
1.44123662e+00 1.82110161e-01 2.73077518e-01 -1.57620981e-01
-9.47894514e-01 2.42102042e-01 2.27113187e-01 1.07316673e-01
3.34959626e-01 1.94811016e-01 -3.14972758e-01 6.33968174e-01
-8.68812025e-01 2.41887882e-01 3.21208000e-01 7.11399376e-01
-5.80576837e-01 -5.80594003e-01 -4.40971971e-01 9.33307290e-01
-8.18605244e-01 -3.97191286e-01 5.24618151e-03 1.00881672e+00
-7.84545019e-02 5.24835944e-01 7.18048096e-01 -5.04867315e-01
2.56650060e-01 3.12357068e-01 -4.66682404e-01 1.13449521e-01
-1.03947401e+00 -1.04407363e-01 -2.01832399e-01 -1.59043595e-01
-1.82420820e-01 -7.82682002e-01 -9.75699842e-01 -6.92438483e-01
-1.45194039e-01 1.50419965e-01 5.79604268e-01 1.13888335e+00
-1.01053938e-01 2.75028497e-01 1.36942601e+00 -1.01488411e+00
-9.60886598e-01 -1.18417335e+00 -9.01050329e-01 1.94292337e-01
6.00562155e-01 -5.41691184e-01 -7.51758218e-01 -1.89028367e-01] | [4.458344459533691, 4.227615833282471] |
ef1c940f-7ddd-4df4-b882-aa5af70deac4 | idms-instance-depth-for-multi-scale-monocular | 2212.01528 | null | https://arxiv.org/abs/2212.01528v2 | https://arxiv.org/pdf/2212.01528v2.pdf | IDMS: Instance Depth for Multi-scale Monocular 3D Object Detection | Due to the lack of depth information of images and poor detection accuracy in monocular 3D object detection, we proposed the instance depth for multi-scale monocular 3D object detection method. Firstly, to enhance the model's processing ability for different scale targets, a multi-scale perception module based on dilated convolution is designed, and the depth features containing multi-scale information are re-refined from both spatial and channel directions considering the inconsistency between feature maps of different scales. Firstly, we designed a multi-scale perception module based on dilated convolution to enhance the model's processing ability for different scale targets. The depth features containing multi-scale information are re-refined from spatial and channel directions considering the inconsistency between feature maps of different scales. Secondly, so as to make the model obtain better 3D perception, this paper proposed to use the instance depth information as an auxiliary learning task to enhance the spatial depth feature of the 3D target and use the sparse instance depth to supervise the auxiliary task. Finally, by verifying the proposed algorithm on the KITTI test set and evaluation set, the experimental results show that compared with the baseline method, the proposed method improves by 5.27\% in AP40 in the car category, effectively improving the detection performance of the monocular 3D object detection algorithm. | ['Weijie Wu', 'Weibing Qiu', 'Liqiang Zhu', 'Chao Hu'] | 2022-12-03 | null | null | null | null | ['monocular-3d-object-detection', 'auxiliary-learning'] | ['computer-vision', 'methodology'] | [-1.71109438e-01 -5.13823748e-01 1.35382310e-01 -6.28169402e-02
-2.16015607e-01 -3.54371279e-01 4.14634466e-01 -3.44659656e-01
-7.54097223e-01 1.91921815e-01 -2.01881185e-01 -1.86637878e-01
2.56076723e-01 -7.23234117e-01 -4.72533494e-01 -7.79329062e-01
-6.07218109e-02 5.54595478e-02 1.04243267e+00 -6.30846061e-03
3.87138724e-01 8.91320229e-01 -1.72558594e+00 4.31089908e-01
5.56249559e-01 1.37552500e+00 9.28719759e-01 5.41346550e-01
2.54877228e-02 3.54205340e-01 -4.85241503e-01 2.04233274e-01
7.73093045e-01 -8.61931220e-03 -4.34035659e-02 3.98375362e-01
5.03669322e-01 -7.04487085e-01 -6.19017601e-01 1.40971816e+00
8.06506753e-01 -9.17219520e-02 5.73163211e-01 -1.04590082e+00
-3.94620836e-01 -1.41345114e-01 -9.16680515e-01 7.18394697e-01
1.02892101e-01 2.77055562e-01 3.20523292e-01 -1.38525093e+00
3.08295161e-01 1.65600932e+00 4.65804607e-01 1.25842795e-01
-5.60293555e-01 -8.52222919e-01 2.59697139e-01 6.44485712e-01
-1.47342956e+00 4.37006261e-03 7.20631123e-01 -3.34598750e-01
9.40993011e-01 -9.47852582e-02 5.44938266e-01 2.83025116e-01
3.95274132e-01 9.78642344e-01 1.25279808e+00 -2.75484622e-01
-1.51986212e-01 3.60128462e-01 3.60912085e-02 7.67547250e-01
3.74863148e-01 7.34947443e-01 -1.37810424e-01 3.74146134e-01
1.05861366e+00 1.61007762e-01 -1.53703138e-01 -4.96059984e-01
-1.19865155e+00 7.90502191e-01 9.63192046e-01 3.38784724e-01
-2.59627998e-01 -3.45026881e-01 1.34082213e-01 1.07816726e-01
4.18732166e-01 2.18629524e-01 -3.97119492e-01 4.27435309e-01
-3.63315582e-01 -7.98585359e-03 7.12584853e-02 1.09031200e+00
8.17761421e-01 4.55038920e-02 -2.04035848e-01 7.65723884e-01
3.31527084e-01 9.99040186e-01 5.42954326e-01 -5.07992208e-01
6.94662511e-01 9.53721225e-01 3.00310612e-01 -9.73751426e-01
-7.61499107e-01 -6.76574886e-01 -7.89624810e-01 6.64865255e-01
4.41787302e-01 -7.41727510e-03 -1.17197204e+00 1.13343108e+00
5.54982960e-01 -9.83086228e-02 2.77436357e-02 1.48329759e+00
1.04812181e+00 7.56649494e-01 -3.13632160e-01 -2.27192268e-01
1.57399166e+00 -7.04127729e-01 -2.79972672e-01 -4.35864419e-01
4.01730061e-01 -1.16409850e+00 6.39399529e-01 2.84288198e-01
-9.60452378e-01 -1.21725762e+00 -1.11965013e+00 1.37455106e-01
-5.10387301e-01 7.37362564e-01 5.58689475e-01 6.70625448e-01
-6.73668921e-01 -1.33625060e-01 -4.50688511e-01 -3.34798008e-01
3.44594806e-01 -2.84179728e-02 -3.23021054e-01 -5.50224185e-01
-1.10318291e+00 1.24514818e+00 8.10246408e-01 2.28071645e-01
-9.44537222e-01 -4.19860542e-01 -8.36584628e-01 -9.89452973e-02
3.93611878e-01 -4.72291052e-01 8.91871691e-01 -3.20400059e-01
-8.51499736e-01 7.23928392e-01 -2.69804001e-01 -1.36010513e-01
5.20377040e-01 1.27461597e-01 -3.62874955e-01 3.62484038e-01
3.18183422e-01 8.19411278e-01 9.42135274e-01 -1.14188135e+00
-1.65733516e+00 -7.57504523e-01 1.51537687e-01 8.30219030e-01
1.41250521e-01 -9.14913695e-03 -6.69325113e-01 -1.83385789e-01
7.36574888e-01 -6.54925108e-01 -8.31886902e-02 5.13540057e-04
-5.64362854e-02 -3.39938223e-01 1.18371356e+00 -5.47909796e-01
6.82866514e-01 -2.31434178e+00 3.36654000e-02 -1.80048555e-01
1.91063166e-01 2.89708406e-01 -1.30160481e-01 -3.12732995e-01
1.00691549e-01 -5.71150243e-01 4.89776656e-02 -1.41850248e-01
-3.73194635e-01 -6.38081506e-02 1.36277169e-01 7.00323105e-01
2.80877322e-01 7.04979956e-01 -6.87673748e-01 -4.60920662e-01
7.81732619e-01 2.94434756e-01 -3.14777017e-01 2.41305068e-01
2.92987645e-01 2.79470205e-01 -5.06610155e-01 9.74527061e-01
1.35654247e+00 2.28598833e-01 -7.21421063e-01 -5.58905363e-01
-5.01779675e-01 -2.30131820e-01 -1.63930595e+00 1.10992503e+00
-4.41720396e-01 5.64226151e-01 1.85226902e-01 -7.81549394e-01
1.17309010e+00 1.04064988e-02 1.85769036e-01 -1.00540173e+00
2.61500269e-01 3.53843451e-01 1.04609199e-01 -3.95517230e-01
2.35612616e-01 -7.56755993e-02 2.72256453e-02 -5.33397719e-02
-2.38992751e-01 -6.74220800e-01 -2.93551292e-02 -2.11626403e-02
7.08812058e-01 -2.88597167e-01 1.72886789e-01 2.86284108e-02
9.00531828e-01 1.52703434e-01 3.79390717e-01 8.33020449e-01
-3.56600612e-01 3.74419868e-01 -1.21861808e-01 -5.17062426e-01
-1.01811481e+00 -1.01956499e+00 -5.17896235e-01 5.91480851e-01
8.93947899e-01 3.83077741e-01 -1.70034811e-01 -7.36331522e-01
1.71709791e-01 3.46552402e-01 -5.01133919e-01 -2.46439725e-01
-3.46625865e-01 -9.90399539e-01 1.32706866e-01 5.86407959e-01
1.28012776e+00 -9.23825264e-01 -9.04978156e-01 2.79344976e-01
-5.32183237e-02 -1.44208384e+00 -3.66507441e-01 3.05966794e-01
-6.90637648e-01 -1.16695762e+00 -7.92042911e-01 -1.01081645e+00
6.50934279e-01 1.32952762e+00 3.60707313e-01 -2.29248300e-01
-6.76022410e-01 2.51436800e-01 -3.19462240e-01 -6.03650510e-01
-4.56112251e-03 -3.84910971e-01 1.98520124e-02 -6.68984726e-02
5.25061846e-01 4.42144237e-02 -6.46157682e-01 7.35906541e-01
-6.62446797e-01 -2.92503759e-02 1.06050062e+00 9.30779338e-01
3.89147699e-01 5.40350020e-01 3.39684874e-01 -9.45262909e-02
1.15093336e-01 -1.74952373e-02 -1.13741946e+00 -2.58185685e-01
-4.00296330e-01 -4.76546496e-01 2.89770812e-01 -5.22690773e-01
-1.29370487e+00 3.21324825e-01 1.51187837e-01 -4.23788249e-01
-2.45648354e-01 1.04691675e-02 -3.70006025e-01 -3.67068887e-01
6.06770575e-01 6.82295620e-01 5.77051542e-04 -3.95556331e-01
2.13026211e-01 8.09941590e-01 3.35042834e-01 2.49358058e-01
1.14644420e+00 6.21443748e-01 1.82467341e-01 -8.72313917e-01
-7.74343073e-01 -1.03161621e+00 -7.85373926e-01 -2.58434922e-01
1.06494892e+00 -1.48732388e+00 -5.53472340e-01 9.44217443e-01
-1.24098778e+00 1.52423427e-01 1.18417203e-01 1.00323725e+00
-2.12585241e-01 6.04534924e-01 -5.15918016e-01 -7.73411274e-01
-7.78664574e-02 -1.16800880e+00 1.32142591e+00 3.67562711e-01
8.14571261e-01 -5.65469384e-01 -6.29301846e-01 2.99629509e-01
2.78385967e-01 -3.17813218e-01 6.61652148e-01 -1.02176100e-01
-9.42202151e-01 -6.38216317e-01 -1.08380795e+00 3.86423230e-01
9.90561396e-02 -6.95628822e-01 -8.70788217e-01 -3.29709053e-01
3.12188864e-01 4.97135660e-03 9.76080000e-01 4.71537709e-01
7.52237380e-01 3.13162684e-01 -4.67420638e-01 6.58574939e-01
1.37562072e+00 5.71033657e-01 4.09287423e-01 4.37584400e-01
8.13868761e-01 4.66610253e-01 1.26569247e+00 3.88612628e-01
2.15585142e-01 7.39745617e-01 7.20487177e-01 -3.54873061e-01
-4.03182179e-01 -7.32476041e-02 1.73952013e-01 9.45341140e-02
9.58077535e-02 3.51200730e-01 -4.67228919e-01 5.50580025e-01
-1.50156975e+00 -9.75572765e-01 -2.92851061e-01 1.95358300e+00
2.80715436e-01 4.18682933e-01 1.06186926e-01 1.83336630e-01
8.38348150e-01 -4.25870419e-02 -6.49893522e-01 -1.17820660e-02
-4.90819871e-01 -3.23396623e-01 8.99278581e-01 4.23940033e-01
-1.26885104e+00 8.60432625e-01 4.84065437e+00 1.07392883e+00
-1.20255172e+00 5.28836139e-02 1.80635020e-01 8.12441483e-02
3.53482217e-01 -2.40748271e-01 -1.59227288e+00 3.54287803e-01
-1.06089108e-01 1.09554380e-01 1.77395329e-01 9.31506157e-01
2.44308189e-01 -5.33764362e-01 -7.02938557e-01 1.33860302e+00
2.87200987e-01 -7.48770297e-01 -1.67370647e-01 8.48404691e-02
6.85786128e-01 2.08442807e-01 2.52942126e-02 3.72922421e-01
-6.97199628e-02 -6.46269798e-01 6.82921231e-01 2.40072403e-02
4.25608605e-01 -7.02940941e-01 1.01460350e+00 7.98169494e-01
-1.62318456e+00 -5.44981956e-01 -1.00115967e+00 -1.69289067e-01
-3.05020493e-02 6.31464660e-01 -1.16937029e+00 4.83477801e-01
8.46025169e-01 7.16376066e-01 -9.55139816e-01 1.47951329e+00
-3.24282408e-01 -6.42628819e-02 -4.22000587e-01 -9.47354212e-02
3.53488564e-01 1.07779510e-01 7.33318269e-01 1.09758842e+00
4.87135351e-01 2.17558801e-01 4.20760483e-01 6.51953161e-01
5.48891425e-01 -1.19660467e-01 -5.91681182e-01 7.63465762e-01
5.06673276e-01 1.28478551e+00 -5.50139904e-01 -4.19189751e-01
-5.38815081e-01 8.06784451e-01 -2.00240627e-01 3.64735991e-01
-7.76188016e-01 -5.84985137e-01 2.52106100e-01 1.59347251e-01
7.23374248e-01 -3.94204289e-01 -2.82326758e-01 -9.97913480e-01
-4.45964262e-02 -5.11605859e-01 3.46280932e-01 -1.07821965e+00
-1.07708681e+00 5.39312601e-01 9.45847202e-03 -1.52554023e+00
1.13254100e-01 -9.70593631e-01 -4.58374888e-01 1.22382855e+00
-1.97257209e+00 -1.28166020e+00 -5.18255174e-01 7.97082961e-01
6.65567756e-01 -4.35083061e-01 1.34617895e-01 3.57677937e-01
-1.85182139e-01 4.39877093e-01 -1.90262213e-01 -1.16576046e-01
5.42253792e-01 -8.80291343e-01 2.85912063e-02 9.85705078e-01
-2.72995234e-01 -1.33829981e-01 3.12133610e-01 -5.50235271e-01
-1.33267426e+00 -1.17480516e+00 5.21381080e-01 -3.23040605e-01
2.99829066e-01 -2.42621824e-01 -7.25670576e-01 2.18670309e-01
-4.05178487e-01 7.47819543e-02 -1.93117157e-01 -3.87367994e-01
-2.14814305e-01 -1.89905345e-01 -1.18535089e+00 2.61969537e-01
1.03263831e+00 -4.99518067e-01 -7.98379004e-01 2.34806612e-01
5.82739353e-01 -5.93376219e-01 -4.26414043e-01 8.12257171e-01
3.33315939e-01 -9.14056003e-01 1.24415541e+00 1.89822733e-01
-7.04142675e-02 -9.19765174e-01 -4.21378464e-01 -1.19603789e+00
-5.62237144e-01 4.69072342e-01 3.10903132e-01 6.80933952e-01
5.51659428e-02 -8.40489566e-01 6.71355605e-01 -1.33513197e-01
-4.26071912e-01 -3.77395034e-01 -1.35525668e+00 -7.71955490e-01
-3.01106125e-01 -3.89971286e-01 2.72305012e-01 4.62005675e-01
-3.55281144e-01 3.31848621e-01 -1.08445965e-01 8.21167946e-01
8.44200969e-01 5.29075146e-01 9.72653687e-01 -1.11711109e+00
-2.58833021e-01 -3.25514793e-01 -7.62676954e-01 -1.77569830e+00
-3.52789730e-01 -6.24840617e-01 -2.34241108e-03 -1.49817371e+00
3.12654376e-01 -2.87568510e-01 -1.67826608e-01 3.97810005e-02
-3.53319645e-01 5.73110759e-01 3.43726784e-01 2.05175191e-01
-3.61981392e-01 5.28923810e-01 1.89904833e+00 -2.83833295e-01
-1.32168427e-01 3.08027118e-01 -4.04246300e-01 6.50847554e-01
4.41098392e-01 -1.03234187e-01 -2.14134336e-01 -2.82573521e-01
-4.12728786e-01 1.86180502e-01 7.47428536e-01 -1.23758256e+00
3.28535885e-01 -1.17314486e-02 1.14726067e+00 -1.52207601e+00
6.35760248e-01 -1.07521760e+00 -5.24849057e-01 8.96807134e-01
3.63302052e-01 -3.61431479e-01 5.37378371e-01 5.28660357e-01
-3.45088392e-01 -2.19043028e-02 1.14702415e+00 -2.24009335e-01
-1.28480685e+00 2.67611414e-01 -2.55805969e-01 -3.87570530e-01
1.19391096e+00 -6.54663086e-01 -3.61540645e-01 -5.73729314e-02
-6.34371996e-01 3.93998057e-01 2.16439962e-01 5.24762332e-01
1.01985049e+00 -1.39145768e+00 -8.02928209e-01 6.48108065e-01
3.40808570e-01 1.42311960e-01 4.36825037e-01 7.23841071e-01
-2.82507002e-01 7.68525839e-01 -4.36987102e-01 -1.02822161e+00
-1.36387825e+00 7.60155976e-01 6.50543869e-01 2.59903878e-01
-3.68252754e-01 9.26232755e-01 8.99615526e-01 -2.26416290e-01
1.91002145e-01 -4.90910143e-01 -3.92423004e-01 2.79840138e-02
6.00912631e-01 2.64325529e-01 -3.02277375e-02 -6.89287603e-01
-3.60304296e-01 9.88740861e-01 -4.28653583e-02 2.01801598e-01
9.89618361e-01 -6.12562358e-01 1.82886034e-01 2.13472143e-01
1.11337495e+00 -6.94805682e-02 -1.53545129e+00 -5.00764370e-01
-6.09531045e-01 -9.41372633e-01 3.35171163e-01 -8.07153106e-01
-8.43484104e-01 1.12913311e+00 1.26986110e+00 -7.66620710e-02
1.20542467e+00 5.58378659e-02 4.99419600e-01 1.93880022e-01
5.85598230e-01 -7.71013439e-01 2.59528518e-01 7.71139324e-01
7.41462648e-01 -1.62563324e+00 -1.82105193e-03 -6.90630257e-01
-4.62657362e-01 1.03788435e+00 9.43892479e-01 -1.16342671e-01
5.48655331e-01 -5.98237440e-02 1.38013372e-02 -1.15484215e-01
-1.79811522e-01 -5.98940432e-01 5.17103136e-01 9.04471219e-01
-2.59579301e-01 5.33348620e-02 1.27775492e-02 3.40410173e-01
1.80024087e-01 -3.78360391e-01 4.22828734e-01 4.63838875e-01
-1.06757784e+00 -5.65114915e-01 -8.40813160e-01 1.52321637e-01
1.43748835e-01 4.70108865e-03 -5.98044060e-02 8.81249666e-01
5.97534299e-01 1.14720345e+00 9.08742249e-02 -3.34894150e-01
7.25332499e-01 -4.11806524e-01 6.07105970e-01 -6.46721900e-01
3.14765982e-02 3.80906969e-01 -2.02043086e-01 -3.43706459e-01
-1.84873745e-01 -5.17807066e-01 -1.14175236e+00 2.82955356e-02
-6.79134607e-01 -1.84823364e-01 7.98990548e-01 7.61879683e-01
2.12520659e-02 3.88352245e-01 1.02223623e+00 -1.22545254e+00
-7.12911308e-01 -1.00677526e+00 -7.79602885e-01 8.48034248e-02
1.66625604e-01 -1.08579421e+00 -7.21376598e-01 -4.79103148e-01] | [7.993961811065674, -2.1741912364959717] |
1b650d7c-061f-41c3-9ee1-885f49ecd3bf | fedmix-mixed-supervised-federated-learning | 2205.01840 | null | https://arxiv.org/abs/2205.01840v1 | https://arxiv.org/pdf/2205.01840v1.pdf | FedMix: Mixed Supervised Federated Learning for Medical Image Segmentation | The purpose of federated learning is to enable multiple clients to jointly train a machine learning model without sharing data. However, the existing methods for training an image segmentation model have been based on an unrealistic assumption that the training set for each local client is annotated in a similar fashion and thus follows the same image supervision level. To relax this assumption, in this work, we propose a label-agnostic unified federated learning framework, named FedMix, for medical image segmentation based on mixed image labels. In FedMix, each client updates the federated model by integrating and effectively making use of all available labeled data ranging from strong pixel-level labels, weak bounding box labels, to weakest image-level class labels. Based on these local models, we further propose an adaptive weight assignment procedure across local clients, where each client learns an aggregation weight during the global model update. Compared to the existing methods, FedMix not only breaks through the constraint of a single level of image supervision, but also can dynamically adjust the aggregation weight of each local client, achieving rich yet discriminative feature representations. To evaluate its effectiveness, experiments have been carried out on two challenging medical image segmentation tasks, i.e., breast tumor segmentation and skin lesion segmentation. The results validate that our proposed FedMix outperforms the state-of-the-art method by a large margin. | ['Kwang-Ting Cheng', 'Xin Yang', 'Huimin Wu', 'Xijie Huang', 'Dong Zhang', 'Zengqiang Yan', 'Jeffry Wicaksana'] | 2022-05-04 | null | null | null | null | ['skin-lesion-segmentation'] | ['medical'] | [ 2.68647462e-01 3.29712331e-01 -7.67781198e-01 -6.96584821e-01
-8.92434299e-01 -4.10745323e-01 1.75884128e-01 8.87313336e-02
-3.67059231e-01 4.77514952e-01 -1.48801908e-01 -1.82941109e-01
-5.02304127e-03 -6.75696552e-01 -5.80884218e-01 -1.05939281e+00
2.72050679e-01 5.60418725e-01 2.63525933e-01 4.02648598e-01
-3.08029413e-01 4.68541533e-01 -1.26766109e+00 3.14913630e-01
8.80044401e-01 1.15750992e+00 1.39303014e-01 3.43609810e-01
-4.73592311e-01 8.83715391e-01 -3.21655989e-01 -4.49002981e-01
2.55800515e-01 -3.59356433e-01 -9.78044152e-01 5.22053301e-01
4.43690538e-01 -4.24196005e-01 -5.33103868e-02 1.15181899e+00
2.95164198e-01 -7.39947185e-02 2.25564331e-01 -1.23357880e+00
-3.05311590e-01 6.22851849e-01 -5.63824594e-01 -1.41741097e-01
-2.29395077e-01 1.72956645e-01 8.90807688e-01 -3.87122035e-01
7.70804346e-01 8.56270969e-01 5.17445743e-01 7.47099936e-01
-1.14496922e+00 -6.80258930e-01 5.74193060e-01 5.59955835e-03
-1.14870882e+00 -1.79008603e-01 8.03282678e-01 -2.02472374e-01
3.21064323e-01 3.42067391e-01 3.39215398e-01 6.35668159e-01
-5.17683476e-02 9.43282962e-01 1.26010299e+00 -3.75686049e-01
3.93811554e-01 2.45778918e-01 3.77519161e-01 8.77994001e-01
1.36656567e-01 -3.57548058e-01 -2.94951022e-01 -4.67451274e-01
5.01440048e-01 4.16775078e-01 -1.88749090e-01 -6.55122757e-01
-9.95955050e-01 7.20841706e-01 6.67971194e-01 4.13088739e-01
-3.15222055e-01 5.42968363e-02 6.09837055e-01 3.68924141e-02
5.18562555e-01 -3.65391165e-01 -5.85540771e-01 4.06964391e-01
-8.20653200e-01 -2.08958596e-01 7.60926664e-01 5.83061099e-01
1.28296185e+00 -5.10302603e-01 -3.52697849e-01 8.64375770e-01
4.84254390e-01 2.56372571e-01 7.99850583e-01 -1.02027798e+00
2.58391559e-01 9.59949970e-01 -1.73148274e-01 -7.37818420e-01
-3.06384772e-01 -5.43588758e-01 -8.58147800e-01 1.68987542e-01
2.35268310e-01 -1.92237943e-01 -1.08007264e+00 1.80253112e+00
9.38630521e-01 4.35669720e-01 -1.20284148e-02 8.87606561e-01
5.74798048e-01 2.76081264e-01 1.57234371e-01 -2.69889772e-01
1.31604552e+00 -1.42436683e+00 -7.42970228e-01 6.67036995e-02
8.11637878e-01 -4.39105362e-01 8.86434197e-01 2.93878734e-01
-8.00856411e-01 -3.11097145e-01 -8.66638362e-01 2.00702876e-01
-1.38532192e-01 3.15969102e-02 8.54731143e-01 6.80763483e-01
-8.72807562e-01 4.00854558e-01 -1.07753396e+00 -1.90625787e-01
8.42764258e-01 5.01493633e-01 -5.36768913e-01 -3.24415922e-01
-5.93113482e-01 3.75187665e-01 2.86320925e-01 -1.15962505e-01
-9.43602860e-01 -5.59103668e-01 -7.08148062e-01 2.50684414e-02
3.92662674e-01 -5.97840786e-01 1.54913604e+00 -1.40790939e+00
-1.43027818e+00 1.00609946e+00 -1.08570769e-01 -3.67177308e-01
4.53984618e-01 2.33476445e-01 -3.05886537e-01 2.89363861e-01
1.71566740e-01 6.36205077e-01 6.97747827e-01 -1.49642766e+00
-9.53072965e-01 -6.42165899e-01 -1.11582791e-02 1.27889991e-01
-8.26316774e-01 -1.59488961e-01 -7.69490957e-01 -3.21082562e-01
2.14580595e-01 -8.02650869e-01 -5.67165256e-01 3.29403877e-01
-3.33163917e-01 -2.33143091e-01 1.10559011e+00 -3.27225506e-01
1.10822856e+00 -2.19496775e+00 -1.56305060e-01 4.64781582e-01
3.54559451e-01 2.48040140e-01 -8.52321014e-02 -4.27885763e-02
2.19931722e-01 -8.59584734e-02 -4.76066351e-01 -7.28535116e-01
-9.09816995e-02 6.02892101e-01 2.07803510e-02 6.35133684e-01
-2.95143992e-01 6.80159211e-01 -8.08515072e-01 -1.09921551e+00
1.08667709e-01 3.96811545e-01 -4.63338792e-01 2.41126627e-01
-3.72281343e-01 6.23263896e-01 -8.51880491e-01 8.69941473e-01
7.98618376e-01 -8.02019835e-01 5.50263762e-01 -1.97218031e-01
7.79764652e-02 -2.26271957e-01 -1.16663229e+00 1.95161188e+00
-5.20377517e-01 -1.55633613e-01 5.65312207e-01 -1.11323261e+00
7.38698781e-01 5.91645718e-01 9.85607028e-01 -5.20958066e-01
1.81885466e-01 2.86958009e-01 -3.10499221e-01 -3.68052602e-01
-1.52685747e-01 -4.08505797e-02 3.40000391e-02 7.41418719e-01
1.89934030e-01 4.50414270e-01 -6.08409569e-02 1.84425667e-01
1.10733855e+00 4.22034115e-02 8.76542553e-02 6.31803349e-02
8.40734959e-01 4.84194309e-02 9.13299561e-01 5.84615290e-01
-4.53339070e-01 4.08717781e-01 2.05836073e-01 -5.39098859e-01
-5.87272108e-01 -8.29927981e-01 -2.57409573e-01 1.30996001e+00
3.49319130e-01 -2.00166106e-01 -1.05879581e+00 -1.34643006e+00
3.45349945e-02 1.09498128e-01 -6.27478123e-01 7.21456409e-02
-5.23750126e-01 -7.53962457e-01 4.65624422e-01 4.21082735e-01
6.51366234e-01 -1.01701152e+00 -6.22483909e-01 2.54558265e-01
-4.98199388e-02 -1.00249052e+00 -6.59348667e-01 2.69169569e-01
-9.28998411e-01 -1.17668247e+00 -6.79653645e-01 -9.55070376e-01
1.12691021e+00 2.91730046e-01 7.87626922e-01 3.25772077e-01
-3.71685505e-01 4.20851380e-01 -2.55954593e-01 -5.81027567e-02
-3.37246895e-01 1.69558018e-01 -4.33812857e-01 6.93609357e-01
1.61177665e-02 -3.63356084e-01 -7.80797124e-01 4.57854956e-01
-1.21178770e+00 1.20495513e-01 6.57995462e-01 9.99969244e-01
9.76244092e-01 -2.17683315e-01 5.58794677e-01 -1.45462310e+00
9.24139246e-02 -6.12903714e-01 -5.34915566e-01 4.78430033e-01
-7.03588605e-01 -2.18635142e-01 7.24906862e-01 -4.13349777e-01
-1.37396252e+00 5.33185124e-01 -5.11276834e-02 -5.90519011e-01
-3.10743749e-01 4.30958092e-01 -3.58526856e-01 -2.88745880e-01
3.07229429e-01 -5.11095040e-02 2.45823130e-01 -6.37848973e-01
4.55490828e-01 8.76034796e-01 6.16710544e-01 -5.89305222e-01
4.70941305e-01 9.35439646e-01 -2.84280896e-01 -1.90354995e-02
-9.58181918e-01 -6.92919910e-01 -4.30155367e-01 -2.33924314e-01
8.43850613e-01 -7.65134931e-01 -6.90943778e-01 5.30132592e-01
-7.99251735e-01 -4.28385228e-01 -5.17919183e-01 3.81736368e-01
-4.28472251e-01 3.00852716e-01 -7.26895392e-01 -3.58255059e-01
-6.20496213e-01 -1.29701817e+00 9.25452828e-01 4.26410824e-01
1.42885908e-01 -1.08150864e+00 1.71734676e-01 7.13477910e-01
4.50030416e-01 4.08551604e-01 7.73335874e-01 -8.05417776e-01
-5.61394989e-01 -2.17532754e-01 -2.17749134e-01 4.28638637e-01
4.80948448e-01 -2.43522763e-01 -1.14899170e+00 -4.06996757e-01
3.47362384e-02 -5.12137115e-01 6.95653439e-01 9.20496285e-02
1.45784104e+00 -4.32644576e-01 -6.70656443e-01 7.40139604e-01
1.49271619e+00 -1.50946677e-01 1.14612408e-01 2.87156731e-01
6.92033768e-01 4.82169390e-01 4.79220331e-01 5.23297250e-01
4.65618223e-01 4.62532222e-01 7.79597044e-01 -5.66345453e-01
-1.42607465e-01 3.29610705e-02 2.68795658e-02 7.08216965e-01
4.61057156e-01 -2.41374206e-02 -7.19112396e-01 5.23492754e-01
-2.14641523e+00 -4.85606462e-01 2.52491564e-01 2.12695813e+00
1.02084804e+00 -1.77907020e-01 -1.44534903e-02 -1.04801312e-01
8.52617741e-01 7.87542239e-02 -1.00568128e+00 -2.44688854e-01
6.49497211e-02 9.85289961e-02 6.01634860e-01 9.50644165e-02
-1.13904881e+00 6.82207406e-01 5.62185240e+00 7.84164250e-01
-1.42466629e+00 6.30914271e-01 9.30348158e-01 -4.94413152e-02
-2.76470184e-01 9.33830738e-02 -5.04637122e-01 4.92349684e-01
5.99867761e-01 1.30347069e-02 2.10163936e-01 1.09634411e+00
-1.79688796e-01 1.99134290e-01 -8.57184410e-01 7.53990412e-01
3.93673033e-02 -1.41212559e+00 -1.72059946e-02 2.60845631e-01
9.85304415e-01 3.49106222e-01 -1.17007598e-01 6.23123497e-02
5.98702133e-01 -7.35229909e-01 3.75180542e-01 4.07502502e-01
7.62365580e-01 -7.09852338e-01 6.65923953e-01 5.80632508e-01
-1.11110938e+00 -2.10850462e-01 3.94141376e-02 4.76511568e-01
-7.58824572e-02 5.41244745e-01 -5.87998807e-01 7.27788389e-01
7.36059904e-01 4.83374417e-01 -3.72468174e-01 9.76176322e-01
1.07673198e-01 6.88845158e-01 -2.22918510e-01 5.26907206e-01
3.96284819e-01 -6.16498198e-03 -3.76678482e-02 1.06946468e+00
-8.44172761e-02 8.43198691e-03 6.66697621e-01 5.33964753e-01
-3.85396898e-01 3.58809233e-01 -6.35057315e-02 4.27150667e-01
3.47917110e-01 1.79696798e+00 -7.98484802e-01 -4.42572534e-01
-6.36862278e-01 8.99945974e-01 4.66479421e-01 2.39186198e-01
-7.52321661e-01 1.20019140e-02 4.69415605e-01 -1.61182463e-01
3.48177016e-01 3.25701028e-01 -1.42948851e-01 -1.01115870e+00
6.12582304e-02 -8.35129559e-01 9.02030647e-01 -8.29033181e-02
-1.48534381e+00 5.73482633e-01 -1.72512144e-01 -1.09235287e+00
2.38469280e-02 -2.32826918e-01 -8.50231469e-01 5.97435474e-01
-1.74148762e+00 -1.60627007e+00 -5.40812016e-01 9.88483191e-01
2.03641593e-01 -3.19145992e-02 9.29217756e-01 4.37217653e-01
-8.52052331e-01 6.92400932e-01 1.78250968e-01 1.74462378e-01
7.87918866e-01 -1.02980328e+00 -3.48084629e-01 4.52847868e-01
-6.03053980e-02 3.76621902e-01 2.83241719e-02 -4.84561801e-01
-1.34524965e+00 -1.36689186e+00 4.50607568e-01 1.72746181e-01
3.58780414e-01 -1.11938559e-01 -1.01086020e+00 7.85075784e-01
2.70822406e-01 9.00893509e-01 9.97569561e-01 -4.24422920e-02
-3.06634516e-01 -5.90919852e-01 -1.64976370e+00 1.87814578e-01
6.56118631e-01 -1.67851150e-01 -2.80903094e-02 7.60527194e-01
7.56964505e-01 -4.82754648e-01 -1.03340352e+00 4.71525759e-01
3.40450823e-01 -7.73414552e-01 5.54414392e-01 -4.47695732e-01
-1.04100913e-01 -3.41645420e-01 -8.16057324e-02 -8.61191869e-01
-1.97729275e-01 -4.17691886e-01 -1.38863981e-01 1.39905691e+00
8.77700523e-02 -9.05390680e-01 1.26724076e+00 7.26192236e-01
-2.33988717e-01 -1.05957007e+00 -1.26846969e+00 -4.90842372e-01
-1.52564883e-01 -3.05278718e-01 8.53616774e-01 9.77723837e-01
-1.90148264e-01 -1.10867605e-01 -1.75629382e-03 3.22870970e-01
8.42782438e-01 3.33463460e-01 6.62410975e-01 -1.04010415e+00
-4.48584199e-01 -3.37533325e-01 -2.24156052e-01 -6.96632504e-01
3.98022890e-01 -1.08557212e+00 -6.46415204e-02 -1.47675788e+00
4.89110172e-01 -9.96404767e-01 -8.21616113e-01 1.13239026e+00
-7.17536062e-02 3.22524786e-01 1.41849175e-01 5.09040773e-01
-9.93672967e-01 2.08544761e-01 1.28811646e+00 -3.60756189e-01
1.15629718e-01 1.44431531e-01 -7.01876044e-01 7.43652225e-01
7.43905902e-01 -5.25796235e-01 -4.41773504e-01 -4.91485983e-01
-4.36375380e-01 -2.47173738e-02 2.80920088e-01 -8.04531991e-01
5.46932220e-01 -2.40694433e-01 2.40929261e-01 -3.68258923e-01
4.03445261e-03 -1.16090524e+00 9.45492461e-02 4.94163752e-01
-3.09561223e-01 -4.92184907e-01 -2.37500053e-02 5.68084896e-01
-2.67657518e-01 -2.57249117e-01 8.78803015e-01 -2.13203102e-01
-5.76597869e-01 6.49851084e-01 4.68388461e-02 -2.13306651e-01
1.33080494e+00 -9.53757912e-02 -2.01805875e-01 1.29863322e-01
-6.94950640e-01 5.54009855e-01 6.19120479e-01 7.02566430e-02
3.55719566e-01 -1.16631150e+00 -4.98907536e-01 2.84514308e-01
2.17167720e-01 3.01947027e-01 4.96227264e-01 8.05818737e-01
-2.49840513e-01 1.35553122e-01 -1.85538586e-02 -7.92917013e-01
-1.29334080e+00 6.82650328e-01 5.47800064e-01 -5.67118704e-01
-6.79478347e-01 6.41480207e-01 3.30845594e-01 -8.74099612e-01
5.23580670e-01 -3.74658294e-02 -3.09202839e-02 -2.30124950e-01
4.58744735e-01 1.76789567e-01 3.63102257e-01 -6.46992385e-01
-3.99640411e-01 4.84824568e-01 -3.38683754e-01 1.98050380e-01
1.14835429e+00 -9.06770602e-02 -3.02708209e-01 1.66521624e-01
1.32832730e+00 -1.41066298e-01 -1.40732026e+00 -6.66500866e-01
-1.74412757e-01 -3.95778388e-01 2.09358633e-01 -8.17787945e-01
-1.75920582e+00 4.06146735e-01 9.06369686e-01 -1.91117764e-01
1.32873034e+00 6.33442551e-02 9.61764276e-01 8.93422961e-02
4.12406087e-01 -9.85236466e-01 8.93946085e-03 -6.86645806e-02
1.64991856e-01 -1.29116726e+00 -7.36853257e-02 -3.41772556e-01
-5.37450910e-01 8.41172457e-01 8.28140378e-01 1.23334378e-01
5.39979160e-01 3.96546632e-01 4.16599393e-01 -1.16320357e-01
-7.43789613e-01 2.27851383e-02 1.76420249e-02 2.60286003e-01
1.64872631e-01 1.02406561e-01 -5.71256392e-02 5.75056255e-01
5.78265011e-01 2.65227139e-01 -5.17798066e-02 1.10843730e+00
-3.33355159e-01 -1.52242923e+00 -5.62811911e-01 4.67731237e-01
-6.13977909e-01 4.17733818e-01 -1.19102001e-01 4.40614194e-01
5.23132443e-01 7.88769782e-01 2.25999318e-02 -1.56712439e-02
5.36436066e-02 -2.79372912e-02 2.15616107e-01 -6.18821859e-01
-8.34366381e-01 2.38259345e-01 -4.23346847e-01 -6.08451426e-01
-5.66941679e-01 -4.30698186e-01 -1.81738997e+00 9.62291434e-02
-5.40847480e-01 2.37249032e-01 8.11248600e-01 8.88953924e-01
3.00100058e-01 4.58061337e-01 9.87207174e-01 -6.09106362e-01
-7.51830578e-01 -5.31908810e-01 -5.89999616e-01 5.88893712e-01
3.29588592e-01 -3.17000985e-01 -1.54055774e-01 2.33453125e-01] | [6.0248589515686035, 6.442208290100098] |
71428bbe-b36c-4d38-b33b-82061ad1282b | multi-perspective-semantic-information | 2008.01526 | null | https://arxiv.org/abs/2008.01526v1 | https://arxiv.org/pdf/2008.01526v1.pdf | Multi-Perspective Semantic Information Retrieval in the Biomedical Domain | Information Retrieval (IR) is the task of obtaining pieces of data (such as documents) that are relevant to a particular query or need from a large repository of information. IR is a valuable component of several downstream Natural Language Processing (NLP) tasks. Practically, IR is at the heart of many widely-used technologies like search engines. While probabilistic ranking functions like the Okapi BM25 function have been utilized in IR systems since the 1970's, modern neural approaches pose certain advantages compared to their classical counterparts. In particular, the release of BERT (Bidirectional Encoder Representations from Transformers) has had a significant impact in the NLP community by demonstrating how the use of a Masked Language Model trained on a large corpus of data can improve a variety of downstream NLP tasks, including sentence classification and passage re-ranking. IR Systems are also important in the biomedical and clinical domains. Given the increasing amount of scientific literature across biomedical domain, the ability find answers to specific clinical queries from a repository of millions of articles is a matter of practical value to medical professionals. Moreover, there are domain-specific challenges present, including handling clinical jargon and evaluating the similarity or relatedness of various medical symptoms when determining the relevance between a query and a sentence. This work presents contributions to several aspects of the Biomedical Semantic Information Retrieval domain. First, it introduces Multi-Perspective Sentence Relevance, a novel methodology of utilizing BERT-based models for contextual IR. The system is evaluated using the BioASQ Biomedical IR Challenge. Finally, practical contributions in the form of a live IR system for medics and a proposed challenge on the Living Systematic Review clinical task are provided. | ['Samarth Rawal'] | 2020-07-17 | null | null | null | null | ['passage-re-ranking'] | ['natural-language-processing'] | [ 6.65799379e-01 2.01444164e-01 -4.61366773e-01 -4.18000847e-01
-1.56684077e+00 -3.44478935e-01 3.93926173e-01 9.41133082e-01
-7.02100158e-01 7.52693117e-01 7.87743092e-01 -2.59147078e-01
-5.76552212e-01 -4.25545126e-01 -3.09549540e-01 -4.57812607e-01
8.94831419e-02 7.45641470e-01 -7.82016292e-02 -6.22135162e-01
5.48001468e-01 4.47166383e-01 -1.14812362e+00 8.83778036e-01
4.83962268e-01 9.39284205e-01 3.86167675e-01 5.58118522e-01
-1.49241537e-02 8.06015909e-01 -5.92618465e-01 -3.08784872e-01
-3.68447125e-01 -2.84989864e-01 -1.41368699e+00 -7.57918894e-01
-6.83820918e-02 1.68494303e-02 3.29358503e-02 7.59587348e-01
1.17464471e+00 1.99450269e-01 4.93388087e-01 -4.09612507e-01
-7.35580564e-01 5.96320987e-01 -1.98343083e-01 6.36496663e-01
6.28721595e-01 -3.84139180e-01 1.21664798e+00 -7.09397733e-01
8.46103847e-01 1.17785335e+00 3.85182738e-01 7.39825845e-01
-9.43318367e-01 -1.20607719e-01 -4.17798132e-01 2.43330747e-01
-1.10068989e+00 -4.95274782e-01 5.04611671e-01 -2.20823020e-01
1.34402621e+00 4.74498808e-01 -4.54494320e-02 1.08997011e+00
4.87152576e-01 7.81367600e-01 6.13085866e-01 -7.82412648e-01
1.79649502e-01 3.21774423e-01 3.75728577e-01 1.55785888e-01
-1.04874246e-01 -1.93907514e-01 -4.35587674e-01 -6.11322343e-01
-5.36419712e-02 -9.53175873e-02 -3.70586276e-01 4.34490442e-01
-1.12025928e+00 9.64626253e-01 3.75577718e-01 7.26482868e-01
-4.85154182e-01 -6.56434447e-02 8.04589391e-01 2.80863881e-01
6.16765976e-01 7.80023932e-01 -6.12528503e-01 -4.16519195e-02
-6.80819452e-01 6.66172430e-02 6.67522132e-01 5.60249388e-01
-4.78047431e-02 -7.40992188e-01 -5.34889400e-01 1.17075372e+00
3.27300280e-01 2.31971741e-01 9.97531414e-01 -6.19056404e-01
5.20530164e-01 4.46825355e-01 -1.33014753e-01 -1.03804266e+00
-4.97815818e-01 -3.88447404e-01 -7.08081186e-01 -5.32247961e-01
-8.79031569e-02 2.24439442e-01 -5.71647048e-01 1.62773788e+00
2.48933688e-01 -4.24913228e-01 5.96306264e-01 7.43903458e-01
1.30183411e+00 6.01861656e-01 4.34136331e-01 -2.71279454e-01
1.87364054e+00 -5.70784807e-01 -8.66106868e-01 -3.90710622e-01
8.42619479e-01 -1.07730007e+00 6.38992608e-01 9.89144146e-02
-1.11378813e+00 -2.47862428e-01 -6.78808153e-01 -6.48195744e-01
-5.09936690e-01 2.13062689e-02 2.64444143e-01 1.84922829e-01
-1.25088584e+00 5.90478957e-01 -3.79169315e-01 -5.77759147e-01
3.04411650e-01 3.28707904e-01 -3.51072192e-01 -3.65741968e-01
-1.69514441e+00 1.39909482e+00 2.56433070e-01 1.92971900e-01
-3.52351338e-01 -7.65965760e-01 -7.01387882e-01 1.15946352e-01
-1.06860861e-01 -9.48060751e-01 1.34642756e+00 -6.22774720e-01
-1.02348459e+00 1.15011537e+00 -2.76885003e-01 -4.81510997e-01
7.14843050e-02 -1.98235005e-01 -5.22486269e-01 5.93997598e-01
2.22802162e-01 6.61162794e-01 3.01627547e-01 -3.56371999e-01
-2.79311150e-01 -5.18779695e-01 -8.19825530e-02 4.95794803e-01
-2.70416528e-01 6.54361129e-01 -2.53751904e-01 -4.78392363e-01
-2.77661622e-01 -7.07357347e-01 -4.28835869e-01 -3.23526829e-01
-3.90300751e-01 -6.24696374e-01 1.79839432e-01 -8.21520746e-01
1.24017978e+00 -2.00565958e+00 -2.32634787e-02 -1.48867875e-01
-7.91458189e-02 2.32094854e-01 -2.08188802e-01 7.62387633e-01
-4.05506343e-01 4.01456743e-01 -5.73527403e-02 1.08993361e-02
-3.48362297e-01 -1.70973092e-01 -3.90937507e-01 1.04234181e-01
2.84968823e-01 1.02223063e+00 -1.07138085e+00 -7.35953212e-01
-1.64952397e-01 7.76439428e-01 -2.83038765e-01 4.73449454e-02
-4.98276800e-02 2.63752848e-01 -6.68273091e-01 3.71985614e-01
-1.13593571e-01 -5.84119320e-01 9.12781805e-02 -1.95068315e-01
3.76268059e-01 7.51427710e-01 -2.91000783e-01 1.75488544e+00
-4.85385090e-01 3.89048427e-01 -4.01451141e-01 -1.10294938e+00
7.06089377e-01 9.59120452e-01 7.71459699e-01 -8.28473568e-01
8.96403417e-02 2.40145236e-01 -1.72031134e-01 -7.67308414e-01
6.12160563e-01 -5.00062227e-01 -1.38731882e-01 4.48151559e-01
-1.65309459e-01 -3.62802073e-02 2.64525890e-01 3.23747098e-01
1.37375963e+00 -1.21676624e-01 7.42023051e-01 -2.25873902e-01
6.76986575e-01 3.07422757e-01 2.12323636e-01 5.65503001e-01
-1.69023145e-02 7.70555973e-01 3.17752212e-01 -3.60629231e-01
-6.28806353e-01 -6.11692786e-01 -6.41682386e-01 9.06161487e-01
-2.45419309e-01 -2.14954600e-01 -3.89525861e-01 -4.60898727e-01
-2.36524895e-01 7.82522440e-01 -6.96754217e-01 -5.70845544e-01
-4.56750184e-01 -9.32152569e-01 6.29082859e-01 2.40537763e-01
-2.10874915e-01 -1.63423324e+00 -6.36489272e-01 3.92063498e-01
-6.64668977e-01 -1.10189795e+00 -4.50142682e-01 1.91203505e-01
-1.10565376e+00 -1.07238126e+00 -9.55176234e-01 -1.01228166e+00
5.17473221e-01 9.18073878e-02 1.36613858e+00 4.76617552e-02
-7.94899642e-01 5.78545392e-01 -5.26153266e-01 -5.64932942e-01
-7.67893493e-01 9.63034332e-02 -2.68624127e-01 -5.76604605e-01
8.35859954e-01 9.08093899e-03 -7.83671200e-01 -9.66469869e-02
-1.16874194e+00 -2.25048423e-01 6.04863226e-01 9.53490257e-01
6.59642696e-01 -3.51651132e-01 1.15894020e+00 -8.94895971e-01
1.26334572e+00 -7.51236081e-01 7.39612989e-03 5.28473854e-01
-6.73806071e-01 1.94379494e-01 3.53793889e-01 -1.40728176e-01
-9.58616972e-01 -3.74803692e-01 -4.89163280e-01 2.25187212e-01
-4.46561202e-02 1.07286298e+00 3.96934360e-01 5.23862362e-01
1.05101013e+00 -2.42879391e-02 1.54025167e-01 -4.97115105e-01
2.42993787e-01 1.04387307e+00 1.13142796e-01 -3.53269219e-01
-1.45866781e-01 1.51842877e-01 -4.69213501e-02 -7.60647833e-01
-1.12477720e+00 -1.03913593e+00 -2.30807692e-01 8.33076984e-02
9.61563826e-01 -7.01229095e-01 -5.86326182e-01 -3.99708122e-01
-1.30015051e+00 2.34524131e-01 -4.15132642e-01 5.77789009e-01
-3.85014325e-01 3.74182969e-01 -8.99442255e-01 -4.89946723e-01
-1.12377095e+00 -1.26727688e+00 1.31791401e+00 6.99594468e-02
-6.16838515e-01 -1.04501903e+00 4.54257578e-01 6.91538632e-01
3.46678764e-01 -1.89654633e-01 1.37966681e+00 -9.79723334e-01
2.41050884e-01 -5.60156345e-01 -1.06135741e-01 3.09491754e-01
2.60536194e-01 -5.01001477e-01 -8.97031069e-01 -8.73672366e-02
1.52032375e-01 -5.00388026e-01 7.32530594e-01 4.40427125e-01
8.72294426e-01 -2.35193729e-01 -4.04961109e-01 -2.58087069e-01
1.32664478e+00 3.16747844e-01 6.34329259e-01 3.51385921e-01
1.17820442e-01 9.81908202e-01 6.51602745e-01 8.08168650e-02
1.60709605e-01 6.75179362e-01 -3.97594459e-02 -2.54000574e-02
-1.31866619e-01 6.12009503e-02 2.08381623e-01 8.57775748e-01
2.33669713e-01 -2.21681118e-01 -9.32000339e-01 6.02227509e-01
-1.61493981e+00 -9.89577651e-01 1.75588742e-01 2.20486999e+00
1.37765276e+00 -2.52461702e-01 -4.87475455e-01 -3.13951448e-02
5.79623938e-01 -3.24136853e-01 -3.87916297e-01 -6.65804148e-01
3.49070989e-02 3.55893880e-01 2.31213346e-02 3.61181259e-01
-8.39193583e-01 5.66623747e-01 5.98191118e+00 8.00884664e-01
-9.46598709e-01 5.79497330e-02 8.95642340e-01 3.50904502e-02
-2.14451432e-01 -3.89080882e-01 -9.64988947e-01 2.63267994e-01
1.36192763e+00 -3.93987507e-01 3.93509641e-02 6.42040849e-01
4.84922737e-01 -1.28797382e-01 -1.35624313e+00 8.22712004e-01
3.96733582e-01 -1.40291512e+00 7.31500238e-02 -1.42602742e-01
2.93071389e-01 3.91814917e-01 1.26175463e-01 2.34735593e-01
2.42696777e-02 -1.12613773e+00 9.51884389e-02 5.12713075e-01
7.91373670e-01 -3.35776895e-01 1.19770622e+00 2.02314407e-01
-4.79890227e-01 2.09017724e-01 -5.97643197e-01 4.87895012e-01
1.73402831e-01 7.99590349e-01 -1.32717669e+00 6.94353461e-01
7.13300705e-01 6.54990613e-01 -3.95253181e-01 1.05184901e+00
-1.42759874e-01 3.37516367e-01 -1.29347846e-01 -1.94050565e-01
3.00402883e-02 2.80889690e-01 3.93315524e-01 1.34336436e+00
2.53435880e-01 1.91047698e-01 -2.59150445e-01 4.57801253e-01
-1.91047832e-01 6.04941785e-01 -6.47683859e-01 -3.21257859e-01
1.01525605e-01 1.20098817e+00 -5.10797381e-01 -2.42959857e-01
-1.61789969e-01 6.96962237e-01 1.87034369e-01 2.72712559e-01
-2.68603116e-01 -3.35926920e-01 2.63590872e-01 -2.50594169e-01
-2.15663269e-01 6.28733277e-01 2.94854352e-03 -7.93334782e-01
3.69441621e-02 -1.04848075e+00 9.31940615e-01 -9.48914766e-01
-1.50407171e+00 8.70185733e-01 -5.77379204e-02 -1.39142227e+00
-5.29300511e-01 -4.94152039e-01 5.21955565e-02 1.04417920e+00
-1.81374300e+00 -7.35352099e-01 3.75960678e-01 2.96307743e-01
6.17130935e-01 1.00501184e-03 1.34341586e+00 4.74404842e-01
-1.17836334e-01 1.36366412e-01 2.55793244e-01 3.23411673e-02
1.13413119e+00 -9.19275343e-01 -8.60267207e-02 2.76159376e-01
1.71396658e-02 1.00733924e+00 5.63396990e-01 -5.72174549e-01
-1.13671803e+00 -7.64605165e-01 1.72885406e+00 -5.66936493e-01
3.69950563e-01 2.88347065e-01 -8.44287276e-01 3.55589837e-01
1.82144731e-01 -1.89968258e-01 1.12606061e+00 1.68454438e-01
-7.60484114e-02 -7.09902272e-02 -1.19989753e+00 5.07352531e-01
3.98823202e-01 -7.60040998e-01 -1.07851219e+00 8.21685672e-01
9.25311327e-01 -1.18991263e-01 -1.09048855e+00 3.81176203e-01
4.23406780e-01 -3.13078940e-01 1.30399776e+00 -1.03527641e+00
8.02884579e-01 4.96945083e-02 -1.65618584e-02 -1.18720400e+00
-1.43813133e-01 -2.66659498e-01 3.37443382e-01 8.60471249e-01
7.39999771e-01 -3.41511369e-01 3.38976502e-01 7.90347934e-01
-1.23958692e-01 -9.07319307e-01 -9.61313665e-01 -1.71047613e-01
2.37447366e-01 -3.34155411e-01 4.59806435e-02 7.41102338e-01
6.87390387e-01 8.18823695e-01 5.43779247e-02 -1.69978648e-01
1.43822715e-01 -3.73065695e-02 -7.03808889e-02 -1.18469465e+00
-2.31721818e-01 -3.95384759e-01 -3.05193782e-01 -5.85388482e-01
-2.23485157e-02 -1.44977367e+00 2.81933486e-01 -1.98466909e+00
5.10423243e-01 -2.92609781e-01 -8.08754027e-01 4.68800604e-01
-3.10837030e-01 1.90827325e-01 -1.61282048e-01 3.78447920e-01
-5.26283920e-01 1.61192685e-01 1.04809105e+00 -4.35508937e-01
-1.41722023e-01 4.77033183e-02 -1.08373392e+00 2.94554144e-01
7.87803590e-01 -8.67800832e-01 -4.59850639e-01 -4.70418274e-01
8.17480922e-01 3.60076994e-01 -2.12333128e-02 -4.21041936e-01
2.83229828e-01 2.41008729e-01 5.45425937e-02 -4.38139111e-01
2.50646591e-01 -5.88173926e-01 -1.20623298e-01 4.71199751e-01
-1.11296201e+00 2.77550608e-01 2.61685580e-01 5.35171390e-01
-4.83172238e-01 -6.51602924e-01 5.46162784e-01 -3.75739425e-01
-3.49652380e-01 -1.04541630e-01 -3.31430256e-01 2.95038462e-01
5.43211758e-01 3.16993326e-01 -3.70335162e-01 -2.05244631e-01
-6.48418188e-01 9.11357701e-02 -2.39361331e-01 6.62511766e-01
8.52883041e-01 -8.63081932e-01 -9.29189742e-01 -4.18446183e-01
4.11560833e-01 -1.11545667e-01 2.97119707e-01 1.07339418e+00
-3.36065531e-01 1.06295502e+00 2.38265395e-01 -4.35036898e-01
-1.50530505e+00 6.52831852e-01 -3.96414138e-02 -8.97401690e-01
-6.90907657e-01 8.33750904e-01 1.90925270e-01 -3.18385243e-01
9.64430720e-02 -1.45491943e-01 -8.25493693e-01 1.76691473e-01
1.07623959e+00 -4.41683345e-02 6.78944767e-01 -4.94632721e-01
-5.09035408e-01 2.94957161e-01 -3.75858635e-01 -2.25412682e-01
1.55779302e+00 -7.98633173e-02 -3.85342956e-01 3.78779620e-01
1.44993246e+00 -3.08674604e-01 6.38517812e-02 -3.67714196e-01
5.11713266e-01 2.51639515e-01 2.55577058e-01 -1.33183539e+00
-5.71370482e-01 1.00146604e+00 4.60500091e-01 -5.30333407e-02
1.03733230e+00 2.78211325e-01 7.19041288e-01 7.61631429e-01
2.27561921e-01 -1.12931120e+00 -3.19025777e-02 4.05975759e-01
1.10531867e+00 -1.24719787e+00 1.13293380e-01 -1.67976674e-02
-7.15133548e-01 1.17174196e+00 -2.17076793e-01 4.55712736e-01
5.30553281e-01 -5.23627363e-03 3.30824494e-01 -5.08818626e-01
-7.47719288e-01 4.67864498e-02 5.22705197e-01 2.54297376e-01
1.06469023e+00 -2.83497304e-01 -8.21843863e-01 5.13243079e-01
9.93688554e-02 3.17856103e-01 2.38765076e-01 9.99385953e-01
-1.53113246e-01 -1.29372442e+00 -1.24660760e-01 6.02291286e-01
-1.19514167e+00 -6.75704181e-01 -1.96353093e-01 1.76673561e-01
-6.09463215e-01 1.22998440e+00 -4.08646286e-01 1.12551332e-01
7.73765817e-02 2.32955649e-01 1.30724698e-01 -9.81954098e-01
-9.21953201e-01 -4.88318540e-02 4.03872281e-01 -4.09772873e-01
-5.67086101e-01 -4.86812055e-01 -1.24621439e+00 4.37702358e-01
-3.02097142e-01 6.73352718e-01 9.51779187e-01 9.81753469e-01
6.22195244e-01 5.95488369e-01 2.41271183e-01 -1.58644587e-01
-7.12445378e-01 -1.01068103e+00 -1.37488678e-01 4.56984907e-01
1.01679772e-01 -8.48341808e-02 -2.03060254e-01 1.71390399e-01] | [8.63451862335205, 8.688395500183105] |
483f9123-b8db-499b-b6c0-765eed7763f0 | understanding-model-complexity-for-temporal | 2303.07925 | null | https://arxiv.org/abs/2303.07925v6 | https://arxiv.org/pdf/2303.07925v6.pdf | Robust incremental learning pipelines for temporal tabular datasets with distribution shifts | In this paper, we present a robust deep incremental learning model for regression tasks on financial temporal tabular datasets. Using commonly available tabular and time-series prediction models as building blocks, a machine-learning model is built incrementally to adapt to distributional shifts in data. Using the concept of self-similarity, the model uses only a basic building block of machine learning methods, decision trees to build models of any required complexity. The model is demonstrated to have robust performances under adverse situations such as regime changes, fat-tailed distributions and low signal-to-noise ratios which is common in financial datasets. Model robustness are studied under different hyper-parameters such as model complexity and data sampling settings using XGBoost models trained on the Numerai dataset as a detailed case study. The two layer deep ensemble of XGBoost models over different model snapshots is demonstrated to deliver high quality predictions under different market regimes. Comparing the XGBoost models with different number of boosting rounds in three scenarios (small, standard and large), we demonstrated the model performances are monotonic increasing with respect to model sizes and converges towards the generalisation upper bound. Our model is efficient with much lower hardware requirement than other machine learning models as no specialised neural architectures are used and each base model can be independently trained in parallel. | ['Mauricio Barahona', 'Thomas Wong'] | 2023-03-14 | null | null | null | null | ['feature-engineering', 'time-series-prediction'] | ['methodology', 'time-series'] | [-3.45224738e-01 -2.52846152e-01 -4.57227081e-01 -6.47067189e-01
-5.87229073e-01 -7.45130956e-01 7.93399870e-01 2.92935997e-01
-1.85273990e-01 7.17334867e-01 -8.81172493e-02 -7.41390705e-01
-5.14488757e-01 -8.04412246e-01 -8.42695057e-01 -5.39273202e-01
-5.68789184e-01 7.59571671e-01 4.89055589e-02 -3.86164993e-01
3.71972322e-01 4.14739639e-01 -1.52873445e+00 5.95259488e-01
4.57072645e-01 1.56595552e+00 -4.46767330e-01 7.88719118e-01
-1.01754196e-01 1.01472914e+00 -4.42731529e-01 -4.49064493e-01
1.01046395e+00 -2.18101129e-01 -3.50335956e-01 -3.40887666e-01
4.79438990e-01 -1.90448686e-01 -1.84456676e-01 5.28263509e-01
7.02143431e-01 -7.42097199e-02 5.64443171e-01 -1.17245460e+00
-3.08341295e-01 8.03658545e-01 -4.88191962e-01 4.75132763e-01
-1.93381533e-01 9.59534869e-02 9.87334847e-01 -3.90688032e-01
3.38843614e-01 1.17965889e+00 1.35590017e+00 1.09003745e-01
-1.51489973e+00 -8.64844859e-01 2.23099798e-01 4.50106472e-01
-4.03517455e-01 -1.65582672e-01 7.18870461e-01 -4.30394500e-01
1.00066745e+00 3.47428024e-01 4.94173318e-01 1.05623305e+00
8.00861597e-01 6.50477827e-01 1.39069235e+00 -3.88443679e-01
5.09112120e-01 2.71344662e-01 3.18958998e-01 4.98335399e-02
1.13456644e-01 6.14391625e-01 -5.59833527e-01 -3.37184638e-01
2.23032370e-01 -1.27148610e-02 3.45797628e-01 -3.43505412e-01
-8.51089656e-01 1.07850027e+00 4.91384208e-01 2.51575857e-01
-5.82554281e-01 2.49135897e-01 1.11632740e+00 9.54235077e-01
9.02501404e-01 4.82065499e-01 -1.23757374e+00 2.38580052e-02
-1.00042379e+00 6.55776620e-01 7.47460783e-01 7.64507651e-01
1.12336926e-01 3.66031677e-01 -1.94067538e-01 7.46583760e-01
5.27806990e-02 2.67337441e-01 1.06551719e+00 -4.97757435e-01
5.44864237e-01 4.67548937e-01 -1.47201698e-02 -9.59741771e-01
-9.88255382e-01 -1.17997479e+00 -8.22985828e-01 1.08173937e-01
5.45948267e-01 -1.56721950e-01 -1.00773573e+00 1.44362867e+00
4.07680273e-01 -2.22129479e-01 -1.47341430e-01 2.24095196e-01
4.15410250e-01 7.54511893e-01 -1.39257267e-01 -4.13749874e-01
1.08372855e+00 -5.59518516e-01 -5.49135029e-01 -9.10067335e-02
6.26970291e-01 -4.98037279e-01 6.19445324e-01 6.21165216e-01
-1.02754343e+00 -6.63481951e-01 -1.08650041e+00 5.02235472e-01
-7.33623683e-01 -5.37140667e-01 7.67528296e-01 8.07426035e-01
-9.13040459e-01 9.42976952e-01 -6.17687047e-01 -1.11702420e-01
3.03321004e-01 4.10721064e-01 1.64384231e-01 3.95413518e-01
-1.46723461e+00 8.15185130e-01 6.23064637e-01 1.45579144e-01
-5.07775187e-01 -9.39600170e-01 -3.93802047e-01 9.44101363e-02
-1.65634066e-01 -6.64845765e-01 1.52547371e+00 -1.40473795e+00
-1.53548944e+00 4.59266096e-01 4.82179672e-01 -1.37547374e+00
1.11523342e+00 -8.65176097e-02 -6.05668008e-01 -4.64506119e-01
-1.28991172e-01 9.23120901e-02 9.29866612e-01 -8.04476678e-01
-8.17099571e-01 -4.49126601e-01 -4.24055666e-01 -1.21237591e-01
-1.07606307e-01 9.39494222e-02 2.42851466e-01 -7.97853291e-01
1.08964369e-01 -7.60701358e-01 -2.02245042e-01 -7.63019502e-01
-3.28455083e-02 9.39738285e-03 5.81872582e-01 -6.98337793e-01
1.25790405e+00 -1.54727614e+00 -5.35511494e-01 4.30148542e-01
-4.35188800e-01 -3.36328521e-02 9.37895253e-02 6.52137995e-01
-5.92421830e-01 -2.25864574e-01 1.02013342e-01 2.51604989e-02
1.40558586e-01 -8.12001824e-02 -7.42642224e-01 4.42376971e-01
-2.60667413e-01 9.90283251e-01 -6.52583897e-01 1.67793021e-01
1.29201293e-01 -1.64326668e-01 -3.32527965e-01 -2.42471080e-02
-3.77605855e-01 1.56160006e-02 7.22160041e-02 5.96695185e-01
7.32278168e-01 -1.05072007e-01 2.41983116e-01 1.65784404e-01
-2.82128677e-02 3.54492307e-01 -9.81132448e-01 9.43712950e-01
-5.52870274e-01 6.69738412e-01 -3.79021257e-01 -1.62827671e+00
1.28526783e+00 1.03042282e-01 4.88431752e-01 -1.43622363e+00
2.41095051e-01 3.99041057e-01 1.34469375e-01 2.04167087e-02
1.64128199e-01 -4.23589617e-01 -2.30973318e-01 4.49618161e-01
-4.23082570e-03 2.37490878e-01 1.22251831e-01 -3.60112786e-01
8.73401761e-01 1.03109203e-01 5.02085209e-01 -6.06067896e-01
2.69119799e-01 1.77233338e-01 5.25672436e-01 9.68656063e-01
-8.09652507e-02 2.21905619e-01 6.03502989e-01 -1.43033075e+00
-1.23243153e+00 -8.06828499e-01 -6.19217336e-01 1.57047057e+00
-4.65471476e-01 -1.58989474e-01 -3.30613315e-01 -4.72342819e-01
5.05001128e-01 9.22186017e-01 -9.86960948e-01 -2.43849099e-01
-5.58514178e-01 -1.57550597e+00 2.09862024e-01 5.71049750e-01
6.04545057e-01 -1.01401424e+00 -9.33603108e-01 6.04760647e-01
5.84232748e-01 -4.44403708e-01 2.58329540e-01 9.22552824e-01
-1.35291541e+00 -1.00432694e+00 -3.33154947e-01 -3.92358243e-01
-3.54311056e-02 -4.33976054e-01 1.48431695e+00 -5.28580308e-01
1.02972142e-01 5.52627034e-02 -1.98753506e-01 -9.51577961e-01
-5.98303080e-01 2.22024992e-01 6.79265633e-02 -4.20256779e-02
3.64358008e-01 -4.30122703e-01 -7.26870179e-01 4.26211566e-01
-7.48601317e-01 -1.19422987e-01 2.84196973e-01 1.32274103e+00
2.00297445e-01 1.59792647e-01 1.02107024e+00 -9.55717444e-01
7.68401325e-01 -8.37050140e-01 -1.02621794e+00 1.22176379e-01
-1.31070280e+00 8.67101029e-02 1.02234912e+00 -4.13462073e-01
-7.51954675e-01 -2.59819359e-01 3.84239584e-01 -9.59423706e-02
1.78383633e-01 8.29852104e-01 4.85379398e-01 5.55909157e-01
8.02906513e-01 2.80647695e-01 1.25624863e-02 -5.70964456e-01
2.27834791e-01 6.13370597e-01 3.87562960e-01 -4.29796040e-01
4.48331922e-01 3.09619904e-01 2.01322615e-01 -8.14116448e-02
-5.19449532e-01 -9.37531739e-02 -6.74482942e-01 -2.44818732e-01
-1.97921954e-02 -8.57260525e-01 -6.48100257e-01 7.33647823e-01
-6.50587082e-01 -5.70949376e-01 -1.93539903e-01 2.47880429e-01
-7.23823190e-01 -3.43622327e-01 -5.78672767e-01 -9.58328843e-01
-7.93616414e-01 -4.44956928e-01 5.43275356e-01 -8.87515321e-02
-1.69561952e-01 -1.32191849e+00 3.10024470e-01 -5.28285205e-02
7.42273211e-01 6.41584933e-01 1.08275175e+00 -1.23502624e+00
-1.48699433e-01 -5.86622536e-01 7.56294876e-02 2.96581745e-01
7.72875026e-02 -7.37108663e-02 -9.56290901e-01 -7.08057761e-01
1.13662980e-01 -2.23176330e-01 8.82319331e-01 7.29203224e-01
1.06694555e+00 -6.65010571e-01 -6.12473823e-02 7.06806660e-01
1.16592646e+00 7.18596101e-01 4.49099541e-01 1.24010384e+00
2.98582818e-02 6.77716255e-01 5.95728278e-01 5.28004527e-01
1.74759373e-01 4.98447239e-01 4.76753145e-01 2.32004933e-02
5.82645416e-01 -1.68962076e-01 3.86618495e-01 5.75634718e-01
2.73859859e-01 -1.24204820e-02 -1.22831416e+00 4.22061086e-01
-2.05985498e+00 -1.14678121e+00 -2.92311199e-02 2.46984673e+00
5.55057704e-01 6.57366991e-01 5.86129367e-01 2.23883703e-01
4.66395169e-01 1.15790283e-02 -9.40433919e-01 -8.60376835e-01
-7.47773349e-02 5.59193715e-02 1.01089466e+00 -1.58571787e-02
-1.30015564e+00 5.15101612e-01 7.14375830e+00 5.61725736e-01
-1.48056161e+00 -9.45503414e-02 1.32052231e+00 -5.68417490e-01
1.37353957e-01 -2.77077675e-01 -5.90815663e-01 8.96748841e-01
1.68190181e+00 -4.05284494e-01 1.47759065e-01 1.01950061e+00
2.19017088e-01 2.85519779e-01 -1.14512026e+00 7.75987864e-01
-3.56552541e-01 -1.52606344e+00 -1.57009467e-01 -6.85977265e-02
7.29631305e-01 5.05630076e-01 4.59457994e-01 7.25727379e-01
6.96698070e-01 -9.68447328e-01 9.27897155e-01 4.84403163e-01
3.44299555e-01 -1.03927648e+00 1.06590307e+00 2.39168718e-01
-7.73206472e-01 -7.81869233e-01 -2.62390524e-01 -2.61241227e-01
-2.59193987e-01 5.22754312e-01 -8.99870753e-01 6.64404035e-01
1.23507488e+00 7.15337873e-01 -6.90390170e-01 1.05347872e+00
5.98413646e-01 9.47812378e-01 -4.64120448e-01 -1.39828064e-02
3.86027247e-01 5.10028601e-02 1.95406854e-01 1.08960819e+00
3.84225368e-01 -4.10761237e-01 -2.12996379e-01 2.78981447e-01
1.84321910e-01 2.81106293e-01 -4.87097144e-01 5.29739559e-01
3.02380502e-01 6.97064042e-01 -6.50147200e-01 -4.11779881e-01
-3.26036245e-01 3.43980402e-01 4.58084233e-02 2.81166285e-01
-6.57560945e-01 -9.56272110e-02 2.59360492e-01 3.73588234e-01
5.32894790e-01 2.81636477e-01 -6.77741647e-01 -7.50249147e-01
-1.16512761e-01 -1.06489182e+00 9.49187160e-01 -5.03925502e-01
-1.48655283e+00 5.48376083e-01 1.89813018e-01 -1.41914916e+00
-9.62402165e-01 -8.27602863e-01 -7.37731576e-01 7.78689444e-01
-1.22460592e+00 -1.00139344e+00 1.37236819e-01 4.45737690e-01
3.30762774e-01 -8.09332430e-01 5.12407839e-01 -7.27636144e-02
-3.63630682e-01 6.75990999e-01 1.06517601e+00 -1.74609661e-01
6.02707922e-01 -1.48870361e+00 7.55491316e-01 5.47499835e-01
-1.10137597e-01 6.65928304e-01 1.00745857e+00 -4.91473019e-01
-1.03512836e+00 -1.19990623e+00 5.89664221e-01 -3.29859763e-01
9.96989846e-01 -6.51142478e-01 -9.15629327e-01 6.79416358e-01
2.65283465e-01 -3.33201326e-02 6.70491755e-01 3.95838171e-01
-4.47595239e-01 -8.39528263e-01 -1.09540761e+00 1.99768364e-01
4.40113634e-01 -1.61105067e-01 -5.43836474e-01 3.54652137e-01
3.12106073e-01 -4.34714735e-01 -9.08689082e-01 5.60371459e-01
9.15349364e-01 -1.19339263e+00 7.83067346e-01 -1.11358237e+00
1.98713586e-01 1.84613258e-01 -9.37383622e-02 -1.45221841e+00
-4.64153677e-01 -8.51292193e-01 -3.58850926e-01 9.59049642e-01
6.13985717e-01 -9.69909966e-01 7.08398402e-01 5.32360435e-01
2.96886235e-01 -7.93487906e-01 -1.27416122e+00 -1.07474005e+00
5.86807668e-01 -6.69100344e-01 8.63359571e-01 1.07123733e+00
9.21603516e-02 2.87186801e-01 -4.15736228e-01 -3.93480331e-01
3.89273107e-01 5.19454598e-01 1.01195395e+00 -1.40488386e+00
-4.64592457e-01 -6.55025959e-01 -5.07775068e-01 -4.76390362e-01
-4.88693714e-02 -8.88614416e-01 -5.35699427e-01 -9.59301591e-01
-1.47697881e-01 -5.06868780e-01 -1.03436434e+00 4.35413241e-01
1.60408601e-01 -1.65640935e-01 1.88002318e-01 2.41813198e-01
-1.80989251e-01 4.08741951e-01 3.44266057e-01 -4.06453103e-01
-4.08780783e-01 5.65971494e-01 -6.09114468e-01 3.35523188e-01
1.04816771e+00 -3.89604717e-01 -4.01493281e-01 1.11741073e-01
6.03418589e-01 1.37724087e-01 1.57220513e-01 -8.46457958e-01
-1.11758569e-03 -3.16989608e-02 8.56861711e-01 -9.25589859e-01
-4.46683675e-01 -9.79738355e-01 3.67627174e-01 7.17775524e-01
-5.34493089e-01 7.30948746e-01 9.12211597e-01 5.68561196e-01
-9.86767933e-02 1.27227902e-01 6.73344076e-01 -2.24059653e-02
-3.72144639e-01 9.44119766e-02 -1.78764850e-01 -3.26737374e-01
1.10543883e+00 -1.81165665e-01 -2.90756166e-01 -5.61598480e-01
-5.53250372e-01 4.87759620e-01 2.17222199e-02 8.03542078e-01
2.48954562e-03 -1.42950463e+00 -9.76176977e-01 2.07685471e-01
9.42973569e-02 -3.39693397e-01 4.74507473e-02 6.79331958e-01
-4.22869235e-01 8.10234547e-01 -3.88299704e-01 -6.26322925e-01
-9.84854043e-01 9.19004798e-01 8.16328287e-01 -6.14743531e-01
-3.81567776e-01 4.71104771e-01 -2.25239526e-02 -6.01557910e-01
1.43558592e-01 -5.70168436e-01 6.47836775e-02 5.48529983e-01
4.22935158e-01 3.91664207e-01 6.18329704e-01 -1.31528959e-01
-1.32296100e-01 1.14223465e-01 -5.15603006e-01 1.36917576e-01
1.76146114e+00 1.67016715e-01 1.24215886e-01 8.55904460e-01
1.10019565e+00 -6.97219729e-01 -1.30607343e+00 -1.92776382e-01
6.23910606e-01 -1.87635496e-01 1.93806499e-01 -1.19482064e+00
-9.42342460e-01 5.37494123e-01 1.01696992e+00 6.17690682e-01
1.11580062e+00 -4.66509551e-01 3.30632687e-01 2.87812620e-01
2.36750320e-01 -1.56663048e+00 -3.25504810e-01 5.49758196e-01
1.01680350e+00 -1.18702698e+00 3.16537917e-01 7.67637432e-01
-3.15653533e-01 1.33392036e+00 2.16282308e-01 -2.36041099e-01
6.86080873e-01 4.90741432e-02 2.63076484e-01 -1.09873258e-01
-1.32195473e+00 4.54418600e-01 3.01310152e-01 4.59840834e-01
2.96271533e-01 5.13320789e-02 -2.10746065e-01 5.46817839e-01
-6.92272723e-01 -1.34585947e-01 1.59209624e-01 6.17606640e-01
-2.40903255e-02 -7.75739193e-01 -4.76229310e-01 9.10210133e-01
-6.77687585e-01 -5.85930198e-02 -2.64363259e-01 1.19431818e+00
-2.08996654e-01 6.35083199e-01 5.12321830e-01 -3.92390907e-01
3.60759765e-01 3.01813543e-01 2.72631235e-02 1.06318392e-01
-1.07903731e+00 -2.91414298e-02 -2.29414762e-03 -5.80199063e-01
-1.12554342e-01 -1.08237052e+00 -5.32134533e-01 -4.94442403e-01
-1.70203552e-01 -2.02981189e-01 7.21370697e-01 6.83257699e-01
3.38210016e-01 4.80039299e-01 9.80538189e-01 -6.64076269e-01
-1.04428899e+00 -1.26298451e+00 -5.60779274e-01 3.86698455e-01
5.01558423e-01 -5.58833301e-01 -3.53306055e-01 -1.56954035e-01] | [7.0923309326171875, 3.299100160598755] |
58b68c92-8271-4225-a8ad-8f7a04f5eca9 | defending-against-adversarial-attacks-by-3 | 1909.06137 | null | https://arxiv.org/abs/1909.06137v1 | https://arxiv.org/pdf/1909.06137v1.pdf | Defending Against Adversarial Attacks by Suppressing the Largest Eigenvalue of Fisher Information Matrix | We propose a scheme for defending against adversarial attacks by suppressing the largest eigenvalue of the Fisher information matrix (FIM). Our starting point is one explanation on the rationale of adversarial examples. Based on the idea of the difference between a benign sample and its adversarial example is measured by the Euclidean norm, while the difference between their classification probability densities at the last (softmax) layer of the network could be measured by the Kullback-Leibler (KL) divergence, the explanation shows that the output difference is a quadratic form of the input difference. If the eigenvalue of this quadratic form (a.k.a. FIM) is large, the output difference becomes large even when the input difference is small, which explains the adversarial phenomenon. This makes the adversarial defense possible by controlling the eigenvalues of the FIM. Our solution is adding one term representing the trace of the FIM to the loss function of the original network, as the largest eigenvalue is bounded by the trace. Our defensive scheme is verified by experiments using a variety of common attacking methods on typical deep neural networks, e.g. LeNet, VGG and ResNet, with datasets MNIST, CIFAR-10, and German Traffic Sign Recognition Benchmark (GTSRB). Our new network, after adopting the novel loss function and retraining, has an effective and robust defensive capability, as it decreases the fooling ratio of the generated adversarial examples, and remains the classification accuracy of the original network. | ['Chaomin Shen', 'Yaxin Peng', 'Jinsong Fan', 'Guixu Zhang'] | 2019-09-13 | null | null | null | null | ['traffic-sign-recognition'] | ['computer-vision'] | [ 1.31276205e-01 4.03613657e-01 2.70023972e-01 -1.09690920e-01
-4.81430180e-02 -9.56183314e-01 4.64001864e-01 -3.73610914e-01
-6.68942153e-01 6.60654426e-01 -3.52312028e-01 -5.69996655e-01
-1.63774326e-01 -9.19989705e-01 -9.82901573e-01 -1.19193029e+00
-2.55539030e-01 -1.17354833e-01 3.94550145e-01 -5.12642682e-01
1.39149830e-01 7.24463284e-01 -1.02238190e+00 6.10821247e-02
7.34439850e-01 1.38885462e+00 -5.93796015e-01 5.35730481e-01
1.49400979e-01 7.79152393e-01 -1.02937424e+00 -7.38959253e-01
8.98566246e-01 -3.60040247e-01 -6.35255456e-01 -6.18694127e-01
4.15919185e-01 -2.34792098e-01 -8.82133424e-01 1.81714880e+00
3.12152028e-01 7.17909215e-03 6.39781654e-01 -1.66419101e+00
-5.27799904e-01 7.65462041e-01 -2.28635177e-01 1.49330094e-01
-1.71405673e-01 2.40931943e-01 6.56485558e-01 -4.72542346e-01
4.18327123e-01 1.27353239e+00 6.17960691e-01 9.02353346e-01
-9.15239811e-01 -9.55546379e-01 2.48424113e-01 1.44724548e-01
-1.36694837e+00 5.60122021e-02 5.58893263e-01 -2.83811271e-01
4.26249325e-01 5.66256285e-01 3.27368528e-01 1.10468173e+00
6.35515988e-01 4.50660855e-01 8.52608263e-01 -2.22265631e-01
1.47952855e-01 3.88850033e-01 3.25517282e-02 5.80213189e-01
3.46303344e-01 2.97198981e-01 6.46320134e-02 -2.73317575e-01
6.38823688e-01 9.21107382e-02 -5.84866166e-01 -2.35081837e-01
-5.93157887e-01 7.92815685e-01 8.24677527e-01 3.35427016e-01
-1.11823611e-01 2.46369407e-01 3.57675105e-01 8.04156363e-01
3.03237643e-02 3.45235080e-01 -3.26518744e-01 3.68665218e-01
-3.14423352e-01 6.74785152e-02 9.64097381e-01 4.78986204e-01
4.09457892e-01 4.50375140e-01 3.04479226e-02 3.47319484e-01
4.79181141e-01 7.37757623e-01 5.80568254e-01 -4.96036351e-01
4.72859204e-01 5.28880060e-01 -1.81508541e-01 -1.08725905e+00
-5.81228398e-02 -6.28656507e-01 -1.03725791e+00 9.92043197e-01
1.07466710e+00 -5.21629274e-01 -8.16940367e-01 2.08757710e+00
1.10320896e-01 4.30842638e-02 1.52071133e-01 8.65603387e-01
3.96654308e-01 4.12290782e-01 2.05284450e-03 4.48448285e-02
9.30070579e-01 -4.35521603e-01 -4.55296129e-01 -2.42537260e-01
3.82619560e-01 -3.61848861e-01 8.56358528e-01 2.85128683e-01
-8.82786989e-01 -3.20600480e-01 -1.37793422e+00 6.51416838e-01
-6.20245159e-01 -4.26744252e-01 2.14089394e-01 1.03619504e+00
-7.21882761e-01 1.01115763e+00 -6.14576280e-01 1.65594116e-01
2.23246604e-01 4.45069849e-01 -6.36520267e-01 4.04591829e-01
-1.80676746e+00 8.23901236e-01 5.93642116e-01 1.99282795e-01
-7.92751372e-01 -7.28640497e-01 -4.78245258e-01 3.41725081e-01
4.67725210e-02 -1.89001873e-01 5.61244667e-01 -1.49537921e+00
-1.46147346e+00 6.69513166e-01 7.95552552e-01 -8.72148097e-01
1.09052050e+00 -6.90278932e-02 -4.32620883e-01 -4.12236117e-02
-4.39424217e-01 2.55643517e-01 1.21329689e+00 -1.05730236e+00
-2.66152263e-01 -4.68621641e-01 2.30823889e-01 -1.61865532e-01
-5.33732057e-01 -1.13690294e-01 3.07169706e-01 -7.56180644e-01
-1.71291791e-02 -1.01755357e+00 -2.07581908e-01 9.23711658e-02
-5.89753032e-01 2.06528664e-01 9.82320726e-01 -4.74034071e-01
1.21599567e+00 -2.65569139e+00 -1.77389428e-01 7.04159975e-01
1.79204300e-01 7.54680872e-01 -1.45912662e-01 1.81924328e-01
-5.03862679e-01 3.15157712e-01 -3.66035134e-01 3.99557650e-01
6.70419261e-02 -2.02417467e-02 -7.44481027e-01 8.58336210e-01
6.87964857e-02 6.16246223e-01 -7.46441185e-01 1.57067478e-01
-1.34358272e-01 4.07461703e-01 -3.50444794e-01 4.20868620e-02
2.41061345e-01 1.21822707e-01 -4.72257286e-01 -1.36380762e-01
1.00790548e+00 2.97378153e-01 -7.09596425e-02 -1.08035319e-01
1.77600726e-01 -9.94885862e-02 -1.41059113e+00 7.18454599e-01
-5.54463863e-02 6.72237933e-01 -4.47863676e-02 -9.82806027e-01
9.93233860e-01 2.15137288e-01 1.17503673e-01 -4.27192450e-01
4.61508900e-01 3.50965738e-01 5.95420539e-01 5.95187694e-02
-2.31922626e-01 -2.19209507e-01 4.11329456e-02 1.56635687e-01
-5.09265997e-02 3.15984070e-01 -1.06097125e-01 3.21643442e-01
1.18382823e+00 -4.16178226e-01 1.01350963e-01 -4.02113408e-01
9.17542398e-01 -4.80501920e-01 4.93526578e-01 8.74599874e-01
-4.37570602e-01 1.72382548e-01 1.07252026e+00 -3.91208351e-01
-8.31089318e-01 -1.36288667e+00 -1.54182091e-01 6.41028225e-01
1.90177098e-01 6.15262538e-02 -1.03274345e+00 -1.20804930e+00
3.46854031e-01 6.86836302e-01 -8.50324869e-01 -8.18177760e-01
-5.42055190e-01 -6.74799800e-01 1.18679798e+00 3.37889016e-01
7.55042791e-01 -8.96594286e-01 -3.12965989e-01 -1.78668424e-01
3.64337951e-01 -8.28430295e-01 -3.37799728e-01 2.72520065e-01
-5.04443109e-01 -1.21115637e+00 -3.85373622e-01 -4.10598069e-01
7.54259586e-01 -1.05787836e-01 4.95765030e-01 2.04701290e-01
-1.84208184e-01 -5.41749597e-02 -7.05372319e-02 -5.57605088e-01
-6.96552575e-01 -3.88189495e-01 3.33408892e-01 4.26544398e-01
-9.69881266e-02 -4.90894884e-01 -4.67400849e-01 5.08577585e-01
-1.07555330e+00 -6.67640328e-01 2.58535236e-01 8.53662431e-01
3.38824034e-01 2.96101391e-01 6.23190582e-01 -7.49554276e-01
6.40550792e-01 -3.46820116e-01 -8.02894473e-01 1.85611099e-01
-4.20144975e-01 2.26298407e-01 1.29879856e+00 -8.75006795e-01
-7.79214799e-01 -2.88727999e-01 -3.22389483e-01 -7.65735507e-01
2.46252585e-02 -3.64588834e-02 -5.01196921e-01 -5.91501951e-01
8.85047436e-01 1.33166268e-01 -1.23190852e-02 -1.52862251e-01
2.63786972e-01 3.56843799e-01 5.58976114e-01 3.50589305e-02
1.38743424e+00 4.31882858e-01 2.80671358e-01 -5.58740377e-01
-5.70231438e-01 2.70700455e-01 -2.76036739e-01 -1.75917536e-01
5.95203876e-01 -2.52685606e-01 -1.13310230e+00 9.44328964e-01
-9.23214555e-01 -1.42952546e-01 -4.89420682e-01 5.38905919e-01
-1.65640563e-01 2.85750002e-01 -4.89749044e-01 -8.07721794e-01
-3.70110750e-01 -1.00452650e+00 1.07598096e-01 2.57398725e-01
1.80834129e-01 -1.09423411e+00 -2.74275601e-01 -1.56371504e-01
6.03387058e-01 7.46818066e-01 1.00308561e+00 -1.30038249e+00
-2.72044152e-01 -7.82101929e-01 -7.72075802e-02 1.04578090e+00
-1.47751734e-01 6.81054518e-02 -9.94196057e-01 -4.27581489e-01
5.43981731e-01 -1.32585764e-01 8.91886592e-01 7.36438185e-02
1.13442886e+00 -7.55399346e-01 6.36288226e-02 8.26521695e-01
1.35806608e+00 4.50166136e-01 7.63996780e-01 4.31172252e-01
7.69363999e-01 3.36329490e-01 2.76548386e-01 2.01553494e-01
-3.69239956e-01 3.27230245e-01 9.99262094e-01 1.77096978e-01
2.42957622e-01 -6.71759099e-02 7.86085427e-01 2.97674298e-01
1.04230680e-01 -2.53685385e-01 -7.60540307e-01 -1.32409126e-01
-1.64024270e+00 -1.18468761e+00 -1.79621577e-01 2.49570799e+00
5.69134951e-01 7.31270671e-01 -2.27179736e-01 3.95547152e-01
5.87089062e-01 2.63181388e-01 -8.48903060e-01 -5.38209081e-01
-2.75217801e-01 8.46822113e-02 9.80827987e-01 5.38733780e-01
-1.16395640e+00 7.60061920e-01 5.60051155e+00 1.06117606e+00
-1.40171182e+00 -1.22332744e-01 6.38316631e-01 -1.85172036e-02
-7.91820064e-02 -2.85934895e-01 -5.20673394e-01 6.02900982e-01
9.08923984e-01 -2.94902474e-01 4.70750034e-01 1.06454074e+00
-2.57761329e-01 3.84339839e-01 -9.41500604e-01 5.61178923e-01
-1.75225198e-01 -1.01089120e+00 5.40197156e-02 4.89481688e-02
2.96154916e-01 9.49833095e-02 4.24330056e-01 2.63333201e-01
3.29489678e-01 -9.95710492e-01 7.60344565e-01 3.46358567e-01
5.51580071e-01 -1.07362902e+00 1.00180757e+00 3.89346957e-01
-8.65051091e-01 -2.64140338e-01 -3.62325549e-01 1.29488721e-01
-2.49886408e-01 3.40488762e-01 -4.19698089e-01 3.17793101e-01
5.77336729e-01 8.30294415e-02 -3.50807399e-01 7.41645515e-01
-3.74172986e-01 4.97244835e-01 -3.16262096e-01 8.98347348e-02
4.44713205e-01 -2.93947995e-01 1.02414095e+00 8.83260310e-01
-7.90023580e-02 -2.44604915e-01 -1.20142780e-01 8.03978145e-01
-3.60474229e-01 -7.82622546e-02 -6.20913863e-01 3.70615646e-02
4.03844446e-01 1.18527198e+00 -5.61066210e-01 -7.92277828e-02
9.35990289e-02 8.45432878e-01 1.45437583e-01 3.41250211e-01
-1.05010366e+00 -8.83313656e-01 9.90179121e-01 -1.45943448e-01
2.94683985e-02 2.15914160e-01 -3.29652101e-01 -8.83279264e-01
2.35727951e-01 -1.03199100e+00 3.12368691e-01 -7.88611770e-02
-1.32853973e+00 8.65664840e-01 -2.35162720e-01 -1.33578312e+00
-2.14156508e-01 -8.34552765e-01 -9.30049300e-01 1.00085151e+00
-9.60058153e-01 -4.81713027e-01 -8.29966143e-02 9.27974820e-01
-1.29047304e-01 -6.11561000e-01 6.17659390e-01 3.99589330e-01
-7.35180676e-01 1.16540492e+00 1.78736746e-01 6.47567451e-01
4.72663730e-01 -1.12451482e+00 5.61553061e-01 1.03200483e+00
-3.68942618e-02 6.88174307e-01 7.87046850e-01 -4.26353604e-01
-1.15060961e+00 -1.04992425e+00 3.44570786e-01 -3.03007722e-01
1.12610698e+00 -3.54030222e-01 -1.05750179e+00 5.67696512e-01
-2.22577214e-01 2.38967344e-01 3.99204254e-01 -6.10927284e-01
-7.15084255e-01 -2.48023316e-01 -1.66169989e+00 7.46163666e-01
7.14540064e-01 -4.23327267e-01 -2.49910742e-01 2.77833581e-01
6.41058803e-01 -2.53118366e-01 -7.28524089e-01 4.44792002e-01
7.39528060e-01 -1.04200613e+00 1.00175703e+00 -1.00535643e+00
8.49210173e-02 -2.32562989e-01 -1.73205763e-01 -1.41317213e+00
-3.07305187e-01 -6.81907594e-01 -1.05571151e-02 9.16905344e-01
4.05487597e-01 -1.13587070e+00 6.48473024e-01 5.00460029e-01
2.46762052e-01 -9.17329550e-01 -1.17773342e+00 -1.06257367e+00
3.74901026e-01 -1.19883195e-01 4.51850832e-01 9.38153625e-01
-2.87707269e-01 -3.19558948e-01 -2.87879765e-01 3.46559137e-01
6.12749159e-01 -6.73990905e-01 4.97007668e-01 -1.23490059e+00
-3.66424710e-01 -5.76885104e-01 -8.84459674e-01 -3.62782508e-01
2.96958417e-01 -9.11143303e-01 -2.34081820e-01 -7.74489403e-01
-4.36062127e-01 -3.05573046e-01 -6.52110696e-01 5.44352889e-01
-1.15009286e-01 1.70314565e-01 5.19704103e-01 3.89852412e-02
1.65760145e-01 2.29511037e-01 1.05136192e+00 -2.35850066e-01
-1.26693398e-01 2.23181903e-01 -5.13517022e-01 1.18242276e+00
9.36590254e-01 -8.19875181e-01 -3.78220588e-01 4.86588897e-03
-3.60998958e-02 -3.38566869e-01 4.36498851e-01 -1.02975893e+00
1.16973326e-01 1.69635974e-02 3.32174420e-01 1.33905441e-01
1.12596035e-01 -1.03563559e+00 1.80210307e-01 1.07959342e+00
-3.23641658e-01 -1.14369147e-01 1.46368533e-01 5.72668552e-01
-1.13439694e-01 -4.75269139e-01 1.41043186e+00 1.69322059e-01
-1.58680961e-01 3.00232768e-01 -3.04021955e-01 1.87620431e-01
1.12946272e+00 -2.20302701e-01 -5.46853364e-01 -2.03351513e-01
-4.68575478e-01 -5.34992591e-02 2.19764963e-01 4.95954573e-01
5.32270968e-01 -1.37901151e+00 -6.39785349e-01 6.03259087e-01
-3.18070769e-01 -5.99750400e-01 2.10752472e-01 6.79479182e-01
-5.17203033e-01 9.32774395e-02 -5.23930907e-01 -1.19359037e-02
-1.34119964e+00 4.94039595e-01 9.31582987e-01 -3.81027520e-01
-3.75300795e-01 9.91876066e-01 4.77475882e-01 -1.74915969e-01
4.82286543e-01 9.75058973e-02 -1.79604605e-01 -8.88612270e-02
6.40883684e-01 4.10249442e-01 -2.28259303e-02 -6.14635408e-01
-5.31449974e-01 2.26451918e-01 -1.42730862e-01 -2.32898630e-02
1.04500091e+00 4.75441217e-01 -1.37020439e-01 7.99059272e-02
1.47566938e+00 7.09359571e-02 -1.18561780e+00 7.63059407e-02
-1.68212011e-01 -5.15062392e-01 -1.66880965e-01 -6.96289361e-01
-1.53237510e+00 7.73576200e-01 8.47839355e-01 6.06263280e-01
1.01701319e+00 -4.57376212e-01 5.50709724e-01 5.28367758e-01
1.82915330e-01 -7.68460155e-01 -4.10326496e-02 7.09294140e-01
9.13505375e-01 -8.05826724e-01 -3.90775472e-01 -2.09093079e-01
-7.02894509e-01 1.12625396e+00 5.90409875e-01 -6.79991543e-01
9.11316216e-01 3.41262430e-01 2.22103015e-01 2.60050923e-01
-3.18542093e-01 4.35816258e-01 3.76564831e-01 3.74037653e-01
-6.43416047e-02 -2.72543579e-02 -3.66008788e-01 6.10645294e-01
-4.00359750e-01 -6.50761724e-01 4.81811821e-01 5.19721627e-01
-4.89054769e-01 -9.75729942e-01 -4.18192327e-01 2.44538993e-01
-8.80334020e-01 -3.75069566e-02 -6.10238671e-01 8.31631958e-01
9.01120603e-02 6.86177433e-01 -1.42397881e-01 -6.26939178e-01
5.15745997e-01 6.20816052e-02 1.64977431e-01 -3.87253352e-02
-9.19636428e-01 -5.19136786e-01 -3.05386126e-01 -7.07229137e-01
3.63834679e-01 -1.40572578e-01 -1.22592700e+00 -4.98630494e-01
-3.43433738e-01 2.34474778e-01 6.65800810e-01 7.63570726e-01
-1.13535061e-01 4.38760251e-01 1.11838078e+00 -5.10652840e-01
-1.21289682e+00 -7.26265609e-01 -8.49272609e-01 7.25963235e-01
3.94205093e-01 -4.91867155e-01 -1.09714139e+00 -4.15340513e-01] | [5.638783931732178, 7.814046382904053] |
e5eb62bb-5e44-4e2c-b94e-be1b23a9c912 | utopia-unconstrained-tracking-objects-without | 2306.09613 | null | https://arxiv.org/abs/2306.09613v1 | https://arxiv.org/pdf/2306.09613v1.pdf | UTOPIA: Unconstrained Tracking Objects without Preliminary Examination via Cross-Domain Adaptation | Multiple Object Tracking (MOT) aims to find bounding boxes and identities of targeted objects in consecutive video frames. While fully-supervised MOT methods have achieved high accuracy on existing datasets, they cannot generalize well on a newly obtained dataset or a new unseen domain. In this work, we first address the MOT problem from the cross-domain point of view, imitating the process of new data acquisition in practice. Then, a new cross-domain MOT adaptation from existing datasets is proposed without any pre-defined human knowledge in understanding and modeling objects. It can also learn and update itself from the target data feedback. The intensive experiments are designed on four challenging settings, including MOTSynth to MOT17, MOT17 to MOT20, MOT17 to VisDrone, and MOT17 to DanceTrack. We then prove the adaptability of the proposed self-supervised learning strategy. The experiments also show superior performance on tracking metrics MOTA and IDF1, compared to fully supervised, unsupervised, and self-supervised state-of-the-art methods. | ['Khoa Luu', 'Bhiksha Raj', 'Samee U. Khan', 'John Gauch', 'Kha Gia Quach', 'Pha Nguyen'] | 2023-06-16 | null | null | null | null | ['object-tracking', 'multiple-object-tracking'] | ['computer-vision', 'computer-vision'] | [ 2.71358311e-01 -3.57773989e-01 -2.56962031e-01 -2.37322375e-01
-4.54285532e-01 -4.59982574e-01 5.34953713e-01 -5.52251041e-02
-5.53639352e-01 1.09254754e+00 -2.62234449e-01 4.48093653e-01
-2.25005701e-01 -3.36481422e-01 -9.05802190e-01 -7.59569764e-01
-2.06546903e-01 9.09940004e-01 9.24396455e-01 3.49705573e-03
1.17264442e-01 -1.29802590e-02 -1.95930171e+00 6.98919445e-02
6.50636375e-01 8.07285249e-01 5.11324883e-01 7.10133135e-01
3.58114094e-02 7.98776686e-01 -5.55527210e-01 -1.28245950e-02
4.01848078e-01 -4.46756214e-01 -7.07212746e-01 2.32993856e-01
6.81973994e-01 -1.16061315e-01 -2.29942545e-01 1.07214093e+00
3.36826354e-01 3.43710870e-01 4.54611391e-01 -1.50715208e+00
-4.56993043e-01 5.92037678e-01 -7.14412987e-01 6.34238303e-01
2.89160401e-01 1.62580311e-01 6.36339784e-01 -6.75712287e-01
8.88369620e-01 1.22077215e+00 6.69821918e-01 8.88532043e-01
-9.69881475e-01 -8.38048697e-01 4.18998480e-01 3.00718725e-01
-1.27519596e+00 -2.72871912e-01 5.30209780e-01 -5.91280043e-01
2.33703688e-01 1.40454814e-01 7.10050344e-01 1.10141134e+00
5.47547005e-02 1.02677190e+00 1.00487494e+00 -1.47611454e-01
-9.22666416e-02 3.01133215e-01 9.67708752e-02 6.99667394e-01
4.79956836e-01 3.39183152e-01 -7.97891021e-01 2.83478171e-01
5.37721336e-01 1.08791634e-01 -1.42840058e-01 -7.80675173e-01
-1.38859069e+00 2.65244573e-01 1.45721942e-01 4.24127072e-01
-4.34064716e-02 1.02798507e-01 3.30138683e-01 1.88778237e-01
1.00477137e-01 1.81941867e-01 -5.33425331e-01 -2.37313688e-01
-9.26439345e-01 -1.62198693e-02 4.71831709e-01 1.40959048e+00
9.19784486e-01 -1.26668951e-02 -6.40630797e-02 3.13886672e-01
2.95619458e-01 8.42206836e-01 7.93337226e-01 -6.25354409e-01
2.30842873e-01 7.15110302e-01 4.11947697e-01 -8.25566173e-01
-4.94040549e-01 -3.58399272e-01 -4.50522482e-01 8.79734606e-02
5.35256386e-01 -1.51765615e-01 -8.52048039e-01 1.65750825e+00
6.96603298e-01 7.33429253e-01 1.16515405e-01 8.19183946e-01
1.03545117e+00 4.60534126e-01 8.28372613e-02 -3.97294223e-01
1.05520892e+00 -1.31159115e+00 -7.95478523e-01 -2.81793088e-01
5.44391870e-01 -5.74761927e-01 8.09986830e-01 3.68331254e-01
-7.50418365e-01 -1.05715370e+00 -8.74959230e-01 3.50321263e-01
-4.95706707e-01 2.97025740e-01 3.20838034e-01 3.53865325e-01
-6.47582948e-01 4.79405403e-01 -9.22559142e-01 -7.36893475e-01
5.59552252e-01 6.53696954e-01 -2.70181090e-01 -1.31963454e-02
-7.59614170e-01 6.67691350e-01 9.20054257e-01 -1.06806658e-01
-1.41614461e+00 -5.51658034e-01 -5.99883854e-01 -4.93648946e-01
8.20130825e-01 -4.80096489e-01 1.12391472e+00 -1.01105523e+00
-1.20836997e+00 8.74910891e-01 -8.66179839e-02 -4.95442837e-01
5.96537828e-01 -4.74565774e-01 -6.63479567e-01 -9.00369436e-02
3.48154873e-01 8.20713580e-01 1.13253319e+00 -1.34087706e+00
-1.22634256e+00 -1.36780575e-01 6.31815642e-02 2.23881170e-01
-4.43479359e-01 -2.12147936e-01 -4.14841682e-01 -4.11364406e-01
-1.88763365e-01 -1.06045794e+00 1.94112450e-01 -2.55440678e-02
-8.52885023e-02 -4.03066367e-01 1.52581179e+00 -6.50300384e-02
1.18252265e+00 -2.12830162e+00 2.30502009e-01 -3.06314141e-01
9.50034410e-02 4.91402566e-01 -3.23852487e-02 7.69548118e-02
2.70959318e-01 -3.81464988e-01 -1.58518441e-02 -2.35793903e-01
-2.07956940e-01 3.65005672e-01 2.05312788e-01 4.38363671e-01
7.95376152e-02 8.46657693e-01 -1.36886096e+00 -1.06366026e+00
2.52795607e-01 -2.16380395e-02 -3.54952604e-01 1.55256748e-01
-4.37694132e-01 1.01841164e+00 -4.08399850e-01 8.77284527e-01
5.61267793e-01 -7.00021327e-01 8.37218091e-02 -1.64116949e-01
-1.34708419e-01 -3.72757822e-01 -1.48235667e+00 1.87083960e+00
7.88929835e-02 6.34519756e-01 -8.64051506e-02 -1.03172362e+00
9.06723797e-01 -5.26731554e-03 8.30306351e-01 -4.19946611e-01
2.84960628e-01 1.61685854e-01 4.13231403e-02 -6.20337486e-01
6.37092769e-01 3.56548056e-02 -2.33239308e-02 3.60572428e-01
3.40601653e-01 5.48781753e-01 7.07472086e-01 -3.86551693e-02
9.57075059e-01 5.09657383e-01 3.40474784e-01 -3.12707692e-01
9.34146583e-01 4.40540880e-01 7.80790210e-01 7.53403962e-01
-6.68811679e-01 2.61450320e-01 -2.39802793e-01 -5.22814870e-01
-6.24250054e-01 -8.84965062e-01 -9.01143029e-02 1.48349011e+00
9.89722490e-01 -3.40642571e-01 -4.48963374e-01 -1.02270818e+00
1.23606950e-01 8.09500664e-02 -7.18924582e-01 -3.91225591e-02
-7.93740094e-01 -6.33067071e-01 4.46019530e-01 5.19385159e-01
8.26543093e-01 -1.09003019e+00 -8.84926438e-01 1.48706421e-01
-2.75681794e-01 -1.33009315e+00 -5.56596875e-01 1.63165499e-02
-8.02021205e-01 -1.53053844e+00 -7.12267756e-01 -1.00150836e+00
7.17551291e-01 3.81020933e-01 8.79121244e-01 2.97978938e-01
-1.96417466e-01 6.79971397e-01 -3.92794400e-01 -4.98172283e-01
-2.42481038e-01 1.94835901e-01 5.15142262e-01 1.36640310e-01
5.58343112e-01 -2.33094424e-01 -3.97415996e-01 8.32261086e-01
-6.99922979e-01 -1.29754245e-01 6.66830480e-01 8.19995761e-01
7.91521311e-01 1.28347725e-01 3.62258583e-01 -7.08458185e-01
2.03003231e-02 -3.90907586e-01 -7.50400364e-01 4.35894787e-01
-5.45449436e-01 -8.50763544e-02 4.70046192e-01 -9.84170198e-01
-9.65964377e-01 4.42661315e-01 5.67234218e-01 -5.48067629e-01
-3.40266734e-01 -1.63214266e-01 -6.52447194e-02 -3.06926638e-01
7.67628968e-01 2.95731723e-01 -2.08139941e-01 -4.91653502e-01
2.30265915e-01 4.82566714e-01 8.39484036e-01 -5.70784271e-01
1.38026607e+00 7.52065659e-01 -3.20539735e-02 -6.56043053e-01
-1.04015732e+00 -8.13102186e-01 -1.17436802e+00 -6.80121183e-01
9.99586225e-01 -9.80364621e-01 -6.93665683e-01 4.69229162e-01
-7.99479246e-01 -2.92836785e-01 -4.26349223e-01 5.61690331e-01
-5.09412766e-01 3.30174446e-01 1.11025296e-01 -6.51231587e-01
-3.35912108e-02 -7.96658635e-01 1.18097818e+00 6.40961766e-01
1.74540848e-01 -1.13873506e+00 1.72035247e-01 3.35779697e-01
1.10188432e-01 4.02941376e-01 7.08322674e-02 -7.93140531e-01
-8.33371222e-01 1.32190898e-01 3.94781716e-02 4.10122313e-02
4.31062669e-01 -2.43128553e-01 -6.02977633e-01 -7.24344492e-01
-2.93529987e-01 -4.29529041e-01 6.77315652e-01 6.72484413e-02
7.18271673e-01 -1.16770379e-01 -9.44932401e-01 5.03376126e-01
1.38837779e+00 3.60727608e-01 4.65717688e-02 5.78791559e-01
6.49779975e-01 2.52512962e-01 1.30334067e+00 3.64530921e-01
3.32085192e-01 6.57016993e-01 4.87936527e-01 2.04878569e-01
-1.34573027e-01 -3.15486759e-01 4.28553134e-01 5.73205650e-01
-1.50729986e-02 -1.62686631e-01 -9.33121443e-01 7.73204625e-01
-2.11549592e+00 -1.00499880e+00 -9.39119160e-02 2.11746550e+00
4.59469378e-01 3.23898017e-01 5.41072249e-01 -6.06270470e-02
1.07181036e+00 -3.99579182e-02 -9.34837043e-01 6.01448834e-01
-2.39000283e-02 -2.94509411e-01 5.27798772e-01 1.83314517e-01
-1.55999172e+00 1.16659427e+00 6.11015129e+00 6.05519295e-01
-9.47705865e-01 3.06445688e-01 -1.62079886e-01 -1.36696517e-01
5.17381966e-01 -1.32697653e-02 -1.30361640e+00 4.95136052e-01
5.46642661e-01 -1.67764351e-01 4.14590746e-01 9.51710105e-01
-2.68587589e-01 -2.01576844e-01 -1.27544785e+00 1.16694045e+00
2.81834990e-01 -1.14227760e+00 -1.32465303e-01 -2.85462379e-01
9.53791261e-01 1.17604844e-02 -7.81362206e-02 5.69167674e-01
3.27001274e-01 -5.40644109e-01 5.61825275e-01 6.86240077e-01
5.99747479e-01 -1.91858023e-01 7.58632421e-01 7.17394054e-01
-1.74830341e+00 -3.36680263e-01 -4.33513612e-01 1.20188221e-01
1.09593853e-01 -2.69272387e-01 -8.27078164e-01 6.83929026e-01
1.17313850e+00 1.31572700e+00 -1.02903593e+00 1.24858487e+00
2.96740998e-02 3.59122604e-01 -4.08306748e-01 -2.05193475e-01
1.20727718e-01 1.40256062e-01 8.01202953e-01 9.82558548e-01
1.02251783e-01 4.07649996e-03 4.24883306e-01 4.64055747e-01
1.60718769e-01 -5.39310910e-02 -5.71856916e-01 -4.19086628e-02
3.95623058e-01 1.28850639e+00 -8.18542600e-01 -5.45848429e-01
-3.06157589e-01 6.93486392e-01 3.28321755e-02 1.25807002e-01
-1.07240355e+00 -4.70411479e-02 4.67163801e-01 2.18167111e-01
8.21193039e-01 -6.01654723e-02 3.06380779e-01 -1.17272592e+00
-4.15450372e-02 -8.32366943e-01 7.79038548e-01 -6.66111529e-01
-1.15829217e+00 4.48948622e-01 3.76855969e-01 -1.83510089e+00
-2.14106634e-01 -6.23305798e-01 -3.06828201e-01 -9.49643180e-02
-1.49111319e+00 -1.21424615e+00 -8.47927988e-01 9.27039862e-01
6.86924815e-01 -5.90844035e-01 2.55654156e-01 5.81039310e-01
-5.43576658e-01 5.96069992e-01 4.15666819e-01 2.83990175e-01
8.67106378e-01 -1.01638329e+00 -7.20755607e-02 9.05912638e-01
2.67570376e-01 3.73006672e-01 6.91668987e-01 -7.98070014e-01
-1.46154809e+00 -1.39249659e+00 1.40480012e-01 -7.93282151e-01
6.34366095e-01 -3.25951755e-01 -7.35778749e-01 9.16870594e-01
7.48163089e-02 4.62695360e-01 4.13220316e-01 -2.63619363e-01
1.81441694e-01 -2.96962649e-01 -1.12286603e+00 9.66923088e-02
1.59175575e+00 1.00521468e-01 -6.82575107e-01 3.24845672e-01
6.37653470e-01 -8.00296605e-01 -8.91749203e-01 5.40425241e-01
4.73189116e-01 -7.64326036e-01 8.69619310e-01 -8.41525137e-01
-2.16854125e-01 -8.66926610e-01 -1.54127702e-01 -9.76724327e-01
-1.72807321e-01 -8.53774011e-01 -4.16740566e-01 1.28658664e+00
4.72327769e-02 -3.37884337e-01 1.00876248e+00 -1.56507015e-01
-9.17404890e-02 -2.91000247e-01 -9.47639167e-01 -1.27336383e+00
-2.24594876e-01 2.13552266e-02 6.17025197e-01 7.89867163e-01
-2.83899069e-01 2.97984838e-01 -7.06994295e-01 3.50253284e-01
9.95357811e-01 2.61681348e-01 1.47033000e+00 -1.62881958e+00
-2.16705903e-01 1.11631997e-01 -7.75655866e-01 -1.44565856e+00
-1.21694811e-01 -7.09006369e-01 2.92603761e-01 -1.32609499e+00
3.19374263e-01 -4.51656699e-01 -4.60373372e-01 4.45603997e-01
-2.32099518e-01 2.63264835e-01 2.72289366e-01 5.80494225e-01
-1.71216679e+00 5.17557204e-01 1.40332556e+00 -3.18852067e-01
-3.44218045e-01 6.13461109e-03 -4.14187908e-01 7.61955202e-01
5.99925280e-01 -9.11762416e-01 -4.10228819e-01 -4.01364475e-01
-3.12896281e-01 -3.15245509e-01 3.96795541e-01 -1.64959633e+00
5.48603892e-01 -2.92588711e-01 2.91075826e-01 -1.17201936e+00
2.90132940e-01 -1.06705725e+00 2.55999297e-01 5.91075003e-01
-4.28407863e-02 2.94749811e-02 3.39658797e-01 1.03886437e+00
-3.23412307e-02 -2.01965645e-01 8.09855521e-01 -1.46718144e-01
-1.46677709e+00 4.24789071e-01 -9.03172269e-02 4.50167835e-01
1.54172218e+00 -4.54640597e-01 -3.72354001e-01 2.59242710e-02
-8.51291478e-01 6.09312952e-01 4.66465235e-01 7.88750470e-01
5.08374989e-01 -1.47957277e+00 -4.88587230e-01 4.68070731e-02
4.14940327e-01 3.97431627e-02 -7.26745185e-03 1.03339851e+00
-1.81945279e-01 2.46527165e-01 -5.08416533e-01 -1.27888441e+00
-1.64383030e+00 7.99198508e-01 3.34844351e-01 -2.43551537e-01
-6.35558188e-01 5.60652196e-01 1.12386748e-01 -3.97044063e-01
5.89683652e-01 -2.97908366e-01 -3.82747710e-01 -1.43802479e-01
4.31926966e-01 5.29149652e-01 -5.06167352e-01 -9.47458684e-01
-5.18535614e-01 9.41937149e-01 -1.81838330e-02 3.01643074e-01
1.10962701e+00 -3.62300009e-01 3.52609247e-01 7.07740664e-01
7.27862418e-01 -3.81550580e-01 -1.41929948e+00 -5.59025288e-01
3.45044464e-01 -6.62033141e-01 -5.79379976e-01 -6.17853224e-01
-9.01168466e-01 6.30962789e-01 1.13481760e+00 -4.80889669e-03
8.93482685e-01 1.62256315e-01 8.80689859e-01 7.81719267e-01
7.01966643e-01 -1.25110674e+00 6.80140138e-01 3.56636852e-01
3.72805208e-01 -1.66762245e+00 1.39327690e-01 -7.44914040e-02
-5.96268117e-01 1.00161159e+00 1.28814483e+00 3.91895771e-02
5.85112393e-01 7.95114040e-02 3.17762196e-02 -2.48978898e-01
-5.64027488e-01 -6.38698041e-01 2.75291204e-01 9.29614604e-01
-8.96826684e-02 -3.37336034e-01 -3.19593363e-02 1.57490984e-01
7.30986521e-03 2.30792046e-01 3.76720697e-01 1.17563057e+00
-7.14429498e-01 -9.49109852e-01 -5.91297507e-01 2.73920983e-01
-3.67916264e-02 5.56534946e-01 -3.91082793e-01 1.46557355e+00
7.07886636e-01 9.39751208e-01 -3.89300101e-02 -6.73143268e-01
3.20880800e-01 -6.65764287e-02 5.09349525e-01 -6.17908061e-01
-2.99030274e-01 -3.75737771e-02 -2.07713634e-01 -1.99939236e-01
-1.09713626e+00 -1.02002656e+00 -1.49512792e+00 -2.76184119e-02
-6.64832532e-01 2.45653600e-01 1.97700992e-01 1.01220238e+00
1.68864220e-01 5.99373162e-01 3.49085152e-01 -6.84747338e-01
-2.06229642e-01 -1.00833476e+00 -3.22489619e-01 6.18203402e-01
5.48873603e-01 -1.28517389e+00 -2.31654555e-01 4.00645167e-01] | [6.385303974151611, -2.1037094593048096] |
996131cf-9699-49eb-ad91-335e14d65612 | improving-language-identification-for | 2001.11019 | null | https://arxiv.org/abs/2001.11019v1 | https://arxiv.org/pdf/2001.11019v1.pdf | Improving Language Identification for Multilingual Speakers | Spoken language identification (LID) technologies have improved in recent years from discriminating largely distinct languages to discriminating highly similar languages or even dialects of the same language. One aspect that has been mostly neglected, however, is discrimination of languages for multilingual speakers, despite being a primary target audience of many systems that utilize LID technologies. As we show in this work, LID systems can have a high average accuracy for most combinations of languages while greatly underperforming for others when accented speech is present. We address this by using coarser-grained targets for the acoustic LID model and integrating its outputs with interaction context signals in a context-aware model to tailor the system to each user. This combined system achieves an average 97% accuracy across all language combinations while improving worst-case accuracy by over 60% relative to our baseline. | ['Andrew Titus', 'Nanxin Chen', 'Jan Silovsky', 'Arnab Ghoshal', 'Roger Hsiao', 'Mary Young'] | 2020-01-29 | null | null | null | null | ['spoken-language-identification'] | ['speech'] | [-2.50857323e-01 -3.64039630e-01 -3.44837494e-02 -5.86608887e-01
-1.38847899e+00 -1.06219256e+00 7.64247596e-01 -1.90749899e-01
-6.30472779e-01 4.74326670e-01 5.22277653e-01 -5.40740669e-01
5.16875565e-01 -2.81410874e-03 -7.90482908e-02 -4.11754221e-01
3.29525262e-01 6.75453544e-01 1.37890667e-01 -1.18861675e-01
-1.74456358e-01 5.14382899e-01 -1.58860314e+00 3.22429776e-01
8.64022732e-01 5.18945515e-01 6.09836020e-02 8.01463902e-01
-2.76281297e-01 3.77246678e-01 -6.70336962e-01 -2.51794636e-01
3.38110961e-02 -3.53518903e-01 -6.34648561e-01 -1.54988289e-01
1.01465535e+00 -7.83132762e-02 -3.56612019e-02 9.49325085e-01
8.47769380e-01 -2.43388470e-02 5.54620743e-01 -8.98831785e-01
-3.42139423e-01 7.25009501e-01 -3.23933721e-01 6.51953667e-02
7.84791291e-01 2.76303351e-01 1.05158949e+00 -9.64572191e-01
2.15486944e-01 1.58779120e+00 6.89625680e-01 6.88501239e-01
-1.57133663e+00 -8.75063777e-01 4.03665602e-01 -2.22947538e-01
-1.57435477e+00 -1.00544059e+00 3.63082439e-01 -4.48631793e-01
1.18747139e+00 3.45981151e-01 1.34380773e-01 9.19622362e-01
-2.28851780e-01 7.92521954e-01 1.32201529e+00 -4.68051106e-01
-2.89509282e-03 5.85741580e-01 2.28049129e-01 2.88422614e-01
-5.80128431e-02 -1.73186526e-01 -5.19593358e-01 -2.23006144e-01
5.28238192e-02 -6.56798482e-01 -2.05559626e-01 1.46110162e-01
-1.03348970e+00 6.06555402e-01 -2.12979764e-01 4.85574126e-01
-7.97292069e-02 -4.25869137e-01 4.19865668e-01 3.25664371e-01
4.12905753e-01 3.76747489e-01 -4.88438189e-01 -3.57143283e-01
-1.07727802e+00 3.35546851e-01 1.17610931e+00 8.06628227e-01
4.32889730e-01 2.38901466e-01 -2.26199031e-01 1.42293155e+00
2.75008976e-01 6.50879085e-01 5.89395404e-01 -5.64277768e-01
4.14828509e-01 4.04647201e-01 9.62665863e-03 -2.39653587e-01
-5.64278185e-01 -4.29128587e-01 -3.41525912e-01 2.03912541e-01
6.76959753e-01 -3.82255822e-01 -8.82584095e-01 2.00837016e+00
1.09710850e-01 -3.64650011e-01 1.79138139e-01 8.13853025e-01
6.26564503e-01 6.19790196e-01 4.91623193e-01 -7.61305094e-02
1.54616594e+00 -6.50699198e-01 -4.36015338e-01 -7.53254592e-01
5.04445076e-01 -1.17050636e+00 1.42169797e+00 4.91575599e-01
-8.77065599e-01 -4.95626837e-01 -8.67336273e-01 -7.98619017e-02
-2.62526780e-01 1.06822304e-01 2.07463995e-01 1.14369118e+00
-1.47839952e+00 -2.31123939e-01 -3.30577075e-01 -6.92212522e-01
-2.80057102e-01 4.56458449e-01 -4.13995385e-01 2.14753374e-02
-1.16702092e+00 9.81879532e-01 -1.32233771e-02 -3.64701420e-01
-6.27095819e-01 -5.70103288e-01 -8.83005023e-01 -2.23807827e-01
3.19304802e-02 -1.40083700e-01 1.64493513e+00 -7.63881564e-01
-1.51744187e+00 1.07282412e+00 -4.18215007e-01 -3.24432492e-01
6.10584617e-01 -2.11612195e-01 -9.05616879e-01 -3.83723050e-01
7.42199272e-02 5.87344348e-01 4.88147110e-01 -1.12811649e+00
-9.44238603e-01 -2.60820299e-01 -2.67063648e-01 4.60869610e-01
-3.06990713e-01 6.08509183e-01 -7.03413427e-01 -4.89687741e-01
-8.79204571e-02 -1.03566301e+00 1.90354019e-01 -3.74990225e-01
-4.93319064e-01 -3.31034124e-01 9.56153631e-01 -9.91442680e-01
1.28763306e+00 -2.13008714e+00 -1.41217694e-01 -2.45095752e-02
-2.76627511e-01 4.55424011e-01 -1.34406969e-01 4.82205868e-01
7.51101226e-02 1.05928801e-01 -9.64326486e-02 -8.18164468e-01
1.83678716e-01 -5.26689813e-02 -1.94418371e-01 3.25342864e-01
-5.72590567e-02 4.03478295e-01 -5.56170166e-01 -3.06926787e-01
2.53557742e-01 6.77593827e-01 -4.34205055e-01 2.84453064e-01
-1.19812198e-01 5.83879590e-01 1.02469631e-01 5.82078516e-01
5.24868011e-01 3.40682238e-01 3.99968117e-01 9.22365561e-02
-5.04810452e-01 7.69757032e-01 -1.32903123e+00 1.16919708e+00
-8.08768272e-01 7.42829680e-01 5.39347649e-01 -3.01766872e-01
9.60004985e-01 5.45020640e-01 2.52490938e-02 -5.26360452e-01
-9.60357562e-02 4.73870754e-01 3.20686698e-01 -1.93105504e-01
4.14719075e-01 -1.29858211e-01 -4.95332628e-01 5.91681242e-01
-9.57118645e-02 -1.62964821e-01 -8.02268758e-02 -5.55520989e-02
7.11121976e-01 -2.49183565e-01 4.40398335e-01 -4.36794072e-01
7.36095250e-01 -1.88059509e-01 4.30745244e-01 8.34616661e-01
-5.12713969e-01 6.58407509e-01 1.60981163e-01 8.67022760e-03
-8.00705969e-01 -1.15406489e+00 -3.36489141e-01 1.57601821e+00
-2.46564582e-01 -1.65841013e-01 -9.51186299e-01 -5.90195179e-01
9.55527797e-02 9.71196353e-01 5.19297235e-02 4.79897819e-02
-5.64364016e-01 -7.21992075e-01 1.01696229e+00 4.17970568e-01
3.60516936e-01 -1.05464089e+00 1.14972942e-01 2.10394815e-01
-1.01930186e-01 -1.38454378e+00 -8.76612306e-01 2.77392209e-01
-4.16651517e-02 -6.13608062e-01 -8.12944353e-01 -1.02081609e+00
1.14404917e-01 1.40562253e-02 1.12461674e+00 -4.18599576e-01
-1.02961957e-01 4.87323314e-01 7.56164789e-02 -3.99670988e-01
-8.79785657e-01 3.11559379e-01 6.17926419e-01 8.12872648e-02
7.69154251e-01 -1.68013468e-01 -2.12157786e-01 3.92373562e-01
-3.55999291e-01 -3.01015496e-01 3.20105731e-01 5.35117328e-01
1.58504561e-01 -3.95177871e-01 6.54908776e-01 -6.48144782e-01
8.07845235e-01 -2.10421801e-01 -4.67338026e-01 2.08266601e-01
-6.27163529e-01 3.64605039e-02 7.17874587e-01 -6.10478401e-01
-1.06774199e+00 2.51269311e-01 -5.51929414e-01 1.18420124e-01
-6.37255073e-01 5.34812100e-02 -6.61763668e-01 2.24150121e-02
4.91599441e-01 3.23096126e-01 -8.67606327e-02 -7.98406065e-01
3.22956473e-01 1.33327293e+00 7.64751792e-01 -5.36452234e-01
4.33601737e-01 1.68150943e-02 -8.88531804e-01 -1.16544807e+00
-4.77548212e-01 -6.94539130e-01 -5.34887552e-01 -1.17335771e-03
7.37596810e-01 -1.17986917e+00 -5.97599745e-01 8.20247471e-01
-8.89976382e-01 -2.62330472e-01 2.71118909e-01 5.06919205e-01
-1.36117294e-01 2.26900294e-01 -4.96220648e-01 -1.11224687e+00
-3.04677665e-01 -1.44527292e+00 1.01980007e+00 1.28995925e-01
-7.95606613e-01 -9.29218948e-01 1.83470622e-01 3.45891118e-01
5.62213540e-01 -5.31031966e-01 7.68092513e-01 -1.11133575e+00
-1.23352548e-02 -3.04958045e-01 -2.48739440e-02 4.45360243e-01
3.92934471e-01 -4.85394299e-02 -1.52406228e+00 -4.28165048e-01
-3.45493704e-01 -2.45888025e-01 7.19851553e-01 3.09234113e-01
5.17137408e-01 -8.12130794e-02 -3.35882783e-01 4.00934309e-01
1.07472146e+00 3.86862665e-01 7.14706704e-02 5.94783612e-02
6.28361762e-01 7.44957209e-01 1.28606409e-01 1.08353816e-01
6.19544864e-01 1.03818905e+00 -2.81377375e-01 -4.64613959e-02
-3.89545381e-01 -1.61784247e-01 8.01180184e-01 7.19048440e-01
5.70463836e-01 -2.28341386e-01 -1.27216423e+00 7.18741715e-01
-1.39708316e+00 -7.60353565e-01 2.31085747e-01 2.47888994e+00
9.21447575e-01 1.24910355e-01 7.07741916e-01 -5.82872480e-02
8.92976522e-01 1.62319079e-01 -5.82292914e-01 -7.23643243e-01
-3.07286739e-01 -1.50866896e-01 2.01080441e-01 1.06262350e+00
-1.23933077e+00 1.20055389e+00 7.28797913e+00 5.49700975e-01
-1.46998382e+00 -2.74417102e-02 6.15110815e-01 2.31597889e-02
-1.80569753e-01 -1.69138342e-01 -1.32359564e+00 4.81129915e-01
1.08146846e+00 -1.91221759e-01 4.49923635e-01 7.69382417e-01
3.16382051e-01 -8.58653784e-02 -1.21951222e+00 1.01388133e+00
2.65418619e-01 -3.93692285e-01 6.15445664e-03 -4.25963961e-02
4.40781772e-01 3.07249129e-01 2.12498859e-01 3.96079600e-01
5.45769930e-01 -1.03936100e+00 8.22640598e-01 3.84150334e-02
9.95808244e-01 -7.23886192e-01 3.47190619e-01 3.72933716e-01
-1.22958434e+00 -3.85420630e-03 1.78727150e-01 -2.52745114e-02
7.43934140e-02 1.11161903e-01 -1.16638637e+00 -1.71959609e-01
6.10991955e-01 5.71148023e-02 -4.99820173e-01 1.00687122e+00
-9.98251587e-02 9.21174943e-01 -5.12453318e-01 1.24284690e-02
1.81850180e-01 1.25621960e-01 6.68612480e-01 1.77365518e+00
2.14160308e-01 -4.15438175e-01 4.73324478e-01 3.95948321e-01
5.43112159e-02 3.53819162e-01 -6.96800888e-01 -4.71605733e-03
7.56957173e-01 1.16122317e+00 -2.41348356e-01 -1.94058195e-01
-6.95002437e-01 1.01010072e+00 2.70437181e-01 4.90456760e-01
-3.35870147e-01 -3.02112430e-01 1.23330617e+00 -4.38672565e-02
-5.11658192e-02 -2.70439893e-01 -2.50169754e-01 -9.26595569e-01
-3.44341397e-02 -1.16155255e+00 4.36554044e-01 -3.28443170e-01
-1.36724639e+00 9.07974780e-01 -2.29529098e-01 -9.58651602e-01
-7.80319750e-01 -6.10156655e-01 -5.08453548e-01 1.35527194e+00
-1.14776075e+00 -1.33972418e+00 3.98837067e-02 4.10579801e-01
5.68663716e-01 -4.17123109e-01 1.06457543e+00 4.19720411e-01
-6.28895760e-01 1.03471053e+00 2.09330276e-01 3.44105601e-01
1.21791649e+00 -1.40485752e+00 6.79628372e-01 9.22920644e-01
2.52501994e-01 6.42074823e-01 7.93409526e-01 -4.41419601e-01
-1.12895072e+00 -9.96694982e-01 1.44592822e+00 -5.18155932e-01
4.57771212e-01 -8.14837992e-01 -9.24441755e-01 6.80132806e-01
2.95115203e-01 -3.23917001e-01 7.20505297e-01 5.28253734e-01
-7.97616839e-01 -1.46139890e-01 -1.08163178e+00 8.36162090e-01
8.35976779e-01 -1.19162083e+00 -3.71791691e-01 9.08335485e-03
5.46472907e-01 -1.83341563e-01 -5.88289618e-01 1.09484732e-01
8.46164346e-01 -9.58709419e-01 7.28101134e-01 -3.72856081e-01
-4.95198309e-01 -3.79703641e-01 -3.81363839e-01 -1.45525849e+00
-1.63939536e-01 -6.07026696e-01 4.13985193e-01 1.74071848e+00
6.55608654e-01 -7.34769881e-01 3.15009147e-01 8.06963086e-01
-5.17766215e-02 -1.99478105e-01 -8.92976522e-01 -8.86422157e-01
3.31550002e-01 -6.90970123e-01 5.79017937e-01 8.37633491e-01
1.21825106e-01 5.29219747e-01 -3.36631894e-01 4.43730563e-01
4.46100742e-01 -5.32118157e-02 7.24270582e-01 -1.13732862e+00
-2.74764061e-01 -8.20859253e-01 -2.22822815e-01 -9.51544106e-01
4.50328946e-01 -9.34568942e-01 2.74562627e-01 -1.26197636e+00
2.25215346e-01 -4.15200710e-01 -1.60708994e-01 5.39175868e-01
-3.34673166e-01 4.55716521e-01 1.24761254e-01 8.51345286e-02
-3.58318955e-01 9.10587236e-03 4.90035087e-01 -2.13230789e-01
-5.04572034e-01 1.61262125e-01 -1.00077450e+00 5.68719566e-01
6.94667995e-01 -1.93491019e-03 -2.34864518e-01 -4.53544796e-01
-4.45406258e-01 -2.46095806e-01 -1.59056094e-02 -1.26247644e+00
2.40134299e-01 8.76597613e-02 1.77926555e-01 -2.53589243e-01
4.62509930e-01 -4.32771057e-01 -5.68561489e-03 1.05228014e-01
-4.55165654e-01 3.64003214e-03 6.44018829e-01 -7.42397234e-02
-3.27175081e-01 1.96243584e-01 1.06794202e+00 2.56151795e-01
-8.64372671e-01 6.27778918e-02 -7.89790690e-01 3.36434931e-01
6.27068341e-01 -6.77528828e-02 -1.42137215e-01 -7.13902295e-01
-6.60197437e-01 1.58570185e-01 5.39904177e-01 7.50605524e-01
-1.16251446e-02 -1.01312602e+00 -9.42708373e-01 4.42588031e-01
3.51519912e-01 -6.35889590e-01 8.53281915e-02 5.09720087e-01
-6.84556365e-02 7.28800654e-01 1.15516260e-01 -5.38827479e-01
-1.69964397e+00 1.38867140e-01 7.35552788e-01 1.75834402e-01
-3.46030623e-01 9.19444561e-01 3.49047363e-01 -8.52161229e-01
6.46317124e-01 1.10368691e-02 -1.80362672e-01 1.12432636e-01
7.37068117e-01 5.71443997e-02 1.74215510e-01 -1.16675270e+00
-7.67615318e-01 4.68273878e-01 -3.57755572e-01 -4.91030842e-01
6.78990960e-01 -3.11397105e-01 2.00598568e-01 8.26168418e-01
1.10304248e+00 5.51365256e-01 -1.02635384e+00 -3.40246797e-01
1.43776774e-01 3.78382318e-02 -2.83268169e-02 -1.15479434e+00
-5.11465490e-01 8.92560601e-01 7.94810772e-01 1.61109999e-01
8.19252074e-01 2.30793267e-01 5.58705389e-01 1.62524179e-01
3.13488781e-01 -1.02125669e+00 -5.66107571e-01 7.23970354e-01
7.36622572e-01 -1.38635480e+00 -4.84813392e-01 -1.69473678e-01
-8.49275172e-01 7.57746160e-01 6.24375582e-01 2.84933478e-01
6.10174596e-01 6.50147915e-01 7.86735833e-01 4.17786419e-01
-6.07577920e-01 -4.19709444e-01 4.93538857e-01 7.18055665e-01
1.00197327e+00 3.70174468e-01 -1.42983079e-01 5.72832644e-01
-3.60701054e-01 -5.62006235e-01 -1.39583424e-02 4.65141207e-01
-4.84821171e-01 -1.37446868e+00 -6.54209316e-01 2.24347368e-01
-5.74993014e-01 -3.28389227e-01 -7.04273820e-01 6.30949438e-01
-6.15474172e-02 1.28197789e+00 1.02823861e-01 -4.81301785e-01
3.64337593e-01 6.99190140e-01 3.63813862e-02 -7.87657619e-01
-8.88058782e-01 2.33066678e-01 4.91280913e-01 -1.33872330e-01
1.75967276e-01 -1.09257793e+00 -9.65218484e-01 -1.17958985e-01
8.90823528e-02 1.11482561e-01 6.45332992e-01 9.19955134e-01
2.68718839e-01 4.82566059e-02 5.10442972e-01 -6.74706817e-01
-5.91247737e-01 -1.07975578e+00 -6.40874982e-01 1.75412938e-01
6.08646631e-01 -3.70747894e-01 -5.21176338e-01 -2.09814683e-01] | [14.19935417175293, 6.628518104553223] |
98381987-01b4-4a55-82b0-a04affe763a7 | towards-good-practices-for-deep-3d-hand-pose | 1707.07248 | null | http://arxiv.org/abs/1707.07248v1 | http://arxiv.org/pdf/1707.07248v1.pdf | Towards Good Practices for Deep 3D Hand Pose Estimation | 3D hand pose estimation from single depth image is an important and
challenging problem for human-computer interaction. Recently deep convolutional
networks (ConvNet) with sophisticated design have been employed to address it,
but the improvement over traditional random forest based methods is not so
apparent. To exploit the good practice and promote the performance for hand
pose estimation, we propose a tree-structured Region Ensemble Network (REN) for
directly 3D coordinate regression. It first partitions the last convolution
outputs of ConvNet into several grid regions. The results from separate
fully-connected (FC) regressors on each regions are then integrated by another
FC layer to perform the estimation. By exploitation of several training
strategies including data augmentation and smooth $L_1$ loss, proposed REN can
significantly improve the performance of ConvNet to localize hand joints. The
experimental results demonstrate that our approach achieves the best
performance among state-of-the-art algorithms on three public hand pose
datasets. We also experiment our methods on fingertip detection and human pose
datasets and obtain state-of-the-art accuracy. | ['Guijin Wang', 'Hengkai Guo', 'Cairong Zhang', 'Xinghao Chen'] | 2017-07-23 | null | null | null | null | ['fingertip-detection'] | ['computer-vision'] | [-2.50894636e-01 -2.14185923e-01 -3.18594903e-01 -2.95864314e-01
-5.40921092e-01 -2.73965508e-01 6.98133484e-02 -9.26958621e-01
-5.10696352e-01 7.06128538e-01 1.67659014e-01 1.38025478e-01
5.60282841e-02 -3.80427986e-01 -7.46240914e-01 -6.54522598e-01
-2.69479543e-01 6.75958812e-01 3.89583260e-01 -1.82395756e-01
1.45415887e-01 9.34452653e-01 -1.22044814e+00 1.05827816e-01
5.07645905e-01 1.10563409e+00 1.72522217e-01 4.94850278e-01
9.43064168e-02 6.29356086e-01 -5.89558959e-01 -1.11140922e-01
3.75557125e-01 1.43968254e-01 -9.22576189e-01 -2.48652294e-01
3.76621753e-01 -7.91442037e-01 -6.23677909e-01 6.20871007e-01
1.18549001e+00 -3.83020043e-02 6.73988819e-01 -9.21924829e-01
-2.06995577e-01 4.86194044e-01 -9.23400044e-01 -1.14790410e-01
4.09825444e-01 2.66059011e-01 7.64087260e-01 -1.02012479e+00
7.77803659e-01 1.50440192e+00 7.13463068e-01 5.35864472e-01
-9.17479873e-01 -8.83971691e-01 4.65277970e-01 3.96112874e-02
-1.47574961e+00 4.01265472e-02 9.57583249e-01 -3.01511317e-01
1.20393467e+00 -1.65043071e-01 6.72187448e-01 1.44706881e+00
3.33637536e-01 1.29526758e+00 1.08008146e+00 -4.31934237e-01
-4.28985268e-01 -3.43972027e-01 1.66851897e-02 9.24225867e-01
-5.17864972e-02 3.65479738e-02 -4.93108153e-01 8.71921405e-02
1.45828903e+00 -9.76959318e-02 -1.44262284e-01 -4.92788285e-01
-1.27808905e+00 5.10072887e-01 1.05805898e+00 1.71079203e-01
-7.03693688e-01 4.33645904e-01 3.30435306e-01 -7.88171869e-03
4.09467727e-01 2.16494188e-01 -7.47819901e-01 3.76461484e-02
-7.10606217e-01 7.83093631e-01 5.77506781e-01 9.34664130e-01
1.67899415e-01 -3.35171789e-01 -4.90152150e-01 8.02754045e-01
4.18063432e-01 3.54645073e-01 -4.63905968e-02 -7.36934066e-01
8.79437685e-01 6.42840803e-01 1.65539399e-01 -6.77712977e-01
-6.97302639e-01 -6.54022694e-01 -7.59703159e-01 6.23720527e-01
9.75724936e-01 -2.12446481e-01 -1.20763016e+00 1.45930922e+00
3.46533448e-01 -4.22712773e-01 -7.08729386e-01 1.21364963e+00
7.46382296e-01 1.78368960e-03 1.59721002e-01 3.28539908e-01
1.29426527e+00 -1.13893354e+00 -5.41433752e-01 -6.62005395e-02
-2.08523124e-02 -7.49635935e-01 1.00311983e+00 6.67719245e-01
-9.55089152e-01 -9.34275389e-01 -8.26755106e-01 -1.10577412e-01
-2.01153338e-01 5.85913360e-01 9.10606802e-01 4.82013166e-01
-7.46915638e-01 7.01455951e-01 -1.10720301e+00 -1.75685864e-02
8.03046465e-01 8.34686697e-01 -3.72880965e-01 1.36926070e-01
-1.03226888e+00 8.96025062e-01 1.25574484e-01 6.98845565e-01
-6.50464475e-01 -3.34314436e-01 -3.98813188e-01 -3.67928952e-01
4.27994341e-01 -6.92746222e-01 9.77749467e-01 -3.01042110e-01
-1.54969001e+00 7.66734600e-01 -1.28662601e-01 2.28393544e-02
9.99060810e-01 -8.57207835e-01 1.78871691e-01 -2.41551884e-02
-5.89459874e-02 9.56870675e-01 1.02143717e+00 -1.12002540e+00
-4.41909492e-01 -9.25092757e-01 -7.02546835e-02 4.83138524e-02
-8.41520801e-02 1.89478219e-01 -7.31488168e-01 -8.80361736e-01
5.25991023e-01 -1.05312514e+00 -3.10571402e-01 9.98746902e-02
-7.66984403e-01 -8.17256391e-01 6.78053677e-01 -1.01925886e+00
1.07738900e+00 -1.39329541e+00 5.88993847e-01 4.56245124e-01
3.87925118e-01 1.80619493e-01 -6.16468191e-02 6.23752326e-02
1.04906902e-01 -4.04141605e-01 1.63295180e-01 -3.54355931e-01
-3.57344747e-02 -2.28838667e-01 -1.26126766e-01 4.55263585e-01
3.58802497e-01 1.14689469e+00 -4.19228673e-01 -6.65457189e-01
2.66754687e-01 9.01072264e-01 -6.59879446e-01 4.29715961e-01
-3.37144285e-01 8.72823358e-01 -7.03201950e-01 9.46524382e-01
6.53509200e-01 -2.11715981e-01 4.44931909e-02 -4.61176336e-01
4.28725593e-02 1.95606112e-01 -1.26589251e+00 2.15191269e+00
-1.37255996e-01 3.35328460e-01 -9.21266004e-02 -5.31492949e-01
9.14858282e-01 2.47237504e-01 5.06854236e-01 -1.66190997e-01
4.93363410e-01 2.02419907e-01 1.04633711e-01 -4.68162894e-01
1.14986993e-01 3.76698881e-01 8.78924727e-02 4.51771289e-01
1.78343713e-01 2.06253514e-01 -5.70349693e-01 -4.14320439e-01
6.80699408e-01 9.68819320e-01 -1.06973402e-01 -1.06369928e-01
4.95753288e-01 -1.81824490e-01 1.87261641e-01 5.99557042e-01
-4.69329119e-01 7.05381215e-01 4.38265294e-01 -6.69415474e-01
-1.07422841e+00 -9.82765138e-01 -9.50716138e-02 1.34783626e+00
-1.23630919e-01 1.86877459e-01 -9.58829522e-01 -8.53142560e-01
4.44889992e-01 -3.34619194e-01 -7.97467828e-01 2.97615975e-01
-1.15429354e+00 -4.53836679e-01 7.63344288e-01 1.27420712e+00
9.52477634e-01 -1.48566091e+00 -5.57876170e-01 3.16318661e-01
-1.83669955e-01 -9.56198156e-01 -4.84976619e-01 1.48959368e-01
-8.18450928e-01 -1.05400372e+00 -1.42814398e+00 -1.06805730e+00
4.33595419e-01 -1.83494985e-01 8.08780491e-01 -2.64265269e-01
-5.54118931e-01 1.12470863e-02 -2.31243342e-01 -4.40364242e-01
3.68026972e-01 7.93029189e-01 2.18874454e-01 -3.86497349e-01
5.02245665e-01 -5.73961079e-01 -1.00137413e+00 4.64508265e-01
2.18493734e-02 -3.97001237e-01 7.90126383e-01 8.19641113e-01
4.45754111e-01 -5.62265456e-01 5.87578118e-01 -3.54287177e-01
7.36529350e-01 3.64440143e-01 -2.77574688e-01 2.55391061e-01
-3.57087702e-01 3.13933581e-01 9.55368802e-02 -6.34168386e-01
-1.13887060e+00 5.06452799e-01 -4.00858670e-01 -5.61025321e-01
-1.25227317e-01 -2.26344354e-02 -3.88034619e-02 -3.30035567e-01
7.04525232e-01 2.62333695e-02 2.91359723e-01 -8.59917700e-01
3.68112624e-01 9.17776942e-01 6.72287285e-01 -7.22414255e-01
5.43370843e-01 3.35586548e-01 3.98274837e-03 -3.71349365e-01
-7.47257173e-01 -2.55161077e-01 -1.29860711e+00 -3.93027157e-01
1.02251267e+00 -8.47135484e-01 -1.28129911e+00 9.16983247e-01
-1.39651871e+00 -2.31292933e-01 3.13824087e-01 5.28939128e-01
-5.54337442e-01 3.42206329e-01 -7.02854633e-01 -1.09271669e+00
-7.56976843e-01 -1.25487471e+00 1.29737008e+00 9.82441530e-02
-2.94284314e-01 -3.89861017e-01 -7.15313554e-02 2.40129203e-01
2.58755207e-01 3.08324784e-01 6.23583496e-01 -2.30446965e-01
-5.33702433e-01 -5.38439512e-01 -4.53086317e-01 3.16907912e-01
1.09622836e-01 -2.73044020e-01 -1.24073565e+00 -5.03323555e-01
-4.91980463e-01 -5.47170997e-01 1.04595232e+00 7.44919658e-01
1.59062469e+00 2.10297957e-01 -5.08257329e-01 5.49160659e-01
8.85069251e-01 -1.55672684e-01 6.15635037e-01 2.91426986e-01
9.86636698e-01 7.72347867e-01 6.01612151e-01 4.36899334e-01
2.67426223e-01 8.57505083e-01 3.90969753e-01 -1.82277247e-01
-2.81830907e-01 -2.07426488e-01 -1.59752406e-02 2.89880872e-01
-1.23933101e+00 2.99258053e-01 -7.15280950e-01 1.80965364e-01
-1.79801011e+00 -6.39833629e-01 1.70140237e-01 1.94443095e+00
8.43291521e-01 2.60826439e-01 7.67692804e-01 2.92313129e-01
6.52795792e-01 2.18630746e-01 -7.35284746e-01 2.22333148e-01
1.66929454e-01 7.38153636e-01 4.38192219e-01 1.02330126e-01
-1.51755357e+00 1.36160123e+00 6.39767885e+00 6.04529262e-01
-1.03838456e+00 -2.64866263e-01 2.62253731e-01 5.44334576e-02
5.24327576e-01 -7.10835516e-01 -9.86961246e-01 -2.27517262e-02
-7.88661316e-02 9.21239138e-01 4.40973401e-01 9.96897697e-01
-1.91894285e-02 1.83802664e-01 -1.04116535e+00 1.01777112e+00
-1.85366094e-01 -7.97330081e-01 -2.18436465e-01 1.38062790e-01
5.47317445e-01 -1.00223953e-02 1.48341238e-01 1.33610964e-01
8.43625590e-02 -1.41171479e+00 6.21490717e-01 6.24956787e-01
1.00461149e+00 -8.20302784e-01 8.02767754e-01 2.45938063e-01
-1.49887025e+00 -1.22434475e-01 -3.67979318e-01 -1.57024309e-01
-9.80806351e-02 2.62615774e-02 -7.34389126e-01 2.33468473e-01
1.01001310e+00 4.34909493e-01 -4.52700317e-01 8.19617093e-01
-3.59388769e-01 6.85892254e-02 -2.21457615e-01 -3.68855506e-01
7.61605799e-02 2.41285905e-01 4.39355075e-01 1.14961886e+00
-2.73950100e-01 -1.30934626e-01 7.70022869e-02 1.02682209e+00
9.49921682e-02 -1.22063883e-01 -3.21881443e-01 1.67077541e-01
2.77192742e-01 9.89120603e-01 -5.11898756e-01 -3.79814045e-03
-1.18003882e-01 1.14922345e+00 5.04175663e-01 3.50283802e-01
-6.27315581e-01 -7.07843840e-01 8.58031929e-01 5.30662686e-02
3.75606209e-01 -6.02145314e-01 -6.13806367e-01 -8.98784876e-01
4.15386021e-01 -6.95087612e-01 1.88521191e-01 -4.58772540e-01
-1.27844298e+00 7.16718316e-01 -1.14754871e-01 -1.11203587e+00
-4.22605842e-01 -1.24365377e+00 -2.15860009e-01 1.27926278e+00
-1.18556988e+00 -1.52761352e+00 -6.93806589e-01 8.82634401e-01
4.63427335e-01 -3.77998412e-01 8.87955308e-01 4.98946905e-02
-3.86744559e-01 9.89605606e-01 -5.08344233e-01 8.02524805e-01
7.59417951e-01 -1.17108822e+00 6.74518406e-01 2.83461213e-01
-1.49412006e-01 7.97007918e-01 2.49222174e-01 -7.45047450e-01
-1.38255882e+00 -7.68366158e-01 7.19433665e-01 -7.30368137e-01
1.22960471e-01 -4.05624866e-01 -4.45073783e-01 6.87968910e-01
-1.45963103e-01 4.56734337e-02 9.70605612e-02 4.06004786e-01
-6.03426337e-01 4.04002257e-02 -1.24243081e+00 5.33263624e-01
1.65570068e+00 -4.53023046e-01 -6.55628979e-01 1.84848875e-01
5.38227737e-01 -7.30491340e-01 -1.05848157e+00 6.54749513e-01
1.45564866e+00 -6.90961897e-01 1.35630870e+00 -1.03404880e+00
5.09411633e-01 -3.08027887e-03 2.03786828e-02 -8.04745197e-01
-4.16895866e-01 -3.31720442e-01 -2.82683611e-01 6.73499584e-01
1.75931782e-01 -4.20882702e-01 1.40392983e+00 3.15828621e-01
2.55941421e-01 -1.06565702e+00 -8.40540648e-01 -6.79181457e-01
3.73108625e-01 -3.70442837e-01 5.98819613e-01 2.86761403e-01
-1.28199577e-01 1.30175933e-01 -4.47861582e-01 1.09730521e-02
9.75880802e-01 6.28601536e-02 9.56583083e-01 -1.33812022e+00
-1.77241132e-01 -4.86199319e-01 -3.91939908e-01 -1.51347995e+00
1.81128129e-01 -5.14272511e-01 9.77403596e-02 -1.40602064e+00
1.79656833e-01 -4.58788216e-01 -2.74423331e-01 5.86197317e-01
-2.31747240e-01 4.49956387e-01 1.10428892e-01 3.69916111e-01
-2.02483237e-01 3.04545254e-01 1.89306509e+00 -1.58896104e-01
-5.19322515e-01 5.22585988e-01 -1.81738287e-01 5.91413438e-01
6.19516790e-01 4.65518869e-02 1.43847466e-01 -2.31894702e-01
-3.43244284e-01 -3.64232212e-02 6.22432053e-01 -1.00838804e+00
9.40817371e-02 1.87428728e-01 1.10852289e+00 -1.13392580e+00
3.71639490e-01 -6.74183011e-01 -3.64476413e-01 7.34851599e-01
-3.28961521e-01 -2.12204188e-01 -6.28065318e-02 2.44625524e-01
1.00499637e-01 3.96396309e-01 6.74713850e-01 -2.89868951e-01
-6.72740757e-01 6.07514441e-01 1.43144161e-01 -3.43095720e-01
6.00567877e-01 -2.58079141e-01 1.45596832e-01 -1.96878150e-01
-9.44195986e-01 1.14550903e-01 -2.48887897e-01 5.38065851e-01
7.00103462e-01 -1.35543513e+00 -7.33455837e-01 2.92841166e-01
-6.47535324e-02 3.22239488e-01 -6.68250397e-02 7.66192198e-01
-5.86625695e-01 8.01087558e-01 -6.13394916e-01 -8.41421664e-01
-1.22127581e+00 1.99808821e-01 4.73118216e-01 -4.17903066e-01
-7.50085890e-01 1.34406376e+00 -1.73854083e-01 -8.15837562e-01
1.04569411e+00 -5.19956470e-01 -2.73252428e-01 -2.03212306e-01
3.37033212e-01 5.59055448e-01 1.14190187e-02 -3.57852101e-01
-5.65295756e-01 9.94900703e-01 -2.20278695e-01 -1.28340991e-02
1.40113449e+00 3.89765739e-01 -7.22032487e-02 -3.14041004e-02
1.14210045e+00 -4.12020683e-01 -1.69737840e+00 -2.52843738e-01
-2.41069347e-01 -4.90501404e-01 -2.10981891e-01 -1.12234306e+00
-1.19623828e+00 1.06751001e+00 9.07006085e-01 -5.38301468e-01
9.12216842e-01 1.95850581e-01 7.48950124e-01 5.53705990e-01
7.49812543e-01 -1.19121444e+00 2.99206376e-01 5.62353730e-01
1.33639216e+00 -1.16026402e+00 -3.42761949e-02 -5.78259885e-01
-3.66359591e-01 1.32831967e+00 9.20380235e-01 -6.57447815e-01
8.29329371e-01 2.67125309e-01 1.96771428e-01 -1.22257091e-01
-7.31815631e-03 -3.75040919e-01 7.36627460e-01 7.28732288e-01
7.38075495e-01 1.13939069e-01 -2.46094495e-01 7.73185492e-01
-1.34009063e-01 3.11749548e-01 -6.07933640e-01 9.34664011e-01
-3.38199377e-01 -1.22546828e+00 -5.53794444e-01 4.96045262e-01
-4.29249555e-01 2.92821288e-01 -6.52745128e-01 1.06831145e+00
8.26247707e-02 4.05665189e-01 -1.19780906e-01 -9.01219368e-01
5.53136230e-01 2.31761649e-01 1.28116071e+00 -2.04707816e-01
-8.11599433e-01 3.27728659e-01 -2.27170095e-01 -7.73345351e-01
-3.72741640e-01 -4.39855993e-01 -1.13413143e+00 -2.54835904e-01
-4.87924248e-01 -5.13930559e-01 5.98602235e-01 9.28685248e-01
3.33694220e-02 7.37693846e-01 4.93488550e-01 -1.62586129e+00
-1.10318780e+00 -1.49960160e+00 -6.24706209e-01 -2.98797674e-02
1.62533477e-01 -1.28134418e+00 1.77953199e-01 -3.74334723e-01] | [6.615572929382324, -0.6621056199073792] |
de4fe771-b516-4ecc-8f8f-17d76b50eac8 | estimating-treatment-effects-from-irregular-1 | 2303.02320 | null | https://arxiv.org/abs/2303.02320v1 | https://arxiv.org/pdf/2303.02320v1.pdf | Estimating Treatment Effects from Irregular Time Series Observations with Hidden Confounders | Causal analysis for time series data, in particular estimating individualized treatment effect (ITE), is a key task in many real-world applications, such as finance, retail, healthcare, etc. Real-world time series can include large-scale, irregular, and intermittent time series observations, raising significant challenges to existing work attempting to estimate treatment effects. Specifically, the existence of hidden confounders can lead to biased treatment estimates and complicate the causal inference process. In particular, anomaly hidden confounders which exceed the typical range can lead to high variance estimates. Moreover, in continuous time settings with irregular samples, it is challenging to directly handle the dynamics of causality. In this paper, we leverage recent advances in Lipschitz regularization and neural controlled differential equations (CDE) to develop an effective and scalable solution, namely LipCDE, to address the above challenges. LipCDE can directly model the dynamic causal relationships between historical data and outcomes with irregular samples by considering the boundary of hidden confounders given by Lipschitz-constrained neural networks. Furthermore, we conduct extensive experiments on both synthetic and real-world datasets to demonstrate the effectiveness and scalability of LipCDE. | ['Yan Liu', 'Hao Niu', 'Chuizheng Meng', 'Xiangchen Song', 'Yujing Wang', 'James Enouen', 'Defu Cao'] | 2023-03-04 | null | null | null | null | ['irregular-time-series'] | ['time-series'] | [ 2.03678861e-01 -2.69633830e-01 -5.54087102e-01 -1.57285497e-01
-7.05875993e-01 -1.42506048e-01 1.86301142e-01 1.69438317e-01
-4.88491133e-02 1.17894006e+00 6.28535509e-01 -4.78563130e-01
-5.38111091e-01 -7.31744051e-01 -7.66871095e-01 -7.63716102e-01
-3.75431508e-01 1.16467141e-01 -4.67200875e-01 1.11713208e-01
-4.41585435e-03 1.24770150e-01 -9.21906471e-01 -2.72727937e-01
1.14602590e+00 6.01627171e-01 -2.18577936e-01 8.61400440e-02
2.73584902e-01 5.91442049e-01 -2.73898691e-01 2.70725135e-02
-1.38525143e-02 -2.33721614e-01 -3.66901338e-01 -1.90946430e-01
-4.34145518e-02 -4.00522411e-01 -1.62401363e-01 7.85731673e-01
5.95535040e-01 1.39950112e-01 6.66923106e-01 -1.29188466e+00
-7.22720921e-01 6.98206842e-01 -9.00384605e-01 2.64464498e-01
7.80182555e-02 2.40379497e-01 7.48193026e-01 -5.54702044e-01
3.56994718e-01 1.53651059e+00 8.83746028e-01 4.09260184e-01
-1.36892807e+00 -9.23537612e-01 5.21309018e-01 -6.66842163e-02
-9.72947776e-01 -2.00827435e-01 6.70017064e-01 -5.48815787e-01
2.67189950e-01 1.05658129e-01 4.23829734e-01 1.47589993e+00
3.16381812e-01 6.22722805e-01 1.01411021e+00 9.05568078e-02
3.47311258e-01 -5.29055178e-01 1.53622359e-01 6.96114972e-02
3.24431270e-01 4.35638428e-01 -1.39615238e-01 -5.01276135e-01
9.66358125e-01 4.14509833e-01 -3.08776200e-01 8.13950878e-03
-1.23865414e+00 9.92667913e-01 3.18886369e-01 -9.99975763e-03
-7.61433065e-01 4.15085852e-01 4.10108060e-01 1.87865511e-01
7.53403664e-01 5.58439679e-02 -4.36775416e-01 1.08309025e-02
-3.95863682e-01 3.07447344e-01 4.11005378e-01 5.94308019e-01
9.89507735e-02 -3.10089383e-02 -4.00189042e-01 5.00841260e-01
1.21589772e-01 6.32379055e-01 2.11840793e-01 -8.13826978e-01
6.26359522e-01 3.85480046e-01 4.76717770e-01 -1.17250431e+00
-5.78168571e-01 -4.04963940e-01 -1.39400935e+00 -5.00782967e-01
6.81258142e-01 -6.14530206e-01 -6.13871992e-01 2.18148994e+00
6.82500482e-01 8.45084727e-01 -3.18391025e-01 1.00224507e+00
3.51060688e-01 5.77427924e-01 3.95143062e-01 -7.54766405e-01
9.74813223e-01 -2.33810261e-01 -1.06976879e+00 8.06921273e-02
6.32605314e-01 -3.30426872e-01 1.04804838e+00 6.02039471e-02
-1.03824472e+00 -1.04796946e-01 -4.64633107e-01 5.62438890e-02
-7.34392330e-02 -1.17841490e-01 7.30785728e-01 1.49626240e-01
-3.69541436e-01 6.07279837e-01 -8.86996627e-01 -4.84566465e-02
4.64872867e-01 4.88303393e-01 3.38613689e-02 -1.75461859e-01
-1.61818492e+00 2.16891333e-01 -9.03867260e-02 5.14721870e-01
-9.75294530e-01 -1.49068916e+00 -6.27043188e-01 1.10773236e-01
6.54895425e-01 -7.68043518e-01 1.06332612e+00 -5.67474186e-01
-1.07179046e+00 1.31595314e-01 -2.25600764e-01 -2.63802469e-01
7.67936945e-01 -1.53156087e-01 -4.18352515e-01 -4.71862197e-01
1.86426267e-01 -7.49082416e-02 7.08568335e-01 -6.92075253e-01
-4.61900264e-01 -6.25935316e-01 -2.09626853e-01 -1.83567747e-01
-2.60136902e-01 -1.04146607e-01 3.33317667e-02 -8.32567334e-01
-1.65891245e-01 -8.60397100e-01 -3.94490570e-01 -1.66464418e-01
-4.35499370e-01 -3.46615702e-01 6.96147323e-01 -6.38130188e-01
1.46782398e+00 -2.30804753e+00 1.75863042e-01 3.14224605e-03
2.76394516e-01 -1.46258295e-01 5.51178232e-02 3.89141291e-01
-2.88376689e-01 2.40128547e-01 -1.89059377e-01 -1.42515644e-01
-1.03097513e-01 3.17985147e-01 -4.70141023e-01 7.92642176e-01
1.85842514e-01 8.55634034e-01 -1.12554419e+00 -2.74590760e-01
-5.68004809e-02 5.31796396e-01 -5.71412086e-01 1.14704512e-01
-1.79449141e-01 8.61640692e-01 -8.91967654e-01 2.96242088e-01
4.69433844e-01 -6.07403278e-01 2.00864792e-01 1.20595068e-01
-2.67655998e-01 1.00603230e-01 -1.34772694e+00 1.10065234e+00
-6.03790462e-01 4.94905174e-01 1.23390868e-01 -1.34099054e+00
3.07823181e-01 6.38471186e-01 7.17144608e-01 -5.37603617e-01
1.80876017e-01 1.59870163e-01 -1.44279867e-01 -7.23804653e-01
4.91984263e-02 -4.89143163e-01 -5.93581684e-02 3.60582858e-01
-7.46324420e-01 6.56938136e-01 -1.42461970e-01 -4.53191884e-02
1.09360266e+00 -4.15542126e-01 1.26611888e-01 -2.58324653e-01
1.89654365e-01 -3.09136361e-01 1.12583184e+00 5.02376199e-01
-1.39215514e-01 3.61466855e-01 9.34325099e-01 -3.92443687e-01
-6.88733697e-01 -8.36111307e-01 -3.23219359e-01 6.66952729e-01
1.83330500e-03 2.19561994e-01 -3.72791678e-01 -4.12070423e-01
3.65591556e-01 4.84924138e-01 -8.91139269e-01 -3.22636575e-01
-6.64995074e-01 -1.26934862e+00 8.54163617e-02 6.82888687e-01
4.65948433e-02 -8.83133173e-01 -2.80365169e-01 6.09214544e-01
-5.27719975e-01 -8.66658151e-01 -8.12188864e-01 -1.87795877e-01
-9.76577997e-01 -1.16928816e+00 -7.08025753e-01 -3.32099676e-01
5.45747042e-01 1.61828667e-01 9.66441154e-01 -1.19864143e-01
-3.26402217e-01 1.85018912e-01 1.63179025e-01 -5.73734939e-01
-1.09712064e-01 -1.53596923e-01 8.68132859e-02 3.23332399e-01
1.25593558e-01 -5.94487667e-01 -8.06373835e-01 2.78524727e-01
-8.95886064e-01 -1.70207307e-01 1.94082141e-01 1.14890468e+00
6.60756230e-01 1.21441700e-01 1.32169008e+00 -1.07062173e+00
7.12778449e-01 -9.88839149e-01 -9.14719939e-01 1.85249910e-01
-7.06156790e-01 -1.32763967e-01 5.77841103e-01 -9.82967436e-01
-1.09929979e+00 -3.59866709e-01 3.35622877e-01 -4.34668899e-01
9.99011546e-02 1.01876760e+00 -4.53745276e-02 5.39726019e-01
2.94336051e-01 -3.98094773e-01 -1.53392494e-01 -3.81193399e-01
1.38086170e-01 5.88120282e-01 3.44137758e-01 -6.16268098e-01
2.59507209e-01 6.26371562e-01 1.46692157e-01 -5.27010918e-01
-9.41537976e-01 -3.03267568e-01 -8.79126936e-02 6.34661093e-02
8.47175062e-01 -1.00995266e+00 -1.12697387e+00 3.26644540e-01
-9.69942272e-01 -6.85537398e-01 -1.10910267e-01 7.43446112e-01
-4.00075823e-01 -3.75775024e-02 -7.68406510e-01 -9.12350714e-01
-2.06311896e-01 -1.06693578e+00 1.01420033e+00 1.44118056e-01
-2.16305479e-01 -1.49334061e+00 1.64549157e-01 1.16479382e-01
1.52320191e-01 7.75156379e-01 1.31302595e+00 1.57357715e-02
-4.10823077e-01 -2.52397001e-01 -3.13732177e-01 -2.24641785e-01
3.06376547e-01 8.85425732e-02 -5.20418584e-01 -2.81621933e-01
5.34850247e-02 -6.21989258e-02 4.83154833e-01 1.20487463e+00
1.39332902e+00 -3.72209996e-01 -5.11197209e-01 5.68980932e-01
1.25337851e+00 1.80870384e-01 3.30712616e-01 -1.81044161e-01
6.83329999e-01 6.66615963e-01 6.24471426e-01 9.19408858e-01
5.17264903e-01 5.03231645e-01 3.81446838e-01 -3.04944277e-01
3.74407291e-01 -3.60834688e-01 6.65483102e-02 6.75020695e-01
1.26170591e-02 -2.12866694e-01 -8.19029272e-01 8.29899311e-01
-2.11111236e+00 -9.26622033e-01 -6.86829031e-01 2.42544079e+00
9.58452404e-01 -1.16263576e-01 1.55675128e-01 -1.36311680e-01
9.60935771e-01 -2.19965115e-01 -8.94066513e-01 -6.21377397e-03
-2.80682873e-02 -6.09137081e-02 6.64427817e-01 2.73370832e-01
-1.04360425e+00 3.65648538e-01 6.26849461e+00 5.00862598e-01
-1.35184085e+00 2.44164407e-01 7.70837665e-01 -1.51724100e-01
-2.58683592e-01 -8.01707655e-02 -5.21058202e-01 7.57428110e-01
1.07114053e+00 -4.57138926e-01 4.49197739e-01 2.59583265e-01
1.12692654e+00 1.62313014e-01 -1.07146597e+00 6.88858807e-01
-6.95855618e-01 -1.04943645e+00 -5.67394853e-01 4.10535485e-01
1.03687620e+00 -8.05426762e-02 3.30159396e-01 1.81322351e-01
5.73868811e-01 -1.13517952e+00 2.15655595e-01 3.25893611e-01
8.44398677e-01 -5.91145158e-01 6.22142851e-01 2.95551091e-01
-1.00357413e+00 -3.75470400e-01 -3.69668186e-01 -2.87249923e-01
4.86761332e-01 9.17959988e-01 -4.18512642e-01 4.89976704e-01
5.30226290e-01 1.16266751e+00 1.69035479e-01 9.05624390e-01
-1.04854658e-01 9.69799638e-01 -3.84225398e-01 2.23745108e-01
4.89949360e-02 -2.67141819e-01 3.66156816e-01 6.36384130e-01
4.15700406e-01 3.20324212e-01 1.22470982e-01 8.14638913e-01
-1.37405276e-01 -9.68993548e-03 -3.86825830e-01 -9.78697240e-02
3.96490484e-01 7.35573173e-01 -3.79268676e-01 -1.67405352e-01
-4.68356162e-01 2.94060409e-01 1.85014069e-01 7.63936698e-01
-1.20231056e+00 1.36934385e-01 7.29816377e-01 1.50392637e-01
-1.34512708e-01 -1.21518083e-01 -4.29177880e-01 -1.29334998e+00
-6.97272569e-02 -8.91912758e-01 8.24534416e-01 -3.13443244e-01
-1.61358988e+00 -2.57933289e-01 1.10518374e-02 -9.94737566e-01
-2.01288313e-02 -9.07242447e-02 -7.13693857e-01 9.14275885e-01
-1.40736830e+00 -7.51978159e-01 -1.56716019e-01 7.70896792e-01
3.69488776e-01 6.80636108e-01 4.23948199e-01 6.79344416e-01
-1.25869191e+00 3.38373840e-01 3.80302161e-01 -1.06191836e-01
8.76546919e-01 -1.13043320e+00 8.24287385e-02 5.51362932e-01
-5.62964916e-01 5.54281831e-01 6.76834047e-01 -9.90533054e-01
-1.53895211e+00 -1.38131762e+00 6.90364301e-01 -2.63480604e-01
1.02589238e+00 -2.66388506e-01 -1.19038522e+00 7.21349776e-01
-2.39895046e-01 1.39810294e-01 4.70782250e-01 4.92654562e-01
7.09774122e-02 -9.96071026e-02 -9.30372119e-01 7.26444125e-01
1.07202876e+00 -2.19130129e-01 -1.43270120e-01 5.98836958e-01
7.66869247e-01 -2.94296861e-01 -1.14121854e+00 4.59953308e-01
5.15352309e-01 -3.44785392e-01 9.77082789e-01 -1.10753119e+00
6.67054832e-01 1.58185527e-01 3.97579074e-01 -1.50130916e+00
-2.01169610e-01 -7.66343892e-01 -1.90996259e-01 1.25149345e+00
3.20465535e-01 -9.51813340e-01 4.41969812e-01 9.80329275e-01
2.97930837e-01 -5.63372135e-01 -9.34080601e-01 -7.39924848e-01
3.97419363e-01 -4.05194968e-01 8.91400397e-01 1.26147306e+00
-7.53474385e-02 1.37566701e-01 -7.34737635e-01 4.02100652e-01
8.89140368e-01 2.10762158e-01 4.49016362e-01 -1.26984513e+00
-1.29223228e-01 -3.17368150e-01 9.09873694e-02 -7.00440228e-01
3.79560769e-01 -3.58824015e-01 5.85323721e-02 -1.28110647e+00
5.95151447e-02 -8.77844036e-01 -3.95054489e-01 2.25948021e-01
-7.39551544e-01 -3.86434644e-01 -2.47879833e-01 4.48365845e-02
-6.84451833e-02 8.21188867e-01 1.38081217e+00 -1.06008984e-01
-4.10114080e-01 2.61966437e-01 -4.79603559e-01 4.70116854e-01
6.26150548e-01 -7.79205978e-01 -6.12805545e-01 -4.54925805e-01
3.08525622e-01 7.74831414e-01 4.84539151e-01 -1.44464269e-01
1.54507449e-02 -8.51449192e-01 3.59767191e-02 -2.14216769e-01
-2.03322291e-01 -8.32343161e-01 4.25174385e-01 5.24461031e-01
-5.69175482e-01 1.19844779e-01 1.29982121e-02 9.93407667e-01
-1.30591854e-01 3.03297549e-01 4.22393411e-01 2.12040454e-01
-5.03158644e-02 6.21056914e-01 -1.74552739e-01 4.77833837e-01
8.29960883e-01 4.67890769e-01 -3.23586702e-01 -5.31272769e-01
-6.93149507e-01 7.79827535e-01 -4.71912045e-03 3.68692189e-01
2.65174031e-01 -1.42205954e+00 -7.67365694e-01 -4.99529876e-02
-1.02253899e-01 2.36424893e-01 6.65183425e-01 1.46234703e+00
2.58308649e-01 3.27108949e-01 4.87969458e-01 -4.38885033e-01
-8.72046173e-01 1.00120008e+00 2.00375184e-01 -3.00804794e-01
-7.35390604e-01 2.46848390e-01 6.97966933e-01 -1.54952168e-01
3.05381685e-01 -7.00156808e-01 -1.00833528e-01 1.01319671e-01
5.26800275e-01 6.16237819e-01 -3.02801222e-01 -7.85790682e-02
-2.18622200e-02 3.74465972e-01 3.03566217e-01 2.17077449e-01
1.49142504e+00 -2.00923219e-01 -4.48548682e-02 6.25085652e-01
1.22081947e+00 -2.36266032e-01 -1.46763349e+00 -2.53805965e-01
3.71666215e-02 -2.52250612e-01 2.14641318e-01 -4.75258708e-01
-1.16788363e+00 8.69315922e-01 5.00809252e-01 4.34102356e-01
1.14076269e+00 -2.34727532e-01 8.50775063e-01 -2.10726559e-01
1.79093242e-01 -8.94055307e-01 -2.62135088e-01 -2.06451723e-03
8.17142308e-01 -1.30333591e+00 -1.82628185e-01 -5.27388632e-01
-2.43069008e-01 7.36023188e-01 2.89064437e-01 -1.70265973e-01
9.14714396e-01 1.61470681e-01 -8.85998234e-02 -4.55544814e-02
-9.50215995e-01 2.04097047e-01 2.04105601e-01 1.35482475e-01
4.93135959e-01 4.18874234e-01 -5.99594474e-01 6.61718845e-01
3.17551166e-01 4.13249582e-01 5.84031224e-01 7.22916126e-01
4.94739145e-01 -8.24213088e-01 -5.05786777e-01 4.88835394e-01
-8.33580554e-01 7.99481124e-02 1.57178432e-01 7.44851887e-01
-2.13593647e-01 1.01418519e+00 9.26473886e-02 3.50736588e-01
4.53786671e-01 -2.57453412e-01 -3.00496500e-02 -1.77933961e-01
-3.07045192e-01 4.29948241e-01 -2.60993659e-01 -4.52293843e-01
-4.05977905e-01 -1.05301905e+00 -1.20947373e+00 -5.20144403e-01
-4.09933597e-01 -6.88872635e-02 4.39366311e-01 1.15057468e+00
3.68098736e-01 8.61619353e-01 7.97899365e-01 -4.00100142e-01
-8.93179715e-01 -8.25350821e-01 -4.79205757e-01 5.55255830e-01
7.86618233e-01 -9.20308948e-01 -5.40673077e-01 1.79073587e-02] | [7.938438892364502, 5.299957275390625] |
7346ab2d-2157-4acf-b562-7a2e6e4f8eb0 | shuttleset-a-human-annotated-stroke-level | 2306.04948 | null | https://arxiv.org/abs/2306.04948v1 | https://arxiv.org/pdf/2306.04948v1.pdf | ShuttleSet: A Human-Annotated Stroke-Level Singles Dataset for Badminton Tactical Analysis | With the recent progress in sports analytics, deep learning approaches have demonstrated the effectiveness of mining insights into players' tactics for improving performance quality and fan engagement. This is attributed to the availability of public ground-truth datasets. While there are a few available datasets for turn-based sports for action detection, these datasets severely lack structured source data and stroke-level records since these require high-cost labeling efforts from domain experts and are hard to detect using automatic techniques. Consequently, the development of artificial intelligence approaches is significantly hindered when existing models are applied to more challenging structured turn-based sequences. In this paper, we present ShuttleSet, the largest publicly-available badminton singles dataset with annotated stroke-level records. It contains 104 sets, 3,685 rallies, and 36,492 strokes in 44 matches between 2018 and 2021 with 27 top-ranking men's singles and women's singles players. ShuttleSet is manually annotated with a computer-aided labeling tool to increase the labeling efficiency and effectiveness of selecting the shot type with a choice of 18 distinct classes, the corresponding hitting locations, and the locations of both players at each stroke. In the experiments, we provide multiple benchmarks (i.e., stroke influence, stroke forecasting, and movement forecasting) with baselines to illustrate the practicability of using ShuttleSet for turn-based analytics, which is expected to stimulate both academic and sports communities. Over the past two years, a visualization platform has been deployed to illustrate the variability of analysis cases from ShuttleSet for coaches to delve into players' tactical preferences with human-interactive interfaces, which was also used by national badminton teams during multiple international high-ranking matches. | ['Wen-Chih Peng', 'Tsi-Ui Ik', 'Yung-Chang Huang', 'Wei-Yao Wang'] | 2023-06-08 | null | null | null | null | ['sports-analytics', 'action-detection'] | ['computer-vision', 'computer-vision'] | [-1.75691187e-01 -1.43790483e-01 -5.77922761e-01 3.71428989e-02
-8.20171893e-01 -6.98843896e-01 3.40109527e-01 3.58624637e-01
-4.51780409e-01 3.56990904e-01 7.90865362e-01 -8.88204277e-02
-3.51353705e-01 -9.65053916e-01 -4.50399280e-01 -1.22776903e-01
-2.54858613e-01 5.74701190e-01 3.98771346e-01 -6.09208643e-01
4.24157560e-01 3.79656613e-01 -1.95263839e+00 6.88574493e-01
6.39740884e-01 6.94362998e-01 -1.69927821e-01 5.24881899e-01
1.29803479e-01 9.39119995e-01 -7.93134093e-01 -6.97168648e-01
7.13682890e-01 -5.85230112e-01 -4.51000810e-01 -3.19789976e-01
5.79595447e-01 -1.90864161e-01 -5.89344203e-01 2.69262552e-01
6.61940336e-01 3.90437454e-01 2.14524925e-01 -1.23280716e+00
2.76428424e-02 7.59135783e-01 -6.81073308e-01 5.91195643e-01
3.94415289e-01 6.57483876e-01 1.04058063e+00 -4.13557768e-01
8.31970274e-01 8.93463135e-01 1.01823449e+00 7.27962479e-02
-8.27326000e-01 -8.41485858e-01 -7.72162229e-02 4.96343106e-01
-1.03713465e+00 -1.18554346e-01 6.75685287e-01 -8.87350142e-01
4.38598990e-01 4.28713530e-01 1.39375699e+00 1.00857460e+00
-1.53356731e-01 9.01703715e-01 1.04434144e+00 -2.06703395e-01
2.68832874e-02 -2.97069013e-01 3.17293286e-01 5.65835774e-01
1.00407831e-01 3.34431052e-01 -1.21158779e+00 1.44501135e-01
6.21082008e-01 -5.69566563e-02 3.08257401e-01 -2.09497854e-01
-1.24992573e+00 7.44560301e-01 4.71243829e-01 2.39468049e-02
-3.25018317e-01 4.92483331e-03 7.36073971e-01 -7.30151236e-02
4.55349594e-01 7.10467875e-01 6.70040920e-02 -1.06922805e+00
-1.16731000e+00 8.81664217e-01 5.71053088e-01 4.41877842e-01
2.42478862e-01 8.10035765e-02 -6.32390738e-01 6.98981404e-01
-5.03498852e-01 1.42948166e-01 1.16329193e-01 -1.09719229e+00
6.78023934e-01 1.01247656e+00 1.06006496e-01 -1.38137090e+00
-7.74800360e-01 -7.10013092e-01 -2.83194214e-01 3.20482016e-01
8.68871987e-01 -1.36373058e-01 -6.70293272e-01 1.38717175e+00
3.31746131e-01 1.32720307e-01 -4.28313226e-01 1.34532452e+00
8.44551384e-01 2.33263478e-01 2.26032343e-02 3.98757577e-01
1.37385273e+00 -6.99438214e-01 -3.41744095e-01 -1.93789765e-01
6.75749958e-01 -3.65494370e-01 1.33156300e+00 5.52258134e-01
-1.05312419e+00 -7.08554804e-01 -9.30783570e-01 5.79675809e-02
-2.67723739e-01 2.93089628e-01 8.83248270e-01 5.53772330e-01
-1.90912351e-01 5.15522182e-01 -8.40503216e-01 -1.25955239e-01
6.90403461e-01 -8.43373388e-02 -1.65019110e-01 3.02716911e-01
-1.27491713e+00 9.39875185e-01 9.85891595e-02 4.63318080e-02
-1.00829852e+00 -1.20519710e+00 -5.83547294e-01 -1.11626908e-02
7.93325663e-01 -1.78252250e-01 1.05509555e+00 -6.84837282e-01
-1.11624801e+00 9.46652174e-01 4.82728302e-01 -5.88520467e-01
9.60238159e-01 -4.86597478e-01 -2.89749742e-01 -6.07077265e-03
5.00010610e-01 3.02679300e-01 -1.53313741e-01 -7.52184689e-01
-9.98488545e-01 -2.62866229e-01 2.92896360e-01 4.12637919e-01
-1.86899051e-01 2.22079650e-01 -4.50323582e-01 -6.80584729e-01
-1.82258427e-01 -9.06562746e-01 -8.78287703e-02 -2.85940170e-01
-4.00037348e-01 -1.97831944e-01 3.23630005e-01 -9.48926687e-01
1.64306557e+00 -2.05294681e+00 1.67147368e-02 1.14321336e-01
1.80526346e-01 2.56032914e-01 2.82657117e-01 7.54646242e-01
1.29409030e-01 2.64510866e-02 1.54939309e-01 1.20584220e-01
5.74438572e-02 -1.06676100e-02 -3.48553509e-01 3.66345823e-01
-3.07237059e-01 8.31556082e-01 -8.88264298e-01 -4.23684716e-01
4.18844491e-01 -2.14199293e-02 -6.48869276e-01 -5.79020679e-02
1.19640760e-01 3.69717091e-01 -2.45328724e-01 7.82041073e-01
-7.77615830e-02 4.02177632e-01 4.02224362e-02 -2.39395536e-02
-6.54354215e-01 3.82076800e-01 -1.17695642e+00 1.71749890e+00
-2.28676442e-02 8.99892628e-01 -1.09865353e-01 -8.43264639e-01
9.81348217e-01 -2.38895610e-01 8.93298149e-01 -8.52900624e-01
-4.98051941e-02 -8.01847503e-02 2.56334901e-01 -5.74630082e-01
7.33846366e-01 1.20886192e-02 -6.29327953e-01 4.39500362e-01
-2.95866936e-01 2.94077456e-01 7.95362592e-01 3.98493439e-01
1.27424407e+00 3.00459236e-01 -2.24552482e-01 -3.59541699e-02
-3.23787540e-01 7.28654683e-01 8.68022680e-01 9.09947813e-01
-2.61787683e-01 6.03567064e-01 6.75424457e-01 -9.63744402e-01
-1.08513749e+00 -9.24951077e-01 2.46034309e-01 1.41410816e+00
8.17380175e-02 -8.05855989e-01 -7.30435848e-01 -2.83731133e-01
1.17227413e-01 4.96407449e-01 -7.41843700e-01 -2.89478809e-01
-7.30086267e-01 -4.39104319e-01 1.10459995e+00 7.79166877e-01
5.04657149e-01 -1.19333684e+00 -1.27404177e+00 1.66355386e-01
-5.19778430e-01 -7.21864164e-01 -3.22638631e-01 -2.49988839e-01
-4.34122473e-01 -1.55691516e+00 -4.99140590e-01 -2.06940040e-01
-1.46935105e-01 6.02912940e-02 1.35016918e+00 2.62735840e-02
-7.56843686e-01 9.66065153e-02 -4.10434216e-01 -7.05671132e-01
4.18516695e-02 4.23175007e-01 1.24164298e-01 -2.24848822e-01
3.78068566e-01 -3.98441046e-01 -7.67610967e-01 5.11246800e-01
-4.04626191e-01 4.59920406e-01 5.10959625e-01 4.79683220e-01
3.86663496e-01 -1.98998407e-01 5.35078943e-01 -6.38898551e-01
6.42681420e-01 -3.71975273e-01 -3.33440065e-01 1.14087090e-01
-3.24364185e-01 -5.47427773e-01 1.23152211e-01 -3.96585554e-01
-7.90428102e-01 -1.38384253e-01 2.06067473e-01 -3.00525635e-01
-1.52309984e-01 6.74084067e-01 1.76943541e-01 3.99788886e-01
1.12042463e+00 -3.26655395e-02 6.69488013e-02 -6.50126874e-01
3.25884342e-01 4.97642994e-01 1.00075591e+00 -6.82946682e-01
7.53858864e-01 3.20245892e-01 -1.64131969e-01 -4.70751286e-01
-1.10873413e+00 -3.52866143e-01 -7.24924266e-01 -1.07763672e+00
8.32560897e-01 -1.00149739e+00 -1.21668911e+00 5.29907286e-01
-3.25289130e-01 -4.20188010e-01 -6.01878464e-01 5.41209102e-01
-4.13623184e-01 5.70187047e-02 -6.02425814e-01 -7.30908215e-01
-1.12273328e-01 -7.82864213e-01 6.45482123e-01 4.36693221e-01
-5.27731240e-01 -3.10522377e-01 2.20055744e-01 1.10503101e+00
6.61370754e-02 6.65063798e-01 5.51365316e-01 -5.79899490e-01
-2.94202119e-01 -5.30670345e-01 1.66524485e-01 -1.24346934e-01
-1.80003807e-01 -5.86508699e-02 -4.48778957e-01 8.67070183e-02
-9.38390672e-01 -4.74419057e-01 6.20373726e-01 4.81484711e-01
1.12897837e+00 9.43062380e-02 -2.11171016e-01 4.02322888e-01
6.39476955e-01 1.96251436e-03 5.34579456e-01 7.78884828e-01
7.24716783e-01 7.86948502e-01 1.13850880e+00 6.05801165e-01
4.04268175e-01 9.76148903e-01 3.03072006e-01 -3.49637061e-01
-2.81610012e-01 -7.79675841e-01 1.24284893e-01 3.25226784e-01
-7.50828981e-01 1.09271571e-01 -1.13384151e+00 7.06326842e-01
-1.78626192e+00 -1.52261281e+00 -3.26947153e-01 2.18421483e+00
7.82689571e-01 4.52656448e-01 9.30312872e-01 2.57546067e-01
6.21072829e-01 1.68818578e-01 -6.48552835e-01 -2.35171735e-01
1.64976493e-02 1.40233502e-01 6.79905236e-01 -8.54885280e-02
-9.49094296e-01 8.89716685e-01 6.11163902e+00 9.56450880e-01
-7.43243098e-01 -1.00961685e-01 4.98103172e-01 -9.32001770e-01
4.18164060e-02 9.41632167e-02 -6.69113398e-01 5.77733934e-01
6.14998281e-01 -1.36405185e-01 1.50131673e-01 7.76442528e-01
5.87701142e-01 -1.86069980e-01 -6.21360362e-01 8.42787743e-01
-6.58800602e-02 -1.84409618e+00 -4.87548739e-01 2.30555803e-01
4.86118466e-01 -1.47464901e-01 -1.78475618e-01 7.98465014e-01
5.03753603e-01 -1.00836170e+00 1.14432430e+00 7.31787562e-01
7.30847180e-01 -1.00525451e+00 3.91271472e-01 5.16480207e-01
-1.14235795e+00 -4.65991348e-01 9.24363881e-02 -7.45460987e-01
3.02282393e-01 2.38130927e-01 -5.73990881e-01 5.04702210e-01
1.04594100e+00 7.85622537e-01 -5.89988589e-01 1.11445224e+00
4.29019853e-02 1.01666129e+00 -4.02382076e-01 4.17505726e-02
2.73149282e-01 -1.72538504e-01 5.87202609e-01 9.38781917e-01
1.45794362e-01 2.59204477e-01 6.58810437e-01 5.67080677e-01
1.90396190e-01 2.76177339e-02 -3.64483684e-01 -7.44114369e-02
5.71317255e-01 1.34953511e+00 -8.25663269e-01 -1.63407117e-01
-2.63296627e-03 1.64715692e-01 1.36877328e-01 -3.03302165e-02
-1.04738128e+00 -2.73904234e-01 8.87858748e-01 9.80344534e-01
-2.94186443e-01 -2.12296873e-01 -7.30488896e-01 -8.17489445e-01
-1.64551377e-01 -1.22198117e+00 7.22230971e-01 -6.91247344e-01
-9.72836256e-01 1.74183235e-01 2.97091812e-01 -1.31362295e+00
-1.27268568e-01 -1.85128510e-01 -7.22884893e-01 7.01147616e-01
-5.79377830e-01 -1.17901623e+00 -5.59685409e-01 2.40139499e-01
6.65416956e-01 -2.97382146e-01 2.63610244e-01 4.65018034e-01
-7.64394999e-01 4.90653962e-01 -3.61323804e-01 6.22083724e-01
6.07208073e-01 -1.02776670e+00 3.28720480e-01 6.94679499e-01
2.65394360e-01 7.22874776e-02 8.16626251e-01 -8.99678648e-01
-1.12847006e+00 -5.82372904e-01 3.01519483e-01 -6.14277840e-01
6.83414698e-01 -4.48232234e-01 -5.35239995e-01 6.40720308e-01
-2.19429195e-01 -3.38229865e-01 8.79402816e-01 6.22212231e-01
2.36579739e-02 -2.59323806e-01 -5.34961462e-01 7.54216790e-01
1.34928250e+00 -2.70052969e-01 -4.95645553e-01 1.58918485e-01
1.51549168e-02 -8.77842784e-01 -8.37059915e-01 4.66889411e-01
8.70268643e-01 -1.08554029e+00 9.86619234e-01 -9.41749871e-01
7.42852151e-01 -1.68332249e-01 3.99657607e-01 -1.06001198e+00
-1.16670229e-01 -3.39455038e-01 1.92418605e-01 1.18286717e+00
2.13236123e-01 1.87303647e-01 1.16809821e+00 7.99540997e-01
-6.62581742e-01 -9.43011701e-01 -8.95520747e-01 -5.05994618e-01
-1.02677688e-01 -6.82546556e-01 3.95638317e-01 7.42910266e-01
2.44402304e-01 -1.12680756e-02 -8.36725116e-01 -3.92576396e-01
4.26190019e-01 4.18297112e-01 1.41126180e+00 -1.35500467e+00
-3.43854606e-01 -5.79689562e-01 -6.47512376e-01 -7.32252479e-01
-3.31713945e-01 -6.94791794e-01 -1.47106126e-01 -1.52343369e+00
2.45446309e-01 -5.86240590e-01 9.53740347e-03 7.89317906e-01
6.67877123e-02 4.44230765e-01 3.77828896e-01 3.21043819e-01
-5.01077831e-01 1.40816689e-01 1.19468844e+00 1.56363901e-02
-3.12153459e-01 2.77213931e-01 -7.03864992e-01 7.87889957e-01
6.96389079e-01 -4.05976117e-01 -2.26327538e-01 -2.13476405e-01
4.07545179e-01 9.62041840e-02 6.30589962e-01 -1.36113775e+00
6.95384443e-02 -4.23381746e-01 4.07913595e-01 -8.54930222e-01
3.59380811e-01 -8.49114135e-02 4.42472905e-01 6.06269717e-01
-7.16582954e-01 -7.10341558e-02 1.67904466e-01 4.32224602e-01
-1.48476794e-01 1.97276622e-01 2.76197344e-01 -1.17537707e-01
-7.72073865e-01 2.26068184e-01 -4.54891235e-01 3.32187682e-01
1.21416926e+00 -5.05853295e-01 -3.15604895e-01 -4.89567995e-01
-8.19782734e-01 3.89016002e-01 2.20221877e-01 7.43555367e-01
1.35138586e-01 -1.18537140e+00 -9.60896313e-01 -1.88841432e-01
1.15453944e-01 -2.40292087e-01 6.41102493e-01 8.24443519e-01
-7.36070812e-01 -7.75465220e-02 -7.12974012e-01 -4.60633457e-01
-1.33497572e+00 2.39712968e-01 2.51364201e-01 -4.34109092e-01
-1.13624513e+00 5.60616553e-01 -4.08617109e-01 -2.14952931e-01
1.43250391e-01 -5.66192083e-02 -4.31305736e-01 3.32052648e-01
5.31751990e-01 9.21377063e-01 -1.33878455e-01 -6.47477984e-01
-2.89509326e-01 1.89093515e-01 3.50642353e-01 -1.07567824e-01
1.28073514e+00 3.95094097e-01 5.92176080e-01 6.36157393e-01
2.60291576e-01 2.72594213e-01 -1.54356432e+00 4.78612669e-02
9.31852758e-02 -6.71070695e-01 -1.46043316e-01 -1.03367293e+00
-1.27934635e+00 9.36852217e-01 6.12318337e-01 5.65222576e-02
8.65566194e-01 -4.46670055e-02 8.82289886e-01 -1.09561719e-01
3.86911690e-01 -1.40051270e+00 1.10606737e-01 1.39162779e-01
7.14769483e-01 -9.30115461e-01 9.30982009e-02 -1.37797460e-01
-1.07278538e+00 8.80322456e-01 9.29012835e-01 -1.13586165e-01
1.67743862e-01 3.36972803e-01 1.48553506e-01 -5.09306550e-01
-5.62886894e-01 -3.50496471e-01 2.86449462e-01 3.71204704e-01
2.40516126e-01 2.54993767e-01 -5.83106220e-01 1.14940870e+00
-7.43660688e-01 1.11064976e-02 4.50974882e-01 9.07950044e-01
-4.42037106e-01 -8.94686520e-01 -3.55632037e-01 7.40050137e-01
-4.66901094e-01 2.85462886e-01 -5.69345832e-01 1.00097239e+00
4.74775016e-01 7.65632153e-01 1.46369740e-01 -6.94942355e-01
7.25678325e-01 -1.29111171e-01 2.03394398e-01 -5.40580451e-01
-1.16282558e+00 -2.79052079e-01 4.88227725e-01 -6.47202075e-01
-1.34721249e-01 -8.72010112e-01 -1.14396274e+00 -7.36203790e-01
9.41419899e-02 1.82352141e-01 3.45898509e-01 7.93754995e-01
4.30919498e-01 6.35992229e-01 1.55951932e-01 -9.41058517e-01
-1.99923590e-01 -1.07679605e+00 -5.53585231e-01 6.35361254e-01
-5.49447119e-01 -1.17301559e+00 1.35362953e-01 -1.75958961e-01] | [6.663517951965332, 0.33256813883781433] |
f53b6c74-0ca7-4871-bd4a-9633aff9ce7e | unifying-vision-and-language-tasks-via-text | 2102.02779 | null | https://arxiv.org/abs/2102.02779v2 | https://arxiv.org/pdf/2102.02779v2.pdf | Unifying Vision-and-Language Tasks via Text Generation | Existing methods for vision-and-language learning typically require designing task-specific architectures and objectives for each task. For example, a multi-label answer classifier for visual question answering, a region scorer for referring expression comprehension, and a language decoder for image captioning, etc. To alleviate these hassles, in this work, we propose a unified framework that learns different tasks in a single architecture with the same language modeling objective, i.e., multimodal conditional text generation, where our models learn to generate labels in text based on the visual and textual inputs. On 7 popular vision-and-language benchmarks, including visual question answering, referring expression comprehension, visual commonsense reasoning, most of which have been previously modeled as discriminative tasks, our generative approach (with a single unified architecture) reaches comparable performance to recent task-specific state-of-the-art vision-and-language models. Moreover, our generative approach shows better generalization ability on questions that have rare answers. Also, we show that our framework allows multi-task learning in a single architecture with a single set of parameters, achieving similar performance to separately optimized single-task models. Our code is publicly available at: https://github.com/j-min/VL-T5 | ['Mohit Bansal', 'Hao Tan', 'Jie Lei', 'Jaemin Cho'] | 2021-02-04 | null | null | null | null | ['conditional-text-generation', 'visual-commonsense-reasoning'] | ['natural-language-processing', 'reasoning'] | [ 3.06746960e-01 2.26171315e-01 -1.02149971e-01 -6.22149169e-01
-1.26335287e+00 -7.04546154e-01 7.68875778e-01 -6.93221688e-02
-3.45196396e-01 4.92171526e-01 2.80709416e-01 -5.26675522e-01
4.93664235e-01 -5.05867958e-01 -1.08292711e+00 -5.04051208e-01
7.27854788e-01 7.15575337e-01 1.17258122e-02 1.59794334e-02
7.69178197e-02 -1.97972119e-01 -1.22460794e+00 8.04396927e-01
8.55893075e-01 9.86636996e-01 4.33065176e-01 9.29128528e-01
-3.49770933e-01 1.42578483e+00 -4.68298078e-01 -6.98767364e-01
-2.65736520e-01 -6.97302401e-01 -1.06429851e+00 1.04128174e-01
1.02460575e+00 -4.86183852e-01 -3.38278502e-01 9.13992882e-01
4.71545845e-01 -9.96212848e-03 1.01671112e+00 -1.36389208e+00
-1.58211470e+00 4.00770515e-01 -5.59563398e-01 -5.86683638e-02
4.96073961e-01 4.52703238e-01 1.30231869e+00 -1.08266640e+00
5.13154447e-01 1.59630358e+00 2.61696249e-01 8.46781731e-01
-1.55616879e+00 -4.40450042e-01 4.45411831e-01 3.62556189e-01
-1.23136199e+00 -4.06275690e-01 6.58353686e-01 -6.36274815e-01
1.04260027e+00 1.58588234e-02 1.75802916e-01 1.51093650e+00
2.42641345e-01 1.29722714e+00 1.04698861e+00 -2.58800209e-01
6.73178807e-02 1.02116689e-01 7.87859857e-02 1.07155693e+00
-8.73131603e-02 -4.01581556e-01 -3.71115923e-01 7.50793666e-02
3.86854887e-01 2.40207203e-02 -3.51706415e-01 -2.94748694e-01
-1.27944112e+00 1.18678570e+00 5.46984971e-01 7.53372461e-02
-1.92327186e-01 6.93916559e-01 3.21354449e-01 1.20982498e-01
4.10710871e-01 9.16083381e-02 -2.84093201e-01 2.68336535e-01
-7.39609897e-01 1.90086812e-01 7.74127126e-01 1.06417274e+00
8.98298383e-01 8.37193355e-02 -1.03407788e+00 8.54793370e-01
5.49112439e-01 8.06162000e-01 2.44263232e-01 -1.01018906e+00
6.97352111e-01 5.75124025e-01 -1.16068251e-01 -5.97115636e-01
-3.49015474e-01 -2.42023334e-01 -8.46979320e-01 4.10699844e-03
4.67902869e-01 -7.00851828e-02 -1.38680291e+00 1.97878802e+00
-1.11248769e-01 -4.63157035e-02 1.85000688e-01 9.33435142e-01
1.46130323e+00 9.01457191e-01 4.18271869e-01 3.15096229e-01
1.75472164e+00 -1.55894601e+00 -6.43862665e-01 -6.52011454e-01
5.62716126e-01 -7.60804236e-01 1.57651031e+00 9.98517498e-02
-1.09541190e+00 -6.43933892e-01 -4.75766033e-01 -8.39241028e-01
-3.73215586e-01 5.52122474e-01 5.27108192e-01 2.33161852e-01
-1.38106787e+00 -3.40532392e-01 -3.42767984e-01 -3.19733679e-01
6.70690000e-01 -1.93254560e-01 -1.01484239e-01 -3.36926967e-01
-9.21416759e-01 9.72117901e-01 1.82857454e-01 -6.50617480e-02
-1.63468230e+00 -5.79722643e-01 -1.02465689e+00 9.81227607e-02
3.42279971e-01 -1.42423737e+00 1.49169075e+00 -9.59015548e-01
-1.18229842e+00 1.40223622e+00 -4.60686237e-01 -4.09807354e-01
4.19297308e-01 -2.81840324e-01 1.29324764e-01 2.81480074e-01
2.90464580e-01 1.31567407e+00 1.09212470e+00 -1.60967815e+00
-2.09761724e-01 -2.54179448e-01 3.61788958e-01 1.96253628e-01
-5.29299527e-02 -1.21437602e-01 -7.79351532e-01 -4.35207546e-01
-4.73097116e-01 -7.77293384e-01 -3.85821871e-02 1.69148594e-01
-7.23828554e-01 -5.78171372e-01 5.48953474e-01 -8.29321861e-01
7.03857780e-01 -1.87058461e+00 5.41665077e-01 -4.81783241e-01
4.30775583e-01 -1.02368832e-01 -5.81832707e-01 2.16148287e-01
-6.82673045e-03 1.31576329e-01 -3.91291499e-01 -8.00047636e-01
2.10883215e-01 2.29163602e-01 -5.46729863e-01 1.71159968e-01
4.56477225e-01 1.59329140e+00 -7.30104566e-01 -7.14728773e-01
8.34561735e-02 4.16410089e-01 -4.38402325e-01 4.83741790e-01
-9.04098272e-01 3.99270177e-01 -5.51676989e-01 8.56178999e-01
3.31800461e-01 -9.14791405e-01 -2.36058056e-01 -2.32556254e-01
3.43002200e-01 -6.20142817e-02 -3.83873016e-01 1.93362176e+00
-8.03686500e-01 8.29122305e-01 5.48332222e-02 -1.04962718e+00
9.15884852e-01 2.80307949e-01 8.36497918e-03 -9.53135490e-01
1.58596903e-01 -3.77270319e-02 -4.38293934e-01 -7.75548875e-01
1.60662264e-01 1.16834335e-01 -1.73988506e-01 5.82452059e-01
4.49017376e-01 -4.18065280e-01 2.52545983e-01 5.90655088e-01
8.64797711e-01 2.80899227e-01 3.24313827e-02 1.31249264e-01
5.48492789e-01 -3.74988541e-02 -1.06930651e-01 9.67190087e-01
-4.98127565e-02 7.91054010e-01 6.81490958e-01 -2.38463715e-01
-1.05064917e+00 -1.26763964e+00 2.80523896e-01 1.50958300e+00
-3.75255570e-02 -2.40356490e-01 -6.10637724e-01 -7.24217772e-01
7.27886781e-02 1.08831346e+00 -7.16593921e-01 -1.60242528e-01
-3.50446433e-01 -4.51871186e-01 6.22126579e-01 5.74235499e-01
4.37311888e-01 -1.25637400e+00 -3.84254009e-01 -2.44519338e-01
-4.38651502e-01 -1.48532152e+00 -5.99673331e-01 8.73918366e-03
-5.79459727e-01 -8.40982795e-01 -1.08625841e+00 -1.07252002e+00
6.21257067e-01 2.30126843e-01 1.67159235e+00 4.74769622e-02
-2.56065279e-01 1.02317691e+00 -2.08693549e-01 -3.41459781e-01
-3.43651265e-01 -3.48405652e-02 -7.59973049e-01 1.06158122e-01
2.34671578e-01 -2.11520910e-01 -6.33245528e-01 -7.27077425e-02
-9.10655499e-01 3.98944825e-01 7.00991631e-01 9.40385520e-01
7.75479555e-01 -1.07007992e+00 6.62748933e-01 -6.08808041e-01
8.77141654e-01 -6.45836532e-01 -4.07537878e-01 7.26621985e-01
-2.40990564e-01 2.10173443e-01 5.23867130e-01 -4.63618398e-01
-1.01963258e+00 1.46862924e-01 -4.86596078e-02 -7.16198087e-01
-3.03738564e-01 2.49465376e-01 3.12602073e-02 8.13010260e-02
6.94998860e-01 5.97077966e-01 -8.53879750e-02 -2.75370657e-01
9.97085869e-01 3.39900255e-01 6.67892277e-01 -7.23876238e-01
5.78371286e-01 4.96857047e-01 -1.40231863e-01 -5.26663065e-01
-1.40245688e+00 -3.54655802e-01 -3.49048615e-01 -1.83266804e-01
1.50905752e+00 -1.27436483e+00 -8.82446349e-01 3.68389249e-01
-1.74240935e+00 -6.31142735e-01 -3.06651369e-02 4.81997840e-02
-8.68020654e-01 3.77994865e-01 -3.98568809e-01 -7.83395410e-01
-6.81993067e-01 -1.23568571e+00 1.57064331e+00 3.58553946e-01
4.61088531e-02 -1.10625732e+00 2.39939075e-02 9.65042710e-01
3.94000322e-01 1.76493227e-01 1.16950285e+00 -5.04634440e-01
-9.35811341e-01 3.44230652e-01 -7.71905720e-01 4.02375042e-01
-2.86895245e-01 -2.75054365e-01 -1.07924485e+00 -1.79131463e-01
-2.09124446e-01 -1.13427806e+00 1.58406997e+00 4.62936461e-01
1.50431073e+00 -1.91653833e-01 -1.59808472e-01 6.72345698e-01
1.30623865e+00 -1.15217097e-01 4.61435884e-01 -7.99129903e-02
1.03285110e+00 5.18250287e-01 1.65582806e-01 1.18321702e-01
1.09814000e+00 5.86935997e-01 7.28030801e-01 -3.44584256e-01
-5.25143385e-01 -2.60637611e-01 5.47067046e-01 3.11526984e-01
1.98043868e-01 -5.52965105e-01 -1.10370481e+00 7.74958551e-01
-2.03084373e+00 -8.59627187e-01 -3.15623015e-01 1.60777724e+00
9.12645996e-01 -4.28728729e-01 -7.78078660e-02 -7.61113644e-01
4.03166145e-01 3.84107232e-01 -7.53162742e-01 -4.26417470e-01
-2.39882752e-01 1.12197451e-01 1.20322332e-01 6.14647210e-01
-1.12405050e+00 1.22700560e+00 5.78498173e+00 6.78656757e-01
-9.11902547e-01 4.83697146e-01 8.01178217e-01 -1.32231474e-01
-7.23464251e-01 -2.13202722e-02 -8.08233440e-01 1.23155981e-01
6.59347951e-01 1.23045772e-01 4.33367670e-01 7.59550810e-01
-3.38676386e-02 -6.99927136e-02 -1.19003093e+00 1.24954486e+00
7.79072583e-01 -1.35731304e+00 5.14194667e-01 -2.88799971e-01
7.76464522e-01 2.13081017e-01 3.07793915e-01 5.58416426e-01
3.55718672e-01 -1.32758760e+00 8.74232233e-01 8.17300797e-01
8.89132321e-01 -2.75766969e-01 4.03968364e-01 1.82670206e-01
-9.55850899e-01 -4.62061204e-02 -2.01114848e-01 3.04372996e-01
3.94063324e-01 3.24448228e-01 -6.63552761e-01 3.59428674e-01
5.24239361e-01 5.57258844e-01 -1.01265132e+00 7.25436032e-01
-6.27737880e-01 5.71705103e-01 1.11434348e-01 -1.32995248e-01
5.44047773e-01 -3.02925240e-02 3.18760037e-01 1.29584873e+00
1.33415461e-01 -3.05571288e-01 3.48747462e-01 1.49726927e+00
-3.52154553e-01 1.06569603e-01 -6.41937912e-01 -1.05175115e-01
5.76100089e-02 1.34946752e+00 -2.06160352e-01 -4.49762374e-01
-7.61002064e-01 1.15082288e+00 6.12625897e-01 8.69502664e-01
-1.03128743e+00 -3.55433002e-02 2.63887733e-01 -2.17726812e-01
2.92659909e-01 -2.20066175e-01 -3.43566477e-01 -1.45644426e+00
2.66971090e-03 -8.75080287e-01 5.52424848e-01 -1.33139491e+00
-1.36047137e+00 5.43278039e-01 -1.35713413e-01 -6.31787717e-01
-4.78280544e-01 -8.44115257e-01 -6.25644982e-01 9.74011600e-01
-1.77413106e+00 -1.72062695e+00 -6.56831026e-01 1.03882968e+00
8.84961426e-01 -3.38381916e-01 7.73179591e-01 5.18615320e-02
-3.82543862e-01 5.52759051e-01 -2.87251264e-01 1.83808938e-01
9.33511376e-01 -1.34120691e+00 2.58524656e-01 6.16030395e-01
5.89305758e-01 2.27303788e-01 4.36422080e-01 -3.38502705e-01
-1.45183814e+00 -1.20022035e+00 8.72915626e-01 -7.75471449e-01
7.59595335e-01 -6.54029548e-01 -9.14047062e-01 9.33678389e-01
7.28600562e-01 -2.10215688e-01 5.42472005e-01 3.65446657e-02
-7.07154930e-01 1.53982058e-01 -6.71218693e-01 7.04648197e-01
7.91498244e-01 -8.25599372e-01 -6.29118860e-01 8.11151385e-01
9.49301958e-01 -3.70602459e-01 -4.25095707e-01 8.89737159e-02
2.67429262e-01 -6.82425618e-01 1.11641049e+00 -7.70317495e-01
8.65885615e-01 -2.64827967e-01 -2.54822850e-01 -1.00119793e+00
-2.47968987e-01 -2.52922416e-01 -2.37945154e-01 1.04510450e+00
6.58815145e-01 -4.14472848e-01 3.17403942e-01 1.68689027e-01
-1.75164595e-01 -7.66434491e-01 -7.11367071e-01 -3.52013826e-01
1.91888809e-01 -2.69418538e-01 2.26040520e-02 6.93932414e-01
-7.34152138e-01 1.04905462e+00 -5.41477621e-01 -4.83346507e-02
7.17711091e-01 4.88668174e-01 8.51854265e-01 -8.23086083e-01
-4.30583268e-01 -5.33590555e-01 1.29589245e-01 -1.36007702e+00
6.47774816e-01 -1.24673283e+00 4.73581925e-02 -2.20728922e+00
7.26413429e-01 3.39487910e-01 7.58530423e-02 8.13884616e-01
-3.80107909e-01 2.02889502e-01 5.08464694e-01 6.32053688e-02
-1.02752221e+00 8.25917840e-01 1.57397854e+00 -6.26193702e-01
1.43420309e-01 -3.05220157e-01 -8.27051461e-01 6.39729619e-01
6.55968487e-01 -2.86399156e-01 -5.46027958e-01 -1.05106723e+00
3.14682215e-01 1.81384888e-02 9.90937591e-01 -5.05195320e-01
2.13843733e-01 -9.35887694e-02 4.41939652e-01 -6.25981808e-01
5.94015777e-01 -3.66775215e-01 -4.37260568e-01 2.29003027e-01
-6.02429688e-01 -4.08880077e-02 1.66667119e-01 6.20395899e-01
-2.96552688e-01 -3.11969548e-01 6.00726366e-01 -1.97737455e-01
-7.99720168e-01 2.91239083e-01 -1.70643494e-01 4.60254967e-01
9.23505008e-01 2.10394457e-01 -9.28962052e-01 -8.57174039e-01
-7.23728955e-01 7.81974375e-01 1.88638389e-01 7.30708301e-01
8.76625001e-01 -1.20272803e+00 -1.18949902e+00 -4.56759006e-01
5.34193575e-01 -8.91891657e-04 5.20932555e-01 7.74884641e-01
-1.46011502e-01 6.10545516e-01 -6.23785481e-02 -9.59198236e-01
-1.10660112e+00 6.45635426e-01 4.49557453e-01 -2.71798372e-01
-3.47472489e-01 1.06365430e+00 8.01900983e-01 -3.78862649e-01
1.55935198e-01 -2.95284271e-01 -1.88312188e-01 8.91409665e-02
3.52938920e-01 -2.03260973e-01 -4.09293562e-01 -5.82289934e-01
-2.51485646e-01 7.26796150e-01 -1.10077180e-01 -5.75104207e-02
8.87206078e-01 -1.75042376e-01 -9.08950046e-02 5.38249850e-01
1.09916258e+00 -4.93958324e-01 -1.29205763e+00 -2.77647376e-01
-2.89501995e-01 4.53150906e-02 -6.79408461e-02 -1.01444411e+00
-1.01384866e+00 1.31762469e+00 2.65003771e-01 -3.52802202e-02
1.12506199e+00 7.79167175e-01 5.14472902e-01 5.53300917e-01
-8.47720131e-02 -6.75689340e-01 7.30980814e-01 7.34573543e-01
1.35679924e+00 -1.57702160e+00 -3.19436044e-01 -3.06018889e-02
-1.11481810e+00 9.25233841e-01 8.24696422e-01 4.75012213e-02
6.27890602e-02 -1.16064265e-01 6.08444661e-02 -3.58890951e-01
-1.18695652e+00 -6.03792191e-01 6.68616712e-01 6.38429403e-01
6.12307549e-01 8.84758905e-02 1.14158407e-01 4.40057129e-01
8.65341574e-02 -2.13669926e-01 2.89622962e-01 3.92233282e-01
-3.93412650e-01 -8.04063439e-01 -1.94892600e-01 3.08060825e-01
-2.69626677e-01 -4.16323334e-01 -6.39117897e-01 5.44122159e-01
-1.96796402e-01 9.39687490e-01 4.59649973e-02 3.67548466e-02
1.95286721e-01 3.97903919e-01 6.56790733e-01 -7.26061642e-01
-4.04962867e-01 -1.37625992e-01 1.98337257e-01 -7.26341903e-01
-3.71171653e-01 -2.96873510e-01 -1.19280374e+00 2.72484541e-01
1.47303447e-01 -3.77773196e-01 5.02342343e-01 1.00905645e+00
4.37848628e-01 6.28474951e-01 1.48573786e-01 -4.71399486e-01
-6.21044815e-01 -1.00370204e+00 -1.64623745e-02 4.88268882e-01
4.18399006e-01 -4.33766693e-01 -1.29803389e-01 4.55272466e-01] | [10.875555992126465, 1.6745952367782593] |
d1d73d63-1723-4124-aa0a-34b4dca97ff8 | spanbert-improving-pre-training-by | 1907.10529 | null | https://arxiv.org/abs/1907.10529v3 | https://arxiv.org/pdf/1907.10529v3.pdf | SpanBERT: Improving Pre-training by Representing and Predicting Spans | We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text. Our approach extends BERT by (1) masking contiguous random spans, rather than random tokens, and (2) training the span boundary representations to predict the entire content of the masked span, without relying on the individual token representations within it. SpanBERT consistently outperforms BERT and our better-tuned baselines, with substantial gains on span selection tasks such as question answering and coreference resolution. In particular, with the same training data and model size as BERT-large, our single model obtains 94.6% and 88.7% F1 on SQuAD 1.1 and 2.0, respectively. We also achieve a new state of the art on the OntoNotes coreference resolution task (79.6\% F1), strong performance on the TACRED relation extraction benchmark, and even show gains on GLUE. | ['Omer Levy', 'Luke Zettlemoyer', 'Mandar Joshi', 'Danqi Chen', 'Yinhan Liu', 'Daniel S. Weld'] | 2019-07-24 | spanbert-improving-pre-training-by-1 | https://aclanthology.org/2020.tacl-1.5 | https://aclanthology.org/2020.tacl-1.5.pdf | tacl-2020-1 | ['linguistic-acceptability'] | ['natural-language-processing'] | [ 3.36957350e-02 7.32801795e-01 -7.02311039e-01 -7.01690018e-02
-1.67623508e+00 -8.44134390e-01 2.84470677e-01 2.50645429e-01
-2.92436659e-01 1.02515101e+00 6.95760489e-01 -1.48536652e-01
-8.68526846e-02 -5.75567901e-01 -6.39384806e-01 -5.88419177e-02
-7.56944120e-02 9.12249029e-01 4.10031259e-01 -3.15457791e-01
1.39355838e-01 1.91550761e-01 -1.16835260e+00 7.24919081e-01
8.80324602e-01 9.35846746e-01 -2.88988769e-01 6.88270688e-01
-2.02161893e-01 1.09110379e+00 -8.63950193e-01 -5.72852492e-01
-2.84411252e-01 -1.35654416e-02 -1.77168298e+00 -4.99071300e-01
7.84239888e-01 -2.32287310e-02 -7.70119905e-01 5.19155264e-01
5.13558626e-01 2.66264468e-01 3.10730428e-01 -9.27911758e-01
-4.57489431e-01 1.60500956e+00 -6.30236089e-01 5.62222660e-01
6.91722333e-01 -3.53024364e-01 1.66877282e+00 -6.88338935e-01
8.49917114e-01 1.14739430e+00 7.87366986e-01 5.12717009e-01
-1.32624340e+00 -7.32779503e-01 1.63135484e-01 5.10847569e-01
-1.39674532e+00 -8.26592207e-01 5.12754560e-01 3.58117521e-02
1.46478820e+00 5.21098793e-01 3.29603180e-02 1.07936811e+00
-2.78858542e-01 7.98568010e-01 7.81453311e-01 -6.72825754e-01
-2.67407417e-01 -3.81654799e-01 8.28063905e-01 4.81407583e-01
5.06435782e-02 -1.86938811e-02 -1.14449024e+00 -3.97179067e-01
-4.69711497e-02 -6.69852853e-01 -7.14783370e-01 3.43768895e-01
-1.21048617e+00 5.61058402e-01 3.38048488e-01 2.92515963e-01
-2.01709136e-01 2.14588806e-01 7.03006148e-01 2.98912942e-01
3.74293774e-01 8.93611372e-01 -7.77546942e-01 -3.46073657e-01
-1.20977950e+00 5.56168556e-01 1.12424481e+00 1.33183217e+00
4.36145812e-01 -3.39450300e-01 -5.97582161e-01 8.56277406e-01
-2.66590416e-01 2.59458542e-01 1.87887684e-01 -1.15343475e+00
1.06623471e+00 4.30015117e-01 1.22737423e-01 -3.60130668e-01
-5.66756010e-01 -3.58950019e-01 -3.30686778e-01 -4.14590776e-01
5.26133537e-01 -1.93878338e-01 -7.31541634e-01 1.99534869e+00
1.99133337e-01 2.08948124e-02 9.36200544e-02 5.35298347e-01
1.27903306e+00 5.33053100e-01 2.65783280e-01 -2.77879089e-01
1.81252813e+00 -1.06954694e+00 -7.62613952e-01 -3.14576417e-01
7.29053140e-01 -8.96601915e-01 8.45756710e-01 2.63202459e-01
-1.45850277e+00 -3.50012958e-01 -1.09501183e+00 -6.04474247e-01
-1.63140193e-01 -1.13281772e-01 4.88881260e-01 1.36699274e-01
-7.19629526e-01 8.04256201e-01 -4.62547660e-01 -6.09552786e-02
2.46938631e-01 1.37460038e-01 -4.01302516e-01 7.96117634e-02
-1.69926894e+00 9.81366336e-01 5.72087824e-01 -4.27506775e-01
-5.24235249e-01 -1.35137010e+00 -8.50211799e-01 4.05996501e-01
6.13186657e-01 -5.47213376e-01 1.58334136e+00 -1.15064330e-01
-1.01395166e+00 1.09688699e+00 -3.73435616e-01 -7.88372278e-01
2.63656706e-01 -1.01135886e+00 -6.69128835e-01 3.77974421e-01
9.62395743e-02 6.02232695e-01 1.58528194e-01 -8.47952962e-01
-3.91790271e-01 1.01237372e-02 9.26898569e-02 2.00898632e-01
6.99420124e-02 3.41877371e-01 -2.95058429e-01 -5.58059394e-01
-5.03466604e-03 -7.72961259e-01 3.08081329e-01 -4.52237904e-01
-9.59953487e-01 -7.27442920e-01 5.27955651e-01 -1.00823641e+00
1.65828407e+00 -1.79388261e+00 8.48989710e-02 -1.55092850e-01
2.44808301e-01 2.59343714e-01 -2.53545254e-01 7.08647847e-01
-3.44716966e-01 3.03928584e-01 -8.71195197e-02 -4.50109899e-01
8.22142363e-02 -1.94463819e-01 -9.14149761e-01 6.83153346e-02
-1.27175981e-02 1.00320148e+00 -7.85411775e-01 -8.45497549e-01
-3.35432380e-01 4.44734059e-02 -4.37464833e-01 2.56823421e-01
-4.24870580e-01 -1.26874477e-01 -7.07836002e-02 5.68905056e-01
4.19431686e-01 -2.28883371e-01 4.19157565e-01 -3.30722272e-01
6.69998229e-02 1.34622979e+00 -7.66549230e-01 1.96447074e+00
-3.41787577e-01 8.03756773e-01 -2.60423664e-02 -5.56113243e-01
8.47205818e-01 7.26508558e-01 4.53799248e-01 -3.43555659e-01
-2.82504439e-01 -3.26084718e-02 -2.38918290e-01 -4.18823719e-01
9.61637139e-01 9.73193422e-02 -6.23184443e-01 6.20410442e-01
2.94348449e-01 5.29279150e-02 2.85544693e-01 6.63362980e-01
1.40024984e+00 9.30869728e-02 3.04563910e-01 -3.00038725e-01
3.07028234e-01 1.58581614e-01 7.76237369e-01 5.25958121e-01
-5.94051667e-02 6.64580464e-01 8.52744937e-01 -1.74464718e-01
-7.63382912e-01 -1.13082325e+00 -2.06088290e-01 1.31344712e+00
4.47915262e-03 -9.08734858e-01 -7.07857490e-01 -1.11787522e+00
1.32457778e-01 1.04816949e+00 -6.14203095e-01 2.47699488e-03
-1.15077519e+00 -3.01677436e-01 1.22988153e+00 7.58045018e-01
3.81368011e-01 -1.30433702e+00 -2.80265719e-01 3.41130733e-01
-9.69866574e-01 -1.25041664e+00 -7.92043924e-01 2.56481200e-01
-7.33303964e-01 -1.45558512e+00 -1.36980578e-01 -8.41180563e-01
-1.12772681e-01 6.66697398e-02 1.90022469e+00 9.82122198e-02
-2.01948926e-01 -2.40366235e-01 -4.70738173e-01 -1.36283979e-01
-4.44010310e-02 8.39746654e-01 -3.51423800e-01 -9.38208520e-01
6.69447482e-01 -5.95197320e-01 -5.21843910e-01 1.89892024e-01
-3.40412855e-01 -1.87579811e-01 2.79280633e-01 9.04893160e-01
4.27291512e-01 -6.17492557e-01 1.05830169e+00 -1.19402516e+00
4.86091882e-01 -3.83564293e-01 -3.92448418e-02 6.28837347e-01
-6.96263373e-01 1.57674968e-01 3.40019584e-01 -3.42197388e-01
-1.00513577e+00 -5.70736945e-01 -6.30234405e-02 -1.91751570e-01
5.65074980e-02 3.57187331e-01 -2.50026882e-01 5.54229677e-01
1.05627906e+00 -3.27170849e-01 -3.00499171e-01 -7.97469854e-01
7.12589145e-01 6.81499183e-01 9.61514115e-01 -8.29230547e-01
5.88935733e-01 2.79605072e-02 -4.79584575e-01 -2.12055787e-01
-1.38513494e+00 -6.94468498e-01 -5.70063531e-01 3.43744278e-01
6.02746010e-01 -9.15666878e-01 -8.06801438e-01 -4.99436371e-02
-1.32732689e+00 -2.92763561e-01 -3.94878000e-01 9.22977179e-02
-3.57987672e-01 4.59347755e-01 -1.14533329e+00 -3.45026642e-01
-1.17246127e+00 -4.68683392e-01 9.21235025e-01 2.88360417e-01
-1.08345532e+00 -5.99491477e-01 3.06175709e-01 7.32084751e-01
-7.16425129e-04 1.94487438e-01 1.10013878e+00 -9.26734984e-01
-2.21801355e-01 -1.87752359e-02 -3.19572300e-01 -1.27361909e-01
-1.91574603e-01 -2.06018969e-01 -1.17030871e+00 -4.47390415e-02
-5.31582594e-01 -7.36538827e-01 1.12843108e+00 6.34716749e-02
1.23940933e+00 -3.79894108e-01 -6.53889418e-01 5.02028763e-01
1.10823321e+00 -1.25030860e-01 6.55062258e-01 3.76195669e-01
3.57458293e-01 6.29438519e-01 6.10215843e-01 1.25287920e-01
4.04781282e-01 7.44367003e-01 1.03331812e-01 2.81772912e-01
-5.98889768e-01 -5.39743423e-01 2.23534718e-01 5.65285802e-01
1.06846709e-02 -1.15501024e-01 -8.48645031e-01 8.78954709e-01
-1.81169295e+00 -1.30955887e+00 -1.37585357e-01 1.79933190e+00
1.48794663e+00 4.08170640e-01 1.18787333e-01 2.56427556e-01
8.18417668e-01 5.65243661e-01 -4.56114262e-01 -5.74589372e-01
-2.82035053e-01 7.41199970e-01 3.44500661e-01 6.02263510e-01
-1.20673966e+00 1.21461344e+00 7.36356306e+00 7.89401114e-01
-5.49474835e-01 -5.29522449e-02 3.76342505e-01 -5.12693703e-01
-3.81800890e-01 3.14619839e-02 -1.13724470e+00 1.81938976e-01
1.10923350e+00 -3.93020838e-01 3.73393744e-01 7.47967005e-01
-3.67910206e-01 1.68796599e-01 -1.27485442e+00 6.68282688e-01
1.24697126e-01 -1.58838475e+00 -2.05231234e-01 -2.41880804e-01
5.14294982e-01 -5.21005504e-02 -3.22119415e-01 8.34607720e-01
5.84543943e-01 -1.32253420e+00 7.01089501e-01 3.85781616e-01
1.02962792e+00 -8.86245430e-01 7.43837118e-01 1.31657749e-01
-1.27723944e+00 2.12604746e-01 -3.64073694e-01 1.30982056e-01
3.78581375e-01 6.14377081e-01 -9.01003420e-01 6.82958066e-01
5.90277255e-01 3.66086215e-01 -3.06046873e-01 9.19881642e-01
-6.34939611e-01 8.93968701e-01 -2.17265874e-01 2.26781696e-01
-1.92324951e-01 6.50988936e-01 6.16098702e-01 1.43289912e+00
-8.58200938e-02 3.04775774e-01 3.65299955e-02 7.45243371e-01
-7.69941926e-01 -5.07686548e-02 -1.32483393e-01 1.95700005e-01
1.42251408e+00 1.25775492e+00 3.67834009e-02 -5.01460671e-01
-1.64582208e-01 6.00708425e-01 8.76931727e-01 3.04630429e-01
-1.02711070e+00 -8.35952342e-01 7.39346445e-01 -1.84777349e-01
4.01497245e-01 1.55986294e-01 -5.56088507e-01 -1.00617135e+00
-2.72697899e-02 -1.01273274e+00 1.06401861e+00 -6.02767169e-01
-1.14398181e+00 9.56257880e-01 1.22611172e-01 -8.57623160e-01
-3.99314195e-01 -1.74038604e-01 -8.76890123e-01 1.06845462e+00
-1.37272203e+00 -1.24231470e+00 -7.94685110e-02 3.29873502e-01
5.09086311e-01 1.61063269e-01 1.20677853e+00 1.89742446e-02
-5.88212788e-01 1.16980588e+00 -3.89749676e-01 5.87676048e-01
1.21756518e+00 -1.51284707e+00 6.48725212e-01 7.11073518e-01
3.45869362e-01 6.69145644e-01 6.44217849e-01 -5.40283740e-01
-8.49543929e-01 -8.92136693e-01 1.62525225e+00 -6.10706806e-01
7.99862981e-01 -2.30970874e-01 -1.10240030e+00 1.04418111e+00
6.45193756e-01 -2.08296478e-01 8.63461792e-01 1.05858266e+00
-1.17984927e+00 -4.63517494e-02 -1.13936329e+00 3.58377129e-01
1.40013158e+00 -7.96525836e-01 -1.45896709e+00 1.50887236e-01
1.15439093e+00 -6.60346568e-01 -1.29927623e+00 5.99371791e-01
3.22669625e-01 -6.40680075e-01 1.15544236e+00 -1.01189065e+00
4.51237768e-01 2.45178223e-01 -7.94211179e-02 -1.07782543e+00
-4.69803184e-01 -9.95222867e-01 -9.56300676e-01 1.57874143e+00
8.21328402e-01 -1.68109670e-01 9.17999923e-01 6.43193781e-01
-3.67410928e-01 -9.80789900e-01 -1.02879500e+00 -6.60133898e-01
4.29759681e-01 -2.19554573e-01 9.66630697e-01 7.86670804e-01
7.61550665e-01 6.88619137e-01 -1.84776589e-01 5.74751720e-02
4.70008433e-01 8.24736655e-01 4.12243038e-01 -1.11452782e+00
-2.28741840e-01 -3.98862213e-01 3.36258471e-01 -1.13845718e+00
5.82100630e-01 -9.49210048e-01 9.87674147e-02 -1.34825659e+00
2.57194430e-01 -5.90307117e-01 -3.84869486e-01 6.68198764e-01
-5.94271302e-01 6.28696978e-02 1.89478219e-01 3.95536125e-01
-7.42221415e-01 2.82245785e-01 8.18936944e-01 -2.41741568e-01
-1.02469780e-01 -2.09547982e-01 -1.06743908e+00 4.09951925e-01
8.89199078e-01 -4.54366654e-01 -1.49997726e-01 -4.17181104e-01
-4.64378390e-03 2.94087917e-01 -1.90327745e-02 -7.44639397e-01
2.04572409e-01 -3.88046205e-02 2.75497913e-01 -8.96919549e-01
3.86584967e-01 1.07315913e-01 -3.32739383e-01 1.62868813e-01
-7.98287213e-01 5.02115265e-02 3.78351718e-01 2.18681663e-01
-2.77256370e-01 -2.93359518e-01 4.72999781e-01 -2.08712500e-02
-6.47274852e-01 8.94237682e-02 5.09480722e-02 9.03719783e-01
5.39372087e-01 3.60027403e-01 -1.11251199e+00 -1.53503418e-01
-8.78880739e-01 4.61870432e-01 1.67798474e-01 4.59451169e-01
2.71964729e-01 -1.22570407e+00 -9.07590210e-01 -4.69389498e-01
1.76498249e-01 2.33543471e-01 2.48351246e-01 5.26733398e-01
-2.79793203e-01 6.05401576e-01 2.71237403e-01 -9.24310833e-03
-1.53542554e+00 6.25146151e-01 2.43736103e-01 -1.01499522e+00
-8.31458449e-01 1.14914608e+00 -2.43073538e-01 -1.84340537e-01
3.91906410e-01 -2.16116443e-01 -3.31318557e-01 3.10866326e-01
8.41989815e-01 5.63589036e-01 2.01205075e-01 -2.73772925e-01
-6.08233094e-01 2.37531066e-01 -4.69676107e-01 -2.19647825e-01
1.05190778e+00 8.17066729e-02 -6.81771040e-02 1.14136077e-01
8.95354390e-01 3.90146822e-01 -9.21408057e-01 -5.17092943e-01
6.40245795e-01 -4.18789461e-02 -2.01823518e-01 -1.25881457e+00
-7.09356904e-01 6.26112342e-01 -3.02451372e-01 1.06634758e-01
7.98365057e-01 4.40013498e-01 1.25153077e+00 3.63988817e-01
2.70478606e-01 -9.99150276e-01 -2.42014639e-02 6.96092069e-01
1.14087594e+00 -7.00382948e-01 1.33409515e-01 -6.87170744e-01
-5.73386610e-01 1.00750315e+00 6.67888343e-01 -1.60010263e-01
3.21217626e-01 5.02559662e-01 1.17625007e-02 -1.91734880e-01
-1.26621544e+00 -1.17011279e-01 5.19455612e-01 4.26966339e-01
8.54038835e-01 1.29686058e-01 -1.79771647e-01 1.00194621e+00
-7.56696403e-01 -2.05170691e-01 2.50614345e-01 6.46935582e-01
-5.42537868e-01 -1.14558053e+00 -3.09795648e-01 3.48484248e-01
-8.43930840e-01 -4.20939207e-01 -6.07199490e-01 8.86442304e-01
-1.73789531e-01 1.16833484e+00 1.08492985e-01 -3.82977694e-01
4.47533906e-01 6.22941136e-01 4.82005596e-01 -7.59909153e-01
-9.09518838e-01 -3.21870983e-01 1.09839463e+00 -7.08164930e-01
6.59455720e-04 -7.16425002e-01 -1.41228843e+00 -7.29451239e-01
-4.62429911e-01 7.27281094e-01 -8.51336271e-02 7.90384173e-01
5.04777551e-01 5.46407402e-01 3.63889664e-01 -1.73744693e-01
-7.87702441e-01 -1.55370128e+00 -3.05065602e-01 4.58219647e-01
1.14180036e-01 -5.21527469e-01 -2.61923641e-01 -6.85823411e-02] | [9.364073753356934, 9.3938570022583] |
75930f4b-ae81-4133-988e-7babc2414ec5 | re-imagine-the-negative-prompt-algorithm | 2304.04968 | null | https://arxiv.org/abs/2304.04968v3 | https://arxiv.org/pdf/2304.04968v3.pdf | Re-imagine the Negative Prompt Algorithm: Transform 2D Diffusion into 3D, alleviate Janus problem and Beyond | Although text-to-image diffusion models have made significant strides in generating images from text, they are sometimes more inclined to generate images like the data on which the model was trained rather than the provided text. This limitation has hindered their usage in both 2D and 3D applications. To address this problem, we explored the use of negative prompts but found that the current implementation fails to produce desired results, particularly when there is an overlap between the main and negative prompts. To overcome this issue, we propose Perp-Neg, a new algorithm that leverages the geometrical properties of the score space to address the shortcomings of the current negative prompts algorithm. Perp-Neg does not require any training or fine-tuning of the model. Moreover, we experimentally demonstrate that Perp-Neg provides greater flexibility in generating images by enabling users to edit out unwanted concepts from the initially generated images in 2D cases. Furthermore, to extend the application of Perp-Neg to 3D, we conducted a thorough exploration of how Perp-Neg can be used in 2D to condition the diffusion model to generate desired views, rather than being biased toward the canonical views. Finally, we applied our 2D intuition to integrate Perp-Neg with the state-of-the-art text-to-3D (DreamFusion) method, effectively addressing its Janus (multi-head) problem. Our project page is available at https://Perp-Neg.github.io/ | ['Mingyuan Zhou', 'Amir Sadeghian', 'Ali Sadeghian', 'Huangjie Zheng', 'Mohammadreza Armandpour'] | 2023-04-11 | null | null | null | null | ['text-to-3d'] | ['computer-vision'] | [ 1.73712954e-01 3.31319481e-01 2.04672173e-01 -2.50012964e-01
-5.99686682e-01 -7.80956686e-01 9.68580425e-01 -4.97744739e-01
-1.42685607e-01 4.18510705e-01 3.95726115e-01 -5.47658563e-01
1.66323408e-01 -6.91834927e-01 -3.91471386e-01 -4.95901138e-01
3.87709051e-01 5.08042455e-01 1.43492356e-01 -1.93381280e-01
4.45957422e-01 4.99893636e-01 -1.39618444e+00 3.61624539e-01
6.99432135e-01 4.52929080e-01 3.55913162e-01 7.77198970e-01
-4.32632565e-02 4.32724446e-01 -6.68097854e-01 -4.98985261e-01
5.03808022e-01 -6.33370459e-01 -4.82720137e-01 2.64034569e-01
6.08650565e-01 -5.45542657e-01 1.98421143e-02 8.32281768e-01
7.75595367e-01 -2.59361416e-02 9.12034929e-01 -1.35800457e+00
-1.06328440e+00 -1.06983408e-02 -8.22774410e-01 2.13103995e-01
5.93040586e-01 3.96491379e-01 7.92024076e-01 -1.32252574e+00
1.10849047e+00 1.33394802e+00 4.34930831e-01 7.87983537e-01
-1.43029261e+00 -5.83066046e-01 4.06866558e-02 -3.43723029e-01
-1.24695587e+00 -4.42155033e-01 8.15045595e-01 -6.80319667e-01
1.07472551e+00 2.58771658e-01 9.31595027e-01 1.45411325e+00
1.96183205e-01 8.12548220e-01 1.37493157e+00 -5.47415733e-01
1.18062176e-01 3.85188520e-01 -3.82301092e-01 5.81775486e-01
1.72999967e-02 3.05681884e-01 -6.32245183e-01 -1.96775377e-01
1.19164860e+00 -4.32424068e-01 -2.27692708e-01 -7.11875319e-01
-1.20462668e+00 9.82889652e-01 1.38141900e-01 2.17695087e-01
-4.21757340e-01 -7.33085051e-02 -1.21636465e-01 1.25307351e-01
7.26789474e-01 6.80665493e-01 -2.11986706e-01 -1.94208741e-01
-1.08888805e+00 5.57639599e-01 6.39958024e-01 8.99726033e-01
5.15546262e-01 -2.52641384e-02 -1.68521956e-01 7.22192764e-01
2.77134985e-01 3.76094759e-01 3.05939972e-01 -1.33017254e+00
4.30261135e-01 4.18063492e-01 1.56409517e-01 -9.95335698e-01
-2.50071347e-01 -2.86168069e-01 -5.15916646e-01 7.40861952e-01
4.09821182e-01 -1.73743173e-01 -1.13133764e+00 1.71032619e+00
3.58045638e-01 -3.73335630e-01 -4.44324762e-02 9.89415109e-01
4.44735378e-01 5.62046468e-01 -6.57624528e-02 2.12607477e-02
9.87945080e-01 -9.97659266e-01 -4.60346907e-01 -2.71279812e-01
7.41960406e-01 -1.05547035e+00 1.38016832e+00 4.46183205e-01
-1.42479348e+00 -3.52786869e-01 -8.31358552e-01 -1.46385923e-01
-3.46073925e-01 8.73116627e-02 5.00909686e-01 6.58592939e-01
-1.45541513e+00 4.05358791e-01 -6.42021775e-01 -6.77945018e-01
3.99859220e-01 6.18549176e-02 -3.79989833e-01 -2.28308327e-02
-8.05558026e-01 1.05539966e+00 1.47854816e-02 -2.31936842e-01
-5.07764995e-01 -7.68991292e-01 -7.37090647e-01 -3.15516889e-01
2.57978261e-01 -1.11606967e+00 1.35843134e+00 -1.17988122e+00
-1.35344839e+00 9.26582098e-01 -1.73721477e-01 -7.02080317e-03
9.49353099e-01 -2.20344812e-01 3.85825075e-02 1.57427162e-01
2.67852247e-01 1.30333281e+00 9.86376345e-01 -1.50825560e+00
-8.60582441e-02 -3.51829916e-01 7.42430165e-02 5.45188010e-01
-2.58080959e-01 -1.62929267e-01 -6.53473794e-01 -9.64276612e-01
1.28138503e-02 -9.94617224e-01 -2.15450093e-01 3.24476987e-01
-4.32384163e-01 1.92321371e-02 1.02012575e+00 -4.10145134e-01
9.14543092e-01 -2.16002941e+00 1.61342278e-01 4.00865316e-01
2.58958876e-01 2.11402774e-01 -3.01065564e-01 5.57125986e-01
-2.24547282e-01 3.97552818e-01 -1.59285247e-01 -5.54235935e-01
-1.20170593e-01 1.52496338e-01 -2.59045362e-01 1.65868729e-01
2.61806637e-01 7.93845475e-01 -8.38111281e-01 -5.52194059e-01
2.58503675e-01 6.65032983e-01 -8.56857955e-01 2.34341994e-02
-2.19253287e-01 4.37375426e-01 -2.28348881e-01 3.52638900e-01
7.49071598e-01 -3.75664741e-01 7.75840431e-02 -2.40952417e-01
-8.63585174e-02 1.99853525e-01 -1.08172631e+00 1.67282772e+00
-3.58413398e-01 6.48128033e-01 2.34275479e-02 -4.35994387e-01
6.68370485e-01 2.41365582e-01 3.99208695e-01 -7.68900812e-01
-1.32323146e-01 1.11645035e-01 -8.97190273e-02 -4.93583500e-01
7.39968836e-01 -5.01117468e-01 4.16837245e-01 8.09374571e-01
-2.20511667e-02 -6.71579599e-01 3.19947034e-01 5.87599576e-01
9.06525552e-01 5.30551314e-01 -3.65635678e-02 -6.46009594e-02
7.40930066e-03 9.91512462e-02 1.21341839e-01 8.09693873e-01
1.84386801e-02 1.29099000e+00 4.53593910e-01 2.53914762e-02
-1.26579475e+00 -1.09578288e+00 7.92148784e-02 6.37525320e-01
-8.20159167e-02 -5.23926079e-01 -8.30987930e-01 -1.09193957e+00
-2.57639140e-01 1.24088097e+00 -7.06744909e-01 1.86591983e-01
-1.71282738e-01 -6.37615442e-01 4.58470494e-01 4.14967448e-01
1.10074230e-01 -1.00131989e+00 -6.68191969e-01 5.55754565e-02
-1.44862920e-01 -8.22623253e-01 -6.46385312e-01 -9.73560214e-02
-7.63808727e-01 -8.39800298e-01 -1.11581564e+00 -4.13072228e-01
1.08855259e+00 4.90521848e-01 1.21231544e+00 -2.19118707e-02
-1.02513090e-01 8.43321264e-01 -3.05731595e-01 -4.46932644e-01
-7.09990203e-01 -1.22651029e-02 -2.15448141e-01 -3.05095077e-01
2.03705907e-01 -5.92505097e-01 -8.18948925e-01 4.39989984e-01
-1.15398371e+00 6.79436982e-01 5.68379104e-01 7.68974185e-01
1.98809683e-01 -1.46783009e-01 3.81011039e-01 -8.64508331e-01
1.02957928e+00 -1.43916726e-01 -2.23055199e-01 -5.88162914e-02
-6.66071773e-01 5.50468266e-02 2.96617925e-01 -5.41197896e-01
-1.28788507e+00 2.53749937e-02 -2.65510082e-01 -4.86255348e-01
6.39998317e-02 3.63021612e-01 2.19956949e-01 1.75687999e-01
9.22170341e-01 1.05267819e-02 2.51199991e-01 -3.15462112e-01
6.06684327e-01 3.51626277e-01 1.02139942e-01 -5.89030266e-01
8.32220972e-01 6.16958678e-01 -2.88778365e-01 -6.31144583e-01
-7.05245197e-01 -5.87710924e-03 -5.62165201e-01 -3.13272059e-01
9.23608005e-01 -7.24936724e-01 -1.19269639e-01 3.61627996e-01
-1.17739475e+00 -5.41476190e-01 -4.17502254e-01 2.92167723e-01
-6.65036201e-01 3.38319808e-01 -3.79651397e-01 -6.71500444e-01
-9.43857133e-02 -1.15499723e+00 1.03896487e+00 7.02867191e-03
-7.63137996e-01 -1.12662661e+00 3.02637443e-02 3.88209969e-01
5.24123013e-01 1.33050233e-02 9.45918441e-01 -4.25937355e-01
-5.40397882e-01 3.02611310e-02 -2.71274030e-01 3.85957599e-01
2.35984717e-02 2.14451879e-01 -8.46626461e-01 -2.72299945e-01
-5.53462431e-02 -4.26075608e-01 5.15623212e-01 3.27484041e-01
7.86504805e-01 -1.86215445e-01 -2.15176642e-01 3.01943183e-01
1.11918902e+00 -2.90404800e-02 5.89549780e-01 3.38557333e-01
5.41979492e-01 8.54987800e-01 4.54006851e-01 2.41431817e-01
3.03405136e-01 7.51672447e-01 2.80223995e-01 -4.46784645e-01
-4.53027993e-01 -5.98032832e-01 3.62038344e-01 4.99668121e-01
2.38671619e-02 -3.25748354e-01 -8.31400573e-01 5.46627998e-01
-1.55554962e+00 -9.42369401e-01 -7.82274306e-02 2.11051798e+00
6.98486507e-01 2.51926422e-01 5.69959991e-02 -2.48823673e-01
4.03465241e-01 2.97284424e-01 -5.21632254e-01 -3.01061600e-01
-1.25443280e-01 -7.14552263e-03 2.12242246e-01 5.48008025e-01
-5.44983208e-01 8.01669955e-01 6.59170151e+00 7.41108954e-01
-1.20713961e+00 1.08867206e-01 7.49122322e-01 -3.70037228e-01
-7.67047346e-01 1.76280245e-01 -5.22319198e-01 1.88180804e-01
4.07079607e-01 -5.22127002e-02 2.51137495e-01 7.58367479e-01
5.55557489e-01 -3.78202945e-01 -1.14492142e+00 9.20401931e-01
2.72829682e-01 -1.37125909e+00 2.75580376e-01 3.76778513e-01
7.47785211e-01 -8.62235799e-02 3.27633202e-01 8.13978389e-02
3.71701181e-01 -1.03010738e+00 9.83815849e-01 3.12303841e-01
8.62717807e-01 -3.32334340e-01 1.91792920e-01 4.22536194e-01
-5.76767027e-01 2.20040739e-01 -4.91317175e-02 8.64639580e-02
4.14020836e-01 6.36613667e-01 -1.13433921e+00 2.82393694e-01
5.83626390e-01 5.77649355e-01 -6.84773207e-01 5.45689225e-01
-1.96481913e-01 3.11608076e-01 -3.83471489e-01 2.71795154e-01
2.57758766e-01 -2.46289611e-01 6.68711364e-01 1.11409295e+00
8.41657102e-01 -1.57016143e-03 -1.22909859e-01 1.19482827e+00
1.42843381e-01 -4.62869735e-04 -1.01595104e+00 -1.80394091e-02
1.62595481e-01 1.23803973e+00 -7.84633338e-01 -3.18680525e-01
-3.45973969e-01 1.24348962e+00 2.58160204e-01 6.15932822e-01
-6.56839550e-01 1.13483146e-02 3.16281378e-01 5.50794840e-01
3.43916416e-01 -4.36337918e-01 -4.50404942e-01 -1.15794742e+00
1.18815713e-01 -1.13302398e+00 1.78020913e-02 -1.50821114e+00
-1.35209382e+00 4.82477009e-01 2.08595231e-01 -1.05630207e+00
-3.61876339e-01 -5.70468962e-01 -5.06241679e-01 1.07693708e+00
-1.11257768e+00 -1.24180055e+00 -1.19703338e-01 4.90529090e-01
6.32360995e-01 1.82949141e-01 6.27371728e-01 7.79804513e-02
-1.91906765e-01 2.95433879e-01 -2.85699219e-01 -1.90625399e-01
1.06850290e+00 -1.19858873e+00 5.95433474e-01 8.42575133e-01
1.65952399e-01 7.79169679e-01 9.03504193e-01 -7.20340312e-01
-1.06172717e+00 -6.52167439e-01 7.76024640e-01 -8.25001299e-01
4.15077209e-01 -3.23557258e-01 -6.58215344e-01 5.94557226e-01
4.45131600e-01 -3.38492423e-01 5.57824790e-01 -1.73070252e-01
-2.13737115e-01 3.57386142e-01 -1.06620681e+00 1.07023180e+00
1.16335678e+00 -4.00279820e-01 -4.55337763e-01 1.85899630e-01
3.82487625e-01 -5.70048094e-01 -4.94883567e-01 2.14873523e-01
4.60766196e-01 -1.44995582e+00 9.51123059e-01 -1.53318718e-01
7.27769673e-01 -2.25531235e-01 6.42694682e-02 -1.59677899e+00
-2.32892290e-01 -6.76428735e-01 5.96731827e-02 1.22925878e+00
7.00776577e-01 -5.77550948e-01 9.02246714e-01 8.68629515e-01
-3.22854891e-02 -8.06255579e-01 -6.34604931e-01 -5.36265910e-01
2.69367129e-01 -5.97195864e-01 1.61871940e-01 9.16416049e-01
-2.86607325e-01 4.26963896e-01 -2.86714047e-01 -1.87413082e-01
4.31323737e-01 -2.11940199e-01 1.04993582e+00 -8.06301594e-01
-4.10446882e-01 -4.50446069e-01 3.73146273e-02 -1.48187625e+00
-2.25187033e-01 -8.57517540e-01 -1.84517130e-01 -1.78685844e+00
2.04452321e-01 -5.51902175e-01 2.68288702e-01 4.86608297e-01
-1.24911621e-01 4.21234339e-01 5.22817194e-01 3.03059667e-01
-7.82613009e-02 5.80297470e-01 1.74952495e+00 9.69843790e-02
-3.02641064e-01 -2.53022313e-01 -1.08474767e+00 7.16136992e-01
7.67181575e-01 -4.43117380e-01 -7.90336847e-01 -5.71274042e-01
5.41572213e-01 -3.33991684e-02 5.29116213e-01 -7.16891766e-01
4.43352610e-02 -1.57746240e-01 6.76277757e-01 -7.02191353e-01
5.70221424e-01 -6.22420728e-01 1.75975218e-01 5.27681559e-02
-2.83462942e-01 4.27871644e-01 1.93857968e-01 3.63341272e-01
-6.75003976e-03 -2.45676056e-01 6.10778332e-01 -2.80516416e-01
-3.15744638e-01 2.55365968e-02 -5.72029173e-01 1.30654261e-01
9.35105622e-01 -5.60272276e-01 -3.21294129e-01 -7.86322176e-01
-6.76967919e-01 4.97514904e-02 1.01858115e+00 4.74771172e-01
7.54118860e-01 -1.17682004e+00 -5.38614988e-01 4.31700170e-01
5.68078458e-03 -3.87520879e-03 2.66688824e-01 9.01272714e-01
-4.33887005e-01 2.05597922e-01 -3.68953794e-02 -6.42159283e-01
-1.32077229e+00 4.65824366e-01 3.14472824e-01 -3.06903988e-01
-7.59545565e-01 7.01387227e-01 6.70627773e-01 -6.35692835e-01
-8.66746306e-02 4.77702171e-02 2.09543481e-01 -4.20031883e-02
3.11255723e-01 1.55444536e-03 -7.13744313e-02 -4.62595552e-01
-6.66159987e-02 4.97847646e-01 -3.19688380e-01 -9.90281522e-01
1.22249842e+00 -2.28685752e-01 1.95390567e-01 1.93877861e-01
9.59127367e-01 3.37261200e-01 -1.61947465e+00 1.70654342e-01
-4.34104830e-01 -6.19292259e-01 2.08167229e-02 -1.04821301e+00
-9.31498766e-01 9.17789221e-01 4.09141868e-01 2.14046746e-01
9.86556470e-01 -4.53994237e-02 6.03959262e-01 -6.00356869e-02
1.37545705e-01 -1.03753328e+00 4.64729190e-01 3.45279396e-01
1.15671360e+00 -1.07438505e+00 1.44223005e-01 -3.52794379e-01
-1.04436970e+00 1.00052750e+00 7.64594436e-01 1.86568618e-01
5.80102444e-01 2.23837227e-01 4.80983436e-01 -5.08531034e-01
-7.96272099e-01 1.54555470e-01 1.44460261e-01 8.43827844e-01
6.13080382e-01 -3.17007154e-01 -2.29995653e-01 -1.20507687e-01
-4.99848127e-01 1.39571697e-01 7.32748210e-01 1.09849918e+00
1.17634535e-01 -1.32319486e+00 -4.62889135e-01 2.58515418e-01
-2.76513964e-01 -2.21626937e-01 -6.62448347e-01 8.68826866e-01
-3.14229913e-02 8.01020741e-01 -5.03730290e-02 -1.71620280e-01
2.93584079e-01 1.43442005e-01 6.87738240e-01 -7.43600786e-01
-4.48803216e-01 2.27127895e-01 2.04921097e-01 -4.58901018e-01
-4.92498279e-01 -7.08382249e-01 -9.46054101e-01 -2.07619935e-01
-2.35803828e-01 -6.83975369e-02 7.22961903e-01 6.66567981e-01
6.97017968e-01 2.66209900e-01 3.37876469e-01 -1.25358331e+00
-5.16384363e-01 -8.74182045e-01 -2.94138789e-01 6.42567217e-01
-4.07743920e-03 -6.93859458e-01 -4.43596661e-01 1.16568595e-01] | [11.287439346313477, -0.1668214201927185] |
c70d3ca1-196f-454b-bf58-e36414dc755e | a-novel-active-solution-for-two-dimensional | 2212.06958 | null | https://arxiv.org/abs/2212.06958v1 | https://arxiv.org/pdf/2212.06958v1.pdf | A Novel Active Solution for Two-Dimensional Face Presentation Attack Detection | Identity authentication is the process of verifying one's identity. There are several identity authentication methods, among which biometric authentication is of utmost importance. Facial recognition is a sort of biometric authentication with various applications, such as unlocking mobile phones and accessing bank accounts. However, presentation attacks pose the greatest threat to facial recognition. A presentation attack is an attempt to present a non-live face, such as a photo, video, mask, and makeup, to the camera. Presentation attack detection is a countermeasure that attempts to identify between a genuine user and a presentation attack. Several industries, such as financial services, healthcare, and education, use biometric authentication services on various devices. This illustrates the significance of presentation attack detection as the verification step. In this paper, we study state-of-the-art to cover the challenges and solutions related to presentation attack detection in a single place. We identify and classify different presentation attack types and identify the state-of-the-art methods that could be used to detect each of them. We compare the state-of-the-art literature regarding attack types, evaluation metrics, accuracy, and datasets and discuss research and industry challenges of presentation attack detection. Most presentation attack detection approaches rely on extensive data training and quality, making them difficult to implement. We introduce an efficient active presentation attack detection approach that overcomes weaknesses in the existing literature. The proposed approach does not require training data, is CPU-light, can process low-quality images, has been tested with users of various ages and is shown to be user-friendly and highly robust to 2-dimensional presentation attacks. | ['Matineh Pooshideh'] | 2022-12-14 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 4.27776128e-01 -5.06377161e-01 -7.56990984e-02 7.55925179e-02
-4.66977417e-01 -7.61493385e-01 5.79085052e-01 5.90843745e-02
-2.23408669e-01 2.58474797e-01 -3.43090415e-01 -4.95848626e-01
1.43513223e-02 -5.70427299e-01 -1.17488228e-01 -8.26374948e-01
-4.07065190e-02 -1.47740217e-02 1.99308507e-02 -2.30070487e-01
5.60421586e-01 1.11727178e+00 -1.72805119e+00 3.08462620e-01
1.45206049e-01 1.37774897e+00 -6.16575956e-01 7.10480392e-01
-5.54793812e-02 1.05639808e-01 -9.29782689e-01 -9.15373623e-01
3.23656499e-01 -2.25135028e-01 -3.80787522e-01 -4.19106334e-02
6.67213440e-01 -6.03613377e-01 -1.16963446e-01 9.59703565e-01
8.36445034e-01 -4.45089161e-01 6.48796499e-01 -1.61226583e+00
-6.16061151e-01 -1.55848891e-01 -9.64398384e-01 3.39784920e-01
8.70021880e-01 1.12750269e-01 2.16485023e-01 -1.11141455e+00
2.74699628e-01 1.15352941e+00 7.15806663e-01 9.74949241e-01
-9.97137964e-01 -1.32682180e+00 -2.91591316e-01 1.61622405e-01
-1.61508656e+00 -6.26511753e-01 6.35888457e-01 -4.44429457e-01
7.31811166e-01 4.70084548e-01 3.83341521e-01 1.01260829e+00
1.97222486e-01 3.59603465e-01 1.19204366e+00 -6.30928040e-01
-5.85072599e-02 1.33573502e-01 2.06228182e-01 6.81924701e-01
6.29142642e-01 6.49271309e-02 -5.76417506e-01 -5.83313286e-01
6.75379872e-01 -6.56872336e-03 -1.16208747e-01 1.36152074e-01
-4.22852218e-01 4.67731386e-01 -2.32597098e-01 3.07603061e-01
-2.79056370e-01 -1.86696067e-01 3.62108409e-01 9.34658423e-02
-1.52775431e-02 -8.63052532e-02 6.35635629e-02 -1.43516779e-01
-7.06670344e-01 3.03237252e-02 8.33373368e-01 3.81090552e-01
4.13018465e-01 1.19649604e-01 5.15204147e-02 5.69506705e-01
5.12848854e-01 8.51632118e-01 2.70783871e-01 -1.68152362e-01
3.47286761e-01 6.78507745e-01 -4.83537000e-03 -1.02959144e+00
-2.03495294e-01 3.69658291e-01 -7.40177095e-01 7.51619220e-01
4.73882645e-01 -8.08514804e-02 -8.36405039e-01 1.22618270e+00
2.47519612e-01 4.58504498e-01 1.26262149e-02 5.10137916e-01
9.80314851e-01 2.09712580e-01 1.11051567e-01 -1.90894723e-01
1.79158509e+00 -1.48779094e-01 -7.26967990e-01 3.23274940e-01
7.15913698e-02 -1.32508469e+00 6.59066975e-01 4.61673558e-01
-1.02680707e+00 -4.48063374e-01 -1.20571411e+00 4.21745449e-01
-6.02176189e-01 1.37485966e-01 4.52537507e-01 1.88974023e+00
-8.72471690e-01 2.32581645e-01 -5.02162993e-01 -4.41614002e-01
4.57155734e-01 8.57952535e-01 -7.62481987e-01 8.79683644e-02
-1.03899121e+00 8.46084297e-01 -2.43532419e-01 1.50866657e-01
-4.62302685e-01 -3.83915365e-01 -7.56116211e-01 -1.55917928e-01
-8.61018822e-02 -2.57352535e-02 8.44276190e-01 -9.03711736e-01
-1.43079543e+00 1.13571882e+00 -2.63266295e-01 -4.05222513e-02
1.25967741e-01 1.32899033e-02 -1.06069899e+00 1.66821688e-01
-4.77693319e-01 -6.14632294e-03 1.23378515e+00 -9.58950698e-01
-4.17279541e-01 -5.91721296e-01 -7.67392069e-02 -1.15091152e-01
-4.37080681e-01 7.62038887e-01 -1.56657159e-01 -4.53684211e-01
-4.47870307e-02 -9.76843596e-01 3.75809938e-01 1.04059726e-01
-1.44808650e-01 1.22585215e-01 1.49914014e+00 -6.96455300e-01
1.29316938e+00 -2.25669646e+00 -5.92803717e-01 6.33376539e-01
-1.86922982e-01 9.23929274e-01 9.83342752e-02 5.81107616e-01
-2.16880679e-01 4.24768209e-01 6.38676882e-02 -4.05895114e-01
-1.10415116e-01 -6.13212250e-02 -2.20628098e-01 7.50699937e-01
1.40055999e-01 4.88599360e-01 -2.73507625e-01 -8.23070779e-02
2.68575251e-01 9.60237622e-01 1.28372461e-01 9.04796347e-02
7.60185480e-01 2.28908882e-01 -2.23130971e-01 1.10846376e+00
1.12344205e+00 2.30929971e-01 2.93828808e-02 -2.60086596e-01
2.67899362e-03 -2.74307411e-02 -1.66264176e+00 9.78739262e-01
-1.27855346e-01 6.50053084e-01 2.59934366e-01 -5.95529437e-01
1.04415250e+00 8.11055481e-01 3.10463995e-01 -6.97937965e-01
3.46852213e-01 4.42494243e-01 3.06064356e-02 -5.60749471e-01
4.40944940e-01 -1.35695515e-02 1.63018018e-01 7.81413674e-01
-2.27776438e-01 5.03769040e-01 2.06390098e-02 -2.15590611e-01
8.56534243e-01 -1.57906964e-01 5.05909562e-01 1.26978040e-01
1.13233054e+00 -8.95041943e-01 1.22516900e-01 4.67267424e-01
-4.70776469e-01 3.98969471e-01 1.67722762e-01 -2.99244374e-01
-5.85452139e-01 -1.08203661e+00 -1.40822366e-01 6.03672683e-01
1.69720739e-01 -2.26356670e-01 -8.45839024e-01 -7.87992418e-01
1.22713186e-01 -1.58836126e-01 -4.89887297e-01 -5.66175021e-03
-5.20483911e-01 -6.96426153e-01 1.22012401e+00 4.54940140e-01
5.81221879e-01 -9.40502107e-01 -6.20517015e-01 -1.67996123e-01
1.79280490e-01 -1.04418945e+00 -2.49081299e-01 -4.80143130e-01
-5.32570779e-01 -1.34145844e+00 -7.22032249e-01 -6.57830656e-01
8.87418151e-01 2.18778610e-01 6.89742923e-01 6.98618293e-01
-7.92518616e-01 6.46580100e-01 -6.72956333e-02 -7.28798509e-01
-3.58623356e-01 -2.16351524e-01 4.50188786e-01 5.05508006e-01
7.90573657e-01 -2.99336046e-01 -6.85111165e-01 4.57575768e-01
-8.46915424e-01 -8.42712581e-01 3.86891603e-01 5.18146694e-01
3.34972367e-02 2.82278322e-02 2.54244864e-01 -6.58957362e-01
7.46022522e-01 -3.56267765e-02 -4.73268926e-01 3.87009621e-01
-5.12431264e-01 -3.44309121e-01 8.87187272e-02 -7.28971660e-01
-8.03507864e-01 1.66433886e-01 -2.80890077e-01 -1.93493083e-01
-3.17495078e-01 6.21210486e-02 -3.88190359e-01 -9.42996919e-01
6.49112046e-01 2.23263085e-01 3.88756007e-01 -3.22819471e-01
-1.25127465e-01 9.04576302e-01 4.65829879e-01 -5.42098701e-01
1.10865653e+00 5.61754763e-01 3.79406005e-01 -1.05514979e+00
6.42518178e-02 -6.65257633e-01 -6.20831251e-01 -5.15647411e-01
4.53910142e-01 -3.39758813e-01 -1.39038706e+00 1.15280211e+00
-9.52949762e-01 5.21506608e-01 4.05509740e-01 1.61722720e-01
3.73249240e-02 7.17250407e-01 -6.26538455e-01 -1.45479345e+00
-7.01738715e-01 -1.17711115e+00 9.01658177e-01 6.21361196e-01
-1.37926877e-01 -8.10123742e-01 -1.45869166e-01 2.12176442e-01
7.53572822e-01 3.78148615e-01 4.03341532e-01 -4.35253471e-01
-2.60639101e-01 -7.64797568e-01 -1.76963300e-01 2.41526335e-01
6.34728193e-01 3.73933464e-01 -1.43552935e+00 -4.55192029e-01
-1.05871312e-01 1.04853790e-02 2.75298685e-01 -1.45760514e-02
8.85788858e-01 -2.77897418e-01 -3.14466089e-01 3.63842964e-01
1.40470123e+00 5.30177295e-01 1.24345350e+00 3.93473297e-01
3.86164635e-01 7.24213302e-01 3.34770501e-01 3.45546603e-01
-9.46148187e-02 7.26753950e-01 3.18384439e-01 -3.11602771e-01
1.06941365e-01 2.43322670e-01 4.34320569e-01 6.59029931e-02
-4.26589310e-01 -9.15788710e-02 -8.11239421e-01 -5.42646684e-02
-1.27553809e+00 -1.26402295e+00 -1.33982643e-01 2.51087666e+00
3.21761012e-01 4.04144451e-02 4.08076435e-01 8.42505097e-01
9.01842296e-01 -2.34622777e-01 -2.02147454e-01 -6.82072043e-01
-6.46397248e-02 7.66048551e-01 5.42403996e-01 1.99636146e-01
-1.22115791e+00 4.55922574e-01 6.38530016e+00 5.29829979e-01
-1.74111450e+00 -1.26452148e-01 5.44400990e-01 3.11673313e-01
3.33752751e-01 -3.35186124e-01 -1.10042191e+00 4.62795675e-01
7.98262179e-01 6.48650229e-02 -7.10114930e-03 6.00793004e-01
-7.29565620e-02 -1.31808922e-01 -9.80858862e-01 1.46396780e+00
5.10299087e-01 -1.04307675e+00 1.90756708e-01 4.22569424e-01
1.18418388e-01 -1.04317856e+00 5.29930711e-01 -2.19970092e-01
-5.81308186e-01 -1.24139810e+00 3.48648131e-01 3.18075091e-01
9.84550893e-01 -8.53579164e-01 8.23608041e-01 -2.12307379e-01
-1.68670118e+00 -1.37581035e-01 1.06475040e-01 -8.02076533e-02
1.67500421e-01 -2.00710058e-01 -5.54890037e-01 5.71343660e-01
5.17298520e-01 1.83624551e-02 -5.60204625e-01 9.85781074e-01
-3.07255555e-02 4.81713861e-01 -4.18890238e-01 1.51175885e-02
-2.63817608e-01 3.40461917e-02 2.34514669e-01 1.19166589e+00
4.72182214e-01 2.88974077e-01 -3.40335406e-02 2.40081221e-01
1.80758789e-01 2.93005139e-01 -6.01711452e-01 3.98278758e-02
4.80001658e-01 1.25266922e+00 -8.10343504e-01 3.30758952e-02
-6.12034380e-01 1.04156899e+00 -5.49102247e-01 9.30558667e-02
-5.25374174e-01 -3.28728646e-01 9.66310918e-01 3.80538791e-01
3.94541584e-02 -1.68048292e-01 -2.11335927e-01 -7.57561088e-01
1.66660815e-01 -1.14434552e+00 5.99621713e-01 -3.61658633e-01
-1.17546487e+00 4.81584191e-01 -2.43144915e-01 -1.45612574e+00
-2.64868617e-01 -9.95862186e-01 -6.87466979e-01 1.35086167e+00
-1.19463193e+00 -1.37262809e+00 -4.84209478e-01 7.49178708e-01
1.98358651e-02 -5.36321163e-01 1.25355136e+00 7.20810235e-01
-6.85573578e-01 1.29755497e+00 -4.65227038e-01 5.09360194e-01
5.79473674e-01 -7.25385129e-01 6.30750537e-01 1.05024850e+00
-4.39807493e-03 1.01203525e+00 3.01696360e-01 -5.30508339e-01
-1.67255688e+00 -1.81521714e-01 6.80210710e-01 -4.37399477e-01
2.93876499e-01 -4.07019138e-01 -8.17679405e-01 8.33756998e-02
1.63909663e-02 -2.34581716e-03 1.18723106e+00 -1.72364756e-01
-6.57224536e-01 -2.03507334e-01 -1.67049098e+00 4.70807552e-01
5.72223604e-01 -8.70332181e-01 -8.93564746e-02 -4.27985132e-01
-5.56032419e-01 -4.51769948e-01 -5.58168769e-01 2.34139070e-01
1.31717885e+00 -9.29890573e-01 1.19856727e+00 -2.28338078e-01
-3.22593391e-01 -3.96940500e-01 -1.12672694e-01 -3.04307431e-01
2.36219391e-02 -8.61048043e-01 -3.26810122e-01 1.60331106e+00
3.07436287e-01 -7.28808939e-01 1.02565217e+00 9.37592924e-01
7.51360178e-01 -4.29239362e-01 -7.88342118e-01 -8.52045834e-01
-4.92723703e-01 -1.31660387e-01 7.60366440e-01 8.82807910e-01
-8.96385610e-02 -7.84698427e-02 -3.76756251e-01 4.79352951e-01
8.80721986e-01 -5.53444147e-01 9.65058327e-01 -1.44710004e+00
1.01250745e-01 -4.88634408e-01 -1.15545428e+00 -2.92179227e-01
-8.24277699e-02 -3.30866665e-01 -5.61487198e-01 -1.02598250e+00
-5.36826588e-02 -1.92534208e-01 -3.69011611e-01 4.95991141e-01
-8.07546005e-02 9.09156561e-01 1.89737678e-01 1.48351729e-01
3.18712413e-01 -3.31707090e-01 4.54490364e-01 -9.94132757e-02
-8.40216428e-02 2.35397667e-01 -5.37123263e-01 5.44816732e-01
6.92438543e-01 -1.16092205e-01 -3.14124137e-01 3.36210757e-01
-9.94046777e-03 -2.83484519e-01 1.97499618e-01 -1.02075255e+00
2.54494697e-01 1.22352645e-01 6.95972800e-01 -4.63629454e-01
4.82078850e-01 -9.48655784e-01 2.10332125e-01 6.59240842e-01
3.35950226e-01 5.16477883e-01 4.45989996e-01 -5.20256460e-02
5.55642545e-02 -3.77863228e-01 8.82343650e-01 8.85637477e-02
-7.30676413e-01 3.92448545e-01 -4.46056306e-01 -5.22635579e-01
1.13572824e+00 -8.10749888e-01 -5.28641522e-01 -2.64533192e-01
-5.33790410e-01 -5.60614347e-01 4.75672305e-01 4.86659050e-01
8.51595163e-01 -1.14918137e+00 -6.27805114e-01 7.72073567e-01
5.73374480e-02 -8.69889081e-01 5.06225564e-02 4.96054560e-01
-6.89423084e-01 1.00613773e-01 -5.56801021e-01 -4.50087041e-01
-2.38258553e+00 4.45415109e-01 3.48751396e-01 1.79765016e-01
-2.76889682e-01 5.21623433e-01 -1.94317594e-01 1.69066712e-01
3.59080791e-01 2.95327783e-01 -4.89073962e-01 2.46754572e-01
1.39152312e+00 4.52845156e-01 3.74666691e-01 -9.70241189e-01
-6.32222950e-01 9.69265342e-01 -1.48056060e-01 -2.86592782e-01
5.44019997e-01 1.42316027e-02 2.55890563e-02 -4.73911911e-02
1.07151282e+00 2.01531246e-01 -6.40586853e-01 1.34896636e-01
3.50056104e-02 -8.25654387e-01 -4.89954174e-01 -5.61352372e-01
-1.08773530e+00 8.37074578e-01 1.28651345e+00 3.68995398e-01
1.22520483e+00 -5.86850047e-01 6.44787908e-01 9.69831347e-02
4.19736385e-01 -8.37093353e-01 1.58651486e-01 1.31638035e-01
5.98190784e-01 -1.02711248e+00 1.48283482e-01 -7.02351689e-01
-2.77431250e-01 1.29434013e+00 5.48049033e-01 1.75893918e-01
1.04044056e+00 6.31341457e-01 4.54092503e-01 -1.52996704e-01
-4.55880016e-02 -1.02388121e-01 5.13562977e-01 1.03622973e+00
6.17606878e-01 -1.58819899e-01 -3.62597823e-01 2.35292867e-01
4.14420478e-03 -1.13113753e-01 3.58541191e-01 1.23459482e+00
-2.30353009e-02 -1.63347960e+00 -9.11292493e-01 3.08714479e-01
-1.07703209e+00 1.05669305e-01 -6.97317481e-01 7.05407739e-01
9.46861356e-02 1.01504493e+00 -9.38785225e-02 -4.37720150e-01
4.44963217e-01 2.53358692e-01 6.49834692e-01 -2.45076269e-01
-9.78401303e-01 -2.24265233e-01 -3.59057039e-02 -2.91098267e-01
-6.00144804e-01 -7.32509017e-01 -9.02559519e-01 -7.07744420e-01
-3.09495866e-01 -1.96635142e-01 1.06742930e+00 9.08078611e-01
2.24907130e-01 -9.40446854e-02 7.70956993e-01 -7.32463300e-01
-1.27896160e-01 -7.05332637e-01 -6.89684570e-01 5.69103658e-01
3.41658473e-01 -8.03048968e-01 -2.50562638e-01 5.65085262e-02] | [13.168493270874023, 1.1412678956985474] |
cb53a284-e850-4a4b-9ecf-e901a69a2d57 | transient-hemodynamics-prediction-using-an | 2302.06557 | null | https://arxiv.org/abs/2302.06557v1 | https://arxiv.org/pdf/2302.06557v1.pdf | Transient Hemodynamics Prediction Using an Efficient Octree-Based Deep Learning Model | Patient-specific hemodynamics assessment could support diagnosis and treatment of neurovascular diseases. Currently, conventional medical imaging modalities are not able to accurately acquire high-resolution hemodynamic information that would be required to assess complex neurovascular pathologies. Therefore, computational fluid dynamics (CFD) simulations can be applied to tomographic reconstructions to obtain clinically relevant information. However, three-dimensional (3D) CFD simulations require enormous computational resources and simulation-related expert knowledge that are usually not available in clinical environments. Recently, deep-learning-based methods have been proposed as CFD surrogates to improve computational efficiency. Nevertheless, the prediction of high-resolution transient CFD simulations for complex vascular geometries poses a challenge to conventional deep learning models. In this work, we present an architecture that is tailored to predict high-resolution (spatial and temporal) velocity fields for complex synthetic vascular geometries. For this, an octree-based spatial discretization is combined with an implicit neural function representation to efficiently handle the prediction of the 3D velocity field for each time step. The presented method is evaluated for the task of cerebral hemodynamics prediction before and during the injection of contrast agent in the internal carotid artery (ICA). Compared to CFD simulations, the velocity field can be estimated with a mean absolute error of 0.024 m/s, whereas the run time reduces from several hours on a high-performance cluster to a few seconds on a consumer graphical processing unit. | ['Andreas Maier', 'Annette Birkhold', 'Markus Kowarschik', 'Laura Pfaff', 'Maximilian Rohleder', 'Mareike Thies', 'Fabian Wagner', 'Katharina Zinn', 'Noah Maul'] | 2023-02-13 | null | null | null | null | ['tomographic-reconstructions'] | ['medical'] | [-3.07215452e-01 -3.91726047e-01 5.75213194e-01 -7.13102743e-02
-3.97996515e-01 -2.60488629e-01 3.54141623e-01 5.08487821e-01
-6.85624361e-01 1.10865283e+00 -3.03590328e-01 -8.17190051e-01
-2.34042071e-02 -8.72746050e-01 -2.08577648e-01 -7.78334379e-01
-4.61185604e-01 8.18855107e-01 2.68664688e-01 2.79749278e-02
4.83820736e-02 1.14279962e+00 -1.39303005e+00 5.03019243e-02
1.24094415e+00 1.06783640e+00 1.51218966e-01 7.89505899e-01
-3.33604902e-01 6.48511946e-01 -2.93481886e-01 2.54800946e-01
3.26139033e-02 -4.23941374e-01 -5.66946685e-01 -2.59448230e-01
2.08570838e-01 -6.02619052e-01 -2.30973572e-01 6.68376863e-01
9.03054833e-01 4.65678215e-01 8.05949271e-01 -5.53432286e-01
4.21709210e-01 -6.50527403e-02 -3.32485884e-02 5.68454444e-01
-1.01137236e-01 4.00055200e-01 -1.84678689e-01 -7.43886352e-01
6.37886703e-01 7.35065460e-01 5.33212841e-01 4.16459084e-01
-1.15491128e+00 -1.04571640e-01 -2.13274330e-01 1.97967649e-01
-1.20846939e+00 -1.04119480e-01 4.76645648e-01 -8.90447021e-01
7.38995969e-01 2.50729293e-01 1.24192703e+00 6.73901975e-01
5.73729634e-01 4.73433203e-04 1.41164720e+00 -1.45256408e-02
6.59406066e-01 5.08596338e-02 2.27335226e-02 5.59386015e-01
3.54514182e-01 4.64341968e-01 8.32685605e-02 -1.91758022e-01
1.34240603e+00 -4.26411986e-01 -2.27554262e-01 -3.86865467e-01
-8.45014930e-01 8.38109493e-01 3.27018827e-01 4.66335535e-01
-9.23663437e-01 1.00820631e-01 7.46692836e-01 -1.49251565e-01
5.76004505e-01 4.11028028e-01 -3.14550340e-01 -5.30510843e-01
-1.11919224e+00 5.02814829e-01 9.00498688e-01 2.79735684e-01
1.95825592e-01 4.22512829e-01 -1.07113630e-01 3.50085139e-01
3.92478891e-02 4.48653638e-01 3.37510914e-01 -1.12735558e+00
-1.76981851e-01 3.93834189e-02 4.49520171e-01 -8.01270008e-01
-7.68714905e-01 -8.36199403e-01 -1.34377587e+00 6.99311376e-01
8.28799009e-01 -4.10446674e-01 -8.61913145e-01 9.31046963e-01
6.83582902e-01 6.52945399e-01 -4.33483049e-02 1.35221994e+00
7.77448297e-01 4.19053972e-01 2.63577014e-01 -4.88412648e-01
1.24914575e+00 -5.72575212e-01 -5.03870249e-01 1.28395125e-01
8.54980707e-01 -5.14892697e-01 6.48543835e-01 3.17955583e-01
-1.37491620e+00 -2.98967183e-01 -6.25779450e-01 3.49505424e-01
-1.76621884e-01 -2.04797357e-01 5.16223848e-01 8.00521731e-01
-9.55304503e-01 9.85540688e-01 -1.30971217e+00 1.67648315e-01
3.90008122e-01 -5.29203936e-02 -1.36800006e-01 1.66294947e-01
-1.21186447e+00 1.31362951e+00 -3.19463052e-02 3.67056906e-01
-8.07580054e-01 -1.38932002e+00 -4.95990038e-01 5.48761487e-02
-1.80050254e-01 -1.15157795e+00 1.07700431e+00 -3.43711853e-01
-1.84780836e+00 4.83161271e-01 -2.34175086e-01 -5.29591799e-01
1.26605284e+00 1.44823361e-02 -1.58452854e-01 5.86257994e-01
-1.58260286e-01 -1.33681610e-01 4.80086476e-01 -1.12453187e+00
-5.07626310e-03 -2.00959623e-01 -2.42642671e-01 2.15800032e-02
3.50258797e-01 -1.25678480e-01 1.49216065e-02 -4.75445986e-01
4.05348204e-02 -7.02545762e-01 -8.38497102e-01 2.40802720e-01
5.72256595e-02 5.84089398e-01 1.89737901e-01 -9.24916148e-01
8.97601545e-01 -1.53391743e+00 5.86704127e-02 3.26998234e-01
3.62893850e-01 6.75403476e-01 5.00070035e-01 6.41177148e-02
-1.34557590e-01 -1.86090082e-01 -3.76085103e-01 9.39983204e-02
-6.52422011e-01 -1.10322563e-02 -7.03433668e-03 6.63584054e-01
-2.49369308e-01 6.85462475e-01 -8.62090290e-01 -3.66168708e-01
6.88723862e-01 9.11003232e-01 -6.16049469e-01 2.01166525e-01
9.81881768e-02 1.49351931e+00 -6.54569030e-01 2.81915236e-02
1.04157996e+00 -1.71751946e-01 1.81384087e-01 -1.32324770e-01
-7.44928718e-01 5.80722801e-02 -1.04506981e+00 1.45970845e+00
-9.81160462e-01 2.44998336e-01 4.75839496e-01 -1.26792693e+00
8.24089825e-01 6.39200389e-01 9.31181014e-01 -1.01506579e+00
6.77340150e-01 5.37851751e-01 3.09635043e-01 -6.88409865e-01
-8.31671581e-02 -6.15548015e-01 5.35650730e-01 3.15734625e-01
-4.79581267e-01 -2.58528709e-01 1.07335851e-01 -1.34237349e-01
9.26879466e-01 6.78296480e-03 -3.23730670e-02 -4.97597903e-01
9.08883274e-01 2.94295579e-01 3.25979710e-01 5.30187428e-01
-2.66216725e-01 2.73375571e-01 5.26986539e-01 -8.53233874e-01
-1.21110225e+00 -1.04661822e+00 -6.71941578e-01 1.85625494e-01
1.11698337e-01 2.57366411e-02 -8.92827094e-01 1.77434012e-01
6.68022484e-02 6.31238639e-01 -3.81708294e-01 1.33013189e-01
-9.84352648e-01 -7.63656199e-01 2.42660865e-01 4.92967367e-01
3.02976161e-01 -7.84044027e-01 -1.25677657e+00 7.06969619e-01
3.50408182e-02 -1.11299646e+00 4.67342883e-01 -1.80502981e-01
-1.25086641e+00 -1.13549781e+00 -1.08105385e+00 -3.72398764e-01
4.61124152e-01 -3.51542950e-01 8.47286403e-01 1.12632483e-01
-6.16143584e-01 1.47409141e-01 -2.55771726e-01 -5.23079969e-02
-5.89627683e-01 -2.50551403e-01 2.60332897e-02 -2.37666458e-01
-2.67714709e-01 -8.91178608e-01 -9.06977713e-01 2.06892118e-02
-4.31539953e-01 1.55229405e-01 3.44519973e-01 7.46377110e-01
6.32580757e-01 -2.86111921e-01 5.20601213e-01 -7.29894400e-01
6.56895697e-01 -5.04249454e-01 -8.57102036e-01 -1.23855442e-01
-4.23539907e-01 -1.46518350e-01 1.06871033e+00 -4.17361885e-01
-1.17418742e+00 -1.33612379e-01 -3.88281792e-01 -5.92354715e-01
-3.80581081e-01 7.51110256e-01 4.87442166e-01 -3.16306949e-01
6.97026134e-01 4.20490950e-01 3.66306663e-01 -6.09835088e-01
1.30559906e-01 2.11796284e-01 4.63033348e-01 -9.78333473e-01
4.03852165e-01 4.71264422e-01 7.17511415e-01 -1.05112946e+00
-5.17988689e-02 -1.55706406e-01 -9.47568893e-01 -5.54466963e-01
7.53275394e-01 -5.46216130e-01 -9.54457164e-01 4.97612715e-01
-1.18813932e+00 -4.88596588e-01 -2.71794736e-01 8.83881390e-01
-8.31456900e-01 4.31903362e-01 -6.97444260e-01 -7.91694105e-01
-3.95691901e-01 -1.36152470e+00 3.45998406e-01 8.45012963e-02
-6.32279143e-02 -1.22437227e+00 -3.76239941e-02 -5.56420572e-02
1.05102301e+00 9.70648527e-01 1.14641690e+00 5.06619513e-02
-4.36241269e-01 -1.56996295e-01 -2.08421588e-01 -6.36711344e-02
-3.85043740e-01 -2.10759774e-01 -7.25271046e-01 -1.38898045e-01
1.69240907e-01 1.96493849e-01 3.89178902e-01 1.08753324e+00
1.22496879e+00 7.35863969e-02 -3.73726159e-01 6.84974432e-01
1.26813424e+00 3.35807562e-01 4.67913508e-01 -3.44563276e-02
4.00446951e-01 5.12098253e-01 3.00057173e-01 8.48284245e-01
3.11321653e-02 5.93297243e-01 1.64286405e-01 -3.88456523e-01
-1.12205409e-01 3.35837394e-01 -3.85207206e-01 4.66142088e-01
-7.28502929e-01 2.56055713e-01 -1.17686832e+00 2.41759658e-01
-1.48641491e+00 -7.09247470e-01 -6.94259048e-01 2.36930251e+00
6.49050355e-01 -6.90912036e-03 1.68476045e-01 -6.81579858e-02
5.12215376e-01 -3.45455974e-01 -6.04486108e-01 -5.04505694e-01
3.11785579e-01 6.55949712e-01 3.46865296e-01 7.93686390e-01
-8.48088443e-01 4.62980360e-01 6.03470898e+00 1.73289210e-01
-1.68150020e+00 3.65093619e-01 5.83254874e-01 -3.92987989e-02
8.02166387e-02 -1.19285502e-01 -1.91473305e-01 4.41634178e-01
1.39230716e+00 -3.99202138e-01 1.44015834e-01 5.69509506e-01
8.60963106e-01 -4.03841734e-01 -6.07440412e-01 8.02654147e-01
-7.02498734e-01 -1.60979331e+00 -2.61965990e-01 5.54589974e-03
3.54520231e-01 -9.95926335e-02 -2.26358831e-01 -1.13173723e-02
-3.54867816e-01 -1.19028819e+00 2.95615077e-01 8.64175022e-01
1.18267441e+00 -6.94037199e-01 7.43522882e-01 5.79939425e-01
-1.07642519e+00 3.62814009e-01 -1.78825289e-01 -1.87928498e-01
7.86845207e-01 8.95300925e-01 -7.33652532e-01 5.36151111e-01
4.71238524e-01 2.92541832e-01 6.93914071e-02 1.47842014e+00
2.75175661e-01 3.93925250e-01 -2.60053337e-01 8.19602460e-02
3.71476144e-01 -4.29226100e-01 6.62217319e-01 9.41932559e-01
5.63373446e-01 5.12197375e-01 5.99751920e-02 9.74926353e-01
6.50853932e-01 3.54958028e-01 -1.41432643e-01 3.30456942e-01
6.46826327e-02 1.10054374e+00 -8.52150559e-01 -4.92759794e-01
-2.20211938e-01 4.43869352e-01 8.89157355e-02 3.99035960e-01
-8.34952891e-01 -1.09262943e-01 7.69236624e-01 8.35913420e-01
3.34345549e-02 -6.55751526e-01 -4.08989310e-01 -1.04987800e+00
-3.26895773e-01 -1.07639104e-01 2.02824106e-03 -4.60295677e-01
-1.07281518e+00 8.87407839e-01 1.36343271e-01 -1.07919836e+00
-4.01030362e-01 -7.64157236e-01 -6.65598691e-01 1.56061769e+00
-1.60277343e+00 -5.12296677e-01 -5.57937860e-01 4.94079590e-01
-9.63259116e-02 9.40751433e-02 1.04684293e+00 5.28406322e-01
-3.37840497e-01 -4.47468758e-02 1.14739977e-01 5.80909662e-02
2.01340064e-01 -1.09537351e+00 1.58553168e-01 6.26950860e-01
-1.00542140e+00 3.49913418e-01 8.79036009e-01 -7.22859263e-01
-9.74193990e-01 -9.46707964e-01 6.62907124e-01 2.80132681e-01
4.31081772e-01 2.11696804e-01 -1.22979438e+00 -1.61842220e-02
-2.23732844e-01 8.73845160e-01 5.54841340e-01 -3.98320615e-01
2.44312778e-01 1.42214969e-01 -1.31801152e+00 2.85742611e-01
6.46853030e-01 -9.72235948e-02 -1.48234805e-02 1.66402429e-01
-3.91090140e-02 -9.07559454e-01 -1.31293559e+00 3.20957333e-01
4.56754982e-01 -8.34692597e-01 1.08083105e+00 -8.15018833e-01
3.18486631e-01 -2.24095479e-01 4.65148807e-01 -1.38626218e+00
-1.79011136e-01 -4.72907364e-01 -1.29221335e-01 2.11829439e-01
-6.34825230e-02 -8.21548879e-01 7.75753319e-01 9.59062099e-01
-2.95125544e-01 -1.01114237e+00 -1.39072657e+00 -5.19010961e-01
7.85454750e-01 -4.49907273e-01 2.14227498e-01 8.09479117e-01
-6.05823845e-02 -3.61186355e-01 1.24010257e-02 1.67025819e-01
6.67778671e-01 2.02304125e-01 2.40544319e-01 -1.57397056e+00
-2.05386076e-02 -6.25431240e-01 -4.99947965e-01 -6.36603951e-01
2.43057549e-01 -8.34620595e-01 -2.78017193e-01 -1.60502696e+00
-4.30810601e-01 -9.48084235e-01 -2.52205580e-02 -2.65002072e-01
9.47054178e-02 -1.45301759e-01 -2.99474299e-01 -1.84689611e-01
4.40627038e-01 4.10239309e-01 1.86059999e+00 4.43000764e-01
-3.02210599e-01 1.81312099e-01 1.93528503e-01 5.41178405e-01
7.82844007e-01 -1.26969248e-01 -3.06187719e-01 -7.13916197e-02
-3.64993393e-01 9.27514732e-01 6.37386143e-01 -1.29447830e+00
2.85019130e-01 -1.10002726e-01 6.00686252e-01 -3.28990400e-01
2.75134176e-01 -6.84143722e-01 1.98996216e-01 9.97200549e-01
-7.61740953e-02 -1.86137736e-01 5.60210526e-01 1.09930247e-01
-1.96352392e-01 -5.00742253e-03 1.22245324e+00 -4.75870550e-01
-3.81547600e-01 5.46990752e-01 -1.09800029e+00 1.25952557e-01
1.11609566e+00 -8.10782686e-02 6.67974586e-03 -1.97365180e-01
-1.28300738e+00 -2.98882902e-01 1.72735110e-01 -4.14166301e-01
7.73063779e-01 -9.14392412e-01 -8.29751015e-01 3.27355415e-01
-4.99593049e-01 1.68809727e-01 1.05315816e+00 1.34655952e+00
-1.45384634e+00 4.79116976e-01 -6.23545408e-01 -4.46343094e-01
-7.86599576e-01 3.73870045e-01 1.13880587e+00 -2.59453386e-01
-1.03464222e+00 5.96932530e-01 -7.88293630e-02 -7.85240233e-02
-3.06048065e-01 -3.00380975e-01 -4.04014051e-01 -6.00672737e-02
6.05330110e-01 6.48220718e-01 3.08001518e-01 -7.22705960e-01
-2.71228403e-01 5.64518094e-01 3.01698923e-01 -7.34795853e-02
1.29968166e+00 1.72557794e-02 1.55873420e-02 -5.05140610e-02
8.52874994e-01 -4.26012814e-01 -1.28330660e+00 5.91675192e-02
-3.54974598e-01 -4.66481835e-01 6.35232806e-01 -8.80202532e-01
-1.24580038e+00 1.35161352e+00 5.98418772e-01 4.74473462e-02
1.04458535e+00 -5.26935041e-01 9.16713595e-01 -2.59956747e-01
3.15005004e-01 -6.84884369e-01 -4.88727659e-01 4.54944909e-01
7.86339581e-01 -6.88444912e-01 -8.69716853e-02 -7.74785697e-01
-2.95672148e-01 1.55873322e+00 5.33867776e-01 -2.24757954e-01
1.10852182e+00 6.48981988e-01 2.13294074e-01 1.20480638e-02
-4.78932917e-01 6.41693100e-02 4.03913595e-02 6.16787970e-01
6.00451887e-01 1.38087511e-01 -6.36332810e-01 2.22213492e-01
-1.58676680e-03 3.36328983e-01 6.83098376e-01 7.41467953e-01
-2.36441195e-01 -9.89984691e-01 -2.58429676e-01 3.62596750e-01
-2.87877172e-01 7.19109597e-03 6.13690078e-01 4.85050976e-01
3.76527347e-02 5.42657495e-01 3.00819904e-01 3.23334068e-01
2.93135136e-01 8.72425064e-02 7.14422047e-01 -2.69539297e-01
-5.47892332e-01 1.60840556e-01 7.95616303e-03 -6.30795419e-01
-1.99879602e-01 -5.74686587e-01 -1.61198366e+00 -5.12041390e-01
4.37142462e-01 2.13762328e-01 8.28542531e-01 8.47728610e-01
3.49693477e-01 8.97767603e-01 2.76873291e-01 -1.06761706e+00
-1.43085152e-01 -7.20606863e-01 -8.12637687e-01 1.83365151e-01
3.09587091e-01 -9.95499671e-01 -1.97171435e-01 -1.35799959e-01] | [6.403521537780762, 3.2640106678009033] |
db2307a8-39c6-4b1a-8c82-ad0cd4a9bf5e | predicting-human-activities-using-stochastic | 1708.00945 | null | http://arxiv.org/abs/1708.00945v1 | http://arxiv.org/pdf/1708.00945v1.pdf | Predicting Human Activities Using Stochastic Grammar | This paper presents a novel method to predict future human activities from
partially observed RGB-D videos. Human activity prediction is generally
difficult due to its non-Markovian property and the rich context between human
and environments.
We use a stochastic grammar model to capture the compositional structure of
events, integrating human actions, objects, and their affordances. We represent
the event by a spatial-temporal And-Or graph (ST-AOG). The ST-AOG is composed
of a temporal stochastic grammar defined on sub-activities, and spatial graphs
representing sub-activities that consist of human actions, objects, and their
affordances. Future sub-activities are predicted using the temporal grammar and
Earley parsing algorithm. The corresponding action, object, and affordance
labels are then inferred accordingly. Extensive experiments are conducted to
show the effectiveness of our model on both semantic event parsing and future
activity prediction. | ['Song-Chun Zhu', 'Siyuan Qi', 'Siyuan Huang', 'Ping Wei'] | 2017-08-02 | predicting-human-activities-using-stochastic-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Qi_Predicting_Human_Activities_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Qi_Predicting_Human_Activities_ICCV_2017_paper.pdf | iccv-2017-10 | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [ 3.75641793e-01 1.05913363e-01 -7.79274479e-02 -6.40829861e-01
-2.80764382e-02 -4.69709635e-01 6.98004484e-01 5.58691733e-02
4.11177753e-03 5.49420893e-01 7.52639711e-01 -1.66087225e-02
-4.12456580e-02 -7.51178086e-01 -6.23707294e-01 -2.50104845e-01
-5.41992605e-01 2.85540104e-01 8.11798096e-01 2.20899269e-01
-3.67373638e-02 5.82414508e-01 -1.79179454e+00 4.35243309e-01
3.78944427e-01 9.40037251e-01 5.54555655e-01 8.98348570e-01
-2.48501018e-01 1.06681681e+00 -3.44819754e-01 -4.11698222e-02
7.98031390e-02 -7.81807482e-01 -9.52040672e-01 5.19132614e-01
-1.77120045e-01 -4.36346471e-01 -4.46036726e-01 5.84694386e-01
-1.32054180e-01 7.48295426e-01 4.26894248e-01 -1.61908472e+00
-4.51165229e-01 3.71560544e-01 -1.15058705e-01 1.05271682e-01
1.12074828e+00 4.97025847e-02 1.03304136e+00 -3.46476704e-01
1.00669003e+00 1.41193807e+00 1.93169966e-01 7.23903179e-01
-7.66539335e-01 -2.11435854e-01 8.33761036e-01 3.36080164e-01
-1.03047192e+00 8.01772345e-03 8.21523130e-01 -4.54353303e-01
1.41215169e+00 1.75805479e-01 1.25953007e+00 1.15574026e+00
2.91179389e-01 1.08064425e+00 6.90284729e-01 -4.06886131e-01
3.51751417e-01 -6.22517288e-01 1.56631127e-01 9.14047658e-01
-1.89026311e-01 2.11570576e-01 -8.08630407e-01 -5.24949543e-02
9.89382446e-01 3.11075062e-01 8.16734657e-02 -5.69382370e-01
-1.41125679e+00 1.27617016e-01 2.22874433e-01 1.83809340e-01
-5.04295588e-01 4.34471518e-01 6.40749037e-02 -3.33825856e-01
2.14665160e-01 -1.81435809e-01 -5.50247014e-01 -2.66666800e-01
-3.20775449e-01 5.03222644e-01 7.01614380e-01 1.19314253e+00
5.77905476e-01 -5.15665889e-01 -1.83127075e-01 1.40192777e-01
3.49860370e-01 3.65843117e-01 2.25705236e-01 -1.24807239e+00
4.80685830e-01 8.72148573e-01 4.61537778e-01 -9.04495656e-01
-5.71125805e-01 3.80711854e-01 -1.25940025e-01 -1.60104394e-01
3.55200112e-01 2.52871633e-01 -1.02062714e+00 1.71435320e+00
6.48197770e-01 5.84384978e-01 -3.26238922e-03 8.68359804e-01
5.51163912e-01 6.30924523e-01 8.43722641e-01 -3.17736328e-01
1.46488178e+00 -9.02969122e-01 -8.50784540e-01 -4.74696785e-01
7.43237436e-01 -1.40775219e-01 9.89083230e-01 -6.97827339e-02
-8.80160987e-01 -6.20785356e-01 -6.61802411e-01 -3.73599052e-01
-2.46466830e-01 -7.50890598e-02 8.69569838e-01 1.30922988e-01
-5.34224033e-01 6.22215807e-01 -1.51638436e+00 -5.51101327e-01
1.94746450e-01 -4.67066616e-02 -4.76262957e-01 1.62362382e-01
-8.52815449e-01 5.41879296e-01 7.30851173e-01 -4.83339187e-03
-9.78133023e-01 4.78048474e-02 -1.29570448e+00 -5.92506230e-02
5.86502492e-01 -6.24616325e-01 1.36890805e+00 -7.42991328e-01
-1.34513354e+00 8.29130650e-01 -6.01937950e-01 -3.42504442e-01
1.40331671e-01 -3.65554631e-01 -6.54556930e-01 2.48553142e-01
1.59842759e-01 5.87084949e-01 4.96581256e-01 -9.56137121e-01
-1.14225936e+00 -7.37013340e-01 1.24913134e-01 5.73069453e-01
2.88467109e-01 5.00150807e-02 -5.93032718e-01 -6.41696513e-01
6.62622809e-01 -1.13104808e+00 -3.57274503e-01 -1.40388444e-01
-3.98475945e-01 -5.60403824e-01 8.07847500e-01 -6.90509379e-01
1.38232100e+00 -2.17268777e+00 2.57070243e-01 -1.61722302e-01
-9.69475508e-02 -5.96642673e-01 -2.38918569e-02 3.03089917e-01
4.54871766e-02 -2.31068954e-01 -2.54557848e-01 -2.26378694e-01
1.36284664e-01 5.80180943e-01 -3.80570829e-01 1.94276288e-01
3.14293317e-02 1.01144969e+00 -1.28675032e+00 -6.52087450e-01
5.47407627e-01 2.52749085e-01 -2.94269979e-01 5.14046550e-01
-7.92199910e-01 6.31198823e-01 -9.87388492e-01 6.89530611e-01
-6.11137785e-02 -2.46640876e-01 5.94714940e-01 -2.49974560e-02
-1.03412725e-01 5.91289699e-01 -1.18293667e+00 2.03080940e+00
-1.41419405e-02 1.94663554e-01 -5.57604969e-01 -4.64289188e-01
5.55313110e-01 3.11356604e-01 7.28484094e-01 -5.16710579e-01
-1.94409147e-01 -1.59098670e-01 -4.39377218e-01 -9.55712616e-01
4.42308486e-01 -6.38942048e-02 -5.85146487e-01 7.10152149e-01
-8.45553353e-03 1.77984443e-02 4.51469451e-01 2.61226028e-01
1.20796156e+00 1.42351723e+00 6.40333951e-01 2.44779065e-01
3.86849016e-01 1.43176302e-01 8.50489795e-01 6.17063105e-01
-5.79657078e-01 1.87821761e-01 4.95524913e-01 -8.39848101e-01
-6.88111305e-01 -1.36641908e+00 5.33652127e-01 1.36470973e+00
6.34660900e-01 -6.12385511e-01 -6.18415892e-01 -8.44410300e-01
-4.22086716e-01 9.23583031e-01 -6.62938535e-01 -1.42479194e-02
-1.02483213e+00 -3.44063282e-01 5.07272370e-02 9.98917282e-01
4.33137655e-01 -1.53850460e+00 -1.42456234e+00 4.11962628e-01
-5.33199608e-01 -1.56243408e+00 -5.00136197e-01 -9.84607358e-03
-1.06727135e+00 -1.28911746e+00 -4.93948571e-02 -7.99283326e-01
5.74978530e-01 2.39960980e-02 1.09144664e+00 -1.26962930e-01
-2.34451249e-01 6.20022297e-01 -4.36193705e-01 -1.89997584e-01
-2.06132814e-01 -5.44973850e-01 -1.40141219e-01 -1.14669828e-02
5.53093433e-01 -6.72352552e-01 -6.86451554e-01 2.94524878e-01
-6.50505543e-01 5.53915501e-01 2.17875212e-01 2.10850745e-01
9.23232853e-01 3.29876840e-02 -3.49771440e-01 -4.92813140e-01
-1.32175479e-02 -1.90204173e-01 -4.20863390e-01 5.80904663e-01
1.21682502e-01 1.49253443e-01 4.48664837e-02 -5.63604116e-01
-1.67024803e+00 6.52785242e-01 3.22321028e-01 -1.65849458e-02
-7.65184641e-01 2.52569199e-01 -5.12096941e-01 7.46205151e-01
3.19595188e-01 1.29400879e-01 -5.82008958e-01 -4.56563890e-01
4.85435873e-01 9.30276886e-02 6.16172254e-01 -6.82247400e-01
1.76973283e-01 7.20269084e-01 1.69461891e-01 -5.16160369e-01
-1.04963517e+00 -2.27530241e-01 -1.03083479e+00 -4.43408996e-01
1.32605946e+00 -6.61083579e-01 -7.49604106e-01 3.70426893e-01
-1.23554742e+00 -6.37803733e-01 -3.70986015e-01 7.24363208e-01
-8.04989934e-01 3.60436291e-01 -6.99482203e-01 -1.00537634e+00
3.54053736e-01 -8.62710357e-01 1.26614869e+00 1.85954988e-01
-5.17949104e-01 -7.95657873e-01 -4.58156653e-02 1.70961618e-01
-6.06617570e-01 6.07233405e-01 8.70805025e-01 -1.84568763e-01
-7.65442491e-01 4.76044696e-03 6.53144866e-02 -2.77283967e-01
1.93083867e-01 -1.49090722e-01 -3.01434129e-01 3.70779783e-01
-1.09378174e-01 1.21982768e-01 2.62724817e-01 5.29829860e-01
1.18741667e+00 3.60415177e-03 -8.49735856e-01 2.16327906e-01
1.15633368e+00 5.32477498e-01 5.46928883e-01 8.50754324e-03
6.91391647e-01 8.23621809e-01 1.01489687e+00 5.52912116e-01
7.61564136e-01 5.88138103e-01 4.33084279e-01 6.41962290e-01
-2.46068642e-01 -8.00486326e-01 3.68059129e-01 1.93919733e-01
-4.69700217e-01 -3.62031907e-01 -9.95740652e-01 4.78233635e-01
-2.16168737e+00 -1.13295841e+00 -2.31762514e-01 1.84135866e+00
3.03377002e-01 1.48234874e-01 2.80686498e-01 -1.22577317e-01
7.74539351e-01 1.61037683e-01 -5.38632095e-01 1.95365548e-02
9.26368982e-02 -1.79229841e-01 7.11020529e-02 3.62980932e-01
-1.04606175e+00 1.19807744e+00 6.45157814e+00 1.10464767e-01
-3.55600417e-01 5.48087060e-02 2.21762016e-01 -1.31349877e-01
-1.08138591e-01 3.35814059e-01 -7.26387918e-01 2.19487637e-01
5.67387998e-01 1.36282682e-01 4.04444814e-01 7.10935652e-01
2.22881123e-01 -6.01343691e-01 -1.37050164e+00 8.22052658e-01
-2.77691018e-02 -8.71094763e-01 6.28341660e-02 -1.63937509e-01
3.51592362e-01 -4.27981555e-01 -5.47562897e-01 2.74778962e-01
4.35045391e-01 -4.65634137e-01 1.23576963e+00 8.26094985e-01
3.47440034e-01 -3.37325335e-01 -3.05807814e-02 5.98996818e-01
-1.70447779e+00 -3.53636891e-01 2.82564819e-01 -4.61788893e-01
8.35795879e-01 1.04660504e-01 -3.70491147e-01 3.54458392e-01
9.60216701e-01 8.66803586e-01 -2.19142422e-01 3.60319734e-01
-8.32111180e-01 1.29666284e-01 -2.62872368e-01 -2.34389365e-01
-4.87205982e-02 -3.90886456e-01 6.33109093e-01 7.33443439e-01
4.44782287e-01 7.51914740e-01 5.15202761e-01 6.04416549e-01
2.96156287e-01 -2.34061584e-01 -5.13169706e-01 -2.13756055e-01
4.03857708e-01 6.76276028e-01 -1.19247437e+00 -5.87400436e-01
-4.80786800e-01 1.21063447e+00 2.06412286e-01 5.41764319e-01
-1.01056814e+00 2.89099813e-02 6.11000419e-01 2.22919777e-01
2.13616759e-01 -6.12047851e-01 -2.85568349e-02 -1.13982689e+00
3.70961487e-01 -2.27389380e-01 7.97351360e-01 -1.25792849e+00
-6.84962034e-01 4.57211107e-01 4.97589231e-01 -1.04246175e+00
-3.48107487e-01 -4.65835541e-01 -4.57742587e-02 2.50287414e-01
-8.03514421e-01 -1.40255940e+00 -3.42475772e-01 7.25295782e-01
6.36666238e-01 4.14060354e-01 8.80106151e-01 -4.34397519e-01
-3.60052556e-01 -4.85738307e-01 -9.76199031e-01 1.16501749e-01
-1.66465655e-01 -1.08772326e+00 5.18684328e-01 1.01121914e+00
4.82172787e-01 2.44212359e-01 6.60630822e-01 -1.31204307e+00
-1.05692959e+00 -8.31016302e-01 1.34652662e+00 -8.21559370e-01
4.84274000e-01 -4.85701591e-01 -5.78354537e-01 1.39895630e+00
-2.44166031e-01 2.91854382e-01 6.43624365e-01 2.95814127e-02
-1.37803346e-01 2.44269371e-01 -8.37863684e-01 9.55409110e-01
2.02292490e+00 -5.75934649e-01 -1.13050973e+00 2.29107752e-01
9.23918843e-01 -6.55275762e-01 -6.27550781e-01 4.75632131e-01
6.75754249e-01 -1.04635513e+00 9.62817132e-01 -7.12826610e-01
2.33780339e-01 -4.30908203e-01 -2.66333669e-01 -6.43512249e-01
-2.52854764e-01 -4.37021106e-01 -6.68268442e-01 6.62134528e-01
1.10497139e-02 2.41455412e-03 9.55153883e-01 8.38006616e-01
-1.39496088e-01 -3.65608782e-01 -8.02343428e-01 -7.59973288e-01
-8.89615715e-01 -1.03016019e+00 9.64023173e-01 6.26526237e-01
2.50469536e-01 1.49943501e-01 -2.34241381e-01 3.09167683e-01
6.56628549e-01 5.99099457e-01 5.28985798e-01 -1.10476673e+00
-2.66279161e-01 -7.57767484e-02 -3.97189438e-01 -1.21659362e+00
3.03111970e-01 -5.54671288e-01 3.21124464e-01 -1.72048008e+00
9.65472162e-02 -1.07620388e-01 -1.49089590e-01 8.42122316e-01
-6.94436207e-02 -2.29546189e-01 6.33740649e-02 1.28492370e-01
-1.06217611e+00 4.50291395e-01 1.34022880e+00 2.23589510e-01
-6.59197569e-01 -1.02831326e-01 2.08229005e-01 1.03213227e+00
5.28454304e-01 -4.81114954e-01 -5.48095703e-01 -4.64803696e-01
2.19511703e-01 7.10252702e-01 5.85820913e-01 -1.08370578e+00
6.94033876e-02 -5.33051193e-01 3.21369916e-01 -6.83663309e-01
4.70229417e-01 -9.26357150e-01 4.93074805e-01 5.11573672e-01
-3.80648911e-01 4.86578122e-02 -1.84001192e-01 9.77299035e-01
-4.38487828e-02 2.64935672e-01 -5.54932468e-02 -4.87680703e-01
-1.35040998e+00 4.58563179e-01 -4.11587924e-01 -3.11417103e-01
1.43556631e+00 -3.76257032e-01 3.27200621e-01 -1.52818099e-01
-1.58803499e+00 6.51650503e-02 3.20840448e-01 7.51997352e-01
7.80970395e-01 -1.38635290e+00 4.39167172e-02 2.18093708e-01
2.02437222e-01 -1.70876062e-03 4.07593787e-01 5.08370280e-01
-3.54538471e-01 1.86863840e-01 -2.34086931e-01 -5.61918139e-01
-1.26614845e+00 7.16334760e-01 1.45643309e-01 -2.80530304e-01
-9.57870543e-01 7.90883780e-01 2.52598166e-01 -3.03653628e-01
3.27090591e-01 -7.02128053e-01 -1.63495556e-01 -4.82580930e-01
3.92589033e-01 5.41198075e-01 -5.71817756e-01 -8.76674712e-01
-5.21575868e-01 3.38326275e-01 6.57904446e-01 -4.25925583e-01
1.20583713e+00 -4.69775140e-01 -2.49740466e-01 6.82740986e-01
6.39898181e-01 -5.03801465e-01 -1.86641264e+00 -1.80536747e-01
4.74956840e-01 -3.84150535e-01 -4.42993492e-01 -5.58398068e-01
-4.85692412e-01 5.99258721e-01 2.71921009e-01 1.40971646e-01
1.23103428e+00 5.48683941e-01 9.66262877e-01 1.46814555e-01
9.34536397e-01 -1.13136053e+00 4.25798595e-01 4.45525438e-01
7.08503962e-01 -7.05040872e-01 -8.18282366e-03 -7.69355476e-01
-7.17570126e-01 8.80359411e-01 6.98265612e-01 9.32085589e-02
6.83590531e-01 8.94935653e-02 -3.49242002e-01 -2.72256345e-01
-6.65638864e-01 -3.89633089e-01 2.25471839e-01 6.42647028e-01
1.45829529e-01 3.47688049e-01 -1.88227862e-01 6.91607118e-01
5.91088831e-02 2.60185659e-01 4.19010073e-02 1.46692920e+00
-3.52849096e-01 -1.03439987e+00 -2.15120882e-01 1.00720689e-01
-2.05582127e-01 4.12078261e-01 -1.49820343e-01 7.86526799e-01
5.17860532e-01 8.24314594e-01 4.80210871e-01 -2.80965358e-01
5.14629662e-01 2.90517181e-01 1.03621364e+00 -7.52499700e-01
-3.98868248e-02 1.49702996e-01 4.85138930e-02 -1.20675385e+00
-9.10636544e-01 -1.16783834e+00 -1.94953156e+00 4.00501907e-01
1.21254191e-01 -1.33718506e-01 4.56722915e-01 1.29127967e+00
3.15883517e-01 4.48628247e-01 1.12935625e-01 -6.74241543e-01
1.71359897e-01 -5.12775242e-01 -7.09079504e-01 8.05175126e-01
-2.16962583e-02 -7.41218269e-01 2.59842575e-02 5.42437911e-01] | [8.051088333129883, 0.5777020454406738] |
c7accbf7-0035-4d4c-95aa-fb5d301ad619 | adaptivepaste-code-adaptation-through | 2205.11023 | null | https://arxiv.org/abs/2205.11023v2 | https://arxiv.org/pdf/2205.11023v2.pdf | AdaptivePaste: Code Adaptation through Learning Semantics-aware Variable Usage Representations | In software development, it is common for programmers to copy-paste or port code snippets and then adapt them to their use case. This scenario motivates the code adaptation task -- a variant of program repair which aims to adapt variable identifiers in a pasted snippet of code to the surrounding, preexisting source code. However, no existing approach has been shown to effectively address this task. In this paper, we introduce AdaptivePaste, a learning-based approach to source code adaptation, based on transformers and a dedicated dataflow-aware deobfuscation pre-training task to learn meaningful representations of variable usage patterns. We evaluate AdaptivePaste on a dataset of code snippets in Python. Results suggest that our model can learn to adapt source code with 79.8% accuracy. To evaluate how valuable is AdaptivePaste in practice, we perform a user study with 10 Python developers on a hundred real-world copy-paste instances. The results show that AdaptivePaste reduces the dwell time to nearly half the time it takes for manual code adaptation, and helps to avoid bugs. In addition, we utilize the participant feedback to identify potential avenues for improvement of AdaptivePaste. | ['Alexey Svyatkovskiy', 'Miltiadis Allamanis', 'Neel Sundaresan', 'Jinu Jang', 'Xiaoyu Liu'] | 2022-05-23 | null | null | null | null | ['program-repair', 'program-repair'] | ['computer-code', 'reasoning'] | [ 1.06137209e-01 1.71612516e-01 -3.01056743e-01 -5.11856735e-01
-6.47324324e-01 -6.36278987e-01 -1.14979044e-01 4.91770446e-01
-6.71680197e-02 2.57011026e-01 6.87655658e-02 -6.89224601e-01
2.29656965e-01 -4.41510439e-01 -9.02316630e-01 1.92144543e-01
-7.56949335e-02 -2.40399599e-01 3.49070489e-01 -1.00156277e-01
7.08250701e-01 1.97176654e-02 -1.47129977e+00 6.92846477e-01
1.16011345e+00 -7.24307224e-02 4.10550982e-01 8.05073977e-01
-5.43485701e-01 7.11277306e-01 -9.19866085e-01 -4.48152602e-01
2.31180966e-01 -4.00430024e-01 -8.89890790e-01 -4.38454151e-01
5.67626178e-01 -2.00716972e-01 2.83994019e-01 9.50576186e-01
4.46198493e-01 -2.21486762e-01 1.13723844e-01 -1.18658650e+00
-8.47225606e-01 1.20389247e+00 -7.06756115e-01 4.69353110e-01
6.02908254e-01 3.03325206e-01 7.60268152e-01 -8.60503674e-01
5.53386152e-01 8.79479706e-01 1.16929948e+00 6.34338200e-01
-1.43852389e+00 -6.16647542e-01 -8.28083605e-02 6.28333390e-02
-1.25343168e+00 -3.49056184e-01 4.81939018e-01 -9.54671264e-01
1.64566123e+00 2.17955232e-01 5.43580353e-01 9.14211333e-01
4.64649886e-01 3.36060017e-01 5.62554598e-01 -8.15487325e-01
1.88831761e-01 4.16398138e-01 4.47171032e-01 7.88706899e-01
3.43158841e-01 -2.12715298e-01 -4.53874528e-01 -6.87775075e-01
2.95952260e-01 -3.62461396e-02 -9.91707072e-02 -3.54202121e-01
-9.19370294e-01 5.24643242e-01 1.95187390e-01 3.82852525e-01
3.37383933e-02 3.88421595e-01 7.23736107e-01 6.74260497e-01
9.52222198e-02 1.06514847e+00 -8.53207707e-01 -7.42444873e-01
-8.66333246e-01 3.04972917e-01 1.08509278e+00 1.29921961e+00
1.09724855e+00 4.60580140e-02 -1.58580095e-01 8.79103720e-01
2.80953139e-01 1.58291608e-01 6.76542342e-01 -7.89626956e-01
6.34012163e-01 1.18885303e+00 -9.80832055e-02 -8.04777801e-01
-1.44942164e-01 -1.26034422e-02 1.06566444e-01 5.88297129e-01
2.27810904e-01 -9.29916129e-02 -5.00808835e-01 1.31577969e+00
6.95957690e-02 -2.01723635e-01 -3.93076003e-01 1.79852843e-01
3.55658501e-01 3.21826249e-01 -4.49667033e-03 1.32571489e-01
9.45222735e-01 -1.01565361e+00 -3.48401546e-01 -4.98644441e-01
1.04301977e+00 -9.22744989e-01 1.78650522e+00 3.25719804e-01
-9.38997984e-01 -4.13688719e-01 -1.17674136e+00 -9.32360440e-03
-2.23861098e-01 1.20080024e-01 5.65296113e-01 9.17233944e-01
-9.52015102e-01 8.15379024e-01 -1.13384795e+00 -5.35129726e-01
3.89402926e-01 1.45631492e-01 -3.05048823e-01 1.05374835e-01
-2.67502397e-01 8.21636200e-01 3.24249774e-01 -4.80672777e-01
-7.93542266e-01 -1.35662186e+00 -8.43982995e-01 2.90077031e-01
2.35234335e-01 -3.74995559e-01 1.85804701e+00 -1.29743659e+00
-1.11935282e+00 4.77582723e-01 -1.85458198e-01 -2.18325723e-02
5.84053993e-02 -4.47426468e-01 -1.70149386e-01 -6.17652833e-01
1.77887082e-01 -7.36958086e-02 8.63000453e-01 -1.05443597e+00
-5.84447920e-01 1.41568277e-02 2.06087515e-01 -4.86196369e-01
-6.39614522e-01 2.96042830e-01 -3.44136357e-01 -7.42364228e-01
-5.23626387e-01 -7.57010579e-01 -1.38356388e-01 5.71348779e-02
-1.74727678e-01 -2.93544643e-02 5.22073448e-01 -9.35055077e-01
2.02956557e+00 -2.36159372e+00 1.46024019e-01 3.55985850e-01
2.06914976e-01 3.97313237e-01 -4.46195662e-01 6.35507286e-01
-3.56026173e-01 4.48200256e-01 -6.76575541e-01 -1.49855390e-01
-6.84731528e-02 1.36179803e-02 -1.59317330e-01 3.11785862e-02
7.02872220e-03 7.11415410e-01 -1.01392734e+00 -1.95189729e-01
-2.27451861e-01 1.99318960e-01 -1.18006158e+00 4.18310642e-01
-5.67498624e-01 -1.98608458e-01 -3.57489973e-01 7.39775360e-01
4.63364244e-01 2.63703968e-02 1.61981553e-01 2.91729271e-01
-2.02810317e-01 4.63456959e-01 -8.80820453e-01 1.89108038e+00
-9.08723652e-01 9.60344791e-01 -3.91871154e-01 -4.60883409e-01
9.83599663e-01 -1.58172939e-02 -1.09983973e-01 -5.83780408e-01
-4.59620118e-01 3.05139899e-01 4.42107543e-02 -1.20421839e+00
4.92388755e-01 4.82544571e-01 -3.02268952e-01 7.99231589e-01
-1.55871108e-01 -5.97625636e-02 1.65051222e-01 1.86748624e-01
1.72994936e+00 4.66814458e-01 7.80856192e-01 -1.78827226e-01
3.87573987e-01 2.87106857e-02 3.85591358e-01 9.89557147e-01
2.11304009e-01 3.76795918e-01 7.17410505e-01 -5.43653369e-01
-1.01734555e+00 -9.21484709e-01 2.22971275e-01 1.50277424e+00
-5.29250622e-01 -1.23933339e+00 -1.17372775e+00 -1.04937565e+00
8.89771134e-02 1.10650539e+00 -6.13901496e-01 -5.12046278e-01
-7.46529222e-01 -3.48524153e-01 6.13118112e-01 6.88126743e-01
1.03314415e-01 -1.21402967e+00 -9.29980755e-01 2.85128534e-01
1.02610618e-01 -1.59581870e-01 -8.93574357e-01 1.33522600e-01
-6.85047626e-01 -1.28809416e+00 -2.66979281e-02 -7.41606712e-01
8.71242464e-01 2.57971473e-02 1.48368490e+00 7.55606592e-01
-5.71612418e-01 6.09065950e-01 -5.79141140e-01 -2.56378859e-01
-1.19234109e+00 4.89483446e-01 -5.52099586e-01 -7.35413492e-01
8.80042672e-01 -6.77647352e-01 -1.88905224e-01 1.48713335e-01
-9.81362164e-01 -2.79170245e-01 5.77592611e-01 6.88210428e-01
6.27910197e-02 -4.43283796e-01 3.28414798e-01 -1.35034823e+00
8.10917318e-01 -7.70244658e-01 -7.47980714e-01 4.88062799e-01
-7.75291622e-01 2.27051407e-01 7.22734809e-01 -5.79157054e-01
-1.10008848e+00 1.92753688e-01 -2.78019786e-01 4.35503982e-02
9.67526659e-02 8.90069187e-01 1.51369140e-01 -3.38249177e-01
1.47574699e+00 -1.42372139e-02 -9.23221782e-02 -8.01394582e-01
2.51968473e-01 9.17497337e-01 4.15295213e-01 -7.31112421e-01
9.96919215e-01 -4.64443952e-01 -8.72896135e-01 -4.90098357e-01
-1.98615462e-01 -3.06168914e-01 -5.30372143e-01 2.61794150e-01
3.45315725e-01 -5.48773587e-01 -3.31725270e-01 3.29218090e-01
-1.16023731e+00 -8.41976702e-01 -3.49129021e-01 -1.61831319e-01
-3.76340717e-01 3.74706686e-01 -2.08533004e-01 -3.60654175e-01
-2.20359892e-01 -1.15319526e+00 5.86541355e-01 2.67474920e-01
-8.43777239e-01 -8.46062660e-01 5.13435721e-01 -1.00684658e-01
9.58170772e-01 5.51369078e-02 1.27695787e+00 -6.70723081e-01
-6.87357664e-01 -1.42210916e-01 1.81520432e-01 3.79509687e-01
4.62564617e-01 3.33998650e-01 -7.02292144e-01 -5.35437047e-01
-2.81578749e-01 8.53383318e-02 3.41512948e-01 -3.22897553e-01
1.39988708e+00 -4.75674242e-01 -4.33079451e-01 6.79242313e-01
1.41164625e+00 1.88095495e-01 6.08749449e-01 7.23597944e-01
6.80262446e-01 1.71745867e-01 5.09008527e-01 7.66296387e-01
3.73830378e-01 5.73405802e-01 4.02308583e-01 3.96059245e-01
-1.61775336e-01 -3.76401126e-01 6.72657430e-01 7.20671237e-01
3.04515243e-01 1.00346752e-01 -1.38143337e+00 8.79248977e-01
-1.58662271e+00 -7.42908239e-01 -1.21048696e-01 2.44909477e+00
1.26737678e+00 -3.08156051e-02 1.01663917e-01 -1.81057438e-01
4.73800153e-01 -4.00381923e-01 -5.46894133e-01 -9.25055027e-01
7.09699929e-01 4.69011545e-01 2.44412631e-01 4.49153215e-01
-6.64367914e-01 6.75765276e-01 6.38055420e+00 3.39400619e-01
-1.08186150e+00 2.58566082e-01 -7.51895308e-02 -8.46609287e-03
-6.85632885e-01 3.50994617e-01 -7.11100817e-01 6.70253575e-01
1.07894397e+00 -5.73739946e-01 8.04088950e-01 1.51624632e+00
-2.18744595e-02 -7.77313188e-02 -1.67353964e+00 5.55574834e-01
1.86497867e-01 -1.24923706e+00 -2.37282231e-01 -4.56160933e-01
6.81193411e-01 -1.25716124e-02 -2.51729757e-01 7.47852504e-01
3.38285387e-01 -9.72236037e-01 5.94200253e-01 4.51433241e-01
7.24314570e-01 -3.55289489e-01 5.30770361e-01 2.56987542e-01
-9.92244124e-01 -4.47519958e-01 -3.21762472e-01 -1.83990106e-01
-3.26325089e-01 2.74731755e-01 -1.39374804e+00 -3.71106565e-02
1.00461328e+00 6.98831677e-01 -1.34115827e+00 1.48197508e+00
-3.08292001e-01 8.23049307e-01 -9.41937119e-02 -1.27341136e-01
-5.59679270e-01 4.56626534e-01 3.51661682e-01 1.79512179e+00
5.72617948e-01 -4.47429150e-01 -1.48994252e-01 1.30999839e+00
-8.25927854e-02 3.40249166e-02 -5.17236590e-01 -2.13322327e-01
9.17704463e-01 9.88419592e-01 -1.74365804e-01 -5.75693771e-02
-5.99575460e-01 9.16087687e-01 5.83811045e-01 4.17555690e-01
-6.74066722e-01 -9.77226317e-01 8.14080179e-01 3.17073256e-01
4.96507198e-01 -1.30048782e-01 -3.61935407e-01 -9.39282417e-01
5.11537611e-01 -1.35985756e+00 1.93356454e-01 -6.76514745e-01
-1.04335332e+00 4.21641290e-01 1.14491738e-01 -9.91456449e-01
-2.92530566e-01 -1.26726225e-01 -1.10453176e+00 8.68255973e-01
-1.05088818e+00 -8.47775698e-01 -5.32442153e-01 1.02695242e-01
6.51927888e-01 -3.34406674e-01 7.93673635e-01 3.57391983e-01
-4.76493031e-01 1.22116399e+00 1.53758690e-01 -4.91187610e-02
7.85996199e-01 -1.48912406e+00 1.12719834e+00 1.17272639e+00
-1.95331857e-01 1.43484044e+00 8.47980320e-01 -7.61139810e-01
-1.31549048e+00 -1.11167228e+00 8.71908367e-01 -7.80476034e-01
6.13234103e-01 -4.36743885e-01 -1.59638786e+00 1.20150006e+00
3.47330093e-01 3.49679552e-02 7.89905250e-01 1.65923849e-01
-8.40411067e-01 -7.29047656e-02 -1.14936256e+00 5.80980659e-01
9.16016221e-01 -6.44849539e-01 -8.15643191e-01 7.06173852e-02
7.80951738e-01 -6.05070472e-01 -1.07491350e+00 -3.30052286e-01
5.02773106e-01 -1.09314585e+00 7.37388968e-01 -5.36481500e-01
6.62208736e-01 -4.39003259e-01 5.02889678e-02 -1.57653713e+00
-2.63365328e-01 -9.53320861e-01 -1.03924550e-01 1.62751400e+00
6.08461916e-01 -6.03363276e-01 4.60999101e-01 9.40176249e-01
-4.14897650e-01 -3.67683560e-01 -6.10079765e-01 -5.93636870e-01
1.34901732e-01 -2.95873135e-01 7.75932550e-01 1.12079012e+00
5.87160587e-01 -1.11146726e-01 4.42038998e-02 3.05217989e-02
1.50300696e-01 -2.31213063e-01 1.14556479e+00 -6.94123626e-01
-8.97871792e-01 -3.79350215e-01 -2.93255985e-01 -7.35653877e-01
1.61831811e-01 -1.00635242e+00 2.16382995e-01 -1.17314351e+00
3.62687528e-01 -5.68404913e-01 -8.88454169e-02 9.59708929e-01
-5.80559731e-01 -3.77776027e-01 1.11562788e-01 8.14267434e-03
-3.20159137e-01 -5.58402613e-02 1.87560692e-01 -1.08478956e-01
-5.19774675e-01 2.19060615e-01 -8.45090270e-01 5.05331576e-01
8.81956458e-01 -1.00901604e+00 -4.72466081e-01 -7.21996486e-01
7.30220199e-01 -1.63539812e-01 1.77308515e-01 -1.00565505e+00
8.56944397e-02 -1.44131690e-01 -1.83746517e-01 -1.11346832e-02
-5.03299594e-01 -6.80539250e-01 2.74336994e-01 5.09516358e-01
-5.93598783e-01 4.55021918e-01 6.99565709e-01 2.26389632e-01
2.94814676e-01 -1.01197457e+00 4.98689622e-01 -2.11442888e-01
-9.29388344e-01 -2.12554187e-01 -6.75312877e-01 1.76629275e-01
1.05462623e+00 -3.47382814e-01 -5.54199815e-01 4.92490351e-01
-2.00364068e-01 -8.32295641e-02 9.82679129e-01 8.65034580e-01
4.87371594e-01 -9.79819715e-01 -3.81803632e-01 5.27396917e-01
6.10895574e-01 -3.14122587e-01 -8.13249126e-02 3.65645349e-01
-7.32862115e-01 -2.36133456e-01 -3.46928269e-01 -4.77364510e-01
-1.48523295e+00 7.53195047e-01 4.64742750e-01 2.23609954e-01
-6.86349690e-01 8.47691357e-01 -2.79731661e-01 -7.32384503e-01
8.71011764e-02 -7.09323466e-01 1.31411821e-01 -5.06954074e-01
7.48945892e-01 3.11755359e-01 2.99030095e-01 2.55654573e-01
-3.82073164e-01 4.61563796e-01 -4.04030323e-01 3.65208238e-01
1.27263951e+00 1.60240605e-01 -3.80576164e-01 5.62856436e-01
1.18100727e+00 3.62443835e-01 -1.26183534e+00 -1.48630038e-01
5.48668742e-01 -9.42942262e-01 -5.23441732e-01 -1.06799376e+00
-7.64963746e-01 6.09839022e-01 5.80694258e-01 1.72978014e-01
8.76600623e-01 -4.50239666e-02 4.12246078e-01 7.65862107e-01
4.59778607e-01 -7.43119895e-01 1.80372685e-01 2.85679638e-01
1.01542175e+00 -1.01917350e+00 -4.01018150e-02 -1.75668895e-01
-2.43647188e-01 1.40921044e+00 1.16224205e+00 9.28814560e-02
3.06086630e-01 6.91953957e-01 2.62211934e-02 2.01327652e-02
-8.20778310e-01 7.04202354e-01 -1.26427561e-01 1.03883719e+00
8.25257301e-01 -2.67496824e-01 -2.69095581e-02 6.27412617e-01
-6.31958395e-02 3.21351022e-01 1.18793857e+00 1.33059919e+00
-3.63982111e-01 -1.44572246e+00 -3.78987134e-01 6.82826936e-01
-2.08410025e-01 -3.41819644e-01 -2.90675581e-01 7.81363964e-01
8.17542672e-02 6.22855246e-01 -1.01073913e-01 -4.02734786e-01
6.98856950e-01 1.45002022e-01 3.63311708e-01 -1.22896588e+00
-1.41735113e+00 -7.69810081e-01 -1.75200328e-02 -7.58474410e-01
1.89422637e-01 -7.36691058e-01 -1.18889368e+00 -3.23192656e-01
-2.29144245e-02 4.38964963e-02 6.02373779e-01 5.42801976e-01
7.71870971e-01 6.83545828e-01 3.13486934e-01 -5.72330296e-01
-8.24885309e-01 -8.44862282e-01 1.19226456e-01 3.76092941e-01
4.71367747e-01 -3.08758527e-01 -2.23875135e-01 5.42620003e-01] | [7.663318157196045, 7.763908863067627] |
a6ce2860-937c-4dca-90c7-210f4d624bab | meta-learning-to-detect-rare-objects | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Wang_Meta-Learning_to_Detect_Rare_Objects_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Wang_Meta-Learning_to_Detect_Rare_Objects_ICCV_2019_paper.pdf | Meta-Learning to Detect Rare Objects | Few-shot learning, i.e., learning novel concepts from few examples, is fundamental to practical visual recognition systems. While most of existing work has focused on few-shot classification, we make a step towards few-shot object detection, a more challenging yet under-explored task. We develop a conceptually simple but powerful meta-learning based framework that simultaneously tackles few-shot classification and few-shot localization in a unified, coherent way. This framework leverages meta-level knowledge about "model parameter generation" from base classes with abundant data to facilitate the generation of a detector for novel classes. Our key insight is to disentangle the learning of category-agnostic and category-specific components in a CNN based detection model. In particular, we introduce a weight prediction meta-model that enables predicting the parameters of category-specific components from few examples. We systematically benchmark the performance of modern detectors in the small-sample size regime. Experiments in a variety of realistic scenarios, including within-domain, cross-domain, and long-tailed settings, demonstrate the effectiveness and generality of our approach under different notions of novel classes.
| [' Martial Hebert', ' Deva Ramanan', 'Yu-Xiong Wang'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['novel-concepts'] | ['reasoning'] | [ 3.65584940e-01 -3.51117343e-01 -2.98355639e-01 -3.60317141e-01
-1.01345432e+00 -3.37674618e-01 8.99617851e-01 2.87561957e-02
-2.66241223e-01 2.90378928e-01 4.81171981e-02 7.18753338e-02
-1.61156468e-02 -5.09971440e-01 -5.69284439e-01 -7.82337725e-01
1.68875679e-01 9.46918130e-02 4.72746074e-01 -1.07226759e-01
2.86413252e-01 2.05407172e-01 -1.80450118e+00 1.51754171e-01
4.85423505e-01 1.00449443e+00 2.42778495e-01 9.46741521e-01
1.15444034e-01 6.52300239e-01 -3.29100847e-01 -4.02939558e-01
3.12432706e-01 -4.00513917e-01 -3.11137915e-01 3.56974363e-01
5.13199151e-01 -2.35185072e-01 -2.97740400e-01 1.16864812e+00
5.29184520e-01 3.37589294e-01 9.80686605e-01 -1.24867320e+00
-5.77215493e-01 3.76894951e-01 -6.55675650e-01 6.90078735e-01
-2.37696618e-01 6.33347154e-01 1.10579193e+00 -1.21024239e+00
5.69143414e-01 1.06337667e+00 6.96832001e-01 7.93717384e-01
-1.18823445e+00 -6.66574657e-01 3.48384172e-01 4.18386310e-01
-1.34457290e+00 -5.62312305e-01 7.07508981e-01 -7.96212912e-01
7.44523168e-01 -2.68445194e-01 4.37844783e-01 1.18433821e+00
2.11708378e-02 8.76059651e-01 8.51839900e-01 -5.13257146e-01
5.97846329e-01 1.72179997e-01 5.55407465e-01 6.00001037e-01
4.14960712e-01 3.42817456e-01 -5.47358692e-01 -1.74632326e-01
3.16729814e-01 4.64644134e-01 1.35674194e-01 -8.46082628e-01
-7.91788399e-01 1.06766641e+00 2.02385038e-01 2.16241062e-01
-8.83660689e-02 1.72403052e-01 4.71423507e-01 2.34279767e-01
6.02480173e-01 2.89642543e-01 -3.52499872e-01 -2.87317298e-02
-7.78634369e-01 1.53940901e-01 7.40191281e-01 1.31236017e+00
9.12003994e-01 1.96921334e-01 -4.65425372e-01 9.27222371e-01
-7.43276030e-02 2.93362319e-01 5.83295405e-01 -4.69433725e-01
6.12218417e-02 2.67218590e-01 8.29001144e-02 -5.09064257e-01
-1.77816719e-01 -6.90435410e-01 -5.73301554e-01 6.27826303e-02
1.75842702e-01 -2.63983637e-01 -1.24764395e+00 1.82933187e+00
4.03053939e-01 6.82211220e-01 -1.64362222e-01 7.84626305e-01
7.53243685e-01 3.80319268e-01 1.89008906e-01 -1.47241443e-01
1.26245165e+00 -1.10948610e+00 -8.65830407e-02 -4.63757664e-01
4.49164808e-01 -3.45104784e-01 9.75874007e-01 2.91633606e-02
-7.00872660e-01 -6.08422458e-01 -1.10333848e+00 1.58140019e-01
-4.96416241e-01 -2.47889385e-01 6.17702246e-01 5.92116654e-01
-4.89356995e-01 5.28762400e-01 -6.68192148e-01 -6.59924567e-01
9.55990732e-01 -3.47521231e-02 -7.22101852e-02 -4.21494991e-01
-7.90160179e-01 7.09032118e-01 4.49780494e-01 -4.18975234e-01
-1.38523722e+00 -1.04402542e+00 -8.77992451e-01 2.88794369e-01
8.26497495e-01 -9.00881588e-01 1.65641809e+00 -6.68388307e-01
-1.02803206e+00 7.52866268e-01 -2.67268112e-03 -6.86084926e-01
3.35433125e-01 -9.54274386e-02 -3.50293934e-01 -8.19515437e-03
8.91094431e-02 4.32969034e-01 1.36419547e+00 -9.97956455e-01
-1.02164352e+00 -2.92960584e-01 -3.44647691e-02 -1.42747443e-02
-4.36542630e-01 9.67049785e-03 -2.65523583e-01 -6.22504711e-01
-2.76968300e-01 -5.93314111e-01 -2.53207982e-01 2.25762993e-01
-2.40934640e-01 -2.57406175e-01 6.47616684e-01 1.34106800e-01
8.08397830e-01 -2.31046176e+00 -1.06883578e-01 -1.91867143e-01
5.95033050e-01 4.65247333e-01 -2.08833754e-01 3.85125726e-01
2.07643844e-02 -2.65654683e-01 -1.11692294e-01 -3.58671993e-01
7.95767233e-02 -1.26496330e-02 -3.96398693e-01 4.84904826e-01
4.75647181e-01 1.12240994e+00 -1.07163799e+00 -2.71211118e-01
3.54456484e-01 2.34308839e-01 -3.94950658e-01 1.72971457e-01
-2.24524036e-01 -1.20597385e-01 -3.42097819e-01 9.31839943e-01
4.00573820e-01 -4.95008200e-01 -2.34315097e-01 7.58971348e-02
5.24065532e-02 -2.98520893e-01 -1.15288174e+00 1.55028868e+00
-3.11288059e-01 6.68074250e-01 -2.71445245e-01 -1.28812921e+00
7.31279433e-01 -8.32018331e-02 1.77455261e-01 -4.27297235e-01
3.68779331e-01 -4.28709760e-02 5.96574247e-02 -3.89179230e-01
2.14589074e-01 -6.84442699e-01 -6.83082268e-02 5.04119456e-01
6.29764855e-01 1.38104543e-01 1.71915218e-01 2.80664265e-01
1.23744893e+00 -2.91776031e-01 9.66887772e-01 -8.48424807e-03
-9.49707255e-02 -9.99816880e-02 5.85080624e-01 1.42023289e+00
-6.69616044e-01 6.34248137e-01 1.81291685e-01 -3.83361012e-01
-1.12836385e+00 -1.15571105e+00 -1.11031212e-01 1.53229320e+00
4.27586101e-02 -2.15492457e-01 -3.83738637e-01 -5.74087739e-01
3.16197485e-01 5.20710111e-01 -8.90232623e-01 -5.68227708e-01
-8.25722739e-02 -9.16005135e-01 6.55802190e-02 6.63442075e-01
3.36183887e-03 -8.24630499e-01 -8.68375301e-01 2.35015094e-01
5.28981805e-01 -1.15019798e+00 -3.00669849e-01 4.33672309e-01
-7.61602223e-01 -1.27616012e+00 -9.72970068e-01 -7.45700896e-01
2.95443356e-01 8.84492695e-01 9.44888711e-01 -2.10550666e-01
-9.83985186e-01 6.27539277e-01 -3.36989760e-01 -7.75016844e-01
-3.27095389e-02 -6.72599897e-02 1.79404080e-01 2.15024278e-01
7.46717930e-01 -4.66389865e-01 -6.00274503e-01 6.03681058e-02
-5.98741472e-01 -1.60724059e-01 7.63210833e-01 1.14246202e+00
3.47111583e-01 -2.37224594e-01 8.18097711e-01 -1.04179454e+00
3.34828883e-01 -8.52495193e-01 -4.93819773e-01 2.93058723e-01
-5.48574626e-01 -1.21120185e-01 5.95366120e-01 -9.29400563e-01
-9.05875683e-01 -2.71688192e-03 2.91292816e-01 -7.50344992e-01
-3.12115967e-01 2.16696337e-01 -4.54760566e-02 -2.00469509e-01
8.45466733e-01 3.70769590e-01 -2.57010877e-01 -4.86280262e-01
7.48633206e-01 4.60269153e-01 6.06567383e-01 -3.38721395e-01
1.08684993e+00 7.54625797e-01 -1.65270582e-01 -1.01481020e+00
-1.13761818e+00 -1.10273659e+00 -7.24945068e-01 -1.32235333e-01
3.80601048e-01 -1.11041260e+00 -3.09657753e-01 4.74654734e-01
-8.11266243e-01 -2.07662687e-01 -6.86663270e-01 3.58737350e-01
-7.04379022e-01 1.73612326e-01 -3.17021847e-01 -9.55896795e-01
-2.32652411e-01 -6.69988155e-01 9.54562485e-01 5.07934809e-01
1.50536940e-01 -8.57330918e-01 2.44180292e-01 -6.45273179e-02
2.89825946e-01 3.76420431e-02 8.64143252e-01 -1.05999589e+00
-5.93246877e-01 -5.22475839e-01 -4.40448701e-01 2.15623707e-01
-2.92126257e-02 -1.79304376e-01 -1.32438648e+00 -3.30445528e-01
-4.22015414e-02 -5.89233518e-01 1.51873195e+00 4.65152562e-01
9.19535935e-01 8.57062638e-02 -4.45444971e-01 6.25314474e-01
1.56177664e+00 -2.18770816e-04 1.97481468e-01 -9.41905677e-02
5.80706000e-01 2.88471788e-01 6.68575764e-01 7.73398936e-01
6.28785938e-02 5.84771633e-01 1.97421417e-01 2.50565946e-01
-2.82078564e-01 -1.46412730e-01 -6.25397936e-02 4.46383059e-01
2.28515327e-01 2.35254504e-02 -8.29989910e-01 7.78271675e-01
-1.92455685e+00 -1.25705802e+00 5.61273634e-01 1.97830057e+00
5.94401896e-01 3.94300044e-01 4.42960054e-01 -1.20397463e-01
9.15429592e-01 2.07830086e-01 -9.98696506e-01 9.39849988e-02
1.12512134e-01 2.21706674e-01 2.47351363e-01 -2.30675824e-02
-1.37344623e+00 9.53041434e-01 6.19073391e+00 9.32017207e-01
-8.79610121e-01 4.28555906e-01 3.78061920e-01 -4.16031241e-01
2.91641235e-01 3.43350954e-02 -1.18883860e+00 3.07984501e-01
8.46848309e-01 -6.34006619e-01 1.32491924e-02 1.40245950e+00
-1.75569087e-01 3.07836048e-02 -1.29595637e+00 1.14199233e+00
4.47253972e-01 -1.61219704e+00 -6.42854348e-02 -1.86056659e-01
1.01560521e+00 3.21038634e-01 3.03638816e-01 6.82061315e-01
3.52909684e-01 -5.29947698e-01 6.58003449e-01 6.17146373e-01
7.72031307e-01 -6.92523599e-01 4.12236780e-01 5.41869938e-01
-1.23904097e+00 -6.46733940e-01 -8.38629723e-01 -2.67081231e-01
-1.28656849e-01 5.48065186e-01 -8.82674158e-01 1.28723353e-01
4.31844056e-01 9.74158823e-01 -7.52863407e-01 1.59700024e+00
1.82709441e-01 5.70404053e-01 -5.36610093e-03 -2.08055705e-01
2.99187094e-01 3.98524135e-01 5.54031730e-01 1.27205086e+00
3.52593139e-02 1.99609891e-01 2.35762417e-01 8.80831242e-01
-2.39297345e-01 -1.73954099e-01 -7.92214274e-01 -1.47029415e-01
5.14427602e-01 1.38516259e+00 -7.88330853e-01 -6.54423058e-01
-7.39516497e-01 4.98099893e-01 5.35353899e-01 2.70192266e-01
-6.58808589e-01 -4.53941286e-01 8.32073629e-01 -1.29807666e-01
8.65512788e-01 8.84313229e-03 -2.33185545e-01 -1.39267290e+00
-2.69040078e-01 -6.06672466e-01 6.46885991e-01 -4.22261387e-01
-1.67076981e+00 9.58127230e-02 7.64260218e-02 -1.42838395e+00
-3.43897730e-01 -6.84880257e-01 -9.96279538e-01 5.37025988e-01
-1.68486595e+00 -1.18207240e+00 -3.35862070e-01 4.31427807e-01
1.01385581e+00 -4.00466979e-01 6.42956674e-01 7.90259615e-02
-6.79315567e-01 7.60120451e-01 4.55075741e-01 1.21220678e-01
6.95563674e-01 -1.07440972e+00 6.32999718e-01 1.05612087e+00
4.32034940e-01 5.95943630e-01 7.68217683e-01 -4.54056561e-01
-1.34128726e+00 -1.28985393e+00 3.56619537e-01 -5.53859234e-01
9.64086890e-01 -5.86093664e-01 -9.42851901e-01 3.57065052e-01
-3.92717391e-01 5.57412982e-01 8.15910280e-01 2.95577407e-01
-8.86493862e-01 -3.79343331e-02 -9.15683925e-01 5.78036904e-01
1.10445201e+00 -7.45095551e-01 -7.05634058e-01 2.78367788e-01
7.21765757e-01 1.63202718e-01 -3.01324099e-01 2.08107352e-01
7.10733831e-01 -7.56908596e-01 1.09126067e+00 -1.27010393e+00
3.52704018e-01 8.34778473e-02 -4.12441224e-01 -1.28787792e+00
-5.40798664e-01 -4.92085576e-01 -7.67965078e-01 8.97585928e-01
7.29925260e-02 -3.06729347e-01 7.57013977e-01 4.09390569e-01
-2.05568492e-01 -6.25441432e-01 -7.97231376e-01 -1.21231771e+00
5.57189249e-02 -6.24725103e-01 2.34245598e-01 5.50288141e-01
7.98586104e-03 5.21045923e-01 -6.41123235e-01 -4.71385308e-02
1.07680607e+00 3.49144906e-01 9.08460975e-01 -1.47451460e+00
-5.19057810e-01 -4.86771464e-01 -7.69954145e-01 -7.78086722e-01
1.07546508e-01 -7.13177204e-01 3.41147959e-01 -1.10365069e+00
8.43654215e-01 7.44858570e-03 -6.65819943e-01 2.18943760e-01
-3.66858721e-01 4.29682821e-01 3.69915813e-01 1.85603976e-01
-1.03885651e+00 5.32667220e-01 7.35056221e-01 -3.31476867e-01
3.78241464e-02 1.75794214e-01 -1.00071871e+00 8.46832871e-01
4.31656748e-01 -5.46160698e-01 -3.83055747e-01 1.00503348e-01
-2.81662822e-01 -2.95619309e-01 7.36498356e-01 -1.19902480e+00
4.77747649e-01 -3.26533228e-01 3.62789989e-01 -4.81123537e-01
4.10924494e-01 -4.27829355e-01 -6.52971506e-01 4.79107410e-01
-3.41043293e-01 -7.79226422e-01 -8.28596726e-02 1.24520504e+00
2.26249754e-01 -3.58798742e-01 1.28481328e+00 -2.25710034e-01
-1.47719777e+00 7.51038671e-01 -7.66663924e-02 4.68005329e-01
1.26611567e+00 -2.65071511e-01 -4.65037405e-01 -1.84568226e-01
-7.67181277e-01 1.36024775e-02 2.12213531e-01 5.62800109e-01
7.40936160e-01 -1.15147138e+00 -5.58642089e-01 2.06283242e-01
1.01713669e+00 -4.55296993e-01 4.94813263e-01 7.16723084e-01
3.91119450e-01 3.22550774e-01 -1.42782792e-01 -7.01253295e-01
-1.18947899e+00 9.58711982e-01 2.58559048e-01 2.97887444e-01
-8.98516715e-01 1.05590332e+00 6.07902646e-01 8.64148587e-02
4.17537928e-01 -3.55921276e-02 4.63621281e-02 1.99835002e-01
1.02891016e+00 3.88953626e-01 -1.81594580e-01 -3.30339074e-01
-2.23456368e-01 3.16832572e-01 -5.21213770e-01 2.08269075e-01
1.38997173e+00 -1.28995761e-01 8.18349957e-01 9.64580595e-01
1.10345376e+00 -7.48368621e-01 -1.73494673e+00 -8.00571918e-01
1.19373195e-01 -5.71758747e-01 -9.07266736e-02 -6.11834586e-01
-4.79055017e-01 1.21491194e+00 6.89221323e-01 -9.93099287e-02
7.50688255e-01 4.15165454e-01 5.89084923e-01 6.61721110e-01
4.32973325e-01 -1.24211442e+00 4.89307165e-01 5.55859268e-01
2.66003430e-01 -1.67501748e+00 -1.65394261e-01 -7.80607574e-03
-4.75563318e-01 9.35527682e-01 6.81525469e-01 -2.60286212e-01
8.88860941e-01 1.73757225e-02 -4.62770998e-01 -1.86696485e-01
-1.19115758e+00 -7.17928708e-01 4.59169060e-01 8.81054103e-01
-1.77763656e-01 1.91127080e-02 3.35942030e-01 9.11448419e-01
3.85766268e-01 1.58145174e-01 4.92413521e-01 1.10002840e+00
-1.15076435e+00 -3.86071533e-01 4.39814739e-02 8.68594170e-01
-4.11159582e-02 -2.15925336e-01 -2.40032807e-01 6.95556700e-01
7.22449049e-02 7.50645936e-01 9.75979194e-02 -3.44805181e-01
1.72731519e-01 2.86524028e-01 4.90970522e-01 -1.31330967e+00
-3.66943516e-02 -4.61511575e-02 -3.90676677e-01 -3.78189027e-01
-2.49045536e-01 -6.79523766e-01 -5.80202341e-01 -1.29930064e-01
-4.57792521e-01 -2.27289870e-01 3.84056091e-01 1.30072415e+00
3.29086423e-01 3.45638484e-01 7.83149004e-01 -8.37014437e-01
-1.10136175e+00 -1.00387621e+00 -8.61348271e-01 2.87415266e-01
5.32816947e-01 -1.10143781e+00 -5.63941360e-01 -3.35490964e-02] | [9.961604118347168, 2.7070326805114746] |
8f6fd620-84f9-4930-8529-75d335217176 | twitmo-a-twitter-data-topic-modeling-and | 2207.11236 | null | https://arxiv.org/abs/2207.11236v1 | https://arxiv.org/pdf/2207.11236v1.pdf | Twitmo: A Twitter Data Topic Modeling and Visualization Package for R | We present Twitmo, a package that provides a broad range of methods to collect, pre-process, analyze and visualize geo-tagged Twitter data. Twitmo enables the user to collect geo-tagged Tweets from Twitter and and provides a comprehensive and user-friendly toolbox to generate topic distributions from Latent Dirichlet Allocations (LDA), correlated topic models (CTM) and structural topic models (STM). Functions are included for pre-processing of text, model building and prediction. In addition, one of the innovations of the package is the automatic pooling of Tweets into longer pseudo-documents using hashtags and cosine similarities for better topic coherence. The package additionally comes with functionality to visualize collected data sets and fitted models in static as well as interactive ways and offers built-in support for model visualizations via LDAvis providing great convenience for researchers in this area. The Twitmo package is an innovative toolbox that can be used to analyze public discourse of various topics, political parties or persons of interest in space and time. | ['Thomas Kneib', 'Krisztina Kis-Katos', 'Benjamin Säfken', 'Christoph Weisser', 'Gillian Kant', 'Andreas Buchmüller'] | 2022-07-08 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [-5.21651089e-01 1.19024366e-01 -3.36325355e-02 -5.89392960e-01
-8.61649275e-01 -7.57018447e-01 1.00158429e+00 5.96959114e-01
-2.00157445e-02 3.74520451e-01 6.89808547e-01 -4.71242517e-01
-1.08087473e-01 -1.08637214e+00 5.92548735e-02 -7.84690380e-01
-2.15479687e-01 8.14108074e-01 2.69997001e-01 -2.40211934e-01
3.34258586e-01 -1.53185548e-02 -1.82473016e+00 5.04878044e-01
8.00531089e-01 4.18926835e-01 3.04784060e-01 5.02962112e-01
-5.87120235e-01 1.89331368e-01 -3.72523367e-01 -2.33681649e-01
-2.63056040e-01 -3.75762992e-02 -5.81114113e-01 -2.36595258e-01
1.84284262e-02 7.43552148e-02 2.08492413e-01 5.31310737e-01
5.50956607e-01 1.47041930e-02 6.43250823e-01 -1.33824003e+00
-4.32090722e-02 8.01559210e-01 -4.44795668e-01 2.34977394e-01
5.39623916e-01 -7.47038331e-03 5.92129052e-01 -1.00793290e+00
8.05109739e-01 1.63508344e+00 6.78489506e-01 -2.89718300e-01
-1.36518073e+00 -7.58645475e-01 -2.34708130e-01 -1.01624951e-01
-1.32982647e+00 -1.85752198e-01 5.42471707e-01 -9.48802114e-01
7.35111415e-01 7.17023313e-01 8.72077405e-01 9.44202602e-01
1.08808920e-01 5.34388423e-01 1.48004794e+00 -2.27337047e-01
2.14583799e-01 5.53362191e-01 6.17610157e-01 -4.99011129e-02
-5.07330485e-02 -6.27681136e-01 -5.72343290e-01 -1.00955021e+00
2.55835235e-01 1.40704408e-01 3.51987451e-01 -1.47685453e-01
-1.41517472e+00 1.22800112e+00 7.66734779e-02 6.02547109e-01
-5.20810843e-01 -2.96667933e-01 4.33167547e-01 -3.72765847e-02
1.31503665e+00 8.45349431e-02 -1.18825547e-01 -5.24802506e-01
-1.24032760e+00 5.10251343e-01 7.68673241e-01 7.09287643e-01
9.15064692e-01 -2.84352809e-01 -2.36044213e-01 7.97250926e-01
6.89349830e-01 5.03424942e-01 5.33436954e-01 -5.29801786e-01
2.72564948e-01 7.04054832e-01 3.05776834e-01 -1.35617959e+00
-6.57817304e-01 1.00841321e-01 -3.50072443e-01 -2.09876612e-01
3.85967225e-01 -3.37369651e-01 -5.64478934e-01 1.13636231e+00
8.37782323e-01 -5.72917685e-02 -3.15482408e-01 4.56823289e-01
1.04251742e+00 1.09030735e+00 5.54114282e-01 -2.98975438e-01
1.85923111e+00 -2.50454694e-01 -1.10223567e+00 2.03209728e-01
8.91332090e-01 -1.02496457e+00 1.16686189e+00 1.49486542e-01
-1.10294425e+00 -1.42699316e-01 -3.04763794e-01 -3.31629813e-01
-1.04231668e+00 -1.53134316e-01 5.54786563e-01 8.44603181e-01
-1.11779392e+00 2.33446538e-01 -1.10883760e+00 -8.07680905e-01
1.69640809e-01 6.23743273e-02 -7.19581991e-02 3.11974555e-01
-1.31459463e+00 7.07104325e-01 3.36203873e-01 -4.78272319e-01
-3.54313284e-01 -9.35090423e-01 -8.69261742e-01 -7.92754367e-02
-2.49866143e-01 -4.28600580e-01 1.01025546e+00 -3.16650093e-01
-1.28966844e+00 1.14254940e+00 -5.31479895e-01 -3.73440802e-01
4.76359040e-01 2.45427489e-02 -2.56003916e-01 -9.26822498e-02
4.23148513e-01 4.63536888e-01 3.65424931e-01 -8.68326187e-01
-6.47603452e-01 -3.31080377e-01 -5.03318369e-01 1.55761689e-01
-4.45168048e-01 6.96064949e-01 -3.15632194e-01 -6.06029689e-01
1.60442322e-01 -8.90265703e-01 -1.29812673e-01 -5.41939855e-01
-4.90037054e-01 -5.91037631e-01 1.26150084e+00 -1.05281901e+00
1.69418252e+00 -2.03978395e+00 -2.80048698e-01 3.51058811e-01
9.22214836e-02 -1.89559162e-01 6.07480347e-01 1.13700438e+00
-9.98636335e-02 4.38524425e-01 6.76309317e-03 -7.45378137e-01
2.78951943e-01 -1.01652466e-01 -3.10935289e-01 5.30657470e-01
-4.70106512e-01 5.18451869e-01 -8.52557659e-01 -6.32452309e-01
4.45991695e-01 6.26807332e-01 -3.27098459e-01 -2.33720139e-01
-1.26068532e-01 4.96150017e-01 -2.91591436e-01 9.64716375e-02
9.05492902e-01 -2.16249660e-01 4.59005892e-01 2.42370665e-01
-9.78234529e-01 7.43537664e-01 -1.09163380e+00 1.34238029e+00
-3.27969819e-01 1.11987948e+00 9.11174342e-02 -4.41921681e-01
1.16977477e+00 4.53169495e-01 6.28705144e-01 -2.33741090e-01
1.67640895e-01 -1.25322357e-01 -6.50799692e-01 -4.33820128e-01
9.97432232e-01 8.83234516e-02 -1.67073801e-01 1.17194188e+00
-2.32168809e-01 -7.39745647e-02 3.46816123e-01 3.02654535e-01
3.45635980e-01 -2.22205892e-01 2.41930231e-01 -7.51136661e-01
1.98528599e-02 3.31693649e-01 -2.92135272e-02 3.40703338e-01
5.24454772e-01 3.26964229e-01 6.84018791e-01 -4.70377028e-01
-1.30696023e+00 -8.17475855e-01 -5.40764093e-01 1.62327921e+00
-4.20269966e-01 -8.49840224e-01 -6.53034091e-01 4.29197960e-02
-1.70279235e-01 1.17650938e+00 -7.11109340e-01 6.13520145e-01
-2.25193873e-01 -1.20535982e+00 -7.21012652e-02 -1.25379860e-01
1.58526838e-01 -7.36947119e-01 -3.34374368e-01 7.84095526e-02
-5.88186979e-01 -3.60863626e-01 5.67416884e-02 -4.28019404e-01
-7.87753463e-01 -7.46445298e-01 -7.59493053e-01 -6.32921755e-01
5.61122239e-01 4.62059945e-01 8.59717906e-01 -4.41255748e-01
-4.96558882e-02 3.00058007e-01 -2.50360042e-01 -5.77938974e-01
-3.44314605e-01 3.36707711e-01 -1.77820042e-01 4.71727811e-02
6.75731122e-01 -5.50680757e-01 -5.04348397e-01 6.26958668e-01
-9.10837531e-01 4.50551838e-01 -5.69567800e-01 2.84780264e-01
1.61400095e-01 9.67142917e-03 3.61593306e-01 -9.02071774e-01
1.08348310e+00 -1.28604138e+00 -6.07898593e-01 -1.14272609e-01
-3.31305236e-01 -5.05957663e-01 -2.29623109e-01 -1.35525301e-01
-1.15555179e+00 -2.67027438e-01 -1.95527598e-02 3.50598097e-01
-2.17376038e-01 8.06475639e-01 8.56255218e-02 7.18841255e-01
6.34315252e-01 1.24625368e-02 1.78063765e-01 -7.52881765e-01
4.17695612e-01 1.02901828e+00 -7.81637728e-02 -1.21311106e-01
3.77642214e-01 9.02055979e-01 -5.57339430e-01 -1.07678843e+00
-4.90020543e-01 -1.01232839e+00 -7.36618638e-01 -3.80632192e-01
8.77540469e-01 -9.55739617e-01 -6.67542100e-01 3.62217516e-01
-9.14653003e-01 -3.15750927e-01 7.88131580e-02 4.79157597e-01
-3.52921426e-01 9.39695630e-03 -2.36612424e-01 -9.06288803e-01
-3.95325840e-01 -7.68598616e-01 8.92753959e-01 1.23531453e-01
-6.82735741e-01 -1.49778831e+00 3.23016107e-01 2.19516531e-02
7.02999830e-01 5.57874024e-01 6.18926585e-01 -1.02709246e+00
-1.35252401e-01 -2.52378285e-01 -6.19364716e-02 -4.73439008e-01
1.84189215e-01 4.61363643e-01 -1.04866540e+00 -1.42607376e-01
-3.12199980e-01 2.25851625e-01 5.60970306e-01 7.23413169e-01
6.10419750e-01 -6.18391454e-01 -9.53814149e-01 2.51478016e-01
8.56997430e-01 -4.48885746e-02 8.00858259e-01 7.06123471e-01
1.60424113e-01 1.06535947e+00 6.74839377e-01 8.15729260e-01
8.80933404e-01 8.11265886e-01 5.93468659e-02 -2.43919909e-01
6.09242678e-01 -1.18198715e-01 1.37785405e-01 7.33703613e-01
5.40315099e-02 -3.50801949e-03 -1.46490908e+00 6.85569286e-01
-1.87960589e+00 -1.16716480e+00 -9.87806559e-01 2.21714950e+00
7.56944895e-01 -1.97742179e-01 6.85252845e-01 -1.73064917e-02
9.38578308e-01 9.31515321e-02 2.63634413e-01 -4.82716918e-01
-6.81130169e-03 -3.10443759e-01 2.47942001e-01 7.23363638e-01
-1.26102757e+00 8.93867970e-01 6.73808765e+00 8.14248025e-01
-1.18667281e+00 5.58536232e-01 6.97351575e-01 -2.52190828e-02
-4.30346787e-01 1.54627070e-01 -9.68543470e-01 6.94335938e-01
1.30270696e+00 -8.93757999e-01 -3.28572065e-01 1.11557317e+00
1.09325564e+00 -4.45771962e-01 -3.70880157e-01 5.32981515e-01
-3.66471291e-01 -1.63824379e+00 -3.66403729e-01 4.31445658e-01
7.09554851e-01 2.30240822e-01 1.84677273e-01 8.74805078e-02
6.23592734e-01 -4.97501165e-01 6.73704803e-01 5.93741179e-01
4.71361846e-01 -6.57776892e-01 6.39246166e-01 3.03312629e-01
-9.62772489e-01 1.93297639e-01 -3.45529377e-01 -1.70447424e-01
6.38071716e-01 8.74444962e-01 -1.47488105e+00 2.61163414e-01
9.69408393e-01 7.65373766e-01 -5.23720860e-01 1.26688516e+00
3.32205981e-01 9.77688432e-01 -7.43259728e-01 -6.26820400e-02
2.64874101e-01 -2.41163313e-01 6.79934323e-01 1.64772367e+00
5.84857821e-01 -2.35154495e-01 3.38088237e-02 5.48508406e-01
8.37751746e-01 3.97492021e-01 -5.48643112e-01 1.62966415e-01
8.49303544e-01 1.48489749e+00 -1.29480767e+00 -6.06716335e-01
8.72159004e-02 1.38713583e-01 -3.06363881e-01 2.23705500e-01
-6.47985935e-01 -1.56638697e-01 3.91571999e-01 8.85088503e-01
-6.99317232e-02 -6.67026162e-01 -3.58082354e-01 -7.35016525e-01
-6.25032723e-01 -4.46326137e-01 6.52318537e-01 -8.32922280e-01
-9.68357265e-01 4.69468415e-01 7.92469978e-01 -1.06206882e+00
-4.37625527e-01 -7.18111638e-03 -9.90900159e-01 9.28417683e-01
-8.17700207e-01 -1.38773644e+00 -4.72878337e-01 5.10871947e-01
2.75388956e-01 1.87712252e-01 9.38228250e-01 2.34428242e-01
-4.08391088e-01 -2.64965773e-01 6.21056378e-01 -2.90066034e-01
6.91160500e-01 -1.24551868e+00 6.29810870e-01 8.32875744e-02
-4.65992182e-01 1.00168121e+00 1.12236738e+00 -7.89420426e-01
-6.93118632e-01 -9.26333308e-01 1.35734487e+00 -5.14268756e-01
1.04524910e+00 -8.12086761e-01 -7.33049452e-01 7.94706643e-01
5.11415005e-01 -8.30469608e-01 1.41956663e+00 4.88462508e-01
9.40212607e-02 2.81090081e-01 -8.65833104e-01 4.70139533e-01
-1.24271112e-02 -2.04727694e-01 -6.05862796e-01 7.54706621e-01
4.98098761e-01 -1.86574221e-01 -9.68489945e-01 -3.85768682e-01
5.50602913e-01 -1.04813421e+00 8.35170746e-01 -2.03671038e-01
7.15885311e-02 -1.05580322e-01 2.11483389e-01 -1.17405832e+00
-2.71024674e-01 -1.10200644e+00 5.27283907e-01 1.69046128e+00
5.32296121e-01 -7.62022018e-01 4.51646894e-01 8.51926804e-01
-7.64899328e-02 -6.61597913e-03 -8.69733930e-01 -1.25379533e-01
2.50898935e-02 -7.61734366e-01 7.98706770e-01 1.23084009e+00
6.09147310e-01 -2.65163016e-02 -2.72621512e-01 -1.31329075e-01
3.67325932e-01 -4.47728708e-02 1.08470881e+00 -1.62781322e+00
6.56340063e-01 -3.54498059e-01 5.40715046e-02 -5.04907191e-01
-4.10882711e-01 -5.90560794e-01 -4.83201772e-01 -1.84496462e+00
2.89021023e-02 -8.36897671e-01 5.14686048e-01 3.97431105e-01
2.17469916e-01 2.87327677e-01 3.45009230e-02 6.76524282e-01
-5.65642953e-01 1.53002113e-01 6.55165672e-01 3.91008914e-01
-6.83117092e-01 2.66845316e-01 -4.38310772e-01 8.05174291e-01
9.48398709e-01 -7.55252540e-01 3.13896313e-03 1.85448527e-02
6.53346360e-01 -4.19582933e-01 3.35968614e-01 -5.23617208e-01
1.67226523e-01 -1.98653638e-02 -2.59029195e-02 -1.40934205e+00
4.23644602e-01 -4.86658335e-01 6.77513242e-01 2.00443745e-01
-1.13727644e-01 2.35611409e-01 3.98762286e-01 2.56006658e-01
-7.79787824e-02 5.67686558e-02 3.15282404e-01 -9.77208316e-02
-1.43939435e-01 -1.70807034e-01 -1.09605420e+00 -4.20360684e-01
1.08205473e+00 -3.54239285e-01 -5.22531390e-01 -7.54654467e-01
-1.17057025e+00 4.86663550e-01 4.16722089e-01 4.75191355e-01
-1.56248391e-01 -1.35362136e+00 -6.66125894e-01 -4.55086939e-02
3.16559523e-02 -1.42624393e-01 6.80145562e-01 1.10527182e+00
-5.30297697e-01 8.71601224e-01 -1.43503444e-02 -7.37781882e-01
-1.41814303e+00 2.24053279e-01 -3.21956664e-01 -3.02752405e-01
-5.66972017e-01 3.62626016e-01 1.22420490e-01 -5.81509233e-01
-1.54623762e-01 -2.19162583e-01 -6.84224069e-01 1.04288280e+00
8.46273303e-01 7.17991769e-01 -1.34641945e-01 -9.07417476e-01
-1.84875816e-01 2.37044886e-01 1.52638942e-01 -3.92464578e-01
1.68689883e+00 -6.60310447e-01 -4.28551942e-01 9.82447505e-01
9.95150924e-01 1.10761538e-01 -8.90033245e-01 -5.07051758e-02
3.31109017e-01 -3.01126808e-01 1.97128579e-02 -3.62648189e-01
-2.46326312e-01 7.99544632e-01 3.77424330e-01 1.16034126e+00
5.29675484e-01 2.65641928e-01 2.22557142e-01 -3.44118834e-01
-4.36293297e-02 -1.08100939e+00 -3.14303666e-01 4.76549476e-01
1.02156663e+00 -8.27090919e-01 7.72006065e-02 -3.75829130e-01
-6.92848623e-01 1.08246768e+00 -2.20326349e-01 1.90403193e-01
1.29414916e+00 4.29831259e-02 6.74627125e-02 -6.23792946e-01
-8.20440471e-01 -5.87189496e-02 3.55505526e-01 6.45238757e-01
8.33377659e-01 9.10835117e-02 -6.52178705e-01 6.28262579e-01
-8.68130147e-01 -2.72983760e-01 4.29828674e-01 8.02577615e-01
-6.73025966e-01 -9.33253706e-01 -9.87039387e-01 2.94131488e-01
-5.80483675e-01 -3.22065540e-02 -1.20881349e-01 8.38364303e-01
-1.87084511e-01 1.02263272e+00 6.05946183e-01 -2.56728455e-02
-4.66918349e-01 2.11719364e-01 -6.69749081e-01 -6.49195552e-01
-7.20753789e-01 5.97045004e-01 4.90728408e-01 -1.67858958e-01
-4.35779810e-01 -1.10953939e+00 -9.37881112e-01 -6.89462543e-01
-4.38480586e-01 6.48278177e-01 1.42041814e+00 6.43125236e-01
6.59232855e-01 -1.55044757e-02 4.67645973e-01 -1.28393793e+00
5.68946362e-01 -1.60055470e+00 -6.45949602e-01 7.15192556e-02
-9.93798003e-02 -6.12838626e-01 -3.26098949e-01 2.73456305e-01] | [10.417296409606934, 7.0645365715026855] |
487541ca-7107-4c45-9863-fc3fe0cfbc02 | spqr-a-sparse-quantized-representation-for | 2306.03078 | null | https://arxiv.org/abs/2306.03078v1 | https://arxiv.org/pdf/2306.03078v1.pdf | SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compression | Recent advances in large language model (LLM) pretraining have led to high-quality LLMs with impressive abilities. By compressing such LLMs via quantization to 3-4 bits per parameter, they can fit into memory-limited devices such as laptops and mobile phones, enabling personalized use. However, quantization down to 3-4 bits per parameter usually leads to moderate-to-high accuracy losses, especially for smaller models in the 1-10B parameter range, which are well-suited for edge deployments. To address this accuracy issue, we introduce the Sparse-Quantized Representation (SpQR), a new compressed format and quantization technique which enables for the first time near-lossless compression of LLMs across model scales, while reaching similar compression levels to previous methods. SpQR works by identifying and isolating outlier weights, which cause particularly-large quantization errors, and storing them in higher precision, while compressing all other weights to 3-4 bits, and achieves relative accuracy losses of less than 1% in perplexity for highly-accurate LLaMA and Falcon LLMs. This makes it possible to run 33B parameter LLM on a single 24 GB consumer GPU without any performance degradation at 15% speedup thus making powerful LLMs available to consumer without any downsides. SpQR comes with efficient algorithms for both encoding weights into its format, as well as decoding them efficiently at runtime. Specifically, we provide an efficient GPU inference algorithm for SpQR which yields faster inference than 16-bit baselines at similar accuracy, while enabling memory compression gains of more than 4x. | ['Dan Alistarh', 'Torsten Hoefler', 'Alexander Borzunov', 'Saleh Ashkboos', 'Elias Frantar', 'Denis Kuznedelev', 'Vage Egiazarian', 'Ruslan Svirschevski', 'Tim Dettmers'] | 2023-06-05 | null | null | null | null | ['quantization'] | ['methodology'] | [ 9.37333480e-02 -7.94870108e-02 -6.65528774e-01 -3.16297829e-01
-1.51610279e+00 -1.88099876e-01 2.59858489e-01 4.55082864e-01
-7.39946842e-01 5.03671467e-01 2.22193241e-01 -7.37741590e-01
1.24931604e-01 -7.61937737e-01 -8.76752257e-01 -4.12182212e-01
-1.58238456e-01 7.87753105e-01 2.02428177e-02 1.28234267e-01
-1.02537185e-01 2.86827713e-01 -1.47869802e+00 4.11768287e-01
6.91415489e-01 1.14425039e+00 3.53559017e-01 9.38677788e-01
-2.15848479e-02 4.39006984e-01 -3.51811588e-01 -6.32851362e-01
2.15480059e-01 1.19101129e-01 -5.80534101e-01 -3.13220710e-01
7.21802473e-01 -9.48515832e-01 -4.24933761e-01 9.02832150e-01
5.10099411e-01 -9.45964009e-02 4.25086230e-01 -8.40989411e-01
-2.79410899e-01 8.85346591e-01 -7.57310748e-01 9.31647718e-02
2.22665668e-01 1.65288866e-01 1.33581424e+00 -6.27153277e-01
3.44134986e-01 1.39809024e+00 9.94165301e-01 2.09561378e-01
-1.32773483e+00 -8.01854014e-01 -2.65424550e-01 -8.25457368e-03
-1.66719735e+00 -8.51889193e-01 -1.13931298e-01 5.34270667e-02
1.55308592e+00 4.09141719e-01 4.92890537e-01 9.15372372e-01
2.52280414e-01 9.39046681e-01 5.89492261e-01 -2.83761203e-01
1.41308814e-01 -1.51848510e-01 1.22140959e-01 7.20784664e-01
4.24208701e-01 -3.18293959e-01 -5.19291341e-01 -4.51105744e-01
5.00295520e-01 -1.42424449e-01 1.66851413e-02 3.90006095e-01
-1.14969528e+00 8.92856061e-01 2.98119694e-01 6.17534071e-02
-2.62064874e-01 6.48396850e-01 5.88175058e-01 1.26630187e-01
4.74631101e-01 1.96479410e-02 -5.54430306e-01 -7.58015215e-01
-1.38972664e+00 3.74520779e-01 6.29668832e-01 1.05067146e+00
6.86158836e-01 -2.84268260e-02 -2.41096139e-01 1.00430393e+00
2.24748209e-01 9.19069648e-01 6.31433606e-01 -1.02295399e+00
9.70335305e-01 2.73340661e-02 -1.20896876e-01 -9.49657261e-01
-4.91478920e-01 -4.77484465e-01 -1.17746079e+00 -5.21051466e-01
2.78996527e-01 4.14388860e-03 -7.65998721e-01 1.79081440e+00
2.58133501e-01 2.07920417e-01 -2.47821584e-01 6.48172796e-01
2.78387666e-01 1.04080927e+00 -4.14420152e-03 -1.88373297e-01
1.66983604e+00 -7.32205868e-01 -4.26612824e-01 -4.10785019e-01
1.25873482e+00 -8.22662652e-01 1.51588440e+00 5.63135386e-01
-1.29705644e+00 -3.46767187e-01 -1.11079037e+00 -6.24805093e-01
1.61710620e-01 -3.86830457e-02 9.17438030e-01 7.41697073e-01
-1.03657210e+00 8.36782336e-01 -1.25054312e+00 -9.13578551e-03
5.18716574e-01 5.11553884e-01 -1.89133495e-01 -4.25662160e-01
-9.31923687e-01 5.93331456e-01 3.58472735e-01 -2.39592627e-01
-4.03408736e-01 -1.09481633e+00 -8.17532837e-01 2.87297934e-01
-7.51519948e-02 -6.91843331e-01 1.33613324e+00 -5.46236992e-01
-1.32307708e+00 6.85874879e-01 -5.65648794e-01 -9.25956070e-01
2.37557545e-01 -3.12164009e-01 -3.74516785e-01 -1.43199954e-02
-1.89649001e-01 7.41633594e-01 8.35071862e-01 -5.99631786e-01
-6.59293115e-01 -2.14527547e-01 -3.08405906e-01 6.38857037e-02
-7.18650877e-01 -6.11452088e-02 -9.30346429e-01 -7.40545332e-01
2.68218517e-01 -7.99988747e-01 -2.31690370e-02 -2.08862916e-01
-1.92313492e-01 -7.35281222e-03 4.21433866e-01 -8.53015840e-01
1.55340815e+00 -2.16316652e+00 -2.27673560e-01 2.07020804e-01
3.05477470e-01 3.52053046e-01 -2.77361035e-01 1.42424732e-01
4.31013137e-01 8.66924450e-02 -1.83703899e-01 -1.02992594e+00
2.76222020e-01 5.44557571e-01 -4.77122009e-01 4.16823059e-01
-2.15007916e-01 9.49722171e-01 -6.80727482e-01 -4.41978157e-01
-5.70973270e-02 5.15236855e-01 -1.04402030e+00 1.38400242e-01
-9.45068896e-02 -2.28609756e-01 -7.77047575e-02 6.79146349e-01
8.80979657e-01 -3.15068722e-01 1.76341891e-01 -2.62690812e-01
3.37741375e-01 7.14062810e-01 -1.07132471e+00 1.62481534e+00
-9.33562577e-01 7.28372931e-01 2.53346294e-01 -6.32509410e-01
5.09585381e-01 -2.67651882e-02 1.14052907e-01 -8.43624175e-01
3.80899571e-02 3.25746506e-01 -2.72970378e-01 -6.23237481e-03
9.73141670e-01 -9.84027684e-02 -1.52800024e-01 5.38333476e-01
-4.58984859e-02 -1.83706626e-01 1.96898133e-01 2.78641135e-01
1.01406968e+00 -5.17035544e-01 4.88117263e-02 3.36503014e-02
-8.09361041e-02 -3.86100113e-01 5.31601965e-01 9.39633191e-01
1.80204332e-01 4.74462181e-01 2.69543469e-01 -3.21989626e-01
-1.42656922e+00 -9.10465121e-01 -4.69433278e-01 1.30730009e+00
-3.66247982e-01 -9.89547491e-01 -8.12545717e-01 -1.17590502e-01
2.92850673e-01 7.18090892e-01 1.57848641e-01 -9.99343246e-02
-4.71416086e-01 -1.03028691e+00 9.08799112e-01 4.56284285e-01
4.72508848e-01 -4.68695194e-01 -2.41710559e-01 2.38889381e-01
-2.38869771e-01 -1.24304879e+00 -4.93074358e-01 2.37131074e-01
-1.20404851e+00 -2.90229380e-01 -5.60029864e-01 -3.55401427e-01
3.70047182e-01 5.83290309e-02 1.28151357e+00 2.61035025e-01
-1.95595458e-01 -3.57155681e-01 -5.97481281e-02 6.43224120e-02
-4.82690156e-01 3.97121221e-01 3.72167051e-01 -4.50326651e-01
2.97637999e-01 -7.03404248e-01 -5.74990690e-01 -2.27533683e-01
-7.27153838e-01 1.93305030e-01 7.37538695e-01 8.25358570e-01
7.90686071e-01 3.26842777e-02 1.73010722e-01 -9.30926144e-01
4.83775318e-01 -4.36233848e-01 -6.08422220e-01 -2.79793143e-03
-8.27298105e-01 2.79076159e-01 8.70840907e-01 -3.79923284e-01
-5.15239656e-01 -3.95324767e-01 -7.95679688e-01 -2.92222291e-01
1.57776669e-01 3.31108630e-01 5.82263581e-02 1.51967093e-01
5.00198245e-01 1.42970353e-01 -2.07463577e-01 -6.23116851e-01
5.31063795e-01 1.09553349e+00 6.52421534e-01 -7.31039107e-01
4.00096565e-01 2.10634425e-01 -1.91870227e-01 -8.83944750e-01
-7.61464417e-01 -3.69832665e-01 -1.51100457e-01 6.34635985e-01
3.24999243e-01 -1.28377724e+00 -7.42593646e-01 3.52919102e-01
-7.55445063e-01 -6.18328273e-01 -7.20687658e-02 4.15360689e-01
-5.11233985e-01 6.38155162e-01 -1.22066951e+00 -6.57344282e-01
-7.54351616e-01 -1.18558311e+00 1.42848575e+00 -2.66186714e-01
-3.48734915e-01 -7.77191877e-01 -3.65930945e-01 4.91576970e-01
4.75538403e-01 -4.60135102e-01 1.12505400e+00 -2.79577345e-01
-4.49843854e-01 -2.73114502e-01 -4.65068489e-01 3.17130595e-01
-4.12665248e-01 -3.38281840e-01 -8.67874563e-01 -8.80895495e-01
-2.54022121e-01 -5.28682411e-01 9.03949022e-01 2.82727182e-01
1.55657256e+00 -5.49427211e-01 -3.65801722e-01 1.27371883e+00
1.28696132e+00 -4.00494188e-01 5.61560392e-01 -5.69545031e-02
6.53033912e-01 -3.36655319e-01 4.33798194e-01 6.27842605e-01
4.73660022e-01 7.88482249e-01 4.12843227e-02 1.43439956e-02
-1.26863018e-01 -4.32957709e-01 4.75397438e-01 1.48692775e+00
3.70987087e-01 -3.00076902e-01 -8.31051171e-01 3.66347849e-01
-1.48891509e+00 -5.01558304e-01 3.33973356e-02 2.56255484e+00
1.25507164e+00 5.73572218e-01 -1.75921053e-01 2.53247797e-01
7.52698854e-02 1.82031646e-01 -4.95547831e-01 -7.54258215e-01
4.50022817e-02 4.95192200e-01 1.06146598e+00 7.51321912e-01
-8.83324981e-01 1.01130450e+00 6.50383997e+00 1.54058695e+00
-1.01140511e+00 3.67155522e-01 1.09465837e+00 -4.91603613e-01
-4.72029656e-01 -1.01164885e-01 -1.32770693e+00 7.26964414e-01
1.70039785e+00 1.77188337e-01 8.26770782e-01 9.28483725e-01
2.06888765e-01 -2.60851204e-01 -9.92666304e-01 1.59466910e+00
-1.46195546e-01 -1.37707555e+00 3.03756595e-01 2.42377996e-01
6.35538220e-01 5.18808067e-01 1.24229409e-01 4.76201385e-01
1.18992642e-01 -1.18883133e+00 5.84136605e-01 1.92805409e-01
1.29850852e+00 -1.11351514e+00 7.29044557e-01 7.15394795e-01
-1.00103056e+00 -6.01239279e-02 -9.69977140e-01 -1.61028460e-01
1.96322396e-01 9.21999693e-01 -8.07862461e-01 2.29766611e-02
6.73877537e-01 3.98594290e-01 -4.68391061e-01 5.51978409e-01
1.76733240e-01 1.09269106e+00 -9.26867247e-01 3.37102152e-02
2.55408227e-01 -1.93471462e-02 1.89709976e-01 1.49685585e+00
6.20304227e-01 3.97710353e-02 -7.98391178e-02 4.21602786e-01
-4.96984720e-01 1.15392469e-01 -1.91727206e-01 -7.86522180e-02
8.85756731e-01 1.01856887e+00 -4.57388878e-01 -6.10610187e-01
-2.78073072e-01 1.19097602e+00 5.46730638e-01 3.21465470e-02
-9.05461431e-01 -4.28189874e-01 9.36524391e-01 2.25546062e-01
2.25622177e-01 -5.16238153e-01 -4.39311296e-01 -1.21828306e+00
-3.27490419e-02 -1.12115157e+00 1.88620985e-01 -3.71159613e-01
-8.85923803e-01 3.14589739e-01 -2.15411633e-01 -7.17920959e-01
-3.28534424e-01 -5.21683335e-01 1.53073249e-02 8.75087678e-01
-1.38992226e+00 -9.47928488e-01 6.87287226e-02 2.41900131e-01
3.84704739e-01 -4.33874279e-02 1.02772176e+00 7.33277082e-01
-5.17904818e-01 1.46993279e+00 6.71607733e-01 -2.45116755e-01
6.48740888e-01 -8.98630977e-01 9.54725921e-01 6.16060853e-01
4.94618952e-01 7.62963116e-01 5.39488256e-01 -3.91181648e-01
-1.78030550e+00 -1.04466987e+00 1.25388587e+00 -3.09138060e-01
5.19699216e-01 -7.22155690e-01 -8.76840949e-01 6.74166977e-01
-2.81866223e-01 1.10319510e-01 5.46217501e-01 4.58726346e-01
-4.49649304e-01 -2.06410229e-01 -1.02905512e+00 7.18096793e-01
1.02641571e+00 -6.88867867e-01 -1.30424738e-01 6.96966648e-01
9.40080643e-01 -8.34282398e-01 -1.06540895e+00 2.10317418e-01
4.91289735e-01 -7.95353472e-01 1.15853989e+00 -2.70735949e-01
2.15433598e-01 1.22205041e-01 -5.50668240e-01 -8.95115316e-01
-3.48567724e-01 -7.81657875e-01 -6.32616341e-01 1.08545673e+00
2.93841720e-01 -5.94941437e-01 7.64066517e-01 4.49358016e-01
8.24240446e-02 -9.95899618e-01 -1.07994008e+00 -6.06152773e-01
1.37606010e-01 -1.21579838e+00 8.00285816e-01 4.92072850e-01
-2.49411047e-01 1.84788957e-01 -3.97598892e-01 1.62268177e-01
5.88072836e-01 -3.40830497e-02 7.07875371e-01 -6.55798852e-01
-7.07455933e-01 -5.06383419e-01 -2.73465753e-01 -1.90309620e+00
1.38199762e-01 -1.16915011e+00 -1.87937811e-01 -1.11645257e+00
3.14034671e-01 -9.15690303e-01 -1.60359894e-03 5.07487476e-01
-8.06327388e-02 8.13195109e-01 2.00197414e-01 1.92756772e-01
-5.96764922e-01 4.07092869e-01 7.06620634e-01 -1.57296527e-02
9.73841026e-02 -2.11903960e-01 -5.44695318e-01 7.61412323e-01
4.20137674e-01 -4.18734908e-01 -1.60244808e-01 -9.79930937e-01
5.21651208e-01 1.28558993e-01 1.40649602e-02 -1.16912079e+00
1.64169952e-01 3.15758258e-01 4.40179110e-01 -6.11286163e-01
5.32942235e-01 -4.73919511e-01 1.54164964e-02 3.38692337e-01
-2.53939658e-01 8.25725198e-02 3.75681907e-01 2.95191854e-01
-1.16959259e-01 -1.46690458e-01 8.23038101e-01 2.27889866e-01
-4.30659056e-01 3.52221280e-01 -2.75697172e-01 2.25911051e-01
3.28468502e-01 1.22630119e-01 1.38830859e-02 -5.55521548e-01
-3.20174456e-01 1.45430833e-01 4.49674487e-01 1.26923949e-01
3.53916556e-01 -1.40194488e+00 -6.26387537e-01 5.44391155e-01
-2.33431205e-01 1.07351653e-01 3.09245497e-01 7.56164134e-01
-7.66288221e-01 5.43849289e-01 3.90651941e-01 -7.69497871e-01
-1.14671826e+00 1.50778681e-01 -3.76754478e-02 -3.69014412e-01
-7.24523425e-01 1.12670481e+00 -2.28952885e-01 -2.03727648e-01
3.41038793e-01 -8.79419088e-01 5.76776445e-01 -1.80697680e-01
8.93165410e-01 3.34472269e-01 4.74928260e-01 -5.46435237e-01
-1.47797063e-01 4.56284612e-01 -8.08802247e-02 7.57565082e-04
1.10252321e+00 -1.52324751e-01 -9.77478325e-02 3.02297026e-01
1.60360241e+00 1.92616746e-01 -1.16653919e+00 -2.82468915e-01
-7.78363794e-02 -4.72027391e-01 4.25271571e-01 -3.69064808e-01
-8.15975785e-01 9.82422471e-01 4.86131221e-01 -2.57847220e-01
1.06715989e+00 -2.47020259e-01 1.80583370e+00 4.94174838e-01
5.87684333e-01 -1.05065131e+00 -3.60307962e-01 7.66545236e-01
2.32887864e-01 -1.14807653e+00 2.46134087e-01 6.14306256e-02
-3.20329070e-01 8.74883533e-01 -8.87066126e-02 8.93863738e-02
3.90153587e-01 7.78844714e-01 -3.39937299e-01 3.96777093e-01
-1.03988135e+00 2.41519943e-01 3.47216159e-01 1.59819186e-01
4.46008265e-01 4.23895597e-01 -1.02476165e-01 6.74535215e-01
-6.07262313e-01 -2.13846713e-01 -5.51605113e-02 5.33425927e-01
-5.47551870e-01 -1.17105663e+00 -2.53569901e-01 9.03890252e-01
-5.88843167e-01 -6.52636349e-01 5.38326740e-01 4.43591416e-01
-6.40335754e-02 7.75264859e-01 6.50216699e-01 -4.36207503e-01
1.02294972e-02 1.28424754e-02 4.28139240e-01 -3.76611114e-01
-4.15147752e-01 1.05156161e-01 1.22052670e-01 -9.54904258e-01
5.12194872e-01 -5.08289039e-01 -1.41808212e+00 -1.07621634e+00
-1.80153757e-01 -9.95288566e-02 6.86512113e-01 7.65449524e-01
6.84741616e-01 2.02513132e-02 2.47138783e-01 -8.86798859e-01
-1.03466320e+00 -1.04026783e+00 -4.60577250e-01 3.33773315e-01
3.07917506e-01 -1.13477252e-01 -1.79362133e-01 -3.60999733e-01] | [8.661556243896484, 3.4595792293548584] |
b83aa113-235f-4598-8e1d-11db7b19c6ab | csi-net-unified-human-body-characterization | 1810.03064 | null | http://arxiv.org/abs/1810.03064v2 | http://arxiv.org/pdf/1810.03064v2.pdf | CSI-Net: Unified Human Body Characterization and Pose Recognition | We build CSI-Net, a unified Deep Neural Network~(DNN), to learn the
representation of WiFi signals. Using CSI-Net, we jointly solved two body
characterization problems: biometrics estimation (including body fat, muscle,
water, and bone rates) and person recognition. We also demonstrated the
application of CSI-Net on two distinctive pose recognition tasks: the hand sign
recognition (fine-scaled action of the hand) and falling detection
(coarse-scaled motion of the body). | ['Fei Wang', 'Shiyuan Zhang', 'Dong Huang', 'Xu He', 'Jinsong Han'] | 2018-10-07 | null | null | null | null | ['person-recognition'] | ['computer-vision'] | [ 2.53031969e-01 -4.97140251e-02 -3.18149507e-01 -4.55245614e-01
-3.94526929e-01 -1.03535742e-01 -2.17636287e-01 -7.81754255e-01
-2.98201740e-01 5.41662693e-01 4.63978618e-01 2.27869704e-01
6.21000305e-02 -7.02547550e-01 -5.79871297e-01 -6.57507896e-01
-4.86184031e-01 4.64582980e-01 -2.68909723e-01 2.35121071e-01
-4.02669936e-01 6.09531581e-01 -1.31176341e+00 -3.05676330e-02
3.75990629e-01 1.46505499e+00 -6.82904124e-01 8.06626379e-01
6.72932208e-01 4.44228381e-01 -4.36085373e-01 -6.83780432e-01
3.85401905e-01 -2.31405780e-01 -4.80232060e-01 -2.58585870e-01
1.02117419e+00 -9.00461495e-01 -8.48521829e-01 8.12186539e-01
1.08286452e+00 -5.23864701e-02 7.25426853e-01 -1.09848058e+00
-3.46624553e-01 7.25161016e-01 -7.73959994e-01 8.77177194e-02
6.93726242e-01 6.70288429e-02 3.88500243e-01 -5.62351584e-01
2.99837261e-01 1.48156929e+00 1.60603535e+00 9.07651007e-01
-7.61603713e-01 -1.21019220e+00 -2.12471679e-01 -1.12612113e-01
-1.75787330e+00 -5.34247041e-01 4.85439271e-01 -1.89447224e-01
1.93598077e-01 2.85633355e-01 1.03582597e+00 1.51146662e+00
2.29410872e-01 9.80754077e-01 8.34103465e-01 2.72571873e-02
-1.63250878e-01 -7.30892837e-01 -1.95324793e-01 9.66067731e-01
6.01587415e-01 1.66506395e-01 -9.23330188e-01 -2.40104169e-01
1.18331623e+00 5.99774309e-02 5.46514168e-02 3.95408124e-02
-9.72375095e-01 5.42861409e-02 3.90284628e-01 -1.36041462e-01
-3.92371237e-01 1.01578450e+00 2.47293636e-01 -9.60474461e-02
2.39631131e-01 -2.57561743e-01 -3.83400321e-01 -2.30580643e-01
-1.19232726e+00 4.90285397e-01 1.00468016e+00 7.85645247e-01
-1.36897758e-01 4.12472636e-01 -5.31471312e-01 4.84178871e-01
9.42008317e-01 1.70411456e+00 1.88125260e-02 -8.15782666e-01
5.50061524e-01 5.65433875e-02 1.90727904e-01 -1.03009140e+00
-8.42728376e-01 -4.57358420e-01 -1.25212526e+00 -2.55914956e-01
8.32933307e-01 -8.32481384e-01 -1.02185738e+00 1.94603157e+00
4.92869705e-01 4.20483500e-01 -3.84560287e-01 1.14770043e+00
1.16903400e+00 -1.67025670e-01 1.94668412e-01 3.41297269e-01
1.40649641e+00 -3.85294616e-01 -5.46309471e-01 -1.57336473e-01
-1.73469797e-01 -1.75684124e-01 3.70469540e-01 4.77182925e-01
-1.16206455e+00 -4.86206263e-01 -8.73449087e-01 1.21437594e-01
5.32568358e-02 2.32956409e-01 8.67438734e-01 1.34682631e+00
-6.05263591e-01 8.80967081e-01 -1.08299816e+00 -1.64049163e-01
4.53724623e-01 8.22606444e-01 -2.52883852e-01 -1.49877414e-01
-1.38599682e+00 4.59371358e-01 -4.99463767e-01 8.81587625e-01
-9.79511082e-01 -5.22642434e-01 -1.01349831e+00 -2.60698348e-01
-1.80819944e-01 -1.02744305e+00 9.97521698e-01 -8.15954983e-01
-1.76616037e+00 9.56502974e-01 1.31163061e-01 -2.69783795e-01
5.26261985e-01 -7.68839002e-01 -8.27635229e-01 -6.41676635e-02
-3.66146952e-01 3.46419007e-01 9.59950030e-01 -3.89214993e-01
7.71760643e-02 -9.28753018e-01 -4.45767462e-01 9.04496410e-04
3.78200375e-02 1.58316195e-01 -5.07869780e-01 -7.60731161e-01
2.51121193e-01 -9.07492280e-01 -1.71912531e-03 5.05609870e-01
-6.89981401e-01 3.61133873e-01 9.86299515e-02 -1.43915832e+00
1.01992369e+00 -1.94334960e+00 3.40746671e-01 5.03345609e-01
1.20444044e-01 -1.02696672e-01 -7.45586306e-02 -2.29236856e-01
3.21728319e-01 -3.42787057e-01 1.64350644e-01 -3.48575234e-01
2.79440522e-01 7.67307132e-02 3.20812166e-01 1.18611193e+00
-4.75264460e-01 9.34267461e-01 -7.40305126e-01 -6.10108376e-01
-1.22364260e-01 6.00049138e-01 -4.56358075e-01 -7.98122883e-02
5.07942557e-01 3.09778929e-01 -2.83594310e-01 1.59356558e+00
8.78034711e-01 2.17781097e-01 3.78818661e-01 -7.66102910e-01
4.46781576e-01 -1.32898241e-01 -1.58311868e+00 1.80468488e+00
-1.45076677e-01 2.24060580e-01 5.03072560e-01 -2.94192702e-01
6.38566256e-01 3.57043892e-01 1.05063283e+00 -6.59169734e-01
3.85722339e-01 -7.58035779e-02 -5.64531907e-02 -5.07530570e-01
4.00007814e-02 -2.33761206e-01 -3.75305533e-01 2.86607951e-01
2.24506319e-01 4.89056438e-01 -5.64384580e-01 -3.07658792e-01
9.38441575e-01 2.68163472e-01 -9.02389586e-02 -1.83700189e-01
1.07368633e-01 -7.85624087e-01 6.78411841e-01 7.86843598e-01
-4.74620461e-01 4.51340914e-01 -1.11676209e-01 -6.03754699e-01
-3.03686023e-01 -1.53877032e+00 -2.55986691e-01 1.34644783e+00
3.44732523e-01 -2.76138425e-01 -6.47194803e-01 -4.42359984e-01
9.21720684e-01 -4.87554073e-01 -8.56183231e-01 -2.33133525e-01
-6.32442653e-01 -8.35022867e-01 1.50903761e+00 1.15893650e+00
6.85557187e-01 -7.28863597e-01 -5.03658652e-01 2.67631281e-02
-2.59839922e-01 -6.42526150e-01 -6.35027528e-01 1.40205353e-01
-7.86562741e-01 -1.12370765e+00 -9.86115813e-01 -3.85999054e-01
1.97822317e-01 -5.50836802e-01 1.14495337e+00 -1.49578884e-01
-7.31713593e-01 7.60190427e-01 3.15893292e-01 -5.04012048e-01
4.92548823e-01 -2.11552277e-01 9.15263593e-01 5.87821454e-02
1.67085558e-01 -6.54207349e-01 -1.13354313e+00 5.61299801e-01
-1.04571849e-01 -3.29318255e-01 3.47883284e-01 4.35093999e-01
5.56680322e-01 -5.65801799e-01 1.00794628e-01 -1.94777578e-01
1.90297008e-01 -2.79575765e-01 3.50123364e-03 3.41912329e-01
5.40373288e-02 -1.28183648e-01 -2.54992038e-01 -5.53298533e-01
-6.27128541e-01 3.21344137e-01 -5.68660319e-01 -8.12852457e-02
1.09327799e-02 1.80221453e-01 -4.92781937e-01 -4.87560749e-01
4.29416925e-01 2.26202592e-01 -1.08249776e-01 -8.07144463e-01
4.56595182e-01 6.06381416e-01 1.34092700e+00 -9.19867516e-01
6.52260482e-01 4.34353054e-01 2.94756532e-01 -7.82659411e-01
-7.03231215e-01 -4.28025097e-01 -8.94574046e-01 -5.42170465e-01
8.65176499e-01 -1.26479185e+00 -1.48786461e+00 1.05393505e+00
-6.36980653e-01 -4.26203281e-01 -1.52926609e-01 4.53632414e-01
-4.37515050e-01 4.56031501e-01 -9.22112703e-01 -1.03171825e+00
-1.04537356e+00 -3.61615807e-01 1.53953660e+00 3.35027516e-01
-4.14483756e-01 -7.14435399e-01 1.96593940e-01 3.42214406e-01
4.10221368e-01 7.52015710e-01 3.96219753e-02 9.37887654e-02
1.27232466e-02 -4.03955877e-01 -1.31133586e-01 6.58826381e-02
6.55836165e-02 -2.89649636e-01 -9.94831860e-01 -3.89312536e-01
-7.61564910e-01 -5.39816856e-01 6.77987695e-01 9.00963366e-01
1.41343689e+00 -1.34948641e-01 -4.04823154e-01 1.36944389e+00
9.95403409e-01 -3.48238379e-01 6.27683938e-01 -1.26937643e-01
7.80811369e-01 -1.83633819e-01 2.13070706e-01 1.00337601e+00
2.73683935e-01 8.75293911e-01 5.29862404e-01 -3.34130585e-01
-5.87568820e-01 -2.20256403e-01 6.70255661e-01 3.61692816e-01
-8.26928914e-01 4.87314165e-02 -6.78035200e-01 1.79582238e-01
-1.46101272e+00 -9.18577015e-01 -2.36217473e-02 1.90185523e+00
5.57904303e-01 -2.41963491e-01 7.25891829e-01 -9.15024206e-02
3.63651574e-01 -1.90352097e-01 -9.11118567e-01 1.53430358e-01
1.06993169e-01 5.46285212e-01 9.91466641e-01 -1.16261113e-02
-1.48650646e+00 4.49518532e-01 8.30497646e+00 3.19027066e-01
-1.17243648e+00 1.37347713e-01 9.48090628e-02 -4.11263883e-01
1.27077579e-01 -1.11497474e+00 -1.05219531e+00 6.03335321e-01
9.53651845e-01 5.77202916e-01 3.44325423e-01 8.45394254e-01
-3.60918462e-01 1.28493413e-01 -1.07985520e+00 1.37485981e+00
1.43233091e-01 -7.41137028e-01 -3.67795557e-01 -7.24696964e-02
1.68540329e-01 2.52387255e-01 2.20365927e-01 2.80287117e-01
2.75269628e-01 -1.17626810e+00 8.25474739e-01 9.08370376e-01
1.55495167e+00 -5.07311940e-01 7.44653940e-01 -2.30404198e-01
-1.67312360e+00 -1.86463855e-02 -2.11585402e-01 -9.86881629e-02
1.10951006e-01 4.89159256e-01 -1.93818688e-01 4.10779744e-01
8.60918581e-01 6.80584431e-01 -3.19906294e-01 1.18017936e+00
-6.94372207e-02 7.66967416e-01 -8.18250060e-01 3.71062934e-01
-5.43532372e-01 2.26427063e-01 7.24639446e-02 1.53491509e+00
1.29231125e-01 5.22957668e-02 2.79446632e-01 7.03873873e-01
-3.27545315e-01 -3.51458877e-01 -8.86978954e-02 2.69328654e-01
1.74264967e-01 1.02494955e+00 -3.22516173e-01 -1.08189853e-02
-6.78768530e-02 1.14182699e+00 -4.32578266e-01 2.37857670e-01
-1.03586960e+00 -1.91199347e-01 1.22316635e+00 1.95846930e-01
1.32488042e-01 -3.44998762e-02 -2.82414824e-01 -1.41051042e+00
-2.31134430e-01 -6.35078728e-01 5.43747485e-01 -4.51289207e-01
-1.38935339e+00 2.07607844e-03 -2.41978809e-01 -8.93475413e-01
-2.04741910e-01 -7.31480777e-01 -3.62635046e-01 9.38075244e-01
-8.05112064e-01 -1.47867322e+00 -6.57676637e-01 9.91736531e-01
-1.43151075e-01 1.34617805e-01 1.00833905e+00 9.40529406e-01
-7.75719941e-01 1.30177259e+00 4.81771529e-02 8.15933824e-01
6.76676631e-01 -1.17056143e+00 6.93057477e-01 6.05916023e-01
-1.63225755e-01 7.22425461e-01 3.84640455e-01 -9.67350781e-01
-2.01874900e+00 -8.85210991e-01 2.95766860e-01 -4.65558380e-01
2.16008976e-01 -4.75865632e-01 7.22369701e-02 6.00341141e-01
-5.38559079e-01 4.81552571e-01 1.03496408e+00 5.44534564e-01
-3.83902609e-01 -5.50394714e-01 -1.43554342e+00 -2.01305095e-02
1.45163023e+00 -5.81585765e-01 -1.99333712e-01 8.60200524e-02
-1.13814361e-01 -8.43735933e-01 -1.27972829e+00 4.13465917e-01
2.04067254e+00 -3.37347895e-01 1.75321114e+00 -8.94176662e-01
-1.13247670e-02 -2.99125612e-02 -3.77595276e-01 -6.74468577e-01
-5.99924743e-01 -1.69181153e-01 -9.41258967e-01 9.61430609e-01
-1.35200977e-01 1.76804259e-01 1.14616632e+00 6.55539036e-01
3.57598722e-01 -4.97021228e-01 -1.17557120e+00 -8.51514637e-01
-2.58879274e-01 -3.67579788e-01 6.23165548e-01 4.74187464e-01
-1.27502546e-01 -2.41903245e-01 -1.24048388e+00 4.10805762e-01
1.16299939e+00 1.45907432e-01 7.50912666e-01 -1.55685115e+00
-5.17140627e-01 1.01650536e-01 -5.66950023e-01 -1.17629015e+00
-1.82433710e-01 -4.82827634e-01 1.03157952e-01 -1.26414859e+00
2.92283118e-01 -1.99106500e-01 -7.35195696e-01 6.67411387e-01
1.01956472e-01 7.87356853e-01 1.70116514e-01 -2.33977661e-01
-7.42098153e-01 1.08186103e-01 1.01756465e+00 -5.14396608e-01
6.04720265e-02 4.23366517e-01 -4.84756619e-01 8.87063384e-01
2.32682049e-01 -2.11109087e-01 3.27404320e-01 -4.85675782e-01
3.97467047e-01 4.22080189e-01 5.84118903e-01 -1.55629313e+00
3.02700531e-02 -1.18364386e-01 1.40782106e+00 -5.96803606e-01
4.92930025e-01 -8.13832700e-01 4.12570804e-01 1.00809062e+00
-5.42449616e-02 -3.28408629e-01 2.80379087e-01 4.24915284e-01
4.28188056e-01 4.54974353e-01 6.27842009e-01 -3.81184034e-02
-4.64747906e-01 6.61728740e-01 -4.41376537e-01 -2.13509679e-01
7.30508566e-02 -3.01825136e-01 5.80447242e-02 -3.92533153e-01
-1.16009843e+00 2.98879184e-02 -9.72459465e-02 2.55674690e-01
6.56534731e-01 -1.67800117e+00 -5.55131435e-01 3.27037841e-01
-2.85415947e-01 -4.30551410e-01 2.35207319e-01 1.01722670e+00
-4.94634211e-01 -2.90074348e-02 -6.33412242e-01 -4.39852327e-01
-1.50654197e+00 -1.84465274e-02 9.20883298e-01 -7.43335858e-02
-6.24793768e-01 1.11385131e+00 -3.45847666e-01 -3.98612797e-01
8.93694401e-01 -7.12007761e-01 1.92669660e-01 -2.01344773e-01
7.86582589e-01 7.15770543e-01 -2.52087891e-01 -3.42905700e-01
-1.04363239e+00 9.03770804e-01 7.60942459e-01 1.47672161e-01
1.35856223e+00 -1.56964272e-01 1.19032018e-01 9.02972296e-02
5.16372263e-01 -2.20534131e-02 -1.15157580e+00 -1.42089143e-01
-3.04978609e-01 -2.08720878e-01 -1.43835591e-02 -1.24783647e+00
-1.34474063e+00 6.59668684e-01 1.31632018e+00 -8.27913761e-01
9.16253865e-01 -2.33268887e-01 1.11532581e+00 4.66811240e-01
5.72076023e-01 -9.68177497e-01 -1.80750579e-01 3.41757298e-01
6.66594446e-01 -9.27508771e-01 4.20503497e-01 -2.55773962e-01
-4.02727909e-02 1.03831840e+00 5.94837725e-01 4.27326653e-03
1.04086804e+00 5.84054828e-01 -4.49945666e-02 -2.90266782e-01
2.18942434e-01 -3.07811022e-01 9.12594259e-01 9.60634530e-01
5.93701243e-01 5.04852176e-01 2.00988770e-01 1.31299031e+00
-7.67206997e-02 3.90263855e-01 -3.30209225e-01 7.19437063e-01
-2.88567990e-01 -5.74671686e-01 -6.69095755e-01 6.32633686e-01
-6.43743575e-01 1.43610612e-01 -6.61927164e-01 5.11949480e-01
7.30283856e-01 3.59921068e-01 -8.14793929e-02 -6.99633062e-01
3.22061121e-01 8.18274170e-02 1.14954293e+00 -3.20517093e-01
-6.77292883e-01 -9.76760462e-02 2.33372435e-01 -1.01647806e+00
-5.92184842e-01 -7.17005908e-01 -9.12331462e-01 -5.16398728e-01
-1.49102569e-01 -5.37223816e-01 6.53210402e-01 9.51247871e-01
-9.69616026e-02 4.62802142e-01 6.78893030e-02 -1.16521537e+00
-6.16858482e-01 -1.10969889e+00 -1.45071781e+00 3.54945987e-01
2.67583042e-01 -8.83161068e-01 3.43970031e-01 -1.03544042e-01] | [6.878854274749756, 0.20472149550914764] |
445983af-1ff0-4d49-ad53-e641c9ae483a | neural-dynamic-focused-topic-model-1 | 2301.10988 | null | https://arxiv.org/abs/2301.10988v1 | https://arxiv.org/pdf/2301.10988v1.pdf | Neural Dynamic Focused Topic Model | Topic models and all their variants analyse text by learning meaningful representations through word co-occurrences. As pointed out by Williamson et al. (2010), such models implicitly assume that the probability of a topic to be active and its proportion within each document are positively correlated. This correlation can be strongly detrimental in the case of documents created over time, simply because recent documents are likely better described by new and hence rare topics. In this work we leverage recent advances in neural variational inference and present an alternative neural approach to the dynamic Focused Topic Model. Indeed, we develop a neural model for topic evolution which exploits sequences of Bernoulli random variables in order to track the appearances of topics, thereby decoupling their activities from their proportions. We evaluate our model on three different datasets (the UN general debates, the collection of NeurIPS papers, and the ACL Anthology dataset) and show that it (i) outperforms state-of-the-art topic models in generalization tasks and (ii) performs comparably to them on prediction tasks, while employing roughly the same number of parameters, and converging about two times faster. Source code to reproduce our experiments is available online. | ['César Ojeda', 'Ramsés J. Sánchez', 'Kostadin Cvejoski'] | 2023-01-26 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [ 1.11348957e-01 3.29389066e-01 -4.49953377e-01 -1.14901885e-01
-6.69779003e-01 -5.80498695e-01 1.42958581e+00 4.13955718e-01
-4.90808368e-01 9.11102295e-01 3.92287850e-01 -3.21418464e-01
-2.42110521e-01 -7.51426578e-01 -8.56779575e-01 -7.49005020e-01
-2.03326806e-01 1.04279828e+00 5.08122444e-01 -4.97963950e-02
4.09935921e-01 -8.81911665e-02 -1.43772686e+00 -1.58555415e-02
8.25271547e-01 5.59793890e-01 2.88643688e-01 5.87599456e-01
-1.99443534e-01 4.03701007e-01 -6.87416434e-01 -5.45450389e-01
-2.26326704e-01 -1.76487282e-01 -7.28199422e-01 -1.03822105e-01
3.71139556e-01 9.40128118e-02 -4.09108758e-01 8.34180057e-01
3.07244472e-02 4.01114702e-01 1.11014307e+00 -1.06072986e+00
-6.20898843e-01 9.64521527e-01 -6.81044221e-01 4.50732261e-01
-2.90251285e-01 -2.28009373e-01 1.38140953e+00 -6.00246310e-01
9.07269955e-01 1.31293559e+00 5.91297090e-01 4.55788463e-01
-1.53602588e+00 -4.78429079e-01 6.89185321e-01 1.84733361e-01
-1.01521397e+00 -2.41750464e-01 7.65430391e-01 -5.27217865e-01
8.87255371e-01 -1.52675584e-01 7.81377077e-01 1.72257984e+00
4.81025785e-01 9.23734188e-01 7.17375815e-01 -3.75365525e-01
4.14833426e-01 2.46151000e-01 4.75604177e-01 1.29274338e-01
2.57617652e-01 -1.21279232e-01 -8.02724183e-01 -4.48480487e-01
4.14712548e-01 -1.46319315e-01 -1.40225902e-01 -5.80545962e-01
-1.17202342e+00 1.24401307e+00 -1.56317558e-03 5.63580632e-01
-4.76887852e-01 3.51914883e-01 3.92348796e-01 -5.98864295e-02
1.05342400e+00 2.42233992e-01 -5.61136842e-01 -4.87649381e-01
-1.21631610e+00 7.90745318e-01 1.01390636e+00 5.74262738e-01
4.39868003e-01 -1.67296156e-01 -2.57310010e-02 9.36408222e-01
4.71364647e-01 2.01420769e-01 7.52367496e-01 -7.65227497e-01
1.31541595e-01 1.49754569e-01 1.23468198e-01 -8.22400928e-01
-2.21440062e-01 -4.81932282e-01 -6.14827037e-01 -2.57735401e-01
4.38559800e-01 -7.90396780e-02 -8.95693302e-01 2.03986359e+00
1.01380311e-01 3.11233044e-01 -2.29693428e-01 3.18375081e-01
3.42789590e-01 1.11474335e+00 3.70953619e-01 -3.17494690e-01
1.26160908e+00 -6.48938477e-01 -7.96774685e-01 -2.80950099e-01
1.93333834e-01 -4.53338087e-01 7.80222476e-01 6.26359642e-01
-1.06766665e+00 -2.45804474e-01 -7.44293213e-01 1.97092928e-02
-4.54839647e-01 -2.87789106e-01 8.32456112e-01 5.27767003e-01
-1.17381001e+00 7.02577233e-01 -1.08677840e+00 -6.61680043e-01
4.52839702e-01 2.55963486e-03 1.87180683e-01 2.95114696e-01
-1.26058412e+00 8.26614380e-01 3.92675221e-01 -2.46989354e-01
-1.09080756e+00 -9.77794528e-01 -4.67311144e-01 2.11841851e-01
4.41218466e-01 -5.15805244e-01 1.45480466e+00 -6.49734974e-01
-1.39904976e+00 7.68984318e-01 -3.13313246e-01 -8.60475600e-01
4.05147910e-01 -4.98220891e-01 -4.88965884e-02 -7.66830370e-02
-6.96645491e-03 7.47396290e-01 8.30693305e-01 -1.15851700e+00
-6.28412068e-01 -3.28689039e-01 -9.88671929e-02 4.06668745e-02
-4.54455405e-01 -2.13366240e-01 -3.91777635e-01 -7.09689558e-01
-9.54467207e-02 -9.15267885e-01 -7.70257339e-02 -1.30214661e-01
-2.87977099e-01 -8.59059453e-01 8.53288233e-01 -5.46703279e-01
1.07025766e+00 -1.82856870e+00 4.81882304e-01 -4.01230194e-02
2.56253481e-01 -1.09492406e-01 2.99222291e-01 4.97505903e-01
7.68598691e-02 1.35895640e-01 -3.09676766e-01 -6.86112702e-01
2.43837714e-01 3.08318138e-01 -8.01098585e-01 5.52511454e-01
6.73489198e-02 6.94992006e-01 -8.62684309e-01 -3.38037580e-01
-2.25297222e-03 6.28189266e-01 -4.89651263e-01 -2.56537169e-01
-7.68564224e-01 9.84438956e-02 -3.23526680e-01 1.75163239e-01
3.08716714e-01 -3.90681267e-01 3.42394590e-01 2.77076691e-01
-8.85627344e-02 6.46194458e-01 -8.91188979e-01 1.49648809e+00
-3.54626387e-01 1.22133124e+00 -1.20759860e-01 -1.22569883e+00
8.37407827e-01 4.26589698e-01 4.05183136e-01 -1.44493818e-01
1.15558475e-01 -8.91905874e-02 -6.44434988e-02 -9.66367219e-03
8.94188285e-01 -1.74562857e-01 1.35137141e-01 7.73771703e-01
3.60466897e-01 -1.06325150e-01 4.04447734e-01 5.57260871e-01
6.83630824e-01 2.11705893e-01 9.41646919e-02 -5.35258412e-01
-1.92007218e-02 -2.47019157e-01 3.03811014e-01 1.14605701e+00
5.58786420e-03 2.88940191e-01 9.55065131e-01 -2.57640541e-01
-1.29717708e+00 -1.25058806e+00 -6.82956338e-01 1.33117497e+00
-2.27931812e-01 -5.41420341e-01 -5.66304982e-01 -2.93804258e-01
-6.37114495e-02 1.11095071e+00 -1.18071282e+00 -8.09061453e-02
-4.70010757e-01 -9.89757001e-01 2.08833754e-01 3.29901338e-01
-7.85474926e-02 -1.11225617e+00 -5.99373817e-01 2.93771297e-01
-9.53136310e-02 -5.41955292e-01 4.56234999e-02 3.46363366e-01
-1.10761380e+00 -7.63871849e-01 -1.07609880e+00 -2.94571340e-01
6.48816898e-02 -1.54829323e-01 1.23319721e+00 -2.38526702e-01
-1.56566828e-01 4.20359731e-01 -1.18744411e-01 -7.17214048e-01
-4.49176639e-01 5.75777233e-01 -3.57062146e-02 -2.40510240e-01
4.17793125e-01 -6.32326424e-01 -3.45729649e-01 -1.49212718e-01
-1.03080070e+00 -1.75074801e-01 1.86300755e-01 8.80901158e-01
2.41663009e-01 -4.37516384e-02 4.17158335e-01 -1.07065392e+00
7.41764724e-01 -9.04543579e-01 -6.10959411e-01 9.68058705e-02
-8.01277697e-01 4.69103567e-02 -5.57183661e-03 -6.99422240e-01
-1.14880025e+00 -7.80333161e-01 2.43613973e-01 -1.87346652e-01
-1.61505923e-01 5.79394698e-01 3.52229208e-01 8.38831842e-01
4.38389927e-01 2.82242388e-01 -2.02968985e-01 -5.07884920e-01
5.47750294e-01 4.16645825e-01 2.29690015e-01 -7.20513999e-01
5.48072994e-01 7.16133535e-01 -1.82687402e-01 -1.07211077e+00
-8.88030410e-01 -4.68596965e-01 -4.47394848e-01 -2.57186741e-01
6.29239976e-01 -8.54825377e-01 -3.46019953e-01 6.08133793e-01
-1.27539468e+00 -4.21617061e-01 -2.67150670e-01 4.76623148e-01
-6.25894547e-01 1.40391082e-01 -5.95125854e-01 -1.01073992e+00
-1.03082210e-01 -7.74363101e-01 1.10293353e+00 3.21041465e-01
-4.68403876e-01 -1.48473859e+00 5.18951476e-01 -1.48748860e-01
5.57333410e-01 -1.12378076e-01 1.13860476e+00 -8.51208329e-01
-3.78538728e-01 2.40070261e-02 6.37629181e-02 -9.71718803e-02
-2.34454930e-01 4.36157942e-01 -9.84391868e-01 -2.54171997e-01
-2.05525942e-02 -8.39199573e-02 1.39081335e+00 8.81129324e-01
9.02919650e-01 -1.50893122e-01 -8.34806979e-01 1.96495727e-01
1.07934618e+00 2.63243914e-02 4.10166830e-01 5.72651982e-01
1.32540286e-01 8.51403236e-01 1.49194583e-01 4.37772751e-01
3.21047872e-01 6.52115583e-01 2.19468713e-01 3.34170103e-01
3.46265286e-01 -2.43942127e-01 1.57196313e-01 7.01334357e-01
-5.22525385e-02 -6.23674393e-01 -9.16831613e-01 9.55520928e-01
-1.97200465e+00 -1.08434570e+00 -1.35953262e-01 1.96892941e+00
9.28608537e-01 4.46543902e-01 3.18038225e-01 -2.81425506e-01
6.73071980e-01 5.49378872e-01 -5.41205108e-01 -2.65482843e-01
-2.28450000e-01 3.71472478e-01 1.70267329e-01 6.14712298e-01
-1.04321229e+00 9.15487349e-01 6.62669468e+00 6.68583155e-01
-7.85350740e-01 2.81158388e-01 7.50591874e-01 -3.30330700e-01
-4.91800785e-01 1.49519928e-02 -8.98621142e-01 5.09511232e-01
1.37892640e+00 -5.31020045e-01 -7.86377192e-02 8.45475018e-01
-1.03006966e-01 -2.80037314e-01 -9.31961000e-01 3.44374835e-01
1.35114878e-01 -1.30299079e+00 1.12546176e-01 2.82563388e-01
8.75130057e-01 2.43640065e-01 4.07501280e-01 4.63630378e-01
6.20099545e-01 -7.26775825e-01 8.42354178e-01 7.63019860e-01
-1.55403586e-02 -5.91856360e-01 4.93671477e-01 3.49676818e-01
-5.69778264e-01 4.74292524e-02 -5.51656723e-01 1.20323069e-01
3.09136033e-01 7.79369056e-01 -5.88424504e-01 2.56490588e-01
8.20230782e-01 8.84954512e-01 -4.52002674e-01 8.74833763e-01
-2.16146216e-01 1.17932880e+00 -4.23059344e-01 -2.59512663e-01
3.92811596e-01 -1.62765026e-01 9.10443962e-01 1.06629300e+00
1.44103050e-01 -2.87340701e-01 -2.66722262e-01 1.15073884e+00
-6.96442463e-03 -1.96803291e-03 -3.00024718e-01 -1.90076217e-01
4.23667759e-01 1.02948320e+00 -9.04261708e-01 -4.62419033e-01
-1.96213946e-01 6.11005664e-01 3.52691323e-01 5.19468367e-01
-6.76027596e-01 -1.58912554e-01 6.28871977e-01 9.92540456e-03
7.75724769e-01 -2.26050124e-01 -1.33792877e-01 -1.02347481e+00
-2.17147395e-01 -4.25513834e-01 2.05222696e-01 -6.08602643e-01
-1.42960310e+00 6.28455400e-01 4.13274348e-01 -4.03066158e-01
-5.06410718e-01 -5.76781332e-01 -7.61469841e-01 8.14491749e-01
-1.32894814e+00 -8.15327108e-01 1.47768795e-01 1.53235361e-01
7.38824308e-01 -1.04157083e-01 6.13380671e-01 -1.43057212e-01
-3.86137635e-01 1.11929953e-01 5.48618257e-01 -3.24134946e-01
6.09433353e-01 -1.37828541e+00 6.92431152e-01 4.78237689e-01
4.88354653e-01 8.50494504e-01 1.09275508e+00 -7.48932838e-01
-8.16505671e-01 -6.45200133e-01 1.03738415e+00 -7.80103326e-01
1.02753425e+00 -6.76392138e-01 -1.20083261e+00 8.36144984e-01
5.11458397e-01 -7.77218163e-01 5.98314285e-01 7.13621080e-01
-3.01077902e-01 4.04396057e-01 -4.28653628e-01 5.57285845e-01
5.58793187e-01 -2.27369413e-01 -9.52213407e-01 5.69690406e-01
8.32213521e-01 -5.65481782e-02 -5.84309459e-01 5.52060641e-02
5.67982733e-01 -9.63089526e-01 7.60618031e-01 -7.81319857e-01
6.34161770e-01 2.61249810e-01 -9.01104137e-02 -1.31806767e+00
-2.42344603e-01 -6.09010637e-01 -5.20933986e-01 1.26396263e+00
5.49190462e-01 -5.67721009e-01 8.87673438e-01 1.93143040e-01
1.69221491e-01 -8.68220925e-01 -1.07657576e+00 -5.86569965e-01
7.36930788e-01 -4.33377117e-01 2.04290792e-01 8.62766802e-01
-1.24684358e-02 3.20374608e-01 -2.10609987e-01 -2.82572985e-01
6.11125112e-01 -8.68259445e-02 6.02999568e-01 -1.85691345e+00
-3.37165982e-01 -9.18124735e-01 -9.45879593e-02 -1.05152416e+00
4.15162116e-01 -5.30426741e-01 9.67015028e-02 -1.32675862e+00
4.33557332e-01 -3.31240863e-01 -2.34932333e-01 9.93384197e-02
-3.42397541e-01 -9.91519541e-02 -1.39701543e-02 3.86240751e-01
-5.22613525e-01 7.98733175e-01 5.83531320e-01 -2.36149188e-02
-6.76512346e-02 1.20035782e-01 -6.58515811e-01 6.78023219e-01
6.54174805e-01 -5.53136230e-01 -3.52390587e-01 -1.82519004e-01
4.53297615e-01 -2.69402415e-01 3.36530864e-01 -7.33980000e-01
1.17811047e-01 1.42045515e-02 2.29184389e-01 -8.06582749e-01
5.95961809e-01 -2.69517958e-01 -1.31923944e-01 2.98144281e-01
-7.12574601e-01 6.30318362e-04 2.55699337e-01 1.05702090e+00
3.22887637e-02 -3.77522469e-01 4.45220172e-01 -1.73309773e-01
-2.40975901e-01 2.04469398e-01 -8.53042424e-01 1.81578875e-01
7.55574465e-01 1.48838475e-01 -2.97605872e-01 -7.16129422e-01
-6.38413191e-01 9.99078527e-02 3.22857976e-01 6.61820710e-01
5.67691065e-02 -7.41857886e-01 -7.58685589e-01 -2.12941617e-01
-1.07008971e-01 -9.60519165e-02 2.73872733e-01 7.12810278e-01
2.78755762e-02 8.32725883e-01 1.98103383e-01 -8.08259666e-01
-8.54220748e-01 4.58224833e-01 1.28336996e-01 -5.18845022e-01
-6.33677721e-01 9.38094378e-01 4.01564837e-01 -2.69926280e-01
3.56312633e-01 -1.09525524e-01 -2.35133782e-01 3.96523386e-01
4.57285941e-01 3.39293987e-01 -2.47147486e-01 -2.77192861e-01
-4.59849648e-02 2.44043544e-01 -7.16026008e-01 -5.43226302e-01
1.65095878e+00 -9.74057540e-02 -1.04361337e-04 1.22412670e+00
8.55682671e-01 -3.09164882e-01 -1.30330896e+00 -4.74802762e-01
1.33120939e-01 -3.75142060e-02 1.46833554e-01 -6.40418887e-01
-5.65832317e-01 1.06908572e+00 2.39295751e-01 6.37881517e-01
4.47488219e-01 5.05318284e-01 4.94309008e-01 1.40307918e-01
-1.91966654e-03 -1.04403746e+00 1.20857120e-01 5.20413518e-01
5.81460416e-01 -9.12651360e-01 1.13560118e-01 5.89019693e-02
-4.78632510e-01 1.09534168e+00 2.32671514e-01 -2.48363316e-01
8.46624374e-01 -1.48419989e-02 -3.43682408e-01 -3.61123145e-01
-1.34394133e+00 2.95791358e-01 2.83541739e-01 5.01940429e-01
5.46188235e-01 -7.55289644e-02 -4.14748341e-01 4.26631361e-01
-4.05188441e-01 -3.83107394e-01 6.11474633e-01 7.33417094e-01
-6.11272216e-01 -9.91702735e-01 -2.05376521e-01 5.26707590e-01
-6.40969574e-01 -2.65901834e-01 -2.93045402e-01 9.85394657e-01
-3.01145673e-01 5.81785798e-01 6.40431106e-01 2.87106425e-01
-2.68612266e-01 5.30609965e-01 3.36776853e-01 -5.31077504e-01
-2.33399138e-01 2.36689642e-01 -2.03405783e-01 -9.85193104e-02
-3.90385389e-01 -1.22357965e+00 -5.78307450e-01 -1.71786144e-01
-4.10956711e-01 3.40980500e-01 9.16351080e-01 1.03612506e+00
1.40773237e-01 7.39312291e-01 8.81078988e-02 -8.81185651e-01
-5.72494030e-01 -1.51948667e+00 -6.85645998e-01 1.00138046e-01
1.85556188e-01 -9.63226199e-01 -5.78840256e-01 2.35337272e-01] | [10.391434669494629, 6.972903728485107] |
d086ae3e-f134-436c-896c-e74bb07e37ff | deep-learning-decoding-of-mental-state-in-non | 1911.05661 | null | http://arxiv.org/abs/1911.05661v1 | http://arxiv.org/pdf/1911.05661v1.pdf | Deep Learning Decoding of Mental State in Non-invasive Brain Computer Interface | Brain computer interface (BCI) has been popular as a key approach to monitor
our brains recent year. Mental states monitoring is one of the most important
BCI applications and becomes increasingly accessible. However, the mental state
prediction accuracy and generality through encephalogram (EEG) are not good
enough for everyday use. Here in this paper we present a deep learning-based
EEG decoding method to read mental states. We propose a novel 1D convolutional
neural network with different filter lengths to capture different frequency
bands information. To improve the prediction accuracy, we also used a
resnet-like structure to train a relatively deep convolutional neural network
to promote feature extraction. Compared with traditional ways of predictions
such as KNN and SVM, we achieved a significantly better result with an accuracy
of 96.40%. Also, in contrast with some already published open source deep
neural network structures, our methods achieved the state of art prediction
accuracy on a mental state recognition dataset. Our results demonstrate using
only 1D convolution could extract the features of EEG and the possibility of
mental state prediction using portable EEG devices. | [] | 2019-11-11 | null | null | null | null | ['eeg-decoding', 'eeg-decoding'] | ['medical', 'time-series'] | [ 1.48199618e-01 -1.10561639e-01 2.84164846e-01 -4.03945863e-01
7.91245624e-02 1.00211866e-01 4.04227674e-01 -1.74716115e-01
-5.04835784e-01 1.03110290e+00 5.24367206e-02 -2.98670650e-01
-1.74727201e-01 -8.18920553e-01 -3.94229650e-01 -5.37460744e-01
-2.19263047e-01 4.13964726e-02 1.14159018e-01 -3.76939178e-01
5.33075511e-01 5.83987057e-01 -1.61504507e+00 5.32412887e-01
9.82307911e-01 1.37748730e+00 3.01536024e-01 2.54244179e-01
5.43937758e-02 7.07253277e-01 -9.56476986e-01 -3.61011550e-02
-1.33289635e-01 -2.03723937e-01 -8.40033412e-01 -5.11595249e-01
-2.71059424e-01 -3.08099151e-01 -7.58441985e-01 9.24023032e-01
8.33865225e-01 -6.89336285e-02 4.27534372e-01 -9.76342022e-01
-5.07549524e-01 4.84563917e-01 -9.71566960e-02 6.42353415e-01
6.48484766e-01 4.55303937e-02 5.40163368e-02 -3.73785764e-01
-7.59976655e-02 6.19633317e-01 5.36984265e-01 7.51778126e-01
-9.54117239e-01 -1.34774959e+00 -2.03712806e-01 8.83479357e-01
-1.25940478e+00 -4.93660659e-01 7.43540347e-01 -2.22337916e-01
1.53416574e+00 1.00438178e-01 1.37670016e+00 1.34855747e+00
7.48144984e-01 4.88011003e-01 1.50127101e+00 -1.28194213e-01
1.57844529e-01 1.36753246e-01 3.75664324e-01 2.88495779e-01
6.16534464e-02 -1.15146220e-01 -7.10833073e-01 1.53073370e-01
8.20174515e-01 7.55922347e-02 -6.77149296e-01 4.35602844e-01
-1.47847044e+00 2.68009543e-01 3.57194096e-01 5.73187709e-01
-4.46960658e-01 -2.91088372e-02 3.32610637e-01 4.04979676e-01
4.54884171e-01 6.15235627e-01 -4.22734737e-01 -5.77328920e-01
-8.62439632e-01 -2.60883659e-01 8.49946320e-01 3.48234653e-01
3.72745365e-01 2.18300670e-01 -2.14442000e-01 8.03048134e-01
-1.11047111e-01 3.81512523e-01 1.16194439e+00 -3.86188775e-01
6.48386851e-02 7.29951680e-01 -8.93418044e-02 -8.52154374e-01
-7.81448007e-01 -5.94509363e-01 -1.29634607e+00 1.23720735e-01
-2.47680908e-03 -2.28101984e-01 -6.68677211e-01 1.44834244e+00
-3.37132007e-01 7.12568820e-01 4.76866923e-02 7.57762015e-01
8.79460275e-01 6.52139366e-01 -7.03826966e-03 -2.46743992e-01
1.27423775e+00 -3.80002260e-01 -9.61817026e-01 -1.50442392e-01
3.96185011e-01 -2.50538528e-01 6.89008117e-01 8.76839578e-01
-9.18375075e-01 -6.40144885e-01 -1.37997353e+00 3.29008490e-01
-5.96841574e-01 1.94812670e-01 9.21449184e-01 7.98100233e-01
-1.38216209e+00 8.59985471e-01 -9.51938391e-01 -3.98689240e-01
6.01638556e-01 9.80213523e-01 -5.84442437e-01 5.76125622e-01
-1.46991813e+00 1.44691300e+00 5.96092045e-01 1.02443516e-01
-6.73478782e-01 -2.92013347e-01 -4.02320534e-01 4.19898868e-01
-3.57045233e-01 -2.50557005e-01 8.25133979e-01 -8.05183709e-01
-2.02850294e+00 5.17142177e-01 -2.58812487e-01 -5.45741081e-01
-1.93159878e-01 -4.60394509e-02 -8.62083375e-01 7.32642561e-02
-5.16794801e-01 5.47167897e-01 5.69533646e-01 -3.10291439e-01
-3.35049629e-01 -3.78906399e-01 -1.21189013e-01 -1.00808151e-01
-9.49682355e-01 8.16213191e-02 1.19341470e-01 -2.50061005e-01
1.93802685e-01 -7.08650827e-01 2.47008592e-01 -5.00765920e-01
-1.43997476e-01 -3.75127047e-01 5.06786585e-01 -8.41700375e-01
1.31877267e+00 -1.84999371e+00 -1.69864550e-01 2.14169815e-01
3.64909500e-01 5.63063741e-01 -2.44608317e-02 1.55316172e-02
-6.15556538e-01 -2.35518977e-01 1.30134225e-01 1.49798125e-01
-2.65328377e-01 -1.34881109e-01 -1.42296776e-01 4.25013214e-01
2.22415432e-01 9.93599713e-01 -6.09221280e-01 3.65343168e-02
3.38338733e-01 6.48966372e-01 -3.82850260e-01 3.40573758e-01
4.71809924e-01 6.12260103e-01 7.46017620e-02 1.95192382e-01
7.14534819e-01 -2.61190206e-01 6.96136951e-02 -3.15155536e-01
-1.05537936e-01 6.05286837e-01 -7.92431712e-01 1.77925873e+00
-3.79942298e-01 9.03878570e-01 -5.32734513e-01 -1.44445372e+00
1.22141874e+00 6.19544744e-01 6.84093416e-01 -1.03202462e+00
4.67416018e-01 -1.48263136e-02 2.18568102e-01 -6.56997919e-01
-9.22480375e-02 -1.63702115e-01 2.77475566e-01 3.79126161e-01
4.54961598e-01 1.97591916e-01 -2.65884846e-01 -2.22002253e-01
1.13349664e+00 -3.91938239e-02 3.60416353e-01 -4.63370323e-01
5.74679315e-01 -5.17840028e-01 4.28595215e-01 3.64367425e-01
-1.66803285e-01 3.68425876e-01 5.19177198e-01 -6.69214249e-01
-4.74570841e-01 -4.79394287e-01 -5.30933797e-01 5.17437756e-01
1.83378905e-01 -3.81858021e-01 -8.12786996e-01 -4.70017456e-02
-4.48819339e-01 6.10676944e-01 -5.67365170e-01 -6.65666759e-01
-1.91326201e-01 -1.01469004e+00 6.53046846e-01 5.89654088e-01
1.10150361e+00 -1.13317156e+00 -8.23755085e-01 5.41092873e-01
-3.28621358e-01 -8.48210931e-01 2.58070976e-01 6.09923661e-01
-9.04098809e-01 -7.95771539e-01 -7.80408561e-01 -9.71862376e-01
2.14866608e-01 -1.70613796e-01 6.06023192e-01 -3.79645713e-02
-1.31896809e-01 -3.07694465e-01 -2.11125568e-01 -6.47866726e-01
5.35196485e-03 3.16835493e-01 3.68494362e-01 7.85709471e-02
8.27113509e-01 -1.09922969e+00 -9.32208419e-01 1.27320915e-01
-5.59409797e-01 3.66361678e-01 5.29126287e-01 5.62333941e-01
8.09097216e-02 3.00612729e-02 1.00423539e+00 -2.30771437e-01
1.06307232e+00 -3.87520999e-01 -1.19440585e-01 3.36596072e-01
-6.93015993e-01 6.21318966e-02 5.67899644e-01 -6.71647668e-01
-1.12749398e+00 -1.26090065e-01 -5.55024028e-01 3.72941829e-02
-4.87833917e-01 4.67537165e-01 1.11843765e-01 -3.58940899e-01
6.45046592e-01 6.70642436e-01 1.36834592e-01 -3.43842119e-01
-6.06990635e-01 1.19972634e+00 3.32208037e-01 -1.42210424e-01
-1.37461916e-01 2.02643424e-02 -1.99365839e-01 -6.81991518e-01
-4.28527683e-01 -1.55941412e-01 -5.13396740e-01 -2.75974363e-01
9.11440372e-01 -1.03453553e+00 -1.26009381e+00 9.02349293e-01
-1.29278934e+00 -3.12662601e-01 2.91934907e-01 8.15490842e-01
-3.53764057e-01 2.69930571e-01 -8.67032528e-01 -7.16099918e-01
-6.83430791e-01 -1.18959582e+00 4.25736666e-01 2.45394111e-01
-3.76597822e-01 -5.58080256e-01 7.94757009e-02 -1.48939369e-02
1.00149000e+00 2.43153907e-02 8.50487828e-01 -6.88318491e-01
6.20836020e-02 -2.36100242e-01 -2.19554797e-01 4.24687862e-01
2.17048049e-01 -5.56763113e-01 -1.28246474e+00 -1.45765215e-01
3.88356477e-01 -2.48746201e-01 5.80222070e-01 2.02532142e-01
1.83070016e+00 -8.17581192e-02 -3.35796773e-01 6.60144448e-01
1.14061606e+00 8.20354939e-01 1.36719596e+00 3.77156854e-01
1.35421976e-01 1.13735303e-01 -2.16855705e-01 7.01513648e-01
2.76731342e-01 2.70176888e-01 2.22772703e-01 9.30491984e-02
8.84860530e-02 2.23347083e-01 3.09316099e-01 9.70121443e-01
-6.30801618e-01 1.32072970e-01 -1.02332556e+00 1.48454845e-01
-1.52219498e+00 -9.71862435e-01 8.45407471e-02 2.10672903e+00
8.66927624e-01 1.18090481e-01 -1.97182164e-01 4.44894820e-01
5.41585624e-01 -2.61386216e-01 -6.31996632e-01 -3.91721606e-01
3.40139046e-02 9.44001794e-01 3.21606584e-02 -2.38901734e-01
-8.15481126e-01 7.22804129e-01 6.50665474e+00 6.35653317e-01
-1.69252110e+00 2.91189551e-01 5.46943188e-01 -3.60110760e-01
3.02345872e-01 -6.54130161e-01 -6.02704227e-01 8.42481613e-01
1.54999435e+00 2.04585120e-02 6.73524559e-01 4.49654967e-01
9.42076370e-02 -1.07671417e-01 -9.57006931e-01 1.63672519e+00
7.63805211e-02 -1.24091971e+00 -1.61449954e-01 1.00847840e-01
3.23488533e-01 2.38804589e-03 -1.47345901e-01 5.00278950e-01
-4.92408633e-01 -1.43700135e+00 3.54031116e-01 7.93170035e-01
1.19163299e+00 -8.96252751e-01 9.70839560e-01 4.74951178e-01
-8.72385442e-01 -2.65518218e-01 -5.31929076e-01 -7.41243541e-01
-1.50404885e-01 5.32190561e-01 -3.36487174e-01 1.10522494e-01
9.65408802e-01 9.64472294e-01 -5.08450091e-01 1.14772844e+00
-6.53243065e-02 7.42554843e-01 -2.17130646e-01 -5.89963078e-01
-3.42527419e-01 -4.71334010e-02 -8.65828395e-02 1.05704415e+00
6.19028628e-01 5.13501108e-01 -6.01368606e-01 6.97952867e-01
8.99464488e-02 -9.20633003e-02 -4.13312972e-01 3.96433584e-02
1.75890326e-01 1.07442653e+00 -5.17451525e-01 -4.42309916e-01
-3.33778203e-01 1.28291261e+00 4.04245287e-01 1.69162944e-01
-9.33541238e-01 -8.36069465e-01 6.86665654e-01 -7.91095942e-02
-2.77052462e-01 -1.37802422e-01 -5.58036029e-01 -1.64015973e+00
-1.43598899e-01 -6.58707500e-01 -1.49236143e-01 -9.49186683e-01
-8.74484062e-01 1.00339603e+00 -1.69566572e-01 -1.24756920e+00
1.01200722e-01 -9.69737470e-01 -7.76928544e-01 1.10535645e+00
-1.46075070e+00 -4.70541418e-01 -6.22223258e-01 1.00809157e+00
2.25949988e-01 -3.01690847e-01 1.39049101e+00 2.35172302e-01
-7.87101686e-01 3.95130932e-01 9.97410715e-03 8.56518000e-02
7.47062624e-01 -7.11929321e-01 1.41341433e-01 6.06199682e-01
-2.59725481e-01 8.62074792e-01 1.69387296e-01 -3.43333900e-01
-1.26801479e+00 -6.35255516e-01 8.84345472e-01 -9.50096995e-02
4.24378008e-01 -2.67657727e-01 -9.70777512e-01 5.04355848e-01
6.40050411e-01 -4.48842108e-01 9.33662355e-01 1.39940891e-03
2.29511812e-01 -4.61757541e-01 -1.15391803e+00 4.57955033e-01
1.06492102e+00 -4.59565490e-01 -8.15603375e-01 1.70055047e-01
1.64398208e-01 -1.65787548e-01 -1.06203425e+00 4.80992168e-01
6.58091784e-01 -1.07476604e+00 7.66480565e-01 -4.07789588e-01
4.75207776e-01 2.29049876e-01 2.31502548e-01 -1.72844350e+00
-7.12366283e-01 -3.05947721e-01 -3.04768115e-01 7.03626275e-01
2.23508015e-01 -1.20020211e+00 6.13188446e-01 9.75576639e-01
-1.44104525e-01 -9.87531960e-01 -8.82032812e-01 -5.03215790e-01
-1.89547643e-01 -4.45230573e-01 1.00177145e+00 6.60931230e-01
1.03737819e+00 2.66014367e-01 -2.27693349e-01 -1.24363497e-01
9.79319885e-02 -1.23503648e-01 1.29902363e-01 -1.52838373e+00
6.32366017e-02 -4.77713257e-01 -9.00421143e-01 -1.02258837e+00
4.05624807e-01 -1.05802417e+00 -2.68380076e-01 -1.72725916e+00
3.32095891e-01 -1.35348141e-01 -1.08003068e+00 7.62874007e-01
6.98737204e-02 3.77395272e-01 -2.73938268e-01 4.12722081e-02
-2.82335460e-01 5.21501541e-01 1.15471649e+00 -1.51852101e-01
-1.97411045e-01 -1.18738629e-01 -7.90884495e-01 4.79669899e-01
1.37915981e+00 -1.74695820e-01 -4.11837399e-01 -3.85860056e-01
1.50209531e-01 1.72150686e-01 2.11684898e-01 -1.84760022e+00
5.03135979e-01 2.70967215e-01 1.12743425e+00 -2.65910983e-01
5.68104863e-01 -5.47087908e-01 2.51129359e-01 6.98397458e-01
-2.55243003e-01 -2.38184392e-01 3.84627223e-01 1.62925988e-01
-1.86388835e-01 5.95606640e-02 5.98780870e-01 -4.71166745e-02
-9.70134556e-01 3.46259326e-01 -8.75885427e-01 -7.09111989e-01
1.06860900e+00 -4.75360096e-01 -4.01087433e-01 -3.23579371e-01
-6.96749389e-01 -2.36628264e-01 -1.33330911e-01 3.13840717e-01
9.82703209e-01 -1.20518911e+00 -4.57135856e-01 7.73380995e-01
-3.46565992e-02 -6.34860039e-01 3.03166807e-01 1.06557453e+00
-4.03570294e-01 6.26993954e-01 -1.02287257e+00 -3.30364347e-01
-8.97975743e-01 2.72391677e-01 4.89489377e-01 9.88498703e-02
-7.09967494e-01 7.39472806e-01 -2.22897291e-01 -5.67155704e-02
2.54983932e-01 -3.34229439e-01 -9.05918777e-01 -8.94666929e-03
8.65502298e-01 2.74793148e-01 6.09837592e-01 -2.63686746e-01
-4.67459381e-01 1.07091203e-01 1.21522270e-01 9.02153179e-02
1.71719825e+00 2.15113655e-01 -4.07244384e-01 2.76579320e-01
1.11946583e+00 -7.69928992e-01 -9.67711747e-01 2.16127083e-01
-3.74818146e-01 -3.98515202e-02 2.83123761e-01 -1.00664437e+00
-1.18644249e+00 1.20994818e+00 1.08837175e+00 1.47651553e-01
1.35422111e+00 -4.17685091e-01 9.61499929e-01 7.12855399e-01
9.01006341e-01 -1.07305920e+00 -1.31773679e-02 5.58698177e-01
7.51408219e-01 -1.13524354e+00 -3.90249997e-01 1.10955089e-01
-4.57521707e-01 1.52175891e+00 7.77822673e-01 -2.19921827e-01
9.46621776e-01 3.28015625e-01 -1.96357295e-01 -1.86967909e-01
-6.63310051e-01 1.11535393e-01 2.13618428e-01 6.10203803e-01
5.91418684e-01 1.38159856e-01 -3.72899234e-01 1.20810187e+00
-4.77706283e-01 4.89939868e-01 4.26980585e-01 7.40034819e-01
-5.61472416e-01 -7.14544952e-01 -1.89964682e-01 1.19512630e+00
-5.48226774e-01 -2.91615516e-01 6.68963864e-02 1.55937910e-01
-2.21058112e-02 1.19015682e+00 1.10358723e-01 -1.08801591e+00
2.75424898e-01 5.54973722e-01 6.73278213e-01 -4.84612763e-01
-6.49999380e-01 -4.59523469e-01 -2.03712001e-01 -7.43600845e-01
-3.87930721e-01 -7.36258999e-02 -1.08134770e+00 -3.95173967e-01
-2.06164137e-01 -1.18947707e-01 8.10229838e-01 1.08395398e+00
6.58061087e-01 7.51217484e-01 3.48666012e-01 -9.94823933e-01
-2.05724776e-01 -1.45868933e+00 -7.89766371e-01 3.18339258e-01
4.73319590e-02 -7.14151919e-01 -1.51320294e-01 -1.12969458e-01] | [13.104564666748047, 3.4445858001708984] |
bd5963dc-ad31-4904-bd5d-c6559f0a5622 | transparency-strategy-based-data-augmentation | 2203.10609 | null | https://arxiv.org/abs/2203.10609v2 | https://arxiv.org/pdf/2203.10609v2.pdf | A Novel Transparency Strategy-based Data Augmentation Approach for BI-RADS Classification of Mammograms | Image augmentation techniques have been widely investigated to improve the performance of deep learning (DL) algorithms on mammography classification tasks. Recent methods have proved the efficiency of image augmentation on data deficiency or data imbalance issues. In this paper, we propose a novel transparency strategy to boost the Breast Imaging Reporting and Data System (BI-RADS) scores of mammogram classifiers. The proposed approach utilizes the Region of Interest (ROI) information to generate more high-risk training examples for breast cancer (BI-RADS 3, 4, 5) from original images. Our extensive experiments on three different datasets show that the proposed approach significantly improves the mammogram classification performance and surpasses a state-of-the-art data augmentation technique called CutMix. This study also highlights that our transparency method is more effective than other augmentation strategies for BI-RADS classification and can be widely applied to other computer vision tasks. | ['Ha Q. Nguyen', 'Chi Phan', 'Hieu H. Pham', 'Huyen T. X. Nguyen', 'Sam B. Tran'] | 2022-03-20 | null | null | null | null | ['image-augmentation'] | ['computer-vision'] | [ 5.05196869e-01 6.04854167e-01 -4.65574473e-01 -7.02304780e-01
-7.84817278e-01 2.29278192e-01 5.43594599e-01 3.08403313e-01
-3.27059925e-01 6.77692533e-01 1.83326676e-01 -7.36077487e-01
5.37095144e-02 -8.50578666e-01 -7.47409940e-01 -8.42395842e-01
1.87634110e-01 2.74447203e-01 4.33959290e-02 2.95038149e-02
-7.42015541e-02 5.40749729e-01 -1.20606923e+00 7.22848117e-01
6.54376149e-01 1.19840765e+00 -2.40374669e-01 5.53658485e-01
-1.07617602e-01 9.47389483e-01 -4.61248338e-01 -6.48673117e-01
3.21625233e-01 -4.67896342e-01 -6.01469934e-01 -1.44981071e-02
4.48796213e-01 -5.35746872e-01 -3.79704893e-01 1.00310183e+00
6.47401392e-01 -6.01644635e-01 6.60175323e-01 -1.14014935e+00
-7.80880094e-01 5.44777453e-01 -1.27652800e+00 6.37237370e-01
-4.26378280e-01 -1.93965822e-01 2.86994010e-01 -7.02251554e-01
3.35027546e-01 9.59926605e-01 8.63372326e-01 6.59705520e-01
-8.08115900e-01 -1.02796435e+00 -2.36154631e-01 6.55633807e-02
-8.77412498e-01 -3.43950123e-01 6.85239375e-01 -3.55638921e-01
3.70476246e-01 6.25518382e-01 6.33364618e-01 8.52321148e-01
3.84221107e-01 8.61698031e-01 1.25132084e+00 -6.05020523e-01
1.59559045e-02 2.93938309e-01 3.97475660e-01 8.70364010e-01
8.25095654e-01 2.09309027e-01 -3.15823495e-01 -1.84852600e-01
8.54710519e-01 9.77249667e-02 2.76012570e-02 -3.60943496e-01
-1.03529215e+00 9.23577666e-01 8.98840487e-01 3.51261407e-01
-4.52073902e-01 -1.99340582e-02 5.73949575e-01 -6.12392388e-02
6.70822501e-01 1.83739558e-01 -1.79876521e-01 5.45054853e-01
-8.01544189e-01 9.16663557e-02 1.43792048e-01 4.82728481e-01
6.93710074e-02 -1.34036383e-02 -5.30045152e-01 7.34028876e-01
1.77735835e-01 3.80344242e-01 7.01575160e-01 -4.35626149e-01
4.73390728e-01 9.31631327e-01 -1.02978975e-01 -9.17015612e-01
-7.58366346e-01 -6.29736483e-01 -1.49731719e+00 2.50756085e-01
1.65465966e-01 8.24271068e-02 -1.34050250e+00 1.27669048e+00
4.25155967e-01 -2.85218030e-01 2.33176813e-01 7.22341776e-01
1.40900767e+00 3.52694988e-01 3.83610100e-01 -1.59286410e-01
1.67358327e+00 -7.04083920e-01 -1.04095101e+00 -1.81709886e-01
1.03377926e+00 -4.48838413e-01 8.92543316e-01 2.52207965e-01
-1.12048960e+00 -6.97892904e-01 -1.29531693e+00 9.30606574e-02
-1.92256361e-01 6.98857427e-01 1.15308952e+00 1.07453561e+00
-5.43021917e-01 1.16233870e-01 -1.10511374e+00 -1.85184251e-03
1.39768589e+00 3.89781922e-01 -4.84349757e-01 -1.58435002e-01
-8.71829569e-01 7.26940155e-01 4.06911552e-01 -3.84281650e-02
-7.05628872e-01 -9.15265858e-01 -8.90585959e-01 -2.88046867e-01
2.09453091e-01 -5.36972344e-01 1.10534668e+00 -1.08363247e+00
-8.97432506e-01 1.33349323e+00 1.95349634e-01 -7.08368778e-01
5.14991999e-01 -1.04603574e-01 -2.76249081e-01 2.41238803e-01
-3.74106551e-03 8.86151969e-01 5.47604799e-01 -1.15127397e+00
-5.45749247e-01 -6.95815563e-01 -2.81371862e-01 3.80712748e-02
-7.39624381e-01 -3.70665900e-02 -2.79218674e-01 -7.76316106e-01
4.29680347e-01 -6.32430613e-01 -3.76721501e-01 2.46091932e-01
-3.68772268e-01 1.91388223e-02 6.08219564e-01 -7.67234325e-01
1.01596344e+00 -1.93880522e+00 -4.18319523e-01 -6.36635199e-02
2.46747226e-01 6.61922872e-01 6.82597086e-02 -4.96538162e-01
-4.29829806e-01 3.76959413e-01 -2.87472248e-01 1.77860651e-02
-5.62854171e-01 -1.46008991e-02 2.53000200e-01 5.13752222e-01
5.71762741e-01 1.19001031e+00 -4.12052214e-01 -7.05685496e-01
2.01914206e-01 3.79389316e-01 -4.30706501e-01 -1.05673559e-02
2.28177905e-01 4.88523960e-01 -3.07191074e-01 8.36527109e-01
1.07528877e+00 -4.26243097e-01 -1.06015038e-02 -6.12431049e-01
4.08264250e-01 -7.60600269e-02 -7.55771220e-01 1.24727452e+00
-2.00243443e-01 4.65812355e-01 -3.32378030e-01 -1.20029783e+00
9.06048834e-01 1.98898837e-01 6.38784826e-01 -1.19944048e+00
2.32761428e-01 -1.12711444e-01 3.73289257e-01 -7.77930915e-01
3.33992332e-01 -1.48519292e-01 2.89272755e-01 3.30256164e-01
-2.06270546e-01 2.28278443e-01 -5.61403930e-02 1.66807801e-01
9.37323987e-01 -1.05311811e-01 5.70974588e-01 -1.31451860e-01
3.93189371e-01 9.56766531e-02 5.29122472e-01 7.17176676e-01
-2.59673834e-01 4.61626917e-01 6.20972574e-01 -6.25870049e-01
-1.27703619e+00 -6.28918648e-01 -5.36810577e-01 5.99875927e-01
-1.41023099e-01 9.40596014e-02 -6.70037448e-01 -1.04552484e+00
1.03513621e-01 4.11438912e-01 -1.07811797e+00 -3.06970090e-01
-5.90355754e-01 -1.69300616e+00 6.26965165e-01 1.00930953e+00
9.07300889e-01 -8.87126923e-01 -7.44838834e-01 -1.23261459e-01
-1.35346889e-01 -1.05694592e+00 1.66018397e-01 9.51567888e-02
-1.26519120e+00 -1.16122115e+00 -9.91264999e-01 -7.29211986e-01
1.23602951e+00 3.09673846e-01 9.47731733e-01 2.09363997e-01
-7.28220046e-01 -5.04101589e-02 -3.98978382e-01 -9.29447234e-01
-8.80146444e-01 -1.60856277e-01 -4.05133635e-01 -1.05262503e-01
4.36034828e-01 1.65567026e-01 -6.80734754e-01 2.45725259e-01
-1.22892904e+00 4.99290854e-01 1.04108381e+00 9.59747791e-01
7.51339793e-01 4.67350669e-02 8.02579701e-01 -1.38377285e+00
3.00588012e-01 -3.69118780e-01 -4.62409347e-01 1.66721910e-01
-6.70100331e-01 -1.48309425e-01 2.22405735e-02 -3.11727077e-01
-1.24830449e+00 4.32661958e-02 -1.10613834e-02 1.40179366e-01
-6.91504925e-02 4.69149709e-01 -4.71438505e-02 -3.05475265e-01
8.44576001e-01 -1.50209919e-01 9.18370858e-02 -3.42990249e-01
-7.75926486e-02 8.16004813e-01 6.37649477e-01 1.18197180e-01
4.02686179e-01 7.99048126e-01 3.50790888e-01 -4.03669238e-01
-9.34138834e-01 -2.60306478e-01 -6.77934110e-01 2.31903028e-02
8.03666115e-01 -9.63273525e-01 -3.91389191e-01 4.29872870e-01
-7.07452118e-01 2.54372712e-02 -2.84227014e-01 5.58587372e-01
4.04587090e-02 1.82525471e-01 -5.47368526e-01 -7.54718065e-01
-6.86709166e-01 -1.05944037e+00 9.63973224e-01 4.39154029e-01
2.03639388e-01 -5.22513866e-01 -3.85989100e-01 7.11643636e-01
3.27405781e-01 4.60821629e-01 9.88246918e-01 -6.91984713e-01
-2.07046062e-01 -5.86137950e-01 -7.93356121e-01 3.81804943e-01
2.31442481e-01 -1.86408177e-01 -1.02304268e+00 -2.01540440e-01
1.53368518e-01 -3.90309125e-01 1.12302816e+00 7.95837760e-01
1.84621859e+00 -2.08138973e-01 -5.82262218e-01 7.06039369e-01
1.24466228e+00 3.84832025e-01 9.45070148e-01 4.44854438e-01
5.27927816e-01 4.68650728e-01 8.93561542e-01 3.20627868e-01
2.15985626e-01 3.39671880e-01 6.22268498e-01 -1.03755581e+00
-3.29229534e-01 1.57122210e-01 -2.77675003e-01 1.08053684e-01
-1.04822993e-01 6.06197864e-02 -1.02033317e+00 4.04640228e-01
-1.45769417e+00 -5.29159248e-01 -4.58749115e-01 1.98600328e+00
8.98027897e-01 3.12531054e-01 -1.61686808e-01 4.33981150e-01
5.71619868e-01 -1.61393851e-01 -5.87813079e-01 -2.53904909e-01
2.13108566e-02 3.80826652e-01 8.59188199e-01 -1.51326209e-01
-1.50868332e+00 4.31958944e-01 6.26376963e+00 4.66196120e-01
-9.53496635e-01 3.72233361e-01 1.58696198e+00 1.30738571e-01
5.40018082e-03 -6.85396969e-01 -7.75409997e-01 1.89290941e-01
7.45951474e-01 2.43491665e-01 -6.71943188e-01 7.68430114e-01
4.27379869e-02 -5.23255289e-01 -1.17516351e+00 8.99707615e-01
2.22346604e-01 -1.68651175e+00 1.57513276e-01 8.22696239e-02
7.73586214e-01 -3.49131107e-01 1.92696273e-01 1.56167224e-01
-3.73293310e-02 -1.08840787e+00 4.17914167e-02 3.11772734e-01
1.05284297e+00 -6.76264286e-01 1.36884797e+00 -2.75099347e-03
-4.82112944e-01 1.20803453e-02 -3.35775346e-01 3.54326606e-01
-3.60609263e-01 6.41885817e-01 -1.25927734e+00 6.86718822e-01
8.86776149e-01 2.57077128e-01 -1.17899883e+00 9.93888915e-01
3.00221741e-01 7.68366039e-01 -9.82124452e-03 1.35254681e-01
1.01227343e-01 4.49220061e-01 -1.34656772e-01 1.10891271e+00
3.39246839e-02 3.89721096e-01 -2.55059630e-01 6.86136484e-01
-3.26600164e-01 1.64028451e-01 -4.74977136e-01 2.48111710e-02
1.14994451e-01 1.40343308e+00 -8.96259964e-01 -6.32161438e-01
-3.69635820e-01 4.85701770e-01 -2.16674089e-01 -2.07874194e-01
-8.88036489e-01 -5.36403544e-02 -2.00878233e-01 3.11108291e-01
1.31185269e-02 4.33727950e-01 -7.08600104e-01 -5.99432766e-01
1.12174459e-01 -9.08102810e-01 5.55632710e-01 -7.17761576e-01
-1.03070831e+00 5.20890057e-01 7.30401948e-02 -1.16311610e+00
1.95717096e-01 -5.77803671e-01 -2.99284995e-01 4.76987898e-01
-1.93983150e+00 -1.39971387e+00 -7.59479523e-01 3.36227179e-01
2.65058756e-01 -5.32938421e-01 8.63296092e-01 5.64958692e-01
-6.15260959e-01 9.51872766e-01 -1.10540107e-01 4.27803576e-01
5.08326292e-01 -9.47616518e-01 3.02602619e-01 7.16078103e-01
-3.39785039e-01 4.20644926e-03 2.03861639e-01 -7.82695353e-01
-1.10451090e+00 -1.34420669e+00 3.47734571e-01 -9.43947881e-02
8.07510614e-02 8.05242732e-02 -7.84969211e-01 5.80300033e-01
-9.03165154e-03 3.78553808e-01 1.00426090e+00 -3.43909472e-01
-4.30154726e-02 -4.36260104e-01 -1.54871690e+00 3.20134312e-01
5.60637951e-01 2.57849157e-01 -2.17103571e-01 4.29503560e-01
3.93522233e-01 -6.24200702e-01 -7.07767189e-01 1.11268723e+00
4.00937408e-01 -8.96451592e-01 9.38740492e-01 -7.86253214e-01
8.77212763e-01 1.70536920e-01 -8.51541851e-03 -7.13874459e-01
-2.16179743e-01 -1.61302704e-02 6.46407437e-03 1.05102181e+00
2.31999546e-01 -3.91087353e-01 1.09317708e+00 5.36557257e-01
-6.19068891e-02 -9.15686607e-01 -9.41834867e-01 -1.79887548e-01
1.12645417e-01 -2.04157710e-01 5.06636977e-01 1.08544350e+00
-5.28457582e-01 -1.94478184e-01 -1.20301768e-01 2.57432848e-01
6.58206880e-01 -1.86904818e-01 6.73832238e-01 -1.14803541e+00
2.43432424e-03 -3.01834852e-01 -5.86118937e-01 -2.47183487e-01
-4.66730535e-01 -1.00820863e+00 -4.00401413e-01 -1.52198398e+00
7.37128854e-01 -7.28125453e-01 -5.00294447e-01 7.87985981e-01
-4.44560260e-01 8.78429592e-01 -6.12803102e-02 2.86470409e-02
-2.84008831e-01 2.30825514e-01 1.64634645e+00 -3.61905694e-01
-6.50540367e-02 2.44100600e-01 -1.12720752e+00 8.45921993e-01
8.86982679e-01 -6.79309428e-01 -2.69774385e-02 -3.10336441e-01
8.91057923e-02 7.98483193e-02 6.20121419e-01 -1.01445079e+00
-1.58154696e-01 8.91585052e-02 1.15618432e+00 -9.88143146e-01
2.31218021e-02 -7.43098140e-01 -1.49095222e-01 1.02622283e+00
-5.64131618e-01 -2.52130866e-01 5.13700724e-01 4.08562332e-01
-4.09383439e-02 -2.46664986e-01 1.11212599e+00 -2.86895365e-01
-3.16956103e-01 1.75208062e-01 -8.22490007e-02 -4.12766904e-01
1.28929961e+00 -2.40750730e-01 -2.63208032e-01 9.78083313e-02
-5.08173048e-01 -3.38178058e-03 -2.18469337e-01 3.71799529e-01
7.77906954e-01 -1.65684319e+00 -1.13667524e+00 4.92451251e-01
3.12506467e-01 1.13744766e-01 1.48086026e-01 1.04839087e+00
-5.51683366e-01 4.70665216e-01 -4.92569953e-01 -8.69552851e-01
-1.75158763e+00 2.78399259e-01 4.22318697e-01 -6.50314629e-01
-7.33175218e-01 9.72439647e-01 4.85881835e-01 -6.67028502e-02
5.06802797e-01 -4.74707514e-01 -4.40021694e-01 -8.20968375e-02
7.80735075e-01 1.95332691e-01 4.73407060e-01 -3.42902809e-01
-2.42149472e-01 5.33159636e-02 -6.41860485e-01 3.37102175e-01
1.53501630e+00 2.88787305e-01 9.32080597e-02 3.37675656e-03
9.73344684e-01 -4.27581042e-01 -8.25266898e-01 -3.74987721e-01
-1.55912131e-01 -4.73083675e-01 4.19111043e-01 -1.13176036e+00
-1.50965548e+00 8.85297596e-01 1.40968871e+00 1.11508898e-01
1.31252563e+00 -3.15982290e-02 5.20230114e-01 1.59924835e-01
-1.83196709e-01 -8.83333325e-01 3.52288723e-01 -4.18055147e-01
9.37776446e-01 -1.91017103e+00 4.39699113e-01 -3.49252045e-01
-9.11921740e-01 1.09002328e+00 8.27993989e-01 7.71904290e-02
5.74519992e-01 5.21203041e-01 6.09104000e-02 -3.11595708e-01
-2.99598932e-01 1.27882734e-01 3.80901396e-01 7.41324604e-01
7.19462693e-01 1.25357613e-01 -5.20698011e-01 8.83444190e-01
1.46781981e-01 3.85686725e-01 5.24278879e-01 9.24190819e-01
-3.02045524e-01 -8.38229954e-01 -6.19380713e-01 9.68818128e-01
-1.00047088e+00 5.58231324e-02 -3.19959998e-01 1.07938957e+00
3.36179882e-01 5.34883797e-01 1.30680874e-01 -8.97973627e-02
2.14392439e-01 -1.84963897e-01 6.07755780e-01 -4.46334600e-01
-5.58974624e-01 1.51155248e-01 -5.22263758e-02 -2.89845139e-01
-7.33932555e-01 -4.71353590e-01 -1.26459873e+00 9.08830762e-02
-5.94910443e-01 -2.88064420e-01 9.91463363e-01 6.26916885e-01
1.58940479e-01 1.24283314e+00 5.35144389e-01 -5.71731329e-01
-4.00240362e-01 -1.09231234e+00 -1.87290072e-01 4.53549385e-01
3.57664198e-01 -6.18061781e-01 2.91097432e-01 2.18894914e-01] | [15.1694917678833, -2.3900599479675293] |
984a8a57-eeae-4942-97e4-49ef15b306d8 | hifi-a-unified-framework-for-neural-vocoding | 2203.13086 | null | https://arxiv.org/abs/2203.13086v2 | https://arxiv.org/pdf/2203.13086v2.pdf | HiFi++: a Unified Framework for Bandwidth Extension and Speech Enhancement | Generative adversarial networks have recently demonstrated outstanding performance in neural vocoding outperforming best autoregressive and flow-based models. In this paper, we show that this success can be extended to other tasks of conditional audio generation. In particular, building upon HiFi vocoders, we propose a novel HiFi++ general framework for bandwidth extension and speech enhancement. We show that with the improved generator architecture and simplified multi-discriminator training, HiFi++ performs better or on par with the state-of-the-art in these tasks while spending significantly less computational resources. The effectiveness of our approach is validated through a series of extensive experiments. | ['Dmitry Vetrov', 'Oleg Ivanov', 'Aibek Alanov', 'Pavel Andreev'] | 2022-03-24 | null | null | null | null | ['bandwidth-extension', 'audio-generation', 'bandwidth-extension'] | ['audio', 'audio', 'speech'] | [ 2.91929960e-01 1.63464189e-01 1.06588118e-01 4.60756905e-02
-8.94375563e-01 -3.14255178e-01 7.22386599e-01 -6.27064943e-01
-1.18999556e-01 1.11652362e+00 5.79844296e-01 -2.21322402e-01
4.17160057e-02 -7.04591095e-01 -7.78544188e-01 -6.28959239e-01
-2.50719607e-01 8.77739191e-02 1.64931953e-01 -3.53896737e-01
-2.69394100e-01 3.81751269e-01 -1.40704119e+00 2.06843957e-01
6.10821128e-01 9.03941095e-01 -1.26496434e-01 1.12171090e+00
3.47805351e-01 1.07131076e+00 -1.03086281e+00 -5.03747284e-01
1.06537215e-01 -6.59494281e-01 -7.39765286e-01 -4.35310453e-01
3.51690918e-01 -5.01115263e-01 -9.98045027e-01 5.48046112e-01
1.09401298e+00 5.21318436e-01 7.37640142e-01 -1.21754313e+00
-7.12244987e-01 7.25752592e-01 1.58135086e-01 2.86769271e-01
3.17322582e-01 7.24278465e-02 8.54153335e-01 -9.66376543e-01
4.92668748e-01 1.28550172e+00 7.36370325e-01 8.58482301e-01
-1.16187894e+00 -7.41602421e-01 -7.69259930e-02 1.36757195e-01
-1.35608649e+00 -7.69416094e-01 9.68057871e-01 -1.69845641e-01
1.08256614e+00 3.03534120e-01 3.81405264e-01 1.49008548e+00
-4.64734947e-03 8.55733991e-01 6.89917088e-01 -4.25583661e-01
1.39101565e-01 -2.89631821e-02 -4.91085500e-01 5.27584732e-01
-3.96828860e-01 6.30819023e-01 -7.45100737e-01 1.02448300e-01
7.60396004e-01 -5.93638897e-01 -5.98813653e-01 1.04774952e-01
-1.01960278e+00 1.00568783e+00 3.75903726e-01 3.40994596e-01
-2.63357818e-01 7.82785356e-01 4.81016487e-01 2.99118668e-01
6.00590587e-01 4.76441056e-01 -5.89365792e-03 -2.98438400e-01
-1.16862535e+00 4.16892201e-01 7.83669710e-01 1.04963815e+00
1.82048157e-01 1.21632206e+00 -5.20189583e-01 8.76388907e-01
2.27425899e-02 4.46387947e-01 6.52029693e-01 -1.18593848e+00
3.28124762e-01 -5.73188305e-01 -2.24736273e-01 -5.37243485e-01
-1.02160044e-01 -7.72085488e-01 -1.10267889e+00 3.19860280e-01
1.14255697e-01 -5.25805295e-01 -9.06855762e-01 1.90292013e+00
-5.54011017e-02 6.85151517e-01 2.30827183e-01 5.98759651e-01
8.64776671e-01 9.56566572e-01 1.39581472e-01 -1.47570357e-01
7.52427280e-01 -1.00724638e+00 -8.80962908e-01 7.55118355e-02
-2.15215757e-02 -8.06942761e-01 7.52771378e-01 5.42695045e-01
-1.44443166e+00 -1.04760349e+00 -1.09081519e+00 -2.41435897e-02
-2.13336229e-01 -9.94539335e-02 9.36547875e-01 1.04577291e+00
-1.41566002e+00 8.30212593e-01 -5.20602286e-01 1.61047667e-01
4.26323682e-01 2.30214581e-01 -1.02701657e-01 2.10429654e-01
-1.63854194e+00 5.27744472e-01 3.00835669e-01 -1.01558596e-01
-1.40460384e+00 -1.10297656e+00 -1.02022600e+00 2.28263512e-01
-4.60987613e-02 -8.94220948e-01 1.51030970e+00 -6.15691185e-01
-2.04405499e+00 3.77954729e-02 -1.63907129e-02 -1.01783490e+00
5.16842842e-01 -4.11772519e-01 -6.48442507e-01 3.50905210e-01
-4.13018376e-01 9.46113527e-01 1.08173990e+00 -1.16681910e+00
-4.07719672e-01 3.38912040e-01 5.17427083e-03 3.48095410e-03
-4.75365102e-01 1.13014176e-01 -1.01395957e-01 -1.05959058e+00
-6.00213110e-01 -7.93069839e-01 -1.72299176e-01 -1.85597077e-01
-3.68526101e-01 -4.18584310e-02 9.81881857e-01 -6.52753294e-01
1.31130862e+00 -2.00185657e+00 2.57602423e-01 -1.56614363e-01
1.07642636e-01 5.43607295e-01 -2.47924924e-01 2.95832723e-01
-2.03413293e-01 3.51262927e-01 -1.13383390e-01 -5.41235209e-01
1.48975343e-01 1.85892224e-01 -8.16736519e-01 1.48131609e-01
4.14960146e-01 9.29516792e-01 -7.01712251e-01 -3.44804764e-01
4.66098815e-01 1.06784582e+00 -9.97086823e-01 3.75193059e-01
7.42251053e-02 5.33201754e-01 6.26216084e-02 3.78130436e-01
3.50011975e-01 3.27098280e-01 -3.23180377e-01 -2.71482859e-03
1.31383464e-01 4.12772894e-01 -1.01316071e+00 1.76415205e+00
-8.85981321e-01 1.07609010e+00 9.56745148e-02 -1.14515555e+00
8.63315523e-01 9.20768321e-01 3.81710708e-01 -3.49168241e-01
-5.22119366e-02 1.19930588e-01 -6.14413619e-02 -2.39528760e-01
5.60706794e-01 -4.40782726e-01 -4.12778705e-02 -2.51181219e-02
8.04050088e-01 -5.94041348e-01 1.40145600e-01 1.91210032e-01
8.95986974e-01 2.07099333e-01 3.42245847e-02 -5.10931090e-02
5.83460510e-01 -5.77499032e-01 2.70953208e-01 9.63281631e-01
-2.28825063e-01 6.92500114e-01 2.64214456e-01 1.42212257e-01
-1.16323209e+00 -1.29942393e+00 -1.07301027e-01 1.14430416e+00
-3.00315499e-01 -2.10411012e-01 -9.28201377e-01 -2.89286673e-01
-1.76922828e-01 9.31721509e-01 -5.78888535e-01 -1.83453798e-01
-7.52537131e-01 -5.48868537e-01 1.15620065e+00 8.54710221e-01
3.78006667e-01 -1.28765917e+00 -3.49984914e-01 4.85561460e-01
-7.73819338e-04 -1.10816288e+00 -4.76461560e-01 2.39792302e-01
-6.18533552e-01 -2.31488496e-01 -1.24486113e+00 -8.34884107e-01
-1.64975151e-01 -1.56424835e-01 1.20292652e+00 -3.12735826e-01
-3.04450125e-01 4.30920303e-01 -3.14413518e-01 -6.26708448e-01
-7.14749515e-01 2.66885877e-01 -3.01167015e-02 -2.42228344e-01
-4.79179382e-01 -9.24449563e-01 -4.91155833e-01 9.66187660e-03
-1.03769124e+00 -2.51910657e-01 1.59168035e-01 1.12611055e+00
3.67621362e-01 1.76325455e-01 1.12804008e+00 -6.47261262e-01
8.85346413e-01 -4.35528427e-01 -1.74110770e-01 -2.60725170e-01
-3.80998731e-01 9.79651436e-02 1.08217561e+00 -5.23856640e-01
-1.30220234e+00 -1.97606459e-01 -8.26298833e-01 -6.32349193e-01
-1.10838056e-01 1.96099043e-01 -1.81068897e-01 -1.68098599e-01
6.59102678e-01 2.10473493e-01 -2.99370080e-01 -2.96289980e-01
5.65278947e-01 5.70623159e-01 1.10710132e+00 -4.21112716e-01
8.62183452e-01 2.01801538e-01 8.99975225e-02 -7.70019174e-01
-4.60236222e-01 -1.11334503e-01 -2.81343311e-01 -3.01091164e-01
8.21670592e-01 -9.82479393e-01 -5.36857665e-01 3.59349370e-01
-1.10149550e+00 -5.65439820e-01 -6.17507815e-01 5.10445893e-01
-9.48896468e-01 1.98689729e-01 -8.38636637e-01 -9.84793186e-01
-4.94612902e-01 -1.10133529e+00 8.62275362e-01 1.88250214e-01
-1.80207547e-02 -1.14585435e+00 2.81928033e-01 8.15902054e-02
8.54183197e-01 2.45709598e-01 5.21232963e-01 -4.53862369e-01
-3.25223029e-01 -9.68567953e-02 2.03923896e-01 7.12028027e-01
-3.62461716e-01 3.57722379e-02 -1.36444592e+00 -1.75398022e-01
-1.43623650e-01 -3.22335988e-01 1.21906173e+00 4.63038087e-01
1.32344782e+00 -1.94474146e-01 8.83849487e-02 1.07987714e+00
1.27787519e+00 4.44710523e-01 8.84001076e-01 -1.73969597e-01
4.91067082e-01 2.27255374e-01 1.40064150e-01 5.38467109e-01
-2.52419144e-01 7.63181567e-01 4.30573463e-01 -2.70185739e-01
-7.64013886e-01 -4.22904521e-01 3.99822861e-01 8.37875128e-01
-2.60979146e-01 -7.06393182e-01 -4.67687488e-01 6.78160429e-01
-1.39150894e+00 -1.40879703e+00 1.82123929e-01 1.85743237e+00
7.35538661e-01 4.64421101e-02 2.16718316e-01 6.71813428e-01
4.96954650e-01 3.73711973e-01 -2.09454492e-01 -7.80055761e-01
-2.40684196e-01 1.19822896e+00 2.93466687e-01 7.16759205e-01
-1.27396810e+00 8.78497064e-01 8.00522137e+00 9.49464381e-01
-9.29554403e-01 2.07812473e-01 4.60476488e-01 -1.63722172e-01
-3.67785215e-01 -4.55803663e-01 -5.30347288e-01 4.00465041e-01
1.33892763e+00 -2.84547478e-01 7.09067345e-01 7.61938691e-01
-3.32221910e-02 5.39861798e-01 -8.88973653e-01 7.81901419e-01
2.01820567e-01 -1.40639448e+00 8.71435925e-02 -1.56758815e-01
6.96279764e-01 -1.38898075e-01 4.45210069e-01 6.47673070e-01
2.96258003e-01 -1.30714965e+00 8.05433571e-01 3.46149415e-01
1.22377193e+00 -1.09325445e+00 6.03908062e-01 -7.41551593e-02
-1.34410858e+00 -1.05426937e-01 -2.75861889e-01 2.86041945e-02
5.25141656e-01 3.22061718e-01 -7.40517914e-01 6.41894221e-01
3.32226127e-01 3.82403135e-01 -1.44778267e-01 1.02034211e+00
-4.94327217e-01 9.94671643e-01 2.72048209e-02 5.58820404e-02
2.25333914e-01 3.98309141e-01 6.96209192e-01 1.77627683e+00
5.99628568e-01 -4.95828055e-02 -1.17893070e-01 7.04249918e-01
-2.44130224e-01 -5.51857911e-02 -7.13732660e-01 1.72000453e-01
5.26351631e-01 8.81797612e-01 6.75974414e-02 -5.02510548e-01
-3.01759392e-01 1.00375855e+00 1.08474895e-01 5.30311465e-01
-1.23387480e+00 -8.79571021e-01 7.33866692e-01 -1.72520593e-01
5.09353697e-01 -1.18411407e-01 6.14929304e-04 -1.00549591e+00
-4.74365920e-01 -6.99340582e-01 1.81550562e-01 -9.52481151e-01
-8.92736256e-01 9.01525259e-01 -6.94806799e-02 -1.01365340e+00
-9.31288838e-01 -4.89713699e-01 -6.87900901e-01 9.37900424e-01
-1.62382841e+00 -8.92764270e-01 -2.46231660e-01 7.59229124e-01
5.71251690e-01 -4.00802493e-01 1.02772999e+00 4.61981237e-01
-4.75276142e-01 1.02966285e+00 8.95738378e-02 9.19025093e-02
5.86511254e-01 -1.29940236e+00 6.38053238e-01 1.12123942e+00
4.68599200e-01 1.78905025e-01 8.47576261e-01 -1.18156731e-01
-1.04678595e+00 -1.25618815e+00 6.27394795e-01 1.37063982e-02
5.85869372e-01 -3.34934622e-01 -7.34075308e-01 5.81142306e-01
6.31671607e-01 1.65247824e-03 6.70445919e-01 -1.94436952e-01
-3.31060559e-01 -6.56704679e-02 -1.09995234e+00 4.54531908e-01
1.16149688e+00 -5.58295906e-01 -3.95520329e-01 1.33373380e-01
1.10017419e+00 -4.35964555e-01 -1.10468769e+00 4.22364265e-01
5.05220294e-01 -9.77430880e-01 1.38961983e+00 -6.57532215e-01
6.33241177e-01 7.42969215e-02 -2.87527442e-01 -1.61482906e+00
-3.74045104e-01 -1.24757528e+00 -4.11630124e-01 1.36853206e+00
3.65678459e-01 -3.86310577e-01 6.30673945e-01 1.04700518e-03
-7.14016259e-01 -5.39798200e-01 -1.24085987e+00 -9.62807357e-01
4.38423872e-01 -7.85074592e-01 5.06235421e-01 4.16421384e-01
-6.14009872e-02 2.77716190e-01 -9.64558780e-01 -6.63750842e-02
4.17653322e-01 -3.43698084e-01 7.42771745e-01 -8.66362512e-01
-7.85789609e-01 -4.55018789e-01 -4.47677791e-01 -1.04813349e+00
4.03368324e-01 -8.71500850e-01 2.23667100e-01 -1.23810935e+00
-2.63429612e-01 -1.20879047e-01 -3.88268083e-01 2.89888352e-01
-2.21223116e-01 5.73707998e-01 5.00779569e-01 -2.59029061e-01
-1.41753331e-01 7.58779705e-01 1.25600612e+00 -2.07424834e-01
-2.47741640e-01 6.82439608e-03 -5.69780529e-01 3.75711709e-01
9.38940167e-01 -1.84202865e-01 -6.03780329e-01 -2.72929668e-01
-4.70359057e-01 3.55913699e-01 4.63611573e-01 -1.67626572e+00
-1.30446464e-01 3.66172314e-01 3.68935674e-01 -1.27148226e-01
8.06143343e-01 -3.14437896e-01 1.59203336e-01 5.02078652e-01
-6.48213387e-01 -2.13947073e-01 3.35402787e-01 5.58900177e-01
-5.27420700e-01 -1.80665433e-01 9.08693433e-01 1.76686421e-01
-4.64890689e-01 2.60795414e-01 -5.54025173e-01 2.87844419e-01
6.65415347e-01 1.96976781e-01 -3.19449097e-01 -8.91184092e-01
-8.33566606e-01 -4.07311022e-01 -1.44116327e-01 3.18799198e-01
7.53768027e-01 -1.49867749e+00 -9.59917963e-01 3.10266197e-01
-3.89600962e-01 -4.47444230e-01 4.25994933e-01 4.61658269e-01
-3.18417490e-01 7.36187518e-01 -1.11683927e-01 -2.35718220e-01
-1.05192018e+00 6.69015408e-01 3.86164725e-01 -3.59216362e-01
-4.73865300e-01 1.08815682e+00 2.71177143e-01 1.30284175e-01
3.40703368e-01 -1.31187275e-01 -5.40903248e-02 -2.50940293e-01
5.11756122e-01 5.45926154e-01 -1.18333355e-01 -4.84685004e-01
-1.15437046e-01 1.14043597e-02 3.00729603e-01 -5.48869252e-01
1.18160450e+00 2.49204069e-01 5.86963117e-01 1.19128361e-01
1.10487056e+00 1.10583320e-01 -1.36183774e+00 1.34970188e-01
-7.67841041e-01 -2.86682695e-01 2.22874269e-01 -6.93975329e-01
-1.36954677e+00 1.21657097e+00 4.14219111e-01 2.10334778e-01
1.40340745e+00 -1.52745917e-01 8.96291375e-01 -1.84582733e-02
2.03437656e-01 -7.93685675e-01 1.15816154e-01 3.86580974e-01
1.13715458e+00 -6.60866678e-01 -3.43654782e-01 -3.24004710e-01
-5.50221026e-01 8.52626979e-01 3.04391861e-01 -3.68747711e-01
7.14643300e-01 6.68941379e-01 -1.71982363e-01 4.46320921e-01
-8.94766867e-01 -1.60843119e-01 3.48467320e-01 1.19039643e+00
7.70236313e-01 3.07533122e-03 -6.08113781e-02 4.90277350e-01
-5.86232901e-01 -6.05002604e-02 5.58715820e-01 4.03970093e-01
-3.01646352e-01 -1.09840882e+00 -3.17483187e-01 1.91166669e-01
-8.01394224e-01 -4.08998191e-01 1.98002718e-02 9.06768560e-01
-1.09846808e-01 1.19903839e+00 3.47124413e-02 -3.95468384e-01
3.13540965e-01 2.28555575e-01 4.87814128e-01 -2.65268356e-01
-7.59349883e-01 1.69037357e-01 3.23552728e-01 -5.42691410e-01
-3.63813370e-01 -4.68155384e-01 -8.63722682e-01 -2.82995909e-01
-1.92058325e-01 7.41969720e-02 5.65408230e-01 4.32970643e-01
3.24852347e-01 1.23673809e+00 6.82459354e-01 -1.06642568e+00
-5.98581791e-01 -1.13660479e+00 -5.30464232e-01 3.30526203e-01
4.65613514e-01 -5.90340376e-01 -4.53484178e-01 1.88270211e-01] | [15.423089981079102, 5.968552589416504] |
7f00dc08-c215-404b-9636-23219ebf77e8 | attention-based-feature-compression-for-cnn | 2211.13745 | null | https://arxiv.org/abs/2211.13745v2 | https://arxiv.org/pdf/2211.13745v2.pdf | Attention-based Feature Compression for CNN Inference Offloading in Edge Computing | This paper studies the computational offloading of CNN inference in device-edge co-inference systems. Inspired by the emerging paradigm semantic communication, we propose a novel autoencoder-based CNN architecture (AECNN), for effective feature extraction at end-device. We design a feature compression module based on the channel attention method in CNN, to compress the intermediate data by selecting the most important features. To further reduce communication overhead, we can use entropy encoding to remove the statistical redundancy in the compressed data. At the receiver, we design a lightweight decoder to reconstruct the intermediate data through learning from the received compressed data to improve accuracy. To fasten the convergence, we use a step-by-step approach to train the neural networks obtained based on ResNet-50 architecture. Experimental results show that AECNN can compress the intermediate data by more than 256x with only about 4% accuracy loss, which outperforms the state-of-the-art work, BottleNet++. Compared to offloading inference task directly to edge server, AECNN can complete inference task earlier, in particular, under poor wireless channel condition, which highlights the effectiveness of AECNN in guaranteeing higher accuracy within time constraint. | ['Qi Zhang', 'Alexandros Iosifidis', 'Nan Li'] | 2022-11-24 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [ 1.44691363e-01 1.26232892e-01 -1.31214336e-01 -3.50366980e-01
-2.28290290e-01 2.11659268e-01 -2.52093300e-02 -3.14119428e-01
-6.58017695e-01 6.61509097e-01 3.59403998e-01 -1.77614987e-01
-2.25608230e-01 -1.01394880e+00 -9.79866862e-01 -5.06273925e-01
1.42316133e-01 -6.18432881e-03 -5.61712682e-02 9.99411345e-02
-1.43830255e-01 1.20639414e-01 -1.68924975e+00 2.40246981e-01
5.98360419e-01 1.59436262e+00 7.21440852e-01 6.64005876e-01
8.01175926e-03 8.58439267e-01 -4.98899907e-01 -5.52888453e-01
-2.97751799e-02 -2.87091821e-01 -4.68029678e-01 -4.15909469e-01
-1.66049972e-01 -9.06989336e-01 -1.02576733e+00 1.14711237e+00
6.86534405e-01 -1.80143148e-01 7.91567564e-02 -1.12884653e+00
-2.82012016e-01 7.82705486e-01 1.68248296e-01 1.54469490e-01
-9.38965306e-02 -2.14646071e-01 7.56414175e-01 -7.21337974e-01
3.43119472e-01 8.89888823e-01 8.00986409e-01 6.67710483e-01
-3.70423377e-01 -1.04100335e+00 -5.81271797e-02 6.33394063e-01
-1.48724246e+00 -7.19235837e-01 6.45544767e-01 3.94154936e-01
1.12677002e+00 2.60802090e-01 8.53806078e-01 1.04248917e+00
1.20995507e-01 9.08749938e-01 5.26280403e-01 -9.96611714e-02
4.24671441e-01 -1.10593140e-02 -3.03339392e-01 6.68962061e-01
1.14072338e-01 3.33788618e-02 -6.70713484e-01 1.20420165e-01
5.40866435e-01 5.99053085e-01 -2.72867501e-01 7.22305417e-01
-9.48015332e-01 5.13594806e-01 6.77563131e-01 2.23370090e-01
-6.35472000e-01 6.26913488e-01 7.15785682e-01 4.35779870e-01
4.44705635e-01 -4.51761127e-01 -7.28072107e-01 -4.59562659e-01
-8.85648251e-01 -9.02232900e-02 9.31613743e-01 1.29746413e+00
4.83963192e-01 1.83317475e-02 -1.49996266e-01 5.36303103e-01
3.38050932e-01 4.83013421e-01 8.05328846e-01 -8.71715307e-01
7.34728158e-01 2.14223593e-01 -6.02861643e-01 -8.82032990e-01
-2.04555795e-01 -8.28128517e-01 -1.30605674e+00 -4.41686064e-01
-4.34379101e-01 -5.28092206e-01 -5.78892350e-01 1.45959532e+00
7.46203512e-02 6.57075465e-01 4.44186211e-01 1.00468266e+00
8.85835648e-01 7.80852497e-01 -2.04264354e-02 -1.25060424e-01
1.48597109e+00 -1.14407957e+00 -9.74206805e-01 -7.45101739e-03
8.11761916e-01 -4.46638674e-01 4.41570222e-01 2.95077860e-01
-1.01355755e+00 -6.35244548e-01 -1.23177254e+00 -2.17535734e-01
-2.41958573e-01 4.44875926e-01 7.45629549e-01 4.11820263e-01
-1.03973651e+00 8.89103830e-01 -9.39704835e-01 -5.66291548e-02
8.47512484e-01 6.84234202e-01 -1.60857767e-01 -1.89150319e-01
-1.21338689e+00 3.64641964e-01 7.97054768e-01 3.44072104e-01
-8.41511130e-01 -7.96536803e-01 -6.68329656e-01 6.33142948e-01
2.60129720e-01 -8.24181855e-01 1.17831802e+00 -8.89464796e-01
-1.84705460e+00 1.14360722e-02 -3.13350201e-01 -8.90605330e-01
1.64976731e-01 -2.90511906e-01 -7.02343583e-01 2.97320426e-01
-4.06860411e-01 6.51795626e-01 6.07240379e-01 -5.10290205e-01
-9.59210694e-01 -2.28759363e-01 -1.13376513e-01 2.03398675e-01
-9.94691193e-01 -3.64730835e-01 -8.74192655e-01 -5.17898381e-01
-3.42707336e-02 -8.45727324e-01 -8.12768266e-02 -1.06814010e-02
-3.81568670e-01 -1.70822397e-01 1.35126281e+00 -9.35721815e-01
1.46142936e+00 -2.35787082e+00 -7.07799271e-02 2.86431015e-01
5.73314250e-01 4.17133510e-01 3.38542342e-01 -2.93720420e-02
4.45668548e-01 -2.15398267e-01 -3.45329195e-02 -5.78854144e-01
-5.21818921e-02 3.90817285e-01 -1.15338817e-01 1.93485886e-01
-6.86389953e-02 1.00645638e+00 -6.17350578e-01 -4.07142729e-01
1.26446962e-01 7.45551765e-01 -8.06579173e-01 1.93416238e-01
-6.23250082e-02 2.08375920e-02 -7.17936337e-01 5.07472456e-01
7.74515808e-01 -4.09340113e-01 4.51760963e-02 -3.96765649e-01
2.10443199e-01 3.57438087e-01 -7.11240053e-01 1.97878218e+00
-1.08093107e+00 8.69362354e-01 1.92238852e-01 -1.02190769e+00
7.11543381e-01 5.41109443e-01 3.07169139e-01 -9.79070544e-01
5.32091200e-01 2.95491874e-01 -1.81905448e-01 -5.74364364e-01
3.04281503e-01 2.69245565e-01 1.47530824e-01 1.68399233e-02
2.36791149e-01 5.95413268e-01 -3.40597898e-01 3.00071687e-02
1.19854951e+00 -2.42683873e-01 -2.70191550e-01 4.52864915e-02
5.38266480e-01 -5.07827103e-01 5.54936588e-01 3.87129426e-01
7.40019605e-02 1.56247348e-01 9.32405442e-02 -3.05133194e-01
-9.62786794e-01 -4.51363832e-01 -5.13914786e-02 8.47391069e-01
1.86681375e-01 -7.12151706e-01 -1.14816129e+00 -5.63019693e-01
-2.17051104e-01 5.04874408e-01 -1.98344886e-01 -5.59043407e-01
-3.12565327e-01 -6.61603391e-01 7.20349252e-01 7.59092569e-01
1.37114429e+00 -9.65218365e-01 -8.78579199e-01 3.92788768e-01
-3.36390942e-01 -1.45455313e+00 -2.91686803e-01 1.15393579e-01
-9.85453725e-01 -5.08200049e-01 -5.54138660e-01 -8.49422157e-01
5.98590612e-01 7.55356848e-02 6.89983845e-01 3.85951251e-01
-7.62371533e-03 5.07410467e-02 -4.73727047e-01 -3.20831478e-01
-7.70212710e-02 4.33729529e-01 -2.07375929e-01 8.30588024e-03
4.85937059e-01 -8.22094858e-01 -8.93578112e-01 -4.46690544e-02
-7.93106139e-01 3.32806706e-01 9.61506605e-01 7.38137901e-01
6.72946095e-01 5.71633339e-01 5.30604959e-01 -6.91026092e-01
3.88768107e-01 -7.16612399e-01 -2.95906693e-01 -7.09792301e-02
-7.32343793e-01 1.33057818e-01 1.21579659e+00 -2.59903371e-01
-9.08201158e-01 -5.25073754e-03 -5.76826751e-01 -7.68818855e-01
1.40953973e-01 5.87210655e-01 -2.25628629e-01 -9.22096819e-02
-3.65408622e-02 5.43477595e-01 -6.63732588e-02 -5.45433581e-01
3.87733839e-02 1.15741909e+00 5.93787253e-01 -4.43055071e-02
2.58728355e-01 5.73841333e-01 8.22227895e-02 -6.69996619e-01
-5.93672097e-01 -1.09473526e-01 1.56645238e-01 -9.64539647e-02
8.94666851e-01 -1.39815331e+00 -1.30605841e+00 2.20957607e-01
-1.14244604e+00 -1.04973659e-01 -1.40939280e-02 9.12236214e-01
-3.23428780e-01 -2.76875515e-02 -8.27260256e-01 -5.16474783e-01
-1.00828445e+00 -1.03332567e+00 1.09579527e+00 4.64707226e-01
3.93707603e-01 -8.09320927e-01 -6.78758383e-01 1.43189847e-01
7.84747660e-01 -1.29345745e-01 5.76622665e-01 -5.52986264e-01
-7.37693071e-01 -3.57343763e-01 -4.65395898e-01 4.85025883e-01
-3.41619134e-01 -5.85024953e-01 -1.18711877e+00 -3.30617815e-01
1.08450755e-01 -3.59516852e-02 9.47591245e-01 2.30059430e-01
2.07323003e+00 -5.75194478e-01 -4.91318047e-01 1.22685611e+00
1.50975215e+00 3.86689045e-02 8.48342717e-01 -2.49577146e-02
7.39637256e-01 -1.05843507e-01 2.44891420e-01 7.62605727e-01
7.27073729e-01 3.68720919e-01 6.66219175e-01 4.16556746e-02
-2.97239453e-01 -3.88011575e-01 2.28221849e-01 1.46397245e+00
-9.12103802e-03 -3.70394349e-01 -7.70992339e-02 2.34378979e-01
-1.89442599e+00 -6.89775765e-01 1.52013555e-01 1.91863418e+00
6.09472632e-01 4.24216121e-01 -4.70211178e-01 3.32776070e-01
5.39695144e-01 -1.02766208e-01 -6.33444011e-01 -1.14623308e-01
2.50210822e-01 3.11005294e-01 8.90680611e-01 1.92709625e-01
-7.30158210e-01 7.76558280e-01 4.88265228e+00 1.24699438e+00
-1.35048032e+00 4.72200722e-01 7.01387763e-01 -2.55359113e-01
-1.91454843e-01 -4.27161530e-02 -1.00182629e+00 1.06090236e+00
1.45911992e+00 3.04113113e-04 7.37981319e-01 1.18557084e+00
6.47477284e-02 2.38585413e-01 -8.86903882e-01 1.45837498e+00
4.53637801e-02 -1.49191821e+00 -2.08049759e-01 2.10986510e-02
3.72810036e-01 1.58842534e-01 -1.46890208e-01 5.15230894e-01
-1.72785699e-01 -9.33208048e-01 5.90317011e-01 7.45798886e-01
9.39542294e-01 -1.10705745e+00 9.72570658e-01 4.31941748e-01
-1.38327742e+00 -2.67614305e-01 -7.36500263e-01 -1.44797534e-01
4.94552553e-02 1.00090706e+00 -4.81893659e-01 5.19978881e-01
1.04080904e+00 9.58166957e-01 -1.58756495e-01 7.71883070e-01
1.24131829e-01 7.09040403e-01 -5.38811028e-01 -3.69817406e-01
-4.07367013e-03 1.35062113e-01 9.21248123e-02 1.16679335e+00
7.14277387e-01 2.46701017e-01 -1.65845931e-01 6.31565034e-01
-6.42726839e-01 -1.62810609e-01 -3.43155026e-01 2.09079877e-01
7.74552464e-01 1.28139853e+00 -4.59761977e-01 -6.15586758e-01
-4.50480878e-01 1.50064933e+00 2.72110730e-01 2.05848575e-01
-9.93756294e-01 -8.45784664e-01 4.76630867e-01 -2.69834936e-01
6.99053109e-01 -2.88857147e-02 -1.15868524e-01 -9.05965269e-01
4.52267259e-01 -4.29558128e-01 3.48731540e-02 -7.63155222e-01
-5.62768340e-01 6.65418029e-01 -3.90527993e-01 -1.16476500e+00
5.73088340e-02 -3.92770022e-01 -5.49902618e-01 5.56500852e-01
-1.83229756e+00 -8.92360389e-01 -6.53783083e-01 7.18591213e-01
5.42652488e-01 -1.79560542e-01 7.80533671e-01 9.59289134e-01
-6.35269225e-01 9.12151337e-01 2.93614745e-01 1.63399890e-01
4.95941639e-02 -5.18415570e-01 3.70988280e-01 4.68698353e-01
-2.65512258e-01 2.70078748e-01 2.66907275e-01 -4.80101794e-01
-1.90220773e+00 -1.48711133e+00 7.91466892e-01 4.65708673e-01
1.70863613e-01 -4.43646997e-01 -5.63666880e-01 5.22434533e-01
1.59436613e-01 4.61896002e-01 5.77826142e-01 -3.61517727e-01
7.28210285e-02 -5.43182313e-01 -1.06759846e+00 4.72951204e-01
1.37705398e+00 -6.52470410e-01 1.63101003e-01 3.89991105e-01
1.12108326e+00 -5.57131529e-01 -1.08758974e+00 3.50211293e-01
5.79584718e-01 -7.45180070e-01 7.53405094e-01 -2.18213797e-01
5.51914990e-01 -7.84773007e-02 -2.54387379e-01 -1.08961248e+00
-1.10594131e-01 -6.33075535e-01 -7.60270476e-01 1.11478674e+00
3.57126534e-01 -5.85866094e-01 1.04506683e+00 2.31659368e-01
-3.33732039e-01 -1.03722036e+00 -1.14305270e+00 -4.95416433e-01
-7.27411389e-01 -7.16626704e-01 9.31804240e-01 3.64652395e-01
-1.90874368e-01 1.78507119e-01 -5.73834479e-01 2.70202577e-01
3.67187381e-01 -2.84247369e-01 4.33438659e-01 -1.06873930e+00
-5.79357624e-01 -1.00553013e-01 -6.22804224e-01 -1.47800887e+00
3.31479937e-01 -9.05590594e-01 -5.67827374e-02 -1.24357104e+00
-5.58807924e-02 -4.26310450e-01 -3.66265088e-01 2.91502148e-01
8.92138332e-02 1.79552794e-01 1.52707815e-01 5.76625578e-02
-9.28581357e-01 8.54031265e-01 1.06949532e+00 2.83002369e-02
7.45669603e-02 1.14767784e-02 -7.80278802e-01 5.33554256e-01
7.77960956e-01 -4.95887429e-01 -5.32644033e-01 -8.39289069e-01
2.49175370e-01 3.01377803e-01 3.73972982e-01 -1.53154230e+00
7.13353992e-01 5.97636461e-01 6.03463173e-01 -4.54863429e-01
3.95696014e-01 -1.49324846e+00 1.46244973e-01 6.81264699e-01
-2.51315147e-01 5.02646994e-03 9.21162963e-02 7.14378417e-01
-1.83138356e-01 -1.64226398e-01 1.60068959e-01 1.61543444e-01
-7.35188723e-01 5.77604413e-01 -1.20385900e-01 -3.28046769e-01
8.31144392e-01 1.31794065e-01 -2.30487749e-01 -3.95321220e-01
-5.17895818e-01 2.83495843e-01 -2.09235713e-01 2.03614429e-01
8.99627924e-01 -1.48727357e+00 -3.86379570e-01 5.18550456e-01
-2.45978653e-01 1.65521771e-01 5.84662974e-01 8.74300063e-01
-6.00018442e-01 4.35853988e-01 7.43130147e-02 -3.73137623e-01
-9.87781763e-01 3.73171389e-01 2.33727530e-01 -1.03038520e-01
-8.55366707e-01 8.53955567e-01 -2.69242018e-01 1.11653104e-01
4.56895739e-01 -4.69726026e-01 1.74967316e-03 -5.24694622e-01
6.87389612e-01 4.86923277e-01 4.11708087e-01 -1.34108499e-01
-1.61372691e-01 2.63659686e-01 -1.96185991e-01 1.62054434e-01
1.25265121e+00 -2.25874469e-01 2.50373278e-02 -2.17215523e-01
1.82097769e+00 -3.41174662e-01 -1.34817719e+00 -3.70828182e-01
-5.12529373e-01 -1.19960085e-01 8.57657492e-01 -3.82442713e-01
-1.65836108e+00 6.56105578e-01 8.87633026e-01 -1.46075860e-01
1.60681105e+00 -1.98092297e-01 1.59475553e+00 6.06516838e-01
1.71606109e-01 -1.26441133e+00 -3.73444974e-01 4.41627502e-01
3.19254398e-01 -9.63974893e-01 -1.86867818e-01 -3.36665511e-01
-2.19083950e-01 1.20590854e+00 3.32928330e-01 -1.17839597e-01
1.19132650e+00 5.19126236e-01 -5.06002069e-01 -1.15671128e-01
-9.12235200e-01 -6.33335561e-02 -1.16398804e-01 2.75252521e-01
-2.66731530e-02 2.40893923e-02 -9.28285792e-02 9.14315164e-01
-5.13751090e-01 3.88273329e-01 -4.96995971e-02 8.14217746e-01
-2.90919781e-01 -6.47558987e-01 1.88600123e-01 8.43034804e-01
-5.11524498e-01 -4.43996698e-01 3.29923004e-01 3.25290650e-01
3.28275889e-01 8.60424340e-01 4.18345362e-01 -9.93641734e-01
1.72300920e-01 -1.86110467e-01 1.11755662e-01 5.48454896e-02
-3.82893950e-01 -1.49402484e-01 -1.68376118e-02 -6.23012722e-01
-2.13681772e-01 -1.40756220e-01 -1.43574739e+00 -7.10606098e-01
-4.37145621e-01 5.35768159e-02 1.12895739e+00 1.19676733e+00
9.39434946e-01 1.17068815e+00 7.07339287e-01 -7.17462957e-01
-2.68010288e-01 -8.53891432e-01 -2.99965054e-01 4.53811586e-02
3.73908132e-01 -1.38877138e-01 -1.83684111e-01 -1.00500114e-01] | [8.458355903625488, 2.839570999145508] |
f163097b-9652-4485-9e4d-1dd89f650177 | hierarchical-sketch-induction-for-paraphrase | 2203.03463 | null | https://arxiv.org/abs/2203.03463v2 | https://arxiv.org/pdf/2203.03463v2.pdf | Hierarchical Sketch Induction for Paraphrase Generation | We propose a generative model of paraphrase generation, that encourages syntactic diversity by conditioning on an explicit syntactic sketch. We introduce Hierarchical Refinement Quantized Variational Autoencoders (HRQ-VAE), a method for learning decompositions of dense encodings as a sequence of discrete latent variables that make iterative refinements of increasing granularity. This hierarchy of codes is learned through end-to-end training, and represents fine-to-coarse grained information about the input. We use HRQ-VAE to encode the syntactic form of an input sentence as a path through the hierarchy, allowing us to more easily predict syntactic sketches at test time. Extensive experiments, including a human evaluation, confirm that HRQ-VAE learns a hierarchical representation of the input space, and generates paraphrases of higher quality than previous systems. | ['Mirella Lapata', 'Hao Tang', 'Tom Hosking'] | 2022-03-07 | null | https://aclanthology.org/2022.acl-long.178 | https://aclanthology.org/2022.acl-long.178.pdf | acl-2022-5 | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [-1.20540798e-01 4.99187559e-01 -1.92535087e-01 -3.80248487e-01
-9.25863385e-01 -8.27140808e-01 6.05271935e-01 -2.71462113e-01
1.39632672e-01 7.31052935e-01 9.71071899e-01 -3.41737598e-01
2.50893176e-01 -1.23569357e+00 -1.04345942e+00 -3.61509502e-01
2.95949101e-01 8.50493908e-01 -2.32936203e-01 -3.42561305e-01
1.45376891e-01 6.54499382e-02 -1.54747903e+00 1.14142179e+00
6.56871259e-01 5.78762770e-01 3.06384712e-01 7.57828116e-01
-3.60146046e-01 7.87886024e-01 -7.68377125e-01 -7.52469122e-01
1.46158099e-01 -6.97863698e-01 -1.00850368e+00 1.87858671e-01
3.64126116e-01 -4.16206896e-01 -2.38679975e-01 7.84371316e-01
-1.43638834e-01 5.76556511e-02 1.11216748e+00 -1.05300200e+00
-1.49365103e+00 8.72805357e-01 -5.96324019e-02 -4.62870263e-02
6.26760185e-01 1.77492499e-01 1.77906072e+00 -1.23918259e+00
9.97426510e-01 1.30014443e+00 6.30165219e-01 8.65993857e-01
-1.89186931e+00 -4.31620359e-01 -6.03540726e-02 -5.91336703e-03
-1.24219227e+00 -9.85845476e-02 8.65113914e-01 -6.60608411e-01
1.31354702e+00 1.36858255e-01 1.13614106e+00 1.56847703e+00
4.61707830e-01 1.13607991e+00 7.03322291e-01 -2.60885328e-01
5.86565852e-01 9.44571272e-02 1.25499979e-01 9.54327524e-01
-3.60479169e-02 -1.13157704e-02 -5.51350057e-01 -6.03630900e-01
1.14867723e+00 -4.30675559e-02 -1.69992581e-01 -6.07383728e-01
-7.05996811e-01 1.51509631e+00 5.00309527e-01 8.16905349e-02
-4.61772591e-01 7.81849772e-02 2.19695598e-01 5.51142097e-01
3.37713510e-01 8.08405936e-01 -2.84559041e-01 -9.08974037e-02
-1.21823823e+00 6.13207400e-01 8.67360592e-01 1.23030114e+00
7.24327087e-01 2.07378827e-02 -5.13534188e-01 9.21160579e-01
7.48899803e-02 2.25935891e-01 6.43584847e-01 -1.17540038e+00
4.98157412e-01 6.26711786e-01 4.78548035e-02 -3.86228204e-01
2.89765984e-01 -2.14750797e-01 -9.31877494e-01 1.61730722e-01
-1.26395702e-01 -4.10085693e-02 -1.03613591e+00 1.79865193e+00
-3.20779860e-01 -2.23111585e-01 2.67417461e-01 5.80437660e-01
6.55483544e-01 1.12952971e+00 1.60035286e-02 -6.04210012e-02
1.04453588e+00 -7.93826699e-01 -4.23411429e-01 3.82232741e-02
3.27260911e-01 -2.35302880e-01 1.45475137e+00 3.38894337e-01
-1.55392361e+00 -9.30125654e-01 -9.61481273e-01 -4.32487667e-01
-1.10507466e-01 -7.39123225e-02 6.02098763e-01 4.30452526e-01
-1.32072997e+00 7.69257605e-01 -6.52349591e-01 8.28995854e-02
4.52600300e-01 8.30357894e-02 -1.06396608e-01 7.50203133e-02
-1.23183036e+00 5.60861230e-01 3.82728130e-01 -3.65887344e-01
-9.34332550e-01 -7.33050585e-01 -1.19766617e+00 4.01072085e-01
-3.12677860e-01 -1.41070616e+00 1.50373483e+00 -8.16879451e-01
-1.54896784e+00 6.98417962e-01 -5.04705906e-01 -4.20727968e-01
1.33161724e-01 7.99036026e-03 2.13214010e-01 -1.03266807e-02
1.59287676e-01 9.78800833e-01 1.14334738e+00 -1.27530432e+00
-3.62251431e-01 -3.37292324e-03 2.55345017e-01 2.21153393e-01
-2.27104440e-01 -4.14902508e-01 -1.79606512e-01 -9.06271636e-01
-2.97457688e-02 -7.20982969e-01 -2.47944012e-01 -2.72077501e-01
-2.61798441e-01 -6.40319347e-01 3.98453385e-01 -6.82605147e-01
1.23910463e+00 -1.95640254e+00 1.02109551e+00 -8.36758614e-02
3.07954431e-01 -4.13148791e-01 -3.02919567e-01 5.34229517e-01
1.45214032e-02 3.75962108e-01 -3.30163330e-01 -6.82584286e-01
2.83883423e-01 5.59231222e-01 -8.66658330e-01 -2.66419470e-01
4.04055178e-01 1.36127579e+00 -1.03386819e+00 -3.79609048e-01
-9.67404023e-02 1.56665221e-01 -1.13185751e+00 6.89996064e-01
-6.35084033e-01 -1.76149663e-02 -1.33989051e-01 3.67186546e-01
1.67928815e-01 -4.86724615e-01 -4.82320264e-02 -1.43550694e-01
2.67903328e-01 4.34778988e-01 -6.43511057e-01 1.82642615e+00
-6.86616421e-01 5.57897270e-01 -2.74906367e-01 -4.98131812e-01
7.33678401e-01 4.68593150e-01 -6.90548718e-02 -4.46402401e-01
-1.49038926e-01 -1.59347266e-01 -5.30924678e-01 -2.51229286e-01
8.32534611e-01 -5.54996014e-01 -5.61961412e-01 6.40914083e-01
5.50017715e-01 -6.15025938e-01 3.02527040e-01 6.79907620e-01
8.95890534e-01 1.70440584e-01 2.51908392e-01 -1.35132924e-01
6.81685358e-02 4.21947800e-02 3.92381340e-01 1.05661881e+00
3.20307940e-01 7.91986346e-01 7.03271747e-01 -5.62508345e-01
-1.89321828e+00 -1.56974959e+00 1.10501051e-03 1.07187819e+00
-2.56755739e-01 -8.37744415e-01 -8.12539041e-01 -6.28672898e-01
1.23037472e-01 1.04728019e+00 -9.64071810e-01 3.26126032e-02
-5.37279308e-01 -3.55853200e-01 3.99691254e-01 9.86773014e-01
6.70681000e-02 -1.29609013e+00 -6.00093901e-01 3.92864943e-01
-2.19546676e-01 -6.62592053e-01 -3.35387230e-01 1.27866134e-01
-1.03019500e+00 -7.11178899e-01 -9.24849272e-01 -1.04625225e+00
6.99250221e-01 -2.03352585e-01 1.74963629e+00 2.19501890e-02
-1.55935928e-01 1.86372072e-01 -4.88038659e-01 2.77878083e-02
-1.06068218e+00 4.05837111e-02 -1.76566049e-01 -3.74930739e-01
2.12882310e-02 -7.98014224e-01 -4.13442105e-01 -3.30048710e-01
-7.82637358e-01 5.12355685e-01 5.03647685e-01 1.35581970e+00
7.16914058e-01 -2.07667038e-01 2.20735729e-01 -1.19096124e+00
1.31419373e+00 -6.06390655e-01 -3.52771401e-01 2.90811956e-01
-2.79479146e-01 5.26064754e-01 1.01825511e+00 -2.64911324e-01
-9.56826270e-01 -4.14095111e-02 -3.71147752e-01 -7.08148658e-01
-1.95913047e-01 4.70147461e-01 2.09245324e-01 6.91111863e-01
9.08223450e-01 6.36911094e-01 -3.71566057e-01 -2.85281181e-01
9.90399122e-01 5.30096769e-01 6.56821311e-01 -8.94733310e-01
7.43692994e-01 -4.03621979e-02 -4.87513989e-01 -6.81733847e-01
-6.86821818e-01 1.83473714e-02 -5.75484157e-01 3.10190350e-01
1.01182640e+00 -9.66333508e-01 -2.37967968e-01 -2.84767710e-03
-1.48089159e+00 -6.24280334e-01 -6.53620362e-01 -2.70524025e-01
-7.94103503e-01 2.38032609e-01 -1.17060947e+00 -4.37595278e-01
-5.28503716e-01 -9.74691093e-01 1.31452012e+00 -6.54044747e-02
-8.41277957e-01 -8.95443857e-01 2.58530498e-01 1.77921683e-01
1.27108768e-01 -7.38596469e-02 1.52109146e+00 -1.67631119e-01
-7.83672452e-01 1.67380154e-01 8.80103260e-02 2.42166564e-01
-2.21212521e-01 1.32926134e-02 -7.16256022e-01 -4.17302042e-01
-4.31032712e-03 -7.93269217e-01 8.77225757e-01 2.68538535e-01
1.19245863e+00 -7.60211885e-01 -2.62888223e-02 7.59045124e-01
1.27322745e+00 -1.61242127e-01 6.55386508e-01 -3.93618047e-02
5.60043097e-01 2.71712840e-01 -9.26777944e-02 5.49660563e-01
2.35591382e-01 4.38773483e-01 1.19459713e-02 1.03957213e-01
-8.64682719e-02 -8.45173061e-01 3.99904102e-01 9.58326638e-01
1.45277247e-01 -1.33534268e-01 -4.93968159e-01 6.69051409e-01
-1.63788748e+00 -1.32245708e+00 4.02628541e-01 1.63042068e+00
9.91707861e-01 1.49102420e-01 2.69643396e-01 -2.36205012e-02
3.41080159e-01 2.28005975e-01 -4.49657947e-01 -8.52732360e-01
8.84093642e-02 6.84532046e-01 -4.99580763e-02 7.43086517e-01
-7.13546634e-01 1.13433492e+00 7.73484898e+00 5.82226336e-01
-6.55589998e-01 -1.80758331e-02 4.78262484e-01 -8.97260532e-02
-1.07387197e+00 -1.04735047e-01 -5.90492547e-01 4.23912674e-01
7.89150715e-01 -3.49566549e-01 6.36047661e-01 1.03399599e+00
-3.05767834e-01 5.38819790e-01 -1.51710415e+00 7.02911854e-01
-9.63891391e-03 -1.94795477e+00 7.07158804e-01 -1.53257668e-01
1.00843453e+00 -2.12615699e-01 1.53857097e-01 8.45328748e-01
9.68389153e-01 -1.11213171e+00 7.96600997e-01 2.57681429e-01
9.12973881e-01 -7.54012406e-01 3.47510010e-01 6.19723618e-01
-1.10980988e+00 -2.73062527e-01 -7.60249317e-01 -2.03266233e-01
2.48528510e-01 4.12987210e-02 -8.38734508e-01 1.63551643e-01
4.24694210e-01 5.67342401e-01 -5.08963823e-01 3.33554804e-01
-8.85912597e-01 6.86515450e-01 1.56096071e-01 -3.87243807e-01
2.45670825e-01 3.65721621e-02 3.80462140e-01 1.41013658e+00
2.79187441e-01 3.98686647e-01 1.45465925e-01 1.67932832e+00
-2.21387312e-01 -1.18737891e-01 -8.46471488e-01 9.05214772e-02
5.72111845e-01 7.86370575e-01 -1.52858958e-01 -7.52267838e-01
-1.52216837e-01 1.54102230e+00 7.65249372e-01 5.76879561e-01
-6.15230620e-01 -2.97011614e-01 6.20188057e-01 -1.99426427e-01
6.79493964e-01 -2.18682840e-01 -4.87719953e-01 -1.64070892e+00
-2.31049851e-01 -1.00237024e+00 3.17156583e-01 -1.19140029e+00
-1.36961305e+00 8.91119659e-01 5.15215807e-02 -8.93622696e-01
-9.65132773e-01 -4.99313354e-01 -7.67694950e-01 1.17501795e+00
-6.15804195e-01 -1.12808967e+00 9.27499160e-02 4.59980100e-01
1.25064790e+00 -4.36510801e-01 1.36348307e+00 -6.42535806e-01
-1.82583332e-01 8.29726160e-01 -1.11024650e-02 2.75002897e-01
-1.26467347e-01 -1.56540620e+00 1.09296858e+00 6.21039212e-01
7.85782516e-01 1.09447694e+00 8.26194048e-01 -6.91890717e-01
-1.07193542e+00 -8.49451602e-01 9.82112706e-01 -9.31100368e-01
5.56385159e-01 -6.31855309e-01 -1.06097341e+00 8.04553866e-01
5.41712284e-01 -6.37223840e-01 8.27439547e-01 5.33836067e-01
-7.74970651e-01 5.90366185e-01 -8.97490621e-01 6.79084063e-01
9.59026992e-01 -1.19372725e+00 -1.50635707e+00 9.68691260e-02
9.56282496e-01 -4.57766950e-01 -9.02491450e-01 -3.52369361e-02
5.68247020e-01 -9.42832232e-01 8.69962394e-01 -9.19029951e-01
1.37205791e+00 2.27919370e-01 -1.72005549e-01 -1.68545187e+00
-1.06433392e+00 -3.15838039e-01 -4.27326739e-01 7.24904180e-01
5.87330699e-01 1.40948728e-01 1.24344301e+00 6.07305646e-01
-1.04561627e-01 -9.86611664e-01 -6.41827285e-01 -4.90028650e-01
5.30169189e-01 -1.39036730e-01 6.72600329e-01 7.65621185e-01
3.52629334e-01 8.77659023e-01 -3.12858820e-01 -4.27632742e-02
8.02954018e-01 5.75966775e-01 7.55665481e-01 -1.02374542e+00
-9.73177135e-01 -5.48717737e-01 -1.47234619e-01 -1.46547723e+00
4.39027607e-01 -1.18968999e+00 -1.46210706e-02 -1.71972942e+00
5.24583995e-01 2.59369820e-01 -1.44862786e-01 4.40118343e-01
-1.97222352e-01 2.17655957e-01 1.97001323e-01 4.08766419e-01
-4.45086479e-01 8.52514446e-01 1.12494826e+00 -4.18334365e-01
-2.19366580e-01 -3.24819624e-01 -6.73570871e-01 4.62993532e-01
5.64786255e-01 -3.76683444e-01 -6.37080669e-01 -7.29617357e-01
2.85964638e-01 5.07480323e-01 2.63876379e-01 -7.50350296e-01
5.27086779e-02 -2.10314110e-01 7.97160387e-01 -6.98995113e-01
4.14068222e-01 -1.71120957e-01 1.20886981e-01 2.20360309e-01
-9.33958709e-01 2.23714426e-01 -6.14316016e-02 4.59179372e-01
-2.73623616e-01 -2.71876454e-01 5.39392650e-01 -4.32918489e-01
-5.97947061e-01 1.51018247e-01 -6.39515281e-01 6.43237233e-02
3.30144823e-01 -2.85669327e-01 -1.27213942e-02 -5.93167067e-01
-9.19200778e-01 1.51605994e-01 7.54678547e-01 4.66854781e-01
9.42882359e-01 -1.59547389e+00 -7.26043105e-01 6.75123036e-01
1.95389435e-01 -7.11134076e-02 1.12513930e-01 -5.22490501e-01
-2.21154764e-01 4.00330216e-01 -4.14957583e-01 -2.81078577e-01
-1.00247085e+00 6.46493316e-01 4.54946943e-02 -3.78084868e-01
-8.85818779e-01 1.25381029e+00 3.60055417e-01 -1.90318942e-01
1.72570020e-01 -4.53834981e-01 -6.61255047e-02 -1.57538190e-01
4.07302529e-01 -2.09054515e-01 -4.48295593e-01 -2.86294639e-01
3.01873118e-01 2.86396354e-01 -2.68469095e-01 -5.24668634e-01
1.30637288e+00 1.85179129e-01 -4.85353963e-03 4.16198015e-01
1.30762923e+00 -1.29315823e-01 -1.36170280e+00 -1.14801683e-01
-2.94658005e-01 -1.97847500e-01 -3.01537544e-01 -4.84762371e-01
-4.98456866e-01 9.34490383e-01 -1.63537800e-01 3.21572572e-01
6.87956870e-01 3.16693157e-01 9.06416237e-01 6.20441258e-01
5.51143110e-01 -8.29450727e-01 5.65708160e-01 7.60585129e-01
1.33117080e+00 -6.29724264e-01 -2.05531716e-01 -6.38203248e-02
-8.65119576e-01 1.07645726e+00 4.80789781e-01 -5.97908854e-01
2.75134325e-01 1.95180848e-01 -7.88023844e-02 -6.62980080e-02
-1.29166770e+00 1.76738396e-01 3.05424273e-01 4.89205092e-01
5.39676785e-01 1.76874503e-01 -5.89223243e-02 9.56765652e-01
-7.44845808e-01 1.14644460e-01 4.06712174e-01 6.81826353e-01
-4.70415026e-01 -1.35193753e+00 -8.17080885e-02 5.17456472e-01
-1.16416672e-02 -3.92079771e-01 -6.06596708e-01 4.30594772e-01
1.10108458e-01 4.83155102e-01 2.84885645e-01 -5.42217076e-01
3.40168804e-01 4.98521537e-01 7.68252790e-01 -1.07589471e+00
-5.74200213e-01 -7.69880116e-02 -6.62085190e-02 -5.70672870e-01
3.11472446e-01 -5.26021838e-01 -1.08399367e+00 -1.08487591e-01
3.05067241e-01 5.36202192e-01 9.77933854e-02 6.36460006e-01
3.06520343e-01 4.26470906e-01 6.44692123e-01 -9.71118569e-01
-9.40845668e-01 -7.98642278e-01 -3.78790200e-01 1.06008077e+00
3.08665097e-01 -2.94603527e-01 -2.79788136e-01 3.09371769e-01] | [11.722087860107422, 9.218084335327148] |
02eb1013-114d-4f09-adc9-06797c1f9b9c | graph-attention-network-based-single-pixel | 2109.05466 | null | https://arxiv.org/abs/2109.05466v2 | https://arxiv.org/pdf/2109.05466v2.pdf | Graph Attention Network Based Single-Pixel Compressive Direction of Arrival Estimation | In this paper, we present a single-pixel compressive direction of arrival (DoA) estimation technique leveraging a graph attention network (GAT)-based deep-learning framework. The physical layer compression is achieved using a coded-aperture technique, probing the spectrum of far-field sources that are incident on the aperture using a set of spatio-temporally incoherent modes. This information is then encoded and compressed into the channel of the coded-aperture. The coded-aperture is based on a metasurface antenna design and it works as a receiver, exhibiting a single-channel and replacing the conventional multichannel raster scan-based solutions for DoA estimation. The GAT network enables the compressive DoA estimation framework to learn the DoA information directly from the measurements acquired using the coded-aperture. This step eliminates the need for an additional reconstruction step and significantly simplifies the processing layer to achieve DoA estimation. We show that the presented GAT integrated single-pixel radar framework can retrieve high fidelity DoA information even under relatively low signal-to-noise ratio (SNR) levels. | ['Güneş Karabulut Kurt', 'Okan Yurduseven', 'Kürşat Tekbıyık'] | 2021-09-12 | null | null | null | null | ['direction-of-arrival-estimation'] | ['audio'] | [ 6.69947803e-01 2.31360510e-01 4.96081620e-01 -2.97460072e-02
-9.03143823e-01 -2.00048134e-01 5.86183012e-01 -3.61330301e-01
-2.86829621e-01 5.40811419e-01 3.96869868e-01 -3.91237617e-01
-6.66733205e-01 -1.05736172e+00 -6.35335267e-01 -1.30455327e+00
-4.27443743e-01 1.43959880e-01 -4.92446840e-01 1.73372686e-01
7.92495906e-02 5.15734375e-01 -1.59125721e+00 3.14873755e-01
5.41242719e-01 1.36346304e+00 6.03650749e-01 8.18760157e-01
3.47383559e-01 7.18872488e-01 -3.21661204e-01 3.03368539e-01
5.18552244e-01 -5.95299602e-01 2.16484427e-01 -7.52048148e-03
7.21964896e-01 -5.58442414e-01 -1.04282653e+00 9.87392902e-01
2.99063802e-01 -2.47317851e-01 6.24181390e-01 -5.42696834e-01
-4.33608741e-01 1.50144398e-01 -1.69151351e-02 -8.60590264e-02
6.46700040e-02 3.67676556e-01 9.01676953e-01 -8.69573653e-01
4.39233929e-01 6.50966704e-01 5.41175604e-01 -9.38270465e-02
-9.34666932e-01 -2.65415370e-01 -8.24531198e-01 1.09499507e-01
-1.22976303e+00 -6.76009417e-01 9.26311076e-01 -2.75417000e-01
7.23342478e-01 1.06105655e-01 8.82704198e-01 6.50426507e-01
6.22909546e-01 2.03396961e-01 1.28337252e+00 -7.42007375e-01
6.27938092e-01 -8.09952974e-01 -2.04011813e-01 8.09766650e-01
5.06758034e-01 5.48357904e-01 -6.26134217e-01 5.67899197e-02
8.59425724e-01 9.63203311e-02 -8.96125257e-01 -4.28947598e-01
-1.32157743e+00 8.43776941e-01 7.67276227e-01 5.21522760e-01
-8.96635115e-01 6.39033258e-01 -4.78578031e-01 4.95242447e-01
1.42218485e-01 8.53279889e-01 6.61606789e-02 4.27747756e-01
-1.42882383e+00 1.29712790e-01 8.93879950e-01 5.06116152e-01
9.14192557e-01 4.97963876e-01 -8.56898203e-02 1.87091172e-01
7.71530867e-01 1.14477837e+00 1.66378133e-02 -1.21001995e+00
1.66653603e-01 -1.35858804e-01 3.08438838e-01 -1.16517115e+00
-3.20355803e-01 -1.31156147e+00 -1.02488577e+00 4.86507207e-01
2.65849531e-01 -4.86076504e-01 -1.00000191e+00 1.46670282e+00
3.76733136e-03 4.31058764e-01 3.12262535e-01 1.17776680e+00
3.02514136e-01 8.08439612e-01 -5.47656000e-01 -5.67233562e-01
1.13957810e+00 -4.67582852e-01 -9.05757785e-01 -5.70251524e-01
3.12745780e-01 -6.34503841e-01 1.70452908e-01 6.75856352e-01
-1.00840771e+00 -3.87032330e-01 -1.51727438e+00 2.09342316e-01
7.73151517e-02 2.73767442e-01 5.94889164e-01 4.53278542e-01
-1.07270467e+00 4.21115190e-01 -6.75928831e-01 3.42100501e-01
4.92223501e-01 -1.22072287e-01 -1.26933947e-01 -7.73707509e-01
-7.85194516e-01 5.32606363e-01 -8.36869422e-03 3.04267466e-01
-1.07981503e+00 -8.12374234e-01 -8.03795218e-01 2.64204174e-01
1.39732718e-01 -9.41775024e-01 4.85917270e-01 -4.15468574e-01
-1.42936563e+00 3.11453223e-01 -6.83298195e-03 -8.00651073e-01
6.50884956e-02 -2.65144318e-01 -3.73014510e-01 7.73114920e-01
7.29491860e-02 1.34194627e-01 1.27717209e+00 -1.18149388e+00
-8.79600123e-02 -4.22110409e-01 -2.94642121e-01 -9.35169384e-02
4.37918633e-01 -9.67942774e-01 1.90632626e-01 -4.81504440e-01
8.80162537e-01 -6.82496786e-01 -2.10691050e-01 3.66601914e-01
-2.08115473e-01 6.67717993e-01 7.88042963e-01 -5.10473371e-01
8.26899767e-01 -2.10455489e+00 1.58609182e-01 4.45249788e-02
3.95845383e-01 1.05938628e-01 -3.12499225e-01 1.02534235e+00
-4.00961079e-02 -6.37513816e-01 -5.93460858e-01 -3.48244846e-01
-3.85979474e-01 -6.19937479e-02 -3.24157953e-01 8.02553177e-01
5.17064743e-02 8.31213415e-01 -7.20698059e-01 1.81072533e-01
3.56666684e-01 4.87484515e-01 -7.59250343e-01 5.39587796e-01
-5.59928834e-01 7.77468503e-01 -3.99879277e-01 5.48539281e-01
1.09701979e+00 -2.67828256e-01 1.16259746e-01 -4.85689312e-01
-4.71539944e-01 7.21875811e-03 -9.03940678e-01 2.19253635e+00
-8.04214478e-01 1.01544440e+00 6.11021876e-01 -1.32934594e+00
9.29691195e-01 2.06897601e-01 4.19482648e-01 -9.47854757e-01
2.04945467e-02 5.46244919e-01 -1.01622425e-01 -8.11715007e-01
9.72171426e-02 -3.92881125e-01 1.43363014e-01 3.95640820e-01
1.05516590e-01 -4.21699941e-01 -4.47511822e-01 1.07396379e-01
1.50690019e+00 -7.28455186e-02 2.21142188e-01 -2.39379883e-01
2.87923306e-01 -9.46008861e-02 1.11294396e-01 1.04801309e+00
4.60837603e-01 6.74830973e-01 -7.21908137e-02 -3.77219737e-01
-1.08022690e+00 -1.05613863e+00 -1.75878912e-01 -5.80863431e-02
-1.13612644e-01 -3.36499065e-02 -3.00686717e-01 2.37419486e-01
-1.04771949e-01 9.49243665e-01 -2.91321158e-01 6.93172812e-02
-3.64143521e-01 -5.21356046e-01 2.01185435e-01 -3.63851428e-01
1.05782652e+00 -4.84237999e-01 -1.02381766e+00 4.45149094e-01
-1.83884650e-01 -1.24319005e+00 4.47076738e-01 2.27273345e-01
-7.78692782e-01 -9.84687805e-01 -7.03358769e-01 -5.10626078e-01
2.54828990e-01 6.40878260e-01 7.05309331e-01 -3.14116925e-01
-4.28318530e-01 8.80750716e-01 -2.97034383e-01 -2.00816289e-01
-1.62878156e-01 -6.53637350e-01 -2.55744338e-01 6.04894102e-01
4.41155508e-02 -1.16484582e+00 -1.02853906e+00 -4.43983018e-01
-6.98232591e-01 2.43949115e-01 1.03872442e+00 9.27885950e-01
4.46513921e-01 7.98657257e-03 4.31812018e-01 -2.05914184e-01
9.28231552e-02 -6.39620185e-01 -9.79732454e-01 -1.98040292e-01
-3.39494139e-01 1.87343918e-02 4.85514671e-01 1.30256057e-01
-7.45839417e-01 5.85104376e-02 3.51647399e-02 -2.92699039e-01
-3.24974388e-01 9.41195190e-01 7.90025946e-03 -6.02691650e-01
5.77124119e-01 7.03000546e-01 -1.39011806e-02 -3.18632931e-01
3.21767360e-01 6.97796404e-01 5.31700253e-01 3.04655671e-01
1.05653679e+00 8.13804924e-01 8.09558809e-01 -1.52947664e+00
-8.32009077e-01 -4.18500334e-01 -3.63208562e-01 -3.77440572e-01
9.53694224e-01 -1.23223054e+00 -6.29564345e-01 5.52687228e-01
-1.17301250e+00 -3.18318039e-01 -2.02015519e-01 1.12183285e+00
-6.87234521e-01 4.48791742e-01 -1.60568371e-01 -9.86368954e-01
-2.29570463e-01 -5.66636980e-01 1.14524984e+00 -4.10357751e-02
6.08737946e-01 -7.15803146e-01 1.80787727e-01 3.72920275e-01
8.34612310e-01 3.36452484e-01 9.07200396e-01 3.28916967e-01
-1.53736341e+00 -3.63141209e-01 -2.04921320e-01 1.69915184e-01
-6.26247972e-02 -1.11699080e+00 -1.04348290e+00 -5.36803007e-01
7.99389899e-01 -3.00411731e-01 9.77337897e-01 9.65884686e-01
7.48869479e-01 -4.24473196e-01 -3.57725322e-02 1.08035445e+00
2.27312684e+00 -1.99680120e-01 7.73307383e-01 -1.20233640e-01
5.27917027e-01 2.75855243e-01 3.22249413e-01 5.25181651e-01
-2.57235970e-02 4.93386090e-01 8.96317601e-01 2.23212659e-01
-5.99906087e-01 3.54020298e-02 -3.51588503e-02 6.58255398e-01
1.37353644e-01 -7.42410421e-01 -6.66914642e-01 6.13025665e-01
-1.36400175e+00 -9.45390940e-01 -3.20287198e-01 2.19640756e+00
1.63059488e-01 -4.27109748e-01 -9.01850760e-01 1.88463867e-01
8.03835243e-02 6.20417655e-01 -2.40737423e-01 4.47472334e-01
-2.97812015e-01 4.86782491e-01 5.98991632e-01 9.35314178e-01
-6.95707083e-01 4.50632244e-01 5.12449265e+00 3.88920814e-01
-1.35622656e+00 -2.09145620e-02 -1.41775414e-01 1.48513541e-01
-7.54927337e-01 1.60452455e-01 -1.02651805e-01 2.58358181e-01
9.61472392e-01 1.83801785e-01 6.36116862e-01 2.01755419e-01
2.76788384e-01 -6.20287061e-01 -8.21975410e-01 1.21891809e+00
7.86368251e-02 -1.75216246e+00 1.54011041e-01 4.24196929e-01
6.08559549e-01 2.26629466e-01 1.29159719e-01 -2.15936825e-01
-3.14318687e-01 -7.81965673e-01 3.47587526e-01 9.74515259e-01
9.14742351e-01 -3.11985761e-01 5.02563417e-01 5.54591000e-01
-7.34232724e-01 -3.98631990e-01 -3.50952893e-01 -2.90303379e-01
3.17386478e-01 1.35860503e+00 -5.97765744e-01 8.47732306e-01
3.65807474e-01 6.70385718e-01 1.26023948e-01 1.09187710e+00
-1.23489857e-01 8.12027991e-01 -3.73095900e-01 2.08389312e-01
5.82301617e-01 -4.50825483e-01 1.15506256e+00 7.23563671e-01
1.12648368e+00 5.81524312e-01 -4.80066985e-02 1.01259375e+00
7.28667155e-02 -5.08914649e-01 -1.00731885e+00 -2.00702414e-01
2.78963715e-01 9.79052305e-01 -2.71393746e-01 1.33900419e-02
-3.48787159e-01 7.23123193e-01 -3.09641957e-01 7.18360543e-01
-2.08791479e-01 -3.86648983e-01 2.72808880e-01 3.82099181e-01
8.94673526e-01 -8.52612257e-01 -4.00491171e-02 -1.02611303e+00
-3.18426564e-02 -6.38618708e-01 -2.13546544e-01 -1.31943953e+00
-7.88005590e-01 2.05329865e-01 -1.99453130e-01 -1.45889735e+00
-6.80148154e-02 -4.93938774e-01 -5.04219711e-01 9.24913704e-01
-2.06173706e+00 -1.15292001e+00 -7.80579150e-01 3.96632582e-01
1.05631702e-01 -2.08161816e-01 9.00182068e-01 1.71973512e-01
4.21661139e-01 -2.67881185e-01 2.94319957e-01 -8.15992728e-02
1.42137602e-01 -8.24371755e-01 -1.31446555e-01 1.00718558e+00
-1.38222473e-02 2.48392031e-01 8.77137899e-01 -7.06240892e-01
-2.06027865e+00 -1.08313620e+00 6.03418291e-01 5.38971961e-01
4.54603553e-01 -2.13889226e-01 -6.28870010e-01 4.50460583e-01
4.96879369e-01 1.82863370e-01 5.86382985e-01 -7.15572119e-01
-8.57471451e-02 -1.84770957e-01 -1.05476570e+00 2.22700730e-01
8.95249188e-01 -5.53696752e-01 -5.32296062e-01 6.39395118e-01
4.76593524e-01 -1.14506692e-01 -5.15565097e-01 3.84689182e-01
3.33474278e-01 -1.20551932e+00 1.00831437e+00 -2.36106850e-02
6.29453182e-01 -3.46061707e-01 -8.17304552e-01 -1.39566052e+00
-6.33437693e-01 -8.47160220e-01 -2.50141144e-01 1.55045331e-01
-2.20911894e-02 -7.62164533e-01 8.74152601e-01 -6.70072258e-01
-5.38404226e-01 -4.95783955e-01 -1.28561509e+00 -5.53965449e-01
-2.81940043e-01 -5.46394825e-01 1.01519056e-01 7.36590922e-01
-4.86546159e-01 1.87227175e-01 -5.14685154e-01 1.15190983e+00
1.55354071e+00 5.68836749e-01 3.84105831e-01 -1.00569177e+00
-8.33910525e-01 2.02681914e-01 -4.25967038e-01 -1.47558665e+00
-4.14162666e-01 -1.07800603e+00 2.29954526e-01 -1.64077508e+00
-6.93740010e-01 -3.91176075e-01 1.41204298e-01 -2.00320333e-01
5.42440891e-01 3.60170335e-01 1.60630159e-02 3.25293094e-01
-5.32290675e-02 8.74631464e-01 1.05890572e+00 -1.46729007e-01
1.08474068e-01 1.45087421e-01 -2.07041148e-02 3.72558981e-01
4.70035285e-01 -5.80003500e-01 -1.96063995e-01 -7.80861437e-01
3.91033053e-01 7.63864875e-01 7.99738407e-01 -1.68983603e+00
6.80574358e-01 2.37644836e-01 3.52952510e-01 -5.77219248e-01
6.40998483e-01 -8.79873514e-01 2.16553405e-01 8.48244786e-01
-1.32628933e-01 -7.52735555e-01 -2.91201651e-01 1.24629056e+00
-2.93041497e-01 -2.37401903e-01 7.96386302e-01 -3.04150194e-01
-3.68030101e-01 1.22515678e-01 -6.02248847e-01 -3.26129466e-01
7.00605452e-01 9.28120036e-03 -2.04565585e-01 -7.89889991e-01
-6.08564138e-01 -3.10859740e-01 1.16096132e-01 -4.81458694e-01
9.09181654e-01 -1.08532012e+00 -8.95575464e-01 6.81258142e-01
-5.17328121e-02 -2.89971292e-01 5.85012078e-01 7.46490479e-01
-8.36467206e-01 6.75813019e-01 -2.65031368e-01 -6.92016125e-01
-6.06878638e-01 2.50288934e-01 6.99034750e-01 4.30577109e-03
-7.29747236e-01 5.98590851e-01 -1.03016654e-02 -3.52263004e-02
-1.98784217e-01 -1.15140505e-01 -3.37095596e-02 -1.40939847e-01
4.99490887e-01 1.95055827e-01 4.54846062e-02 -2.34880641e-01
4.62978194e-03 9.53384697e-01 7.06419647e-01 -4.03372586e-01
1.76438010e+00 -7.32137635e-02 -9.61160287e-02 1.51982963e-01
1.31059158e+00 3.25537860e-01 -1.28012705e+00 -3.55415493e-01
-5.88224471e-01 -5.68865240e-01 9.48397338e-01 -6.59164548e-01
-1.02513218e+00 1.20717800e+00 8.75329971e-01 3.10181230e-01
1.12256300e+00 -1.22060522e-01 7.51761496e-01 8.01294088e-01
6.86397195e-01 -5.58713257e-01 1.82679594e-01 3.34056228e-01
9.87919331e-01 -8.08552027e-01 2.40623087e-01 -2.05486998e-01
3.40257078e-01 1.30140662e+00 -4.30263728e-01 -4.08421099e-01
9.85006392e-01 1.37276053e-01 -6.77568838e-02 -7.54094720e-01
-4.54289764e-01 -4.56590354e-02 9.71246064e-02 6.79472685e-01
-1.78585589e-01 -2.88821589e-02 2.81732343e-02 -1.64046004e-01
2.21266262e-02 7.36520812e-02 7.72263944e-01 7.52620578e-01
-8.20796430e-01 -6.12163424e-01 -3.12398404e-01 3.71193498e-01
-9.28626731e-02 -3.11220586e-01 3.62655640e-01 3.39932323e-01
3.40416729e-02 8.65783036e-01 4.07639183e-02 -8.85720253e-02
1.25562161e-01 -8.01248029e-02 6.22794807e-01 -5.60753644e-01
1.61931738e-01 2.32489139e-01 -4.33217473e-02 -6.47445977e-01
-4.04584736e-01 -5.21786392e-01 -7.31274068e-01 2.93365806e-01
-1.82930216e-01 1.17567785e-01 9.17770147e-01 8.62464666e-01
5.87210536e-01 4.04355437e-01 8.60345066e-01 -9.88750994e-01
-3.86184067e-01 -7.31861532e-01 -7.80579329e-01 -4.54458386e-01
1.00566137e+00 -3.04890633e-01 -9.02273118e-01 -3.37835193e-01] | [10.632256507873535, -2.141759157180786] |
07ff3aa8-28a4-4080-97e7-21bd52ac9ce1 | self-attentive-multi-context-one-class | null | null | https://aclanthology.org/P19-1398 | https://aclanthology.org/P19-1398.pdf | Self-Attentive, Multi-Context One-Class Classification for Unsupervised Anomaly Detection on Text | There exist few text-specific methods for unsupervised anomaly detection, and for those that do exist, none utilize pre-trained models for distributed vector representations of words. In this paper we introduce a new anomaly detection method{---}Context Vector Data Description (CVDD){---}which builds upon word embedding models to learn multiple sentence representations that capture multiple semantic contexts via the self-attention mechanism. Modeling multiple contexts enables us to perform contextual anomaly detection of sentences and phrases with respect to the multiple themes and concepts present in an unlabeled text corpus. These contexts in combination with the self-attention weights make our method highly interpretable. We demonstrate the effectiveness of CVDD quantitatively as well as qualitatively on the well-known Reuters, 20 Newsgroups, and IMDB Movie Reviews datasets. | ['V', 'Yury Zemlyanskiy', 'Thomas Schnake', 'Robert ermeulen', 'Marius Kloft', 'Lukas Ruff'] | 2019-07-01 | null | null | null | acl-2019-7 | ['contextual-anomaly-detection'] | ['miscellaneous'] | [-8.50599818e-03 -9.47733670e-02 -8.27370286e-02 -5.13404191e-01
-3.89653057e-01 -3.34675223e-01 7.35098302e-01 1.06279206e+00
-4.92866367e-01 2.14332983e-01 5.85942864e-01 -4.01054144e-01
2.16722876e-01 -4.93801326e-01 -2.57842422e-01 -3.88815880e-01
-1.87948421e-01 2.15188354e-01 9.13619027e-02 -4.13018823e-01
6.49911880e-01 2.98484951e-01 -1.48498929e+00 2.33147234e-01
6.35037780e-01 8.05531323e-01 -2.67522246e-01 7.85855889e-01
-8.03281844e-01 5.72424591e-01 -6.30614758e-01 -3.67328376e-01
-2.03357618e-02 -2.36059129e-01 -5.82901299e-01 7.50880241e-02
6.26561105e-01 -3.79210413e-01 -5.08349836e-01 1.19486201e+00
1.38786495e-01 4.56711560e-01 1.00765705e+00 -1.14326012e+00
-1.25581932e+00 3.51047248e-01 -6.68331385e-01 9.74507332e-01
3.74334991e-01 -9.27756578e-02 1.58268917e+00 -1.30609024e+00
2.88657308e-01 1.14637804e+00 5.40556431e-01 8.21928263e-01
-1.15104496e+00 -2.55309999e-01 6.92043006e-01 3.58496875e-01
-1.03418028e+00 -1.87396809e-01 9.88225281e-01 -4.33539927e-01
1.53387558e+00 9.33056399e-02 3.60616565e-01 1.50759780e+00
3.76987636e-01 8.85500848e-01 1.25255138e-01 -7.02097893e-01
3.92918319e-01 -1.66132376e-01 9.42484200e-01 5.80456853e-01
5.38860202e-01 -3.70144665e-01 -5.61771631e-01 -5.96774995e-01
1.95019558e-01 3.86964053e-01 1.27839416e-01 -2.67890185e-01
-7.61493981e-01 1.14369678e+00 -7.12972656e-02 6.83826029e-01
-3.75965089e-01 1.30789161e-01 1.02775669e+00 5.75905442e-01
1.07222641e+00 5.61397076e-01 -4.56437230e-01 -3.68060321e-02
-4.44608301e-01 2.73497492e-01 5.66964090e-01 8.13790143e-01
7.19408333e-01 4.20990169e-01 -2.39612505e-01 8.46179545e-01
5.72701573e-01 2.20200494e-01 1.13389218e+00 -3.00821453e-01
3.22773784e-01 8.40883434e-01 -1.75095834e-02 -1.39589024e+00
-3.53916347e-01 -3.17912921e-02 -6.54713273e-01 -1.66749328e-01
1.65242687e-01 1.62409946e-01 -8.71887982e-01 1.63576066e+00
1.19984329e-01 3.35364372e-01 2.54354179e-01 3.91526401e-01
6.55544817e-01 5.74450076e-01 3.09889704e-01 -8.16540122e-02
1.14514601e+00 -7.62123644e-01 -1.02843618e+00 -5.90380788e-01
9.64439809e-01 -4.42056239e-01 1.38257670e+00 2.65970584e-02
-6.68331981e-01 -2.62762189e-01 -1.19088376e+00 -1.15309052e-01
-7.94612229e-01 -5.53814709e-01 3.84365618e-01 3.33723098e-01
-8.74330223e-01 4.13441002e-01 -8.64605367e-01 -7.25315750e-01
3.70026827e-01 -4.42182571e-02 -3.84861022e-01 -7.15138987e-02
-1.07017148e+00 8.47504497e-01 3.20061266e-01 -2.85239309e-01
-6.37783706e-01 -4.06812519e-01 -1.38513148e+00 3.46402004e-02
1.21687017e-01 -2.15695113e-01 1.19770575e+00 -1.00843525e+00
-8.72188807e-01 8.84391904e-01 -4.45013911e-01 -6.03969276e-01
-3.48115057e-01 -5.09151697e-01 -9.10704792e-01 7.43541941e-02
2.25424379e-01 3.93345281e-02 9.62718964e-01 -8.41898322e-01
-3.34507793e-01 -5.09433329e-01 -3.14224988e-01 1.38881326e-01
-1.09269273e+00 2.21347123e-01 -2.09310606e-01 -1.18902695e+00
2.06588224e-01 -2.65276104e-01 -4.20446604e-01 4.41534407e-02
-4.83433545e-01 -7.13550150e-01 1.16399026e+00 -5.26825130e-01
1.50171793e+00 -2.51439309e+00 -2.68412568e-02 3.36265385e-01
2.39606872e-01 1.61786720e-01 -4.67919320e-01 5.62350154e-01
-3.02491575e-01 2.36171514e-01 -3.25425029e-01 -5.93960464e-01
6.94525894e-03 6.48561656e-01 -5.64582944e-01 4.81731206e-01
5.09703398e-01 6.23731375e-01 -1.13823915e+00 -2.07971513e-01
1.28505170e-01 7.39209130e-02 -7.00028300e-01 1.42103896e-01
-3.45880389e-01 -6.75578415e-02 -6.00221574e-01 6.84693992e-01
3.13091427e-01 -1.20889433e-01 -8.28856379e-02 2.81005651e-01
2.76488751e-01 2.40779027e-01 -7.99727201e-01 1.67443919e+00
-1.26554742e-01 7.03865588e-01 -3.57763171e-01 -1.23268437e+00
8.88004124e-01 2.97499835e-01 4.28199500e-01 -5.40056825e-01
1.14905499e-02 1.87011674e-01 -1.64885938e-01 -6.87847495e-01
8.91419470e-01 -4.81857173e-02 -2.32041582e-01 7.16857910e-01
4.30102378e-01 3.37997645e-01 2.52987772e-01 6.69567704e-01
1.39163291e+00 -4.83279645e-01 5.12861967e-01 -2.89371818e-01
5.32341897e-01 -1.44518986e-01 3.75247657e-01 9.58056986e-01
-4.52575415e-01 6.42350733e-01 5.55987298e-01 -6.87117994e-01
-1.22431123e+00 -9.87380624e-01 -1.47468492e-01 1.37835300e+00
-1.10449180e-01 -8.84917855e-01 -2.61002719e-01 -1.00124466e+00
3.44927996e-01 1.10410357e+00 -8.08231473e-01 -4.56613541e-01
-3.17392647e-01 -5.66015422e-01 4.51514214e-01 7.07700551e-01
-2.81789899e-01 -1.29316103e+00 5.14627919e-02 2.74333924e-01
7.99000114e-02 -9.11989152e-01 -5.46347916e-01 2.66635031e-01
-8.49837422e-01 -9.54280734e-01 -2.56310552e-01 -7.69985437e-01
7.12900043e-01 1.96931615e-01 1.15781724e+00 2.26302803e-01
-4.77215439e-01 8.08443666e-01 -6.01185441e-01 -5.52999318e-01
-4.36138511e-01 -2.87074447e-01 6.41760826e-01 -2.93303877e-02
1.34122777e+00 -4.88533258e-01 -2.49359906e-01 -8.92701894e-02
-1.14494443e+00 -8.72442722e-01 1.11027256e-01 1.02171314e+00
5.54439485e-01 -3.59564662e-01 7.49652445e-01 -9.24458504e-01
9.55144942e-01 -1.04261458e+00 -9.96584296e-02 -6.45857975e-02
-8.12711000e-01 5.67825176e-02 5.50187945e-01 -4.42756891e-01
-6.59078181e-01 -5.26223123e-01 -3.11703593e-01 -5.00255942e-01
-5.24402380e-01 5.38532674e-01 6.84417859e-02 4.28533733e-01
1.00084984e+00 2.55636901e-01 -2.03350652e-02 -6.25616372e-01
4.98828828e-01 7.35255778e-01 4.97337401e-01 -4.17455465e-01
7.44626760e-01 3.55482966e-01 -6.02248073e-01 -1.13950992e+00
-9.23818946e-01 -8.84489179e-01 -5.25961459e-01 2.75391877e-01
8.82944107e-01 -7.74176955e-01 1.55341983e-01 1.82075530e-01
-1.09371841e+00 3.01675022e-01 -5.98002076e-01 2.60675311e-01
-2.27085486e-01 9.60619032e-01 -7.06540704e-01 -7.91819394e-01
-4.58416909e-01 -5.08835673e-01 7.03839600e-01 -8.93512815e-02
-7.33943939e-01 -1.28530347e+00 4.43250656e-01 -2.29388461e-01
5.26315749e-01 2.93896180e-02 1.28089499e+00 -1.61674058e+00
2.32389808e-01 -6.65268421e-01 1.19078629e-01 7.01462686e-01
2.74450809e-01 3.49517874e-02 -9.75887179e-01 -2.96631515e-01
-8.38642791e-02 -1.86496049e-01 9.65208471e-01 5.84371500e-02
1.53980494e+00 -4.57956433e-01 -1.89276204e-01 1.66461155e-01
1.16133142e+00 -4.75819111e-02 4.38853920e-01 3.95043552e-01
8.50420594e-01 4.12611037e-01 3.14929485e-01 6.15523636e-01
6.57695904e-02 1.72502905e-01 4.58842546e-01 2.89293259e-01
5.58162868e-01 -2.46926531e-01 4.81417149e-01 1.12441504e+00
4.11073178e-01 -5.43385923e-01 -9.08917367e-01 1.03828704e+00
-1.89121366e+00 -1.06091869e+00 -2.24951915e-02 1.84746563e+00
4.63624835e-01 2.46511057e-01 -1.75283775e-01 1.21427715e-01
6.88809931e-01 4.80183274e-01 -6.64610386e-01 -8.78413856e-01
-1.71377912e-01 1.73983678e-01 -1.17114246e-01 3.34320217e-01
-1.26249444e+00 7.84123361e-01 6.79630899e+00 5.56653798e-01
-4.92709100e-01 -1.44543210e-02 3.35122049e-01 -1.18232280e-01
-7.48076856e-01 -2.17986271e-01 -5.37595272e-01 4.12988216e-01
9.80417848e-01 -3.84867072e-01 -2.84293532e-01 1.14478040e+00
-1.59098998e-01 2.76944727e-01 -1.28561330e+00 8.47574830e-01
5.37464797e-01 -1.08085108e+00 4.37061459e-01 -1.41328350e-01
6.02864921e-01 2.86711276e-01 7.66235292e-02 4.29450393e-01
1.68845728e-01 -7.87987709e-01 6.27281293e-02 2.85457551e-01
2.76652902e-01 -7.34710038e-01 8.81732583e-01 1.94383860e-01
-7.35821784e-01 -2.08676249e-01 -7.40059197e-01 -9.67594013e-02
1.28200389e-02 7.28312552e-01 -5.10559082e-01 3.25211614e-01
7.64919341e-01 1.03144634e+00 -6.26240075e-01 5.72756290e-01
-1.35425538e-01 8.30356896e-01 -1.50250122e-01 -1.63901657e-01
3.96817714e-01 -8.26512426e-02 9.91051674e-01 1.28301632e+00
1.91228941e-01 -2.09067285e-01 6.25052080e-02 5.81977904e-01
-2.09522039e-01 5.94934165e-01 -1.09278238e+00 -4.07712787e-01
2.82829076e-01 9.99469995e-01 -3.15950185e-01 -3.29208225e-01
-8.75046134e-01 1.16645610e+00 4.38019365e-01 5.17194331e-01
-1.77840322e-01 -4.80230659e-01 1.22340238e+00 -1.23364337e-01
2.53200263e-01 -2.61650980e-01 -1.59535810e-01 -1.53895748e+00
1.86181128e-01 -7.39164591e-01 7.76203573e-01 -3.35358500e-01
-2.28338599e+00 5.55965662e-01 -7.83467442e-02 -9.97864544e-01
-4.89048392e-01 -8.34091008e-01 -1.10555935e+00 5.88091493e-01
-1.35042524e+00 -6.26927316e-01 1.68292776e-01 7.44636834e-01
8.02495122e-01 -7.61785030e-01 1.36662745e+00 4.12730090e-02
-7.34764218e-01 7.53746748e-01 3.86322647e-01 3.62661541e-01
7.14556634e-01 -1.46376014e+00 9.16107178e-01 1.01137424e+00
5.34780264e-01 7.54903674e-01 7.67747402e-01 -5.89106202e-01
-9.42927361e-01 -9.14297640e-01 1.14394975e+00 -7.38264382e-01
1.03988516e+00 -4.81738985e-01 -1.60228145e+00 8.63387883e-01
2.17230633e-01 3.37825745e-01 1.22190893e+00 6.03902638e-01
-6.81322575e-01 2.19411984e-01 -1.10289061e+00 6.26658797e-01
8.39027286e-01 -8.68477464e-01 -1.27292621e+00 4.09527928e-01
8.46487522e-01 3.28600965e-02 -2.55331784e-01 8.47150981e-02
-6.97656646e-02 -5.26396096e-01 8.57036114e-01 -1.36788344e+00
4.05369252e-01 -3.30252163e-02 -5.45612574e-01 -1.27541196e+00
-3.97789389e-01 -2.48053059e-01 -6.87308550e-01 1.26568413e+00
3.97395045e-01 -6.12737834e-01 6.18271649e-01 5.57318687e-01
-3.41484904e-01 -6.57627344e-01 -1.02296197e+00 -6.91836953e-01
1.36068419e-01 -7.48803079e-01 2.75288224e-01 1.22289801e+00
4.31746513e-01 3.14284861e-01 -3.04985464e-01 1.72361672e-01
4.73501563e-01 -2.69774228e-01 3.06886792e-01 -1.25262952e+00
4.56998684e-03 -2.94210583e-01 -1.01114249e+00 -8.89482677e-01
5.90030313e-01 -9.62717414e-01 -1.07272096e-01 -1.20061958e+00
6.44939914e-02 1.36160403e-01 -8.22014570e-01 6.22098744e-01
-4.76843387e-01 3.33779007e-02 -4.85727429e-01 1.12359278e-01
-8.10222626e-01 8.74639988e-01 3.41203153e-01 -1.67609930e-01
8.47244933e-02 -3.41599822e-01 -8.70325565e-01 9.76166725e-01
8.42271090e-01 -5.67553818e-01 -2.65159905e-01 -5.61175644e-01
2.26452604e-01 -9.20930088e-01 1.48979828e-01 -6.30852938e-01
1.00257941e-01 1.60840318e-01 4.87308532e-01 -5.99726140e-01
8.64463334e-04 -6.40397549e-01 -1.08867693e+00 1.29020661e-01
-4.90963846e-01 7.46337891e-01 4.25551623e-01 1.31136131e+00
-4.48833346e-01 -4.71967459e-01 4.38489139e-01 -8.78531933e-02
-1.10103369e+00 3.27508301e-01 -8.72063220e-01 3.99706334e-01
8.44156623e-01 1.07957125e-01 -3.69181246e-01 -4.04247165e-01
-7.12412000e-01 4.21081930e-01 2.55425870e-01 9.13367987e-01
9.17107165e-01 -1.54413867e+00 -5.78182518e-01 5.75342715e-01
8.58721614e-01 -9.24700350e-02 2.52871424e-01 9.68898088e-02
-2.19060540e-01 5.15199453e-02 -2.14489903e-02 -5.04857719e-01
-9.53726768e-01 8.27003777e-01 1.60270222e-02 -3.12855430e-02
-8.53339612e-01 7.76461363e-01 1.59953963e-02 -4.34496641e-01
3.76911461e-01 -2.59070367e-01 -3.56450051e-01 9.62983742e-02
7.93021560e-01 -4.58471775e-02 7.67774507e-02 -5.42712748e-01
-5.19168973e-01 1.76301345e-01 -6.25780046e-01 6.95072040e-02
1.22103143e+00 -1.82842463e-01 -7.94469416e-02 9.03017282e-01
1.22861147e+00 -1.27508342e-01 -6.72695518e-01 -7.70023167e-01
5.70338011e-01 -5.15140474e-01 -1.10006250e-01 -2.36026853e-01
-5.42789221e-01 8.26427460e-01 4.47065026e-01 4.38564420e-01
6.91181481e-01 1.81824058e-01 6.71182334e-01 8.05584490e-01
-4.11758035e-01 -1.32902384e+00 6.53358519e-01 9.38307524e-01
6.92138851e-01 -1.37286997e+00 -2.19947934e-01 1.33819774e-01
-7.25838959e-01 1.08106256e+00 9.08536732e-01 -3.37021947e-01
8.14816892e-01 -1.24898277e-01 1.77675068e-01 -4.51785237e-01
-8.32418799e-01 -6.66648448e-02 4.10045981e-01 4.85549331e-01
3.62536669e-01 -3.99389625e-01 -1.60546511e-01 5.57744920e-01
4.28755492e-01 -6.81465924e-01 4.26095426e-01 1.09483230e+00
-6.96552515e-01 -1.06739664e+00 7.20130140e-03 8.20417821e-01
-7.45241642e-01 -2.77446777e-01 -4.17707205e-01 5.02473533e-01
-4.15086955e-01 9.61297691e-01 6.57506227e-01 -2.05768511e-01
3.55851471e-01 8.14686954e-01 -1.65160850e-01 -1.03943396e+00
-3.15618277e-01 -3.30430329e-01 -5.66938370e-02 -5.31672597e-01
-7.91037530e-02 -5.34645319e-01 -1.19190264e+00 -8.72901082e-02
-1.58287451e-01 1.72462016e-01 4.74641949e-01 1.18905926e+00
4.69888151e-01 2.86840618e-01 6.03365481e-01 -4.94489521e-01
-7.58493721e-01 -1.18811667e+00 -6.97669327e-01 1.04299951e+00
5.84380984e-01 -4.07018661e-01 -8.95303726e-01 -3.91998678e-01] | [10.518207550048828, 8.612540245056152] |
14301513-8c41-4566-98b2-93d7809011af | ontology-matching-techniques-a-gold-standard | 1811.10191 | null | http://arxiv.org/abs/1811.10191v1 | http://arxiv.org/pdf/1811.10191v1.pdf | Ontology Matching Techniques: A Gold Standard Model | Typically an ontology matching technique is a combination of much different
type of matchers operating at various abstraction levels such as structure,
semantic, syntax, instance etc. An ontology matching technique which employs
matchers at all possible abstraction levels is expected to give, in general,
best results in terms of precision, recall and F-measure due to improvement in
matching opportunities and if we discount efficiency issues which may improve
with better computing resources such as parallel processing. A gold standard
ontology matching model is derived from a model classification of ontology
matching techniques. A suitable metric is also defined based on gold standard
ontology matching model. A review of various ontology matching techniques
specified in recent research papers in the area was undertaken to categorize an
ontology matching technique as per newly proposed gold standard model and a
metric value for the whole group was computed. The results of the above study
support proposed gold standard ontology matching model. | ['Alok Chauhan', 'Vijayakumar V', 'Layth Sliman'] | 2018-11-26 | null | null | null | null | ['ontology-matching'] | ['knowledge-base'] | [ 3.39670658e-01 1.11946210e-01 -1.44552559e-01 -4.57552105e-01
-2.72172898e-01 -3.85898560e-01 6.80489182e-01 8.12658131e-01
-4.08473432e-01 3.20276648e-01 1.97353378e-01 -9.10520628e-02
-9.64664519e-01 -1.23745906e+00 -4.92572924e-03 3.61568481e-02
-8.65909010e-02 7.47391164e-01 4.62179154e-01 -6.08853638e-01
7.14875400e-01 5.53772271e-01 -2.48403835e+00 3.04315567e-01
6.76051676e-01 8.99859667e-01 1.71412230e-01 2.64618129e-01
-6.88557923e-01 2.46294484e-01 -4.38282549e-01 -4.93474752e-01
4.43129689e-01 -1.37219489e-01 -1.09029388e+00 -3.56157303e-01
3.06041181e-01 5.38046479e-01 -3.69909406e-02 1.33131242e+00
2.80354679e-01 1.41063124e-01 6.30391240e-01 -1.50183868e+00
-3.79393756e-01 3.02838087e-01 1.20933697e-01 3.23930293e-01
1.05893111e+00 -4.93786693e-01 9.68062818e-01 -4.40251231e-01
5.99724650e-01 1.22923148e+00 5.56216598e-01 4.94441278e-02
-8.00543308e-01 -7.88334668e-01 -4.21959132e-01 3.18367958e-01
-1.45793104e+00 -6.16249032e-02 1.24704912e-01 -4.60519820e-01
1.36653996e+00 5.22064924e-01 3.93384784e-01 -8.60523954e-02
4.75015640e-01 -1.86169744e-01 1.05783594e+00 -8.71612310e-01
5.42921126e-02 2.87297726e-01 4.64212269e-01 6.36050344e-01
7.08099425e-01 3.93520236e-01 -3.74985307e-01 -4.10754114e-01
4.15230058e-02 4.20663729e-02 1.27687365e-01 -2.55016088e-01
-7.44910181e-01 6.33258164e-01 1.20694287e-01 8.28984380e-01
-2.53256470e-01 -4.20835942e-01 7.09434271e-01 8.74290526e-01
-2.38277331e-01 4.96707946e-01 -3.96516353e-01 6.16718568e-02
-6.58418536e-01 6.56330287e-01 1.01738083e+00 1.23727417e+00
1.02971101e+00 -3.04684669e-01 2.40040407e-01 7.48984098e-01
5.39139092e-01 5.42123169e-02 6.35227084e-01 -8.62389326e-01
2.66904444e-01 1.46294510e+00 8.35790113e-02 -9.06840324e-01
-4.56794828e-01 -4.24952544e-02 -1.17556944e-01 4.90433723e-01
1.03789970e-01 3.80466610e-01 -8.65481973e-01 1.43636274e+00
3.52170944e-01 -2.43481874e-01 5.83300412e-01 3.01200032e-01
1.09462905e+00 3.03425372e-01 5.70075333e-01 1.11222818e-01
1.87138879e+00 -3.64026725e-01 -6.74889207e-01 2.24528864e-01
8.19554746e-01 -1.24062467e+00 4.57092404e-01 2.49219596e-01
-8.25520217e-01 -5.51739693e-01 -1.35489237e+00 1.32527038e-01
-1.26640713e+00 -8.38380635e-01 7.02359557e-01 9.59966123e-01
-9.39710379e-01 8.10392737e-01 -5.10086678e-02 -1.23667395e+00
-1.83038399e-01 6.63388729e-01 -6.56387508e-01 -4.50837761e-02
-1.48260224e+00 1.36087859e+00 9.97998178e-01 -6.53974354e-01
3.64933386e-02 -5.14289618e-01 -7.63335228e-01 5.52182943e-02
-2.34333426e-02 -6.82042360e-01 8.68542254e-01 -6.87274992e-01
-8.89197826e-01 1.27030540e+00 3.34281311e-03 -4.31288540e-01
1.28570497e-01 4.76580173e-01 -1.14436734e+00 -5.90049895e-03
2.25339323e-01 3.18433851e-01 -5.42153306e-02 -7.90761650e-01
-1.44579566e+00 -4.84517515e-01 2.59045184e-01 1.14971250e-01
-1.29352659e-01 5.46693623e-01 2.59640157e-01 -2.92845100e-01
3.95770609e-01 -4.93611753e-01 -4.46751937e-02 -3.04829717e-01
7.22233117e-01 -4.96561557e-01 5.60025513e-01 -2.47950673e-01
1.55565715e+00 -1.72307026e+00 -2.99615920e-01 6.13736629e-01
-1.70241818e-01 2.05988809e-01 6.42099008e-02 1.00863910e+00
-3.32155496e-01 1.58779994e-01 -5.89449704e-02 5.46380997e-01
3.82847875e-01 4.74640191e-01 2.20030323e-01 1.99062109e-01
-3.44377100e-01 3.16772372e-01 -7.74635255e-01 -7.20740438e-01
4.60260928e-01 1.71424061e-01 -2.64878362e-01 -1.27448200e-03
1.82010457e-01 -3.45077336e-01 -4.18031812e-01 9.13023710e-01
4.82534558e-01 1.39754012e-01 3.09220046e-01 -1.26380786e-01
-1.17615640e-01 5.25755100e-02 -1.58856547e+00 1.52109432e+00
-5.99531651e-01 9.89803597e-02 -3.74941796e-01 -1.29080427e+00
1.56250381e+00 1.19581127e+00 8.00401568e-01 -9.41355586e-01
2.17691585e-01 9.01522040e-01 2.56944686e-01 -7.03440964e-01
3.96449715e-01 -3.05488288e-01 -3.54717448e-02 2.08750144e-01
2.06698939e-01 1.81990508e-02 4.75965261e-01 -3.21848214e-01
1.12777054e+00 9.83686298e-02 1.06215382e+00 -9.48289096e-01
1.00903928e+00 1.92808285e-01 3.43849212e-01 5.92294216e-01
-3.59875321e-01 -7.52544776e-02 -9.84974876e-02 -8.01411271e-01
-1.12072551e+00 -8.13779175e-01 -6.97272956e-01 8.39035630e-01
2.50461787e-01 -6.19230866e-01 -4.64433759e-01 -1.68869063e-01
3.68900374e-02 2.22122520e-01 -2.64408380e-01 -3.52469869e-02
-2.81910032e-01 -5.64441562e-01 5.35706699e-01 -5.61019108e-02
3.99503559e-01 -1.20659602e+00 -9.27221119e-01 5.03824890e-01
3.81049365e-01 -1.10269976e+00 5.25312185e-01 -5.97423092e-02
-1.09367573e+00 -1.21195817e+00 -1.01012580e-01 -9.04685140e-01
2.79409587e-01 1.40127420e-01 1.34638774e+00 6.90262318e-01
-3.11404854e-01 6.66510910e-02 -7.47043669e-01 -8.46676767e-01
-5.57730377e-01 1.02345958e-01 -8.27272683e-02 -3.57234329e-01
1.24333405e+00 -8.16752911e-01 -2.53753304e-01 1.17122672e-01
-1.11636555e+00 -5.81713140e-01 2.84807503e-01 4.18880135e-01
4.15056348e-02 5.53336203e-01 5.89458883e-01 -6.78063631e-01
6.83708787e-01 -7.08222985e-01 -6.79628670e-01 3.76390845e-01
-1.07111824e+00 2.21257567e-01 1.63793623e-01 1.04704089e-01
-7.50326216e-01 -2.40229934e-01 -4.67997715e-02 5.73527098e-01
-3.95866096e-01 4.71698523e-01 -2.14832827e-01 -5.18036306e-01
5.64299226e-01 -2.04968899e-01 -7.86825493e-02 -5.81354439e-01
-3.54217559e-01 9.18259203e-01 3.26360852e-01 -1.04905438e+00
7.19661832e-01 3.32210511e-01 3.48194629e-01 -5.29156029e-01
-2.69558698e-01 -8.29497933e-01 -3.90823722e-01 -1.16213202e-01
7.50727415e-01 -4.23242509e-01 -7.90824831e-01 -1.80887822e-02
-9.84169126e-01 7.19021976e-01 6.76956624e-02 5.50154567e-01
-5.41655481e-01 2.66457230e-01 8.08814093e-02 -8.24419498e-01
-5.37143886e-01 -9.22384858e-01 6.82743549e-01 4.34343666e-01
-6.43619359e-01 -9.66089964e-01 5.57262525e-02 2.56935239e-01
4.26376253e-01 3.19527209e-01 1.31173444e+00 -9.62791860e-01
-2.70324856e-01 -9.06077743e-01 -1.57017097e-01 -1.03639126e-01
2.89712369e-01 -2.26370454e-01 -5.86349487e-01 9.79518145e-02
-2.83912927e-01 3.89190108e-01 6.01052903e-02 -2.42583737e-01
4.98422652e-01 1.06573664e-01 -4.89001244e-01 2.38204002e-01
2.25461721e+00 8.80198359e-01 8.95009100e-01 1.15017807e+00
-8.13862085e-02 1.02277589e+00 9.35607374e-01 1.30314440e-01
2.21969604e-01 8.67626965e-01 1.41830727e-01 3.28896314e-01
1.90804482e-01 1.34342358e-01 -3.17424417e-01 5.34018576e-01
-2.99431205e-01 -5.16438447e-02 -1.23112416e+00 4.96580213e-01
-1.68811071e+00 -1.14974928e+00 -1.95794702e-01 2.39366603e+00
2.97981232e-01 1.79346263e-01 2.38177627e-01 6.73849523e-01
8.44426751e-01 -2.69329607e-01 3.83447945e-01 -9.50128853e-01
3.97261642e-02 8.70373845e-01 4.59064811e-01 7.44099736e-01
-5.76883197e-01 6.89585567e-01 6.75917244e+00 3.63062888e-01
-6.76358521e-01 1.63377076e-01 -5.82521439e-01 4.43005651e-01
-2.12705001e-01 5.47532499e-01 -7.34836102e-01 5.72769463e-01
1.05840254e+00 -9.02079344e-01 1.99635029e-01 5.42742729e-01
-6.51147738e-02 -9.40055102e-02 -8.47743690e-01 9.29954112e-01
-2.34854922e-01 -1.28035617e+00 3.85688871e-01 3.31928521e-01
5.12601256e-01 -1.66033104e-01 -7.78355718e-01 1.10615186e-01
2.94633508e-01 -9.46998537e-01 3.70017588e-01 5.63376904e-01
1.67831331e-01 -7.98713446e-01 1.18093407e+00 -6.76348731e-02
-1.53986800e+00 -1.46769270e-01 -2.61009634e-01 -3.30573946e-01
1.05071642e-01 -4.67992947e-02 -2.78881788e-01 1.20835304e+00
6.84344292e-01 1.16293527e-01 -3.05438370e-01 1.33427095e+00
4.12309200e-01 -1.72116801e-01 -4.26086932e-01 4.00763564e-02
2.28980616e-01 -4.64389652e-01 4.79019940e-01 1.29801846e+00
6.47867978e-01 1.90222815e-01 -7.83426538e-02 3.49456519e-01
6.52029932e-01 6.53515160e-01 -7.27011025e-01 1.94746226e-01
9.27506983e-01 8.69906306e-01 -6.26060188e-01 -5.87202072e-01
-6.59915030e-01 3.04997474e-01 -3.00960988e-01 -9.92696658e-02
-2.48467714e-01 -7.98844397e-01 9.17197764e-01 3.88715714e-01
-2.94377357e-01 1.96100920e-01 -2.30386168e-01 -6.43596649e-01
-6.16100291e-03 -9.12828147e-01 9.44467127e-01 -4.29594010e-01
-9.65771317e-01 6.99700415e-01 5.06146371e-01 -1.20871353e+00
-2.17806160e-01 -7.73425937e-01 -2.85207897e-01 1.22132516e+00
-1.23109698e+00 -8.40729952e-01 -4.83352005e-01 3.00468475e-01
1.83802634e-01 -6.39648080e-01 1.42816484e+00 7.62672186e-01
-2.61653005e-03 2.19061077e-01 -3.69305760e-01 -3.47163707e-01
3.32597345e-01 -1.00365508e+00 1.08685851e-01 6.43741667e-01
-2.02557176e-01 7.81531572e-01 1.01615131e+00 -4.70472604e-01
-7.73868382e-01 -3.50810885e-01 1.59309173e+00 -2.92498022e-01
5.73269784e-01 2.59317696e-01 -9.48555708e-01 4.01732534e-01
1.39966786e-01 -2.16202989e-01 9.21963990e-01 -4.83450256e-02
-3.22105974e-01 -3.64001781e-01 -1.59733057e+00 3.11790198e-01
1.02223849e+00 -4.03096557e-01 -1.15127647e+00 1.11174203e-01
1.66503504e-01 2.57390877e-03 -1.71352923e+00 6.48798883e-01
9.14612174e-01 -1.17260325e+00 9.87980962e-01 -8.21935892e-01
-1.41249791e-01 -6.77326441e-01 -7.05300570e-01 -4.24475551e-01
-2.45224252e-01 -5.28060317e-01 3.33404124e-01 1.21784210e+00
3.12045544e-01 -9.31245983e-01 5.28322339e-01 6.83046341e-01
8.89724717e-02 -4.00433868e-01 -9.40062344e-01 -1.05734551e+00
-9.62888673e-02 -2.93786436e-01 1.30244184e+00 1.06593406e+00
3.83405179e-01 8.78023505e-02 2.81711817e-01 1.11601681e-01
7.74219751e-01 -1.10980719e-01 4.83400673e-01 -2.15128899e+00
1.89450040e-01 -5.59563875e-01 -1.40456879e+00 2.84204096e-01
1.32382706e-01 -9.53574121e-01 -6.35218322e-01 -1.75845873e+00
-6.82210252e-02 -5.77069283e-01 -5.70396304e-01 1.91230044e-01
1.48028180e-01 3.81678231e-02 1.10248663e-01 1.95112839e-01
1.64522350e-01 -4.53565508e-01 7.00417399e-01 1.84524283e-01
1.91730946e-01 -1.56730622e-01 -4.71647710e-01 5.91630399e-01
7.34949768e-01 -7.89734662e-01 -5.48233449e-01 -1.57629065e-02
3.61402094e-01 -6.44889250e-02 -6.89851567e-02 -1.36807668e+00
1.19063951e-01 -3.68255734e-01 -2.80436069e-01 3.70963253e-02
5.55761196e-02 -1.53109741e+00 9.67434704e-01 7.48020053e-01
-8.37136582e-02 4.43646371e-01 -1.21681862e-01 4.10548635e-02
-4.72089469e-01 -1.00922954e+00 8.90952647e-01 -4.68924493e-01
-1.34922957e+00 9.35968533e-02 -2.04278812e-01 -3.24343830e-01
1.15519869e+00 -1.13720655e+00 3.01685389e-02 1.58401594e-01
-7.12166786e-01 -7.51523599e-02 6.18361831e-01 6.98251247e-01
-1.40281469e-01 -1.46621382e+00 -4.76076722e-01 -1.52125046e-01
8.35171998e-01 -8.50934565e-01 -2.30844751e-01 2.11601436e-01
-8.79544258e-01 7.66621232e-01 -8.28278720e-01 -1.09683514e-01
-1.41260934e+00 5.97869277e-01 5.41716576e-01 -2.03739181e-01
-6.46587849e-01 5.64156882e-02 -4.00278568e-01 -5.23276985e-01
3.82885523e-02 7.91617297e-03 -5.24803698e-01 4.68413997e-03
5.42155743e-01 5.51225841e-01 4.06537801e-01 -8.31574023e-01
-6.50875330e-01 1.02120852e+00 4.38057065e-01 -2.02427924e-01
9.80733931e-01 -1.14917144e-01 -3.83103549e-01 1.48319960e-01
1.05636895e+00 -4.12207603e-01 1.54857188e-01 -1.77259818e-01
1.00648606e+00 -6.71346724e-01 -2.05751836e-01 -5.89873135e-01
-5.03957391e-01 2.83175051e-01 1.09010017e+00 6.22646928e-01
1.06871974e+00 -3.38619083e-01 4.86513913e-01 1.89446792e-01
8.19376290e-01 -1.27074742e+00 -7.93717265e-01 2.57423460e-01
5.05698740e-01 -9.76669550e-01 1.40400201e-01 -5.35836101e-01
4.08979692e-02 1.48196948e+00 3.88721615e-01 -2.98253298e-01
9.69188035e-01 3.56343031e-01 1.44742519e-01 -7.07141280e-01
-4.62493330e-01 -4.71879959e-01 2.34017059e-01 9.01053965e-01
8.72530639e-01 -2.00655200e-02 -1.37781870e+00 1.26755595e-01
-3.77474546e-01 3.17038894e-01 8.80573913e-02 1.22580385e+00
-9.48656857e-01 -1.74989223e+00 -5.50652862e-01 3.87201846e-01
-4.87663388e-01 1.10549606e-01 -8.47949311e-02 1.11742997e+00
6.16448641e-01 1.08255672e+00 2.18382016e-01 -2.74140537e-01
7.58846104e-01 2.57605314e-01 6.53638959e-01 -5.82060754e-01
-9.11486804e-01 -7.75275290e-01 4.12851453e-01 -3.97082090e-01
-8.77370000e-01 -4.98808682e-01 -1.30254281e+00 -5.66837549e-01
-2.01467782e-01 6.10629082e-01 8.20985675e-01 1.37435555e+00
1.64414376e-01 2.10422143e-01 -1.02564707e-01 -1.60335019e-01
-1.65472776e-01 -8.10940504e-01 -5.92132986e-01 7.26904154e-01
-2.15698630e-01 -9.77608025e-01 -7.52651542e-02 -3.80284414e-02] | [9.205137252807617, 8.0642671585083] |
0c8f74fd-4998-4a37-8a39-85fc891c292e | label-only-membership-inference-attack | 2207.13766 | null | https://arxiv.org/abs/2207.13766v1 | https://arxiv.org/pdf/2207.13766v1.pdf | Label-Only Membership Inference Attack against Node-Level Graph Neural Networks | Graph Neural Networks (GNNs), inspired by Convolutional Neural Networks (CNNs), aggregate the message of nodes' neighbors and structure information to acquire expressive representations of nodes for node classification, graph classification, and link prediction. Previous studies have indicated that GNNs are vulnerable to Membership Inference Attacks (MIAs), which infer whether a node is in the training data of GNNs and leak the node's private information, like the patient's disease history. The implementation of previous MIAs takes advantage of the models' probability output, which is infeasible if GNNs only provide the prediction label (label-only) for the input. In this paper, we propose a label-only MIA against GNNs for node classification with the help of GNNs' flexible prediction mechanism, e.g., obtaining the prediction label of one node even when neighbors' information is unavailable. Our attacking method achieves around 60\% accuracy, precision, and Area Under the Curve (AUC) for most datasets and GNN models, some of which are competitive or even better than state-of-the-art probability-based MIAs implemented under our environment and settings. Additionally, we analyze the influence of the sampling method, model selection approach, and overfitting level on the attack performance of our label-only MIA. Both of those factors have an impact on the attack performance. Then, we consider scenarios where assumptions about the adversary's additional dataset (shadow dataset) and extra information about the target model are relaxed. Even in those scenarios, our label-only MIA achieves a better attack performance in most cases. Finally, we explore the effectiveness of possible defenses, including Dropout, Regularization, Normalization, and Jumping knowledge. None of those four defenses prevent our attack completely. | ['Jing Xu', 'Stjepan Picek', 'Jiaxin Li', 'Mauro Conti'] | 2022-07-27 | null | null | null | null | ['inference-attack', 'membership-inference-attack'] | ['adversarial', 'computer-vision'] | [ 1.72583729e-01 4.74540353e-01 -6.10126972e-01 -1.32519737e-01
-3.37616801e-01 -7.39313245e-01 2.42353037e-01 3.59014571e-01
-3.58159155e-01 6.50351644e-01 -3.85099918e-01 -9.36118484e-01
-1.67870313e-01 -1.30497694e+00 -9.81045306e-01 -5.39037287e-01
-4.57231164e-01 2.85180181e-01 3.57297093e-01 2.31036246e-02
-1.73096642e-01 7.99038708e-01 -7.25046933e-01 1.15396053e-01
5.89535713e-01 8.86807680e-01 -6.17515326e-01 6.32570326e-01
1.86675683e-01 6.84178591e-01 -7.00243711e-01 -8.02289307e-01
3.64772499e-01 -1.26266927e-02 -7.46664762e-01 -4.69023466e-01
2.03469649e-01 -3.39322001e-01 -9.44462061e-01 1.17571330e+00
2.37083390e-01 -4.66062814e-01 5.52403927e-01 -1.74609613e+00
-3.35598171e-01 1.13993073e+00 -3.39619815e-01 1.70329753e-02
6.18824959e-02 3.95377636e-01 1.00893986e+00 3.10698077e-02
6.65950179e-01 1.06399512e+00 8.41279030e-01 6.35831952e-01
-1.34669590e+00 -1.21265507e+00 6.06574528e-02 -1.11134984e-01
-1.52864361e+00 -1.09098569e-01 6.54110372e-01 -3.79702449e-02
4.35386330e-01 5.46298504e-01 2.11383298e-01 1.41737890e+00
1.04596995e-01 5.54696739e-01 5.36927640e-01 -6.72876015e-02
2.57330418e-01 3.57872188e-01 3.13040406e-01 7.74660408e-01
6.06767893e-01 4.04911548e-01 -5.99814765e-03 -9.46595669e-01
5.10279536e-01 1.19741865e-01 -2.89880008e-01 -1.93651676e-01
-7.28391171e-01 9.70502198e-01 7.52098083e-01 -7.17433617e-02
-2.00059459e-01 4.25292283e-01 4.30988550e-01 4.45981592e-01
2.06114292e-01 3.01933050e-01 -6.18713260e-01 4.84681606e-01
-7.42960274e-01 -8.77982602e-02 1.20190191e+00 6.45744324e-01
6.30124688e-01 -1.02194041e-01 -2.60583580e-01 -1.37788281e-01
2.14336172e-01 4.88973469e-01 -5.00399098e-02 -4.55525190e-01
4.30239111e-01 6.81367338e-01 -1.62466198e-01 -1.10642350e+00
-6.19676590e-01 -8.83135319e-01 -1.01134729e+00 -1.12437375e-01
4.67769802e-01 -6.40279830e-01 -7.93046832e-01 2.16462421e+00
2.76780188e-01 6.64598346e-01 6.07392602e-02 5.35603344e-01
8.58734190e-01 3.44633371e-01 3.63197654e-01 2.70538274e-02
1.21129167e+00 -3.84665370e-01 -2.22708151e-01 5.99797070e-03
1.09174109e+00 -9.14185569e-02 5.21817386e-01 5.31933978e-02
-5.80140710e-01 -1.13149416e-02 -9.69531953e-01 4.91844118e-01
-4.59003001e-01 -3.98152173e-02 8.73953581e-01 1.16895807e+00
-9.38550115e-01 6.98602378e-01 -9.66956496e-01 -1.85306832e-01
6.26538634e-01 7.27225602e-01 -4.52305973e-01 1.41626047e-02
-1.65964866e+00 3.76163542e-01 3.72027695e-01 -9.56011713e-02
-1.11215925e+00 -7.10466206e-01 -7.33401656e-01 4.19992477e-01
5.11850834e-01 -5.94750762e-01 6.49948180e-01 -7.03798413e-01
-7.76075006e-01 3.93486589e-01 9.36816633e-02 -8.37532878e-01
5.71863413e-01 2.86942810e-01 -5.50926685e-01 2.09767565e-01
-3.71817112e-01 4.36167300e-01 5.77279747e-01 -1.03122735e+00
-1.98658824e-01 -1.75534382e-01 5.66404581e-01 -4.46428984e-01
-3.87005001e-01 -1.43019333e-01 -3.07239056e-01 -2.52842009e-01
-1.18029818e-01 -1.00570869e+00 -4.99987394e-01 1.56694755e-01
-1.02707303e+00 -1.33168206e-01 9.39703465e-01 -3.33211213e-01
1.36281157e+00 -2.25654483e+00 -3.54459167e-01 9.39336896e-01
3.69004607e-01 6.06037021e-01 -2.41196141e-01 4.79934096e-01
-5.08035682e-02 6.14497304e-01 2.53174175e-02 -2.40048304e-01
-1.31071895e-01 2.25703448e-01 -3.39171082e-01 5.75712740e-01
5.57558669e-04 8.68010402e-01 -5.02019465e-01 -3.42363477e-01
-9.34877992e-02 4.18228358e-01 -7.07531869e-01 -5.63336574e-02
-4.02885884e-01 3.30514967e-01 -6.47054970e-01 4.59476292e-01
6.84544563e-01 -5.86562455e-01 6.46321893e-01 -3.01708311e-01
7.88943827e-01 4.03956354e-01 -1.18719864e+00 7.38490760e-01
-2.06922203e-01 3.06563467e-01 -1.88065153e-02 -7.75008857e-01
7.67086983e-01 4.16595995e-01 2.13386491e-01 -5.74789718e-02
1.52570829e-01 4.39435430e-02 3.18416089e-01 -1.06408574e-01
-1.52883261e-01 2.33836636e-01 9.88193452e-02 6.52453959e-01
-3.12061608e-01 6.94047093e-01 -1.41600579e-01 6.28682733e-01
1.51334691e+00 -6.99035287e-01 5.21704331e-02 -3.57038341e-02
3.41228276e-01 -3.99826765e-01 6.54010355e-01 1.31237531e+00
-1.24266371e-01 1.55114338e-01 1.05195069e+00 -4.99200672e-01
-6.21601045e-01 -9.89346087e-01 -9.46727842e-02 7.02400625e-01
2.77103573e-01 -5.95867097e-01 -6.40875638e-01 -1.23278701e+00
1.95071831e-01 5.98476350e-01 -7.96205997e-01 -6.21598601e-01
-3.40903461e-01 -1.02485490e+00 1.38354909e+00 3.52119178e-01
4.43169594e-01 -8.22546959e-01 -1.30660713e-01 8.12139809e-02
-9.46600735e-02 -1.08949804e+00 -1.58079863e-01 1.97003037e-01
-5.63379824e-01 -1.46747673e+00 2.12485865e-01 -3.59688282e-01
8.09993088e-01 -1.92910641e-01 8.33291769e-01 5.57169080e-01
-6.44955784e-02 1.91051394e-01 -8.09113905e-02 -2.26973012e-01
-7.06888556e-01 5.76302171e-01 1.03121608e-01 7.85304680e-02
2.13982791e-01 -8.77897978e-01 -5.94525099e-01 3.34823132e-01
-9.46288228e-01 -3.88027787e-01 5.55143297e-01 4.33380634e-01
1.06809631e-01 3.90519619e-01 6.18359685e-01 -1.54005873e+00
5.38809717e-01 -7.89231300e-01 -6.03903294e-01 3.64372522e-01
-5.55812657e-01 1.00813329e-01 8.78284156e-01 -8.42416525e-01
-2.85990834e-01 -3.91943380e-02 -1.61741063e-01 -3.85265738e-01
-1.41494051e-01 5.37824571e-01 -4.73574430e-01 -4.41056699e-01
8.31479073e-01 -3.75674516e-02 9.41994190e-02 -2.15940729e-01
4.07094285e-02 3.78879875e-01 1.39867485e-01 -4.84710693e-01
1.06446517e+00 4.80312616e-01 4.48941588e-01 -3.73552471e-01
-7.08156586e-01 4.94853407e-02 -4.00220044e-02 1.56951770e-01
4.07576114e-01 -6.24203444e-01 -1.25040758e+00 5.03611088e-01
-9.63670194e-01 -2.80547172e-01 5.53582646e-02 4.03175831e-01
3.07163745e-01 4.67798293e-01 -1.06252337e+00 -6.28465474e-01
-4.89381939e-01 -1.22410083e+00 2.97836095e-01 2.45400220e-01
-6.06135316e-02 -1.22222805e+00 -5.30706584e-01 4.59186323e-02
4.27076906e-01 4.57602113e-01 1.23117900e+00 -1.39919949e+00
-6.64220035e-01 -5.56253731e-01 -4.27983254e-01 1.16609633e-01
-1.28298894e-01 8.02122429e-02 -9.61529076e-01 -5.70908666e-01
-3.51281255e-01 -2.04014510e-01 8.15702140e-01 3.39500219e-01
1.42889071e+00 -8.21377635e-01 -7.23207176e-01 9.07065034e-01
1.26821339e+00 -9.57835913e-02 5.87675989e-01 2.36229822e-02
7.95324206e-01 1.80639669e-01 9.32427198e-02 3.50597829e-01
2.49922141e-01 4.14099872e-01 9.40479398e-01 -2.57445991e-01
2.66640037e-01 -5.62058747e-01 3.06055546e-01 -7.18020052e-02
3.63330960e-01 -6.72934890e-01 -6.88405812e-01 4.67995740e-02
-1.44935977e+00 -7.38626480e-01 6.36876468e-03 2.14012766e+00
7.83854067e-01 5.55218756e-01 8.90356451e-02 1.82370231e-01
7.51987696e-01 2.39527836e-01 -7.17559397e-01 -3.61194223e-01
6.70508519e-02 1.90604165e-01 1.16111410e+00 5.31984508e-01
-1.19467676e+00 7.35272110e-01 5.56852579e+00 1.01385713e+00
-1.14550471e+00 -1.53141662e-01 8.46590221e-01 2.21886709e-01
-4.23634440e-01 1.39280751e-01 -8.50884199e-01 3.99554908e-01
1.17739713e+00 1.99363511e-02 4.89533275e-01 9.21381295e-01
-2.99947292e-01 3.31068158e-01 -1.14293587e+00 4.24879551e-01
-2.36645490e-01 -1.43012738e+00 7.39094317e-02 5.72246075e-01
4.04253542e-01 1.14854820e-01 2.36936048e-01 3.82191747e-01
7.81272650e-01 -1.16486335e+00 2.31681764e-01 6.17242120e-02
8.83214533e-01 -9.62498367e-01 9.27206337e-01 7.33259976e-01
-9.11702931e-01 -2.23087445e-01 -2.87096202e-01 1.38940051e-01
-1.76483005e-01 5.55830777e-01 -9.54318583e-01 6.41327441e-01
2.59146184e-01 2.12981075e-01 -7.31799662e-01 7.41944611e-01
-3.50674838e-01 1.26289713e+00 -7.78672636e-01 -6.55477270e-02
4.01043832e-01 2.61412352e-01 6.86562479e-01 9.57788885e-01
-1.22834869e-01 1.12807952e-01 4.13471133e-01 8.09024692e-01
-4.50457633e-01 -2.54620790e-01 -4.79933053e-01 -2.10777372e-02
8.52181792e-01 1.14803767e+00 -5.35668552e-01 -8.62435251e-02
-7.23376572e-02 3.81001472e-01 3.63222122e-01 4.79846656e-01
-9.89671946e-01 -4.06461775e-01 7.29108036e-01 3.17202866e-01
9.35673192e-02 1.99537814e-01 -1.71687081e-01 -9.24712300e-01
-2.15465054e-01 -8.89108241e-01 8.25156271e-01 -1.97213724e-01
-1.31196165e+00 5.74894190e-01 -1.44574478e-01 -8.42992485e-01
-1.11290112e-01 -4.60041106e-01 -9.06473517e-01 7.02660203e-01
-1.23113883e+00 -1.34028471e+00 1.66319206e-01 6.62729263e-01
-4.72436756e-01 -6.30303696e-02 8.76580000e-01 4.04672891e-01
-6.70323193e-01 1.60486817e+00 -1.40928254e-01 8.99373472e-01
3.98626179e-01 -7.49004006e-01 3.92389566e-01 9.88151968e-01
2.95778751e-01 6.25781775e-01 5.56430101e-01 -8.21358383e-01
-1.24368215e+00 -1.28925204e+00 6.17961943e-01 -3.24739903e-01
5.90923131e-01 -4.87509608e-01 -9.03519630e-01 9.46819782e-01
-7.19875515e-01 4.27332073e-01 7.80976117e-01 2.50289947e-01
-8.17644358e-01 -2.02659547e-01 -1.60235953e+00 6.81206346e-01
9.31363702e-01 -5.54052472e-01 2.50959814e-01 2.34352127e-01
9.86824512e-01 -3.31548393e-01 -7.90059268e-01 6.16153657e-01
5.11048198e-01 -6.63155675e-01 1.06347179e+00 -1.07641363e+00
-4.02282290e-02 -7.95737654e-02 9.63425487e-02 -9.20930207e-01
-2.21861169e-01 -5.84096074e-01 -2.52511829e-01 1.17166543e+00
8.13039601e-01 -1.13390779e+00 1.14307177e+00 7.49866605e-01
5.41702509e-01 -7.66949654e-01 -9.47966933e-01 -5.94897270e-01
-9.71679389e-03 -3.24344277e-01 1.03147233e+00 1.00401318e+00
-2.25725099e-01 4.50682975e-02 -4.84565288e-01 8.34592521e-01
7.01826513e-01 -3.35574076e-02 7.37816870e-01 -1.34014177e+00
-3.97657871e-01 -1.11418687e-01 -5.94823837e-01 -4.12913054e-01
4.79568094e-01 -1.09160864e+00 -7.44884729e-01 -9.58097696e-01
9.78624374e-02 -9.00844276e-01 -5.11187196e-01 8.69166672e-01
-4.21414748e-02 1.49052009e-01 2.50944138e-01 9.52241868e-02
-3.78904194e-01 -4.43238020e-02 8.61488342e-01 -2.77365655e-01
-2.85700232e-01 6.14666641e-01 -7.56500483e-01 7.00525463e-01
1.03287125e+00 -8.35143685e-01 -5.34934402e-01 1.05979413e-01
2.87593424e-01 2.12678254e-01 6.87601626e-01 -8.11772645e-01
3.38354945e-01 3.16865146e-02 2.97073096e-01 -3.30189615e-01
1.87000439e-01 -9.85301077e-01 4.99367297e-01 9.65782940e-01
-5.33368409e-01 -3.32396626e-01 -3.07060666e-02 8.59163642e-01
3.53150368e-01 -1.39962047e-01 6.46984816e-01 1.23384342e-01
2.38565039e-02 8.15707803e-01 -1.29352078e-01 6.12610690e-02
9.02997732e-01 3.19817886e-02 -8.31270516e-01 -5.79574347e-01
-7.76866436e-01 2.45547980e-01 3.89572948e-01 -4.64293640e-03
3.18777204e-01 -8.36541593e-01 -6.61898196e-01 4.20357108e-01
-6.28663078e-02 -2.62800127e-01 -5.93382446e-03 6.01113737e-01
-3.29263657e-01 1.12270743e-01 1.14853621e-01 -2.13918760e-01
-1.30866778e+00 6.59252107e-01 4.61677283e-01 -6.53360188e-01
-4.89033699e-01 6.43564641e-01 1.10930040e-01 -6.49022639e-01
5.27841866e-01 -1.33539326e-02 1.88427549e-02 -4.25189078e-01
2.76119888e-01 2.25501448e-01 -1.76050663e-01 -2.95884013e-01
-4.68626320e-01 -1.47538185e-01 -3.37904960e-01 3.88627738e-01
9.06147480e-01 2.70536989e-01 -1.17304344e-02 -1.25568181e-01
1.04254889e+00 1.16599098e-01 -8.26428175e-01 -3.86205941e-01
-1.92952737e-01 -1.52104318e-01 -1.77858174e-01 -8.38296413e-01
-1.41213834e+00 7.67289162e-01 2.97475755e-01 2.72152543e-01
8.72629702e-01 5.04577681e-02 8.34913790e-01 3.69604051e-01
4.89518434e-01 -2.53399193e-01 -4.09121394e-01 1.48467600e-01
-1.74117293e-02 -9.94794428e-01 4.55814525e-02 -7.18756676e-01
-2.52248585e-01 7.69753575e-01 7.46660113e-01 -5.19922841e-03
1.01575732e+00 2.04672739e-01 -1.59263760e-01 -2.76317596e-02
-9.94620442e-01 1.52380213e-01 4.55397144e-02 5.19938290e-01
-2.67248482e-01 3.14900935e-01 2.18465943e-02 9.41195905e-01
-3.74044925e-02 -3.19906503e-01 4.80345905e-01 4.98518765e-01
-1.53391615e-01 -1.02313972e+00 1.83747839e-02 6.85814738e-01
-8.97603154e-01 -1.95688650e-01 -3.90040249e-01 7.93078899e-01
-7.07892627e-02 7.99582839e-01 -2.27186382e-01 -7.44949281e-01
-1.75743401e-02 -9.53075588e-02 -7.13720843e-02 -5.32644033e-01
-9.46550965e-01 -5.66439569e-01 2.88115263e-01 -4.18520927e-01
1.64958462e-01 -5.26306331e-02 -1.06750536e+00 -8.09843123e-01
-5.48597157e-01 4.09088820e-01 3.64805788e-01 7.68449605e-01
5.03513813e-01 1.16891421e-01 8.41123283e-01 1.04256785e-02
-8.54752719e-01 -6.96274579e-01 -6.27190948e-01 2.31874973e-01
1.81505784e-01 -4.11986470e-01 -7.86762834e-01 -6.19345307e-01] | [6.025822162628174, 7.241660118103027] |
a868e647-19d5-437d-865a-64ed1c6e483b | human-action-recognition-in-egocentric | 2306.05147 | null | https://arxiv.org/abs/2306.05147v1 | https://arxiv.org/pdf/2306.05147v1.pdf | Human Action Recognition in Egocentric Perspective Using 2D Object and Hands Pose | Egocentric action recognition is essential for healthcare and assistive technology that relies on egocentric cameras because it allows for the automatic and continuous monitoring of activities of daily living (ADLs) without requiring any conscious effort from the user. This study explores the feasibility of using 2D hand and object pose information for egocentric action recognition. While current literature focuses on 3D hand pose information, our work shows that using 2D skeleton data is a promising approach for hand-based action classification, might offer privacy enhancement, and could be less computationally demanding. The study uses a state-of-the-art transformer-based method to classify sequences and achieves validation results of 94%, outperforming other existing solutions. The accuracy of the test subset drops to 76%, indicating the need for further generalization improvement. This research highlights the potential of 2D hand and object pose information for action recognition tasks and offers a promising alternative to 3D-based methods. | ['Martin Kampel', 'Wiktor Mucha'] | 2023-06-08 | null | null | null | null | ['action-classification', 'action-recognition-in-videos', 'action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.27680409e-01 3.93088907e-02 -5.60399532e-01 -2.55333096e-01
-4.40304071e-01 -2.37009287e-01 4.23391879e-01 -3.34528118e-01
-7.09573388e-01 7.31905222e-01 5.10110438e-01 1.65698171e-01
-5.72668985e-02 -4.13941622e-01 -1.33555740e-01 -9.40173984e-01
1.52588142e-02 3.78127158e-01 2.17813358e-01 2.33285889e-01
3.29948068e-01 8.94005477e-01 -1.54419959e+00 3.49568099e-01
2.34737277e-01 1.09407187e+00 -7.68641606e-02 6.26578569e-01
3.33895653e-01 8.27978373e-01 -5.23953497e-01 -4.73984718e-01
4.41226155e-01 -3.54904741e-01 -7.38231540e-01 4.09697026e-01
6.23575807e-01 -9.56802607e-01 -4.63368237e-01 5.98743021e-01
9.03152287e-01 1.22639790e-01 3.89308602e-01 -1.23819029e+00
-7.93205649e-02 -2.91096181e-01 -3.33709329e-01 2.56503671e-01
8.03372622e-01 1.78973824e-01 4.78396446e-01 -3.87278914e-01
5.78184605e-01 8.10555220e-01 6.15977228e-01 7.21751511e-01
-1.01341522e+00 -7.68883824e-01 -1.16481900e-01 6.78267300e-01
-1.34707606e+00 -5.59160113e-01 8.29119861e-01 -4.82271135e-01
1.25705922e+00 3.49463940e-01 1.13970518e+00 1.33383381e+00
3.50910425e-01 9.76271868e-01 1.14638436e+00 -3.90631497e-01
2.44773448e-01 7.30681121e-02 -9.85157043e-02 5.69113493e-01
3.06796372e-01 -7.74478391e-02 -8.76012206e-01 -1.61960945e-01
8.79800975e-01 2.33327597e-01 -3.16700995e-01 -1.09066689e+00
-1.31094718e+00 4.70603138e-01 1.14454031e-01 2.90856451e-01
-7.58425891e-01 1.21134624e-01 6.83166683e-01 -1.18308710e-02
5.72224557e-01 2.06805676e-01 -3.12496632e-01 -8.43368471e-01
-7.94981956e-01 -7.77822733e-02 4.84228045e-01 7.76772618e-01
4.33722734e-02 -1.66020706e-01 -2.21238822e-01 4.08928156e-01
1.03238568e-01 3.82306308e-01 6.03602111e-01 -1.16985476e+00
4.04624850e-01 8.26522768e-01 1.68344881e-02 -7.54418969e-01
-4.30319220e-01 -1.28284320e-01 -3.22828919e-01 3.72066736e-01
5.10891259e-01 1.33762851e-01 -6.95197344e-01 1.29447782e+00
3.37537438e-01 -2.39586294e-01 -1.37534559e-01 9.30388927e-01
3.67709786e-01 -3.71079773e-01 2.36507371e-01 -3.47864658e-01
1.33928001e+00 -6.52190447e-01 -7.58972347e-01 8.10867026e-02
7.58881569e-01 -4.88362432e-01 8.72076750e-01 6.34927154e-01
-8.81897271e-01 -4.83227782e-02 -8.58615041e-01 1.32442221e-01
-2.04366505e-01 3.25690746e-01 8.88875842e-01 1.34410203e+00
-7.02396810e-01 5.31505525e-01 -1.15078652e+00 -1.01901901e+00
8.11300993e-01 5.40607929e-01 -8.88877332e-01 -2.63857897e-02
-6.21549428e-01 1.12757337e+00 3.05304110e-01 -2.44308978e-01
-6.09395146e-01 -3.64795208e-01 -8.78165781e-01 -3.35731924e-01
1.99586421e-01 -4.67838079e-01 9.78943050e-01 -6.09370291e-01
-1.62802720e+00 1.09333849e+00 -3.36230665e-01 -4.81012762e-01
8.84097934e-01 -5.06026745e-01 -2.23854080e-01 6.96206927e-01
1.21886814e-02 4.18008745e-01 7.64267683e-01 -3.77351791e-01
-4.55878645e-01 -1.04348946e+00 4.47206944e-02 4.78553802e-01
-5.89073420e-01 1.91139653e-02 -2.20459744e-01 -5.21857142e-01
2.99609423e-01 -1.05207407e+00 2.71668494e-01 3.99405658e-01
6.83203563e-02 -2.70971000e-01 1.00660896e+00 -9.09803271e-01
8.19827437e-01 -2.07124829e+00 -1.73236981e-01 -9.88470167e-02
-5.40048406e-02 4.55236107e-01 3.45381409e-01 1.48370415e-01
-2.11164549e-01 -3.79842728e-01 2.26853937e-01 -6.50270879e-02
-3.21648777e-01 7.65411630e-02 3.95390034e-01 1.03158689e+00
-1.69265389e-01 8.17972898e-01 -9.48391616e-01 -5.50717711e-01
8.52410555e-01 5.95020175e-01 -3.92265469e-01 -1.39388412e-01
5.74173152e-01 6.17283225e-01 -3.22684526e-01 8.55536759e-01
2.69153923e-01 9.42266360e-02 1.80586398e-01 -3.53011370e-01
2.27897123e-01 1.92703575e-01 -9.90730226e-01 1.76716805e+00
-1.50321767e-01 7.46619463e-01 -3.51271510e-01 -9.39043939e-01
7.50787854e-01 7.28164732e-01 1.05464494e+00 -6.36885643e-01
3.23057026e-01 -1.24450125e-01 -2.71861762e-01 -8.41214776e-01
1.23536341e-01 -1.19306043e-01 3.15121293e-01 3.31727326e-01
-1.63190395e-01 2.13543847e-01 -1.08941257e-01 -1.45889476e-01
1.22993720e+00 6.60892844e-01 7.10150242e-01 -1.37864882e-02
6.48198009e-01 -8.57541114e-02 3.15729409e-01 3.78964484e-01
-7.53850698e-01 4.25836414e-01 2.43676491e-02 -5.02551258e-01
-7.39727557e-01 -9.92071152e-01 -4.31405976e-02 6.46576524e-01
-1.33313313e-01 -3.11121523e-01 -7.76839554e-01 -9.65246022e-01
-4.68340218e-02 8.06063175e-01 -6.28206611e-01 -1.92417905e-01
-4.08496261e-01 -2.32057050e-01 5.10652304e-01 1.01893997e+00
9.41453457e-01 -7.86325395e-01 -1.17964351e+00 -7.60512054e-02
-4.88411844e-01 -1.10825920e+00 -3.15377653e-01 -2.85660893e-01
-1.22407174e+00 -1.37576747e+00 -1.15112758e+00 -1.60207242e-01
7.26598144e-01 3.88821214e-01 4.89308476e-01 -5.43807566e-01
-4.34228033e-01 1.30266440e+00 -4.25205231e-01 -4.09053206e-01
1.62026539e-01 -8.82653370e-02 4.63990629e-01 1.14957675e-01
9.05197442e-01 -6.39148057e-01 -8.68236661e-01 5.90691090e-01
-1.93195179e-01 -2.92896450e-01 4.96808410e-01 5.11623919e-01
3.56438279e-01 -1.83720887e-01 1.65999457e-01 -3.29496980e-01
1.97414845e-01 4.59293984e-02 -2.24774163e-02 1.98677629e-01
-7.25262523e-01 -2.14661106e-01 2.32660361e-02 -4.82628912e-01
-1.22159410e+00 5.68443000e-01 2.87970096e-01 -5.45474470e-01
-3.77581239e-01 -2.86528587e-01 -3.77466053e-01 -3.07270378e-01
6.07265890e-01 2.96897054e-01 3.66057545e-01 -4.51644450e-01
-1.33944184e-01 9.82909977e-01 3.58038366e-01 -7.90660530e-02
1.78453431e-01 9.85260308e-01 2.66712725e-01 -1.05294669e+00
-3.84372026e-01 -7.92448580e-01 -1.21792817e+00 -6.65897787e-01
9.94360685e-01 -9.25462246e-01 -9.17764604e-01 6.29585564e-01
-7.41020739e-01 1.45334840e-01 -4.64946419e-01 1.13971043e+00
-1.12845993e+00 7.56820381e-01 2.02069022e-02 -1.12630701e+00
-3.74526441e-01 -9.92952049e-01 1.13493931e+00 -2.34414607e-01
-8.83199573e-01 -6.13117397e-01 -1.57728985e-01 1.07890832e+00
1.12657987e-01 4.10011530e-01 2.29807004e-01 -4.79284674e-01
-4.04051155e-01 -9.06089127e-01 1.63219064e-01 3.89580160e-01
6.02853537e-01 -7.39257991e-01 -1.03385830e+00 -3.16234261e-01
2.04056412e-01 -2.48690069e-01 2.63231456e-01 2.47995317e-01
9.37565625e-01 -1.95917428e-01 -2.78083891e-01 3.60643804e-01
8.25529099e-01 4.27597374e-01 1.00641131e+00 5.88970125e-01
5.15176058e-01 6.16810620e-01 8.75949442e-01 6.02904081e-01
2.30934337e-01 9.30947721e-01 3.93230468e-01 3.84637445e-01
-8.86992514e-02 4.28077094e-02 3.98477346e-01 -7.02261701e-02
-7.56731749e-01 2.50288606e-01 -5.81972063e-01 4.64491844e-01
-1.65192354e+00 -1.24585187e+00 4.15669791e-02 2.46920466e+00
2.35694781e-01 -3.15375328e-01 4.86482948e-01 6.16747499e-01
6.10465229e-01 -3.51056159e-02 -6.61834776e-01 -2.77396023e-01
3.75990629e-01 1.43339351e-01 8.84599626e-01 -1.79596707e-01
-1.21313524e+00 6.09238863e-01 6.81634283e+00 2.86728054e-01
-1.06246614e+00 6.61691353e-02 -1.62646640e-04 -4.82257247e-01
6.21838689e-01 -3.26182365e-01 -4.02772725e-01 3.04127544e-01
6.18487000e-01 1.28812417e-01 2.06872821e-01 1.16603172e+00
4.83676314e-01 -5.82959294e-01 -1.16694951e+00 1.30716634e+00
5.55882275e-01 -7.98977733e-01 -8.14632401e-02 3.84667158e-01
1.13335893e-01 -1.72537670e-01 -3.54985923e-01 -2.75234431e-01
-4.11774844e-01 -5.86865067e-01 3.55281383e-01 5.71341038e-01
8.38116586e-01 -6.36402488e-01 7.07555354e-01 3.05860460e-01
-1.01694322e+00 -1.56152278e-01 1.28354952e-01 -3.19307983e-01
1.24267034e-01 -2.63656884e-01 -9.84749794e-01 3.72790217e-01
8.98511410e-01 8.66669774e-01 -4.70229030e-01 1.01231146e+00
-1.28851354e-01 3.22783262e-01 -3.36771607e-01 -1.07173160e-01
-1.78772420e-01 1.08811669e-01 7.72459745e-01 9.14085150e-01
2.58793414e-01 2.41505355e-01 -2.39335552e-01 -1.58257205e-02
5.07794201e-01 1.15684308e-01 -9.23846424e-01 -4.15540999e-03
-1.10088900e-01 7.73512363e-01 -7.55286992e-01 -1.22647248e-01
-3.79267007e-01 1.41968560e+00 -1.34892941e-01 -5.45516014e-02
-6.00767672e-01 -1.98112488e-01 7.40519166e-01 3.82227272e-01
3.22811753e-01 -3.65280539e-01 -2.65586853e-01 -1.14727962e+00
3.32041919e-01 -7.92652547e-01 7.12867200e-01 -9.03286815e-01
-6.55079424e-01 -7.17227906e-02 2.37194389e-01 -1.79134202e+00
-5.47191918e-01 -6.75564408e-01 -1.61864936e-01 5.35760522e-01
-7.49665439e-01 -1.29724181e+00 -5.98236680e-01 9.79489148e-01
5.83412051e-01 -2.75660455e-01 9.98344183e-01 1.57894999e-01
-2.72249430e-01 7.65140831e-01 -1.00528277e-01 2.20244601e-01
6.56260669e-01 -7.55059063e-01 -1.01880513e-01 6.23188436e-01
-5.75848222e-02 5.57370424e-01 5.10549724e-01 -7.51325607e-01
-1.50422144e+00 -7.14535058e-01 7.95630038e-01 -9.84099686e-01
1.53818771e-01 1.03110291e-01 -7.28051737e-02 6.69980645e-01
-2.42050096e-01 -2.22108260e-01 9.01288629e-01 -2.14025676e-01
-7.49745145e-02 -2.50627428e-01 -1.74945986e+00 6.12927318e-01
1.37290967e+00 -5.84856629e-01 -6.31758749e-01 2.84298778e-01
-2.14474067e-01 -2.20136255e-01 -9.90558028e-01 3.53015125e-01
1.22405672e+00 -9.95239437e-01 1.04906917e+00 -6.12015784e-01
-7.94073716e-02 -6.81361780e-02 -2.96526790e-01 -7.60585189e-01
-2.15591624e-01 -2.89808422e-01 -3.36818725e-01 8.84705365e-01
-2.64340460e-01 -8.53537083e-01 1.44750428e+00 1.10768116e+00
3.21088672e-01 -4.06242251e-01 -1.25161457e+00 -1.12294829e+00
-5.80072880e-01 -6.08502030e-01 1.51698992e-01 7.38726676e-01
3.46856326e-01 -1.96941420e-01 -4.23006326e-01 -1.49823740e-01
7.20368147e-01 -4.49746579e-01 9.34679210e-01 -1.00368857e+00
1.39505073e-01 -4.28324230e-02 -1.52746677e+00 -6.28182590e-01
1.12956859e-01 -7.21644521e-01 -4.60526317e-01 -1.56177461e+00
2.00804919e-01 3.51774424e-01 -2.31096208e-01 5.64617097e-01
3.91172230e-01 6.82797551e-01 -5.86101902e-04 1.64573669e-01
-2.78481901e-01 4.28601086e-01 1.13478541e+00 -1.41349524e-01
4.17922810e-03 3.49083990e-01 -4.15371537e-01 7.59089768e-01
8.45189393e-01 -3.00488591e-01 -6.10901594e-01 9.70827788e-02
-4.39688504e-01 -2.80612409e-01 5.38633525e-01 -1.15944958e+00
1.30025759e-01 -5.89632578e-02 7.22695529e-01 -8.12978745e-01
8.44688952e-01 -1.32583904e+00 1.88895702e-01 7.91426003e-01
1.27415508e-02 -2.80457675e-01 -5.49867842e-03 5.46486557e-01
1.48953959e-01 -9.56817344e-02 5.88311911e-01 -4.71254975e-01
-7.93918669e-01 4.77661751e-02 -6.07701123e-01 -2.83609658e-01
1.55449176e+00 -1.09192324e+00 5.38472869e-02 -5.12117624e-01
-8.35305452e-01 -2.71001637e-01 6.29309297e-01 3.61628234e-01
5.68431437e-01 -1.23972666e+00 -1.79688841e-01 2.02732205e-01
3.55446517e-01 -6.19447827e-01 4.70252894e-02 1.23275948e+00
-5.59480548e-01 7.63714194e-01 -6.28347278e-01 -7.14243114e-01
-1.95599556e+00 3.11038077e-01 3.28739196e-01 1.76737696e-01
-1.03914154e+00 4.74007607e-01 -1.62822977e-01 -2.20093671e-02
4.43047881e-01 1.07219316e-01 -2.33675659e-01 1.97205752e-01
5.70341587e-01 1.17099106e+00 6.68692961e-02 -8.23130906e-01
-8.77370834e-01 6.14722311e-01 3.65038328e-02 -1.78115681e-01
1.19352424e+00 -1.42002672e-01 1.58035889e-01 3.43215913e-01
1.02865732e+00 -3.84595811e-01 -1.11659789e+00 6.82690144e-02
-1.71280101e-01 -1.04301512e+00 1.10534288e-01 -6.61692739e-01
-9.43497181e-01 7.96361685e-01 1.17228222e+00 -4.18445945e-01
1.14594185e+00 -4.29444760e-02 5.43125808e-01 7.12514639e-01
9.46254373e-01 -1.32305276e+00 1.40708968e-01 1.05545834e-01
7.77954400e-01 -1.13027775e+00 4.99709517e-01 -2.56886303e-01
-7.99134254e-01 1.03783309e+00 4.34725881e-01 1.96578249e-01
5.32002032e-01 -1.65197611e-01 -2.59468071e-02 -1.58360094e-01
1.82186905e-02 -1.35685414e-01 2.15028137e-01 1.02533114e+00
3.78507078e-01 4.76062261e-02 -5.89753568e-01 -2.92783529e-02
-6.39205948e-02 5.49728036e-01 2.42000967e-01 1.45937610e+00
3.55203897e-02 -1.01007485e+00 -2.89325804e-01 6.29029870e-01
-5.02758563e-01 3.76948148e-01 -5.61230063e-01 8.98203015e-01
-1.27101596e-02 8.56013894e-01 3.76931690e-02 -3.87746155e-01
4.89765763e-01 5.08152544e-01 1.01099575e+00 -3.63689542e-01
-3.12976688e-01 -3.19115341e-01 2.88802654e-01 -1.13427031e+00
-7.51709282e-01 -1.30129957e+00 -8.00540984e-01 -1.33014724e-01
-1.99019328e-01 -3.71068865e-01 7.51690686e-01 1.00797093e+00
4.22497571e-01 7.94588029e-02 3.68465066e-01 -9.80964422e-01
-4.05924350e-01 -8.48929644e-01 -5.92211843e-01 3.12087297e-01
1.92841694e-01 -1.03338683e+00 -1.80943936e-01 1.37892067e-01] | [7.501162528991699, 0.41396650671958923] |
be7921f3-dd0d-4ec1-938c-08563771831d | federated-td-learning-over-finite-rate | 2305.08104 | null | https://arxiv.org/abs/2305.08104v1 | https://arxiv.org/pdf/2305.08104v1.pdf | Federated TD Learning over Finite-Rate Erasure Channels: Linear Speedup under Markovian Sampling | Federated learning (FL) has recently gained much attention due to its effectiveness in speeding up supervised learning tasks under communication and privacy constraints. However, whether similar speedups can be established for reinforcement learning remains much less understood theoretically. Towards this direction, we study a federated policy evaluation problem where agents communicate via a central aggregator to expedite the evaluation of a common policy. To capture typical communication constraints in FL, we consider finite capacity up-link channels that can drop packets based on a Bernoulli erasure model. Given this setting, we propose and analyze QFedTD - a quantized federated temporal difference learning algorithm with linear function approximation. Our main technical contribution is to provide a finite-sample analysis of QFedTD that (i) highlights the effect of quantization and erasures on the convergence rate; and (ii) establishes a linear speedup w.r.t. the number of agents under Markovian sampling. Notably, while different quantization mechanisms and packet drop models have been extensively studied in the federated learning, distributed optimization, and networked control systems literature, our work is the first to provide a non-asymptotic analysis of their effects in multi-agent and federated reinforcement learning. | ['George J. Pappas', 'Aritra Mitra', 'Nicolò Dal Fabbro'] | 2023-05-14 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-2.49024466e-01 6.11382984e-02 -2.64824480e-01 -8.38499144e-02
-8.76211464e-01 -6.86456621e-01 5.21035671e-01 4.77210701e-01
-7.85300970e-01 1.14564514e+00 -8.64003524e-02 -4.90592331e-01
-3.29899102e-01 -7.24713862e-01 -9.68310237e-01 -9.53937650e-01
-9.46041584e-01 4.68842834e-01 -2.44713366e-01 4.91834395e-02
-1.32699192e-01 3.49624872e-01 -1.11946106e+00 -2.74996579e-01
7.60351598e-01 1.19772065e+00 -3.53651345e-01 8.72456968e-01
1.11280897e-04 9.25605714e-01 -8.10632944e-01 -4.83965755e-01
6.70883179e-01 -4.87501204e-01 -6.64100051e-01 1.04489140e-01
1.89283222e-01 -8.90345216e-01 -4.30339158e-01 1.27628970e+00
6.29305840e-01 7.20793828e-02 3.25664759e-01 -1.75048006e+00
-3.52484584e-01 1.07194459e+00 -2.80641437e-01 1.16079316e-01
3.01760454e-02 1.78069115e-01 9.56814826e-01 1.09304547e-01
4.99122232e-01 1.26138246e+00 3.75481963e-01 6.78432345e-01
-1.08551013e+00 -8.40517104e-01 1.98838040e-01 8.58315974e-02
-7.11005807e-01 -3.46773446e-01 3.52671295e-01 2.15156190e-02
4.80668813e-01 -1.29627492e-02 6.26000702e-01 8.49861622e-01
2.51025110e-01 1.05110872e+00 1.26200783e+00 -4.05938685e-01
8.09001029e-01 6.01238981e-02 -1.62485182e-01 7.51256645e-01
3.32155526e-01 5.16772687e-01 -5.32055974e-01 -5.87533653e-01
6.15713894e-01 1.18039779e-01 -3.28425050e-01 -4.97532457e-01
-8.54772925e-01 1.10052764e+00 1.20015688e-01 -1.89182490e-01
-6.85229361e-01 5.06759703e-01 7.13286102e-01 1.09017217e+00
5.77982128e-01 8.13383982e-02 -6.17098808e-01 -2.98555225e-01
-5.56113422e-01 2.76491672e-01 1.41571307e+00 1.02027857e+00
7.36653388e-01 8.18206742e-02 -7.77675286e-02 2.73897469e-01
-3.63048725e-03 8.36912930e-01 9.17876884e-02 -1.58457696e+00
3.63988698e-01 -1.12866603e-01 5.82626581e-01 -3.69863421e-01
-1.67880654e-01 -3.68758380e-01 -6.43422961e-01 4.65929896e-01
6.61561012e-01 -9.89015043e-01 8.53898078e-02 1.92520404e+00
5.09649396e-01 9.31285024e-02 2.94581860e-01 9.37539697e-01
-3.05129617e-01 4.97727752e-01 -1.50118068e-01 -7.79000461e-01
9.94698286e-01 -8.42716753e-01 -6.51864409e-01 5.25158823e-01
7.77450442e-01 -3.22216779e-01 3.89100939e-01 4.78496492e-01
-1.20682287e+00 2.38467678e-01 -6.71229422e-01 5.58637857e-01
-3.43818888e-02 -5.60506761e-01 6.25561774e-01 8.03829432e-01
-1.13689148e+00 7.93624997e-01 -9.95202422e-01 -4.02654856e-01
6.89686894e-01 4.43940431e-01 -3.59014310e-02 -1.05729260e-01
-9.53155577e-01 4.44169939e-01 -8.36218521e-02 -5.07742465e-01
-1.32022274e+00 -7.68188715e-01 -4.51331168e-01 7.80853927e-02
8.18529904e-01 -8.67924452e-01 1.77309060e+00 -1.06925368e+00
-1.80963874e+00 5.55805154e-02 1.83585107e-01 -1.13882995e+00
9.03892219e-01 -7.66802281e-02 -6.37876838e-02 4.99911547e-01
-8.02380294e-02 2.44939737e-02 8.40914130e-01 -1.00133085e+00
-1.07275832e+00 -5.21226704e-01 4.52182859e-01 1.98485434e-01
-7.77328134e-01 -1.75663322e-01 2.85417020e-01 -4.38295007e-01
-9.01716530e-01 -8.61274838e-01 -3.37742716e-01 3.84228230e-01
1.97144747e-02 -3.59422028e-01 9.54664409e-01 -5.53540587e-02
7.10862935e-01 -1.92678130e+00 -2.20706269e-01 7.51182213e-02
2.25720331e-01 1.88754648e-01 -2.23107249e-01 8.19281399e-01
7.73172736e-01 2.40060780e-03 6.19531162e-02 -6.37750864e-01
4.95974630e-01 3.90985966e-01 -3.11194658e-01 8.93125832e-01
-5.58086455e-01 5.51145434e-01 -1.09866858e+00 -3.59449685e-01
4.04328294e-02 2.41339102e-01 -5.92443824e-01 2.81835973e-01
-4.27702188e-01 3.64462256e-01 -8.81934881e-01 3.40219468e-01
5.45067728e-01 -2.37868637e-01 2.62498677e-01 2.62909144e-01
-2.41099313e-01 -7.18453825e-02 -9.82652068e-01 1.45845258e+00
-6.00611210e-01 2.77942359e-01 1.09823275e+00 -1.12352931e+00
4.69234914e-01 6.05777025e-01 1.00342107e+00 -7.16454804e-01
3.26251835e-01 2.09749103e-01 -3.87065709e-01 -4.29875672e-01
2.31151626e-01 -3.29245925e-01 8.32850337e-02 1.13159490e+00
-9.16517079e-02 4.55752403e-01 1.15544915e-01 3.55685532e-01
1.45373487e+00 -3.22760612e-01 1.02023482e-01 -2.35974118e-01
3.02729309e-01 -6.46053478e-02 4.97555465e-01 1.15421236e+00
-7.33930349e-01 -4.50190306e-01 4.28996950e-01 -3.34288538e-01
-9.11292136e-01 -8.68038058e-01 1.42058924e-01 1.24339974e+00
2.63917089e-01 -2.02867448e-01 -6.92341924e-01 -9.39023614e-01
5.64090252e-01 4.84351873e-01 -4.50691909e-01 8.57526734e-02
-2.75275439e-01 -4.58391964e-01 6.83468282e-01 -3.73409421e-04
6.66923642e-01 -8.12313557e-01 -8.63326609e-01 4.86173779e-01
1.85055852e-01 -1.01933038e+00 -7.16261864e-01 2.47266904e-01
-6.62385464e-01 -1.31998277e+00 -6.31391168e-01 -4.23291683e-01
4.51224476e-01 3.63731891e-01 7.49395311e-01 -2.27342099e-01
-3.98695618e-02 1.27191925e+00 -2.95358092e-01 -4.89885688e-01
-5.79681873e-01 -1.52178124e-01 3.87849957e-01 2.23182470e-01
1.10821374e-01 -5.50519168e-01 -8.30914855e-01 1.61787584e-01
-1.11987054e+00 -5.85850894e-01 3.36019695e-01 6.79677486e-01
2.92936116e-01 1.65277049e-01 8.23999107e-01 -8.20760489e-01
9.78882730e-01 -6.57703996e-01 -9.80470121e-01 1.95781842e-01
-7.97545075e-01 2.81239331e-01 1.24858868e+00 -3.14605296e-01
-8.45661521e-01 -1.80303320e-01 3.63972276e-01 -4.30129260e-01
-2.72414670e-03 1.02625221e-01 4.24032182e-01 -3.35358143e-01
3.03989023e-01 1.42880172e-01 6.99688792e-01 -2.99397767e-01
5.00465155e-01 8.24770868e-01 8.19288641e-02 -1.01456439e+00
6.50169790e-01 8.07999074e-01 8.42945054e-02 -6.30026579e-01
-3.64300519e-01 -1.91648290e-01 2.56053805e-01 -1.46065354e-01
2.93403506e-01 -8.31427515e-01 -1.58051407e+00 2.91248769e-01
-9.46518898e-01 -6.61372781e-01 -6.51761472e-01 7.28249907e-01
-9.10776079e-01 4.59529787e-01 -8.80645156e-01 -1.35884571e+00
-6.06901348e-01 -8.36033165e-01 3.80523503e-01 1.09122746e-01
4.24495071e-01 -1.27590442e+00 2.27822572e-01 -7.07451627e-02
7.82341897e-01 1.21344537e-01 5.18435180e-01 -7.08609402e-01
-6.02481008e-01 8.65787417e-02 1.39409021e-01 2.73362905e-01
9.24127325e-02 -4.34179515e-01 -5.40297151e-01 -1.01353669e+00
3.36934142e-02 -7.21580148e-01 4.67346400e-01 4.36831921e-01
9.79557812e-01 -1.07991397e+00 -1.18826881e-01 4.74283874e-01
1.53914082e+00 2.64644530e-02 -3.04806978e-01 2.55409002e-01
-1.03012502e-01 4.60998327e-01 5.50004184e-01 1.37076318e+00
4.52482760e-01 2.71565795e-01 8.60206664e-01 3.32929403e-01
2.60166645e-01 -2.17018843e-01 7.89000273e-01 3.84132892e-01
3.48262221e-01 -3.61598670e-01 -2.36330301e-01 3.98351192e-01
-1.80158436e+00 -9.64426756e-01 4.92941082e-01 2.44181371e+00
7.35965967e-01 -2.16639131e-01 6.71535552e-01 -1.78167045e-01
7.33055651e-01 -9.96450707e-02 -1.00211632e+00 -4.09161687e-01
-1.22125290e-01 5.17042428e-02 1.11734784e+00 5.41481614e-01
-7.94452429e-01 4.35065806e-01 5.49362135e+00 6.60311401e-01
-1.08755386e+00 3.24349135e-01 4.90683317e-01 -4.24475491e-01
-1.08414225e-01 -1.78993106e-01 -5.62516034e-01 4.46514428e-01
1.13407016e+00 -6.51416719e-01 1.03879344e+00 8.46216798e-01
3.77946436e-01 1.16780326e-01 -1.16000175e+00 8.69771123e-01
-4.68967885e-01 -1.19162297e+00 -1.76043808e-01 5.26440740e-01
8.60752940e-01 4.46471423e-01 7.06028566e-02 2.25465223e-01
1.00607264e+00 -6.14105463e-01 5.65106273e-01 1.29201919e-01
5.60712397e-01 -9.78120506e-01 4.52048749e-01 4.92817551e-01
-9.54273403e-01 -6.29879415e-01 -3.11462790e-01 -1.61515117e-01
1.44855559e-01 3.64027470e-01 -4.13031906e-01 5.15028954e-01
4.57895190e-01 5.01116633e-01 2.18153939e-01 1.05066562e+00
3.35361868e-01 7.63962924e-01 -6.10508680e-01 -2.85735011e-01
4.23872471e-01 -3.40976655e-01 7.25610316e-01 8.82415175e-01
2.62370646e-01 -7.76977912e-02 4.81816053e-01 4.97841030e-01
-4.48291987e-01 1.29146412e-01 -3.95210087e-01 -2.81690974e-02
8.04547787e-01 1.12664473e+00 -1.36522189e-01 -2.68611521e-01
-6.12950683e-01 8.45079303e-01 4.09089833e-01 4.99845535e-01
-5.85772395e-01 -2.64904708e-01 1.14880931e+00 -8.81808400e-02
5.50965130e-01 -1.36329740e-01 2.87941605e-01 -9.50956762e-01
-7.12474734e-02 -9.05839860e-01 7.03945279e-01 2.57088959e-01
-1.69311917e+00 1.88214704e-01 -2.69680947e-01 -9.91063058e-01
-3.61775577e-01 -3.01230222e-01 -4.98923808e-01 3.10219049e-01
-1.82065547e+00 -5.43847620e-01 2.21735254e-01 1.06777179e+00
1.84665799e-01 -1.68741137e-01 8.71611536e-01 2.70359665e-01
-5.95887303e-01 7.97943532e-01 8.91783178e-01 7.81473443e-02
6.62244201e-01 -1.17732596e+00 -3.36755782e-01 5.98250866e-01
-1.16981506e-01 2.16240883e-01 7.04457045e-01 -2.35954612e-01
-2.12744713e+00 -1.22614181e+00 1.00157589e-01 1.47262543e-01
8.70859921e-01 5.40444478e-02 -4.30803180e-01 5.30494750e-01
2.96178550e-01 6.19473755e-01 4.71396327e-01 -3.21639955e-01
-2.24935189e-01 -5.84605873e-01 -1.56584811e+00 2.61087507e-01
7.04819858e-01 -3.92228931e-01 2.45148674e-01 4.25484180e-01
6.04537547e-01 -2.56213406e-03 -1.06273198e+00 -3.84654999e-01
3.65248948e-01 -7.81862617e-01 3.68504435e-01 -8.24974656e-01
-2.44353861e-01 1.23967811e-01 -1.25778228e-01 -1.60697985e+00
2.11007878e-01 -1.65396762e+00 -5.23333311e-01 9.41059291e-01
1.35485272e-04 -1.12210810e+00 9.78300571e-01 5.40556371e-01
3.06287497e-01 -7.46616662e-01 -1.40845275e+00 -1.25609732e+00
5.48837483e-01 -1.39225289e-01 5.78593493e-01 6.35353982e-01
4.42583621e-01 -1.62584424e-01 -4.37840313e-01 1.75074533e-01
1.23233831e+00 1.14868805e-01 7.51044452e-01 -8.09196413e-01
-5.77140808e-01 -3.21184725e-01 1.36347160e-01 -1.35896754e+00
3.47247809e-01 -6.86486185e-01 -1.13530390e-01 -9.24181521e-01
-7.31502995e-02 -7.96253145e-01 -3.39118034e-01 2.96499670e-01
3.58667344e-01 -3.96103650e-01 5.18869996e-01 2.99699813e-01
-1.21533322e+00 8.65664244e-01 1.06386304e+00 -9.46620628e-02
-1.70574430e-02 2.88186818e-01 -5.33375263e-01 1.45347714e-01
1.06099534e+00 -5.41125536e-01 -4.31971788e-01 -3.85494471e-01
-1.32965803e-01 8.20516229e-01 3.34546804e-01 -6.76468372e-01
5.28790116e-01 -1.22259870e-01 -3.61424357e-01 1.02320872e-01
-3.44156995e-02 -1.10904109e+00 -2.96571165e-01 9.01881993e-01
-6.87897444e-01 1.22206025e-01 -2.48846889e-01 1.22502851e+00
1.31379202e-01 -3.27922553e-02 6.65544689e-01 -1.64773598e-01
-8.99108425e-02 8.76279593e-01 -5.99325776e-01 4.33097392e-01
1.22676194e+00 4.21286106e-01 -4.40556139e-01 -9.19936776e-01
-4.06145602e-01 7.62522876e-01 3.81399870e-01 -5.78578822e-02
4.24864590e-01 -9.87345636e-01 -7.03435361e-01 -2.86247022e-02
-4.22261477e-01 -3.16352606e-01 -1.29830480e-01 8.89869630e-01
3.74743827e-02 3.08351874e-01 -9.28804204e-02 -1.80033386e-01
-9.74785089e-01 6.71349943e-01 5.27203500e-01 -2.14679912e-01
-4.89177346e-01 4.07494754e-01 -4.18719947e-01 -9.25844386e-02
8.12347233e-01 -4.46439497e-02 4.72886562e-01 -2.43747592e-01
5.95988691e-01 8.40877712e-01 -1.40930057e-01 6.03344142e-02
-8.80268142e-02 -1.34247229e-01 -1.49452776e-01 -3.61172736e-01
1.22975910e+00 -3.83086115e-01 9.33485553e-02 1.57954991e-01
1.36129558e+00 -1.55956239e-01 -1.83787251e+00 -6.85015440e-01
-3.03324163e-01 -2.95179904e-01 2.15888917e-01 -5.85035443e-01
-1.37204957e+00 4.33979243e-01 5.58468819e-01 7.20182657e-01
8.95919979e-01 -1.74872085e-01 1.05551732e+00 5.10293424e-01
9.55927491e-01 -1.01392770e+00 -9.36747938e-02 3.64368677e-01
2.06416383e-01 -1.15073073e+00 -2.86075562e-01 1.82767034e-01
-3.65068227e-01 1.23283398e+00 2.90468425e-01 -1.29802525e-01
7.06775308e-01 5.21204770e-01 -1.01520881e-01 1.49438232e-01
-1.04104948e+00 -9.55022648e-02 -7.33126044e-01 6.02971315e-01
8.07700679e-03 1.72278509e-01 -2.86708385e-01 2.38711670e-01
1.71808988e-01 1.85722128e-01 6.28861368e-01 1.18180728e+00
-5.38196504e-01 -1.27678132e+00 -2.71518260e-01 3.98704350e-01
-7.85090327e-01 2.50378966e-01 -2.72084344e-02 4.20774251e-01
-3.47307235e-01 1.29958677e+00 5.49119599e-02 1.42498419e-01
-9.07449573e-02 -2.72555768e-01 5.01821935e-01 -2.25785952e-02
-7.12313294e-01 -1.93808347e-01 -1.32523775e-01 -7.37985075e-01
-5.12486815e-01 -6.95779264e-01 -1.29824233e+00 -9.02603626e-01
-1.29914388e-01 8.21822762e-01 6.51898503e-01 8.23270857e-01
5.27347386e-01 5.66743761e-02 1.11870027e+00 -3.70321661e-01
-1.71588743e+00 -5.60803235e-01 -8.76662076e-01 2.30070025e-01
9.34504330e-01 -2.88214892e-01 -7.43518770e-01 -4.73733246e-01] | [4.222477912902832, 2.573914051055908] |
3832a388-9221-4d4f-8439-1188742655da | low-rank-laplacian-uniform-mixed-model-for | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Dong_Low-Rank_Laplacian-Uniform_Mixed_Model_for_Robust_Face_Recognition_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Dong_Low-Rank_Laplacian-Uniform_Mixed_Model_for_Robust_Face_Recognition_CVPR_2019_paper.pdf | Low-Rank Laplacian-Uniform Mixed Model for Robust Face Recognition | Sparse representation based methods have successfully put forward a general framework for robust face recognition through linear reconstruction and sparsity constraints. However, residual modeling in existing works is not yet robust enough when dealing with dense noise. In this paper, we aim at recognizing identities from faces with varying levels of noises of various forms such as occlusion, pixel corruption, or disguise, and take improving the fitting ability of the error model as the key to addressing this problem. To fully capture the characteristics of different noises, we propose a mixed model combining robust sparsity constraint and low-rank constraint, which can deal with random errors and structured errors simultaneously. For random noises such as pixel corruption, we adopt a Laplacian-uniform mixed function for fitting the error distribution. For structured errors like continuous occlusion or disguise, we utilize robust nuclear norm to constrain the rank of the error matrix. An effective iterative reweighted algorithm is then developed to solve the proposed model. Comprehensive experiments were conducted on several benchmark databases for robust face recognition, and the overall results demonstrate that our model is most robust against various kinds of noises, when compared with state-of-the-art methods.
| [' Lina Lian', ' Huicheng Zheng', 'Jiayu Dong'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['robust-face-recognition'] | ['computer-vision'] | [ 3.41324657e-01 -6.92978561e-01 1.01860173e-01 -3.83623332e-01
-5.79794466e-01 -1.01530209e-01 2.89259821e-01 -7.61100054e-01
-8.20183381e-02 6.28690720e-01 2.08081514e-01 2.25662500e-01
-2.63870418e-01 -4.38300222e-01 -5.17219126e-01 -9.78129923e-01
4.22208160e-01 -7.86786247e-03 -2.11962596e-01 1.57089494e-02
3.13091397e-01 6.42605245e-01 -1.73327589e+00 -1.66826900e-02
9.92572308e-01 1.20796824e+00 9.43390355e-02 -5.27751409e-02
-1.28803685e-01 8.36902261e-01 -5.23825169e-01 -3.36778253e-01
3.88668001e-01 -3.64972651e-01 -7.23745003e-02 5.75857937e-01
4.73093569e-01 -1.71768740e-01 -6.68389499e-01 1.69331908e+00
7.21747994e-01 2.39373043e-01 5.55002213e-01 -1.04280472e+00
-8.43292177e-01 2.67358631e-01 -8.60354483e-01 1.84914902e-01
4.11133111e-01 -1.57757979e-02 3.57312530e-01 -1.51452971e+00
4.67101455e-01 1.58370817e+00 7.84339845e-01 5.08161664e-01
-1.11925936e+00 -8.52958262e-01 2.93303519e-01 4.64389920e-01
-1.78250313e+00 -1.01856017e+00 1.05039227e+00 -3.60471040e-01
2.48824582e-01 4.24595118e-01 3.51730362e-02 1.03201771e+00
-2.72793204e-01 5.55809081e-01 1.32154298e+00 -1.97471946e-01
1.91544350e-02 -1.26088619e-01 8.51537138e-02 4.63038892e-01
5.72769642e-01 -1.09728858e-01 -3.45540971e-01 -2.02352270e-01
7.28348732e-01 3.18058908e-01 -6.78070128e-01 -1.59826338e-01
-8.78801227e-01 4.96549755e-01 1.69144824e-01 3.77093375e-01
-3.73477966e-01 -1.91875696e-01 1.15219764e-01 4.58710790e-02
4.29583549e-01 -1.82805002e-01 -6.29306063e-02 3.73381108e-01
-9.68412995e-01 -2.52542682e-02 4.98339772e-01 8.55683446e-01
7.23828793e-01 5.78557789e-01 -2.90004313e-01 1.29779077e+00
6.85417652e-01 7.57817745e-01 4.03330117e-01 -8.84740949e-01
3.54331017e-01 4.29945201e-01 9.41463858e-02 -1.67253792e+00
-1.72312066e-01 -6.11015797e-01 -1.38942528e+00 4.17838385e-03
2.69699395e-01 -4.62385900e-02 -9.99456346e-01 1.52960908e+00
4.40836310e-01 8.23040783e-01 -4.71656211e-02 1.06729627e+00
1.21770608e+00 6.10979855e-01 -3.11728895e-01 -8.71309638e-01
1.17907047e+00 -5.06790698e-01 -1.20183766e+00 -1.02039285e-01
2.34141611e-02 -1.05655611e+00 4.51945752e-01 4.66630965e-01
-9.99855816e-01 -6.41350150e-01 -8.03740978e-01 1.43105179e-01
2.31695727e-01 4.59959835e-01 1.97304890e-01 5.94940662e-01
-7.30962098e-01 2.04561561e-01 -6.02202177e-01 7.26169124e-02
5.47976136e-01 1.63503364e-01 -5.93099415e-01 -7.08282530e-01
-9.86452699e-01 6.84955120e-01 -5.97702861e-02 1.01992321e+00
-7.15315700e-01 -5.31657338e-01 -8.55532348e-01 -2.24188298e-01
5.79329014e-01 -2.56717771e-01 4.27909255e-01 -8.32306147e-01
-1.30098641e+00 6.23685062e-01 -4.81809676e-01 2.28632510e-01
3.72608542e-01 -1.31942660e-01 -6.61407590e-01 -1.02831021e-01
-8.28087330e-02 -1.98528558e-01 1.24197769e+00 -1.49011612e+00
2.81529557e-02 -7.21793354e-01 -4.37867284e-01 2.17769928e-02
-3.28575730e-01 5.28137147e-01 -7.91972816e-01 -8.62418890e-01
6.59229994e-01 -7.51632273e-01 -2.83299297e-01 9.21398774e-02
-3.19015026e-01 5.65156899e-03 9.14650619e-01 -8.98098290e-01
1.28689206e+00 -2.43235755e+00 2.27430478e-01 3.83700043e-01
4.83073518e-02 5.10887563e-01 -2.11438090e-01 -8.70260298e-02
-2.08006248e-01 -5.06922789e-02 -4.64195907e-01 -3.60201001e-01
-1.76843122e-01 3.21919531e-01 -7.14986026e-02 8.00314486e-01
1.42069161e-01 4.08653438e-01 -5.15723765e-01 -2.65172750e-01
-2.53966507e-02 8.93390298e-01 -2.71695256e-01 3.13025445e-01
2.35481650e-01 5.36266744e-01 -5.16720414e-01 8.27741146e-01
1.36678302e+00 4.11776192e-02 -3.11416443e-02 -6.61187947e-01
2.89923921e-02 -3.82580996e-01 -2.06383896e+00 1.31163895e+00
-1.51048616e-01 1.56887457e-01 8.77603829e-01 -1.07257926e+00
1.17638052e+00 2.48879462e-01 3.75048161e-01 -4.56061423e-01
2.94545293e-01 4.00627404e-01 -1.40325502e-01 -7.60435939e-01
3.19359228e-02 -6.02868199e-02 5.87166667e-01 -1.32406399e-01
-1.88733265e-01 2.50272602e-01 7.98650607e-02 -2.07811564e-01
9.27126288e-01 -3.53514478e-02 1.23715606e-02 -1.76903039e-01
1.10912108e+00 -7.17934310e-01 1.20756435e+00 4.61916834e-01
-1.36558846e-01 8.36777568e-01 2.19291285e-01 -2.95838118e-01
-5.69716096e-01 -5.42401612e-01 -4.21981275e-01 5.82672656e-01
1.78373218e-01 -4.52772975e-02 -5.12001097e-01 -4.09521163e-01
4.31765690e-02 1.81328595e-01 -3.41342002e-01 6.44354224e-02
-5.61605036e-01 -1.18156815e+00 3.26718301e-01 1.96884543e-01
8.05978835e-01 -7.57579744e-01 3.92657548e-01 1.26602836e-02
-2.40269169e-01 -1.30572593e+00 -5.60830712e-01 -3.33640933e-01
-5.21582246e-01 -1.16257679e+00 -7.78377473e-01 -8.77979755e-01
1.20210040e+00 5.77840388e-01 5.58476210e-01 3.45277071e-01
-5.42858362e-01 3.56149018e-01 -1.65532678e-01 1.44626379e-01
7.30356872e-02 -8.98036659e-01 2.71587402e-01 8.26049030e-01
-3.23878005e-02 -4.46696162e-01 -4.80521828e-01 7.69531727e-01
-1.02081335e+00 -5.73683262e-01 3.86697531e-01 1.09107375e+00
6.69351339e-01 4.09844071e-01 4.30646241e-01 -7.62602329e-01
4.90182608e-01 -6.66319072e-01 -5.12487352e-01 1.39772147e-01
-3.63478541e-01 -2.91316748e-01 5.19928992e-01 -6.47881687e-01
-1.29729402e+00 2.80980859e-02 -2.20413879e-01 -8.46269250e-01
-1.15822218e-01 5.68876505e-01 -8.00063848e-01 -5.19172728e-01
3.07117045e-01 5.05888283e-01 7.09989294e-02 -6.65469289e-01
1.30711108e-01 6.82854533e-01 5.21507502e-01 -6.14645839e-01
1.08042932e+00 4.17178899e-01 1.93746015e-01 -9.80812192e-01
-7.05464542e-01 -4.28037822e-01 -2.21255958e-01 -1.80511504e-01
4.61100549e-01 -1.07376170e+00 -6.41292691e-01 1.04196322e+00
-1.02959311e+00 3.89862299e-01 1.25521332e-01 4.94968891e-01
-1.51468311e-02 8.37907255e-01 -4.26383376e-01 -1.06889534e+00
-8.59362483e-02 -1.34000611e+00 6.64702177e-01 3.22379500e-01
5.90292513e-01 -6.72495484e-01 -3.52384120e-01 5.51551461e-01
6.46895826e-01 2.14902624e-01 3.22840720e-01 -3.72133076e-01
-5.40439248e-01 -2.89675683e-01 -3.37503463e-01 7.09785104e-01
2.16109991e-01 2.36717220e-02 -8.21893990e-01 -3.88394028e-01
5.54621756e-01 -2.76735127e-01 8.25434506e-01 1.58564851e-01
1.39131236e+00 -5.50315559e-01 -1.08086854e-01 9.96834755e-01
1.31293070e+00 1.37246236e-01 9.59874451e-01 -4.89039794e-02
9.27288294e-01 6.12782836e-01 5.49699485e-01 6.49910688e-01
9.12674814e-02 6.80867314e-01 3.55446815e-01 8.71293098e-02
-8.42874870e-03 1.09989047e-01 3.42439115e-01 9.26888227e-01
-7.90441632e-02 -4.03922796e-02 -4.89161849e-01 2.53573537e-01
-1.77647114e+00 -1.14572942e+00 -2.78852284e-01 2.14210725e+00
7.30791450e-01 -2.90409386e-01 -4.00917321e-01 2.99129814e-01
1.07562935e+00 2.93755800e-01 -5.73908687e-01 3.18347901e-01
-5.45529783e-01 -3.92055921e-02 2.04932332e-01 3.02971959e-01
-9.00051892e-01 6.91748202e-01 5.90405750e+00 1.22053802e+00
-1.05174565e+00 8.66805960e-04 5.64895511e-01 1.22628599e-01
-3.51638675e-01 -1.66150033e-01 -8.92448902e-01 7.53034532e-01
3.01741749e-01 1.19349502e-01 6.36619329e-01 5.87841570e-01
3.84840250e-01 2.90207118e-01 -5.12659848e-01 1.54558468e+00
6.16895914e-01 -8.23116481e-01 -5.37356995e-02 -1.93749860e-01
8.62997830e-01 -5.09991705e-01 2.28932589e-01 1.10427216e-01
-4.41511013e-02 -1.17160845e+00 3.44287694e-01 9.91417110e-01
6.72581553e-01 -4.36188042e-01 7.21181571e-01 3.12649131e-01
-1.23008227e+00 -3.33318417e-03 -5.98847389e-01 7.94746876e-02
-2.45802160e-02 1.00515163e+00 2.35318884e-01 5.99672496e-01
4.34532851e-01 9.71353710e-01 -3.06840569e-01 1.12757969e+00
-1.97570369e-01 4.99697000e-01 -3.75845134e-01 5.58650434e-01
-2.85734713e-01 -5.30256867e-01 6.94225252e-01 8.32884014e-01
6.39465511e-01 4.61305231e-01 4.26748544e-01 6.95393980e-01
-2.29811937e-01 3.28693748e-01 -4.62755293e-01 1.56280577e-01
6.12204969e-01 1.41176069e+00 -3.20173323e-01 4.49593691e-03
-5.98435104e-01 8.02980244e-01 1.04289547e-01 6.25375926e-01
-6.36111081e-01 -8.86410773e-02 7.39202023e-01 2.96685137e-02
3.77068609e-01 -2.84981728e-01 -1.11816511e-01 -1.48992574e+00
3.22482556e-01 -1.23122001e+00 1.86239719e-01 -6.61514461e-01
-1.42126012e+00 5.02860010e-01 -4.15527582e-01 -1.08485281e+00
4.44202542e-01 -5.63414097e-01 -7.09147871e-01 8.62187922e-01
-1.69852781e+00 -1.12482285e+00 -5.38852155e-01 8.84264648e-01
3.74066561e-01 -4.01386023e-01 4.13999796e-01 7.86389291e-01
-1.29546201e+00 5.96556365e-01 2.25019768e-01 5.08105271e-02
7.27451622e-01 -4.34909791e-01 -4.26227450e-01 1.16081011e+00
-1.09202534e-01 9.22620356e-01 6.11629963e-01 -6.52052760e-01
-1.95599580e+00 -1.14427865e+00 4.48538214e-01 1.38693094e-01
5.48422635e-01 -2.15695687e-02 -1.19043791e+00 3.69797140e-01
-2.93814570e-01 7.84403801e-01 5.77226639e-01 -1.29000425e-01
-5.10412276e-01 -3.65270793e-01 -1.33092976e+00 2.33252928e-01
1.22975540e+00 -3.54664564e-01 -3.00818801e-01 2.52962083e-01
2.32073471e-01 -4.68357325e-01 -8.28632772e-01 8.92070591e-01
2.95603603e-01 -7.60740340e-01 1.11555564e+00 -3.05356234e-01
7.38018677e-02 -7.90074885e-01 -4.55554783e-01 -1.01630580e+00
-6.59812868e-01 -7.21480012e-01 -1.85345024e-01 1.70315051e+00
-7.00816661e-02 -5.52805364e-01 4.31559741e-01 4.53052759e-01
-9.99293402e-02 -5.49385846e-01 -1.00762272e+00 -7.25989461e-01
-4.65952665e-01 -3.38056803e-01 3.52553517e-01 1.07988226e+00
-5.94403088e-01 -1.20499641e-01 -9.18069303e-01 5.27934551e-01
9.50312912e-01 -2.33810991e-01 6.93260610e-01 -1.13968277e+00
6.12980127e-02 -2.27378294e-01 -6.59976482e-01 -1.05828094e+00
5.52145541e-01 -6.88184142e-01 1.49737313e-01 -1.36718702e+00
2.56379962e-01 -3.43777984e-01 -3.16861868e-01 3.00339729e-01
-6.10878766e-01 5.49412549e-01 8.91269371e-02 5.02234697e-01
-4.72125053e-01 8.59018207e-01 1.21384096e+00 -2.54773617e-01
1.59170061e-01 -5.27311750e-02 -8.42978239e-01 9.63613093e-01
2.76359618e-01 -2.62274653e-01 -1.69044226e-01 -4.54039007e-01
9.17697139e-03 2.34776251e-02 2.33093396e-01 -1.03415143e+00
4.40592259e-01 -2.73045719e-01 4.56521004e-01 -3.09518695e-01
4.05624062e-01 -1.09945142e+00 4.56339449e-01 2.51406915e-02
-2.33456101e-02 -3.64102691e-01 7.65281990e-02 7.67135739e-01
-4.27088410e-01 -2.54592776e-01 1.06671965e+00 1.34471506e-02
-5.84706724e-01 5.64157307e-01 -7.01134279e-02 1.38161033e-01
8.89088750e-01 -2.37824157e-01 -1.05830342e-01 -2.26740435e-01
-8.31701815e-01 2.46455610e-01 1.94332689e-01 2.95981437e-01
9.95487511e-01 -1.58018935e+00 -1.07394731e+00 6.14934921e-01
-1.03732280e-01 -2.73550838e-01 6.03695393e-01 9.61390615e-01
-1.11808628e-01 -1.66638777e-01 6.45390227e-02 -4.27007347e-01
-1.50684261e+00 5.36701083e-01 3.34372431e-01 8.72716382e-02
-2.90352941e-01 9.68060315e-01 3.83582190e-02 -3.69834363e-01
4.24636543e-01 1.78716376e-01 -4.66670781e-01 1.51144937e-01
7.87427485e-01 5.89247942e-01 -3.95044424e-02 -1.24406445e+00
-4.43051815e-01 1.04055679e+00 7.59766474e-02 4.60661560e-01
1.25110209e+00 -2.98991263e-01 -5.22932231e-01 1.33714125e-01
1.12239003e+00 2.36002788e-01 -1.05209804e+00 -5.65875888e-01
-2.06908092e-01 -9.79448617e-01 2.02949315e-01 -3.89635324e-01
-1.50381041e+00 5.15695214e-01 5.40809989e-01 -1.76418066e-01
1.27692151e+00 -5.85918367e-01 6.18317604e-01 1.49032652e-01
4.17469412e-01 -8.00994277e-01 1.51959602e-02 5.87752700e-01
1.12397754e+00 -1.33856952e+00 2.64272511e-01 -8.08811367e-01
-3.02268922e-01 9.24728930e-01 6.48851335e-01 -1.53427660e-01
8.58403981e-01 2.93830276e-01 -2.57038157e-02 7.71144480e-02
-3.65921408e-01 -2.23702669e-01 4.53468204e-01 4.42488849e-01
3.07082534e-01 -2.83263087e-01 -5.30532658e-01 7.08499014e-01
4.20300186e-01 6.11582436e-02 2.78811783e-01 6.52165949e-01
-4.87087607e-01 -8.84518623e-01 -1.05923188e+00 3.05947691e-01
-6.57044530e-01 4.53340672e-02 -3.51040028e-02 1.05119184e-01
3.97046596e-01 1.30824792e+00 -3.44568640e-01 -2.20326945e-01
4.53693300e-01 -8.19239840e-02 3.35733503e-01 -4.56663489e-01
-1.40975893e-01 4.17562276e-01 -1.23644605e-01 -5.46189964e-01
-5.47408104e-01 -6.86732888e-01 -9.11212862e-01 -2.29356974e-01
-6.12923145e-01 5.08903600e-02 5.54764569e-01 8.43922138e-01
2.59758949e-01 3.35489541e-01 9.01467383e-01 -7.66325116e-01
-8.29292774e-01 -9.56276536e-01 -8.83047879e-01 6.31251514e-01
2.29323789e-01 -9.60743070e-01 -9.03934181e-01 -4.09067720e-02] | [12.5476713180542, 0.3956843316555023] |
421f0ca7-beec-41fe-990e-59530468ea1d | debiased-batch-normalization-via-gaussian | 2203.01723 | null | https://arxiv.org/abs/2203.01723v2 | https://arxiv.org/pdf/2203.01723v2.pdf | Debiased Batch Normalization via Gaussian Process for Generalizable Person Re-Identification | Generalizable person re-identification aims to learn a model with only several labeled source domains that can perform well on unseen domains. Without access to the unseen domain, the feature statistics of the batch normalization (BN) layer learned from a limited number of source domains is doubtlessly biased for unseen domain. This would mislead the feature representation learning for unseen domain and deteriorate the generalizaiton ability of the model. In this paper, we propose a novel Debiased Batch Normalization via Gaussian Process approach (GDNorm) for generalizable person re-identification, which models the feature statistic estimation from BN layers as a dynamically self-refining Gaussian process to alleviate the bias to unseen domain for improving the generalization. Specifically, we establish a lightweight model with multiple set of domain-specific BN layers to capture the discriminability of individual source domain, and learn the corresponding parameters of the domain-specific BN layers. These parameters of different source domains are employed to deduce a Gaussian process. We randomly sample several paths from this Gaussian process served as the BN estimations of potential new domains outside of existing source domains, which can further optimize these learned parameters from source domains, and estimate more accurate Gaussian process by them in return, tending to real data distribution. Even without a large number of source domains, GDNorm can still provide debiased BN estimation by using the mean path of the Gaussian process, while maintaining low computational cost during testing. Extensive experiments demonstrate that our GDNorm effectively improves the generalization ability of the model on unseen domain. | ['Zheng-Jun Zha', 'Kecheng Zheng', 'Liang Li', 'Zhipeng Huang', 'Jiawei Liu'] | 2022-03-03 | null | null | null | null | ['generalizable-person-re-identification'] | ['computer-vision'] | [-7.12197274e-02 -6.55635670e-02 2.00012416e-01 -5.50857425e-01
-3.83530587e-01 -8.06945622e-01 4.06169295e-01 -2.57816583e-01
-5.16562819e-01 7.29962230e-01 1.29981022e-02 3.45156521e-01
-2.37257779e-01 -8.76175582e-01 -4.55458552e-01 -8.20479929e-01
4.14665014e-01 9.43685949e-01 7.70483017e-02 -1.49376288e-01
-2.15479657e-01 3.74429196e-01 -1.22598350e+00 -7.20969066e-02
1.08033586e+00 8.61864388e-01 -9.39947069e-02 3.68384004e-01
-8.48027691e-02 2.13628691e-02 -9.91099000e-01 -6.20449424e-01
4.01425153e-01 -2.92256117e-01 -3.92403215e-01 6.12446330e-02
3.96873891e-01 -4.67860520e-01 -4.60607886e-01 1.41841424e+00
6.49947166e-01 3.77988249e-01 9.71027553e-01 -1.21820533e+00
-1.26773810e+00 4.66133237e-01 -2.51978487e-01 2.85997331e-01
1.29146546e-01 3.10354561e-01 4.08582181e-01 -6.67540610e-01
1.90673336e-01 1.50657022e+00 7.60747194e-01 1.06156564e+00
-1.43204415e+00 -1.33537960e+00 4.10010278e-01 -1.98530570e-01
-1.63129044e+00 -4.45474207e-01 7.10433304e-01 -3.77948433e-01
2.76166409e-01 1.73610896e-02 3.27123404e-01 1.58172250e+00
-2.24465221e-01 5.75083971e-01 8.08224916e-01 2.07543984e-01
2.99203485e-01 5.23464680e-01 4.33646113e-01 1.79075107e-01
5.26211977e-01 2.32151657e-01 -4.11704391e-01 -3.95064950e-01
8.88306260e-01 1.76443905e-01 -2.73270547e-01 -3.35216939e-01
-1.07372069e+00 4.93023157e-01 9.15222317e-02 1.02827325e-02
-2.72922277e-01 -2.64249921e-01 2.39659220e-01 3.72101068e-01
4.61893946e-01 1.55841157e-01 -6.80770040e-01 7.19416142e-02
-6.34981215e-01 5.26747644e-01 7.90574670e-01 1.27828217e+00
9.21552777e-01 -5.23368046e-02 -4.46438402e-01 9.56847191e-01
2.10798696e-01 9.90181386e-01 9.00860369e-01 -4.83395606e-01
6.62896574e-01 7.71862745e-01 3.13688576e-01 -9.26292121e-01
-2.04712138e-01 -6.63290620e-01 -1.17112863e+00 -2.53285348e-01
7.37514019e-01 -4.24578220e-01 -1.00207937e+00 1.87576973e+00
3.43675733e-01 4.83111233e-01 1.92712307e-01 7.31007159e-01
5.42522073e-01 4.78726894e-01 3.02914500e-01 2.53541209e-02
1.14696598e+00 -5.60811579e-01 -4.42064345e-01 -2.84370065e-01
2.71383762e-01 -2.10925534e-01 1.02567708e+00 5.35621047e-01
-7.12786317e-01 -1.04175222e+00 -9.79385257e-01 3.20147693e-01
-2.66707897e-01 1.79507330e-01 3.84047627e-01 9.44780469e-01
-7.32160151e-01 5.77097058e-01 -4.19260800e-01 -3.73523980e-01
5.89752436e-01 6.03377581e-01 -4.75620925e-01 -1.71882525e-01
-1.31219995e+00 4.44233209e-01 9.25933123e-01 -3.21403495e-03
-1.02674294e+00 -9.65444684e-01 -7.00963914e-01 1.79144561e-01
1.49791464e-01 -7.69749105e-01 9.86085057e-01 -9.78096247e-01
-1.34512579e+00 7.40519524e-01 -1.51190013e-01 -3.62852305e-01
6.61174059e-01 -2.93968916e-01 -8.69900703e-01 -2.12661207e-01
2.36637518e-01 4.62928981e-01 1.43195224e+00 -1.17467844e+00
-7.66651511e-01 -6.20676517e-01 -4.61341083e-01 2.82867968e-01
-7.80406892e-01 -2.30613440e-01 -2.31613651e-01 -6.68510854e-01
2.45906770e-01 -6.74757957e-01 -1.79307945e-02 -1.74211547e-01
-3.03087294e-01 -4.00771946e-01 5.46552479e-01 -7.12144375e-01
1.20160377e+00 -2.46925664e+00 -1.79544538e-01 4.58811879e-01
4.16279286e-01 1.62220448e-01 -1.13049842e-01 -3.74990910e-01
-8.96418989e-02 1.25443995e-01 -1.37733482e-02 -1.52882665e-01
5.54466993e-02 1.84927642e-01 -3.50119948e-01 5.46900690e-01
1.07601225e-01 6.97802961e-01 -9.56835806e-01 -1.62843168e-01
-8.44313279e-02 2.13726968e-01 -4.47538525e-01 4.25000072e-01
1.62306920e-01 6.27830684e-01 -5.96013725e-01 3.45175385e-01
1.07329357e+00 3.93682718e-03 -2.51361698e-01 -2.50475585e-01
5.28307915e-01 -2.84866124e-01 -1.68941236e+00 1.19397509e+00
5.47678061e-02 -1.79279998e-01 -1.18878096e-01 -9.04465973e-01
1.36703813e+00 5.64389043e-02 2.26333842e-01 -2.86076188e-01
4.94638868e-02 1.03718797e-02 -8.19019526e-02 -2.08629057e-01
1.95633933e-01 -3.99809420e-01 -2.57489175e-01 1.98274910e-01
4.03183192e-01 3.95238042e-01 -2.67879337e-01 -2.07852229e-01
7.51237631e-01 -7.87180066e-02 1.86663955e-01 -2.17858687e-01
7.89139748e-01 -3.25780213e-01 7.45770335e-01 1.08736992e+00
-5.02887487e-01 6.19294465e-01 1.37034759e-01 -4.82698411e-01
-1.14033794e+00 -1.54112756e+00 -2.40407765e-01 1.35074043e+00
2.94297844e-01 5.73071428e-02 -6.82159603e-01 -9.91563618e-01
2.86400169e-01 7.81847179e-01 -5.90682864e-01 -6.33590281e-01
-4.35442328e-01 -1.03119111e+00 8.91537666e-01 5.55253029e-01
8.74625564e-01 -6.82670474e-01 4.55721051e-01 2.21847147e-01
-7.88412988e-03 -8.56225491e-01 -6.86203420e-01 1.81949530e-02
-7.26503074e-01 -9.44956899e-01 -1.04440844e+00 -7.87573755e-01
7.64172852e-01 -9.70770642e-02 6.88595474e-01 -4.55748320e-01
3.66905004e-01 4.07029390e-01 -9.82826054e-02 -3.12317431e-01
-5.98697484e-01 6.84197396e-02 6.45309329e-01 4.12740827e-01
1.06643045e+00 -5.99113524e-01 -4.48105991e-01 5.95647514e-01
-7.14229107e-01 -4.73401219e-01 3.35819244e-01 8.84415030e-01
3.43142271e-01 6.54317796e-01 8.61319602e-01 -8.23775828e-01
9.69499290e-01 -7.13090897e-01 -3.96312803e-01 2.23792493e-01
-7.43005514e-01 1.92810014e-01 8.73349905e-01 -1.22798586e+00
-1.31821394e+00 -2.10622832e-01 1.53613552e-01 -4.42163587e-01
-6.63612068e-01 -1.04233988e-01 -7.41123378e-01 1.82207137e-01
9.19629872e-01 5.23672342e-01 -6.37648255e-02 -6.87182784e-01
2.30682835e-01 7.77279973e-01 7.92198181e-01 -9.16674554e-01
1.34164643e+00 4.62340087e-01 -3.52789909e-01 -2.70575374e-01
-8.13015223e-01 -5.96361279e-01 -7.67350256e-01 1.72528282e-01
6.33927464e-01 -1.02335119e+00 -7.96913564e-01 7.90944636e-01
-9.94708478e-01 -4.28033574e-03 -3.62206131e-01 3.18736255e-01
-1.04229473e-01 4.92091626e-01 -3.05817902e-01 -7.53580153e-01
-2.89132863e-01 -6.72043145e-01 8.05645943e-01 5.70276558e-01
-2.32992142e-01 -1.01842046e+00 -9.50043574e-02 2.73858756e-01
2.90468425e-01 -9.66050178e-02 7.93954074e-01 -1.54804540e+00
-9.63185877e-02 -4.39570874e-01 -2.69257039e-01 8.67405355e-01
2.58846432e-01 -7.04677284e-01 -1.26168418e+00 -5.14555633e-01
3.44187856e-01 6.71535656e-02 6.45162404e-01 7.42249861e-02
1.15815854e+00 -4.57676381e-01 -4.33683783e-01 8.93567026e-01
1.07014120e+00 1.39325157e-01 4.03279483e-01 4.35161680e-01
9.38840389e-01 3.17029059e-01 3.29165012e-01 4.75515664e-01
2.51808614e-01 1.00070789e-01 -9.15886015e-02 1.77736387e-01
1.03249237e-01 -5.76383412e-01 4.03102428e-01 2.44036347e-01
-6.73696175e-02 -1.02832749e-01 -7.36201882e-01 4.38205242e-01
-1.55425870e+00 -8.93289745e-01 3.46929640e-01 2.54709339e+00
9.56626177e-01 2.36265391e-01 2.39063740e-01 -1.92725196e-01
1.18817747e+00 -3.44104469e-01 -1.16431630e+00 2.03016564e-01
-2.85576284e-01 4.49391007e-02 6.45211816e-01 1.36455506e-01
-1.23125744e+00 9.48533058e-01 5.73154354e+00 1.00528538e+00
-7.59547830e-01 7.05415681e-02 5.46747148e-01 1.83239609e-01
-1.78675026e-01 -2.18534172e-01 -1.46051109e+00 8.22749674e-01
7.60656834e-01 -4.03150499e-01 6.11774385e-01 1.24462414e+00
-6.05306104e-02 4.47008640e-01 -1.44921398e+00 1.37557006e+00
3.58051807e-03 -6.01481915e-01 4.21010584e-01 1.50225759e-01
6.92556620e-01 -3.85181427e-01 3.27628046e-01 8.55881870e-01
5.21774173e-01 -1.10665917e+00 4.97248918e-01 6.71512783e-01
7.58649349e-01 -8.25954616e-01 7.62016833e-01 6.05693042e-01
-8.93884122e-01 -4.58779424e-01 -9.32787955e-01 2.32001260e-01
-1.87209114e-01 4.03118014e-01 -9.94425476e-01 3.46610874e-01
6.96907461e-01 5.79453051e-01 -7.69888163e-01 8.12590301e-01
2.69820984e-03 6.68459415e-01 -3.03325653e-01 3.37523937e-01
-2.25124866e-01 -1.89949065e-01 7.33022392e-01 9.24515605e-01
4.17359978e-01 -6.79032579e-02 2.52258152e-01 1.14398646e+00
-9.80951190e-02 1.83059108e-02 -3.36706370e-01 9.21305493e-02
7.02374935e-01 1.01347113e+00 -1.75797209e-01 -3.89567286e-01
-1.39056712e-01 1.33485711e+00 3.42286378e-01 9.10418272e-01
-8.59908819e-01 -1.40211120e-01 8.14028680e-01 2.93915629e-01
4.82964180e-02 1.03942849e-01 -2.11564675e-01 -1.34975839e+00
-7.52781555e-02 -9.39501882e-01 7.19266951e-01 -3.38504821e-01
-2.34176517e+00 4.89086986e-01 1.84792876e-01 -1.23072588e+00
-2.45741326e-02 -5.93361437e-01 -3.74687552e-01 1.42220652e+00
-1.23905492e+00 -1.16276777e+00 -4.01578218e-01 1.02354157e+00
1.56814083e-01 -6.98252082e-01 7.18218744e-01 2.54505813e-01
-5.02185583e-01 1.14841664e+00 3.50301415e-01 4.66337830e-01
1.19871414e+00 -1.26510179e+00 4.04085398e-01 7.00948119e-01
-4.51129198e-01 9.87039745e-01 6.24269128e-01 -9.13228035e-01
-8.23112786e-01 -1.29386497e+00 4.09302801e-01 -6.94166839e-01
5.27454257e-01 -2.89301187e-01 -1.12374926e+00 8.25895727e-01
-6.39573991e-01 -1.51902482e-01 8.16687167e-01 2.67963171e-01
-3.53375793e-01 -5.56373060e-01 -1.62278855e+00 4.61651266e-01
1.09423935e+00 -3.38967711e-01 -8.49225104e-01 8.00228193e-02
6.26951754e-01 -2.12469861e-01 -8.98076475e-01 2.16788381e-01
3.21799755e-01 -4.83300716e-01 1.25890744e+00 -9.49850798e-01
-3.32097083e-01 -4.12635446e-01 2.80445572e-02 -1.38060701e+00
-7.24623084e-01 -4.81712282e-01 -1.05610706e-01 1.71123898e+00
2.04441436e-02 -1.12780762e+00 9.16345358e-01 1.24530005e+00
3.06867748e-01 3.53373475e-02 -8.58109176e-01 -1.03086901e+00
2.55782485e-01 -3.63051444e-01 1.17665601e+00 9.86013114e-01
-4.71801072e-01 2.89988220e-01 -4.03056413e-01 6.74556434e-01
1.03513098e+00 -4.73035306e-01 9.65982318e-01 -1.72791111e+00
-2.21853837e-01 -7.45303407e-02 -4.87873137e-01 -1.36841619e+00
3.50499034e-01 -9.64864075e-01 -1.44080773e-01 -8.83401275e-01
3.27901512e-01 -6.17232382e-01 -5.48544407e-01 3.11277151e-01
-4.95687783e-01 -1.79594427e-01 -1.02225244e-01 3.47153217e-01
-3.94147873e-01 6.44231558e-01 1.27982461e+00 -2.97421426e-01
-4.15616781e-01 4.66250896e-01 -1.13857746e+00 8.29104900e-01
7.51276135e-01 -6.16684377e-01 -6.33379996e-01 -1.83043092e-01
-1.22596376e-01 -5.24617076e-01 6.36124909e-01 -1.17705202e+00
4.29752171e-01 -1.80144399e-01 1.30229306e+00 -3.35478812e-01
1.73110604e-01 -8.38623822e-01 1.33740291e-01 -9.75712091e-02
-2.60237426e-01 -3.40880573e-01 1.08984396e-01 8.77466738e-01
1.13975219e-01 -3.82427067e-01 7.76013017e-01 -1.95192799e-01
-6.69031560e-01 6.73044503e-01 8.12287182e-02 1.85787439e-01
8.55207562e-01 -5.58901370e-01 -1.31594539e-01 -2.34869123e-01
-1.11226797e+00 2.53613383e-01 4.35920745e-01 3.95096928e-01
5.42134464e-01 -1.32104909e+00 -7.98115849e-01 7.48703539e-01
4.58800681e-02 3.05221647e-01 4.93828773e-01 8.96417946e-02
4.22738940e-02 1.21213291e-02 -2.02831864e-01 -5.81107676e-01
-7.05632031e-01 8.93929899e-01 6.14615858e-01 -1.21215872e-01
-4.45942342e-01 8.29344034e-01 6.93179905e-01 -7.80091882e-01
1.40016243e-01 -1.90621674e-01 -2.88044661e-01 -1.15321428e-01
7.75505066e-01 5.22980034e-01 -3.76143366e-01 -6.36932433e-01
-1.97041422e-01 3.72785687e-01 -3.77061874e-01 4.68167141e-02
1.02384079e+00 -4.83909965e-01 7.15184435e-02 1.95695043e-01
9.58768904e-01 -8.37159231e-02 -1.46969962e+00 -4.06270772e-01
-7.02868849e-02 -3.51806372e-01 -3.70255321e-01 -7.52327263e-01
-6.21955514e-01 4.59603757e-01 9.14552331e-01 -1.76189300e-02
1.08706069e+00 -1.67576484e-02 7.91989326e-01 3.47801030e-01
2.56815851e-01 -1.11310458e+00 5.18742343e-03 4.62596208e-01
5.24196148e-01 -1.11024773e+00 -3.88453215e-01 -1.62022874e-01
-7.35334873e-01 1.19609904e+00 9.51696634e-01 -2.20130742e-01
6.94243014e-01 -1.17585495e-01 -1.06653489e-01 9.73362103e-02
-2.21663881e-02 1.12483159e-01 4.83319730e-01 1.25628078e+00
-3.51919621e-01 9.86172855e-02 3.64269346e-01 1.63625455e+00
-3.63031209e-01 -5.58626615e-02 3.46850716e-02 1.20844066e-01
-4.60325629e-01 -9.69444931e-01 -7.31763840e-01 4.99293238e-01
-2.46244550e-01 -1.41590104e-01 1.91982035e-02 5.36041856e-01
5.93500972e-01 7.49043465e-01 2.01184135e-02 -4.06792462e-01
4.91263241e-01 3.85888398e-01 1.88136548e-01 -6.88436210e-01
-3.96774173e-01 -2.61406064e-01 -2.67074883e-01 -8.49093497e-02
1.28991529e-01 -5.93794942e-01 -9.21982825e-01 -3.79417628e-01
-6.37160838e-02 3.47264037e-02 -5.08081680e-03 9.35275078e-01
2.04505458e-01 8.52248967e-02 4.85430747e-01 -4.21538413e-01
-1.23094583e+00 -1.11924303e+00 -9.02894855e-01 8.41631114e-01
3.12934846e-01 -6.60134494e-01 -4.55348462e-01 2.55763084e-01] | [14.717079162597656, 1.1177418231964111] |
a1b773fc-2a20-4862-8fe5-d1819a6f71aa | an-asymmetric-modeling-for-action-assessment | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/7352_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123750222.pdf | An Asymmetric Modeling for Action Assessment | Action assessment is a task of assessing the performance of an action. It is widely applicable to many real-world scenarios such as medical treatment and sporting events. However, existing methods for action assessment are mostly limited to individual actions, especially lacking modeling of the asymmetric relations among agents (e.g., between persons and objects); and this limitation undermines their ability to assess actions containing asymmetrically interactive motion patterns, since there always exists subordination between agents in many interactive actions. In this work, we model the asymmetric interactions among agents for action assessment. In particular, we propose an asymmetric interaction module (AIM), to explicitly model asymmetric interactions between intelligent agents within an action, where we group these agents into a primary one (e.g., human) and secondary ones (e.g., objects). We perform experiments on JIGSAWS dataset containing surgical actions, and additionally collect a new dataset, TASD-2, for interactive sporting actions. The experimental results on two interactive action datasets show the effectiveness of our model, and our method achieves state-of-the-art performance. The extended experiment on AQA-7 dataset also demonstrates the generalization capability of our framework to conventional action assessment. | ['Yao-Wei Wang', 'Jian-Huang Lai', 'Jia-Hui Pan', 'Wei-Shi Zheng', 'Chengying Gao', 'Wei Zeng', 'Jibin Gao'] | null | null | null | null | eccv-2020-8 | ['action-assessment'] | ['computer-vision'] | [ 1.08362727e-01 -1.02495447e-01 -1.20552972e-01 -2.09917650e-01
-2.40817189e-01 -5.05735397e-01 8.30485821e-01 -1.84377208e-01
-5.84871829e-01 5.62495112e-01 6.20065808e-01 -1.04345664e-01
-3.20594847e-01 -8.57274532e-01 -2.76495159e-01 -8.62488091e-01
-2.03240424e-01 7.69071162e-01 8.59911323e-01 -5.30229151e-01
1.94751993e-01 4.89670277e-01 -1.78853512e+00 3.89110357e-01
8.59084666e-01 7.16154039e-01 -2.30559856e-02 6.31123602e-01
2.19780773e-01 1.27642584e+00 -6.88395858e-01 -3.15140426e-01
4.00591254e-01 -6.17515266e-01 -8.87801290e-01 2.70655364e-01
2.36367881e-01 -8.65594029e-01 -5.03415883e-01 7.19553530e-01
5.12014389e-01 4.90504503e-01 7.08260417e-01 -1.82563210e+00
-1.46186680e-01 5.22671878e-01 -3.89534026e-01 1.85313761e-01
6.06335998e-01 6.01059139e-01 9.57150221e-01 -5.46464622e-01
4.55370039e-01 1.58184838e+00 2.24320531e-01 5.86045563e-01
-6.41624749e-01 -6.10074818e-01 4.98271108e-01 5.40826082e-01
-8.17433417e-01 -1.24630488e-01 5.62240839e-01 -5.53092003e-01
6.09389365e-01 3.22666973e-01 9.67766523e-01 1.08733797e+00
1.26878500e-01 1.36001956e+00 8.49198043e-01 1.04982436e-01
3.34527612e-01 -5.59686661e-01 6.88972920e-02 3.16475421e-01
7.46974200e-02 3.51442732e-02 -4.84647483e-01 -9.98572707e-02
8.38186741e-01 7.53171593e-02 1.60688609e-01 -4.89114404e-01
-1.54410863e+00 5.98031461e-01 1.38344020e-01 4.59656864e-02
-5.96393108e-01 1.44995645e-01 4.67830032e-01 9.58352387e-02
2.30599359e-01 1.63167030e-01 -1.10859998e-01 -5.13777912e-01
-9.10857469e-02 7.05954432e-01 4.62409139e-01 6.69105589e-01
4.87776458e-01 -2.30535239e-01 -5.69507003e-01 6.77871823e-01
6.01998717e-02 4.67910886e-01 1.91956863e-01 -1.43627644e+00
6.66420400e-01 1.09735441e+00 4.50276136e-01 -1.16242301e+00
-6.99254572e-01 -8.05417821e-02 -7.85492480e-01 6.15551114e-01
4.14147228e-01 -9.70616862e-02 -5.40814102e-01 1.73362517e+00
7.39014924e-01 2.73129880e-01 1.83441684e-01 1.26803827e+00
8.29059601e-01 2.25123346e-01 2.90671498e-01 -1.54392689e-01
1.26593459e+00 -1.35274792e+00 -7.52762854e-01 -2.35910371e-01
8.55971396e-01 -2.77633220e-01 1.19765198e+00 3.08778107e-01
-1.10748982e+00 -6.13043904e-01 -6.32259667e-01 2.91499168e-01
-1.07522398e-01 3.90701834e-03 7.73133934e-01 1.92230523e-01
-5.85777164e-01 4.57221478e-01 -8.20845485e-01 -3.36428881e-01
2.15280324e-01 3.62296969e-01 -5.32301843e-01 1.45305648e-01
-1.38323462e+00 8.26074958e-01 4.64246064e-01 1.03835978e-01
-1.01628602e+00 -2.02047274e-01 -7.75292814e-01 -2.71174967e-01
9.80789721e-01 -5.93475342e-01 1.28794515e+00 -7.57614911e-01
-1.42669559e+00 4.80135024e-01 3.48288238e-01 -2.39332855e-01
8.85588884e-01 -3.96262914e-01 -3.73439908e-01 2.94904262e-01
2.38924950e-01 5.29095232e-01 2.27987692e-01 -1.14837122e+00
-1.22329438e+00 -5.29175282e-01 1.06186414e+00 8.64119530e-01
-4.74712819e-01 1.17893673e-01 -5.83654225e-01 -5.90662181e-01
8.27435777e-02 -1.27493322e+00 -2.72492021e-01 3.66653912e-02
-4.00739878e-01 -5.69518983e-01 9.00176525e-01 -2.92591840e-01
1.32693350e+00 -1.85982442e+00 5.66159487e-01 -9.72025171e-02
1.91549897e-01 2.34001517e-01 -2.92318016e-01 5.87454736e-01
2.38765121e-01 -1.22189783e-01 -3.53619605e-01 -2.25918725e-01
2.10823998e-01 5.13717830e-01 2.10871547e-01 3.44024062e-01
-6.93347678e-02 7.75970221e-01 -1.29577780e+00 -8.71390641e-01
4.03115273e-01 5.86774349e-02 -4.32312697e-01 2.96946019e-01
-7.35937729e-02 7.28524923e-01 -9.37948823e-01 5.66058099e-01
4.62363660e-01 1.17967464e-01 6.32570162e-02 -1.55430675e-01
-6.31228238e-02 1.02532625e-01 -1.62113917e+00 1.43847728e+00
-6.06315136e-02 1.97363898e-01 -1.74551830e-01 -7.20473588e-01
4.79850620e-01 1.81698084e-01 1.03047371e+00 -6.69109583e-01
-5.49725667e-02 -6.68594167e-02 3.39145035e-01 -7.80219436e-01
3.95661741e-01 1.23897716e-01 -3.21987689e-01 7.61908889e-01
-3.63718390e-01 -4.25329059e-02 7.52208054e-01 3.98672819e-01
1.26667261e+00 3.91206712e-01 3.57971668e-01 -1.74238998e-02
6.65535569e-01 7.63255656e-02 6.20368838e-01 8.52012336e-01
-5.33384144e-01 3.15639347e-01 5.82382202e-01 -6.60622478e-01
-5.78725100e-01 -9.21391606e-01 3.87499809e-01 1.31076670e+00
9.13709641e-01 -4.22586113e-01 -7.59761691e-01 -1.06633914e+00
-1.61155939e-01 5.03973842e-01 -9.15571630e-01 -3.38866025e-01
-9.70844805e-01 -7.05355942e-01 5.01386285e-01 9.08516467e-01
1.00010908e+00 -1.61167288e+00 -9.97619033e-01 2.95585364e-01
-5.96060455e-01 -1.21410477e+00 -4.64504212e-01 -6.03172660e-01
-5.25185704e-01 -1.49094927e+00 -4.53029126e-01 -3.75993103e-01
3.83774668e-01 3.52532357e-01 1.10438108e+00 3.79285097e-01
1.12876728e-01 6.34206355e-01 -6.73645496e-01 -2.66879082e-01
-3.70988935e-01 -2.52218843e-01 4.14431930e-01 2.29713619e-01
2.84405351e-01 -3.25048655e-01 -1.01953447e+00 1.13808107e+00
-9.70648646e-01 2.08293349e-01 7.30176508e-01 5.46628296e-01
3.16765457e-01 2.94714749e-01 2.41594955e-01 -5.72696328e-01
5.81368685e-01 -1.94039345e-01 3.08672618e-02 2.18304917e-01
6.11651056e-02 -3.50891769e-01 3.20547521e-01 -7.60819852e-01
-1.22414875e+00 -9.08088088e-02 1.11271508e-01 -3.85201760e-02
-2.89977908e-01 3.82730305e-01 -4.63049352e-01 2.36813918e-01
5.49540758e-01 4.97035831e-02 -7.25967288e-02 -1.45422190e-01
1.11978374e-01 4.94485438e-01 5.72744787e-01 -6.15949392e-01
6.55384183e-01 6.79381132e-01 1.57622740e-01 -4.95080084e-01
-6.91867769e-01 -4.65173602e-01 -7.11096168e-01 -9.55145299e-01
1.00909960e+00 -5.49433172e-01 -1.21841979e+00 9.55770969e-01
-9.05739069e-01 -6.54261351e-01 -2.77093261e-01 7.64590800e-01
-8.70353162e-01 5.35916507e-01 -6.87860131e-01 -8.08249235e-01
9.33476686e-02 -1.37895334e+00 1.26248741e+00 8.04997087e-02
-2.04985559e-01 -7.40583420e-01 1.45728394e-01 7.05885947e-01
-1.03544988e-01 4.32790011e-01 3.85760635e-01 -4.82918113e-01
-3.86799186e-01 -6.55944124e-02 1.19810537e-01 1.78152829e-01
3.59328121e-01 -1.95979029e-01 -3.25019091e-01 -9.09475610e-02
-3.14334422e-01 -5.00367105e-01 5.67800701e-01 1.32757843e-01
1.00525367e+00 -1.97110489e-01 -3.10967118e-01 -2.38244534e-01
5.71638048e-01 4.41850483e-01 1.06749833e+00 3.81244183e-01
7.61632800e-01 1.03550887e+00 1.52109623e+00 5.83549142e-01
4.48645145e-01 1.15895832e+00 7.76710629e-01 -1.42947599e-01
-1.08991869e-01 1.79949161e-02 4.66284662e-01 5.31293213e-01
-9.24813330e-01 -5.50501466e-01 -7.68633008e-01 3.41393203e-01
-2.39402318e+00 -1.29064476e+00 -4.34201360e-01 2.04181361e+00
5.10942698e-01 2.26780728e-01 6.06504858e-01 2.40924194e-01
6.22853696e-01 1.62031576e-01 -7.67825902e-01 2.84254164e-01
4.61656116e-02 -5.27031422e-01 2.26676296e-02 3.12044650e-01
-1.30767703e+00 9.57392991e-01 5.61690855e+00 1.03053474e+00
-3.32024932e-01 -1.21179134e-01 3.42486560e-01 -7.80226067e-02
9.40250084e-02 -2.04249620e-01 -5.74708343e-01 5.44284999e-01
3.53188291e-02 2.37528961e-02 5.40989190e-02 5.32636881e-01
4.68094796e-01 -2.47229174e-01 -1.09316218e+00 5.95352829e-01
-4.12437990e-02 -6.77020133e-01 3.37487519e-01 6.07814454e-02
3.91438216e-01 -6.35096431e-01 -1.83426142e-01 4.93054390e-01
5.97175479e-01 -7.47357845e-01 7.58053899e-01 3.88992310e-01
2.51285642e-01 -6.32952452e-01 9.73390996e-01 6.65615439e-01
-1.50694728e+00 -2.23916411e-01 1.78179387e-02 -4.33152586e-01
5.84543288e-01 -1.47723973e-01 -2.66232789e-01 7.15262473e-01
8.03892136e-01 7.74924338e-01 -4.82889473e-01 8.08045268e-01
-3.26979280e-01 2.65047550e-01 -1.42926946e-01 -6.51803613e-02
3.98970842e-01 -5.39606690e-01 7.28974819e-01 6.88562632e-01
6.46623522e-02 6.50970340e-01 5.82790971e-01 1.16874382e-01
5.25951803e-01 -3.73966852e-03 -4.68261421e-01 9.54337418e-02
2.24774882e-01 1.08434784e+00 -7.59287059e-01 -5.50297201e-01
-3.17794442e-01 8.89335990e-01 1.42606512e-01 2.70054787e-01
-1.17952287e+00 1.22855119e-02 9.81257558e-01 2.06993788e-01
-1.92013055e-01 -2.16496930e-01 1.64188087e-01 -9.25918102e-01
1.63510725e-01 -1.19822145e+00 7.32295036e-01 -8.39015543e-01
-1.01875603e+00 3.33741248e-01 4.85313147e-01 -1.86559379e+00
-4.14347738e-01 -4.52648669e-01 -6.35542750e-01 2.55729407e-01
-6.24527752e-01 -1.25613236e+00 -7.29263425e-01 7.44369030e-01
7.80471265e-01 -7.20790699e-02 3.41300070e-01 2.54902422e-01
-3.77584875e-01 2.27230802e-01 -5.89805484e-01 2.10068345e-01
4.79559124e-01 -1.18875158e+00 6.11852445e-02 6.26842618e-01
-1.08106593e-02 3.01441401e-01 7.69646883e-01 -8.28787148e-01
-1.11061442e+00 -8.60122025e-01 3.33361357e-01 -5.69760203e-01
5.90995312e-01 -1.64606959e-01 -7.35112131e-01 7.01202393e-01
-4.35141567e-03 -2.89605319e-01 5.21477103e-01 -6.19851798e-02
2.08887234e-01 1.15121953e-01 -9.96187627e-01 1.19707906e+00
1.90259492e+00 2.05128323e-02 -5.95595241e-01 3.99517864e-01
6.45475328e-01 -5.14106929e-01 -5.74234903e-01 8.52590621e-01
8.05951118e-01 -1.25534081e+00 1.24539387e+00 -9.88995552e-01
5.77064037e-01 -5.78530788e-01 9.66536719e-03 -1.31494987e+00
-3.14772636e-01 -2.08600357e-01 -1.68498099e-01 9.17427182e-01
-3.62736583e-02 -5.09772003e-01 8.88127208e-01 6.37927949e-01
-2.46915102e-01 -7.41301179e-01 -7.96799123e-01 -9.79866087e-01
-2.14968860e-01 -2.90498883e-01 7.39413023e-01 8.05159867e-01
-4.59565856e-02 1.59713253e-01 -6.04069591e-01 1.37538776e-01
5.26036561e-01 4.24871184e-02 1.25423110e+00 -1.15865934e+00
-6.21050239e-01 -4.56117898e-01 -6.87075377e-01 -1.20354104e+00
1.16458267e-01 -1.69712052e-01 9.24099758e-02 -1.93139446e+00
3.67877394e-01 -3.49153578e-01 -7.86281079e-02 5.55520415e-01
-4.27047640e-01 4.61312950e-01 2.63840228e-01 3.34148884e-01
-1.06761849e+00 7.05054820e-01 1.82537460e+00 -2.44667023e-01
-2.90109545e-01 3.71144891e-01 -2.28978202e-01 1.13591707e+00
8.14100564e-01 -1.98627919e-01 -7.49556959e-01 -3.35209221e-01
1.24574520e-01 1.16956465e-01 5.01416922e-01 -1.13423991e+00
2.08256677e-01 -6.78244233e-01 -2.02580824e-01 -6.21110976e-01
5.95353246e-01 -7.13957667e-01 2.88815379e-01 7.26245284e-01
-3.46100450e-01 -9.61283669e-02 -1.88928545e-01 5.87999284e-01
-2.74317026e-01 -1.42887473e-01 2.39408121e-01 -1.17651954e-01
-9.98311102e-01 5.84197640e-01 -5.52060068e-01 1.39656469e-01
1.42524707e+00 -4.03031677e-01 -3.98187101e-01 -6.40514791e-01
-9.05315161e-01 8.05772126e-01 2.81010419e-01 4.71084893e-01
4.92835373e-01 -1.67360580e+00 -9.34524596e-01 -4.06973422e-01
4.02756035e-01 8.47066566e-02 5.61141789e-01 9.04698849e-01
-3.55605930e-01 -1.53183788e-01 -3.95240694e-01 -5.33471346e-01
-1.69384420e+00 3.02162528e-01 2.71198809e-01 -4.43335801e-01
-6.88134193e-01 3.24020386e-01 7.50383496e-01 -5.64494967e-01
2.34048322e-01 -2.30327714e-02 -8.03290844e-01 8.12796727e-02
6.96487725e-01 8.56939137e-01 -4.39937443e-01 -9.26891983e-01
-4.49185282e-01 6.29213989e-01 1.61649823e-01 -2.08055913e-01
1.13091934e+00 -4.74333279e-02 3.02992947e-02 3.90760571e-01
5.20922840e-01 -2.54869670e-01 -1.54196835e+00 -4.39776704e-02
-3.14028263e-01 -5.66788852e-01 -5.25484741e-01 -5.60689747e-01
-9.77486134e-01 7.06996262e-01 4.33848649e-01 2.83086002e-01
1.14310527e+00 2.16475740e-01 6.22246325e-01 5.67984343e-01
8.07073891e-01 -1.34935939e+00 7.73114026e-01 3.51249695e-01
1.22674143e+00 -1.18004048e+00 8.15888792e-02 -6.02775633e-01
-1.05946577e+00 6.94943130e-01 1.16178763e+00 1.23480052e-01
2.04835832e-01 5.97995482e-02 3.72770019e-02 -4.14074421e-01
-5.03023684e-01 -4.48322117e-01 1.19487420e-01 5.49573064e-01
-1.83903560e-01 2.39138439e-01 -5.98130703e-01 4.15610403e-01
1.28941908e-01 -5.55014648e-02 5.67581356e-01 1.16483366e+00
-3.26904535e-01 -9.99544680e-01 -2.53826708e-01 3.65221918e-01
-1.11856483e-01 4.32483941e-01 -5.73233664e-01 1.03725958e+00
3.00710857e-01 1.14220953e+00 2.23307714e-01 -5.17037690e-01
7.64434814e-01 -5.16973138e-01 3.44108790e-01 -4.05554086e-01
-6.09712183e-01 -3.09944779e-01 3.55431914e-01 -8.90788019e-01
-9.96902883e-01 -8.76454115e-01 -1.35349417e+00 -4.20776546e-01
4.41580527e-02 -4.60718796e-02 -5.82568385e-02 9.59987581e-01
5.51631451e-02 5.71143270e-01 4.78167385e-01 -9.04717445e-01
-3.89273137e-01 -1.03212631e+00 -5.89236557e-01 1.04599857e+00
-1.53272375e-01 -1.36028886e+00 -3.74465585e-01 -3.49450149e-02] | [8.220709800720215, 0.5701557397842407] |
24001142-4d57-4025-b36c-1486e8d54fbd | edu-ap-elementary-discourse-unit-based | null | null | https://aclanthology.org/2022.sigdial-1.19 | https://aclanthology.org/2022.sigdial-1.19.pdf | EDU-AP: Elementary Discourse Unit based Argument Parser | Neural approaches to end-to-end argument mining (AM) are often formulated as dependency parsing (DP), which relies on token-level sequence labeling and intricate post-processing for extracting argumentative structures from text. Although such methods yield reasonable results, operating solely with tokens increases the possibility of discontinuous and overly segmented structures due to minor inconsistencies in token level predictions. In this paper, we propose EDU-AP, an end-to-end argument parser, that alleviates such problems in dependency-based methods by exploiting the intrinsic relationship between elementary discourse units (EDUs) and argumentative discourse units (ADUs) and operates at both token and EDU level granularity. Further, appropriately using contextual information, along with optimizing a novel objective function during training, EDU-AP achieves significant improvements across all four tasks of AM compared to existing dependency-based methods. | ['Rohini Srihari', 'Souvik Das', 'Sougata Saha'] | null | null | null | null | sigdial-acl-2022-9 | ['dependency-parsing', 'argument-mining'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.63004231e-01 8.40775907e-01 -5.10762572e-01 -3.95364732e-01
-1.22993600e+00 -8.80087197e-01 6.01058900e-01 6.76209390e-01
-5.11840105e-01 1.04630089e+00 6.94210649e-01 -8.92452717e-01
1.46402881e-01 -7.54618645e-01 -7.80286670e-01 -1.34714186e-01
-2.05235481e-02 4.86044586e-01 1.51460052e-01 -1.64750755e-01
3.87556493e-01 -1.81164742e-01 -1.23132694e+00 9.60301518e-01
1.03509116e+00 6.20355070e-01 -1.37734879e-02 8.78062904e-01
-5.85575879e-01 1.29213011e+00 -7.99791932e-01 -5.58547080e-01
-2.24440306e-01 -6.74893439e-01 -1.59299695e+00 -4.42138582e-01
2.22336426e-01 -2.37444431e-01 3.42185497e-01 7.29256570e-01
2.28460476e-01 1.47464750e-02 6.08523190e-01 -5.97318053e-01
-5.50441980e-01 1.42426109e+00 -4.38585937e-01 2.74338901e-01
2.86521226e-01 -1.69472262e-01 1.71912742e+00 -8.89155984e-01
7.72659361e-01 1.48643994e+00 8.00450802e-01 4.08557832e-01
-1.26242208e+00 -4.50801626e-02 4.85417306e-01 3.12738419e-02
-3.11722308e-01 -3.54699343e-01 8.73467147e-01 -3.18322629e-01
1.51863790e+00 1.43240809e-01 5.23084879e-01 1.04953158e+00
-1.40515313e-01 1.36775708e+00 7.41983891e-01 -7.49512613e-01
2.10422099e-01 -5.60256720e-01 6.41563177e-01 5.74690104e-01
1.37169555e-01 -3.01027179e-01 -3.50651562e-01 -2.50942111e-01
3.05211931e-01 -8.35658193e-01 2.33935371e-01 3.69800419e-01
-1.01503754e+00 1.10991275e+00 1.96893051e-01 3.88206452e-01
-4.56923902e-01 6.69569373e-02 9.54409778e-01 1.88451499e-01
5.15850246e-01 6.40732706e-01 -6.06240809e-01 -4.27073270e-01
-7.70220757e-01 3.56243730e-01 9.84869063e-01 6.53753877e-01
3.20871592e-01 8.77398998e-02 -3.86965424e-01 8.53679657e-01
1.55726641e-01 -3.03009152e-01 5.03978670e-01 -9.40961421e-01
1.00360858e+00 8.09377611e-01 -3.30664068e-02 -3.90148789e-01
-4.49456513e-01 -2.75901318e-01 -1.95254803e-01 7.85999522e-02
8.24126601e-01 -5.74236214e-01 -7.33161032e-01 1.99414790e+00
5.07012963e-01 -3.01277161e-01 4.07270998e-01 5.74154496e-01
9.63126421e-01 5.63029110e-01 6.62665904e-01 -1.70490444e-01
1.63158727e+00 -1.08529639e+00 -6.96301997e-01 -5.41125298e-01
1.21707177e+00 -6.20200634e-01 1.20239413e+00 1.96408685e-02
-1.60676169e+00 -3.69538844e-01 -1.02025163e+00 -5.29337406e-01
-5.57821319e-02 -1.89979933e-02 5.90872645e-01 3.67573977e-01
-5.72095096e-01 6.35770261e-01 -7.77293026e-01 2.75613368e-01
5.24910390e-01 3.10301244e-01 -8.91585946e-02 3.44242513e-01
-1.33304656e+00 8.74040723e-01 9.61202681e-01 -8.36946890e-02
-1.98051319e-01 -6.02613091e-01 -1.11330366e+00 2.81351000e-01
4.92788881e-01 -6.45135641e-01 1.64007807e+00 -1.12749219e+00
-1.51129639e+00 8.22412550e-01 -3.32708299e-01 -8.27988803e-01
4.06882495e-01 -7.13675022e-01 1.30323350e-01 1.60472453e-01
2.55224079e-01 5.43874502e-01 5.14216840e-01 -9.59213078e-01
-7.83757865e-01 -2.09386766e-01 2.37345889e-01 4.19864625e-01
-1.27253670e-03 2.06862748e-01 1.08446658e-01 -6.41785264e-01
1.05812289e-01 -5.22495151e-01 -1.99432656e-01 -6.87666118e-01
-6.00615859e-01 -9.31151271e-01 6.83055103e-01 -9.29173052e-01
1.30076766e+00 -1.80348980e+00 1.23377889e-01 -1.53131768e-01
1.00711450e-01 4.28422838e-01 -1.06915548e-01 4.17807579e-01
-1.35534495e-01 3.21448147e-01 -4.47711796e-01 -4.31059480e-01
7.87463188e-02 4.02685970e-01 -1.05060816e-01 2.95489170e-02
7.12992549e-01 1.12117732e+00 -1.04508471e+00 -7.92487979e-01
-3.60302590e-02 1.18029706e-01 -7.40101814e-01 2.28724316e-01
-7.88247406e-01 3.87584239e-01 -5.63791871e-01 6.48463309e-01
2.06904784e-01 -2.82967389e-01 7.30348766e-01 1.75239459e-01
-2.93357104e-01 1.41171396e+00 -7.01611757e-01 1.61133230e+00
-3.39871258e-01 5.67634642e-01 5.29064499e-02 -1.42265856e+00
6.14588022e-01 4.50099647e-01 1.72099650e-01 -6.19543433e-01
2.91037053e-01 3.86826009e-01 3.62040162e-01 -3.97568792e-01
6.77257001e-01 -1.02781586e-01 -3.01912129e-01 5.96377730e-01
-7.76687413e-02 3.29050332e-01 5.42835176e-01 5.00013940e-02
1.28432786e+00 4.49594110e-01 5.22324324e-01 -2.47713387e-01
4.27118212e-01 4.02580053e-01 7.81693101e-01 7.18017519e-01
1.71964709e-02 3.87242854e-01 9.15550113e-01 -4.44137394e-01
-1.40997815e+00 -9.14836049e-01 -1.26512498e-01 1.23082292e+00
-3.76552284e-01 -5.42792559e-01 -1.09960413e+00 -1.03936398e+00
-1.45142928e-01 8.94817054e-01 -6.18962705e-01 3.59923720e-01
-1.49431014e+00 -8.01509142e-01 9.36277211e-01 9.64901567e-01
4.31199431e-01 -1.47823727e+00 -9.27174747e-01 9.01568890e-01
-4.91671324e-01 -1.15080929e+00 1.01268105e-01 6.15667284e-01
-9.37962890e-01 -1.15201485e+00 -3.38722676e-01 -9.27928269e-01
4.20861632e-01 -4.25289840e-01 1.49602258e+00 3.45627040e-01
-5.75650076e-04 -3.68485600e-01 -6.21293485e-01 -4.61660326e-01
-6.66116059e-01 4.12325233e-01 -5.82367122e-01 -6.49797916e-01
2.94146687e-01 -4.33257520e-01 -2.79636443e-01 -3.58513445e-01
-4.94478494e-01 2.00003728e-01 5.57796955e-01 1.15048230e+00
5.14938414e-01 -4.81698215e-01 9.41833854e-01 -1.36651754e+00
8.92295957e-01 -4.80377197e-01 -2.93916315e-01 9.11729336e-02
-4.04301375e-01 3.13420862e-01 7.00283468e-01 -2.57756054e-01
-1.33932686e+00 -2.83122480e-01 -6.29869640e-01 4.67810422e-01
-1.14043668e-01 7.42219150e-01 9.93336551e-03 7.84715474e-01
7.49181449e-01 -3.02016497e-01 -7.55532309e-02 -3.87488782e-01
4.00873452e-01 4.22061503e-01 6.38832688e-01 -1.09115338e+00
3.35413903e-01 8.37299153e-02 -1.95180833e-01 -7.18035042e-01
-1.44556725e+00 -7.91158825e-02 -7.38523722e-01 3.19436252e-01
1.05732822e+00 -7.01179504e-01 -7.27914155e-01 -4.59246263e-02
-1.40859807e+00 -6.23969495e-01 -5.00026226e-01 2.84347534e-01
-5.67819297e-01 5.63176095e-01 -1.16242135e+00 -8.57272148e-01
-7.43020952e-01 -7.49567151e-01 8.37234914e-01 2.24301368e-01
-9.05261397e-01 -1.13156796e+00 -7.09747672e-02 5.07238746e-01
-4.89790849e-02 5.19656479e-01 1.51666534e+00 -9.33728755e-01
-1.44503579e-01 1.25218406e-01 -1.86451048e-01 3.19905490e-01
-1.16652295e-01 -1.20374098e-01 -8.25055778e-01 1.74106658e-01
-1.72758609e-01 -5.94683111e-01 7.85553694e-01 5.68035781e-01
8.94110084e-01 -6.30283475e-01 -1.08640581e-01 2.45867521e-01
9.40775990e-01 1.42642543e-01 5.01531005e-01 9.37106907e-01
5.78981161e-01 8.83983135e-01 7.73183763e-01 2.45820269e-01
4.96571481e-01 1.75045013e-01 4.01377201e-01 -2.17048168e-01
-7.43182823e-02 -3.72073442e-01 2.74425149e-01 4.60755616e-01
-5.68382591e-02 -4.03706044e-01 -7.89012253e-01 8.10726464e-01
-2.02900028e+00 -1.11440790e+00 -6.07441902e-01 1.65238595e+00
1.17762768e+00 5.49451709e-01 3.44270647e-01 4.92508769e-01
5.47295988e-01 2.03006074e-01 -5.10252714e-01 -9.98634636e-01
-2.26777032e-01 5.09947538e-01 1.40024930e-01 5.29354930e-01
-1.12538755e+00 1.20136428e+00 6.54058790e+00 4.70017165e-01
-4.41970378e-01 -9.66238081e-02 6.50590062e-01 2.67113358e-01
-4.16461051e-01 3.67450923e-01 -9.18717325e-01 2.85382330e-01
9.33967650e-01 1.80240616e-01 -7.76968077e-02 8.60370636e-01
-7.06703663e-02 -9.33568478e-02 -1.17120326e+00 1.29829824e-01
-3.86929005e-01 -1.62988317e+00 -1.23636082e-01 -5.92115037e-02
4.15956169e-01 -6.12136349e-02 -2.58399367e-01 4.37682569e-01
6.75475836e-01 -1.00772595e+00 8.38205218e-01 -3.14209104e-01
3.71282339e-01 -8.02460253e-01 7.70322621e-01 5.24349034e-01
-1.13717639e+00 -1.99129984e-01 -1.20088480e-01 -4.62225944e-01
3.80270660e-01 4.88940835e-01 -1.08904672e+00 5.45377970e-01
2.51451850e-01 3.48165542e-01 -9.79369879e-02 2.68148452e-01
-8.20936620e-01 1.12298214e+00 -3.58300477e-01 -1.58601180e-01
7.64913321e-01 2.80922186e-02 6.01639509e-01 1.61157501e+00
-1.79930300e-01 2.71462351e-01 1.86265886e-01 7.77318656e-01
-1.94131687e-01 2.46366709e-01 -2.72920698e-01 -1.88656285e-01
6.65026426e-01 1.02041602e+00 -7.59904802e-01 -6.13137722e-01
-2.96882927e-01 6.68576539e-01 9.10991371e-01 -2.01533064e-02
-7.35362351e-01 -1.83683902e-01 6.14545524e-01 3.69688235e-02
3.65797848e-01 -2.42027551e-01 -6.96917534e-01 -7.77269363e-01
2.33217075e-01 -8.46473098e-01 8.57645988e-01 -1.77446052e-01
-1.12429297e+00 4.97354209e-01 -6.46453574e-02 -7.48259783e-01
-7.69548535e-01 -6.71298742e-01 -9.56522584e-01 8.02840233e-01
-1.60234320e+00 -1.31791711e+00 4.28997815e-01 1.14843503e-01
9.53182042e-01 1.56977028e-01 8.32592309e-01 -4.73317765e-02
-4.05619591e-01 6.87959254e-01 -2.84837574e-01 6.23271048e-01
2.57343352e-01 -1.58796680e+00 8.50942731e-01 8.78620625e-01
1.10637341e-02 5.85269809e-01 5.74368358e-01 -7.28044868e-01
-9.09188747e-01 -8.40030372e-01 1.28166366e+00 -3.97070080e-01
5.93715131e-01 -1.20358720e-01 -1.06436634e+00 7.68717051e-01
3.23729396e-01 -3.02134663e-01 7.69593179e-01 6.98864937e-01
-2.96600401e-01 8.85846972e-01 -8.79100800e-01 6.00293338e-01
1.13258481e+00 -3.88910621e-01 -1.25168407e+00 1.53527781e-01
8.88226688e-01 -7.53527045e-01 -8.72024357e-01 3.42578024e-01
4.92611945e-01 -9.72310066e-01 9.94313419e-01 -1.02239621e+00
1.00710630e+00 3.99820507e-02 1.08577587e-01 -9.99612331e-01
-1.65346354e-01 -4.73523110e-01 -3.63355994e-01 1.37756276e+00
8.23975682e-01 -4.17251766e-01 8.30329061e-01 5.03074169e-01
-6.51400328e-01 -8.45377386e-01 -7.52917051e-01 -5.56606472e-01
4.74440217e-01 -4.05225396e-01 4.27665263e-01 1.01066768e+00
4.14067030e-01 9.14411128e-01 -4.03028205e-02 -1.51831135e-01
5.79184830e-01 2.35141560e-01 4.54919875e-01 -1.29890084e+00
-4.35520470e-01 -6.21261775e-01 3.55215222e-01 -1.00767040e+00
5.93934238e-01 -9.73120928e-01 8.39400887e-02 -1.76134551e+00
-2.85858989e-01 -4.74942505e-01 3.03649530e-02 6.02612197e-01
-5.51907420e-01 -1.28484502e-01 9.13950894e-03 1.75306827e-01
-3.79672766e-01 2.93018669e-01 1.06858611e+00 1.06674455e-01
-4.49007392e-01 -1.92493066e-01 -6.76179469e-01 1.12074530e+00
8.92941713e-01 -5.93081295e-01 -1.39940813e-01 -5.84415615e-01
3.26214612e-01 2.09877715e-01 -8.44189245e-03 -4.82363790e-01
-4.69203629e-02 6.97536720e-03 3.74438375e-01 -8.99632633e-01
1.98469877e-01 -1.23570599e-01 -5.65079212e-01 4.90801841e-01
-7.82351792e-01 2.87517995e-01 1.18313275e-01 2.87461907e-01
-4.75212663e-01 -6.37001932e-01 6.27611518e-01 -3.50705832e-01
-5.07858813e-01 -3.82982701e-01 -4.40304905e-01 6.03087306e-01
6.55430734e-01 -3.91659379e-01 -5.62104642e-01 1.19530030e-01
-6.53622985e-01 3.60379159e-01 3.74636389e-02 1.15301691e-01
2.59646863e-01 -8.53963614e-01 -9.54880536e-01 -8.45425129e-02
-2.61282623e-01 5.87644935e-01 -8.03878382e-02 6.46637380e-01
-4.10558790e-01 2.82073796e-01 9.45485160e-02 -3.08927536e-01
-1.46319473e+00 1.08546078e-01 -8.99233595e-02 -9.93271589e-01
-1.01508880e+00 1.01480806e+00 -1.14036398e-02 -4.08276707e-01
5.61766066e-02 -3.68653566e-01 -4.52314973e-01 2.15025127e-01
3.34335297e-01 1.04893997e-01 -7.81572834e-02 -3.58896792e-01
-1.65331885e-01 2.00936291e-02 -3.39092284e-01 -2.26429641e-01
1.39735556e+00 1.25574619e-01 -8.56908970e-03 2.64886767e-01
8.30444694e-01 8.12478289e-02 -1.22680891e+00 -1.59687340e-01
7.54492342e-01 1.61229968e-01 -3.18734407e-01 -8.24604392e-01
-2.23664686e-01 8.81796539e-01 -2.37384900e-01 3.98928672e-01
5.83099902e-01 5.29337302e-02 1.17195320e+00 4.69775051e-01
-2.46827915e-01 -1.48331940e+00 2.44280156e-02 1.13180411e+00
3.06826591e-01 -1.00842369e+00 -1.08736478e-01 -5.08526206e-01
-6.27863884e-01 1.16494620e+00 6.50922954e-01 -2.47174427e-01
7.11029302e-03 5.49584270e-01 -7.79433846e-02 -1.01262569e-01
-8.66352439e-01 -2.72784412e-01 3.81742902e-02 3.89859110e-01
1.05402696e+00 -1.55440876e-02 -7.99491286e-01 6.37152791e-01
-4.27134573e-01 -5.29319465e-01 1.91158101e-01 1.31397462e+00
-6.05976403e-01 -1.52769446e+00 -2.53980756e-01 3.05775940e-01
-1.05395591e+00 -4.31045175e-01 -5.53685367e-01 1.00312793e+00
-1.17373616e-01 9.31748033e-01 2.63410985e-01 9.97503996e-02
1.84212968e-01 4.08587724e-01 4.24089223e-01 -7.62863517e-01
-1.06880212e+00 4.04589623e-02 1.12709844e+00 -1.50447130e-01
-4.03416574e-01 -8.07776034e-01 -2.06735516e+00 -8.52138549e-02
-2.06161931e-01 3.94820571e-01 5.17915726e-01 1.35220337e+00
1.61194310e-01 7.66956449e-01 1.17876403e-01 -6.49510205e-01
-6.29797995e-01 -1.04071426e+00 8.43265876e-02 5.10667980e-01
2.89447814e-01 -3.68712127e-01 -1.60971522e-01 5.90390936e-02] | [10.57597541809082, 9.464433670043945] |
c9d1fad2-46d7-4b41-b464-34c6c45aa5ac | modeling-global-and-local-node-contexts-for | 2001.11003 | null | https://arxiv.org/abs/2001.11003v2 | https://arxiv.org/pdf/2001.11003v2.pdf | Modeling Global and Local Node Contexts for Text Generation from Knowledge Graphs | Recent graph-to-text models generate text from graph-based data using either global or local aggregation to learn node representations. Global node encoding allows explicit communication between two distant nodes, thereby neglecting graph topology as all nodes are directly connected. In contrast, local node encoding considers the relations between neighbor nodes capturing the graph structure, but it can fail to capture long-range relations. In this work, we gather both encoding strategies, proposing novel neural models which encode an input graph combining both global and local node contexts, in order to learn better contextualized node embeddings. In our experiments, we demonstrate that our approaches lead to significant improvements on two graph-to-text datasets achieving BLEU scores of 18.01 on AGENDA dataset, and 63.69 on the WebNLG dataset for seen categories, outperforming state-of-the-art models by 3.7 and 3.1 points, respectively. | ['Leonardo F. R. Ribeiro', 'Yue Zhang', 'Iryna Gurevych', 'Claire Gardent'] | 2020-01-29 | null | null | null | null | ['graph-to-sequence', 'kg-to-text'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.34205604e-01 7.28065431e-01 -3.61969531e-01 -3.20461720e-01
-4.84932393e-01 -6.83606982e-01 1.11220860e+00 9.43905354e-01
-1.99275509e-01 5.54086626e-01 5.14178872e-01 -4.69350249e-01
-1.96941182e-01 -1.36615431e+00 -6.17077529e-01 -3.87601227e-01
-3.54808897e-01 7.15920091e-01 2.07389891e-01 -3.42833340e-01
2.66970918e-02 2.49248296e-01 -1.02601564e+00 1.86960891e-01
8.19819152e-01 5.97172737e-01 -4.96975482e-02 7.99609065e-01
-4.60185707e-01 7.42819846e-01 -5.15927911e-01 -6.43562257e-01
-8.28139335e-02 -3.84827942e-01 -7.08536923e-01 -1.14302196e-01
7.04897463e-01 -7.49241114e-02 -8.99864793e-01 8.01186383e-01
3.43674928e-01 8.52602348e-02 8.71832848e-01 -1.12443006e+00
-1.20366311e+00 1.29644954e+00 -4.99130577e-01 -4.31941301e-02
5.41875958e-01 -2.09119678e-01 1.88085032e+00 -5.82169354e-01
6.79833412e-01 1.20574105e+00 5.48368454e-01 4.61073458e-01
-1.19085395e+00 -2.37291977e-01 4.70661670e-01 -3.86975966e-02
-1.34975946e+00 -1.18865050e-01 8.87965977e-01 -1.44691810e-01
1.24934626e+00 1.73166499e-01 7.47017622e-01 1.25107896e+00
3.21031094e-01 5.60704291e-01 5.33739090e-01 -4.58484590e-01
5.01002967e-02 -3.49632174e-01 2.90541798e-01 9.32112575e-01
7.81103849e-01 -2.27090135e-01 -4.08363312e-01 -1.79071680e-01
6.28266335e-01 -1.18627558e-02 -1.41918480e-01 -2.77260065e-01
-1.11431658e+00 1.01231897e+00 1.11381900e+00 4.03303206e-01
-3.51470947e-01 5.76183736e-01 3.45450282e-01 2.41978586e-01
6.92845762e-01 5.05676329e-01 -1.51849300e-01 1.68514609e-01
-6.58843696e-01 1.89353541e-01 8.69826436e-01 1.19088709e+00
7.79283106e-01 -1.92853697e-02 -3.42238307e-01 6.77573144e-01
4.93490458e-01 2.83761561e-01 1.99619383e-01 -3.54084104e-01
7.73201704e-01 1.00379121e+00 -4.68642682e-01 -1.39088821e+00
-5.66175997e-01 -6.90177321e-01 -1.08526266e+00 -4.70705301e-01
1.68647572e-01 3.36610340e-02 -1.19033778e+00 1.53012407e+00
-1.96086448e-02 2.67924607e-01 7.12345093e-02 3.82988930e-01
1.27402747e+00 7.16378450e-01 1.41991392e-01 1.87248319e-01
1.25283027e+00 -1.11082053e+00 -7.48600960e-01 -4.72346663e-01
1.03873098e+00 -2.61199743e-01 1.05829060e+00 -3.48654203e-02
-9.33845699e-01 -4.00276005e-01 -1.02008140e+00 -2.22964510e-01
-7.58666635e-01 -2.25875914e-01 7.52714872e-01 4.46979433e-01
-1.49353862e+00 6.31688714e-01 -6.81452155e-01 -6.11199260e-01
2.61642486e-01 1.10927120e-01 -3.97225469e-01 -2.89330482e-01
-1.39983976e+00 5.44211984e-01 5.54004788e-01 -2.49982953e-01
-6.40304208e-01 -5.41650116e-01 -1.20868659e+00 3.71718436e-01
3.98926735e-01 -8.97420108e-01 7.60886788e-01 -1.92376316e-01
-9.35408294e-01 7.35608220e-01 -2.04095349e-01 -6.27910316e-01
1.46067306e-01 6.68283403e-02 -4.83056575e-01 2.21181616e-01
-1.18842006e-01 8.65284085e-01 5.33198416e-01 -1.20476091e+00
-2.20496625e-01 -3.37400079e-01 3.98226768e-01 2.25850135e-01
-6.66440785e-01 -4.65829849e-01 -6.56337440e-01 -7.45820165e-01
1.67459741e-01 -8.27329397e-01 -2.77689457e-01 -1.15794435e-01
-6.77051842e-01 -7.05409229e-01 6.95787966e-01 -4.11586076e-01
1.56741214e+00 -1.73452020e+00 2.47149527e-01 2.04961270e-01
8.06971788e-01 7.20752627e-02 -5.64170897e-01 9.95074809e-01
1.25475660e-01 6.48159981e-01 -6.89076856e-02 -4.28981751e-01
2.07540706e-01 4.98592526e-01 -9.43272039e-02 1.09676525e-01
2.21443757e-01 1.43848038e+00 -1.25159526e+00 -4.27265525e-01
4.80835177e-02 5.61822593e-01 -6.11876667e-01 6.30851910e-02
-3.35098922e-01 -2.15174004e-01 -4.97347504e-01 3.44057500e-01
4.45248723e-01 -5.84359229e-01 6.71241939e-01 -9.67809930e-03
6.04993701e-01 7.08694935e-01 -6.50071561e-01 1.75063503e+00
-6.03975832e-01 6.94587469e-01 -4.35276404e-02 -7.50710130e-01
1.08450723e+00 9.73140001e-02 2.78660327e-01 -5.39195597e-01
-6.93395659e-02 -7.29473308e-02 -4.16772859e-03 -1.10831298e-01
7.55729973e-01 4.24423188e-01 -2.98402667e-01 3.95018786e-01
1.63617224e-01 -1.18156493e-01 4.23641473e-01 1.05560327e+00
1.50362313e+00 -1.49169296e-01 2.47968063e-01 -2.13143647e-01
2.79035270e-01 -4.78964627e-01 9.24649239e-02 8.14062297e-01
1.61748081e-01 7.10239410e-01 8.95690441e-01 -3.59051466e-01
-8.09385359e-01 -1.12337267e+00 3.06004763e-01 1.05547059e+00
-5.49252192e-03 -1.29607368e+00 -5.47881603e-01 -1.13725555e+00
1.94099963e-01 8.53841245e-01 -9.05618131e-01 -2.86770731e-01
-6.67847872e-01 -5.01748085e-01 4.99998003e-01 6.49912655e-01
5.65335751e-02 -7.69035459e-01 1.53217092e-01 2.13099033e-01
-1.87130067e-02 -1.21026480e+00 -5.81954658e-01 1.37744180e-03
-9.11195040e-01 -9.71962988e-01 -4.51720804e-01 -7.36716092e-01
8.09553325e-01 3.10799241e-01 1.54404533e+00 5.81227720e-01
-1.76450666e-02 2.01569736e-01 -7.60349572e-01 8.55796933e-02
-3.53086114e-01 7.10027337e-01 -3.52944851e-01 -1.91020951e-01
2.20645159e-01 -6.59824848e-01 -4.82613325e-01 -5.09967804e-02
-9.03985023e-01 1.20687023e-01 5.77623188e-01 9.19886410e-01
3.09328079e-01 -1.42687917e-01 5.58555782e-01 -1.25039005e+00
9.28935170e-01 -7.01638162e-01 -8.78777504e-02 2.92654872e-01
-8.72121990e-01 2.11764231e-01 6.85418963e-01 -1.22978859e-01
-5.04809141e-01 -2.54838377e-01 -1.75075024e-01 -2.73736473e-02
2.34052818e-02 9.15753543e-01 -1.45693839e-01 1.84324667e-01
6.79062366e-01 8.10360909e-02 -3.88184011e-01 -3.22229356e-01
8.24078977e-01 5.23610950e-01 2.16963097e-01 -6.14304066e-01
8.51498783e-01 1.71842650e-01 1.86717778e-01 -7.35421419e-01
-8.03319216e-01 -3.06046993e-01 -7.67081022e-01 7.89555833e-02
7.13170648e-01 -7.35881209e-01 -1.77451923e-01 7.19040111e-02
-1.12502348e+00 -2.40246668e-01 -2.16504127e-01 1.36753246e-01
-9.32554677e-02 4.87535208e-01 -8.04491401e-01 -4.32772160e-01
-4.23406959e-01 -7.05078483e-01 1.16736495e+00 -7.20392689e-02
-2.95213878e-01 -1.61529279e+00 -5.25050983e-02 9.21820626e-02
4.44463849e-01 4.20894474e-01 1.20962083e+00 -9.73086834e-01
-5.67772090e-01 -3.37433577e-01 -4.49398637e-01 -1.29610717e-01
2.42813170e-01 1.29010249e-02 -6.00941539e-01 -5.57240427e-01
-9.99206543e-01 -2.21942544e-01 1.10828161e+00 -2.35889778e-02
1.11988842e+00 -5.78418970e-01 -7.06986070e-01 5.78122318e-01
1.25562286e+00 -3.00270081e-01 6.04764044e-01 -1.11831203e-01
1.11805618e+00 5.19748271e-01 6.72404990e-02 3.05489868e-01
7.29520559e-01 5.36289871e-01 6.88364029e-01 -8.63246620e-02
-6.53471529e-01 -9.15660441e-01 1.82606250e-01 1.01997805e+00
1.21899232e-01 -1.00351048e+00 -9.64621484e-01 7.48924851e-01
-1.74725437e+00 -5.90012312e-01 -6.56263411e-01 1.90342271e+00
5.73889494e-01 3.11312914e-01 -1.19575135e-01 1.12049161e-02
6.13867700e-01 7.74191916e-01 -2.24727854e-01 -3.45865935e-01
1.82780772e-02 2.62964755e-01 5.22386909e-01 6.66413367e-01
-8.50562274e-01 1.15143073e+00 6.48524332e+00 6.62215948e-01
-6.92466438e-01 -6.77437857e-02 4.92171913e-01 9.74308923e-02
-9.73819435e-01 9.18120891e-02 -6.18424058e-01 2.30449483e-01
1.20081258e+00 -3.53089869e-01 3.43394965e-01 5.23163915e-01
-3.06185305e-01 5.33181965e-01 -1.11352968e+00 6.39266789e-01
2.69492805e-01 -1.41982687e+00 5.97644269e-01 2.14711428e-01
8.18812966e-01 1.14888757e-01 -1.54176250e-01 4.28711444e-01
6.99530184e-01 -1.34160256e+00 2.53645569e-01 3.33378494e-01
8.58008862e-01 -6.26658916e-01 7.09412813e-01 2.50624806e-01
-1.55708933e+00 2.38430277e-01 -3.48416448e-01 -1.13613345e-01
1.17395602e-01 7.51975060e-01 -9.93707895e-01 1.08267391e+00
3.47861111e-01 1.00545788e+00 -9.74393249e-01 5.07849514e-01
-5.83257735e-01 7.98886836e-01 -2.00267524e-01 -4.68792677e-01
6.09419346e-01 -9.53538939e-02 4.98585254e-01 1.18938279e+00
2.49558598e-01 -1.88912541e-01 3.38199884e-01 8.29781532e-01
-5.66488981e-01 3.99606414e-02 -9.14936364e-01 -3.51285338e-01
6.60917699e-01 1.43483627e+00 -6.85880005e-01 -3.62605274e-01
-4.26959097e-01 1.00269353e+00 8.35308135e-01 4.79033977e-01
-6.45069063e-01 -8.37132633e-01 4.33009624e-01 3.18505973e-01
1.82023048e-01 -6.07903123e-01 -9.69806761e-02 -1.08209729e+00
-1.03390470e-01 -4.00920779e-01 6.59456074e-01 -6.22063160e-01
-1.46743453e+00 7.55191922e-01 6.09076358e-02 -6.78433120e-01
-3.10188740e-01 -5.77367485e-01 -7.41567791e-01 6.93152308e-01
-1.44370413e+00 -1.46805489e+00 -3.97153288e-01 2.31853068e-01
2.19520241e-01 9.57606174e-03 8.90767813e-01 -3.09716575e-02
-1.92098290e-01 8.68213654e-01 -6.45230636e-02 3.95992517e-01
4.52106327e-01 -1.57084692e+00 1.12069023e+00 6.72958195e-01
7.48814404e-01 7.95976400e-01 3.51125300e-01 -7.61276424e-01
-1.55628777e+00 -1.31966758e+00 1.42006111e+00 -4.51699227e-01
8.15304101e-01 -7.79268205e-01 -1.00145745e+00 9.27096069e-01
5.14233530e-01 2.67632157e-01 5.39227843e-01 4.94142473e-01
-7.60275245e-01 7.97930509e-02 -9.23102856e-01 8.00244868e-01
1.58638597e+00 -4.57377046e-01 -4.67310131e-01 4.34794068e-01
1.14286900e+00 -2.53798366e-01 -1.17239976e+00 2.48378769e-01
1.34769559e-01 -6.15747690e-01 8.41418147e-01 -6.70239627e-01
4.87963736e-01 1.06716715e-02 -6.64275661e-02 -1.75775719e+00
-6.91622734e-01 -6.85817897e-01 -4.87216592e-01 1.29441071e+00
7.83712089e-01 -7.94372499e-01 9.42367494e-01 6.19100817e-02
-2.08529234e-01 -9.13961470e-01 -7.12563992e-01 -6.25755668e-01
3.92770559e-01 -1.53586715e-01 7.78500080e-01 1.02933609e+00
9.14418921e-02 8.17667723e-01 -1.18000314e-01 -3.72253843e-02
5.05949974e-01 -3.58260795e-02 6.24784291e-01 -1.42458117e+00
-1.34691224e-01 -6.02441549e-01 -4.69739556e-01 -1.41244519e+00
4.19995010e-01 -1.45673895e+00 -4.16351497e-01 -2.46502924e+00
1.22453324e-01 -5.41413367e-01 -2.62579650e-01 6.03302419e-01
-4.60513562e-01 1.85453981e-01 1.39822453e-01 -2.61802971e-01
-6.79636359e-01 7.47188807e-01 1.21091485e+00 -4.72046822e-01
1.13100253e-01 -4.28421617e-01 -9.86077487e-01 3.00336927e-01
8.65023732e-01 -4.17399377e-01 -7.69652665e-01 -7.93547153e-01
5.45313179e-01 -2.04047546e-01 2.28604764e-01 -7.86112964e-01
2.43261620e-01 -2.77850516e-02 2.15901613e-01 -4.76094723e-01
2.79526085e-01 -6.50588512e-01 -1.52289078e-01 3.04771036e-01
-7.62992680e-01 9.43983719e-02 -7.66162127e-02 1.01741517e+00
-1.04135297e-01 -1.00015149e-01 2.27583155e-01 -5.52172028e-02
-3.89464498e-01 4.98172373e-01 -9.60701108e-02 2.26185784e-01
8.18992496e-01 1.12809956e-01 -6.52311802e-01 -8.10760498e-01
-5.26722789e-01 4.62811083e-01 4.19754654e-01 7.41711378e-01
7.34334230e-01 -1.46368873e+00 -1.03595853e+00 1.21098086e-01
4.00442600e-01 6.62807822e-02 -3.06501519e-02 3.44230711e-01
-3.90451163e-01 5.19689083e-01 3.34493995e-01 -4.72027063e-01
-1.10236609e+00 4.78745013e-01 2.92878486e-02 -6.42166734e-01
-7.37960875e-01 8.25536668e-01 -9.11457986e-02 -7.44953454e-01
1.19186483e-01 -5.04204392e-01 -2.29940534e-01 -1.14539869e-01
2.04167545e-01 2.76228905e-01 1.20653212e-01 -5.67693949e-01
-2.12342232e-01 4.84434456e-01 -2.36061633e-01 2.16429576e-01
1.09080791e+00 -3.10030393e-02 -1.61849752e-01 2.82045841e-01
1.34469008e+00 1.88517600e-01 -7.39129245e-01 -5.22878647e-01
9.52885747e-02 -4.38844979e-01 9.48214531e-02 -7.13520169e-01
-1.02970612e+00 9.98262167e-01 -2.34903038e-01 6.95876181e-01
8.33817363e-01 3.28799129e-01 9.02139366e-01 4.77583170e-01
3.20724249e-01 -6.28833115e-01 2.05848023e-01 7.59108424e-01
8.51084411e-01 -9.17393088e-01 1.45178229e-01 -7.36366987e-01
-4.52966899e-01 1.08356225e+00 5.98862469e-01 -1.75046623e-01
5.73961616e-01 1.11146532e-01 -3.96170527e-01 -3.34364742e-01
-1.15344846e+00 -2.23831505e-01 5.01285553e-01 6.71565592e-01
6.36325777e-01 2.95598984e-01 -2.75750101e-01 2.45421275e-01
-2.89158523e-01 -6.00599587e-01 5.06991088e-01 6.99115932e-01
-3.86442184e-01 -1.36567533e+00 2.55561441e-01 7.35433757e-01
-2.57737726e-01 -3.91140372e-01 -7.93535411e-01 9.10625994e-01
-3.50791186e-01 1.03491247e+00 3.93260777e-01 -5.67151010e-01
1.97743237e-01 -1.00219082e-02 3.56553376e-01 -9.16446805e-01
-5.17638147e-01 -2.53177673e-01 5.23279369e-01 -2.54248530e-01
6.51068753e-03 -2.92666465e-01 -1.44614160e+00 -4.91465837e-01
-2.72671610e-01 1.73935041e-01 4.24914241e-01 3.47916961e-01
7.17271984e-01 8.57926071e-01 4.10765111e-01 -4.89735365e-01
-3.61064047e-01 -1.24013126e+00 -5.70835948e-01 5.46185136e-01
1.54787213e-01 -3.03388029e-01 -3.84325027e-01 -3.66887480e-01] | [10.108146667480469, 8.15916633605957] |
f1665956-cf32-463c-8f48-f2d4b382488d | mmwave-mapping-and-slam-for-5g-and-beyond | 2211.16024 | null | https://arxiv.org/abs/2211.16024v1 | https://arxiv.org/pdf/2211.16024v1.pdf | MmWave Mapping and SLAM for 5G and Beyond | Device localization and radar-like mapping are at the heart of integrated sensing and communication, enabling not only new services and applications, but can also improve communication quality with reduced overheads. These forms of sensing are however susceptible to data association problems, due to the unknown relation between measurements and detected objects or targets. In this chapter, we provide an overview of the fundamental tools used to solve mapping, tracking, and simultaneous localization and mapping (SLAM) problems. We distinguish the different types of sensing problems and then focus on mapping and SLAM as running examples. Starting from the applicable models and definitions, we describe the different algorithmic approaches, with a particular focus on how to deal with data association problems. In particular, methods based on random finite set theory and Bayesian graphical models are introduced in detail. A numerical study with synthetic and experimental data is then used to compare these approaches in a variety of scenarios. | ['Henk Wymeersch', 'Mikko Valkama', 'Lennart Svensson', 'Sunwoo Kim', 'Jukka Talvitie', 'Hyowon Kim', 'Ossi Kaltiokallio', 'Yu Ge'] | 2022-11-29 | null | null | null | null | ['simultaneous-localization-and-mapping'] | ['computer-vision'] | [ 6.05726898e-01 -1.99969485e-01 -5.26033789e-02 -5.29642045e-01
-7.20996618e-01 -5.71137011e-01 6.62998736e-01 1.65928841e-01
-2.68947363e-01 1.13358819e+00 -3.46099228e-01 -1.76915705e-01
-7.64448702e-01 -9.33499813e-01 -4.47939128e-01 -7.44767606e-01
-5.84902108e-01 7.77121246e-01 1.48989081e-01 2.95566060e-02
1.74194366e-01 9.12851095e-01 -1.45389366e+00 -7.19620585e-01
8.06283414e-01 1.11057532e+00 4.34200764e-01 6.69536650e-01
-5.95152341e-02 3.12470526e-01 -6.92944705e-01 8.32336023e-02
2.04610959e-01 -2.22853258e-01 -2.25243986e-01 -1.49546415e-01
1.38926134e-01 -6.31606132e-02 -1.14929393e-01 1.14473975e+00
6.50349021e-01 1.12560326e-02 5.12534738e-01 -1.29182804e+00
-5.74746504e-02 5.36837518e-01 -3.90698761e-01 -6.11609668e-02
6.32708013e-01 -6.32064462e-01 4.05866295e-01 -7.09567785e-01
2.99042631e-02 1.03952575e+00 1.16580009e+00 1.52042329e-01
-1.41493225e+00 -8.00060213e-01 -1.56784564e-01 4.29297090e-02
-1.86037648e+00 -6.84382856e-01 5.53727031e-01 -2.35484987e-01
2.90455848e-01 4.84693408e-01 5.88858306e-01 6.08306527e-01
3.32598269e-01 1.93911016e-01 1.15269637e+00 -6.71396434e-01
3.65626484e-01 3.60763967e-02 4.01095837e-01 4.43627328e-01
9.36025500e-01 4.48872834e-01 -5.26195645e-01 -4.99581099e-01
4.63372499e-01 1.06614158e-01 -2.38779455e-01 -8.50885153e-01
-1.10004437e+00 7.64203072e-01 3.39592487e-01 2.87968099e-01
-4.95833218e-01 3.91446561e-01 -3.28497708e-01 1.03995614e-02
9.18198973e-02 5.60972653e-02 -9.70313847e-02 2.79751718e-01
-9.16814685e-01 -1.52957337e-02 7.70029366e-01 1.02058077e+00
1.04133940e+00 8.32390860e-02 3.98285687e-01 5.23854375e-01
7.59537399e-01 1.53165627e+00 -4.56149668e-01 -6.41365826e-01
1.09810852e-01 -2.72294041e-02 5.50490141e-01 -1.23443794e+00
-6.25806868e-01 -4.55758363e-01 -9.61082637e-01 7.41411224e-02
2.49178968e-02 -2.99889743e-01 -6.86437726e-01 1.51501155e+00
1.86444014e-01 2.31460989e-01 2.78698772e-01 4.20374036e-01
5.52198648e-01 4.95229602e-01 -3.76666427e-01 -5.56273580e-01
1.14579439e+00 -8.92693326e-02 -1.10638523e+00 -6.38578057e-01
2.53190160e-01 -7.86604583e-01 2.97909789e-02 1.38930678e-01
-5.43760002e-01 -3.77810359e-01 -1.24897766e+00 6.47593141e-01
-2.97411144e-01 -3.47610824e-02 6.93899095e-01 9.74213481e-01
-9.59606946e-01 -6.38207793e-03 -1.06979764e+00 -7.10202098e-01
2.53185667e-02 4.99901325e-01 2.81030964e-02 -2.78240710e-01
-1.06632781e+00 1.05378854e+00 3.69335003e-02 4.33818221e-01
-5.72723925e-01 -3.50093812e-01 -9.50240731e-01 -4.67871726e-01
1.63750827e-01 -4.16476935e-01 9.90521014e-01 -3.92153487e-02
-1.23495495e+00 3.19061011e-01 -5.55615127e-01 -6.99099839e-01
1.59623519e-01 -2.51887888e-01 -7.50909328e-01 -3.83307427e-01
2.89324939e-01 1.36661455e-01 3.95154357e-01 -1.44595909e+00
-7.44499743e-01 -4.86485451e-01 -3.29006344e-01 2.51160245e-02
4.59067494e-01 -2.23391339e-01 -4.87057120e-02 -1.12393111e-01
9.59784925e-01 -8.46120715e-01 -5.27043641e-01 -1.20437719e-01
-2.90603220e-01 4.90594059e-01 7.21495390e-01 -7.09382519e-02
8.21054816e-01 -1.98399866e+00 -2.23155484e-01 7.59225607e-01
-1.25322804e-01 -2.18991265e-01 2.48807758e-01 8.44485164e-01
6.22509658e-01 -3.31071526e-01 -2.86170781e-01 -4.96372491e-01
-8.10913891e-02 5.91258645e-01 -4.81436580e-01 1.01103640e+00
-6.07624054e-01 6.12171412e-01 -9.69653845e-01 -2.58738816e-01
6.19847655e-01 5.83883703e-01 3.14522386e-01 -2.23604888e-01
2.40583420e-01 8.28853548e-01 -6.41698480e-01 8.21746469e-01
1.01394606e+00 5.55829257e-02 2.87325680e-01 -1.23319522e-01
-4.93703544e-01 2.20695406e-01 -1.47911549e+00 1.50050282e+00
-6.35070145e-01 6.21700525e-01 5.91171622e-01 -9.24840093e-01
1.28257942e+00 -2.80079097e-02 5.98414779e-01 -6.36636436e-01
-2.84921601e-02 4.34752613e-01 -1.99440390e-01 -6.76775649e-02
6.49059713e-01 -1.67827442e-01 -6.23346865e-01 4.96906549e-01
-4.49598610e-01 -2.50748515e-01 -2.99342990e-01 1.22275382e-01
1.13713396e+00 -2.74640769e-01 8.26407731e-01 -5.26130676e-01
5.02937436e-01 2.59321034e-01 5.85799634e-01 1.18228543e+00
6.96581528e-02 -1.86380133e-01 -5.82253337e-01 -3.43611568e-01
-3.54175538e-01 -1.23087418e+00 -5.45933306e-01 4.74326998e-01
1.03745592e+00 -2.06553966e-01 2.02373162e-01 9.06566605e-02
3.96572530e-01 5.36524117e-01 -4.42699015e-01 2.21025914e-01
-1.21746168e-01 -9.51562226e-01 5.64085424e-01 1.57497689e-01
3.96337718e-01 -4.53698784e-01 -9.81613696e-01 2.47186065e-01
-2.71891922e-01 -1.13475192e+00 5.18010736e-01 4.45171952e-01
-7.67790198e-01 -9.63239968e-01 -7.43166134e-02 -2.14989617e-01
6.18146896e-01 8.81310880e-01 7.93508053e-01 -1.25270098e-01
-2.15048134e-01 7.09920287e-01 -4.82250094e-01 -5.84760904e-01
-5.98119907e-02 -3.61250311e-01 5.60274184e-01 1.04306564e-01
1.87049463e-01 -6.82884753e-01 -4.46095876e-02 6.33477986e-01
-2.98021525e-01 -3.27849209e-01 7.24600732e-01 5.21156132e-01
8.21447432e-01 1.79797530e-01 1.13543063e-01 -5.05351901e-01
1.51167855e-01 -3.40888619e-01 -1.30013704e+00 2.37212688e-01
-5.93468547e-01 -4.12297063e-02 -1.46722227e-01 -1.16428845e-01
-7.80048192e-01 5.28267324e-01 1.93546027e-01 1.48002580e-01
-6.44518659e-02 5.58557034e-01 -4.50185269e-01 -1.03859472e+00
6.34057343e-01 3.63038808e-01 1.46025550e-02 -3.67538035e-01
2.42268920e-01 6.90345466e-01 5.59727728e-01 -3.06986332e-01
1.11222172e+00 1.16564000e+00 5.43002665e-01 -1.15775108e+00
-8.78312528e-01 -6.03228152e-01 -6.82776332e-01 -3.81388336e-01
3.36558878e-01 -7.62400150e-01 -5.35556972e-01 1.86807930e-01
-1.02103794e+00 -2.49108925e-01 -1.95026383e-01 8.37197959e-01
-5.30920863e-01 2.77007729e-01 1.63434997e-01 -1.53358591e+00
2.28714123e-01 -8.54940712e-01 1.05692363e+00 1.07446648e-01
-7.40719261e-03 -1.07642341e+00 4.53920841e-01 -3.16199772e-02
5.71022570e-01 4.39050794e-01 3.33000571e-02 -3.77463520e-01
-9.12188828e-01 -5.86896718e-01 -9.02675018e-02 -4.68338192e-01
3.47406656e-01 -5.94431162e-01 -7.39858687e-01 -4.32054341e-01
7.62681291e-02 2.53591955e-01 4.84297425e-01 5.34958661e-01
2.65253335e-01 4.06037420e-02 -1.22932351e+00 8.28422010e-01
1.56524777e+00 2.01164663e-01 5.86316168e-01 -1.64162845e-03
3.28136444e-01 5.68311632e-01 1.27510619e+00 3.96327585e-01
2.71586508e-01 9.39806819e-01 8.08106422e-01 8.37089196e-02
2.16250271e-01 2.45104041e-02 6.50639459e-02 3.12055111e-01
8.98366421e-02 -2.76033610e-01 -9.07488704e-01 3.05294067e-01
-1.93764448e+00 -9.13100004e-01 -6.99023008e-01 2.56370401e+00
-1.44508645e-01 -3.00546110e-01 -3.92527729e-01 1.85715824e-01
1.01137877e+00 9.30658132e-02 -2.33458623e-01 4.09657598e-01
-2.29525760e-01 -4.67573591e-02 1.34026551e+00 1.17588854e+00
-9.89221632e-01 5.88870049e-01 6.95274639e+00 4.88040924e-01
-8.25062752e-01 2.68048525e-01 -6.21639490e-01 3.53374422e-01
-6.43728673e-02 2.83770770e-01 -1.14878249e+00 3.47144246e-01
8.24212909e-01 1.95595279e-01 1.98147625e-01 2.47333467e-01
1.16766848e-01 -8.49465251e-01 -6.50198519e-01 1.16958225e+00
3.69130410e-02 -1.26595151e+00 -4.27065432e-01 4.50451583e-01
4.95480508e-01 4.30016547e-01 -4.38351482e-01 -1.94338232e-01
5.42873442e-01 -6.12298667e-01 8.03571939e-01 6.84049666e-01
6.82140589e-01 -5.19169092e-01 9.51271296e-01 5.91759920e-01
-1.49656475e+00 1.05345510e-02 -3.93644810e-01 -3.76422107e-01
7.84376502e-01 1.06953549e+00 -5.30369997e-01 1.05921364e+00
6.39685988e-01 6.07299745e-01 2.75245570e-02 1.39143300e+00
-3.32426906e-01 2.71126479e-02 -9.96683061e-01 -3.26654434e-01
-1.32027090e-01 -2.21435800e-01 7.58287966e-01 8.28767061e-01
7.54117489e-01 2.57242680e-01 3.38553250e-01 6.43741190e-01
6.88371718e-01 -4.32838976e-01 -9.59742248e-01 5.05070269e-01
1.23639917e+00 1.11269891e+00 -6.86817706e-01 2.31935401e-02
-2.33893201e-01 2.96025962e-01 -4.28310275e-01 1.25031710e-01
-6.93718016e-01 -3.37988317e-01 6.35400832e-01 7.64190033e-02
-3.33462469e-02 -7.94766724e-01 -2.38200933e-01 -6.24242485e-01
-3.01532537e-01 -1.16347857e-01 1.22199737e-01 -3.27895343e-01
-8.78823340e-01 3.60841751e-01 3.24121982e-01 -1.46411407e+00
-1.59943804e-01 -1.42853647e-01 -1.47603497e-01 9.05320764e-01
-1.44746804e+00 -9.68592823e-01 -7.64721334e-01 4.91751909e-01
-1.52815551e-01 1.26586318e-01 9.03985083e-01 2.77228892e-01
7.93530196e-02 -1.69328541e-01 4.68516827e-01 -3.91998067e-02
3.15848887e-01 -6.87882900e-01 1.68859750e-01 1.12821150e+00
3.91295910e-01 4.89607930e-01 1.08210588e+00 -9.46628332e-01
-1.47683847e+00 -1.07723320e+00 8.94026697e-01 -4.73878473e-01
6.27155304e-01 -6.53436840e-01 -1.57289773e-01 7.52822876e-01
-2.36456811e-01 4.58488204e-02 6.80075467e-01 1.67618647e-01
1.51560500e-01 -3.37103426e-01 -1.21430445e+00 -4.05464396e-02
1.05602062e+00 -3.12542647e-01 -2.79444635e-01 5.30242860e-01
6.52060751e-03 -3.27478230e-01 -5.47784030e-01 8.05918396e-01
7.32891500e-01 -9.96629000e-01 1.07415938e+00 5.30807972e-01
-1.21760488e+00 -8.47175598e-01 -8.56839299e-01 -9.43781734e-01
-1.84240356e-01 -7.26397872e-01 9.35479477e-02 1.07266879e+00
1.93324327e-01 -1.19249511e+00 6.84128344e-01 -2.24047557e-01
5.38751334e-02 1.20687969e-01 -1.17534792e+00 -1.23669350e+00
-7.85015166e-01 -8.33552957e-01 6.45367324e-01 6.42274320e-01
-3.79261941e-01 4.03852165e-01 -5.71060658e-01 1.08869362e+00
1.46951330e+00 3.28453273e-01 1.06165326e+00 -1.85231030e+00
-1.45078767e-02 1.29205436e-01 -7.35512972e-01 -1.29818332e+00
-6.57716990e-02 -3.66813242e-01 6.79551959e-01 -1.71856141e+00
-1.93393365e-01 -1.31923628e+00 2.43281618e-01 1.26469791e-01
6.00329936e-01 5.30112624e-01 -1.91219687e-01 5.68625689e-01
-6.14335835e-01 3.05359334e-01 3.00832629e-01 -1.30243093e-01
-1.22862309e-01 7.51496792e-01 -2.62404621e-01 6.06843114e-01
8.31635356e-01 -6.28315330e-01 -2.84243256e-01 -5.89825094e-01
5.55259347e-01 2.82275379e-01 7.06639111e-01 -1.57023776e+00
6.95401490e-01 -1.26935601e-01 3.81825306e-02 -1.14787972e+00
9.16846037e-01 -1.37982380e+00 8.98870945e-01 9.44705367e-01
2.94904828e-01 -6.66581392e-02 4.32971902e-02 1.01518846e+00
-6.60498813e-02 -2.93397784e-01 7.79170573e-01 3.55829269e-01
-9.91135716e-01 1.21731311e-01 -4.21709180e-01 -6.54752612e-01
1.25108552e+00 -4.65565950e-01 -1.41187653e-01 -8.30861092e-01
-9.36098218e-01 4.02320951e-01 4.18095142e-01 -1.53414682e-01
5.37726521e-01 -1.29671919e+00 -5.25615752e-01 2.95685589e-01
3.58155012e-01 -1.39311299e-01 1.91194594e-01 1.06425226e+00
-4.01725262e-01 7.12058485e-01 4.79070283e-03 -1.02408373e+00
-1.17043698e+00 2.19748855e-01 4.05366689e-01 1.42686814e-01
-7.09672421e-02 6.59407258e-01 -3.11459571e-01 -6.48733497e-01
3.32628936e-01 -7.95289651e-02 9.34613794e-02 -2.97119170e-01
5.27805150e-01 4.16001588e-01 -3.61174867e-02 -8.49249601e-01
-9.77109253e-01 1.04934084e+00 6.33427501e-01 -3.77646983e-01
9.35919523e-01 -9.16837037e-01 -2.99184591e-01 6.17401063e-01
1.85774744e-01 5.02029061e-01 -8.13858688e-01 -5.32877028e-01
1.97455749e-01 -4.10066634e-01 7.04322234e-02 -5.78375399e-01
-5.55993080e-01 6.77192330e-01 8.19754958e-01 6.07132018e-01
8.06382895e-01 3.90597701e-01 8.91621709e-02 4.78749156e-01
1.56797218e+00 -4.39500779e-01 -6.02673650e-01 5.60231864e-01
3.14855129e-01 -9.86354589e-01 2.89199024e-01 -8.35900366e-01
1.96310893e-01 7.09813833e-01 -1.44133911e-01 -1.10537648e-01
8.06192160e-01 7.42022932e-01 1.52401730e-01 -1.90789089e-01
-1.07285783e-01 -5.94214976e-01 -1.05765261e-01 1.26245844e+00
-9.16409716e-02 2.69908935e-01 3.27428728e-02 1.39701396e-01
-1.70594707e-01 -2.75976002e-01 2.78314352e-01 1.07988548e+00
-8.37483704e-01 -1.17796171e+00 -1.05670059e+00 3.26038003e-01
1.07012995e-01 1.36494756e-01 -3.84291679e-01 1.00015450e+00
2.42832348e-01 1.27519417e+00 1.75800905e-01 -4.87306923e-01
1.07631125e-01 -5.35028577e-01 7.17300415e-01 -4.74879533e-01
3.04437459e-01 -8.51127654e-02 1.05309315e-01 -5.89068890e-01
-8.61071944e-01 -1.04464304e+00 -1.06579196e+00 -2.78069913e-01
-6.76780462e-01 6.25995398e-01 1.05294371e+00 9.15375412e-01
3.22570890e-01 3.51675302e-01 3.51494342e-01 -1.03895736e+00
-3.53299409e-01 -8.11590850e-01 -9.12681043e-01 -7.22066224e-01
4.96917903e-01 -1.16883337e+00 -3.60382080e-01 -5.55985510e-01] | [6.188792705535889, 0.8990471363067627] |
4de1f278-b77b-4416-b89e-f1adff2db370 | efficiently-leveraging-multi-level-user | 2206.12781 | null | https://arxiv.org/abs/2206.12781v2 | https://arxiv.org/pdf/2206.12781v2.pdf | Efficiently Leveraging Multi-level User Intent for Session-based Recommendation via Atten-Mixer Network | Session-based recommendation (SBR) aims to predict the user's next action based on short and dynamic sessions. Recently, there has been an increasing interest in utilizing various elaborately designed graph neural networks (GNNs) to capture the pair-wise relationships among items, seemingly suggesting the design of more complicated models is the panacea for improving the empirical performance. However, these models achieve relatively marginal improvements with exponential growth in model complexity. In this paper, we dissect the classical GNN-based SBR models and empirically find that some sophisticated GNN propagations are redundant, given the readout module plays a significant role in GNN-based models. Based on this observation, we intuitively propose to remove the GNN propagation part, while the readout module will take on more responsibility in the model reasoning process. To this end, we propose the Multi-Level Attention Mixture Network (Atten-Mixer), which leverages both concept-view and instance-view readouts to achieve multi-level reasoning over item transitions. As simply enumerating all possible high-level concepts is infeasible for large real-world recommender systems, we further incorporate SBR-related inductive biases, i.e., local invariance and inherent priority to prune the search space. Experiments on three benchmarks demonstrate the effectiveness and efficiency of our proposal. We also have already launched the proposed techniques to a large-scale e-commercial online service since April 2021, with significant improvements of top-tier business metrics demonstrated in the online experiments on live traffic. | ['Sunghun Kim', 'Haohan Wang', 'Xing Xie', 'Yan Zhang', 'Jaeboum Kim', 'Yueqi Xie', 'Chaozhuo Li', 'Jiayan Guo', 'Peiyan Zhang'] | 2022-06-26 | null | null | null | null | ['session-based-recommendations'] | ['miscellaneous'] | [-1.00892549e-02 -1.27101481e-01 -2.27602527e-01 -4.09723580e-01
-1.32581159e-01 -5.07527351e-01 4.32426065e-01 -1.18916016e-02
-1.48617342e-01 3.75380963e-01 1.23528741e-01 -6.93493247e-01
-5.56697130e-01 -8.64677429e-01 -7.89227843e-01 -4.16469723e-01
-3.43517989e-01 2.89405376e-01 3.50378752e-01 -6.72414780e-01
2.69771218e-01 2.54428715e-01 -1.39777267e+00 2.84984112e-01
1.12447989e+00 1.00546002e+00 1.43975243e-01 6.31103694e-01
-6.80208951e-02 1.04790688e+00 -1.60779610e-01 -7.75878072e-01
3.35790455e-01 -4.73745286e-01 -5.71635544e-01 -2.41099387e-01
1.73453912e-01 -3.08105260e-01 -6.80420220e-01 9.86380339e-01
2.51365662e-01 6.35638177e-01 3.87765288e-01 -1.25387263e+00
-9.31171179e-01 1.15533710e+00 -5.44895530e-01 2.88597941e-01
1.34643212e-01 2.43759111e-01 1.56781399e+00 -6.62055790e-01
8.75240788e-02 1.03007627e+00 4.38580394e-01 6.13436043e-01
-1.15321159e+00 -5.12313962e-01 9.71775770e-01 4.66979742e-01
-1.20228255e+00 -1.62260085e-01 9.50109780e-01 3.18491198e-02
9.20083761e-01 5.64518213e-01 6.49666011e-01 1.29194677e+00
-4.21441235e-02 8.77668202e-01 8.63008559e-01 -6.22894727e-02
2.10425466e-01 2.14586109e-01 5.20843983e-01 5.44710517e-01
4.84675914e-01 -1.82784170e-01 -5.51934958e-01 -5.92117570e-02
7.50578165e-01 4.24584121e-01 -2.70361722e-01 -1.78783581e-01
-7.29222476e-01 7.41517603e-01 8.17739308e-01 5.94393611e-02
-3.91868651e-01 1.59697622e-01 4.89024259e-02 3.21043789e-01
2.44748741e-01 4.66024131e-01 -3.57161522e-01 6.00850582e-03
-6.75211906e-01 4.31716405e-02 8.06946695e-01 9.52340841e-01
4.65363055e-01 -1.30063459e-01 -3.39043468e-01 8.96265805e-01
5.42363822e-01 4.60712835e-02 2.51812994e-01 -6.57453835e-01
4.10548806e-01 7.41967440e-01 -3.41375009e-03 -1.11160278e+00
-3.24194551e-01 -1.00709045e+00 -8.95663083e-01 -3.19232583e-01
3.01850855e-01 1.02503952e-02 -8.62700403e-01 1.78440773e+00
8.33030865e-02 3.95935953e-01 -2.52719194e-01 1.22460318e+00
5.90628386e-01 4.61157113e-01 9.18649286e-02 -1.61852278e-02
1.21962321e+00 -1.26035929e+00 -3.57128829e-01 -3.70205909e-01
2.91915327e-01 -2.34747648e-01 1.15487528e+00 5.25071502e-01
-1.12654436e+00 -5.08280933e-01 -1.10988379e+00 5.95048554e-02
-3.21964473e-01 -2.25785688e-01 1.10297596e+00 6.58649862e-01
-1.06633627e+00 6.68113887e-01 -6.60873294e-01 -3.90032023e-01
2.89015621e-01 5.02423763e-01 1.52374372e-01 -2.06180498e-01
-1.56629014e+00 5.23288786e-01 -1.56103186e-02 6.61790907e-01
-7.89441645e-01 -5.22505462e-01 -2.53963113e-01 5.30157924e-01
7.58503020e-01 -7.22683191e-01 1.13403964e+00 -7.93626070e-01
-1.58720362e+00 1.13005541e-01 4.58327904e-02 -5.62195420e-01
2.24547148e-01 -1.05483517e-01 -7.49056458e-01 -2.45974332e-01
-6.53116941e-01 2.57271022e-01 6.37833118e-01 -1.22440064e+00
-7.09843457e-01 -2.28879482e-01 9.61319983e-01 2.95224756e-01
-6.10959709e-01 -2.76664168e-01 -7.27162242e-01 -4.29944426e-01
1.72296524e-01 -9.92949843e-01 -3.75363231e-01 -5.08471668e-01
-2.91580051e-01 -3.27174246e-01 8.05509090e-02 -4.83824581e-01
1.67508650e+00 -1.88002038e+00 -7.76561582e-03 4.53409374e-01
4.02296692e-01 2.59434909e-01 -2.50093877e-01 4.75438267e-01
8.23725164e-02 1.48735225e-01 2.40813270e-01 -8.50102454e-02
1.24476418e-01 1.53089821e-01 -3.71963978e-01 8.92191008e-02
-1.84005909e-02 1.03821063e+00 -8.44241261e-01 3.30001302e-02
-4.64907251e-02 2.91278571e-01 -1.00726771e+00 1.27590999e-01
-4.04299855e-01 2.27038443e-01 -5.41969121e-01 4.95247483e-01
5.67727029e-01 -7.57105827e-01 5.11362970e-01 -2.76956290e-01
2.78558135e-01 8.05938601e-01 -1.10940289e+00 1.47445536e+00
-3.54521990e-01 1.95290312e-01 -2.70413179e-02 -1.04640341e+00
6.67432189e-01 -7.02606291e-02 3.15467209e-01 -9.46582139e-01
-1.69118810e-02 -6.74499571e-02 5.76717913e-01 -1.43879592e-01
7.57182062e-01 2.42045224e-01 1.35620788e-01 3.74221146e-01
7.00353310e-02 9.92605507e-01 1.02355920e-01 6.32697284e-01
1.25775957e+00 3.02162822e-02 -7.13443607e-02 -1.32103369e-01
4.86890227e-01 -3.56223673e-01 4.34173822e-01 1.17967999e+00
-1.16201125e-01 2.03402117e-01 4.87923801e-01 -6.15677349e-02
-4.78156805e-01 -8.57427716e-01 2.92741418e-01 1.60827839e+00
4.83469218e-01 -5.62769294e-01 -4.15364236e-01 -8.83916557e-01
-3.11283562e-02 9.39581335e-01 -6.51084721e-01 -4.92157370e-01
-5.80413043e-01 -9.49475944e-01 3.06710333e-01 5.43651819e-01
2.68492162e-01 -8.84173036e-01 1.86579564e-04 2.57899970e-01
-1.87400192e-01 -8.21638584e-01 -5.25930524e-01 2.27740496e-01
-8.52344155e-01 -8.52841854e-01 -3.86635989e-01 -3.30968380e-01
5.96374810e-01 7.14040160e-01 1.09069872e+00 1.31191283e-01
1.34585664e-01 4.09997821e-01 -5.28521359e-01 8.91827643e-02
-4.21973653e-02 2.51619697e-01 -1.37384124e-02 2.87222922e-01
4.52475995e-01 -9.16371942e-01 -1.12121141e+00 5.05511582e-01
-6.48537099e-01 6.22583479e-02 7.01046407e-01 6.44935191e-01
2.01058507e-01 1.51334286e-01 7.55330622e-01 -1.17457199e+00
8.25398564e-01 -7.75152802e-01 -3.70074183e-01 3.66772950e-01
-1.02254355e+00 7.77859427e-03 8.71268332e-01 -5.68649709e-01
-1.10877001e+00 -7.10585833e-01 -2.44838864e-01 -3.60492915e-01
9.42706168e-02 7.26854384e-01 -1.39164507e-01 6.11098893e-02
4.85558301e-01 2.91198999e-01 -4.97040331e-01 -4.60495502e-01
4.94270086e-01 5.77471554e-01 1.23205952e-01 -6.36329293e-01
7.26469338e-01 2.08286762e-01 -1.83765560e-01 -3.25859517e-01
-9.91669595e-01 -5.10603845e-01 1.34323582e-01 -3.32174152e-01
3.57317984e-01 -9.13672626e-01 -1.21093833e+00 1.14909731e-01
-6.95992887e-01 -3.33697081e-01 2.72034537e-02 3.37220579e-01
-7.00084120e-02 1.57008424e-01 -7.67608047e-01 -9.00142908e-01
-4.69440401e-01 -8.74491990e-01 3.27969104e-01 5.14944494e-01
1.49150595e-01 -9.44597244e-01 -2.18946233e-01 4.92894113e-01
6.97554946e-01 -5.74993670e-01 9.46347117e-01 -1.04956198e+00
-9.12158132e-01 -9.30078030e-02 -3.13466907e-01 2.45739013e-01
-9.24745128e-02 -2.06293285e-01 -9.06927645e-01 -3.04447383e-01
-2.02835172e-01 8.95642787e-02 9.26975787e-01 2.45052934e-01
1.44151103e+00 -2.78429925e-01 -2.42822871e-01 3.68134379e-01
1.39912939e+00 2.85434067e-01 6.71056151e-01 1.43136634e-02
9.64264214e-01 3.22875738e-01 4.44915324e-01 2.68608898e-01
6.91381037e-01 6.22133613e-01 5.45233071e-01 2.11610615e-01
-5.13365231e-02 -5.88103354e-01 5.83117425e-01 1.11004603e+00
-4.11801904e-01 -7.82825589e-01 -4.18328613e-01 1.57768175e-01
-2.16291714e+00 -9.53992963e-01 -2.84109145e-01 2.39349866e+00
5.57007611e-01 6.27938449e-01 1.81848794e-01 -3.06244105e-01
6.68439567e-01 2.41450965e-01 -8.26129377e-01 -2.96007097e-01
1.65596187e-01 4.00715582e-02 5.67239404e-01 2.84524590e-01
-6.96211278e-01 7.37633467e-01 5.73631144e+00 9.24680829e-01
-9.46170390e-01 7.13291438e-03 5.52192271e-01 -2.71024466e-01
-5.97311437e-01 1.33074299e-01 -9.29713666e-01 5.93694210e-01
1.18738449e+00 -2.32633064e-03 1.01550913e+00 7.25244999e-01
2.01673180e-01 1.16000041e-01 -1.11134505e+00 7.21194565e-01
1.23691283e-01 -1.20142317e+00 2.12671578e-01 1.85841665e-01
5.37078857e-01 8.00019205e-02 8.48526061e-02 8.48424911e-01
5.36329269e-01 -8.04174066e-01 4.68191177e-01 6.93034410e-01
1.72593892e-01 -7.92269588e-01 4.01315868e-01 3.78007025e-01
-1.12221384e+00 -4.13285434e-01 -4.46753472e-01 -1.77627891e-01
7.23157451e-02 5.15249491e-01 -4.98978645e-01 8.77739668e-01
6.13454819e-01 6.51536822e-01 -7.88032532e-01 1.03229105e+00
-2.14521810e-01 1.11740446e+00 -2.18137115e-01 -2.86999643e-01
2.31497481e-01 -4.59646344e-01 5.09687364e-01 9.58826661e-01
1.18680581e-01 2.74790749e-02 -7.41277188e-02 8.58590484e-01
-4.06491280e-01 -3.00701689e-02 -9.52509716e-02 -1.10611767e-01
3.81068021e-01 1.62692857e+00 -6.66789532e-01 -1.68582574e-01
-6.43162787e-01 8.27677608e-01 5.24211526e-01 6.83491766e-01
-1.42384756e+00 -1.27515823e-01 6.64282680e-01 2.87652314e-01
5.13030291e-01 -2.73385178e-02 -2.02749491e-01 -1.24462497e+00
2.90419087e-02 -9.66392994e-01 4.86491412e-01 -5.58094740e-01
-1.64563763e+00 4.39132482e-01 -3.06830853e-01 -1.04920769e+00
1.62956119e-01 -4.40666288e-01 -7.10014880e-01 8.45643997e-01
-1.55516326e+00 -9.50963914e-01 -2.37613082e-01 5.83379745e-01
4.36057836e-01 1.34718806e-01 5.12416422e-01 6.68152869e-01
-8.85992944e-01 1.03719461e+00 1.92996114e-02 -2.70071533e-02
6.00115478e-01 -1.17352498e+00 4.15333807e-01 8.43151748e-01
4.01749164e-01 1.26380444e+00 5.30812025e-01 -4.15907621e-01
-1.71585584e+00 -1.12247491e+00 4.20488238e-01 -3.54513019e-01
7.99509406e-01 -5.73207617e-01 -9.45033312e-01 7.34485090e-01
5.86742423e-02 -1.61459550e-01 7.63823450e-01 7.89881647e-01
-4.71156240e-01 -4.37164217e-01 -7.15924263e-01 9.34270859e-01
1.54475701e+00 -4.67823923e-01 -2.95017451e-01 1.13576613e-01
7.95071602e-01 -5.43117896e-02 -7.16090143e-01 3.51451069e-01
5.68449616e-01 -9.19528544e-01 9.54251587e-01 -9.63259816e-01
3.01777393e-01 -3.67584437e-01 -1.23119637e-01 -1.13933659e+00
-7.99781084e-01 -6.66731298e-01 -7.27910101e-01 1.22903538e+00
6.36554778e-01 -7.44620323e-01 7.59377599e-01 8.31382275e-01
-2.15531603e-01 -9.56358075e-01 -3.83546233e-01 -5.80733418e-01
-3.37829351e-01 -5.67183733e-01 5.93242109e-01 8.45979989e-01
3.95201594e-02 7.36871123e-01 -6.16191328e-01 3.77949446e-01
4.85824585e-01 3.38652760e-01 4.71091837e-01 -1.22617590e+00
-9.62560475e-01 -4.65776503e-01 3.94917530e-04 -1.62464595e+00
-2.31862009e-01 -9.59869266e-01 -1.22773357e-01 -1.62667584e+00
3.56136769e-01 -6.18692458e-01 -1.14113986e+00 3.08837950e-01
-3.59352320e-01 7.29909167e-04 6.06947346e-03 1.39579803e-01
-1.18503857e+00 3.88004899e-01 1.31776989e+00 -1.21059949e-02
-2.13024989e-01 3.02895635e-01 -1.38138306e+00 5.08438289e-01
7.50241637e-01 -3.12950730e-01 -8.14564168e-01 -2.99580067e-01
6.57829523e-01 -7.15346783e-02 3.14391285e-01 -7.74594963e-01
5.03371418e-01 -2.43802458e-01 1.02462098e-01 -3.76866728e-01
2.24105462e-01 -8.48243713e-01 6.67202845e-02 5.74686676e-02
-6.58965468e-01 -2.87429214e-01 -2.11998284e-01 1.01273119e+00
2.09451407e-01 3.89171466e-02 4.16296124e-01 1.78282801e-02
-7.02457786e-01 5.58566988e-01 -1.55960009e-01 -1.27819330e-01
5.89486837e-01 -8.69947597e-02 -4.21604097e-01 -4.54725534e-01
-6.24297678e-01 3.79922450e-01 8.89915880e-03 7.26369739e-01
3.32398564e-01 -1.15605617e+00 -3.14382195e-01 5.60887493e-02
1.19069852e-01 -2.90094614e-01 6.79178417e-01 1.05830431e+00
3.79643440e-02 4.88023728e-01 1.50040984e-01 -2.04194322e-01
-7.37755537e-01 7.42391527e-01 3.51147503e-01 -6.58581316e-01
-4.52809632e-01 8.61309707e-01 2.13578701e-01 -3.48721445e-01
3.31299275e-01 -2.14230970e-01 -3.85052353e-01 -9.02873948e-02
3.42052907e-01 4.48084086e-01 1.74318880e-01 -5.10877445e-02
-3.46232563e-01 -1.30368173e-01 -5.01246691e-01 2.53809303e-01
1.25654066e+00 -3.53229642e-01 4.71001677e-02 4.01584715e-01
7.70754337e-01 7.74110183e-02 -1.12237883e+00 -2.79877961e-01
-1.56658351e-01 -2.80334115e-01 1.70849472e-01 -1.07707334e+00
-1.10084212e+00 6.92967176e-01 3.47604483e-01 6.36362910e-01
1.17601585e+00 -1.92726150e-01 8.33823740e-01 4.90957677e-01
4.93041247e-01 -1.02707207e+00 -2.36220136e-02 4.83919501e-01
4.60309774e-01 -1.04709852e+00 -1.21053025e-01 -3.46392900e-01
-5.79839170e-01 7.12308645e-01 9.48200583e-01 -2.75622070e-01
7.36866534e-01 -2.09010527e-01 -4.73514497e-01 -1.18856207e-01
-1.17947483e+00 -2.24757105e-01 4.55323249e-01 1.74094379e-01
4.23369020e-01 2.16191769e-01 -3.82334054e-01 1.17596579e+00
3.34661484e-01 -8.82310569e-02 3.61986339e-01 5.01085401e-01
-3.95690352e-01 -1.12275302e+00 2.23063082e-01 8.56474638e-01
-2.72520125e-01 -3.80779058e-01 -7.19324052e-02 5.02731919e-01
-1.78374901e-01 1.11093390e+00 -7.82072023e-02 -7.58604527e-01
4.43090379e-01 6.01703785e-02 3.65510076e-01 -3.48872751e-01
-8.07154953e-01 3.41734325e-04 1.60822153e-01 -7.54088581e-01
-1.25143066e-01 -2.73338526e-01 -1.12356806e+00 -5.39888978e-01
-4.90872115e-01 2.27252185e-01 3.84382367e-01 7.92269230e-01
7.25528657e-01 9.58528697e-01 7.41723239e-01 -3.88648480e-01
-8.30046892e-01 -9.71359253e-01 -6.04311526e-01 2.86731064e-01
-6.34156242e-02 -5.62991440e-01 -4.30454046e-01 -4.96461064e-01] | [10.139397621154785, 5.578834533691406] |
338a93ca-13b6-4421-ac7d-19a240d04fc8 | german-dialect-identification-and-mapping-for | null | null | https://aclanthology.org/2022.eurali-1.10 | https://aclanthology.org/2022.eurali-1.10.pdf | German Dialect Identification and Mapping for Preservation and Recovery | Many linguistic projects which focus on dialects do collection of audio data, analysis, and linguistic interpretation on the data. The outcomes of such projects are good language resources because dialects are among less-resources languages as most of them are oral traditions. Our project Dialektatlas Mittleres Westdeutschland (DMW) 1 focuses on the study of German language varieties through collection of audio data of words and phrases which are selected by linguistic experts based on the linguistic significance of the words (and phrases) to distinguish dialects among each other. We used a total of 7,814 audio snippets of the words and phrases of eight different dialects from middle west Germany. We employed a multilabel classification approach to address the problem of dialect mapping using Support Vector Machine (SVM) algorithm. The experimental result showed a promising accuracy of 87%. | ['Sabine Roller', 'Aynalem Tesfaye Misganaw'] | null | null | null | null | eurali-lrec-2022-6 | ['dialect-identification'] | ['natural-language-processing'] | [-2.35362872e-01 -1.96818173e-01 -1.70290813e-01 -6.97959363e-01
-1.12761188e+00 -8.15213561e-01 4.70257491e-01 4.02843386e-01
-4.99819756e-01 5.92604995e-01 6.61379695e-01 -4.09744948e-01
-1.78053707e-01 -6.98745966e-01 -1.36926115e-01 -5.08066893e-01
-1.60596147e-01 5.60116231e-01 -7.00621307e-03 -6.15608454e-01
2.96663225e-01 2.31729552e-01 -1.36941004e+00 4.87529665e-01
7.02669263e-01 3.98706049e-01 3.74901593e-01 3.62483889e-01
-5.22553980e-01 2.12607160e-01 -6.06955171e-01 -3.73990893e-01
1.30506158e-01 -2.99612582e-01 -1.07853174e+00 -7.65302777e-02
5.59478700e-01 2.47636750e-01 1.32074282e-01 1.11424458e+00
5.63835859e-01 -5.87300994e-02 6.34524286e-01 -7.72656381e-01
-4.87413794e-01 1.31756485e+00 -3.07973027e-01 2.23319203e-01
6.78851902e-01 -2.47105733e-01 1.50606644e+00 -8.64907622e-01
7.75594890e-01 1.33580983e+00 7.31722891e-01 1.48602873e-01
-1.45214009e+00 -7.52930224e-01 -1.50640253e-02 1.44362196e-01
-1.58151615e+00 -5.77870190e-01 6.53829813e-01 -9.01996255e-01
6.85784280e-01 2.37193599e-01 6.28940940e-01 6.25810862e-01
-1.61284924e-01 3.63094836e-01 1.20882463e+00 -8.78966570e-01
-5.21206520e-02 8.57853532e-01 4.18009669e-01 5.51171958e-01
-3.13475311e-01 -7.24739879e-02 -7.71830738e-01 -2.08455265e-01
3.09606135e-01 -7.98019350e-01 -2.32841238e-01 1.55046150e-01
-1.22257757e+00 1.10155380e+00 1.24892734e-01 8.51825356e-01
-1.67466387e-01 -4.47481245e-01 7.89086282e-01 6.77786767e-01
6.02482617e-01 1.81099787e-01 -5.85342824e-01 -1.21501563e-02
-6.57608867e-01 2.56146848e-01 6.35625780e-01 6.60862088e-01
6.61621869e-01 1.59760058e-01 3.72490048e-01 1.47592890e+00
5.62038183e-01 5.69195688e-01 6.28966153e-01 -4.98802274e-01
5.57462692e-01 4.49740499e-01 -2.70439655e-01 -1.00661707e+00
-3.18744868e-01 1.23820655e-01 -3.78343016e-01 1.33721173e-01
6.05592668e-01 -2.36967310e-01 -3.51388037e-01 1.57709575e+00
2.05437720e-01 -5.73221803e-01 2.36730948e-02 6.84878886e-01
9.66173172e-01 8.54637563e-01 1.92436785e-01 -9.51622203e-02
1.29090929e+00 -2.50749588e-01 -5.20829380e-01 1.74977154e-01
8.12734425e-01 -1.22165763e+00 1.42281616e+00 5.41820526e-01
-9.20671284e-01 -4.70091939e-01 -7.62798071e-01 2.05533072e-01
-6.12462640e-01 1.76253483e-01 6.01945698e-01 9.53633249e-01
-1.10459375e+00 2.40533054e-02 -4.01203744e-02 -4.39951718e-01
-2.58111298e-01 3.66399735e-01 -7.91382611e-01 4.04633701e-01
-1.29180253e+00 6.40126169e-01 2.41301715e-01 -2.00149998e-01
-4.34340127e-02 -8.56517375e-01 -9.39827979e-01 -2.68696278e-01
-3.48845750e-01 2.34530598e-01 9.02618170e-01 -1.23619592e+00
-1.23987734e+00 1.67454708e+00 5.58758117e-02 -2.34822795e-01
1.52698323e-01 1.77959442e-01 -9.27751780e-01 1.76871549e-02
4.62308615e-01 4.36080575e-01 2.18596637e-01 -8.50149095e-01
-1.02211797e+00 -3.50612670e-01 -3.66393179e-01 -2.36233711e-01
-3.77717793e-01 4.78006363e-01 2.86982983e-01 -9.05789614e-01
3.32478255e-01 -7.38155186e-01 3.64682019e-01 -4.48680788e-01
-2.68166095e-01 -3.16009879e-01 2.39485592e-01 -1.17386234e+00
1.12439060e+00 -2.36696410e+00 -2.28463840e-02 5.78365505e-01
-1.12745851e-01 -1.09495781e-01 6.68812320e-02 6.70495927e-01
-2.49227479e-01 8.74917880e-02 2.33272791e-01 3.55313458e-02
2.41749585e-01 -5.35108894e-02 -3.98542583e-01 4.73972291e-01
-6.81766421e-02 3.18334848e-02 -7.79158771e-01 -5.55129111e-01
3.17951888e-01 2.66201675e-01 -3.15120101e-01 -3.54138345e-01
4.30432819e-02 8.95532072e-02 -7.40951598e-02 6.36824608e-01
3.68556201e-01 8.21122408e-01 1.29703149e-01 -2.02582311e-02
-4.86349672e-01 7.11641550e-01 -1.18502164e+00 1.30146277e+00
-6.98818982e-01 1.12717521e+00 1.71508208e-01 -8.49257469e-01
1.21345985e+00 4.52808887e-01 3.59326214e-01 -2.26724580e-01
4.45965379e-02 4.96520132e-01 3.84529978e-02 -4.97451007e-01
6.45183623e-01 -2.91090757e-01 -5.39713860e-01 2.76873827e-01
1.89612046e-01 -1.77103877e-01 1.62243277e-01 -5.72115630e-02
5.63007772e-01 -2.42267430e-01 4.97809440e-01 -8.71120989e-01
7.39098132e-01 2.40543455e-01 5.52177489e-01 1.41555771e-01
-1.96073905e-01 3.35753232e-01 3.29882443e-01 -3.90706509e-01
-8.09936762e-01 -1.17976427e+00 -7.34151661e-01 1.49367917e+00
-4.65768605e-01 -4.32864577e-01 -6.15795255e-01 -5.28117977e-02
-1.14653222e-01 6.53655291e-01 -1.37963235e-01 3.15757960e-01
-3.97630244e-01 -3.72248828e-01 6.97243154e-01 1.07580751e-01
2.42502257e-01 -1.05709982e+00 1.93110764e-01 3.10468048e-01
-6.75863400e-02 -9.25700605e-01 -3.46746713e-01 7.97233433e-02
-1.44511506e-01 -7.90890455e-01 -4.54799920e-01 -1.44120276e+00
2.19923154e-01 -1.01441853e-01 8.54027987e-01 -3.78712147e-01
-2.24810690e-01 1.94425017e-01 -4.29726779e-01 -3.68626416e-01
-7.61343896e-01 3.21679711e-01 3.27226549e-01 -4.63409200e-02
8.21932197e-01 -5.79442739e-01 4.29491043e-01 2.94816494e-01
-2.31227651e-01 -3.87765795e-01 1.26732454e-01 7.40016520e-01
4.63573277e-01 7.06596449e-02 7.30177462e-01 -9.33186293e-01
6.05454862e-01 -4.80448157e-01 -7.01637268e-01 1.75517917e-01
-1.98580176e-01 -2.71923900e-01 5.88583052e-01 -3.19246322e-01
-8.45101297e-01 1.23412177e-01 -3.70110601e-01 5.34027040e-01
-4.63876545e-01 6.12052262e-01 -3.74457657e-01 -1.15302570e-01
7.77828395e-01 1.25785917e-02 -2.49412313e-01 -5.70358515e-01
1.28826275e-01 1.40330887e+00 6.89890921e-01 -8.38595808e-01
3.77362549e-01 9.77982022e-03 -3.62170100e-01 -1.45282853e+00
-2.18941391e-01 -7.26975560e-01 -6.90418065e-01 -4.85611320e-01
6.38765395e-01 -9.98159230e-01 -4.92999464e-01 4.50187773e-01
-8.54308367e-01 -9.10915807e-02 4.75727879e-02 7.87751019e-01
-3.37136447e-01 -1.39768988e-01 -5.38444102e-01 -8.94252062e-01
-1.76121928e-02 -8.76448393e-01 8.32244098e-01 -2.76188076e-01
-8.46499681e-01 -1.12338901e+00 2.30793223e-01 2.50542194e-01
7.68261589e-03 -2.05346346e-02 1.43494201e+00 -7.72780836e-01
-4.36067395e-02 -1.92612469e-01 1.46296635e-01 2.91673034e-01
1.96347252e-01 1.96423426e-01 -1.19811606e+00 1.26203120e-01
-3.72638047e-01 -1.83596119e-01 4.70973283e-01 3.36023718e-01
3.09569567e-01 6.07381091e-02 -7.60952570e-03 1.37077570e-01
1.35820532e+00 7.11380988e-02 1.95921168e-01 3.17553788e-01
3.50036800e-01 1.08122122e+00 6.10575616e-01 2.81544089e-01
3.19699466e-01 8.29750776e-01 -3.58480334e-01 2.51889914e-01
-5.21497466e-02 -2.58370996e-01 8.47981811e-01 1.25084162e+00
6.64012879e-02 1.92613557e-01 -1.17594612e+00 1.12228608e+00
-1.28786325e+00 -8.86023760e-01 -5.59078813e-01 2.22800899e+00
9.27067339e-01 1.42375410e-01 6.18152797e-01 6.81206524e-01
1.00907636e+00 5.24029415e-03 3.98918599e-01 -5.72263658e-01
-2.91619658e-01 3.06880981e-01 3.96776378e-01 8.61867368e-01
-8.48191857e-01 1.30514669e+00 6.39048338e+00 1.01200855e+00
-1.11048174e+00 -1.36266246e-01 1.89979702e-01 6.36225007e-03
-3.35966855e-01 -2.26565316e-01 -9.95199919e-01 3.71094257e-01
1.07133448e+00 -3.31177205e-01 4.26496774e-01 6.25554979e-01
5.05811632e-01 -2.38896325e-01 -9.10967410e-01 8.27850282e-01
-5.52476682e-02 -1.04454446e+00 -9.53664780e-02 -1.00063421e-01
5.20983338e-01 9.11861882e-02 6.00643158e-02 9.26117003e-02
3.57216746e-01 -8.19490433e-01 1.08444560e+00 8.69981498e-02
6.74086988e-01 -9.43972945e-01 6.01950765e-01 8.03551357e-03
-1.17873657e+00 1.35112405e-01 -5.11294544e-01 -1.52003273e-01
1.02164149e-01 7.31178284e-01 -8.09588492e-01 2.24029362e-01
6.40299737e-01 4.88575846e-01 -2.71725148e-01 7.64920950e-01
-2.28667587e-01 1.07857966e+00 -3.54359001e-01 -3.31531584e-01
3.27519536e-01 -2.86730975e-01 6.82894945e-01 1.43460917e+00
1.88188300e-01 -5.08242846e-01 2.75689006e-01 4.78717685e-01
3.12990516e-01 8.91903937e-01 -6.56850815e-01 -4.24570125e-03
6.92570150e-01 9.69235659e-01 -5.73893905e-01 5.62488586e-02
-5.31169593e-01 2.87478477e-01 2.59217042e-02 8.15191194e-02
-2.03934863e-01 -6.71878278e-01 9.11931813e-01 4.05059844e-01
-1.22762710e-01 -2.70379722e-01 -4.60892171e-01 -5.50038755e-01
8.56071413e-02 -6.95144475e-01 4.18080539e-01 -3.64505589e-01
-1.54851997e+00 7.47939527e-01 -2.37010829e-02 -1.01135814e+00
-4.38587844e-01 -7.63614833e-01 -2.70921201e-01 1.21128488e+00
-8.76421750e-01 -1.17163920e+00 2.07055911e-01 6.15791261e-01
6.02264404e-01 -9.05462027e-01 1.22130001e+00 4.29732233e-01
-9.63694826e-02 4.69144553e-01 2.58776009e-01 3.58199894e-01
9.86984432e-01 -1.40648043e+00 7.33258277e-02 1.85287625e-01
4.25118536e-01 2.38231272e-01 6.64090097e-01 -4.86486554e-01
-6.65778935e-01 -7.84620464e-01 1.71275330e+00 -2.87983492e-02
1.14747334e+00 -5.90750039e-01 -5.52350044e-01 5.09782553e-01
3.32344979e-01 -7.91145146e-01 1.13949883e+00 5.40182352e-01
-3.22874039e-01 -1.74699560e-01 -1.04114926e+00 4.17405635e-01
6.58527553e-01 -9.41232502e-01 -6.83387697e-01 4.44852382e-01
4.78147626e-01 1.53900295e-01 -9.40823317e-01 -4.68887329e-01
6.02968574e-01 -8.90363157e-01 7.68840969e-01 -2.99045593e-01
1.96643397e-01 -1.20943330e-01 -5.98712683e-01 -1.63196325e+00
-4.49591316e-02 -6.58945203e-01 1.29343271e+00 1.91458690e+00
6.21975482e-01 -8.30642521e-01 2.49881178e-01 2.34468117e-01
2.51366720e-02 2.67062306e-01 -8.83707285e-01 -7.31494009e-01
2.89908916e-01 -6.23649776e-01 4.43429977e-01 1.17760122e+00
5.93758166e-01 5.67687213e-01 -3.45617533e-03 5.72017990e-02
2.14507252e-01 8.26239586e-03 2.69641578e-01 -1.43562460e+00
-2.25253537e-01 -5.74848413e-01 -9.83672380e-01 -1.35109305e-01
7.74743497e-01 -1.23315668e+00 5.14264368e-02 -8.44068706e-01
-5.97684920e-01 -6.00870192e-01 1.73792228e-01 3.96089762e-01
3.20308656e-01 1.20425344e-01 -3.21520120e-02 -9.08384472e-02
3.60173494e-01 -4.80054989e-02 4.22692806e-01 -2.52251178e-01
-5.08131802e-01 3.24703276e-01 -5.34123957e-01 7.43361294e-01
8.40733945e-01 -4.38438326e-01 -2.00522840e-01 -4.21152979e-01
7.43547156e-02 -2.18142897e-01 8.13139975e-02 -7.46366858e-01
3.83552574e-02 -4.56825159e-02 -7.25185499e-02 -4.33576077e-01
1.18072815e-01 -6.48862123e-01 9.04546455e-02 1.71858609e-01
-5.49245000e-01 1.07688755e-01 2.27678299e-01 -1.78772155e-02
-7.16172457e-01 -5.20874619e-01 8.33276212e-01 2.24834681e-02
-8.30282032e-01 -3.09182882e-01 -1.00191414e+00 5.87364770e-02
8.68696809e-01 -1.21743590e-01 5.76106943e-02 -4.17567074e-01
-7.45473921e-01 -1.88685045e-01 2.66216040e-01 1.78550944e-01
5.08890040e-02 -1.36470616e+00 -1.09999084e+00 4.40917969e-01
3.22085679e-01 -6.64582610e-01 -4.33330163e-02 4.88901824e-01
-6.93377554e-01 4.51323777e-01 -4.00197387e-01 -3.08969647e-01
-1.67176259e+00 5.86691126e-02 4.52966318e-02 3.04132968e-01
-5.21569312e-01 8.95454466e-01 -4.79126945e-02 -6.85099423e-01
2.94996828e-01 -6.47171661e-02 -4.71497625e-01 6.06970251e-01
3.86254817e-01 2.04285607e-01 2.06555113e-01 -1.20092762e+00
-4.33765799e-01 6.25885904e-01 9.01083872e-02 -7.02673018e-01
1.29640758e+00 -5.22214286e-02 -3.57461095e-01 9.82091725e-01
1.20134842e+00 5.71000159e-01 -1.43839940e-01 -2.81503320e-01
3.93763930e-01 -5.12537718e-01 2.53666401e-01 -5.09335101e-01
-6.97158217e-01 7.54607618e-01 4.77533519e-01 4.85063553e-01
7.83821464e-01 2.00636964e-02 4.04630691e-01 1.50479674e-01
5.61027884e-01 -1.48265839e+00 -7.81595945e-01 5.28687477e-01
7.24561453e-01 -9.34479475e-01 -4.06806380e-01 -6.32289767e-01
-5.04330337e-01 1.18906343e+00 -2.22769752e-02 -9.53393057e-02
9.71762478e-01 2.47024968e-01 5.01763523e-01 4.78338003e-02
-4.26590621e-01 -3.16814154e-01 1.65380806e-01 1.04881322e+00
8.67736459e-01 4.51430887e-01 -6.58819139e-01 8.06965292e-01
-9.73345995e-01 -3.95218879e-01 3.46312732e-01 4.60604787e-01
-4.80760485e-01 -1.36924350e+00 -6.45521224e-01 2.99692839e-01
-4.04870778e-01 -3.04816842e-01 -6.12300456e-01 8.35337698e-01
2.10394815e-01 1.02182913e+00 9.93710011e-02 -5.76705933e-01
2.24646792e-01 1.92535147e-01 1.78074792e-01 -6.16058290e-01
-8.56614351e-01 2.29206339e-01 6.65568709e-01 1.31875277e-01
-4.41633195e-01 -1.11039114e+00 -1.18388367e+00 -6.43091917e-01
-1.14211790e-01 3.83713901e-01 8.60486805e-01 8.76193285e-01
-4.51470882e-01 1.08813263e-01 9.15830791e-01 -6.29189551e-01
-1.65504858e-01 -9.64602530e-01 -1.28912759e+00 1.36386910e-02
-5.65962233e-02 -3.16436648e-01 -3.59151959e-01 4.37331468e-01] | [10.341818809509277, 10.422983169555664] |
12af87c1-87d7-47df-a3dd-a92f4e386b84 | hidden-poison-machine-unlearning-enables | 2212.10717 | null | https://arxiv.org/abs/2212.10717v1 | https://arxiv.org/pdf/2212.10717v1.pdf | Hidden Poison: Machine Unlearning Enables Camouflaged Poisoning Attacks | We introduce camouflaged data poisoning attacks, a new attack vector that arises in the context of machine unlearning and other settings when model retraining may be induced. An adversary first adds a few carefully crafted points to the training dataset such that the impact on the model's predictions is minimal. The adversary subsequently triggers a request to remove a subset of the introduced points at which point the attack is unleashed and the model's predictions are negatively affected. In particular, we consider clean-label targeted attacks (in which the goal is to cause the model to misclassify a specific test point) on datasets including CIFAR-10, Imagenette, and Imagewoof. This attack is realized by constructing camouflage datapoints that mask the effect of a poisoned dataset. | ['Ayush Sekhari', 'Gautam Kamath', 'Jayadev Acharya', 'Jack Douglas', 'Jimmy Z. Di'] | 2022-12-21 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [ 6.48816943e-01 2.61991888e-01 -2.23448440e-01 -9.82844532e-02
-6.59516990e-01 -1.15105164e+00 5.46626210e-01 2.94130534e-01
-5.00748992e-01 8.17358196e-01 -1.77495360e-01 -4.46865231e-01
5.75768501e-02 -7.37698078e-01 -1.01358593e+00 -9.65915740e-01
3.04679815e-02 3.98043454e-01 4.97830473e-02 8.71393159e-02
2.86251187e-01 7.05561996e-01 -7.15751588e-01 2.57249504e-01
4.02048379e-01 5.59999287e-01 -5.22887945e-01 5.07824063e-01
4.39243376e-01 8.95527244e-01 -9.60194051e-01 -5.88945568e-01
5.96472442e-01 -1.95354536e-01 -7.20302880e-01 1.20899662e-01
4.99682724e-01 -4.07240063e-01 -4.87826735e-01 1.41546857e+00
1.91506371e-02 -1.82180420e-01 3.82517189e-01 -1.68641949e+00
-4.90871876e-01 5.11381507e-01 -5.30324757e-01 3.11149955e-01
6.68969899e-02 6.24319434e-01 6.21007562e-01 -5.60306132e-01
4.38136518e-01 1.07291222e+00 3.74932200e-01 9.93691385e-01
-1.39526033e+00 -1.10064018e+00 1.21856607e-01 -2.79712409e-01
-1.13059318e+00 -4.21293169e-01 7.20987141e-01 -3.24223876e-01
2.11026832e-01 5.65334141e-01 8.94273594e-02 1.52164841e+00
3.20515633e-01 4.55552995e-01 1.04267991e+00 -1.94065586e-01
7.53434300e-01 1.15967169e-01 3.26758325e-01 4.12956923e-01
5.58785796e-01 5.47985256e-01 -3.27375561e-01 -1.05973768e+00
2.77244300e-01 2.61975199e-01 -3.11825573e-01 -3.67710918e-01
-7.99674094e-01 9.58780766e-01 4.31465328e-01 -1.87578246e-01
-3.85073662e-01 1.98005393e-01 3.90523225e-01 3.35862815e-01
3.63476664e-01 9.03827488e-01 -5.82785249e-01 3.04482967e-01
-4.19026613e-01 3.88393342e-01 4.60601389e-01 3.75972688e-01
7.67995656e-01 5.94376475e-02 -1.60374999e-01 1.74089327e-01
-1.85226217e-01 4.45287347e-01 3.69396836e-01 -6.48518622e-01
2.94744432e-01 5.28109372e-01 3.51313770e-01 -7.27701366e-01
9.51884314e-02 -3.52529526e-01 -4.10513312e-01 5.12358785e-01
4.67007071e-01 -3.37598920e-01 -1.11398673e+00 1.91674435e+00
4.66439277e-01 6.59774721e-01 1.28934011e-01 8.70772660e-01
1.11396201e-01 3.41732413e-01 3.52141678e-01 -1.33321779e-02
9.28721547e-01 -5.01299500e-01 -3.40033829e-01 -3.44282866e-01
8.72152805e-01 -1.42558455e-01 1.24634886e+00 5.31567574e-01
-6.82654381e-01 -1.34813994e-01 -1.21992147e+00 4.21744436e-01
-5.38714409e-01 -7.41283715e-01 3.67943078e-01 7.52959073e-01
-3.00143182e-01 5.37666559e-01 -7.57884562e-01 5.22976108e-02
7.69947052e-01 4.96292591e-01 -5.64686298e-01 -2.87192732e-01
-1.25173986e+00 7.05348194e-01 2.23391950e-01 -2.64395654e-01
-1.57757306e+00 -7.86483645e-01 -5.36485732e-01 1.37701735e-01
6.34726226e-01 -4.56998795e-01 1.10690975e+00 -1.02612448e+00
-6.14323497e-01 6.91506028e-01 2.96421438e-01 -9.24608529e-01
7.77424574e-01 -1.70798287e-01 -3.79078686e-01 6.05022646e-02
2.66186800e-02 5.29931486e-01 1.36644626e+00 -1.39168715e+00
-3.74381363e-01 -5.20986617e-01 1.46939963e-01 7.34981615e-03
-3.84221554e-01 -4.89755236e-02 2.94662528e-02 -8.18581223e-01
-4.83977526e-01 -1.08539236e+00 -4.31941211e-01 -3.01354855e-01
-8.95936012e-01 1.88845292e-01 1.46357405e+00 -1.75258771e-01
8.87256145e-01 -2.18709469e+00 -3.08348268e-01 3.96636456e-01
3.23221415e-01 7.28509724e-01 -4.86861527e-01 2.48246908e-01
-2.10533723e-01 4.97585654e-01 -4.13852483e-01 -1.24431521e-01
-3.16991448e-01 3.00368249e-01 -8.70549917e-01 6.52404010e-01
-9.47970226e-02 5.94889879e-01 -7.45700359e-01 2.95298874e-01
5.56894355e-02 1.03852093e-01 -8.26817751e-01 3.35016668e-01
-5.41016161e-01 4.87285554e-01 -6.38734758e-01 3.67356449e-01
5.70412219e-01 4.36607748e-02 -1.87358007e-01 3.56841236e-01
4.23184305e-01 2.39014894e-01 -7.42365062e-01 8.11297238e-01
2.40949132e-02 1.16172306e-01 -2.20155194e-01 -4.19010699e-01
6.05593204e-01 2.84123987e-01 1.76272437e-01 -1.07625790e-01
2.28069812e-01 -1.91336632e-01 1.02698192e-01 -2.85577357e-01
2.97783241e-02 -1.23449638e-01 -2.62490898e-01 6.83592618e-01
-4.19110477e-01 2.33411387e-01 -3.47130209e-01 3.90414894e-01
1.56830466e+00 -5.98897278e-01 1.16070338e-01 -1.06199563e-01
2.46477857e-01 2.71553844e-01 6.54051542e-01 1.13125372e+00
-3.65041018e-01 1.88088208e-01 5.18134475e-01 -5.74701488e-01
-1.16045797e+00 -1.05385840e+00 1.13515869e-01 9.87679839e-01
-6.84671253e-02 -2.01866239e-01 -1.09746659e+00 -1.70625627e+00
4.37313735e-01 9.87019300e-01 -8.94437015e-01 -7.97902167e-01
-2.77149469e-01 -5.63325703e-01 8.28690588e-01 1.21076100e-01
6.35490119e-01 -9.44517195e-01 -3.34262043e-01 -1.82647318e-01
2.26743519e-01 -6.97854161e-01 -7.56649137e-01 2.58091033e-01
-7.55477071e-01 -1.59262729e+00 1.03784941e-01 -4.09920156e-01
9.81583834e-01 1.63419545e-01 5.60918570e-01 4.65899795e-01
8.22304636e-02 -1.89476889e-02 -2.55551875e-01 -4.89291757e-01
-6.89486444e-01 2.77543068e-02 2.19048977e-01 2.02793881e-01
6.49444699e-01 -2.14703411e-01 -1.70078546e-01 3.11886251e-01
-1.24820006e+00 -4.71335381e-01 4.28223163e-02 7.67567217e-01
4.28666651e-01 3.77151847e-01 9.13730204e-01 -1.55603421e+00
5.91116726e-01 -6.34985745e-01 -5.95369816e-01 1.98640302e-01
-4.42799419e-01 -1.58908248e-01 9.03710008e-01 -1.13211465e+00
-5.27621806e-01 2.05207437e-01 2.05758289e-02 -9.63030279e-01
-4.37548935e-01 5.46702519e-02 -6.10437393e-01 -3.68300617e-01
9.90972340e-01 5.50023615e-02 -2.22048253e-01 -5.37950099e-01
1.59996733e-01 5.82022250e-01 6.66764677e-01 -4.09901857e-01
1.18811631e+00 4.77361143e-01 -5.12836240e-02 -4.58924413e-01
-8.47374022e-01 1.28314449e-02 -3.82278681e-01 7.08210021e-02
5.48607409e-01 -5.19033432e-01 -5.96283257e-01 5.31839550e-01
-8.79603505e-01 -4.23045397e-01 -4.01670247e-01 7.36436173e-02
-1.22854844e-01 8.28282163e-02 -3.81006688e-01 -6.24676585e-01
-1.17710739e-01 -1.15290356e+00 4.62947488e-01 6.93885908e-02
-3.04854810e-01 -8.05695713e-01 -7.36373216e-02 3.78840595e-01
4.38106880e-02 5.09870708e-01 1.16733134e+00 -1.58506930e+00
-3.69754225e-01 -6.80760145e-01 2.84388751e-01 4.68209654e-01
1.58717737e-01 -2.63881564e-01 -1.06908023e+00 -6.62581086e-01
4.68713045e-01 -5.19291997e-01 5.49304187e-01 -1.87684789e-01
1.57026219e+00 -9.72074032e-01 -4.38623428e-01 3.68229449e-01
9.90173519e-01 6.99709833e-01 5.00574946e-01 2.93785602e-01
7.58699179e-01 1.82973012e-01 3.88922781e-01 2.58212149e-01
-2.80068457e-01 3.95979643e-01 7.57283211e-01 -2.03108162e-01
4.05725002e-01 -7.33680427e-01 2.04031318e-01 -3.46488386e-01
8.74935448e-01 -4.32469189e-01 -7.50145137e-01 2.58969575e-01
-1.55736601e+00 -8.41768742e-01 2.22799167e-01 2.52167845e+00
7.83804774e-01 4.33877140e-01 6.33723065e-02 3.29093665e-01
7.82714665e-01 2.06757396e-01 -9.83241200e-01 -4.09982383e-01
1.15995832e-01 1.58022359e-01 7.57703960e-01 6.40852749e-01
-1.14031589e+00 1.19677448e+00 6.00635338e+00 6.88179314e-01
-1.08672702e+00 1.44517466e-01 1.07585442e+00 -3.72890830e-01
-2.28908047e-01 2.10036889e-01 -6.17308438e-01 5.90450466e-01
8.56156409e-01 4.39806841e-03 6.28924251e-01 9.57769752e-01
5.80044016e-02 8.54346380e-02 -1.19672966e+00 3.46071452e-01
-3.40056382e-02 -1.20697510e+00 1.03707917e-01 4.00468200e-01
5.57368338e-01 -3.42309959e-02 3.44920307e-01 3.27704191e-01
8.78808141e-01 -1.07064855e+00 4.02915090e-01 5.49550243e-02
6.48785472e-01 -1.13380980e+00 3.00287962e-01 7.75312841e-01
-1.40323356e-01 -3.83702427e-01 -2.36682087e-01 -8.04550126e-02
-2.88717628e-01 4.54010665e-01 -1.17980564e+00 -2.46794268e-01
2.84540892e-01 -7.69864246e-02 -5.59953868e-01 7.15628445e-01
-5.25021017e-01 1.16815412e+00 -1.57655746e-01 2.84665555e-01
3.42484295e-01 1.29985809e-01 6.85262799e-01 6.96589708e-01
-3.10892493e-01 1.86640069e-01 3.60879481e-01 8.08818221e-01
-6.15546703e-01 -3.70471984e-01 -1.15557969e+00 4.40926291e-02
8.34807694e-01 1.01543677e+00 -3.87027055e-01 -3.35238427e-01
-9.58534479e-02 1.04070246e+00 2.90248990e-01 4.72226828e-01
-8.05888116e-01 -1.92221537e-01 9.24398363e-01 1.54456422e-01
-5.53410910e-02 1.26305893e-01 -3.39822263e-01 -9.62609589e-01
-4.54794765e-01 -1.21488595e+00 7.13469326e-01 -7.52693951e-01
-1.37299240e+00 5.17356932e-01 -1.61341920e-01 -8.02514553e-01
-7.23403618e-02 -1.55306593e-01 -7.72864878e-01 7.68278062e-01
-1.00610638e+00 -6.78028166e-01 3.04399490e-01 6.60209596e-01
3.14510107e-01 -2.33057499e-01 7.33097255e-01 -1.28788844e-01
-7.29357779e-01 7.33551860e-01 -2.74575680e-01 3.00328344e-01
4.70350504e-01 -8.81765842e-01 5.71430147e-01 1.23145413e+00
3.03879589e-01 7.52904892e-01 7.96053708e-01 -1.10292161e+00
-1.04126787e+00 -1.48560798e+00 8.07987213e-01 -8.07576120e-01
5.16156673e-01 -6.49640739e-01 -1.22241819e+00 9.97308910e-01
-2.69466698e-01 2.97572255e-01 5.75190365e-01 -3.25561702e-01
-7.79811621e-01 1.73086911e-01 -1.87634921e+00 7.08738923e-01
5.48952162e-01 -6.19310677e-01 -5.50102592e-01 4.90489990e-01
8.09967995e-01 -2.18067050e-01 -2.75695711e-01 4.64677721e-01
8.37424323e-02 -5.03956556e-01 8.50678980e-01 -1.50700831e+00
-1.46454111e-01 -2.93126225e-01 -1.66698486e-01 -1.32196641e+00
-2.17042744e-01 -7.64277399e-01 -1.14150181e-01 9.12748575e-01
2.65290260e-01 -6.99850440e-01 9.91922855e-01 7.84996986e-01
3.00963670e-01 -4.20225382e-01 -8.94724369e-01 -6.56929255e-01
3.39555264e-01 -1.77535623e-01 9.45273459e-01 1.39081073e+00
-4.73786667e-02 2.08799586e-01 -5.38863599e-01 7.07914948e-01
6.77506864e-01 -4.21818137e-01 9.54055309e-01 -9.17948008e-01
-1.01620480e-01 1.73009723e-01 -7.14546666e-02 -5.17331600e-01
2.15196118e-01 -7.06138432e-01 -1.76701799e-01 -2.58521140e-01
2.03542933e-01 -4.72838759e-01 -4.59016919e-01 9.94774461e-01
-3.69175613e-01 3.58418077e-01 3.70526642e-01 4.33219999e-01
-2.98177004e-01 4.29100953e-02 9.30451751e-01 -1.47037417e-01
-8.27084109e-02 3.76553088e-01 -8.04883718e-01 6.47644997e-01
9.33394730e-01 -1.21181083e+00 -5.38452327e-01 -5.52142002e-02
-9.65047926e-02 1.49915973e-02 5.91755509e-01 -7.69810200e-01
9.02301520e-02 -3.22875559e-01 3.95131618e-01 -1.13085501e-01
3.50083709e-02 -1.07522738e+00 1.76552147e-01 8.36820662e-01
-8.68117988e-01 1.07643798e-01 1.16480760e-01 6.33801937e-01
3.02008688e-01 -3.78193974e-01 1.22187054e+00 6.76368400e-02
-4.43923585e-02 5.28359771e-01 -3.78750384e-01 5.82606271e-02
1.28768051e+00 1.11562684e-01 -7.15515256e-01 -2.27084816e-01
-6.38212681e-01 4.43574339e-02 7.50521600e-01 3.95226181e-01
5.30590355e-01 -1.01685870e+00 -3.97062808e-01 7.03013122e-01
5.57603911e-02 -1.33974940e-01 -5.32113612e-02 7.57982135e-02
7.77154192e-02 1.15759164e-01 -4.14992403e-03 -1.48149073e-01
-1.27552235e+00 1.17791140e+00 4.97424692e-01 -1.80368021e-01
-5.22247434e-01 6.73361361e-01 5.47063589e-01 -1.93560734e-01
4.28937703e-01 -2.63067298e-02 2.34714761e-01 -6.04115307e-01
6.16016984e-01 2.26890653e-01 2.35452041e-01 -4.19814974e-01
-2.59508222e-01 -3.42029423e-01 -5.48775613e-01 -5.25055565e-02
9.19015229e-01 3.33994508e-01 2.81166673e-01 2.91546900e-03
1.03810728e+00 3.52800816e-01 -1.70421863e+00 -2.16678962e-01
-6.83888933e-03 -6.42500460e-01 -5.50214760e-02 -1.25811124e+00
-1.04632497e+00 5.43448687e-01 4.25765902e-01 3.31313103e-01
1.05614591e+00 -2.34712735e-01 7.34819949e-01 4.07659620e-01
3.61174285e-01 -5.36430836e-01 2.21358240e-01 2.38411084e-01
5.32150388e-01 -9.52934682e-01 -2.54654944e-01 -6.67992681e-02
-6.99732125e-01 4.32866335e-01 8.09407711e-01 -4.13626611e-01
4.10281837e-01 3.92213345e-01 4.21874821e-02 -1.93434104e-01
-8.80611598e-01 5.20501554e-01 -1.03483416e-01 6.48158789e-01
-6.18177652e-01 -9.44885798e-03 1.30133092e-01 7.02948272e-01
-1.18199065e-01 -1.13272630e-01 6.89822555e-01 9.41321015e-01
-5.06608784e-01 -9.38108325e-01 -6.45009995e-01 4.74480540e-01
-6.82714760e-01 -7.92716220e-02 -8.76853645e-01 7.79374361e-01
1.95673272e-01 9.16629791e-01 1.75413070e-03 -6.29692316e-01
2.61285067e-01 3.24867755e-01 -1.12710893e-01 -7.54168034e-01
-8.47804189e-01 -1.03163451e-01 -6.76787868e-02 -4.46360648e-01
3.17188472e-01 -3.23579222e-01 -1.19056308e+00 -3.84551138e-01
-2.70958632e-01 5.41860998e-01 4.37418759e-01 9.92470860e-01
4.80821878e-01 3.48757096e-02 1.04006147e+00 -1.92939430e-01
-8.41312647e-01 -7.89107442e-01 -5.32752275e-01 6.37892723e-01
6.44149840e-01 -4.81678158e-01 -7.51927614e-01 4.29945439e-02] | [5.8069963455200195, 7.656182289123535] |
0c8f6e26-935f-498e-a3a2-104adc43b61e | i-vector-based-features-embedding-for-heart | 1904.11914 | null | https://arxiv.org/abs/1904.11914v3 | https://arxiv.org/pdf/1904.11914v3.pdf | Statistical feature embedding for heart sound classification | Cardiovascular Disease (CVD) is considered as one of the principal causes of death in the world. Over recent years, this field of study has attracted researchers' attention to investigate heart sounds' patterns for disease diagnostics. In this study, an approach is proposed for normal/abnormal heart sound classification on the Physionet challenge 2016 dataset. For the first time, a fixed-length feature vector; called i-vector; is extracted from each heart sound using Mel Frequency Cepstral Coefficient (MFCC) features. Afterwards, Principal Component Analysis (PCA) transform and Variational Autoencoder (VAE) are applied on the i-vector to achieve dimension reduction. Eventually, the reduced size vector is fed to Gaussian Mixture Models (GMMs) and Support Vector Machine (SVM) for classification purpose. Experimental results demonstrate the proposed method could achieve a performance improvement of 16% based on Modified Accuracy (MAcc) compared with the baseline system on the Physoinet dataset. | ['Bagher BabaAli', 'Saeedreza Shehnepoor', 'Mohammad Adiban'] | 2019-04-26 | null | null | null | null | ['sound-classification'] | ['audio'] | [ 1.65846989e-01 -1.07122645e-01 8.07674304e-02 3.70484293e-02
-6.54453695e-01 -7.12667406e-02 3.38162303e-01 4.30014700e-01
-2.25522116e-01 3.56667340e-01 3.23340654e-01 -3.25094461e-01
4.15550694e-02 -5.35988390e-01 1.24986567e-01 -7.81946957e-01
-2.03513861e-01 3.98075068e-03 1.80360526e-02 3.42120886e-01
1.22474119e-01 3.28679800e-01 -1.53215861e+00 1.47589236e-01
7.58107901e-01 1.05181444e+00 -3.66675220e-02 1.06088662e+00
1.67015232e-02 6.00178063e-01 -6.55183256e-01 -1.72115862e-01
-2.36882772e-02 -7.17292488e-01 -6.26260042e-01 -8.96841735e-02
-3.63393575e-02 1.69019252e-02 -2.02409849e-01 7.93067515e-01
8.62684906e-01 3.50989938e-01 7.22320080e-01 -8.24203253e-01
-1.48221955e-01 2.55370438e-01 -2.02805042e-01 6.86132550e-01
5.56206219e-02 -7.48432204e-02 8.67130578e-01 -9.26267028e-01
1.75613090e-01 9.82615829e-01 7.36092925e-01 4.80270714e-01
-1.11097586e+00 -4.72797304e-01 -6.29420340e-01 5.93305051e-01
-1.49217927e+00 -3.57524395e-01 1.23488522e+00 -4.95246142e-01
7.60472298e-01 6.76775455e-01 9.56645846e-01 1.01006782e+00
4.16591197e-01 5.61799347e-01 9.01267350e-01 -5.76587021e-01
4.65674400e-01 3.68597597e-01 2.09265232e-01 3.97147894e-01
-1.07790858e-01 6.46615252e-02 -5.57666160e-02 -4.71674412e-01
5.05797327e-01 -4.65051793e-02 -2.91854143e-01 1.52468951e-02
-1.07394278e+00 1.01685917e+00 -5.84692461e-03 5.45131505e-01
-8.02299917e-01 -2.56050944e-01 5.89708567e-01 -1.43951908e-01
3.81967247e-01 1.80670828e-01 -3.58573198e-01 -4.24752533e-01
-8.67881119e-01 1.82362780e-01 7.95134425e-01 9.08946097e-02
3.30898799e-02 2.15536430e-01 -3.10532510e-01 1.09826887e+00
4.66338962e-01 5.70613801e-01 8.45413744e-01 -9.54852521e-01
7.66471475e-02 2.75274485e-01 -2.64872611e-01 -1.32686806e+00
-5.10010302e-01 -8.39268684e-01 -1.30744505e+00 -3.66720915e-01
1.69329867e-01 -1.99539021e-01 -2.39783809e-01 1.20652747e+00
6.15708411e-01 7.61030376e-01 2.30999649e-01 7.34130561e-01
9.27711070e-01 7.09457576e-01 1.76443428e-01 -4.09990042e-01
1.61022663e+00 -5.04663348e-01 -6.85488701e-01 2.83972710e-01
3.65986139e-01 -7.45174766e-01 7.12175250e-01 5.46476603e-01
-6.54484153e-01 -8.47563446e-01 -7.50605941e-01 4.25693482e-01
-5.74758165e-02 3.20139855e-01 2.99122781e-01 1.18905509e+00
-4.64711249e-01 6.09041929e-01 -9.99988854e-01 -1.98278651e-01
4.40695196e-01 -1.29840776e-01 -1.09079398e-01 3.09350967e-01
-1.23865330e+00 4.73653734e-01 7.60460347e-02 -4.40961123e-02
-5.70644319e-01 -9.40616727e-01 -7.80402720e-01 1.38705760e-01
-4.09756839e-01 -5.42412519e-01 9.16978657e-01 -3.88250679e-01
-1.72802603e+00 4.70594108e-01 5.77005697e-03 -6.15000904e-01
8.67534652e-02 -2.32334271e-01 -8.49131942e-01 4.81094718e-01
-1.43941000e-01 5.61794303e-02 9.45978940e-01 -7.45086849e-01
-6.57107234e-01 -3.24983805e-01 -7.47076690e-01 -1.37243241e-01
-5.47674656e-01 4.58554663e-02 -1.25764549e-01 -8.50733399e-01
-3.54143279e-03 -9.80870187e-01 -3.72497179e-02 -4.63936985e-01
-5.23648739e-01 -4.63154972e-01 7.00458467e-01 -1.16205263e+00
1.71176457e+00 -2.51052046e+00 5.10657504e-02 2.89057612e-01
1.53797776e-01 6.19472444e-01 1.77947447e-01 1.22895725e-01
-1.39690265e-01 -3.64831984e-02 -2.34430969e-01 -2.31330201e-01
-2.88988501e-01 1.47910997e-01 -1.75413266e-01 5.92651546e-01
2.48830751e-01 4.35407549e-01 -6.73973501e-01 -6.47334874e-01
6.46531224e-01 8.77630472e-01 -7.28816450e-01 1.21237054e-01
3.86839002e-01 7.45454431e-01 -4.84214872e-01 2.59991556e-01
6.30653977e-01 1.14366710e-01 3.48050073e-02 -4.23772097e-01
-6.81475922e-02 -5.65924169e-03 -1.13678133e+00 1.17438519e+00
-3.25271457e-01 4.47882861e-01 -3.34046185e-01 -1.26995862e+00
9.50556755e-01 7.91355968e-01 8.24744284e-01 -3.86487365e-01
1.93276435e-01 -7.52147436e-02 3.05934012e-01 -8.62492800e-01
-3.16564709e-01 -2.44543493e-01 4.31255884e-02 -6.79769889e-02
-7.21711740e-02 2.67005414e-01 -8.50860924e-02 -1.65029585e-01
1.04381943e+00 -4.64958668e-01 3.67188960e-01 -7.00890571e-02
1.04381788e+00 -1.01630948e-01 6.32919073e-01 3.93764943e-01
-6.35296106e-01 5.00946105e-01 1.88219637e-01 -4.29685056e-01
-7.07132459e-01 -9.01154101e-01 -5.34285724e-01 5.59538364e-01
-4.05813128e-01 -5.04815042e-01 -7.49967575e-01 -4.46014047e-01
-9.25321504e-02 8.74176800e-01 -4.45594251e-01 -1.68375894e-01
-6.57016218e-01 -9.67839897e-01 8.41943622e-01 5.09561419e-01
3.86895984e-01 -1.06702602e+00 -1.03396654e+00 3.66554290e-01
-2.11779878e-01 -8.81250143e-01 -1.31957829e-01 -1.76773310e-01
-9.79056597e-01 -1.01011050e+00 -9.60326433e-01 -5.27285993e-01
4.50138450e-02 -2.08488852e-01 6.97967529e-01 -3.18028361e-01
-7.90901840e-01 3.49915892e-01 -5.61938703e-01 -5.40077507e-01
-6.21765792e-01 -2.24787131e-01 4.81840596e-02 3.93222064e-01
2.80814469e-01 -6.52160227e-01 -8.95876169e-01 -1.66585267e-01
-4.62573349e-01 -3.62831503e-01 5.79379737e-01 8.60515952e-01
6.07817054e-01 4.87203538e-01 7.34084010e-01 -6.31228507e-01
7.12889671e-01 -5.26951194e-01 -2.04420388e-01 -2.68544197e-01
-6.45384014e-01 -3.26509148e-01 7.87696362e-01 -3.35103422e-01
-9.36777890e-01 -1.48215458e-01 -5.05610466e-01 -5.17986417e-01
-4.78779078e-01 5.44223785e-01 9.37349573e-02 3.76023352e-01
6.57475531e-01 6.89912796e-01 1.62641212e-01 -7.99315631e-01
7.11629242e-02 9.98057425e-01 6.20916963e-01 -8.23338628e-02
4.06241596e-01 2.60851979e-01 5.78419156e-02 -1.29756498e+00
-5.23306012e-01 -7.27101326e-01 -4.72101301e-01 -2.01577261e-01
1.12786293e+00 -7.44183242e-01 -5.79840481e-01 3.14667642e-01
-8.83100510e-01 3.28862727e-01 -1.32986352e-01 1.02633858e+00
-1.75617650e-01 5.69637895e-01 -7.24396467e-01 -1.10861742e+00
-7.63340950e-01 -7.72289932e-01 6.59848094e-01 2.99162000e-01
-4.32647586e-01 -9.28814769e-01 4.13460165e-01 4.45187002e-01
5.23527861e-01 3.02965432e-01 1.05694246e+00 -8.15783918e-01
4.01207596e-01 -6.04378760e-01 1.08279362e-01 9.65537071e-01
1.53801039e-01 -1.42900184e-01 -9.98770654e-01 -1.58298403e-01
3.31710190e-01 3.66715103e-01 6.18713260e-01 7.15493202e-01
1.21450877e+00 -3.15409809e-01 -2.60515600e-01 3.88926148e-01
9.91769075e-01 3.62282008e-01 4.83483315e-01 -8.31905231e-02
5.76588988e-01 2.24687278e-01 3.48309219e-01 8.30473661e-01
3.03637236e-01 3.96558166e-01 6.29230142e-02 1.30619019e-01
-1.83236107e-01 1.46239266e-01 1.43606141e-01 1.24187028e+00
-2.45464057e-01 2.95317620e-01 -8.92358363e-01 4.94785845e-01
-1.25232148e+00 -1.04412341e+00 -4.53142256e-01 2.22949290e+00
5.90359628e-01 -9.02320817e-02 2.94051260e-01 8.87680471e-01
6.60048962e-01 9.59430728e-03 -1.65344283e-01 -2.14244723e-01
3.54050905e-01 5.18309772e-01 -1.39950067e-01 1.49156600e-01
-1.33471179e+00 1.62318692e-01 5.28588343e+00 7.58659661e-01
-1.43926620e+00 1.56251311e-01 5.40381730e-01 2.49229237e-01
3.09730738e-01 -3.17688286e-01 -3.68598700e-01 7.45237529e-01
1.44355118e+00 1.17965169e-01 2.16479763e-01 8.13037217e-01
3.65111768e-01 1.01887144e-01 -5.11289775e-01 1.25197065e+00
5.20430058e-02 -8.67636442e-01 -3.46548349e-01 -4.76068212e-03
2.45529890e-01 -1.23689555e-01 1.58920392e-01 5.14238834e-01
-6.22923851e-01 -6.96047843e-01 1.78999320e-01 6.17964268e-01
6.15553319e-01 -1.00200534e+00 7.94044912e-01 5.13881326e-01
-1.12608147e+00 -1.67588964e-01 -3.73942256e-01 2.20544338e-01
5.44589339e-03 9.85528052e-01 -9.15927947e-01 4.06000435e-01
5.86777449e-01 5.63690186e-01 -4.25258726e-01 1.20137465e+00
2.89837360e-01 1.32747030e+00 -1.79188713e-01 -3.15024965e-02
-2.33723432e-01 4.21541631e-02 9.36331809e-01 1.32777488e+00
3.62877131e-01 1.55223906e-01 5.03721200e-02 5.29147685e-01
4.52335566e-01 4.79719937e-01 -4.55084816e-02 -1.18059278e-01
3.86943161e-01 1.27424359e+00 -5.93461990e-01 -4.80335742e-01
-2.44906634e-01 8.51006746e-01 -4.61721808e-01 4.81144823e-02
-8.17920804e-01 -5.82834840e-01 3.55667114e-01 -1.73933014e-01
3.00990850e-01 2.17538297e-01 -1.04378909e-01 -9.80714619e-01
-2.79338628e-01 -7.93841660e-01 5.82859278e-01 -1.03826113e-01
-1.10219705e+00 5.42853415e-01 -2.69238204e-01 -1.31112659e+00
-2.96364903e-01 -2.23993689e-01 -7.18292236e-01 1.06923378e+00
-1.31271410e+00 -6.20271742e-01 -2.16575637e-01 4.78408724e-01
5.99031210e-01 -3.97051811e-01 1.37624276e+00 5.63801229e-01
-8.43122184e-01 4.80043709e-01 1.48784935e-01 3.74276817e-01
2.40473524e-01 -1.23126745e+00 -1.96351111e-02 5.49292505e-01
1.04978353e-01 5.10610521e-01 7.66224265e-01 -4.62091118e-01
-1.22486341e+00 -1.36670649e+00 1.21118963e+00 -2.05861375e-01
2.78658181e-01 2.02401206e-01 -7.86750078e-01 -1.25012212e-02
-1.42607763e-01 3.31552714e-01 1.28285265e+00 -1.83528662e-01
-7.47709945e-02 -2.02782497e-01 -1.24000943e+00 5.81621677e-02
1.03337556e-01 -4.52353299e-01 -5.70962250e-01 -1.52009614e-02
2.89191872e-01 -1.43255249e-01 -1.42604125e+00 3.64734262e-01
7.09920645e-01 -9.28870142e-01 1.23336136e+00 -5.55288076e-01
2.96068579e-01 -1.44542679e-01 -1.96345255e-01 -1.24136162e+00
-4.09894884e-01 -3.95710200e-01 -3.82838726e-01 1.20804119e+00
-8.19520727e-02 -4.76338863e-01 4.82355088e-01 1.42566130e-01
-2.65668537e-02 -7.76031315e-01 -1.10552824e+00 -4.63728845e-01
1.13079352e-02 -7.88792551e-01 2.37456083e-01 7.36886859e-01
1.26495585e-01 4.45290565e-01 -3.87025684e-01 2.70660341e-01
7.50956714e-01 -3.30966408e-03 4.75087702e-01 -1.51096523e+00
-5.21703839e-01 -4.64645654e-01 -5.82099497e-01 -3.93482417e-01
-1.52962178e-01 -9.23972428e-01 -1.85043171e-01 -1.11865258e+00
6.97369128e-02 -2.18716815e-01 -7.08281755e-01 -1.46443509e-02
-3.19549114e-01 1.96948841e-01 9.58704129e-02 4.30197157e-02
-1.39957428e-01 4.03548211e-01 9.07135546e-01 -6.52887076e-02
-4.60935771e-01 5.49714565e-01 -4.61483270e-01 6.30708694e-01
8.55844378e-01 -4.93133515e-01 -2.38499880e-01 4.40478176e-01
-5.58590829e-01 5.69363952e-01 5.60105860e-01 -1.37702298e+00
-1.78043604e-01 1.91350952e-01 5.53511381e-01 -5.25940716e-01
3.39260340e-01 -6.05545223e-01 2.83358663e-01 9.26711380e-01
-3.40065539e-01 -4.26871419e-01 9.80124101e-02 6.68891490e-01
-1.87996849e-01 2.64229830e-02 8.76148164e-01 1.99284196e-01
-1.47400469e-01 2.19927430e-02 -6.65328979e-01 -2.49825660e-02
8.40562940e-01 2.70053416e-01 1.96379825e-01 1.77545249e-02
-9.81930435e-01 -3.89388919e-01 -4.89969552e-01 2.34549299e-01
7.56549060e-01 -1.24133956e+00 -9.85460997e-01 4.54161495e-01
2.73479093e-02 -3.52225900e-01 7.51744688e-01 1.27999306e+00
-4.74548370e-01 5.91612637e-01 6.01893887e-02 -6.45119131e-01
-1.34803534e+00 4.27502513e-01 2.18750596e-01 -2.30842546e-01
-1.12269044e+00 7.62132227e-01 -2.64807940e-01 2.04921383e-02
1.80685177e-01 -3.06053519e-01 -8.63832474e-01 9.06392485e-02
8.03666890e-01 7.79182494e-01 4.03595231e-02 -8.73679936e-01
-4.84331399e-01 3.91108006e-01 2.63935715e-01 -1.85903516e-02
1.31460762e+00 4.56139678e-03 5.30038327e-02 4.19830650e-01
1.38811028e+00 -2.50865258e-02 -4.62266952e-01 -9.57756639e-02
-3.96114960e-02 -1.93707243e-01 4.97403651e-01 -5.59218049e-01
-1.06642091e+00 1.12898636e+00 1.26904118e+00 4.25691932e-01
1.02797151e+00 -1.01822831e-01 1.20440686e+00 8.71068686e-02
-3.06019485e-01 -8.62305164e-01 -2.16101885e-01 -1.40373390e-02
6.68682873e-01 -7.85416961e-01 -5.06437540e-01 -1.79841116e-01
-6.82434082e-01 1.21730936e+00 -1.50348529e-01 -3.33118767e-01
1.23466516e+00 -1.19652927e-01 2.27741495e-01 -3.86725813e-02
-4.36232746e-01 -7.45098069e-02 4.72889602e-01 6.41457796e-01
4.37360853e-01 3.26814502e-01 -4.02037382e-01 1.15683377e+00
-2.87705749e-01 -1.00628421e-01 1.00923136e-01 5.70325375e-01
-4.91451293e-01 -8.05539787e-01 -5.92804432e-01 7.45586574e-01
-8.54605973e-01 -4.88647856e-02 2.38529086e-01 1.42591342e-01
2.32175648e-01 1.29383540e+00 1.21292770e-02 -4.48966503e-01
1.93567649e-01 4.74812001e-01 2.04350576e-01 -4.53979254e-01
-4.61487293e-01 4.88532394e-01 -2.10512385e-01 -3.01426589e-01
-2.62293816e-01 -8.91457915e-01 -1.26768064e+00 2.60678828e-01
-1.14795327e-01 3.98081094e-01 7.48632312e-01 8.21948946e-01
3.67290378e-01 9.14614320e-01 9.21939015e-01 -5.07734478e-01
-7.07045496e-01 -1.06733119e+00 -6.82078183e-01 5.60675025e-01
4.26239967e-01 -3.90436113e-01 -4.49119717e-01 3.64313722e-01] | [14.355836868286133, 3.3724446296691895] |
0ab87949-5be8-4342-88e1-0269f1284852 | financial-news-annotation-by-weakly | null | null | https://aclanthology.org/2020.finnlp-1.1 | https://aclanthology.org/2020.finnlp-1.1.pdf | Financial News Annotation by Weakly-Supervised Hierarchical Multi-label Learning | null | ['Guangwei Shi', 'Jidong Lu', 'Jian Gao', 'Mengjun Ni', 'Chenyu Wang', 'Yuefeng Lin', 'Zhongchen Miao', 'Hang Jiang'] | null | null | null | null | finnlp-coling-2020-1 | ['news-annotation'] | ['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] | [-7.465348720550537, 3.877047061920166] |
79cfd587-208c-445a-89d1-ffae4420b874 | evaluating-community-detection-algorithms-for | 2007.08635 | null | https://arxiv.org/abs/2007.08635v1 | https://arxiv.org/pdf/2007.08635v1.pdf | Evaluating Community Detection Algorithms for Progressively Evolving Graphs | Many algorithms have been proposed in the last ten years for the discovery of dynamic communities. However, these methods are seldom compared between themselves. In this article, we propose a generator of dynamic graphs with planted evolving community structure, as a benchmark to compare and evaluate such algorithms. Unlike previously proposed benchmarks, it is able to specify any desired evolving community structure through a descriptive language, and then to generate the corresponding progressively evolving network. We empirically evaluate six existing algorithms for dynamic community detection in terms of instantaneous and longitudinal similarity with the planted ground truth, smoothness of dynamic partitions, and scalability. We notably observe different types of weaknesses depending on their approach to ensure smoothness, namely Glitches, Oversimplification and Identity loss. Although no method arises as a clear winner, we observe clear differences between methods, and we identified the fastest, those yielding the most smoothed or the most accurate solutions at each step. | ['Souaad Boudebza', 'Remy Cazabet', 'Giulio Rossetti'] | 2020-07-16 | null | null | null | null | ['dynamic-community-detection'] | ['graphs'] | [ 1.40892386e-01 1.31803066e-01 2.18052149e-01 2.66790777e-01
-2.00518236e-01 -1.09738696e+00 8.78653705e-01 4.04200137e-01
-1.07481107e-02 7.68738389e-01 1.62027441e-02 -2.05772191e-01
-5.90186417e-01 -7.79444575e-01 -1.80792034e-01 -7.39058018e-01
-8.53971303e-01 8.22858274e-01 6.47289574e-01 -2.84930855e-01
2.96734661e-01 2.07203373e-01 -1.41619468e+00 9.07094330e-02
1.17790437e+00 2.77239710e-01 -1.25317365e-01 6.77243948e-01
-1.76149234e-02 4.38826382e-01 -5.45665562e-01 -5.12411118e-01
3.15739870e-01 -5.32999635e-01 -8.95571828e-01 2.85377532e-01
3.57861631e-02 3.89944255e-01 -1.28025692e-02 1.01370084e+00
4.11197722e-01 -4.18105692e-01 7.63296247e-01 -1.57413709e+00
-4.69775826e-01 1.10199451e+00 -8.00028920e-01 2.93293178e-01
7.03549027e-01 3.09731513e-01 9.69385445e-01 -6.99782133e-01
1.36530626e+00 1.13123345e+00 1.03236341e+00 3.88541937e-01
-1.84012902e+00 -5.44677675e-01 1.89938247e-01 6.11239895e-02
-1.75506377e+00 -2.50733554e-01 6.85881197e-01 -8.97951305e-01
4.11838919e-01 3.99600297e-01 1.03979254e+00 1.21732700e+00
-4.89855818e-02 2.81915486e-01 1.30741549e+00 -3.00131470e-01
2.93333888e-01 1.50888026e-01 8.19687322e-02 4.30487365e-01
7.45762289e-01 -6.50458187e-02 -1.64418861e-01 -7.16338515e-01
4.81310129e-01 -3.48820895e-01 -4.11585599e-01 -9.64601219e-01
-1.21672130e+00 6.59740329e-01 1.90594465e-01 6.31230772e-01
-1.29125938e-01 -1.44212946e-01 3.89349580e-01 4.24220890e-01
4.19733971e-01 3.33420902e-01 7.88597390e-03 -1.52893225e-03
-1.13001621e+00 4.21237171e-01 1.07999551e+00 8.60423625e-01
6.32996798e-01 -1.40644848e-01 -7.85128847e-02 2.78224558e-01
1.33651972e-01 1.37381658e-01 9.25078839e-02 -6.59130931e-01
1.77555993e-01 8.31866860e-01 1.42047731e-02 -1.40784168e+00
-3.34385276e-01 -5.26340544e-01 -1.20298183e+00 7.50976801e-02
5.23507297e-01 -3.13156247e-02 -5.65585554e-01 1.48482132e+00
3.37406188e-01 3.29459965e-01 9.14851949e-02 4.81693745e-01
4.16902274e-01 2.39407152e-01 -3.11964095e-01 -5.30744195e-01
7.67385006e-01 -6.74310744e-01 -3.73017460e-01 1.56335130e-01
1.49491563e-01 -6.70554340e-01 5.61347067e-01 3.01416397e-01
-1.04064107e+00 -7.53831565e-02 -1.03720665e+00 7.99993873e-01
-1.92021281e-01 -4.64100063e-01 4.80123103e-01 7.46935964e-01
-1.65005398e+00 8.70537877e-01 -7.08290756e-01 -6.94362164e-01
-1.09575670e-02 2.50462919e-01 -2.62814671e-01 2.36366868e-01
-8.66968334e-01 5.80947161e-01 5.65382421e-01 -3.28053534e-02
-9.17355359e-01 -5.26287198e-01 -3.32757354e-01 -7.77121354e-03
4.25065100e-01 -8.36046755e-01 6.48620367e-01 -1.15003121e+00
-1.01452434e+00 1.02925527e+00 2.37282202e-01 -7.01352894e-01
1.01185393e+00 6.67580009e-01 -4.36926812e-01 2.49862932e-02
-2.98079625e-02 3.55185777e-01 6.55870199e-01 -1.74700153e+00
-3.67664367e-01 1.51694089e-01 6.26186281e-02 -1.43467903e-01
-2.06833243e-01 -1.69878881e-02 -1.98724523e-01 -5.76992154e-01
2.73049891e-01 -8.96097064e-01 -4.70769823e-01 -2.36043647e-01
-5.20141244e-01 -6.59356192e-02 5.82657337e-01 -3.93616408e-01
1.60658622e+00 -1.93062043e+00 3.66150320e-01 6.42624319e-01
8.15974474e-01 -1.35461897e-01 -1.11795217e-01 8.02694738e-01
-1.73842371e-01 6.60167396e-01 -5.71703613e-01 -1.41782999e-01
-1.02180667e-01 4.84920945e-03 2.48733070e-03 5.42483389e-01
1.69282794e-01 5.67136288e-01 -1.16703403e+00 -4.70417589e-01
-3.76487046e-01 3.14539135e-01 -5.13877451e-01 -2.85719991e-01
-6.59391731e-02 3.17098647e-01 -5.63269556e-02 4.52346206e-01
7.48136044e-01 -4.29847211e-01 8.59484196e-01 2.61524171e-01
-2.54992783e-01 -1.25774890e-01 -1.68653727e+00 1.19560242e+00
2.81431049e-01 6.53179348e-01 2.36533463e-01 -7.48127639e-01
1.10226047e+00 5.05364299e-01 7.73255646e-01 -7.97414780e-02
-1.60940647e-01 3.26830506e-01 4.66208071e-01 -1.93307504e-01
5.57459950e-01 1.78467423e-01 1.90685362e-01 7.61543393e-01
-2.16472223e-01 1.55041620e-01 7.89682567e-01 5.50125301e-01
1.57551360e+00 -3.58643681e-01 3.33273679e-01 -8.65411758e-01
4.67045486e-01 1.74404740e-01 3.18262249e-01 6.63748443e-01
-1.10532612e-01 7.68416286e-01 8.29549730e-01 -3.45162272e-01
-1.28714263e+00 -1.31137645e+00 2.14137375e-01 4.65081602e-01
2.09461525e-01 -5.10149181e-01 -7.76094019e-01 -3.72636527e-01
-3.41239907e-02 3.21811959e-02 -6.73226357e-01 2.05705360e-01
-4.02825922e-01 -6.98547184e-01 5.12712181e-01 -8.77782702e-02
1.77401860e-04 -1.06904900e+00 -3.92507643e-01 2.88176000e-01
-3.70687991e-01 -8.19096029e-01 -4.19760466e-01 -1.52180925e-01
-1.00935221e+00 -1.46806240e+00 -6.68499589e-01 -7.59092391e-01
8.54802668e-01 2.65833765e-01 1.41326213e+00 6.24723375e-01
-3.06925029e-01 3.91748428e-01 -3.74214709e-01 1.75092220e-01
-8.16305757e-01 3.50212306e-01 8.10497999e-02 2.02236116e-01
-1.14628553e-01 -1.06055593e+00 -5.06079197e-01 2.19875649e-01
-8.35371971e-01 -6.66968971e-02 2.80038565e-01 6.15760446e-01
1.21555433e-01 2.48967245e-01 1.61745816e-01 -9.04911876e-01
1.13946068e+00 -6.68024540e-01 -5.34638405e-01 3.43167633e-01
-9.88914371e-01 1.26665071e-01 3.17415446e-01 -5.23483217e-01
-4.95509028e-01 -2.06112102e-01 3.72045726e-01 -2.22127587e-01
8.43267366e-02 8.17420661e-01 4.38479424e-01 -4.07127142e-02
9.00955617e-01 2.93766528e-01 1.03104293e-01 -2.42230326e-01
1.99534670e-01 2.13229567e-01 4.43490297e-01 -7.13646829e-01
1.26377976e+00 6.28418207e-01 -4.20276791e-01 -4.97465134e-01
5.53812925e-03 -5.34117043e-01 -9.67657626e-01 -5.48265517e-01
3.29001099e-01 -6.04981661e-01 -4.80723560e-01 5.84088683e-01
-1.10945320e+00 -2.17512220e-01 -2.94286400e-01 -1.57258511e-01
-3.60240966e-01 6.68893993e-01 -4.83732492e-01 -8.94507408e-01
-4.57048953e-01 -7.63340652e-01 4.93654191e-01 -2.42106561e-02
-3.46260190e-01 -1.11583507e+00 5.59737742e-01 -3.12013656e-01
5.59802592e-01 9.37390566e-01 8.04286718e-01 -3.65183443e-01
-6.29942596e-01 -4.30371836e-02 1.83465220e-02 -3.53886336e-01
1.10433586e-01 9.15534437e-01 -6.16216242e-01 -7.59728789e-01
-5.40471733e-01 3.55114132e-01 6.17273331e-01 2.07667440e-01
5.11443317e-01 -3.78816634e-01 -6.74185038e-01 3.55835974e-01
1.66912198e+00 9.19590071e-02 8.77554536e-01 3.87589008e-01
2.32510388e-01 7.42771387e-01 1.27514020e-01 3.80241066e-01
2.86603570e-01 5.11085391e-01 3.91261816e-01 2.30732327e-03
6.30423650e-02 -1.25984959e-02 3.94508958e-01 1.07502580e+00
-5.52900136e-01 -2.78328627e-01 -1.42692173e+00 7.21982360e-01
-1.86869955e+00 -1.43661010e+00 -4.88088965e-01 2.27452564e+00
7.70886004e-01 5.97658515e-01 6.58159614e-01 2.51556516e-01
1.18307281e+00 -9.60421711e-02 -9.90296006e-02 2.17897631e-02
-2.71264613e-01 6.28565401e-02 2.16628268e-01 4.80779320e-01
-7.72090435e-01 5.44303834e-01 7.18238306e+00 5.10968983e-01
-1.00959277e+00 9.68970172e-03 4.54184264e-01 1.16424434e-01
-6.99669123e-01 3.47546190e-01 -3.54184180e-01 2.80430913e-01
6.62692606e-01 -8.77223313e-01 4.46518123e-01 5.92971325e-01
2.20558837e-01 1.59572616e-01 -8.56693506e-01 5.06302536e-01
-2.40247563e-01 -1.43324840e+00 -1.30147979e-01 2.35730544e-01
1.02043259e+00 -6.02585189e-02 -3.72826338e-01 -5.91759011e-02
7.68924117e-01 -9.91079092e-01 9.54089344e-01 6.82002902e-01
5.15781105e-01 -6.47814870e-01 3.95455927e-01 3.03067803e-01
-1.66249144e+00 -5.79889044e-02 -8.80818535e-03 -1.75384566e-01
6.77144378e-02 6.85948014e-01 -7.99469173e-01 9.12210345e-01
7.07473099e-01 6.25663102e-01 -8.84542406e-01 1.54651117e+00
2.09437415e-01 6.70530140e-01 -4.24576312e-01 -1.85560316e-01
7.27673396e-02 -5.40015399e-01 1.02904642e+00 1.23271012e+00
4.83696103e-01 -3.59938771e-01 3.81511033e-01 1.00584948e+00
3.69812429e-01 1.42541915e-01 -6.08926237e-01 -1.55109957e-01
5.55827796e-01 1.28141057e+00 -1.45762825e+00 -2.63963997e-01
5.45475148e-02 7.21540272e-01 2.73869038e-01 3.36216450e-01
-8.29873085e-01 -1.68680586e-02 5.70767283e-01 7.02274144e-01
2.78683364e-01 -4.78166610e-01 -1.29797503e-01 -1.08826184e+00
9.67605934e-02 -1.05132163e+00 4.97055680e-01 -5.09919882e-01
-1.48529100e+00 9.21142161e-01 1.59618199e-01 -1.20548856e+00
-2.20161319e-01 -7.99364746e-02 -8.39872658e-01 5.16487241e-01
-8.37246120e-01 -8.53408754e-01 -4.97295260e-01 3.87306601e-01
1.64290160e-01 2.48140823e-02 6.05858982e-01 3.78144801e-01
-4.11507875e-01 3.34028006e-01 2.36706272e-01 -2.72219209e-03
5.48654199e-01 -1.45131719e+00 6.46442711e-01 1.07385612e+00
1.16055585e-01 7.01641083e-01 1.11820531e+00 -8.13303232e-01
-8.06591630e-01 -8.26472521e-01 7.87604213e-01 -4.96490240e-01
1.10915732e+00 -5.59018791e-01 -9.89869237e-01 3.64507884e-01
3.19159985e-01 -2.86672860e-01 6.23997003e-02 2.10208997e-01
-3.22996467e-01 8.42212737e-02 -9.10988748e-01 7.94387639e-01
1.48510981e+00 -1.87613055e-01 -3.38810012e-02 1.31345809e-01
5.11401296e-01 -1.24621421e-01 -9.13701832e-01 3.35537791e-01
5.64568341e-01 -1.36263144e+00 8.10600817e-01 -2.15241238e-01
2.62371421e-01 -5.41872799e-01 4.47166681e-01 -1.19575226e+00
-7.08662689e-01 -9.54854310e-01 3.24148089e-02 1.56734419e+00
6.51430666e-01 -7.37768173e-01 8.11996520e-01 -1.44796103e-01
2.35542178e-01 -5.03029943e-01 -7.51056790e-01 -9.61604476e-01
-5.26549108e-02 9.05486718e-02 5.36650538e-01 1.35212517e+00
1.11426018e-01 2.95974016e-01 -1.35493025e-01 1.55064734e-02
7.85121202e-01 2.93646425e-01 9.19475436e-01 -1.70996225e+00
-3.28041881e-01 -1.03443718e+00 -6.19424522e-01 -2.59934574e-01
-3.32129091e-01 -9.26209092e-01 -1.41678303e-01 -1.54565382e+00
6.99815154e-01 -5.50088525e-01 7.18592806e-03 3.94993424e-02
-1.79767847e-01 1.75703645e-01 1.60270929e-01 2.64897645e-01
-5.51379979e-01 -7.05394447e-02 1.02260184e+00 -3.76021236e-01
-3.89019370e-01 1.65285602e-01 -6.17795467e-01 3.79830718e-01
7.06896245e-01 -7.01738179e-01 -4.08635437e-01 1.82373926e-01
5.97355187e-01 -1.25360996e-01 4.40309763e-01 -1.11061120e+00
3.47250342e-01 -9.39629599e-02 -1.46104202e-01 -4.38860774e-01
-3.42432171e-01 -5.90637147e-01 1.23094738e+00 9.28914249e-01
2.50291429e-03 5.06247461e-01 -3.28269601e-02 7.15578854e-01
-1.44400343e-01 -2.46179268e-01 6.38942599e-01 -2.67489851e-01
-5.68471491e-01 2.88609475e-01 -5.36510229e-01 3.19760919e-01
1.19464898e+00 -6.34902358e-01 -2.61182427e-01 -3.97430509e-01
-9.78546798e-01 3.00305158e-01 9.76409078e-01 3.30432534e-01
2.30100602e-01 -1.12112796e+00 -1.11329103e+00 -2.84684628e-01
8.70576426e-02 -2.34442756e-01 -2.89599597e-02 9.93889213e-01
-7.56466508e-01 -1.04827650e-01 -1.81743264e-01 -9.03876185e-01
-1.59989583e+00 6.83021367e-01 5.13437688e-01 -7.83900142e-01
-6.17708802e-01 4.09600616e-01 -3.08069259e-01 -3.73470753e-01
1.77607968e-01 -2.15017989e-01 -1.87383831e-01 2.60195524e-01
6.53104708e-02 3.57699275e-01 -6.55656308e-02 -3.80546987e-01
-3.51336330e-01 4.92869079e-01 3.46724778e-01 -7.61377206e-03
1.31437302e+00 -1.82840779e-01 -4.96850193e-01 4.37543422e-01
4.85782176e-01 1.37605831e-01 -1.13517904e+00 4.22319025e-02
6.42828703e-01 -2.32356414e-01 -7.39924788e-01 -4.17814046e-01
-9.27669823e-01 2.15197563e-01 4.15227652e-01 1.20029080e+00
9.99849796e-01 -8.74781013e-02 5.63825713e-04 -7.67186955e-02
5.90643466e-01 -7.13906825e-01 1.67112634e-01 5.18799841e-01
8.10872555e-01 -7.60899484e-01 2.02352032e-01 -8.23006153e-01
-3.59145284e-01 1.12052536e+00 3.24630260e-01 -2.25849360e-01
7.08721995e-01 4.28170055e-01 -1.32191226e-01 -4.90531385e-01
-9.45775151e-01 -2.57308424e-01 1.22113667e-01 9.94408190e-01
3.13466221e-01 -6.00529313e-02 -5.81806064e-01 -1.16542587e-02
-3.27335268e-01 -3.90311211e-01 8.50581884e-01 6.43603981e-01
-3.53655487e-01 -1.11043382e+00 -4.79098946e-01 3.39008063e-01
3.01444728e-04 6.18544221e-03 -8.73182416e-01 1.06405199e+00
1.07955292e-01 8.27967882e-01 4.25962359e-02 -4.11751628e-01
2.26874590e-01 -1.35216163e-02 4.71936941e-01 -5.13274312e-01
-8.36331725e-01 -1.53144643e-01 2.04116568e-01 -5.96799068e-02
-6.39206290e-01 -8.42605531e-01 -7.42036939e-01 -8.07395041e-01
-6.25122011e-01 4.61949646e-01 8.25783312e-02 5.16121924e-01
2.82938451e-01 4.81042504e-01 6.51309907e-01 -7.62959242e-01
-3.54063421e-01 -9.99198377e-01 -6.12624884e-01 6.12683654e-01
-7.03735501e-02 -6.59501910e-01 -5.80408335e-01 2.50412166e-01] | [6.973922252655029, 5.282011985778809] |
9ea46082-2b05-4a06-a12c-8d5b768f588f | patch-based-object-centric-transformers-for | 2206.04003 | null | https://arxiv.org/abs/2206.04003v2 | https://arxiv.org/pdf/2206.04003v2.pdf | Patch-based Object-centric Transformers for Efficient Video Generation | In this work, we present Patch-based Object-centric Video Transformer (POVT), a novel region-based video generation architecture that leverages object-centric information to efficiently model temporal dynamics in videos. We build upon prior work in video prediction via an autoregressive transformer over the discrete latent space of compressed videos, with an added modification to model object-centric information via bounding boxes. Due to better compressibility of object-centric representations, we can improve training efficiency by allowing the model to only access object information for longer horizon temporal information. When evaluated on various difficult object-centric datasets, our method achieves better or equal performance to other video generation models, while remaining computationally more efficient and scalable. In addition, we show that our method is able to perform object-centric controllability through bounding box manipulation, which may aid downstream tasks such as video editing, or visual planning. Samples are available at https://sites.google.com/view/povt-public | ['Pieter Abbeel', 'Stephen James', 'Ryo Okumura', 'Wilson Yan'] | 2022-06-08 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 5.06565906e-02 1.53580233e-01 -3.73384506e-01 5.91104515e-02
-6.78649306e-01 -5.81748188e-01 7.52827883e-01 -2.52454668e-01
2.16674238e-01 5.32407582e-01 7.43934929e-01 -1.30509257e-01
4.68468703e-02 -7.13131309e-01 -1.17338634e+00 -3.95286947e-01
-1.70791268e-01 4.78538483e-01 1.61162108e-01 -1.45546615e-01
2.44875371e-01 5.10984242e-01 -1.44467628e+00 5.79860628e-01
5.42541564e-01 9.22308505e-01 4.10955608e-01 8.71842086e-01
4.56890106e-01 1.25037074e+00 -6.20011166e-02 1.80420876e-01
4.48095888e-01 -1.47660300e-01 -6.44918323e-01 3.92163038e-01
6.09674633e-01 -7.90542543e-01 -8.40496123e-01 2.32548594e-01
1.85323000e-01 4.22834963e-01 6.00882113e-01 -1.26184142e+00
-6.41373754e-01 4.24443126e-01 -1.80263251e-01 5.00544459e-02
4.53814000e-01 5.02858937e-01 1.08154058e+00 -8.91969264e-01
1.16865790e+00 1.24719322e+00 3.27338994e-01 5.70511162e-01
-1.41191471e+00 -3.58732253e-01 4.63972002e-01 1.05828717e-01
-1.17372918e+00 -7.07869053e-01 7.47011065e-01 -7.52024949e-01
1.18816209e+00 2.09557772e-01 1.08830380e+00 1.06994236e+00
3.95809084e-01 1.08459282e+00 5.39628625e-01 2.21535508e-02
3.74060944e-02 -2.72699863e-01 -6.32480860e-01 8.60360980e-01
-3.41085166e-01 2.88377583e-01 -7.46071935e-01 -1.33698341e-02
1.26362967e+00 2.76433796e-01 -4.22647357e-01 -9.03917074e-01
-1.56964624e+00 8.30337226e-01 4.31689173e-01 -1.34961188e-01
-4.31915045e-01 8.07504833e-01 1.58414483e-01 9.41954032e-02
5.73044002e-01 6.45984590e-01 -3.43821943e-01 -3.27200860e-01
-1.14331639e+00 6.87206030e-01 4.45020765e-01 1.52781034e+00
5.18268466e-01 4.48865741e-01 -7.24944115e-01 3.34687769e-01
1.10658139e-01 3.89325351e-01 1.89158604e-01 -1.61354113e+00
4.48804289e-01 2.55748123e-01 3.11801225e-01 -9.21311080e-01
-3.91533487e-02 -1.72699392e-01 -3.97937208e-01 -5.88939041e-02
-9.52515304e-02 -2.29856954e-03 -1.10352159e+00 1.57799697e+00
2.62606055e-01 4.17745471e-01 -2.27251932e-01 8.91586006e-01
5.56741655e-01 1.13663852e+00 -4.98204418e-02 -1.46479115e-01
9.97555077e-01 -1.40384340e+00 -4.60259199e-01 2.03652099e-01
5.98957360e-01 -6.18785799e-01 8.93688560e-01 4.08024132e-01
-1.34578121e+00 -3.66194695e-01 -6.01453900e-01 -2.89331406e-01
9.68853384e-02 1.67253718e-01 8.29964042e-01 -6.78375363e-02
-1.38860440e+00 7.12632418e-01 -1.23043346e+00 -1.49343595e-01
6.48935080e-01 2.64282078e-01 -3.97992462e-01 -2.15421289e-01
-6.52518868e-01 4.58148807e-01 2.89817542e-01 -2.99391925e-01
-1.64968002e+00 -1.21965718e+00 -1.04555988e+00 5.99109717e-02
5.57557106e-01 -9.85076845e-01 1.47552538e+00 -7.00770736e-01
-1.61447370e+00 3.44221532e-01 -3.77029717e-01 -7.06295252e-01
5.17392159e-01 -4.47422355e-01 2.52783448e-01 4.89772201e-01
2.19769273e-02 1.23813736e+00 1.08474696e+00 -1.08309269e+00
-3.48133385e-01 1.95342228e-01 3.37764651e-01 3.06910604e-01
-2.52581865e-01 -1.37211367e-01 -8.01757276e-01 -1.10514617e+00
-1.29428580e-01 -1.13849390e+00 -3.34883034e-01 4.72354561e-01
-3.19944561e-01 -9.73559394e-02 1.09004986e+00 -7.13198960e-01
1.05017555e+00 -2.01417232e+00 6.67130291e-01 -1.99859560e-01
3.67422670e-01 -8.32601637e-02 -2.04585433e-01 7.25471199e-01
4.88045961e-02 1.40433431e-01 1.37933314e-01 -4.77477252e-01
-1.97188318e-01 1.08103886e-01 -5.59305012e-01 1.67402551e-01
3.86877686e-01 1.08860242e+00 -9.79530513e-01 -5.94344199e-01
3.87618572e-01 5.75594008e-01 -1.37676728e+00 1.85958415e-01
-7.66640902e-01 7.55156100e-01 -6.41928375e-01 7.37707555e-01
1.68203220e-01 -3.34726959e-01 1.72345750e-02 -3.78117502e-01
-1.73376232e-01 2.42028520e-01 -6.50429666e-01 1.89526522e+00
-6.32809222e-01 8.36735725e-01 2.43796082e-03 -8.04646909e-01
4.41686302e-01 5.24969041e-01 9.82755184e-01 -3.16812038e-01
-1.11868009e-01 -2.70978868e-01 -3.96568388e-01 -2.75489599e-01
7.85931706e-01 1.77608356e-01 1.80646226e-01 4.40526247e-01
8.48092735e-02 -4.06569749e-01 2.51879901e-01 5.94602644e-01
1.12347639e+00 7.77024090e-01 -1.91384647e-02 -1.28734395e-01
-1.24696150e-01 2.27956101e-01 4.81962770e-01 6.00602388e-01
1.68214515e-01 7.72869647e-01 4.42838252e-01 -3.80370468e-01
-1.39802670e+00 -1.12746119e+00 1.42121091e-01 9.52955961e-01
-1.27232494e-02 -8.09010029e-01 -3.81460458e-01 -3.35990995e-01
1.29064023e-01 7.46217847e-01 -4.92297739e-01 -5.95686994e-02
-8.18191171e-01 1.05400234e-01 4.69069444e-02 7.44528770e-01
4.71155643e-02 -8.44245017e-01 -6.59401834e-01 3.83723527e-01
-2.70514756e-01 -1.12793839e+00 -8.94508958e-01 -2.41623715e-01
-1.24374926e+00 -7.75538862e-01 -8.66839111e-01 -4.25422281e-01
6.61450684e-01 4.38173741e-01 1.05190873e+00 -4.28378470e-02
-3.22534084e-01 9.22104836e-01 -4.60540861e-01 -1.08648330e-01
-2.22323209e-01 -1.31882995e-01 -1.07744159e-02 -1.67789400e-01
-5.81898272e-01 -6.26339495e-01 -7.85441399e-01 2.83078760e-01
-8.25099111e-01 6.93688810e-01 1.44245222e-01 8.14698040e-01
7.85220146e-01 -2.33711377e-01 -3.10487095e-02 -5.64344585e-01
8.82333666e-02 -5.16986072e-01 -7.61615932e-01 -1.19837023e-01
-3.32899719e-01 1.54268518e-01 5.47201097e-01 -5.30463696e-01
-1.03403223e+00 1.40280455e-01 4.14323330e-01 -1.21488237e+00
2.86043614e-01 3.36459786e-01 1.19206071e-01 7.77994916e-02
3.15298855e-01 2.99096793e-01 6.93199486e-02 -2.90212780e-01
5.71379840e-01 -6.65517673e-02 2.53643423e-01 -8.15007448e-01
7.90969074e-01 6.35708332e-01 3.80485132e-02 -5.16500592e-01
-5.27495921e-01 -2.02138320e-01 -2.50665098e-01 -3.89295310e-01
8.40605855e-01 -1.28617609e+00 -6.65582895e-01 1.85847521e-01
-1.01517034e+00 -8.66659641e-01 -3.44204485e-01 4.57080930e-01
-1.20530808e+00 6.91858083e-02 -8.42252374e-01 -5.24620473e-01
-1.60947204e-01 -1.20906675e+00 1.64191210e+00 -1.64103419e-01
-1.65759996e-01 -8.87568831e-01 -1.53874487e-01 4.40026522e-01
3.73093873e-01 3.40281218e-01 7.36596704e-01 -3.47102024e-02
-1.56792974e+00 7.37306401e-02 1.67313695e-01 4.47341129e-02
-6.53328225e-02 3.08132410e-01 -4.72150177e-01 -5.41095197e-01
-4.55439925e-01 -4.02979910e-01 9.62686896e-01 5.31384110e-01
1.38868880e+00 -6.87751949e-01 -5.30134499e-01 9.01004553e-01
1.20294666e+00 1.32725999e-01 5.30988872e-01 7.53015131e-02
9.10893083e-01 3.55556101e-01 1.02362919e+00 7.82642365e-01
4.67520028e-01 1.09868288e+00 5.11049211e-01 1.64264753e-01
-3.06594759e-01 -7.26168931e-01 5.14288068e-01 6.34969354e-01
-2.94354498e-01 -5.18349648e-01 -7.59063780e-01 7.14259446e-01
-2.02439737e+00 -1.17965031e+00 2.74759203e-01 1.83990884e+00
6.17416263e-01 -1.85011715e-01 1.99072480e-01 -4.11336124e-01
3.67508084e-01 3.82338107e-01 -5.53067207e-01 -1.27427489e-01
1.59535721e-01 -3.99736455e-03 5.82167864e-01 5.16382694e-01
-1.15657878e+00 1.39046192e+00 6.29851198e+00 6.93848312e-01
-1.06770718e+00 1.71061367e-01 5.76355457e-01 -8.44401181e-01
-6.07413292e-01 2.17215315e-01 -6.63271725e-01 3.33531708e-01
6.86097085e-01 -3.06242228e-01 6.06688380e-01 8.78321648e-01
7.49981999e-01 5.03397770e-02 -1.33741856e+00 8.50541890e-01
-3.92885804e-02 -2.02873540e+00 4.03835684e-01 2.42385447e-01
1.00101054e+00 -3.63378823e-02 2.52011240e-01 2.51608968e-01
3.19351017e-01 -8.94855559e-01 1.10540104e+00 6.88704371e-01
8.64545703e-01 -5.07145822e-01 -1.21479101e-01 1.96434826e-01
-1.34936857e+00 -2.09974438e-01 -1.91252172e-01 2.95145139e-02
4.50131536e-01 2.06674114e-01 -8.61582458e-01 3.50645155e-01
6.76065207e-01 1.22845280e+00 -1.15335874e-01 8.82678211e-01
5.57754748e-02 6.04439914e-01 -3.99311393e-01 3.79743487e-01
2.83930331e-01 -3.84775810e-02 8.06079566e-01 7.96507299e-01
5.13489604e-01 2.74962634e-01 4.78864491e-01 9.78469849e-01
2.05472223e-02 -1.81536615e-01 -9.20022130e-01 -3.25354606e-01
4.18718249e-01 9.34119701e-01 -4.96780634e-01 -4.61895287e-01
-2.03658700e-01 1.06080008e+00 2.01552585e-01 5.15214264e-01
-1.09563863e+00 2.22380728e-01 7.29993045e-01 4.37154502e-01
7.57717311e-01 -6.89486504e-01 1.33638278e-01 -1.44239640e+00
8.77961796e-03 -7.53630817e-01 -2.51105819e-02 -1.18015742e+00
-5.87739587e-01 2.86108941e-01 3.41416746e-01 -1.53128684e+00
-5.12898266e-01 -5.98333836e-01 -3.86253893e-01 3.84973675e-01
-1.04775941e+00 -1.48726356e+00 -1.91280603e-01 7.15216517e-01
9.34434056e-01 -1.55599385e-01 4.82999623e-01 1.31414771e-01
-2.90002823e-01 3.92237931e-01 2.42859032e-02 -1.37468204e-01
5.92343867e-01 -9.67821538e-01 4.60710943e-01 7.65703678e-01
1.00592285e-01 5.46298504e-01 6.60132766e-01 -8.97043705e-01
-1.92633128e+00 -1.46605909e+00 3.63121480e-01 -6.37388945e-01
5.42877257e-01 -3.54773581e-01 -2.77817070e-01 1.20668566e+00
-4.28119004e-02 9.46936682e-02 2.26889297e-01 -1.63256332e-01
-2.76164263e-01 8.21384043e-03 -7.49529660e-01 8.91627967e-01
1.43013573e+00 -4.15580839e-01 -8.63211676e-02 6.51867270e-01
1.21966136e+00 -7.88271606e-01 -1.05256128e+00 3.60457689e-01
5.76187193e-01 -6.48380578e-01 1.15379083e+00 -6.34775639e-01
9.09570277e-01 -3.63096714e-01 -2.30369344e-01 -1.20647824e+00
-5.57192564e-01 -9.09962654e-01 -6.73881173e-01 8.32393646e-01
1.80144250e-01 -3.11034471e-01 1.06383991e+00 4.80105519e-01
-4.86062229e-01 -1.06622148e+00 -6.46837115e-01 -8.35923910e-01
-2.34408244e-01 -3.40342164e-01 4.19141501e-01 5.84324241e-01
-1.49829924e-01 -2.03909338e-01 -8.85810912e-01 1.08609028e-01
3.63956630e-01 2.96994090e-01 8.90708447e-01 -5.10985255e-01
-6.21487796e-01 -2.12425426e-01 -4.18333471e-01 -1.52493441e+00
7.49770999e-02 -7.68847287e-01 -2.42293123e-02 -1.51644862e+00
1.84597954e-01 -3.28199565e-01 8.15595910e-02 6.24186039e-01
2.72157788e-01 2.55147755e-01 6.86059237e-01 3.38615030e-01
-7.26276636e-01 1.12213767e+00 1.65355015e+00 -1.62478298e-01
-2.10895970e-01 -2.21429676e-01 -3.54039222e-01 4.66806889e-01
6.94873333e-01 -1.84855446e-01 -7.85426855e-01 -7.38468707e-01
-8.91788527e-02 5.56066990e-01 4.92766887e-01 -9.66526866e-01
-7.20782205e-03 -4.56210107e-01 3.10393453e-01 -5.40905476e-01
8.67704630e-01 -6.21876180e-01 3.84018242e-01 3.24071914e-01
-4.42924052e-01 1.34610012e-01 1.37482569e-01 7.16498852e-01
-6.34034649e-02 3.69812697e-01 4.86792028e-01 -2.18990192e-01
-6.63843393e-01 9.29093659e-01 -6.40078068e-01 -1.69403970e-01
1.26734293e+00 -2.38158986e-01 -1.86356470e-01 -7.42308378e-01
-8.05669606e-01 3.52385908e-01 8.19168389e-01 6.14497483e-01
8.33272815e-01 -1.41752410e+00 -5.45770824e-01 1.84602011e-03
1.28857195e-01 3.77895534e-02 4.05798972e-01 8.03687871e-01
-7.64750779e-01 6.99593782e-01 -1.30329967e-01 -8.99365425e-01
-1.15419161e+00 7.12335646e-01 1.83006406e-01 -1.46150097e-01
-8.30740511e-01 7.58349419e-01 6.30057335e-01 -1.45744815e-01
6.30556196e-02 -4.12463009e-01 2.18994260e-01 -5.17624676e-01
3.00768107e-01 3.26536924e-01 -4.15381491e-01 -4.88107711e-01
-6.83123469e-02 4.85499144e-01 -2.49377608e-01 -2.22693026e-01
1.30739474e+00 3.14942151e-02 1.34809211e-01 1.60277531e-01
1.02381885e+00 -1.58527717e-01 -1.91520119e+00 6.05198182e-02
-5.68111658e-01 -6.72235727e-01 1.45051211e-01 -3.74351263e-01
-1.17401278e+00 6.04002476e-01 1.29893973e-01 -3.90235811e-01
1.01532578e+00 5.78180142e-02 7.90871739e-01 4.82574552e-01
7.14625478e-01 -8.19908500e-01 4.31413949e-01 4.69573528e-01
1.24509823e+00 -8.40825260e-01 1.39042944e-01 -5.99035501e-01
-6.99152231e-01 9.00089800e-01 6.99233055e-01 -2.95267940e-01
6.42726362e-01 2.36296147e-01 -5.00280201e-01 -1.26048282e-01
-1.45881069e+00 1.96485236e-01 2.86958635e-01 5.77943146e-01
4.19913828e-01 -4.00800817e-02 1.52959958e-01 7.84052163e-03
-1.67402953e-01 1.91614747e-01 5.39740741e-01 1.13058865e+00
-2.75150776e-01 -1.05814898e+00 -1.12553574e-01 5.76848209e-01
-1.79249823e-01 -2.65750736e-01 1.50689721e-01 7.00349987e-01
-1.93274260e-01 7.00189710e-01 3.09065789e-01 -2.90472358e-01
-1.17221391e-02 -3.78589243e-01 7.00074434e-01 -8.21720183e-01
-1.56718090e-01 2.25439504e-01 1.83578610e-01 -1.26945746e+00
-2.94561356e-01 -8.54407132e-01 -1.05336821e+00 -3.77458453e-01
6.74657226e-02 -1.31961405e-01 2.60535121e-01 5.24652958e-01
8.67835701e-01 4.97040689e-01 5.43126583e-01 -1.44252408e+00
-2.42767379e-01 -5.71518421e-01 -1.76029772e-01 2.76874155e-01
4.01194930e-01 -8.64572108e-01 1.29347488e-01 5.50839126e-01] | [10.64059066772461, -0.46211811900138855] |
fc47d5af-31d9-42fa-bc82-ff475f32cfee | bevscope-enhancing-self-supervised-depth | 2306.11598 | null | https://arxiv.org/abs/2306.11598v1 | https://arxiv.org/pdf/2306.11598v1.pdf | BEVScope: Enhancing Self-Supervised Depth Estimation Leveraging Bird's-Eye-View in Dynamic Scenarios | Depth estimation is a cornerstone of perception in autonomous driving and robotic systems. The considerable cost and relatively sparse data acquisition of LiDAR systems have led to the exploration of cost-effective alternatives, notably, self-supervised depth estimation. Nevertheless, current self-supervised depth estimation methods grapple with several limitations: (1) the failure to adequately leverage informative multi-camera views. (2) the limited capacity to handle dynamic objects effectively. To address these challenges, we present BEVScope, an innovative approach to self-supervised depth estimation that harnesses Bird's-Eye-View (BEV) features. Concurrently, we propose an adaptive loss function, specifically designed to mitigate the complexities associated with moving objects. Empirical evaluations conducted on the Nuscenes dataset validate our approach, demonstrating competitive performance. Code will be released at https://github.com/myc634/BEVScope. | ['Hang Zhao', 'Tianbao Zhang', 'Ruowen Zhao', 'Yucheng Mao'] | 2023-06-20 | null | null | null | null | ['depth-estimation'] | ['computer-vision'] | [ 1.61873579e-01 -5.34612052e-02 -2.77180701e-01 -6.20886326e-01
-7.94294715e-01 -4.87772018e-01 6.36240661e-01 -7.14385659e-02
-5.37556112e-01 5.98984420e-01 -1.75638739e-02 -1.14921279e-01
-6.95378408e-02 -6.59833729e-01 -5.47320127e-01 -5.28883100e-01
1.04278000e-02 2.13768423e-01 5.34923136e-01 -1.61951128e-02
5.56708336e-01 6.50718391e-01 -2.18254137e+00 -2.60745227e-01
7.86576450e-01 1.01288486e+00 5.41670501e-01 7.19745040e-01
8.70500281e-02 5.20000935e-01 2.45942008e-02 4.24081786e-03
4.02165264e-01 -9.74711478e-02 -3.70724499e-01 1.19460978e-01
7.22964287e-01 -6.73479259e-01 -4.11060631e-01 8.87598336e-01
4.58299845e-01 1.86558634e-01 5.18312514e-01 -1.49995351e+00
1.17203355e-01 -2.58405656e-01 -6.95149779e-01 5.09213150e-01
2.23433271e-01 1.83658287e-01 1.03684843e+00 -1.08846068e+00
6.06922567e-01 9.84344482e-01 5.78284442e-01 5.91472864e-01
-1.04879916e+00 -4.36979145e-01 1.00082152e-01 3.58947873e-01
-1.24058414e+00 -7.85854697e-01 7.75888383e-01 -5.05085111e-01
1.02767742e+00 -3.76600251e-02 6.41613245e-01 8.28882158e-01
2.97221780e-01 7.66406000e-01 1.17765903e+00 -1.36045277e-01
3.42511356e-01 7.71983191e-02 -1.48661539e-01 7.91993678e-01
3.36332113e-01 4.72040385e-01 -9.21728373e-01 8.62628445e-02
5.12919307e-01 -4.19763438e-02 1.03881776e-01 -1.05005300e+00
-1.07502460e+00 8.55124891e-01 2.83162624e-01 -3.60895395e-01
-1.45749763e-01 2.52446681e-01 2.95309961e-01 3.77049819e-02
5.03891170e-01 1.05961166e-01 -2.80714422e-01 -4.73765224e-01
-7.72747993e-01 2.68545151e-01 3.75996411e-01 1.09740818e+00
1.17249060e+00 -9.03382003e-02 6.18088186e-01 5.82187116e-01
5.34970403e-01 6.11415148e-01 1.55541092e-01 -1.17198884e+00
4.98412728e-01 5.34129083e-01 7.09184632e-02 -6.23034418e-01
-5.47509432e-01 -4.71194610e-02 -4.18716341e-01 7.68590093e-01
3.24835032e-01 4.32098210e-02 -6.61333263e-01 1.47320294e+00
4.47774410e-01 -9.94023308e-02 1.73383251e-01 9.31197524e-01
7.60454357e-01 2.00538263e-01 -2.25264743e-01 1.25824988e-01
1.00411010e+00 -8.94581258e-01 -3.56966257e-01 -7.49743402e-01
4.23485339e-01 -5.06662011e-01 8.96527112e-01 4.95002538e-01
-9.56565082e-01 -4.48058903e-01 -1.31679130e+00 -3.34191024e-01
-3.29665363e-01 3.81501392e-02 7.40483463e-01 6.70539081e-01
-9.24495161e-01 3.30155313e-01 -1.05392718e+00 -4.47344393e-01
5.88011563e-01 2.70706803e-01 -4.48059082e-01 -1.52932420e-01
-5.88999450e-01 9.42228556e-01 2.38288030e-01 1.78298969e-02
-8.91258717e-01 -5.46920538e-01 -1.10927916e+00 -4.46938843e-01
4.63803768e-01 -5.67147970e-01 1.32548130e+00 -4.10328954e-01
-1.39638472e+00 1.01950753e+00 -4.85712588e-01 -6.25618458e-01
7.30477810e-01 -4.44423825e-01 -7.60060130e-03 2.91818291e-01
2.84188628e-01 1.04476011e+00 6.63820446e-01 -1.27497792e+00
-9.74182725e-01 -5.62187016e-01 -7.95389041e-02 4.38530266e-01
-2.67576780e-02 -4.36889142e-01 -2.72723436e-01 -3.86926793e-02
5.47554255e-01 -1.03742087e+00 -1.93666160e-01 3.86086375e-01
-1.63011208e-01 1.35817789e-02 9.63846087e-01 -6.61915317e-02
6.79886401e-01 -2.12437606e+00 3.70708890e-02 -2.43095770e-01
3.97193134e-01 -1.50912749e-02 2.23023281e-01 3.37207526e-01
3.37958306e-01 -4.18490678e-01 -3.17709923e-01 -6.30657315e-01
-2.70934433e-01 3.07108581e-01 -1.21045232e-01 6.60354555e-01
3.37511122e-01 7.16835141e-01 -1.04113293e+00 -4.51128572e-01
7.47965813e-01 3.22693437e-01 -4.46375430e-01 1.13966227e-01
-1.93104431e-01 4.96944994e-01 -3.41827601e-01 7.89939106e-01
7.78435469e-01 5.56531735e-02 -2.67612487e-01 -1.05488770e-01
-5.99262118e-01 3.27341437e-01 -1.07869697e+00 1.77119040e+00
-3.25302631e-01 9.02622879e-01 1.41608819e-01 -7.47198641e-01
9.75317955e-01 -1.68067753e-01 6.07962310e-01 -6.86202526e-01
8.46845210e-02 2.92810708e-01 -2.17155680e-01 -4.79433835e-01
7.44004548e-01 -2.47380897e-01 1.08487524e-01 7.00717643e-02
2.86623463e-03 -6.99544549e-01 2.47048978e-02 1.64190695e-01
1.09078050e+00 5.08721650e-01 4.15846676e-01 -1.04732841e-01
3.05457920e-01 3.01574141e-01 5.17275929e-01 3.79238069e-01
-6.96100831e-01 7.78549612e-01 1.07145526e-01 -2.86133856e-01
-1.04068601e+00 -1.22862899e+00 -3.42899680e-01 6.32544041e-01
4.80050772e-01 -3.01178187e-01 -2.49509797e-01 -4.69710141e-01
3.50779027e-01 4.93929237e-01 -3.99932534e-01 3.51864845e-03
-3.06216717e-01 -2.98255265e-01 3.60893488e-01 7.52827406e-01
5.76380193e-01 -6.89372778e-01 -1.38004720e+00 1.04052290e-01
-6.90244287e-02 -1.37342227e+00 2.78316233e-02 4.48357582e-01
-1.05926442e+00 -1.09812689e+00 -3.25064659e-01 -3.90092850e-01
4.72861201e-01 8.95932913e-01 9.63958085e-01 -3.59268099e-01
-5.30235350e-01 7.77331233e-01 -1.36721119e-01 -6.23489857e-01
-3.19153108e-02 -5.95002249e-02 2.50623584e-01 -3.83217812e-01
6.26081526e-01 -5.72009504e-01 -7.83396482e-01 3.81388485e-01
-5.71389139e-01 4.98978384e-02 4.93238956e-01 5.54957747e-01
8.69710743e-01 -2.75683962e-02 4.01145488e-01 -5.83723664e-01
-7.27989599e-02 -4.95132238e-01 -1.06006324e+00 -3.91687661e-01
-8.25655699e-01 -9.11005139e-02 1.71941742e-01 9.93139297e-02
-1.01815975e+00 4.56148028e-01 -9.01573822e-02 -3.01814526e-01
-2.98782110e-01 1.62753344e-01 -5.19960672e-02 -2.80646503e-01
6.24065638e-01 1.11972302e-01 4.19354826e-01 -2.00736016e-01
3.12655568e-01 5.67105591e-01 6.27455413e-01 -2.22136095e-01
8.22948217e-01 1.09903979e+00 2.90723324e-01 -1.17382419e+00
-8.50706339e-01 -8.79320979e-01 -8.22896600e-01 -4.36781585e-01
7.90721178e-01 -1.24887288e+00 -5.79473734e-01 5.69463551e-01
-7.39717722e-01 -3.65992367e-01 -1.37392223e-01 6.55524850e-01
-9.52275097e-01 5.24017811e-01 -7.00516403e-02 -1.08603716e+00
-9.23115462e-02 -1.02130544e+00 1.12309301e+00 2.94309884e-01
-9.26171616e-02 -8.36859941e-01 5.68649806e-02 6.89748168e-01
1.31908327e-01 3.09839845e-01 1.84993684e-01 -4.77275392e-03
-9.95824695e-01 -1.31183177e-01 -2.77724773e-01 4.12882954e-01
1.60303339e-01 5.30973896e-02 -1.37811136e+00 -2.81860262e-01
-1.01864822e-01 -6.40256405e-01 8.84618998e-01 4.08576667e-01
8.69859517e-01 3.73053730e-01 -3.00613374e-01 7.87265360e-01
1.50027287e+00 -8.42138156e-02 5.15615702e-01 5.91549039e-01
5.65144598e-01 7.69928992e-01 1.08128166e+00 6.38392985e-01
7.99272716e-01 5.01538754e-01 8.09113681e-01 1.74582466e-01
-1.53038859e-01 -4.29180652e-01 9.78698283e-02 4.95205164e-01
6.88808411e-02 -1.01403922e-01 -1.03664446e+00 6.33574426e-01
-1.65656173e+00 -8.01793277e-01 -2.92991102e-01 2.16713190e+00
1.93235978e-01 4.93871093e-01 8.67171735e-02 1.70504168e-01
1.18180215e-01 2.27981120e-01 -9.36000586e-01 -1.02437921e-01
-2.18024343e-01 -2.74655610e-01 8.30187559e-01 4.29619998e-01
-1.18312883e+00 8.76369476e-01 6.16772270e+00 2.65949160e-01
-1.07495165e+00 -3.01203448e-02 1.10146463e-01 -3.05302918e-01
-2.27875710e-01 -1.80570467e-03 -1.11240494e+00 2.54885823e-01
6.44878924e-01 -1.19745232e-01 2.48832420e-01 9.20870543e-01
2.26479813e-01 -7.89677978e-01 -1.07921493e+00 1.06346416e+00
1.35535166e-01 -1.13158119e+00 -5.40382862e-01 2.04134822e-01
5.77384412e-01 7.03722894e-01 1.04412615e-01 -6.50202781e-02
3.12323291e-02 -6.70770288e-01 7.93595791e-01 2.23469138e-01
8.62430036e-01 -6.32671595e-01 3.63578588e-01 5.42526901e-01
-1.20165741e+00 -1.11421555e-01 -4.64368612e-01 -2.99039572e-01
5.12026660e-02 5.66424012e-01 -8.48063529e-01 2.53738195e-01
8.77844155e-01 9.38353837e-01 -5.02561271e-01 9.98039722e-01
-1.67496949e-01 2.34409541e-01 -5.58759451e-01 9.61841866e-02
2.85286397e-01 -2.02319726e-01 6.60942793e-01 8.19219649e-01
2.94835031e-01 -3.22374888e-02 -6.63967207e-02 6.90033138e-01
4.00635004e-01 -1.52602449e-01 -1.08242011e+00 1.32286727e-01
5.41799128e-01 1.24964154e+00 -6.61770225e-01 4.24066596e-02
-8.54434133e-01 7.92528033e-01 4.25311327e-01 7.55126029e-03
-4.95384753e-01 -1.31708220e-01 9.12205517e-01 3.28407466e-01
3.79174471e-01 -7.22865939e-01 -4.96854722e-01 -9.91897762e-01
2.27023005e-01 -4.27484632e-01 1.96341291e-01 -7.47778833e-01
-9.77331281e-01 3.23174149e-01 1.25887960e-01 -1.60597730e+00
-2.20863461e-01 -5.90507686e-01 -3.42952102e-01 5.31748593e-01
-2.07928443e+00 -1.01825070e+00 -8.31432462e-01 2.69414961e-01
6.88354254e-01 -1.13236256e-01 5.77377141e-01 1.53864831e-01
-3.28456700e-01 1.84238330e-01 1.11805499e-01 -4.60776299e-01
6.84079826e-01 -1.27023947e+00 5.48940361e-01 7.93435395e-01
-1.56301349e-01 1.90621063e-01 7.54057109e-01 -5.09347618e-01
-1.63352084e+00 -8.15382659e-01 4.41434652e-01 -6.12828195e-01
5.69042802e-01 -4.25637692e-01 -7.98829675e-01 5.52996635e-01
-2.79694229e-01 1.98783919e-01 5.54032028e-01 -1.19937681e-01
-2.71827072e-01 -2.05852732e-01 -1.12862372e+00 4.69946921e-01
1.08435535e+00 -5.27223527e-01 -4.44243401e-01 -1.77095663e-02
2.84687430e-01 -5.46069443e-01 -5.47764540e-01 6.21058345e-01
6.81164980e-01 -1.49773955e+00 1.01591802e+00 1.16399907e-01
4.84538525e-01 -3.87808472e-01 -2.48303801e-01 -9.54504371e-01
-1.28997751e-02 -3.78594726e-01 -2.41564319e-01 8.86505485e-01
1.13620616e-01 -8.40082943e-01 1.23714447e+00 5.21142185e-01
-2.02657953e-01 -7.08562553e-01 -1.22448039e+00 -7.67549932e-01
-1.43021330e-01 -5.97636402e-01 1.56573832e-01 4.68477458e-01
-1.31657466e-01 2.93998957e-01 -2.66087890e-01 2.77718842e-01
1.06886172e+00 4.33190726e-02 1.02111316e+00 -1.30570149e+00
-8.84153247e-02 -2.57803887e-01 -7.40977824e-01 -1.29507399e+00
2.04704814e-02 -4.91288185e-01 3.72384816e-01 -1.56121612e+00
5.65850437e-02 -5.88284075e-01 7.14056194e-02 1.60298407e-01
6.22068420e-02 3.04572195e-01 -1.58702224e-01 2.77671903e-01
-6.08775854e-01 7.02043533e-01 9.89852071e-01 1.88481793e-01
-8.41127783e-02 1.18485354e-01 -3.91193777e-01 8.35572183e-01
1.09694231e+00 -1.82515249e-01 -7.01122701e-01 -5.01846135e-01
2.03307986e-01 3.86449322e-02 5.06471157e-01 -1.25927663e+00
4.01783615e-01 -1.65629461e-01 2.35720456e-01 -1.13910544e+00
7.99950242e-01 -7.37306178e-01 -2.70096064e-01 3.82001400e-01
1.67403072e-01 7.72649646e-02 3.81798029e-01 7.91607857e-01
-2.73283482e-01 -1.80157781e-01 8.73193800e-01 -3.75618488e-02
-1.23674130e+00 3.14905643e-01 -4.02530402e-01 1.18897207e-01
1.10381961e+00 -7.86121607e-01 -2.80941725e-01 -3.00390095e-01
-8.03752020e-02 4.40909714e-01 8.35496306e-01 5.46010077e-01
9.61634815e-01 -1.03526938e+00 -4.45988923e-01 5.29425323e-01
6.62945569e-01 4.67594057e-01 2.21810669e-01 7.10445940e-01
-6.79557741e-01 4.85175729e-01 -2.52840906e-01 -8.91669750e-01
-1.29675496e+00 2.81535715e-01 2.21010730e-01 4.76908714e-01
-8.78565133e-01 8.49673152e-01 1.50080666e-01 -4.66337532e-01
2.22403660e-01 -2.22154036e-01 7.99141750e-02 -2.16091592e-02
3.56552213e-01 6.47066355e-01 9.30827409e-02 -6.68245792e-01
-5.33726513e-01 7.00783551e-01 3.84309962e-02 -2.80519605e-01
1.25641048e+00 -5.15647829e-01 2.61090130e-01 6.34831548e-01
1.09846127e+00 -5.23452759e-02 -1.88555038e+00 -3.28361720e-01
1.01024866e-01 -7.38030195e-01 4.47389662e-01 -3.00669819e-01
-6.46817863e-01 9.79102969e-01 7.12345123e-01 -1.39116570e-01
8.24475944e-01 -5.01891077e-02 7.18326211e-01 4.51504797e-01
6.78189218e-01 -1.21468759e+00 1.94483191e-01 5.12447774e-01
5.30437171e-01 -1.81866682e+00 3.50819677e-01 -4.94211555e-01
-6.73417687e-01 9.65816677e-01 8.26208770e-01 8.88778269e-02
6.22276664e-01 3.99896979e-01 3.23537290e-01 -3.32205981e-01
-7.74616957e-01 -3.80349994e-01 -7.48595921e-03 7.14685380e-01
1.62363365e-01 -1.16863877e-01 9.44178104e-02 -2.01842740e-01
-1.76806003e-01 -2.38954034e-02 5.47032058e-01 1.34088624e+00
-7.46980727e-01 -8.31342459e-01 -1.33778945e-01 4.76394117e-01
1.65671036e-02 1.89603120e-01 -1.79138839e-01 7.28980124e-01
-5.26600704e-03 1.03639066e+00 2.00616613e-01 -3.28644156e-01
3.15433830e-01 -1.37294471e-01 6.29400373e-01 -6.97624028e-01
1.47072822e-01 -4.50104959e-02 -6.42357320e-02 -8.20892572e-01
-5.10287821e-01 -1.03563821e+00 -1.26366472e+00 -1.44669533e-01
-2.38832518e-01 -4.24300164e-01 9.93388712e-01 7.10395217e-01
3.66083354e-01 1.14308618e-01 7.25407183e-01 -1.02325881e+00
-5.27220607e-01 -6.53275132e-01 -5.73555231e-01 2.55613681e-02
7.11456478e-01 -9.68861341e-01 -5.87424159e-01 -2.33200058e-01] | [8.12161636352539, -2.3675172328948975] |
0b8b5195-1ac5-4b8a-90e2-9e0d04ce71f0 | fenerf-face-editing-in-neural-radiance-fields | 2111.15490 | null | https://arxiv.org/abs/2111.15490v2 | https://arxiv.org/pdf/2111.15490v2.pdf | FENeRF: Face Editing in Neural Radiance Fields | Previous portrait image generation methods roughly fall into two categories: 2D GANs and 3D-aware GANs. 2D GANs can generate high fidelity portraits but with low view consistency. 3D-aware GAN methods can maintain view consistency but their generated images are not locally editable. To overcome these limitations, we propose FENeRF, a 3D-aware generator that can produce view-consistent and locally-editable portrait images. Our method uses two decoupled latent codes to generate corresponding facial semantics and texture in a spatial aligned 3D volume with shared geometry. Benefiting from such underlying 3D representation, FENeRF can jointly render the boundary-aligned image and semantic mask and use the semantic mask to edit the 3D volume via GAN inversion. We further show such 3D representation can be learned from widely available monocular image and semantic mask pairs. Moreover, we reveal that joint learning semantics and texture helps to generate finer geometry. Our experiments demonstrate that FENeRF outperforms state-of-the-art methods in various face editing tasks. | ['Jue Wang', 'Yebin Liu', 'Qi Zhang', 'Xiaoyu Li', 'Yong Zhang', 'Xuan Wang', 'Jingxiang Sun'] | 2021-11-30 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Sun_FENeRF_Face_Editing_in_Neural_Radiance_Fields_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Sun_FENeRF_Face_Editing_in_Neural_Radiance_Fields_CVPR_2022_paper.pdf | cvpr-2022-1 | ['3d-aware-image-synthesis'] | ['computer-vision'] | [ 5.14329493e-01 4.91457880e-01 9.74607095e-02 -4.93729115e-01
-6.69907987e-01 -7.88584650e-01 7.48366654e-01 -9.05121565e-01
5.79715133e-01 8.61459672e-01 4.39072669e-01 3.13667893e-01
2.28959396e-01 -1.17616832e+00 -8.88309896e-01 -6.10267282e-01
6.88616276e-01 5.92505395e-01 -3.80959630e-01 -2.47702524e-01
-1.46808654e-01 5.66181481e-01 -1.51797187e+00 2.53756613e-01
1.05776787e+00 9.01391983e-01 6.92623630e-02 6.16390824e-01
-6.69018850e-02 4.00074542e-01 -3.65211517e-01 -5.80137193e-01
7.16027081e-01 -9.30908978e-01 -4.80650783e-01 4.54281449e-01
1.02147996e+00 -7.37384975e-01 -3.38658303e-01 6.79664373e-01
3.56381238e-01 -3.33499789e-01 8.41763914e-01 -1.36515796e+00
-1.22396898e+00 1.95055604e-01 -8.30728054e-01 -6.27297759e-01
5.21468759e-01 2.51481235e-01 6.40588045e-01 -8.60510170e-01
1.15894377e+00 1.45984197e+00 5.55370569e-01 8.95276070e-01
-1.46721101e+00 -7.18756974e-01 1.59552395e-01 -5.17409682e-01
-1.23472643e+00 -5.23030877e-01 1.10460651e+00 -4.11697030e-01
4.17955846e-01 3.98411781e-01 8.71462345e-01 1.23376954e+00
1.39865056e-01 4.30387378e-01 1.49551225e+00 -1.29044265e-01
8.75296593e-02 -2.07990497e-01 -1.01597154e+00 1.04307365e+00
-6.08870313e-02 1.81154296e-01 -7.89369226e-01 -3.86412404e-02
1.50277328e+00 5.51144704e-02 -3.04054260e-01 -4.31701690e-01
-1.22555697e+00 6.76191509e-01 3.57720017e-01 -1.20326318e-01
-2.20889300e-01 3.84580195e-01 -2.01608568e-01 2.62980402e-01
8.89083922e-01 3.84059668e-01 -1.19354026e-02 5.43594435e-02
-1.02497256e+00 4.15313452e-01 3.80958438e-01 1.37599444e+00
8.06664109e-01 3.68136823e-01 -3.74998480e-01 8.05884004e-01
1.56760588e-01 8.77145588e-01 -2.01380074e-01 -1.27540982e+00
2.98481435e-01 4.71866041e-01 -6.84962422e-03 -9.66224313e-01
2.58278459e-01 -9.80681553e-02 -1.06143892e+00 4.23635066e-01
-1.63680071e-03 -4.72565228e-03 -1.18016016e+00 1.87791824e+00
5.33466637e-01 -6.49131238e-02 -3.30663443e-01 8.11923444e-01
8.07177842e-01 5.05188286e-01 -4.56508279e-01 9.44106653e-03
1.04050994e+00 -9.35652554e-01 -9.01750386e-01 -1.66777782e-02
8.75681713e-02 -7.87103295e-01 9.88481343e-01 1.11007176e-01
-1.54516125e+00 -4.06557500e-01 -8.89954329e-01 -3.73394817e-01
6.77282810e-02 1.09681636e-01 5.49026370e-01 5.33737302e-01
-1.37780809e+00 3.10796291e-01 -5.38263261e-01 -2.34773487e-01
8.27490985e-01 6.60331920e-02 -5.51377118e-01 -1.83817163e-01
-7.89510667e-01 5.94098032e-01 -2.57444501e-01 -3.40166278e-02
-1.27022409e+00 -9.82908845e-01 -1.01646411e+00 -3.37671906e-01
1.33569255e-01 -1.43872511e+00 8.85051966e-01 -9.82593536e-01
-1.84231746e+00 1.36142135e+00 -3.12474698e-01 5.52341081e-02
8.08261275e-01 1.17063604e-01 -4.84355502e-02 1.83214754e-01
2.31568515e-01 1.01656425e+00 1.19764507e+00 -1.62578928e+00
-1.32204100e-01 -4.00499880e-01 9.85861663e-03 3.74147654e-01
-6.01992523e-03 -5.68875611e-01 -3.80426973e-01 -1.03426766e+00
1.61118731e-01 -8.61028790e-01 1.24549970e-01 6.64883316e-01
-6.57613277e-01 4.39974636e-01 1.05456471e+00 -7.51164854e-01
6.55839741e-01 -1.78944850e+00 6.21517003e-01 7.83560574e-02
5.35621405e-01 -1.96470916e-01 -2.66696781e-01 4.17143732e-01
1.29581526e-01 2.79784083e-01 -4.63217616e-01 -7.14899600e-01
9.69500244e-02 2.31639221e-01 -4.95120883e-01 2.35176191e-01
1.82710037e-01 1.25187624e+00 -8.46888661e-01 -3.74906093e-01
2.96517104e-01 8.37864697e-01 -8.12819719e-01 4.53656107e-01
-3.93211424e-01 1.05699790e+00 -3.06852818e-01 6.84133470e-01
9.63449538e-01 -9.69477072e-02 1.06172711e-01 -3.76790673e-01
9.17855874e-02 -1.93469767e-02 -5.99683225e-01 2.13961887e+00
-6.56584978e-01 5.49153090e-01 2.14091018e-01 -5.88409066e-01
1.09201050e+00 1.72738686e-01 3.77205700e-01 -6.74465954e-01
-8.03964511e-02 1.07259601e-01 -7.34137297e-01 -5.46443351e-02
4.32254255e-01 -3.73094380e-01 -4.90520475e-03 6.00950658e-01
1.21342540e-01 -9.25129354e-01 -4.65934217e-01 2.53927201e-01
7.03383565e-01 6.86606288e-01 -1.41623914e-01 -1.73127726e-01
1.68113261e-01 -4.39830333e-01 4.40590143e-01 3.98563832e-01
4.08779204e-01 1.26801229e+00 4.54224110e-01 -5.51027298e-01
-1.42426121e+00 -1.52106786e+00 1.24406267e-03 4.17303085e-01
1.26991361e-01 -3.91229510e-01 -1.04068339e+00 -6.09691322e-01
5.26608946e-03 5.42169690e-01 -9.37479854e-01 -1.07659385e-01
-3.99367869e-01 -2.12518081e-01 3.35785896e-01 3.02575022e-01
7.96791971e-01 -6.06650293e-01 -9.82319266e-02 -1.12284496e-01
-2.53563076e-01 -9.88777339e-01 -1.06790090e+00 -6.68290555e-01
-6.75401926e-01 -8.57487321e-01 -1.03684938e+00 -6.67679965e-01
1.09185851e+00 2.81221241e-01 1.31582630e+00 1.06729388e-01
-3.10924083e-01 6.39518678e-01 -1.20710999e-01 -1.60535872e-01
-6.04080856e-01 -1.01595990e-01 -1.36980880e-02 2.33443826e-01
-5.02634048e-01 -1.19393528e+00 -8.83015096e-01 4.45089817e-01
-9.00545657e-01 9.53543842e-01 7.39501193e-02 8.30751538e-01
8.83084536e-01 -4.11832899e-01 4.30268228e-01 -1.02560091e+00
2.39995793e-01 -8.17407444e-02 -5.67909658e-01 2.37731412e-01
-3.80888522e-01 8.66041481e-02 5.55505693e-01 -8.52087885e-02
-1.36955106e+00 -8.86921212e-02 -1.12141244e-01 -6.74863935e-01
3.72601040e-02 -3.01159978e-01 -5.19467831e-01 -2.22667411e-01
4.29180622e-01 4.30126011e-01 1.92303911e-01 -3.15223932e-01
7.82696426e-01 3.23595166e-01 5.58596671e-01 -6.68812394e-01
1.15033972e+00 9.69332755e-01 1.55634284e-01 -5.75427949e-01
-8.87203693e-01 5.40593266e-01 -7.06536114e-01 -3.99787724e-01
1.00565147e+00 -1.14142990e+00 -4.56946313e-01 8.04939330e-01
-1.00909543e+00 -6.58511102e-01 -6.36792839e-01 -2.91284293e-01
-9.84182298e-01 1.07366778e-01 -5.19397676e-01 -5.40286422e-01
-3.69693905e-01 -9.73340034e-01 1.83708203e+00 6.92672981e-03
-1.27036020e-01 -8.67010117e-01 -1.15724564e-01 5.58147907e-01
4.69849288e-01 9.58199620e-01 8.66020739e-01 7.12309659e-01
-9.51256514e-01 2.88086534e-01 -1.47378892e-01 1.64121926e-01
6.64775252e-01 -2.76133176e-02 -1.02729869e+00 -2.66561419e-01
-1.20337188e-01 -3.43167573e-01 6.34813309e-01 3.96859318e-01
1.37822902e+00 -6.21746719e-01 -2.09619671e-01 1.24679470e+00
1.27051413e+00 -6.55415729e-02 7.34103739e-01 -3.81838799e-01
1.28233540e+00 4.61754471e-01 1.74001366e-01 4.06978160e-01
5.21911204e-01 9.91550803e-01 3.71974647e-01 -3.17759812e-01
-7.06903338e-01 -1.11310112e+00 2.73225814e-01 7.42216766e-01
-2.31494293e-01 -2.50293612e-01 -3.11075300e-01 3.45988333e-01
-1.56071711e+00 -8.95805061e-01 1.84397653e-01 1.97308290e+00
1.00646675e+00 -4.03044969e-01 -8.52520764e-02 -3.88448775e-01
6.00440085e-01 3.63663971e-01 -8.05277109e-01 -3.19427639e-01
-4.28903162e-01 2.65220374e-01 2.61495560e-01 7.19550610e-01
-5.55898845e-01 1.09841931e+00 6.20564556e+00 8.61209929e-01
-1.01518142e+00 2.50438631e-01 1.08759820e+00 -4.39869165e-01
-1.35159135e+00 1.95934214e-02 -4.95147109e-01 4.88523334e-01
1.83320448e-01 -5.10604605e-02 5.89381158e-01 5.02883852e-01
1.16580404e-01 8.08494762e-02 -1.08726859e+00 1.43697655e+00
4.68784362e-01 -1.95377278e+00 6.41778231e-01 4.10556674e-01
1.55804694e+00 -6.54360771e-01 2.66956985e-01 -2.97845244e-01
4.57484215e-01 -1.39520848e+00 1.01682687e+00 8.63612950e-01
1.84954154e+00 -7.48273313e-01 -2.37872321e-02 -6.20924234e-02
-1.13265860e+00 4.45096791e-01 -1.76508725e-01 2.06688106e-01
5.23654044e-01 6.76262975e-01 -3.09994251e-01 7.25892723e-01
5.15137911e-01 1.00520372e+00 -2.18426496e-01 1.70096382e-01
-4.28590506e-01 1.48721606e-01 -2.55502671e-01 5.91539085e-01
-1.02523334e-01 -6.97444618e-01 6.04970038e-01 3.32913399e-01
6.73419118e-01 2.55065143e-01 -1.32994592e-01 1.62116551e+00
-4.73588705e-01 -3.87229174e-01 -1.01871717e+00 8.59944522e-02
5.73112011e-01 1.09780693e+00 -4.24835563e-01 -2.01151624e-01
5.23842834e-02 1.74123323e+00 2.94382006e-01 3.62842649e-01
-8.96584868e-01 4.53340597e-02 9.60056603e-01 4.59109157e-01
3.27269465e-01 -2.85283774e-01 -4.46072072e-01 -1.38722849e+00
1.65558264e-01 -7.80381501e-01 -2.89731711e-01 -1.28069651e+00
-1.35295165e+00 7.61533082e-01 -9.09716040e-02 -1.12216830e+00
-1.56295717e-01 -1.61633268e-01 -5.50351262e-01 9.15405929e-01
-1.20835841e+00 -1.87996507e+00 -6.20753765e-01 6.00129843e-01
3.84108663e-01 -1.47929221e-01 8.43437135e-01 -2.91355196e-02
-1.47725880e-01 8.10383618e-01 -6.92353696e-02 -4.82174084e-02
7.64627934e-01 -1.05221784e+00 8.04916918e-01 5.42134941e-01
1.04651652e-01 2.55293578e-01 2.32673302e-01 -8.68916214e-01
-1.62739384e+00 -1.37658668e+00 4.99321610e-01 -8.43186736e-01
-2.02371985e-01 -7.10186243e-01 -3.70355606e-01 7.38044202e-01
4.07058299e-01 -9.85834301e-02 3.33837092e-01 -4.24713373e-01
-4.85549569e-01 -1.91399470e-01 -1.49243808e+00 8.86089504e-01
2.02128172e+00 -7.14212477e-01 -6.86235353e-02 3.77977997e-01
7.76208282e-01 -8.20592582e-01 -9.03205335e-01 2.14994296e-01
6.18919075e-01 -1.20215976e+00 9.12859440e-01 -6.54342771e-02
8.34488153e-01 -3.65965396e-01 -3.25993448e-02 -1.53948057e+00
-7.93707594e-02 -1.14278519e+00 6.58923611e-02 1.43350399e+00
2.23441012e-02 -6.30769432e-01 7.40130305e-01 4.67919052e-01
-1.62435308e-01 -7.79733956e-01 -8.25456500e-01 -6.16708457e-01
3.39649282e-02 -2.85136234e-02 1.18156958e+00 9.30823565e-01
-7.12716162e-01 6.54157400e-02 -7.77355433e-01 -3.01203310e-01
9.03963745e-01 5.29384971e-01 1.05856907e+00 -8.92287076e-01
-2.44910317e-03 -1.79849088e-01 -2.31023714e-01 -1.21689034e+00
2.12235630e-01 -1.07226002e+00 -2.10665807e-01 -1.51750028e+00
1.83077082e-01 -5.91865599e-01 7.11379647e-01 4.79241192e-01
2.20490456e-01 7.97841728e-01 1.76791251e-01 8.27443749e-02
6.26415852e-03 1.12745523e+00 2.04626203e+00 -9.56022218e-02
4.15258333e-02 -6.93616033e-01 -1.00033224e+00 4.40793991e-01
3.88054162e-01 -1.01179898e-01 -6.97425246e-01 -7.91339517e-01
3.34275365e-01 4.52985568e-03 5.89731336e-01 -7.64685631e-01
-2.76683480e-01 -2.64412194e-01 8.00305784e-01 -4.08661306e-01
5.60809672e-01 -5.60724437e-01 8.34237218e-01 -1.24674581e-01
-2.63083935e-01 -1.87014878e-01 -8.51362497e-02 5.73083758e-01
-4.96514328e-02 5.80532312e-01 8.62993777e-01 -2.88870573e-01
-1.29620016e-01 9.61883903e-01 9.70613733e-02 2.77816027e-01
9.14729774e-01 -5.40882587e-01 -1.45712659e-01 -8.75185728e-01
-4.35436368e-01 -4.69936766e-02 1.28820252e+00 5.54535508e-01
8.49691451e-01 -1.95894623e+00 -8.81237626e-01 6.81730628e-01
-6.75770193e-02 3.98767263e-01 6.10168159e-01 3.33269477e-01
-6.73810005e-01 9.43662077e-02 -5.05397201e-01 -7.26844132e-01
-1.04989493e+00 1.30539507e-01 4.40410972e-01 1.23217784e-01
-7.46793866e-01 9.59118307e-01 8.28881264e-01 -7.49325037e-01
-3.08211893e-01 -1.21608518e-01 5.13640463e-01 -2.44010746e-01
3.85587335e-01 -8.51221979e-02 -2.25162178e-01 -6.84888601e-01
-3.57441530e-02 1.11889052e+00 2.74948299e-01 -3.04816753e-01
1.48787642e+00 -2.87451237e-01 -2.63119370e-01 6.67808577e-02
1.05101907e+00 2.95723915e-01 -1.92356920e+00 4.60061878e-02
-1.02777302e+00 -1.06282735e+00 -1.44703150e-01 -6.54151022e-01
-1.53549540e+00 6.78252339e-01 2.59246141e-01 -2.89328784e-01
1.27889752e+00 2.76950505e-02 1.06986189e+00 -4.05733019e-01
7.80624628e-01 -8.33894789e-01 1.80891737e-01 1.09359428e-01
1.51536548e+00 -8.47428083e-01 -3.02525442e-02 -9.19535041e-01
-6.75369024e-01 8.52666318e-01 7.42908359e-01 -2.25701705e-01
5.75416863e-01 1.18303239e-01 -1.78219467e-01 -3.19080055e-01
-5.72939098e-01 2.57447392e-01 4.28469479e-01 9.53405499e-01
2.28796199e-01 1.15280159e-01 9.94701087e-02 2.04886794e-01
-6.32656097e-01 -7.98277482e-02 3.79696429e-01 5.19998193e-01
3.64591837e-01 -1.19318283e+00 -2.81291932e-01 3.51260245e-01
1.76459983e-01 -1.06119867e-02 -6.45645082e-01 5.04092813e-01
2.71855861e-01 5.78095138e-01 3.15103680e-01 -2.98582792e-01
8.60525295e-02 -1.26734555e-01 1.09065342e+00 -5.96992373e-01
-1.29818425e-01 -6.53889179e-02 -8.81374069e-03 -7.96937168e-01
-3.72359544e-01 -3.46882999e-01 -6.55540287e-01 -6.28825009e-01
1.46815836e-01 -2.99835742e-01 2.71623015e-01 6.41717851e-01
8.38207424e-01 3.94889474e-01 8.80167782e-01 -1.05474055e+00
1.15865596e-01 -5.60432792e-01 -6.99497998e-01 6.02182031e-01
3.41674626e-01 -6.54422939e-01 -2.86855161e-01 4.31959450e-01] | [12.42754077911377, -0.4876845180988312] |
1f61f5c0-0593-4d2f-894c-52041c87b598 | extreme-face-inpainting-with-sketch-guided | 2105.06033 | null | https://arxiv.org/abs/2105.06033v1 | https://arxiv.org/pdf/2105.06033v1.pdf | Extreme Face Inpainting with Sketch-Guided Conditional GAN | Recovering badly damaged face images is a useful yet challenging task, especially in extreme cases where the masked or damaged region is very large. One of the major challenges is the ability of the system to generalize on faces outside the training dataset. We propose to tackle this extreme inpainting task with a conditional Generative Adversarial Network (GAN) that utilizes structural information, such as edges, as a prior condition. Edge information can be obtained from the partially masked image and a structurally similar image or a hand drawing. In our proposed conditional GAN, we pass the conditional input in every layer of the encoder while maintaining consistency in the distributions between the learned weights and the incoming conditional input. We demonstrate the effectiveness of our method with badly damaged face examples. | ['Andreas Savakis', 'Nilesh Pandey'] | 2021-05-13 | null | null | null | null | ['facial-inpainting'] | ['computer-vision'] | [ 6.35141015e-01 3.27120781e-01 2.48366773e-01 -3.36459219e-01
-6.48302197e-01 -4.70098466e-01 2.09915370e-01 -5.99226475e-01
-2.92337481e-02 9.01982605e-01 2.90558249e-01 2.12360486e-01
3.10195565e-01 -8.53210151e-01 -1.06512475e+00 -6.99540019e-01
2.89382160e-01 1.53187349e-01 -1.52613193e-01 3.83325517e-02
-5.43860756e-02 7.52917588e-01 -1.36188269e+00 4.28425550e-01
8.71622026e-01 1.02781188e+00 1.25933424e-01 5.48209429e-01
1.46487087e-01 7.02755749e-01 -7.77476370e-01 -6.53361380e-01
5.12565553e-01 -7.09611177e-01 -4.49617505e-01 4.01745588e-01
5.82543314e-01 -6.39594197e-01 -4.06013906e-01 1.09790754e+00
3.04448813e-01 9.91582870e-03 7.04107583e-01 -1.17672300e+00
-8.06326985e-01 1.66895777e-01 -8.78243387e-01 -9.02130529e-02
2.82139093e-01 1.94146320e-01 3.98050398e-01 -1.01076412e+00
5.49166977e-01 1.15433466e+00 4.98290420e-01 8.18399131e-01
-1.33794594e+00 -7.75311470e-01 7.38849789e-02 1.00528739e-01
-1.32332075e+00 -6.76607192e-01 1.24221694e+00 -3.86311948e-01
4.05482799e-01 -3.26054208e-02 2.54889131e-01 1.06783247e+00
1.08305484e-01 4.23235953e-01 1.01253927e+00 -4.06604111e-01
3.64448667e-01 8.14464390e-02 -6.56590283e-01 6.30122125e-01
1.00193307e-01 1.42093882e-01 -5.16631305e-01 -1.21528730e-01
8.71105134e-01 2.52223015e-01 -4.80263770e-01 -1.51041627e-01
-4.75709826e-01 6.92662656e-01 6.59926176e-01 -3.93806472e-02
-5.61714470e-01 2.50893414e-01 -1.91717505e-01 3.39856774e-01
6.87604904e-01 1.69298679e-01 -1.52559325e-01 3.08063924e-01
-1.24720931e+00 -1.35601714e-01 4.11356479e-01 7.47984111e-01
8.62482727e-01 3.04293215e-01 -8.74715522e-02 7.71878660e-01
6.08135527e-03 3.03773224e-01 1.95884481e-02 -9.32497442e-01
4.36333984e-01 3.92602652e-01 4.50008102e-02 -8.78692806e-01
3.33775759e-01 -2.92369634e-01 -8.86412382e-01 5.90855718e-01
2.96019226e-01 -4.01371717e-01 -1.16085386e+00 2.12753773e+00
2.36674905e-01 2.91136980e-01 -1.03509622e-02 6.67464733e-01
3.99819523e-01 4.85924542e-01 2.74010524e-02 -3.34165663e-01
8.18785429e-01 -7.22027838e-01 -6.80501580e-01 -6.90629780e-01
-2.00778425e-01 -7.21266150e-01 8.13638628e-01 1.75364748e-01
-1.34755266e+00 -4.92934525e-01 -9.97367561e-01 -1.31304771e-01
-7.64756575e-02 2.63519555e-01 1.51023790e-01 4.32869464e-01
-9.89808559e-01 5.47270536e-01 -7.13955700e-01 -5.74530661e-02
8.83979261e-01 4.04203922e-01 -8.03712368e-01 -6.00991189e-01
-8.87622058e-01 7.51090884e-01 4.13163267e-02 2.57993996e-01
-1.11501145e+00 -7.28414953e-01 -9.46659327e-01 3.71214241e-01
1.33653209e-01 -5.94870925e-01 6.69584453e-01 -1.33907712e+00
-1.26639080e+00 8.34130645e-01 -2.46347606e-01 -2.05257282e-01
6.47716761e-01 -1.39710203e-01 -1.07603431e-01 4.44705099e-01
-8.39776024e-02 6.24365687e-01 1.50690186e+00 -1.55898809e+00
1.88973937e-02 -6.34824574e-01 -1.63036332e-01 -1.86666921e-02
-2.04525679e-01 -1.10540502e-01 -2.49896273e-01 -1.01156235e+00
7.96392467e-03 -6.54004037e-01 -2.39754207e-02 4.88965958e-01
-2.80428469e-01 4.04189199e-01 1.09771788e+00 -1.19368041e+00
6.75580084e-01 -2.51822019e+00 3.35446060e-01 1.01335980e-01
-1.36015639e-01 9.07142162e-02 -1.56422436e-01 3.48764390e-01
-3.73741746e-01 -5.27872667e-02 -7.00804055e-01 -4.90272582e-01
-3.09485734e-01 3.49125624e-01 -5.07067204e-01 5.12702882e-01
6.35532200e-01 7.52990901e-01 -5.71436703e-01 -2.05881700e-01
-1.17920727e-01 9.45433438e-01 -5.05976140e-01 6.14208400e-01
-2.22968236e-01 8.18935812e-01 -6.90196827e-02 4.86773849e-01
1.05575407e+00 1.68020964e-01 -5.19211628e-02 -1.55584618e-01
5.03647745e-01 -1.88399032e-01 -9.57415998e-01 1.62273741e+00
-5.37228584e-01 4.44069862e-01 3.06835830e-01 -8.28852952e-01
7.95386732e-01 4.37035918e-01 2.08458781e-01 -3.99003714e-01
7.55804870e-03 -1.25083998e-01 -2.54738629e-01 -4.80917454e-01
4.88259457e-02 -6.66303515e-01 2.74021596e-01 6.06910229e-01
1.31289959e-01 -2.45984986e-01 -2.54806787e-01 2.35159352e-01
1.10828435e+00 1.48121879e-01 6.80489838e-03 2.21608400e-01
2.86796927e-01 -6.91375554e-01 6.42043352e-01 1.36257768e-01
1.13675982e-01 1.16020346e+00 7.36282170e-01 -1.48801506e-01
-1.01500630e+00 -1.13187969e+00 -7.35517638e-03 6.56578660e-01
-2.11841211e-01 8.39571729e-02 -1.01410091e+00 -9.60936904e-01
8.34601447e-02 5.80188930e-01 -9.39082444e-01 -5.66635430e-01
-5.17055035e-01 -3.58584583e-01 3.08958948e-01 8.07723343e-01
5.98733604e-01 -1.18756425e+00 -2.21314773e-01 -7.14713037e-02
-2.86148172e-02 -1.19630742e+00 -5.60214460e-01 7.72296414e-02
-7.75743783e-01 -9.71710205e-01 -7.25251079e-01 -8.90280128e-01
1.36971724e+00 -5.80425449e-02 9.69529629e-01 2.03249738e-01
-2.96904355e-01 1.90405369e-01 -8.16093944e-03 -5.61714545e-02
-2.94199705e-01 -5.36038935e-01 -2.10235611e-01 5.73752105e-01
-1.73263669e-01 -8.62171829e-01 -6.71620607e-01 2.04865843e-01
-1.15765107e+00 -7.77211860e-02 6.73148930e-01 1.00242078e+00
3.64727527e-01 3.98610801e-01 6.71423852e-01 -1.08785629e+00
1.90335751e-01 -4.15646881e-01 -3.26406509e-01 2.02395752e-01
-5.80049120e-03 8.40170607e-02 7.05763042e-01 -4.49405611e-01
-1.34262025e+00 3.76068413e-01 -4.04154837e-01 -6.10337734e-01
-1.27229661e-01 1.46566257e-01 -7.88377821e-01 -2.74329782e-01
5.63203275e-01 1.88424706e-01 5.80858178e-02 -5.55600405e-01
3.19320142e-01 4.27025855e-01 7.90394068e-01 -5.92600584e-01
9.76124883e-01 6.87379181e-01 1.13947041e-01 -5.59745252e-01
-6.96111679e-01 1.64307818e-01 -6.89265430e-01 -1.99769348e-01
5.65113306e-01 -8.80580246e-01 -3.14548343e-01 5.22398293e-01
-1.02561915e+00 -4.88255113e-01 -5.44271350e-01 6.29052222e-02
-5.40655613e-01 3.21902454e-01 -6.44815028e-01 -7.79491603e-01
-1.11561917e-01 -9.10543084e-01 9.32026565e-01 2.66863137e-01
2.03499317e-01 -8.74232054e-01 -3.70057106e-01 4.45373446e-01
4.54957724e-01 6.85418725e-01 1.04543006e+00 -5.50727025e-02
-5.51650167e-01 -4.84377027e-01 -1.85749561e-01 9.57717478e-01
4.55002040e-01 -1.32233500e-01 -1.15814841e+00 -4.86544430e-01
4.31684405e-01 -5.42364120e-01 1.13453734e+00 2.50471503e-01
1.31855953e+00 -4.66052800e-01 -1.14207178e-01 6.50544763e-01
1.41906106e+00 -1.22980133e-01 9.47263777e-01 -4.89144683e-01
6.58919454e-01 7.11127460e-01 1.75252631e-01 3.34352672e-01
-1.83371037e-01 3.67678136e-01 5.49748123e-01 -2.02877939e-01
-5.13821244e-01 -5.25536418e-01 4.18957382e-01 2.92525798e-01
3.48950140e-02 -1.64017081e-01 -5.12183845e-01 5.32891452e-01
-1.40416789e+00 -1.01877856e+00 4.94337618e-01 2.29844570e+00
1.01558995e+00 -3.14302603e-03 -4.59856063e-01 2.11438805e-01
8.25957000e-01 6.47752583e-02 -6.95846140e-01 -2.14305311e-01
-3.09594721e-02 6.30717337e-01 8.11999813e-02 6.07282221e-01
-7.62839079e-01 7.62868464e-01 6.40157604e+00 5.77827156e-01
-1.10212445e+00 1.25570685e-01 7.46616066e-01 -2.02376157e-01
-1.83962911e-01 4.16000513e-03 -2.48013869e-01 7.41360962e-01
4.65930074e-01 1.49834469e-01 6.43301129e-01 7.31312156e-01
-4.47163545e-02 -2.57210404e-01 -1.18470120e+00 7.47110188e-01
5.27618587e-01 -8.64443839e-01 1.04894027e-01 7.90879726e-02
9.47939873e-01 -6.39331877e-01 5.24367392e-01 -7.45707229e-02
1.55604929e-01 -1.33774388e+00 4.79922473e-01 6.17294490e-01
1.27254570e+00 -8.77699196e-01 5.07249475e-01 3.03559065e-01
-7.74223864e-01 -6.17349111e-02 -3.44244659e-01 -6.43843934e-02
1.92677617e-01 7.49472976e-01 -5.47339678e-01 2.64491588e-01
4.87111509e-01 4.42170322e-01 -4.10867959e-01 7.45801508e-01
-7.00359523e-01 4.46943671e-01 -2.21311778e-01 1.01658595e+00
-3.43546450e-01 -7.81547651e-02 2.65316278e-01 7.00274408e-01
3.41602981e-01 2.07950562e-01 -1.92638576e-01 9.84829426e-01
-5.98918378e-01 -3.55966300e-01 -8.09222817e-01 2.21409664e-01
1.15509987e-01 1.10404110e+00 -5.28985262e-01 -1.14208810e-01
-4.20486182e-01 1.35771000e+00 5.35656273e-01 6.59914374e-01
-6.41252160e-01 -1.08524039e-01 6.15440905e-01 2.07254127e-01
6.61171138e-01 -1.20342106e-01 -4.08006489e-01 -1.01710653e+00
3.54029626e-01 -6.56121016e-01 2.46875331e-01 -9.91608858e-01
-1.34970331e+00 6.88975811e-01 -4.28759366e-01 -9.45329785e-01
-1.00648925e-01 -3.64872426e-01 -8.76109004e-01 1.07280838e+00
-1.55468142e+00 -1.32427013e+00 -5.17107129e-01 7.13018954e-01
4.06612992e-01 -1.63855299e-01 6.55908942e-01 3.25459123e-01
-4.97832537e-01 6.85081244e-01 -1.87464833e-01 2.41904601e-01
8.20409656e-01 -1.02197409e+00 1.86423942e-01 1.05865800e+00
4.94396538e-02 4.79185134e-01 6.16104364e-01 -6.71167195e-01
-1.07213199e+00 -1.25093913e+00 4.36289966e-01 -2.48937711e-01
3.75721417e-02 -5.11279881e-01 -1.09976339e+00 9.12096739e-01
4.12426293e-01 4.75688130e-01 5.61567068e-01 -4.64180857e-01
-4.48454857e-01 -1.26720294e-01 -1.63985717e+00 3.25977027e-01
8.42238069e-01 -6.85743213e-01 -4.22777712e-01 2.87730038e-01
3.92622143e-01 -3.84503961e-01 -4.88582343e-01 3.88206035e-01
3.60969067e-01 -9.60597754e-01 8.66907775e-01 -5.72342515e-01
6.78726375e-01 -3.15114021e-01 -1.53442964e-01 -1.55653691e+00
2.10791714e-02 -4.87357169e-01 -2.78563589e-01 1.32353866e+00
2.75432348e-01 -3.74241680e-01 9.67549920e-01 8.24634731e-01
8.08124393e-02 -5.23637533e-01 -9.33393836e-01 -5.54033756e-01
1.04117900e-01 -1.52557911e-02 5.39916098e-01 7.05211818e-01
-4.58086401e-01 1.07178994e-01 -6.81384146e-01 2.79510945e-01
6.33960903e-01 2.00060830e-02 5.84040821e-01 -9.88742650e-01
-5.41730165e-01 1.37695804e-01 -3.17850679e-01 -7.78464854e-01
5.07212818e-01 -6.19718850e-01 2.16092587e-01 -1.22609568e+00
3.02032977e-01 -1.24296211e-01 -7.76111633e-02 7.06016421e-01
-3.04349929e-01 6.67496920e-01 7.06363246e-02 -5.50994612e-02
1.93729416e-01 8.30212712e-01 1.38557744e+00 -1.86015606e-01
3.06324158e-02 3.43536660e-02 -7.20777869e-01 6.83608830e-01
5.96594632e-01 -7.20038712e-01 -4.26605016e-01 -5.42438328e-01
1.00729391e-01 1.78360760e-01 5.45084834e-01 -9.82366920e-01
1.95200220e-01 -1.24020077e-01 9.18048918e-01 -7.01933578e-02
6.74011528e-01 -1.05143416e+00 2.84623891e-01 1.80713773e-01
-2.21790269e-01 -1.52412847e-01 2.11693764e-01 6.49182916e-01
-4.04626727e-01 -9.40073282e-02 1.17078090e+00 -1.40847355e-01
-1.83187097e-01 3.42918962e-01 2.36220926e-01 1.66040994e-02
1.14340413e+00 -1.86932862e-01 -2.31714204e-01 -5.01075923e-01
-9.42499399e-01 -1.21729597e-01 8.00192773e-01 1.26860335e-01
9.35527146e-01 -1.30145299e+00 -7.16119587e-01 7.93737054e-01
-2.30790809e-01 1.45427257e-01 4.21514660e-01 5.16668200e-01
-4.13273156e-01 -4.00485039e-01 -5.95476985e-01 -1.20907336e-01
-1.11751175e+00 8.42819989e-01 3.08509737e-01 2.07756937e-01
-6.36818767e-01 8.03713083e-01 5.07971704e-01 1.24032632e-01
2.77935296e-01 1.53354883e-01 8.71615484e-02 -2.06910580e-01
5.61875224e-01 -1.01038933e-01 6.78395852e-02 -7.21415401e-01
-1.32211596e-01 5.61719894e-01 -1.24875896e-01 -1.83786824e-01
1.48335433e+00 6.04774915e-02 -2.49200717e-01 -9.38991010e-02
1.25304902e+00 2.80777425e-01 -1.92001426e+00 -1.68472856e-01
-5.71501374e-01 -8.63512993e-01 8.79455730e-02 -7.77602553e-01
-1.72213089e+00 9.69008803e-01 4.79149371e-01 -1.42732441e-01
1.51610804e+00 9.67793018e-02 7.91340232e-01 -2.99332798e-01
2.12886080e-01 -7.96731174e-01 4.08755541e-01 -1.14032350e-01
1.29855561e+00 -1.06585586e+00 3.18019502e-02 -5.33825576e-01
-6.39039516e-01 8.72024536e-01 6.42198324e-01 -4.07079011e-01
8.03063512e-01 4.75640804e-01 -2.00462248e-02 -1.50112119e-02
-6.04278088e-01 3.05233765e-02 1.70544863e-01 7.14119196e-01
1.78276628e-01 -2.06202716e-01 6.20047487e-02 4.97738302e-01
-3.81726660e-02 -1.25431731e-01 4.31331724e-01 1.09007275e+00
-9.52238664e-02 -1.08998513e+00 -3.33243012e-01 3.05341929e-01
-5.73853552e-01 1.12304434e-01 -5.10437429e-01 2.30983093e-01
2.80137628e-01 7.03814685e-01 5.00902347e-03 -1.26979083e-01
2.43821591e-01 2.66796708e-01 6.88011229e-01 -7.58338392e-01
-3.61005604e-01 -1.34297356e-01 -4.51726019e-01 -4.72759515e-01
-1.94510549e-01 -5.58960915e-01 -9.78398860e-01 -2.57898152e-01
-8.17689374e-02 -8.74236077e-02 6.33223057e-01 9.23917055e-01
4.25254345e-01 4.70689684e-01 9.17331040e-01 -8.53438914e-01
-2.71243036e-01 -9.80090916e-01 -7.15210855e-01 6.32356346e-01
6.62936985e-01 -6.86166584e-01 -4.75882471e-01 4.49734032e-01] | [12.562405586242676, -0.21204376220703125] |
cfc5ee02-aba4-4217-8c9b-6310d8cb0455 | neural-implicit-surface-reconstruction-from | 2210.01548 | null | https://arxiv.org/abs/2210.01548v1 | https://arxiv.org/pdf/2210.01548v1.pdf | Neural Implicit Surface Reconstruction from Noisy Camera Observations | Representing 3D objects and scenes with neural radiance fields has become very popular over the last years. Recently, surface-based representations have been proposed, that allow to reconstruct 3D objects from simple photographs. However, most current techniques require an accurate camera calibration, i.e. camera parameters corresponding to each image, which is often a difficult task to do in real-life situations. To this end, we propose a method for learning 3D surfaces from noisy camera parameters. We show that we can learn camera parameters together with learning the surface representation, and demonstrate good quality 3D surface reconstruction even with noisy camera observations. | ['Patrik Huber', 'Sarthak Gupta'] | 2022-10-02 | null | null | null | null | ['camera-calibration'] | ['computer-vision'] | [ 2.71578729e-01 -2.75028050e-01 2.03055695e-01 -5.37292361e-01
-7.37596273e-01 -4.97449130e-01 3.03757995e-01 -4.20474559e-01
-8.34872574e-02 3.66562426e-01 -1.29941881e-01 1.30772546e-01
7.31873140e-03 -8.15047741e-01 -1.01820624e+00 -6.12411380e-01
3.76916796e-01 4.17575836e-01 7.86075220e-02 3.31529714e-02
1.60594240e-01 7.69864857e-01 -1.55166292e+00 -1.90148041e-01
5.46239674e-01 9.81284559e-01 4.80871230e-01 6.74347579e-01
-2.45762601e-01 3.57709348e-01 -5.32937586e-01 -2.60886908e-01
4.52786952e-01 -1.90299571e-01 -3.36969256e-01 7.34105825e-01
8.42362046e-01 -5.13676584e-01 -3.38585615e-01 1.27655494e+00
8.95413756e-02 4.54903878e-02 8.11671436e-01 -7.75636733e-01
-6.50054097e-01 -2.55191505e-01 -5.24488091e-01 -2.43405253e-01
4.31164712e-01 -6.81333467e-02 7.30116367e-01 -6.57867849e-01
5.13666630e-01 1.32245696e+00 6.62983894e-01 5.13540506e-01
-1.39011300e+00 -1.74974531e-01 5.77806830e-02 6.94922879e-02
-1.31800985e+00 -3.58972281e-01 1.29833817e+00 -3.55531842e-01
6.18901253e-01 5.16255992e-03 6.85245037e-01 1.11688590e+00
8.67970958e-02 6.69270575e-01 1.12280202e+00 -4.22933906e-01
2.67590165e-01 1.54492661e-01 -1.41939610e-01 5.64216852e-01
3.77193689e-01 -1.96486190e-01 -1.81980774e-01 6.41893819e-02
1.38946092e+00 3.24217618e-01 -6.10446930e-01 -7.19293356e-01
-1.03913605e+00 5.61376631e-01 6.24871850e-01 2.19549447e-01
-4.75452363e-01 4.54562604e-01 -1.67243004e-01 1.30747989e-01
6.16867721e-01 3.52515012e-01 -2.01015741e-01 1.08763531e-01
-4.94541496e-01 4.49117040e-03 6.93292141e-01 1.03668916e+00
9.94639695e-01 8.67986381e-02 7.55986273e-01 9.62653279e-01
3.25598270e-01 1.01355839e+00 1.31978229e-01 -1.30542064e+00
2.48887435e-01 4.74083930e-01 2.89373964e-01 -1.07601404e+00
-1.94386020e-01 -1.76171958e-01 -1.13955104e+00 3.14336926e-01
2.36377060e-01 7.98088163e-02 -7.90047348e-01 1.47261381e+00
1.96370319e-01 5.02124846e-01 2.23418102e-01 1.13028729e+00
5.39223552e-01 8.91064525e-01 -6.40632570e-01 1.03672349e-03
8.81379724e-01 -4.81470615e-01 -6.34469926e-01 -4.31258768e-01
2.76893526e-01 -6.73188627e-01 8.83231640e-01 7.23106921e-01
-1.06781268e+00 -4.81870741e-01 -9.67442274e-01 -1.12542696e-01
1.54886174e-03 5.08559197e-02 6.34402335e-01 5.70299685e-01
-9.68842208e-01 4.58999246e-01 -8.26787412e-01 -2.72064924e-01
2.11457938e-01 2.87310213e-01 -6.50832593e-01 -2.05977753e-01
-7.01877475e-01 9.54636335e-01 -4.43799570e-02 3.80925894e-01
-9.31554615e-01 -3.18377614e-01 -8.00468743e-01 -4.54556793e-02
4.42250848e-01 -7.78880656e-01 1.32972193e+00 -1.01988614e+00
-1.94395149e+00 1.07295132e+00 -1.76223367e-01 -9.86539945e-03
1.99679598e-01 -5.55768788e-01 -9.43614542e-02 2.08277404e-01
-4.35645163e-01 2.99588978e-01 9.11144316e-01 -1.62047803e+00
-6.38498664e-02 -5.95876276e-01 4.46428865e-01 3.21858346e-01
-8.22885986e-03 -3.70111287e-01 -5.23356616e-01 -1.72351956e-01
5.97779095e-01 -9.36887026e-01 -4.42072600e-01 4.83851939e-01
-2.27508590e-01 1.03855453e-01 6.99936509e-01 -3.95168871e-01
2.13843912e-01 -2.10059214e+00 4.87228334e-01 6.90916553e-02
-7.96242338e-03 1.38939783e-01 -8.32581446e-02 1.63551327e-02
1.88610435e-01 -7.75564164e-02 -3.46228153e-01 -5.67895830e-01
-2.24443614e-01 3.06034744e-01 -2.39219919e-01 6.81034625e-01
1.38000548e-01 5.52966297e-01 -6.32617533e-01 -9.22015756e-02
6.50194466e-01 9.27925646e-01 -5.06856382e-01 3.98034662e-01
-4.25287008e-01 5.48643172e-01 -5.91889143e-01 3.48738253e-01
8.43326688e-01 -3.84787172e-01 2.05594171e-02 -9.22058448e-02
3.41015309e-02 -8.90751854e-02 -1.32268775e+00 1.87716067e+00
-8.26401591e-01 7.07193494e-01 3.76138091e-01 -1.10944116e+00
1.32036996e+00 3.08557212e-01 3.88687104e-01 -4.63300079e-01
2.57043213e-01 2.73837328e-01 -7.17803359e-01 -4.77213442e-01
3.87120068e-01 -4.09383565e-01 2.51112860e-02 2.61574388e-01
-1.45615235e-01 -1.15333605e+00 -5.02926946e-01 -2.34886900e-01
6.54356956e-01 2.39693135e-01 3.46149862e-01 1.57941371e-01
5.80165803e-01 -2.29033083e-01 3.81446868e-01 3.32537204e-01
3.63870710e-01 9.75479066e-01 1.66485041e-01 -5.67774773e-01
-1.18690157e+00 -9.82690752e-01 -6.92322776e-02 -2.13074759e-02
5.35165191e-01 7.42761195e-02 -8.26359212e-01 -2.96668112e-01
-7.57835954e-02 4.29923415e-01 -3.09960842e-01 -2.77647525e-02
-6.09601378e-01 -3.08911830e-01 -7.91288819e-03 3.41970116e-01
5.09445786e-01 -5.41808128e-01 -6.62444949e-01 4.46199551e-02
-1.13722689e-01 -1.32364762e+00 -1.57953143e-01 8.35102573e-02
-1.27009892e+00 -1.01020801e+00 -9.00801837e-01 -6.20791495e-01
8.30023170e-01 9.88376558e-01 1.04613721e+00 9.48285684e-02
-2.39184484e-01 7.40630388e-01 -7.59958103e-02 -2.91111141e-01
-2.74893165e-01 -3.43464196e-01 1.02018960e-01 2.52123505e-01
1.00841038e-01 -8.08360755e-01 -3.19778919e-01 4.62517142e-01
-1.01336575e+00 2.07165375e-01 6.66123033e-01 5.41609168e-01
1.03427231e+00 1.24847135e-02 -7.78848957e-03 -8.56631458e-01
-8.62878636e-02 -2.16015130e-01 -1.18094468e+00 1.97782382e-01
9.65424925e-02 6.50625974e-02 7.89490819e-01 -3.65103096e-01
-1.20024705e+00 4.01517153e-01 -2.38159850e-01 -7.86128223e-01
-5.35658121e-01 1.65271461e-01 -3.69495153e-01 -2.37019554e-01
6.82740808e-01 1.48986936e-01 -5.48824146e-02 -7.64671803e-01
3.47162306e-01 7.99916565e-01 6.28687739e-01 -4.54771519e-01
9.63712275e-01 8.33407044e-01 2.67870724e-01 -1.21790540e+00
-1.18088078e+00 -4.59070742e-01 -9.43292677e-01 -2.43664369e-01
8.52756560e-01 -1.03113496e+00 -4.96481448e-01 6.48997724e-01
-1.48884726e+00 -4.02588934e-01 -9.07676071e-02 6.86039031e-01
-8.82824719e-01 4.27140325e-01 -1.93115532e-01 -9.07765031e-01
3.25794667e-02 -1.14258397e+00 1.28218997e+00 2.91785777e-01
1.67923391e-01 -1.04827750e+00 1.29932702e-01 2.59922206e-01
1.72946200e-01 3.12478006e-01 6.94665194e-01 4.26378757e-01
-1.16997707e+00 -2.48797968e-01 -2.76939869e-01 6.78987920e-01
3.39934260e-01 -1.03740767e-01 -1.26335764e+00 -1.29990563e-01
4.57544595e-01 -2.61520892e-01 5.67146838e-01 5.59968710e-01
1.29553151e+00 1.72684565e-01 -1.84904948e-01 1.06580853e+00
1.70546937e+00 7.46668279e-02 4.66981381e-01 1.80944074e-02
6.98804677e-01 5.40110290e-01 2.03813553e-01 1.70107573e-01
1.99083358e-01 9.07338083e-01 7.59371936e-01 4.82254773e-02
-1.72855288e-01 -2.74165988e-01 2.41277874e-01 9.31568325e-01
-9.63944495e-02 -2.55282074e-01 -7.01261342e-01 2.70324022e-01
-1.45321608e+00 -4.81110781e-01 -2.02129573e-01 2.24920440e+00
4.12925929e-01 1.04034483e-01 -4.41747338e-01 1.15062065e-01
6.01034343e-01 2.95051813e-01 -6.76406920e-01 -2.29360871e-02
-3.75635177e-01 -1.24963537e-01 3.71186823e-01 6.70586944e-01
-7.99414992e-01 8.29264343e-01 6.81489849e+00 8.77269804e-02
-1.21935081e+00 -5.93520962e-02 2.74655879e-01 1.75331622e-01
-6.37613714e-01 8.20349157e-02 -7.52011776e-01 1.64662257e-01
5.33763945e-01 1.93333533e-02 4.67680961e-01 8.05788934e-01
9.88211259e-02 -2.28967085e-01 -1.14289296e+00 1.57279694e+00
4.74667460e-01 -1.13427496e+00 1.39444873e-01 6.77939579e-02
8.57537389e-01 -5.68209961e-02 -5.11004962e-02 -2.47460276e-01
2.02317774e-01 -7.62986422e-01 5.62771678e-01 8.69963288e-01
7.30131388e-01 -4.05439824e-01 5.85766852e-01 6.47102714e-01
-7.05155373e-01 6.58678412e-02 -9.19073641e-01 3.82333472e-02
4.40413088e-01 1.03222215e+00 -3.66298676e-01 4.56045836e-01
7.99743652e-01 1.00198960e+00 -2.43872330e-01 1.28019428e+00
-4.49499160e-01 3.85191441e-01 -6.16783261e-01 4.84321779e-03
4.00113240e-02 -6.72119915e-01 5.57470679e-01 5.94157457e-01
6.71802223e-01 3.96237463e-01 -7.81363528e-03 9.42089856e-01
-2.79815972e-01 -1.49461240e-01 -1.03096247e+00 3.02108288e-01
2.15177741e-02 1.10137844e+00 -4.90548819e-01 -2.63600871e-02
-4.78198975e-01 1.11443877e+00 2.91294813e-01 4.00970995e-01
-4.27260578e-01 -8.91053751e-02 5.92866540e-01 3.80982794e-02
1.82267338e-01 -7.53042459e-01 1.97002869e-02 -1.47808087e+00
1.74534112e-01 -4.12158042e-01 -2.00560704e-01 -1.37724936e+00
-1.28776979e+00 3.93997818e-01 2.40237936e-02 -1.35173273e+00
-5.13160019e-04 -8.71100545e-01 -3.68408918e-01 6.70268178e-01
-1.70652437e+00 -7.09287941e-01 -6.45840049e-01 5.76239586e-01
5.52043617e-01 2.32457474e-01 7.72308886e-01 -1.48955658e-02
-2.92146415e-01 -5.97905517e-02 3.14565629e-01 -1.93463549e-01
5.43080568e-01 -1.20818424e+00 4.61833894e-01 5.06609678e-01
5.09502828e-01 3.60175788e-01 5.09684145e-01 -2.76070625e-01
-1.90686643e+00 -6.97411299e-01 4.41782802e-01 -4.70353931e-01
2.37575442e-01 -5.86218238e-01 -1.09036756e+00 7.32421398e-01
-7.59415478e-02 8.88883322e-02 3.05501908e-01 -9.63227898e-02
-3.52673888e-01 -4.20864046e-01 -1.00620866e+00 4.32093471e-01
1.13420463e+00 -7.05902576e-01 -5.59067965e-01 4.58324790e-01
5.45490384e-01 -5.91976047e-01 -7.50446141e-01 1.49694994e-01
3.09045225e-01 -9.41381633e-01 1.04009283e+00 -9.49678272e-02
4.16056626e-02 -3.80155861e-01 -4.13339466e-01 -1.54132986e+00
-4.01810184e-02 -5.24693668e-01 7.32542053e-02 7.61074066e-01
7.59048089e-02 -5.09238720e-01 9.18000221e-01 3.79753590e-01
-4.55440059e-02 -1.54350221e-01 -8.68809879e-01 -7.75174081e-01
-8.78998041e-02 -5.10430038e-01 5.46683550e-01 5.69390774e-01
-7.46563315e-01 3.23956579e-01 -6.00434124e-01 5.35688043e-01
8.16041350e-01 3.20386022e-01 1.05092800e+00 -1.66338122e+00
-3.00053656e-01 -2.88768530e-01 -6.25342250e-01 -1.72403049e+00
4.07187521e-01 -5.78066528e-01 3.10106099e-01 -1.63536823e+00
5.57572953e-02 -4.84558046e-01 2.92534113e-01 -1.04697555e-01
1.25366062e-01 1.51430160e-01 4.00145128e-02 3.12228680e-01
-4.36513633e-01 7.32160032e-01 1.31074917e+00 -1.10472016e-01
-2.18255818e-02 2.56463468e-01 -3.34714592e-01 1.03960145e+00
7.20176816e-01 -2.85062045e-01 -4.29513305e-01 -1.13804948e+00
4.01196241e-01 1.74903288e-01 5.67774236e-01 -1.05968583e+00
1.08597547e-01 -2.18767285e-01 3.65531921e-01 -5.45402646e-01
8.95004749e-01 -1.31473207e+00 3.87429804e-01 1.19968019e-02
-2.38020122e-02 -2.26288021e-01 -8.32188651e-02 8.37403476e-01
-2.93855101e-01 -5.16019166e-01 8.96210194e-01 -3.72323245e-01
-6.29436791e-01 2.97880799e-01 -1.28916234e-01 -3.30846995e-01
8.03909898e-01 -2.25820035e-01 1.47886360e-02 -6.26222134e-01
-5.02209783e-01 -1.85324684e-01 9.31278825e-01 3.08257580e-01
8.43124688e-01 -1.18178666e+00 -4.25927728e-01 3.83505046e-01
1.76570192e-02 6.52358055e-01 3.68723184e-01 2.94594347e-01
-8.87501597e-01 1.77412242e-01 -9.05802473e-02 -9.92946863e-01
-1.07091308e+00 4.67530787e-01 6.68133497e-01 3.60735178e-01
-9.51929510e-01 5.54858088e-01 3.44662458e-01 -5.58653235e-01
2.54508674e-01 -5.41599333e-01 -2.97539588e-02 -4.69450593e-01
4.27844226e-01 5.47421128e-02 -7.00108781e-02 -6.53745115e-01
-4.62640040e-02 1.51861203e+00 3.60078782e-01 -5.37996143e-02
1.57069182e+00 -2.62988687e-01 8.56508017e-02 7.16281354e-01
1.32315624e+00 -2.03983635e-01 -1.65586340e+00 -5.70431709e-01
-1.52695343e-01 -8.42218459e-01 4.19533849e-01 -2.09305272e-01
-1.25519395e+00 1.34307909e+00 4.56948698e-01 2.09457353e-02
1.05835927e+00 1.27409950e-01 8.44150424e-01 1.00994778e+00
7.66294837e-01 -6.83900177e-01 1.12293661e-01 2.81061381e-01
8.83050442e-01 -1.25982451e+00 3.51455547e-02 -6.12434685e-01
-2.81354994e-01 1.25096548e+00 2.79264927e-01 -3.98282200e-01
7.41016626e-01 9.04697254e-02 1.42144889e-01 -1.35502025e-01
-5.14252603e-01 2.02380624e-02 1.19494781e-01 6.07974529e-01
5.67794219e-03 1.81840709e-03 5.59993804e-01 2.99119502e-02
4.29926664e-02 -1.69114605e-01 8.24700594e-01 5.18451273e-01
-5.02961636e-01 -9.92042482e-01 -6.11804307e-01 5.20605184e-02
-1.23761594e-01 3.76695722e-01 -4.11168218e-01 6.23124540e-01
-2.61940718e-01 5.99070072e-01 1.19615614e-01 -6.30299374e-02
7.01654017e-01 -2.16397777e-01 8.46700311e-01 -6.98856413e-01
6.63793758e-02 9.55151170e-02 -2.52349615e-01 -4.74241316e-01
-7.23772287e-01 -6.74743533e-01 -9.70527172e-01 -4.42603901e-02
-4.64468151e-01 -5.96072932e-04 1.02857363e+00 6.32245839e-01
6.99753687e-02 -1.68765336e-02 9.01113391e-01 -1.29921830e+00
-4.44695204e-01 -6.25316858e-01 -8.43509793e-01 4.62391973e-01
5.69841802e-01 -8.81581604e-01 -7.00745940e-01 1.18927196e-01] | [9.050745010375977, -2.887497663497925] |
720db600-684a-4268-b379-2a134f9811c6 | deep-bayes-factor-scoring-for-authorship | 2008.10105 | null | https://arxiv.org/abs/2008.10105v1 | https://arxiv.org/pdf/2008.10105v1.pdf | Deep Bayes Factor Scoring for Authorship Verification | The PAN 2020 authorship verification (AV) challenge focuses on a cross-topic/closed-set AV task over a collection of fanfiction texts. Fanfiction is a fan-written extension of a storyline in which a so-called fandom topic describes the principal subject of the document. The data provided in the PAN 2020 AV task is quite challenging because authors of texts across multiple/different fandom topics are included. In this work, we present a hierarchical fusion of two well-known approaches into a single end-to-end learning procedure: A deep metric learning framework at the bottom aims to learn a pseudo-metric that maps a document of variable length onto a fixed-sized feature vector. At the top, we incorporate a probabilistic layer to perform Bayes factor scoring in the learned metric space. We also provide text preprocessing strategies to deal with the cross-topic issue. | ['Dorothea Kolossa', 'Robert M. Nickel', 'Julian Rupp', 'Benedikt Boenninghoff'] | 2020-08-23 | null | null | null | null | ['authorship-verification'] | ['natural-language-processing'] | [-5.55977151e-02 -2.80201316e-01 -1.35518357e-01 -4.30032790e-01
-1.25840020e+00 -7.66474247e-01 1.11016166e+00 2.76699424e-01
-1.92958206e-01 3.45207214e-01 5.13391078e-01 1.04081407e-02
-1.87782541e-01 -3.68557006e-01 -4.31606770e-01 -3.19276899e-01
3.03276449e-01 7.34930873e-01 -1.47061497e-01 1.80547684e-01
7.83257902e-01 1.18961543e-01 -1.13903534e+00 4.45816517e-01
5.02776027e-01 8.66368473e-01 -1.63494311e-02 9.85555589e-01
-2.24029452e-01 6.33292794e-01 -1.08134818e+00 -8.80070329e-01
4.29524451e-01 -3.91729444e-01 -8.77375841e-01 -9.54397023e-02
1.20069408e+00 -2.43033871e-01 -4.79725629e-01 9.42638338e-01
3.82985741e-01 1.80227645e-02 1.08149242e+00 -1.62334669e+00
-5.35787940e-01 1.29649293e+00 -6.68118894e-01 -1.94936898e-02
3.89662057e-01 -9.42804292e-02 1.63263619e+00 -1.10135543e+00
7.82840610e-01 1.31541014e+00 8.39722157e-01 3.85569841e-01
-1.27720213e+00 -9.69304562e-01 -1.94804341e-01 1.91644996e-01
-1.42826331e+00 -1.69573933e-01 8.94280732e-01 -9.25106347e-01
4.83838499e-01 9.21526328e-02 2.80078858e-01 1.47792840e+00
2.60334760e-01 9.69333529e-01 7.53791928e-01 -2.92881489e-01
1.02142632e-01 -4.10247035e-02 6.36268377e-01 4.59856063e-01
2.12125838e-01 -3.46632063e-01 -1.05629075e+00 -4.13054377e-01
-6.04799055e-02 -1.20517120e-01 -2.56017059e-01 -2.45841429e-01
-1.48648834e+00 8.23463738e-01 -3.83506306e-02 3.37922990e-01
2.22898677e-01 2.97325462e-01 9.07541871e-01 2.52989382e-01
2.03724891e-01 7.44233847e-01 -5.77301025e-01 -5.47166228e-01
-1.69263065e+00 7.80808687e-01 9.53440726e-01 8.70997548e-01
4.22065020e-01 -6.01200938e-01 -6.80093527e-01 6.86579168e-01
6.25496149e-01 3.32387567e-01 5.03758311e-01 -8.71787727e-01
9.66078341e-01 6.07380569e-01 9.20893401e-02 -9.40967619e-01
-2.38460988e-01 -2.88108796e-01 -6.42694473e-01 1.24387458e-01
6.21003985e-01 5.21299522e-03 -5.25656998e-01 1.56230676e+00
1.76390226e-04 -8.47068280e-02 -1.90078139e-01 7.06577301e-01
8.37943077e-01 6.45874083e-01 -2.50522226e-01 9.68969986e-02
1.46202600e+00 -1.14118743e+00 -9.05525684e-01 2.10471764e-01
5.70447326e-01 -9.79897618e-01 1.29535913e+00 6.94192469e-01
-5.98461688e-01 -4.30864781e-01 -1.50809741e+00 -6.25899732e-01
-6.18478000e-01 3.53480935e-01 -4.92166430e-02 5.90445399e-01
-7.04662442e-01 8.35380852e-01 -4.17196363e-01 -6.02213815e-02
5.69456518e-01 -1.61816016e-01 -4.46425140e-01 1.24653369e-01
-1.12909055e+00 7.08205223e-01 1.99315324e-01 -4.37137663e-01
-1.00721264e+00 -1.27578032e+00 -7.81856656e-01 8.59554484e-02
-4.70650904e-02 -3.90685916e-01 1.41863942e+00 -4.65648264e-01
-1.43545139e+00 1.05779576e+00 2.91274767e-02 -3.46470088e-01
9.90888655e-01 -6.78637028e-01 -3.08322310e-01 -1.10304795e-01
2.82782763e-01 2.80118465e-01 1.07332408e+00 -1.00347710e+00
-5.14146388e-01 -3.16652119e-01 -5.32381654e-01 -1.28793150e-01
-6.42464161e-01 3.59360725e-01 -3.89665663e-01 -1.27885306e+00
-3.28233421e-01 -7.22309589e-01 5.47933340e-01 1.47796065e-01
-8.84741604e-01 -7.23796248e-01 1.21865451e+00 -8.57200861e-01
1.76017404e+00 -2.14215183e+00 3.35769624e-01 1.42787933e-01
6.16578102e-01 -2.77362734e-01 -2.20167043e-04 4.59208608e-01
8.98307860e-02 1.97772890e-01 -2.70362169e-01 -9.90530133e-01
5.09885311e-01 -5.27264237e-01 -5.61214566e-01 7.13053524e-01
-1.57356411e-01 6.45871580e-01 -8.65815222e-01 -5.79211652e-01
-2.06969708e-01 2.79733598e-01 -5.85521944e-02 2.63267398e-01
-2.71433532e-01 -2.64803857e-01 -1.72440354e-02 4.90006119e-01
7.97326863e-01 -1.95296824e-01 -1.89675435e-01 -1.13127589e-01
2.92301551e-02 5.38045347e-01 -1.16688108e+00 2.07056212e+00
-7.34163746e-02 1.33491874e+00 -1.41737595e-01 -2.75215685e-01
9.87102211e-01 1.65822789e-01 3.51164073e-01 8.50738212e-02
3.09571832e-01 2.70172924e-01 -2.26293445e-01 -4.57088202e-02
9.59516227e-01 2.66320974e-01 -4.94074613e-01 1.06103849e+00
2.59042203e-01 -1.54623806e-01 3.95296603e-01 5.52228868e-01
1.24360156e+00 6.67322502e-02 1.29780725e-01 -5.29504836e-01
6.59914196e-01 -3.22431624e-01 3.59885246e-01 7.81375051e-01
-2.48288557e-01 1.16149259e+00 8.93055856e-01 -3.30954492e-01
-1.28952122e+00 -1.03108323e+00 -3.90752792e-01 1.18596995e+00
-1.47894636e-01 -1.15174353e+00 -7.05850065e-01 -1.13870513e+00
3.89910489e-01 8.80595684e-01 -1.03764427e+00 -1.17513938e-02
-4.23015028e-01 -3.86077553e-01 9.03987765e-01 3.54754031e-01
1.62100166e-01 -6.28246844e-01 -5.58849990e-01 -1.71131402e-01
-4.11500335e-02 -7.49791205e-01 -1.15996778e+00 2.24633858e-01
-3.22156876e-01 -1.23756826e+00 -9.29720044e-01 -5.88720024e-01
2.39629716e-01 -9.37649831e-02 1.13623726e+00 -3.33677590e-01
-2.51642823e-01 1.08545728e-01 -3.57378542e-01 -3.60487461e-01
-7.66430914e-01 4.03301686e-01 -2.51552463e-02 5.85325770e-02
5.84459841e-01 -3.72263938e-02 -3.25512111e-01 1.02168173e-01
-5.69306254e-01 -3.20307389e-02 5.80754243e-02 1.00752974e+00
2.00091615e-01 -4.05382276e-01 7.09759220e-02 -8.44477177e-01
8.46678853e-01 -3.59534085e-01 -6.87578380e-01 3.55721861e-01
-8.54179084e-01 1.46815509e-01 6.28819942e-01 -5.06629467e-01
-4.88526940e-01 -1.88521639e-01 3.81174505e-01 -4.70809728e-01
4.00886506e-01 2.89858907e-01 -5.12140095e-01 4.92489040e-01
7.80237079e-01 -5.23490049e-02 -1.03315888e-02 -7.51962245e-01
5.38131356e-01 1.06466198e+00 7.26195812e-01 -5.67132831e-01
1.06192112e+00 6.32768571e-02 -5.81075884e-02 -3.08746547e-01
-9.82658744e-01 -4.73927438e-01 -7.39817739e-01 -1.27489716e-01
5.93392372e-01 -7.50035822e-01 -5.41450202e-01 6.85856700e-01
-1.35551345e+00 -3.61479491e-01 -3.54210615e-01 1.26940414e-01
-5.57745755e-01 5.09919405e-01 -6.08892620e-01 -3.58324170e-01
-7.37762213e-01 -8.99043620e-01 1.43529165e+00 1.60064682e-01
-8.41132402e-01 -8.28691244e-01 4.39837456e-01 4.11158174e-01
2.98246201e-02 9.06376094e-02 8.37147415e-01 -1.13435686e+00
-4.51663174e-02 -3.80186558e-01 -4.16419506e-01 3.08851838e-01
2.11994983e-02 4.90854174e-01 -1.04333222e+00 -3.53019804e-01
-3.41071337e-01 -3.04289669e-01 9.72766757e-01 2.68922299e-02
1.30096090e+00 -2.61960089e-01 -2.68814772e-01 7.02044010e-01
9.66028988e-01 -3.81677836e-01 1.27825469e-01 6.64454401e-01
9.02146041e-01 3.85017425e-01 6.49097502e-01 8.11266243e-01
4.17946100e-01 8.44667315e-01 1.27139306e-02 4.38111454e-01
-4.74305540e-01 -5.26611328e-01 4.58863497e-01 5.01114547e-01
6.87480509e-01 -5.90097666e-01 -1.13792527e+00 3.92299414e-01
-1.67376781e+00 -9.59175885e-01 -2.56167829e-01 2.17511773e+00
1.15953636e+00 8.80567953e-02 1.39147565e-01 5.61449945e-01
7.46389806e-01 2.68916339e-01 -4.38071400e-01 -3.23559374e-01
-4.21817213e-01 -2.06811219e-01 2.88727790e-01 5.11583209e-01
-1.48335731e+00 7.51830697e-01 6.48169613e+00 9.48467970e-01
-9.05175149e-01 1.70573205e-01 3.59150827e-01 -1.76682845e-01
-3.09034109e-01 -9.88158882e-02 -1.07335711e+00 7.21097648e-01
1.05512500e+00 -6.03013039e-01 7.90508389e-02 8.27128053e-01
-8.70332196e-02 2.17218518e-01 -1.59110260e+00 1.24143970e+00
5.28831303e-01 -1.30521214e+00 9.20667052e-02 1.77516952e-01
7.49173582e-01 -1.47595167e-01 2.53606796e-01 2.20405743e-01
3.49760383e-01 -1.05320966e+00 1.13207603e+00 5.01817584e-01
1.13665724e+00 -6.80812776e-01 5.95271051e-01 -6.29444420e-03
-8.47240448e-01 -1.32738516e-01 -9.99818444e-02 3.12920421e-01
1.59263089e-02 7.70355403e-01 -8.23331058e-01 2.21995547e-01
5.42150199e-01 1.14356685e+00 -8.86638224e-01 1.03571343e+00
-3.59103203e-01 5.90415597e-01 -3.60406819e-03 -9.43033397e-02
-3.07257511e-02 1.24322154e-01 8.74040067e-01 1.49815381e+00
4.15290862e-01 -7.21469164e-01 -3.15119445e-01 9.65440810e-01
-7.60620475e-01 1.64758265e-01 -2.63034701e-01 -2.45060876e-01
4.19785261e-01 1.34764576e+00 -4.08348292e-01 -4.09964770e-01
-1.61660910e-01 1.22454298e+00 2.63844937e-01 -1.78940035e-02
-7.31653452e-01 -7.45323241e-01 8.51122439e-01 1.08045273e-01
2.94018537e-01 -8.94584432e-02 -8.01314056e-01 -1.28364837e+00
-1.14080727e-01 -9.69486296e-01 6.16655290e-01 -5.74641526e-01
-1.62396491e+00 3.29186589e-01 -1.78771064e-01 -1.22518444e+00
-1.97204381e-01 -7.02378392e-01 -5.46582103e-01 1.03414226e+00
-1.07370603e+00 -1.14956427e+00 -3.39162380e-01 2.98502535e-01
7.21957862e-01 -7.20053256e-01 7.83189833e-01 1.97884277e-01
-6.27536774e-01 1.32549465e+00 5.55675387e-01 5.60247421e-01
1.40062881e+00 -1.69927168e+00 6.21883750e-01 9.86028790e-01
3.78501773e-01 6.55638278e-01 8.98647428e-01 -8.13453197e-01
-1.23474073e+00 -1.10702443e+00 1.50168049e+00 -1.08739007e+00
1.01446688e+00 -7.65263319e-01 -7.99570560e-01 3.87061954e-01
1.87821910e-01 -4.90494147e-02 9.82825160e-01 3.20985943e-01
-1.06080580e+00 -5.33304885e-02 -1.04613352e+00 4.61827397e-01
5.99733770e-01 -7.85978973e-01 -9.70771730e-01 3.26136500e-01
4.94460136e-01 -2.96797901e-01 -8.94261539e-01 -1.33893102e-01
7.92775214e-01 -3.86173278e-01 5.24084866e-01 -7.10419297e-01
9.50058937e-01 -4.22906786e-01 -3.09459329e-01 -1.00326180e+00
-2.46385306e-01 -1.05719328e+00 -7.54873037e-01 1.67168343e+00
3.22331131e-01 1.15695290e-01 7.97763288e-01 4.82161731e-01
6.84059188e-02 -2.74792075e-01 -1.02249646e+00 -7.36533642e-01
4.52084273e-01 -4.85154033e-01 5.68023086e-01 9.72565174e-01
3.12883526e-01 4.98798490e-01 -3.43465716e-01 -2.83975393e-01
8.67628515e-01 2.05589548e-01 9.99720812e-01 -1.59047353e+00
-1.73357651e-01 -9.91217852e-01 -4.57688391e-01 -8.53279054e-01
2.83696711e-01 -1.19105935e+00 1.83427900e-01 -1.17997301e+00
7.59989023e-01 -1.31738007e-01 -2.18926400e-01 2.34674886e-01
-3.30961287e-01 1.22971751e-01 2.87308127e-01 4.53886032e-01
-6.42351151e-01 6.19545817e-01 6.64086461e-01 -5.69142938e-01
8.74955282e-02 -8.39617103e-02 -6.69986427e-01 2.88838476e-01
3.74089628e-01 -6.32170081e-01 -5.19637801e-02 -2.32589453e-01
3.07512432e-01 -4.15022075e-01 4.48413454e-02 -7.99242079e-01
3.10750812e-01 2.22711638e-01 2.65945017e-01 -9.70269978e-01
1.16642928e-02 -5.32464147e-01 -2.64808863e-01 1.96037337e-01
-8.88330221e-01 2.07760096e-01 -1.24343790e-01 5.50445557e-01
-1.01243496e-01 -3.25492948e-01 7.56654263e-01 4.13765073e-01
4.05983627e-03 1.35148719e-01 -1.85386822e-01 3.54305238e-01
9.09714162e-01 3.64543311e-02 -5.25267899e-01 -9.20244902e-02
-1.52241558e-01 2.62902647e-01 5.30444264e-01 8.91617417e-01
3.56661916e-01 -1.32814789e+00 -1.05760241e+00 -4.22964506e-02
3.75879705e-01 -2.22125709e-01 -1.90130010e-01 4.11970288e-01
-6.14848435e-01 4.04544652e-01 -3.70010622e-02 -4.84793872e-01
-1.32709134e+00 5.30432284e-01 1.92290023e-01 -4.66536045e-01
-7.05434322e-01 1.07537234e+00 -1.16736293e-01 -3.37946653e-01
6.74125314e-01 -1.02967806e-01 -1.74895510e-01 4.46168542e-01
9.71556962e-01 7.21495211e-01 -1.68836359e-02 -7.74919987e-01
-3.67594838e-01 3.84754121e-01 -4.10014838e-01 -4.86571848e-01
1.22109485e+00 -2.29705684e-02 4.20652283e-03 8.43848526e-01
1.51015449e+00 2.50212401e-01 -1.33416367e+00 -3.53239924e-01
4.12397087e-01 -4.18180406e-01 2.13286832e-01 -1.04392409e+00
-7.08666980e-01 9.68031049e-01 4.61872786e-01 -2.11498007e-01
6.26798570e-01 1.79814976e-02 6.79802418e-01 2.51202255e-01
-3.10848176e-01 -1.32979810e+00 1.29112154e-01 6.17093444e-01
1.18724048e+00 -1.16783464e+00 2.59202272e-01 1.05986536e-01
-4.87111211e-01 1.49809206e+00 3.70907068e-01 1.42594382e-01
7.22053826e-01 4.04477060e-01 6.47418872e-02 -5.62956594e-02
-7.04862475e-01 5.71293712e-01 6.20483339e-01 2.52996445e-01
4.85204875e-01 -4.73024845e-02 -2.59416729e-01 1.23916328e+00
-7.67859638e-01 -3.19209009e-01 7.19335794e-01 5.08271337e-01
-1.07292093e-01 -1.09455013e+00 -3.76852006e-01 5.65018713e-01
-5.11588335e-01 -1.19715430e-01 -8.37881446e-01 4.51015919e-01
4.41938685e-03 6.74897492e-01 1.42354116e-01 -5.62621057e-01
8.03200826e-02 2.92047769e-01 1.66214168e-01 -5.37446022e-01
-9.76360083e-01 -1.83570296e-01 -2.40950823e-01 -2.47893542e-01
2.76249886e-01 -1.32063794e+00 -9.86065269e-01 -4.20657367e-01
-2.57004239e-02 2.59709269e-01 7.95325518e-01 8.07805359e-01
3.16760689e-01 2.42461488e-01 5.69127023e-01 -5.29337108e-01
-8.26227009e-01 -1.38582444e+00 -6.68064833e-01 5.66467345e-01
4.66815680e-01 -5.07500052e-01 -6.66864097e-01 1.54509827e-01] | [9.611775398254395, 10.559517860412598] |
90b61869-f2f4-4065-9682-988c4dfc629c | patchnets-patch-based-generalizable-deep | 2008.01639 | null | https://arxiv.org/abs/2008.01639v2 | https://arxiv.org/pdf/2008.01639v2.pdf | PatchNets: Patch-Based Generalizable Deep Implicit 3D Shape Representations | Implicit surface representations, such as signed-distance functions, combined with deep learning have led to impressive models which can represent detailed shapes of objects with arbitrary topology. Since a continuous function is learned, the reconstructions can also be extracted at any arbitrary resolution. However, large datasets such as ShapeNet are required to train such models. In this paper, we present a new mid-level patch-based surface representation. At the level of patches, objects across different categories share similarities, which leads to more generalizable models. We then introduce a novel method to learn this patch-based representation in a canonical space, such that it is as object-agnostic as possible. We show that our representation trained on one category of objects from ShapeNet can also well represent detailed shapes from any other category. In addition, it can be trained using much fewer shapes, compared to existing approaches. We show several applications of our new representation, including shape interpolation and partial point cloud completion. Due to explicit control over positions, orientations and scales of patches, our representation is also more controllable compared to object-level representations, which enables us to deform encoded shapes non-rigidly. | ['Michael Zollhöfer', 'Christian Theobalt', 'Carsten Stoll', 'Edgar Tretschk', 'Ayush Tewari', 'Vladislav Golyanik'] | 2020-08-04 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2547_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123610290.pdf | eccv-2020-8 | ['point-cloud-completion'] | ['computer-vision'] | [ 1.89730134e-02 4.00303483e-01 2.36970410e-01 -3.15926433e-01
-7.15436518e-01 -7.69379497e-01 6.16991162e-01 5.34025952e-02
3.52584600e-01 4.79174584e-01 -4.96959835e-02 2.54783064e-01
-3.03220819e-03 -1.16652274e+00 -1.19915223e+00 -6.14896297e-01
1.60066634e-01 7.61015832e-01 4.46087599e-01 -3.35469425e-01
3.59989442e-02 1.18775415e+00 -1.47770715e+00 4.02636766e-01
8.60943973e-01 9.23503816e-01 1.61048859e-01 2.63964772e-01
-2.17144683e-01 8.74566510e-02 -1.33029357e-01 -1.90556169e-01
4.37790900e-01 3.68131362e-02 -6.93535745e-01 4.28596705e-01
8.43616605e-01 -2.20781431e-01 -3.58784646e-01 8.24393153e-01
9.17106643e-02 1.09224625e-01 9.47748125e-01 -7.99519360e-01
-1.10320103e+00 -6.66431487e-02 -3.31758201e-01 -7.98905313e-01
1.75745443e-01 1.23610431e-02 9.05956805e-01 -1.20638430e+00
8.21262240e-01 1.42976010e+00 8.81945312e-01 5.97369373e-01
-1.79345369e+00 -3.12712312e-01 2.23132238e-01 -3.68590564e-01
-1.43996215e+00 -2.81155169e-01 1.09753251e+00 -5.39281785e-01
7.08271503e-01 2.88329870e-01 7.02689469e-01 5.82949936e-01
-8.75046030e-02 6.50507569e-01 8.76231015e-01 -2.83172607e-01
2.06743330e-01 -2.31730901e-02 -2.85089910e-01 6.59974992e-01
3.31935138e-01 -3.33650410e-02 7.40028471e-02 -3.59783053e-01
1.62799704e+00 4.40225661e-01 -2.78225482e-01 -1.14104700e+00
-1.26152372e+00 7.74265528e-01 9.71870303e-01 2.87265778e-01
-2.54086614e-01 4.39521998e-01 -1.18467599e-01 5.32827228e-02
6.16919041e-01 4.53530252e-01 -6.17855728e-01 4.46882933e-01
-7.22154379e-01 4.26284790e-01 7.49158859e-01 1.22194183e+00
1.24243093e+00 2.25069344e-01 1.30108103e-01 8.88749063e-01
3.75261188e-01 6.25178099e-01 -1.42054200e-01 -1.20258570e+00
8.02734420e-02 7.66116083e-01 2.64580667e-01 -1.22805059e+00
-1.70181081e-01 -2.13491306e-01 -9.78996992e-01 6.30658090e-01
1.84947401e-01 3.69441479e-01 -1.17889190e+00 1.67078507e+00
3.22443962e-01 2.48644769e-01 -2.96975642e-01 7.67199039e-01
7.56832302e-01 6.52709186e-01 -2.91886449e-01 2.20024228e-01
1.16980219e+00 -5.33644855e-01 -2.15811074e-01 1.28509119e-01
2.73816705e-01 -6.06184483e-01 1.01645529e+00 2.77440757e-01
-1.20209002e+00 -4.32746172e-01 -8.22455347e-01 -3.44100416e-01
-2.59362996e-01 -7.47134760e-02 7.07925975e-01 1.10405348e-01
-1.24656916e+00 9.54519928e-01 -1.00325859e+00 -2.60800570e-01
7.50466883e-01 4.17132676e-01 -4.90454257e-01 -2.68457178e-02
-4.92910564e-01 6.42400384e-01 -3.39350998e-02 -1.13885909e-01
-7.13817954e-01 -8.92681539e-01 -9.63648856e-01 6.51048571e-02
-5.71296290e-02 -8.75202417e-01 8.47821593e-01 -9.98406351e-01
-1.41507864e+00 9.57369030e-01 -2.21807808e-01 5.78088202e-02
2.84223229e-01 2.55178779e-01 1.30082831e-01 4.53549959e-02
-8.09904039e-02 8.47211123e-01 1.00467122e+00 -1.71374393e+00
-1.95310600e-02 -4.03330892e-01 2.15273798e-01 1.42179485e-02
2.79470757e-02 -4.25506175e-01 -2.59682298e-01 -9.70265508e-01
5.83063483e-01 -8.83711219e-01 -4.80708808e-01 9.12653744e-01
2.03791019e-02 -2.02294484e-01 8.14750016e-01 -5.69772243e-01
4.31390554e-01 -2.16687536e+00 3.87698591e-01 2.83631772e-01
3.19469899e-01 9.60909948e-03 -3.62215102e-01 2.95619398e-01
1.53071620e-02 2.54209429e-01 -8.08861852e-01 -3.64556074e-01
2.97260750e-02 6.92164898e-01 -3.45350295e-01 5.02602041e-01
6.62867963e-01 1.01593578e+00 -7.34612942e-01 -1.32373214e-01
1.57629684e-01 1.00364566e+00 -9.58951771e-01 1.01975568e-01
-4.97519583e-01 8.11650038e-01 -5.90449393e-01 6.36637986e-01
1.04640806e+00 -2.86723971e-01 -1.09303810e-01 -3.27879876e-01
-1.38751678e-02 2.56184697e-01 -1.23599660e+00 1.91918194e+00
-6.69458330e-01 3.49966884e-01 4.42725867e-01 -1.22537577e+00
1.18301177e+00 4.12757218e-01 5.58368802e-01 -2.82157660e-01
-2.34470114e-01 3.42489868e-01 -3.62456620e-01 6.01517409e-02
2.49253541e-01 -2.79391974e-01 1.49115041e-01 2.42382690e-01
-5.16450815e-02 -7.19049335e-01 -4.01103139e-01 -8.24686959e-02
7.62092769e-01 4.51374769e-01 5.17838821e-02 -4.02224302e-01
3.07558239e-01 -6.98505864e-02 5.84023535e-01 2.40502745e-01
4.81878340e-01 1.12253952e+00 1.90828983e-02 -7.09829926e-01
-1.23880684e+00 -1.34915042e+00 -6.47976995e-01 5.79937398e-01
2.01624647e-01 -1.05029404e-01 -4.39247012e-01 -3.86906058e-01
5.77639997e-01 3.06140989e-01 -6.27646685e-01 1.03465132e-01
-9.05397296e-01 -1.20347433e-01 -1.18749719e-02 8.35853577e-01
9.94884074e-02 -1.06879616e+00 -1.37489021e-01 2.53361225e-01
2.93297201e-01 -8.33799422e-01 -4.90899324e-01 -1.97471187e-01
-1.46106958e+00 -1.03193605e+00 -8.48276317e-01 -1.00577724e+00
1.26091719e+00 2.35896304e-01 1.13595939e+00 3.43255162e-01
-2.18707740e-01 6.44117773e-01 -1.41175255e-01 -4.12805527e-02
-3.27910602e-01 -2.35762239e-01 -1.55146986e-01 2.77919114e-01
-1.91154659e-01 -1.13986158e+00 -6.66804075e-01 3.33920062e-01
-1.03073466e+00 -2.07505561e-02 4.64079529e-01 7.13720322e-01
1.14248240e+00 -3.50861907e-01 4.78000611e-01 -9.00095582e-01
1.14367068e-01 -3.17901939e-01 -5.82606733e-01 1.82867631e-01
-1.33883044e-01 2.27834836e-01 6.18188739e-01 -5.94098032e-01
-8.82110357e-01 3.39038074e-01 -2.08078563e-01 -7.26707339e-01
-2.70132810e-01 1.84645504e-01 -2.14077264e-01 -5.01762331e-01
6.98188424e-01 2.23665103e-01 2.56806254e-01 -9.61924970e-01
7.18210757e-01 2.70987928e-01 4.16200429e-01 -9.94244277e-01
1.10188556e+00 8.36263001e-01 1.68555036e-01 -1.09350789e+00
-3.97942930e-01 -1.74520954e-01 -1.02359593e+00 3.38564306e-01
5.91353714e-01 -8.41086805e-01 -4.96756554e-01 2.32833236e-01
-1.44440830e+00 -3.82034570e-01 -7.92086542e-01 1.24406688e-01
-8.95525157e-01 3.12531501e-01 -5.26569784e-01 -4.74956244e-01
-1.13256566e-01 -1.00414324e+00 1.40374625e+00 -1.15297385e-01
-1.09106526e-01 -1.18246174e+00 -1.99593790e-02 -1.27823606e-01
4.58913118e-01 6.79471135e-01 9.31658506e-01 -7.70706683e-02
-1.07510877e+00 -1.77072465e-01 -1.81142151e-01 3.67216110e-01
4.23460662e-01 7.94715583e-02 -7.27062106e-01 -4.95223731e-01
-9.42957215e-03 -1.47998184e-01 7.18659103e-01 2.95319229e-01
1.57463622e+00 -5.39444327e-01 -4.10050362e-01 1.03314817e+00
1.37669718e+00 -2.38741770e-01 7.82172620e-01 -1.20241612e-01
1.02944708e+00 4.69934732e-01 5.21869101e-02 1.90557420e-01
3.73322397e-01 8.10777664e-01 5.22113383e-01 -1.43561274e-01
-3.60297024e-01 -3.18494678e-01 3.32482834e-03 9.45032716e-01
-4.24985021e-01 3.93804908e-01 -6.47596121e-01 5.52295506e-01
-1.66856039e+00 -7.07664609e-01 -1.55319914e-01 2.30889463e+00
7.79591680e-01 -3.88696790e-01 4.51041060e-03 -6.34405464e-02
5.34824371e-01 -3.07252631e-02 -6.22520208e-01 -3.47664893e-01
-9.60737169e-02 6.60340726e-01 2.70876110e-01 5.18596232e-01
-8.44062150e-01 8.40795994e-01 6.84321117e+00 7.23291993e-01
-1.21053672e+00 5.56504577e-02 3.02152425e-01 2.22913519e-01
-9.54912961e-01 1.49967849e-01 -3.71179938e-01 1.44697160e-01
2.37129331e-01 -1.58149973e-01 5.36794543e-01 9.67746258e-01
-2.14211896e-01 5.78896463e-01 -1.30537522e+00 9.61066604e-01
-1.29030332e-01 -1.63834858e+00 4.66255903e-01 1.50270462e-01
9.87481952e-01 -8.99182186e-02 -7.23642483e-02 1.03061780e-01
2.24386960e-01 -1.14520907e+00 7.86102295e-01 6.85129881e-01
1.12120712e+00 -5.73182344e-01 -8.42484925e-03 4.83045697e-01
-1.47932565e+00 3.99383932e-01 -9.98228729e-01 8.82331282e-02
-3.77183333e-02 4.67579156e-01 -4.77756709e-01 6.52091265e-01
4.30546165e-01 9.89240468e-01 -2.48714387e-01 9.91721034e-01
-5.05426899e-02 2.66052306e-01 -5.85271657e-01 3.63220483e-01
-2.22121879e-01 -5.41551709e-01 4.77621078e-01 8.48219931e-01
4.91222471e-01 4.19683158e-01 3.32496434e-01 1.53460741e+00
-1.89971909e-01 -8.52126442e-03 -8.49877298e-01 1.93760708e-01
3.01964939e-01 1.27328360e+00 -5.07707655e-01 -2.23834142e-01
-5.15988708e-01 7.72121310e-01 7.16384649e-01 4.56657350e-01
-4.38197076e-01 -1.02721639e-01 8.42356980e-01 6.85643554e-01
5.84665120e-01 -5.67725778e-01 -4.99891222e-01 -1.25412881e+00
1.02067292e-01 -4.98838544e-01 -2.09788531e-01 -7.69735694e-01
-1.54448569e+00 5.21847069e-01 -1.08896978e-01 -1.43129838e+00
2.21610010e-01 -7.44285047e-01 -5.99916637e-01 9.25246656e-01
-1.41561007e+00 -1.33697903e+00 -1.85664669e-01 6.80046797e-01
2.94331491e-01 2.52679169e-01 1.02159607e+00 1.60919502e-01
1.62616700e-01 3.66962731e-01 1.31933033e-01 2.36204237e-01
3.17158192e-01 -1.20032084e+00 5.48304677e-01 3.27101260e-01
2.92497098e-01 6.88488424e-01 1.79900944e-01 -5.50823331e-01
-1.67373502e+00 -1.24955249e+00 2.70693690e-01 -7.15937793e-01
2.84705788e-01 -3.42856050e-01 -1.47531497e+00 9.15242493e-01
-3.06079000e-01 6.38321042e-01 1.57350749e-01 8.53073224e-03
-5.81910551e-01 -1.44511849e-01 -1.36250985e+00 4.75597411e-01
1.28178322e+00 -5.89634717e-01 -7.40374744e-01 3.51083219e-01
7.75748551e-01 -5.33801198e-01 -1.11574709e+00 5.92323899e-01
3.47898602e-01 -7.14867711e-01 1.34201133e+00 -6.62103653e-01
2.69805610e-01 -3.38278145e-01 -2.61348158e-01 -1.39153302e+00
-7.69024014e-01 -4.67137665e-01 -2.39941135e-01 9.82197940e-01
2.07487732e-01 -8.73652697e-01 7.18378782e-01 8.25255454e-01
-3.78832281e-01 -9.31669652e-01 -8.49879682e-01 -1.03222871e+00
6.33786976e-01 -1.23662606e-01 9.18567419e-01 1.04503477e+00
-4.49448287e-01 -1.87328830e-01 -1.52676105e-02 2.31341660e-01
7.44476497e-01 5.46609759e-01 6.74639165e-01 -1.71139920e+00
-1.77078217e-01 -4.01013345e-01 -5.69501519e-01 -1.54517317e+00
1.80116028e-01 -1.37862849e+00 -2.69800853e-02 -1.76189244e+00
-9.23340246e-02 -1.07983184e+00 1.54932681e-02 5.98632634e-01
1.90459877e-01 4.89814848e-01 1.82959642e-02 3.96759570e-01
3.32488790e-02 8.80073547e-01 1.93302238e+00 -2.15424865e-01
-8.66067708e-02 -6.49409518e-02 -6.69611812e-01 8.93198490e-01
5.42454302e-01 -3.18685144e-01 -1.20581366e-01 -7.24869251e-01
5.35184853e-02 -3.14337686e-02 5.70230842e-01 -6.91516817e-01
-6.72735944e-02 -2.42609531e-01 4.57037956e-01 -4.49510127e-01
6.54439211e-01 -9.66173887e-01 5.21177649e-01 1.09608404e-01
-5.20160645e-02 -3.73036623e-01 1.71694294e-01 6.12694025e-01
-1.31782293e-01 -2.01534815e-02 9.46525097e-01 -3.42487842e-01
-1.30769208e-01 8.36770236e-01 2.23056048e-01 -6.33998262e-03
9.32696939e-01 -4.81698722e-01 8.82731751e-03 -2.06535965e-01
-9.40286219e-01 -1.68992937e-01 1.08370113e+00 2.02824563e-01
8.45467508e-01 -1.91070628e+00 -7.60027826e-01 4.90179092e-01
6.35948777e-02 8.37274909e-01 5.07934503e-02 5.55622935e-01
-7.00201750e-01 1.10226013e-01 -2.13507265e-01 -9.60357428e-01
-6.71666026e-01 5.90118289e-01 3.89931649e-01 2.32077137e-01
-1.08929515e+00 5.93493104e-01 7.85728514e-01 -9.04526651e-01
-1.46214306e-01 -5.60932696e-01 2.87640989e-01 -4.21663225e-01
2.44688228e-01 5.29227369e-02 7.47880340e-02 -7.79943526e-01
-2.59292811e-01 1.23153162e+00 2.57082880e-01 3.32174301e-01
1.62017822e+00 3.02356511e-01 -4.34128672e-01 3.06952566e-01
1.20372915e+00 1.30280524e-01 -1.68709695e+00 -3.23232055e-01
-2.83960879e-01 -7.82723784e-01 -2.90851772e-01 -3.13491106e-01
-1.34590435e+00 9.53557253e-01 2.51856446e-03 2.30716109e-01
7.87703812e-01 3.38930100e-01 6.71915114e-01 2.34153003e-01
8.59667897e-01 -5.26251853e-01 3.08994446e-02 5.00687838e-01
1.61522388e+00 -1.03921092e+00 -3.65533195e-02 -8.05174053e-01
-2.31215581e-01 1.22246170e+00 2.76148885e-01 -8.30358922e-01
9.16688561e-01 1.88072816e-01 -3.60768676e-01 -2.51112819e-01
-5.35087585e-01 -7.61694163e-02 6.76388860e-01 9.34630275e-01
3.96238595e-01 2.93870062e-01 8.22145194e-02 3.41882318e-01
-4.76676747e-02 -2.44969755e-01 3.09795588e-01 7.65368879e-01
-4.95999545e-01 -1.34380972e+00 -4.71748471e-01 4.45619315e-01
7.86602423e-02 2.40407899e-01 -2.91451514e-01 7.14286387e-01
-5.68416193e-02 3.20776641e-01 2.64712572e-01 3.43053974e-02
5.40381134e-01 -1.75053030e-01 9.34562027e-01 -9.78136957e-01
-6.85594752e-02 -5.12410253e-02 -3.44119161e-01 -6.47659302e-01
-3.76348943e-01 -5.01103461e-01 -1.44833219e+00 -2.53621131e-01
-1.02405407e-01 -8.22815746e-02 4.79023576e-01 5.47019780e-01
7.18645573e-01 1.68904081e-01 6.57596171e-01 -1.60778201e+00
-5.02750039e-01 -5.78674793e-01 -6.94039464e-01 8.36566031e-01
2.60833770e-01 -9.55562592e-01 -3.35042030e-01 1.36759788e-01] | [8.668328285217285, -3.637852191925049] |
14a3377e-bb87-497a-b43c-3194887c6fe1 | scalable-privacy-preserving-cancer-type | 2204.05496 | null | https://arxiv.org/abs/2204.05496v1 | https://arxiv.org/pdf/2204.05496v1.pdf | Scalable privacy-preserving cancer type prediction with homomorphic encryption | Machine Learning (ML) alleviates the challenges of high-dimensional data analysis and improves decision making in critical applications like healthcare. Effective cancer type from high-dimensional genetic mutation data can be useful for cancer diagnosis and treatment, if the distinguishable patterns between cancer types are identified. At the same time, analysis of high-dimensional data is computationally expensive and is often outsourced to cloud services. Privacy concerns in outsourced ML, especially in the field of genetics, motivate the use of encrypted computation, like Homomorphic Encryption (HE). But restrictive overheads of encrypted computation deter its usage. In this work, we explore the challenges of privacy preserving cancer detection using a real-world dataset consisting of more than 2 million genetic information for several cancer types. Since the data is inherently high-dimensional, we explore smaller ML models for cancer prediction to enable fast inference in the privacy preserving domain. We develop a solution for privacy preserving cancer inference which first leverages the domain knowledge on somatic mutations to efficiently encode genetic mutations and then uses statistical tests for feature selection. Our logistic regression model, built using our novel encoding scheme, achieves 0.98 micro-average area under curve with 13% higher test accuracy than similar studies. We exhaustively test our model's predictive capabilities by analyzing the genes used by the model. Furthermore, we propose a fast matrix multiplication algorithm that can efficiently handle high-dimensional data. Experimental results show that, even with 40,000 features, our proposed matrix multiplication algorithm can speed up concurrent inference of multiple individuals by approximately 10x and inference of a single individual by approximately 550x, in comparison to standard matrix multiplication. | ['Michail Maniatakos', 'Mark Gerstein', 'Leo Chen', 'Gamze Gursoy', 'Eduardo Chielle', 'Esha Sarkar'] | 2022-04-12 | null | null | null | null | ['type-prediction'] | ['computer-code'] | [ 2.93313295e-01 -8.25806856e-02 -4.36830163e-01 -4.41203088e-01
-1.17797947e+00 -8.32839429e-01 1.00960284e-02 5.66702783e-01
-7.64521658e-01 8.95454049e-01 6.25354722e-02 -8.31060469e-01
5.10144718e-02 -9.07893538e-01 -7.19499767e-01 -9.20419574e-01
-1.57124341e-01 3.08110088e-01 -3.35461199e-01 2.69889146e-01
-1.95954651e-01 4.56274927e-01 -1.23852813e+00 5.85799992e-01
8.38658392e-01 9.18095887e-01 -1.92812905e-01 7.71178663e-01
1.67007908e-01 8.38447809e-02 -2.95973063e-01 -8.53453457e-01
4.24039334e-01 -1.01650178e-01 -5.26032686e-01 -5.26537955e-01
5.48992492e-02 -5.12331784e-01 -3.88342381e-01 1.11755419e+00
5.99412978e-01 -5.97223520e-01 2.54368842e-01 -1.33427835e+00
-2.12559309e-02 3.14431608e-01 -2.12794915e-01 -2.09813476e-01
2.93414772e-01 1.87194452e-01 9.11823630e-01 -3.52684408e-01
9.92058158e-01 7.06779420e-01 9.19962108e-01 6.78621590e-01
-1.34098125e+00 -6.73031032e-01 -6.55304074e-01 -2.77693570e-02
-1.66111302e+00 -2.79101253e-01 3.69071960e-03 -1.26729295e-01
9.19002056e-01 1.00068593e+00 7.46064186e-01 7.78837025e-01
6.60231173e-01 6.04841530e-01 9.46640313e-01 -3.42666246e-02
4.12055373e-01 2.09646985e-01 9.55315009e-02 8.51158023e-01
7.27608800e-01 4.97340001e-02 -4.91499394e-01 -1.12857902e+00
-4.90395203e-02 5.33759713e-01 -3.80608201e-01 -2.97687829e-01
-1.02833247e+00 8.09580207e-01 6.05373681e-02 -2.28096083e-01
-1.44648463e-01 1.05225377e-01 7.13432491e-01 4.02153611e-01
3.20900977e-01 1.68382257e-01 -6.26083970e-01 -3.83469239e-02
-7.54885495e-01 3.60647351e-01 9.36669588e-01 8.19114447e-01
4.88115489e-01 -8.35898101e-01 -2.51933068e-01 2.62732506e-01
-1.48943424e-01 3.91702265e-01 4.55623209e-01 -4.45713997e-01
2.85057336e-01 7.22706437e-01 -2.67522275e-01 -7.44465947e-01
-5.00313759e-01 -2.88165271e-01 -1.31311178e+00 -2.93333143e-01
1.32852256e-01 -2.40255252e-01 -4.34505224e-01 1.65024674e+00
7.16434300e-01 3.47315744e-02 4.55925256e-01 6.16976321e-01
4.78291631e-01 2.12555885e-01 -2.37448681e-02 -2.49013871e-01
1.86861634e+00 -1.92279994e-01 -6.24622107e-01 4.58795309e-01
1.39446580e+00 -2.41295770e-01 6.94013238e-01 1.35248750e-01
-8.45347464e-01 4.72506523e-01 -8.09992969e-01 -4.63558555e-01
-5.02090216e-01 -5.49732596e-02 1.28955877e+00 9.91040170e-01
-8.80071163e-01 3.60648155e-01 -9.80770767e-01 -2.44624406e-01
1.00492084e+00 7.84529924e-01 -8.23645175e-01 -5.72200835e-01
-1.11548746e+00 2.70105630e-01 2.64332592e-01 -3.43705148e-01
-5.70307851e-01 -1.35380495e+00 -8.14658284e-01 1.35824457e-01
1.43103108e-01 -9.91081476e-01 6.36036396e-01 -4.39284742e-02
-8.78222823e-01 9.55633223e-01 -4.19868678e-01 -7.19103813e-01
5.02983093e-01 1.85726762e-01 -3.70194703e-01 4.32221293e-02
-1.24517061e-01 2.69184262e-01 2.44285673e-01 -2.85753280e-01
-7.82828450e-01 -7.81214356e-01 -4.37831134e-01 -1.21558867e-02
-7.27955580e-01 -1.57259449e-01 -4.11549658e-01 -3.98864180e-01
-1.08425289e-01 -1.14221776e+00 -3.68704379e-01 4.47452873e-01
-6.19954884e-01 2.90899336e-01 8.22614551e-01 -7.58413672e-01
1.11393070e+00 -2.38408875e+00 8.34815726e-02 5.75415134e-01
5.23953199e-01 1.34286746e-01 2.54837543e-01 2.49293402e-01
3.32194746e-01 2.10613236e-01 -4.67927784e-01 -2.87306346e-02
-1.26679614e-01 1.27576962e-01 -8.87789577e-03 7.72529602e-01
-7.76415616e-02 1.16768670e+00 -5.45993745e-01 -4.06601846e-01
-4.56722915e-01 6.03738487e-01 -1.01864386e+00 -3.90877277e-02
-7.73826241e-02 1.63137948e-03 -5.34415781e-01 9.41057980e-01
1.08783579e+00 -4.09961581e-01 8.41708839e-01 -1.88597232e-01
2.64304370e-01 -2.95384731e-02 -6.77764714e-01 1.76621532e+00
1.44786097e-03 2.58941084e-01 2.69570529e-01 -5.34151196e-01
4.86469150e-01 3.18548411e-01 5.93494236e-01 -1.71276554e-01
-1.67736724e-01 2.69758552e-01 4.86848131e-02 -3.10578287e-01
8.49446505e-02 6.23012409e-02 -2.67509162e-01 5.50823689e-01
-3.93749237e-01 -1.41176963e-02 -3.41586232e-01 2.28689998e-01
1.74582124e+00 -6.16795182e-01 4.83171940e-01 -2.33120099e-01
3.68901342e-01 1.39457017e-01 9.25898194e-01 4.73676860e-01
-1.02632575e-01 -1.86271872e-02 7.84097373e-01 -8.02543640e-01
-7.77295947e-01 -8.24647784e-01 -4.84959126e-01 4.03425962e-01
-2.30860278e-01 -8.32436740e-01 -5.48124075e-01 -1.00669444e+00
7.44421422e-01 2.88491964e-01 -5.32261074e-01 -2.66146630e-01
-1.69022933e-01 -1.48293769e+00 9.88092661e-01 3.31792720e-02
3.23470175e-01 -8.34351182e-02 -5.69317997e-01 -9.37723890e-02
5.54628335e-02 -8.65272820e-01 -4.77274716e-01 2.66799510e-01
-8.43044341e-01 -1.13076723e+00 -2.05811247e-01 -3.03494751e-01
1.03054333e+00 -6.28094673e-02 7.42460132e-01 1.73081025e-01
-9.87196624e-01 -9.38198064e-03 1.47254199e-01 -5.35269618e-01
-3.79867315e-01 1.90456659e-02 1.49164528e-01 -8.89248475e-02
7.83472657e-01 -3.35726291e-01 -6.86568022e-01 -2.72892602e-02
-1.07451165e+00 -2.55959183e-02 8.28820109e-01 1.11330891e+00
9.66801047e-01 3.09637547e-01 1.97988227e-01 -1.37340367e+00
5.13169467e-01 -6.12525523e-01 -7.46999979e-01 3.97687733e-01
-6.70514703e-01 1.76250190e-01 6.77298009e-01 -1.20917320e-01
-6.81828558e-01 4.95064259e-01 -7.21794516e-02 -1.87396270e-03
1.14953302e-01 4.43093807e-01 -2.84041822e-01 -4.34484512e-01
3.03942472e-01 3.48928034e-01 5.85362434e-01 -1.19966701e-01
7.93391373e-03 8.67955029e-01 2.02210799e-01 -2.38624379e-01
4.74573106e-01 6.87903464e-01 7.12867498e-01 -6.03246868e-01
-1.97661489e-01 -1.24648631e-01 -2.35380024e-01 6.05972409e-01
5.35150766e-01 -1.14317203e+00 -1.41316998e+00 2.73931712e-01
-5.68371058e-01 -1.59354452e-02 -5.08123748e-02 6.08160853e-01
-1.89767078e-01 3.59394372e-01 -9.36945915e-01 -4.95005101e-01
-6.77202046e-01 -1.13331294e+00 1.09403372e+00 -4.00987267e-01
-2.31633574e-01 -7.23541379e-01 -1.16516344e-01 3.42787296e-01
2.45218754e-01 3.97456825e-01 1.28878677e+00 -7.54369974e-01
-6.92339659e-01 -6.12988591e-01 -7.17876107e-02 -3.07315379e-01
2.04978272e-01 -4.60615009e-01 -9.03376877e-01 -7.29702532e-01
-1.73225299e-01 -3.24404359e-01 6.99434876e-01 1.55771792e-01
1.87578249e+00 -5.03471732e-01 -8.53162766e-01 1.31844902e+00
1.44198632e+00 -1.47665367e-01 6.37376904e-01 -1.35330483e-01
4.87283975e-01 4.28374827e-01 7.33697116e-01 7.83210397e-01
3.60401988e-01 6.70398891e-01 2.06908971e-01 -2.16439277e-01
4.82478708e-01 -2.04650462e-01 5.73896877e-02 4.57133055e-01
3.48398328e-01 -5.77914380e-02 -6.96695983e-01 1.26091346e-01
-1.50492477e+00 -5.50640047e-01 7.61701688e-02 2.55922103e+00
1.24939263e+00 -3.79482806e-01 -1.57113582e-01 6.30543800e-03
2.55855262e-01 -4.23021466e-01 -8.70756865e-01 -4.28708881e-01
-2.71570534e-01 2.64974087e-01 1.03193927e+00 2.24309638e-01
-8.54944468e-01 5.32542527e-01 6.19825172e+00 8.58072340e-01
-1.02317250e+00 4.16527502e-02 1.02606356e+00 -5.80773711e-01
-6.20922506e-01 -2.26065859e-01 -9.71229851e-01 5.59496522e-01
1.02744138e+00 -4.11658794e-01 2.88847208e-01 6.88072920e-01
-1.98182762e-01 1.06573403e-01 -1.28738880e+00 1.07574785e+00
-1.81896031e-01 -1.71508324e+00 -8.55467170e-02 8.20950449e-01
4.37220633e-01 -8.94208103e-02 3.19434375e-01 -1.41384631e-01
3.07676762e-01 -1.07907724e+00 -3.97633731e-01 4.10762966e-01
1.31619811e+00 -1.16843390e+00 9.41638649e-01 3.06384534e-01
-6.86602473e-01 -2.15712070e-01 -5.69851696e-01 3.08146626e-01
-3.57756972e-01 1.04958522e+00 -1.38946259e+00 5.94383240e-01
6.95502341e-01 4.33967322e-01 -4.03051853e-01 4.69037443e-01
4.70146298e-01 3.68335813e-01 -5.52070379e-01 -1.94186434e-01
-2.35356718e-01 4.85687377e-03 1.89624771e-01 1.09245086e+00
6.40456438e-01 4.92032021e-01 -1.97065979e-01 4.53724146e-01
-2.99532950e-01 2.95648456e-01 -8.48014534e-01 -1.56369671e-01
6.94399953e-01 1.18389916e+00 -9.79981497e-02 -7.30178431e-02
-3.50375473e-01 1.21757913e+00 3.29708964e-01 -1.42241940e-01
-5.22900343e-01 -4.39289719e-01 1.31987667e+00 -3.58628966e-02
1.13449462e-01 2.13082135e-01 -2.31679350e-01 -1.18602037e+00
6.33823127e-02 -1.17626405e+00 8.82471681e-01 1.48641273e-01
-1.11891675e+00 1.73695326e-01 -4.68520790e-01 -9.90432143e-01
-3.41900080e-01 -5.17036378e-01 -9.23554972e-02 8.36010873e-01
-1.35032928e+00 -1.02769351e+00 7.92722777e-03 6.85145199e-01
-3.95387799e-01 -2.60681391e-01 1.48916435e+00 2.59018540e-01
-7.22769082e-01 1.43925416e+00 5.10700166e-01 -9.47258398e-02
5.84637761e-01 -8.95707726e-01 3.14227939e-01 5.42486906e-01
-2.94582069e-01 7.88352132e-01 1.78299248e-01 -8.30806255e-01
-2.47626328e+00 -1.34197044e+00 1.05730104e+00 -3.01979810e-01
1.59937233e-01 -8.66969943e-01 -7.62951076e-01 6.38443947e-01
-5.17946661e-01 4.70721334e-01 1.51151574e+00 -5.69496565e-02
-3.85209113e-01 -4.42549318e-01 -2.06009936e+00 6.30133450e-01
8.91568542e-01 -6.93866253e-01 4.77723300e-01 6.82789028e-01
8.37303519e-01 -4.39333588e-01 -1.43211997e+00 1.85378075e-01
6.56814873e-01 -6.08342052e-01 1.02815020e+00 -8.54256332e-01
1.17930025e-01 -1.47639796e-01 -2.89045662e-01 -9.83618140e-01
-1.85535416e-01 -7.82682538e-01 -2.10721269e-01 8.03816974e-01
4.88918304e-01 -1.04847193e+00 1.18668723e+00 1.20578575e+00
4.54276741e-01 -9.24267530e-01 -1.19489729e+00 -3.79009008e-01
-1.80232942e-01 -2.31051102e-01 1.13482940e+00 1.12092507e+00
2.49220639e-01 -4.63645250e-01 -2.67960936e-01 4.38463539e-01
9.45690453e-01 8.36117566e-02 8.06126654e-01 -9.55643237e-01
-4.95083630e-01 1.25170559e-01 -9.96420026e-01 -2.27860287e-01
1.89331546e-01 -1.44492960e+00 -5.83709180e-01 -6.04550183e-01
6.41606033e-01 -6.60329640e-01 -3.14032972e-01 7.76023567e-01
-1.79767057e-01 1.85371265e-01 -3.23272675e-01 -1.17351897e-01
-1.43319115e-01 1.03990793e-01 8.19665015e-01 -1.19070329e-01
7.59844556e-02 -6.08194098e-02 -8.83515775e-01 1.44176215e-01
6.82690442e-01 -5.72124481e-01 -3.06942403e-01 -1.26224756e-01
3.09641510e-01 1.66076317e-01 3.34459633e-01 -6.68273389e-01
4.72330958e-01 -8.19777176e-02 6.72841549e-01 -5.94289601e-01
4.25619245e-01 -1.00787127e+00 7.30477512e-01 1.18317842e+00
-3.63812745e-01 -6.47671346e-04 2.81951725e-01 6.84231818e-01
3.26691009e-02 3.24622691e-01 3.14171463e-01 8.71903449e-02
-1.98876292e-01 6.93103194e-01 -4.90781702e-02 -3.16929489e-01
1.40201211e+00 1.00128993e-01 -4.28083986e-01 -3.01218182e-02
-3.23767424e-01 2.86787271e-01 8.44670653e-01 -3.16842616e-01
7.89846063e-01 -1.10105371e+00 -8.92034948e-01 7.28171170e-01
4.18259919e-01 -1.35977998e-01 4.13517982e-01 7.61785269e-01
-4.84358519e-01 5.80266833e-01 -4.81757373e-02 -4.99107391e-01
-1.83941627e+00 6.70531452e-01 -3.73241752e-02 -2.65122354e-01
-6.94148064e-01 9.05614197e-01 1.52785018e-01 -5.37401378e-01
1.22779198e-02 -1.50587454e-01 6.13049150e-01 -3.59193057e-01
8.27309132e-01 2.92147994e-01 3.12868357e-01 -6.85304552e-02
-5.44493556e-01 1.79037794e-01 -4.65948254e-01 1.65570065e-01
1.38439953e+00 3.35637599e-01 -6.47978365e-01 -2.56546825e-01
1.72318304e+00 2.37165168e-01 -8.09133947e-01 -2.43767072e-02
-2.37008825e-01 -8.02452981e-01 3.33257206e-02 -6.92364037e-01
-1.11834025e+00 4.73520577e-01 6.49495840e-01 -4.30942744e-01
1.37238181e+00 -2.75868952e-01 9.47581351e-01 5.96670210e-01
5.46725571e-01 -5.52361071e-01 -1.02904582e+00 -1.76720768e-01
1.96025655e-01 -1.06916726e+00 4.45807934e-01 -7.32976973e-01
-4.96123195e-01 8.22752237e-01 1.40270337e-01 4.88480687e-01
6.49038136e-01 8.77490878e-01 -2.56174862e-01 -6.66567609e-02
-1.31624317e+00 6.49885118e-01 -8.36494714e-02 5.77844560e-01
1.86272830e-01 4.75680023e-01 -4.66909528e-01 7.62925267e-01
-2.53319710e-01 2.78148651e-01 2.01327309e-01 1.16723335e+00
2.27380738e-01 -1.47252405e+00 -2.39665210e-01 1.11888921e+00
-8.15695643e-01 -2.55984902e-01 -3.69994938e-01 3.90503019e-01
-8.07152987e-02 3.68438989e-01 1.46234319e-01 -3.48653853e-01
-1.41614571e-01 1.49684131e-01 1.49099633e-01 -1.91145942e-01
-7.31819749e-01 -4.21018273e-01 2.75475159e-03 -8.38592649e-01
3.26144189e-01 -8.59308183e-01 -1.18777704e+00 -6.87089026e-01
7.40922913e-02 1.02440685e-01 9.30423081e-01 3.26682031e-01
1.20826721e+00 7.56376684e-02 5.93631029e-01 1.18900999e-01
-7.62870669e-01 -4.61267941e-02 -7.17918456e-01 3.10365438e-01
3.47710669e-01 4.12437245e-02 -1.05784625e-01 -3.42035055e-01] | [6.121723651885986, 6.532205104827881] |
a1670a6c-768b-41a6-8d4c-2549122d2b5a | entitycs-improving-zero-shot-cross-lingual | 2210.12540 | null | https://arxiv.org/abs/2210.12540v2 | https://arxiv.org/pdf/2210.12540v2.pdf | EntityCS: Improving Zero-Shot Cross-lingual Transfer with Entity-Centric Code Switching | Accurate alignment between languages is fundamental for improving cross-lingual pre-trained language models (XLMs). Motivated by the natural phenomenon of code-switching (CS) in multilingual speakers, CS has been used as an effective data augmentation method that offers language alignment at the word- or phrase-level, in contrast to sentence-level via parallel instances. Existing approaches either use dictionaries or parallel sentences with word alignment to generate CS data by randomly switching words in a sentence. However, such methods can be suboptimal as dictionaries disregard semantics, and syntax might become invalid after random word switching. In this work, we propose EntityCS, a method that focuses on Entity-level Code-Switching to capture fine-grained cross-lingual semantics without corrupting syntax. We use Wikidata and English Wikipedia to construct an entity-centric CS corpus by switching entities to their counterparts in other languages. We further propose entity-oriented masking strategies during intermediate model training on the EntityCS corpus for improving entity prediction. Evaluation of the trained models on four entity-centric downstream tasks shows consistent improvements over the baseline with a notable increase of 10% in Fact Retrieval. We release the corpus and models to assist research on code-switching and enriching XLMs with external knowledge. | ['Ignacio Iacobacci', 'Fenia Christopoulou', 'Chenxi Whitehouse'] | 2022-10-22 | null | null | null | null | ['zero-shot-cross-lingual-transfer', 'word-alignment'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.70649216e-02 6.68593273e-02 -5.26225805e-01 -4.13639724e-01
-9.03135061e-01 -6.26441002e-01 4.45900112e-01 5.27526319e-01
-5.30138135e-01 5.10585487e-01 4.42787707e-01 -7.36871898e-01
3.88511568e-01 -5.02330661e-01 -8.30627859e-01 -7.11894706e-02
2.79971343e-02 2.77337492e-01 -4.59179888e-03 -6.19104207e-01
2.55383641e-01 -9.22671407e-02 -1.22605371e+00 7.65650928e-01
1.60429049e+00 6.51936978e-02 6.09824896e-01 1.72320604e-01
-7.84337163e-01 7.74283051e-01 -4.85509574e-01 -1.09494174e+00
2.20154583e-01 -4.12263215e-01 -7.80629396e-01 -3.89637291e-01
6.52592599e-01 3.55815262e-01 1.15735367e-01 1.21117544e+00
3.47597569e-01 -3.01991284e-01 4.10485476e-01 -9.39814031e-01
-1.09433985e+00 1.50604713e+00 -3.67779404e-01 2.45766416e-01
5.58530271e-01 -2.35377401e-01 1.33231878e+00 -1.18631732e+00
9.34268296e-01 8.70266616e-01 8.43095958e-01 7.23037422e-01
-1.34709299e+00 -7.70898938e-01 3.58812749e-01 1.47160724e-01
-1.65417218e+00 -6.77927732e-01 5.97434402e-01 -6.66090488e-01
1.58461761e+00 1.98535115e-01 3.22117895e-01 1.07831585e+00
3.02732050e-01 8.50481272e-01 8.64354789e-01 -1.00240135e+00
-2.79786617e-01 6.01598501e-01 2.86640394e-02 6.52360618e-01
3.16510767e-01 -1.50440648e-01 -6.43090427e-01 -8.04926977e-02
1.52195290e-01 -3.70433211e-01 -2.39733011e-01 -3.04767281e-01
-1.38546991e+00 7.99581826e-01 2.18594238e-01 5.18878222e-01
-6.51145279e-02 -2.19840050e-01 6.81123257e-01 5.57839036e-01
4.52021927e-01 9.64936256e-01 -8.98006141e-01 2.04128861e-01
-1.10496104e+00 1.75770208e-01 6.88825011e-01 1.48296416e+00
6.77947998e-01 1.82243213e-01 -1.76862031e-01 1.11691546e+00
2.54273325e-01 6.59805238e-01 1.10949051e+00 -3.66202563e-01
1.00467193e+00 7.10057259e-01 -3.25131327e-01 -8.91627669e-01
-1.15195155e-01 -8.32175851e-01 -4.57998276e-01 -4.65291500e-01
4.47811075e-02 -2.20350502e-03 -5.78167915e-01 2.13206387e+00
-4.84983437e-02 3.04087158e-02 4.64817464e-01 2.02593267e-01
7.78920412e-01 2.95775205e-01 3.84666353e-01 3.91835580e-03
1.25670767e+00 -1.17370749e+00 -8.01037788e-01 -6.61873639e-01
1.45852375e+00 -9.36038435e-01 1.53403199e+00 -2.71818526e-02
-9.18935001e-01 -5.85347593e-01 -1.05488074e+00 -1.14319623e-02
-6.57989383e-01 3.72293711e-01 4.98135656e-01 7.72123337e-01
-1.02801752e+00 2.38176510e-01 -7.88362026e-01 -3.29832226e-01
-3.78753617e-02 -7.40528032e-02 -3.07407439e-01 6.27612099e-02
-1.52269995e+00 1.16449499e+00 4.63402122e-01 -4.32714373e-01
-2.87865847e-01 -1.00349426e+00 -1.26530802e+00 -2.40999348e-02
7.45356455e-02 -5.05637944e-01 1.11829603e+00 -1.06761408e+00
-1.07541633e+00 1.30360615e+00 -3.21868151e-01 -5.54054022e-01
-1.03111686e-02 -2.61821777e-01 -8.46030831e-01 -5.17746150e-01
5.31160533e-01 4.54836279e-01 4.20402855e-01 -1.07631910e+00
-5.91462910e-01 -7.42058381e-02 -7.82262683e-02 3.73622924e-01
-5.11193454e-01 3.29765230e-01 -4.01165098e-01 -1.03383875e+00
1.68729909e-02 -9.65563476e-01 -2.13823855e-01 -8.18976760e-01
-4.59195226e-01 -1.56750277e-01 -2.94310469e-02 -9.87447143e-01
1.90448916e+00 -2.07444239e+00 1.42626211e-01 -2.45427117e-02
-1.08766787e-01 3.77524734e-01 -4.03485060e-01 7.02740192e-01
-2.80373275e-01 3.73769641e-01 -4.81841207e-01 -7.01676667e-01
-1.34228736e-01 8.10051262e-02 -6.13039374e-01 -2.39998251e-02
3.19796473e-01 9.64330018e-01 -9.15158093e-01 -4.93243009e-01
-2.15388790e-01 -3.69239710e-02 -1.01541281e+00 1.87731963e-02
-1.09764144e-01 2.86354035e-01 -3.52553427e-02 5.67765474e-01
3.55699748e-01 -8.63041449e-03 3.56822252e-01 1.34818271e-01
-2.60891110e-01 1.14085639e+00 -9.57616448e-01 2.02476931e+00
-1.01787496e+00 4.34202492e-01 -3.70532781e-01 -7.39654899e-01
7.14954495e-01 3.33744526e-01 -7.12712947e-03 -7.41438627e-01
-3.77972335e-01 6.42210364e-01 1.59868225e-01 -5.41507304e-01
8.55846763e-01 -1.37305826e-01 -5.17104328e-01 1.41125754e-01
1.49201646e-01 -1.09132037e-01 3.52208942e-01 4.12297368e-01
9.70693588e-01 1.52920589e-01 5.63128233e-01 -5.58271468e-01
7.88901091e-01 4.08482969e-01 5.98367691e-01 5.89214861e-01
3.21460932e-01 3.29418570e-01 -3.28528732e-02 7.27555901e-02
-9.55902338e-01 -9.20884132e-01 -4.21475232e-01 1.28721583e+00
2.39949040e-02 -1.01323819e+00 -7.22881556e-01 -8.04081857e-01
1.93819124e-02 1.26325500e+00 -2.17103735e-01 -3.77543986e-01
-9.36813176e-01 -5.50581515e-01 9.11422431e-01 2.78527379e-01
1.57380804e-01 -9.91271257e-01 -2.27414090e-02 3.72049689e-01
-5.44966400e-01 -1.18187869e+00 -6.51149929e-01 1.84990197e-01
-5.48413217e-01 -7.43087828e-01 -1.25113025e-01 -1.17854989e+00
8.15408766e-01 -1.60815790e-02 1.63093841e+00 3.86139244e-01
2.15094179e-01 3.53383832e-03 -5.59401453e-01 -2.60432124e-01
-1.02570093e+00 5.89556694e-01 2.94366837e-01 -3.60333443e-01
7.08669066e-01 -3.62253189e-01 -7.65029043e-02 -1.17354676e-01
-9.12693918e-01 1.70216888e-01 5.23658991e-01 7.78109610e-01
3.29848349e-01 -5.21320343e-01 6.90556109e-01 -1.51317978e+00
7.01162279e-01 -6.74557149e-01 -4.96981233e-01 5.21390140e-01
-7.02985048e-01 3.76986533e-01 9.50024545e-01 -3.16118389e-01
-1.09862769e+00 5.95793985e-02 -3.01953405e-01 1.01084717e-01
9.09116641e-02 1.13928163e+00 -1.63286597e-01 1.42699927e-01
9.37067509e-01 4.08154339e-01 -3.70226294e-01 -5.38900256e-01
6.07473731e-01 7.93302536e-01 5.86259604e-01 -8.41000617e-01
7.53036737e-01 -5.50511107e-02 -8.23259652e-01 -4.89129394e-01
-8.67172182e-01 -5.00244260e-01 -8.86681855e-01 2.25798741e-01
6.40999079e-01 -1.38982141e+00 2.88379967e-01 2.01683789e-01
-1.22049034e+00 -3.47829700e-01 -1.48551911e-01 4.01006520e-01
-8.17399621e-02 4.05230522e-01 -5.87316632e-01 -1.13028668e-01
-1.38082415e-01 -1.12804866e+00 1.00376403e+00 -2.78790623e-01
-5.40552199e-01 -1.24851501e+00 4.59339470e-01 2.18358785e-01
5.70839226e-01 -4.54561859e-01 1.21574843e+00 -1.00724375e+00
-4.07362938e-01 -9.37622339e-02 3.75453413e-01 3.05905789e-01
1.83687449e-01 7.71232396e-02 -4.57428724e-01 -3.53780001e-01
-2.14074150e-01 -1.66646048e-01 4.71658528e-01 -8.93649608e-02
7.93708622e-01 -4.24214810e-01 -4.39229518e-01 7.58468926e-01
1.54947567e+00 1.05974441e-02 4.22507256e-01 6.45451725e-01
6.40101612e-01 6.09000504e-01 4.23708528e-01 5.21272328e-03
8.23111534e-01 7.24049687e-01 -6.42679781e-02 -4.98171002e-02
-3.92352998e-01 -4.91075426e-01 7.38518357e-01 1.83385229e+00
5.84646583e-01 -3.77207226e-03 -1.23908246e+00 9.31254387e-01
-1.39080846e+00 -6.41325712e-01 -3.86381716e-01 2.23779202e+00
1.57235074e+00 -4.21567028e-03 -4.86926466e-01 -4.97295469e-01
5.96395433e-01 -1.45004839e-01 -8.46366119e-03 -3.70223284e-01
-4.25461322e-01 2.59617448e-01 5.85918725e-01 7.84260094e-01
-9.06361997e-01 1.56246316e+00 5.56611681e+00 8.70239019e-01
-1.17446578e+00 4.42543119e-01 -9.53575037e-03 2.42787480e-01
-7.80238926e-01 3.68782073e-01 -1.37586749e+00 6.80530071e-01
9.62354124e-01 -4.31050956e-01 3.15397143e-01 9.99100745e-01
-3.02242279e-01 2.52167553e-01 -1.27214229e+00 7.49000669e-01
2.99312562e-01 -1.31579423e+00 2.04173356e-01 -3.76164973e-01
9.94694769e-01 4.40239966e-01 -8.45479965e-02 8.60478044e-01
6.25664890e-01 -7.91351020e-01 1.01946580e+00 8.02300423e-02
9.51666296e-01 -4.78897601e-01 7.27038622e-01 4.08573389e-01
-1.31212747e+00 2.81126529e-01 -3.25282097e-01 9.72716324e-03
1.66833028e-01 4.87631232e-01 -7.72544861e-01 5.67607343e-01
4.58505332e-01 7.83471346e-01 -1.01951838e+00 7.78195977e-01
-4.87160385e-01 5.84550440e-01 -1.78794041e-02 1.78945467e-01
8.97659775e-05 -7.60982484e-02 7.19284654e-01 1.77260208e+00
4.23183888e-01 -3.81981939e-01 1.19723275e-01 8.62433136e-01
-1.86410025e-01 6.17884696e-01 -8.39112639e-01 6.65806681e-02
9.65226233e-01 1.01178169e+00 -3.50313246e-01 -5.84002376e-01
-8.77528250e-01 1.15663147e+00 6.80880070e-01 2.30488256e-01
-6.74993277e-01 -5.28938651e-01 8.01199794e-01 -3.64541821e-03
1.45309150e-01 -3.08635950e-01 -2.83140957e-01 -1.74116290e+00
1.88107818e-01 -1.26613867e+00 3.05514842e-01 -3.47705930e-01
-1.33976722e+00 1.02397966e+00 -1.53998122e-01 -1.32227731e+00
-2.72565752e-01 -4.68214303e-01 -2.47992769e-01 9.19995964e-01
-1.68750489e+00 -1.14625061e+00 2.15980023e-01 4.85669881e-01
8.04590285e-01 -4.91574705e-01 9.81686890e-01 7.71211624e-01
-5.31112492e-01 1.14542460e+00 2.15758815e-01 4.80231732e-01
1.07068074e+00 -1.34113193e+00 7.67412126e-01 1.40016329e+00
5.57504892e-01 1.33029938e+00 6.68866694e-01 -1.06086743e+00
-9.12715971e-01 -1.30165112e+00 1.79233515e+00 -9.88619447e-01
9.18526649e-01 -6.68097734e-01 -1.18933332e+00 1.04543257e+00
3.95502448e-01 -2.60815412e-01 9.65328991e-01 3.31853569e-01
-5.38924754e-01 1.78405404e-01 -5.53928673e-01 1.03239298e+00
1.20181477e+00 -8.58281910e-01 -1.05305684e+00 2.49772057e-01
9.68413055e-01 -6.34398401e-01 -7.52772033e-01 3.76451761e-01
5.59960827e-02 -4.86724794e-01 8.14579189e-01 -8.97596657e-01
6.03892863e-01 -3.29963773e-01 -3.28894198e-01 -1.67438662e+00
-2.35190839e-01 -3.22042167e-01 1.99312598e-01 1.54850924e+00
1.08286238e+00 -5.40184319e-01 2.63950944e-01 4.14971024e-01
-6.33765459e-01 -3.18529069e-01 -7.33242214e-01 -1.08373439e+00
4.96811450e-01 -5.59804380e-01 6.98174179e-01 1.65565276e+00
5.26614010e-01 5.18104911e-01 1.80900784e-03 2.62990773e-01
1.76719159e-01 -9.13225987e-04 6.67219937e-01 -8.21582437e-01
-2.95720637e-01 -5.16217887e-01 -1.08886734e-01 -1.03215945e+00
7.02979267e-01 -1.86023211e+00 1.75964862e-01 -1.00009429e+00
1.97567791e-01 -8.83126855e-01 -9.69343856e-02 5.71989119e-01
-5.28099120e-01 -5.43911308e-02 -3.14112334e-03 3.13496798e-01
-4.58022445e-01 5.18224776e-01 5.47958076e-01 6.41754121e-02
-1.48248360e-01 -2.41772652e-01 -7.43416369e-01 5.42099297e-01
6.39219761e-01 -9.00760770e-01 -2.21177518e-01 -8.01125824e-01
7.72628069e-01 -2.59952813e-01 -3.08077484e-01 -9.03100431e-01
2.42245063e-01 -2.13572830e-02 -3.41461837e-01 -1.43552095e-01
-3.43500137e-01 -7.28483319e-01 8.32152590e-02 4.94307190e-01
-6.94706976e-01 3.92752737e-01 4.87339467e-01 9.19088572e-02
-4.27520275e-01 -5.78787863e-01 5.20181000e-01 -2.27705374e-01
-7.56896138e-01 -8.70266780e-02 -4.89995807e-01 4.80804056e-01
6.56750977e-01 -1.28452089e-02 -2.86582977e-01 8.21498856e-02
-4.89457130e-01 1.64430514e-01 6.65360212e-01 8.04379761e-01
1.53487176e-01 -1.46000350e+00 -8.94758046e-01 4.99094486e-01
6.44165277e-01 -2.99029440e-01 -1.48224235e-01 7.67501235e-01
-4.16786462e-01 5.88323414e-01 -5.06959856e-02 -3.86673391e-01
-1.25196433e+00 4.63667691e-01 2.25691900e-01 -5.40269196e-01
-2.38365844e-01 1.13098919e+00 1.16513893e-01 -1.26011980e+00
-5.49367294e-02 -3.39292496e-01 -1.99956417e-01 -1.11633226e-01
2.07863733e-01 -2.57206082e-01 6.19844854e-01 -7.60234118e-01
-4.68456596e-01 3.05010349e-01 -3.61569077e-01 7.40612298e-02
1.16290069e+00 -4.45557028e-01 -4.93135542e-01 5.05526483e-01
9.89261806e-01 9.68197405e-01 -4.78937894e-01 -6.95817173e-01
7.74986804e-01 -3.31758142e-01 -3.46735537e-01 -7.75349617e-01
-7.66012669e-01 6.28840148e-01 3.14655416e-02 -9.36659276e-02
6.91761672e-01 1.16696805e-01 7.78851867e-01 2.88457632e-01
4.85064238e-01 -1.07447314e+00 -1.60491273e-01 8.77870440e-01
8.02117646e-01 -1.27402663e+00 -4.07877922e-01 -4.97787893e-01
-8.11961651e-01 7.10520506e-01 9.61599529e-01 1.05519801e-01
4.97645378e-01 4.06709224e-01 2.79039919e-01 -6.42282069e-02
-6.85265958e-01 -1.83421507e-01 4.25985068e-01 2.63874173e-01
1.10875547e+00 8.15321803e-02 -5.23526788e-01 8.78441751e-01
-6.52753770e-01 -5.84086061e-01 4.83978629e-01 9.40842330e-01
-7.03934506e-02 -1.74042225e+00 -8.35525710e-03 4.27375764e-01
-7.35468090e-01 -1.20652533e+00 -1.81240007e-01 7.12737322e-01
1.71285510e-01 5.95516145e-01 7.85082579e-02 -2.81164259e-01
3.40670586e-01 5.71087658e-01 1.42289281e-01 -1.36506140e+00
-8.59649539e-01 -3.95044982e-01 2.40043074e-01 -3.55844140e-01
-2.38577127e-01 -8.87565374e-01 -1.33932769e+00 -1.16469674e-01
-4.60197866e-01 2.52105057e-01 6.52165473e-01 9.56825376e-01
5.63574135e-01 5.28887630e-01 3.45644385e-01 -1.78933725e-01
-3.34651709e-01 -9.82404768e-01 -1.40768051e-01 6.22358024e-01
1.79033235e-01 -3.44231904e-01 -3.46173257e-01 3.95795971e-01] | [11.021403312683105, 9.95093059539795] |
936be60a-54f1-4c0c-ac8b-acaa49c0c0b0 | uni6dv2-noise-elimination-for-6d-pose | 2208.06416 | null | https://arxiv.org/abs/2208.06416v2 | https://arxiv.org/pdf/2208.06416v2.pdf | Uni6Dv2: Noise Elimination for 6D Pose Estimation | Uni6D is the first 6D pose estimation approach to employ a unified backbone network to extract features from both RGB and depth images. We discover that the principal reasons of Uni6D performance limitations are Instance-Outside and Instance-Inside noise. Uni6D's simple pipeline design inherently introduces Instance-Outside noise from background pixels in the receptive field, while ignoring Instance-Inside noise in the input depth data. In this paper, we propose a two-step denoising approach for dealing with the aforementioned noise in Uni6D. To reduce noise from non-instance regions, an instance segmentation network is utilized in the first step to crop and mask the instance. A lightweight depth denoising module is proposed in the second step to calibrate the depth feature before feeding it into the pose regression network. Extensive experiments show that our Uni6Dv2 reliably and robustly eliminates noise, outperforming Uni6D without sacrificing too much inference efficiency. It also reduces the need for annotated real data that requires costly labeling. | ['Xiaoke Jiang', 'Rui Zhao', 'Liwei Wu', 'Guoqiang Jin', 'Jianqiu Chen', 'Tianpeng Bao', 'Ye Zheng', 'Mingshan Sun'] | 2022-08-15 | null | null | null | null | ['6d-pose-estimation-1'] | ['computer-vision'] | [ 3.34522903e-01 2.79679179e-01 1.81165159e-01 -6.59418821e-01
-7.62583971e-01 -5.80184639e-01 2.39815101e-01 -1.39408052e-01
-5.45182705e-01 3.93674195e-01 -1.18549310e-01 -2.08714560e-01
1.09202944e-01 -9.15358067e-01 -7.99556792e-01 -7.23792553e-01
3.97550911e-01 3.57311428e-01 4.64534014e-01 2.24418640e-01
1.31682381e-01 6.61205888e-01 -1.68142164e+00 8.66688564e-02
6.93291903e-01 1.39493668e+00 1.85446098e-01 4.86141741e-01
-2.56523252e-01 3.16570252e-01 -7.27825224e-01 -2.82573819e-01
7.41043329e-01 -1.70626968e-01 -4.46905404e-01 1.39142945e-01
8.25313270e-01 -7.76247680e-01 -3.37641269e-01 9.88601744e-01
8.19407880e-01 -1.22226886e-01 4.01627481e-01 -9.45677340e-01
6.70178831e-02 2.62956470e-02 -8.20302606e-01 -8.47223848e-02
2.14638501e-01 3.56593728e-01 6.31459594e-01 -8.21674585e-01
6.90468311e-01 1.20852840e+00 7.65244961e-01 7.34991491e-01
-1.18418550e+00 -5.40405571e-01 1.22345872e-01 -4.46617901e-01
-1.25193000e+00 -2.81289637e-01 7.92671084e-01 -2.24674642e-01
8.91619742e-01 1.61742672e-01 6.62569225e-01 1.13475883e+00
-1.26298545e-02 8.49801898e-01 1.16752326e+00 -1.22988001e-01
4.69924897e-01 -2.83044755e-01 4.84105647e-02 7.11699247e-01
2.81761736e-01 -1.46046681e-02 -6.10103250e-01 1.41223192e-01
1.06313491e+00 -1.68544829e-01 9.47488844e-02 -4.81818438e-01
-7.45283246e-01 4.78836685e-01 6.40173852e-01 -3.17839719e-02
-2.98447758e-01 3.93369764e-01 2.09420338e-01 -3.54164653e-02
5.82093120e-01 1.60538405e-02 -8.15071285e-01 -1.97473869e-01
-9.85555649e-01 1.28332078e-01 4.88376260e-01 9.95359242e-01
1.02142024e+00 -2.96569288e-01 8.71058702e-02 6.33561850e-01
3.77026111e-01 4.98261541e-01 -5.70190698e-03 -1.21763992e+00
6.37660205e-01 8.75368118e-01 -1.52028620e-01 -8.02741349e-01
-6.08282447e-01 -4.84540850e-01 -6.46344066e-01 3.71150047e-01
6.13065660e-01 -1.80926278e-01 -1.41915107e+00 1.45969820e+00
6.96363032e-01 9.52212214e-02 -1.61045074e-01 9.47816730e-01
1.09305334e+00 1.97901472e-01 -1.21469900e-01 1.48065612e-01
1.12101018e+00 -5.94039321e-01 -3.01229894e-01 -6.04337633e-01
4.12763387e-01 -5.99624336e-01 8.85049462e-01 3.98442239e-01
-9.38287079e-01 -4.69886631e-01 -1.07841730e+00 -5.17304122e-01
-2.25521311e-01 2.04028130e-01 8.08198929e-01 5.82435310e-01
-9.10495102e-01 5.28915286e-01 -9.60604787e-01 -2.29679853e-01
6.00065351e-01 6.27596200e-01 -4.43035156e-01 -2.54226357e-01
-6.93845451e-01 6.26166940e-01 2.66831636e-01 4.49069649e-01
-7.29755938e-01 -6.91030800e-01 -1.10274911e+00 -3.55635464e-01
4.53420341e-01 -6.51823759e-01 1.04354417e+00 -5.22649884e-01
-1.36201215e+00 9.59597826e-01 -2.08786234e-01 -1.78048670e-01
6.74861372e-01 -3.24024260e-01 2.17252538e-01 1.22606717e-01
1.55802995e-01 9.66124296e-01 8.97204816e-01 -1.40521574e+00
-6.05580151e-01 -8.49664092e-01 6.80539384e-02 2.06132933e-01
9.78758708e-02 -5.42386413e-01 -1.12978601e+00 -4.01650190e-01
9.41660941e-01 -7.79803216e-01 -2.77671963e-01 2.21335128e-01
-4.09788072e-01 -2.56031118e-02 8.61802638e-01 -4.49844271e-01
7.48673081e-01 -2.23286653e+00 -8.60797688e-02 4.61396843e-01
4.48839486e-01 4.42850217e-02 -5.44025972e-02 -1.54116124e-01
3.16525847e-01 -1.77936465e-01 -4.14403975e-01 -7.24373400e-01
-1.24934249e-01 6.07142687e-01 9.33362618e-02 4.32297140e-01
3.38337034e-01 7.91740894e-01 -8.20423305e-01 -4.87605721e-01
4.75588053e-01 7.10896611e-01 -6.52994514e-01 2.02551350e-01
-3.24690640e-01 5.51281393e-01 -3.01993698e-01 9.45030332e-01
1.06431282e+00 1.22091502e-01 -3.31616253e-02 -5.25922894e-01
-6.12822510e-02 4.59430754e-01 -1.31661761e+00 2.09429336e+00
-4.12389696e-01 2.88137615e-01 3.85461390e-01 -6.05986416e-01
1.06497371e+00 -1.42860487e-01 4.73357260e-01 -8.21118355e-01
3.64083469e-01 3.38530093e-01 -2.26911098e-01 -5.04758179e-01
2.64479786e-01 1.04442425e-01 -1.58767775e-01 1.32046774e-01
1.86518878e-01 -4.50224727e-01 -3.33947465e-02 -1.80941485e-02
1.10680532e+00 8.18821788e-01 -1.04647338e-01 -6.47432879e-02
3.12132239e-01 -6.73452914e-02 9.74117041e-01 7.03635752e-01
-3.66107196e-01 1.07144475e+00 4.46887791e-01 -4.26770657e-01
-6.10695064e-01 -1.01978874e+00 -1.63039789e-01 4.61870313e-01
3.57704639e-01 -2.49569029e-01 -8.40359867e-01 -8.34222198e-01
2.07439587e-01 3.41491103e-01 -7.33965337e-01 1.48763165e-01
-4.78657663e-01 -8.02999020e-01 5.91942489e-01 7.25780487e-01
7.53475308e-01 -4.99691993e-01 -7.46795058e-01 2.03387901e-01
-1.73808429e-02 -1.24134934e+00 -9.52647701e-02 8.79761338e-01
-1.10556018e+00 -9.29088891e-01 -4.68971878e-01 -4.42948997e-01
8.81343007e-01 1.09596647e-01 9.64326024e-01 -6.81844875e-02
-4.19106483e-01 2.89493471e-01 3.39396819e-02 -3.09060693e-01
1.77733868e-01 1.68545842e-01 -1.17333859e-01 -3.47379029e-01
6.53862476e-01 -6.45549893e-01 -7.94973135e-01 2.19162792e-01
-7.49578774e-01 -1.70840994e-01 6.59677982e-01 4.97778445e-01
9.17391002e-01 1.18902046e-02 5.58218658e-02 -5.97243190e-01
-5.23218885e-02 -6.59323558e-02 -7.62433112e-01 -2.55910069e-01
-4.10548955e-01 -1.64535944e-03 1.44619280e-02 -4.92189899e-02
-1.08029354e+00 6.07081831e-01 -4.59785998e-01 -2.85575688e-01
-3.58579129e-01 6.27704561e-02 -5.91810942e-01 -1.00948706e-01
4.09105957e-01 -7.04124523e-03 -5.69331683e-02 -7.11805820e-01
1.47764802e-01 5.13092339e-01 6.57401860e-01 -5.25707603e-01
9.28613901e-01 6.40461445e-01 1.97068244e-01 -8.16129506e-01
-1.04348183e+00 -4.66380835e-01 -9.85756278e-01 -1.56828925e-01
9.05901015e-01 -9.67633188e-01 -6.93811119e-01 6.89378381e-01
-1.10356164e+00 -3.19712609e-01 -2.56950647e-01 2.65401274e-01
-2.37548083e-01 1.26455337e-01 -6.00126207e-01 -6.70580626e-01
-1.60199568e-01 -1.24961352e+00 1.57469130e+00 3.80250841e-01
-5.96335642e-02 -5.90344310e-01 -2.85989165e-01 5.94943464e-01
8.88040289e-02 3.34714621e-01 5.12845397e-01 -1.63953915e-01
-8.26378405e-01 -3.24174166e-01 -4.13739324e-01 4.03630972e-01
1.28500387e-01 5.41767329e-02 -1.29417682e+00 9.25469622e-02
4.95463889e-03 -2.36510441e-01 9.22859609e-01 4.94877547e-01
1.27756989e+00 3.35317224e-01 -1.45167131e-02 1.07399666e+00
1.52749014e+00 3.56687903e-02 7.21278310e-01 4.75651622e-01
1.06179512e+00 7.34792769e-01 6.93989038e-01 1.37106612e-01
4.45379317e-01 3.46070975e-01 8.45990241e-01 -2.70777136e-01
-2.37997800e-01 -2.81428397e-01 1.51640818e-01 4.43465799e-01
-5.27681373e-02 -6.36614263e-02 -8.16828847e-01 3.52303684e-01
-1.47898650e+00 -3.22645277e-01 -2.92770267e-01 1.98672473e+00
8.25086951e-01 4.91895586e-01 -9.49304923e-02 2.10964352e-01
3.41068357e-01 1.55544683e-01 -6.43149972e-01 -2.77019173e-01
-2.63339221e-01 3.91754687e-01 7.81888247e-01 4.91250098e-01
-1.15657282e+00 9.46187198e-01 5.49690676e+00 6.37829721e-01
-9.34614301e-01 -1.74277321e-01 6.63143694e-01 -2.27489680e-01
-2.35712364e-01 -1.93799630e-01 -1.05418885e+00 2.18899131e-01
3.85357231e-01 9.00152564e-01 2.56440882e-02 9.28260684e-01
1.91646561e-01 -6.32728696e-01 -1.02972519e+00 1.13978112e+00
-8.24898854e-02 -9.23326015e-01 -1.60681799e-01 1.09489344e-01
5.34422219e-01 8.49001929e-02 -1.28277481e-01 -1.50557935e-01
-2.75039244e-02 -8.35764527e-01 6.54931426e-01 2.78562337e-01
6.70226872e-01 -9.71634448e-01 8.88145030e-01 2.95151889e-01
-9.31187570e-01 2.20332861e-01 -3.86486530e-01 -1.40133038e-01
4.77326773e-02 1.15542877e+00 -7.29109049e-01 3.83543074e-01
9.97457027e-01 5.16260028e-01 -5.88045716e-01 8.20087373e-01
-5.80073059e-01 3.59184116e-01 -6.95587397e-01 3.93376231e-01
2.38673612e-01 -2.65184402e-01 4.20464754e-01 9.95639145e-01
1.60873786e-01 -3.80555540e-02 -7.66354427e-02 8.15239966e-01
-9.39866975e-02 -3.88950586e-01 -4.45130587e-01 3.95567805e-01
3.30481023e-01 1.21823466e+00 -1.06187391e+00 -1.06617972e-01
-2.78729498e-01 1.24351418e+00 5.38934255e-03 9.58940759e-02
-8.09166193e-01 -3.73528689e-01 7.33920395e-01 8.01101252e-02
4.69361097e-01 -4.16185170e-01 -7.92537749e-01 -8.76661420e-01
2.49632284e-01 -5.48008800e-01 7.09087700e-02 -5.92304647e-01
-1.15114474e+00 3.78393918e-01 -2.36243337e-01 -9.94770050e-01
1.24601044e-01 -7.09815800e-01 -2.29261130e-01 7.88024366e-01
-1.49326432e+00 -1.12170506e+00 -7.79653907e-01 4.68090177e-01
3.99693310e-01 5.32237887e-01 5.27220845e-01 5.48610091e-01
-7.79163241e-01 5.39312303e-01 -4.08327103e-01 3.58533651e-01
6.02854252e-01 -1.39843988e+00 5.65844297e-01 8.81072402e-01
-2.17053041e-01 5.73097408e-01 4.13918108e-01 -8.45586181e-01
-1.51921761e+00 -9.91715670e-01 6.85300231e-01 -4.71328437e-01
2.43500397e-01 -6.39872670e-01 -5.52025914e-01 4.58056778e-01
-4.49596912e-01 2.92577893e-01 2.86002934e-01 -2.06046313e-01
-3.15028220e-01 -2.92071253e-01 -1.41610789e+00 4.18139964e-01
1.41089773e+00 -5.08792818e-01 -5.33746302e-01 -5.00387549e-02
7.27044046e-01 -8.40218961e-01 -7.92190075e-01 6.88727856e-01
6.13848388e-01 -1.17435944e+00 1.12854278e+00 1.87341094e-01
4.32983577e-01 -6.75282598e-01 -3.94262820e-01 -7.39812255e-01
1.56356871e-01 -4.29917276e-01 -2.97814518e-01 1.41607451e+00
2.52205551e-01 -4.44935769e-01 1.28291166e+00 7.20550358e-01
-1.53684378e-01 -6.75235212e-01 -1.11432457e+00 -5.64144075e-01
-1.72216892e-01 -7.31238663e-01 5.11939645e-01 4.40980494e-01
-9.22497690e-01 1.49031803e-01 -5.70072494e-02 4.43406194e-01
1.04924464e+00 -6.58751652e-02 9.55362380e-01 -1.24256253e+00
-1.21669263e-01 2.69706175e-02 -5.72588444e-01 -1.58145547e+00
-2.69263387e-01 -4.71686184e-01 4.36106950e-01 -1.63052654e+00
-2.38574624e-01 -6.79265141e-01 6.31228834e-02 5.34729660e-01
-1.13408476e-01 7.58321583e-01 -2.28576884e-01 -6.06002137e-02
-3.08626831e-01 3.59407991e-01 1.39118075e+00 -3.07883322e-02
-3.78869414e-01 9.42270309e-02 -4.68913376e-01 1.07415557e+00
6.61158085e-01 -5.67248046e-01 -4.18369681e-01 -6.46152139e-01
1.58969089e-01 -4.83731240e-01 5.41923583e-01 -1.21090806e+00
2.39537191e-02 1.09693490e-01 8.41047764e-01 -1.20860028e+00
5.57070255e-01 -1.10307157e+00 -1.85326368e-01 2.96846181e-01
1.84453726e-01 -1.69321746e-01 1.32592201e-01 4.99824405e-01
-5.82460389e-02 -1.75881952e-01 7.59053826e-01 -2.59162039e-01
-8.84486556e-01 2.79871583e-01 -1.49690121e-01 -3.01008746e-02
9.33249891e-01 -6.78534448e-01 -1.54159814e-01 6.20539114e-02
-7.30453670e-01 1.94205135e-01 5.44314742e-01 2.51930177e-01
5.33398271e-01 -9.24269557e-01 -2.08789513e-01 4.67389643e-01
-1.43134203e-02 8.69667709e-01 2.12789938e-01 7.43507445e-01
-8.09705436e-01 -4.87670302e-03 5.32061160e-02 -1.07541883e+00
-1.01276422e+00 1.09491244e-01 4.51555371e-01 1.06041573e-01
-6.84631169e-01 1.34925854e+00 3.49389203e-02 -7.49661028e-01
6.10582530e-01 -6.34367824e-01 2.03391045e-01 2.86756158e-01
2.50698179e-01 3.12423080e-01 4.19920683e-01 -4.76676702e-01
-5.85837007e-01 7.77239561e-01 6.74943253e-02 -4.49090935e-02
1.62284005e+00 -2.47452155e-01 -2.62258381e-01 2.74240971e-01
1.22340214e+00 -7.88833573e-02 -1.70778286e+00 1.03183933e-01
-2.37503741e-02 -5.24195075e-01 3.16066116e-01 -7.71929502e-01
-1.33046806e+00 8.13400745e-01 7.01819956e-01 -2.97799706e-01
1.23255205e+00 -2.05570161e-02 9.70357358e-01 2.64414102e-01
4.51065093e-01 -1.27323544e+00 -1.32083803e-01 5.43841839e-01
4.24150527e-01 -1.27978969e+00 1.65410161e-01 -6.76947534e-01
-2.23104581e-01 1.07363582e+00 1.00547242e+00 -9.13082212e-02
5.36569774e-01 6.92876041e-01 3.63853365e-01 -4.20677930e-01
-3.73015478e-02 -4.00485605e-01 2.05928758e-01 7.55773544e-01
1.91761941e-01 -3.90777439e-01 -9.06685516e-02 3.20542365e-01
-1.62148625e-01 -2.51531750e-01 8.71973932e-02 1.21294057e+00
-3.83730143e-01 -1.18661392e+00 -4.66367334e-01 3.75684470e-01
-3.44988674e-01 8.37981999e-02 -4.79403555e-01 8.13599110e-01
8.11755180e-01 7.47908175e-01 2.39510790e-01 -5.18325984e-01
4.74593401e-01 -6.04861155e-02 7.03412712e-01 -6.23020887e-01
-6.80784345e-01 3.62577766e-01 1.52057149e-02 -1.10166216e+00
-6.11647010e-01 -5.68330944e-01 -1.47649240e+00 -3.84088680e-02
-2.14961275e-01 -3.52341831e-01 1.10025954e+00 9.31878686e-01
4.16688740e-01 5.99656820e-01 3.84841740e-01 -1.08695459e+00
-3.21668327e-01 -6.86460137e-01 -4.19442862e-01 5.55744134e-02
4.63736773e-01 -6.84064984e-01 -5.95214069e-01 -2.23952159e-01] | [8.034953117370605, -2.860666275024414] |
070f7a93-9827-4ffb-ab06-2e849dfeed37 | a-novel-filter-based-on-three-variables | 2210.14609 | null | https://arxiv.org/abs/2210.14609v1 | https://arxiv.org/pdf/2210.14609v1.pdf | A novel filter based on three variables mutual information for dimensionality reduction and classification of hyperspectral images | The high dimensionality of hyperspectral images (HSI) that contains more than hundred bands (images) for the same region called Ground Truth Map, often imposes a heavy computational burden for image processing and complicates the learning process. In fact, the removal of irrelevant, noisy and redundant bands helps increase the classification accuracy. Band selection filter based on "Mutual Information" is a common technique for dimensionality reduction. In this paper, a categorization of dimensionality reduction methods according to the evaluation process is presented. Moreover, a new filter approach based on three variables mutual information is developed in order to measure band correlation for classification, it considers not only bands relevance but also bands interaction. The proposed approach is compared to a reproduced filter algorithm based on mutual information. Experimental results on HSI AVIRIS 92AV3C have shown that the proposed approach is very competitive, effective and outperforms the reproduced filter strategy performance. Keywords - Hyperspectral images, Classification, band Selection, Three variables Mutual Information, information gain. | ['Chafik Nacir', 'Ahmed Hammouch', 'Elkebir Sarhrouni', 'Asma Elmaizi'] | 2022-10-26 | null | null | null | null | ['classification-of-hyperspectral-images'] | ['computer-vision'] | [ 6.06416702e-01 -4.89020109e-01 2.96669513e-01 -2.36255199e-01
-2.17988163e-01 -4.34427679e-01 2.96666861e-01 1.01797774e-01
-3.79345179e-01 8.35086107e-01 3.35649066e-02 4.70086224e-02
-9.81312156e-01 -1.04756999e+00 1.33780137e-01 -1.08781040e+00
-5.12362830e-02 -1.74519401e-02 -1.30568817e-01 -6.71608225e-02
3.38817775e-01 5.48779607e-01 -2.30045867e+00 1.74951077e-01
1.18882346e+00 9.80317116e-01 5.99205613e-01 4.00180191e-01
-1.75838619e-01 2.44845912e-01 -3.32277238e-01 2.52147943e-01
4.36016440e-01 -5.56420803e-01 -7.36063480e-01 4.42223877e-01
-1.36183202e-01 3.04162372e-02 1.81796119e-01 1.44885612e+00
4.97405916e-01 3.94452631e-01 1.11623132e+00 -6.75876081e-01
-2.21555337e-01 5.78436077e-01 -8.43413174e-01 1.59201086e-01
-1.04606703e-01 -1.58291116e-01 8.90542030e-01 -6.96278393e-01
2.31070817e-01 9.25647497e-01 5.69250166e-01 -2.20210880e-01
-1.28385782e+00 -6.18225694e-01 -1.51123494e-01 5.16138256e-01
-1.57015693e+00 1.61666244e-01 7.54063427e-01 -8.29621196e-01
5.89008868e-01 8.37188780e-01 8.24507833e-01 -1.07163571e-01
5.69238812e-02 2.43433788e-01 1.55625892e+00 -6.49910450e-01
1.63735691e-02 3.31466645e-01 6.43314958e-01 3.44766438e-01
4.45524424e-01 3.43399346e-01 -1.88482076e-01 -1.05879635e-01
5.35106659e-03 -8.42510909e-02 -6.73578084e-01 -2.94095725e-01
-7.39882886e-01 8.75050366e-01 3.89711171e-01 7.18130350e-01
-5.28587937e-01 -8.00530612e-01 1.45971403e-01 4.14400369e-01
4.58938837e-01 3.49404901e-01 -3.11761677e-01 3.45620155e-01
-7.66605675e-01 -7.44472519e-02 6.89371169e-01 4.13132459e-01
1.17680597e+00 -6.52746037e-02 4.32535820e-03 8.72811258e-01
4.35405314e-01 6.36645854e-01 5.13934970e-01 -2.17801645e-01
-1.27008215e-01 9.99327540e-01 -7.20208287e-02 -1.54193735e+00
-6.91476941e-01 -6.60690904e-01 -1.04830790e+00 3.30705911e-01
-2.74145365e-01 -1.68857232e-01 -7.24446356e-01 1.15978086e+00
2.17297181e-01 -2.34992325e-01 3.12690765e-01 1.15783095e+00
7.87912190e-01 7.57192254e-01 1.51862241e-02 -6.17163181e-01
1.29644215e+00 -1.61007151e-01 -9.02468979e-01 7.82228783e-02
2.66877860e-01 -1.05633044e+00 5.39973080e-01 6.72529817e-01
-5.64281642e-01 -4.75273252e-01 -1.21411562e+00 6.51966631e-01
-7.70060122e-01 4.17633474e-01 6.88887477e-01 8.59273553e-01
-5.75397730e-01 7.02068686e-01 -3.15967411e-01 -4.64250237e-01
1.38016969e-01 3.31098914e-01 -4.55698401e-01 4.22334373e-02
-1.04935300e+00 9.00120318e-01 1.00908148e+00 1.31301448e-01
8.96520838e-02 -3.59245718e-01 -5.16353488e-01 1.08123552e-02
1.32538974e-02 -1.92508981e-01 3.34431142e-01 -1.04414403e+00
-1.26572192e+00 6.51288033e-01 5.73278144e-02 -2.69038171e-01
2.64000595e-01 -2.37489454e-02 -6.71071291e-01 2.57187247e-01
-1.51621625e-01 4.08103876e-02 6.36577487e-01 -1.12078953e+00
-9.17772293e-01 -7.86650360e-01 -4.88437116e-01 4.92623031e-01
-5.47853053e-01 -2.44039908e-01 2.23331973e-01 -3.59080672e-01
8.77447546e-01 -6.90361619e-01 -8.57679024e-02 -4.64277536e-01
-1.67713806e-01 2.73384571e-01 9.93532956e-01 -7.11088300e-01
1.16279328e+00 -2.22417498e+00 8.81701708e-02 7.85637259e-01
-1.99490383e-01 1.91144839e-01 1.79402381e-01 3.06006521e-01
-3.97523046e-01 -3.48969770e-04 -6.86228573e-01 5.98946929e-01
-2.52799124e-01 -7.54049420e-02 1.54429153e-01 6.94816172e-01
-9.17420164e-02 -2.81801403e-01 -5.63481987e-01 -2.88187295e-01
5.77310145e-01 6.06574595e-01 -2.89263934e-01 3.69915292e-02
3.58170629e-01 3.72336805e-01 -1.67442262e-01 3.21555525e-01
1.30275989e+00 -6.18798025e-02 9.99556184e-02 -1.04103982e+00
-6.37694538e-01 -5.73090434e-01 -1.57060885e+00 1.31192172e+00
-1.12597622e-01 6.35302782e-01 -7.78917521e-02 -1.13046920e+00
1.19044018e+00 1.93389684e-01 7.02625275e-01 -2.57206023e-01
3.75075012e-01 2.01209709e-01 5.89930378e-02 -7.11807311e-01
2.57401586e-01 -9.08329412e-02 7.30084598e-01 9.07809287e-02
-2.10997108e-02 -3.10632676e-01 4.60103840e-01 -3.42647344e-01
3.80201608e-01 -3.76378037e-02 6.92091763e-01 -6.87554538e-01
1.02103949e+00 1.55834153e-01 4.15028989e-01 2.73042887e-01
4.72626649e-02 2.22282723e-01 9.64201707e-03 -2.45678216e-01
-6.84332550e-01 -4.99324054e-01 -5.10128260e-01 4.23249990e-01
3.52954537e-01 1.51621476e-01 -3.33693445e-01 -1.71125069e-01
5.57878017e-02 5.47035336e-01 -3.56832176e-01 -2.46909499e-01
2.67955393e-01 -1.40953720e+00 -4.06494737e-02 -3.75329018e-01
1.11222732e+00 -3.67894590e-01 -5.78308165e-01 2.98644602e-01
-1.38448805e-01 -3.34367245e-01 3.04892451e-01 1.59530655e-01
-9.93553579e-01 -1.30699277e+00 -3.25820357e-01 -3.67393136e-01
4.56668764e-01 7.66898274e-01 7.12674499e-01 -1.63040236e-01
-8.54133427e-01 1.32158428e-01 -7.91774213e-01 -5.81468284e-01
-1.11817539e-01 -2.96704024e-01 -3.03299516e-01 3.56011569e-01
7.64409423e-01 -5.67309201e-01 -5.54885566e-01 3.30186576e-01
-9.49146032e-01 -2.44118217e-02 9.20457661e-01 9.76255774e-01
5.80744386e-01 1.00431788e+00 3.11244965e-01 -7.66092002e-01
4.78608876e-01 -2.35247359e-01 -9.40200508e-01 1.90170690e-01
-9.06243980e-01 -9.79704857e-02 1.29008293e-01 3.22751164e-01
-1.46857035e+00 2.91003227e-01 1.82729229e-01 2.81277239e-01
-2.91569322e-01 8.30567956e-01 -1.85864523e-01 -5.19698501e-01
8.99716377e-01 5.08933067e-01 2.38072798e-01 -6.31318986e-01
9.20186490e-02 8.51856291e-01 1.66342542e-01 1.20833322e-01
8.62840772e-01 4.11585450e-01 1.98913231e-01 -1.25695753e+00
-4.98203725e-01 -8.69409680e-01 -6.69770420e-01 -3.24718446e-01
9.19970691e-01 -8.30212116e-01 -6.85479283e-01 4.02978718e-01
-8.62858951e-01 6.21844232e-01 2.27950037e-01 1.16466868e+00
-2.62597948e-01 4.00655359e-01 -1.21699631e-01 -1.18475676e+00
-4.17646796e-01 -1.00144506e+00 4.23386812e-01 4.16361690e-01
4.25088257e-01 -7.40707636e-01 -1.82062581e-01 1.71433076e-01
3.76462996e-01 3.75792593e-01 9.71360087e-01 -4.25619364e-01
-2.68327475e-01 -1.35008395e-01 -6.39557183e-01 6.02216840e-01
6.39601409e-01 5.51104583e-02 -9.00182426e-01 -8.01172778e-02
2.90944159e-01 2.49141783e-01 9.71855223e-01 5.36681831e-01
9.40616667e-01 -1.15669347e-01 -2.58574426e-01 6.37179613e-01
2.18397570e+00 6.34406447e-01 6.58854842e-01 4.63132471e-01
3.82802427e-01 8.02636921e-01 9.84173715e-01 7.82163024e-01
-2.69522786e-01 -4.69622649e-02 5.93426108e-01 -3.23179573e-01
2.52002746e-01 7.25838602e-01 -3.44725102e-01 7.63313890e-01
-3.40648681e-01 -6.35090098e-02 -7.44195938e-01 8.00420865e-02
-1.51613283e+00 -1.18985343e+00 -6.15215182e-01 2.20497251e+00
4.81114507e-01 -4.06895876e-01 -2.09237635e-01 8.80231977e-01
9.78452444e-01 9.30026546e-02 -1.72818020e-01 9.89023373e-02
-5.65336347e-01 1.29605815e-01 9.07962799e-01 5.75971961e-01
-1.52942920e+00 3.97285640e-01 5.32561731e+00 5.30560851e-01
-1.13467288e+00 -6.12944476e-02 2.72849381e-01 4.22997981e-01
-1.85514218e-03 -9.38488543e-02 -3.70856345e-01 3.58303696e-01
2.96465069e-01 -2.73410231e-01 4.20290649e-01 4.78739947e-01
3.56865585e-01 -5.34067869e-01 -1.78411335e-01 1.31842351e+00
1.35309860e-01 -9.21505392e-01 1.45384744e-01 1.27810806e-01
8.24449122e-01 -2.23420724e-01 -1.15049064e-01 -3.43371928e-01
7.91166257e-03 -5.30930161e-01 1.01196937e-01 8.66884708e-01
1.36496782e-01 -1.22321606e+00 1.06792784e+00 1.87549457e-01
-1.24322999e+00 -3.59509766e-01 -6.78099215e-01 -7.41540939e-02
-3.80631953e-01 1.02878714e+00 -6.09540224e-01 1.17123950e+00
6.73858345e-01 7.78238654e-01 -5.21824002e-01 1.46243536e+00
2.07352549e-01 2.91001022e-01 -2.25261167e-01 -2.06931978e-02
2.54621714e-01 -8.34293783e-01 7.11776555e-01 1.04410303e+00
6.70315325e-01 5.14981449e-01 -8.03148653e-03 7.43656039e-01
6.74007118e-01 9.30797696e-01 -6.34876609e-01 -1.17716581e-01
3.73538762e-01 1.16675639e+00 -7.16786802e-01 -1.94722131e-01
-2.82818645e-01 7.25792110e-01 -5.76212406e-01 1.47139847e-01
-4.61948216e-01 -6.98351324e-01 3.95047188e-01 -4.14339453e-01
1.87002774e-02 -2.34013181e-02 -1.10938154e-01 -8.10031176e-01
-3.23874652e-01 -7.92510271e-01 6.48616672e-01 -5.82716107e-01
-1.19486237e+00 6.57751262e-01 -2.70059537e-02 -1.70864415e+00
2.21758574e-01 -8.70416343e-01 -3.08649745e-02 1.22901380e+00
-1.56706405e+00 -6.96553051e-01 -8.76375914e-01 6.23368800e-01
3.08434725e-01 -3.19906622e-01 1.02046847e+00 5.03112614e-01
-6.27108097e-01 -2.09118843e-01 6.32131338e-01 -3.78271222e-01
3.96521002e-01 -1.05634880e+00 -8.97083044e-01 8.81116211e-01
-4.45189685e-01 1.01325870e-01 8.49480808e-01 -6.64090216e-01
-1.18596125e+00 -9.03373301e-01 2.62081057e-01 5.89366376e-01
3.92114401e-01 5.94712198e-01 -8.12245488e-01 8.30409229e-02
1.41147017e-01 -4.54223394e-01 1.28999889e+00 -1.09628670e-01
1.33527562e-01 -4.41454351e-01 -1.20858312e+00 2.88942665e-01
5.75061381e-01 -1.42988220e-01 -3.19066346e-01 7.67721176e-01
1.18158408e-01 2.36127242e-01 -1.08877778e+00 6.55320168e-01
5.93469501e-01 -1.36887515e+00 7.54835963e-01 -3.50728631e-02
-1.23813406e-01 -6.54246390e-01 -3.13134491e-01 -1.29968631e+00
-7.06740022e-01 -5.42532429e-02 7.00108349e-01 1.18159342e+00
4.70942527e-01 -7.51421690e-01 5.46111882e-01 2.51894176e-01
1.67213306e-01 2.22600903e-03 -5.23500919e-01 -8.12967241e-01
-6.19586945e-01 1.67060792e-02 7.63919890e-01 1.12975085e+00
-9.64147821e-02 1.45833984e-01 -2.75109500e-01 6.40600562e-01
9.68382061e-01 3.31100941e-01 4.38392371e-01 -1.99512744e+00
-2.21165106e-01 -6.82392776e-01 -8.36125255e-01 -1.21655993e-01
-1.73279434e-01 -7.27485597e-01 -1.40767500e-01 -1.55438662e+00
1.20286874e-01 -2.10557505e-01 -3.18681896e-01 -8.52147043e-02
-1.51436746e-01 3.04962397e-01 9.97051895e-02 3.55865985e-01
3.04435581e-01 3.52252215e-01 9.48128462e-01 -3.85293305e-01
-4.60331112e-01 1.03103712e-01 -4.60987121e-01 6.61044955e-01
8.69448125e-01 -2.02784821e-01 -5.78133702e-01 -1.78485394e-01
2.19645258e-02 -1.15241095e-01 -7.84480646e-02 -1.47163272e+00
1.71492055e-01 -2.04140678e-01 6.12422228e-01 -9.13302720e-01
2.42579818e-01 -1.29165447e+00 6.86869502e-01 7.02067196e-01
-3.94441709e-02 -4.39155698e-01 6.76572844e-02 4.13992196e-01
-5.16869426e-01 -5.60725808e-01 1.04879189e+00 -3.70462626e-01
-9.38635290e-01 1.12788178e-01 -3.81177396e-01 -9.43836808e-01
9.71190155e-01 -5.73276937e-01 3.15301906e-04 -2.99476348e-02
-9.65539932e-01 -1.36720717e-01 4.43933792e-02 -1.42187640e-01
3.98719907e-01 -1.15259552e+00 -8.89110446e-01 2.22728133e-01
2.67231047e-01 -5.80381095e-01 6.54683530e-01 8.06478024e-01
-8.67236197e-01 2.73117155e-01 -5.24931490e-01 -6.17355525e-01
-1.62792099e+00 2.20581964e-01 3.02547693e-01 -1.35015706e-02
-9.74663198e-02 7.82904983e-01 -2.61532515e-02 -9.69271958e-02
-1.59978136e-01 8.16999972e-02 -1.06997156e+00 6.83527291e-01
5.76305270e-01 5.85853279e-01 3.10515493e-01 -9.46884215e-01
-1.52852446e-01 8.76625121e-01 4.27677840e-01 1.88131388e-02
1.29275084e+00 -3.62255156e-01 -6.23399556e-01 2.53718853e-01
1.12802303e+00 -1.85203537e-01 -6.16859138e-01 -2.83959866e-01
-1.86498668e-02 -9.86568332e-01 5.13093352e-01 -8.50689292e-01
-9.99726653e-01 6.26246393e-01 1.50330615e+00 5.11874616e-01
1.67951179e+00 -8.99631023e-01 -1.10181347e-01 6.09981298e-01
2.81375021e-01 -1.24897480e+00 -5.87101638e-01 3.01500142e-01
7.80084670e-01 -1.38844180e+00 3.44627947e-01 -8.06798577e-01
-2.69515604e-01 1.49001312e+00 4.11598951e-01 1.13058791e-01
1.12583339e+00 -2.79292047e-01 -1.33775994e-01 -1.71803042e-01
-4.51054564e-03 -7.53441989e-01 2.79608995e-01 5.73424518e-01
6.75325632e-01 2.82652497e-01 -1.13711178e+00 2.28527769e-01
-2.31865585e-01 -2.23115176e-01 2.42622301e-01 9.20608461e-01
-1.09431338e+00 -8.09983432e-01 -8.90076518e-01 7.34274328e-01
-1.69530913e-01 -7.54233375e-02 -2.10909709e-01 7.35853612e-01
4.46851611e-01 1.19267929e+00 -8.65030363e-02 -5.32561004e-01
3.87190700e-01 -1.54725369e-02 3.01691949e-01 -3.38686407e-01
-2.63479888e-01 1.69638872e-01 -9.72935185e-02 -1.88291281e-01
-8.69964421e-01 -5.82557619e-01 -8.39663744e-01 5.25332941e-03
-9.62209702e-01 5.22336185e-01 1.15319574e+00 8.16428840e-01
7.91424885e-02 4.12700683e-01 9.65051711e-01 -7.96242177e-01
-3.66578072e-01 -1.09741795e+00 -1.18106341e+00 2.90580004e-01
7.63150901e-02 -9.29346383e-01 -6.92690313e-01 1.45375028e-01] | [9.774955749511719, -1.8509254455566406] |
7e77afd9-4de5-4dce-a44e-e802d0c5a469 | extending-word-level-quality-estimation-for | 2209.11378 | null | https://arxiv.org/abs/2209.11378v1 | https://arxiv.org/pdf/2209.11378v1.pdf | Extending Word-Level Quality Estimation for Post-Editing Assistance | We define a novel concept called extended word alignment in order to improve post-editing assistance efficiency. Based on extended word alignment, we further propose a novel task called refined word-level QE that outputs refined tags and word-level correspondences. Compared to original word-level QE, the new task is able to directly point out editing operations, thus improves efficiency. To extract extended word alignment, we adopt a supervised method based on mBERT. To solve refined word-level QE, we firstly predict original QE tags by training a regression model for sequence tagging based on mBERT and XLM-R. Then, we refine original word tags with extended word alignment. In addition, we extract source-gap correspondences, meanwhile, obtaining gap tags. Experiments on two language pairs show the feasibility of our method and give us inspirations for further improvement. | ['Masaaki Nagata', 'Takehito Utsuro', 'Yizhen Wei'] | 2022-09-23 | null | null | null | null | ['word-alignment', 'xlm-r'] | ['natural-language-processing', 'natural-language-processing'] | [ 5.33548534e-01 -9.43932161e-02 -2.57225454e-01 -2.70308018e-01
-1.03062212e+00 -6.32850528e-01 1.10213973e-01 2.61640191e-01
-7.98253000e-01 7.06401825e-01 3.95021558e-01 -4.42662448e-01
-1.76107883e-02 -7.80805588e-01 -5.10991216e-01 -2.48328283e-01
3.31262767e-01 2.97846645e-01 1.93665728e-01 -5.75757504e-01
5.57719171e-01 -8.03717077e-02 -9.70426142e-01 1.60397023e-01
1.51135528e+00 2.98724562e-01 8.02076876e-01 3.13033253e-01
-5.66357374e-01 1.42347112e-01 -6.04295194e-01 -6.98462307e-01
9.52094495e-02 -4.87991869e-01 -8.04249704e-01 -1.16131775e-01
-1.26117200e-01 6.44269437e-02 1.07071333e-01 1.16173005e+00
4.51701432e-01 1.85055360e-01 3.32269192e-01 -8.65875542e-01
-8.42806458e-01 8.90956879e-01 -4.59846914e-01 1.82402462e-01
5.74262798e-01 -1.89407349e-01 1.41279161e+00 -1.22648227e+00
5.46581328e-01 7.12907255e-01 8.00722897e-01 5.92417717e-01
-8.93258512e-01 -5.72360694e-01 1.88087538e-01 2.06075117e-01
-1.38171887e+00 -2.14047536e-01 4.22210604e-01 -1.08727768e-01
1.21318424e+00 2.25231513e-01 4.48234886e-01 7.89509714e-01
-3.06405406e-03 8.20957720e-01 7.65149832e-01 -7.96715021e-01
-2.14282557e-01 -1.42917514e-01 2.94285148e-01 9.97060478e-01
4.99826595e-02 -3.20901990e-01 -4.49967563e-01 -6.13175184e-02
4.22862113e-01 -8.77775773e-02 -3.42981905e-01 -5.81048802e-02
-1.58175516e+00 5.75924754e-01 -1.28233269e-01 6.06844902e-01
-1.66402385e-01 -1.73187092e-01 4.16849166e-01 1.97773099e-01
5.53967834e-01 7.93175519e-01 -6.63090289e-01 -3.33609343e-01
-9.02141154e-01 -1.57056436e-01 4.04187173e-01 1.30945814e+00
1.05667675e+00 -4.09419626e-01 -5.97872674e-01 1.12557077e+00
3.17549594e-02 5.29459894e-01 7.98807085e-01 -5.40447712e-01
6.96150839e-01 6.84734702e-01 1.58978179e-01 -9.00708199e-01
-2.80671567e-01 -7.61605144e-01 -7.35375106e-01 -6.38173223e-01
-3.56612471e-03 2.33748052e-02 -5.74143887e-01 1.76321912e+00
-4.31685820e-02 3.52144092e-01 -3.50891016e-02 6.53191388e-01
3.55683357e-01 7.33719707e-01 -2.37729140e-02 -3.07706296e-01
1.51911497e+00 -1.36467981e+00 -9.29326355e-01 -3.62203091e-01
1.33939862e+00 -8.12568367e-01 1.30308950e+00 2.10467771e-01
-1.02258468e+00 -6.21619761e-01 -9.82581675e-01 -1.81818828e-01
-1.91972405e-01 6.56187713e-01 3.50909680e-01 5.35812855e-01
-7.36863792e-01 6.36004269e-01 -7.23886788e-01 -2.79294044e-01
-8.33412707e-02 1.98544413e-01 -4.04804826e-01 5.42814843e-02
-1.51453483e+00 1.10857201e+00 5.62807620e-01 2.10217103e-01
-2.58141756e-02 -6.08641922e-01 -1.06356430e+00 1.59762979e-01
4.93793339e-01 -5.51744342e-01 1.12757957e+00 -5.78982711e-01
-1.34185135e+00 9.24856901e-01 -5.39340436e-01 -6.47034720e-02
9.56418663e-02 -3.44549090e-01 -8.18296731e-01 -3.03835809e-01
3.81250113e-01 3.00281376e-01 3.15631300e-01 -7.92298913e-01
-1.11361444e+00 -1.12049825e-01 -2.80137420e-01 1.53928265e-01
-7.47935653e-01 2.48832256e-01 -9.01614785e-01 -1.07459867e+00
-2.06155442e-02 -7.63938069e-01 -5.20637572e-01 -6.51707649e-01
-3.77128273e-01 -3.19408625e-01 -4.57932837e-02 -1.21102834e+00
1.98958552e+00 -1.96688151e+00 3.12973797e-01 4.43740934e-01
3.46381277e-01 5.75888932e-01 -5.75496674e-01 5.25953233e-01
-3.82274762e-02 3.06584418e-01 -6.73174679e-01 -3.92618835e-01
7.41463378e-02 2.70624340e-01 -2.34896034e-01 7.28499517e-03
4.11473215e-01 1.07307327e+00 -1.15382588e+00 -3.74695510e-01
-1.48413837e-01 -1.14760630e-01 -4.52738732e-01 3.25599045e-01
-7.93733299e-02 3.16962510e-01 -5.26299298e-01 2.32068241e-01
6.79691851e-01 -2.34837551e-02 3.58794719e-01 -2.36141592e-01
-2.02613056e-01 4.81400430e-01 -8.94055784e-01 2.09660482e+00
-8.43040526e-01 2.94834435e-01 -3.97196114e-01 -1.17872298e+00
1.23896658e+00 8.14282894e-02 4.02240008e-01 -7.48066843e-01
-1.46059409e-01 3.78502071e-01 -2.05367878e-01 -4.99482274e-01
1.17613065e+00 2.05347076e-01 -4.28830087e-01 4.45113510e-01
9.95067134e-02 3.30153883e-01 2.02892438e-01 2.22150460e-01
1.20208418e+00 4.88890886e-01 5.80707073e-01 -1.68344826e-01
7.91359067e-01 -2.31787544e-02 8.29364300e-01 5.05817413e-01
1.73875764e-01 2.17841744e-01 1.46315306e-01 1.94545686e-02
-1.11430097e+00 -7.24761844e-01 2.39226893e-01 1.29044676e+00
4.22429770e-01 -1.21872532e+00 -8.20682347e-01 -8.95451128e-01
-3.94046038e-01 7.16330469e-01 -2.30732128e-01 -3.37689131e-01
-1.03636050e+00 -6.49553359e-01 8.57643247e-01 5.57446063e-01
2.15808794e-01 -7.75953233e-01 -2.86640860e-02 5.82717896e-01
-8.14193845e-01 -1.11770499e+00 -1.14670968e+00 -1.01643600e-01
-6.67677999e-01 -7.40583718e-01 -7.64184713e-01 -1.06064010e+00
7.25890577e-01 2.53756166e-01 1.11449993e+00 4.20244843e-01
2.01781858e-02 -7.80568570e-02 -8.03687811e-01 3.56419943e-02
-4.52843606e-01 6.44911289e-01 1.85703989e-02 -3.34866971e-01
6.29631162e-01 -4.13190812e-01 -2.58869022e-01 5.15002370e-01
-7.48229027e-01 2.47831672e-01 8.53149414e-01 1.04615784e+00
5.94744325e-01 -3.75185817e-01 5.19659340e-01 -9.83587384e-01
7.70830333e-01 -2.16337547e-01 -4.72301334e-01 8.44194651e-01
-7.97294021e-01 5.18903017e-01 6.24270380e-01 -2.86997437e-01
-1.11201215e+00 -1.90372020e-01 -5.63833654e-01 1.83947399e-01
1.88163400e-01 7.86869049e-01 -3.90671819e-01 3.07726711e-01
3.41441691e-01 3.65234226e-01 -3.92810404e-01 -7.59357691e-01
6.32582724e-01 1.01974845e+00 8.24284673e-01 -7.62583196e-01
9.77306664e-01 -2.16123000e-01 -4.03008223e-01 -2.86869198e-01
-8.52290690e-01 -5.52418709e-01 -7.63971806e-01 2.90909231e-01
7.97850013e-01 -7.39702821e-01 -4.55405951e-01 1.35846511e-01
-1.27358425e+00 -1.45304263e-01 -3.35261300e-02 4.48070735e-01
-3.84418905e-01 8.61503303e-01 -5.57099700e-01 -4.15512234e-01
-5.21153867e-01 -9.35525358e-01 1.12954485e+00 -7.82516822e-02
-3.27284724e-01 -9.59105432e-01 3.22197497e-01 2.29734018e-01
1.71357557e-01 -4.55671221e-01 1.06113291e+00 -8.20760131e-01
-3.17427337e-01 2.64503770e-02 -2.04235345e-01 2.61760861e-01
3.06751937e-01 -2.01919153e-01 -8.58598202e-02 -2.07816899e-01
-5.45305133e-01 2.09313482e-01 7.96308517e-01 -2.37489417e-01
1.16276360e+00 -1.61335990e-01 -4.46952581e-01 6.44562244e-01
1.15079474e+00 -1.16516635e-01 5.81555367e-01 5.69728315e-01
7.97574043e-01 3.44406843e-01 1.28761089e+00 5.28455019e-01
5.00885129e-01 9.99378443e-01 -1.29318848e-01 -2.08298475e-01
8.22265819e-02 -4.63447839e-01 5.95999002e-01 1.62548327e+00
-1.38699323e-01 -2.42505178e-01 -8.68450284e-01 5.35054922e-01
-1.97982049e+00 -6.70704067e-01 -2.18760788e-01 2.20676446e+00
1.03031158e+00 -2.70977803e-02 -7.95196816e-02 -3.86665724e-02
9.87274885e-01 -1.48915082e-01 2.08066069e-02 -3.11032951e-01
-1.19858965e-01 7.15471208e-01 7.08721995e-01 6.27884805e-01
-6.61258042e-01 1.49799705e+00 5.51774120e+00 1.25743937e+00
-6.67986870e-01 3.55769813e-01 -8.92821699e-02 4.56758499e-01
-8.10569465e-01 1.60770565e-01 -1.12095630e+00 6.06720448e-01
6.40886366e-01 -3.20991457e-01 4.81635541e-01 4.85490978e-01
6.33150563e-02 3.26465130e-01 -8.74027252e-01 8.02804410e-01
2.33078450e-01 -1.12160087e+00 8.11031535e-02 -1.71966642e-01
4.51120138e-01 -2.17412040e-01 -3.75844032e-01 5.52631617e-01
4.64825600e-01 -6.40379786e-01 5.29723465e-01 5.74434698e-01
1.16467392e+00 -7.50527024e-01 8.65296304e-01 4.30427134e-01
-1.44707489e+00 1.78988591e-01 -6.20708227e-01 1.06582507e-01
4.28912133e-01 6.96541190e-01 -6.33203447e-01 1.09753680e+00
1.07852392e-01 9.63856637e-01 -5.06298423e-01 8.58805895e-01
-6.64946079e-01 5.33268750e-01 1.81740999e-01 -9.74348411e-02
6.33543879e-02 -5.33636272e-01 4.98243123e-01 1.73071909e+00
9.05732989e-01 3.00226714e-02 1.85916722e-01 4.57720637e-01
-1.64277956e-01 5.28720081e-01 -2.14817794e-03 -6.00376874e-02
7.87913322e-01 1.42378008e+00 -3.72432202e-01 -3.43371600e-01
-4.63139057e-01 1.68514621e+00 7.43539751e-01 1.60215005e-01
-9.05937433e-01 -8.54323089e-01 7.31725037e-01 -4.41750467e-01
3.26113373e-01 -4.72581595e-01 1.31576091e-01 -1.59890330e+00
1.14241563e-01 -8.94608259e-01 2.90877759e-01 -7.68624961e-01
-1.06758618e+00 6.23923659e-01 -4.12855625e-01 -1.32410407e+00
-1.69071123e-01 -5.84480762e-01 -3.73145610e-01 1.07300794e+00
-1.58368719e+00 -1.10682213e+00 -1.77229479e-01 3.55169088e-01
4.89629388e-01 -2.54342526e-01 1.03796184e+00 4.65183824e-01
-7.50603497e-01 1.02065694e+00 2.12595284e-01 4.04306233e-01
1.15181732e+00 -1.21423554e+00 8.62828732e-01 1.04999840e+00
4.91743833e-01 9.43064690e-01 3.82229179e-01 -8.31583798e-01
-9.89950061e-01 -1.38692844e+00 1.58389843e+00 -3.89378399e-01
5.90613186e-01 -5.17015994e-01 -9.22070563e-01 7.37755537e-01
1.56256631e-02 -4.69432503e-01 9.99955058e-01 2.56451488e-01
-3.42561990e-01 -1.49554629e-02 -6.00621581e-01 7.62723327e-01
1.64168429e+00 -6.20315671e-01 -1.05855227e+00 1.62442669e-01
1.09138262e+00 -4.21421349e-01 -1.03113008e+00 4.77641732e-01
3.42714369e-01 -2.93706715e-01 7.25482941e-01 -6.90633476e-01
3.27948123e-01 -4.08563286e-01 9.93206818e-03 -1.76410127e+00
-6.89233840e-01 -6.37982368e-01 9.77392793e-02 1.54252613e+00
7.09721029e-01 -4.63394970e-01 5.02663851e-01 6.38140086e-03
-5.86086094e-01 -6.79700494e-01 -4.91896868e-01 -1.02638721e+00
-2.02321798e-01 -4.87282962e-01 9.98306572e-01 1.15126646e+00
3.76831979e-01 4.40933168e-01 -5.88991940e-01 2.06818759e-01
1.82700068e-01 7.57559240e-02 5.75522602e-01 -9.69854414e-01
-4.96333778e-01 -2.93150365e-01 -1.08347446e-01 -1.59510291e+00
5.84098220e-01 -1.10576296e+00 2.93807685e-01 -1.46077156e+00
3.33599120e-01 -6.78819418e-01 -6.16246343e-01 6.20589316e-01
-8.45477223e-01 3.65472049e-01 -9.27109718e-02 2.27192327e-01
-6.74775004e-01 6.07834399e-01 8.53977025e-01 4.81330603e-02
-9.90916118e-02 -1.72437683e-01 -6.66067421e-01 2.83841521e-01
7.83455372e-01 -4.61897999e-01 -5.81101701e-03 -7.39930809e-01
6.77649081e-01 -1.11267656e-01 -2.59777904e-01 -5.57753980e-01
1.13218926e-01 -1.60952821e-01 -1.17819816e-01 -3.61000866e-01
-1.03008382e-01 -5.12803495e-01 -1.57817543e-01 2.24502906e-01
-5.11726856e-01 3.47286165e-01 -1.08377770e-01 2.80173808e-01
-2.17536971e-01 -5.74140012e-01 8.46468657e-02 5.22723012e-02
-9.39539969e-01 2.81152546e-01 -4.07173872e-01 5.19673228e-02
6.84684634e-01 -2.24468678e-01 -1.44388691e-01 -6.61505833e-02
-6.38267040e-01 4.18967694e-01 4.99321610e-01 5.41701257e-01
3.84138674e-01 -1.44581628e+00 -8.32031965e-01 1.39351115e-01
6.36156380e-01 -4.48672593e-01 -9.15279761e-02 7.32479095e-01
-8.11903551e-02 3.31519037e-01 -3.02800000e-01 -1.74031749e-01
-1.54613590e+00 4.84163284e-01 -9.55437571e-02 -5.79168260e-01
-3.90159458e-01 7.32227504e-01 -2.28600949e-01 -9.39499021e-01
-1.26138926e-01 -2.51231074e-01 -2.87916273e-01 -2.78500766e-01
4.47999030e-01 2.68267423e-01 3.90133739e-01 -4.88065183e-01
-3.55794340e-01 6.45064831e-01 -3.93316746e-02 -2.08778948e-01
1.16802967e+00 -4.16229427e-01 -4.73353773e-01 4.99002934e-02
9.54467118e-01 6.98057890e-01 -5.42767704e-01 -6.54951096e-01
6.71988666e-01 -4.86265779e-01 -3.22828293e-01 -6.25482857e-01
-6.81797802e-01 8.63988280e-01 1.01000272e-01 -4.92145240e-01
1.25475049e+00 -1.57072142e-01 1.19506264e+00 6.66781008e-01
2.72690177e-01 -1.17806208e+00 -8.62203836e-02 8.09668660e-01
3.63648713e-01 -8.35555375e-01 -3.93567234e-01 -7.34026730e-01
-4.78976399e-01 1.02169490e+00 6.12985671e-01 3.91543329e-01
6.84676087e-03 2.36086294e-01 -1.01446569e-01 2.78135002e-01
-2.69009233e-01 -4.22053725e-01 3.90409291e-01 4.84011322e-01
6.32283747e-01 1.68317668e-02 -1.02904487e+00 1.04838395e+00
-2.70201087e-01 -8.55855793e-02 3.21446031e-01 6.49060428e-01
-5.77846110e-01 -1.98066521e+00 8.27020779e-02 3.77899706e-01
-2.12268054e-01 -5.80339432e-01 -1.50218278e-01 1.92682430e-01
2.98942715e-01 8.70312750e-01 1.48746341e-01 -6.95520341e-01
4.34650213e-01 1.13344833e-01 4.56198245e-01 -8.31234634e-01
-6.19879067e-01 -1.59332395e-01 4.49544787e-01 -5.50764143e-01
-1.99693888e-01 -3.67583901e-01 -1.32076001e+00 -1.85326174e-01
-7.67979622e-01 7.55719841e-01 4.19299871e-01 1.20372963e+00
4.86289918e-01 5.78023076e-01 6.98516726e-01 -1.62115395e-01
-4.78006482e-01 -1.09790456e+00 -3.14903259e-01 5.05207956e-01
-9.75869745e-02 -2.24475607e-01 2.12717175e-01 -3.95217091e-02] | [11.187239646911621, 10.190994262695312] |
22fca94a-7733-414f-b6d3-610f4a501cc3 | experiments-on-paraphrase-identification | 2006.02648 | null | https://arxiv.org/abs/2006.02648v2 | https://arxiv.org/pdf/2006.02648v2.pdf | Experiments on Paraphrase Identification Using Quora Question Pairs Dataset | We modeled the Quora question pairs dataset to identify a similar question. The dataset that we use is provided by Quora. The task is a binary classification. We tried several methods and algorithms and different approach from previous works. For feature extraction, we used Bag of Words including Count Vectorizer, and Term Frequency-Inverse Document Frequency with unigram for XGBoost and CatBoost. Furthermore, we also experimented with WordPiece tokenizer which improves the model performance significantly. We achieved up to 97 percent accuracy. Code and Dataset. | ['Ruben Stefanus', 'Andreas Chandra'] | 2020-06-04 | null | null | null | null | ['paraphrase-identification'] | ['natural-language-processing'] | [-5.78313172e-01 -3.00499737e-01 -4.37570393e-01 -4.00462747e-01
-1.23133206e+00 -6.23818994e-01 4.99285102e-01 3.50473017e-01
-4.42917526e-01 7.88896859e-01 5.76396883e-01 -2.53839761e-01
-1.94148362e-01 -1.06310678e+00 -4.17938292e-01 -2.58081168e-01
2.54521072e-01 1.76936805e-01 2.62946218e-01 -4.14975315e-01
1.02915299e+00 -4.66760360e-02 -1.45514917e+00 6.07064426e-01
8.91383529e-01 1.04346669e+00 8.66012052e-02 8.18227887e-01
-4.53870982e-01 1.27734435e+00 -8.96802008e-01 -7.05192924e-01
2.18194813e-01 -8.61553371e-01 -1.22375751e+00 -5.26406586e-01
4.91494209e-01 1.19020358e-01 -2.02904269e-01 8.92884135e-01
3.78883779e-01 3.00833195e-01 8.42065692e-01 -1.16943932e+00
-1.16404533e+00 6.82317555e-01 -4.32488620e-01 5.50512671e-01
8.36554945e-01 -2.45067358e-01 1.57401705e+00 -7.90361166e-01
5.14640033e-01 1.15995038e+00 6.31250262e-01 3.96275192e-01
-5.11543453e-01 -7.67900705e-01 -8.05386782e-01 8.87253523e-01
-1.01662028e+00 8.41950923e-02 7.52278686e-01 -5.97766817e-01
1.05088794e+00 5.35349369e-01 5.07426202e-01 6.85468197e-01
6.34829283e-01 7.75603354e-01 1.54972780e+00 -9.04666781e-01
2.53626108e-02 3.49222988e-01 9.69026208e-01 7.50496328e-01
-3.68856013e-01 -3.04872900e-01 -5.84022582e-01 -6.24908328e-01
-1.18453234e-01 3.06568649e-02 -1.01925209e-01 4.14738953e-01
-7.81374693e-01 1.51819706e+00 5.28087139e-01 7.42759526e-01
-8.69363993e-02 5.49446285e-01 3.43769640e-01 5.16346335e-01
1.36461511e-01 9.71000969e-01 -5.28434694e-01 -4.22798365e-01
-5.75628102e-01 6.65252805e-01 9.46527362e-01 8.66952121e-01
7.66049027e-01 -5.75968325e-01 -4.52645808e-01 1.22244072e+00
3.43432069e-01 2.46083274e-01 1.26505482e+00 -7.17183590e-01
3.13997984e-01 7.86866605e-01 -1.98992506e-01 -1.19686568e+00
-2.86026567e-01 -3.52198705e-02 -4.63203602e-02 -2.41059914e-01
2.35826388e-01 -1.28591835e-01 -7.55975604e-01 1.06694067e+00
8.47381800e-02 -2.13339999e-01 -6.29297048e-02 5.33746481e-01
1.22418857e+00 7.47196317e-01 -1.16745695e-01 1.76864248e-02
1.95512390e+00 -1.27400315e+00 -1.05849481e+00 1.29690126e-01
8.56975496e-01 -1.19345701e+00 1.49983418e+00 3.05443704e-01
-6.42877460e-01 -4.99518514e-01 -1.03205371e+00 -2.56748676e-01
-6.60418153e-01 -1.67606339e-01 7.09308147e-01 1.16084385e+00
-3.50518793e-01 7.93631315e-01 -3.37524235e-01 -3.64655644e-01
3.78469855e-01 -1.76431030e-01 -2.21002534e-01 6.16130233e-02
-1.44194651e+00 1.08585596e+00 5.41723311e-01 -9.61071908e-01
-4.94271964e-01 -5.61795175e-01 -8.49949479e-01 1.54434917e-02
-4.19382676e-02 -1.29071742e-01 1.28036737e+00 -6.07866943e-01
-1.46764803e+00 7.16249526e-01 -7.43542379e-03 -4.10953164e-01
4.80283312e-02 -1.77615032e-01 -5.76954901e-01 9.56213027e-02
1.41245142e-01 1.68221235e-01 4.58009571e-01 -5.47575474e-01
-5.23944378e-01 -2.60242045e-01 1.38814270e-01 9.75605566e-03
-7.10166395e-01 6.25755608e-01 5.40599376e-02 -6.04142189e-01
1.92068085e-01 -7.06842065e-01 2.92315125e-01 -4.17528868e-01
1.82956047e-02 -6.63095832e-01 6.41879082e-01 -9.94023144e-01
1.46651065e+00 -1.99052322e+00 -5.85990310e-01 -1.40272915e-01
2.69305222e-02 -2.20750391e-01 1.01485655e-01 7.13625908e-01
1.45674184e-01 3.06674689e-01 -1.94066405e-01 1.04986213e-01
-1.26479298e-01 1.47513717e-01 -9.85752642e-02 3.56819093e-01
-1.71129093e-01 9.22895670e-01 -7.68625855e-01 -9.01541293e-01
-3.35510045e-01 -2.13969633e-01 -6.13898933e-01 4.88555133e-01
-2.83411175e-01 -3.91344786e-01 -3.19967180e-01 7.49861896e-01
3.48182619e-01 -4.76592146e-02 -2.40201145e-01 6.80421740e-02
2.47549191e-01 6.51147306e-01 -7.90287435e-01 1.52340102e+00
-7.42393076e-01 8.13385546e-01 -9.47017848e-01 -7.82589853e-01
1.19811046e+00 1.89418033e-01 2.11923048e-01 -8.60460877e-01
4.18360770e-01 3.37105393e-01 4.00981717e-02 -1.24497426e+00
6.78899884e-01 -3.61128926e-01 -2.45529205e-01 4.63350326e-01
3.97580713e-01 -4.94129956e-01 2.54014462e-01 5.40176705e-02
1.03798604e+00 -4.91205305e-02 7.80557752e-01 -5.74370205e-01
3.47413868e-01 4.63232875e-01 1.98220029e-01 7.02023983e-01
-2.04196781e-01 7.95523822e-01 6.99654043e-01 -1.63683772e-01
-6.40451491e-01 -8.97038698e-01 -4.93647218e-01 1.45256150e+00
1.08753189e-01 -6.95191562e-01 -3.16768080e-01 -1.21401453e+00
1.49583086e-01 1.09371746e+00 -1.04304326e+00 -6.96826633e-03
-5.52433491e-01 -9.48046327e-01 5.77256858e-01 3.89390796e-01
5.42512894e-01 -9.67571557e-01 -4.92955565e-01 2.93778598e-01
-1.86875165e-02 -5.11888206e-01 -4.73768890e-01 1.36207566e-01
-5.54238617e-01 -1.45725656e+00 -3.60175520e-01 -8.46168876e-01
-2.00853348e-01 -1.34333417e-01 1.27338850e+00 2.55179375e-01
-2.49231651e-01 -9.54132453e-02 -9.24654484e-01 -5.69652498e-01
-4.83858496e-01 8.09426699e-03 -5.23374498e-01 -4.90656555e-01
6.27723157e-01 -2.68100262e-01 -6.42766058e-01 1.81528121e-01
-6.91441000e-01 -4.71607983e-01 2.83634849e-02 1.06496692e+00
-1.81095749e-01 -4.86433029e-01 8.26423109e-01 -9.89793241e-01
1.13587153e+00 -9.29514527e-01 -2.79818892e-01 3.90716285e-01
-7.91076303e-01 -6.43280298e-02 5.03428996e-01 -3.04539531e-01
-6.65215790e-01 -5.19159436e-01 -4.80692476e-01 3.07576478e-01
1.51402384e-01 6.16162360e-01 1.86602235e-01 -1.82000905e-01
1.24302399e+00 6.30768389e-02 -1.31309003e-01 -6.02040112e-01
4.56368268e-01 1.45831192e+00 5.70994988e-02 -3.59217405e-01
1.41851917e-01 -4.88929525e-02 -4.53028649e-01 -4.88398820e-01
-7.72843242e-01 -7.96758890e-01 -1.10987186e-01 -1.05294086e-01
1.06164205e+00 -4.83376533e-01 -7.60798216e-01 2.01355264e-01
-1.08159792e+00 3.58461976e-01 7.42609352e-02 6.34339333e-01
-2.49763489e-01 2.85337240e-01 -6.47580206e-01 -7.47981131e-01
-4.93240625e-01 -7.46309876e-01 1.92039460e-01 4.77576315e-01
-2.75686443e-01 -8.71821404e-01 6.52842104e-01 8.26547444e-01
5.33511758e-01 2.00885907e-01 9.53058004e-01 -1.42110705e+00
1.86039001e-01 -3.35715413e-01 -9.20552686e-02 5.05338550e-01
1.62147701e-01 1.19126096e-01 -8.38795602e-01 1.91010565e-01
3.50025415e-01 -7.41595447e-01 8.36516023e-01 -2.04702199e-01
1.25896049e+00 -4.86471206e-01 -7.38184825e-02 -1.11292573e-02
1.54088819e+00 2.76112169e-01 4.55406815e-01 5.08205533e-01
4.59064454e-01 4.73944277e-01 7.20588267e-01 2.31075391e-01
6.65935874e-01 7.23210156e-01 1.28193736e-01 6.45785809e-01
-7.04593211e-02 -4.27301116e-02 -1.08777313e-02 1.17245340e+00
3.64249349e-01 -8.55847374e-02 -1.44411743e+00 8.22636962e-01
-1.72702444e+00 -8.89874399e-01 -5.34302354e-01 1.57118738e+00
1.04984570e+00 2.44888617e-03 1.95350885e-01 2.51838326e-01
3.98684829e-01 3.81684564e-02 2.69398659e-01 -7.81225562e-01
-2.81217191e-02 4.23975289e-01 2.62517065e-01 4.37521875e-01
-7.99216390e-01 6.39633060e-01 6.78387213e+00 1.09976315e+00
-8.75649452e-01 6.69589102e-01 5.04962325e-01 1.87955022e-01
-2.81801134e-01 5.33522330e-02 -5.07551789e-01 7.00246692e-01
8.54287148e-01 -4.00309891e-01 1.75055694e-02 1.08015692e+00
-6.65329874e-01 -5.34964740e-01 -6.56449318e-01 1.00726306e+00
7.48433828e-01 -1.41351628e+00 -3.66523743e-01 -5.13380647e-01
8.68443847e-01 1.85509905e-01 -4.51940298e-01 6.54963315e-01
4.22416836e-01 -1.19902790e+00 5.79625189e-01 2.88138062e-01
6.67087594e-03 -6.29088640e-01 9.72353041e-01 4.25166577e-01
-5.80212235e-01 -2.79080212e-01 -5.28075874e-01 -4.52495724e-01
-3.00733387e-01 3.67906332e-01 -9.24374938e-01 5.91496408e-01
8.96781504e-01 3.32542479e-01 -9.23633277e-01 1.41831625e+00
-7.72038996e-02 1.03472638e+00 -1.47845358e-01 -9.48822856e-01
2.34101295e-01 1.53798744e-01 -6.07111827e-02 1.15041244e+00
1.44607514e-01 -1.29645020e-01 -1.40106052e-01 4.46142107e-01
-2.04417631e-01 7.72956789e-01 -7.97781274e-02 3.44405115e-01
6.82431877e-01 1.17857778e+00 -5.70631981e-01 -4.73437071e-01
-4.76585865e-01 5.37243903e-01 3.72018009e-01 -2.90809453e-01
-9.01481211e-01 -8.72999847e-01 1.02236994e-01 -6.71780407e-02
-2.62631141e-02 1.56486541e-01 -4.60192174e-01 -9.64765191e-01
-2.64072716e-01 -8.83526444e-01 7.49591649e-01 -6.96789384e-01
-1.67533600e+00 5.13770640e-01 1.34402707e-01 -1.01138425e+00
-8.62288028e-02 -5.05289435e-01 -7.81914353e-01 7.94627428e-01
-1.25603986e+00 -9.13101912e-01 -3.32409799e-01 4.12793964e-01
4.24747974e-01 -3.09938878e-01 8.44512105e-01 3.09652418e-01
-1.33381888e-01 6.61343634e-01 4.03964728e-01 3.82727742e-01
8.88175547e-01 -1.41584301e+00 1.25901774e-01 3.11612695e-01
4.20611650e-01 5.61302125e-01 8.17447841e-01 -5.08260250e-01
-1.08847845e+00 -6.66084230e-01 1.09017575e+00 -7.32431352e-01
9.04645145e-01 -3.39317799e-01 -9.23418701e-01 3.93861264e-01
6.39685214e-01 -3.51261906e-02 1.18697369e+00 3.18610016e-03
-4.60306227e-01 -5.53135872e-02 -1.44239783e+00 1.20210558e-01
2.84931868e-01 -5.36696494e-01 -1.23504555e+00 8.50210726e-01
5.72676837e-01 -2.99995631e-01 -8.91307473e-01 1.87173635e-01
5.55237353e-01 -7.47249305e-01 4.74643648e-01 -9.59456146e-01
1.03672123e+00 6.13638461e-02 -5.43303490e-01 -1.14864254e+00
-1.33662552e-01 -1.84006035e-01 -1.27044663e-01 1.36530650e+00
5.86715162e-01 -5.58896303e-01 6.13616109e-01 3.30877066e-01
1.62875891e-01 -9.94948149e-01 -1.01419330e+00 -9.66755509e-01
6.54236138e-01 -3.09667200e-01 7.13038802e-01 1.27074087e+00
5.64583540e-01 5.75555205e-01 -1.91986203e-01 -6.25191033e-01
6.41371831e-02 4.56755787e-01 5.80364585e-01 -6.25238895e-01
-5.96405804e-01 -3.82105738e-01 -4.96102154e-01 -8.15923095e-01
3.69652500e-03 -1.18963313e+00 3.91974021e-03 -1.40251350e+00
4.04422581e-01 -5.69912046e-02 -2.24868014e-01 4.04093385e-01
-7.74268568e-01 5.56772053e-01 1.24074958e-01 6.82059228e-02
-2.20531389e-01 5.29141784e-01 9.50486183e-01 -2.01095231e-02
1.71556503e-01 -1.38848662e-01 -5.74134111e-01 5.31794250e-01
1.18110883e+00 -1.12269187e+00 2.58740429e-02 -2.09535673e-01
3.34745258e-01 2.31307879e-01 8.99934024e-02 -7.91211188e-01
2.20024034e-01 -3.98688912e-01 1.28587961e-01 -4.75105107e-01
1.76543221e-01 -4.94358420e-01 -4.01793867e-01 6.37872875e-01
-4.60037589e-01 3.97566617e-01 -1.70011267e-01 8.63635167e-02
-3.25753450e-01 -1.06846118e+00 8.60473335e-01 -2.47141302e-01
-4.38054800e-01 -3.81829858e-01 -4.00370747e-01 3.53410810e-01
1.18360066e+00 4.07641158e-02 -5.46151221e-01 -2.54070520e-01
-8.16202909e-02 2.43460014e-01 1.65170103e-01 7.27395773e-01
3.66433024e-01 -1.47766197e+00 -9.07374859e-01 -2.68481314e-01
6.22322917e-01 -1.00032413e+00 -1.63789466e-01 4.98134136e-01
-1.06269205e+00 8.22386742e-02 -2.91862220e-01 -1.99442267e-01
-1.37023818e+00 2.72023350e-01 3.79471242e-01 -5.51553816e-02
-1.07267730e-01 1.06493568e+00 -8.23211193e-01 -7.19350517e-01
-2.18478709e-01 -1.32233528e-02 -8.20871830e-01 4.66375351e-01
5.56036830e-01 7.27057219e-01 1.62917539e-01 -3.73982310e-01
-5.86801410e-01 4.93203700e-01 -1.01373397e-01 -2.16775864e-01
1.28142619e+00 2.77927488e-01 -5.52635550e-01 7.42505312e-01
1.70250511e+00 2.54023165e-01 -9.50141177e-02 2.38689291e-03
4.53306854e-01 -9.28468585e-01 -1.82055146e-01 -8.44485939e-01
-6.78133190e-01 5.03384769e-01 5.56125522e-01 5.54352701e-01
6.03632152e-01 1.00899205e-01 6.83344364e-01 4.92217153e-01
6.86026588e-02 -1.12645483e+00 3.42729539e-01 7.50799716e-01
9.84876037e-01 -1.47974801e+00 3.29061955e-01 -5.96501231e-02
-3.71921808e-01 1.27390170e+00 8.17166388e-01 -5.88922024e-01
1.10679400e+00 -2.73653176e-02 4.25504893e-01 -3.08287233e-01
-6.97420835e-01 1.49655370e-02 4.31558847e-01 1.71948791e-01
8.61739099e-01 -1.06769100e-01 -1.39380646e+00 9.08711076e-01
-1.00608242e+00 -2.89145172e-01 8.37653339e-01 9.78374958e-01
-6.20201409e-01 -9.55007911e-01 -4.20294613e-01 9.66541648e-01
-1.01326609e+00 -3.91995102e-01 -3.74540418e-01 7.52360046e-01
9.21184719e-02 1.30557573e+00 -3.80070098e-02 -5.94646633e-01
1.97318062e-01 2.08874702e-01 3.11628252e-01 -7.25468814e-01
-1.47031915e+00 -7.25117505e-01 2.68761456e-01 -2.99835980e-01
-4.10238877e-02 -4.21229601e-01 -1.16016984e+00 -1.58500046e-01
-9.07586813e-01 6.31296337e-01 1.04636621e+00 9.54280317e-01
-2.14333788e-01 3.62256914e-01 9.60008681e-01 2.48912483e-01
-6.90850854e-01 -1.60949898e+00 -4.47197169e-01 7.12207615e-01
6.55557290e-02 -3.88473570e-01 -4.27297741e-01 -1.35013446e-01] | [11.37753677368164, 8.509770393371582] |
8de5c9d5-ace1-4215-98e0-acab0c3a7080 | trading-off-price-for-data-quality-to-achieve | 2306.13440 | null | https://arxiv.org/abs/2306.13440v1 | https://arxiv.org/pdf/2306.13440v1.pdf | Trading-off price for data quality to achieve fair online allocation | We consider the problem of online allocation subject to a long-term fairness penalty. Contrary to existing works, however, we do not assume that the decision-maker observes the protected attributes -- which is often unrealistic in practice. Instead they can purchase data that help estimate them from sources of different quality; and hence reduce the fairness penalty at some cost. We model this problem as a multi-armed bandit problem where each arm corresponds to the choice of a data source, coupled with the online allocation problem. We propose an algorithm that jointly solves both problems and show that it has a regret bounded by $\mathcal{O}(\sqrt{T})$. A key difficulty is that the rewards received by selecting a source are correlated by the fairness penalty, which leads to a need for randomization (despite a stochastic setting). Our algorithm takes into account contextual information available before the source selection, and can adapt to many different fairness notions. We also show that in some instances, the estimates used can be learned on the fly. | ['Vianney Perchet', 'Patrick Loiseau', 'Nicolas Gast', 'Mathieu Molina'] | 2023-06-23 | null | null | null | null | ['fairness', 'fairness'] | ['computer-vision', 'miscellaneous'] | [ 4.02589003e-03 1.70474514e-01 -6.57139957e-01 -4.19003397e-01
-1.06816316e+00 -8.53571534e-01 1.39703164e-02 4.56256002e-01
-7.97694504e-01 1.04860139e+00 9.43026096e-02 -3.85172129e-01
-4.98957306e-01 -8.66629064e-01 -7.03843951e-01 -6.68169081e-01
2.19341703e-02 7.18198061e-01 -2.33541518e-01 7.57000549e-03
3.01327080e-01 3.35328728e-01 -1.16683865e+00 -1.02663048e-01
8.45537007e-01 1.29451346e+00 -1.13117941e-01 5.69170892e-01
-1.43576324e-01 6.31423652e-01 -5.10876894e-01 -9.32542622e-01
7.72649050e-01 -6.64987981e-01 -9.50507700e-01 2.11328015e-01
7.44166821e-02 -5.59890985e-01 1.95208676e-02 1.12422979e+00
5.12306154e-01 2.73134530e-01 5.40018320e-01 -1.37762237e+00
-2.95658827e-01 8.73235643e-01 -8.22577536e-01 2.36465037e-02
4.92014410e-03 -1.79681823e-01 1.35803211e+00 -9.52652916e-02
4.56086427e-01 9.68637168e-01 3.63406867e-01 4.63454902e-01
-1.57942355e+00 -5.43377399e-01 3.59800041e-01 -5.68761751e-02
-9.39143002e-01 -6.96778655e-01 5.41533828e-01 -4.14324433e-01
6.60278723e-02 5.67509472e-01 4.64173317e-01 5.56424320e-01
-4.45549667e-01 8.90521348e-01 1.35462725e+00 -5.96186578e-01
6.34503722e-01 3.22895586e-01 1.92070365e-01 2.36340314e-01
4.65608686e-01 1.20515361e-01 -4.95470673e-01 -4.93946999e-01
5.09556174e-01 1.73625290e-01 -3.51749718e-01 -7.07566619e-01
-7.67603278e-01 1.03179777e+00 2.55001724e-01 -3.70308608e-02
-4.84394521e-01 3.02858591e-01 2.76500314e-01 7.75618911e-01
5.73130369e-01 2.34053046e-01 -5.74047029e-01 -2.80939527e-02
-1.04750240e+00 3.60105544e-01 1.00504482e+00 9.16840315e-01
8.00496280e-01 -3.39288175e-01 -3.80435884e-01 6.43508673e-01
8.04268718e-02 6.33960009e-01 -8.51142332e-02 -1.21006835e+00
8.78990889e-01 -8.95892307e-02 9.85571504e-01 -8.80331874e-01
-1.67261451e-01 -3.61318499e-01 -5.92015505e-01 4.22299564e-01
1.02774668e+00 -3.39941859e-01 -3.63200039e-01 1.95468593e+00
4.43963230e-01 -4.44640458e-01 -4.27809089e-01 1.04419088e+00
-3.54025394e-01 2.76864409e-01 -9.01671052e-02 -7.32074261e-01
1.00899267e+00 -7.66955614e-01 -7.91804075e-01 -6.92967027e-02
3.76197517e-01 -4.42661077e-01 7.21270084e-01 5.04141271e-01
-1.38393426e+00 1.59689844e-01 -6.78032041e-01 2.38632813e-01
-1.30492682e-02 -2.86199093e-01 6.45374000e-01 1.14738083e+00
-7.94245541e-01 7.85319805e-01 -3.89666080e-01 -2.76466031e-02
5.91128469e-01 4.63821709e-01 -7.76877999e-02 -4.26841080e-02
-7.83703268e-01 6.54749453e-01 4.23823521e-02 -7.35725984e-02
-6.69101655e-01 -5.06393909e-01 -2.91436970e-01 3.14075410e-01
9.60852385e-01 -6.24238789e-01 1.33115697e+00 -1.37562144e+00
-1.57613301e+00 6.87952816e-01 -2.09324375e-01 -4.84815329e-01
1.23287868e+00 -7.52656683e-02 2.49216706e-01 -8.13386217e-02
3.25779378e-01 -3.66545580e-02 7.81170785e-01 -1.10572147e+00
-7.98552334e-01 -5.24198711e-01 5.67092538e-01 1.96996972e-01
-2.84643352e-01 2.15648204e-01 -6.35411292e-02 -5.20606220e-01
-1.54579386e-01 -8.62658143e-01 -5.63433588e-01 2.50488281e-01
-4.38214123e-01 1.04263276e-01 -1.99087426e-01 -3.02962631e-01
1.06425261e+00 -2.01957035e+00 8.78601521e-02 5.09074211e-01
5.69946244e-02 -2.08013326e-01 3.85181569e-02 3.64518791e-01
1.70790866e-01 4.52498645e-01 -2.07203284e-01 -4.35871929e-01
3.74506772e-01 6.11512223e-03 -2.75027573e-01 8.62287343e-01
-5.97536445e-01 5.34091651e-01 -7.42073476e-01 -2.11037025e-01
-1.19395070e-01 -3.71630192e-01 -7.59971321e-01 1.41768888e-01
-2.46373728e-01 1.82115018e-01 -6.34103775e-01 2.35100552e-01
8.50430727e-01 -2.65564054e-01 3.62340063e-01 4.18153256e-01
-9.28366557e-03 1.56443179e-01 -1.76507866e+00 1.41957331e+00
-5.94402909e-01 5.66906594e-02 5.23672462e-01 -1.19233119e+00
2.37675741e-01 1.94773927e-01 4.88095582e-01 -4.90130037e-01
1.84692189e-01 3.06358814e-01 -2.82647431e-01 -1.67841583e-01
2.48328686e-01 -4.77855027e-01 -2.70515651e-01 9.52823639e-01
-2.13156626e-01 3.33467394e-01 -5.06647341e-02 6.34911051e-03
7.79217601e-01 -3.18344563e-01 4.13829744e-01 -3.26037884e-01
2.55093053e-02 -2.25518867e-01 7.73171365e-01 1.42745030e+00
-1.80012733e-01 3.08361530e-01 8.51052284e-01 -5.09287536e-01
-9.97856081e-01 -8.57968152e-01 1.36772722e-01 1.39426351e+00
1.65560320e-01 1.60768867e-01 -6.26942813e-01 -8.26517284e-01
4.27218139e-01 6.53680027e-01 -7.32236743e-01 3.68582547e-01
5.24307378e-02 -7.54708052e-01 -4.50865142e-02 4.47593518e-02
1.71015307e-01 -5.20077348e-01 -7.07080007e-01 3.23827237e-01
-2.55978584e-01 -6.10103071e-01 -6.70484126e-01 2.12762535e-01
-6.86689913e-01 -9.06428337e-01 -8.70960712e-01 1.43130526e-01
6.01446569e-01 2.72127956e-01 1.09808278e+00 1.62972473e-02
-3.25012505e-02 3.51590246e-01 -2.85157293e-01 -6.46973908e-01
-3.16535346e-02 6.92736506e-02 -8.43010321e-02 3.47693950e-01
5.07758260e-02 -4.64776576e-01 -7.49069154e-01 2.35840589e-01
-6.70804441e-01 -3.04127336e-01 2.19862461e-01 8.16198528e-01
4.31695431e-01 6.53191134e-02 8.00091624e-01 -1.32890308e+00
4.60893810e-01 -5.49785972e-01 -1.17408609e+00 3.90228093e-01
-6.46739662e-01 2.18318403e-01 7.71811843e-01 -1.56375036e-01
-7.75650322e-01 -4.84850071e-02 4.18196380e-01 -2.16520876e-01
5.00758737e-02 1.90318704e-01 -3.86650026e-01 -8.90476070e-03
4.34733003e-01 -9.13902596e-02 1.62237361e-02 -5.76283216e-01
5.41454613e-01 6.10441387e-01 -1.48853838e-01 -8.35050046e-01
6.24124229e-01 4.22780663e-01 1.28380135e-01 -1.63786426e-01
-1.18566775e+00 -1.30595565e-01 -1.33521155e-01 3.02233975e-02
3.97528768e-01 -7.38625348e-01 -1.20889664e+00 1.93072353e-02
-9.60532904e-01 -3.56791794e-01 -6.65455818e-01 2.64818579e-01
-7.78131425e-01 1.49717569e-01 -3.16828281e-01 -1.55584776e+00
-1.43183442e-02 -9.09542739e-01 4.08545613e-01 4.81575690e-02
1.46261275e-01 -6.90664113e-01 -2.54908413e-01 5.13034403e-01
5.38229108e-01 9.67965499e-02 6.40521884e-01 -5.93337417e-01
-7.50495195e-01 -2.78529972e-01 -2.00188205e-01 3.94985974e-02
-6.13557734e-02 -5.00141144e-01 -8.39877903e-01 -3.49775046e-01
1.29776999e-01 -3.68713349e-01 6.43265784e-01 6.13818228e-01
1.48032928e+00 -1.06424403e+00 -4.84422361e-03 3.52616817e-01
1.61245239e+00 -1.00047216e-02 1.82562053e-01 3.24502587e-01
2.17820331e-01 6.65795863e-01 5.74467957e-01 9.78896499e-01
2.63189882e-01 8.30847144e-01 4.45675820e-01 1.23789638e-01
5.70221066e-01 3.54693783e-03 -8.19108915e-03 -8.58046189e-02
-6.36878014e-02 -3.45120013e-01 -5.29130161e-01 6.86788857e-01
-2.09409642e+00 -1.02403927e+00 1.64178744e-01 2.85205841e+00
9.76561487e-01 -6.43251911e-02 7.31549859e-01 1.88490182e-01
7.27216005e-01 4.99109849e-02 -6.60582840e-01 -5.85958004e-01
7.94654265e-02 1.34135440e-01 1.16911626e+00 7.31187284e-01
-8.62405360e-01 3.46261472e-01 6.55802631e+00 9.61358726e-01
-7.01995850e-01 2.96879292e-01 1.22912335e+00 -6.92285895e-01
-5.28631270e-01 1.49103343e-01 -3.27458590e-01 7.43607342e-01
9.05178607e-01 -5.04676104e-01 1.04528141e+00 8.82478178e-01
2.16718018e-01 -3.91593575e-01 -1.24318635e+00 7.49900341e-01
-3.59900415e-01 -1.08659840e+00 -2.96300948e-01 3.63195181e-01
7.32193768e-01 -3.89587969e-01 6.11458533e-02 -1.38432935e-01
8.41680229e-01 -9.11774397e-01 1.00893939e+00 4.36454624e-01
7.03350067e-01 -1.18111622e+00 5.85014164e-01 6.26326442e-01
-6.93674684e-01 -4.58343804e-01 -4.08182591e-01 -1.78897977e-01
1.31015182e-01 1.05661547e+00 -1.04042165e-01 7.50412643e-01
4.72089648e-01 4.28602025e-02 1.96188599e-01 1.23476636e+00
-2.74657216e-02 4.57911909e-01 -4.24030215e-01 -1.12006150e-01
-5.63493483e-02 -3.76086265e-01 2.94992119e-01 7.86415219e-01
3.60597521e-01 2.97852516e-01 3.03357393e-01 7.83792555e-01
-3.78242940e-01 3.04628462e-01 -3.27413470e-01 1.92976862e-01
5.78989804e-01 9.71082330e-01 -5.07906020e-01 -2.03370616e-01
-3.75120044e-01 9.89649177e-01 5.38460851e-01 2.95312196e-01
-6.54473245e-01 -2.84926862e-01 6.28500104e-01 8.18140656e-02
3.61913443e-01 1.73074692e-01 -5.67146122e-01 -1.11765146e+00
2.35260695e-01 -7.35568106e-01 6.18336618e-01 -2.44098216e-01
-1.57300329e+00 -1.26472581e-02 -1.43579334e-01 -9.04243529e-01
-1.13040544e-02 -2.28956446e-01 -1.57445595e-01 1.16224456e+00
-1.70290935e+00 -5.54857850e-01 2.89523482e-01 4.77341861e-01
7.59290457e-02 1.97090551e-01 7.80383945e-01 2.68128395e-01
-4.26111877e-01 7.06165314e-01 6.33206785e-01 -1.22721285e-01
5.40725231e-01 -1.37131417e+00 -2.37371981e-01 6.79257929e-01
-3.97506356e-02 3.33601892e-01 8.23493242e-01 -1.07636631e-01
-1.23747969e+00 -8.23334157e-01 9.56368387e-01 -2.26880386e-01
6.67301297e-01 -3.71111840e-01 -2.78942734e-01 5.34551024e-01
1.11124136e-01 3.53550404e-01 9.29234684e-01 4.24940497e-01
-3.27387452e-01 -5.07913172e-01 -1.53362513e+00 3.99178743e-01
1.16261506e+00 -3.54866922e-01 1.90126728e-02 3.76949579e-01
4.21127349e-01 -1.66987583e-01 -6.98111296e-01 -3.06698143e-01
6.43862605e-01 -1.14434469e+00 6.21317089e-01 -9.10335422e-01
2.52443135e-01 1.15121894e-01 -1.13819890e-01 -1.27955270e+00
-2.44331121e-01 -9.96872365e-01 6.40336797e-02 1.19054508e+00
5.06066322e-01 -7.73502588e-01 8.21460426e-01 1.44715190e+00
6.98324323e-01 -4.37094986e-01 -1.28673160e+00 -8.61227810e-01
1.77248076e-01 -2.13127628e-01 8.00731778e-01 8.98934484e-01
8.81289616e-02 -4.75516766e-02 -8.26430500e-01 -7.35712284e-03
9.97272789e-01 6.12477601e-01 7.30934441e-01 -1.17669678e+00
-7.58092344e-01 -4.90418196e-01 1.67274907e-01 -9.83814955e-01
6.28394559e-02 -6.94357872e-01 2.55601481e-02 -1.17484748e+00
3.87431711e-01 -1.05698454e+00 -5.87270379e-01 5.25380790e-01
-2.49968201e-01 -2.02070668e-01 5.64675272e-01 1.00535177e-01
-8.75161409e-01 3.19370359e-01 8.18846703e-01 8.15809742e-02
-8.92785043e-02 5.28754234e-01 -1.29586494e+00 3.47107947e-01
7.18567371e-01 -8.46468508e-01 -3.50402266e-01 -5.60370266e-01
4.85174745e-01 8.17750037e-01 2.58171260e-01 -4.16805327e-01
1.12585478e-01 -8.05327654e-01 9.41887647e-02 -7.34817907e-02
-2.85117999e-02 -1.21779776e+00 2.70286500e-01 3.29828113e-01
-8.78404438e-01 -2.39418864e-01 -3.48491222e-01 1.00279212e+00
3.86502028e-01 -5.41859865e-01 7.28669226e-01 -3.82446229e-01
3.17808151e-01 5.21610081e-01 -2.80042022e-01 2.25865066e-01
1.03762579e+00 2.01135188e-01 -6.22690395e-02 -7.39459872e-01
-7.58018911e-01 3.09696108e-01 6.02217317e-01 -2.45833933e-01
-1.02474391e-01 -1.22869933e+00 -6.59013808e-01 -2.60254562e-01
-1.34149313e-01 -2.09842846e-01 2.38873258e-01 6.70425296e-01
1.07759826e-01 4.20852304e-02 5.37887565e-04 -8.55383128e-02
-9.30259109e-01 7.95482278e-01 4.00375426e-01 -4.33928281e-01
-6.05126470e-02 7.52864301e-01 -6.46761358e-02 2.35574003e-02
3.23762655e-01 1.35598198e-01 2.22091690e-01 3.06973130e-01
5.33635557e-01 5.50754309e-01 -9.24288034e-02 -6.56347945e-02
-1.96562082e-01 8.93233940e-02 -6.01159036e-02 -5.53987324e-01
1.47683644e+00 -5.78456223e-01 -1.54274151e-01 4.22177970e-01
8.89046907e-01 3.33291620e-01 -1.27156329e+00 -5.01415908e-01
1.85116008e-01 -1.16403592e+00 -1.01521477e-01 -8.93978000e-01
-1.32847750e+00 6.04274869e-01 4.10574883e-01 6.57970250e-01
1.15644324e+00 -4.00950640e-01 4.29272592e-01 1.88919410e-01
9.04133141e-01 -1.35825551e+00 -4.19989288e-01 -4.30445634e-02
4.87957567e-01 -1.25923133e+00 9.74574313e-02 -2.23543718e-01
-5.73053122e-01 8.89082372e-01 -1.18896827e-01 -3.01072121e-01
5.51319838e-01 2.92760786e-03 -2.38965541e-01 3.42882842e-01
-6.21287286e-01 -2.04457268e-01 -5.67956343e-02 3.54590535e-01
1.31893039e-01 3.79937202e-01 -6.36917114e-01 8.23656917e-01
5.71767613e-02 1.38310477e-01 7.85048187e-01 9.21864271e-01
-2.99848229e-01 -1.48971856e+00 -3.02448422e-01 7.89978445e-01
-1.04073822e+00 1.64713431e-02 -2.01439276e-01 2.20995396e-01
3.96148721e-03 1.16940546e+00 -1.79368585e-01 2.75316864e-01
1.47296190e-01 -7.63421282e-02 3.80639464e-01 -3.97333443e-01
-6.24863446e-01 1.48949280e-01 2.24920571e-01 -5.91092050e-01
-5.74518144e-01 -7.18089104e-01 -4.50226367e-01 -7.06152320e-01
-4.35367972e-01 5.39080083e-01 4.75477040e-01 8.70401978e-01
3.32314938e-01 1.04964502e-01 1.19746089e+00 -5.63256264e-01
-1.04115069e+00 -3.54663610e-01 -1.03354824e+00 4.47141379e-01
6.16768718e-01 -4.66493577e-01 -4.94858980e-01 -2.89205611e-01] | [4.558324337005615, 3.3769919872283936] |
e945bdf2-8801-4cf5-a85e-70141b876ed7 | image-as-a-foreign-language-beit-pretraining-1 | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wang_Image_as_a_Foreign_Language_BEiT_Pretraining_for_Vision_and_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wang_Image_as_a_Foreign_Language_BEiT_Pretraining_for_Vision_and_CVPR_2023_paper.pdf | Image as a Foreign Language: BEiT Pretraining for Vision and Vision-Language Tasks | A big convergence of language, vision, and multimodal pretraining is emerging. In this work, we introduce a general-purpose multimodal foundation model BEiT-3, which achieves excellent transfer performance on both vision and vision-language tasks. Specifically, we advance the big convergence from three aspects: backbone architecture, pretraining task, and model scaling up. We use Multiway Transformers for general-purpose modeling, where the modular architecture enables both deep fusion and modality-specific encoding. Based on the shared backbone, we perform masked "language" modeling on images (Imglish), texts (English), and image-text pairs ("parallel sentences") in a unified manner. Experimental results show that BEiT-3 obtains remarkable performance on object detection (COCO), semantic segmentation (ADE20K), image classification (ImageNet), visual reasoning (NLVR2), visual question answering (VQAv2), image captioning (COCO), and cross-modal retrieval (Flickr30K, COCO). | ['Furu Wei', 'Subhojit Som', 'Saksham Singhal', 'Owais Khan Mohammed', 'Kriti Aggarwal', 'Qiang Liu', 'Zhiliang Peng', 'Johan Bjorck', 'Li Dong', 'Hangbo Bao', 'Wenhui Wang'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['visual-reasoning', 'cross-modal-retrieval', 'visual-reasoning'] | ['computer-vision', 'miscellaneous', 'reasoning'] | [-1.00523099e-01 -1.01490512e-01 -2.94486014e-03 -4.92204100e-01
-1.07548952e+00 -6.53170824e-01 7.51402676e-01 -8.34468976e-02
-6.39135659e-01 1.92372441e-01 -1.37753342e-03 -4.63072181e-01
4.61253375e-01 -3.84171724e-01 -9.82024014e-01 -3.71095449e-01
4.12746072e-01 5.42064130e-01 9.59696323e-02 -3.44659030e-01
2.07302533e-02 1.12760700e-01 -1.29568994e+00 8.30245435e-01
7.52043784e-01 1.19444704e+00 3.65238816e-01 9.10161853e-01
-3.17316890e-01 1.06368995e+00 -6.49737641e-02 -8.64755988e-01
-3.28662187e-01 -9.32779759e-02 -1.34429216e+00 2.41743475e-01
7.88228929e-01 -4.87589091e-01 -4.49757904e-01 1.01025343e+00
5.29482543e-01 -1.82534471e-01 6.74018145e-01 -1.47757435e+00
-1.26953065e+00 4.41214114e-01 -6.54696703e-01 -2.49432087e-01
3.10368717e-01 5.30015588e-01 1.00931525e+00 -1.29523945e+00
5.63011050e-01 1.69667053e+00 4.16947335e-01 7.69077003e-01
-1.01944864e+00 -3.73565942e-01 1.56054586e-01 4.11434054e-01
-1.30478311e+00 -4.56648111e-01 3.80872488e-01 -4.49133456e-01
9.26493764e-01 3.22961390e-01 4.68051136e-01 1.29586339e+00
-7.74777755e-02 1.52053726e+00 1.04775333e+00 -3.55542511e-01
-6.55869693e-02 3.23509902e-01 2.49371275e-01 8.61298621e-01
-2.97039598e-01 -3.52037460e-01 -5.26816607e-01 3.17593247e-01
7.17622817e-01 -1.27411112e-01 -3.50462198e-01 -1.62616134e-01
-1.43994653e+00 6.38936520e-01 6.55039489e-01 1.75444320e-01
-1.12509832e-01 3.55593443e-01 6.03553474e-01 3.85423303e-01
-8.12727213e-03 1.04740247e-01 -3.81118208e-01 3.09619904e-01
-6.68141484e-01 -1.25660479e-01 4.66524899e-01 1.21117711e+00
7.82816052e-01 -9.51527134e-02 -6.34269178e-01 1.08470452e+00
7.08688796e-01 1.10477531e+00 5.07182717e-01 -9.94327784e-01
5.72863758e-01 6.44132674e-01 -1.23757236e-01 -6.75519466e-01
-2.14756355e-01 4.07237932e-02 -1.06546772e+00 -3.72901231e-01
8.25614929e-02 -3.51368114e-02 -1.38745105e+00 1.72194505e+00
2.08387226e-02 -2.69617289e-01 4.24356431e-01 1.05993092e+00
1.65434504e+00 8.49327624e-01 5.55882215e-01 2.20694125e-01
1.84470963e+00 -1.48376465e+00 -5.11890471e-01 -4.17579085e-01
4.46212053e-01 -8.01513374e-01 1.43319082e+00 2.23228652e-02
-1.25512826e+00 -9.19273078e-01 -4.85367388e-01 -6.17431998e-01
-5.92538118e-01 3.09848845e-01 5.20561993e-01 2.09799483e-01
-1.58185208e+00 -4.86457109e-01 -2.98058182e-01 -7.99197078e-01
4.81353611e-01 3.25347632e-02 -5.71831644e-01 -4.97835428e-01
-1.16372955e+00 7.60995448e-01 4.00679201e-01 2.06506535e-01
-1.47779059e+00 -4.93381053e-01 -1.06723785e+00 -1.03526182e-01
2.42089108e-01 -1.22824013e+00 1.24638832e+00 -1.06294882e+00
-1.24018526e+00 1.63867104e+00 -1.50685430e-01 -3.81776303e-01
4.26974297e-01 -1.25504330e-01 -3.49175334e-01 7.12667346e-01
-1.01636229e-02 1.65098429e+00 9.57476556e-01 -1.46178997e+00
-4.23850775e-01 -3.83503854e-01 1.63264722e-01 4.17568833e-01
-3.66953820e-01 4.92061153e-02 -1.21097291e+00 -2.98419803e-01
-2.12309495e-01 -7.24243164e-01 -1.13369182e-01 -7.80957565e-02
-6.55595124e-01 -4.72815752e-01 6.28220022e-01 -8.67605865e-01
6.24540865e-01 -2.24086928e+00 2.82047689e-01 -2.11082190e-01
2.43221164e-01 2.58945316e-01 -7.70111799e-01 4.65761244e-01
1.56147495e-01 -2.42071506e-02 -1.90922394e-01 -5.17650723e-01
2.04908580e-01 3.13783139e-01 -4.45165813e-01 4.24375236e-02
3.31082910e-01 1.91168368e+00 -6.74990594e-01 -1.01774573e+00
2.90566951e-01 4.23726201e-01 -4.52525109e-01 4.21721697e-01
-6.08110309e-01 2.72133440e-01 -3.65824729e-01 1.03167450e+00
6.06546640e-01 -6.92800879e-01 -6.95400536e-02 -7.07279146e-01
1.20099336e-01 -5.09915411e-01 -4.85276610e-01 2.05697966e+00
-4.72278148e-01 7.01710463e-01 3.58400881e-01 -8.06489825e-01
5.61462283e-01 2.58531153e-01 8.23178291e-02 -1.10084820e+00
2.07206905e-01 -1.38319895e-01 -5.91932058e-01 -1.03272653e+00
5.61113775e-01 3.40448022e-01 -7.33145326e-02 1.25896528e-01
5.16752660e-01 -1.80748850e-01 2.18970746e-01 5.28567553e-01
5.49147844e-01 -4.90782335e-02 -2.20358074e-01 1.48139121e-02
7.11491644e-01 2.09048584e-01 -2.23035753e-01 8.87614310e-01
-1.83213770e-01 7.22065389e-01 3.73720765e-01 -4.50986773e-02
-8.70706320e-01 -1.22670639e+00 6.78040758e-02 1.46452332e+00
3.78004581e-01 -3.17063689e-01 -8.50550056e-01 -5.65197289e-01
-1.56320222e-02 5.78167617e-01 -6.37259901e-01 -1.33738920e-01
-1.62520185e-01 -6.07635498e-01 7.70838261e-01 6.09129071e-01
8.93348038e-01 -1.27794695e+00 -2.32439607e-01 -3.23970646e-01
-6.21712148e-01 -1.64553499e+00 -5.62431633e-01 -1.88320652e-01
-6.18906796e-01 -9.07181621e-01 -1.04496276e+00 -1.26515448e+00
5.84940970e-01 5.65591872e-01 1.36835814e+00 -8.95839259e-02
-3.75379145e-01 1.35494626e+00 -1.97630197e-01 -1.23352319e-01
-2.85611659e-01 -2.44226351e-01 -4.44231749e-01 1.79593369e-01
1.42134354e-01 -7.10740918e-03 -6.47892714e-01 1.90578282e-01
-1.19483459e+00 3.99712324e-01 9.42238629e-01 8.41698349e-01
6.72037363e-01 -6.54680014e-01 3.88650268e-01 -3.44487190e-01
5.58792591e-01 -3.39609355e-01 -3.51423830e-01 9.97783899e-01
-2.20523730e-01 -2.57012784e-01 6.42780364e-02 -4.92219090e-01
-1.07877576e+00 -9.76447165e-02 -2.26038724e-01 -8.06789100e-01
-4.45568591e-01 5.17527759e-01 -3.83085787e-01 -7.63586238e-02
2.85265446e-01 8.28167796e-01 7.44300708e-02 -4.34050769e-01
1.05545092e+00 7.36125112e-01 9.83537734e-01 -6.07035816e-01
4.59641248e-01 5.13266027e-01 -4.84251648e-01 -9.98730958e-01
-7.46195912e-01 -3.22524101e-01 -5.22935510e-01 -3.14183474e-01
1.57241797e+00 -1.42313552e+00 -1.10411799e+00 6.76484406e-01
-1.41715002e+00 -4.87832069e-01 4.31817211e-03 1.94789514e-01
-3.88232946e-01 5.06992579e-01 -8.32425535e-01 -5.84497273e-01
-7.35433519e-01 -1.21889496e+00 1.43611515e+00 3.43170553e-01
4.18426871e-01 -8.80299866e-01 -2.42684811e-01 1.14759576e+00
3.37169439e-01 -2.11178973e-01 1.00861180e+00 -4.81937736e-01
-7.66167819e-01 4.62078564e-02 -1.04423416e+00 6.40582561e-01
-6.20911479e-01 -1.20476253e-01 -1.06999910e+00 -2.57631570e-01
-3.45993400e-01 -1.15607142e+00 1.09985197e+00 2.97184050e-01
1.18461585e+00 7.94729777e-03 -2.06189170e-01 6.07206464e-01
1.34946585e+00 -1.68197244e-01 7.40519464e-01 3.70633323e-03
1.09871662e+00 5.57369947e-01 2.87631303e-01 -1.66164264e-01
1.18390107e+00 3.09527159e-01 6.46078467e-01 -5.50751925e-01
-3.73905391e-01 -1.99197769e-01 5.76143384e-01 1.07498622e+00
2.65259951e-01 -2.22279355e-01 -1.14924753e+00 7.56615877e-01
-1.94385207e+00 -6.77379310e-01 -2.12213904e-01 1.55850363e+00
7.35152364e-01 -3.62170398e-01 -1.51020721e-01 -8.57219458e-01
5.85823476e-01 -1.64255630e-02 -5.61538696e-01 -2.19673023e-01
-4.01336610e-01 -3.00606340e-01 1.44324183e-01 2.42833838e-01
-1.15865433e+00 1.39647579e+00 6.56165886e+00 9.70000803e-01
-1.13641059e+00 2.73189276e-01 8.86988461e-01 3.34569156e-01
-4.47403163e-01 -2.82784730e-01 -7.01266825e-01 1.61857575e-01
5.40437937e-01 4.03882623e-01 2.06680045e-01 6.24001443e-01
-3.66329461e-01 -2.45731305e-02 -1.03008509e+00 1.57435405e+00
4.79215235e-01 -1.42013526e+00 7.18277812e-01 -2.23508254e-01
5.72446525e-01 5.21431684e-01 3.29883218e-01 6.51202023e-01
1.64424688e-01 -1.18017697e+00 7.22288370e-01 7.04443216e-01
7.95618713e-01 -2.74036914e-01 5.88633239e-01 1.04635343e-01
-9.65903580e-01 6.15110919e-02 -1.82432532e-01 6.17325485e-01
1.96873486e-01 3.02032307e-02 -4.70272064e-01 7.72412717e-01
9.51591611e-01 6.83480024e-01 -1.00204706e+00 6.49080396e-01
-1.75410807e-01 4.55439895e-01 -6.24653585e-02 5.56455292e-02
5.26714206e-01 -1.06546171e-01 2.41816387e-01 1.53104568e+00
-2.58332998e-01 -1.47732675e-01 2.91465163e-01 8.67693067e-01
-4.43579644e-01 1.48762599e-01 -3.21048051e-01 -1.70908436e-01
1.15870371e-01 1.51491702e+00 -4.63911384e-01 -6.01719260e-01
-6.34108663e-01 1.24785948e+00 3.30368549e-01 8.16590488e-01
-9.42190170e-01 -1.05985709e-01 3.23102236e-01 -4.03952628e-01
3.97127926e-01 -1.06405407e-01 3.90851647e-02 -1.47122133e+00
-9.18857232e-02 -9.48765516e-01 5.69573879e-01 -1.47576904e+00
-1.51785171e+00 6.89901352e-01 -2.17264127e-02 -6.78234220e-01
1.46565050e-01 -8.11534166e-01 -2.78891891e-01 6.32321119e-01
-1.73408496e+00 -1.87700760e+00 -4.90695387e-01 1.11541677e+00
5.84800482e-01 -2.38278583e-01 5.51841199e-01 5.17652631e-01
-6.05866194e-01 6.77198350e-01 -2.78791606e-01 5.85032761e-01
7.64600098e-01 -1.06661677e+00 1.66778594e-01 4.80762839e-01
3.15586716e-01 3.70062023e-01 8.21503624e-02 -2.82575369e-01
-1.84984255e+00 -1.32513893e+00 6.90340281e-01 -6.05679691e-01
8.21661115e-01 -3.52409482e-01 -7.26899981e-01 6.96218610e-01
7.99525917e-01 -4.70578149e-02 5.26893735e-01 -1.87158048e-01
-7.80858338e-01 2.33985428e-02 -8.49228442e-01 8.38383734e-01
7.32753634e-01 -1.00513411e+00 -6.08923435e-01 5.03443539e-01
1.24670315e+00 -1.73125923e-01 -8.61025929e-01 4.05596346e-01
4.33563292e-01 -5.89476347e-01 1.28627992e+00 -9.88338053e-01
5.84044218e-01 -2.93539047e-01 -4.66167808e-01 -6.95343733e-01
5.76010533e-02 -3.23679358e-01 8.13244954e-02 1.29119110e+00
3.11452359e-01 -2.27906153e-01 1.65275574e-01 2.43111610e-01
-2.32856959e-01 -5.37200391e-01 -6.55134439e-01 -3.19078624e-01
9.14215669e-02 -6.05542898e-01 2.98761100e-01 1.05585194e+00
-3.29875559e-01 8.76014948e-01 -3.60345572e-01 1.17635392e-01
6.36808157e-01 2.47363389e-01 8.94386590e-01 -6.66231632e-01
-2.31939211e-01 -5.36257803e-01 -1.32416859e-01 -1.41262960e+00
1.71561301e-01 -1.10359406e+00 -1.74300037e-02 -1.93703008e+00
6.33368433e-01 1.90293208e-01 -2.48602405e-01 7.75239706e-01
-1.50853470e-01 5.35984576e-01 4.55896586e-01 1.97965205e-01
-1.41644967e+00 6.94301248e-01 1.62056303e+00 -5.33777595e-01
2.43736565e-01 -6.35772586e-01 -6.48059011e-01 6.25279307e-01
3.91416371e-01 3.24841768e-01 -3.36500525e-01 -1.11496508e+00
2.76657701e-01 1.72228158e-01 1.03001642e+00 -6.82157159e-01
3.17077130e-01 4.86365333e-02 4.37479526e-01 -8.03068995e-01
4.32620853e-01 -7.00234950e-01 -4.06182081e-01 2.60033816e-01
-5.20087898e-01 8.10804814e-02 3.43752891e-01 4.99980360e-01
-4.90900666e-01 1.19824544e-01 6.64808989e-01 -2.16671869e-01
-1.18111241e+00 3.89326483e-01 -1.78242192e-01 2.37448141e-01
7.87109137e-01 1.87783375e-01 -7.84661531e-01 -4.71031219e-01
-8.01662564e-01 9.46063876e-01 2.11607650e-01 5.99682271e-01
9.38005090e-01 -1.27877045e+00 -8.58368397e-01 -7.59783685e-02
6.87059045e-01 -9.01537612e-02 6.43849492e-01 1.20895672e+00
-4.61280495e-01 6.91852152e-01 -1.15204565e-01 -9.93036509e-01
-1.07710755e+00 7.61093020e-01 3.35961878e-01 2.91636735e-02
-3.29881668e-01 1.08192694e+00 5.34741282e-01 -6.27409458e-01
4.73862469e-01 -1.96819440e-01 -2.25976884e-01 2.06394941e-01
4.33386475e-01 -1.72529683e-01 -2.78537333e-01 -7.62920141e-01
-3.86049122e-01 6.73707724e-01 -2.54050761e-01 -1.39519796e-01
8.03120375e-01 -4.45056558e-01 -5.97658932e-01 3.22281897e-01
1.27023232e+00 -6.58020496e-01 -8.91160965e-01 -4.10920918e-01
-2.29979023e-01 2.12856874e-01 -8.94757211e-02 -1.00569367e+00
-1.09416747e+00 1.31338596e+00 7.23321438e-01 7.18735307e-02
1.08573365e+00 5.45886338e-01 1.01412487e+00 5.93336940e-01
5.59540428e-02 -1.02881682e+00 4.36712414e-01 7.36499906e-01
9.41796362e-01 -1.58896577e+00 -5.56249559e-01 -2.85389684e-02
-1.05326700e+00 7.68540561e-01 7.84580112e-01 3.22062492e-01
4.55172718e-01 -3.58401686e-01 3.74730915e-01 -2.88315535e-01
-8.96707594e-01 -6.73455834e-01 6.84500039e-01 5.97471118e-01
1.45266101e-01 5.96718444e-03 2.65213072e-01 4.59249467e-01
3.12285393e-01 -2.43308887e-01 -6.08283132e-02 6.15203142e-01
-4.00226235e-01 -5.84970891e-01 -3.16403896e-01 -2.37275716e-02
-7.27069303e-02 -5.21623075e-01 -5.02372444e-01 8.43840539e-01
-4.90221614e-03 8.91670525e-01 7.78956711e-02 -2.74529457e-01
2.12329760e-01 1.61828965e-01 5.64892292e-01 -2.21251681e-01
-5.91099203e-01 1.91062063e-01 -9.03438851e-02 -6.33245528e-01
-4.73184079e-01 -1.01720177e-01 -1.12731302e+00 7.76551738e-02
1.30069330e-01 2.50014979e-02 8.01174879e-01 1.02487576e+00
5.02649665e-01 4.65541869e-01 1.49856970e-01 -5.25940359e-01
-2.20091954e-01 -8.23585629e-01 -1.12365283e-01 3.92596066e-01
2.98960418e-01 -9.90020111e-02 2.66397037e-02 4.35169369e-01] | [10.884016036987305, 1.6042437553405762] |
17126d66-ab45-484b-b58a-b9f1af76b708 | truly-unordered-probabilistic-rule-sets-for | 2206.08804 | null | https://arxiv.org/abs/2206.08804v3 | https://arxiv.org/pdf/2206.08804v3.pdf | Truly Unordered Probabilistic Rule Sets for Multi-class Classification | Rule set learning has long been studied and has recently been frequently revisited due to the need for interpretable models. Still, existing methods have several shortcomings: 1) most recent methods require a binary feature matrix as input, while learning rules directly from numeric variables is understudied; 2) existing methods impose orders among rules, either explicitly or implicitly, which harms interpretability; and 3) currently no method exists for learning probabilistic rule sets for multi-class target variables (there is only one for probabilistic rule lists). We propose TURS, for Truly Unordered Rule Sets, which addresses these shortcomings. We first formalize the problem of learning truly unordered rule sets. To resolve conflicts caused by overlapping rules, i.e., instances covered by multiple rules, we propose a novel approach that exploits the probabilistic properties of our rule sets. We next develop a two-phase heuristic algorithm that learns rule sets by carefully growing rules. An important innovation is that we use a surrogate score to take the global potential of the rule set into account when learning a local rule. Finally, we empirically demonstrate that, compared to non-probabilistic and (explicitly or implicitly) ordered state-of-the-art methods, our method learns rule sets that not only have better interpretability but also better predictive performance. | ['Matthijs van Leeuwen', 'Lincen Yang'] | 2022-06-17 | null | null | null | null | ['classification'] | ['methodology'] | [ 4.38655466e-01 3.26055974e-01 -5.99877298e-01 -4.37680125e-01
-6.81980848e-01 -7.27837861e-01 5.26663959e-01 2.93938309e-01
-7.25329891e-02 1.10728705e+00 5.41312918e-02 -5.74899793e-01
-8.96270931e-01 -9.50302660e-01 -7.12239742e-01 -5.71729362e-01
9.53539275e-03 9.69471574e-01 2.78247029e-01 5.37957251e-02
1.80277646e-01 3.99518728e-01 -1.90288568e+00 4.53844845e-01
1.15837646e+00 9.19245839e-01 -2.16325477e-01 1.55291721e-01
-1.01865627e-01 9.04096127e-01 -3.91015679e-01 -3.70309919e-01
1.85092568e-01 -3.27048033e-01 -8.81349862e-01 -1.04753055e-01
2.14884892e-01 -1.24305375e-02 8.75521228e-02 9.47015584e-01
-1.54502377e-01 2.02450559e-01 1.01930511e+00 -1.24712300e+00
-5.18865049e-01 1.00092614e+00 -3.43345970e-01 -1.97048858e-01
2.14329109e-01 -6.83087111e-02 1.36580372e+00 -6.64642215e-01
4.20541376e-01 1.35450208e+00 5.59485972e-01 4.27538604e-01
-1.28032446e+00 -6.25893474e-01 5.39281726e-01 4.24266160e-01
-1.40475821e+00 -1.52035192e-01 6.87288582e-01 -3.50658327e-01
7.40424097e-01 6.45068049e-01 5.42139828e-01 8.39938641e-01
1.96462989e-01 7.00437427e-01 1.24953747e+00 -5.79295635e-01
4.99375314e-01 1.68810084e-01 3.77631903e-01 8.48437369e-01
6.02715969e-01 4.09973145e-01 -4.72279847e-01 -4.72314388e-01
4.66139942e-01 2.34239977e-02 -4.75535402e-03 -6.46165550e-01
-8.76472414e-01 1.07696557e+00 3.22189555e-02 1.23077638e-01
-2.48787329e-01 -2.15116888e-02 2.10801914e-01 1.14344470e-01
1.94731787e-01 6.53878391e-01 -8.57239366e-01 -2.01680381e-02
-7.74779081e-01 3.62858832e-01 9.68696415e-01 6.12924814e-01
9.60899115e-01 -3.51304144e-01 -1.39703557e-01 6.82186544e-01
4.97640818e-01 3.40324074e-01 1.79521307e-01 -7.25951076e-01
3.61568362e-01 8.75244498e-01 9.95718539e-02 -9.35233653e-01
-2.79355288e-01 -1.96414977e-01 -6.39671981e-01 2.49512509e-01
4.66022968e-01 -1.43330349e-02 -9.35800433e-01 1.66862535e+00
2.62542188e-01 -9.97346453e-03 1.01171434e-02 6.47579432e-01
3.99018347e-01 4.22826260e-01 -9.02700648e-02 -4.08062071e-01
1.09818900e+00 -4.97140199e-01 -5.47116220e-01 -1.17700614e-01
4.33657140e-01 -3.15205872e-01 9.13003683e-01 8.16392720e-01
-5.73332489e-01 -1.69196337e-01 -1.01362693e+00 2.42469639e-01
-5.87677181e-01 -2.85268445e-02 1.09616423e+00 7.64054179e-01
-5.35739601e-01 5.63021481e-01 -7.59465814e-01 1.98317450e-02
3.48904461e-01 6.75628126e-01 -2.29166955e-01 -1.86007079e-02
-1.36205244e+00 1.08367407e+00 7.73551047e-01 -1.06515959e-01
-5.32020509e-01 -5.54513812e-01 -1.00612891e+00 1.47298470e-01
1.07963276e+00 -6.01592183e-01 1.03873980e+00 -8.94282281e-01
-1.49040914e+00 3.20852160e-01 -3.13708812e-01 -5.01139760e-01
3.90069783e-01 -2.48917118e-01 -4.54800546e-01 -1.89170927e-01
-5.79082184e-02 4.05903995e-01 6.06955469e-01 -1.50390542e+00
-7.84812987e-01 -2.80369282e-01 2.85213351e-01 4.12423760e-02
-2.68700510e-01 -2.09613800e-01 -2.85915047e-01 -5.50068498e-01
1.68638565e-02 -8.89734447e-01 -3.07511628e-01 -3.00865710e-01
-7.26360917e-01 -4.46084470e-01 6.72262669e-01 -1.05488356e-02
1.71845782e+00 -1.81721282e+00 2.17594709e-02 8.31745207e-01
1.68800741e-01 2.74292856e-01 1.53907478e-01 1.83622301e-01
-1.99540630e-01 3.77386302e-01 -7.05026686e-01 1.70237407e-01
8.89986902e-02 7.08186448e-01 -5.40571749e-01 -1.01847143e-03
3.53327096e-01 6.22368336e-01 -1.09414089e+00 -5.20428598e-01
3.98714483e-01 9.82702151e-02 -6.11119449e-01 -1.68875962e-01
-3.36250424e-01 -2.59848144e-02 -6.02669179e-01 5.01181304e-01
4.89929259e-01 -3.50985944e-01 6.74514711e-01 -7.89696798e-02
-6.29465953e-02 2.93415487e-01 -1.69657576e+00 8.75076532e-01
-2.85103947e-01 4.89381626e-02 -5.14540851e-01 -1.00149071e+00
8.34935963e-01 1.27193332e-01 4.40967321e-01 -7.25356117e-02
-7.63607398e-02 2.75991648e-01 4.01421674e-02 -3.70413154e-01
6.63179085e-02 -3.42456281e-01 -9.39832777e-02 3.37755710e-01
-1.01563804e-01 -2.20473617e-01 4.32450026e-01 -1.25630021e-01
1.12780547e+00 1.71024472e-01 8.64440739e-01 -2.28541359e-01
5.43015182e-01 -2.69668195e-02 9.29370761e-01 1.13445771e+00
2.51133174e-01 5.22743762e-01 7.36085951e-01 -7.96247840e-01
-3.51781756e-01 -1.08942115e+00 -3.13552856e-01 9.99316990e-01
8.90146643e-02 -6.03507698e-01 -3.13868463e-01 -1.20046484e+00
3.59588802e-01 9.05310750e-01 -8.66169214e-01 -1.77577391e-01
-4.06490028e-01 -8.68005097e-01 2.76609123e-01 5.43082654e-01
-3.56891262e-03 -9.05569017e-01 -5.11085093e-01 2.42792696e-01
3.09599917e-02 -7.81735480e-01 -1.80642888e-01 6.40239060e-01
-9.29562271e-01 -1.49978733e+00 2.48801634e-01 -2.44454190e-01
6.84660017e-01 -1.09913088e-01 1.17930698e+00 1.71629384e-01
1.33062541e-01 1.17464900e-01 -3.27582449e-01 -6.05756581e-01
-4.04261917e-01 8.24382454e-02 2.57994741e-01 7.06600100e-02
5.03356695e-01 -4.49860007e-01 -3.90849560e-02 4.66434062e-01
-1.02844083e+00 8.52190182e-02 9.04194057e-01 9.68122602e-01
7.72070765e-01 3.37368667e-01 6.44068241e-01 -1.40963125e+00
4.92064387e-01 -4.20191854e-01 -6.88802719e-01 6.50262356e-01
-1.03434289e+00 6.49788380e-01 7.80564070e-01 -2.92446107e-01
-9.61351573e-01 4.01448339e-01 2.16424644e-01 -2.42941007e-01
-2.88080692e-01 6.98380172e-01 -2.73589104e-01 1.41212076e-01
7.30724931e-01 -8.34065676e-02 -3.58781815e-02 -2.30983719e-01
3.80761534e-01 5.33967257e-01 1.85030147e-01 -9.63833749e-01
8.53497803e-01 3.72828633e-01 1.04957327e-01 -2.93639034e-01
-1.12572360e+00 -1.99245334e-01 -6.47599518e-01 7.50289634e-02
3.37509722e-01 -3.94768715e-01 -8.36379528e-01 -1.58546731e-01
-1.00735378e+00 -1.08271837e-01 -3.56091589e-01 4.60019171e-01
-5.74524999e-01 3.51051122e-01 -8.43572766e-02 -1.01546872e+00
-5.95231466e-02 -1.02366173e+00 7.40383685e-01 -3.15196626e-02
-6.03072882e-01 -1.02668822e+00 7.71826832e-03 2.14665323e-01
-1.94365121e-02 2.79493332e-01 1.25394726e+00 -8.60222638e-01
-4.09320176e-01 5.17438948e-02 4.89740893e-02 1.85012385e-01
4.34630096e-01 2.37705395e-01 -6.56365693e-01 1.70951441e-01
-1.82859674e-01 -2.91496128e-01 9.15276706e-01 3.71892691e-01
1.48234487e+00 -6.75146937e-01 -5.79097927e-01 2.25397304e-01
1.09982252e+00 2.49297693e-01 4.34779346e-01 2.38973558e-01
5.81832409e-01 7.14251459e-01 7.91278422e-01 5.62247455e-01
4.48881984e-01 6.88262582e-01 3.23045641e-01 1.77945226e-01
2.29723334e-01 -2.59034842e-01 2.72854805e-01 3.71294141e-01
-2.23632142e-01 -2.18363956e-01 -1.01097727e+00 3.71980220e-01
-2.20552301e+00 -8.67012024e-01 -4.11303937e-02 2.36676264e+00
1.20219421e+00 4.51901585e-01 1.10646442e-01 5.11686862e-01
7.43011951e-01 -1.84061244e-01 -6.34601653e-01 -3.60563695e-01
-1.10108323e-01 3.51374418e-01 3.99212599e-01 7.32701302e-01
-1.20959735e+00 7.75296390e-01 6.83903885e+00 9.15545166e-01
-8.56677711e-01 -3.53837818e-01 6.23383045e-01 -1.38874307e-01
-7.07678616e-01 2.64642954e-01 -8.37130964e-01 4.43407238e-01
4.99312550e-01 -8.04766268e-02 4.11288559e-01 8.74671936e-01
4.80561797e-03 -9.31852460e-02 -1.47014630e+00 7.85015166e-01
-1.35662109e-01 -1.38122535e+00 5.16159058e-01 1.00855559e-01
7.62748003e-01 -3.97170395e-01 -4.21354994e-02 4.05014664e-01
7.26956904e-01 -1.24322236e+00 7.60079861e-01 5.91010511e-01
5.76758206e-01 -9.78308737e-01 7.21696556e-01 3.60145330e-01
-1.04091251e+00 -3.64213943e-01 -1.09962739e-01 -3.06770384e-01
-1.55520082e-01 1.15895295e+00 -8.06702077e-01 7.90849268e-01
3.93375427e-01 6.88072085e-01 -5.29548526e-01 8.31450224e-01
-5.95761240e-01 8.24051201e-01 -5.82263887e-01 -1.15148678e-01
1.19989999e-01 -1.11666307e-01 4.35045242e-01 9.41011846e-01
6.25876859e-02 1.47910580e-01 2.93593049e-01 8.88846695e-01
2.57627189e-01 -1.88233674e-01 -6.90227389e-01 2.43957698e-01
6.39967740e-01 9.36641574e-01 -6.61213577e-01 -1.56271785e-01
-2.73464352e-01 1.72244012e-01 4.22327578e-01 3.25267762e-01
-7.43213236e-01 -1.51915878e-01 5.35388291e-01 1.35602271e-02
3.47653866e-01 3.47214611e-03 -6.22381806e-01 -1.02226412e+00
1.04566820e-01 -9.72306252e-01 7.21167266e-01 -1.53341979e-01
-1.44310606e+00 2.62403756e-01 2.75869906e-01 -1.13770676e+00
-5.47841728e-01 -8.47413480e-01 -2.67085522e-01 3.94913077e-01
-1.43884540e+00 -9.11004543e-01 1.85570151e-01 3.14987332e-01
2.71105468e-01 -2.22621206e-02 9.35011566e-01 -2.55607124e-02
-4.96297240e-01 6.34402394e-01 3.84432599e-02 -9.20562074e-02
5.43752372e-01 -1.42630816e+00 -2.21265256e-01 6.01861417e-01
3.38437736e-01 8.95989895e-01 6.71672881e-01 -6.82171524e-01
-1.28186524e+00 -9.76528168e-01 8.88105035e-01 -7.76957750e-01
5.17722309e-01 -2.57424027e-01 -9.21934545e-01 6.82608426e-01
-4.03445721e-01 6.95002601e-02 9.29558098e-01 7.18148708e-01
-4.97099251e-01 -3.19728822e-01 -1.10257971e+00 5.32784283e-01
1.00503635e+00 -1.09446727e-01 -8.90880764e-01 1.22791290e-01
3.72004569e-01 -2.09227905e-01 -6.52222693e-01 8.33826184e-01
7.55514503e-01 -9.54547226e-01 8.20903838e-01 -7.49068141e-01
3.61426711e-01 -6.66397572e-01 -1.18754292e-02 -1.18258750e+00
-4.30366695e-01 -5.66525042e-01 -6.68701768e-01 1.01081085e+00
7.93188810e-01 -7.83476532e-01 6.52254641e-01 6.79188371e-01
6.96198791e-02 -1.10818410e+00 -9.42976177e-01 -9.34732497e-01
-2.45523322e-02 -5.60822904e-01 7.76391089e-01 9.63605464e-01
2.02157378e-01 3.49887580e-01 -3.76721919e-01 9.46702436e-02
4.17310864e-01 4.47886080e-01 4.57879156e-01 -1.64442337e+00
-4.46852297e-01 -5.79271138e-01 -3.08113158e-01 -7.17543662e-01
2.70775586e-01 -7.98514426e-01 1.26498789e-01 -1.34344125e+00
3.62727344e-01 -7.97514498e-01 -5.39165556e-01 1.02798796e+00
-4.94224608e-01 -1.51812077e-01 -1.81844365e-02 7.96304941e-02
-7.52989769e-01 4.28936720e-01 9.23269331e-01 1.04649365e-02
-2.19337493e-01 1.26176476e-02 -1.04116440e+00 9.46201563e-01
6.22280121e-01 -6.78801239e-01 -5.02373159e-01 -6.46010563e-02
5.47392130e-01 -2.26367384e-01 3.04322094e-01 -8.92745852e-01
8.92062262e-02 -7.62681901e-01 3.29437792e-01 -4.97425586e-01
-5.04404865e-02 -9.14384186e-01 2.53419697e-01 3.74440342e-01
-3.49177331e-01 -3.23454052e-01 4.98529449e-02 7.34414756e-01
-2.31543928e-01 -3.67349774e-01 5.84733546e-01 1.28747135e-01
-5.28915286e-01 1.05732128e-01 -4.46190327e-01 -1.98721290e-02
9.96520996e-01 -1.24718703e-01 -5.04863635e-03 -1.23126835e-01
-5.99553823e-01 2.66900450e-01 2.69595325e-01 4.59396541e-01
5.37099481e-01 -1.24179792e+00 -6.16323173e-01 -1.70709522e-04
2.83680290e-01 2.25701883e-01 -2.35554203e-01 4.23873484e-01
-6.12395704e-02 4.85609055e-01 2.07583398e-01 -4.62793976e-01
-9.85432863e-01 5.94421148e-01 1.32459551e-01 -7.29119599e-01
-2.38028377e-01 5.28609037e-01 1.24196470e-01 -6.96366489e-01
2.49181949e-02 -5.62752485e-01 -3.24750364e-01 -1.69216227e-02
3.88705492e-01 3.30215991e-01 8.57991055e-02 -8.33085775e-02
-6.04466498e-01 5.57417870e-01 -2.69241720e-01 5.10050170e-02
1.29809892e+00 3.28048468e-01 -8.40075091e-02 5.57853699e-01
5.53235948e-01 8.67266133e-02 -1.10810530e+00 -1.72729120e-01
3.37523669e-01 -3.55713934e-01 -2.18061090e-01 -1.13777363e+00
-6.69938982e-01 4.43031281e-01 1.53715894e-01 2.64572918e-01
1.05528951e+00 5.81031740e-02 3.04942071e-01 6.23447120e-01
4.45796698e-01 -1.16014326e+00 -2.62105584e-01 9.35803592e-01
7.23442674e-01 -1.29458380e+00 2.56163657e-01 -7.42741227e-01
-5.47827184e-01 1.23483539e+00 5.18275380e-01 1.78355902e-01
5.50426245e-01 3.32603961e-01 -2.15871304e-01 -1.05359830e-01
-8.98264468e-01 -2.38297373e-01 5.48364699e-01 4.77304369e-01
3.45021337e-01 3.24072808e-01 -4.09789532e-01 9.53267992e-01
-1.86642513e-01 1.80891603e-01 7.71216229e-02 8.95746469e-01
-3.82879257e-01 -1.43333173e+00 -5.02017558e-01 8.82464826e-01
-1.93415746e-01 -1.71708226e-01 -6.74073458e-01 7.61545956e-01
4.41540897e-01 1.06920326e+00 -3.02080244e-01 -4.32802945e-01
2.53349245e-01 8.79956633e-02 5.03358364e-01 -9.10624146e-01
-4.60697144e-01 -4.79633778e-01 1.47314787e-01 -3.36187869e-01
-2.93197870e-01 -7.70845830e-01 -1.37552869e+00 -3.17694023e-02
-4.71097350e-01 3.76252919e-01 1.37451231e-01 1.32091427e+00
3.93216074e-01 3.37693721e-01 5.64470112e-01 -3.75645101e-01
-6.63227081e-01 -6.19396746e-01 -3.81866217e-01 3.00906450e-01
1.10808767e-01 -1.16731811e+00 -3.62093538e-01 -7.58511499e-02] | [8.808249473571777, 6.39182186126709] |
06711db8-b679-4b07-bf8b-59951eb03f71 | attacking-point-cloud-segmentation-with-color | 2112.05871 | null | https://arxiv.org/abs/2112.05871v4 | https://arxiv.org/pdf/2112.05871v4.pdf | On Adversarial Robustness of Point Cloud Semantic Segmentation | Recent research efforts on 3D point cloud semantic segmentation (PCSS) have achieved outstanding performance by adopting neural networks. However, the robustness of these complex models have not been systematically analyzed. Given that PCSS has been applied in many safety-critical applications like autonomous driving, it is important to fill this knowledge gap, especially, how these models are affected under adversarial samples. As such, we present a comparative study of PCSS robustness. First, we formally define the attacker's objective under performance degradation and object hiding. Then, we develop new attack by whether to bound the norm. We evaluate different attack options on two datasets and three PCSS models. We found all the models are vulnerable and attacking point color is more effective. With this study, we call the attention of the research community to develop new approaches to harden PCSS models. | ['Yufei Ding', 'Boyuan Feng', 'Zhou Li', 'Zhe Zhou', 'Jiacen Xu'] | 2021-12-11 | null | null | null | null | ['point-cloud-segmentation'] | ['computer-vision'] | [ 2.92663306e-01 6.41667545e-02 8.89899433e-02 -1.22616343e-01
-3.24073493e-01 -9.34035122e-01 3.47843230e-01 -8.50174874e-02
-2.70899743e-01 2.15550423e-01 -4.98185515e-01 -7.61880159e-01
1.14744768e-01 -8.93916488e-01 -8.61258626e-01 -6.09993577e-01
-5.38114607e-02 -4.74467911e-02 7.58587897e-01 -1.85688585e-01
6.98162198e-01 9.26427662e-01 -1.29162920e+00 -1.32962421e-01
7.41720676e-01 1.05513573e+00 -3.28537196e-01 5.65188289e-01
4.67290655e-02 1.79855600e-01 -8.96547198e-01 -6.93161070e-01
7.59130418e-01 7.83535168e-02 -7.46104836e-01 -3.08008045e-01
1.31245926e-01 -2.78486580e-01 -1.81777835e-01 1.55499482e+00
4.78368431e-01 -1.32418171e-01 3.79181176e-01 -1.80646741e+00
-3.77767473e-01 4.99345183e-01 -6.23108804e-01 2.19740361e-01
-1.05766498e-01 4.34056550e-01 4.76705372e-01 -5.25621958e-02
1.76499709e-01 1.38885427e+00 5.41249514e-01 7.81627774e-01
-7.38479078e-01 -1.13472259e+00 2.04302818e-01 3.87675732e-01
-1.27130222e+00 2.57738191e-03 8.99544418e-01 -2.11713597e-01
5.56977272e-01 5.05507529e-01 2.70786941e-01 1.12692451e+00
1.20282553e-01 9.24611688e-01 1.35153794e+00 -1.25809893e-01
3.44746232e-01 2.63221711e-01 3.39936078e-01 3.87475699e-01
5.70562243e-01 4.18496579e-01 -9.06045064e-02 -2.65815347e-01
2.52118379e-01 -4.27597612e-01 -1.27929136e-01 -2.30564401e-01
-6.19022548e-01 8.03072691e-01 3.53954464e-01 -1.72880664e-01
2.95673013e-01 2.09017456e-01 4.29700524e-01 1.14881694e-01
1.71255797e-01 4.29024756e-01 -5.08600652e-01 -1.71311386e-02
-4.79126066e-01 2.53910959e-01 6.18947744e-01 1.07763898e+00
4.04757559e-01 1.40691906e-01 2.69749165e-01 2.63710976e-01
2.75187731e-01 5.70832729e-01 -6.05308404e-03 -9.38114345e-01
4.51187849e-01 4.06272441e-01 1.43897738e-02 -1.25906289e+00
-1.82687223e-01 -2.30420321e-01 -6.80982888e-01 5.45140982e-01
1.90466329e-01 -2.06890166e-01 -7.48088598e-01 1.52742314e+00
3.83285522e-01 6.08109653e-01 2.07204878e-01 7.97330976e-01
4.36671883e-01 5.46859384e-01 -2.24252138e-02 2.21252531e-01
1.19928312e+00 -6.76155865e-01 -4.90112215e-01 -1.17265368e-02
4.31993872e-01 -6.54695809e-01 1.01604640e+00 3.51083815e-01
-8.16391587e-01 -4.68326867e-01 -1.54856110e+00 3.02105576e-01
-6.90784097e-01 -4.51947898e-01 5.41375279e-01 1.49648654e+00
-6.26863420e-01 5.96536100e-01 -1.10243344e+00 -3.96627821e-02
6.26733065e-01 5.80511212e-01 2.69352198e-02 3.25084805e-01
-1.45199430e+00 9.41603482e-01 3.71633261e-01 8.97596478e-02
-7.87313938e-01 -6.61780238e-01 -6.63083553e-01 -1.79586515e-01
4.71454084e-01 -5.05331099e-01 1.16640222e+00 -4.53493059e-01
-1.59751904e+00 6.15490198e-01 2.96684563e-01 -9.53954995e-01
6.62474632e-01 -2.50274301e-01 -5.62369764e-01 2.61873633e-01
-1.48260638e-01 6.33453131e-01 9.85023737e-01 -1.50180495e+00
-6.91631854e-01 -4.86380696e-01 3.49067599e-01 -9.74697024e-02
-2.95128286e-01 3.09389234e-01 -4.31893408e-01 -5.08464873e-01
-2.05441024e-02 -1.26991320e+00 -4.49342221e-01 7.60203376e-02
-8.72098565e-01 7.00044557e-02 1.27500129e+00 -2.15515226e-01
7.85454988e-01 -2.65077376e+00 -3.41627121e-01 4.35129672e-01
-2.83136815e-02 7.70558774e-01 3.74972299e-02 1.06167398e-01
1.96746379e-01 6.69470310e-01 -5.15853107e-01 -7.25279823e-02
2.04856053e-01 1.85454443e-01 -8.44273150e-01 6.07998013e-01
4.05836552e-01 6.26680255e-01 -4.15745199e-01 -3.47150475e-01
3.93812895e-01 2.14526206e-01 -6.23474538e-01 6.76560327e-02
-1.48487195e-01 4.39532340e-01 -7.46962965e-01 6.98333025e-01
1.19789076e+00 2.06096545e-01 -4.60457057e-01 -1.23195626e-01
1.60851687e-01 -2.66443282e-01 -1.20900917e+00 9.06809926e-01
8.32076296e-02 4.81472135e-01 1.79821141e-02 -8.33318889e-01
9.04785693e-01 -1.02980256e-01 2.22298235e-01 -4.34042156e-01
3.97994369e-01 1.44047618e-01 4.06200662e-02 -3.85410249e-01
5.50142586e-01 1.23824865e-01 -1.80362985e-01 4.74063912e-03
-6.51108265e-01 -5.03254533e-01 -3.51726532e-01 1.30380750e-01
8.61298323e-01 -1.40475869e-01 -1.39238119e-01 -4.13201638e-02
5.73253274e-01 1.27178401e-01 5.17748892e-01 8.31437230e-01
-6.29823327e-01 5.79297364e-01 6.61066592e-01 -1.82595387e-01
-6.15842462e-01 -1.08089316e+00 -1.21141531e-01 3.61798137e-01
9.43640888e-01 -1.59335732e-01 -1.05659378e+00 -8.89467478e-01
6.38976544e-02 8.09731364e-01 -4.59566146e-01 -5.72814524e-01
-5.18472791e-01 -7.01574862e-01 1.32973540e+00 4.50320899e-01
9.15877044e-01 -7.38181114e-01 -7.72854149e-01 -2.81066805e-01
9.61346030e-02 -1.42654896e+00 -2.08652481e-01 -9.49927345e-02
-7.16045856e-01 -1.30443668e+00 -2.41829231e-01 -3.64924282e-01
4.65507120e-01 3.80903482e-01 5.84057748e-01 2.25964010e-01
-1.00070216e-01 2.31737331e-01 -5.18391192e-01 -9.34829473e-01
-6.19150698e-01 9.90226269e-02 8.69611725e-02 -6.38573021e-02
3.52624267e-01 -5.08904517e-01 -4.64985400e-01 5.86229622e-01
-1.08276522e+00 -2.22944409e-01 3.61979544e-01 2.46570393e-01
3.39316338e-01 4.31897283e-01 2.70324737e-01 -9.06226873e-01
6.02474689e-01 -4.03109372e-01 -1.12253332e+00 5.43058664e-03
-5.97760558e-01 -3.88376713e-02 6.39372945e-01 -3.84630501e-01
-7.32058227e-01 -2.24612541e-02 -5.58948576e-01 -4.76554245e-01
-3.61565858e-01 -3.31420749e-02 -5.92610359e-01 -5.74185014e-01
4.04408902e-01 -7.57297352e-02 -1.28747031e-01 -1.95279419e-01
2.70238191e-01 7.04313755e-01 6.45038068e-01 -6.82764947e-01
1.26143086e+00 7.39858568e-01 3.12945664e-01 -8.97196770e-01
-3.85971904e-01 -1.37256637e-01 -2.32331604e-01 -1.00423954e-01
9.15196598e-01 -4.32797164e-01 -1.06365097e+00 7.44351149e-01
-1.27083933e+00 -7.03368429e-03 4.36154343e-02 1.49850890e-01
-4.68658000e-01 8.47082794e-01 -3.93893033e-01 -7.97278941e-01
-2.41441414e-01 -1.60997367e+00 9.19806361e-01 1.94092542e-01
1.43010885e-01 -5.17430246e-01 -2.58735955e-01 5.14325023e-01
2.20216036e-01 6.02052629e-01 8.36856723e-01 -8.58191013e-01
-8.05258930e-01 -2.57563978e-01 -3.08606565e-01 5.87034225e-01
-1.52447194e-01 2.93850154e-01 -1.08134699e+00 -3.40154082e-01
3.26298684e-01 1.70415014e-01 6.92671299e-01 1.40351616e-02
1.67768168e+00 -2.28600323e-01 -2.50047863e-01 9.93658721e-01
1.16443217e+00 4.73991841e-01 9.40555036e-01 6.73832297e-01
6.63478196e-01 5.50168157e-01 7.29110539e-01 3.84781621e-02
-3.88281606e-03 4.99097258e-01 1.06291521e+00 4.98049036e-02
2.20999300e-01 -1.10245302e-01 4.00924683e-01 3.02266985e-01
1.37757093e-01 -2.20828608e-01 -8.49610269e-01 1.86045438e-01
-1.44278252e+00 -7.29505002e-01 -2.78822780e-01 1.84408140e+00
4.26142693e-01 6.02031469e-01 -7.11039379e-02 4.06116009e-01
8.86981308e-01 3.05715352e-01 -7.16977537e-01 -4.37251985e-01
-6.93611652e-02 1.22930981e-01 1.07128608e+00 2.93173105e-01
-1.19882846e+00 1.14329970e+00 6.22935677e+00 8.79091620e-01
-1.27351940e+00 -1.42989472e-01 5.51194370e-01 2.84407914e-01
-6.28103837e-02 1.57914996e-01 -8.70254457e-01 7.40766168e-01
6.60558283e-01 3.21744233e-02 1.60325855e-01 9.49421465e-01
-7.99898133e-02 1.66287810e-01 -7.10069180e-01 7.76141942e-01
-1.16560757e-01 -1.07581651e+00 3.96567620e-02 1.46628708e-01
4.70589817e-01 -1.47433460e-01 3.97547275e-01 1.02030836e-01
4.18520927e-01 -9.36719775e-01 8.70082438e-01 -1.19061738e-01
3.86894256e-01 -9.64801610e-01 7.73167253e-01 3.77225369e-01
-9.02949870e-01 -1.45057691e-02 -3.26089054e-01 1.82396248e-01
1.42293081e-01 1.07432306e-01 -6.80548370e-01 6.85678780e-01
8.82168472e-01 1.28976285e-01 -7.64765024e-01 1.08204865e+00
-3.49114835e-01 8.20511401e-01 -4.17900920e-01 -3.46035361e-02
3.55660826e-01 4.76875864e-02 8.08478713e-01 9.99689519e-01
1.24570176e-01 1.67589083e-01 -2.10958086e-02 7.94263601e-01
2.04379678e-01 -3.24222624e-01 -6.74854934e-01 9.55553204e-02
6.34658217e-01 7.85000741e-01 -1.05690241e+00 6.37569502e-02
-2.76217520e-01 7.62870669e-01 -2.20481962e-01 2.69752115e-01
-1.08286309e+00 -4.07517105e-01 1.07832289e+00 -1.04629971e-01
2.55362451e-01 -4.57219303e-01 -7.28318334e-01 -7.57186294e-01
-1.02424003e-01 -9.19303596e-01 2.39951119e-01 -5.10563195e-01
-1.08445585e+00 4.79409248e-01 1.96103886e-01 -1.23133135e+00
4.42620963e-01 -7.98194230e-01 -6.58618629e-01 5.31059384e-01
-1.66800725e+00 -9.10666466e-01 -4.40882295e-02 6.30369723e-01
2.07048580e-01 -1.85982674e-01 3.74590486e-01 2.08553299e-01
-6.51511610e-01 8.19907308e-01 -2.72854090e-01 1.98430806e-01
3.29020590e-01 -8.88701677e-01 9.64646876e-01 1.23692405e+00
-1.33265480e-02 5.81384957e-01 9.04379845e-01 -6.92603648e-01
-1.29630792e+00 -1.11939371e+00 -3.00544649e-02 -4.39323634e-01
5.35319567e-01 -2.06205711e-01 -9.31472003e-01 4.45953995e-01
1.27559379e-01 1.11328112e-02 3.44501287e-01 -4.27275717e-01
-3.76159698e-01 4.16010357e-02 -1.49235177e+00 8.85809064e-01
8.27328086e-01 -3.41948837e-01 -3.98521632e-01 -4.39271554e-02
1.19198084e+00 -6.20517194e-01 -4.19610739e-01 7.22445607e-01
2.06607491e-01 -1.04023278e+00 1.12782717e+00 -4.54590201e-01
-5.91384945e-03 -6.00272655e-01 -3.42117131e-01 -8.46803248e-01
2.73385197e-01 -6.33298934e-01 1.15730196e-01 1.10857761e+00
1.77262649e-01 -9.54584181e-01 9.88426089e-01 6.00645721e-01
-3.47977310e-01 -5.46398461e-01 -9.85028505e-01 -1.01810551e+00
1.92629769e-01 -8.45483661e-01 1.04485524e+00 8.64734650e-01
-5.79222858e-01 -2.57061779e-01 -3.29443604e-01 1.17248893e+00
6.90631807e-01 -1.13050826e-01 1.02337778e+00 -8.32870841e-01
3.52570601e-02 -5.33086658e-01 -7.89304197e-01 -8.16654027e-01
1.60189405e-01 -4.90518898e-01 -1.84046105e-01 -9.00314510e-01
-2.75297165e-01 -4.55885768e-01 -2.08941028e-01 4.10697103e-01
-1.46205112e-01 2.37764165e-01 4.91908699e-01 2.68265251e-02
-2.65491456e-01 4.29264754e-01 1.08631957e+00 -4.40165341e-01
1.67858582e-02 5.23300946e-01 -6.95051074e-01 9.02299464e-01
1.31985331e+00 -6.30275071e-01 -5.34077048e-01 -3.72808516e-01
4.43762951e-02 -3.82756442e-01 5.94437122e-01 -1.14586306e+00
1.93839148e-01 -5.27692258e-01 -1.89996004e-01 -8.25773120e-01
1.87356234e-01 -1.25071144e+00 9.85138863e-02 7.04573393e-01
6.58695912e-03 1.03034917e-02 4.67319608e-01 5.53726137e-01
-1.35161519e-01 -3.42962086e-01 9.98435020e-01 2.88365949e-02
-8.20478380e-01 3.51578921e-01 -2.08298221e-01 -7.05375597e-02
1.48627031e+00 -5.83415151e-01 -4.25233781e-01 -2.57785395e-02
-3.22295517e-01 2.87132978e-01 6.37022257e-01 6.84536815e-01
6.25373781e-01 -1.00280452e+00 -3.48529071e-01 3.52219880e-01
-4.68456857e-02 2.75885522e-01 2.20824182e-01 2.76968390e-01
-8.69184732e-01 9.08276886e-02 -2.02218801e-01 -6.44716978e-01
-1.34512246e+00 9.09679055e-01 2.12025687e-01 2.20371783e-01
-4.49031144e-01 7.16651797e-01 1.22483179e-01 -3.80068690e-01
4.94989038e-01 -3.75179976e-01 -7.52743185e-02 -4.30236191e-01
1.92739233e-01 4.75067407e-01 3.98566611e-02 -5.28971791e-01
-4.85577524e-01 5.97770333e-01 2.11619660e-02 -6.30727923e-03
9.21840072e-01 -7.85904005e-02 9.94199663e-02 -1.15439177e-01
1.13102508e+00 1.49002090e-01 -1.28642654e+00 2.57269114e-01
1.49056345e-01 -6.93939507e-01 -2.67347664e-01 -4.73839998e-01
-1.32857275e+00 1.04597259e+00 8.50563884e-01 4.35116976e-01
1.08858931e+00 -2.51567066e-01 1.04313111e+00 3.14233363e-01
5.10374188e-01 -9.34662938e-01 -1.24179222e-01 4.93511230e-01
5.31075776e-01 -1.07512021e+00 -1.59255669e-01 -8.23520005e-01
-5.50125897e-01 7.29436338e-01 6.04175806e-01 -2.43194357e-01
7.17154026e-01 5.75568199e-01 1.69278771e-01 -8.25747475e-02
-1.28263727e-01 6.77269623e-02 -4.10340540e-02 8.34655404e-01
-5.85239589e-01 3.49696390e-02 3.61975841e-02 6.94478273e-01
-6.34219408e-01 -2.73783028e-01 5.43523252e-01 9.47878182e-01
-3.44057232e-01 -1.12885046e+00 -6.58687592e-01 -5.77197585e-04
-6.02362871e-01 1.58092216e-01 -4.90495712e-01 7.97208071e-01
2.09867671e-01 1.13643169e+00 -3.63687038e-01 -7.86310256e-01
4.42017585e-01 -3.12102854e-01 6.35716841e-02 -2.07540199e-01
-4.48414385e-01 -5.01419306e-01 -9.76234823e-02 -6.03998721e-01
-2.89768815e-01 -5.66435218e-01 -1.23185384e+00 -5.59445918e-01
-4.70889360e-01 1.32018313e-01 8.18363607e-01 8.54550064e-01
3.02671701e-01 3.71210188e-01 8.34622264e-01 -4.46099430e-01
-8.15125585e-01 -3.30970585e-01 -3.89877588e-01 1.97597519e-01
2.90043265e-01 -6.22992396e-01 -4.83118981e-01 -3.01119894e-01] | [7.699694633483887, -4.4511237144470215] |
6ff90d4d-7687-4488-946e-45a71e9a3d2a | cnn-based-cost-volume-analysis-as-confidence | 1905.07287 | null | https://arxiv.org/abs/1905.07287v2 | https://arxiv.org/pdf/1905.07287v2.pdf | CNN-based Cost Volume Analysis as Confidence Measure for Dense Matching | Due to its capability to identify erroneous disparity assignments in dense stereo matching, confidence estimation is beneficial for a wide range of applications, e.g. autonomous driving, which needs a high degree of confidence as mandatory prerequisite. Especially, the introduction of deep learning based methods resulted in an increasing popularity of this field in recent years, caused by a significantly improved accuracy. Despite this remarkable development, most of these methods rely on features learned from disparity maps only, not taking into account the corresponding 3-dimensional cost volumes. However, it was already demonstrated that with conventional methods based on hand-crafted features this additional information can be used to further increase the accuracy. In order to combine the advantages of deep learning and cost volume based features, in this paper, we propose a novel Convolutional Neural Network (CNN) architecture to directly learn features for confidence estimation from volumetric 3D data. An extensive evaluation on three datasets using three common dense stereo matching techniques demonstrates the generality and state-of-the-art accuracy of the proposed method. | ['Christian Heipke', 'Max Mehltretter'] | 2019-05-17 | null | null | null | null | ['stereo-matching'] | ['computer-vision'] | [-7.66759813e-02 5.63538559e-02 4.23788615e-02 -5.99017739e-01
-4.93323505e-01 1.06030144e-01 7.13593960e-01 5.04246235e-01
-7.21999109e-01 9.09950197e-01 -7.35615045e-02 -7.01239109e-02
-1.00454383e-01 -1.01071513e+00 -6.15426898e-01 -5.20566642e-01
-8.94238800e-02 4.73006099e-01 4.43042785e-01 -6.67895749e-02
4.60289448e-01 6.50059521e-01 -2.31310630e+00 -1.97980210e-01
1.14824808e+00 1.40753829e+00 2.87268162e-01 -2.57387031e-02
-2.84101903e-01 4.99517560e-01 -2.68959850e-01 -4.53608453e-01
2.86343157e-01 5.39596863e-02 -4.08823103e-01 -1.97165295e-01
4.11627859e-01 -4.52271730e-01 -3.95165145e-01 9.25740600e-01
3.32885325e-01 1.81418303e-02 5.89773178e-01 -1.00042582e+00
1.94318518e-01 -3.63352112e-02 -3.65652651e-01 1.48045108e-01
1.70334339e-01 1.04061134e-01 8.71501148e-01 -8.79948378e-01
3.08103919e-01 7.86806703e-01 7.23559082e-01 2.15823874e-01
-1.03639901e+00 -8.73740494e-01 -8.01361129e-02 4.90120053e-01
-1.38761115e+00 -1.67561591e-01 8.77314806e-01 -6.23962820e-01
9.87828791e-01 -1.04638554e-01 9.22114849e-01 7.00815618e-01
2.82697111e-01 5.30259073e-01 1.48517025e+00 -3.01475525e-01
3.05066913e-01 2.81562179e-01 -1.44821525e-01 5.58715880e-01
4.98012513e-01 5.09154558e-01 -4.04853344e-01 1.94856972e-01
5.58469951e-01 2.57634930e-02 -2.37415344e-01 -8.61106217e-01
-8.73984039e-01 9.46008801e-01 7.79251695e-01 3.62348616e-01
-1.92845315e-01 1.21784694e-01 3.50190163e-01 -3.34446728e-02
6.62640452e-01 2.30115041e-01 -1.95604876e-01 -4.08054799e-01
-1.08873761e+00 1.42545074e-01 4.11085546e-01 4.93086427e-01
1.20053947e+00 -8.20523053e-02 1.10013768e-01 5.96887767e-01
2.20845893e-01 3.85340065e-01 4.52728063e-01 -5.69804370e-01
2.73475707e-01 8.83249521e-01 2.96476148e-02 -1.09127676e+00
-5.95210075e-01 -6.89528227e-01 -1.02089441e+00 7.58599281e-01
3.95429522e-01 3.57200682e-01 -7.82100558e-01 1.54554200e+00
3.16202730e-01 2.96730608e-01 -2.33196005e-01 9.97509837e-01
5.89655161e-01 1.53914347e-01 -1.86406627e-01 1.13421336e-01
8.21250796e-01 -3.27352434e-01 -2.18422621e-01 -8.28588381e-02
5.41704059e-01 -5.59490621e-01 6.63126945e-01 4.88199174e-01
-6.25473440e-01 -7.52877116e-01 -1.50676346e+00 2.86809728e-02
-5.32964826e-01 -9.40004587e-02 8.33954573e-01 8.46477926e-01
-9.44895446e-01 9.02533233e-01 -7.17029989e-01 -2.28933871e-01
5.33145487e-01 5.32277048e-01 -5.00430167e-01 -1.59629792e-01
-1.21774924e+00 1.07799900e+00 4.46184099e-01 2.74594456e-01
-4.99860495e-01 -5.49550414e-01 -9.59380150e-01 7.48961791e-02
5.51293790e-02 -5.54054916e-01 7.66906083e-01 -6.58096075e-01
-1.39684045e+00 1.05227280e+00 -7.01230345e-03 -6.92033291e-01
1.09638166e+00 -2.72260159e-01 -1.05064094e-01 5.57011552e-02
2.75824275e-02 7.66837239e-01 7.37038016e-01 -1.07599866e+00
-8.40689778e-01 -5.16159058e-01 2.45795265e-01 -1.65378541e-01
-3.74346763e-01 -6.42399013e-01 -2.46672809e-01 -2.54544199e-01
2.72708714e-01 -7.24541247e-01 -1.51607603e-01 1.06268279e-01
-8.29021335e-02 -1.55701905e-01 5.30018866e-01 -3.83257866e-01
7.77385175e-01 -2.00392127e+00 4.57985699e-02 2.51085758e-01
4.20455575e-01 3.20106149e-01 4.06989276e-01 2.18080133e-01
1.48687392e-01 -3.47463548e-01 -3.59119296e-01 -3.92658889e-01
-1.83056414e-01 8.78542736e-02 3.78442146e-02 6.62689626e-01
4.86887634e-01 5.82908869e-01 -9.39426601e-01 -5.72592914e-01
1.00365210e+00 7.24245250e-01 -5.05286932e-01 3.00741553e-01
-2.08998974e-02 7.84613371e-01 -2.69953132e-01 2.66560704e-01
9.11864042e-01 -7.20359292e-03 -2.68317401e-01 -1.47015229e-01
-3.01616818e-01 9.27146524e-02 -9.96743977e-01 1.71274698e+00
-7.81866550e-01 7.55326211e-01 -3.49346042e-01 -1.20882630e+00
1.16048408e+00 1.09258220e-01 6.06619060e-01 -1.02942371e+00
3.37910920e-01 6.91076100e-01 3.67605723e-02 -1.48893997e-01
5.67885220e-01 -8.72676447e-02 8.10165703e-02 1.71058811e-02
-5.45223765e-02 -5.07377684e-01 9.40407515e-02 -8.21130797e-02
8.58751595e-01 1.25586301e-01 2.62099475e-01 -3.56713235e-01
1.01822102e+00 -1.59825116e-01 4.23161298e-01 3.65060270e-01
-1.22074038e-01 7.61674225e-01 1.91651508e-01 -6.11382544e-01
-1.03219926e+00 -7.59302139e-01 -6.87352836e-01 1.22643977e-01
4.73591477e-01 -9.62066129e-02 -5.14697611e-01 -4.07046020e-01
3.82421464e-01 3.87253255e-01 -7.24074423e-01 -1.79442436e-01
-5.83221555e-01 -4.42559689e-01 1.91333696e-01 6.16753876e-01
9.38145816e-01 -6.91889584e-01 -1.07345235e+00 3.05217683e-01
1.39425933e-01 -1.17021072e+00 3.43501478e-01 2.79166698e-01
-1.15398479e+00 -1.09781134e+00 -7.92303741e-01 -3.65555167e-01
4.50706989e-01 1.58621460e-01 1.08491015e+00 3.17981929e-01
-5.06707132e-01 1.37694448e-01 -2.27647573e-01 -1.66780248e-01
-2.97419041e-01 2.89265841e-01 -2.83271912e-02 1.74048729e-02
4.50362980e-01 -7.34807730e-01 -6.87832117e-01 2.60694116e-01
-7.83893466e-01 2.12183058e-01 6.69347823e-01 8.76235723e-01
4.76217389e-01 -9.60481614e-02 5.08262873e-01 -5.97478569e-01
1.51243895e-01 -1.72379240e-01 -1.03928542e+00 -2.06457421e-01
-8.35017920e-01 3.91505122e-01 4.02721733e-01 1.52223125e-01
-8.23091507e-01 -3.40855517e-03 -6.31016433e-01 -3.24247569e-01
-1.85457215e-01 5.14976144e-01 3.06730922e-02 -2.72944063e-01
3.69842649e-01 1.13756701e-01 1.34590745e-01 -3.44246924e-01
-4.82786959e-03 6.53871715e-01 4.38532352e-01 -4.59067404e-01
7.65249550e-01 6.60204828e-01 4.00934011e-01 -7.27470815e-01
-6.39008999e-01 -5.56721568e-01 -8.36736202e-01 -4.76535827e-01
6.93064988e-01 -1.09144390e+00 -7.39612639e-01 6.06673300e-01
-1.00482476e+00 -2.18116436e-02 -2.50159800e-01 7.13062167e-01
-5.34668207e-01 4.53285784e-01 -2.05699250e-01 -8.29802752e-01
-1.42331973e-01 -1.36924171e+00 1.02147949e+00 3.42164278e-01
2.38751490e-02 -9.17018652e-01 2.68942006e-02 2.06452504e-01
6.82936907e-01 3.77907455e-01 7.69334614e-01 -2.55767643e-01
-8.68689597e-01 -5.57637751e-01 -5.41640341e-01 4.08900768e-01
1.31268948e-01 -2.27359191e-01 -1.35318553e+00 -2.70004719e-01
-1.22469766e-02 -3.89229327e-01 1.10225892e+00 3.21748912e-01
1.15190721e+00 7.27422118e-01 -3.12620580e-01 6.15922272e-01
1.64902306e+00 -3.01193267e-01 7.75891602e-01 5.80270588e-01
4.02222723e-01 5.95678210e-01 8.05382431e-01 6.21039808e-01
2.39831388e-01 8.04021597e-01 9.56328034e-01 -2.51739416e-02
-1.89051107e-01 -1.11693963e-01 -2.63756603e-01 5.62251091e-01
-3.57791126e-01 3.09131294e-01 -8.12828064e-01 4.28430021e-01
-1.75798523e+00 -7.97600031e-01 -8.15306380e-02 2.73083997e+00
4.17427421e-01 6.38865352e-01 -2.05861077e-01 5.62342823e-01
6.54755533e-01 1.39064208e-01 -4.57529306e-01 -1.49541616e-01
-5.32298833e-02 3.61827672e-01 4.59999055e-01 4.13907766e-01
-9.91117597e-01 6.15692675e-01 4.97270060e+00 8.07179272e-01
-1.45974660e+00 -1.27882257e-01 6.34104431e-01 2.21946701e-01
-3.55843067e-01 -2.19392166e-01 -7.74767697e-01 4.53990191e-01
6.07159913e-01 -1.50370151e-01 1.92324165e-03 8.47859502e-01
2.85059363e-02 -6.33926034e-01 -1.03760195e+00 1.26633716e+00
-1.03743710e-01 -1.29367006e+00 -2.63529986e-01 1.79931164e-01
5.94919741e-01 1.74500555e-01 5.47923855e-02 2.93161601e-01
-3.57804209e-01 -1.05755568e+00 6.82238698e-01 2.47595727e-01
8.74621630e-01 -9.09371972e-01 1.17614555e+00 2.59896964e-01
-1.22751296e+00 1.35032699e-01 -4.70036924e-01 -4.24658865e-01
9.05573219e-02 1.18687356e+00 -7.53719151e-01 6.49564147e-01
7.82457471e-01 7.74729848e-01 -4.44616675e-01 1.44883358e+00
-2.38833830e-01 8.90585557e-02 -4.25312400e-01 -6.08250089e-02
1.65134564e-01 -1.37555197e-01 2.87468433e-01 1.07761681e+00
3.83383155e-01 -3.74176681e-01 -2.58499473e-01 9.23918843e-01
9.22632962e-02 1.45960137e-01 -6.47288620e-01 3.45531762e-01
5.84534965e-02 1.34501994e+00 -7.64639258e-01 -2.86883712e-01
-6.18604898e-01 6.78139865e-01 4.78582263e-01 -1.76478341e-01
-8.07228744e-01 -3.65011871e-01 7.74039268e-01 3.88714284e-01
4.83493477e-01 -5.42756557e-01 -3.20797145e-01 -1.08798552e+00
1.22482069e-01 -2.39871755e-01 8.26727785e-03 -3.91748637e-01
-1.03058887e+00 7.64213264e-01 -3.16831730e-02 -1.51038516e+00
-4.10968155e-01 -6.69113100e-01 -3.22929591e-01 1.05734384e+00
-2.07216859e+00 -8.13940048e-01 -8.86449873e-01 4.40446913e-01
3.92213762e-01 1.49020180e-01 6.82119370e-01 5.03846824e-01
-1.13069691e-01 3.04914445e-01 9.17636752e-02 -7.85020068e-02
6.00049078e-01 -1.13065457e+00 2.53541201e-01 5.96563816e-01
-6.14269674e-02 2.15383694e-01 7.74052143e-01 -4.11743134e-01
-9.80952740e-01 -8.68247330e-01 7.14259624e-01 -1.00731507e-01
3.29149455e-01 -2.39154384e-01 -8.80767107e-01 3.80384922e-02
-8.00486803e-02 2.02172920e-01 2.90743083e-01 1.56448856e-02
-2.70900697e-01 -4.22020108e-01 -1.18660021e+00 1.94129691e-01
9.19213474e-01 -4.93274271e-01 -5.52449524e-01 -1.23632610e-01
2.59767115e-01 -5.51407576e-01 -6.94590449e-01 9.96034324e-01
6.64323807e-01 -1.69741893e+00 7.68340707e-01 1.69608220e-01
4.72242922e-01 -3.01771969e-01 -1.25637516e-01 -1.15877354e+00
9.25477147e-02 2.08461210e-01 8.73626396e-03 7.93976247e-01
1.75826967e-01 -8.83738995e-01 1.15512002e+00 5.34234345e-01
-2.71598548e-01 -7.26002395e-01 -1.44375384e+00 -7.20179737e-01
-3.60586122e-02 -6.16370499e-01 5.62356293e-01 4.81552333e-01
-6.57455996e-02 -2.27106791e-02 -2.21682802e-01 5.30700684e-02
6.37122750e-01 4.02979881e-01 8.19935203e-01 -1.77673423e+00
-1.59787595e-01 -4.53467369e-01 -1.15401101e+00 -1.10683060e+00
8.22278336e-02 -7.00406194e-01 1.33302078e-01 -1.42748797e+00
4.26583253e-02 -8.70257258e-01 -4.03898954e-01 -1.03063233e-01
-1.34160116e-01 5.50988197e-01 -1.16386682e-01 5.84911406e-02
-2.59000093e-01 8.43434751e-01 8.79736602e-01 -2.78450817e-01
6.07858300e-02 1.76505342e-01 -3.40938754e-02 5.21727264e-01
7.31446624e-01 -2.99627006e-01 -2.68261522e-01 -3.32735926e-01
3.30202937e-01 -1.25136778e-01 3.88862967e-01 -1.70247817e+00
9.12308469e-02 2.98063934e-01 5.36290705e-01 -7.31506348e-01
6.49783075e-01 -1.05530655e+00 -6.21693162e-03 7.11938500e-01
1.95599243e-01 -2.16174781e-01 4.32697952e-01 6.49519265e-01
-7.23746121e-01 -1.51163340e-01 8.50913107e-01 2.40653791e-02
-1.03606772e+00 2.78999507e-01 -7.26274922e-02 -2.72472054e-01
1.01967645e+00 -5.77775776e-01 6.02013953e-02 -2.97336429e-01
-2.50460833e-01 -1.10585086e-01 7.20891416e-01 3.62998754e-01
7.23154306e-01 -1.15877485e+00 -4.43218201e-01 3.43891561e-01
5.18017292e-01 1.82580724e-01 3.44679683e-01 9.60503459e-01
-6.59969747e-01 6.57493174e-01 -5.50115049e-01 -1.14245176e+00
-8.66958857e-01 4.07878667e-01 3.40330958e-01 -2.34009385e-01
-6.46214902e-01 6.45485640e-01 8.93803388e-02 -1.24561071e-01
1.74612567e-01 -6.12987876e-01 -3.01381826e-01 -1.40882865e-01
2.99526870e-01 1.65847778e-01 5.88448584e-01 -6.42956734e-01
-4.14307088e-01 8.28470349e-01 1.54600739e-01 1.06773570e-01
1.28810298e+00 -8.89268816e-02 2.05047905e-01 4.09525156e-01
1.13268566e+00 -6.41686097e-02 -1.39009762e+00 -9.22664180e-02
7.78338164e-02 -9.38991487e-01 3.12102169e-01 -3.08749735e-01
-1.28953755e+00 1.32745564e+00 7.84076273e-01 -7.48337479e-03
8.41913342e-01 -2.01606780e-01 5.69779456e-01 7.18572363e-02
1.13418925e+00 -9.01242256e-01 -6.42421022e-02 2.54513144e-01
5.91442049e-01 -1.76653671e+00 7.88124129e-02 -4.48657066e-01
-1.17379509e-01 1.35657692e+00 6.21168792e-01 -1.90259501e-01
8.03385198e-01 7.44917616e-02 -9.63822678e-02 9.24526155e-02
-1.55116498e-01 -4.62624103e-01 1.37470186e-01 5.53243756e-01
5.37289917e-01 -1.05626337e-01 -3.63533080e-01 4.65335026e-02
-5.92305362e-02 2.51152605e-01 3.43058258e-01 8.67881238e-01
-6.19077861e-01 -1.15426028e+00 -1.82551131e-01 5.87166607e-01
-1.51273012e-01 1.13069445e-01 2.54729807e-01 8.94396722e-01
2.48827487e-01 8.71645749e-01 2.69271195e-01 -4.41804081e-01
2.78969884e-01 -2.48074621e-01 6.94959939e-01 -3.97481203e-01
-2.94620037e-01 -5.10863543e-01 -1.12996593e-01 -7.20539510e-01
-5.97522974e-01 -5.18447638e-01 -1.04831803e+00 -3.15365911e-01
-4.48231369e-01 6.39597028e-02 9.53581870e-01 9.92536008e-01
1.98234186e-01 3.63346905e-01 4.45355237e-01 -1.16212964e+00
-3.02891254e-01 -9.36725736e-01 -5.64431250e-01 4.47514027e-01
2.99549967e-01 -1.23014843e+00 -4.95701760e-01 -4.91789371e-01] | [8.66506290435791, -2.121777296066284] |
75181cb9-9f73-426e-9d98-2764428d575c | training-generative-adversarial-networks-from | 1905.12660 | null | https://arxiv.org/abs/1905.12660v2 | https://arxiv.org/pdf/1905.12660v2.pdf | Training Generative Adversarial Networks from Incomplete Observations using Factorised Discriminators | Generative adversarial networks (GANs) have shown great success in applications such as image generation and inpainting. However, they typically require large datasets, which are often not available, especially in the context of prediction tasks such as image segmentation that require labels. Therefore, methods such as the CycleGAN use more easily available unlabelled data, but do not offer a way to leverage additional labelled data for improved performance. To address this shortcoming, we show how to factorise the joint data distribution into a set of lower-dimensional distributions along with their dependencies. This allows splitting the discriminator in a GAN into multiple "sub-discriminators" that can be independently trained from incomplete observations. Their outputs can be combined to estimate the density ratio between the joint real and the generator distribution, which enables training generators as in the original GAN framework. We apply our method to image generation, image segmentation and audio source separation, and obtain improved performance over a standard GAN when additional incomplete training examples are available. For the Cityscapes segmentation task in particular, our method also improves accuracy by an absolute 14.9% over CycleGAN while using only 25 additional paired examples. | ['Simon Dixon', 'Sebastian Ewert', 'Daniel Stoller'] | 2019-05-29 | null | https://openreview.net/forum?id=Hye1RJHKwB | https://openreview.net/pdf?id=Hye1RJHKwB | iclr-2020-1 | ['audio-source-separation'] | ['audio'] | [ 6.68783009e-01 4.11605716e-01 5.29370876e-03 -2.25138083e-01
-1.36829710e+00 -7.58059502e-01 6.87039316e-01 -1.75394878e-01
-3.48446310e-01 9.01529431e-01 1.25118226e-01 -1.89831167e-01
4.31746542e-01 -8.19854736e-01 -7.70440578e-01 -1.00007343e+00
4.04364735e-01 6.43350482e-01 -1.55368149e-01 1.19015230e-02
-4.53388870e-01 1.44530147e-01 -1.38606298e+00 1.85149997e-01
9.86925185e-01 8.37086856e-01 1.23681329e-01 7.46268749e-01
9.51253343e-03 5.97916424e-01 -9.12753105e-01 -3.43502969e-01
3.24448407e-01 -1.04773092e+00 -5.72873652e-01 2.80983031e-01
1.42445520e-01 -3.74486953e-01 -1.67002693e-01 9.07869697e-01
5.87730169e-01 2.41073091e-02 8.33837628e-01 -1.21984732e+00
-1.74266353e-01 5.83003521e-01 -5.84616423e-01 -3.54987144e-01
5.61743267e-02 1.32647112e-01 7.94372797e-01 -2.92236209e-01
4.57895935e-01 1.06704307e+00 5.15748203e-01 7.33056009e-01
-1.61537111e+00 -7.99784303e-01 -1.76445529e-01 -1.92402348e-01
-1.10779250e+00 -5.05944669e-01 7.06517279e-01 -3.38678539e-01
4.04391110e-01 2.50165284e-01 4.24189448e-01 1.32538724e+00
-4.02998090e-01 9.73399937e-01 1.24330854e+00 -4.99316037e-01
2.25943893e-01 2.65216678e-01 -7.00606227e-01 3.06610823e-01
-2.31185034e-01 -2.36685351e-02 -2.85659850e-01 4.87672910e-02
8.05382907e-01 -2.42271185e-01 -3.73918593e-01 -3.53525430e-02
-1.15996957e+00 1.20732808e+00 3.87167126e-01 1.80805951e-01
-1.90344319e-01 2.21945733e-01 1.42391130e-01 3.92601222e-01
7.77926445e-01 5.16051114e-01 -2.87909061e-01 -2.33409181e-02
-1.13908744e+00 2.19699070e-01 7.04427481e-01 8.22131813e-01
8.80183041e-01 3.49527270e-01 -1.59261554e-01 1.03935254e+00
1.24339916e-01 6.72057450e-01 6.40159011e-01 -1.04703856e+00
5.77820361e-01 1.85489833e-01 -1.05392165e-01 -5.34259558e-01
-8.49438179e-03 -4.88981336e-01 -1.08113182e+00 3.37156028e-01
6.81879938e-01 -4.83738661e-01 -1.32312119e+00 1.87080085e+00
2.24174798e-01 2.98110366e-01 1.72571346e-01 6.86123431e-01
3.86822701e-01 8.07724655e-01 -1.15698494e-01 -7.55151957e-02
1.09102094e+00 -9.78954494e-01 -4.57240611e-01 -5.37784159e-01
6.29390597e-01 -8.71796608e-01 8.70206952e-01 4.35591996e-01
-1.00563657e+00 -5.86207449e-01 -9.37519968e-01 1.08064406e-01
-1.02709450e-01 2.46368542e-01 4.15148735e-01 6.67367935e-01
-9.47320402e-01 6.17470622e-01 -7.45879471e-01 1.03078611e-01
4.72402483e-01 3.43032837e-01 -3.42688769e-01 -1.57583252e-01
-1.07598150e+00 5.82090735e-01 2.72043675e-01 -2.12205783e-03
-9.00733292e-01 -5.18724024e-01 -9.69386101e-01 -8.62533376e-02
3.63840878e-01 -5.65607965e-01 1.32457924e+00 -1.32664609e+00
-1.62717044e+00 7.71187544e-01 4.93185297e-02 -5.88600457e-01
7.66131282e-01 1.62092950e-02 -1.47493199e-01 6.66603148e-02
1.99114934e-01 1.10478318e+00 1.21564078e+00 -1.25929582e+00
-5.64258516e-01 -9.46575850e-02 -7.77158514e-02 2.36015394e-01
-9.45426226e-02 -3.47092003e-01 -4.30856407e-01 -9.75582659e-01
-2.03936890e-01 -1.15387440e+00 -3.39400142e-01 -2.29849145e-01
-4.94611531e-01 1.40470222e-01 6.15392625e-01 -9.24087644e-01
7.04916120e-01 -2.19478202e+00 3.05248171e-01 2.25453168e-01
8.50926191e-02 2.56832659e-01 -2.32153535e-01 2.30988458e-01
-4.15106975e-02 7.13357553e-02 -6.64268434e-01 -7.06431031e-01
-5.67155220e-02 4.29826289e-01 -4.48328078e-01 9.96268839e-02
3.73037308e-01 8.15221906e-01 -8.54774773e-01 -3.91277850e-01
1.88815594e-01 5.49456120e-01 -4.35167730e-01 3.84903550e-01
-4.56476450e-01 1.06557906e+00 -1.45748198e-01 2.46493384e-01
5.21695733e-01 5.67885190e-02 1.89220816e-01 1.46296665e-01
4.65263605e-01 2.40333974e-01 -1.09288728e+00 1.77350247e+00
-8.44940841e-01 8.48978937e-01 2.64509767e-02 -1.06306326e+00
9.01648641e-01 5.23496985e-01 3.77467752e-01 -5.80140650e-01
-2.96374615e-02 2.73650140e-01 4.07459699e-02 -3.23763862e-02
1.88560620e-01 -5.02731323e-01 -2.50946850e-01 6.21069193e-01
3.41602564e-01 -5.22366405e-01 1.95089638e-01 5.32119069e-03
9.92563426e-01 2.47804254e-01 -8.56212601e-02 3.42702955e-01
2.11925387e-01 -2.71585494e-01 4.78402615e-01 5.94109058e-01
4.03032094e-01 1.25548196e+00 7.83341110e-01 6.36696070e-02
-1.30100763e+00 -8.91920149e-01 1.82043001e-01 6.56651616e-01
-3.29508632e-01 -2.44747534e-01 -1.10227847e+00 -8.78446341e-01
-2.26648927e-01 7.62882054e-01 -4.81260240e-01 -8.22584853e-02
-4.59557265e-01 -8.98884594e-01 6.73126698e-01 5.20563722e-01
3.87925684e-01 -1.10901856e+00 -2.03081414e-01 2.29082704e-01
-3.30766529e-01 -1.29708838e+00 -3.47110003e-01 3.76038194e-01
-7.33803093e-01 -7.68409491e-01 -1.04826713e+00 -6.10353649e-01
7.63699234e-01 -2.94944435e-01 1.10375404e+00 -3.16767186e-01
-5.97202443e-02 1.00724764e-01 -2.68154502e-01 -4.64967579e-01
-9.93155837e-01 2.99811631e-01 -3.83580714e-01 2.26100564e-01
-1.91275448e-01 -7.40315974e-01 -4.96925324e-01 1.64547533e-01
-1.11516798e+00 2.76431233e-01 7.24276662e-01 1.07517827e+00
6.21170700e-01 6.41363263e-02 7.98492849e-01 -1.25284207e+00
3.21727455e-01 -4.93679285e-01 -5.27196765e-01 -1.02980331e-01
-4.68663007e-01 2.46946558e-01 8.85664225e-01 -4.48414057e-01
-1.18935072e+00 2.18149632e-01 -3.96655142e-01 -4.50340927e-01
-4.52810615e-01 3.66075099e-01 -2.90574193e-01 2.49147013e-01
6.10397875e-01 8.23926553e-02 1.37815177e-01 -4.94508028e-01
5.36034822e-01 7.19518483e-01 7.25415468e-01 -3.63106221e-01
8.32371116e-01 2.35189706e-01 2.79716272e-02 -6.44351482e-01
-7.11715639e-01 -1.78974316e-01 -3.58644992e-01 1.06139630e-01
8.06263804e-01 -1.05051231e+00 -1.08396240e-01 6.00409150e-01
-1.02425027e+00 -6.76683187e-01 -6.89088166e-01 4.71129060e-01
-7.50030935e-01 1.18438125e-01 -5.26213109e-01 -7.65063107e-01
-7.85655901e-02 -1.03777754e+00 1.28488314e+00 1.93804055e-01
-1.12761557e-01 -1.09956551e+00 4.44660261e-02 4.43135440e-01
2.33249277e-01 4.94326502e-01 8.46712708e-01 -6.40538454e-01
-4.38295156e-01 -3.91606480e-01 1.79720372e-01 8.31931889e-01
3.20405871e-01 -3.09972584e-01 -1.25680304e+00 -2.01513052e-01
6.92306012e-02 -4.47549611e-01 9.65577066e-01 2.04455838e-01
1.13020062e+00 -2.70099252e-01 -1.21230312e-01 6.52293265e-01
1.19135988e+00 1.07027426e-01 7.79841900e-01 -1.64999142e-01
8.28231037e-01 6.31242394e-01 2.80230880e-01 2.01396555e-01
6.39173726e-04 7.50817478e-01 3.18354428e-01 -3.62527579e-01
-4.28779483e-01 -5.28310001e-01 3.79059434e-01 7.80897737e-01
1.23371102e-01 -5.65749824e-01 -6.56595349e-01 5.21859169e-01
-1.64059579e+00 -7.37785578e-01 6.49338737e-02 2.27318430e+00
1.11163175e+00 -4.08501588e-02 2.08949134e-01 4.02000695e-01
6.52461112e-01 1.40807450e-01 -5.14874399e-01 -1.79154351e-01
-1.18741520e-01 6.71245456e-01 5.50629437e-01 5.20028651e-01
-1.01034594e+00 6.55272007e-01 5.92817688e+00 1.10603213e+00
-1.06481576e+00 1.85991049e-01 1.01008415e+00 -2.11820956e-02
-4.67657506e-01 7.02121109e-02 -4.06202346e-01 7.74780691e-01
1.06826019e+00 2.55002111e-01 5.47238052e-01 5.55514753e-01
-1.04498379e-01 -2.15760484e-01 -1.06780028e+00 9.93408024e-01
5.57699129e-02 -9.28925157e-01 -1.53619304e-01 1.98445216e-01
1.01594782e+00 -2.34740272e-01 2.18385592e-01 2.34957293e-01
3.40235144e-01 -1.22210455e+00 5.50001383e-01 2.41103247e-01
1.05001843e+00 -7.87820041e-01 7.76689947e-01 5.70146799e-01
-6.65667951e-01 1.59605071e-01 -1.10193364e-01 1.66430250e-01
3.55730891e-01 7.87949383e-01 -1.05283797e+00 5.68375409e-01
2.51897246e-01 4.31173801e-01 -3.45727742e-01 7.62728572e-01
-6.65865958e-01 9.30777133e-01 -4.70959425e-01 4.84655797e-01
4.99454737e-02 -3.83016169e-01 3.69869888e-01 8.70000482e-01
5.57154715e-01 -2.76909977e-01 -9.06288326e-02 8.82229447e-01
-3.18506837e-01 -1.89432815e-01 -5.77236950e-01 -2.03003258e-01
1.35926917e-01 1.30712903e+00 -8.78604233e-01 -3.43300134e-01
-2.44546339e-01 1.27602935e+00 1.79860182e-02 3.93163294e-01
-8.07533979e-01 -3.27299327e-01 3.43293577e-01 1.23270869e-01
4.50418204e-01 -1.76259596e-02 -1.72212124e-01 -1.13818717e+00
-1.51871629e-02 -1.13792121e+00 1.61015406e-01 -6.98593140e-01
-1.13144112e+00 6.04094803e-01 1.06735351e-02 -1.26675093e+00
-1.10712719e+00 -3.71152639e-01 -6.12553895e-01 1.02915156e+00
-1.29460335e+00 -1.37457299e+00 -2.11322382e-01 4.28041965e-01
5.14773548e-01 -1.89941108e-01 1.01377428e+00 3.95168453e-01
-4.19051141e-01 6.44023836e-01 2.54197508e-01 1.69264674e-01
8.41853499e-01 -1.49453497e+00 4.71766293e-01 8.55018020e-01
5.18462598e-01 -1.47278771e-01 6.20737970e-01 -3.94228101e-01
-7.92983055e-01 -1.11829948e+00 5.06330311e-01 -2.12154061e-01
3.58224988e-01 -5.89004695e-01 -6.21138215e-01 6.68724000e-01
1.46117255e-01 2.72820480e-02 5.89403093e-01 -2.38970965e-01
-2.07581028e-01 -1.63167134e-01 -1.10059941e+00 4.08861279e-01
6.39684677e-01 -5.89662492e-01 -1.22681484e-02 2.38620549e-01
5.56469619e-01 -4.85803068e-01 -6.41155362e-01 2.58549660e-01
3.16104889e-01 -9.39247608e-01 6.88329399e-01 -2.94425428e-01
6.27198935e-01 -2.86666662e-01 -9.75653529e-02 -1.71660745e+00
2.88811535e-01 -8.49157393e-01 2.39972547e-02 1.67331564e+00
5.55985093e-01 -7.51071572e-01 9.18453038e-01 3.32264751e-01
1.06666880e-02 -5.23113787e-01 -7.81769454e-01 -6.29459858e-01
8.97777081e-02 -4.65010613e-01 6.34045482e-01 7.75572360e-01
-5.33953428e-01 4.87503171e-01 -6.84513748e-01 -1.99616730e-01
4.73439515e-01 1.40734032e-01 9.23953831e-01 -8.68596256e-01
-8.79530311e-01 -1.62703007e-01 -4.00378048e-01 -1.04583228e+00
3.68977487e-01 -8.92331362e-01 3.09123367e-01 -1.28972709e+00
-6.42462820e-02 -7.47195423e-01 7.57361948e-02 5.26455045e-01
-1.64244965e-01 7.09030926e-01 2.56422937e-01 4.22068797e-02
-1.02934033e-01 5.62662601e-01 1.12535501e+00 -2.89390087e-01
-1.05630212e-01 3.41159672e-01 -7.36686528e-01 5.97647965e-01
8.01429689e-01 -6.26942098e-01 -6.29649580e-01 -4.47602630e-01
1.48311518e-02 2.73402452e-01 3.56801480e-01 -1.03934646e+00
-2.26860434e-01 1.60188362e-01 4.68549401e-01 2.14361027e-03
5.42567074e-01 -6.92435324e-01 4.73344743e-01 7.98051879e-02
-3.34322602e-01 -4.11788940e-01 4.94399592e-02 6.00455344e-01
-5.16161978e-01 -3.96024793e-01 6.33755088e-01 -7.00570494e-02
-4.01679985e-02 1.63236216e-01 -1.76561579e-01 1.63309842e-01
8.44683409e-01 2.95225307e-02 3.35029215e-02 -9.13063526e-01
-8.57215047e-01 -5.59526719e-02 4.72032785e-01 1.26412854e-01
2.33997330e-01 -1.44386899e+00 -7.93184876e-01 4.08565044e-01
-2.80787289e-01 6.12939119e-01 1.76349059e-01 7.11395264e-01
-2.61622310e-01 5.24229649e-03 -5.80423921e-02 -6.08676076e-01
-9.39928710e-01 2.94810146e-01 2.40756467e-01 -4.88832444e-01
-2.99658298e-01 7.18397200e-01 4.35377866e-01 -2.90444463e-01
-9.98720527e-03 -1.19217932e-01 9.49801281e-02 3.00301194e-01
1.73048392e-01 1.92008674e-01 1.68195680e-01 -5.56854069e-01
9.83903483e-02 3.45317036e-01 1.88199639e-01 -5.54758668e-01
1.32308853e+00 7.07414076e-02 1.16558753e-01 4.46849853e-01
1.34730422e+00 1.36744052e-01 -1.66277158e+00 1.13538913e-02
-4.72943425e-01 -4.42223251e-01 -1.83511779e-01 -6.72496438e-01
-1.37856722e+00 1.11774123e+00 3.63106519e-01 4.21786100e-01
1.36965823e+00 2.06465960e-01 9.68657970e-01 -2.53298342e-01
5.98191246e-02 -8.05049956e-01 -4.36261520e-02 3.92545648e-02
7.15607643e-01 -1.18671584e+00 -2.95143247e-01 -3.21693301e-01
-7.47528672e-01 9.04326081e-01 2.89180458e-01 -2.09779620e-01
2.03327745e-01 3.53989482e-01 1.69528320e-01 3.59139293e-01
-5.46764791e-01 -3.12774032e-01 3.06896895e-01 8.41549337e-01
2.18767121e-01 1.62327081e-01 7.26653934e-02 5.09179354e-01
-3.48582476e-01 -2.42795244e-01 4.78499115e-01 5.72508216e-01
1.32321477e-01 -1.60394025e+00 -2.85637975e-01 6.09895289e-01
-6.59040153e-01 -2.85193045e-02 -2.47368708e-01 5.62772691e-01
3.18237752e-01 9.06062961e-01 2.18462214e-01 -3.08394969e-01
4.85672103e-03 1.37035295e-01 5.11584938e-01 -7.07319498e-01
-3.15744787e-01 4.40264225e-01 1.62503853e-01 -2.56166518e-01
-4.54847217e-01 -5.92350125e-01 -8.64030600e-01 -2.56809574e-02
-4.33594435e-01 2.46192083e-01 6.94515884e-01 9.53926921e-01
1.44135192e-01 4.76405114e-01 7.51391530e-01 -1.01859474e+00
-5.00862300e-01 -9.68487799e-01 -4.53490764e-01 5.10527253e-01
4.67338920e-01 -2.30420202e-01 -5.32130003e-01 4.14275765e-01] | [11.636958122253418, -0.20047180354595184] |
6c38993d-70d4-4efe-b9f5-5e31bb6ce918 | reproducible-scaling-laws-for-contrastive | 2212.07143 | null | https://arxiv.org/abs/2212.07143v1 | https://arxiv.org/pdf/2212.07143v1.pdf | Reproducible scaling laws for contrastive language-image learning | Scaling up neural networks has led to remarkable performance across a wide range of tasks. Moreover, performance often follows reliable scaling laws as a function of training set size, model size, and compute, which offers valuable guidance as large-scale experiments are becoming increasingly expensive. However, previous work on scaling laws has primarily used private data \& models or focused on uni-modal language or vision learning. To address these limitations, we investigate scaling laws for contrastive language-image pre-training (CLIP) with the public LAION dataset and the open-source OpenCLIP repository. Our large-scale experiments involve models trained on up to two billion image-text pairs and identify power law scaling for multiple downstream tasks including zero-shot classification, retrieval, linear probing, and end-to-end fine-tuning. We find that the training distribution plays a key role in scaling laws as the OpenAI and OpenCLIP models exhibit different scaling behavior despite identical model architectures and similar training recipes. We open-source our evaluation workflow and all models, including the largest public CLIP models, to ensure reproducibility and make scaling laws research more accessible. Source code and instructions to reproduce this study will be available at https://github.com/LAION-AI/scaling-laws-openclip | ['Jenia Jitsev', 'Ludwig Schmidt', 'Christoph Schuhmann', 'Cade Gordon', 'Gabriel Ilharco', 'Mitchell Wortsman', 'Ross Wightman', 'Romain Beaumont', 'Mehdi Cherti'] | 2022-12-14 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Cherti_Reproducible_Scaling_Laws_for_Contrastive_Language-Image_Learning_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Cherti_Reproducible_Scaling_Laws_for_Contrastive_Language-Image_Learning_CVPR_2023_paper.pdf | cvpr-2023-1 | ['open-vocabulary-attribute-detection', 'zero-shot-cross-modal-retrieval'] | ['computer-vision', 'miscellaneous'] | [-2.14370772e-01 -7.46006072e-01 -4.07268912e-01 -4.55914855e-01
-1.10411358e+00 -8.96987379e-01 4.85561639e-01 1.39536276e-01
-8.23067844e-01 2.56391406e-01 2.05604613e-01 -4.32539493e-01
-5.23721203e-02 -3.82020622e-01 -7.63162196e-01 -4.64364797e-01
1.54558420e-01 4.00868863e-01 1.95548370e-01 -6.45308048e-02
1.68077424e-01 3.36320192e-01 -1.32541251e+00 5.04904449e-01
4.37837660e-01 9.27735209e-01 2.44110629e-01 9.90667164e-01
-2.97360718e-02 4.24789250e-01 -1.41249403e-01 -2.55748838e-01
4.14585888e-01 -1.78718597e-01 -7.34287858e-01 -4.74662572e-01
8.77191603e-01 -4.66555148e-01 -4.76367533e-01 9.14986134e-01
9.08476412e-01 3.18241380e-02 4.52749193e-01 -1.23209536e+00
-1.10353160e+00 5.82847357e-01 -5.87574959e-01 6.37273431e-01
-2.20727831e-01 6.19605005e-01 1.24993646e+00 -7.83579648e-01
4.38731968e-01 1.13134670e+00 8.10208261e-01 6.59023643e-01
-1.34423172e+00 -8.45097423e-01 -2.56894659e-02 1.11596428e-01
-1.34394610e+00 -8.07119250e-01 1.33475557e-01 -3.72450560e-01
1.25693047e+00 3.86510976e-02 3.40578198e-01 1.28108239e+00
1.51427120e-01 6.66784286e-01 9.63303268e-01 -4.30124670e-01
6.35197014e-02 2.13206150e-02 3.58686954e-01 7.26128519e-01
5.81246018e-02 -4.72723059e-02 -6.08235359e-01 -3.72271299e-01
7.22820461e-01 -8.25501978e-02 -1.81793794e-01 -1.32738993e-01
-1.15360188e+00 7.49226749e-01 4.61008400e-01 2.80854285e-01
1.23961139e-02 6.43027425e-01 7.14401543e-01 5.08214355e-01
5.99719524e-01 5.73623359e-01 -8.43326509e-01 -4.16930109e-01
-8.35776150e-01 2.95237839e-01 7.15821445e-01 7.90571451e-01
5.31007946e-01 -3.21521580e-01 -1.53457373e-01 1.27822387e+00
6.79771826e-02 5.83595395e-01 6.88749373e-01 -1.21538341e+00
3.14067483e-01 1.70609012e-01 -3.32255751e-01 -5.33543408e-01
-4.30501193e-01 -4.23358798e-01 -7.04479754e-01 -6.20296597e-02
5.50236642e-01 -4.90001962e-02 -9.61425781e-01 1.98991668e+00
4.73172963e-03 1.54930120e-02 -1.97938949e-01 8.11624825e-01
8.90099525e-01 5.97723901e-01 2.79719144e-01 2.84350365e-01
1.51954794e+00 -1.16594541e+00 -9.01899021e-03 -4.13766444e-01
9.38650846e-01 -9.06551898e-01 1.67939067e+00 2.08853051e-01
-1.12490284e+00 -4.05976892e-01 -8.00741434e-01 -5.81188738e-01
-4.07011509e-01 9.08128396e-02 7.38529921e-01 1.41676351e-01
-1.31318784e+00 6.52648032e-01 -9.32669520e-01 -7.62626529e-01
5.47843456e-01 1.30260944e-01 -2.58675635e-01 -2.83770472e-01
-9.41853225e-01 5.88030338e-01 6.66182786e-02 -3.62028062e-01
-7.72236645e-01 -1.12975419e+00 -4.63656247e-01 1.73776969e-01
1.29478633e-01 -7.56635249e-01 1.58067608e+00 -7.95583963e-01
-1.00889730e+00 1.09375668e+00 -7.89589509e-02 -3.64889294e-01
3.78706098e-01 -3.81204516e-01 7.22974017e-02 2.14635432e-02
1.24686481e-02 1.00996351e+00 6.26705825e-01 -7.36810744e-01
-3.56283545e-01 -4.32167113e-01 -6.41971305e-02 -2.40983791e-03
-5.98717391e-01 2.48795286e-01 -7.86268711e-01 -3.70495766e-01
-1.94539979e-01 -9.94447649e-01 6.19912446e-02 2.13008374e-01
-6.60313014e-03 -1.02706946e-01 3.89450312e-01 -4.23798203e-01
1.21869314e+00 -2.21839547e+00 -1.07157342e-01 -2.02223361e-01
2.82993108e-01 9.39665735e-02 -7.37209082e-01 5.36061168e-01
-3.63606438e-02 3.16676736e-01 -9.45056695e-03 -4.49447304e-01
8.46439078e-02 -1.21705234e-01 -3.59230369e-01 4.97783303e-01
-1.89869385e-02 9.38218951e-01 -5.07537127e-01 -4.21043068e-01
-1.42337114e-01 4.01264161e-01 -8.05943668e-01 2.38032136e-02
-4.05231506e-01 2.34397762e-02 -6.92669079e-02 5.69822788e-01
5.05998731e-01 -7.72698343e-01 -2.78153330e-01 -2.31614381e-01
-1.44780457e-01 3.69989686e-02 -5.62706649e-01 1.83825243e+00
-5.40139794e-01 7.88538694e-01 1.46083891e-01 -7.12786973e-01
3.40365887e-01 5.56557328e-02 3.00096124e-01 -8.37690473e-01
3.04215223e-01 1.94472224e-01 8.93819854e-02 -5.12918115e-01
2.85000473e-01 5.01821116e-02 1.08201817e-01 6.67489469e-01
2.58348823e-01 8.53343215e-03 4.54009145e-01 4.15590525e-01
1.32007086e+00 -7.54286200e-02 -7.81337321e-02 -2.74198920e-01
-8.15381408e-02 8.65729749e-02 1.65965557e-01 1.03474891e+00
-3.86357814e-01 6.93261087e-01 6.55826986e-01 -2.96120524e-01
-1.59598267e+00 -8.83822680e-01 -4.69194710e-01 1.91922963e+00
-3.29661220e-01 -4.71198648e-01 -7.32757449e-01 -7.36993924e-02
1.52363822e-01 4.95432049e-01 -6.55401647e-01 -1.31841600e-01
-2.31232926e-01 -8.84307146e-01 9.02633488e-01 6.23380959e-01
4.25827980e-01 -7.96146512e-01 -2.47310862e-01 -1.37375548e-01
3.41912434e-02 -1.30281115e+00 -7.26219833e-01 1.82633892e-01
-6.59705460e-01 -8.78455818e-01 -7.63084590e-01 -7.68976271e-01
3.26249748e-01 4.40045446e-01 1.25091577e+00 4.02427286e-01
-6.34843469e-01 5.48651934e-01 -1.20181710e-01 -3.19552124e-01
-4.26353127e-01 5.79322040e-01 3.62719968e-02 -5.38117290e-01
3.96463931e-01 -6.28083766e-01 -8.55240524e-01 1.99403808e-01
-8.84428799e-01 1.33724779e-01 6.16112232e-01 7.72381246e-01
4.90362465e-01 -5.09049535e-01 3.81837279e-01 -6.92100823e-01
9.00479734e-01 -6.28865421e-01 -8.05843830e-01 2.91205645e-01
-8.63777995e-01 1.73112676e-01 6.12742782e-01 -5.04123688e-01
-6.61463082e-01 -3.55151206e-01 -3.19607586e-01 -6.88660264e-01
-2.55033016e-01 4.99704510e-01 5.10160267e-01 3.51421498e-02
1.05446362e+00 1.64557308e-01 8.58730376e-02 -5.46973765e-01
4.30081755e-01 7.10693479e-01 4.54717308e-01 -7.26548254e-01
5.02455115e-01 3.88635278e-01 -2.75394052e-01 -7.72722960e-01
-9.99947309e-01 -6.79068387e-01 -4.06415403e-01 2.14604497e-01
7.69756913e-01 -1.18618023e+00 -8.25919151e-01 4.95629847e-01
-9.84821498e-01 -8.77341211e-01 -6.96009919e-02 3.93911153e-01
-4.35538292e-01 1.24122515e-01 -1.08854353e+00 -4.29755718e-01
-7.41768122e-01 -1.04973280e+00 1.20453691e+00 1.49171367e-01
-5.48235103e-02 -9.66246665e-01 2.89791375e-01 3.95520538e-01
8.21579397e-01 -3.75918895e-01 1.26211929e+00 -5.92692673e-01
-5.16597033e-01 -1.64730594e-01 -7.51715362e-01 4.41448689e-01
-5.51748753e-01 2.52233326e-01 -1.09497941e+00 -5.47410607e-01
-4.35528815e-01 -9.67954218e-01 1.11958849e+00 4.83316869e-01
1.31662714e+00 7.97579363e-02 -1.17264174e-01 9.07589495e-01
1.50872409e+00 -3.66594493e-01 2.63352782e-01 2.71819025e-01
5.33426523e-01 1.23014674e-01 1.00777395e-01 4.69135791e-01
2.21387833e-01 3.80632102e-01 6.56718984e-02 -3.05813942e-02
-1.12669103e-01 -1.33038446e-01 2.86896080e-01 7.93069601e-01
2.39998817e-01 -2.33010009e-01 -1.22851264e+00 3.48444700e-01
-1.85609603e+00 -8.81106675e-01 1.27898246e-01 2.18941140e+00
1.00848222e+00 1.44809410e-01 1.76492825e-01 -5.27464211e-01
3.33330393e-01 2.14845851e-01 -9.43748951e-01 -1.78432241e-01
-8.15157741e-02 3.34633142e-02 6.52495980e-01 5.76784492e-01
-8.75423014e-01 1.09395945e+00 6.54382992e+00 1.05426395e+00
-1.59245956e+00 5.61476886e-01 1.00515556e+00 -8.28535616e-01
-2.11694926e-01 -7.40606189e-02 -1.02402401e+00 2.18081996e-01
1.18591845e+00 -3.29073846e-01 8.98007572e-01 9.07038510e-01
1.11938357e-01 -1.44420326e-01 -1.19100237e+00 1.07828331e+00
-6.64982125e-02 -1.46192348e+00 -2.01119974e-01 8.26870129e-02
5.14052808e-01 1.24108291e+00 2.07908571e-01 5.51663458e-01
3.74090046e-01 -9.17229712e-01 5.38036406e-01 2.69194096e-01
1.16402936e+00 -3.01621586e-01 2.80398846e-01 3.31786931e-01
-8.89804602e-01 -1.49283692e-01 -5.28576732e-01 7.68028796e-02
-2.38582879e-01 3.97388726e-01 -4.69607264e-01 -2.79530525e-01
9.92476225e-01 3.90853643e-01 -9.13047791e-01 9.19842362e-01
3.41370314e-01 8.31375539e-01 -5.52797377e-01 -6.21345416e-02
3.26899320e-01 2.73364652e-02 9.68788043e-02 1.32363713e+00
9.42487940e-02 -1.19376674e-01 -4.58933227e-03 9.05552149e-01
-4.03614461e-01 1.87214851e-01 -4.13585782e-01 -4.61833030e-01
4.51959401e-01 1.35288858e+00 -6.14680052e-01 -2.41857499e-01
-6.56115890e-01 8.58150125e-01 8.29670727e-01 4.32739794e-01
-9.04254377e-01 -2.19928473e-01 7.81667590e-01 2.48415351e-01
-2.08047424e-02 -4.99346554e-01 -2.29670346e-01 -1.29790032e+00
-1.59376830e-01 -7.73220837e-01 3.62292737e-01 -7.99317539e-01
-1.51996231e+00 5.44864893e-01 1.68011770e-01 -6.18758202e-01
9.83814374e-02 -7.50066161e-01 -5.43087900e-01 7.21465528e-01
-1.30473685e+00 -1.05612040e+00 -3.81813109e-01 4.36657280e-01
5.48911929e-01 -2.53316432e-01 7.58744717e-01 4.69103575e-01
-8.27412844e-01 8.81067276e-01 5.76803923e-01 9.03459564e-02
1.02406478e+00 -9.05592978e-01 6.20088995e-01 5.42185068e-01
8.39317366e-02 7.72981286e-01 6.11499846e-01 -3.44200164e-01
-1.48865891e+00 -8.29226077e-01 4.96061593e-01 -5.68659902e-01
1.12869251e+00 -6.00212097e-01 -1.01104033e+00 7.37260103e-01
4.51296836e-01 2.49248341e-01 9.24083650e-01 5.68767428e-01
-8.14630091e-01 -1.93123132e-01 -6.66479766e-01 6.06187344e-01
1.26850259e+00 -7.13634253e-01 6.65636063e-02 7.66153574e-01
9.29781199e-01 -2.15193883e-01 -8.74616981e-01 1.97094768e-01
7.81556308e-01 -9.01854694e-01 7.56196976e-01 -8.75445127e-01
7.17410505e-01 4.03163791e-01 -2.61970311e-01 -1.06271398e+00
-6.15766287e-01 -2.88920969e-01 1.31854907e-01 1.01655519e+00
6.10229194e-01 -6.66981876e-01 6.80886924e-01 6.35667264e-01
3.62642668e-02 -1.09796631e+00 -6.95620179e-01 -7.33239174e-01
5.44329882e-01 -5.42822540e-01 2.87354916e-01 7.83899605e-01
-1.98461413e-01 6.79129362e-01 -6.02925979e-02 -9.67756733e-02
5.56099653e-01 2.09122039e-02 7.99559832e-01 -9.02748764e-01
-7.81063318e-01 -7.62833118e-01 1.79884642e-01 -9.77497160e-01
2.28908584e-01 -1.06266785e+00 -2.05663010e-01 -1.20269167e+00
6.33624732e-01 -7.07056880e-01 -2.50605911e-01 7.38991559e-01
-9.20201540e-02 4.60785538e-01 3.07570666e-01 5.98134875e-01
-8.69345427e-01 4.22513604e-01 9.38870251e-01 1.14436641e-01
-3.85601632e-02 -3.12849045e-01 -8.71343315e-01 5.37095129e-01
8.13747287e-01 -3.48861575e-01 -3.98766339e-01 -1.02618587e+00
4.51914907e-01 -2.68714279e-01 3.50400239e-01 -1.03958750e+00
2.63274163e-01 -1.17095187e-01 2.49135986e-01 -1.11165836e-01
3.34443897e-01 -4.45466757e-01 -3.03061813e-01 3.09324056e-01
-7.84507573e-01 2.91128963e-01 5.55705726e-01 2.50573754e-01
2.25582227e-01 -2.72335559e-01 1.00341690e+00 -1.62892386e-01
-5.01739502e-01 5.31192780e-01 -1.18459247e-01 4.44461823e-01
6.68355405e-01 4.01642442e-01 -8.02659154e-01 -4.49324638e-01
-5.18803239e-01 2.70329356e-01 6.79003835e-01 4.73930150e-01
7.77436197e-02 -1.02786517e+00 -7.32809603e-01 1.06682152e-01
3.51079792e-01 -1.79641575e-01 3.67306620e-01 9.92111623e-01
-6.13121212e-01 5.10464847e-01 4.99977358e-03 -6.19275212e-01
-1.23519278e+00 5.16825259e-01 4.16464508e-01 -2.37599388e-01
-3.18089932e-01 1.03080928e+00 2.21046135e-01 -4.51668531e-01
3.69632661e-01 -3.39864105e-01 5.29628158e-01 -1.59834042e-01
6.03106916e-01 1.69808388e-01 -3.73123176e-02 -2.47128308e-01
-1.41736194e-01 5.07265151e-01 -4.92294610e-01 -9.26083922e-02
1.31072164e+00 4.45912767e-04 -4.68087904e-02 5.69290459e-01
1.53961420e+00 -2.90494353e-01 -1.20870757e+00 -5.03770709e-01
-5.15154064e-01 -2.25210816e-01 2.67950803e-01 -6.75374210e-01
-9.12343085e-01 9.96284962e-01 7.59985149e-01 1.36813730e-01
9.60856259e-01 3.42976034e-01 8.05428386e-01 6.19618237e-01
1.66169167e-01 -1.06224370e+00 1.55638993e-01 7.64520586e-01
9.17233884e-01 -1.50066531e+00 8.64996687e-02 1.14717044e-01
-4.43326294e-01 7.21800327e-01 8.14025283e-01 -8.94472748e-03
8.04311275e-01 5.35384715e-01 1.92401111e-01 -1.22385569e-01
-1.36577857e+00 8.03510919e-02 -3.89091522e-02 1.13958761e-01
7.56295323e-01 -2.80017197e-01 -2.49343421e-02 3.86482298e-01
-2.12994948e-01 2.39672124e-01 1.10267736e-01 7.50545263e-01
-3.28768641e-01 -1.00364888e+00 -1.60326481e-01 7.67442763e-01
-4.23019558e-01 -6.63613558e-01 -1.85904667e-01 5.12249708e-01
-3.65666658e-01 4.92624491e-01 2.69918710e-01 -2.67659575e-01
1.43008366e-01 2.37184972e-01 5.71816862e-01 -5.99059343e-01
-4.93206173e-01 -1.27388770e-02 -1.17121011e-01 -6.78074718e-01
8.88861865e-02 -5.06331444e-01 -1.15271068e+00 -8.70265305e-01
-1.60885707e-01 -1.39256537e-01 8.04680109e-01 6.98667705e-01
9.13550735e-01 1.53395340e-01 1.58445761e-01 -1.00692022e+00
-9.60634053e-01 -1.04530084e+00 -2.64951527e-01 4.57582772e-01
1.84874594e-01 -2.01308250e-01 -4.13516432e-01 1.47952586e-02] | [9.406370162963867, 2.221642255783081] |
dd56b7a0-5f43-4503-9f7f-8093c22541ae | convergence-and-price-of-anarchy-guarantees | 2206.07642 | null | https://arxiv.org/abs/2206.07642v1 | https://arxiv.org/pdf/2206.07642v1.pdf | Convergence and Price of Anarchy Guarantees of the Softmax Policy Gradient in Markov Potential Games | We study the performance of policy gradient methods for the subclass of Markov games known as Markov potential games (MPGs), which extends the notion of normal-form potential games to the stateful setting and includes the important special case of the fully cooperative setting where the agents share an identical reward function. Our focus in this paper is to study the convergence of the policy gradient method for solving MPGs under softmax policy parameterization, both tabular and parameterized with general function approximators such as neural networks. We first show the asymptotic convergence of this method to a Nash equilibrium of MPGs for tabular softmax policies. Second, we derive the finite-time performance of the policy gradient in two settings: 1) using the log-barrier regularization, and 2) using the natural policy gradient under the best-response dynamics (NPG-BR). Finally, extending the notion of price of anarchy (POA) and smoothness in normal-form games, we introduce the POA for MPGs and provide a POA bound for NPG-BR. To our knowledge, this is the first POA bound for solving MPGs. To support our theoretical results, we empirically compare the convergence rates and POA of policy gradient variants for both tabular and neural softmax policies. | ['Thinh T. Doan', 'Qi Zhang', 'Dingyang Chen'] | 2022-06-15 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-2.49752522e-01 4.91936266e-01 -4.53725159e-01 9.92931649e-02
-7.57124722e-01 -7.22685277e-01 2.41436169e-01 -1.72312811e-01
-7.80491114e-01 1.15588164e+00 9.27353743e-03 -8.11042547e-01
-5.67956507e-01 -6.52802527e-01 -9.24093366e-01 -9.45209801e-01
-4.47334051e-01 5.08786738e-01 7.15989619e-02 -4.49425042e-01
-3.28770056e-02 3.41126591e-01 -8.17250192e-01 -3.47754836e-01
1.07413805e+00 1.11003745e+00 1.54953867e-01 8.13079953e-01
1.43160865e-01 8.06977510e-01 -3.86025637e-01 -4.19771791e-01
7.65828907e-01 -2.10236445e-01 -9.05299664e-01 -1.61378741e-01
2.11169809e-01 -4.83758152e-01 -3.96218300e-01 1.39012742e+00
6.16082072e-01 5.71216762e-01 3.49573225e-01 -1.54749787e+00
-4.05314445e-01 8.97225142e-01 -6.20990753e-01 2.77625054e-01
-6.67129830e-02 2.71332622e-01 1.28477609e+00 -1.35168713e-02
3.75167131e-01 1.64804053e+00 4.98876601e-01 1.00237334e+00
-1.32284272e+00 -5.18791199e-01 6.04956567e-01 -1.49629831e-01
-8.11563790e-01 -1.59391016e-01 6.47658017e-03 -3.27465683e-01
9.58030581e-01 5.90700470e-02 5.64459920e-01 1.02908790e+00
-9.02661830e-02 1.02426755e+00 1.23015070e+00 -2.67247945e-01
5.32106817e-01 -9.53086391e-02 1.83109134e-01 7.01637447e-01
2.77667344e-01 4.94628489e-01 -3.17886062e-02 -6.12155974e-01
9.92205799e-01 -2.58864701e-01 -3.49683493e-01 -2.89032727e-01
-6.16836190e-01 1.07454145e+00 4.98404056e-02 -1.74761638e-01
-7.87664533e-01 7.15226293e-01 1.57817021e-01 6.75388455e-01
5.01251698e-01 6.43275142e-01 -5.26374578e-01 -5.33389091e-01
-6.45013332e-01 4.18389618e-01 1.31930959e+00 8.88220906e-01
5.14589787e-01 4.06305104e-01 -4.90014076e-01 6.06037259e-01
2.30254069e-01 6.82065010e-01 1.56702518e-01 -1.52757406e+00
6.63265407e-01 -1.23072475e-01 8.72331798e-01 -3.06080014e-01
-5.34960806e-01 -3.73218149e-01 -3.59084308e-01 3.14158350e-01
9.04970527e-01 -8.43475461e-01 -7.20330596e-01 2.44897413e+00
8.90438929e-02 2.71145254e-01 1.78127959e-01 9.15141225e-01
1.68969631e-01 8.99788201e-01 -1.91936176e-02 -7.76658237e-01
8.91447186e-01 -6.46679223e-01 -4.57811534e-01 -1.93937197e-01
6.05054855e-01 -2.29567483e-01 1.11620927e+00 2.98616946e-01
-1.47106397e+00 2.40320429e-01 -5.59683681e-01 5.61251819e-01
2.33138561e-01 -4.72066909e-01 6.46156549e-01 6.93978190e-01
-1.45926297e+00 6.23280942e-01 -8.96083415e-01 -3.82909745e-01
1.52542964e-01 6.19323909e-01 2.30689272e-01 3.65104288e-01
-1.37527168e+00 9.25275266e-01 4.67468083e-01 1.69904958e-02
-1.32146525e+00 -7.52596438e-01 -4.80191439e-01 5.47786832e-01
1.29766345e+00 -4.79117751e-01 1.86121738e+00 -1.10418749e+00
-2.07920361e+00 3.30291241e-01 2.30396420e-01 -6.99572504e-01
7.35229850e-01 -1.05319314e-01 3.94727856e-01 -9.39983409e-03
-6.60335198e-02 3.37260604e-01 5.54598451e-01 -1.04566300e+00
-8.02769244e-01 -1.14712348e-04 8.21918666e-01 4.41934168e-01
-2.63045847e-01 -2.15368867e-02 6.44787550e-02 -1.96789727e-01
-6.38571262e-01 -1.03974819e+00 -7.13708758e-01 -3.53509575e-01
-2.76123792e-01 -5.25776267e-01 2.70240456e-01 -3.73959005e-01
9.84326541e-01 -1.72451711e+00 1.84703708e-01 4.37446594e-01
1.89434960e-02 1.35419846e-01 -2.84608454e-01 3.60697955e-01
2.14627534e-01 2.69827992e-01 -2.29210723e-02 -2.19580501e-01
5.00592172e-01 6.21244371e-01 -5.27955413e-01 4.66252327e-01
-3.28243256e-01 1.12283349e+00 -9.81202364e-01 -1.21523045e-01
-3.44389945e-01 -4.47196007e-01 -7.48316944e-01 6.38191551e-02
-5.46313465e-01 -1.50104344e-01 -6.75828099e-01 3.82835642e-02
4.95047122e-01 -3.11860383e-01 3.04408222e-01 6.56330168e-01
-1.88287914e-01 9.60869715e-02 -1.10916352e+00 1.08685279e+00
-4.36974794e-01 1.90713063e-01 7.44571865e-01 -1.30350292e+00
2.98414022e-01 4.04322267e-01 6.77315116e-01 -4.38272029e-01
2.45251685e-01 1.94029033e-01 -1.36397034e-01 -1.66181773e-01
5.93169332e-01 -4.75167602e-01 -1.29058868e-01 5.77197015e-01
6.22300953e-02 7.59168640e-02 3.57332021e-01 3.61250401e-01
7.39769757e-01 1.45128489e-01 2.53127933e-01 -6.76200211e-01
-1.59008652e-02 -1.82755157e-01 6.60059512e-01 1.45752215e+00
-3.73680860e-01 -1.84424266e-01 1.32860827e+00 2.30623797e-01
-9.66094792e-01 -9.63319242e-01 2.79972464e-01 1.50213182e+00
9.50976759e-02 -8.15458298e-02 -6.98715091e-01 -4.86172825e-01
3.09150040e-01 5.95860302e-01 -6.74792528e-01 -1.41630530e-01
-4.68662530e-01 -9.02160048e-01 4.84202743e-01 3.87674689e-01
6.08348668e-01 -9.01482999e-01 -3.72773528e-01 4.42472845e-01
8.82002059e-05 -1.10993052e+00 -8.04484665e-01 2.03292295e-01
-7.08312750e-01 -8.40353429e-01 -9.80879188e-01 -4.13962185e-01
2.40310997e-01 -6.87489733e-02 7.50089526e-01 -4.27832246e-01
6.26197100e-01 7.88753092e-01 -2.53756978e-02 -2.59296387e-01
-4.40535843e-01 1.49195313e-01 2.02092215e-01 8.62516686e-02
-3.13852996e-01 -4.17435318e-01 -5.23818016e-01 3.38772774e-01
-6.29859865e-01 -2.66958377e-03 2.31122911e-01 7.20476925e-01
3.58737528e-01 -2.58886665e-01 4.94958133e-01 -5.61538339e-01
1.34286571e+00 -6.72725260e-01 -1.39798796e+00 4.14201200e-01
-7.19831645e-01 3.92173976e-01 8.32052231e-01 -7.80930161e-01
-1.10052228e+00 -1.47582710e-01 1.64430410e-01 -4.28931892e-01
4.63309675e-01 6.81453407e-01 2.41967246e-01 -1.05355613e-01
6.03666127e-01 1.17254727e-01 1.76345587e-01 -3.52335781e-01
4.05746311e-01 4.28626359e-01 3.99896175e-01 -1.31830621e+00
5.73061109e-01 4.27933514e-01 2.49784023e-01 -5.91864109e-01
-8.87269378e-01 -2.65916079e-01 3.65252554e-01 -2.35041589e-01
6.61582708e-01 -4.27000731e-01 -1.73101687e+00 3.87260437e-01
-8.94878209e-01 -1.13468421e+00 -6.25442624e-01 5.92439532e-01
-1.26321769e+00 5.13894200e-01 -8.36261868e-01 -1.42545164e+00
-1.69880122e-01 -9.63966489e-01 4.23455387e-01 5.45290649e-01
6.19499743e-01 -1.11694825e+00 2.61484414e-01 -2.01969802e-01
5.32645106e-01 7.34377373e-03 6.93841338e-01 -4.19697762e-01
-3.59851599e-01 1.98338613e-01 -1.09731883e-01 2.92197078e-01
-3.33542168e-01 -3.40886623e-01 -5.04668832e-01 -6.08681500e-01
-8.74357522e-02 -5.41483402e-01 9.06949222e-01 1.11564398e+00
7.58418620e-01 -8.87339056e-01 -9.10807997e-02 5.76722980e-01
1.39377487e+00 4.62361276e-01 3.56055290e-01 4.08655912e-01
2.50960827e-01 3.29031706e-01 4.12629932e-01 8.19750547e-01
2.37115294e-01 5.58942378e-01 7.50340760e-01 1.48206994e-01
8.19534481e-01 -3.29446912e-01 7.95088887e-01 1.48729742e-01
-5.49311697e-01 -3.33724976e-01 -5.01321554e-01 2.79825777e-01
-2.57500887e+00 -1.16629994e+00 3.28132689e-01 2.50036097e+00
1.09052539e+00 7.42288008e-02 7.17523396e-01 -6.69691145e-01
8.08516979e-01 -2.99535096e-02 -8.07118833e-01 -9.20458257e-01
-1.38347924e-01 3.47731024e-01 1.28285170e+00 9.58525717e-01
-8.20282102e-01 1.24568057e+00 6.66193008e+00 1.11978388e+00
-9.48442161e-01 2.97695100e-01 5.91836452e-01 -4.77078438e-01
2.58374438e-02 8.70816484e-02 -7.13437200e-01 4.19185251e-01
1.03653264e+00 -6.25352740e-01 1.18124151e+00 1.07995486e+00
7.09990501e-01 -6.29878119e-02 -6.47740781e-01 5.71838677e-01
-7.77478993e-01 -1.14832675e+00 -6.23842537e-01 5.48535168e-01
9.09011126e-01 1.54322371e-01 8.89963135e-02 6.18832290e-01
1.31283903e+00 -8.94963801e-01 7.44941533e-01 2.31399953e-01
6.40322983e-01 -8.08639288e-01 3.91111046e-01 4.06959474e-01
-9.09311056e-01 -5.48820734e-01 -5.39250910e-01 -2.19050720e-01
3.64194661e-01 -1.00207239e-01 -4.18562084e-01 4.02263701e-01
4.03839618e-01 2.15625063e-01 3.16124260e-01 1.22397554e+00
-1.24716125e-01 9.38028991e-01 -7.50237346e-01 -2.38446951e-01
9.74664330e-01 -4.59542602e-01 7.89837182e-01 9.07731175e-01
8.99249613e-02 9.96990800e-02 5.27006090e-01 8.59566391e-01
-9.64093804e-02 5.19840047e-02 -4.35248911e-02 -3.50550652e-01
9.03848261e-02 1.02503192e+00 -4.65459198e-01 -1.38640746e-01
-1.97618008e-01 6.04394913e-01 4.50069815e-01 7.66674221e-01
-7.63132572e-01 -6.89672157e-02 1.00986862e+00 -1.89055637e-01
3.69861782e-01 -2.13414446e-01 2.17418730e-01 -1.09390473e+00
-7.65358657e-02 -5.92482984e-01 6.35382473e-01 -5.16890049e-01
-1.35203230e+00 1.54380873e-02 4.66620326e-01 -5.16435027e-01
-5.88840127e-01 -7.28765965e-01 -7.16974020e-01 8.85469973e-01
-1.34729290e+00 -4.82253641e-01 4.57625777e-01 7.42584705e-01
1.28836393e-01 7.14753792e-02 3.20421606e-01 -4.30641249e-02
-5.74278474e-01 6.48048639e-01 7.02423453e-01 -2.37937689e-01
1.08253136e-01 -1.68046856e+00 5.35989832e-03 8.09193790e-01
-5.64792097e-01 3.26393783e-01 8.04980397e-01 -4.39510524e-01
-1.33725166e+00 -6.29327834e-01 -1.77040789e-02 3.42475891e-01
1.16483855e+00 -2.61154294e-01 -5.94261289e-01 7.92209685e-01
4.60290462e-02 -1.94448262e-01 -1.62925929e-01 1.96924582e-01
1.57964304e-01 7.43819401e-02 -1.00044250e+00 9.28140938e-01
1.05548203e+00 -1.09200716e-01 -2.69357227e-02 4.67596233e-01
8.57643962e-01 -4.83796746e-01 -6.88824832e-01 -1.15470119e-01
3.61078143e-01 -3.90673816e-01 6.49519086e-01 -1.25259125e+00
-1.85650438e-02 2.67029911e-01 -4.96658795e-02 -1.64908183e+00
-3.60451967e-01 -1.74879611e+00 -2.48121679e-01 5.84507227e-01
4.85020280e-01 -1.14632034e+00 9.77796495e-01 7.80461073e-01
-1.52984589e-01 -8.59647274e-01 -1.39386010e+00 -1.30017209e+00
7.46032119e-01 -3.51836205e-01 2.30006918e-01 6.72417045e-01
2.59399861e-01 -4.02185656e-02 -8.50532770e-01 -1.90770086e-02
6.34014666e-01 -3.40066403e-02 3.60688657e-01 -8.17003667e-01
-9.71634090e-01 -9.70921576e-01 1.14433616e-01 -1.50772333e+00
4.31241900e-01 -7.73089290e-01 1.08168252e-01 -1.45767164e+00
3.40884387e-01 -5.26992440e-01 -3.03270429e-01 6.27418399e-01
-8.36672559e-02 -4.34587121e-01 5.91496766e-01 6.80293664e-02
-7.33560205e-01 5.54913700e-01 1.46866226e+00 5.45872077e-02
-5.15038490e-01 5.61660767e-01 -8.23904455e-01 3.23576927e-01
9.58016098e-01 -4.58885372e-01 -5.75914323e-01 7.84794539e-02
5.47011375e-01 5.28624237e-01 1.98644266e-01 -1.96389273e-01
6.54229447e-02 -1.08327591e+00 -6.42550707e-01 2.23725364e-02
1.93837658e-01 -3.86799574e-01 -2.73862630e-01 7.13509440e-01
-5.86625814e-01 -7.85378590e-02 1.14871703e-01 6.07453227e-01
4.72883135e-01 -4.53699797e-01 9.13343012e-01 -1.41661897e-01
-4.37902898e-01 6.68387055e-01 -7.27459431e-01 6.77704930e-01
8.82421672e-01 -7.70652294e-02 -5.58036685e-01 -1.18066311e+00
-9.41075027e-01 7.91877866e-01 2.68505681e-02 -1.80946797e-01
2.97761202e-01 -1.00793314e+00 -6.41130388e-01 -5.45698404e-01
-5.31601191e-01 -4.21801925e-01 1.58607680e-02 1.06424475e+00
-9.30239260e-02 4.28997278e-01 -4.21851762e-02 -3.52086648e-02
-8.77965629e-01 5.19057155e-01 9.75891113e-01 -6.90830529e-01
-2.83605069e-01 6.16630316e-01 4.25121546e-01 -1.16680160e-01
3.96304935e-01 -4.31144536e-01 1.44726440e-01 -2.84255713e-01
1.57000065e-01 5.30292809e-01 -4.94271368e-01 -8.31098855e-02
8.52471218e-02 -1.46597758e-01 -6.34172233e-03 -7.80478477e-01
1.19637144e+00 3.84266078e-02 9.60302949e-02 -1.44678140e-02
7.23496258e-01 -3.45861942e-01 -1.84707785e+00 -3.42778355e-01
-8.72111246e-02 -1.14515379e-01 7.55693614e-02 -7.28233993e-01
-1.23233759e+00 4.48977649e-01 4.60580081e-01 6.65236950e-01
7.63661683e-01 -5.55224940e-02 4.87606496e-01 6.39745951e-01
4.47986424e-01 -1.36582434e+00 -1.50528476e-01 8.37922037e-01
5.28571844e-01 -7.41756022e-01 -3.32755059e-01 1.35748297e-01
-7.20740497e-01 8.24197769e-01 4.65233654e-01 -2.25494087e-01
6.24309838e-01 2.48568967e-01 -3.48043799e-01 1.42783716e-01
-9.79306936e-01 -6.79964483e-01 -1.45278007e-01 5.24441242e-01
-1.75588608e-01 3.01970094e-01 -6.53320312e-01 6.34914577e-01
-1.12032384e-01 1.64679333e-03 8.04515064e-01 9.54038382e-01
-6.03883088e-01 -9.95476604e-01 -1.35428324e-01 4.55573231e-01
-8.41881633e-01 -1.86822727e-01 -2.84514338e-01 7.25610018e-01
-8.28605592e-01 1.03436744e+00 -8.29842389e-02 1.02685221e-01
2.30093852e-01 -1.30953103e-01 7.07606614e-01 -3.29387128e-01
-6.66013718e-01 1.84965655e-01 1.11013353e-01 -5.53731859e-01
-2.83940136e-01 -2.43559599e-01 -1.00914919e+00 -7.60902166e-01
-4.59942609e-01 3.98997992e-01 3.97046238e-01 1.03228366e+00
4.19449918e-02 -4.20327485e-02 5.98499656e-01 -7.06993341e-01
-1.50750446e+00 -8.57670724e-01 -8.40496302e-01 1.76553965e-01
2.20969513e-01 -7.06906438e-01 -4.48385805e-01 -7.22771466e-01] | [4.222933769226074, 2.6328647136688232] |
44dbfced-f434-4ed9-ad3f-c1cd4bc1d424 | weakly-and-self-supervised-learning-for | 1708.02731 | null | http://arxiv.org/abs/1708.02731v1 | http://arxiv.org/pdf/1708.02731v1.pdf | Weakly- and Self-Supervised Learning for Content-Aware Deep Image Retargeting | This paper proposes a weakly- and self-supervised deep convolutional neural
network (WSSDCNN) for content-aware image retargeting. Our network takes a
source image and a target aspect ratio, and then directly outputs a retargeted
image. Retargeting is performed through a shift map, which is a pixel-wise
mapping from the source to the target grid. Our method implicitly learns an
attention map, which leads to a content-aware shift map for image retargeting.
As a result, discriminative parts in an image are preserved, while background
regions are adjusted seamlessly. In the training phase, pairs of an image and
its image-level annotation are used to compute content and structure losses. We
demonstrate the effectiveness of our proposed method for a retargeting
application with insightful analyses. | ['Yu-Wing Tai', 'Tae-Hyun Oh', 'Jinsun Park', 'In So Kweon', 'Donghyeon Cho'] | 2017-08-09 | weakly-and-self-supervised-learning-for-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Cho_Weakly-_and_Self-Supervised_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Cho_Weakly-_and_Self-Supervised_ICCV_2017_paper.pdf | iccv-2017-10 | ['image-retargeting'] | ['computer-vision'] | [ 6.76483631e-01 3.05510581e-01 -2.12562278e-01 -3.99457455e-01
-3.97113532e-01 -5.40908277e-01 3.51004928e-01 9.18337330e-02
-5.24395466e-01 3.77831787e-01 1.53457865e-01 5.54719418e-02
3.03295314e-01 -8.87830019e-01 -9.76223707e-01 -7.08703041e-01
5.69618940e-01 -2.03872502e-01 5.54449439e-01 -2.23488271e-01
3.67513597e-01 6.63520455e-01 -1.12596071e+00 1.38990298e-01
6.62472069e-01 8.41614425e-01 4.51121718e-01 6.39747500e-01
-3.22105214e-02 5.31261086e-01 -3.55981320e-01 -1.90089151e-01
4.68044937e-01 -4.47607279e-01 -8.16614270e-01 3.85724425e-01
7.50826538e-01 -3.66899341e-01 -5.97909987e-01 1.23397279e+00
4.61670905e-01 2.55343407e-01 5.03394425e-01 -9.57965732e-01
-1.02616787e+00 6.17043674e-01 -1.00464725e+00 3.11817706e-01
-1.80745810e-01 4.60912377e-01 7.81241357e-01 -8.63083839e-01
6.15098357e-01 9.39645708e-01 4.54404682e-01 2.46006146e-01
-1.58064830e+00 -7.46066928e-01 3.26148719e-01 6.29819930e-02
-1.31737256e+00 -2.91669905e-01 1.10618317e+00 -3.40701640e-01
5.05782604e-01 2.64129758e-01 7.88209140e-01 5.69958031e-01
4.84334260e-01 5.88399231e-01 7.35062838e-01 -4.54170585e-01
2.32606724e-01 3.57679650e-02 -4.50648904e-01 6.45426035e-01
1.88103095e-01 6.31224588e-02 -5.17655075e-01 3.51616442e-01
1.07633770e+00 1.21369012e-01 -2.90778935e-01 -7.35589862e-01
-1.37265038e+00 5.16961336e-01 9.97276127e-01 2.76192188e-01
-2.77789801e-01 1.82594612e-01 2.95064032e-01 7.41437310e-03
6.64707005e-01 6.19240642e-01 -4.46984321e-01 3.94349366e-01
-8.44926715e-01 1.08268380e-01 3.57332975e-02 7.42066622e-01
1.19465840e+00 1.38412520e-01 -7.54641712e-01 7.95634687e-01
-2.51774210e-02 4.31005478e-01 6.46519423e-01 -9.57890391e-01
1.19599320e-01 7.00851917e-01 1.05267942e-01 -1.34026933e+00
-3.37768376e-01 -6.69187844e-01 -8.77880394e-01 5.37571073e-01
8.23597237e-02 -4.14478481e-02 -1.01018953e+00 1.67904758e+00
5.93276918e-01 1.11247107e-01 -1.79817766e-01 1.09562349e+00
7.47775376e-01 3.39851797e-01 2.83305526e-01 -2.66953297e-02
1.12469459e+00 -1.32951725e+00 -4.92323667e-01 -3.25672209e-01
3.76291752e-01 -8.12583208e-01 1.36963356e+00 -5.86751103e-02
-1.25897372e+00 -6.93001270e-01 -9.76878643e-01 -3.61373395e-01
-2.75042146e-01 1.45857796e-01 1.66383803e-01 3.65463585e-01
-1.13679242e+00 5.27206481e-01 -4.76923138e-01 -3.54191840e-01
4.47942227e-01 2.39898771e-01 -3.91365796e-01 4.18032482e-02
-9.06257331e-01 6.00147426e-01 5.72995961e-01 -8.22850987e-02
-8.50843370e-01 -9.88551259e-01 -1.00740898e+00 2.68649697e-01
2.68831104e-01 -5.76378226e-01 1.08526886e+00 -1.60225940e+00
-1.60778475e+00 1.15527570e+00 1.26716435e-01 -2.65265852e-01
4.15132284e-01 8.14124942e-02 -3.62370729e-01 1.58860832e-01
4.27354753e-01 9.78513956e-01 1.26930928e+00 -1.32702410e+00
-8.39413345e-01 -1.02348372e-01 4.72298004e-02 3.72844160e-01
-4.21657711e-01 -2.24057272e-01 -7.30387628e-01 -1.31437540e+00
1.49218351e-01 -7.01226592e-01 -2.44160950e-01 3.98338675e-01
-7.39139855e-01 5.43246508e-01 8.63765359e-01 -7.95248568e-01
1.14882123e+00 -2.31736088e+00 2.66330361e-01 9.98865217e-02
3.99268299e-01 1.57367632e-01 -6.36159182e-01 -5.60923778e-02
-4.31991100e-01 4.41140272e-02 -5.45311809e-01 -1.67553663e-01
-4.98782635e-01 -1.81871459e-01 -3.00578386e-01 6.26992226e-01
2.11915806e-01 1.21541786e+00 -1.05387902e+00 -4.52359736e-01
4.97503936e-01 3.46873552e-01 -7.07311511e-01 3.06495905e-01
-3.00039381e-01 7.21208930e-01 -1.17235489e-01 2.95423925e-01
8.83600116e-01 1.39633729e-03 -4.84811813e-02 -7.68004835e-01
-3.31678748e-01 -3.61079156e-01 -7.29890525e-01 1.77561462e+00
-5.82405031e-01 7.68783689e-01 -2.56482393e-01 -9.10847723e-01
9.31901693e-01 -3.19279730e-01 5.18824100e-01 -9.89101708e-01
1.74750626e-01 -1.37631088e-01 -2.26177871e-01 -4.34926480e-01
6.47070527e-01 1.97794452e-01 3.78706045e-02 6.01879239e-01
-1.82928622e-01 -9.84779596e-02 -2.54010469e-01 1.40792921e-01
7.43957162e-01 2.98652083e-01 3.56583476e-01 -5.02676785e-01
4.88984555e-01 -1.53836250e-01 4.69416410e-01 5.24206281e-01
-7.03344420e-02 1.05655861e+00 3.19930583e-01 -7.54811049e-01
-1.29304969e+00 -1.04708266e+00 7.75430053e-02 1.28153121e+00
5.89823425e-01 -3.19364332e-02 -8.49234521e-01 -7.88882732e-01
-1.47415251e-01 7.40131974e-01 -9.27532732e-01 -6.67957664e-01
-6.27102256e-01 -3.87221247e-01 1.03832118e-01 4.60485876e-01
8.78031373e-01 -1.10572779e+00 -5.72319031e-01 2.60420322e-01
-1.28150091e-01 -7.43521690e-01 -1.29710853e+00 1.00791119e-01
-7.56865740e-01 -9.30533051e-01 -9.87641692e-01 -9.99365628e-01
1.14620233e+00 6.47831798e-01 8.97354007e-01 1.24692872e-01
-3.43168944e-01 3.25380601e-02 -2.43783206e-01 1.26553535e-01
-3.95392895e-01 2.22752661e-01 -2.58674026e-01 4.24044967e-01
-2.32816517e-01 -6.41014338e-01 -9.86539125e-01 4.73839164e-01
-1.29784918e+00 3.95914823e-01 5.91366947e-01 8.45628977e-01
9.40082252e-01 9.84251648e-02 3.54798675e-01 -9.72197294e-01
2.21978679e-01 -2.20932782e-01 -6.48270607e-01 2.42091477e-01
-5.33944786e-01 1.05698280e-01 4.86697048e-01 -5.93584836e-01
-1.19885314e+00 3.36514115e-01 9.11117196e-02 -4.21472281e-01
2.19191285e-03 8.07606056e-02 -4.86792833e-01 -4.32764858e-01
8.04540932e-01 3.19384307e-01 -1.47018135e-01 -2.65602797e-01
8.54576230e-01 3.99235696e-01 9.19443846e-01 -1.85631543e-01
9.65013802e-01 6.49389684e-01 -2.61891484e-01 -3.69938344e-01
-9.61831629e-01 -3.85387018e-02 -9.38322663e-01 -1.96285784e-01
9.04227078e-01 -6.93340361e-01 -3.51599127e-01 5.48635900e-01
-9.96982217e-01 -7.88983226e-01 -8.19686174e-01 9.00696144e-02
-4.22054976e-01 2.64977366e-01 -2.87675291e-01 3.32381316e-02
-3.46449405e-01 -9.82330620e-01 1.14120340e+00 4.76808965e-01
-3.60944718e-02 -8.57761085e-01 9.50910747e-02 1.76575650e-02
5.36892056e-01 2.21438438e-01 1.03794837e+00 -3.08831811e-01
-5.90252936e-01 5.31313056e-03 -7.13151217e-01 2.00843304e-01
5.82504988e-01 -1.17499948e-01 -8.93261492e-01 -4.80265558e-01
-4.91069794e-01 -2.83173118e-02 8.02113891e-01 5.46469867e-01
1.62148178e+00 -5.84120870e-01 -4.09267604e-01 1.10245836e+00
1.18352854e+00 1.19629107e-01 8.21075320e-01 5.42512894e-01
1.01766539e+00 3.11076224e-01 4.90478873e-01 1.91860154e-01
2.61986166e-01 8.18418741e-01 6.45606339e-01 -7.65284240e-01
-6.73202276e-01 -3.58398080e-01 -1.07351653e-01 2.23495245e-01
3.98432434e-01 -2.83167720e-01 -4.18468207e-01 6.59737170e-01
-1.66188145e+00 -6.34653091e-01 2.74183214e-01 2.37156820e+00
9.94750977e-01 -3.02092358e-02 -9.94957238e-02 -3.50854874e-01
1.06431878e+00 2.04725727e-01 -1.16249549e+00 -2.16752794e-02
-1.92482069e-01 4.24931124e-02 7.61579871e-01 6.60485685e-01
-1.20783031e+00 1.25362194e+00 6.35019159e+00 8.90153587e-01
-1.52935982e+00 1.20078065e-01 1.05197763e+00 -1.29072770e-01
-6.07502520e-01 -1.47062272e-01 -2.82064378e-01 5.64404190e-01
1.61270261e-01 -2.21090689e-01 5.66768408e-01 7.71163285e-01
2.03098804e-01 -2.03484893e-02 -8.04264247e-01 9.76640463e-01
3.08347940e-01 -1.49979436e+00 2.70282298e-01 -2.51678884e-01
1.11213541e+00 -2.84297884e-01 3.44521970e-01 -1.79620460e-02
2.51409084e-01 -8.62114906e-01 1.00874698e+00 4.16378945e-01
1.24848199e+00 -8.11840355e-01 4.46078092e-01 -3.24700564e-01
-1.20731366e+00 -4.94532995e-02 -4.36916083e-01 3.46879601e-01
8.83559603e-03 6.19817436e-01 -5.38444221e-01 2.43161738e-01
8.08171690e-01 8.17051053e-01 -8.75509977e-01 1.00042820e+00
-3.04781407e-01 1.20344728e-01 1.34835914e-01 5.70633173e-01
-1.36627361e-01 -1.98758796e-01 5.85764468e-01 1.03995037e+00
2.03921854e-01 -9.95937213e-02 1.25393420e-02 1.08664036e+00
-3.57215434e-01 1.55874014e-01 -4.40013796e-01 3.83534789e-01
3.80637109e-01 1.46271348e+00 -1.04529595e+00 -2.56895423e-01
-8.90643001e-02 1.60605490e+00 4.66325343e-01 6.56089485e-01
-7.05045164e-01 -5.08895159e-01 6.32274330e-01 3.14357758e-01
4.29633439e-01 2.87015975e-01 -3.71193677e-01 -1.09527469e+00
-1.66825980e-01 -5.40704012e-01 2.08394602e-01 -1.17890060e+00
-1.13858294e+00 7.28578329e-01 -1.34165838e-01 -1.30099344e+00
3.28721493e-01 -1.42088458e-01 -7.49695897e-01 7.68734396e-01
-1.40530300e+00 -1.35383248e+00 -7.52424777e-01 3.58933270e-01
5.92483103e-01 -2.82371283e-01 3.50847989e-01 2.38830730e-01
-5.20687103e-01 7.62918353e-01 -7.60618644e-03 5.19324206e-02
9.40418780e-01 -1.05831122e+00 7.22608864e-01 1.08570600e+00
-7.43964538e-02 3.02829891e-01 7.53881991e-01 -6.60338402e-01
-8.17390323e-01 -1.72244191e+00 3.38469118e-01 -2.93024480e-01
3.29217345e-01 -1.10130586e-01 -9.65717793e-01 7.12843359e-01
3.12344640e-01 2.67300010e-01 9.19865295e-02 -6.29239082e-01
-3.86311114e-01 -2.84476578e-01 -1.24938142e+00 9.66083646e-01
1.10082865e+00 -3.90098691e-01 -1.16224684e-01 2.46892303e-01
1.25739253e+00 -7.61698961e-01 -5.18231928e-01 1.97198570e-01
6.04971647e-01 -9.79064286e-01 1.08438492e+00 -3.83114845e-01
4.35348332e-01 -6.08458638e-01 5.23805842e-02 -1.58629835e+00
-7.95621753e-01 -4.44920987e-01 2.88018078e-01 1.05676937e+00
1.14831552e-01 -5.21560311e-01 6.57120287e-01 4.04746115e-01
-1.62994683e-01 -5.02866149e-01 -8.17472517e-01 -2.69434065e-01
3.65237147e-02 -4.65249866e-02 8.17051649e-01 9.12837803e-01
-5.13800323e-01 1.19942918e-01 -3.33032846e-01 2.34247029e-01
5.77441752e-01 1.62649319e-01 7.28708804e-01 -6.67235374e-01
-9.20121446e-02 -5.56703210e-01 -3.94565672e-01 -9.73569155e-01
1.42458826e-01 -7.95225263e-01 4.88304980e-02 -1.23312342e+00
4.38950598e-01 -3.91676098e-01 -2.46246427e-01 7.22999871e-01
-3.08922291e-01 9.19995964e-01 -1.10734608e-02 1.24945201e-01
-3.07174385e-01 6.94369435e-01 1.44804585e+00 -4.94302660e-01
-4.17780071e-01 -2.53859162e-01 -9.70386863e-01 5.05113184e-01
9.69121039e-01 -3.98250252e-01 -3.92023683e-01 -5.79854906e-01
1.64553940e-01 -4.45856005e-01 6.08608961e-01 -7.34365642e-01
-3.06425728e-02 -3.61957073e-01 7.18576610e-01 -2.96593040e-01
-1.18954964e-01 -7.34377205e-01 5.14490083e-02 3.91214520e-01
-6.25147462e-01 -1.05504014e-01 3.84871811e-01 5.68061829e-01
8.03904980e-02 -6.33791760e-02 1.20413971e+00 1.93712398e-01
-7.84458458e-01 5.58661699e-01 -1.78226642e-02 -1.84799314e-01
1.16924345e+00 -2.39925429e-01 -1.74630567e-01 -2.97952086e-01
-5.95037520e-01 -1.25599653e-01 8.96454692e-01 5.37560821e-01
6.36089385e-01 -1.50521445e+00 -5.60846090e-01 4.83556896e-01
3.24572086e-01 1.75734833e-01 6.80318713e-01 6.53697431e-01
-5.73860407e-01 -9.69680920e-02 -5.76411009e-01 -5.04668534e-01
-9.94285047e-01 8.60057175e-01 7.33101189e-01 1.50488511e-01
-9.50453341e-01 7.70782650e-01 9.26069498e-01 -4.27773774e-01
4.05739099e-02 -3.79658729e-01 -5.80009408e-02 -3.93125623e-01
6.35318398e-01 1.81294009e-02 -7.34544965e-03 -7.03475952e-01
-1.20294541e-02 7.59967864e-01 -1.27926245e-01 -8.20824951e-02
1.03384852e+00 -5.30881286e-01 -1.35539576e-01 9.23895165e-02
1.25121999e+00 8.50432515e-02 -1.78449416e+00 -5.09924114e-01
-5.26418090e-01 -7.97966957e-01 2.96633989e-01 -7.72551596e-01
-1.68022561e+00 4.62680399e-01 9.80606079e-01 -2.29889527e-01
1.33998883e+00 2.26493720e-02 7.23303974e-01 -5.74150532e-02
-7.46110603e-02 -8.75575006e-01 5.98360419e-01 8.09340775e-02
1.11295879e+00 -1.13803017e+00 -2.93360464e-03 -2.13230044e-01
-6.21196210e-01 8.10741901e-01 9.28377569e-01 2.65041329e-02
4.20462161e-01 1.18086271e-01 9.58032608e-02 -9.01174694e-02
-1.07048593e-01 -1.25630032e-02 3.69471431e-01 8.71001065e-01
6.12629205e-02 -7.86286294e-02 2.41803408e-01 2.55274475e-01
1.18596470e-02 -1.59856379e-01 4.16215807e-01 4.91872430e-01
-3.99196118e-01 -7.78867841e-01 -4.63033020e-01 2.97721535e-01
5.38346693e-02 -1.77368402e-01 -1.86790332e-01 4.77975160e-01
3.11531454e-01 4.78012830e-01 3.37925732e-01 -4.26074564e-01
3.73246133e-01 -6.33614242e-01 2.75589883e-01 -7.53713906e-01
-4.00059640e-01 -7.48181194e-02 -5.93833148e-01 -7.02990830e-01
-1.64307877e-01 -1.85977757e-01 -1.17300928e+00 -1.18507035e-01
-1.15290284e-01 -3.09978634e-01 3.59746665e-01 6.07042909e-01
6.06080174e-01 9.53556061e-01 7.80109525e-01 -1.12239373e+00
1.46475375e-01 -7.37708390e-01 -5.47812104e-01 4.18684900e-01
4.15553868e-01 -4.91241097e-01 -1.04127258e-01 5.47222435e-01] | [11.285279273986816, -1.001785397529602] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.