Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files Community
1
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
9a9742ab-45c1-4b00-a493-baece26f0253
node-embedding-from-neural-hamiltonian-orbits
2305.18965
null
https://arxiv.org/abs/2305.18965v1
https://arxiv.org/pdf/2305.18965v1.pdf
Node Embedding from Neural Hamiltonian Orbits in Graph Neural Networks
In the graph node embedding problem, embedding spaces can vary significantly for different data types, leading to the need for different GNN model types. In this paper, we model the embedding update of a node feature as a Hamiltonian orbit over time. Since the Hamiltonian orbits generalize the exponential maps, this approach allows us to learn the underlying manifold of the graph in training, in contrast to most of the existing literature that assumes a fixed graph embedding manifold with a closed exponential map solution. Our proposed node embedding strategy can automatically learn, without extensive tuning, the underlying geometry of any given graph dataset even if it has diverse geometries. We test Hamiltonian functions of different forms and verify the performance of our approach on two graph node embedding downstream tasks: node classification and link prediction. Numerical experiments demonstrate that our approach adapts better to different types of graph datasets than popular state-of-the-art graph node embedding GNNs. The code is available at \url{https://github.com/zknus/Hamiltonian-GNN}.
['Wee Peng Tay', 'Sijie Wang', 'Yang song', 'Kai Zhao', 'Qiyu Kang']
2023-05-30
null
null
null
null
['graph-embedding', 'link-prediction']
['graphs', 'graphs']
[-4.27913696e-01 5.87152839e-01 -2.56221384e-01 5.51310778e-02 2.66147666e-02 -8.08930337e-01 5.05704880e-01 3.03213924e-01 -1.84468683e-02 4.35121000e-01 -1.22540362e-01 -5.54471314e-01 -1.95766181e-01 -1.15178740e+00 -6.24576390e-01 -8.06684434e-01 -4.41063434e-01 5.96797645e-01 2.59378731e-01 -4.68924046e-01 9.43306908e-02 6.11626565e-01 -9.23162103e-01 -6.97940886e-01 6.15456641e-01 4.72596884e-01 -8.32769573e-02 9.31544125e-01 1.14987694e-01 1.07253172e-01 -6.05715439e-03 -4.93856907e-01 4.96879369e-01 -2.17342898e-01 -8.60065699e-01 -1.94669545e-01 2.70152986e-01 3.54675390e-02 -1.11729479e+00 1.00350380e+00 3.29079986e-01 -3.02779805e-02 8.28514457e-01 -1.55642211e+00 -9.83124197e-01 5.72929561e-01 -1.36856556e-01 1.83387980e-01 2.27169767e-01 -1.39166296e-01 1.51632977e+00 -7.36252189e-01 8.52277935e-01 1.04215288e+00 8.86815548e-01 4.67052490e-01 -1.73731756e+00 -2.60291874e-01 -1.67839617e-01 2.25990623e-01 -1.55446923e+00 1.04829513e-01 1.13806367e+00 -5.89096010e-01 7.38502264e-01 2.26421267e-01 9.91507888e-01 8.79076481e-01 4.64079142e-01 3.08292687e-01 5.54605782e-01 -4.02445138e-01 -1.74193885e-02 1.56025961e-01 3.40258181e-01 1.04047406e+00 5.53099990e-01 -4.68379781e-02 -4.34281044e-02 -3.09172392e-01 7.01152086e-01 9.95151699e-02 -4.02108341e-01 -9.81269598e-01 -1.03950942e+00 1.22602093e+00 8.05798054e-01 4.27026719e-01 5.79309687e-02 5.28854251e-01 2.77966529e-01 6.93158686e-01 5.00615954e-01 3.18137735e-01 -2.57052660e-01 2.04357117e-01 -3.44243586e-01 1.94067851e-01 1.28445148e+00 9.58815157e-01 1.11814368e+00 -3.12962681e-01 3.89328659e-01 3.62128913e-01 4.21117008e-01 3.36078465e-01 4.76537533e-02 -6.50838971e-01 1.84908047e-01 7.40943551e-01 -3.05524290e-01 -1.41434050e+00 -6.03532493e-01 -3.09349567e-01 -9.23868358e-01 -9.50359330e-02 4.07572567e-01 3.31866555e-02 -5.20829797e-01 1.88054717e+00 5.13925254e-01 2.55416185e-01 -1.07557282e-01 5.81410825e-01 6.84796989e-01 5.76367199e-01 -3.05408508e-01 2.53243446e-01 1.00916862e+00 -8.20335150e-01 -4.50459301e-01 1.11377172e-01 1.13806367e+00 -3.84250522e-01 8.39628398e-01 -2.08426669e-01 -7.13194191e-01 -1.84156254e-01 -1.04000366e+00 -2.80111223e-01 -6.54946923e-01 -2.53197908e-01 9.20507669e-01 6.30705714e-01 -1.53965080e+00 9.88352180e-01 -8.94116044e-01 -7.20809817e-01 9.00343657e-02 5.14887333e-01 -4.62067872e-01 -2.15034969e-02 -1.33001423e+00 7.83840954e-01 4.02175844e-01 -7.18608592e-03 -5.03085256e-01 -8.81560862e-01 -1.02600074e+00 1.31082401e-01 2.13600993e-01 -7.51474202e-01 7.64316797e-01 -4.44813937e-01 -1.10275340e+00 7.20244646e-01 2.24021897e-01 -3.75812531e-01 4.29024100e-01 3.83063763e-01 -2.58153290e-01 1.91868916e-01 -2.74077207e-01 4.89601761e-01 7.27986157e-01 -8.37036014e-01 2.46632501e-01 -2.52635658e-01 4.55201656e-01 -9.59670991e-02 -5.80468357e-01 -6.51669145e-01 -1.84835345e-01 -4.58361030e-01 1.82601348e-01 -1.41004670e+00 -2.31043935e-01 3.26453507e-01 -4.66437995e-01 -2.94810653e-01 9.70172107e-01 -4.25419331e-01 1.39526224e+00 -2.09767795e+00 7.85585940e-01 4.11974549e-01 6.30001247e-01 -1.71598211e-01 -8.63385946e-02 1.11839473e+00 -2.99962968e-01 4.61668909e-01 -3.01732600e-01 -2.31525734e-01 2.25523293e-01 3.42842460e-01 -1.00563578e-01 9.18961346e-01 6.84034526e-02 9.70082164e-01 -1.00410998e+00 -4.59207684e-01 2.12873548e-01 8.19835067e-01 -7.49947846e-01 -1.48494482e-01 -3.72736119e-02 2.54192233e-01 -4.85209882e-01 2.82935143e-01 5.41418195e-01 -7.60011077e-01 3.71399313e-01 -1.80469021e-01 2.23511994e-01 1.51278511e-01 -1.16457200e+00 1.63160348e+00 -2.88217127e-01 7.76638806e-01 -1.19428344e-01 -1.28081560e+00 7.40468144e-01 2.92106658e-01 6.49868608e-01 -4.64217141e-02 8.38000551e-02 1.94522560e-01 2.04366043e-01 -1.40733555e-01 3.23739707e-01 1.54867440e-01 -5.67734614e-02 4.36675370e-01 2.52978563e-01 6.66086674e-02 2.61609972e-01 7.03654408e-01 1.54502594e+00 -3.92946154e-01 2.00106740e-01 -6.76393211e-01 4.36792254e-01 -2.46418983e-01 1.64930984e-01 3.70689183e-01 2.77281948e-03 2.66953051e-01 8.57244313e-01 -4.46884960e-01 -1.20066476e+00 -1.15272641e+00 -4.33428437e-01 5.37937880e-01 3.68709087e-01 -8.35569978e-01 -6.22240841e-01 -6.45813346e-01 2.81535149e-01 2.63091654e-01 -8.66114140e-01 -3.88372004e-01 -5.01868129e-01 -6.52317762e-01 2.61512667e-01 1.65813744e-01 5.86995110e-02 -7.58887351e-01 -8.95681083e-02 7.26660937e-02 1.52980983e-01 -8.79661024e-01 -6.50222480e-01 -8.61851871e-03 -1.03968036e+00 -1.29745722e+00 -3.64592999e-01 -9.16925907e-01 9.91094232e-01 1.10039271e-01 1.05812061e+00 7.50264287e-01 -5.59256196e-01 8.28269303e-01 -3.29666615e-01 2.24193469e-01 -5.74115455e-01 6.95309460e-01 5.49940281e-02 -2.11124614e-01 1.16975039e-01 -8.84287655e-01 -7.52888143e-01 2.28060409e-01 -8.50798547e-01 7.13139325e-02 3.04129481e-01 7.74196088e-01 3.71032059e-01 1.82510599e-01 3.53570998e-01 -8.87915194e-01 4.65307474e-01 -7.20854521e-01 -7.39748478e-01 2.06908420e-01 -1.03496885e+00 4.23688263e-01 7.04626322e-01 -3.89219880e-01 5.21513559e-02 -9.82218087e-02 2.25327119e-01 -2.80566543e-01 3.32983792e-01 6.37776613e-01 6.48327395e-02 -5.98485470e-01 5.07970512e-01 -1.87230129e-02 2.87212253e-01 -4.00217474e-01 4.87524271e-01 -5.32697933e-03 5.36070671e-03 -3.53592902e-01 1.39709020e+00 3.86803329e-01 5.52872717e-01 -9.32686210e-01 -2.47892708e-01 -4.23730969e-01 -9.57186878e-01 -1.65010974e-01 7.08772957e-01 -4.89798278e-01 -7.03002930e-01 1.18683591e-01 -1.08771741e+00 -4.87156183e-01 -1.91534594e-01 2.65584826e-01 -4.79494184e-01 4.99971956e-01 -8.70451450e-01 -4.34810817e-01 -2.50108838e-01 -8.92156959e-01 9.32852149e-01 -5.50242141e-02 -4.97221351e-02 -1.81756210e+00 5.03541708e-01 -3.10332268e-01 3.06552589e-01 4.81263816e-01 1.05102456e+00 -3.61665040e-01 -7.48610735e-01 -4.69233990e-01 -1.14868088e-02 -5.31658009e-02 3.06945909e-02 2.76544690e-01 -4.68785822e-01 -6.75109446e-01 -3.52589130e-01 1.07243821e-01 7.25919604e-01 1.75568134e-01 8.72364640e-01 -3.92362893e-01 -5.96767962e-01 8.27678263e-01 1.74402094e+00 -5.55885613e-01 5.31632245e-01 9.56463739e-02 9.76779819e-01 4.43538487e-01 9.78380963e-02 1.17460988e-01 4.40923959e-01 7.56015182e-01 5.78795791e-01 -7.46880053e-03 1.26595423e-01 -3.63079697e-01 3.08242142e-01 1.15699351e+00 -1.28644541e-01 -3.03002864e-01 -1.03679252e+00 5.02906263e-01 -1.90065420e+00 -9.17003334e-01 -2.68299550e-01 2.03169703e+00 4.48211193e-01 -9.72277205e-03 2.26880506e-01 1.72464132e-01 6.75081968e-01 3.59260708e-01 -4.97831225e-01 -2.61440009e-01 1.66029662e-01 9.16784927e-02 8.87705028e-01 8.96182179e-01 -8.93596351e-01 7.84043908e-01 5.82602644e+00 4.12309408e-01 -1.03138125e+00 1.65473938e-01 1.46363154e-01 3.35204601e-01 -6.28297150e-01 3.83706361e-01 -5.93319833e-01 2.66569674e-01 9.38095808e-01 -6.46630585e-01 7.28728890e-01 7.49179780e-01 -2.24618152e-01 5.53141177e-01 -1.23219311e+00 7.58888602e-01 -8.96672085e-02 -1.42143989e+00 -1.53814629e-01 5.86494863e-01 5.31714320e-01 1.60188377e-01 1.22830644e-02 2.35367671e-01 1.54985815e-01 -9.78842616e-01 3.93927634e-01 4.40048248e-01 6.22253537e-01 -4.37425315e-01 4.76910233e-01 6.20005578e-02 -1.75910437e+00 2.28652805e-01 -4.58221912e-01 1.59402657e-02 6.48064762e-02 4.63197798e-01 -8.80165875e-01 7.94248164e-01 4.76877511e-01 1.10851741e+00 -9.16083336e-01 8.87895703e-01 -1.26075581e-01 5.86450815e-01 -5.05907178e-01 -1.76510960e-01 -1.89404469e-02 -6.58159554e-01 8.70575309e-01 7.35739112e-01 3.26680541e-01 -3.78274210e-02 1.86763540e-01 1.00200963e+00 -1.15422234e-01 2.44659871e-01 -1.08781075e+00 -4.67101991e-01 4.76384342e-01 1.50518274e+00 -8.79233241e-01 2.04762638e-01 -3.85055602e-01 9.57993746e-01 6.10790133e-01 4.99886513e-01 -9.09174502e-01 -4.21514124e-01 6.82116568e-01 4.86537665e-01 4.49152797e-01 -6.08309150e-01 1.87848300e-01 -1.23614454e+00 7.73347318e-02 -2.03793287e-01 3.20749342e-01 -3.25157017e-01 -1.17691743e+00 3.99996489e-01 8.91957507e-02 -1.11058128e+00 -1.15035318e-01 -9.59723055e-01 -7.99810290e-01 4.75766867e-01 -1.23201799e+00 -1.07087421e+00 -2.80460417e-01 3.74075979e-01 -1.92218170e-01 8.57209489e-02 9.77490842e-01 2.57907987e-01 -4.42497462e-01 5.00230253e-01 2.33871773e-01 2.68792391e-01 3.00037891e-01 -1.56764698e+00 6.42201304e-01 5.49424171e-01 2.93152928e-01 5.47552407e-01 8.46062303e-01 -5.15558779e-01 -1.94892359e+00 -9.22892988e-01 7.47071087e-01 -5.63921154e-01 1.28841853e+00 -7.82526970e-01 -1.01286364e+00 9.45331514e-01 -3.42883915e-02 4.60389048e-01 5.79236269e-01 2.21286699e-01 -3.05598736e-01 6.16632625e-02 -9.97803092e-01 7.30254412e-01 1.34330678e+00 -6.29038572e-01 1.42019212e-01 6.45125091e-01 8.79188299e-01 -2.50252426e-01 -1.35209703e+00 2.57813722e-01 4.56504852e-01 -5.92274547e-01 1.07825744e+00 -5.71385562e-01 1.54580086e-01 -1.29552871e-01 -5.41233867e-02 -1.31066549e+00 -4.66924101e-01 -7.16396213e-01 -5.51730335e-01 8.21944594e-01 5.42415142e-01 -1.13295901e+00 7.48305380e-01 4.51196790e-01 1.49316370e-01 -9.78676558e-01 -9.12322938e-01 -8.48704040e-01 2.62857825e-01 -2.64544711e-02 4.83288646e-01 1.01069069e+00 1.05579033e-01 4.08232599e-01 -7.53683690e-03 4.13141400e-01 8.26400876e-01 3.04333959e-02 7.67023742e-01 -1.66858530e+00 -4.24917907e-01 -5.84131658e-01 -1.21221089e+00 -8.17170620e-01 5.03493011e-01 -1.73770964e+00 -6.41563654e-01 -1.47596490e+00 1.18710101e-02 -5.92756271e-01 2.75498349e-03 3.49536061e-01 1.52736694e-01 1.56343654e-01 1.09410405e-01 6.04144223e-02 -4.03176814e-01 6.93979979e-01 1.08103657e+00 -8.95835161e-02 -6.64766282e-02 -2.34262377e-01 -4.35045362e-01 3.19760352e-01 9.15055037e-01 -5.36056459e-01 -5.79961300e-01 -1.98699757e-01 5.58333814e-01 -9.23563987e-02 6.74086511e-01 -8.44685197e-01 1.24673791e-01 2.29882464e-01 -1.90915868e-01 -1.63474604e-01 1.88957885e-01 -9.39619720e-01 5.89240193e-01 6.80889904e-01 -9.65591967e-02 2.68751830e-01 8.83734450e-02 8.55405033e-01 1.72930330e-01 -3.44663709e-01 5.94591796e-01 3.31134111e-01 -2.71165907e-01 8.62471640e-01 6.78019673e-02 1.75029203e-01 1.04372287e+00 -6.68349192e-02 -4.99380261e-01 -5.33570886e-01 -8.25643897e-01 2.42602274e-01 1.00453210e+00 3.02216470e-01 4.31564838e-01 -1.69576824e+00 -4.48580325e-01 1.52924240e-01 1.43072948e-01 -1.34839952e-01 3.25896293e-02 1.10150027e+00 -7.39426672e-01 2.97563970e-01 2.11458564e-01 -6.62123322e-01 -1.12459886e+00 6.52387142e-01 4.96183425e-01 -3.17003459e-01 -9.64614570e-01 4.02456582e-01 6.88586384e-02 -7.32881665e-01 -8.07956904e-02 -2.06012547e-01 1.74773440e-01 -1.99481055e-01 -2.73317434e-02 3.34628522e-01 -1.83173984e-01 -7.41466105e-01 -1.91068098e-01 7.89626479e-01 4.33714315e-02 2.01924562e-01 1.32392538e+00 5.44232130e-02 -3.39303583e-01 7.11011052e-01 1.66574168e+00 -4.36869897e-02 -8.11778486e-01 -1.73808366e-01 -1.38284087e-01 -2.55968094e-01 -1.58299893e-01 1.49894610e-01 -1.11901736e+00 6.88490808e-01 4.26241636e-01 8.00264180e-01 5.09419382e-01 3.56524229e-01 5.53680480e-01 4.97095466e-01 4.55617964e-01 -6.30540013e-01 2.00176872e-02 3.81051660e-01 9.28370595e-01 -1.18729711e+00 4.41737622e-02 -7.24852204e-01 1.02990540e-02 1.26623094e+00 2.91437268e-01 -6.18798077e-01 1.34817648e+00 -8.40778649e-02 -4.36759233e-01 -4.99982208e-01 -6.15552604e-01 6.03539236e-02 4.09660667e-01 3.47913802e-01 3.67632031e-01 1.92602471e-01 -2.90790081e-01 -5.49585931e-02 -5.20060718e-01 -5.79894245e-01 6.60251737e-01 6.56451285e-01 -1.23740360e-01 -1.40652025e+00 2.02140898e-01 4.91037905e-01 5.41718025e-03 -4.20039371e-02 -4.84050393e-01 1.16054654e+00 -5.29780447e-01 3.86974514e-01 -9.86237898e-02 -4.79654282e-01 -4.45384905e-02 5.60001358e-02 6.03493035e-01 -6.52279437e-01 3.33529618e-03 -5.26874661e-01 -2.23525405e-01 -4.96393740e-01 -2.56135821e-01 -7.37153709e-01 -1.44958258e+00 -7.17522025e-01 -5.18730044e-01 2.09199131e-01 5.39311707e-01 4.79832411e-01 3.79151493e-01 2.76134789e-01 6.35301471e-01 -9.73749638e-01 -5.33377349e-01 -7.61466444e-01 -7.58990407e-01 5.56277394e-01 3.70971084e-01 -8.63994896e-01 -9.03434217e-01 -3.18276495e-01]
[7.027192115783691, 5.8788628578186035]
90a177a2-e4fb-4961-9c92-5903a5504a39
automated-pulmonary-embolism-detection-from
2111.05506
null
https://arxiv.org/abs/2111.05506v1
https://arxiv.org/pdf/2111.05506v1.pdf
Automated Pulmonary Embolism Detection from CTPA Images Using an End-to-End Convolutional Neural Network
Automated methods for detecting pulmonary embolisms (PEs) on CT pulmonary angiography (CTPA) images are of high demand. Existing methods typically employ separate steps for PE candidate detection and false positive removal, without considering the ability of the other step. As a result, most existing methods usually suffer from a high false positive rate in order to achieve an acceptable sensitivity. This study presents an end-to-end trainable convolutional neural network (CNN) where the two steps are optimized jointly. The proposed CNN consists of three concatenated subnets: 1) a novel 3D candidate proposal network for detecting cubes containing suspected PEs, 2) a 3D spatial transformation subnet for generating fixed-sized vessel-aligned image representation for candidates, and 3) a 2D classification network which takes the three cross-sections of the transformed cubes as input and eliminates false positives. We have evaluated our approach using the 20 CTPA test dataset from the PE challenge, achieving a sensitivity of 78.9%, 80.7% and 80.7% at 2 false positives per volume at 0mm, 2mm and 5mm localization error, which is superior to the state-of-the-art methods. We have further evaluated our system on our own dataset consisting of 129 CTPA data with a total of 269 emboli. Our system achieves a sensitivity of 63.2%, 78.9% and 86.8% at 2 false positives per volume at 0mm, 2mm and 5mm localization error.
['Xin Yang', 'Kwang-Ting Cheng', 'Jingen Liu', 'Xiang Li', 'Xiang Wang', 'Jianchao Su', 'Yi Lin']
2021-11-10
null
null
null
null
['pulmonary-embolism-detection']
['medical']
[-1.01566382e-01 4.24616560e-02 1.85681269e-01 1.40731141e-01 -1.12060797e+00 -5.45129895e-01 2.52004087e-01 2.37457097e-01 -5.26578784e-01 5.59054375e-01 -1.00999303e-01 -7.20468521e-01 -1.40051901e-01 -7.25805461e-01 -5.78850329e-01 -5.29963493e-01 -2.66288579e-01 7.14572251e-01 9.36694384e-01 4.02629912e-01 -7.59267733e-02 1.01406896e+00 -1.07327580e+00 5.07415771e-01 7.21450269e-01 1.13975918e+00 -1.20127965e-02 9.14048791e-01 8.67117848e-03 8.21697474e-01 -7.29369521e-01 -2.78400213e-01 6.00724339e-01 -3.21139604e-01 -6.07781708e-01 -1.23516805e-01 4.60054368e-01 -3.89359951e-01 -3.23059738e-01 6.06408477e-01 9.66478646e-01 -1.74210906e-01 7.52367795e-01 -8.39359164e-01 -7.69777000e-02 3.38374704e-01 -5.13257027e-01 7.71434724e-01 -3.43969971e-01 3.61385018e-01 5.68216443e-01 -9.97293770e-01 2.98506111e-01 8.46054375e-01 6.74004614e-01 1.89636990e-01 -8.03949773e-01 -6.81889176e-01 -5.53933859e-01 -1.93903374e-03 -1.39833176e+00 1.28534678e-02 4.28035796e-01 -5.71658373e-01 9.45547342e-01 3.15050036e-01 7.24515796e-01 6.32990420e-01 4.92580026e-01 2.01390028e-01 8.36887598e-01 -2.05958620e-01 -1.25532433e-01 1.34058490e-01 -2.16869161e-01 7.68089831e-01 5.18578112e-01 4.52752560e-01 2.91356951e-01 -2.66022652e-01 1.23580003e+00 2.57673804e-02 -1.49140000e-01 -4.60565090e-01 -1.13806796e+00 8.81374240e-01 7.32367516e-01 2.81994313e-01 -6.82133079e-01 2.81393398e-02 2.63344795e-01 -2.14213014e-01 -1.27356714e-02 5.43142438e-01 -1.63958207e-01 -1.95322614e-02 -7.82805920e-01 1.54113993e-01 7.14740336e-01 7.18461931e-01 -1.40065506e-01 1.92055553e-01 -6.56150460e-01 6.73309803e-01 1.44204479e-02 4.93610889e-01 4.82295394e-01 -5.58054090e-01 5.08362830e-01 5.32814503e-01 2.05602676e-01 -9.78936791e-01 -6.53621137e-01 -1.22633255e+00 -7.76066244e-01 3.05095434e-01 4.88825917e-01 -2.63559729e-01 -1.09349728e+00 1.16927135e+00 2.53920704e-01 4.24627423e-01 -2.86481649e-01 1.02703750e+00 1.14193511e+00 3.24788243e-01 1.43300071e-01 -1.16644442e-01 1.50399745e+00 -1.03010154e+00 -1.79957137e-01 1.55582726e-01 6.66872144e-01 -8.60797882e-01 8.99011135e-01 3.20296943e-01 -1.38101196e+00 -5.63519478e-01 -1.07688248e+00 3.88697922e-01 -1.02625019e-03 5.70040703e-01 -5.22577800e-02 7.61334896e-01 -8.79404068e-01 5.24396122e-01 -8.98085356e-01 1.09539866e-01 8.18678975e-01 5.54203153e-01 -5.66386208e-02 4.00996447e-01 -7.69200206e-01 1.03750932e+00 1.45655677e-01 -9.78991538e-02 -1.04454923e+00 -1.05019510e+00 -2.66493678e-01 5.90612292e-01 4.38171297e-01 -9.51965034e-01 1.24384761e+00 -3.12032908e-01 -1.20329487e+00 7.29784012e-01 5.80830425e-02 -6.79837346e-01 9.12784100e-01 -1.60942182e-01 -3.38180035e-01 5.85251629e-01 1.22772448e-01 3.93259704e-01 7.27473080e-01 -8.43116343e-01 -1.08708847e+00 -1.27475426e-01 -2.28558660e-01 -6.65701032e-02 1.23766586e-01 -2.76541449e-02 -2.77584434e-01 -7.81976581e-01 1.12283841e-01 -8.05898905e-01 -3.01403880e-01 8.18646047e-03 -3.15547287e-01 -1.64671410e-02 6.89630985e-01 -6.06762648e-01 1.19856060e+00 -1.68632948e+00 -3.60910296e-01 4.23184991e-01 9.19887304e-01 7.91629314e-01 -1.21161947e-02 -3.19202155e-01 -3.83578062e-01 3.07255805e-01 -2.45442495e-01 5.52368797e-02 -3.87219846e-01 -3.62747461e-01 1.33085340e-01 2.13860348e-01 4.88395929e-01 9.87156272e-01 -8.16617668e-01 -5.67818761e-01 6.62260532e-01 5.76048255e-01 -4.19732422e-01 2.25952804e-01 3.95888507e-01 5.12353659e-01 -5.09183168e-01 4.24120367e-01 7.50909150e-01 -5.34664869e-01 -7.43792579e-02 -1.36519164e-01 -1.05597131e-01 1.97401479e-01 -1.26891804e+00 9.08649147e-01 -3.96630287e-01 4.10909623e-01 -2.15354621e-01 -6.67919517e-01 8.35530400e-01 6.64224148e-01 9.46346462e-01 -4.76668447e-01 4.09089386e-01 5.04794180e-01 6.14279687e-01 -5.45512736e-01 -9.41395611e-02 -3.07779819e-01 4.09638673e-01 2.82634974e-01 -2.97936231e-01 1.87909547e-02 7.26374462e-02 -1.73562597e-02 1.41951287e+00 -5.47546566e-01 3.29696417e-01 -2.39800990e-01 7.66817927e-01 8.49046409e-02 2.92501867e-01 1.08094251e+00 -4.99533474e-01 9.82993484e-01 6.34906113e-01 -7.64271379e-01 -9.74302173e-01 -1.38379490e+00 -3.77529293e-01 1.17955036e-01 -1.10192254e-01 -1.92231014e-01 -3.76623631e-01 -1.03893638e+00 -2.08858371e-01 2.76503414e-01 -4.05749947e-01 -2.87013482e-02 -1.09323692e+00 -7.44553268e-01 6.14935815e-01 8.33602071e-01 6.95605516e-01 -1.05443621e+00 -1.03569865e+00 3.85125101e-01 -1.68818682e-01 -1.15568876e+00 -1.45101175e-01 1.05016604e-02 -9.63403344e-01 -1.47413588e+00 -7.47056305e-01 -7.63093710e-01 5.34011066e-01 2.98875779e-01 1.26241958e+00 2.75099844e-01 -5.47083378e-01 -8.29531103e-02 -2.35608831e-01 -3.61502558e-01 -3.31374854e-01 1.65969878e-01 -4.11656797e-01 -2.61043459e-01 1.00712240e-01 -4.72373366e-01 -9.55266714e-01 3.90301406e-01 -5.23980439e-01 -2.94487149e-01 1.00803256e+00 8.11735511e-01 8.22695971e-01 -4.87111621e-02 6.67496085e-01 -9.22292113e-01 5.97485363e-01 -4.21990871e-01 -6.20746374e-01 7.02867359e-02 -3.00173700e-01 -4.06078309e-01 8.43526721e-01 -4.54460233e-01 -6.73871577e-01 3.06441993e-01 -2.52189994e-01 -8.04432869e-01 -2.63828754e-01 1.12221882e-01 3.24585050e-01 -1.71765730e-01 1.05652022e+00 -3.14348824e-02 -1.30792215e-01 -1.26252040e-01 4.16249484e-02 5.90904951e-01 5.66350102e-01 -2.27995202e-01 7.93321073e-01 4.93688583e-01 4.62141335e-01 -4.72611904e-01 -5.49079239e-01 -5.27103782e-01 -7.87919581e-01 -5.94535582e-02 8.49489152e-01 -7.40907967e-01 -3.55779767e-01 -2.35057231e-02 -8.50275159e-01 1.44759223e-01 -6.44660532e-01 8.07382882e-01 -2.10692927e-01 4.16913003e-01 -4.27279055e-01 -3.95269960e-01 -8.86197209e-01 -1.54328585e+00 7.81387806e-01 7.57285729e-02 -8.50715339e-02 -6.42571032e-01 -2.33217314e-01 1.05858214e-01 7.86163926e-01 4.30723429e-01 9.63606417e-01 -9.35256779e-01 -7.30182409e-01 -5.94742477e-01 -5.65138996e-01 2.35683143e-01 9.81907174e-02 -1.66343838e-01 -6.60940826e-01 -2.42489815e-01 2.02061590e-02 1.68065339e-01 6.35208964e-01 7.86956072e-01 1.32243681e+00 4.06180918e-02 -4.83367145e-01 6.80473447e-01 1.14436662e+00 5.00638723e-01 6.21729970e-01 2.37580508e-01 6.05716944e-01 4.74638194e-02 3.22010636e-01 4.89383012e-01 -1.08768538e-01 3.79911870e-01 6.36148334e-01 -3.78174096e-01 -4.19552326e-01 2.36793235e-01 -4.62635040e-01 2.77461559e-01 -2.26172596e-01 -3.24831903e-01 -1.16975331e+00 5.20579576e-01 -1.43772209e+00 -6.91522241e-01 -6.31579936e-01 2.33646846e+00 1.62310600e-01 3.81577134e-01 3.45105112e-01 -3.29904445e-02 7.49575377e-01 -1.74342737e-01 -4.46052939e-01 -1.18658006e-01 3.06945950e-01 1.00206339e+00 4.52016383e-01 2.39553764e-01 -1.56835556e+00 5.17668128e-01 5.44807863e+00 7.24811733e-01 -1.32960379e+00 2.02432513e-01 6.18825078e-01 -2.57137060e-01 3.08967322e-01 -1.71050072e-01 -5.20676672e-01 4.66569573e-01 8.31347406e-01 2.08057284e-01 -2.33165398e-01 6.93952382e-01 1.96429908e-01 6.94587901e-02 -7.46591747e-01 9.36962843e-01 -1.07437856e-01 -1.40932727e+00 -1.72027946e-02 -1.11331396e-01 3.70416701e-01 2.24167928e-01 3.27646695e-02 2.49738574e-01 -2.20752005e-02 -1.25052524e+00 3.13195050e-01 2.20383093e-01 9.12298560e-01 -6.65516198e-01 1.08372951e+00 2.96364516e-01 -1.28725004e+00 2.98928171e-02 -2.91246831e-01 3.90048951e-01 1.95127949e-01 4.75073934e-01 -1.30876625e+00 4.36367661e-01 7.73782372e-01 3.81923348e-01 -6.66444719e-01 1.64669001e+00 -3.20225954e-01 6.08545482e-01 -6.64898515e-01 5.10921888e-02 2.43520662e-01 1.78604454e-01 9.11054969e-01 1.18123472e+00 5.09848237e-01 2.21897066e-01 8.83821100e-02 9.06994998e-01 -8.96638408e-02 3.57093930e-01 -2.24492773e-01 4.81430829e-01 6.84151709e-01 1.40540588e+00 -9.45210695e-01 -6.62455976e-01 -2.37642184e-01 3.35720837e-01 1.85438991e-01 -5.23460768e-02 -1.23800457e+00 -5.42134821e-01 -1.42386004e-01 6.33134127e-01 5.86624980e-01 2.14662090e-01 -3.84913296e-01 -8.75648260e-01 2.36026675e-01 -5.29060543e-01 6.88084960e-01 -5.68201840e-01 -1.04956234e+00 9.51138020e-01 -1.71793044e-01 -1.62935102e+00 -1.28123030e-01 -7.53106773e-01 -9.75771606e-01 1.05430996e+00 -1.35610950e+00 -8.55538309e-01 -7.10991561e-01 2.72264779e-01 3.50856751e-01 -2.06483483e-01 6.17963493e-01 5.08628190e-01 -4.15558785e-01 6.28660858e-01 -2.07112342e-01 2.43635595e-01 6.35447383e-01 -1.15703464e+00 2.11809382e-01 8.91798675e-01 -1.97148085e-01 3.46531689e-01 7.49682263e-02 -5.84294140e-01 -6.80321813e-01 -1.46140778e+00 8.01774859e-01 -6.78178966e-01 2.79937655e-01 2.72618562e-01 -8.44416976e-01 4.70619529e-01 -2.53748715e-01 5.17955363e-01 3.32356513e-01 -4.94535923e-01 8.85883644e-02 1.58347394e-02 -1.35811949e+00 3.71825308e-01 8.41487288e-01 1.60230994e-01 -3.80264193e-01 3.93641204e-01 2.45117739e-01 -8.94148588e-01 -1.14814961e+00 9.75090146e-01 6.33685529e-01 -1.02254903e+00 1.25303972e+00 -5.85893333e-01 5.25815010e-01 -2.42502585e-01 4.27203804e-01 -9.91162062e-01 -7.11943209e-01 -8.37920159e-02 -4.31872606e-02 2.61114597e-01 6.36310220e-01 -7.69818902e-01 1.00924957e+00 1.87560245e-01 -5.50615251e-01 -1.23252141e+00 -1.07526827e+00 -5.15134513e-01 4.74987358e-01 -2.82631844e-01 5.33949971e-01 5.87858796e-01 -5.21952450e-01 2.42753297e-01 6.32190108e-02 3.30699742e-01 2.99156576e-01 2.71878503e-02 6.31737947e-01 -1.26086330e+00 -2.81134367e-01 -8.23226452e-01 -3.16902637e-01 -7.61999488e-01 -5.67949653e-01 -1.00956035e+00 -3.53987426e-01 -1.51337683e+00 2.51569033e-01 -4.76198912e-01 -3.67937297e-01 2.37066403e-01 -2.19616026e-01 8.88586044e-02 1.01449989e-01 3.98229808e-01 -1.56558841e-01 2.99387891e-02 1.64622140e+00 1.46791607e-01 -3.37153673e-01 2.39469409e-01 -4.72642362e-01 7.75723755e-01 1.13680387e+00 -7.18869209e-01 -3.89489651e-01 -2.58140594e-01 -4.52264488e-01 3.41365218e-01 6.48683965e-01 -1.29725218e+00 1.44468978e-01 2.10150540e-01 8.57512414e-01 -9.79989171e-01 2.52328515e-01 -8.56406987e-01 -3.10213976e-02 8.83208096e-01 -6.65443838e-02 1.85677081e-01 2.75545895e-01 2.19737008e-01 -1.12261713e-01 -1.90351337e-01 1.14463735e+00 -3.97626549e-01 -1.57292575e-01 4.71180707e-01 -4.54480499e-01 1.70704037e-01 1.17556477e+00 -3.27082902e-01 -3.17814887e-01 -1.03728890e-01 -1.04888260e+00 1.27600998e-01 -2.48810068e-01 1.62159488e-01 8.52795780e-01 -1.01362717e+00 -9.54539657e-01 4.40274268e-01 -2.37570569e-01 2.45688245e-01 3.10718834e-01 1.36685777e+00 -1.04119301e+00 7.04654813e-01 -1.99464217e-01 -9.74397182e-01 -1.28265619e+00 3.05661440e-01 1.02452493e+00 -7.40614831e-01 -8.82206500e-01 8.00389767e-01 2.73185492e-01 -5.00054434e-02 1.46625131e-01 -5.75706124e-01 -2.15676919e-01 -4.27695125e-01 1.78978398e-01 5.87038040e-01 5.39539635e-01 -5.00491321e-01 -3.37959915e-01 5.00165939e-01 -1.19882740e-01 2.48346731e-01 9.46091175e-01 5.14606237e-01 3.23507667e-01 -4.28242028e-01 9.39081132e-01 7.19830347e-03 -6.99436724e-01 -1.11454733e-01 -1.47755802e-01 -6.29744411e-01 -7.72125795e-02 -9.29914534e-01 -1.37960660e+00 1.19903672e+00 1.02594268e+00 -3.81341204e-02 9.77322876e-01 -3.76331760e-03 8.67424011e-01 1.61095746e-02 -1.48554251e-01 -4.21841711e-01 1.12506464e-01 2.88468510e-01 8.18949223e-01 -1.18992376e+00 -3.61843742e-02 -6.17981553e-01 -4.07362819e-01 1.19706738e+00 8.98828447e-01 -3.67105633e-01 8.42229187e-01 2.08587974e-01 5.30876964e-02 -5.10707021e-01 -5.42294860e-01 -1.10500060e-01 5.23401201e-01 4.29780155e-01 4.06599253e-01 3.15393478e-01 -3.34302366e-01 9.17823672e-01 -2.46871844e-01 -1.27390712e-01 3.90775919e-01 1.00327659e+00 -6.11405194e-01 -7.63251901e-01 -3.85648698e-01 9.80735898e-01 -8.98295164e-01 2.94667147e-02 -2.87161857e-01 1.18599558e+00 3.24693829e-01 7.62368679e-01 -5.33328718e-03 -2.27966636e-01 8.46374393e-01 -2.06564710e-01 2.42268458e-01 -6.84635043e-01 -1.02666605e+00 4.58110005e-01 1.20941110e-01 -4.43948448e-01 -2.85952613e-02 -5.26693583e-01 -1.42075789e+00 2.92882510e-02 -4.53149736e-01 1.64272934e-01 1.83611184e-01 5.24269998e-01 5.89337647e-01 8.91375065e-01 4.76402551e-01 -3.67492259e-01 -7.16120303e-01 -1.00755501e+00 -2.77598113e-01 3.35701525e-01 1.96535915e-01 -7.44864762e-01 -3.89845818e-01 -1.21117778e-01]
[15.183398246765137, -2.0896756649017334]
7688ebe2-2be4-48cf-b3f5-12f3a2b689ea
multi-object-video-generation-from-single
2305.03983
null
https://arxiv.org/abs/2305.03983v2
https://arxiv.org/pdf/2305.03983v2.pdf
Multi-object Video Generation from Single Frame Layouts
In this paper, we study video synthesis with emphasis on simplifying the generation conditions. Most existing video synthesis models or datasets are designed to address complex motions of a single object, lacking the ability of comprehensively understanding the spatio-temporal relationships among multiple objects. Besides, current methods are usually conditioned on intricate annotations (e.g. video segmentations) to generate new videos, being fundamentally less practical. These motivate us to generate multi-object videos conditioning exclusively on object layouts from a single frame. To solve above challenges and inspired by recent research on image generation from layouts, we have proposed a novel video generative framework capable of synthesizing global scenes with local objects, via implicit neural representations and layout motion self-inference. Our framework is a non-trivial adaptation from image generation methods, and is new to this field. In addition, our model has been evaluated on two widely-used video recognition benchmarks, demonstrating effectiveness compared to the baseline model.
['Liang Lin', 'Hefeng Wu', 'Zhibin Liu', 'Yang Wu']
2023-05-06
null
null
null
null
['video-recognition', 'video-generation']
['computer-vision', 'computer-vision']
[ 6.98727429e-01 -4.19188403e-02 -1.98914409e-01 -9.75410938e-02 -4.68544692e-01 -6.03479207e-01 8.44731152e-01 -3.69619608e-01 -4.40163910e-02 7.42044687e-01 2.56928235e-01 -4.25661094e-02 1.57401159e-01 -8.21569443e-01 -1.24433243e+00 -6.21251583e-01 2.52181560e-01 2.11152032e-01 2.25631073e-01 -8.91035423e-02 3.15092653e-01 4.63261634e-01 -1.73371077e+00 3.21337312e-01 6.47923946e-01 7.25553095e-01 3.97133201e-01 9.02927876e-01 2.19225828e-02 9.93727982e-01 -7.37570763e-01 -3.11794609e-01 2.36924693e-01 -8.12162399e-01 -7.72884369e-01 6.23205125e-01 8.80704165e-01 -6.99215651e-01 -5.96859753e-01 9.27015483e-01 1.40817359e-01 2.54690051e-01 6.70580864e-01 -1.32315123e+00 -9.55301344e-01 6.70386732e-01 -3.07636648e-01 -1.29763680e-02 4.74452347e-01 3.56161118e-01 8.84302914e-01 -7.13069439e-01 9.63891923e-01 1.22777998e+00 3.55585605e-01 7.30784595e-01 -1.39710426e+00 -3.51890713e-01 6.12776697e-01 3.40326786e-01 -1.33291423e+00 -6.13753676e-01 8.55313480e-01 -6.08004510e-01 7.94564962e-01 2.95159608e-01 8.01472723e-01 1.53138590e+00 -1.06581427e-01 9.17580128e-01 7.36227691e-01 -3.63515049e-01 2.61583328e-01 -2.59601265e-01 -3.46831113e-01 4.24298197e-01 2.48814031e-01 3.99091281e-02 -5.19223273e-01 2.65795678e-01 1.11950529e+00 -8.75781775e-02 -4.86001581e-01 -5.66615283e-01 -1.43116379e+00 6.69191778e-01 1.86606124e-01 2.24291727e-01 -3.08646560e-01 4.30886030e-01 2.15446040e-01 -1.34224087e-01 1.28464937e-01 3.67484570e-01 1.38593717e-02 -1.48051485e-01 -1.21069705e+00 7.12426841e-01 7.39083827e-01 1.36277807e+00 7.57920504e-01 2.98788220e-01 -4.85825390e-01 4.01935071e-01 1.30367503e-01 3.93194109e-01 2.73654312e-01 -1.19955218e+00 3.82181674e-01 3.08814973e-01 2.27814510e-01 -1.21759546e+00 4.01564986e-02 -3.19934845e-01 -8.01603556e-01 -1.36120155e-01 3.02690864e-01 1.83959436e-02 -9.70511377e-01 1.70265007e+00 9.72662270e-02 7.52562523e-01 -1.00252368e-01 9.98807251e-01 7.44142950e-01 8.34547043e-01 -6.65649697e-02 -2.66030878e-01 9.64492500e-01 -1.43148839e+00 -8.10668945e-01 -1.09249778e-01 2.70422906e-01 -6.76225483e-01 9.39004183e-01 5.52301943e-01 -1.34838140e+00 -8.09376717e-01 -9.81135905e-01 -1.30611017e-01 -2.61491895e-01 -1.27809085e-02 6.83492243e-01 5.72866023e-01 -1.19996834e+00 4.54629093e-01 -6.70878172e-01 -3.01913083e-01 3.60924244e-01 1.63560063e-01 -1.61053658e-01 -2.04321861e-01 -8.82669270e-01 3.96190733e-01 6.33534908e-01 3.15074623e-01 -1.20329654e+00 -6.74160480e-01 -9.02720928e-01 -3.90329324e-02 6.82073295e-01 -1.15021682e+00 1.21209872e+00 -1.38160264e+00 -1.77417135e+00 4.01532888e-01 -1.78912804e-01 -4.62355465e-01 5.69093227e-01 -4.19694662e-01 -1.24529295e-01 2.52213299e-01 1.13660820e-01 1.12798762e+00 1.05295563e+00 -1.57835531e+00 -5.27874887e-01 1.56761065e-01 3.46542329e-01 1.70456961e-01 -2.63381213e-01 -1.43483043e-01 -9.17729676e-01 -1.13829660e+00 -3.22767496e-01 -9.00790393e-01 -2.93979198e-01 -1.68929294e-01 -5.06883621e-01 1.39757946e-01 1.03063238e+00 -5.58920562e-01 1.37537849e+00 -1.90437257e+00 6.58394217e-01 -1.35504022e-01 6.51856288e-02 3.33419472e-01 -4.47638303e-01 3.18588018e-01 7.85799325e-03 2.44837284e-01 -3.11083943e-01 -3.56747389e-01 -4.92706709e-02 2.60022193e-01 -5.70645571e-01 8.61485824e-02 5.91646731e-01 9.94727671e-01 -1.08328927e+00 -4.60085541e-01 4.26795006e-01 4.46882606e-01 -1.02672672e+00 3.82201642e-01 -6.68533266e-01 6.42933428e-01 -3.46057802e-01 6.27974749e-01 5.79942465e-01 -3.37881774e-01 3.13510984e-01 -3.50608587e-01 2.02907138e-02 -7.94417486e-02 -1.18809998e+00 2.06863666e+00 -2.29570016e-01 4.67563182e-01 -3.31625998e-01 -1.05700457e+00 5.68334401e-01 2.86459297e-01 6.35559916e-01 -4.03313667e-01 1.74802616e-01 -1.42242819e-01 -2.07785651e-01 -6.12878799e-01 7.08605587e-01 2.41583422e-01 7.12499842e-02 2.32577443e-01 1.53674960e-01 -1.69497132e-01 7.18194187e-01 1.58461362e-01 1.00328338e+00 8.13782513e-01 2.40295470e-01 1.15366556e-01 4.16585535e-01 -1.07294835e-01 5.00207424e-01 9.16357815e-01 6.16693795e-02 9.99897420e-01 4.05588299e-01 -3.60585093e-01 -1.14464462e+00 -1.17910814e+00 2.18775809e-01 6.78924561e-01 2.85251468e-01 -6.64290190e-01 -1.03596985e+00 -6.82875693e-01 -3.59841228e-01 5.21962464e-01 -5.53570211e-01 5.71580464e-03 -8.38949859e-01 -5.95606863e-01 3.85522842e-01 6.16804302e-01 3.07992369e-01 -1.14330196e+00 -5.84682643e-01 3.79199892e-01 -4.83799309e-01 -1.63570023e+00 -5.78626692e-01 -5.08671880e-01 -7.87638426e-01 -9.56839740e-01 -8.51350605e-01 -7.67490327e-01 7.60728180e-01 3.79414797e-01 1.27026510e+00 1.31491438e-01 -3.98215979e-01 5.92725754e-01 -4.92516488e-01 4.21635769e-02 -3.56705427e-01 7.83535764e-02 -1.69273019e-01 3.73077184e-01 -1.64418519e-01 -5.71660697e-01 -5.55841148e-01 2.61261195e-01 -1.35533583e+00 6.65561497e-01 6.85284793e-01 8.25399697e-01 5.63609660e-01 -7.60829961e-03 5.64104974e-01 -8.19126666e-01 1.76809356e-01 -4.88260508e-01 -5.52983105e-01 2.30451748e-01 1.36790304e-02 -3.99707407e-02 7.81643808e-01 -6.20740473e-01 -1.26201475e+00 2.56251484e-01 1.01985373e-01 -6.28454208e-01 -5.42687595e-01 2.93473482e-01 -5.04878581e-01 1.24589331e-01 2.97062635e-01 3.81796598e-01 -3.40123624e-01 -2.94755906e-01 7.23092556e-01 2.50892997e-01 8.70205879e-01 -8.29172552e-01 8.63863289e-01 5.24979770e-01 -3.79830338e-02 -9.56499636e-01 -5.27717650e-01 -8.66057873e-02 -8.74353230e-01 -3.33318532e-01 1.06060266e+00 -8.48472118e-01 -3.56424659e-01 5.81128538e-01 -1.26517737e+00 -5.21703124e-01 -1.26830667e-01 3.12377214e-01 -9.16504085e-01 6.27859414e-01 -4.53432173e-01 -4.75866795e-01 1.52760103e-01 -1.32219970e+00 1.24043894e+00 1.47193313e-01 -2.38521978e-01 -8.66101444e-01 -8.36469308e-02 3.95676762e-01 2.86851525e-01 4.68441159e-01 6.16658747e-01 1.49740115e-01 -1.41255605e+00 1.98700711e-01 -1.80511430e-01 3.27193737e-01 3.02496195e-01 4.02474880e-01 -8.24101150e-01 -2.52372354e-01 -2.75596619e-01 -9.06274095e-02 7.72380531e-01 2.87698865e-01 1.51508963e+00 -4.49635237e-01 -3.94097239e-01 7.29303360e-01 1.39576304e+00 2.62516946e-01 8.52251768e-01 2.61878341e-01 1.05722439e+00 5.07596731e-01 4.53053981e-01 5.22993386e-01 3.19900244e-01 1.01854157e+00 3.29964250e-01 9.11068618e-02 -4.35011059e-01 -4.22405869e-01 5.22851646e-01 6.47184193e-01 -3.07858348e-01 -6.51446760e-01 -6.16265714e-01 6.27510250e-01 -2.05350018e+00 -1.10050976e+00 -6.28060326e-02 1.83950806e+00 6.20592892e-01 -9.52552781e-02 1.22808360e-01 -8.25865418e-02 6.61516249e-01 2.41487145e-01 -3.57134670e-01 1.34422123e-01 -2.84398288e-01 1.50303543e-01 2.41738558e-01 3.11081678e-01 -1.25768936e+00 1.21888685e+00 6.49137640e+00 7.31746912e-01 -1.04392362e+00 -2.49667212e-01 5.85417986e-01 -2.75768131e-01 -3.39446157e-01 1.71494469e-01 -8.10412943e-01 4.45661575e-01 6.71616256e-01 3.53592485e-02 6.90951586e-01 5.62401474e-01 3.06961417e-01 8.09439167e-04 -1.31977999e+00 1.06420338e+00 4.58039254e-01 -1.67961228e+00 6.24940276e-01 -2.17834990e-02 1.13817489e+00 -6.27030611e-01 5.12344874e-02 1.43615410e-01 9.70450416e-02 -9.99944985e-01 1.19413793e+00 7.50897110e-01 5.85461795e-01 -5.88258684e-01 3.01572949e-01 1.01941645e-01 -1.22195518e+00 5.70573620e-02 -4.10658084e-02 -4.70528491e-02 5.41495740e-01 3.35190445e-01 -6.20198965e-01 7.74698913e-01 6.32984877e-01 9.55279350e-01 -7.31379032e-01 9.52513039e-01 -1.28494963e-01 3.92848521e-01 -1.30685372e-02 3.32594693e-01 2.51470655e-01 -2.16770992e-01 4.23825353e-01 1.24414575e+00 4.21474338e-01 -1.34940907e-01 2.57454067e-01 1.15592349e+00 -1.02015674e-01 2.45983303e-02 -8.08040857e-01 -3.75357866e-01 1.70449898e-01 1.26972866e+00 -9.55125034e-01 -5.58101833e-01 -7.00392067e-01 1.06745136e+00 1.92618728e-01 5.66821635e-01 -1.26213396e+00 -8.95593539e-02 5.64905941e-01 1.49244294e-01 6.40912950e-01 -4.52860504e-01 -1.11633547e-01 -1.38394105e+00 4.43029553e-02 -1.13334405e+00 -1.43154994e-01 -8.09108257e-01 -1.02372217e+00 4.70342636e-01 1.76731631e-01 -1.22847784e+00 -5.01749694e-01 -6.61445439e-01 -3.73932749e-01 3.90606880e-01 -1.40917575e+00 -1.24557638e+00 -4.30017740e-01 5.00361443e-01 8.13159645e-01 1.04735317e-02 5.48246562e-01 5.17334342e-01 -6.65398419e-01 4.14407969e-01 -3.39856178e-01 1.36458755e-01 6.73931599e-01 -1.02598917e+00 6.40938044e-01 1.21719968e+00 3.71327728e-01 6.71341240e-01 6.17455661e-01 -6.76928520e-01 -1.63859713e+00 -1.29707539e+00 3.86025876e-01 -5.37781358e-01 5.09859025e-01 -5.76336086e-01 -8.11805904e-01 7.43888199e-01 5.02532542e-01 -9.94858071e-02 4.15742010e-01 -4.05934334e-01 -1.71348259e-01 1.00359164e-01 -5.43053031e-01 9.77236450e-01 1.51466763e+00 -2.80275434e-01 -2.94914693e-01 1.47078201e-01 7.11187601e-01 -4.19526666e-01 -7.32014239e-01 6.16854072e-01 4.55142021e-01 -1.06332111e+00 1.21576357e+00 -6.69181108e-01 6.94783866e-01 -7.19582379e-01 -2.19373062e-01 -9.00327981e-01 -2.99703270e-01 -7.46126354e-01 -4.69772190e-01 1.37254155e+00 -4.26032906e-03 -2.50169244e-02 8.52893114e-01 3.26811641e-01 -3.54282051e-01 -3.52742404e-01 -2.88735807e-01 -7.57094383e-01 -3.04702371e-01 -3.72131824e-01 5.96785367e-01 7.86926746e-01 -6.10410154e-01 2.36754492e-01 -7.63723552e-01 2.25033209e-01 5.68784297e-01 2.53254443e-01 1.26678026e+00 -7.13097513e-01 -7.69017220e-01 -7.20272839e-01 -3.89249653e-01 -1.47830713e+00 3.79701465e-01 -5.68395197e-01 2.60457844e-01 -1.42575657e+00 3.07513088e-01 -2.84411255e-02 3.89902256e-02 1.68598160e-01 -3.18279147e-01 5.47905326e-01 3.75803232e-01 6.11547418e-02 -7.18338490e-01 6.66006982e-01 1.49841213e+00 -2.63856769e-01 3.24994209e-03 -3.14071625e-01 -5.76586723e-01 7.09979773e-01 5.72682798e-01 -8.12678486e-02 -7.62993813e-01 -5.74059963e-01 8.55290592e-02 5.86617589e-02 6.78628027e-01 -1.08079064e+00 9.42015052e-02 -4.45057631e-01 3.38383049e-01 -5.54251492e-01 3.43990088e-01 -6.01153076e-01 5.18547237e-01 1.60109669e-01 -1.51705444e-01 1.98870469e-02 9.86816287e-02 6.95844412e-01 -3.28864247e-01 -2.40792632e-01 4.69545752e-01 -2.51792818e-01 -1.14540398e+00 4.31557328e-01 -4.46602970e-01 3.78376134e-02 1.12034845e+00 -3.45143318e-01 -2.62130409e-01 -3.43029648e-01 -7.44850934e-01 -2.09649086e-01 7.40130603e-01 7.92465985e-01 6.73028767e-01 -1.37654591e+00 -4.17935252e-01 2.07895681e-01 7.04749227e-02 2.42685869e-01 4.30169821e-01 5.48624933e-01 -8.09211791e-01 6.16425276e-01 -3.30145270e-01 -7.00679958e-01 -1.01358891e+00 9.39273179e-01 1.02828771e-01 1.28618954e-02 -5.79361975e-01 5.34123242e-01 7.57125556e-01 -4.04653661e-02 1.35487959e-01 -5.98598540e-01 1.07018426e-02 -1.60311386e-01 4.13909167e-01 3.32927793e-01 -2.12548479e-01 -7.04788208e-01 1.01334490e-01 6.93189919e-01 -1.70284715e-02 -1.27245888e-01 1.05696344e+00 -1.12545811e-01 -6.43372675e-03 3.16413611e-01 9.47743118e-01 -1.41388789e-01 -1.67824519e+00 4.20414805e-02 2.45454460e-02 -6.51421070e-01 -2.55553693e-01 -4.27072525e-01 -1.02700019e+00 7.55868673e-01 8.90896320e-02 1.12028904e-02 1.24035633e+00 -2.27729127e-01 8.21887493e-01 3.14256161e-01 4.68972236e-01 -9.55481708e-01 5.69078147e-01 4.01487201e-01 9.28940594e-01 -9.87689316e-01 -1.95814893e-01 -6.20107234e-01 -4.40667480e-01 1.14741933e+00 8.75793219e-01 -2.07987830e-01 2.51831472e-01 2.89019436e-01 -2.10466206e-01 2.90380239e-01 -6.99678123e-01 -7.62718245e-02 3.99944574e-01 6.18954659e-01 4.27052826e-01 -1.80647418e-01 -2.11998373e-02 2.68908411e-01 1.61766505e-03 2.07392395e-01 6.69504344e-01 9.16118920e-01 -1.22724354e-01 -1.32786751e+00 -3.06908935e-01 3.88749465e-02 -2.71122098e-01 -2.53306869e-02 -2.79438525e-01 7.35089958e-01 3.38802308e-01 7.23889172e-01 1.37843713e-01 -1.51838854e-01 5.83171658e-02 -1.74789503e-01 8.50752413e-01 -6.99570000e-01 -9.02842507e-02 3.41382742e-01 -1.00081839e-01 -6.72031641e-01 -8.61302137e-01 -8.37946177e-01 -8.64985466e-01 -4.65610586e-02 -1.98975727e-02 -2.30466381e-01 3.69074374e-01 7.74273694e-01 3.13606858e-01 9.03021038e-01 3.25100571e-01 -1.29201138e+00 -4.41355594e-02 -6.63483620e-01 -3.45393747e-01 6.43206596e-01 2.26932719e-01 -7.42004633e-01 4.25606817e-02 7.46369839e-01]
[10.789131164550781, -0.543785810470581]
c0b3ebe1-5b9d-44e3-a5ce-475a718189a9
learning-generalisable-omni-scale
1910.06827
null
https://arxiv.org/abs/1910.06827v5
https://arxiv.org/pdf/1910.06827v5.pdf
Learning Generalisable Omni-Scale Representations for Person Re-Identification
An effective person re-identification (re-ID) model should learn feature representations that are both discriminative, for distinguishing similar-looking people, and generalisable, for deployment across datasets without any adaptation. In this paper, we develop novel CNN architectures to address both challenges. First, we present a re-ID CNN termed omni-scale network (OSNet) to learn features that not only capture different spatial scales but also encapsulate a synergistic combination of multiple scales, namely omni-scale features. The basic building block consists of multiple convolutional streams, each detecting features at a certain scale. For omni-scale feature learning, a unified aggregation gate is introduced to dynamically fuse multi-scale features with channel-wise weights. OSNet is lightweight as its building blocks comprise factorised convolutions. Second, to improve generalisable feature learning, we introduce instance normalisation (IN) layers into OSNet to cope with cross-dataset discrepancies. Further, to determine the optimal placements of these IN layers in the architecture, we formulate an efficient differentiable architecture search algorithm. Extensive experiments show that, in the conventional same-dataset setting, OSNet achieves state-of-the-art performance, despite being much smaller than existing re-ID models. In the more challenging yet practical cross-dataset setting, OSNet beats most recent unsupervised domain adaptation methods without using any target data. Our code and models are released at \texttt{https://github.com/KaiyangZhou/deep-person-reid}.
['Andrea Cavallaro', 'Yongxin Yang', 'Tao Xiang', 'Kaiyang Zhou']
2019-10-15
null
null
null
null
['unsupervised-person-re-identification']
['computer-vision']
[-2.40823016e-01 -4.62574542e-01 7.68525293e-03 -7.42796183e-01 -5.46011567e-01 -6.23181522e-01 6.38518572e-01 -1.31608322e-01 -6.97556198e-01 5.74217379e-01 2.66990602e-01 1.16635725e-01 -2.14820087e-01 -7.11896002e-01 -6.46240234e-01 -2.52420068e-01 -2.10877389e-01 4.57376122e-01 -1.01531891e-03 -2.96879619e-01 -2.35837653e-01 7.13957787e-01 -1.60683632e+00 6.29561394e-02 8.07951331e-01 8.95808876e-01 -1.05706334e-01 6.58500195e-01 9.76660252e-02 7.82472789e-02 -7.18660176e-01 -8.03437650e-01 6.07764781e-01 -5.03596663e-02 -7.23589182e-01 -3.03406090e-01 9.40136790e-01 -4.83304888e-01 -6.98428631e-01 1.02011061e+00 8.94135177e-01 2.27355465e-01 3.73159587e-01 -1.40027273e+00 -8.97812128e-01 4.37181175e-01 -4.10477012e-01 3.85804594e-01 2.51727611e-01 2.26200610e-01 7.51029551e-01 -7.66544342e-01 2.76297659e-01 1.35166562e+00 1.07497394e+00 8.82082820e-01 -1.10161889e+00 -1.12192547e+00 3.41364920e-01 1.99105754e-01 -1.59754896e+00 -4.54918712e-01 5.59484541e-01 -1.80682093e-01 1.01793563e+00 1.73148453e-01 5.28853416e-01 1.23643756e+00 -1.82429135e-01 8.31806421e-01 7.53140569e-01 -1.61604565e-02 -2.04890534e-01 -2.82187387e-02 1.50383055e-01 4.94986534e-01 4.56313759e-01 1.60349146e-01 -6.41924202e-01 -1.02167428e-01 8.17088068e-01 3.33309650e-01 -3.99375707e-02 -1.66877508e-01 -1.05693638e+00 6.22861326e-01 8.04128706e-01 3.01940531e-01 -1.35753661e-01 2.54691750e-01 4.58291918e-01 4.45712060e-01 3.94904763e-01 3.57060194e-01 -7.10951269e-01 -5.50665073e-02 -7.99921453e-01 5.68315148e-01 4.46661651e-01 1.04728103e+00 8.07763577e-01 -1.33270502e-01 -2.22275153e-01 9.51159716e-01 6.10766411e-02 5.46064794e-01 6.02737188e-01 -5.79562843e-01 5.06958306e-01 7.57000625e-01 1.15240589e-01 -7.50156403e-01 -8.66991699e-01 -7.43004382e-01 -1.01834166e+00 -1.01460638e-02 4.79594857e-01 -2.92286098e-01 -9.53605473e-01 2.01558876e+00 3.53944868e-01 4.61465359e-01 -9.92483571e-02 1.00433302e+00 1.11190808e+00 1.37260243e-01 2.58542836e-01 6.85776234e-01 1.66417050e+00 -9.12224352e-01 -1.98584542e-01 -3.29294920e-01 4.98559296e-01 -4.52916980e-01 6.96745574e-01 -6.70332313e-02 -9.64537621e-01 -9.97238815e-01 -1.12013507e+00 -2.29675829e-01 -8.34592760e-01 3.48326772e-01 6.35108769e-01 7.39308834e-01 -1.28335285e+00 4.97702599e-01 -4.66070741e-01 -6.76736951e-01 5.30746341e-01 6.69789493e-01 -6.65101886e-01 -5.55940829e-02 -1.44574642e+00 7.67499089e-01 2.16482028e-01 1.13527536e-01 -6.21493399e-01 -9.26907003e-01 -9.93621349e-01 1.55298755e-01 -1.25458598e-01 -9.63708341e-01 1.04246855e+00 -1.07330441e+00 -1.15850914e+00 9.77867663e-01 -3.33113998e-01 -4.48645055e-01 5.29447973e-01 -2.99097985e-01 -8.35511804e-01 -5.01874723e-02 3.95213962e-01 9.03386414e-01 8.16348016e-01 -9.99307752e-01 -8.23803127e-01 -5.27710915e-01 2.20683768e-01 1.34928867e-01 -5.89216411e-01 2.31545597e-01 -7.24620223e-01 -7.53195047e-01 -4.36777145e-01 -8.78460884e-01 -1.60823151e-01 5.34349196e-02 -3.67961884e-01 -4.38165247e-01 5.45935988e-01 -6.34508550e-01 1.13133848e+00 -2.21141124e+00 -1.65199280e-01 3.31686199e-01 3.63621533e-01 4.35528547e-01 -4.61317033e-01 1.47889480e-01 -2.55718380e-01 2.35274136e-02 7.42714014e-03 -7.75685787e-01 1.28691658e-01 -8.07600021e-02 1.24648355e-01 5.70192695e-01 3.28400224e-01 1.24657214e+00 -7.62531102e-01 -1.18994594e-01 2.18613937e-01 5.74635983e-01 -4.26442832e-01 1.06972665e-01 4.55904156e-01 4.56123382e-01 -2.73692578e-01 7.14256525e-01 1.01676476e+00 -2.41761431e-01 -7.03259706e-02 -1.89481303e-01 -1.03433639e-01 1.23579301e-01 -1.31344092e+00 1.75462830e+00 -3.83459121e-01 4.54442948e-01 7.72403181e-02 -9.60668802e-01 8.42764556e-01 -3.79456393e-02 4.68975723e-01 -8.69748235e-01 8.20082501e-02 2.06314698e-01 -2.63860971e-01 -5.18269055e-02 5.13255477e-01 2.77382463e-01 -3.86195183e-01 2.46515095e-01 3.51009667e-01 7.20436215e-01 1.36722445e-01 6.90988451e-03 9.19548631e-01 -1.40034273e-01 1.56767607e-01 -3.02998751e-01 6.49282336e-01 -3.32385749e-01 5.90445161e-01 1.01919162e+00 -4.44625318e-01 7.56441653e-01 -9.86441877e-03 -8.42581928e-01 -9.73627746e-01 -1.03729653e+00 -2.83309788e-01 1.33005691e+00 3.76258612e-01 -3.36117923e-01 -6.36699677e-01 -7.49718606e-01 5.19401491e-01 4.59879227e-02 -9.08519030e-01 -5.39044514e-02 -7.27432370e-01 -7.74373412e-01 9.12409782e-01 9.13908362e-01 9.49425936e-01 -7.43994772e-01 -3.52379143e-01 1.72922835e-01 -5.91555163e-02 -1.06046498e+00 -9.80639935e-01 -1.72590632e-02 -3.84209156e-01 -1.11498344e+00 -1.00133145e+00 -8.59186232e-01 5.42608976e-01 5.55486739e-01 1.13726377e+00 2.34087661e-01 -3.62133622e-01 4.94668871e-01 -2.25999832e-01 -3.79982740e-01 3.14617842e-01 4.77924198e-01 3.11262369e-01 1.93351239e-01 8.61560941e-01 -5.22046089e-01 -9.40493464e-01 5.03793478e-01 -6.92207336e-01 -2.78041393e-01 4.56099242e-01 9.24610078e-01 2.15568885e-01 -1.65247560e-01 6.68272734e-01 -3.91648591e-01 5.44696152e-01 -4.69665527e-01 -3.12084317e-01 2.62282401e-01 -3.46564919e-01 -1.60247907e-01 6.67003036e-01 -4.67183262e-01 -8.57814610e-01 -8.48949030e-02 -2.41091013e-01 -2.56279200e-01 -3.96955848e-01 5.58226816e-02 -3.60669255e-01 -1.96884930e-01 6.46142304e-01 1.84205323e-01 -4.35405113e-02 -6.66845560e-01 3.49621356e-01 6.65928423e-01 7.66236603e-01 -5.73044479e-01 1.09647024e+00 6.14965796e-01 -3.78776520e-01 -4.82705742e-01 -6.18637502e-01 -7.57161856e-01 -8.38543892e-01 1.82193406e-02 8.53272855e-01 -1.32279778e+00 -8.77554893e-01 7.95297325e-01 -9.45897520e-01 -3.34671229e-01 -1.28223315e-01 3.75572264e-01 -8.78438726e-02 1.45204693e-01 -4.92735356e-01 -3.42526615e-01 -4.77648407e-01 -8.60336244e-01 1.14002502e+00 7.76807427e-01 -1.76466703e-01 -1.02391028e+00 -1.12177983e-01 2.42450342e-01 6.57335460e-01 -1.13371322e-02 2.59226292e-01 -9.54569876e-01 -2.90688664e-01 -3.11076045e-01 -6.23320103e-01 7.76155069e-02 1.20904468e-01 -4.89175707e-01 -1.08596539e+00 -7.99890161e-01 -7.79212415e-01 -2.28246018e-01 9.84602153e-01 3.23456496e-01 1.30610228e+00 -1.48315534e-01 -4.88005072e-01 1.22759342e+00 1.20658255e+00 -2.37806231e-01 5.12453973e-01 8.10571134e-01 7.58162558e-01 4.27844256e-01 1.88776031e-01 5.28205574e-01 8.75830114e-01 9.04936790e-01 1.43182993e-01 -4.38761413e-01 -3.70868295e-01 -2.74959922e-01 1.37994677e-01 -1.78354681e-02 -1.96917117e-01 -3.88379535e-03 -7.20125437e-01 7.64293432e-01 -1.85827792e+00 -1.11249232e+00 2.34690711e-01 2.07084489e+00 5.72913349e-01 -1.66284800e-01 6.22815728e-01 -3.87146443e-01 8.45914125e-01 5.22354916e-02 -8.50445330e-01 -1.53651178e-01 -2.01907352e-01 3.23233157e-01 8.54951620e-01 2.34751940e-01 -1.50519848e+00 9.22762632e-01 5.64033413e+00 6.48455262e-01 -1.10248160e+00 2.72627413e-01 3.58985037e-01 -2.03394219e-01 -9.46009085e-02 -4.39874917e-01 -1.20028973e+00 5.97273707e-01 7.88022637e-01 8.71423706e-02 4.42701936e-01 8.45484912e-01 -1.45966277e-01 4.45129067e-01 -9.59886909e-01 1.27118194e+00 1.06859155e-01 -1.16304874e+00 -1.34652555e-01 -1.41951926e-02 5.93605578e-01 2.48093963e-01 2.19637766e-01 4.92086589e-01 5.35643816e-01 -1.18369043e+00 7.91009367e-01 3.37090999e-01 1.08791399e+00 -9.53802824e-01 8.62726390e-01 -1.48092031e-01 -1.72033811e+00 -3.78952503e-01 -3.88566136e-01 1.03031561e-01 8.73560533e-02 2.13493422e-01 -2.81519741e-01 6.73677027e-01 1.28719699e+00 7.92597353e-01 -9.57201123e-01 1.14830399e+00 5.58639988e-02 8.23610574e-02 -4.25021768e-01 2.79834002e-01 1.35814145e-01 2.96003103e-01 2.54144102e-01 1.59739470e+00 3.71849507e-01 -1.64764673e-01 1.39146551e-01 7.06361890e-01 -2.45803729e-01 -2.23165691e-01 -3.18567187e-01 5.85184395e-01 6.69776917e-01 1.25331378e+00 -2.34613970e-01 -3.09305936e-01 -6.77342892e-01 1.29888546e+00 5.21885395e-01 5.56077957e-01 -8.30041945e-01 -4.83331174e-01 1.43752646e+00 5.80350608e-02 3.99433255e-01 -2.08406717e-01 -3.47645342e-01 -1.40169573e+00 1.83189765e-01 -7.02556133e-01 7.19083428e-01 -2.04753399e-01 -1.84363842e+00 5.67215919e-01 3.85157727e-02 -1.19916022e+00 -8.61078799e-02 -8.07760119e-01 -6.79953456e-01 1.21076500e+00 -1.93827903e+00 -1.81083739e+00 -5.82161009e-01 1.07131720e+00 3.89686853e-01 -4.71021563e-01 7.78692961e-01 5.59279203e-01 -6.48500383e-01 1.47964060e+00 5.98355383e-02 6.93911314e-01 1.02550387e+00 -1.18296802e+00 1.00505841e+00 9.53180075e-01 -2.52170444e-01 1.06373513e+00 1.92614526e-01 -5.87879539e-01 -1.21221435e+00 -1.32113290e+00 9.37534928e-01 -4.56099451e-01 4.31080103e-01 -4.93329436e-01 -7.42397606e-01 7.91907847e-01 -1.77146509e-01 3.39146852e-01 8.31269383e-01 4.22969699e-01 -7.87514210e-01 -3.85845870e-01 -1.31459403e+00 4.36880887e-01 1.56537580e+00 -5.81832647e-01 -3.05064350e-01 1.34732574e-02 3.48513603e-01 -3.53082478e-01 -9.19842243e-01 2.94716060e-01 9.02814865e-01 -8.66111994e-01 1.49205029e+00 -7.02828884e-01 -1.29688755e-01 -3.43617052e-01 4.70558889e-02 -1.28061366e+00 -7.74556875e-01 -5.01336515e-01 -3.34702199e-03 1.23301589e+00 2.61894315e-01 -1.16953373e+00 7.02561140e-01 7.35960960e-01 -2.93774344e-03 -4.70728666e-01 -1.08004570e+00 -9.76758838e-01 2.00877592e-01 -2.28783116e-01 1.19158113e+00 9.54436660e-01 -2.68266678e-01 1.38655743e-02 -4.71796483e-01 4.33379382e-01 8.45073104e-01 -1.12152368e-01 1.05145288e+00 -1.41978908e+00 -1.27186731e-01 -6.41895115e-01 -6.86028421e-01 -1.28429914e+00 2.68415809e-01 -1.05987740e+00 -3.94880652e-01 -1.28985775e+00 3.48119974e-01 -7.33078241e-01 -5.58859229e-01 6.74371898e-01 -4.13736254e-01 5.19263446e-01 3.53506058e-01 2.88348317e-01 -6.29011989e-01 3.48352998e-01 8.71348202e-01 -2.81297982e-01 -1.44756690e-01 -1.07800514e-02 -1.08085680e+00 3.77090245e-01 8.72911811e-01 -1.43778652e-01 -1.32012516e-01 -6.94180787e-01 -1.37353659e-01 -6.51071548e-01 7.96472549e-01 -1.12498307e+00 4.41028595e-01 1.58725530e-01 1.10154700e+00 -4.43609715e-01 2.50686884e-01 -6.27136707e-01 1.29617944e-01 1.82397619e-01 -2.76017636e-01 3.21984917e-01 2.67212272e-01 3.36222053e-01 -2.51950715e-02 4.35029417e-02 8.10243428e-01 -7.00870231e-02 -9.82108831e-01 7.84975171e-01 1.13852724e-01 -9.48222131e-02 7.85316110e-01 -3.86179149e-01 -4.04060423e-01 -2.50563890e-01 -4.75665987e-01 5.76991379e-01 6.18406713e-01 7.74322927e-01 4.98055160e-01 -1.52855837e+00 -8.63634706e-01 4.78308737e-01 3.83986950e-01 -9.53222141e-02 6.25151873e-01 5.00671864e-01 -1.97181687e-01 4.85304564e-01 -4.14223462e-01 -3.72404546e-01 -1.08578217e+00 4.02998596e-01 5.33442557e-01 -2.64634550e-01 -6.24777615e-01 1.02461612e+00 3.25099826e-01 -9.22748625e-01 1.79979414e-01 7.92224556e-02 -2.15252861e-01 -5.61970100e-02 9.95454431e-01 2.70036608e-01 1.72020979e-02 -8.52467895e-01 -5.86918473e-01 6.83396578e-01 -2.34188020e-01 2.28523523e-01 1.28009486e+00 -1.78100467e-01 1.44252181e-01 -2.37443924e-01 1.39352262e+00 -2.50543982e-01 -1.46841729e+00 -4.88129407e-01 -1.47307694e-01 -4.82100427e-01 -3.54351670e-01 -7.83042490e-01 -1.03727019e+00 4.78553593e-01 9.72113788e-01 -7.04447106e-02 1.01847184e+00 1.12448186e-01 1.01379311e+00 1.45497069e-01 3.21002901e-01 -1.16388798e+00 -1.72388032e-01 5.04086375e-01 7.19439209e-01 -1.47815239e+00 -1.49036169e-01 2.38702781e-02 -3.78487200e-01 1.12021315e+00 9.37103271e-01 -1.30297646e-01 7.28026152e-01 1.51308134e-01 1.31992921e-01 -1.59721792e-01 -1.95741192e-01 -5.95515907e-01 4.45602179e-01 9.38549697e-01 2.13852078e-01 2.68017322e-01 1.83988865e-02 8.42328668e-01 -4.04712677e-01 -9.40623581e-02 -8.39434341e-02 6.21889353e-01 -8.78813416e-02 -1.13589668e+00 -3.67576808e-01 3.87114108e-01 -3.61256987e-01 9.08895803e-04 -1.97958320e-01 8.19221497e-01 4.86561775e-01 8.27330947e-01 3.98012847e-01 -5.49695730e-01 5.74152708e-01 -1.43480361e-01 2.98527420e-01 -2.15052083e-01 -8.29620361e-01 -5.55182993e-01 1.48035828e-02 -6.55410051e-01 -4.26001489e-01 -6.89117730e-01 -9.70475733e-01 -6.22703969e-01 5.18888049e-02 -1.45595074e-01 5.87167919e-01 7.96581209e-01 6.94682777e-01 3.13268751e-01 4.70744550e-01 -1.15522802e+00 -4.36454207e-01 -1.04489613e+00 -4.34312522e-01 6.78205609e-01 4.77750301e-01 -7.76217103e-01 -3.03804260e-02 -2.66621143e-01]
[14.702816009521484, 0.9752671718597412]
2f2564db-7e70-432b-a478-1f88db9adb82
refer-itts-a-system-for-referring-in-spoken
null
null
https://aclanthology.org/W17-3509
https://aclanthology.org/W17-3509.pdf
Refer-iTTS: A System for Referring in Spoken Installments to Objects in Real-World Images
Current referring expression generation systems mostly deliver their output as one-shot, written expressions. We present on-going work on incremental generation of spoken expressions referring to objects in real-world images. This approach extends upon previous work using the words-as-classifier model for generation. We implement this generator in an incremental dialogue processing framework such that we can exploit an existing interface to incremental text-to-speech synthesis. Our system generates and synthesizes referring expressions while continuously observing non-verbal user reactions.
["M. Soledad L{\\'o}pez Gambino", 'Sina Zarrie{\\ss}', 'David Schlangen']
2017-09-01
null
null
null
ws-2017-9
['referring-expression-generation']
['computer-vision']
[ 4.19733405e-01 7.27003038e-01 6.07459173e-02 -8.05811524e-01 -1.18342233e+00 -5.84482551e-01 1.20429015e+00 -2.16336846e-01 -4.74591069e-02 7.59467781e-01 7.64917731e-01 -2.64467597e-02 5.36502421e-01 -6.30361676e-01 -1.68792158e-01 -3.51726934e-02 4.12374586e-01 6.96864843e-01 5.32725081e-03 -8.24302197e-01 1.33053511e-01 4.65904146e-01 -1.57661450e+00 8.14922094e-01 6.12358339e-02 6.27690613e-01 -2.07468688e-01 1.49956334e+00 -5.22215426e-01 1.66942120e+00 -1.22803056e+00 -4.59574848e-01 -1.84353307e-01 -1.11878622e+00 -1.28537798e+00 4.54173028e-01 1.75556481e-01 -3.87483954e-01 4.25325669e-02 7.18890071e-01 7.36722291e-01 5.49546182e-01 6.00420833e-01 -1.37240660e+00 -9.95745718e-01 7.58189559e-01 3.62914205e-01 -7.37328231e-02 1.42487597e+00 2.98629373e-01 6.08957171e-01 -7.68702090e-01 1.02091682e+00 1.73381662e+00 1.05035119e-01 1.21362793e+00 -1.12622523e+00 -2.43387699e-01 5.54412715e-02 -1.40946329e-01 -1.25644076e+00 -9.66592789e-01 6.46471441e-01 -2.58967429e-01 1.43355811e+00 5.95569968e-01 4.30366606e-01 1.47335756e+00 -1.98252574e-01 1.08344865e+00 1.13649368e+00 -9.03497815e-01 2.04272777e-01 3.58091593e-01 -2.11081401e-01 7.13361979e-01 -1.21450567e+00 -1.98198795e-01 -7.68766999e-01 -1.42806113e-01 7.62197971e-01 -7.10430741e-01 -6.02679811e-02 4.57923979e-01 -1.31334996e+00 8.16214442e-01 1.14503652e-01 2.60687768e-01 -5.74707627e-01 3.35247219e-01 5.21787107e-01 6.31850600e-01 9.18843567e-01 7.01947033e-01 1.02422051e-01 -8.01650703e-01 -8.21016192e-01 6.56301677e-01 1.19527245e+00 1.45567226e+00 3.96534681e-01 2.74793178e-01 -7.66514540e-01 9.87232089e-01 6.20312840e-02 3.28385085e-01 5.95914245e-01 -1.44896090e+00 -1.24631673e-01 3.73062313e-01 4.90772963e-01 -5.64715266e-01 -9.98077765e-02 4.25808817e-01 -1.03141077e-01 7.57652670e-02 -1.19687781e-01 -5.39876699e-01 -6.24624133e-01 1.62217939e+00 2.48857260e-01 -2.23738909e-01 4.73137885e-01 8.54161322e-01 1.44481349e+00 9.03223455e-01 2.66278774e-01 -3.50051582e-01 1.42200160e+00 -1.14258087e+00 -1.19643545e+00 5.81795163e-02 6.68840587e-01 -8.77084136e-01 1.26967716e+00 4.15485293e-01 -1.49773955e+00 -5.89165807e-01 -6.44827366e-01 -3.91198725e-01 -4.96259660e-01 1.51137680e-01 4.96670246e-01 3.40057075e-01 -1.47793090e+00 1.68502077e-01 -4.76866871e-01 -6.25892460e-01 -1.43589243e-01 1.43212572e-01 -2.04900458e-01 5.25683284e-01 -1.21404672e+00 1.08743918e+00 3.56046259e-02 -3.40357572e-01 -5.33544958e-01 -2.61935472e-01 -1.05888748e+00 -5.47310174e-01 3.75770390e-01 -7.62485147e-01 2.49135208e+00 -1.38899946e+00 -2.52478909e+00 1.23595917e+00 -5.40683687e-01 -2.91091204e-01 3.88737738e-01 -2.33997077e-01 -3.85642171e-01 4.04853225e-01 2.17792019e-02 1.30560577e+00 5.93334377e-01 -1.29632604e+00 -3.70771646e-01 -7.43122920e-02 4.25781429e-01 5.25772154e-01 2.76720792e-01 8.12627673e-01 -1.06723353e-01 -6.75108671e-01 -2.67310172e-01 -8.46860468e-01 -1.45898089e-01 -7.62408152e-02 -3.74034405e-01 -7.95362353e-01 8.34278286e-01 -1.69240788e-01 9.57708597e-01 -1.60582769e+00 2.47006029e-01 -4.38071251e-01 -1.58342794e-01 1.66521773e-01 -1.91532627e-01 8.42413902e-01 -2.02698946e-01 1.43433332e-01 1.97620302e-01 -6.68919921e-01 2.57629156e-01 4.60050076e-01 -5.86031139e-01 -1.23424426e-01 6.57844007e-01 1.19171548e+00 -1.18793190e+00 -7.33418584e-01 3.21306527e-01 2.99635768e-01 -1.24412932e-01 9.97365832e-01 -6.68132842e-01 2.61533558e-01 -5.81802189e-01 3.96650791e-01 1.49543909e-02 2.21382290e-01 -3.54471385e-01 2.26856530e-01 -3.56757581e-01 3.39884013e-01 -6.47050977e-01 1.79857314e+00 -9.42469239e-01 7.11146891e-01 3.72938626e-02 -6.25530183e-01 1.17046309e+00 8.91323268e-01 -5.34675196e-02 -4.54768568e-01 3.15147221e-01 -9.30224732e-03 -4.09235448e-01 -9.08924878e-01 7.88282633e-01 -5.73785484e-01 -5.42426765e-01 9.59734976e-01 4.17126596e-01 -1.14161098e+00 1.53956935e-01 4.06407714e-01 8.85876417e-01 6.28135026e-01 7.48051286e-01 2.25092515e-01 7.10788906e-01 2.20247358e-01 -5.77215791e-01 7.28949904e-01 -4.55172807e-02 6.42133057e-01 5.67673385e-01 -4.33642328e-01 -7.70515800e-01 -8.04347754e-01 4.24377024e-01 1.63215768e+00 -4.57434714e-01 -3.09587598e-01 -1.10162449e+00 -3.56210381e-01 -7.98523545e-01 1.34846151e+00 -5.56000054e-01 1.63650215e-01 -4.15222287e-01 9.50122699e-02 7.93039739e-01 4.27956909e-01 -1.54990889e-02 -2.04613423e+00 -1.01536548e+00 4.56319183e-01 -1.52197510e-01 -1.27213049e+00 -2.31607839e-01 2.33420376e-02 -3.26082110e-01 -3.90254706e-01 -7.89072633e-01 -7.78773308e-01 4.94960934e-01 -2.22065881e-01 1.49303567e+00 -2.45781671e-02 -1.78563803e-01 9.33722615e-01 -9.04590726e-01 -9.35349286e-01 -1.28808773e+00 -1.71232507e-01 -1.57861620e-01 -9.27996859e-02 3.96596879e-01 -1.56830698e-01 -1.99308340e-02 9.86688137e-02 -8.11004639e-01 4.77069318e-01 -1.10563850e-02 6.08545303e-01 2.67798543e-01 -1.00696778e+00 7.25628257e-01 -7.34708190e-01 1.57479680e+00 -2.72140622e-01 1.13702249e-02 2.74238765e-01 1.46788359e-01 1.12864681e-01 6.45560563e-01 -5.34003019e-01 -1.43025589e+00 2.98568636e-01 -5.10842621e-01 -3.00958276e-01 -8.64954591e-01 1.98304325e-01 1.94658399e-01 3.20799559e-01 8.29454124e-01 2.88750231e-01 1.47681341e-01 2.27550074e-01 1.19001818e+00 1.17230093e+00 9.37339842e-01 -9.18653667e-01 2.56023735e-01 1.41967908e-01 -4.37615007e-01 -1.03190315e+00 -8.07878196e-01 -4.81004179e-01 -5.21863103e-01 -6.51914597e-01 7.98927963e-01 -8.17321241e-01 -8.39774430e-01 9.54453871e-02 -1.74286187e+00 -8.20562005e-01 -6.98792696e-01 3.37907262e-02 -1.27182508e+00 7.65867457e-02 -6.58204079e-01 -1.49238181e+00 -5.39248705e-01 -8.77655029e-01 1.70656049e+00 3.42584252e-01 -1.23976803e+00 -8.71867657e-01 1.69615120e-01 -2.80537978e-02 6.80594563e-01 2.16952875e-01 3.34682792e-01 -8.25296283e-01 -9.10016056e-03 -2.55351990e-01 -1.39670465e-02 -1.91047844e-02 1.05860256e-01 2.71882802e-01 -1.06545389e+00 5.86419880e-01 4.19759145e-03 -1.11012959e+00 1.30028948e-01 -1.31213292e-01 6.87916219e-01 -5.98852992e-01 1.61604568e-01 1.62041411e-01 6.59559548e-01 4.21704590e-01 4.91452307e-01 -1.81903288e-01 1.94085181e-01 1.12985814e+00 7.43347168e-01 4.87890035e-01 4.69085395e-01 9.08560872e-01 -2.98035424e-02 -2.03083724e-01 -5.72279282e-02 -1.82319164e-01 5.27287483e-01 3.91156346e-01 3.51733565e-02 -6.06487870e-01 -6.97030783e-01 5.80375075e-01 -1.88120770e+00 -1.40635514e+00 -1.32337138e-01 1.28952646e+00 1.22012949e+00 -2.45567769e-01 1.33651927e-01 -3.58066976e-01 3.15385371e-01 2.38327324e-01 -9.71465334e-02 -1.37341762e+00 1.00759484e-01 7.06731677e-01 -3.71078491e-01 7.85190940e-01 -8.39967608e-01 1.51050103e+00 6.86733150e+00 3.15781116e-01 -1.28954744e+00 8.67076889e-02 5.67508757e-01 -1.71157897e-01 -1.37341633e-01 -2.87648529e-01 -5.16719878e-01 -3.58669400e-01 1.39743173e+00 -5.04174888e-01 3.12856615e-01 9.46585536e-01 6.17931306e-01 -1.89024940e-01 -1.33745611e+00 8.31373692e-01 6.43474400e-01 -1.12760997e+00 1.60015583e-01 -7.10225761e-01 3.80265117e-01 -3.25726032e-01 -2.86462456e-01 4.05389696e-01 6.28129125e-01 -9.98465180e-01 9.35887992e-01 7.69433796e-01 1.03172827e+00 -5.88440835e-01 6.85069144e-01 5.17307281e-01 -7.42631376e-01 4.75218475e-01 2.26194590e-01 -4.85889465e-01 7.61011243e-01 -2.99397945e-01 -1.39227819e+00 1.33350745e-01 7.65756667e-02 1.49632365e-01 -4.31461066e-01 -3.66112702e-02 -5.70434570e-01 4.21959817e-01 -2.28321791e-01 -6.68819904e-01 4.10991371e-01 4.22926806e-02 4.25362170e-01 1.83725548e+00 2.12969482e-01 7.44988978e-01 3.66978347e-01 1.07837296e+00 1.55747473e-01 4.48339224e-01 -1.00239849e+00 -1.48862749e-01 2.60471553e-01 1.45171571e+00 -6.23447716e-01 -1.08762753e+00 -2.95320600e-01 1.41061246e+00 -2.14969814e-02 2.43585587e-01 -5.09196758e-01 -4.41314936e-01 2.19706833e-01 -6.60355538e-02 -2.02506900e-01 1.64744686e-02 2.74129063e-01 -7.93089449e-01 -3.60830128e-01 -1.19544256e+00 -2.08309945e-02 -1.70048022e+00 -1.21837533e+00 1.07293522e+00 3.96024734e-01 -7.81417191e-01 -1.33809185e+00 -4.29117978e-01 -8.27239215e-01 1.05754924e+00 -7.45585382e-01 -1.43845272e+00 -5.08537889e-01 4.34345812e-01 1.37352991e+00 -1.43964842e-01 1.63934362e+00 -5.30069053e-01 -2.33869419e-01 2.36571774e-01 -1.26755714e+00 1.13551527e-01 8.26081693e-01 -1.39456975e+00 7.09361136e-01 4.40296590e-01 4.15956557e-01 7.49389350e-01 1.06990027e+00 -3.89828205e-01 -1.04409719e+00 -7.51376331e-01 1.04742706e+00 -5.73122680e-01 7.72743046e-01 -4.15845424e-01 -7.98305392e-01 7.17243016e-01 1.10668993e+00 -1.69144690e-01 8.15212607e-01 -3.33082497e-01 -5.17259538e-02 6.74466431e-01 -9.25746202e-01 9.98239934e-01 9.28674161e-01 -8.28038454e-01 -9.58727717e-01 6.70423448e-01 8.57105196e-01 -8.12776983e-01 -5.98563969e-01 -1.70547619e-01 4.05739218e-01 -7.72288322e-01 5.90955198e-01 -1.00870287e+00 7.57337511e-01 -1.88444018e-01 -1.59048870e-01 -1.34503043e+00 4.77323860e-01 -1.51558709e+00 1.61717311e-01 1.31273293e+00 3.41190606e-01 -1.87701464e-01 5.05824685e-01 9.80877042e-01 -2.42626563e-01 -4.17072326e-01 -6.32410467e-01 -1.95752606e-01 -2.80542076e-01 -6.33486867e-01 4.14664984e-01 5.93735814e-01 7.09086657e-01 9.67271268e-01 -2.73002207e-01 -4.94361192e-01 -2.47365624e-01 -1.78182319e-01 1.31937671e+00 -5.86911142e-01 -2.68489391e-01 -4.04356599e-01 -3.74051601e-01 -1.23160934e+00 7.19363213e-01 -7.06852078e-01 6.50092423e-01 -1.35142076e+00 -2.29933441e-01 4.49482977e-01 5.85008681e-01 5.44351280e-01 -1.13713905e-01 1.30614161e-01 3.17409635e-01 -2.16833130e-01 -8.47121596e-01 6.17981136e-01 1.03824461e+00 1.39452204e-01 -2.70029813e-01 -2.58667115e-02 -5.01987278e-01 7.31198907e-01 6.27793431e-01 -2.08908081e-01 -4.78358686e-01 -9.09706652e-02 -6.94851801e-02 6.45620286e-01 3.57805908e-01 -6.62359238e-01 2.58920252e-01 -3.06657821e-01 -9.61913988e-02 -5.08574128e-01 6.20533824e-01 -1.15384601e-01 -7.17623532e-02 -1.19940728e-01 -1.03595722e+00 4.37016964e-01 1.00990675e-01 -1.04085952e-01 -4.93804544e-01 -5.15225649e-01 4.66331035e-01 -7.35446870e-01 -6.33053064e-01 -3.42601866e-01 -1.05358148e+00 7.36651197e-03 1.01284504e+00 -2.02918217e-01 -2.76046932e-01 -1.50076866e+00 -9.57717121e-01 4.08563502e-02 2.57365376e-01 6.01032972e-01 9.20485854e-01 -1.17015266e+00 -8.83368313e-01 -9.81275663e-02 3.51988614e-01 -1.62832588e-01 -2.75853783e-01 2.22192243e-01 -5.16588211e-01 3.95868659e-01 -3.26585062e-02 -4.76177633e-01 -1.47543061e+00 4.48357761e-01 4.15902466e-01 1.42339647e-01 -4.06508118e-01 1.11467731e+00 -3.08530718e-01 -3.67345184e-01 -3.56372111e-02 -2.18281895e-01 -3.35492790e-01 2.98010945e-01 9.00175512e-01 -1.63415506e-01 -1.87417895e-01 -9.73072946e-01 9.21534896e-02 2.53803879e-01 2.26031393e-01 -1.05009377e+00 9.75962579e-01 -1.68887928e-01 -2.61928793e-02 1.16447675e+00 1.01395178e+00 2.58584581e-02 -7.48866141e-01 -5.58393821e-02 -4.15809005e-02 1.20735034e-01 -3.89987707e-01 -6.15580857e-01 -1.97790012e-01 8.38444948e-01 1.18029095e-01 4.69575644e-01 9.08790648e-01 3.43170851e-01 5.14589429e-01 7.31120348e-01 3.39800686e-01 -1.10663736e+00 4.92981404e-01 6.20750666e-01 1.63203025e+00 -1.10177195e+00 9.18674096e-02 -2.29062304e-01 -1.26746023e+00 1.53968477e+00 6.79750979e-01 -8.84246826e-02 1.26822606e-01 6.03924751e-01 9.45463359e-01 -4.91891414e-01 -1.54426503e+00 -3.28732818e-01 -6.92284480e-02 7.68614650e-01 1.13235128e+00 1.92695469e-01 -1.87955230e-01 3.05928349e-01 -6.80695117e-01 2.18466088e-01 9.63836133e-01 9.70433056e-01 -1.72572732e-01 -1.12576509e+00 -3.21365952e-01 -4.35315557e-02 -4.13817137e-01 -1.43382311e-01 -1.40639448e+00 6.89594030e-01 -3.51747483e-01 1.13923371e+00 3.17896187e-01 -2.87184604e-02 7.99010456e-01 6.17489517e-01 5.66467822e-01 -1.32723391e+00 -9.50400233e-01 -9.00942385e-02 7.98878610e-01 -6.58351302e-01 -8.35704744e-01 -5.90885758e-01 -1.45183861e+00 2.99627542e-01 -1.28452227e-01 1.49721026e-01 4.78724629e-01 8.90124500e-01 1.77789330e-01 4.71894294e-01 5.65762877e-01 -1.29159701e+00 -5.10848165e-01 -1.38957942e+00 8.84936750e-02 6.49833143e-01 1.10900857e-01 -1.46516740e-01 -1.40688820e-02 5.56717515e-01]
[13.06035041809082, 7.686365604400635]
0db96baa-fd54-48fd-a891-76d1907d1f46
eica-team-at-semeval-2017-task-3-semantic-and
null
null
https://aclanthology.org/S17-2047
https://aclanthology.org/S17-2047.pdf
EICA Team at SemEval-2017 Task 3: Semantic and Metadata-based Features for Community Question Answering
We describe our system for participating in SemEval-2017 Task 3 on Community Question Answering. Our approach relies on combining a rich set of various types of features: semantic and metadata. The most important group turned out to be the metadata feature and the semantic vectors trained on QatarLiving data. In the main Subtask C, our primary submission was ranked fourth, with a MAP of 13.48 and accuracy of 97.08. In Subtask A, our primary submission get into the top 50{\%}.
['Jian Jiang', 'Yufei Xie', 'Maoquan Wang', 'Zhao Lu', 'Jing Ma']
2017-08-01
null
null
null
semeval-2017-8
['question-similarity']
['natural-language-processing']
[-3.17372799e-01 -8.24790541e-03 -6.08530305e-02 -2.85938025e-01 -1.22717857e+00 -6.06968045e-01 8.26988161e-01 4.45272386e-01 -7.12176025e-01 8.36228788e-01 7.09480643e-01 -8.62956718e-02 -2.33895347e-01 -5.44510007e-01 -5.30242622e-01 -2.14751706e-01 1.72137380e-01 8.92474115e-01 5.22819936e-01 -5.55283010e-01 5.08173227e-01 -3.84940118e-01 -1.54145789e+00 7.00084925e-01 8.98898065e-01 1.31946945e+00 -5.89654744e-02 6.54872119e-01 -4.99157429e-01 8.57988894e-01 -6.90895855e-01 -8.07398498e-01 -2.22831920e-01 -1.64720953e-01 -1.43435180e+00 -7.39251316e-01 9.55042899e-01 2.71661431e-01 -1.82221636e-01 7.95851171e-01 5.60139716e-01 1.27283916e-01 7.67933369e-01 -1.16021228e+00 -8.47268701e-01 5.92867911e-01 1.09653622e-01 6.91594183e-01 8.56442630e-01 -1.02010332e-01 1.60770428e+00 -1.24375165e+00 9.55172241e-01 1.12317812e+00 6.00147903e-01 4.82383519e-01 -7.67779827e-01 -5.45443475e-01 -1.74333021e-01 9.70541000e-01 -1.33558071e+00 -4.68983740e-01 4.91196513e-01 -6.53648376e-01 1.07040429e+00 5.36836267e-01 3.56264174e-01 1.06639910e+00 -3.74107361e-01 8.98016930e-01 1.21925914e+00 -2.76460052e-01 3.50263566e-01 2.53739148e-01 6.38957322e-01 4.81999755e-01 1.38323918e-01 -5.87352395e-01 -7.43588805e-01 -5.63341796e-01 -3.23367827e-02 -3.96035850e-01 -4.38812107e-01 -6.58815503e-02 -1.10795724e+00 1.04479551e+00 4.76817667e-01 3.64350289e-01 -2.16988802e-01 -8.23917612e-03 3.37363124e-01 4.33309257e-01 3.84957224e-01 9.72927928e-01 -8.40018570e-01 -2.16115966e-01 -7.43406057e-01 8.35577846e-01 1.07286894e+00 5.84147453e-01 5.05055070e-01 -7.53088534e-01 -7.44093299e-01 1.00344729e+00 2.91410655e-01 4.73761767e-01 1.15238957e-01 -1.03413975e+00 8.43319416e-01 6.53117299e-01 1.27125174e-01 -7.30843544e-01 -2.82802552e-01 -5.32973826e-01 -5.86532235e-01 -6.42801225e-01 7.45636702e-01 2.60386560e-02 -7.07197011e-01 1.49204421e+00 2.08273664e-01 -1.90707445e-01 1.89002268e-02 9.01515424e-01 1.66545260e+00 4.91243571e-01 2.56875843e-01 2.43756860e-01 1.65153933e+00 -1.11257911e+00 -7.52021849e-01 1.23181894e-01 5.74390352e-01 -8.34976912e-01 9.56997573e-01 1.90388933e-01 -8.47790182e-01 -2.32565016e-01 -6.32869542e-01 -1.64626434e-01 -7.75408566e-01 -1.05596118e-01 5.36402106e-01 4.72377181e-01 -1.32559288e+00 2.90468335e-01 1.13607291e-02 -6.29615843e-01 4.28431153e-01 -5.34241125e-02 -4.14109349e-01 -3.30459088e-01 -1.56394863e+00 1.02539217e+00 2.14873552e-01 -5.13792753e-01 -6.52029037e-01 -1.06718814e+00 -6.29362464e-01 1.47822261e-01 3.08351964e-01 -9.33727682e-01 1.11027479e+00 -3.95135581e-01 -9.40570354e-01 1.19739377e+00 -3.32272321e-01 -6.30315065e-01 1.81527302e-01 -2.43410349e-01 -4.76738274e-01 3.17096323e-01 3.30001056e-01 4.93758470e-01 4.61677551e-01 -7.64361084e-01 -4.82068598e-01 -5.20393014e-01 7.15930983e-02 1.68906704e-01 -4.61925805e-01 5.18361628e-01 -4.19366926e-01 -3.29469413e-01 -8.55260715e-02 -6.53764606e-01 -2.34871563e-02 -5.37311375e-01 -1.19146042e-01 -1.05452168e+00 4.04274553e-01 -1.11805236e+00 1.32229877e+00 -1.46197248e+00 1.41958654e-01 -4.26029973e-02 3.95411551e-01 2.26049766e-01 -7.44333640e-02 7.11788774e-01 4.72133420e-02 2.50916481e-01 1.01395212e-01 -1.71138242e-01 1.83780104e-01 -2.42382705e-01 -5.10671258e-01 4.38542143e-02 1.37189299e-01 1.29820049e+00 -9.35691774e-01 -4.85889912e-01 -1.68614253e-01 2.78982282e-01 -5.31675100e-01 1.23019680e-01 -4.55597699e-01 4.02616829e-01 -6.57357454e-01 7.16752887e-01 4.06421453e-01 -6.39775932e-01 -2.37671912e-01 3.33721712e-02 1.19054690e-01 8.41454506e-01 -7.57425725e-01 1.78438044e+00 -3.84003557e-02 5.21606386e-01 1.48618221e-01 -7.49810040e-01 4.93746787e-01 4.57417637e-01 3.59562695e-01 -8.66204083e-01 -8.58844295e-02 2.35738844e-01 -2.97460407e-01 -6.75264299e-01 5.20954669e-01 4.14207995e-01 -4.05895561e-01 4.17286754e-01 3.90884548e-01 -6.89241216e-02 2.05091387e-01 5.87296367e-01 1.23092425e+00 -2.43825212e-01 7.65415430e-02 -7.84096479e-01 7.57013321e-01 1.99436620e-01 5.62126748e-03 8.69326413e-01 -3.79440308e-01 8.09771895e-01 4.22268510e-01 -5.10519743e-01 -7.08965182e-01 -1.04237616e+00 -5.64455241e-02 1.25685287e+00 -2.51697004e-01 -8.33265066e-01 -6.39767945e-01 -9.31639910e-01 -1.09829390e-02 6.06604934e-01 -8.78665686e-01 1.55823663e-01 -4.44492519e-01 -5.09661973e-01 6.46613300e-01 2.80363560e-01 4.39445436e-01 -8.83338511e-01 -2.68005412e-02 -1.05834924e-01 -7.61464536e-01 -1.00939274e+00 -3.92306566e-01 -2.41819546e-01 -7.80337334e-01 -1.24755704e+00 -8.59965861e-01 -8.18267822e-01 6.68860367e-03 4.77221198e-02 1.74508238e+00 1.76231816e-01 -1.37131542e-01 5.38300574e-01 -7.03812122e-01 -4.21441615e-01 3.16150337e-01 3.73657018e-01 -3.16039324e-01 -1.42968193e-01 5.19643903e-01 -2.92148888e-01 -6.79979980e-01 7.67988563e-02 -3.64925534e-01 -4.42886859e-01 1.94369242e-01 7.59436071e-01 2.83416867e-01 -6.76325977e-01 8.08902919e-01 -6.44412339e-01 5.43196499e-01 -7.65043318e-01 -6.42309114e-02 5.79623401e-01 -5.30226648e-01 -1.06911492e-02 3.99107605e-01 7.80396536e-02 -6.48707449e-01 -4.87226635e-01 -6.14585876e-01 4.01313789e-02 -1.58080623e-01 3.30589771e-01 -6.49803281e-02 1.72248676e-01 7.01498270e-01 1.10400960e-01 -4.45227921e-01 -1.02602923e+00 4.12155926e-01 7.86146760e-01 2.76582479e-01 -7.69831121e-01 6.32754683e-01 1.85267583e-01 -4.83692884e-01 -8.54599953e-01 -1.43061495e+00 -1.04069829e+00 -8.19499269e-02 -1.19890839e-01 1.08471322e+00 -9.54904854e-01 -8.95544887e-01 3.79169345e-01 -1.12665069e+00 9.24408957e-02 -2.83462703e-01 2.21591696e-01 -1.70963407e-01 2.49560982e-01 -5.01293182e-01 -5.62105238e-01 -7.00467825e-01 -7.61472762e-01 1.18129420e+00 2.81719267e-01 -2.85826534e-01 -9.17928040e-01 3.26601177e-01 1.41232717e+00 8.82669747e-01 -2.17778504e-01 8.51815581e-01 -1.21415043e+00 -3.06697965e-01 -9.50289071e-02 -4.90059823e-01 -8.60016420e-02 -3.53436023e-01 -6.71948254e-01 -1.03784406e+00 -1.29275188e-01 -2.94641078e-01 -7.04396784e-01 1.40250778e+00 5.16360886e-02 1.37090278e+00 -1.26367077e-01 -1.80115193e-01 5.84966280e-02 1.24706662e+00 -3.97466153e-01 5.44726193e-01 4.01079774e-01 6.52456522e-01 6.67970061e-01 4.50390697e-01 6.37278259e-02 1.03092027e+00 1.07067382e+00 3.63971025e-01 5.42244375e-01 -3.95591021e-01 -3.35380137e-01 1.72577396e-01 1.07889760e+00 -2.82070696e-01 -2.01671526e-01 -1.18050599e+00 7.56895661e-01 -1.86298931e+00 -1.10380971e+00 -5.80115616e-01 2.13886476e+00 8.56408656e-01 -3.08673978e-01 6.70548752e-02 -2.97375679e-01 4.50344771e-01 1.99676439e-01 -1.38032101e-02 -1.36666864e-01 -3.29333216e-01 8.43139172e-01 -2.67112032e-02 4.17651802e-01 -1.26420307e+00 9.90561843e-01 7.22451067e+00 1.09323788e+00 -2.79609561e-01 6.24922991e-01 2.61417538e-01 1.12602897e-01 -5.39916217e-01 4.02824767e-02 -9.62295830e-01 5.98716617e-01 1.17792940e+00 -1.32002523e-02 1.50121242e-01 4.81730998e-01 -5.24348259e-01 -1.90936312e-01 -7.28079259e-01 9.20986474e-01 6.65127039e-01 -1.54570425e+00 2.45128661e-01 -2.24595189e-01 8.36119592e-01 3.81478250e-01 -1.34639904e-01 7.84483433e-01 2.98050046e-01 -1.57738173e+00 4.82098669e-01 6.60898507e-01 3.22954834e-01 -3.72734815e-01 1.11474514e+00 3.17623168e-01 -9.09447134e-01 5.26607670e-02 -8.36069509e-02 -1.77766889e-01 1.77910343e-01 5.15292704e-01 -6.15886450e-01 6.97898507e-01 1.18106663e+00 6.16947949e-01 -8.81521642e-01 1.26211345e+00 -2.82425344e-01 1.03425026e+00 -3.35917264e-01 -4.29048330e-01 3.29254614e-03 1.30776614e-01 6.57517254e-01 1.33114290e+00 -1.16727334e-02 -5.35744652e-02 1.73380394e-02 5.22243381e-01 -4.29570287e-01 4.18506652e-01 -1.33119091e-01 -6.84630647e-02 3.88817489e-01 1.24400485e+00 -1.21996969e-01 -4.99877602e-01 -1.78193986e-01 8.14729214e-01 6.68825746e-01 2.13796809e-01 -6.71323419e-01 -2.39541769e-01 4.37377572e-01 -4.91323099e-02 4.33361202e-01 -4.85741496e-02 -2.30314702e-01 -1.52496600e+00 1.26340672e-01 -6.86144054e-01 9.26721811e-01 -6.80346072e-01 -1.67771411e+00 5.34672379e-01 -2.67023861e-01 -5.75411260e-01 8.08499679e-02 -5.10708988e-01 -4.16631132e-01 8.01357388e-01 -1.60667360e+00 -1.25178313e+00 -4.19982970e-01 5.84708154e-01 4.77959931e-01 -5.10709345e-01 1.03053629e+00 5.96729219e-01 -2.34909087e-01 8.02755356e-01 1.01928845e-01 2.04771623e-01 7.88656592e-01 -1.44570553e+00 2.39884883e-01 2.49262765e-01 3.11022252e-01 6.07764900e-01 5.93396902e-01 -3.47792655e-01 -1.23047554e+00 -8.57153177e-01 1.78094435e+00 -1.18238378e+00 9.15124059e-01 -3.61207753e-01 -8.03315222e-01 3.38209391e-01 4.76793766e-01 -2.31065214e-01 8.84952426e-01 5.30081928e-01 -8.18699360e-01 1.81230351e-01 -1.02417183e+00 1.48202628e-01 1.05450678e+00 -7.94671178e-01 -1.11548853e+00 6.56190097e-01 8.22649658e-01 -1.84829324e-01 -1.23742187e+00 4.92308646e-01 4.04994726e-01 -6.30687416e-01 1.22119844e+00 -1.07547951e+00 4.66115564e-01 -3.16743910e-01 -4.65683460e-01 -1.00265384e+00 -2.07684532e-01 -1.11167036e-01 -3.77775431e-01 1.25520563e+00 6.87230468e-01 -5.72592974e-01 8.51828516e-01 3.27232927e-01 9.31736156e-02 -7.04878688e-01 -1.18134594e+00 -4.93769646e-01 4.40337837e-01 -4.70730036e-01 3.37658972e-01 1.20697677e+00 -1.39643654e-01 8.94470632e-01 2.71072481e-02 -3.76601189e-01 5.83818376e-01 2.29562849e-01 4.75690693e-01 -1.52748334e+00 -1.51176453e-01 -5.81407785e-01 -2.82337606e-01 -8.83676410e-01 5.90665229e-02 -1.32010877e+00 -3.97737026e-01 -1.93011343e+00 8.13108087e-01 -3.20617169e-01 -5.18715203e-01 2.78713375e-01 -4.48017150e-01 4.41358984e-01 9.54237953e-02 1.69987366e-01 -1.50657213e+00 6.62204981e-01 9.70702529e-01 -2.09950164e-01 2.39986986e-01 -2.26526875e-02 -8.90544832e-01 3.21308851e-01 8.08797538e-01 -3.33494842e-01 8.00227597e-02 -4.29195225e-01 5.56237519e-01 -3.70035440e-01 6.10911131e-01 -6.49996161e-01 1.36247769e-01 1.75406504e-02 1.67327434e-01 -7.15549529e-01 4.58868235e-01 -2.74102658e-01 -1.09120093e-01 3.02716404e-01 -4.90188897e-01 -3.85579802e-02 -1.47159517e-01 3.06130111e-01 -3.96957964e-01 -1.01564936e-01 2.95900911e-01 -9.07783508e-02 -7.31102049e-01 1.67194307e-01 -4.88437861e-02 9.47296143e-01 4.71579343e-01 3.60056579e-01 -8.95829678e-01 -4.24341470e-01 -7.65923738e-01 5.99404693e-01 7.27267936e-02 9.93533432e-01 3.57170820e-01 -1.31007302e+00 -1.05613363e+00 -4.46605176e-01 6.33241951e-01 -3.98147047e-01 4.49053854e-01 9.89001095e-01 -3.53238255e-01 1.03805339e+00 1.18339218e-01 -4.41127598e-01 -1.14251208e+00 1.24099150e-01 3.89394090e-02 -6.95721507e-01 -1.06890619e-01 1.03896224e+00 -4.55967993e-01 -7.48091221e-01 1.83114111e-01 1.87304467e-01 -1.02388334e+00 4.27771926e-01 8.68300974e-01 6.28896654e-01 3.41880500e-01 -7.18969524e-01 -7.63265431e-01 5.57357788e-01 2.52992269e-02 -2.52045095e-01 1.45965159e+00 2.52466708e-01 -5.13938308e-01 3.74799639e-01 1.29999411e+00 5.30987866e-02 -2.16531381e-01 -3.56640548e-01 3.28400999e-01 -2.70412058e-01 -6.76273927e-02 -1.27686405e+00 -6.17482543e-01 8.06406677e-01 4.66622889e-01 2.52843112e-01 6.23308957e-01 6.95188165e-01 1.01231408e+00 5.71765661e-01 4.79759902e-01 -1.15582216e+00 6.07526228e-02 1.13450158e+00 1.06766856e+00 -1.33259058e+00 -1.30289108e-01 -3.38247061e-01 -4.67196345e-01 6.39085174e-01 4.34093416e-01 6.72673360e-02 7.13990927e-01 -4.63174582e-01 -8.18116963e-02 -6.48148715e-01 -7.93593585e-01 -6.02434814e-01 9.82402146e-01 5.42843282e-01 5.01931787e-01 2.36680925e-01 -6.49927735e-01 1.05727708e+00 -5.33569932e-01 -2.53263712e-01 -5.60692661e-02 5.30304551e-01 -6.11880064e-01 -9.10329819e-01 -1.14383504e-01 6.52376711e-01 -8.11839998e-01 -2.60639429e-01 -6.52008116e-01 3.57562512e-01 1.06429689e-01 1.19822419e+00 -5.22190109e-02 -5.44539213e-01 2.69917220e-01 2.89725870e-01 4.53817129e-01 -5.44571698e-01 -1.04225063e+00 -9.20005918e-01 6.26937985e-01 -6.75408483e-01 -4.15690064e-01 -9.14699137e-01 -8.62562358e-01 -1.40756473e-01 -1.07616514e-01 6.22409403e-01 6.47024870e-01 1.14169335e+00 4.99744087e-01 1.12167764e-02 2.72803158e-01 7.95389805e-03 -4.13316935e-01 -1.29503083e+00 -5.03081307e-02 6.27357006e-01 1.80713087e-01 -5.39106965e-01 -4.55223650e-01 -2.63260782e-01]
[11.383204460144043, 7.999159812927246]
19fbe579-d7fe-4bbd-88c8-de6557c0eaa6
vren-volleyball-rally-dataset-with-expression
2209.13846
null
https://arxiv.org/abs/2209.13846v1
https://arxiv.org/pdf/2209.13846v1.pdf
VREN: Volleyball Rally Dataset with Expression Notation Language
This research is intended to accomplish two goals: The first goal is to curate a large and information rich dataset that contains crucial and succinct summaries on the players' actions and positions and the back-and-forth travel patterns of the volleyball in professional and NCAA Div-I indoor volleyball games. While several prior studies have aimed to create similar datasets for other sports (e.g. badminton and soccer), creating such a dataset for indoor volleyball is not yet realized. The second goal is to introduce a volleyball descriptive language to fully describe the rally processes in the games and apply the language to our dataset. Based on the curated dataset and our descriptive sports language, we introduce three tasks for automated volleyball action and tactic analysis using our dataset: (1) Volleyball Rally Prediction, aimed at predicting the outcome of a rally and helping players and coaches improve decision-making in practice, (2) Setting Type and Hitting Type Prediction, to help coaches and players prepare more effectively for the game, and (3) Volleyball Tactics and Attacking Zone Statistics, to provide advanced volleyball statistics and help coaches understand the game and opponent's tactics better. We conducted case studies to show how experimental results can provide insights to the volleyball analysis community. Furthermore, experimental evaluation based on real-world data establishes a baseline for future studies and applications of our dataset and language. This study bridges the gap between the indoor volleyball field and computer science.
['Linda Petzold', 'Yuan-Fang Wang', 'Erwan Fraisse', 'Yun Zhao', 'Rhys Tracy', 'Haotian Xia']
2022-09-28
null
null
null
null
['type-prediction']
['computer-code']
[-2.33808070e-01 -5.60299277e-01 -5.52738488e-01 -7.71271512e-02 -7.28347898e-01 -6.68921649e-01 1.15234137e-01 3.04529607e-01 -5.18268347e-01 5.42457163e-01 5.77480674e-01 -5.40526271e-01 -5.82214236e-01 -1.03035057e+00 -6.37447298e-01 -2.77474970e-01 -3.07540059e-01 5.89471459e-01 3.70947480e-01 -8.41065705e-01 7.14271903e-01 4.73981410e-01 -1.72228730e+00 4.96994555e-01 3.29241544e-01 8.68263245e-01 -1.66147575e-02 7.80567229e-01 4.35620517e-01 1.45691240e+00 -6.68131351e-01 -3.71232629e-01 7.53657699e-01 -5.07808268e-01 -8.74042213e-01 -1.29348874e-01 2.12025151e-01 -5.13972864e-02 -5.04022479e-01 2.53431052e-01 4.57752466e-01 6.92700207e-01 2.50679076e-01 -1.05277431e+00 8.67889673e-02 8.40248764e-01 -2.96508729e-01 6.98274434e-01 7.20093012e-01 4.46272016e-01 9.85762656e-01 -2.77816445e-01 5.16070366e-01 7.06335306e-01 8.07216108e-01 1.15033448e-01 -7.17567563e-01 -1.07180393e+00 1.56254396e-01 3.99012893e-01 -1.35057271e+00 -1.63873809e-03 5.48078656e-01 -6.85913146e-01 1.76678821e-01 6.31890774e-01 1.33690345e+00 9.45779800e-01 9.83993188e-02 8.77557337e-01 1.01339495e+00 -2.07827479e-01 7.07648098e-02 -3.08222950e-01 3.10778826e-01 1.88261271e-01 7.12609366e-02 2.09545419e-01 -9.29438055e-01 1.16278492e-01 8.15134227e-01 -2.61496473e-02 2.88976669e-01 4.36730534e-02 -1.02167022e+00 9.72316980e-01 2.58832395e-01 6.66850582e-02 -4.56482321e-01 2.36657396e-01 4.84835714e-01 1.05456762e-01 3.62854242e-01 8.26654971e-01 1.83281794e-01 -1.07148969e+00 -1.16298950e+00 1.15851641e+00 9.03117597e-01 5.10030389e-01 4.97064620e-01 -1.56818494e-01 -1.53222039e-01 6.94818079e-01 -2.60096878e-01 1.00990899e-01 6.75633997e-02 -1.06353235e+00 7.64720082e-01 5.93532801e-01 2.13830203e-01 -1.23697186e+00 -5.73484778e-01 -3.81950170e-01 9.31267664e-02 4.34280001e-02 7.46206701e-01 -1.77422673e-01 -5.72668493e-01 1.36484587e+00 5.07350452e-02 2.75363684e-01 -3.83210093e-01 1.29211545e+00 8.88576388e-01 4.52912182e-01 -2.10768636e-03 2.07823768e-01 1.46480811e+00 -6.51441216e-01 -2.73363471e-01 -4.98772860e-01 5.81368506e-01 -7.64719367e-01 1.27303374e+00 5.11924386e-01 -1.29696774e+00 -6.19217277e-01 -6.26479864e-01 1.99426502e-01 -3.39453528e-03 -2.84458809e-02 1.09961832e+00 7.17788160e-01 -2.03550920e-01 3.94898713e-01 -9.96102691e-01 -1.09776713e-01 3.17964464e-01 1.13820262e-01 -1.07753925e-01 1.10954709e-01 -1.43663323e+00 8.70171666e-01 4.44214433e-01 -1.50358498e-01 -9.12418783e-01 -1.09461391e+00 -8.53355050e-01 -2.00486615e-01 1.06725729e+00 -1.59142748e-01 1.24422479e+00 -4.31502044e-01 -1.17294765e+00 9.13709104e-01 4.78845388e-01 -5.27068615e-01 4.43876833e-01 -2.29687899e-01 -3.01150861e-03 -2.95835704e-01 5.99523544e-01 -1.21233404e-01 -3.89145464e-01 -1.01643097e+00 -1.18581700e+00 -2.90432632e-01 5.26673973e-01 7.15007424e-01 8.61822367e-02 3.22442710e-01 -6.81329072e-01 -7.60143697e-01 2.52004594e-01 -1.05426705e+00 -3.58056664e-01 -8.17537129e-01 -3.47560555e-01 -1.80785172e-02 4.98096608e-02 -6.19381666e-01 1.88948596e+00 -1.90242803e+00 5.64538129e-02 5.46513617e-01 1.51291013e-01 3.32448352e-03 3.64900380e-01 1.05291772e+00 1.15155362e-01 1.81532325e-03 5.18305004e-01 3.47324669e-01 -9.11699161e-02 1.04159795e-01 -3.77395570e-01 3.52854282e-01 -7.25327909e-01 6.62512720e-01 -9.19181049e-01 -4.63970482e-01 2.56561995e-01 -3.59528154e-01 -8.25082481e-01 -3.07735186e-02 3.29465598e-01 3.88443530e-01 -7.16993272e-01 7.12840438e-01 1.01571210e-01 5.01218319e-01 1.13755308e-01 -5.91424033e-02 -6.32045448e-01 7.49144137e-01 -1.50149953e+00 1.44671452e+00 -4.12652254e-01 7.41761804e-01 -3.23986374e-02 -9.72212911e-01 6.67217910e-01 -2.17692718e-01 9.38685834e-01 -6.31339371e-01 2.92328149e-01 -8.27001706e-02 2.29715973e-01 -6.40500963e-01 1.09165061e+00 -2.89873213e-01 -8.98150325e-01 7.80797601e-01 -4.34613407e-01 -2.10202798e-01 8.34329188e-01 2.48379245e-01 1.05593503e+00 2.30443642e-01 9.30096656e-02 -1.73861712e-01 -1.83011413e-01 7.22341418e-01 6.74849451e-01 1.15028071e+00 -1.55563891e-01 4.03421044e-01 4.05817419e-01 -8.82062435e-01 -7.81571567e-01 -1.08021307e+00 2.03546092e-01 1.46309233e+00 5.39351225e-01 -7.66235352e-01 -5.22470832e-01 1.87512785e-02 -3.78799364e-02 5.95716536e-01 -7.30668604e-01 -1.80225194e-01 -7.56855547e-01 -5.67870915e-01 8.71567190e-01 7.39628613e-01 6.39021993e-01 -7.81768620e-01 -8.07151139e-01 3.13587308e-01 -8.67140591e-01 -9.55531359e-01 -5.37613392e-01 -9.48617011e-02 -3.29230249e-01 -1.40512431e+00 5.97571628e-03 -5.57936609e-01 -2.33850136e-01 2.40411729e-01 1.17188692e+00 -1.72716286e-02 -4.68527615e-01 3.82638663e-01 -6.36150718e-01 -7.71281481e-01 -1.93225205e-01 2.45914251e-01 3.86666089e-01 -3.38232130e-01 5.60255647e-01 -4.43792820e-01 -6.44338369e-01 7.99958944e-01 -5.52010059e-01 3.23962063e-01 3.90602797e-01 3.45119923e-01 4.32960778e-01 1.73861921e-01 1.97593924e-02 -7.47674584e-01 9.62966025e-01 -5.40416718e-01 -2.11803570e-01 -2.08671186e-02 -2.48533368e-01 -5.53406656e-01 1.51893660e-01 -6.07812345e-01 -5.65063119e-01 -3.54802907e-01 -2.45538056e-01 2.91654486e-02 -5.34769595e-02 8.67606282e-01 3.07511270e-01 1.76807269e-01 1.13393342e+00 3.00470322e-01 -2.21039176e-01 -4.07235622e-01 -6.54842798e-03 6.18045866e-01 8.87833953e-01 -1.19432151e+00 7.97283649e-01 4.25491393e-01 -9.80355516e-02 -5.39907336e-01 -1.01624966e+00 -8.02482128e-01 -4.64176118e-01 -9.21817899e-01 7.77445614e-01 -9.95751143e-01 -1.57725990e+00 3.07981700e-01 -2.21810728e-01 -4.48223859e-01 -5.93826175e-01 1.03811586e+00 -7.40800083e-01 9.41729024e-02 -8.52496326e-01 -1.02891541e+00 2.94138223e-01 -1.00826216e+00 5.59593439e-01 2.92519242e-01 -4.68996167e-01 -7.00720727e-01 2.91835815e-01 1.14804661e+00 1.02445707e-02 3.33336949e-01 3.59820694e-01 -7.31198192e-01 -4.42371547e-01 -4.30183113e-01 4.22811717e-01 -4.77239005e-02 -6.41543418e-02 -2.98953652e-01 -2.56840646e-01 5.28663509e-02 -3.70528996e-01 -5.06162763e-01 4.00917917e-01 5.72795451e-01 1.04962265e+00 5.89504056e-02 -1.23252064e-01 5.93348742e-01 8.24699044e-01 1.64670512e-01 6.31986141e-01 8.98711264e-01 4.66960430e-01 7.23257244e-01 1.35663712e+00 5.51137269e-01 6.84809327e-01 1.01862431e+00 2.44222626e-01 -1.38382733e-01 -5.22745587e-03 -8.19330335e-01 2.01838613e-01 6.38170421e-01 -1.11107934e+00 7.36007690e-02 -1.20268154e+00 5.10021925e-01 -1.88990319e+00 -1.57885468e+00 -2.31652364e-01 2.05765271e+00 7.82000780e-01 3.48068327e-01 8.74631763e-01 3.42861861e-01 4.18030739e-01 2.10249990e-01 -1.37379974e-01 -4.07174468e-01 3.17608625e-01 3.43149811e-01 8.76209915e-01 2.15810269e-01 -1.02408731e+00 1.17330682e+00 6.03650618e+00 1.35691559e+00 -8.38823855e-01 -1.25188544e-01 3.37064952e-01 -5.38196981e-01 1.51579455e-01 2.79639810e-01 -9.00335968e-01 2.64728904e-01 6.36837363e-01 -3.34941834e-01 4.32956159e-01 8.14637601e-01 5.88446736e-01 -3.54315281e-01 -8.76672387e-01 1.03401911e+00 -5.93774952e-02 -1.73034978e+00 -2.95791358e-01 1.84331328e-01 3.70329171e-01 -2.68612325e-01 -7.05387965e-02 8.40980530e-01 7.91857421e-01 -9.56075311e-01 1.21329725e+00 4.22390223e-01 4.03636664e-01 -8.61178696e-01 5.01435220e-01 5.95188737e-01 -1.28871441e+00 -4.57268447e-01 5.61696105e-03 -1.04841006e+00 1.98202208e-01 -5.92835136e-02 -6.18229628e-01 7.36338615e-01 9.83259439e-01 5.34106255e-01 -9.25903991e-02 1.05684757e+00 -8.78235791e-03 1.08880246e+00 -3.25247794e-01 -1.17209643e-01 5.03656507e-01 -2.81776220e-01 6.23469234e-01 7.64246047e-01 1.80534609e-02 5.74915648e-01 8.73101532e-01 6.33280396e-01 2.90423036e-01 3.69084567e-01 -3.10821921e-01 -1.04709707e-01 5.43476403e-01 1.02207243e+00 -8.30209851e-01 -1.55530591e-03 -9.02958810e-02 7.47343451e-02 -1.72962040e-01 1.32232845e-01 -1.01215577e+00 -1.59685075e-01 1.03249073e+00 1.04570723e+00 -3.67125779e-01 -3.81649554e-01 -6.38244748e-01 -9.67124581e-01 -3.82980466e-01 -1.23885000e+00 6.14003778e-01 -5.87982953e-01 -9.99451458e-01 3.32744122e-01 6.05115533e-01 -1.42658341e+00 -3.61427486e-01 -3.48771036e-01 -6.89901114e-01 7.42648542e-01 -6.01513863e-01 -1.14660668e+00 -3.25523585e-01 4.93437290e-01 7.48722017e-01 -2.53566861e-01 2.96135008e-01 3.95566493e-01 -4.70473230e-01 4.74688232e-01 -4.47572827e-01 5.46171963e-01 4.22875047e-01 -7.72894621e-01 1.39571711e-01 6.40724242e-01 3.08469862e-01 5.21373212e-01 1.05561793e+00 -8.93133342e-01 -1.35271835e+00 -4.32499439e-01 2.15458572e-01 -7.21319199e-01 8.38921666e-01 -4.04186755e-01 -2.82895952e-01 8.29629898e-01 -6.52807415e-01 -6.33608878e-01 1.07745039e+00 8.51059377e-01 2.87743807e-01 -1.13537908e-01 -3.59184921e-01 9.22881782e-01 1.11070979e+00 -3.46162438e-01 -6.59417868e-01 3.31578672e-01 -1.04646772e-01 -1.16179550e+00 -8.53506386e-01 3.34602267e-01 9.06570017e-01 -9.31147814e-01 1.25414848e+00 -9.41369832e-01 5.91790497e-01 2.16567200e-02 -9.29721147e-02 -1.10104549e+00 -2.60840088e-01 -3.66085529e-01 5.81408203e-01 7.69676328e-01 1.75857395e-01 1.63232848e-01 1.21130800e+00 8.77291203e-01 -4.99970406e-01 -9.56270635e-01 -6.90312207e-01 -7.03746736e-01 7.43870661e-02 -1.28409338e+00 4.35908467e-01 7.76270807e-01 1.82668835e-01 -1.16183937e-01 -8.04997504e-01 -9.12045985e-02 3.80693465e-01 2.93485582e-01 1.49584401e+00 -1.03430247e+00 -5.07589996e-01 -4.50915426e-01 -7.55006731e-01 -1.11693239e+00 -3.37382317e-01 -8.07110608e-01 -1.17242977e-01 -1.52139854e+00 2.56607413e-01 -8.61424804e-01 1.15266711e-01 4.37217265e-01 -1.05209760e-02 6.21802807e-01 4.44330156e-01 3.11037898e-01 -6.46848083e-01 -1.21925630e-01 1.15232027e+00 2.38772660e-01 -5.42423844e-01 7.26248980e-01 -9.43139315e-01 8.46279979e-01 5.34609735e-01 -4.43217278e-01 -4.11452144e-01 -7.36814504e-03 5.61049581e-01 3.11831295e-01 4.72207189e-01 -1.18789315e+00 3.65249008e-01 -9.93335247e-01 -3.47104557e-02 -5.39277911e-01 4.65064019e-01 -1.85651526e-01 1.73876569e-01 4.30793554e-01 -3.90366018e-01 -1.48634538e-01 3.05301338e-01 2.58344233e-01 -3.47134620e-01 -6.48614094e-02 2.70137489e-01 -2.05164045e-01 -8.30430567e-01 3.86950374e-01 -8.70930254e-01 5.04694939e-01 1.37314570e+00 -7.75531352e-01 -5.70305847e-02 -1.02301145e+00 -1.00667167e+00 5.04241765e-01 2.30803847e-01 5.37968993e-01 2.64494658e-01 -1.17788589e+00 -9.75594401e-01 8.35592076e-02 2.48356014e-02 -4.74932402e-01 6.32849097e-01 8.57111871e-01 -9.68207598e-01 7.46795237e-02 -4.83029485e-01 -3.63713324e-01 -1.28056884e+00 3.92655171e-02 2.88008153e-01 -5.81177056e-01 -6.97439075e-01 8.35418582e-01 -4.41021845e-02 -2.84699976e-01 9.83661711e-02 -1.75495908e-01 -2.93289870e-01 -1.31304273e-02 4.32083130e-01 6.06716871e-01 -2.05521137e-01 -6.65679574e-01 -3.49182516e-01 3.37642908e-01 1.86063483e-01 -1.36197999e-01 1.35335720e+00 1.59865338e-02 3.84118050e-01 8.23343277e-01 3.20799232e-01 5.64013779e-01 -1.12762010e+00 -2.79105958e-02 -1.01123206e-01 -8.68022740e-01 -9.09770429e-02 -6.14977360e-01 -8.34862530e-01 5.89404762e-01 6.50980622e-02 2.76734382e-01 9.60198700e-01 5.93463238e-03 8.95355344e-01 4.68589067e-02 7.34878063e-01 -1.43505538e+00 6.87257648e-02 4.75550652e-01 5.43845475e-01 -8.26983631e-01 3.29591781e-01 -3.68846506e-01 -1.12664199e+00 8.77142012e-01 6.23261273e-01 -1.53945833e-01 4.76967424e-01 4.82709229e-01 2.31532395e-01 -4.35403615e-01 -4.98095781e-01 -4.01858360e-01 3.39505851e-01 2.84137994e-01 2.47664556e-01 1.51214302e-01 -7.67637908e-01 1.38031876e+00 -1.00071013e+00 1.11354649e-01 6.55982494e-01 1.10215151e+00 -5.01293123e-01 -1.17182910e+00 -7.06368983e-01 6.74300432e-01 -7.79389560e-01 1.47767052e-01 -3.43176126e-01 1.20595741e+00 5.23779929e-01 1.05681002e+00 4.96932976e-02 -9.53588247e-01 8.03925216e-01 -3.87100697e-01 3.77714276e-01 -8.85276258e-01 -1.20071268e+00 -1.33447543e-01 3.82774174e-01 -7.47833133e-01 -1.79199770e-01 -7.61032224e-01 -1.07688260e+00 -9.84173238e-01 -4.01968658e-02 6.62922382e-01 5.73772907e-01 8.21443796e-01 -2.56181300e-01 5.91243982e-01 3.91183585e-01 -9.83577490e-01 -2.51273215e-01 -8.82892966e-01 -9.61419344e-01 6.31786168e-01 -5.65922797e-01 -1.02287006e+00 -3.89232635e-02 -2.41433725e-01]
[6.635614395141602, 0.37314194440841675]
b937b9fb-6b4c-4eec-93a7-3d949ba0f4a2
design-challenges-for-entity-linking
null
null
https://aclanthology.org/Q15-1023
https://aclanthology.org/Q15-1023.pdf
Design Challenges for Entity Linking
Recent research on entity linking (EL) has introduced a plethora of promising techniques, ranging from deep neural networks to joint inference. But despite numerous papers there is surprisingly little understanding of the state of the art in EL. We attack this confusion by analyzing differences between several versions of the EL problem and presenting a simple yet effective, modular, unsupervised system, called Vinculum, for entity linking. We conduct an extensive evaluation on nine data sets, comparing Vinculum with two state-of-the-art systems, and elucidate key aspects of the system that include mention extraction, candidate generation, entity type prediction, entity coreference, and coherence.
['Xiao Ling', 'Sameer Singh', 'Daniel S. Weld']
2015-01-01
null
null
null
tacl-2015-1
['type-prediction']
['computer-code']
[-4.13627326e-01 6.32521451e-01 -8.05267155e-01 -2.44251937e-01 -8.33544910e-01 -6.19076610e-01 5.55634081e-01 5.04299939e-01 -5.51922202e-01 1.44570267e+00 4.16578084e-01 -1.61741495e-01 -1.85917109e-01 -7.51393020e-01 -6.10088348e-01 -2.35456284e-02 -4.59665567e-01 1.22511768e+00 2.90579557e-01 -3.47866535e-01 -6.71421513e-02 3.68134886e-01 -1.18189859e+00 1.96354896e-01 9.23853219e-01 5.83167076e-01 -4.10724044e-01 3.93602163e-01 -5.05422831e-01 7.74661243e-01 -8.54418457e-01 -1.19323003e+00 -2.12054446e-01 -4.46963795e-02 -1.31959558e+00 -7.90192187e-01 5.66374481e-01 1.51742116e-01 -3.99885982e-01 1.01949131e+00 8.96120250e-01 -1.53420670e-02 5.59770346e-01 -1.60993731e+00 -8.94124568e-01 1.46635747e+00 -3.26103210e-01 2.31636524e-01 6.18460596e-01 -6.32264733e-01 1.47861254e+00 -6.83808744e-01 1.39901125e+00 1.08808661e+00 1.35982168e+00 3.56994271e-01 -1.14488053e+00 -8.32611620e-01 -2.37123087e-01 3.73682350e-01 -1.45815551e+00 -5.00723004e-01 3.29058886e-01 -1.85748413e-01 1.35100245e+00 -8.70836433e-03 1.95113659e-01 9.83194470e-01 -2.58616935e-02 1.01330721e+00 6.01026773e-01 -5.81149518e-01 -1.03518337e-01 2.23652586e-01 4.49163258e-01 5.76912522e-01 7.35943854e-01 -1.17758147e-01 -5.02261221e-01 -4.54327792e-01 5.22289813e-01 -9.96692598e-01 -2.86893636e-01 -5.25760651e-01 -1.28798270e+00 8.47230375e-01 4.67584401e-01 4.71641839e-01 -2.45544255e-01 -7.80850500e-02 5.64473927e-01 1.55393332e-01 2.34782025e-01 8.59634101e-01 -8.05774510e-01 1.60088360e-01 -9.15694952e-01 5.24533391e-01 1.52796423e+00 1.27403235e+00 5.52226245e-01 -4.49548334e-01 -2.21068516e-01 7.40580678e-01 2.91779339e-01 3.52945328e-02 2.84852177e-01 -8.58596861e-01 7.72479355e-01 6.14997745e-01 1.44196361e-01 -1.03675437e+00 -7.99319446e-01 -2.92463422e-01 -6.82549953e-01 -4.04814810e-01 1.79849416e-01 -7.40536094e-01 -2.27560163e-01 2.10883331e+00 2.97719210e-01 7.80759677e-02 4.13406342e-01 5.15873730e-01 1.61790228e+00 2.65924245e-01 5.80865800e-01 -2.62647629e-01 1.35934544e+00 -9.16870832e-01 -1.03964019e+00 -1.33118033e-01 8.25106561e-01 -8.37474942e-01 -2.64804095e-01 -3.38872731e-01 -1.53807735e+00 -4.13955569e-01 -1.02237034e+00 -3.32723171e-01 -7.88922608e-01 6.30344599e-02 1.20063639e+00 4.77739066e-01 -1.00367415e+00 7.21483529e-01 -3.88579160e-01 -7.79962420e-01 2.74634719e-01 7.68144906e-01 -8.22301865e-01 6.74959779e-01 -2.15514970e+00 1.42282820e+00 1.08631444e+00 -2.48599529e-01 1.31195188e-01 -6.74387634e-01 -1.02174616e+00 1.26438007e-01 4.32906628e-01 -1.26066732e+00 1.47378063e+00 -3.63977283e-01 -9.59085464e-01 9.74978805e-01 -5.37672713e-02 -8.39945972e-01 2.59468347e-01 -4.18708622e-01 -8.93150687e-01 -3.75160240e-02 3.61007988e-01 9.94603574e-01 -9.17748734e-02 -1.22822738e+00 -8.81847501e-01 1.44827619e-01 -3.39123383e-02 2.42689982e-01 -6.88016117e-02 2.95703292e-01 -3.84791344e-01 -5.75648487e-01 -2.63161749e-01 -8.59174192e-01 8.63420814e-02 -4.51063603e-01 -8.28593671e-01 -7.98380315e-01 4.96559858e-01 -5.17773390e-01 1.71707082e+00 -1.55845535e+00 2.25976199e-01 -1.64502248e-01 2.60899991e-01 4.20968950e-01 4.97767441e-02 8.38419259e-01 -4.59449857e-01 2.34773055e-01 -1.95070848e-01 -4.58383262e-01 4.13413018e-01 1.67473275e-02 -2.07509026e-01 8.82281587e-02 2.37268418e-01 1.34280014e+00 -1.20316219e+00 -9.95252192e-01 -2.72615671e-01 3.12978953e-01 -2.75240898e-01 1.69911742e-01 -2.69260645e-01 2.23361831e-02 -3.42046231e-01 7.27280259e-01 5.19131720e-01 -2.78019667e-01 8.36706400e-01 -5.53228080e-01 -1.25868216e-01 6.98184192e-01 -1.31660581e+00 1.48381233e+00 1.72506183e-01 7.83764780e-01 -1.62617534e-01 -7.68891215e-01 7.55392313e-01 7.90702701e-01 5.21786034e-01 -3.05949837e-01 1.88605934e-01 4.26216692e-01 -1.75565600e-01 -6.14722729e-01 1.06343293e+00 1.74560696e-01 -3.86050940e-01 3.70105863e-01 6.63203776e-01 6.47324085e-01 6.36498153e-01 4.20272708e-01 9.19853210e-01 4.10227418e-01 8.08813453e-01 -1.83288683e-03 2.72871107e-01 2.15329677e-01 6.92140400e-01 6.97498739e-01 -2.23662302e-01 1.88902184e-01 4.39295381e-01 -1.71282664e-01 -1.03063154e+00 -1.04369318e+00 -3.30461144e-01 7.68189967e-01 2.33444661e-01 -6.27602339e-01 -6.88199818e-01 -9.31254566e-01 2.21702620e-01 6.10993862e-01 -4.81365800e-01 2.86472384e-02 -9.74164367e-01 -9.80792999e-01 1.20923293e+00 8.17326844e-01 7.41399467e-01 -1.36036026e+00 -1.94988608e-01 4.96982992e-01 -7.72877395e-01 -1.34557140e+00 -3.30829732e-02 1.37412757e-01 -6.51787102e-01 -1.07586002e+00 -3.39126796e-01 -1.10042834e+00 1.10135548e-01 -3.16527694e-01 2.02490997e+00 -7.69632682e-02 7.05929399e-02 5.07200174e-02 -8.77926350e-02 -1.97541922e-01 -4.26023155e-01 7.98481405e-01 8.40103105e-02 -7.71394789e-01 9.58574593e-01 -4.61691648e-01 -2.12043360e-01 -7.88144618e-02 -3.18076283e-01 -3.50650489e-01 9.46735740e-01 7.72083580e-01 2.70949721e-01 -3.23777586e-01 9.62899983e-01 -1.35795665e+00 8.11755478e-01 -9.50275123e-01 -1.82233408e-01 5.46208382e-01 -6.40472353e-01 1.58620536e-01 1.20107152e-01 -5.60765192e-02 -1.16453886e+00 -1.98847309e-01 -3.03693593e-01 1.55939996e-01 -3.98098469e-01 7.94226944e-01 -2.15821281e-01 7.05501214e-02 5.36531210e-01 -4.23064083e-01 -5.97259521e-01 -5.40968001e-01 9.09807563e-01 5.19666255e-01 1.03639102e+00 -6.80414617e-01 6.30192339e-01 6.21001609e-02 -1.32243752e-01 -2.47383595e-01 -9.52111483e-01 -5.00815332e-01 -1.14635801e+00 3.36056501e-01 8.86558890e-01 -9.88912702e-01 -7.32227921e-01 1.12228274e-01 -1.56640112e+00 1.79383263e-01 -1.34628534e-01 2.99389392e-01 -4.15009588e-01 3.91313881e-01 -8.55949163e-01 -2.94141471e-01 -6.06388271e-01 -5.51405251e-01 7.34581411e-01 4.79617029e-01 -7.37689912e-01 -1.43150091e+00 6.37097359e-01 1.74659178e-01 7.33702183e-02 4.26712513e-01 9.33801293e-01 -1.26264393e+00 -5.85977197e-01 -2.46643513e-01 -4.25447971e-01 -4.19548243e-01 -1.53681546e-01 6.73596859e-02 -6.38413787e-01 5.07801175e-02 -8.80842090e-01 -2.54953027e-01 7.41057038e-01 2.50004172e-01 2.88848370e-01 -9.07398760e-02 -1.26004207e+00 4.69909012e-01 1.38722277e+00 1.18480928e-01 5.16312897e-01 7.71115363e-01 5.32459795e-01 4.49315786e-01 4.63173568e-01 -9.93147865e-02 1.05093086e+00 9.62090135e-01 4.21330929e-02 -9.06569362e-02 -3.56113106e-01 -2.78877437e-01 -1.42106503e-01 8.59868765e-01 -8.77267122e-02 -5.06920457e-01 -8.60629618e-01 8.52111936e-01 -2.13682342e+00 -1.16626358e+00 -2.26796076e-01 1.71889830e+00 1.10459971e+00 4.13698815e-02 1.07614338e-01 -2.92779982e-01 1.13148439e+00 -3.33669186e-02 -5.05250506e-02 -1.14264004e-01 -5.28719783e-01 8.27722251e-02 3.37745816e-01 2.64447451e-01 -1.69165862e+00 1.03171551e+00 7.73881340e+00 4.35089827e-01 -5.31726301e-01 -3.68175516e-03 1.60965756e-01 5.65261364e-01 1.94104575e-02 1.18342258e-01 -1.32947814e+00 8.32625702e-02 8.56937766e-01 -4.46082324e-01 -3.51260394e-01 7.54212379e-01 -7.90474772e-01 1.28016889e-01 -1.44854259e+00 5.27258754e-01 1.66648164e-01 -1.55453634e+00 -2.60987375e-02 -4.00988579e-01 8.88198137e-01 7.70946369e-02 -3.79469723e-01 7.43491828e-01 7.43905962e-01 -7.88536131e-01 2.88690537e-01 5.38745105e-01 4.88185704e-01 -7.40476906e-01 1.24660838e+00 -1.09098919e-01 -1.35214329e+00 1.03406951e-01 -2.19664440e-01 3.57926697e-01 5.99009693e-01 5.45099318e-01 -7.63632417e-01 1.27210772e+00 5.43064713e-01 6.71397150e-01 -5.40982306e-01 1.37103999e+00 -4.39419359e-01 5.64920567e-02 -1.44516289e-01 7.37108141e-02 9.19120312e-02 2.63347566e-01 7.51993060e-01 1.69595695e+00 4.09747101e-02 1.06337480e-01 1.46682248e-01 7.47282863e-01 -7.93990433e-01 4.59481217e-02 -4.87654746e-01 7.88010508e-02 1.24816239e+00 1.37365234e+00 -4.25811648e-01 -4.75287110e-01 -5.05506754e-01 5.82122982e-01 8.48139346e-01 4.74541746e-02 -8.56878221e-01 -6.67302847e-01 4.13522333e-01 -5.12181461e-01 3.86000931e-01 -9.87036750e-02 -2.88047701e-01 -1.10558164e+00 -3.08664948e-01 -6.44677758e-01 9.22458410e-01 -5.72204590e-01 -1.53006685e+00 6.70715451e-01 3.58649909e-01 -8.79434228e-01 -3.94842476e-01 -2.91629136e-01 -5.11888683e-01 6.38153672e-01 -1.61730206e+00 -1.03348684e+00 -1.91894174e-02 1.32864386e-01 3.95571962e-02 -2.91406840e-01 1.09693575e+00 8.46300423e-01 -6.77797616e-01 1.01111424e+00 -1.19549735e-02 7.32671797e-01 1.27245998e+00 -1.50620985e+00 7.64320731e-01 5.28774202e-01 1.85997441e-01 8.79281223e-01 6.96319938e-01 -8.80355179e-01 -9.01284754e-01 -1.01597679e+00 1.81798744e+00 -8.02348435e-01 9.70758855e-01 5.98806255e-02 -8.04685295e-01 1.08749592e+00 6.97249770e-01 -2.79927522e-01 7.92133749e-01 6.79048419e-01 -3.83624613e-01 4.30370837e-01 -1.08961773e+00 6.88680887e-01 1.20755839e+00 -1.46742046e-01 -1.29499960e+00 2.35780105e-01 6.46125972e-01 -5.47858536e-01 -1.38730741e+00 8.42043221e-01 4.86406952e-01 -7.50074446e-01 1.25115895e+00 -9.25949574e-01 2.65964389e-01 -1.26257062e-01 5.82629628e-02 -1.34438026e+00 -5.57667673e-01 -6.34344935e-01 -7.85279989e-01 1.78322196e+00 7.80786633e-01 -4.41638261e-01 6.72647893e-01 4.08338338e-01 -1.35650948e-01 -6.60070896e-01 -6.78397059e-01 -5.90701759e-01 2.62466878e-01 1.71845138e-01 7.20177233e-01 1.53228593e+00 5.66228807e-01 8.79244864e-01 -1.38367891e-01 3.57930869e-01 5.63219726e-01 3.91461670e-01 5.41146994e-01 -1.67426193e+00 -4.16097678e-02 -5.53650975e-01 -2.90909320e-01 -8.16279888e-01 6.54126108e-01 -1.07353890e+00 -2.88155079e-01 -1.83281088e+00 3.17796618e-01 -5.83196998e-01 -2.29994699e-01 5.60191333e-01 -4.78531241e-01 2.05095097e-01 -1.63185913e-02 3.10328484e-01 -1.02073109e+00 2.75710166e-01 5.31756043e-01 1.90085229e-02 -1.82786267e-02 -4.29614037e-01 -6.88739419e-01 7.36449420e-01 5.38093567e-01 -5.48635304e-01 1.04448475e-01 -4.38593179e-01 6.75912321e-01 8.32220465e-02 -1.33640915e-01 -7.63795018e-01 6.53708339e-01 4.29427147e-01 4.15458977e-01 -9.48077500e-01 -2.92707253e-02 -3.84547859e-01 3.04306895e-01 1.50942221e-01 -4.24857795e-01 3.53022248e-01 1.71595797e-01 2.13494331e-01 -4.40472752e-01 -4.59086537e-01 4.07967865e-01 -8.69110525e-02 -1.19703841e+00 2.09877014e-01 7.95940235e-02 3.94455314e-01 8.63071144e-01 2.59737641e-01 -8.64533424e-01 -7.97218755e-02 -9.19885755e-01 5.66493571e-01 5.91105297e-02 5.83101153e-01 4.24916595e-02 -1.63803697e+00 -9.50503886e-01 -3.17024648e-01 5.44081964e-02 -2.88503230e-01 -1.79633364e-01 7.63058722e-01 -1.95850223e-01 8.23254168e-01 -2.81197578e-01 -9.21042189e-02 -1.21339679e+00 6.27747655e-01 2.67704606e-01 -7.99254537e-01 -4.86001670e-01 7.90998876e-01 -3.66912395e-01 -7.56856918e-01 1.97618634e-01 4.87528950e-01 -8.05846095e-01 5.83837450e-01 1.81835294e-01 4.95097071e-01 -6.50335252e-02 -7.26395488e-01 -5.32380700e-01 3.29898477e-01 -2.67323196e-01 2.04694375e-01 1.10077190e+00 -1.54080853e-01 -3.27416986e-01 2.65461653e-01 8.61053526e-01 -7.86938891e-02 -4.25879210e-01 -3.06871921e-01 6.93728626e-01 2.36565277e-01 -3.26275706e-01 -9.23433483e-01 -8.01195920e-01 3.43938679e-01 2.21996725e-01 4.63501960e-01 6.34475350e-01 1.63252726e-01 9.51705754e-01 8.13744783e-01 4.95065182e-01 -9.93001938e-01 -5.51908493e-01 8.67828786e-01 4.74038303e-01 -1.21195841e+00 2.41558120e-01 -6.63987696e-01 -4.40810502e-01 1.13176322e+00 7.09824741e-01 -1.06130458e-01 5.40746987e-01 5.23283243e-01 -9.77713093e-02 -2.71669835e-01 -8.11653972e-01 -2.86615461e-01 2.64344811e-01 5.72848678e-01 9.45775390e-01 -1.08120948e-01 -5.84506691e-01 7.48834431e-01 -3.22725385e-01 -9.63141862e-03 2.76916742e-01 7.50306189e-01 -2.99926162e-01 -1.37016082e+00 -1.26994133e-01 1.34804219e-01 -8.07081461e-01 -3.74484360e-01 -6.51329517e-01 1.37786996e+00 3.32217932e-01 6.54169619e-01 -2.83055790e-02 -4.97774482e-02 4.74285007e-01 4.36495006e-01 5.15996337e-01 -5.30551136e-01 -7.69808352e-01 -4.02674347e-01 9.38653588e-01 -2.20245942e-01 -9.15137708e-01 -8.12926412e-01 -1.24080932e+00 -4.51962501e-01 -6.74125433e-01 5.40422380e-01 3.08141142e-01 1.00527692e+00 4.60333586e-01 4.76552159e-01 1.01151511e-01 -5.75249195e-01 1.00243893e-02 -1.07575703e+00 -2.10915312e-01 3.79569709e-01 -3.93111110e-02 -9.69970405e-01 1.30101740e-01 -1.49571821e-01]
[9.402423858642578, 8.75474739074707]
40aaad8b-dc56-4f5b-b512-e86147a077ff
conformalized-fairness-via-quantile
2210.02015
null
https://arxiv.org/abs/2210.02015v2
https://arxiv.org/pdf/2210.02015v2.pdf
Conformalized Fairness via Quantile Regression
Algorithmic fairness has received increased attention in socially sensitive domains. While rich literature on mean fairness has been established, research on quantile fairness remains sparse but vital. To fulfill great needs and advocate the significance of quantile fairness, we propose a novel framework to learn a real-valued quantile function under the fairness requirement of Demographic Parity with respect to sensitive attributes, such as race or gender, and thereby derive a reliable fair prediction interval. Using optimal transport and functional synchronization techniques, we establish theoretical guarantees of distribution-free coverage and exact fairness for the induced prediction interval constructed by fair quantiles. A hands-on pipeline is provided to incorporate flexible quantile regressions with an efficient fairness adjustment post-processing algorithm. We demonstrate the superior empirical performance of this approach on several benchmark datasets. Our results show the model's ability to uncover the mechanism underlying the fairness-accuracy trade-off in a wide range of societal and medical applications.
['Bei Jiang', 'Linglong Kong', 'Wulong Liu', 'Dengdeng Yu', 'Lei Ding', 'Meichen Liu']
2022-10-05
null
null
null
null
['prediction-intervals']
['miscellaneous']
[ 1.61626905e-01 2.26361811e-01 -6.99761927e-01 -1.02979541e+00 -8.92576635e-01 -5.34286499e-01 2.26925477e-01 5.72659671e-01 -4.59302366e-01 1.20472252e+00 5.46214223e-01 -4.99391913e-01 -4.76227641e-01 -8.63516986e-01 -3.16151202e-01 -3.21946323e-01 -3.61827463e-01 3.71786803e-01 -3.52174073e-01 -3.01165376e-02 6.37890637e-01 2.46008232e-01 -1.40965796e+00 -4.20676880e-02 1.31555045e+00 1.03558886e+00 -9.11677957e-01 4.37106878e-01 2.83187926e-01 5.64727426e-01 -3.79513919e-01 -1.00171733e+00 4.74958241e-01 -6.81200445e-01 -6.83876038e-01 -5.62108696e-01 2.99376398e-01 -6.65009797e-01 2.73934036e-01 8.67659688e-01 6.86331272e-01 2.09616885e-01 8.52790594e-01 -1.44120300e+00 -6.24143720e-01 9.42450583e-01 -7.38107562e-01 4.41935658e-02 2.20076159e-01 3.91028747e-02 1.34048903e+00 -2.19385445e-01 2.76723355e-01 1.23252547e+00 8.27476323e-01 7.04490185e-01 -1.44616044e+00 -8.19429517e-01 -3.03929746e-01 -3.58192354e-01 -1.27312362e+00 -7.37260580e-01 5.79244159e-02 -2.19538629e-01 3.07832599e-01 7.31073141e-01 5.58057010e-01 5.16953588e-01 3.22070837e-01 3.57738405e-01 1.11777449e+00 -2.02972859e-01 4.71278340e-01 2.74739158e-03 1.43711463e-01 4.29369748e-01 5.24881244e-01 2.66539991e-01 -7.16665685e-01 -7.39013314e-01 5.45550644e-01 -1.37619853e-01 6.21996149e-02 -7.50873536e-02 -6.58090293e-01 9.79138374e-01 1.91586629e-01 -4.70267951e-01 -1.82993293e-01 4.02523994e-01 5.83957374e-01 3.45393538e-01 1.06405485e+00 2.44868711e-01 -2.79753953e-01 -2.33835697e-01 -1.21420979e+00 5.60345769e-01 9.53550041e-01 7.56380796e-01 4.25156385e-01 -3.74680251e-01 -8.08966517e-01 5.74270904e-01 1.80719078e-01 5.24732411e-01 -3.03708345e-01 -1.57176864e+00 2.95929730e-01 2.88305819e-01 4.62490261e-01 -9.68696594e-01 -4.74273264e-01 -2.19812796e-01 -7.94150889e-01 1.08235620e-01 8.88533294e-01 -1.99723527e-01 -2.17996240e-01 2.01902390e+00 6.02884710e-01 1.37838162e-02 -3.33677113e-01 8.53394032e-01 2.93226272e-01 1.54056355e-01 5.80967844e-01 -4.33258474e-01 1.28820086e+00 -1.21571779e-01 -3.35743368e-01 3.77710521e-01 4.20721024e-01 -4.01811957e-01 9.18925881e-01 -7.60813728e-02 -1.23581684e+00 8.43470544e-02 -4.19824958e-01 -3.08617979e-01 1.08123362e-01 -6.72563910e-01 8.01213682e-01 1.44900358e+00 -8.14585328e-01 6.92727566e-01 -4.69381392e-01 -3.43334705e-01 1.00848961e+00 3.69906247e-01 1.47142224e-02 4.05830033e-02 -1.23931134e+00 6.27457917e-01 -2.45200902e-01 -3.14970195e-01 -4.40358162e-01 -1.52272081e+00 -6.12959087e-01 3.33996177e-01 1.27186432e-01 -1.11433244e+00 9.96627986e-01 -1.11402023e+00 -1.30472791e+00 1.11700928e+00 8.15419182e-02 -5.46189427e-01 1.19920290e+00 -5.74841276e-02 -3.37237343e-02 -3.37195434e-02 4.56788272e-01 2.75870740e-01 3.99840176e-01 -8.31992030e-01 -7.62157202e-01 -4.47338730e-01 -2.98340190e-02 3.26569051e-01 -4.76532578e-01 4.96215701e-01 3.36946487e-01 -3.37697595e-01 -5.78154922e-01 -3.86818022e-01 -4.10670847e-01 3.52313906e-01 -5.38478494e-01 -4.45250012e-02 -2.09017366e-01 -5.11045039e-01 1.45344102e+00 -1.96467412e+00 -7.25281775e-01 6.75611556e-01 2.18914226e-01 -3.63795727e-01 1.05065286e-01 2.67297775e-01 3.21562290e-01 4.51925337e-01 -4.92029220e-01 -2.71963030e-01 4.30718303e-01 -1.62050650e-01 -3.12101841e-01 1.06867814e+00 -1.14260398e-01 7.46002674e-01 -8.36327672e-01 -6.31493390e-01 -1.31151721e-01 2.70018488e-01 -1.16417420e+00 2.33604342e-01 4.74728905e-02 3.66265059e-01 -3.93872052e-01 6.50214076e-01 8.36489737e-01 -9.35754329e-02 1.85079306e-01 3.03308815e-01 -2.02157319e-01 1.35366321e-01 -7.91723549e-01 1.18951237e+00 -8.76055509e-02 2.43340898e-02 1.70121118e-01 -7.52693594e-01 8.01485777e-01 -5.57562746e-02 6.88502908e-01 -6.61098838e-01 3.01640987e-01 2.84418881e-01 -3.34362000e-01 -1.51419550e-01 7.19927609e-01 -6.93732560e-01 -4.41023678e-01 8.42975676e-01 -6.10726476e-01 1.83460116e-01 -2.56734878e-01 6.90738782e-02 7.83278584e-01 -2.85730362e-02 5.70967615e-01 -1.00297201e+00 1.10427603e-01 -2.06927240e-01 6.67918801e-01 9.52002048e-01 -7.38638043e-01 6.18846953e-01 1.03898168e+00 -4.82738286e-01 -1.03868103e+00 -1.16554928e+00 -5.29107213e-01 1.65887952e+00 1.92542106e-01 4.57858741e-02 -7.49552965e-01 -4.69001591e-01 7.43748784e-01 8.24750721e-01 -1.04924810e+00 -1.09575778e-01 -1.84945032e-01 -1.16687465e+00 9.47016358e-01 1.83425561e-01 2.40383089e-01 -4.50936675e-01 -8.49593937e-01 -8.33579972e-02 -5.91650419e-02 -5.93877077e-01 -8.04479897e-01 -4.15591300e-01 -6.32227719e-01 -1.07599175e+00 -6.32210314e-01 5.22411615e-02 5.16896904e-01 -2.23025501e-01 1.32259011e+00 1.76636055e-01 -2.15454891e-01 3.56042624e-01 1.77695706e-01 -4.80949372e-01 -2.01291874e-01 8.67248550e-02 8.71589333e-02 1.16169497e-01 7.86440000e-02 -4.99983490e-01 -1.16661131e+00 2.26385802e-01 -6.96338177e-01 -2.47222677e-01 -1.67275518e-01 6.27816737e-01 4.45079088e-01 -5.56616366e-01 1.39068305e+00 -1.39351380e+00 9.73120213e-01 -9.17931378e-01 -6.91946983e-01 1.31947547e-01 -1.12043369e+00 -1.29543573e-01 5.14181256e-01 1.87718466e-01 -1.05840087e+00 -3.82698387e-01 3.66550945e-02 2.96042591e-01 2.65728414e-01 3.42665911e-01 -3.35960723e-02 1.49637848e-01 8.53458524e-01 -3.94878805e-01 1.54336810e-01 -1.55232819e-02 5.00292301e-01 6.42112494e-01 4.99969780e-01 -1.27495027e+00 1.93461999e-01 6.54968381e-01 2.73562849e-01 -3.04533958e-01 -1.06294858e+00 -8.19091499e-02 -6.48823828e-02 -1.91466913e-01 6.08054519e-01 -7.00583160e-01 -1.37802339e+00 -9.80604738e-02 -6.48998141e-01 -4.54245061e-01 -6.98259890e-01 1.59417540e-01 -7.35924661e-01 3.40702236e-01 -5.61255515e-01 -1.40740263e+00 -7.33726680e-01 -4.43440706e-01 8.92949760e-01 3.67663413e-01 -4.84541208e-01 -1.00784183e+00 1.77976593e-01 3.37272972e-01 6.50598943e-01 6.15282416e-01 8.65981460e-01 -6.18811131e-01 -3.47783804e-01 -6.30499274e-02 -5.67979395e-01 -1.32744595e-01 -3.91677395e-02 1.57032505e-01 -1.10757542e+00 -2.06987649e-01 -6.49768114e-01 -3.08427304e-01 7.62669146e-01 8.15167427e-01 1.38591492e+00 -5.11791885e-01 -4.61961813e-02 7.72358477e-01 1.24130988e+00 -5.53031504e-01 4.65595126e-01 9.72154438e-02 2.01760933e-01 9.49602842e-01 6.50507271e-01 1.22872961e+00 8.98793459e-01 2.85020024e-01 3.67710024e-01 1.03621103e-03 4.25030202e-01 -2.10465550e-01 5.79873696e-02 -5.39298952e-02 -2.41192520e-01 -4.65015844e-02 -8.13786626e-01 5.63535571e-01 -1.80037308e+00 -1.17862821e+00 -2.57083401e-02 2.67324638e+00 1.12785745e+00 -1.03282928e-01 6.65240765e-01 -8.43926966e-02 8.01329911e-01 -9.01675522e-02 -7.56127417e-01 -9.05904591e-01 4.60632257e-02 2.35031351e-01 8.98674667e-01 7.01376438e-01 -8.85605693e-01 5.65044165e-01 7.31448936e+00 6.84262395e-01 -6.78053200e-01 -2.01700881e-01 1.55067170e+00 -3.18142742e-01 -7.68522978e-01 -5.31282052e-02 -4.17732477e-01 5.59287250e-01 1.25064003e+00 -9.72458243e-01 4.09385502e-01 5.51511168e-01 6.47738039e-01 -2.17209205e-01 -1.10808814e+00 4.63389724e-01 -4.15553510e-01 -1.17177427e+00 -3.88833225e-01 1.79092094e-01 8.18627298e-01 -3.71441334e-01 2.76394933e-01 3.25337760e-02 6.56743646e-01 -1.44295120e+00 7.49936938e-01 7.35236526e-01 1.45178223e+00 -1.34342456e+00 5.14654994e-01 2.20890716e-01 -7.46643245e-01 -4.72322665e-02 -4.77061003e-01 -4.08366352e-01 -9.33269132e-03 1.16446328e+00 -5.40276349e-01 5.18903315e-01 4.99474674e-01 5.01842558e-01 -7.73156062e-02 1.10196197e+00 3.17886949e-01 7.62421310e-01 -3.03264081e-01 1.74797252e-01 -3.49280894e-01 -7.45157376e-02 1.14495205e-02 1.38291717e+00 3.27388138e-01 3.01390052e-01 -2.14252979e-01 9.68011975e-01 -4.84846801e-01 6.07241273e-01 -3.10031325e-01 3.32382023e-01 7.49093115e-01 1.07594514e+00 -5.91996193e-01 -1.02251060e-01 -9.12770256e-02 4.98449683e-01 1.73591316e-01 1.63949549e-01 -9.01242793e-01 -2.03386784e-01 1.19803870e+00 1.68197453e-01 -4.02490139e-01 3.47095519e-01 -1.15456128e+00 -9.31064248e-01 -3.42076451e-01 -5.69641113e-01 1.02908719e+00 -3.45658101e-02 -1.68503594e+00 -2.00671270e-01 1.16024846e-02 -8.00094545e-01 -6.15618452e-02 -1.11224167e-01 -5.68813086e-01 1.19506657e+00 -1.70172870e+00 -9.24287558e-01 -1.03758201e-01 5.03959775e-01 -3.50847214e-01 1.77044258e-01 8.36983562e-01 3.69194657e-01 -3.57880801e-01 1.12635326e+00 1.42814800e-01 -1.63224429e-01 1.08094442e+00 -1.19871056e+00 1.15897357e-01 5.37069023e-01 -7.31805682e-01 5.67618310e-01 8.25159073e-01 -5.68100750e-01 -1.00616622e+00 -1.10319543e+00 1.08887100e+00 -5.97786963e-01 3.68336141e-01 -1.53756976e-01 -4.49895740e-01 3.79715413e-01 -7.71464705e-02 2.66734630e-01 1.30911553e+00 2.45367423e-01 -5.41482508e-01 -5.54297090e-01 -1.86428177e+00 4.52089459e-01 1.07137406e+00 -3.21776748e-01 1.40417546e-01 1.55463805e-02 4.35214669e-01 -3.28118563e-01 -1.07378006e+00 2.39705369e-01 1.20062351e+00 -1.38167214e+00 8.71763110e-01 -8.82922411e-01 8.42900395e-01 1.98928609e-01 -3.11263621e-01 -9.66624081e-01 -1.31899700e-01 -1.01434278e+00 2.94147938e-01 1.33783233e+00 3.11674356e-01 -9.05092835e-01 7.88008630e-01 1.38794041e+00 2.84768254e-01 -8.89176786e-01 -1.23546910e+00 -3.98554504e-01 6.16072357e-01 -3.66997600e-01 1.10525465e+00 1.14005661e+00 1.96125433e-01 -2.40573615e-01 -7.43192136e-01 -1.21483169e-02 1.09628308e+00 4.27206486e-01 6.97698057e-01 -1.27257836e+00 -6.55942857e-02 -7.08358288e-01 -1.46603107e-01 -3.73088747e-01 2.68625587e-01 -9.53655958e-01 -6.04755469e-02 -1.13428640e+00 6.60431325e-01 -7.97832787e-01 -4.80554849e-01 3.60894352e-01 -5.17933011e-01 3.63832504e-01 -2.32586614e-03 -1.03268780e-01 -7.14415014e-01 4.98462051e-01 1.02406311e+00 3.11298490e-01 -4.65847440e-02 2.69100994e-01 -1.59802449e+00 2.18125075e-01 8.86950135e-01 -6.05394602e-01 -3.15066665e-01 -1.36052156e-02 5.33263862e-01 2.92513341e-01 3.36595953e-01 -2.66984224e-01 -3.11867278e-02 -8.31047833e-01 3.37675780e-01 2.77241856e-01 -3.13447595e-01 -5.04181802e-01 2.03717910e-02 4.39513922e-01 -9.24853325e-01 -1.96085438e-01 -4.69237678e-02 6.13781989e-01 4.25458103e-01 2.82427281e-01 1.01102436e+00 2.57648349e-01 9.17954519e-02 6.01221800e-01 1.65310383e-01 7.09271133e-01 9.78477538e-01 -2.61204801e-02 -5.53897560e-01 -6.01284266e-01 -1.47744358e-01 5.75681150e-01 6.56104505e-01 -2.02661246e-01 2.42968589e-01 -1.25035787e+00 -1.33436739e+00 -5.29182516e-02 5.66736013e-02 -4.81619030e-01 9.87041146e-02 7.84393132e-01 -4.17371511e-01 -1.08957477e-01 -2.92641312e-01 -1.54451177e-01 -9.48402822e-01 1.16021127e-01 5.47699809e-01 -1.50771275e-01 -2.34876424e-01 7.15109527e-01 2.53885597e-01 -4.15651083e-01 -1.02478033e-02 -5.35961203e-02 2.67178327e-01 -1.52819967e-02 5.54888308e-01 8.91201675e-01 -4.20454413e-01 -4.33317661e-01 -5.54302454e-01 1.62327319e-01 6.84650064e-01 -1.24581508e-01 1.29484940e+00 -4.99726236e-01 -2.71808058e-01 2.02402130e-01 9.08974171e-01 3.68094653e-01 -1.40630770e+00 1.66240320e-01 1.00698143e-01 -8.84616554e-01 -4.86107975e-01 -8.41633558e-01 -8.68019581e-01 6.22905493e-01 4.23888326e-01 3.80483687e-01 1.09231269e+00 -3.29512447e-01 5.62990129e-01 -3.48669291e-01 2.49890000e-01 -1.16947615e+00 -6.17313743e-01 8.90962556e-02 4.25313890e-01 -1.22479522e+00 4.42643970e-01 -3.09369922e-01 -5.64392686e-01 8.56827796e-01 1.90149933e-01 -1.23636782e-01 7.70388424e-01 2.32840404e-01 6.90008029e-02 2.28867456e-01 -6.53006554e-01 1.76384434e-01 1.98694855e-01 5.37287593e-01 7.73835242e-01 5.96335709e-01 -8.07041049e-01 1.08179212e+00 -4.18731987e-01 1.86545253e-01 5.28299272e-01 3.47529739e-01 -3.67728770e-01 -9.49285388e-01 -1.53164104e-01 9.45035815e-01 -1.09253967e+00 -2.05035374e-01 1.62982158e-02 3.68233740e-01 -1.74932510e-01 9.11166370e-01 1.69365272e-01 7.27558732e-02 2.35362753e-01 -1.44822299e-01 3.49378914e-01 -3.75578165e-01 -8.85117352e-01 -3.73637944e-01 2.90603250e-01 -7.84351110e-01 -3.95984471e-01 -7.70661235e-01 -1.23596156e+00 -1.25917709e+00 1.48052260e-01 1.41097039e-01 1.62690744e-01 7.69316256e-01 3.91588062e-01 -3.79667692e-02 8.84279251e-01 -3.58759612e-01 -8.74947488e-01 -1.90104157e-01 -9.71843481e-01 4.54923987e-01 4.85226929e-01 -2.78131813e-01 -2.62497276e-01 -3.01589727e-01]
[8.892563819885254, 5.2859015464782715]
40b066e5-8496-430b-84b6-d10470e0250c
adversarial-inference-for-multi-sentence
1812.05634
null
http://arxiv.org/abs/1812.05634v2
http://arxiv.org/pdf/1812.05634v2.pdf
Adversarial Inference for Multi-Sentence Video Description
While significant progress has been made in the image captioning task, video description is still in its infancy due to the complex nature of video data. Generating multi-sentence descriptions for long videos is even more challenging. Among the main issues are the fluency and coherence of the generated descriptions, and their relevance to the video. Recently, reinforcement and adversarial learning based methods have been explored to improve the image captioning models; however, both types of methods suffer from a number of issues, e.g. poor readability and high redundancy for RL and stability issues for GANs. In this work, we instead propose to apply adversarial techniques during inference, designing a discriminator which encourages better multi-sentence video description. In addition, we find that a multi-discriminator "hybrid" design, where each discriminator targets one aspect of a description, leads to the best results. Specifically, we decouple the discriminator to evaluate on three criteria: 1) visual relevance to the video, 2) language diversity and fluency, and 3) coherence across sentences. Our approach results in more accurate, diverse, and coherent multi-sentence video descriptions, as shown by automatic as well as human evaluation on the popular ActivityNet Captions dataset.
['Anna Rohrbach', 'Jae Sung Park', 'Trevor Darrell', 'Marcus Rohrbach']
2018-12-13
adversarial-inference-for-multi-sentence-1
http://openaccess.thecvf.com/content_CVPR_2019/html/Park_Adversarial_Inference_for_Multi-Sentence_Video_Description_CVPR_2019_paper.html
http://openaccess.thecvf.com/content_CVPR_2019/papers/Park_Adversarial_Inference_for_Multi-Sentence_Video_Description_CVPR_2019_paper.pdf
cvpr-2019-6
['video-description']
['computer-vision']
[ 3.03255886e-01 -4.69142906e-02 -1.92299187e-01 -2.23033711e-01 -9.96331871e-01 -6.28961921e-01 7.63206959e-01 -2.13104606e-01 -1.40676692e-01 9.91968751e-01 5.48189163e-01 2.62396634e-01 2.71826893e-01 -3.46469164e-01 -7.44970560e-01 -6.68058217e-01 3.22491705e-01 2.42015630e-01 5.78808673e-02 -1.60704717e-01 7.18274191e-02 2.90388912e-01 -1.52467906e+00 3.39200258e-01 8.93667996e-01 9.21439409e-01 3.84927124e-01 6.83772624e-01 2.99400222e-02 9.89532053e-01 -7.13169277e-01 -6.08926654e-01 4.61348034e-02 -1.04123247e+00 -5.30418515e-01 1.68479919e-01 6.29349172e-01 -3.76710683e-01 -3.52437168e-01 1.01978779e+00 6.15775824e-01 6.81136129e-03 6.58406317e-01 -1.37436187e+00 -7.78752983e-01 4.36483622e-01 -3.98652494e-01 8.50387588e-02 6.97130799e-01 4.91322309e-01 8.42642605e-01 -5.87434351e-01 8.42152476e-01 1.11289144e+00 3.26922238e-01 8.48081768e-01 -1.26428235e+00 -5.76505363e-01 9.39858407e-02 3.20542216e-01 -1.25428879e+00 -6.89148903e-01 8.74240994e-01 -5.00952780e-01 5.93989134e-01 2.72442341e-01 7.36789405e-01 1.76823127e+00 2.01034099e-01 7.15129435e-01 1.11878812e+00 -2.39719748e-01 3.13933283e-01 3.72227341e-01 -6.34124279e-01 3.70189279e-01 8.52084234e-02 -4.08086814e-02 -5.37982821e-01 2.37809956e-01 8.06689262e-01 -3.65897387e-01 -5.40513754e-01 -2.83678383e-01 -1.15496290e+00 9.22191501e-01 1.47940904e-01 4.43425983e-01 -5.16441882e-01 2.24684164e-01 4.15855438e-01 1.09158397e-01 3.59680772e-01 7.07668781e-01 1.17221169e-01 -4.24180120e-01 -1.10944676e+00 4.08289999e-01 5.45788646e-01 8.01514328e-01 4.82339978e-01 2.44741708e-01 -5.90790987e-01 7.22390652e-01 9.08666551e-02 5.91673493e-01 4.51605499e-01 -8.93391550e-01 5.94502389e-01 1.24653220e-01 4.97825025e-03 -1.16576171e+00 -3.39122564e-02 -3.78744394e-01 -9.13054705e-01 2.34135818e-02 1.91568181e-01 -1.49440572e-01 -7.19381034e-01 2.10074115e+00 -8.58459920e-02 2.15281919e-02 1.32523373e-01 1.08431840e+00 8.82838905e-01 8.73588443e-01 2.78175533e-01 -2.77565211e-01 1.08395779e+00 -1.07346690e+00 -9.55021799e-01 -3.95758957e-01 1.64825767e-01 -7.79453456e-01 1.04608643e+00 1.07890993e-01 -1.33404684e+00 -6.62733972e-01 -9.44173336e-01 1.37844175e-01 -1.54781726e-03 6.72999863e-03 2.83793151e-01 5.05205512e-01 -1.17300034e+00 1.59268558e-01 -4.16324258e-01 -3.07868630e-01 3.55144292e-01 1.35955602e-01 -3.92954677e-01 -1.82360813e-01 -1.30702400e+00 9.89265501e-01 4.47980344e-01 -1.88118786e-01 -1.00257254e+00 -4.24270689e-01 -9.15994883e-01 1.03998721e-01 3.09887439e-01 -9.13641751e-01 1.00149596e+00 -1.62853861e+00 -1.51424432e+00 6.47608519e-01 1.02560230e-01 -3.87418658e-01 7.15477824e-01 3.25379930e-02 -3.34139764e-01 4.00284529e-01 3.46389636e-02 1.27744913e+00 9.92184639e-01 -1.52234292e+00 -4.04079854e-01 4.45432924e-02 2.76021898e-01 4.48197305e-01 -3.83048922e-01 -2.47214222e-03 -4.65843588e-01 -9.12487388e-01 -5.36573946e-01 -1.01538956e+00 5.46475826e-03 -9.35825929e-02 -3.28543931e-01 9.35345888e-02 5.78469694e-01 -8.25730503e-01 1.07701814e+00 -2.23501801e+00 4.12346154e-01 -2.85247207e-01 1.30441725e-01 3.69548351e-01 -2.64988154e-01 3.82238448e-01 -1.27864527e-02 2.39402264e-01 -1.34411240e-02 -4.09231126e-01 -1.65161610e-01 1.25195965e-01 -1.85205773e-01 2.16325700e-01 4.55984950e-01 1.02498960e+00 -9.44634855e-01 -6.42960131e-01 1.50162786e-01 6.88008845e-01 -6.55458748e-01 5.15989482e-01 -3.68222713e-01 8.29896092e-01 -3.94740015e-01 4.45577025e-01 4.42419797e-01 -2.60708809e-01 2.57985704e-02 -2.85961717e-01 1.21872060e-01 -2.35661436e-02 -7.80851245e-01 1.78075421e+00 -4.49142873e-01 8.87799859e-01 -1.25862211e-01 -8.00725758e-01 7.51417398e-01 5.24381340e-01 4.07457769e-01 -9.12836134e-01 5.61129265e-02 1.16982751e-01 -3.84485088e-02 -8.21083963e-01 7.03585505e-01 -2.34914929e-01 -2.84318887e-02 2.25303873e-01 -1.23292319e-02 -1.21729769e-01 4.00443643e-01 2.23921075e-01 9.47297871e-01 3.72913182e-01 3.39548439e-01 1.12947069e-01 4.21980679e-01 -5.95727414e-02 3.28509510e-01 5.40314317e-01 -2.35560343e-01 1.10169065e+00 6.69983804e-01 6.03399146e-03 -1.35039866e+00 -9.42491353e-01 3.02226186e-01 7.23138928e-01 2.74325371e-01 -2.59327739e-01 -8.91331553e-01 -5.28383732e-01 -4.16747242e-01 7.94222414e-01 -4.88081843e-01 -3.46312076e-01 -5.61013877e-01 -3.15364003e-01 5.64778090e-01 4.26091641e-01 5.43039024e-01 -1.28357446e+00 -5.89627087e-01 1.81086048e-01 -5.69818020e-01 -1.53251088e+00 -7.07433581e-01 -3.55780661e-01 -4.98731196e-01 -7.04314470e-01 -1.08708060e+00 -6.63194418e-01 6.34265423e-01 1.62745297e-01 1.16213930e+00 -2.66856015e-01 -6.38039270e-03 4.06905681e-01 -6.40264988e-01 2.13913973e-02 -8.36226761e-01 -8.91385153e-02 -1.19902305e-01 3.61495391e-02 -2.56680846e-01 -3.58208209e-01 -6.57458425e-01 2.02452719e-01 -1.10833395e+00 4.30491179e-01 8.98854196e-01 9.17135239e-01 3.74177724e-01 -2.77188659e-01 7.37339199e-01 -4.80405957e-01 7.37409353e-01 -5.19320905e-01 -2.13824719e-01 3.44089538e-01 -4.11530524e-01 1.83306634e-02 9.33262587e-01 -6.73471272e-01 -9.20978427e-01 1.94766801e-02 -2.04140633e-01 -6.03374004e-01 -2.17297733e-01 3.20054412e-01 -2.35113412e-01 7.41573870e-02 5.23121059e-01 5.11352539e-01 2.24789619e-01 5.50375320e-02 3.90660197e-01 5.20237863e-01 5.14101386e-01 -3.76958400e-01 6.73303783e-01 1.29355565e-01 -2.50105619e-01 -7.24017262e-01 -5.64157069e-01 -1.09958947e-02 -2.05332994e-01 -5.25819719e-01 1.18716002e+00 -1.03620720e+00 -4.97217983e-01 2.92451054e-01 -1.34601963e+00 -1.08431861e-01 -1.86403245e-01 4.79707897e-01 -8.32495630e-01 3.76787782e-01 -3.64116341e-01 -5.72248340e-01 -2.68519580e-01 -1.53197038e+00 9.60751116e-01 2.95140743e-01 -3.42778713e-01 -8.05488288e-01 5.33714145e-03 6.00370884e-01 5.50901592e-01 4.96012807e-01 7.70867229e-01 -4.98362511e-01 -7.29769945e-01 -1.60422161e-01 -3.03511977e-01 5.60126066e-01 -8.65067318e-02 -1.44728914e-01 -8.14549744e-01 -3.29409957e-01 -7.01245442e-02 -6.28858626e-01 4.61971700e-01 3.14027697e-01 9.49459136e-01 -5.57425857e-01 1.85910184e-02 4.04081672e-01 1.43995678e+00 3.07798356e-01 9.19014871e-01 2.82686502e-01 6.46775961e-01 5.47031462e-01 6.04016364e-01 2.90761441e-01 3.57641160e-01 8.61203492e-01 5.53262532e-01 -1.39586002e-01 -4.56203938e-01 -4.60833967e-01 6.63983226e-01 7.38041818e-01 -4.52030376e-02 -6.82177424e-01 -5.74711084e-01 4.53906417e-01 -1.80386853e+00 -1.22528195e+00 2.48463243e-01 1.96490395e+00 7.09488690e-01 4.54533361e-02 2.21447647e-01 -1.72316223e-01 8.26496542e-01 3.36530566e-01 -4.96808112e-01 -4.49107856e-01 -3.79950225e-01 -1.72145560e-01 2.62129128e-01 1.13522008e-01 -7.81587005e-01 7.74501264e-01 6.18911648e+00 8.41998816e-01 -1.26612520e+00 1.41381398e-01 8.79335821e-01 -1.45064607e-01 -4.55902576e-01 -1.79585218e-01 -5.16147673e-01 7.88874745e-01 7.04616487e-01 -6.86528608e-02 5.15559077e-01 5.33211648e-01 3.68156523e-01 -1.02691315e-01 -1.06149447e+00 1.24659050e+00 6.45740926e-01 -1.31953216e+00 4.00702924e-01 -5.17683588e-02 9.92222905e-01 -3.82997662e-01 1.59891799e-01 2.52676666e-01 -3.31777275e-01 -1.15926290e+00 1.05696452e+00 4.70927626e-01 9.82794285e-01 -6.24039650e-01 7.38796890e-01 1.66188449e-01 -9.48627114e-01 5.02939783e-02 1.93194002e-02 2.05537543e-01 4.49429572e-01 2.01957881e-01 -6.08875036e-01 5.25689244e-01 3.37443888e-01 5.70372701e-01 -6.68555260e-01 9.75768328e-01 -2.51203209e-01 3.44805896e-01 1.89965755e-01 -2.79585332e-01 3.83879632e-01 -1.78294525e-01 6.12583697e-01 1.15233934e+00 4.13488418e-01 -1.21134609e-01 1.09718136e-01 1.03646839e+00 -1.28611401e-01 1.93436831e-01 -7.76234746e-01 -3.68021101e-01 3.02899539e-01 1.05921745e+00 -5.27552366e-01 -3.38687003e-01 -4.18519855e-01 1.08622372e+00 2.23964959e-01 4.17207330e-01 -1.21380663e+00 4.31574043e-03 4.41043735e-01 6.09373488e-02 2.86849439e-01 -1.81465790e-01 -6.60513192e-02 -1.20918834e+00 2.41137013e-01 -1.29668057e+00 -3.91209731e-04 -1.05434954e+00 -1.02039087e+00 9.39524293e-01 9.75652561e-02 -1.35535121e+00 -5.97064018e-01 -1.71418741e-01 -4.21161801e-01 5.88166595e-01 -1.32981801e+00 -1.19892597e+00 -3.88258725e-01 4.43554878e-01 8.53314996e-01 -2.87820220e-01 6.02045953e-01 4.99435902e-01 -4.58532453e-01 6.78522468e-01 -3.35879661e-02 -1.38556525e-01 7.83460617e-01 -8.52618933e-01 1.00373983e-01 7.73578703e-01 3.99134271e-02 8.15958157e-02 1.00563884e+00 -5.09639442e-01 -1.37605703e+00 -9.99962389e-01 7.11284399e-01 -9.31080803e-02 4.49686736e-01 -3.14385146e-01 -6.93797171e-01 3.75454307e-01 5.29833913e-01 -3.44214082e-01 4.87174600e-01 -4.22585934e-01 -1.63488880e-01 -9.20871180e-03 -1.05167007e+00 8.78316939e-01 8.43496919e-01 -4.11136121e-01 -2.67422229e-01 2.03442603e-01 6.54790223e-01 -2.77381867e-01 -6.61833286e-01 1.96336955e-01 5.80676913e-01 -1.09650147e+00 8.55556190e-01 -2.53959179e-01 1.00923550e+00 -1.68258101e-01 -9.45576951e-02 -1.35549939e+00 -2.62412339e-01 -6.18315458e-01 5.62678576e-02 1.50972056e+00 3.57114106e-01 -2.27653548e-01 6.43809795e-01 3.92544627e-01 -4.22256850e-02 -7.55234301e-01 -7.75451303e-01 -7.55081832e-01 -2.93092817e-01 -4.28322256e-02 3.00479680e-01 7.54859626e-01 -1.86375067e-01 6.27570033e-01 -9.56457019e-01 -2.95459211e-01 3.62945855e-01 -2.66136732e-02 8.26986730e-01 -5.76863945e-01 -3.66456479e-01 -5.51271737e-01 -5.47762394e-01 -1.02588069e+00 2.65365958e-01 -7.27766335e-01 1.99060842e-01 -1.58553600e+00 6.13143623e-01 -1.67845160e-01 1.48395270e-01 2.38587216e-01 -1.58735603e-01 4.46751386e-01 5.80331147e-01 2.47823074e-01 -7.95014381e-01 6.88584030e-01 1.53772604e+00 -1.67817324e-01 -3.83507013e-02 -3.31486970e-01 -6.73361957e-01 2.69984126e-01 7.09609270e-01 -2.53106207e-01 -4.69258130e-01 -4.96274918e-01 8.84905681e-02 5.02777159e-01 3.93699139e-01 -1.12370431e+00 1.56448353e-02 -1.54450998e-01 3.62858564e-01 -2.06421152e-01 7.26719260e-01 -5.81948340e-01 5.07898688e-01 4.36889112e-01 -4.92681861e-01 3.15047741e-01 1.09574199e-01 5.78721464e-01 -4.73016441e-01 -1.98548809e-01 9.26991463e-01 -1.29990309e-01 -6.73218548e-01 1.85038701e-01 -2.63524324e-01 1.83054388e-01 1.12396824e+00 -3.50939721e-01 -2.22567722e-01 -9.18330133e-01 -4.38752562e-01 8.99958462e-02 6.36507809e-01 6.72001183e-01 6.42213106e-01 -1.44998014e+00 -1.02501464e+00 -1.36937514e-01 1.28829777e-01 -3.62409443e-01 3.66967320e-01 7.97159493e-01 -4.95735079e-01 3.87646735e-01 -5.65376520e-01 -6.77256644e-01 -1.20096147e+00 5.16409755e-01 1.02154305e-02 -2.07183838e-01 -4.41232294e-01 4.79656041e-01 2.67881900e-01 4.09712702e-01 2.42627457e-01 1.70309454e-01 -3.28760594e-01 9.50462297e-02 3.47037435e-01 9.99958590e-02 -3.52868646e-01 -8.93892944e-01 -1.36891603e-01 6.35152876e-01 -5.65281734e-02 -4.88219634e-02 1.02942336e+00 -2.27562994e-01 2.16617018e-01 2.44787723e-01 1.27071393e+00 -1.95800170e-01 -1.37014568e+00 2.60823995e-01 -3.50349963e-01 -5.69996953e-01 -2.41454110e-01 -8.29707623e-01 -1.11046410e+00 5.72802424e-01 5.08658946e-01 2.79853314e-01 1.25708878e+00 3.44593525e-02 9.68273640e-01 -1.39713436e-01 1.52601600e-01 -8.43433678e-01 4.93923098e-01 2.38498762e-01 1.16325772e+00 -1.33208191e+00 -1.15908414e-01 -1.65094808e-01 -1.11260176e+00 9.35797095e-01 6.41427457e-01 6.81041107e-02 -9.69815776e-02 -3.17122340e-02 7.58614689e-02 1.77877843e-01 -7.18020976e-01 -6.79188147e-02 3.13571006e-01 5.34524143e-01 4.48851109e-01 -1.47174761e-01 -5.22765398e-01 2.86178559e-01 -8.02115351e-02 3.23333405e-03 5.58798969e-01 4.92504776e-01 -9.14566666e-02 -9.77255702e-01 -3.53631616e-01 1.69744968e-01 -4.85879779e-01 -3.72147653e-04 -3.04884881e-01 7.29339957e-01 9.24946219e-02 9.86255288e-01 -6.33446947e-02 -3.42765957e-01 1.78966582e-01 -2.20124349e-01 6.11438632e-01 -3.79716605e-01 -5.70336461e-01 2.68623196e-02 1.67203888e-01 -4.84336764e-01 -6.11959159e-01 -5.98696649e-01 -8.83552194e-01 -2.12906644e-01 -2.08817378e-01 1.29068449e-01 8.61463428e-01 8.48843336e-01 3.65961939e-01 5.41488826e-01 7.10509419e-01 -9.25109327e-01 -5.68642855e-01 -9.19059217e-01 -2.45232821e-01 7.93169022e-01 1.43190995e-01 -5.51645577e-01 -3.28189462e-01 2.07206234e-01]
[10.96462345123291, 0.6095175743103027]
61ff6a9d-3664-4f8a-888f-4af579d49a66
hierarchical-aggregation-of-dialectal-data
null
null
https://aclanthology.org/2022.lrec-1.489
https://aclanthology.org/2022.lrec-1.489.pdf
Hierarchical Aggregation of Dialectal Data for Arabic Dialect Identification
Arabic is a collection of dialectal variants that are historically related but significantly different. These differences can be seen across regions, countries, and even cities in the same countries. Previous work on Arabic Dialect identification has focused mainly on specific dialect levels (region, country, province, or city) using level-specific resources; and different efforts used different schemas and labels. In this paper, we present the first effort aiming at defining a standard unified three-level hierarchical schema (region-country-city) for dialectal Arabic classification. We map 29 different data sets to this unified schema, and use the common mapping to facilitate aggregating these data sets. We test the value of such aggregation by building language models and using them in dialect identification. We make our label mapping code and aggregated language models publicly available.
['Nizar Habash', 'Houda Bouamor', 'Nurpeiis Baimukan']
null
null
null
null
lrec-2022-6
['dialect-identification']
['natural-language-processing']
[-7.34740555e-01 -4.10936832e-01 -9.29156095e-02 -4.98303950e-01 -7.79339612e-01 -1.29030859e+00 9.54029202e-01 4.42343354e-01 -2.55050719e-01 4.27598089e-01 6.20009303e-01 -3.82370710e-01 -2.63105128e-02 -1.12840080e+00 1.95184653e-03 -3.97614717e-01 5.99531308e-02 8.05545568e-01 3.37530106e-01 -1.06416845e+00 3.48211586e-01 5.11504769e-01 -1.20153701e+00 4.65003639e-01 1.22210908e+00 2.18657508e-01 -1.87718436e-01 2.51496136e-01 -4.76425171e-01 5.34354866e-01 -5.66090107e-01 -6.32520378e-01 -5.78285158e-02 -5.82868099e-01 -1.22469497e+00 -1.03797726e-01 3.79638940e-01 -1.42512038e-01 1.35952562e-01 9.08535540e-01 2.16228515e-01 -3.82369965e-01 1.01940155e+00 -9.83517110e-01 -8.87095749e-01 9.89799976e-01 -3.93935889e-01 -3.85309994e-01 7.52878070e-01 -4.89166349e-01 8.69014740e-01 -7.81169295e-01 7.97633290e-01 1.43795812e+00 9.47443604e-01 1.91014349e-01 -1.24150300e+00 -3.04481953e-01 2.25637153e-01 -6.14774413e-04 -1.98053348e+00 -2.65745819e-01 3.23037505e-01 -6.29950702e-01 6.86674774e-01 2.90589929e-01 4.97565478e-01 4.32831556e-01 -2.19938815e-01 1.40117943e-01 1.54583228e+00 -8.50080490e-01 -2.13924363e-01 4.74820703e-01 5.16983509e-01 8.49946558e-01 1.56576648e-01 -6.39618278e-01 -2.52402216e-01 -2.00464875e-01 4.58480567e-01 -6.00323141e-01 -1.87269803e-02 -3.33455950e-02 -1.30690885e+00 8.72449160e-01 3.21928471e-01 6.20042980e-01 1.48476020e-01 -6.48134530e-01 4.49689031e-01 6.68180883e-01 3.72010708e-01 3.23249876e-01 -3.05954635e-01 5.66949844e-02 -6.94240630e-01 2.00772792e-01 9.07131553e-01 1.01784492e+00 9.55543220e-01 -6.02189638e-02 3.51748496e-01 1.46571589e+00 4.65783387e-01 4.90631938e-01 3.16278756e-01 -8.41713727e-01 3.82645249e-01 7.72224486e-01 3.26143593e-01 -7.33716846e-01 -7.13583648e-01 1.02856860e-01 -3.32249522e-01 1.53951332e-01 1.01522660e+00 -2.15196371e-01 -7.20869124e-01 1.49559891e+00 1.24456257e-01 -7.35061288e-01 1.37144804e-01 6.80652916e-01 8.72435629e-01 7.56343007e-01 -2.74503650e-03 3.16517621e-01 1.49898922e+00 -6.70668721e-01 -4.31906015e-01 3.02696675e-01 1.13400733e+00 -1.07334781e+00 1.12726736e+00 3.07150185e-01 -9.08415020e-01 -1.52825981e-01 -1.00892782e+00 -1.71727300e-01 -1.13143849e+00 1.09557807e-01 4.24167782e-01 1.46009886e+00 -1.32473636e+00 -1.85479105e-01 -4.56619799e-01 -8.50065649e-01 -4.36066121e-01 1.65988818e-01 -4.87005591e-01 2.46029329e-02 -1.39749789e+00 1.24922121e+00 3.59321147e-01 -2.56011903e-01 -3.69095802e-01 -2.31857002e-01 -9.28517342e-01 -3.60982507e-01 -3.21499318e-01 -8.85690451e-02 7.18785107e-01 -1.00484943e+00 -1.31769609e+00 1.25188541e+00 -1.06586469e-02 8.59847069e-02 5.77093177e-02 2.81160116e-01 -1.22524345e+00 -1.73359558e-01 3.19036126e-01 5.07740021e-01 1.02625981e-01 -1.63016236e+00 -1.07523537e+00 -5.16740143e-01 5.43912709e-01 2.58309633e-01 -4.67857182e-01 7.62167394e-01 -4.02383804e-01 -8.39439809e-01 4.67535228e-01 -1.13245153e+00 1.40136927e-01 -6.45766973e-01 1.72386304e-01 -2.26524040e-01 1.91006735e-01 -1.21322739e+00 1.63386238e+00 -1.83358157e+00 1.10719085e-01 6.50932193e-01 -1.88779816e-01 -2.94887751e-01 -1.35821432e-01 7.12848246e-01 6.05350770e-02 1.30578473e-01 -3.04696023e-01 5.44472411e-02 2.82809645e-01 1.89854443e-01 -1.08942702e-01 5.41164815e-01 -1.33115414e-03 3.35222602e-01 -7.82844424e-01 -5.93981624e-01 -2.74908513e-01 2.87580580e-01 -3.74521494e-01 -5.00244200e-01 1.49338186e-01 1.90364674e-01 1.99056774e-01 1.13041306e+00 6.98804140e-01 4.03386414e-01 6.00444496e-01 -1.05080642e-01 -5.03793418e-01 4.08269465e-01 -1.26411140e+00 1.42419982e+00 -7.39205539e-01 5.81702709e-01 1.01164877e-01 -4.34001446e-01 1.01784122e+00 2.71309942e-01 3.60659659e-01 -4.96959209e-01 2.74042860e-02 7.10814714e-01 1.21605955e-01 -8.89303442e-03 8.94912064e-01 8.71440768e-02 -7.00501740e-01 8.67988825e-01 1.32892773e-01 -2.33825475e-01 5.90097189e-01 1.06197312e-01 2.36244723e-02 2.06718996e-01 2.66733259e-01 -8.05971444e-01 9.77624416e-01 4.49461967e-01 1.30495116e-01 5.01418769e-01 -9.27267075e-02 7.01964974e-01 5.26451766e-01 -3.43556672e-01 -9.46147442e-01 -1.42715383e+00 -7.09627867e-01 1.63851857e+00 1.82402488e-02 -6.48885131e-01 -8.89316142e-01 -6.62303329e-01 -2.31830046e-01 6.02138221e-01 -4.94209349e-01 4.09399897e-01 -7.62862861e-01 -1.30923557e+00 1.17962623e+00 8.67812112e-02 3.59867483e-01 -6.79673970e-01 2.75755614e-01 5.06246202e-02 -5.07404327e-01 -6.03179276e-01 -3.67044657e-01 -2.21721038e-01 -2.73330629e-01 -8.36517155e-01 -9.08956647e-01 -1.38947761e+00 5.79001725e-01 1.97586864e-02 1.55099964e+00 -4.27157246e-02 5.08501649e-01 4.35842544e-01 -6.99320078e-01 -3.10050577e-01 -8.63256097e-01 5.13439775e-01 5.29950708e-02 -7.84628764e-02 4.35441703e-01 8.28782842e-02 1.05140619e-01 6.47393346e-01 -7.25478828e-01 -5.42250574e-01 -1.91685244e-01 4.43393946e-01 -8.28737691e-02 -1.63403153e-01 6.32399976e-01 -9.21406925e-01 8.16538453e-01 -6.98451996e-01 -5.75411022e-01 6.88893378e-01 -5.19152999e-01 -5.54660633e-02 3.48871619e-01 1.34227529e-01 -1.10312653e+00 1.33978780e-02 -5.43016195e-01 1.10062492e+00 -3.05971235e-01 9.95658636e-01 -2.99927086e-01 -2.61185318e-01 8.07243824e-01 1.94872946e-01 -5.61525077e-02 -3.11045676e-01 5.57831824e-01 1.20278728e+00 4.13597435e-01 -1.01152742e+00 3.70551527e-01 3.32994759e-01 -4.76632297e-01 -7.50434041e-01 -1.46921828e-01 -4.25334007e-01 -1.33842647e+00 -4.04280156e-01 6.89884007e-01 -1.25500274e+00 -1.89563319e-01 8.37609589e-01 -6.91714048e-01 -3.38620603e-01 4.42159265e-01 1.09746799e-01 -2.76321650e-01 6.32229805e-01 -8.18911970e-01 -3.21533948e-01 1.06054753e-01 -1.13001609e+00 6.11778378e-01 -2.92706266e-02 -5.50703466e-01 -1.43227661e+00 3.31776828e-01 -1.41333401e-01 1.29223615e-01 1.31149516e-01 1.27494144e+00 -2.59324849e-01 4.94435243e-02 8.43569562e-02 -2.69723773e-01 -1.46398246e-01 7.00446725e-01 3.72741610e-01 -5.66941500e-01 -2.77859777e-01 -4.05987233e-01 -2.08975285e-01 5.71617663e-01 -8.51667598e-02 2.26640254e-01 2.33780779e-02 9.78404507e-02 3.97342741e-01 1.56734073e+00 8.95843655e-02 4.47356731e-01 8.63454700e-01 7.52050698e-01 1.12702703e+00 5.54599464e-01 5.34934327e-02 1.06465232e+00 9.13391113e-01 -1.99465811e-01 3.61632439e-03 -2.17237845e-02 3.81796002e-01 7.87861347e-01 1.21451569e+00 -3.41760099e-01 1.32645115e-01 -1.59592700e+00 9.49025333e-01 -1.49244153e+00 -7.96575189e-01 -5.95643461e-01 2.20147347e+00 1.17027855e+00 -1.86401844e-01 8.00259233e-01 1.10302670e-02 6.11019552e-01 -1.63495168e-01 3.39326173e-01 -7.67222941e-01 -7.00803399e-01 -1.00023210e-01 4.80348319e-01 1.10780036e+00 -8.79461706e-01 1.28587830e+00 7.14287949e+00 6.36839271e-01 -1.12515223e+00 -4.24392708e-02 3.70971501e-01 4.30746287e-01 -3.28839213e-01 5.98277375e-02 -8.79320621e-01 3.67606968e-01 9.78053093e-01 -1.73660919e-01 5.42692721e-01 1.87826276e-01 -1.33715510e-01 -4.69879694e-02 -6.17886126e-01 6.21892989e-01 2.36047089e-01 -1.04934168e+00 3.89305055e-01 -1.50635317e-01 7.65369594e-01 -3.34490687e-02 6.39967397e-02 7.90603235e-02 1.04230881e+00 -8.81274760e-01 1.17915380e+00 1.42570153e-01 1.12615824e+00 -1.03016090e+00 6.26104832e-01 -2.63174355e-01 -1.38432837e+00 2.62204967e-02 -1.82156891e-01 2.38106749e-03 5.25469631e-02 -1.91554695e-01 -4.87936497e-01 8.03255081e-01 7.92872965e-01 6.97213173e-01 -1.11186564e+00 5.44738114e-01 1.11062661e-01 6.56593025e-01 -2.70236939e-01 2.45039731e-01 5.73040068e-01 -8.65293920e-01 3.44607756e-02 1.59750819e+00 5.76849401e-01 -3.83480400e-01 2.92673633e-02 1.56700596e-01 5.01258910e-01 4.90271747e-01 -3.24257612e-01 2.99205214e-01 5.71211338e-01 1.04470897e+00 -1.12454724e+00 -3.50402832e-01 -9.15549636e-01 8.28225315e-01 1.89068407e-01 2.68918216e-01 -7.50662744e-01 -6.18658304e-01 7.72625208e-01 5.97598031e-03 -2.22682014e-01 -6.22558057e-01 -6.12410545e-01 -1.07687390e+00 -4.54142720e-01 -1.05617464e+00 9.60445404e-01 -4.34276909e-01 -1.24524868e+00 7.81546056e-01 8.27258378e-02 -1.26914203e+00 -3.57397765e-01 -8.80161762e-01 3.94951999e-02 1.22750604e+00 -9.57703352e-01 -1.64305651e+00 -1.39757380e-01 9.28272545e-01 -3.93845960e-02 -6.36802971e-01 1.40499938e+00 4.65754330e-01 -3.60247254e-01 8.60871255e-01 5.38072765e-01 6.15004420e-01 1.23720241e+00 -1.59930122e+00 3.49515826e-01 5.94526827e-01 2.41865605e-01 8.89331579e-01 3.14452440e-01 -4.61046964e-01 -5.31310678e-01 -8.72595608e-01 1.30881536e+00 -1.08598113e+00 1.02142751e+00 -6.68151736e-01 -7.87489653e-01 7.94757605e-01 6.53499007e-01 -9.62031901e-01 9.80197132e-01 6.17516279e-01 -7.42740750e-01 -8.54158178e-02 -1.15899050e+00 8.53663087e-01 6.57131970e-01 -8.45469594e-01 -4.18467849e-01 3.10064673e-01 2.49288321e-01 -5.73960654e-02 -1.48896658e+00 -3.28052431e-01 8.82187188e-01 -9.24510777e-01 1.04117537e+00 -3.14654261e-01 7.59683326e-02 -5.16199648e-01 -3.59525442e-01 -1.68715775e+00 -4.87633586e-01 -1.68125466e-01 8.48456442e-01 1.66273952e+00 7.99699247e-01 -1.01243353e+00 3.53095829e-01 2.16846436e-01 -3.03646147e-01 2.11990401e-01 -5.21143675e-01 -7.14255333e-01 1.06268704e+00 -1.71332527e-02 1.19622600e+00 1.49942529e+00 5.48175633e-01 1.53090283e-01 7.22232312e-02 3.25703830e-01 1.74149796e-01 1.78368032e-01 4.02701914e-01 -1.23273027e+00 3.43286842e-01 -9.01767373e-01 -1.28586724e-01 -3.34207833e-01 1.66415110e-01 -1.11014581e+00 -1.47314340e-01 -1.65753174e+00 -4.17152733e-01 -1.02132571e+00 -1.78440660e-02 4.92638141e-01 -2.12220941e-02 6.29671037e-01 2.44639963e-01 5.52420139e-01 -2.39510447e-01 -1.60277471e-01 5.42491317e-01 -2.60027796e-01 -3.80079269e-01 -3.37726802e-01 -6.94783866e-01 6.64805710e-01 1.11082625e+00 -3.43775868e-01 -3.15362103e-02 -7.41545618e-01 5.61749458e-01 -3.90998155e-01 -3.41720521e-01 -9.89850998e-01 8.52761269e-02 -1.53418243e-01 9.80739743e-02 -3.83433074e-01 -1.82004925e-02 -8.18901658e-01 7.88459554e-02 4.34806496e-01 5.26421182e-02 6.49910629e-01 1.66562274e-01 -4.41914231e-01 -5.46035349e-01 -4.94263470e-01 8.68318141e-01 -1.29038677e-01 -9.42545652e-01 -1.46763310e-01 -1.05032218e+00 -3.67228724e-02 7.34074116e-01 -2.44378969e-01 -3.97043198e-01 -3.37078303e-01 -8.39321494e-01 2.53183693e-02 1.03213501e+00 4.91213053e-01 -4.22043130e-02 -1.45083642e+00 -1.26250780e+00 1.60580710e-01 6.91182852e-01 -7.80144036e-01 -2.62391090e-01 4.12383944e-01 -1.31188595e+00 5.57602227e-01 -8.53652596e-01 -3.55656177e-01 -1.22568941e+00 2.01538727e-01 5.55795193e-01 9.50413523e-04 1.39320299e-01 3.17239285e-01 -2.53905833e-01 -1.34287393e+00 -2.55228639e-01 7.96100050e-02 -9.01209831e-01 8.12285185e-01 2.49530628e-01 4.47099149e-01 1.93845481e-01 -1.51354146e+00 -4.05277222e-01 8.80483091e-01 -4.49763983e-02 -6.09819770e-01 8.24437678e-01 -5.18657565e-01 -9.51103151e-01 7.07762420e-01 7.50594854e-01 8.76207709e-01 -4.28733468e-01 -5.39548211e-02 4.57080543e-01 -9.06768627e-03 -5.78691900e-01 -9.15536761e-01 -5.42276502e-01 2.87215084e-01 6.75472915e-01 5.75601339e-01 1.01683223e+00 -7.97429234e-02 2.84306288e-01 9.49833542e-02 5.80702662e-01 -1.25201881e+00 -7.02252507e-01 1.28983498e+00 6.88931823e-01 -8.41095150e-01 -1.21814474e-01 -6.27791882e-01 -5.41799605e-01 1.14374602e+00 4.17326242e-01 1.53231233e-01 9.76616859e-01 -5.63141424e-03 8.74175310e-01 3.84455696e-02 2.11429566e-01 -5.33487141e-01 2.11334884e-01 9.82666552e-01 1.04073513e+00 3.70810419e-01 -7.16386974e-01 5.45027316e-01 -6.29820287e-01 -4.99904871e-01 9.09228504e-01 9.15391564e-01 -3.18548381e-01 -1.64603460e+00 -9.16001558e-01 1.68339595e-01 -2.84949988e-01 -4.08106923e-01 -7.55336821e-01 1.03152585e+00 3.72553021e-01 1.12920618e+00 4.48746741e-01 -3.83930355e-01 1.63448915e-01 9.04350504e-02 6.45952702e-01 -4.43130344e-01 -9.91754293e-01 -2.02821836e-01 6.22711241e-01 1.75235778e-01 -5.79299569e-01 -9.50690329e-01 -1.23338044e+00 -7.09102213e-01 1.65247619e-01 2.26666227e-01 3.98942500e-01 6.11841559e-01 -2.10557029e-01 -2.18013972e-01 6.72503233e-01 -4.50117409e-01 1.94530606e-01 -9.27881002e-01 -8.26988697e-01 4.19916838e-01 -4.04889211e-02 -4.04634476e-01 4.59010862e-02 3.44191045e-01]
[10.199626922607422, 10.711024284362793]
cd85f778-c3b3-4a05-b5a6-07f30f58ec70
can-domain-pre-training-help
null
null
https://aclanthology.org/2021.nlp4dh-1.14
https://aclanthology.org/2021.nlp4dh-1.14.pdf
Can Domain Pre-training Help Interdisciplinary Researchers from Data Annotation Poverty? A Case Study of Legal Argument Mining with BERT-based Transformers
Interdisciplinary Natural Language Processing (NLP) research traditionally suffers from the requirement for costly data annotation. However, transformer frameworks with pre-training have shown their ability on many downstream tasks including digital humanities tasks with limited small datasets. Considering the fact that many digital humanities fields (e.g. law) feature an abundance of non-annotated textual resources, and the recent achievements led by transformer models, we pay special attention to whether domain pre-training will enhance transformer’s performance on interdisciplinary tasks and how. In this work, we use legal argument mining as our case study. This aims to automatically identify text segments with particular linguistic structures (i.e., arguments) from legal documents and to predict the reasoning relations between marked arguments. Our work includes a broad survey of a wide range of BERT variants with different pre-training strategies. Our case study focuses on: the comparison of general pre-training and domain pre-training; the generalisability of different domain pre-trained transformers; and the potential of merging general pre-training with domain pre-training. We also achieve better results than the current transformer baseline in legal argument mining.
['Paul Nulty', 'David Lillis', 'Gechuan Zhang']
null
null
null
null
nlp4dh-icon-2021-12
['argument-mining']
['natural-language-processing']
[ 2.35531226e-01 6.97341621e-01 -6.04824603e-01 -2.00904325e-01 -9.96060789e-01 -8.64110053e-01 1.10848880e+00 4.96302783e-01 -3.97233963e-01 8.50890219e-01 6.77812397e-01 -1.08576119e+00 -5.48533738e-01 -8.06372225e-01 -6.58819914e-01 -1.05648793e-01 3.30019653e-01 1.01980865e+00 4.43268210e-01 -6.53596103e-01 2.83585846e-01 2.13706106e-01 -1.43351710e+00 7.68628180e-01 1.43524253e+00 7.28390396e-01 -3.25003326e-01 6.22735508e-02 -6.57038748e-01 1.15127397e+00 -6.39634252e-01 -1.23552191e+00 1.45234242e-01 -1.77019641e-01 -1.26922798e+00 -3.75696123e-01 4.58883315e-01 -1.12954415e-02 7.42310733e-02 8.17071378e-01 4.50572997e-01 -3.12150180e-01 6.20870292e-01 -1.07132185e+00 -7.44368851e-01 1.27921021e+00 -3.89203936e-01 5.05284905e-01 5.40260434e-01 -2.29708403e-01 1.40235662e+00 -5.70979357e-01 1.26370251e+00 1.25353885e+00 9.11681950e-01 3.57754558e-01 -1.02624667e+00 -4.64509159e-01 9.96762216e-02 5.04828811e-01 -4.65162337e-01 -3.25531960e-01 9.16725516e-01 -6.62434638e-01 1.14445841e+00 -9.40040573e-02 3.75261426e-01 1.31158531e+00 -1.06068641e-01 1.05429220e+00 1.01238334e+00 -6.67883873e-01 -1.38762608e-01 5.14406264e-02 4.27608043e-01 3.87833297e-01 2.14632422e-01 -1.73596889e-01 -4.06472832e-01 -4.04654562e-01 3.21065038e-01 -5.14820755e-01 1.37539074e-01 -1.28972158e-01 -1.01530755e+00 1.14519739e+00 -1.88748792e-01 7.85189271e-01 -2.81448096e-01 -3.96600574e-01 8.61656427e-01 5.07736742e-01 6.29053712e-01 5.72463930e-01 -7.96524107e-01 -3.70363891e-01 -7.72572875e-01 5.51854074e-01 9.17006910e-01 8.13378930e-01 3.77528697e-01 -5.22343397e-01 -3.55523437e-01 1.12538826e+00 -1.82062257e-02 1.35490134e-01 2.93105930e-01 -6.98175132e-01 1.25357091e+00 1.09301519e+00 -1.57984361e-01 -6.73187137e-01 -3.39215398e-01 -9.31491032e-02 -3.86195689e-01 -1.81993768e-01 1.06699455e+00 -2.77414113e-01 -6.06625140e-01 1.48809993e+00 3.41687441e-01 -3.19710255e-01 3.13260436e-01 3.57612759e-01 8.85999620e-01 3.24241281e-01 4.56974596e-01 6.28050789e-02 1.54636371e+00 -4.40960109e-01 -6.89037323e-01 -3.46370429e-01 1.04242551e+00 -8.76153469e-01 1.37457359e+00 3.45373124e-01 -1.05346489e+00 -1.70358777e-01 -5.86688399e-01 -6.37322009e-01 -6.56429112e-01 1.44242883e-01 9.10974085e-01 6.96491957e-01 -3.86266708e-01 5.29753506e-01 -2.51119435e-01 -4.08385515e-01 9.03994977e-01 6.42515942e-02 -2.31005132e-01 5.30113429e-02 -1.53538704e+00 1.08651555e+00 3.48845184e-01 -4.71914083e-01 -9.86431241e-02 -1.22605753e+00 -7.53429294e-01 -1.65640280e-01 7.58775711e-01 -4.83208448e-01 1.28407788e+00 -8.04644108e-01 -1.00219786e+00 1.49982476e+00 1.57628243e-03 -9.70952034e-01 8.92431557e-01 -4.79588926e-01 -3.86328876e-01 -1.30725816e-01 6.69737399e-01 1.87016249e-01 3.10844839e-01 -6.58970952e-01 -8.13814998e-01 -3.35327744e-01 3.52283835e-01 -1.70917869e-01 -5.00337780e-01 5.88978589e-01 5.63992448e-02 -7.10785985e-01 -3.87368679e-01 -5.27104318e-01 -5.13216779e-02 -3.20987403e-01 -2.31594920e-01 -1.04101408e+00 8.28191936e-01 -5.85636020e-01 1.16743422e+00 -1.95300865e+00 -2.34802604e-01 3.12624685e-02 1.60040893e-02 4.38627809e-01 1.04917042e-01 5.65777361e-01 -1.20770030e-01 4.16894138e-01 -7.98947737e-02 1.72270551e-01 2.95302033e-01 5.40944219e-01 -8.07533085e-01 2.12908417e-01 3.67023051e-01 9.87844646e-01 -8.10144007e-01 -1.00660539e+00 -8.26987475e-02 1.38285130e-01 -6.05723381e-01 -3.11325848e-01 -6.11159682e-01 3.37456137e-01 -7.85866797e-01 6.17285609e-01 3.85552198e-01 -1.76053450e-01 2.62252063e-01 -8.44803751e-02 -3.66026521e-01 1.14889121e+00 -6.74794555e-01 1.60884428e+00 -2.81913072e-01 8.34989488e-01 -1.53386211e-02 -1.39684689e+00 7.60789514e-01 5.17826080e-01 5.30006170e-01 -1.13628519e+00 2.33399302e-01 4.90663916e-01 3.35744828e-01 -7.34852135e-01 2.30146348e-01 -3.42405885e-01 -2.18919396e-01 6.62841856e-01 -1.14700504e-01 7.38444999e-02 7.53204584e-01 2.79667467e-01 1.05862665e+00 2.45960906e-01 3.19205523e-01 -3.50764692e-01 6.65624321e-01 5.81166267e-01 6.18419886e-01 5.12562454e-01 -7.86084235e-02 7.74800852e-02 1.04754519e+00 -5.52218676e-01 -1.16408122e+00 -6.02639675e-01 -6.12013638e-01 1.21022236e+00 -2.43195340e-01 -4.11583304e-01 -6.48338079e-01 -1.19230103e+00 2.27987558e-01 8.33278120e-01 -7.22068191e-01 3.90693426e-01 -1.03541553e+00 -7.31935382e-01 1.00231826e+00 5.29306233e-01 5.55639327e-01 -1.28921366e+00 -7.16564894e-01 3.08881402e-01 -4.02462870e-01 -1.49984097e+00 3.12898815e-01 1.24325618e-01 -7.52623320e-01 -1.42992878e+00 -3.00566971e-01 -7.06105947e-01 1.15432255e-01 -4.87573117e-01 1.46035719e+00 -2.08104234e-02 -5.73698357e-02 1.67854056e-02 -5.91362178e-01 -8.22012663e-01 -6.39817834e-01 2.97156960e-01 -4.14956063e-01 -4.97481078e-01 7.63673723e-01 -4.89001185e-01 -5.28840832e-02 1.78372383e-01 -7.95266032e-01 -1.17606953e-01 5.08001506e-01 7.73081481e-01 2.81323224e-01 -1.57202229e-01 7.45698512e-01 -1.59540594e+00 1.12888491e+00 -5.23932874e-01 -3.84457767e-01 5.72783351e-01 -6.93969429e-01 2.49674767e-01 6.32286012e-01 -4.12726998e-01 -1.39997137e+00 -6.42746866e-01 -3.16355467e-01 3.27812552e-01 -1.31725505e-01 8.07255447e-01 -2.79113799e-01 6.02452636e-01 1.03918087e+00 -3.01270306e-01 -1.03269435e-01 -5.84521949e-01 4.60349500e-01 4.88451183e-01 3.37708563e-01 -1.32874978e+00 8.15026581e-01 3.95525724e-01 -1.79417476e-01 -5.18241167e-01 -1.31087637e+00 -1.66894928e-01 -8.39910686e-01 2.31078137e-02 7.13512838e-01 -4.56822187e-01 -3.51297051e-01 -1.43232986e-01 -1.24019682e+00 -4.01291817e-01 -5.41312695e-01 1.37609586e-01 -2.55583405e-01 6.00537777e-01 -6.44837081e-01 -5.56500375e-01 -4.56109226e-01 -7.16311932e-01 8.19667578e-01 -1.02213234e-01 -4.66378301e-01 -1.14162791e+00 1.21997148e-01 9.24739301e-01 8.10051784e-02 3.74108672e-01 1.53153443e+00 -1.17426324e+00 5.26707135e-02 -8.05981681e-02 -2.39120483e-01 9.27777961e-02 -1.33107662e-01 -6.25043362e-02 -8.10857952e-01 4.23084736e-01 -1.90253243e-01 -4.09114718e-01 6.29893422e-01 2.06704721e-01 7.39788353e-01 -3.04332644e-01 -4.52046573e-01 5.11329472e-02 8.58479440e-01 2.69802753e-02 8.04905772e-01 9.94166017e-01 3.11443537e-01 1.03952706e+00 1.10826802e+00 2.18082443e-01 3.09210330e-01 5.39751351e-01 -6.25733379e-03 -1.24086887e-01 -6.60000890e-02 -3.85504663e-01 1.25505894e-01 -7.89647270e-03 -3.38266999e-01 -1.78136304e-01 -1.28454638e+00 8.33627045e-01 -2.05221319e+00 -1.31012499e+00 -4.95177120e-01 1.66804910e+00 1.03406119e+00 4.57704872e-01 2.76488811e-01 4.55155194e-01 4.29534465e-01 -1.00456744e-01 -3.76598239e-01 -5.76936841e-01 -5.03551900e-01 6.12143576e-01 2.58029342e-01 6.92989379e-02 -1.17937958e+00 1.11973524e+00 5.57730532e+00 1.05029535e+00 -6.62563622e-01 2.58806527e-01 5.18974483e-01 2.68087417e-01 -4.99655932e-01 2.18763098e-01 -8.12093735e-01 2.83381969e-01 9.07341301e-01 -5.17476737e-01 -1.40996486e-01 8.67169797e-01 9.33386832e-02 1.75670519e-01 -1.10379279e+00 4.89728779e-01 -1.63837925e-01 -1.67304122e+00 1.19492173e-01 4.13997620e-01 4.58201587e-01 5.01884408e-02 -5.65714724e-02 5.75383127e-01 6.09462619e-01 -9.15596485e-01 8.92469406e-01 -1.71591401e-01 6.05491281e-01 -4.16839749e-01 8.92025411e-01 4.42155778e-01 -7.87140429e-01 -4.03183609e-01 -3.21258932e-01 -2.16222450e-01 2.89779872e-01 8.40206683e-01 -8.99241984e-01 7.13473320e-01 5.83800793e-01 1.05348265e+00 -4.32207882e-01 6.41921282e-01 -6.78858519e-01 8.49153996e-01 -2.84940660e-01 2.33267136e-02 4.14279163e-01 -2.40697220e-01 3.80826622e-01 1.34534132e+00 8.92783105e-02 2.40605295e-01 3.63777913e-02 7.97677636e-01 -1.81671008e-01 4.77497280e-01 -7.55775988e-01 -7.69282505e-02 3.52998525e-01 9.28291559e-01 -4.83505547e-01 -5.41341662e-01 -6.90255702e-01 2.32248560e-01 3.54170769e-01 -1.45725328e-02 -8.49273264e-01 -2.48592161e-02 4.83599246e-01 6.83033705e-01 2.07852304e-01 2.68788636e-01 -5.47413230e-01 -1.05732036e+00 1.06352173e-01 -1.07534873e+00 1.12275398e+00 -4.52228278e-01 -1.62470865e+00 4.81314093e-01 2.87450939e-01 -1.17903018e+00 -3.77648860e-01 -8.31315458e-01 -5.49626231e-01 6.26370430e-01 -1.76476359e+00 -1.45151711e+00 3.17763954e-01 7.83585310e-01 4.38988388e-01 -2.52286136e-01 4.93437290e-01 6.43292904e-01 -3.72581482e-01 6.14039361e-01 -3.32392573e-01 5.19353926e-01 8.18304777e-01 -1.19853377e+00 3.06842119e-01 7.19822347e-01 7.82552585e-02 7.58251131e-01 5.85138023e-01 -7.82316208e-01 -1.00464666e+00 -7.25753307e-01 1.56084204e+00 -8.36816370e-01 1.22032368e+00 -3.09577972e-01 -8.98092508e-01 8.22914898e-01 3.16877604e-01 -4.98133659e-01 7.80245960e-01 7.42817640e-01 -6.61313176e-01 2.57814020e-01 -1.26213300e+00 5.57013214e-01 1.27181089e+00 -5.81598639e-01 -1.31005752e+00 4.96250629e-01 3.09699446e-01 -3.51475805e-01 -8.93739879e-01 3.30933988e-01 3.82981241e-01 -9.18958485e-01 1.05369794e+00 -8.97446513e-01 9.29694235e-01 7.99181014e-02 2.53719687e-01 -6.50233328e-01 3.68753374e-02 -4.78605330e-01 3.15485030e-01 1.72868240e+00 9.84951138e-01 -7.50069022e-01 8.05898666e-01 8.00466001e-01 -2.10499361e-01 -5.71815073e-01 -1.03660381e+00 -7.15430975e-01 7.11401761e-01 -7.21435964e-01 5.45884490e-01 1.26412141e+00 5.11919379e-01 8.99320662e-01 1.25704244e-01 -2.67083764e-01 4.68443722e-01 4.03066009e-01 8.05324852e-01 -1.91896868e+00 7.55842477e-02 -6.22915983e-01 -1.20262448e-02 -7.96565235e-01 5.70064008e-01 -1.07427108e+00 -6.01750851e-01 -1.81447029e+00 1.71770483e-01 -5.09204209e-01 3.01792592e-01 8.36192489e-01 5.53934984e-02 -2.28637233e-01 -2.51933128e-01 1.78059921e-01 -3.95662695e-01 1.73732072e-01 1.12176299e+00 -1.73751175e-01 -1.36423215e-01 -1.11884311e-01 -9.33471799e-01 7.78180301e-01 6.79933369e-01 -5.16894698e-01 -4.55805600e-01 -5.44131339e-01 6.64984465e-01 -1.87958479e-01 1.92889333e-01 -4.13121492e-01 1.47657007e-01 -1.30420446e-01 -9.31726396e-02 -5.25499523e-01 -1.98942095e-01 -5.41595399e-01 -3.31883639e-01 2.47152120e-01 -4.26313281e-01 -3.08509432e-02 3.86470497e-01 2.65134305e-01 -4.83841389e-01 -4.93578762e-01 3.49765569e-01 -2.13794634e-01 -5.64758360e-01 -1.75951451e-01 -3.70968640e-01 6.72611237e-01 8.02432001e-01 -4.70485240e-01 -8.29390585e-01 -3.64569738e-03 -4.83428746e-01 2.38751292e-01 2.10823879e-01 4.12871718e-01 1.24710135e-01 -7.35185742e-01 -1.01057005e+00 -1.63885623e-01 8.99819657e-02 -4.52935211e-02 -2.04891324e-01 9.01427031e-01 -1.36536866e-01 6.14561379e-01 -5.08184098e-02 -2.42806450e-01 -1.52605104e+00 4.53031003e-01 -3.91399078e-02 -1.00197995e+00 -8.85380149e-01 6.39526486e-01 -1.41765490e-01 -5.64158678e-01 -3.98733690e-02 -4.50050741e-01 -6.54514909e-01 3.43924284e-01 1.93076655e-01 2.26288959e-01 1.26296386e-01 -4.04260129e-01 -3.81889194e-01 4.44507986e-01 -2.90878445e-01 -6.72508106e-02 1.80827153e+00 4.50445026e-01 -3.66987526e-01 -4.70832326e-02 7.51578033e-01 4.26316381e-01 -5.16989768e-01 -3.75247985e-01 8.00240993e-01 -2.39256531e-01 -2.58679301e-01 -9.26918149e-01 -7.73264170e-01 7.89161503e-01 -6.11673780e-02 3.33878011e-01 7.79903233e-01 5.48293471e-01 6.79682672e-01 3.59261900e-01 2.76666135e-01 -1.35710704e+00 -1.63199529e-01 9.41499949e-01 8.09415877e-01 -9.29767251e-01 1.20819606e-01 -8.46052706e-01 -8.10175180e-01 1.05322778e+00 2.79544860e-01 1.02615289e-01 5.88023722e-01 4.55696732e-01 2.08636716e-01 -7.69907594e-01 -6.46695316e-01 -4.74479258e-01 2.02314362e-01 6.08057499e-01 8.06913197e-01 -4.02477533e-01 -9.29746926e-01 7.26184070e-01 -6.14390492e-01 2.34594971e-01 2.31651276e-01 9.02922750e-01 -1.09947056e-01 -1.67725658e+00 -2.78289795e-01 6.00225568e-01 -1.02392554e+00 -3.33105862e-01 -9.13436174e-01 1.11498868e+00 4.29759443e-01 1.03352535e+00 -1.67647853e-01 2.03237563e-01 6.15532458e-01 2.43990362e-01 3.52065146e-01 -6.46379054e-01 -1.17761457e+00 -1.59576043e-01 9.23227966e-01 -1.38708398e-01 -7.33722687e-01 -1.03549290e+00 -1.30203283e+00 -3.17170918e-01 -1.44458234e-01 5.36568105e-01 1.12267725e-01 1.30075884e+00 2.48444200e-01 2.28685603e-01 -3.50482464e-01 6.37655705e-02 -3.11986506e-01 -9.99553025e-01 -1.40260339e-01 7.77799129e-01 -1.73239812e-01 -7.27920651e-01 -1.79957196e-01 1.58314645e-01]
[9.705301284790039, 9.414767265319824]
582194da-f026-4eea-9280-5666f1e3a68f
one-shot-doc-snippet-detection-powering
2209.06584
null
https://arxiv.org/abs/2209.06584v1
https://arxiv.org/pdf/2209.06584v1.pdf
One-Shot Doc Snippet Detection: Powering Search in Document Beyond Text
Active consumption of digital documents has yielded scope for research in various applications, including search. Traditionally, searching within a document has been cast as a text matching problem ignoring the rich layout and visual cues commonly present in structured documents, forms, etc. To that end, we ask a mostly unexplored question: "Can we search for other similar snippets present in a target document page given a single query instance of a document snippet?". We propose MONOMER to solve this as a one-shot snippet detection task. MONOMER fuses context from visual, textual, and spatial modalities of snippets and documents to find query snippet in target documents. We conduct extensive ablations and experiments showing MONOMER outperforms several baselines from one-shot object detection (BHRL), template matching, and document understanding (LayoutLMv3). Due to the scarcity of relevant data for the task at hand, we train MONOMER on programmatically generated data having many visually similar query snippets and target document pairs from two datasets - Flamingo Forms and PubLayNet. We also do a human study to validate the generated data.
['Balaji Krishnamurthy', 'Mausoom Sarkar', 'Surgan Jandial', 'Milan Aggarwal', 'Shripad Deshmukh', 'Abhinav Java']
2022-09-12
null
null
null
null
['one-shot-object-detection', 'template-matching']
['computer-vision', 'computer-vision']
[ 4.81995285e-01 -3.05893958e-01 -1.58909589e-01 -2.36648902e-01 -1.01502323e+00 -1.02422452e+00 8.54820967e-01 6.78717434e-01 -1.60216510e-01 3.57282581e-03 3.64764035e-01 -4.70003188e-01 -3.51330966e-01 -4.59001154e-01 -8.23029876e-01 -1.34249359e-01 3.02398741e-01 4.97619569e-01 6.30976737e-01 -6.67911395e-02 8.62246037e-01 4.28685933e-01 -1.58505774e+00 8.72713804e-01 8.85333180e-01 8.82850051e-01 9.09501791e-01 8.00685227e-01 -6.86316967e-01 3.95397693e-01 -6.98705316e-01 -2.64332175e-01 1.57985330e-01 -4.51967008e-02 -9.52955186e-01 2.07697973e-01 1.28900993e+00 -3.11628073e-01 -2.80932605e-01 9.36504781e-01 4.83891129e-01 1.33096531e-01 5.63576341e-01 -1.10687971e+00 -9.24045503e-01 5.39875507e-01 -8.55098903e-01 3.56504530e-01 8.86274159e-01 1.68196693e-01 1.16304159e+00 -1.31617689e+00 9.82151449e-01 1.60710192e+00 3.10920298e-01 7.58718625e-02 -1.28920364e+00 -3.88803154e-01 2.48674810e-01 1.86550081e-01 -1.35447907e+00 -5.73955715e-01 6.01494253e-01 -4.60926175e-01 1.04100418e+00 5.74597299e-01 3.61746877e-01 1.24549282e+00 -2.95937628e-01 1.39554620e+00 4.41602260e-01 -8.23261976e-01 8.21234584e-02 4.71775383e-01 3.83905619e-01 6.32032275e-01 5.52552976e-02 -2.81762332e-01 -7.54721045e-01 -1.25327870e-01 4.09032255e-01 3.38105634e-02 -2.78154314e-01 -6.23559058e-01 -1.14797199e+00 4.78904516e-01 3.62554282e-01 3.46568674e-01 5.53659908e-02 -7.72146061e-02 2.76168585e-01 1.16093561e-01 1.90832868e-01 4.76529509e-01 -6.22816235e-02 -2.76551358e-02 -1.13453937e+00 5.86103797e-01 8.31643581e-01 1.48265111e+00 6.90594912e-01 -5.89957297e-01 -6.55007601e-01 9.12656009e-01 2.61169076e-01 6.67913318e-01 3.66371274e-01 -4.98784155e-01 1.07657778e+00 9.73398030e-01 2.74965137e-01 -1.27291977e+00 -1.65493801e-01 -7.78768137e-02 -1.04896456e-01 -5.22418693e-02 5.20811021e-01 3.36495936e-01 -9.83605862e-01 9.69191849e-01 1.90655738e-01 -6.08146250e-01 -2.93942243e-01 8.65108132e-01 7.65956044e-01 6.90122664e-01 -1.70719832e-01 3.69904429e-01 1.54184628e+00 -1.13616490e+00 -6.29059196e-01 -6.54835999e-01 7.26906955e-01 -1.29313564e+00 1.62252605e+00 2.84178346e-01 -1.02773213e+00 -6.22217000e-01 -1.04362190e+00 -5.52696884e-01 -7.79802918e-01 2.59986997e-01 2.12884054e-01 4.35245395e-01 -7.83172846e-01 3.96553963e-01 -3.47448021e-01 -9.43819225e-01 3.83039534e-01 -8.70182365e-02 -1.14114523e-01 -4.64232743e-01 -5.15397668e-01 5.41786075e-01 3.90312850e-01 -1.42734587e-01 -7.43756592e-01 -7.97977984e-01 -6.11682057e-01 2.52086341e-01 9.13499117e-01 -4.83704358e-01 1.36900508e+00 -8.46643388e-01 -5.86767316e-01 8.93001556e-01 -4.01207685e-01 7.22116753e-02 4.60862517e-01 -4.56612706e-01 -4.87757057e-01 2.04216853e-01 4.71353829e-01 4.67745900e-01 8.86363924e-01 -1.33718312e+00 -7.43112028e-01 -9.04576778e-01 -1.68644875e-01 2.84808576e-01 -5.01182616e-01 3.18790883e-01 -1.10707831e+00 -6.86152101e-01 2.22985730e-01 -6.10743046e-01 4.28653061e-01 4.58086789e-01 -9.21755373e-01 -2.93207288e-01 1.27021074e+00 -7.24476755e-01 1.42718506e+00 -2.29754019e+00 -1.60519570e-01 4.71431673e-01 -6.63130209e-02 6.47958517e-02 -4.80973482e-01 7.91447580e-01 2.22315535e-01 3.03400964e-01 2.30928987e-01 -3.23048770e-01 1.96792603e-01 -3.46738935e-01 -6.54148638e-01 7.20209926e-02 -2.91869432e-01 1.05119848e+00 -9.26184535e-01 -8.12368512e-01 -7.80198723e-02 1.40249714e-01 -2.20688179e-01 1.69908419e-01 -7.52493083e-01 -3.34642678e-01 -5.02550542e-01 1.08832455e+00 6.52428865e-01 -4.40912575e-01 1.49089515e-01 -4.45269287e-01 -5.99748679e-02 3.86924408e-02 -1.26905060e+00 1.96074557e+00 -2.02310920e-01 9.18995321e-01 9.25789550e-02 -3.49739462e-01 6.62643433e-01 -1.15544327e-01 -6.95738941e-02 -1.19558477e+00 -4.51336801e-01 1.44467846e-01 -4.80533868e-01 -8.43485534e-01 9.50214326e-01 7.30228543e-01 7.70290643e-02 6.51567221e-01 -2.62757540e-01 1.83176380e-02 4.92152900e-01 7.24644423e-01 1.18773150e+00 1.97424904e-01 -2.01234326e-01 -1.70369893e-01 1.10359482e-01 3.56076926e-01 -2.79215276e-01 1.36577344e+00 1.79588616e-01 7.25137532e-01 3.04205358e-01 -1.28345713e-01 -1.09929705e+00 -1.14875138e+00 9.89161283e-02 1.66073525e+00 4.37040240e-01 -8.40563416e-01 -5.96252561e-01 -7.08220184e-01 2.38348618e-01 5.52300572e-01 -4.24914718e-01 2.49465406e-01 -6.04499996e-01 -1.65034920e-01 2.87893951e-01 6.70986414e-01 1.92292437e-01 -1.05591583e+00 -7.16982245e-01 -6.60616681e-02 3.06787044e-02 -1.03226495e+00 -9.26167309e-01 5.56195108e-03 -5.78881621e-01 -1.16137719e+00 -8.42739344e-01 -9.02856886e-01 7.35034466e-01 7.53280759e-01 1.35718250e+00 2.16803774e-01 -8.56895804e-01 6.33856297e-01 -3.25908065e-01 -4.44689393e-01 -1.15625136e-01 1.78978220e-01 -4.90993112e-01 -3.62734467e-01 4.03050840e-01 -4.00335006e-02 -8.65202606e-01 4.74955797e-01 -1.14283526e+00 -2.29596142e-02 5.45019925e-01 2.97313392e-01 3.57435495e-01 -4.62536775e-02 -1.71408489e-01 -8.57891083e-01 1.01245522e+00 -3.12429219e-01 -5.58450520e-01 8.17300975e-01 -6.06050849e-01 1.08332179e-01 3.43842417e-01 -3.69115174e-01 -1.05810523e+00 2.56626476e-02 5.77768683e-01 -3.14801037e-01 -2.61594713e-01 3.64612281e-01 -3.39840263e-01 4.07781601e-01 7.87429392e-01 2.30951935e-01 -3.37446302e-01 -9.65584040e-01 6.12994611e-01 7.66265631e-01 7.08430529e-01 -6.25281811e-01 8.30407500e-01 5.49439490e-01 -5.20313621e-01 -8.92804027e-01 -7.18532145e-01 -1.02050138e+00 -6.15375042e-01 -5.03495298e-02 5.80834091e-01 -4.80600893e-01 -5.14687419e-01 5.81278373e-03 -1.32732248e+00 -1.03574127e-01 -4.27699052e-02 -2.45363012e-01 -9.36920941e-02 5.10368466e-01 -3.91600765e-02 -8.20313096e-01 -4.52878743e-01 -8.71570051e-01 1.53013504e+00 1.68290228e-01 -3.73493850e-01 -6.73080981e-01 1.69788599e-02 3.22459310e-01 1.48463085e-01 -2.21525982e-01 1.26176405e+00 -7.71571696e-01 -8.93303156e-01 -4.43787068e-01 -6.68790877e-01 -4.20121819e-01 5.59593216e-02 2.37690195e-01 -9.83439505e-01 -4.40918833e-01 -6.09047472e-01 -2.97333151e-01 8.89177263e-01 -7.39363134e-02 1.38914943e+00 -4.53136206e-01 -1.00779927e+00 3.62971485e-01 1.37823093e+00 4.12803441e-01 2.79542059e-01 5.60281932e-01 7.55123675e-01 9.12161231e-01 6.45277858e-01 5.15645921e-01 1.03427425e-01 9.06823337e-01 3.96543860e-01 -1.54195735e-02 -1.60970971e-01 -6.21488631e-01 -4.82806191e-02 -7.39226937e-02 6.97368622e-01 -8.63487899e-01 -1.18844426e+00 8.34139049e-01 -1.98652101e+00 -9.50128198e-01 -6.21688999e-02 1.88405514e+00 4.92113560e-01 1.63200185e-01 1.13612995e-01 -6.10708036e-02 7.53906071e-01 2.38181010e-01 -5.92729390e-01 -1.46843130e-02 -2.67048955e-01 -2.78563321e-01 2.01306716e-01 1.74405202e-01 -9.79612947e-01 9.02196169e-01 5.49357128e+00 9.49503303e-01 -7.00590014e-01 -3.37080061e-01 3.35989386e-01 -2.63283253e-01 -5.07453501e-01 1.85393631e-01 -1.14774036e+00 5.20689368e-01 3.50485414e-01 -4.38440219e-02 3.50069255e-01 8.46299887e-01 2.47249976e-01 -4.89786744e-01 -1.62087202e+00 1.09163749e+00 4.20925021e-01 -1.36037993e+00 3.11576396e-01 -1.48929164e-01 4.30453807e-01 -3.89226586e-01 2.86313623e-01 9.00660083e-02 1.66708812e-01 -1.15442657e+00 9.24474776e-01 4.68573391e-01 5.87006867e-01 -2.58157700e-01 -4.22845855e-02 3.16789836e-01 -1.12410259e+00 -2.34419614e-01 -2.41841882e-01 5.17844975e-01 -8.63170251e-03 3.23116839e-01 -1.19240868e+00 1.95598155e-01 8.28485548e-01 5.06312609e-01 -1.25546288e+00 1.39360130e+00 1.49156183e-01 -6.59827515e-02 -2.24900872e-01 -3.82032186e-01 2.85709381e-01 1.32068679e-01 5.36131740e-01 1.38701963e+00 2.78254002e-01 -4.48753297e-01 8.60612839e-02 1.20685315e+00 -1.18736170e-01 3.31562549e-01 -5.55037439e-01 -3.41973811e-01 7.54777491e-01 1.24761021e+00 -8.37796926e-01 -3.08317006e-01 -3.75471860e-01 1.29496849e+00 2.10942462e-01 4.65399712e-01 -1.76733643e-01 -7.82639444e-01 6.53569624e-02 5.54024935e-01 4.30150449e-01 3.22400443e-02 -1.25003442e-01 -7.90213823e-01 7.21503973e-01 -9.00093615e-01 4.76258397e-01 -1.26131654e+00 -1.15501904e+00 2.75861204e-01 5.74353077e-02 -1.09503734e+00 -2.13348344e-01 -5.78527808e-01 -7.46184409e-01 8.47711086e-01 -1.03568733e+00 -1.20576227e+00 -6.66142285e-01 4.63770241e-01 1.10043669e+00 -5.02115786e-02 5.19857883e-01 4.64106426e-02 -5.64298749e-01 4.78870094e-01 4.92309511e-01 2.38904133e-01 9.48294759e-01 -1.40514410e+00 6.66197896e-01 7.73678958e-01 5.64842284e-01 8.88854444e-01 6.21465087e-01 -8.62157404e-01 -1.64627707e+00 -7.52171457e-01 7.83935964e-01 -7.73862720e-01 6.31692052e-01 -8.76011550e-01 -9.93349433e-01 6.22188687e-01 3.49127591e-01 -2.21599966e-01 3.93899351e-01 8.34673792e-02 -5.70592463e-01 9.59588811e-02 -6.45691037e-01 8.04880083e-01 1.09135318e+00 -8.24875414e-01 -5.37200212e-01 7.30261981e-01 5.43294132e-01 -2.53630757e-01 -2.15684757e-01 -5.95271327e-02 7.41829455e-01 -7.84801602e-01 1.20421183e+00 -6.34120584e-01 3.55225980e-01 -1.16175942e-01 -1.11277975e-01 -9.32725072e-01 6.06022365e-02 -6.24787748e-01 -2.72088021e-01 1.43837357e+00 5.82079053e-01 3.96047324e-01 8.63868713e-01 6.53478444e-01 3.99562018e-03 -3.99257660e-01 -4.11844909e-01 -7.68171370e-01 -3.05558443e-01 -3.38548452e-01 6.14035189e-01 5.41377664e-01 3.08474898e-01 4.72496390e-01 6.05179816e-02 -9.20785293e-02 4.51893359e-01 5.08545339e-01 1.02368343e+00 -1.07252157e+00 -1.94955990e-01 -5.40564477e-01 1.83616951e-01 -1.45630777e+00 -1.51653826e-01 -9.42983806e-01 5.41236736e-02 -1.75186312e+00 6.02238595e-01 -2.57225990e-01 1.36824429e-01 2.36798555e-01 -5.64119918e-03 -2.69151032e-01 3.31252873e-01 4.80247796e-01 -9.74028111e-01 1.67699590e-01 9.71362293e-01 -6.54964507e-01 -2.85886168e-01 -1.93251997e-01 -6.20709300e-01 5.74691117e-01 2.25104526e-01 -2.10212275e-01 -3.97493243e-01 -7.81663537e-01 4.59737420e-01 -9.83635895e-03 4.70258832e-01 -8.66371334e-01 5.02310812e-01 -9.53051299e-02 6.90378428e-01 -1.22999692e+00 2.56250948e-01 -8.57337177e-01 -4.66725945e-01 -1.41971949e-02 -5.97504735e-01 3.74314338e-01 2.96536207e-01 8.00375819e-01 2.33623773e-01 -6.27162874e-01 1.08405299e-01 -2.67968267e-01 -1.03910685e+00 -6.78876713e-02 -2.13544190e-01 1.53363869e-01 7.15515137e-01 -6.22017801e-01 -8.93025517e-01 -1.14336543e-01 -4.91473228e-01 4.44887102e-01 5.56129038e-01 8.69078755e-01 8.63983750e-01 -9.54703510e-01 -2.42977485e-01 7.87985995e-02 7.07165837e-01 -1.46503642e-01 1.39871221e-02 2.20172450e-01 -3.89451593e-01 8.39682460e-01 1.29809096e-01 -6.08121812e-01 -1.39146662e+00 9.72161472e-01 -5.88347912e-02 1.04135029e-01 -6.74522817e-01 7.23452389e-01 2.71815807e-01 -1.69948891e-01 8.24044585e-01 -2.07422033e-01 -4.27432880e-02 2.42587820e-01 6.77408695e-01 5.20689785e-01 2.01918766e-01 -1.82765275e-01 -3.59376907e-01 5.63145995e-01 -5.28377116e-01 -2.18894839e-01 9.16012168e-01 -3.85461390e-01 2.44646668e-01 2.60104060e-01 1.38532889e+00 8.23192075e-02 -1.04566598e+00 -4.62091267e-01 7.30959535e-01 -7.57203937e-01 -9.67058167e-02 -1.06324482e+00 -4.72837776e-01 8.39957654e-01 8.26619983e-01 3.13859493e-01 6.05300248e-01 3.79487425e-01 6.43444598e-01 7.42017865e-01 -1.14136122e-01 -1.34413779e+00 7.56280124e-01 1.98258787e-01 1.14712918e+00 -1.37493503e+00 1.42829210e-01 -3.94786835e-01 -2.31422484e-01 1.12537467e+00 9.54597116e-01 4.46201861e-01 2.44306341e-01 1.75339535e-01 -1.30071357e-01 -5.62106550e-01 -6.71806872e-01 -8.74437615e-02 6.91831350e-01 4.81595933e-01 3.86728406e-01 -5.23942530e-01 3.41305435e-01 2.51398444e-01 -2.88020782e-02 -3.83557707e-01 1.46778166e-01 1.38610303e+00 -5.97665727e-01 -8.45547915e-01 -6.07132077e-01 5.87027252e-01 -1.59131378e-01 -3.23777527e-01 -7.98434317e-01 7.69211709e-01 -3.87689710e-01 7.98743486e-01 4.00147647e-01 4.65474203e-02 4.94545519e-01 2.61972863e-02 3.17219168e-01 -8.40397060e-01 -5.19469857e-01 2.77983218e-01 -1.34885117e-01 -6.77881896e-01 2.73567010e-02 -5.41974187e-01 -8.93372774e-01 1.33107349e-01 -3.44049662e-01 -8.36173519e-02 7.58189261e-01 7.24569261e-01 5.37687659e-01 3.19002748e-01 5.47293276e-02 -5.90346515e-01 -4.97963220e-01 -8.58497202e-01 -4.09348071e-01 7.30231166e-01 2.53542185e-01 -3.24144959e-01 -5.40635623e-02 1.17188990e-01]
[11.548048973083496, 2.2812037467956543]
71fb3303-68a2-42a8-8b5e-ee387ce13bb2
semantic-and-syntactic-enhanced-aspect
2106.03315
null
https://arxiv.org/abs/2106.03315v1
https://arxiv.org/pdf/2106.03315v1.pdf
Semantic and Syntactic Enhanced Aspect Sentiment Triplet Extraction
Aspect Sentiment Triplet Extraction (ASTE) aims to extract triplets from sentences, where each triplet includes an entity, its associated sentiment, and the opinion span explaining the reason for the sentiment. Most existing research addresses this problem in a multi-stage pipeline manner, which neglects the mutual information between such three elements and has the problem of error propagation. In this paper, we propose a Semantic and Syntactic Enhanced aspect Sentiment triplet Extraction model (S3E2) to fully exploit the syntactic and semantic relationships between the triplet elements and jointly extract them. Specifically, we design a Graph-Sequence duel representation and modeling paradigm for the task of ASTE: we represent the semantic and syntactic relationships between word pairs in a sentence by graph and encode it by Graph Neural Networks (GNNs), as well as modeling the original sentence by LSTM to preserve the sequential information. Under this setting, we further apply a more efficient inference strategy for the extraction of triplets. Extensive evaluations on four benchmark datasets show that S3E2 significantly outperforms existing approaches, which proves our S3E2's superiority and flexibility in an end-to-end fashion.
['Hai Jin', 'Xuanhua Shi', 'Bang Liu', 'Hong Huang', 'Zhexue Chen']
2021-06-07
null
https://aclanthology.org/2021.findings-acl.128
https://aclanthology.org/2021.findings-acl.128.pdf
findings-acl-2021-8
['aspect-sentiment-triplet-extraction']
['natural-language-processing']
[ 1.81466490e-01 1.14383437e-01 -1.22764066e-01 -6.02934480e-01 -4.15840030e-01 -5.14232457e-01 3.06982249e-01 4.07593548e-01 -3.06840360e-01 3.09198350e-01 5.08181989e-01 -4.61225957e-01 1.07411638e-01 -8.86173368e-01 -7.08166063e-01 -3.38194549e-01 2.17348039e-01 2.06854522e-01 1.03009155e-03 -3.29800963e-01 1.12846181e-01 -1.18535191e-01 -9.75556612e-01 3.69770795e-01 8.50147486e-01 1.17219126e+00 -1.16377771e-01 4.00528550e-01 -6.36933565e-01 1.07451928e+00 -4.70968843e-01 -9.55439091e-01 -4.50173318e-02 -4.18774098e-01 -6.68066502e-01 2.93433011e-01 1.05879955e-01 2.00810842e-02 -2.96690226e-01 1.15800047e+00 1.29302487e-01 -8.00681785e-02 4.24366206e-01 -1.21868873e+00 -7.98669636e-01 9.84926641e-01 -8.70222628e-01 -1.39931157e-01 2.55045533e-01 3.34861167e-02 1.68896306e+00 -9.45731461e-01 4.27626848e-01 1.18309247e+00 6.96235418e-01 2.15304419e-01 -8.28971386e-01 -3.87870729e-01 5.40214539e-01 7.80859590e-02 -9.42091703e-01 -1.83610424e-01 9.70172226e-01 -1.11746296e-01 1.31154752e+00 1.40988246e-01 8.73329937e-01 9.96583104e-01 3.18774730e-01 1.12039948e+00 8.31145465e-01 -2.97668785e-01 2.78023425e-02 3.64322625e-02 7.38794982e-01 9.60132778e-01 3.73860866e-01 -3.75729740e-01 -6.19414866e-01 -1.88976191e-02 1.24504328e-01 1.32900933e-02 -2.10139006e-02 -7.05375522e-02 -9.84315515e-01 7.40605712e-01 6.06758535e-01 4.13780183e-01 -5.43004572e-01 6.17822036e-02 5.60112000e-01 3.66247803e-01 7.04151928e-01 2.04029232e-01 -6.72087550e-01 2.52015233e-01 -5.07055879e-01 -7.21589774e-02 1.03372729e+00 1.03167355e+00 8.41576934e-01 -8.68196189e-02 -3.74050856e-01 6.30375564e-01 5.38859248e-01 4.01157200e-01 4.95331258e-01 -1.68781668e-01 8.17918122e-01 1.14846563e+00 -3.30242336e-01 -1.43993664e+00 -4.40986127e-01 -6.00230455e-01 -9.18526232e-01 -5.35875797e-01 -3.18316549e-01 -3.94712329e-01 -8.93372715e-01 1.74362874e+00 5.44341505e-01 8.72434899e-02 1.70250058e-01 5.68508744e-01 1.20033538e+00 5.01352966e-01 3.23213786e-02 -1.13366812e-01 1.68005919e+00 -1.26560175e+00 -8.64783108e-01 -7.21923053e-01 7.45364070e-01 -6.62823737e-01 7.90535748e-01 -1.65332332e-01 -7.45153487e-01 -1.15300231e-01 -1.01087546e+00 -2.82397777e-01 -3.62163514e-01 2.17695832e-01 7.92452931e-01 3.00078362e-01 -9.11164522e-01 4.60127652e-01 -7.57163525e-01 -5.63344173e-02 3.77954543e-01 3.20208043e-01 -3.41653377e-01 2.08456535e-02 -1.32146561e+00 6.46448731e-01 6.40318274e-01 6.32038653e-01 -1.84608661e-02 -4.47756350e-01 -1.49397099e+00 2.95060456e-01 6.80447161e-01 -1.17635953e+00 9.90551770e-01 -8.81204545e-01 -1.29150546e+00 4.91993368e-01 -5.31159997e-01 -3.71158212e-01 -2.37691570e-02 -9.69725624e-02 -4.33832616e-01 -3.97092737e-02 3.04105878e-01 1.74989969e-01 6.66060090e-01 -9.74170625e-01 -4.70146567e-01 -6.14131272e-01 3.88033956e-01 2.26218686e-01 -6.71493769e-01 1.60728544e-01 -7.79268384e-01 -5.72685421e-01 1.48966178e-01 -9.02701795e-01 -3.16316545e-01 -5.59237659e-01 -9.00911450e-01 -4.35981303e-01 6.01068437e-01 -7.50591278e-01 1.43020117e+00 -2.10704565e+00 1.84823081e-01 2.34855816e-01 6.06607556e-01 2.77629197e-01 -3.73154044e-01 5.58996201e-01 5.01674823e-02 9.99293029e-02 -5.45993328e-01 -8.32978427e-01 2.81328589e-01 1.75176948e-01 -2.33076945e-01 -3.56191769e-02 5.01433730e-01 1.39207852e+00 -9.23781514e-01 -6.26275659e-01 -1.48039207e-01 4.83194381e-01 -5.23245871e-01 1.99971884e-01 -3.64427119e-01 8.62409302e-04 -7.66025543e-01 5.11741579e-01 5.86817861e-01 -6.88267529e-01 4.46039975e-01 -5.42666733e-01 1.60541371e-01 6.51765347e-01 -8.04859757e-01 1.52122164e+00 -5.58623731e-01 3.25883478e-01 -2.66622365e-01 -6.86660886e-01 9.54104125e-01 9.85984728e-02 4.03367132e-01 -7.53314555e-01 2.91952819e-01 1.94514215e-01 -1.26173168e-01 -6.51577830e-01 6.91530406e-01 -2.54616171e-01 -2.40919709e-01 5.48600495e-01 1.96732983e-01 2.01053202e-01 5.48699200e-01 5.46557248e-01 1.01708591e+00 1.66283771e-01 3.56141865e-01 1.46208435e-01 6.33961201e-01 -2.18286201e-01 6.59840524e-01 3.42543930e-01 2.18188256e-01 2.39548683e-01 9.06028032e-01 -2.74286956e-01 -6.76557899e-01 -3.99954766e-01 6.05791628e-01 7.00378418e-01 9.42625403e-02 -1.00649488e+00 -5.86251199e-01 -1.16656709e+00 -9.62709412e-02 8.29642415e-01 -7.68602729e-01 -3.65938067e-01 -7.07407653e-01 -9.23529506e-01 1.25667721e-01 6.75085664e-01 5.45620382e-01 -1.01610565e+00 -5.12863509e-02 2.52777427e-01 -5.24963140e-01 -1.55941987e+00 -6.64705634e-01 9.63535567e-04 -5.77160358e-01 -1.11983120e+00 -1.62781775e-01 -8.03026736e-01 7.07481623e-01 4.05789018e-01 1.33829021e+00 1.45754874e-01 2.54608244e-01 1.01572294e-02 -5.37328184e-01 -2.24006906e-01 -9.32125375e-02 1.64533034e-01 -4.57136601e-01 5.07945478e-01 7.23766923e-01 -5.90092182e-01 -4.60669219e-01 -2.09304109e-01 -1.00574934e+00 2.15136796e-01 8.05380762e-01 7.48106360e-01 6.09869480e-01 7.87628740e-02 3.39761287e-01 -1.43391585e+00 8.30034018e-01 -6.00054204e-01 -3.95123482e-01 4.24981177e-01 -5.58679402e-01 1.77450180e-01 7.86256194e-01 1.46230564e-01 -9.21432436e-01 -2.14032933e-01 -2.24357963e-01 -2.30675444e-01 2.74399340e-01 1.18596768e+00 -4.03447390e-01 3.24446172e-01 -1.18211761e-01 3.85371923e-01 -2.97646940e-01 -4.05655950e-01 3.76043111e-01 5.84285319e-01 2.34148487e-01 -2.99910039e-01 7.07580924e-01 3.00366044e-01 3.92559431e-02 -5.94942808e-01 -1.63187146e+00 -4.26076889e-01 -5.62801600e-01 1.00270763e-01 7.11151183e-01 -8.43263566e-01 -6.91498280e-01 4.40503359e-01 -1.28946030e+00 2.53100604e-01 -1.20881066e-01 3.20682853e-01 -1.20094590e-01 5.22430599e-01 -6.63745284e-01 -6.11070871e-01 -8.23449910e-01 -1.07100952e+00 1.38578117e+00 1.34532064e-01 -3.00819930e-02 -1.20211542e+00 5.90214990e-02 4.82691854e-01 2.28926033e-01 3.73641729e-01 1.17755544e+00 -7.78808475e-01 -4.22130346e-01 -2.92841882e-01 -4.54294860e-01 4.48143512e-01 2.53207803e-01 -1.23830222e-01 -6.41787648e-01 -2.07382768e-01 1.08773544e-01 -2.28056796e-02 1.05102849e+00 2.59449109e-02 7.67969310e-01 -6.12346113e-01 -1.21591337e-01 4.76063550e-01 1.48394918e+00 -2.17525184e-01 2.58051515e-01 3.19943249e-01 1.28896928e+00 6.64573491e-01 2.98633844e-01 1.31315961e-01 1.06994915e+00 3.72497916e-01 4.21507746e-01 -1.43203527e-01 -1.47871017e-01 -4.71386015e-01 5.83298743e-01 1.67603827e+00 4.10069287e-01 -5.97589731e-01 -5.56629837e-01 4.70155329e-01 -1.99375713e+00 -4.29894835e-01 -2.92963862e-01 1.64703083e+00 5.81694901e-01 3.44672173e-01 -2.83565391e-02 -2.92614326e-02 6.90213680e-01 5.64489722e-01 -4.86587793e-01 -3.40799302e-01 -2.16606572e-01 -3.14709358e-02 2.04535753e-01 3.61435115e-01 -9.27493870e-01 1.10991287e+00 4.86342287e+00 6.87460661e-01 -9.00689840e-01 -6.66557401e-02 4.31438416e-01 2.39824399e-01 -8.08900595e-01 3.44099045e-01 -7.42379725e-01 3.79502684e-01 8.21647763e-01 -2.38212481e-01 1.30644143e-01 4.07522857e-01 -3.14717293e-02 3.36602837e-01 -9.36340809e-01 6.10275686e-01 3.56178194e-01 -1.05396545e+00 4.49412137e-01 -1.49129719e-01 5.73888659e-01 4.93045412e-02 -1.41805828e-01 4.30904925e-01 2.44482905e-01 -5.67877889e-01 5.79469025e-01 4.59505469e-01 3.28180522e-01 -7.29663849e-01 9.93617535e-01 1.34306386e-01 -1.53236628e+00 3.50825042e-02 -1.85804397e-01 1.09334417e-01 3.19701225e-01 8.64321589e-01 -6.96725965e-01 1.24725878e+00 3.65575761e-01 1.30452168e+00 -7.55176961e-01 6.68117881e-01 -7.49695241e-01 6.10539317e-01 -1.14755780e-01 -4.19615060e-01 6.47465169e-01 -5.09603560e-01 6.13936603e-01 1.21583915e+00 1.62434623e-01 -1.65934756e-01 1.42033145e-01 7.91961432e-01 -3.94976884e-01 3.49577188e-01 -5.16821921e-01 -4.32763219e-01 2.37551883e-01 1.51974332e+00 -7.63229251e-01 -4.43546355e-01 -7.97830522e-01 9.37224269e-01 7.84680009e-01 2.98010916e-01 -6.32019699e-01 -6.03313863e-01 3.18452597e-01 -4.62014109e-01 6.54306233e-01 -1.35578781e-01 -2.83011943e-01 -1.53581583e+00 6.47763729e-01 -8.87181222e-01 4.94829923e-01 -7.08520412e-01 -1.35459554e+00 9.75146472e-01 -3.95792872e-01 -1.05980992e+00 -2.30256245e-01 -6.29952788e-01 -6.66582942e-01 5.49232125e-01 -1.85888648e+00 -1.57157350e+00 1.88032463e-02 4.81254905e-01 3.41004401e-01 9.39615518e-02 6.02319360e-01 2.49372348e-01 -9.00030911e-01 5.85388660e-01 -3.00792366e-01 4.57682371e-01 1.09870814e-01 -1.34793377e+00 9.77060139e-01 1.03545213e+00 3.94656897e-01 8.07417929e-01 4.42535192e-01 -7.11514711e-01 -1.70367801e+00 -1.33990085e+00 1.63523412e+00 -3.38988662e-01 1.08905947e+00 -5.33358574e-01 -8.51878226e-01 9.78791714e-01 4.18856919e-01 -2.67916888e-01 7.32584417e-01 3.27656388e-01 -5.98384678e-01 -5.77675877e-04 -5.66570401e-01 7.38733232e-01 1.03369570e+00 -6.14695072e-01 -7.54410982e-01 3.90915632e-01 1.37873507e+00 -3.08865368e-01 -6.80332184e-01 5.29143333e-01 2.93590873e-01 -8.96489918e-01 5.32647610e-01 -8.42402756e-01 7.45768726e-01 -2.13527769e-01 -1.01779364e-01 -1.44952524e+00 -1.01253569e-01 -5.59963226e-01 -3.98364961e-01 1.59259748e+00 7.93102145e-01 -7.41460204e-01 4.60758269e-01 4.47497547e-01 -1.87822804e-01 -1.11128104e+00 -5.00757873e-01 -5.37386537e-01 -3.23814809e-01 -3.26492578e-01 1.02431774e+00 7.15790033e-01 -4.68305610e-02 1.16569340e+00 -4.70639050e-01 8.34154263e-02 4.96611029e-01 8.32603872e-01 4.79894280e-01 -9.11814868e-01 -3.31356764e-01 -3.94280523e-01 -1.99102864e-01 -1.06183803e+00 6.01886511e-01 -1.04568148e+00 -4.98986170e-02 -1.89557004e+00 4.95458484e-01 -4.45279144e-02 -5.02047479e-01 6.03714883e-01 -6.01121187e-01 1.28065972e-02 1.91510320e-01 -6.40528426e-02 -9.19415355e-01 1.05971980e+00 1.31614029e+00 -3.20652992e-01 5.90680949e-02 -1.35253683e-01 -1.20393527e+00 7.88213611e-01 5.27935028e-01 -5.34630179e-01 -3.38852644e-01 -8.29855978e-01 6.93327725e-01 -4.76044528e-02 1.09418489e-01 -4.14435714e-01 5.02629757e-01 1.92262188e-01 -5.07816561e-02 -6.46286547e-01 1.16893157e-01 -9.41616297e-01 -2.89949864e-01 2.24345550e-01 -2.50049055e-01 3.16059977e-01 8.15146323e-03 7.36503422e-01 -5.95509231e-01 -1.03365533e-01 4.62201014e-02 -6.01963028e-02 -5.02797067e-01 6.40646994e-01 1.99413270e-01 1.80193931e-01 5.17388225e-01 2.93835342e-01 -4.06054825e-01 -2.56262064e-01 -4.68734741e-01 5.98414719e-01 1.06319517e-01 6.37047350e-01 7.10482955e-01 -1.44352365e+00 -8.16033423e-01 2.47480646e-01 2.78499156e-01 -4.45109867e-02 2.44567111e-01 8.80443156e-01 8.71129185e-02 2.64022261e-01 3.33258331e-01 -2.16206685e-01 -1.21968865e+00 5.50682902e-01 1.01931341e-01 -8.01630855e-01 -6.23395681e-01 7.73685575e-01 1.75810501e-01 -5.84190607e-01 -2.61682391e-01 -3.20965409e-01 -6.38000071e-01 1.23114191e-01 1.48238182e-01 -1.81082845e-01 1.74401119e-01 -8.79338622e-01 -4.64837253e-01 5.24740577e-01 -4.02600318e-01 1.69127762e-01 1.46906388e+00 -1.89757943e-01 -7.12464571e-01 3.43403578e-01 1.44685411e+00 4.87963669e-02 -6.58579111e-01 -5.38969576e-01 3.01569372e-01 -1.95611522e-01 -4.19605002e-02 -4.65794593e-01 -1.39523554e+00 6.16104007e-01 -4.85182583e-01 2.96170264e-01 1.09739077e+00 -2.28344873e-02 1.36333549e+00 3.83421510e-01 -2.53328378e-03 -6.56478107e-01 -6.79346323e-02 9.08075392e-01 6.72920227e-01 -1.11686051e+00 -1.03776693e-01 -6.74080431e-01 -7.42194772e-01 9.70399618e-01 5.50125837e-01 -5.44091314e-02 6.06495321e-01 8.09255466e-02 -7.66220242e-02 -5.90680718e-01 -9.35001314e-01 -3.61829877e-01 4.46952820e-01 2.13530213e-02 3.71456683e-01 4.90343235e-02 -3.69021088e-01 7.87070632e-01 -3.38540375e-01 -1.41973317e-01 2.61698544e-01 8.24207842e-01 -9.57079008e-02 -1.13622212e+00 4.63892221e-01 4.62118149e-01 -4.41889971e-01 -6.45601988e-01 -6.73560560e-01 4.92390275e-01 -9.32605490e-02 1.09220648e+00 -2.68236041e-01 -6.45556331e-01 5.75366378e-01 -4.76304144e-02 1.31058753e-01 -6.43709898e-01 -8.98522615e-01 1.66290879e-01 3.91349792e-01 -5.65236568e-01 -5.72363377e-01 -3.65188479e-01 -1.27291369e+00 -6.11977279e-02 -4.44014788e-01 3.92322332e-01 6.82871461e-01 1.30045974e+00 6.80695117e-01 9.95284319e-01 9.61548328e-01 -1.24571390e-01 -4.34285492e-01 -9.57004309e-01 -5.29295743e-01 4.75160867e-01 4.02453125e-01 -2.67116845e-01 -2.56330997e-01 -3.00506651e-01]
[11.519247055053711, 6.619039535522461]
5c600441-50aa-46f0-b140-bb66842aa4b4
draformer-differentially-reconstructed
2206.05495
null
https://arxiv.org/abs/2206.05495v1
https://arxiv.org/pdf/2206.05495v1.pdf
DRAformer: Differentially Reconstructed Attention Transformer for Time-Series Forecasting
Time-series forecasting plays an important role in many real-world scenarios, such as equipment life cycle forecasting, weather forecasting, and traffic flow forecasting. It can be observed from recent research that a variety of transformer-based models have shown remarkable results in time-series forecasting. However, there are still some issues that limit the ability of transformer-based models on time-series forecasting tasks: (i) learning directly on raw data is susceptible to noise due to its complex and unstable feature representation; (ii) the self-attention mechanisms pay insufficient attention to changing features and temporal dependencies. In order to solve these two problems, we propose a transformer-based differentially reconstructed attention model DRAformer. Specifically, DRAformer has the following innovations: (i) learning against differenced sequences, which preserves clear and stable sequence features by differencing and highlights the changing properties of sequences; (ii) the reconstructed attention: integrated distance attention exhibits sequential distance through a learnable Gaussian kernel, distributed difference attention calculates distribution difference by mapping the difference sequence to the adaptive feature space, and the combination of the two effectively focuses on the sequences with prominent associations; (iii) the reconstructed decoder input, which extracts sequence features by integrating variation information and temporal correlations, thereby obtaining a more comprehensive sequence representation. Extensive experiments on four large-scale datasets demonstrate that DRAformer outperforms state-of-the-art baselines.
['Zhen Jia', 'Jie Hu', 'Tianrui Li', 'Shengdong Du', 'Benhan Li']
2022-06-11
null
null
null
null
['weather-forecasting']
['miscellaneous']
[ 3.29614848e-01 -6.78442955e-01 -5.48830703e-02 -2.16809273e-01 -4.51980114e-01 -4.83548939e-01 4.81084675e-01 -1.69433281e-01 -7.09845275e-02 4.66922790e-01 4.80290234e-01 -4.37734872e-01 -1.19066603e-01 -5.14680803e-01 -5.96832037e-01 -9.80459988e-01 -1.71934292e-01 -1.34591945e-02 3.10114861e-01 -5.66691816e-01 4.21840072e-01 4.37212318e-01 -1.53487051e+00 2.90425897e-01 1.06909823e+00 1.43364549e+00 3.49602371e-01 4.16638374e-01 -2.87579566e-01 9.30442512e-01 -6.40168786e-01 -1.67485565e-01 5.70640825e-02 -5.50980449e-01 -3.74838352e-01 -2.02052385e-01 -2.02486441e-01 -2.08817497e-01 -5.73355675e-01 9.24356043e-01 4.26581413e-01 2.28655681e-01 5.48042059e-01 -1.31927919e+00 -1.23438716e+00 4.53935266e-01 -5.86431861e-01 7.98159957e-01 1.55907452e-01 4.66015309e-01 7.78894603e-01 -8.65499556e-01 1.71665013e-01 1.31371391e+00 8.07768404e-01 2.28300124e-01 -8.95125687e-01 -7.64691293e-01 4.34121370e-01 6.38611555e-01 -1.15525055e+00 -1.26058295e-01 9.95984733e-01 -5.22073686e-01 1.29610336e+00 4.41852361e-01 4.62675720e-01 1.25022697e+00 8.40547323e-01 8.94200206e-01 7.06141770e-01 -9.81630236e-02 8.30020010e-02 -2.53332406e-01 1.43697798e-01 1.06469877e-01 -5.11050045e-01 2.46829599e-01 -3.34289134e-01 2.75205970e-02 5.31913221e-01 4.38452154e-01 -5.70184052e-01 9.30860043e-02 -1.49183714e+00 5.31854153e-01 2.45631516e-01 7.52953708e-01 -6.09496415e-01 -4.20955308e-02 7.14359820e-01 6.03465438e-01 6.87631845e-01 1.64180607e-01 -5.97063005e-01 -4.75992948e-01 -5.75782359e-01 4.77677286e-02 3.04527491e-01 8.22929740e-01 5.04309952e-01 4.66326535e-01 -5.32233298e-01 6.56922340e-01 -2.97780354e-02 5.62709987e-01 1.01012766e+00 -3.74876916e-01 5.64850748e-01 3.47558677e-01 -8.06107074e-02 -1.26703632e+00 -3.17474216e-01 -6.12784624e-01 -1.11208355e+00 -3.36691201e-01 1.52884156e-01 5.62395602e-02 -7.09019899e-01 1.75504088e+00 -4.29567285e-02 5.97600698e-01 -2.75650680e-01 7.19240904e-01 5.81950009e-01 1.01284206e+00 -1.44087791e-01 -5.17767072e-01 1.04523206e+00 -8.96025658e-01 -1.15235198e+00 8.21530744e-02 3.76991391e-01 -7.14131951e-01 1.00533879e+00 2.02835545e-01 -7.48141110e-01 -9.55937088e-01 -1.05350101e+00 3.80611718e-02 -3.65566224e-01 -1.44020855e-01 1.83704227e-01 4.90034252e-01 -8.93615723e-01 7.05352366e-01 -6.60944760e-01 -2.25667343e-01 1.41727388e-01 3.99748161e-02 -4.72993478e-02 1.07288130e-01 -1.59506583e+00 7.80121386e-01 2.77342200e-01 3.16126198e-01 -6.35782123e-01 -1.02631712e+00 -8.78163278e-01 3.33589405e-01 1.55017093e-01 -3.15148741e-01 1.15287423e+00 -9.59238172e-01 -1.65940011e+00 1.30341470e-01 -3.24174672e-01 -3.83283406e-01 5.42435408e-01 -2.33756885e-01 -1.01401901e+00 -1.94451764e-01 8.49030167e-03 -2.05295011e-02 9.19881284e-01 -5.30930936e-01 -6.72679901e-01 -2.35357076e-01 -5.46308696e-01 -1.22520074e-01 -4.11066324e-01 -5.00083528e-02 -2.82151192e-01 -1.20148504e+00 -3.19763534e-02 -7.62475491e-01 -1.05456904e-01 -3.25696647e-01 -1.64646834e-01 -4.16340649e-01 1.28079093e+00 -9.10435915e-01 1.81874120e+00 -2.37993574e+00 1.91477928e-02 1.23964949e-02 -5.97791839e-03 2.86610872e-01 -2.96821356e-01 7.48678684e-01 -4.44650739e-01 3.44081520e-04 -1.26998127e-01 4.38976102e-02 -3.38598490e-02 4.27615307e-02 -9.25516844e-01 4.51019704e-01 3.12544167e-01 1.04016066e+00 -1.09830785e+00 -1.33247105e-02 3.90812695e-01 5.20649612e-01 -1.90745890e-01 9.85882133e-02 -1.02663741e-01 6.49264872e-01 -4.40983355e-01 3.49064529e-01 6.47283375e-01 -2.67085761e-01 -3.18138786e-02 -2.80414402e-01 -2.29196429e-01 3.62311393e-01 -6.77193820e-01 1.40025747e+00 -1.92100659e-01 7.37958848e-01 -6.46572948e-01 -1.11852109e+00 1.15506160e+00 3.34961027e-01 7.45681405e-01 -1.13411582e+00 4.21853773e-02 1.12864198e-02 5.29077416e-03 -7.04986691e-01 4.29554731e-01 -7.37416968e-02 3.09523791e-02 4.61452007e-01 -7.82542080e-02 3.40362281e-01 -1.58823282e-01 -5.59060695e-03 1.16986573e+00 1.05470985e-01 1.11267105e-01 -1.39753625e-01 6.66730702e-01 -5.83678782e-01 8.17334473e-01 3.22772473e-01 -3.00523132e-01 5.13833165e-01 4.43365902e-01 -6.63752794e-01 -8.96005988e-01 -8.19751382e-01 9.88427326e-02 1.14894199e+00 1.06252238e-01 -2.06325859e-01 -3.36704612e-01 -5.40785193e-01 -4.78249788e-02 9.39010680e-01 -7.84495234e-01 -5.33830523e-01 -6.98854864e-01 -4.87790197e-01 3.84164363e-01 7.81830490e-01 3.19361657e-01 -1.18623328e+00 -3.16287488e-01 4.94182825e-01 -5.13004422e-01 -7.16192186e-01 -1.13210583e+00 2.16500357e-01 -7.80679584e-01 -7.98926890e-01 -8.24761868e-01 -6.75985456e-01 2.15385437e-01 4.40398395e-01 1.01461053e+00 -1.57433718e-01 1.10943638e-01 1.45635024e-01 -4.73760992e-01 -3.42408359e-01 -2.54544020e-01 -2.90136095e-02 3.30538377e-02 3.50703329e-01 5.36803126e-01 -7.04143763e-01 -4.95125264e-01 4.42245185e-01 -9.11743820e-01 -2.96461821e-01 3.33093375e-01 9.58297968e-01 3.88662726e-01 2.52473384e-01 9.01993394e-01 -5.17736614e-01 8.08346331e-01 -8.12116504e-01 -3.46555799e-01 3.60063851e-01 -5.48989892e-01 3.48119065e-02 9.11614239e-01 -8.00126970e-01 -9.83921170e-01 -3.88916939e-01 -3.16607356e-01 -7.99536586e-01 1.49537906e-01 6.22338355e-01 -2.04394504e-01 3.60680550e-01 3.49565476e-01 9.87362504e-01 1.48550406e-01 -4.08544153e-01 6.18991032e-02 6.14399374e-01 5.30201018e-01 -3.53921294e-01 6.47780359e-01 2.52882749e-01 -3.54068249e-01 -7.18642414e-01 -4.27795291e-01 -2.56868273e-01 -4.55578983e-01 -2.35310331e-01 6.34981096e-01 -6.41084075e-01 -7.97776103e-01 7.97502458e-01 -1.08322680e+00 -1.91820279e-01 -1.48121640e-01 4.62280750e-01 -5.16048312e-01 6.99892282e-01 -8.18873346e-01 -8.39265943e-01 -3.03829968e-01 -1.05826747e+00 9.28000987e-01 9.71327499e-02 -2.79626369e-01 -1.12066245e+00 8.90067145e-02 -1.75521716e-01 6.97181106e-01 1.68843806e-01 1.23078740e+00 -6.23750806e-01 -4.30006325e-01 -5.40288761e-02 8.79853442e-02 3.40595782e-01 4.01823729e-01 3.39747630e-02 -8.16590548e-01 -3.06234002e-01 3.00022542e-01 1.42015055e-01 7.15100288e-01 3.83664489e-01 1.50314355e+00 -2.05007032e-01 -3.48898411e-01 5.55518627e-01 9.22709763e-01 7.63057232e-01 9.69716072e-01 2.57299364e-01 4.87870961e-01 4.13128585e-01 6.09904706e-01 5.62588990e-01 4.05530095e-01 4.89764631e-01 2.86809176e-01 1.23673506e-01 1.86480686e-01 -4.26066667e-01 4.85089213e-01 1.37569869e+00 4.46074642e-02 -4.74632084e-01 -6.32515371e-01 5.79090476e-01 -2.02673125e+00 -1.42028666e+00 6.04220070e-02 2.12881041e+00 5.97058475e-01 1.95256978e-01 9.07438174e-02 2.52978206e-01 7.99546063e-01 4.20775443e-01 -9.91255045e-01 -1.60916686e-01 -2.14130268e-01 -2.16680229e-01 9.67368409e-02 -9.51875150e-02 -9.02669609e-01 3.92482013e-01 6.09421587e+00 1.00837946e+00 -1.52620375e+00 5.25261723e-02 7.23142922e-01 7.86156431e-02 -5.73321342e-01 -4.46143180e-01 -3.21052134e-01 1.18788135e+00 8.65017712e-01 -4.83500332e-01 3.86275202e-01 5.63433230e-01 2.31574818e-01 6.03911579e-01 -1.04402602e+00 1.08464801e+00 9.47187394e-02 -1.20068884e+00 4.68603782e-02 -1.23148911e-01 5.38914859e-01 4.20146063e-02 3.17333370e-01 6.70332849e-01 7.36424476e-02 -8.55857432e-01 9.09339368e-01 7.03746498e-01 6.34708762e-01 -7.54653573e-01 7.61824250e-01 4.18161333e-01 -1.48204970e+00 -2.50064969e-01 -1.63251698e-01 1.12362122e-02 2.52063990e-01 6.57776654e-01 -3.42353374e-01 7.83131599e-01 7.75104046e-01 1.20091057e+00 -2.27966741e-01 9.20416236e-01 2.22683772e-01 8.92866135e-01 -2.01124772e-02 -1.05259113e-01 2.50076026e-01 -3.04328620e-01 5.15758932e-01 1.24969184e+00 6.39417887e-01 9.49179903e-02 -1.76595021e-02 6.75558925e-01 2.16001555e-01 -1.65387973e-01 -3.34507078e-01 -1.88026577e-01 4.81422067e-01 6.70353830e-01 -5.83356977e-01 -1.21464267e-01 -5.96598387e-01 1.05600238e+00 5.47987074e-02 6.87871575e-01 -1.06064498e+00 -4.40484196e-01 9.04570878e-01 -3.62662673e-02 7.88367212e-01 -1.67696655e-01 -1.06963880e-01 -1.13229871e+00 2.23683223e-01 -9.30069447e-01 4.90159839e-01 -8.70278478e-01 -1.61124766e+00 8.77235472e-01 -4.12922949e-01 -1.67200029e+00 -2.78256953e-01 -3.06665510e-01 -6.67732775e-01 1.07764482e+00 -1.44029093e+00 -7.59198606e-01 -6.90382048e-02 6.81559622e-01 9.05036390e-01 -2.63719141e-01 4.78961676e-01 5.52154064e-01 -5.51238358e-01 6.96411610e-01 4.07013506e-01 9.27907303e-02 5.17508566e-01 -9.12066758e-01 8.66807997e-01 8.46104145e-01 -1.94975555e-01 6.30468130e-01 6.21651471e-01 -6.67426050e-01 -1.51808608e+00 -1.23079860e+00 8.42156112e-01 -3.29123467e-01 7.94931114e-01 -1.63199365e-01 -1.43532968e+00 6.48728728e-01 1.04388721e-01 6.06389642e-02 5.89349687e-01 -1.86938539e-01 -5.08957565e-01 -2.63798028e-01 -7.05756366e-01 4.81569320e-01 1.00459576e+00 -7.93350458e-01 -4.90786165e-01 1.00019850e-01 9.80284810e-01 -2.80867666e-01 -7.07815945e-01 3.07068259e-01 5.69346905e-01 -9.14315641e-01 7.39476383e-01 -5.87819934e-01 4.38483000e-01 -4.34586793e-01 -1.01604395e-01 -1.51023483e+00 -9.27414000e-01 -8.86497080e-01 -2.70565689e-01 1.41124785e+00 3.24737057e-02 -1.03139246e+00 1.56517044e-01 2.00874180e-01 -3.31653982e-01 -8.75876844e-01 -7.99426138e-01 -9.79823053e-01 6.66196644e-02 -4.24698502e-01 1.06815994e+00 1.05915093e+00 -8.97060335e-03 1.79860994e-01 -6.71214402e-01 1.17275022e-01 4.39714603e-02 3.02706420e-01 3.44667971e-01 -1.00798464e+00 -1.91644937e-01 -6.05942011e-01 -4.13542688e-01 -1.44799244e+00 1.63069904e-01 -5.40634751e-01 9.96863246e-02 -9.53762770e-01 -4.47745733e-02 -2.26841599e-01 -7.83447564e-01 1.28303885e-01 -4.48976785e-01 -3.13137114e-01 1.02829330e-01 2.59978056e-01 -4.74834621e-01 1.03383088e+00 1.15205348e+00 -2.09486589e-01 6.01155870e-02 4.37790230e-02 -6.11129701e-01 3.15491557e-01 5.77894151e-01 -2.13751823e-01 -5.36269426e-01 -3.57235909e-01 -2.00324226e-02 3.52956295e-01 7.25551695e-02 -7.75902569e-01 1.89260870e-01 -2.72344559e-01 3.76451820e-01 -9.38844264e-01 -2.84375604e-02 -9.31922913e-01 3.28995049e-01 5.80681503e-01 -3.08431149e-01 5.11924744e-01 1.94433808e-01 8.17666173e-01 -3.13927352e-01 2.36076564e-01 3.93801451e-01 1.77728042e-01 -1.02390516e+00 4.90816712e-01 -6.09422565e-01 -1.16287880e-02 8.85460436e-01 -3.27154428e-01 -3.15867007e-01 -6.56230807e-01 -1.18770048e-01 2.49447897e-01 1.27366766e-01 1.04879236e+00 4.20674920e-01 -1.63298309e+00 -7.38396227e-01 5.51084578e-01 7.28467926e-02 -4.73992705e-01 6.48692846e-01 7.87868023e-01 -6.84060380e-02 4.39722121e-01 -1.88503414e-01 -7.47739434e-01 -9.21393812e-01 1.08115065e+00 2.91603327e-01 -2.13533044e-01 -6.85979009e-01 5.69960237e-01 4.35313404e-01 -3.49122316e-01 1.87386975e-01 -6.58935666e-01 -1.96512833e-01 4.74649779e-02 8.83836687e-01 3.74477446e-01 2.39357188e-01 -6.12240970e-01 -4.09112126e-01 5.51870048e-01 -1.66648105e-01 3.47432286e-01 1.36217368e+00 -4.42619860e-01 9.36331525e-02 7.11026609e-01 1.31757498e+00 -2.70912617e-01 -1.49037588e+00 -5.09529173e-01 2.16935147e-02 -5.59982717e-01 -1.43149599e-01 -6.33319974e-01 -1.05638587e+00 9.53500509e-01 5.46122968e-01 6.07557654e-01 1.46245801e+00 -4.80606407e-01 1.13929284e+00 -6.50771111e-02 3.51077110e-01 -7.81163156e-01 8.72891620e-02 8.87262046e-01 1.00993133e+00 -9.87744808e-01 -4.24183637e-01 -1.97352186e-01 -7.19669998e-01 1.17272055e+00 3.42801720e-01 1.44085348e-01 7.44137704e-01 3.68765622e-01 5.11678122e-02 1.64520308e-01 -1.11337328e+00 8.31417888e-02 3.70709330e-01 6.14443302e-01 5.08617759e-01 -1.52116343e-01 -4.70934473e-02 8.46804678e-01 -1.40253566e-02 8.60408172e-02 -5.27516603e-02 6.30625486e-01 -1.78790122e-01 -6.56616151e-01 -3.44591528e-01 3.79263580e-01 -3.89181852e-01 -5.43303825e-02 3.75711024e-02 4.03156161e-01 -3.74607779e-02 9.64275599e-01 3.81051421e-01 -6.86309755e-01 3.63221169e-01 6.90058693e-02 -8.76354277e-02 -4.51985225e-02 -7.08903968e-01 9.77157280e-02 -3.83458316e-01 -6.25010371e-01 -1.00628138e-01 -7.33368754e-01 -9.79357779e-01 -3.64338040e-01 -2.13314146e-01 5.37719466e-02 2.47023150e-01 1.08065701e+00 7.08215356e-01 1.05037236e+00 9.50862944e-01 -7.55222082e-01 -7.26961195e-01 -9.82104480e-01 -5.01563251e-01 5.76027095e-01 6.90866470e-01 -5.59842288e-01 -3.02286714e-01 1.20379120e-01]
[7.0132951736450195, 2.9655921459198]
43a81ecf-dcae-4bdc-9694-04ed07e6dbf2
learn-the-big-picture-representation-learning
null
null
https://aclanthology.org/2021.repl4nlp-1.15
https://aclanthology.org/2021.repl4nlp-1.15.pdf
Learn The Big Picture: Representation Learning for Clustering
Existing supervised models for text clustering find it difficult to directly optimize for clustering results. This is because clustering is a discrete process and it is difficult to estimate meaningful gradient of any discrete function that can drive gradient based optimization algorithms. So, existing supervised clustering algorithms indirectly optimize for some continuous function that approximates the clustering process. We propose a scalable training strategy that directly optimizes for a discrete clustering metric. We train a BERT-based embedding model using our method and evaluate it on two publicly available datasets. We show that our method outperforms another BERT-based embedding model employing Triplet loss and other unsupervised baselines. This suggests that optimizing directly for the clustering outcome indeed yields better representations suitable for clustering.
['Laura Dietz', 'Sumanta Kashyapi']
null
null
null
null
acl-repl4nlp-2021-8
['text-clustering']
['natural-language-processing']
[-2.93001205e-01 -1.39797181e-01 -3.39254797e-01 -8.82251799e-01 -1.19871581e+00 -5.51712096e-01 6.93450332e-01 5.60609877e-01 -5.69750130e-01 1.57370493e-01 4.09048378e-01 -7.95868412e-02 -3.17947954e-01 -4.85801488e-01 -4.79822904e-01 -7.84904122e-01 -2.73236692e-01 1.16744649e+00 -1.92998409e-01 3.40348631e-01 2.03372672e-01 2.86860913e-01 -1.14761543e+00 2.54847705e-01 7.16711819e-01 4.58717197e-01 8.67146924e-02 8.45892251e-01 -4.17182185e-02 6.39402807e-01 -6.63915455e-01 -2.80304998e-01 2.49361366e-01 -5.42914629e-01 -7.40238547e-01 8.33478048e-02 3.47189099e-01 -9.53602567e-02 -4.36386317e-01 8.82285416e-01 3.48013937e-01 4.46095020e-01 1.32077134e+00 -1.49704230e+00 -9.51475263e-01 8.47089469e-01 -3.66373360e-01 -4.60913628e-02 8.25500116e-02 -4.35635746e-02 1.66030157e+00 -7.42288470e-01 5.77692211e-01 1.44281030e+00 7.93689251e-01 2.20101237e-01 -1.65933776e+00 -1.71038181e-01 1.66455090e-01 1.48760125e-01 -1.41221082e+00 -2.41136223e-01 7.85512924e-01 -3.74659985e-01 8.17417860e-01 5.73798008e-02 4.73146468e-01 9.36189413e-01 -2.69951373e-01 1.03558552e+00 8.59394789e-01 -4.55254018e-01 3.77768517e-01 3.55443835e-01 5.11242628e-01 7.57531643e-01 7.40839913e-02 -2.47993823e-02 -2.70288795e-01 -1.18137747e-01 2.66400665e-01 1.41343623e-01 -2.74871048e-02 -6.92860842e-01 -1.17507541e+00 1.18967593e+00 7.06199050e-01 2.59385496e-01 6.82762042e-02 6.36945665e-01 1.40020818e-01 4.08534229e-01 5.34293711e-01 8.27498972e-01 -3.42042595e-01 -4.65134889e-01 -1.33924294e+00 -4.72536199e-02 1.06618500e+00 7.82132030e-01 1.09484136e+00 -3.44242036e-01 -3.19317222e-01 8.77302647e-01 5.36606848e-01 -4.00123820e-02 5.12332618e-01 -1.30535078e+00 1.88901842e-01 6.39697492e-01 -7.77082331e-03 -1.10127556e+00 -4.80305672e-01 3.65602039e-02 -3.58457714e-01 5.62417731e-02 6.71430290e-01 -1.06797881e-01 -6.02190435e-01 1.54386067e+00 1.88995734e-01 4.71666530e-02 -2.42084995e-01 6.98898673e-01 5.12812376e-01 6.86404943e-01 -5.71108051e-02 -8.99307132e-02 7.52773345e-01 -1.23557734e+00 -6.85681880e-01 -4.45155136e-04 1.05984640e+00 -5.53697586e-01 1.61882567e+00 3.42386484e-01 -7.46949375e-01 -1.97422162e-01 -9.70115125e-01 -2.94088006e-01 -6.16692126e-01 3.38209510e-01 8.21716845e-01 7.58352280e-01 -1.39698458e+00 9.38900948e-01 -1.23948741e+00 -4.70779032e-01 3.31507862e-01 5.57023108e-01 -5.00640608e-02 6.52228966e-02 -6.41969740e-01 6.50168240e-01 3.68979245e-01 -1.96614698e-01 -5.96146643e-01 -6.24197602e-01 -6.56975925e-01 2.64831334e-01 -4.43293825e-02 -3.35211873e-01 9.47280228e-01 -7.81045496e-01 -1.46573317e+00 8.69960904e-01 -3.01062584e-01 -4.82627749e-01 2.41460890e-01 -6.89163804e-02 5.09384088e-02 3.60844344e-01 9.61417407e-02 8.02364409e-01 8.30079436e-01 -1.41929984e+00 -1.30456775e-01 -3.02997142e-01 -2.27170959e-01 5.20695090e-01 -9.38412309e-01 -1.93555281e-02 -5.78806818e-01 -4.54783589e-01 5.99591620e-03 -6.28752232e-01 -3.08013499e-01 7.47648403e-02 -4.87618059e-01 -4.50633347e-01 7.85481274e-01 -3.59358370e-01 1.43390250e+00 -2.22060132e+00 4.52105291e-02 2.95191735e-01 3.94169420e-01 -4.40304071e-01 -1.48284554e-01 4.79904622e-01 2.12679699e-01 6.40307903e-01 -3.87354076e-01 -8.62930954e-01 4.89021748e-01 2.04001114e-01 1.86823964e-01 6.69676900e-01 1.22721449e-01 1.00142062e+00 -1.02790952e+00 -8.56527090e-01 2.63447851e-01 5.20111144e-01 -8.22876215e-01 2.17066407e-01 -1.46909714e-01 -5.99159859e-02 -1.74458221e-01 4.99636382e-01 4.48510677e-01 -4.23130810e-01 3.38734239e-01 -6.43624961e-02 1.83323398e-01 2.80761838e-01 -1.05701387e+00 1.71298814e+00 -2.59152591e-01 9.67980206e-01 2.01680213e-02 -1.51090264e+00 8.19443703e-01 5.18984236e-02 1.04502511e+00 2.31832445e-01 7.46751428e-02 -2.80865997e-01 -2.06182167e-01 -6.79102242e-02 3.50965679e-01 -5.57427555e-02 -6.80414140e-02 8.61829698e-01 2.12228283e-01 -5.09833097e-01 1.71523631e-01 5.69105864e-01 1.38105106e+00 -1.16613828e-01 -2.41728798e-01 -4.61860895e-01 1.59834772e-02 2.13420004e-01 2.91483283e-01 8.84608269e-01 -4.11493570e-01 9.46928382e-01 6.00923657e-01 6.19751809e-04 -9.36361670e-01 -1.25040591e+00 -2.25504175e-01 1.19210279e+00 9.04534906e-02 -8.43214869e-01 -8.01136315e-01 -9.94027078e-01 1.24422900e-01 6.37260318e-01 -6.83457077e-01 -3.20776641e-01 -2.77320474e-01 -1.03799760e+00 5.79107046e-01 6.11330390e-01 9.09090713e-02 -7.00797439e-01 1.63673863e-01 2.40132049e-01 -6.56559169e-02 -7.14610040e-01 -8.58378947e-01 5.32803714e-01 -1.06447196e+00 -1.15272033e+00 -1.17049262e-01 -1.02950323e+00 7.56951034e-01 2.62201458e-01 1.34816706e+00 3.03998768e-01 -3.12322229e-01 6.93347216e-01 -3.90817910e-01 -2.70310342e-02 -3.19190681e-01 3.37755054e-01 -1.30538970e-01 -1.39165968e-01 1.05989289e+00 -2.84685314e-01 -4.78908360e-01 1.48269042e-01 -6.51425898e-01 -5.83067656e-01 2.43949473e-01 1.07479894e+00 5.29670000e-01 3.95953804e-01 3.83762240e-01 -8.21501911e-01 9.53071058e-01 -5.07640958e-01 -3.76752555e-01 3.88343722e-01 -1.01919365e+00 3.71564627e-01 8.46748412e-01 -4.94732797e-01 -7.92264760e-01 3.83948058e-01 1.18366323e-01 -4.74994391e-01 -2.20830902e-01 4.37512189e-01 -4.12782095e-02 3.58961403e-01 8.49612713e-01 -2.29488626e-01 3.50867137e-02 -5.00171363e-01 7.08989978e-01 8.03527296e-01 1.85656622e-01 -8.26891124e-01 9.42646861e-01 7.07075715e-01 -2.74213791e-01 -8.62891734e-01 -8.20711672e-01 -9.39731658e-01 -8.68056178e-01 -5.32603450e-03 1.03785634e+00 -7.42499530e-01 -7.06990004e-01 -2.61810124e-01 -7.60782540e-01 -6.85683668e-01 -3.97036403e-01 7.78603733e-01 -7.60966539e-01 4.33993429e-01 -7.16578841e-01 -6.90503001e-01 -5.60221225e-02 -1.05390751e+00 1.15642309e+00 -2.67761111e-01 -4.24908251e-01 -1.72893047e+00 2.94564635e-01 1.63055986e-01 7.81285837e-02 7.35509172e-02 7.65923798e-01 -7.20019042e-01 -4.41596866e-01 -2.23973662e-01 -2.13859946e-01 4.04526323e-01 2.23364443e-01 6.31048977e-01 -8.33602965e-01 -4.43979830e-01 -7.23939016e-03 -6.23235822e-01 1.23775983e+00 6.36668563e-01 1.32207525e+00 -2.56064236e-01 -4.51666296e-01 9.57842827e-01 1.39361572e+00 -1.65488616e-01 3.62380266e-01 2.14927956e-01 9.16791081e-01 3.55623513e-01 4.01856363e-01 2.34103799e-01 5.15093386e-01 4.91925389e-01 -1.45940512e-01 -1.11835383e-01 6.32590950e-02 -3.58934164e-01 6.14742637e-01 1.02692664e+00 3.35715652e-01 -1.08563259e-01 -1.00595677e+00 6.05923414e-01 -2.00204420e+00 -1.03157794e+00 -1.76607877e-01 2.02453613e+00 1.28733230e+00 -8.49159714e-03 2.35135943e-01 1.29338995e-01 6.84615731e-01 6.51374906e-02 -4.35453385e-01 -4.45034504e-01 8.43416154e-02 2.49229535e-01 3.79474312e-01 8.67726445e-01 -1.30827487e+00 1.19838369e+00 7.83877325e+00 6.08976841e-01 -6.29060924e-01 5.72472997e-02 6.48901463e-01 -2.48494074e-01 -3.23259830e-01 2.82184601e-01 -6.56590343e-01 3.85116994e-01 9.97087896e-01 -1.58616811e-01 6.88204646e-01 8.91281009e-01 3.91624629e-01 2.28594705e-01 -1.62611616e+00 8.83607090e-01 -7.06201792e-02 -1.10275459e+00 -9.45499763e-02 1.70998901e-01 8.48169565e-01 -2.78757196e-02 1.30429149e-01 2.75436521e-01 1.06813157e+00 -1.29183853e+00 3.37852985e-01 4.47857082e-01 3.92858028e-01 -7.83338249e-01 2.60680765e-01 1.89347610e-01 -1.01513827e+00 2.14386769e-02 -6.45989478e-01 1.25985041e-01 -1.50870889e-01 8.59434545e-01 -1.08758163e+00 -1.36252493e-01 6.23981416e-01 1.09119976e+00 -9.59182322e-01 9.12786365e-01 -1.91757575e-01 1.06454134e+00 -5.64777851e-01 -3.06448251e-01 4.45453376e-01 -7.05640554e-01 3.49349916e-01 1.59906590e+00 2.20895447e-02 -3.49298656e-01 3.60157251e-01 1.11609149e+00 -4.55324084e-01 1.15628511e-01 -6.97230518e-01 -5.28840721e-01 7.15291321e-01 1.38344514e+00 -9.85923767e-01 -3.06381017e-01 -3.25433135e-01 1.01774180e+00 7.14363337e-01 5.71586072e-01 -6.94668770e-01 -5.11962473e-01 6.79025412e-01 -2.41877958e-01 2.97642499e-01 -5.62633395e-01 -5.99490166e-01 -1.42575383e+00 -2.21520707e-01 -7.43548870e-01 5.28690398e-01 -3.78921002e-01 -1.60100186e+00 1.38234556e-01 -6.76941797e-02 -7.89047837e-01 -3.15239787e-01 -5.84313631e-01 -7.84228683e-01 3.52351904e-01 -9.61290002e-01 -7.85810947e-01 -3.36803570e-02 6.61060154e-01 2.86678761e-01 -1.90726146e-02 7.56187677e-01 -2.65679834e-03 -7.23782480e-01 9.54571724e-01 7.87042677e-01 3.11766654e-01 9.67896819e-01 -2.07263970e+00 6.71437150e-03 5.89190722e-01 7.10875332e-01 8.11297238e-01 5.43890715e-01 -3.44645083e-01 -1.28220737e+00 -9.88070071e-01 7.58487403e-01 -1.03083360e+00 7.66117632e-01 -5.84768116e-01 -8.61374438e-01 7.85289824e-01 1.91571042e-01 -4.02900308e-01 9.72297132e-01 4.91466254e-01 -4.54535425e-01 -1.82173565e-01 -1.18173647e+00 4.90803152e-01 8.66420448e-01 -5.89270651e-01 -4.85523164e-01 7.59505332e-01 6.62839890e-01 4.48097050e-01 -1.21087098e+00 -1.80088341e-01 5.96107244e-02 -7.64931738e-01 1.06152701e+00 -6.52124524e-01 3.57930928e-01 -1.75141811e-01 -2.68792987e-01 -1.53634536e+00 -3.13387066e-01 -5.67296326e-01 -2.50929415e-01 1.22442937e+00 5.36708891e-01 -2.94872731e-01 1.03245282e+00 9.25965011e-01 3.95091921e-02 -5.95630348e-01 -5.42914450e-01 -8.65353763e-01 6.11156285e-01 -3.08763146e-01 5.04453123e-01 1.21944642e+00 3.37297887e-01 3.96920979e-01 2.11564660e-01 -9.89203677e-02 1.00559533e+00 -7.14183897e-02 7.28687763e-01 -1.26464951e+00 -3.33136857e-01 -7.69915581e-01 -3.91524494e-01 -1.22199965e+00 5.39157987e-01 -1.33255017e+00 1.58420220e-01 -1.76620233e+00 1.99899226e-01 -6.68742836e-01 -3.40776742e-01 3.16823661e-01 -3.52434516e-01 2.85238698e-02 -3.07973735e-02 3.85885596e-01 -7.98215926e-01 7.09489346e-01 7.84721017e-01 -3.81424010e-01 -4.04610515e-01 -3.51014525e-01 -8.27697396e-01 5.26986361e-01 9.14728999e-01 -7.08293438e-01 -4.35569823e-01 -4.51579034e-01 3.75448130e-02 -5.71488321e-01 1.29799202e-01 -7.66990125e-01 4.86172229e-01 -1.41761690e-01 3.93315315e-01 -5.73218703e-01 4.31041330e-01 -5.86396813e-01 -5.01408577e-01 -4.19464670e-02 -7.29306519e-01 -7.34571666e-02 -4.15087163e-01 7.59872258e-01 -3.69893402e-01 -3.87771785e-01 7.04611719e-01 -1.36805428e-02 -2.29205236e-01 3.19500864e-01 -2.75909543e-01 4.64114636e-01 7.35193133e-01 -2.06826374e-01 1.07425272e-01 -5.77727020e-01 -8.58865738e-01 5.55333912e-01 8.41099203e-01 3.63730825e-02 5.73362410e-01 -1.44051838e+00 -7.03066468e-01 -1.08196065e-01 -1.28882498e-01 -1.06995694e-01 -8.04941118e-01 7.18114853e-01 -5.63813567e-01 6.14310987e-02 5.12258589e-01 -7.33195543e-01 -8.72462153e-01 7.13984191e-01 3.20834368e-01 -2.64571577e-01 -4.75683123e-01 7.88195848e-01 -3.19816738e-01 -9.49818730e-01 5.93471587e-01 -2.04992220e-01 1.46310955e-01 1.37619257e-01 1.37543157e-01 4.92761463e-01 -5.38601652e-02 -3.57673168e-01 -3.29208404e-01 3.28011990e-01 3.91935632e-02 -5.00657797e-01 1.33394921e+00 -9.13079008e-02 -1.71579689e-01 8.00262213e-01 2.01975584e+00 -3.77062619e-01 -1.28097355e+00 -6.66439608e-02 3.55525196e-01 -5.09321511e-01 2.74381489e-01 -3.80731285e-01 -8.86236727e-01 1.02572656e+00 2.52434433e-01 3.97674829e-01 7.97010243e-01 1.91668168e-01 3.45592380e-01 9.63128567e-01 -7.11795017e-02 -1.72626400e+00 5.25416076e-01 3.46287131e-01 3.51225674e-01 -1.52385676e+00 1.09119631e-01 7.97397345e-02 -7.43273020e-01 1.14132953e+00 3.65272224e-01 -2.91019887e-01 1.20890832e+00 1.13185540e-01 2.01894209e-01 -4.46356028e-01 -8.08489084e-01 -2.09490404e-01 2.61992991e-01 7.32299805e-01 8.02807868e-01 1.75363019e-01 -1.51439667e-01 1.55955628e-01 -4.19399321e-01 -5.45645893e-01 3.55178684e-01 5.15810013e-01 -3.55866939e-01 -1.22080183e+00 -2.32521728e-01 8.56064618e-01 -2.31616393e-01 -1.58463210e-01 -8.65693569e-01 6.39805794e-01 -3.61339658e-01 1.31112230e+00 1.43512055e-01 -3.95697355e-01 -3.13874096e-01 2.64564455e-01 2.80137777e-01 -8.84652853e-01 -5.22813797e-01 -1.03634767e-01 -2.43631169e-01 -5.82749188e-01 -4.41322535e-01 -8.94990742e-01 -1.39045691e+00 -3.61848682e-01 -5.98903835e-01 4.06730771e-01 6.78645372e-01 5.22107899e-01 8.76785219e-02 8.60068724e-02 1.07332814e+00 -7.41708636e-01 -9.39339042e-01 -8.27985525e-01 -7.46439993e-01 9.45505559e-01 4.70536128e-02 -4.92676497e-01 -8.79370093e-01 2.01060280e-01]
[9.133063316345215, 3.2735660076141357]
23a576a3-1565-4fef-bbb7-e625fba4f5c7
explanatory-analysis-and-rectification-of-the
2111.05679
null
https://arxiv.org/abs/2111.05679v1
https://arxiv.org/pdf/2111.05679v1.pdf
Explanatory Analysis and Rectification of the Pitfalls in COVID-19 Datasets
Since the onset of the COVID-19 pandemic in 2020, millions of people have succumbed to this deadly virus. Many attempts have been made to devise an automated method of testing that could detect the virus. Various researchers around the globe have proposed deep learning based methodologies to detect the COVID-19 using Chest X-Rays. However, questions have been raised on the presence of bias in the publicly available Chest X-Ray datasets which have been used by the majority of the researchers. In this paper, we propose a 2 staged methodology to address this topical issue. Two experiments have been conducted as a part of stage 1 of the methodology to exhibit the presence of bias in the datasets. Subsequently, an image segmentation, super-resolution and CNN based pipeline along with different image augmentation techniques have been proposed in stage 2 of the methodology to reduce the effect of bias. InceptionResNetV2 trained on Chest X-Ray images that were augmented with Histogram Equalization followed by Gamma Correction when passed through the pipeline proposed in stage 2, yielded a top accuracy of 90.47% for 3-class (Normal, Pneumonia, and COVID-19) classification task.
['Chandra Prakash', 'Yuvraj Singh Champawat', 'Amrit Raj', 'Shaanya Singh', 'Japman Singh Monga', 'Samyak Prajapati']
2021-11-10
null
null
null
null
['image-augmentation']
['computer-vision']
[ 3.69918942e-01 -8.62112194e-02 2.47272179e-01 -3.94114941e-01 -4.76473421e-01 -3.08650821e-01 5.49232483e-01 1.98542595e-01 -6.21001065e-01 6.87903821e-01 -2.11667176e-02 -4.41454947e-01 2.18260989e-01 -7.37796128e-01 -5.34924567e-01 -5.75566530e-01 -9.25175995e-02 6.44218326e-01 3.21309626e-01 -1.38512452e-03 2.82695025e-01 4.90388870e-01 -1.24809599e+00 5.75665295e-01 4.34368193e-01 7.53535628e-01 1.50049403e-01 1.19660389e+00 1.84581392e-02 6.35715485e-01 -5.89524984e-01 -8.74914452e-02 3.27449113e-01 -5.05140722e-01 -8.06502581e-01 -2.14814767e-01 4.04508770e-01 -8.98441255e-01 2.70932112e-02 7.25907624e-01 6.81400716e-01 -2.53443688e-01 7.07748294e-01 -9.22179461e-01 -2.92582184e-01 -5.87876365e-02 -7.88403928e-01 9.61542368e-01 1.61396638e-01 2.46025786e-01 4.26981419e-01 -7.48157740e-01 6.26404941e-01 9.80266273e-01 1.01449323e+00 5.79015374e-01 -7.84441352e-01 -7.41765380e-01 -5.66146731e-01 1.55911177e-01 -1.03173602e+00 1.00411274e-01 2.56089926e-01 -6.99250638e-01 1.26575911e+00 3.14977914e-01 6.81096017e-01 1.15925729e+00 5.44543445e-01 2.43365511e-01 1.40502095e+00 -2.28984356e-01 -9.86627024e-03 3.42742145e-01 6.51454106e-02 7.05065966e-01 4.82582420e-01 1.22200986e-02 5.35250343e-02 -3.12252104e-01 8.59552145e-01 2.96692342e-01 3.27603109e-02 1.13653786e-01 -8.25568795e-01 1.07498169e+00 4.61419970e-01 4.50707465e-01 -6.28840864e-01 -2.17134744e-01 6.71595514e-01 -4.94771712e-02 5.42875588e-01 2.87214011e-01 -4.75418091e-01 3.50640774e-01 -1.04104817e+00 4.14287686e-01 1.90583333e-01 1.50744632e-01 3.06985676e-01 -2.24563822e-01 -4.30104524e-01 5.23369312e-01 5.05753338e-01 4.33387965e-01 5.15465975e-01 -3.14242750e-01 3.05693686e-01 7.32709944e-01 -3.12217940e-02 -9.38894212e-01 -7.12335944e-01 -4.06548321e-01 -8.78524542e-01 1.07617281e-01 8.27182159e-02 -4.27701831e-01 -1.50411856e+00 1.27184081e+00 4.79060739e-01 3.21157962e-01 -2.24635974e-01 8.55262816e-01 7.34191120e-01 6.35134339e-01 1.73772171e-01 1.10864215e-01 1.62363994e+00 -6.83490872e-01 -5.38372755e-01 2.01883793e-01 5.27956963e-01 -7.04875052e-01 8.61939013e-01 3.44989717e-01 -8.21834624e-01 -4.80827242e-01 -1.16978991e+00 2.99124509e-01 -5.37989557e-01 -3.21729541e-01 2.66139805e-01 1.09880984e+00 -1.06178212e+00 2.55461931e-01 -9.10247624e-01 -8.02707851e-01 6.78594947e-01 3.38564098e-01 -4.29272242e-02 -6.81651905e-02 -1.10227346e+00 1.07626414e+00 2.93716997e-01 2.03835107e-02 -1.07638121e+00 -5.65948129e-01 -3.91288579e-01 -2.83665746e-01 -5.14144748e-02 -7.44772673e-01 8.86685967e-01 -7.81773150e-01 -7.42286801e-01 1.06245148e+00 1.77967265e-01 -4.17839736e-01 5.69262505e-01 -2.04043478e-01 -3.79300117e-01 3.43797684e-01 1.38123125e-01 8.31395566e-01 5.64926684e-01 -9.64983523e-01 -8.02917600e-01 -6.35560751e-01 -1.10003047e-01 -2.81229224e-02 9.78159532e-02 6.29191935e-01 -2.12690890e-01 -7.20821917e-01 -4.08511996e-01 -9.93724108e-01 -5.15719205e-02 -3.47176909e-01 -3.04834485e-01 -5.74662499e-02 1.09486842e+00 -1.00330901e+00 7.99982250e-01 -1.67787194e+00 -6.76792920e-01 1.90749511e-01 1.57341897e-01 6.74866080e-01 1.09395757e-01 2.21646145e-01 -8.33594427e-02 4.04121459e-01 -5.14496267e-01 8.34560022e-03 -5.06862104e-01 7.64513984e-02 -1.51782125e-01 6.60503626e-01 5.06917655e-01 6.63427353e-01 -5.62730253e-01 -4.72980320e-01 1.45253375e-01 9.74530101e-01 -4.36924487e-01 5.19203901e-01 -1.43924847e-01 6.66080832e-01 -3.70880991e-01 7.64570415e-01 9.39453900e-01 -3.09188455e-01 -1.78967446e-01 1.06649898e-01 -5.89705855e-02 1.71256028e-02 -6.24809146e-01 1.10624468e+00 -1.66840628e-02 5.25370955e-01 1.73836544e-01 -7.03553796e-01 3.29517394e-01 6.34504557e-01 4.51425940e-01 -5.20493090e-01 4.95688409e-01 8.48169997e-02 1.59047946e-01 -8.86682153e-01 2.11134911e-01 -3.61253053e-01 4.76704121e-01 3.90365034e-01 -2.21395358e-01 8.32258984e-02 -2.97177266e-02 -1.21917807e-01 1.22377670e+00 7.03384057e-02 1.01462126e-01 -1.09818801e-01 6.16788030e-01 4.61454958e-01 2.24606410e-01 8.13973725e-01 -5.73972821e-01 9.30591047e-01 6.49898723e-02 -5.87120175e-01 -1.18079197e+00 -8.35119367e-01 -3.17584544e-01 7.62892962e-01 -3.33090872e-01 3.18307847e-01 -1.04051292e+00 -6.44449294e-01 -3.92835468e-01 2.65624672e-01 -8.74150276e-01 4.94421691e-01 -6.67625606e-01 -1.29136467e+00 1.01968980e+00 5.44666648e-01 7.77680814e-01 -1.28931272e+00 -1.37392175e+00 6.14404827e-02 6.73417747e-02 -1.02886701e+00 -7.60932639e-02 9.55673233e-02 -7.26255655e-01 -1.32270777e+00 -9.20876086e-01 -6.27212107e-01 6.71273947e-01 -2.30653118e-02 6.29439354e-01 5.18053770e-01 -8.57679725e-01 1.21818453e-01 -2.67582774e-01 -6.94983423e-01 -3.39644611e-01 -8.03848431e-02 -2.44974092e-01 -4.44275469e-01 4.76774842e-01 1.14600593e-02 -1.08323860e+00 -1.20722227e-01 -1.03398907e+00 -1.00358009e-01 6.16998732e-01 4.69272316e-01 4.88542527e-01 -4.61789034e-02 5.81515849e-01 -1.18585229e+00 5.73855460e-01 -7.72686422e-01 -3.76341999e-01 -7.05503970e-02 -5.42273343e-01 -4.86196429e-01 4.37802702e-01 -1.00756129e-02 -9.14835751e-01 -3.76904011e-02 -6.26009345e-01 -1.42905220e-01 -5.26373565e-01 1.79001778e-01 4.55471128e-01 1.88795730e-01 4.87068921e-01 -1.08325243e-01 2.35354714e-02 -2.54809201e-01 -1.13017425e-01 8.92416656e-01 4.39921021e-01 1.75875664e-01 5.58157206e-01 7.27245450e-01 -2.78042164e-02 -9.01133358e-01 -8.52315068e-01 -5.91367602e-01 -4.89675820e-01 -2.84229785e-01 1.60911191e+00 -8.01498830e-01 -1.60805732e-01 8.19009662e-01 -1.08215380e+00 -1.07552245e-01 3.95505279e-01 4.81420159e-01 8.49448796e-03 1.81002691e-01 -5.95350146e-01 -6.60291791e-01 -9.28764641e-01 -1.16521835e+00 6.35200918e-01 2.85629541e-01 -2.82827586e-01 -8.26643527e-01 5.71613371e-01 6.52111232e-01 7.47184753e-01 6.04907990e-01 9.08073187e-01 -1.02576423e+00 -2.26458713e-01 -2.01764166e-01 -4.85655129e-01 3.73133451e-01 1.71758667e-01 -6.71946332e-02 -1.08143389e+00 -4.34128940e-01 3.39560837e-01 -2.83828825e-01 6.63836241e-01 5.21131396e-01 9.52254236e-01 -1.69233739e-01 -3.17093492e-01 5.24512470e-01 1.53308034e+00 4.50738758e-01 5.36729991e-01 4.49551255e-01 5.64404070e-01 4.37780559e-01 2.96296179e-01 1.75228909e-01 2.92378783e-01 2.39974499e-01 5.62598288e-01 -4.01661992e-01 -7.75562599e-02 3.45978588e-01 -6.59313500e-02 4.87984478e-01 -6.05493374e-02 -1.87973559e-01 -1.18825769e+00 5.98455191e-01 -1.11536002e+00 -7.22162008e-01 -3.90207231e-01 1.80162454e+00 5.07902980e-01 2.83460826e-01 2.49480337e-01 3.68368551e-02 6.77164078e-01 -8.44814777e-02 -5.10043383e-01 -6.46003425e-01 1.47396773e-01 4.52962726e-01 4.34788972e-01 2.77840197e-01 -1.27494180e+00 3.10156226e-01 6.70728445e+00 8.22310448e-02 -1.53096187e+00 3.02824020e-01 8.93164337e-01 2.68351734e-01 6.74598068e-02 -4.61754233e-01 -6.03301466e-01 4.31251019e-01 1.00049829e+00 7.01950610e-01 8.56974870e-02 4.95053917e-01 2.00615749e-01 -3.59900475e-01 -4.58269477e-01 4.27358687e-01 3.31342250e-01 -1.20884931e+00 -1.65809602e-01 -7.44001716e-02 6.90245509e-01 6.13843679e-01 -6.26342744e-02 1.90661833e-01 -7.53229111e-02 -1.11414337e+00 9.60949138e-02 2.53388971e-01 8.26605260e-01 -8.03092599e-01 1.16851735e+00 2.00716048e-01 -8.59587371e-01 1.82625860e-01 -1.49842933e-01 1.03803001e-01 -4.01816778e-02 1.72960296e-01 -1.58522367e+00 1.95377350e-01 1.08396316e+00 -1.90955088e-01 -6.56264961e-01 9.50805902e-01 5.81798032e-02 8.89410794e-01 -2.68600732e-01 9.97365788e-02 3.96968514e-01 1.45849511e-01 3.21625888e-01 1.55459392e+00 1.22063003e-01 1.85049474e-01 -1.01392426e-01 7.16425836e-01 6.30345643e-02 1.39487013e-01 -4.50658053e-01 7.02598915e-02 5.56011833e-02 1.27405059e+00 -1.25859237e+00 -4.14097726e-01 -4.91417497e-01 7.40846455e-01 -8.85922089e-02 -8.90751742e-03 -1.04933560e+00 -1.94987476e-01 7.54234493e-02 5.16430616e-01 4.33904052e-01 1.72138348e-01 -4.15037423e-01 -5.00482678e-01 -3.97682011e-01 -9.88516867e-01 5.12431204e-01 -7.38052726e-01 -8.83902252e-01 8.46729159e-01 7.47744069e-02 -3.84493142e-01 -1.28446996e-01 -5.89957893e-01 -8.90434265e-01 9.66833889e-01 -1.34997225e+00 -1.05201519e+00 -4.33301121e-01 4.20232356e-01 5.82168996e-01 -2.21315578e-01 9.08472598e-01 3.43061328e-01 -6.88225567e-01 1.17489338e-01 -7.19773248e-02 1.48803055e-01 4.50937837e-01 -1.05506182e+00 2.77370721e-01 9.97812152e-01 -5.59620440e-01 7.29571104e-01 6.35580361e-01 -1.07327449e+00 -8.84421945e-01 -1.31276643e+00 7.52770245e-01 -4.26174432e-01 2.28585422e-01 -2.31588393e-01 -1.00959468e+00 5.63211322e-01 5.45664966e-01 -1.47524640e-01 6.68317676e-01 -5.74885964e-01 5.28478213e-02 3.50256413e-01 -1.74936140e+00 9.30027962e-02 1.70670062e-01 -3.46948177e-01 -7.91568160e-01 3.74807030e-01 4.21523482e-01 -3.19337577e-01 -5.17036557e-01 7.53263533e-01 4.56446379e-01 -1.04346025e+00 9.06809211e-01 -5.13041258e-01 4.99421448e-01 -2.50483781e-01 2.96629332e-02 -7.18236923e-01 2.39280220e-02 -6.77044317e-02 1.01154648e-01 8.54805827e-01 2.06900105e-01 -5.71420133e-01 7.76274323e-01 1.67929456e-01 1.38959020e-01 -1.02947259e+00 -6.89896882e-01 -2.07592711e-01 7.14362189e-02 -2.30337560e-01 3.48830611e-01 8.36723447e-01 -7.95617819e-01 8.85442123e-02 -2.18825385e-01 2.79635429e-01 3.79692107e-01 -3.14439267e-01 5.50050735e-01 -9.70093012e-01 -3.01426947e-02 -1.56378046e-01 -1.12073861e-01 -2.02851370e-01 -5.71265399e-01 -6.49536729e-01 -6.94146678e-02 -1.87548280e+00 6.78276360e-01 -2.11424351e-01 -5.50971448e-01 3.97040725e-01 -3.60695273e-01 7.42192328e-01 -5.87181784e-02 2.50181615e-01 -4.17301729e-02 -3.38449866e-01 9.76112068e-01 1.90934017e-01 1.58608090e-02 -1.18760064e-01 -4.38809752e-01 7.58654058e-01 1.16747439e+00 -7.31225967e-01 -3.68016541e-01 -3.09249252e-01 2.72659749e-01 -2.54517764e-01 5.14357209e-01 -1.02127481e+00 -2.32053965e-01 6.48933351e-02 8.40999365e-01 -1.33114624e+00 1.39021367e-01 -7.79156387e-01 1.34038150e-01 9.65605021e-01 5.46734594e-02 5.69993854e-01 2.88146138e-01 2.10142896e-01 1.64391965e-01 -1.82324156e-01 1.09469199e+00 -2.25733817e-01 -2.17644781e-01 4.44480032e-02 -7.14505255e-01 1.48068994e-01 1.21049273e+00 -2.56115109e-01 -3.15447420e-01 1.66632801e-01 -3.32430512e-01 1.16096117e-01 3.32485050e-01 2.16140583e-01 5.72119176e-01 -6.26222491e-01 -8.74144256e-01 2.19008490e-01 -3.25844765e-01 2.54228413e-02 2.69249797e-01 1.01786852e+00 -1.29737103e+00 6.51140749e-01 -6.08078122e-01 -7.07113922e-01 -1.48511016e+00 6.15072250e-01 3.81042361e-01 -5.35417438e-01 -5.94396055e-01 9.17516947e-01 -4.43517119e-02 -3.52224469e-01 1.18704513e-01 -3.42432171e-01 -5.14372408e-01 7.88238868e-02 5.91437399e-01 4.61249292e-01 2.97951818e-01 -8.10842633e-01 -7.16067433e-01 3.36879045e-01 -3.12328160e-01 1.27201118e-02 1.57399929e+00 -1.21790888e-02 -1.35906279e-01 1.19408406e-01 1.17941213e+00 -1.77739069e-01 -8.69086564e-01 4.13481206e-01 -7.58012235e-02 -1.21027313e-01 -9.18051787e-03 -1.05778265e+00 -8.82256866e-01 9.27608252e-01 1.38133931e+00 2.31888950e-01 1.02591920e+00 -2.36962065e-01 9.24462020e-01 -7.58914202e-02 -2.67154127e-01 -8.76203895e-01 -6.41758367e-02 4.83998626e-01 5.17679393e-01 -1.39647043e+00 7.28733391e-02 -8.83539487e-03 -4.10221219e-01 7.56746709e-01 5.21749616e-01 -2.69396871e-01 5.57232559e-01 3.48447055e-01 4.13218617e-01 -5.71278751e-01 -5.22758782e-01 1.63977280e-01 1.24731459e-01 7.74879813e-01 6.95219576e-01 -5.47621474e-02 -4.17931378e-01 2.34835640e-01 7.80269280e-02 1.34253040e-01 3.53423119e-01 1.16763079e+00 -5.29802501e-01 -6.23118639e-01 -7.00780630e-01 7.24778593e-01 -1.21102726e+00 -1.23021312e-01 -4.17767555e-01 8.99497747e-01 6.45411789e-01 8.52477014e-01 1.15674749e-01 -1.52287424e-01 -1.33245075e-02 -4.11319733e-03 3.45025837e-01 -5.86911142e-01 -1.04856944e+00 -2.30006352e-02 -1.11440420e-01 -1.38865739e-01 -5.56396365e-01 -4.52554584e-01 -1.51340210e+00 1.33282347e-02 -1.92238390e-01 -3.09401453e-01 8.62488329e-01 9.82385695e-01 5.23492023e-02 6.64261460e-01 2.71409214e-01 -4.79061961e-01 -2.16470778e-01 -1.01468194e+00 -2.54606791e-02 3.76450717e-01 4.82630193e-01 -3.40858370e-01 -1.71113580e-01 9.13497992e-03]
[15.561485290527344, -1.7564095258712769]
76d2a872-9c60-4198-b317-43a3e97da3b8
gender-prediction-in-english-hindi-code-mixed
1806.056
null
http://arxiv.org/abs/1806.05600v1
http://arxiv.org/pdf/1806.05600v1.pdf
Gender Prediction in English-Hindi Code-Mixed Social Media Content : Corpus and Baseline System
The rapid expansion in the usage of social media networking sites leads to a huge amount of unprocessed user generated data which can be used for text mining. Author profiling is the problem of automatically determining profiling aspects like the author's gender and age group through a text is gaining much popularity in computational linguistics. Most of the past research in author profiling is concentrated on English texts \cite{1,2}. However many users often change the language while posting on social media which is called code-mixing, and it develops some challenges in the field of text classification and author profiling like variations in spelling, non-grammatical structure and transliteration \cite{3}. There are very few English-Hindi code-mixed annotated datasets of social media content present online \cite{4}. In this paper, we analyze the task of author's gender prediction in code-mixed content and present a corpus of English-Hindi texts collected from Twitter which is annotated with author's gender. We also explore language identification of every word in this corpus. We present a supervised classification baseline system which uses various machine learning algorithms to identify the gender of an author using a text, based on character and word level features.
['Ankush Khandelwal', 'Manish Shrivastava', 'Syed Sarfaraz Akhtar', 'Sahil Swami']
2018-06-14
null
null
null
null
['gender-prediction']
['computer-vision']
[-1.79481357e-01 4.41500619e-02 -2.81474710e-01 -3.16278219e-01 -4.44905937e-01 -6.83956027e-01 7.31829703e-01 9.06027198e-01 -5.90801239e-01 5.30781984e-01 3.64494950e-01 -4.99319643e-01 1.17819890e-01 -5.89544237e-01 1.03995241e-01 -3.69136006e-01 5.24015129e-02 5.31412125e-01 6.13026088e-04 -2.97963053e-01 1.04287827e+00 5.45705296e-02 -1.80709779e+00 7.14828968e-02 1.00444043e+00 4.41385537e-01 2.04845443e-01 8.98170948e-01 -8.30157995e-01 8.15135241e-01 -5.43342233e-01 -7.32718050e-01 -1.92038611e-01 -3.88598263e-01 -1.01609516e+00 5.61794862e-02 3.50101650e-01 2.37263009e-01 1.62947401e-01 1.44313669e+00 3.92113388e-01 -1.40844345e-01 9.07864988e-01 -1.25719726e+00 -2.37086698e-01 1.11929595e+00 -9.59217906e-01 4.34441596e-01 6.93439126e-01 -6.87841833e-01 1.18453646e+00 -6.51873171e-01 7.85775483e-01 1.24257147e+00 7.83716679e-01 3.47109556e-01 -8.03917408e-01 -6.96327329e-01 -1.83127716e-01 -1.80547982e-01 -1.47593915e+00 -1.41427234e-01 7.05657244e-01 -1.16698825e+00 3.79198253e-01 3.27239364e-01 3.40838760e-01 9.44947422e-01 1.66884452e-01 6.71873391e-01 1.17729974e+00 -7.94536829e-01 -2.08291009e-01 4.80933875e-01 5.33700407e-01 8.70081484e-01 1.71508461e-01 -7.16490090e-01 -5.54031730e-01 -3.45049292e-01 -1.57772660e-01 -4.36541811e-02 3.47419739e-01 9.79502872e-02 -1.14036286e+00 1.15832961e+00 -4.88752544e-01 4.67829436e-01 2.90116400e-01 -4.35974061e-01 1.06969845e+00 2.37526268e-01 7.56758153e-01 5.24380326e-01 -3.51355553e-01 -5.99317491e-01 -1.33128846e+00 5.14240742e-01 9.45180058e-01 1.01541853e+00 8.00455511e-01 -3.44983548e-01 2.68816371e-02 1.36993968e+00 6.08709753e-01 3.04114372e-01 1.12803876e+00 -6.11619771e-01 5.81844509e-01 1.14541173e+00 -1.19891353e-01 -1.19844425e+00 -5.94782531e-01 -2.82686293e-01 -5.35016298e-01 -2.12806299e-01 5.87757468e-01 -2.05585092e-01 -3.68964225e-01 1.20678794e+00 1.54777557e-01 -7.20761180e-01 -2.83079326e-01 1.40513420e-01 1.16535830e+00 6.02053404e-01 1.06270634e-01 -2.32033178e-01 1.61328721e+00 -7.88186908e-01 -6.71421647e-01 -2.79314071e-02 8.96491170e-01 -1.00785565e+00 8.75309289e-01 1.80816829e-01 -8.06932807e-01 -3.87939453e-01 -5.12753069e-01 -3.15829627e-02 -9.00982916e-01 9.39182490e-02 4.14484829e-01 1.22617280e+00 -6.62410617e-01 5.10034859e-01 -2.14168236e-01 -5.81739128e-01 2.71217436e-01 3.63979310e-01 -1.10960238e-01 3.45958203e-01 -9.33489621e-01 3.39701682e-01 2.01267079e-01 -7.53100336e-01 1.84376929e-02 -5.54566860e-01 -9.46260393e-01 -4.41921532e-01 -1.11072265e-01 1.17253913e-02 1.28699207e+00 -1.12445116e+00 -1.21337247e+00 1.69427788e+00 -2.25503549e-01 1.63859315e-02 7.67940581e-01 2.05661714e-01 -8.06356370e-01 -5.02001882e-01 7.75470138e-01 3.31327081e-01 6.77318811e-01 -8.61947596e-01 -1.30711830e+00 -6.37540519e-01 -3.48769963e-01 -1.64686009e-01 -7.32053816e-01 7.33743250e-01 -1.88744903e-01 -8.51515532e-01 8.33141580e-02 -1.06276846e+00 -2.32161600e-02 -8.86101782e-01 -6.23062372e-01 -7.73939013e-01 4.95860875e-01 -8.63937140e-01 1.85364163e+00 -2.14190674e+00 2.05635559e-02 4.00717527e-01 5.91375768e-01 -2.20426768e-01 4.81689394e-01 7.14670241e-01 1.44502118e-01 4.44410682e-01 -6.99822232e-02 -5.89617193e-01 1.61218554e-01 -9.38235745e-02 -1.33997396e-01 4.97261852e-01 -6.45328403e-01 2.63495386e-01 -1.00301051e+00 -1.04200041e+00 -3.42425555e-01 -6.32249787e-02 -5.70964873e-01 2.20663920e-01 1.53438419e-01 3.36803734e-01 -2.87415951e-01 7.59252369e-01 5.83825052e-01 1.42485127e-01 1.69626288e-02 6.32829845e-01 -8.48258913e-01 5.77971004e-02 -8.99882734e-01 1.26136184e+00 -3.85786563e-01 1.13506365e+00 -1.04531152e-02 -5.64569056e-01 1.11705315e+00 -1.38047740e-01 5.84833205e-01 -2.37734362e-01 6.75042331e-01 5.70600092e-01 1.92318007e-01 -6.75924718e-01 9.79919553e-01 2.66923726e-01 -7.45659113e-01 5.59153497e-01 -2.35183403e-01 1.90429419e-01 1.02480185e+00 3.36142004e-01 7.33165622e-01 -1.27755120e-01 3.92186850e-01 -8.38390350e-01 9.82402325e-01 -3.43476348e-02 1.92047104e-01 6.82988822e-01 -3.35532159e-01 5.33266306e-01 8.22928488e-01 -3.07214051e-01 -1.12041461e+00 -3.96455318e-01 -4.07089651e-01 1.91801274e+00 -3.51676941e-01 -7.38471091e-01 -1.03833747e+00 -4.12888259e-01 2.11539745e-01 2.53705591e-01 -6.27443254e-01 3.93497467e-01 -4.02198672e-01 -8.44537020e-01 8.15033078e-01 -1.54230986e-02 3.95609111e-01 -9.02584255e-01 -3.71169776e-01 9.78331715e-02 -1.90437004e-01 -9.46935594e-01 -5.00597954e-01 6.05509169e-02 -4.90491062e-01 -9.33448076e-01 -6.98419631e-01 -1.00272751e+00 7.58265376e-01 -1.99854821e-01 1.24857152e+00 3.05694401e-01 -2.15311542e-01 8.62760469e-02 -6.94073319e-01 -9.24908996e-01 -7.25196242e-01 1.01804340e+00 -6.78497506e-03 4.83251177e-02 7.29821503e-01 -2.83119768e-01 -1.48779377e-01 1.79902581e-03 -4.90982026e-01 -1.05408356e-01 8.51486400e-02 3.20891440e-01 -3.09166908e-01 1.05116116e-02 1.44512296e-01 -1.73622787e+00 9.83640432e-01 -7.37874448e-01 -1.58173099e-01 -3.55021417e-01 -7.45261133e-01 6.59785867e-02 6.93091273e-01 -2.12030008e-01 -6.43531740e-01 -1.79786131e-01 -4.01833832e-01 6.68549240e-01 -5.76753318e-02 6.25190020e-01 1.23066623e-02 1.70469701e-01 5.58201849e-01 8.87019858e-02 3.03208139e-02 -6.09874010e-01 -3.04346621e-01 1.47049594e+00 2.54089355e-01 -3.97184283e-01 7.49438703e-01 -5.65361581e-04 -2.98668265e-01 -9.94984686e-01 -9.28456366e-01 -1.04119086e+00 -1.01446450e+00 -3.30963254e-01 8.02766085e-01 -8.24921250e-01 -8.54936481e-01 8.63869309e-01 -7.47374177e-01 5.69160879e-02 3.78852874e-01 8.09106454e-02 -1.13751762e-01 5.70617437e-01 -6.31357551e-01 -8.60706508e-01 -2.27251798e-01 -9.09168422e-01 9.96895075e-01 1.33688897e-01 -1.05226088e+00 -1.14556873e+00 2.67096668e-01 6.40292525e-01 9.05451253e-02 3.00914109e-01 9.85525131e-01 -9.41430748e-01 3.86160761e-01 -3.56598079e-01 -1.67488575e-01 -8.87373388e-02 -1.96842983e-01 5.45146108e-01 -5.95686018e-01 -1.54763535e-01 -5.66556513e-01 -2.46000979e-02 6.10584497e-01 -3.83422002e-02 1.33895612e+00 -3.59366268e-01 -2.68833399e-01 4.36608523e-01 1.22731626e+00 -1.77358817e-02 1.90762490e-01 7.38463640e-01 1.07509053e+00 6.56066656e-01 3.42768639e-01 1.02211475e+00 7.90396392e-01 2.44572774e-01 7.75302052e-02 3.99432361e-01 4.89710927e-01 -1.88117757e-01 2.94968694e-01 1.29845119e+00 -3.75298947e-01 -1.10728331e-01 -1.52705896e+00 4.15962815e-01 -1.54618418e+00 -1.01279092e+00 -7.45445490e-01 1.81556106e+00 1.07697701e+00 1.63137630e-01 7.11848199e-01 7.09031641e-01 9.75239575e-01 2.52534389e-01 7.65015259e-02 -1.03151417e+00 1.64288804e-01 -5.89273870e-02 8.53265345e-01 5.69183469e-01 -1.20272791e+00 6.13151073e-01 5.51896238e+00 7.60543466e-01 -8.57805610e-01 1.24491446e-01 6.09955370e-01 4.81175333e-01 -5.72328344e-02 -2.39054531e-01 -1.36685240e+00 1.08586943e+00 1.10055864e+00 -2.66071945e-01 2.31563821e-01 9.17283952e-01 -7.68014509e-03 -4.34008360e-01 -8.49762321e-01 1.38096082e+00 4.07120377e-01 -9.30019379e-01 -2.31964290e-01 4.12714154e-01 8.08450162e-01 -2.41633624e-01 4.81387377e-02 5.38514078e-01 2.26031095e-01 -8.42034161e-01 1.12658620e+00 -1.97151136e-02 8.12760592e-01 -9.59635556e-01 7.76939034e-01 7.18683898e-01 -9.42540944e-01 -3.83935899e-01 -9.17138159e-02 -4.92369622e-01 -3.77096176e-01 6.34237468e-01 -5.75192869e-01 -1.01897590e-01 6.96596622e-01 7.39033103e-01 -1.16437948e+00 6.68238103e-01 2.66058624e-01 6.31911218e-01 2.70351261e-01 -6.71567500e-01 1.68290690e-01 -2.65962303e-01 4.02635276e-01 1.65668237e+00 3.58176947e-01 -4.79718924e-01 3.18053871e-01 1.66166916e-01 -2.46413231e-01 7.82020032e-01 -3.50133598e-01 -3.18116456e-01 3.69248807e-01 1.44217122e+00 -1.20442867e+00 -1.99672610e-01 -4.77283597e-01 7.99178362e-01 2.31089145e-01 -2.14600146e-01 -5.94937205e-01 -7.55426049e-01 3.33176076e-01 6.28949583e-01 -3.55668575e-01 -3.09079081e-01 -3.52860898e-01 -9.61615026e-01 -4.05712634e-01 -8.84944797e-01 5.68813086e-01 -1.57279208e-01 -1.27910328e+00 3.80720913e-01 -2.70405531e-01 -1.02553928e+00 -2.56339163e-01 -8.60669553e-01 -2.51371652e-01 7.86112309e-01 -9.68004465e-01 -9.46733236e-01 -3.91421020e-01 2.28365898e-01 6.85859203e-01 -9.16517317e-01 6.00791931e-01 6.43922389e-01 -7.02659369e-01 8.21471035e-01 4.80298966e-01 7.28569925e-01 7.84974873e-01 -1.57399738e+00 1.19809628e-01 3.89312893e-01 -1.26432195e-01 5.39296806e-01 1.01231074e+00 -8.77317488e-01 -1.03084767e+00 -9.80197668e-01 1.77353442e+00 -9.43209589e-01 9.84530032e-01 -7.23146260e-01 -2.24736482e-01 4.26871389e-01 2.83203930e-01 -4.49544966e-01 1.13987458e+00 3.48192781e-01 -3.01262110e-01 2.47290611e-01 -1.03797877e+00 5.26260555e-01 9.98529494e-01 -6.84644878e-01 -3.14449131e-01 4.19512898e-01 -9.27248448e-02 -3.39013368e-01 -8.39904547e-01 -2.55557239e-01 5.42320967e-01 -9.83381331e-01 4.02333975e-01 -3.78378600e-01 1.04308486e+00 1.92423269e-01 1.41512915e-01 -1.05351019e+00 -3.91501099e-01 -7.12183177e-01 2.89776206e-01 1.85227668e+00 4.43581432e-01 -4.00254577e-01 7.72512138e-01 6.01552188e-01 3.16962600e-02 -4.18494165e-01 -5.54623425e-01 -3.70965809e-01 3.98318559e-01 3.47142783e-03 4.09077466e-01 1.14339924e+00 3.69958043e-01 4.81212616e-01 -6.45707473e-02 -4.74604666e-01 4.44898367e-01 5.00448653e-03 9.31312382e-01 -1.87541687e+00 3.30339938e-01 -8.71546566e-01 -6.14953101e-01 -3.80232543e-01 5.30277252e-01 -1.17589033e+00 -1.69506103e-01 -1.19761014e+00 6.40458763e-01 -3.61083746e-01 5.22487402e-01 -1.26154274e-01 -2.24471483e-02 3.24414968e-01 -1.41357809e-01 4.62813497e-01 -6.79548085e-01 -2.52248496e-01 7.82645702e-01 -2.48798653e-02 -4.93341275e-02 1.18770286e-01 -6.73104942e-01 9.58569765e-01 6.37594044e-01 -6.10190272e-01 9.60307568e-02 1.26526475e-01 1.10106790e+00 -4.36692774e-01 -3.85380894e-01 -9.28794026e-01 1.56299725e-01 -2.62439568e-02 1.93781093e-01 -5.90841770e-01 -4.35268611e-01 -5.24779201e-01 -3.32428694e-01 4.99573469e-01 -3.93905103e-01 5.70044219e-01 -3.55644345e-01 2.29605153e-01 -2.31398523e-01 -9.10288274e-01 8.40893805e-01 -4.59354490e-01 -5.19944072e-01 3.17888886e-01 -1.07961452e+00 5.65563917e-01 9.20541108e-01 -5.71482241e-01 -2.46802978e-02 8.79692212e-02 -4.05922562e-01 1.62537172e-02 8.30317140e-01 4.80425715e-01 -2.05964178e-01 -9.18826818e-01 -8.50986481e-01 7.83398747e-02 4.39202964e-01 -6.64083183e-01 -2.05264837e-01 5.93367696e-01 -8.93973947e-01 1.29828706e-01 -2.99775451e-01 -3.97283852e-01 -1.61432421e+00 2.43117616e-01 -1.00633413e-01 -2.73411900e-01 -5.54760881e-02 7.90269911e-01 -6.07061028e-01 -7.20541298e-01 1.47203326e-01 2.38143265e-01 -1.04363513e+00 1.18307137e+00 2.77369529e-01 7.33468115e-01 -3.88817564e-02 -1.22949946e+00 -3.74191821e-01 3.73378783e-01 -8.73179287e-02 -1.37976304e-01 1.18355978e+00 -4.44678634e-01 -9.25098658e-01 1.11684906e+00 1.49478745e+00 8.18648994e-01 -2.11710736e-01 5.77776581e-02 6.36670768e-01 -7.63665974e-01 -2.54651874e-01 -9.60947648e-02 -7.29829431e-01 6.91438735e-01 4.01149392e-01 7.60185480e-01 3.09319109e-01 -2.78841928e-02 7.77015924e-01 2.65235100e-02 1.61095902e-01 -1.92366314e+00 -1.38603345e-01 1.12083757e+00 2.75532067e-01 -1.49037647e+00 5.81350401e-02 -2.87686914e-01 -4.57185119e-01 1.18988752e+00 4.43869621e-01 2.82744765e-01 1.01906359e+00 3.30535978e-01 1.55724153e-01 -1.27175674e-01 7.60648772e-02 -1.72032699e-01 7.47940615e-02 1.72287464e-01 1.13435245e+00 2.39035487e-01 -9.29069400e-01 8.61420810e-01 -1.02260149e+00 -5.40390372e-01 9.87898946e-01 8.28841031e-01 -6.57759428e-01 -1.43571496e+00 -4.25701082e-01 9.73527730e-01 -1.30420661e+00 -1.31971359e-01 -5.77428401e-01 4.85306233e-01 3.72087896e-01 8.47678304e-01 2.59457737e-01 -3.07639450e-01 -2.39282057e-01 2.75803655e-01 1.80886611e-01 -9.35052812e-01 -9.99041498e-01 -4.28408176e-01 2.57399172e-01 3.59628260e-01 -4.69879836e-01 -1.37948728e+00 -1.28047848e+00 -8.69229496e-01 1.09879509e-01 5.44349194e-01 8.31420302e-01 8.36238325e-01 -8.70003179e-02 1.93048611e-01 7.17647612e-01 -6.16764784e-01 -2.79384032e-02 -1.00512397e+00 -9.04679954e-01 7.17695117e-01 9.59447548e-02 -5.69229245e-01 -5.50637007e-01 2.84632355e-01]
[9.473895072937012, 10.343740463256836]
4b59b1eb-742b-4dde-8bea-fa828d726c49
counterfactual-reasoning-do-language-models
2212.03278
null
https://arxiv.org/abs/2212.03278v1
https://arxiv.org/pdf/2212.03278v1.pdf
Counterfactual reasoning: Do language models need world knowledge for causal understanding?
Current pre-trained language models have enabled remarkable improvements in downstream tasks, but it remains difficult to distinguish effects of statistical correlation from more systematic logical reasoning grounded on understanding of the real world. In this paper we tease these factors apart by leveraging counterfactual conditionals, which force language models to predict unusual consequences based on hypothetical propositions. We introduce a set of tests drawn from psycholinguistic experiments, as well as larger-scale controlled datasets, to probe counterfactual predictions from a variety of popular pre-trained language models. We find that models are consistently able to override real-world knowledge in counterfactual scenarios, and that this effect is more robust in case of stronger baseline world knowledge -- however, we also find that for most models this effect appears largely to be driven by simple lexical cues. When we mitigate effects of both world knowledge and lexical cues to test knowledge of linguistic nuances of counterfactuals, we find that only GPT-3 shows sensitivity to these nuances, though this sensitivity is also non-trivially impacted by lexical associative factors.
['Allyson Ettinger', 'Lang Yu', 'Jiaxuan Li']
2022-12-06
null
null
null
null
['logical-reasoning']
['reasoning']
[ 4.35587168e-02 5.93979418e-01 -2.86750495e-01 -3.43212247e-01 -6.15637124e-01 -6.65194035e-01 1.26976728e+00 3.67468059e-01 -5.55547059e-01 1.27322888e+00 1.00996780e+00 -9.35688674e-01 -2.80433059e-01 -1.08292401e+00 -1.03818440e+00 -2.01939717e-01 -3.70994151e-01 2.59503454e-01 1.37595400e-01 -4.81288671e-01 4.91562486e-01 2.60444969e-01 -1.18684864e+00 4.56262439e-01 6.58868194e-01 2.46456385e-01 -9.26288962e-02 1.78333178e-01 5.73819317e-02 1.17599940e+00 -2.80413628e-01 -7.73358285e-01 2.35338271e-01 -4.33244854e-01 -8.94088328e-01 -6.09441757e-01 3.27839732e-01 -2.46029884e-01 -3.43189895e-01 9.55813706e-01 5.32239854e-01 4.97127585e-02 7.25815654e-01 -9.37059343e-01 -8.60212684e-01 1.42865229e+00 -3.11314374e-01 5.62135935e-01 6.29838586e-01 8.47039998e-01 1.26161110e+00 -4.67955500e-01 1.00434244e+00 1.80141139e+00 7.72565067e-01 1.06198117e-01 -1.58177090e+00 -6.37340367e-01 4.33457047e-01 1.30225629e-01 -9.50977266e-01 -6.74957275e-01 6.77246749e-01 -3.50333899e-01 1.27811480e+00 -1.23822475e-02 6.05047405e-01 1.54390705e+00 6.63732886e-01 3.23779821e-01 1.83092952e+00 -5.70083141e-01 1.44578904e-01 3.88207622e-02 7.63465688e-02 3.20940524e-01 7.28298903e-01 8.53231847e-01 -8.03467453e-01 -3.37035567e-01 6.27530396e-01 -6.80017471e-01 -5.77810109e-01 -2.96498001e-01 -1.56307316e+00 7.31081009e-01 3.59815896e-01 3.52887064e-01 -5.97171724e-01 4.05147374e-02 3.98084968e-01 3.47865611e-01 3.31849813e-01 9.62877035e-01 -8.30014706e-01 1.20200641e-01 -6.93519115e-01 6.43831670e-01 9.82147336e-01 3.84397805e-01 5.08531153e-01 -1.25794550e-02 -3.35189342e-01 4.06569034e-01 3.14626664e-01 3.76166642e-01 7.90861189e-01 -8.93893719e-01 5.95134199e-01 6.52866289e-02 4.84144866e-01 -1.01186657e+00 -5.35096407e-01 -2.11783066e-01 -2.93543853e-04 8.60984176e-02 7.16957390e-01 -2.74539292e-01 -5.58879793e-01 2.33816624e+00 4.89901640e-02 6.80255368e-02 2.13507518e-01 6.84161246e-01 2.53075063e-01 2.26330966e-01 6.48959398e-01 -6.39687896e-01 1.16073143e+00 3.94064263e-02 -5.05550027e-01 -5.83920002e-01 1.02707803e+00 -4.57185090e-01 1.35408485e+00 3.90668064e-02 -1.07809138e+00 -1.77678034e-01 -1.14499116e+00 -7.79671595e-02 -3.47931117e-01 -6.79571033e-01 1.19628513e+00 6.80018306e-01 -1.06333852e+00 7.33565450e-01 -5.33154666e-01 -3.59074891e-01 3.12055320e-01 -4.22190651e-02 -1.51254982e-01 1.53890355e-02 -1.76038265e+00 1.59358943e+00 8.98889244e-01 -5.72468899e-02 -8.17051470e-01 -7.02704132e-01 -9.30739820e-01 -4.75588366e-02 6.77235305e-01 -8.48347008e-01 1.14503610e+00 -1.01353014e+00 -1.26648116e+00 1.01024926e+00 -1.59128174e-01 -8.53017628e-01 6.63369060e-01 -3.83514166e-02 -5.84695399e-01 -2.92840272e-01 4.30256844e-01 5.36835134e-01 3.97123814e-01 -1.10685706e+00 -7.36975893e-02 -4.39564288e-01 4.18357283e-01 3.77435654e-01 2.85516798e-01 1.44096330e-01 6.88026667e-01 -5.60955048e-01 -1.08043268e-01 -7.44668365e-01 -9.00784656e-02 -2.41112843e-01 -5.42411089e-01 -1.68838874e-01 -1.35203805e-02 -3.64778221e-01 1.01119435e+00 -1.68790460e+00 -6.00691020e-01 4.63826992e-02 -1.26829773e-01 -1.40545383e-01 -2.94826310e-02 3.61905575e-01 -8.25518668e-01 5.16645730e-01 -9.64597240e-02 4.87998605e-01 3.88017029e-01 1.05211148e-02 -8.92921507e-01 4.10902143e-01 3.84888977e-01 1.31459415e+00 -8.73195410e-01 -4.44256991e-01 1.13951914e-01 -1.61161438e-01 -6.33191049e-01 -3.57400000e-01 -5.32674730e-01 -1.37590751e-01 -1.20804198e-01 1.75567731e-01 5.04889667e-01 4.99682128e-02 6.38306499e-01 1.16058881e-03 -3.17687154e-01 1.40242493e+00 -9.67785597e-01 1.29906952e+00 -3.65303189e-01 4.84565705e-01 -3.27974290e-01 -9.55639958e-01 3.55124861e-01 2.51766086e-01 -5.94498515e-01 -7.35038817e-01 7.57161379e-02 1.90295473e-01 9.48835373e-01 -6.15383744e-01 2.70974368e-01 -1.23930466e+00 -1.04204580e-01 7.28723466e-01 -7.20564201e-02 -3.43946815e-01 1.33549035e-01 2.42315158e-01 7.85130024e-01 5.69437861e-01 8.68142724e-01 -8.11854482e-01 8.79074410e-02 2.80452877e-01 6.35906816e-01 1.14319682e+00 -2.20970333e-01 -1.56580135e-02 7.13036060e-01 -3.94117653e-01 -8.74049604e-01 -1.10715389e+00 -4.35806096e-01 8.98327827e-01 -4.13816012e-02 -1.06642261e-01 -1.83572039e-01 -5.85614741e-01 1.48522720e-01 1.93284452e+00 -7.46728182e-01 -4.69672382e-01 -3.76221865e-01 -1.28586709e+00 6.13839924e-01 4.50944364e-01 1.66770428e-01 -1.19465661e+00 -8.53523016e-01 1.56066194e-01 -7.10399449e-02 -8.71748328e-01 2.39194870e-01 -8.07629153e-02 -9.08532917e-01 -9.01230812e-01 2.48508751e-02 -7.45940208e-03 -1.35908261e-01 -9.04831663e-02 1.38011181e+00 2.09217612e-02 1.67687625e-01 6.38705343e-02 7.55946636e-02 -8.35464120e-01 -6.30236626e-01 -5.57362497e-01 2.65576869e-01 -6.24475241e-01 6.24928892e-01 -8.35906446e-01 -2.28551477e-01 -1.59977078e-01 -7.30640352e-01 9.60857049e-02 6.21638238e-01 7.32984841e-01 4.17138301e-02 -9.82666016e-02 7.48523831e-01 -1.08652508e+00 9.47010398e-01 -8.07400465e-01 -3.79515558e-01 2.43881732e-01 -4.96262640e-01 4.52245653e-01 6.15369737e-01 -4.60692316e-01 -1.58215666e+00 -8.15894008e-01 1.73152849e-01 4.10609871e-01 -5.04990637e-01 8.20078552e-01 -2.12456793e-01 4.96417463e-01 1.01626956e+00 -6.20176494e-02 -2.41092786e-01 1.58761255e-02 6.26512885e-01 1.41881868e-01 2.84262151e-01 -1.25295436e+00 6.63641393e-01 5.55801332e-01 -1.80356488e-01 -5.55918157e-01 -1.20556176e+00 5.92233241e-01 -4.01046485e-01 2.39006236e-01 6.27107799e-01 -1.05103147e+00 -5.95429599e-01 -6.22044057e-02 -1.07364821e+00 -7.53853261e-01 -2.69770235e-01 8.43955219e-01 -7.76763499e-01 2.20560014e-01 -5.24920046e-01 -8.31554651e-01 3.04616153e-01 -8.18060100e-01 4.48019147e-01 -2.95273602e-01 -7.94487238e-01 -1.27618134e+00 -6.64940244e-03 1.97138581e-02 2.77935117e-01 3.02059561e-01 1.36246037e+00 -1.02973008e+00 -4.47760969e-01 5.75632229e-02 -2.52078861e-01 -1.57482818e-01 -1.40265971e-02 -4.97883894e-02 -1.06295717e+00 2.55314916e-01 3.66979092e-01 -5.98096371e-01 1.02703309e+00 4.06396419e-01 5.88916600e-01 -5.39421320e-01 -3.60290945e-01 8.06860328e-02 1.41681457e+00 -5.42249866e-02 6.83160067e-01 3.25576633e-01 1.14056639e-01 9.19216454e-01 4.62796897e-01 2.23900061e-02 6.09545231e-01 4.69272852e-01 1.49330571e-02 4.33335900e-01 3.17727774e-02 -6.69213951e-01 6.08436525e-01 -1.56266093e-01 2.25294624e-02 -1.23145223e-01 -1.11499524e+00 8.55580330e-01 -1.45895529e+00 -1.49002898e+00 -2.20150635e-01 2.30603814e+00 1.22518611e+00 8.26641738e-01 -4.69127744e-02 -1.21337056e-01 4.57678080e-01 3.76335561e-01 -3.69473368e-01 -4.63017106e-01 -4.66982365e-01 1.79309741e-01 3.93904269e-01 7.97073781e-01 -6.43381357e-01 1.14365935e+00 7.44775105e+00 5.64724207e-01 -1.03882015e+00 -1.00825511e-01 7.58562863e-01 -3.20140034e-01 -8.83774579e-01 3.07911396e-01 -2.42131069e-01 3.17991316e-01 1.23529696e+00 -6.04769289e-01 9.73152742e-02 4.19558048e-01 4.12975699e-01 -4.40137118e-01 -1.54535949e+00 1.36332452e-01 -1.96367487e-01 -1.28171337e+00 3.21699917e-01 -1.25959069e-01 5.93163013e-01 1.60423741e-01 -4.31037769e-02 4.41356629e-01 1.03151155e+00 -1.22903037e+00 1.14289272e+00 2.32095614e-01 4.57257509e-01 -3.93115342e-01 7.62782931e-01 5.84559083e-01 -1.85273319e-01 -1.49802923e-01 -4.57022548e-01 -9.47637558e-01 2.94829667e-01 6.42093956e-01 -8.06606233e-01 3.80436331e-01 1.73703581e-01 2.53414959e-01 -3.67358565e-01 4.37079191e-01 -6.59379005e-01 7.43967593e-01 -2.53662080e-01 -1.17359877e-01 1.08560860e-01 2.08346099e-01 4.54966217e-01 1.24797750e+00 -3.65636833e-02 3.71783078e-01 -4.03809279e-01 1.29710340e+00 4.36765216e-02 -1.21619832e-02 -1.16209686e+00 2.14546639e-02 5.65658510e-01 5.79886496e-01 -5.20997405e-01 -4.40574288e-01 -5.48079193e-01 4.80347157e-01 5.95061719e-01 4.29779619e-01 -8.33741307e-01 1.98167399e-01 5.40882945e-01 5.79867652e-03 -6.33573458e-02 -6.43103346e-02 -5.90336502e-01 -1.51750088e+00 1.88995767e-02 -8.50971162e-01 2.42247418e-01 -1.02149475e+00 -1.32838953e+00 -6.11844733e-02 4.30060804e-01 -3.70061457e-01 -5.30594945e-01 -8.80658627e-01 -7.70184219e-01 1.09828353e+00 -1.40146196e+00 -8.70302796e-01 7.79387355e-01 3.21978718e-01 1.33948401e-01 4.61047322e-01 8.10065687e-01 -5.98334074e-01 -2.90419787e-01 9.39741954e-02 -5.41792989e-01 -5.11759780e-02 9.66300428e-01 -1.15683937e+00 5.15409827e-01 1.07818663e+00 2.50765532e-01 1.35067332e+00 1.18202078e+00 -8.30242634e-01 -8.74297678e-01 -4.88880962e-01 1.19476771e+00 -1.01518679e+00 1.07985246e+00 -2.07575664e-01 -8.45379472e-01 1.21746111e+00 3.02230150e-01 -2.20588788e-01 7.50251710e-01 7.14256942e-01 -9.84783530e-01 4.45921719e-01 -1.01647139e+00 1.26590681e+00 1.52248919e+00 -6.82989120e-01 -1.65005016e+00 2.04615161e-01 8.43441129e-01 -1.04564756e-01 -4.47407305e-01 4.05320257e-01 5.55338562e-01 -1.25065315e+00 8.89806747e-01 -1.25937402e+00 8.60475361e-01 -7.05475137e-02 -2.48106807e-01 -1.59274292e+00 -3.46934050e-01 -4.35438514e-01 4.72302705e-01 1.02794778e+00 9.37279820e-01 -1.02244735e+00 9.72205102e-02 1.15544200e+00 3.71174403e-02 -3.76631230e-01 -9.66995656e-01 -6.75701797e-01 5.68679094e-01 -8.83037448e-01 6.21793807e-01 1.33962905e+00 6.44541681e-01 5.30834019e-01 2.45297849e-01 6.11057803e-02 5.54572940e-01 2.49897480e-01 3.63275170e-01 -8.64026189e-01 -4.90705013e-01 -5.40552557e-01 -5.38169444e-02 -6.38981581e-01 6.01951540e-01 -9.47362959e-01 -1.32213682e-01 -1.27695882e+00 4.34244424e-01 -2.05878332e-01 -8.16272125e-02 3.54966283e-01 -4.35556442e-01 -1.79252759e-01 2.44353339e-01 -2.03322902e-01 -2.58073837e-01 4.47339982e-01 1.07079482e+00 3.39687616e-01 -7.87381679e-02 -4.82488990e-01 -1.36693716e+00 1.21832502e+00 7.35656798e-01 -3.02573681e-01 -5.28830767e-01 -4.67136979e-01 6.12190783e-01 1.55787900e-01 7.41150618e-01 -4.06681746e-01 -2.17407823e-01 -7.37719417e-01 5.35703599e-01 1.75469711e-01 1.10054798e-01 -3.22732478e-01 -1.10183932e-01 5.02770007e-01 -7.34645605e-01 -6.22921437e-02 7.13525891e-01 4.84567821e-01 1.97810560e-01 7.57356435e-02 3.62530202e-01 -4.71419901e-01 -6.71481729e-01 -4.27506417e-01 -4.44756955e-01 4.77721155e-01 6.89401627e-01 -6.29425645e-02 -6.63599432e-01 -4.06682909e-01 -4.97492373e-01 -4.33751848e-03 5.73143661e-01 3.21892112e-01 2.27974623e-01 -1.00305653e+00 -1.04693174e+00 -2.35939309e-01 -5.30296452e-02 -7.61000752e-01 1.68589458e-01 9.35771823e-01 -1.69839770e-01 7.80520797e-01 -5.99488765e-02 2.20241204e-01 -4.73345220e-01 9.17515874e-01 3.69754821e-01 -2.39461437e-01 -2.12299839e-01 8.29875767e-01 4.96664286e-01 -3.38785917e-01 -6.25488639e-01 -5.83071828e-01 1.27635635e-02 2.76070298e-03 4.75925773e-01 1.71830673e-02 -3.26864123e-01 -4.99822706e-01 -4.44557607e-01 -5.56904152e-02 -8.23720470e-02 -5.20596385e-01 1.16758060e+00 -7.00612739e-02 -2.01291963e-01 7.39092886e-01 5.13397157e-01 4.30930823e-01 -9.53162313e-01 -1.62627865e-02 2.57039696e-01 -3.17028075e-01 -2.89392740e-01 -1.34749484e+00 -4.42985557e-02 6.24893486e-01 -2.12829322e-01 1.51750937e-01 6.13142431e-01 6.77688494e-02 3.33147012e-02 3.90745103e-01 6.89128518e-01 -9.08605933e-01 -5.16574025e-01 3.79100442e-01 1.10386658e+00 -1.11365962e+00 2.06729308e-01 -2.63740361e-01 -5.97703218e-01 6.96221173e-01 5.55654347e-01 -2.21850634e-01 3.40313494e-01 1.52769998e-01 -7.32290745e-02 -1.39598116e-01 -1.27023900e+00 -1.18972570e-01 -1.68278545e-01 5.38557768e-01 9.41376567e-01 2.89993852e-01 -7.35030472e-01 7.51753092e-01 -9.49797750e-01 -1.35687470e-01 6.95556104e-01 3.58239025e-01 -6.37933165e-02 -7.00578749e-01 -5.42277992e-01 6.13392889e-01 -7.52453446e-01 -7.31921673e-01 -6.13770247e-01 1.32428288e+00 8.21151212e-02 7.91566789e-01 9.47722346e-02 1.17030144e-01 9.04989541e-02 4.91760731e-01 6.68207407e-01 -7.73249745e-01 -2.46165201e-01 -2.99363971e-01 7.44753659e-01 -6.35714471e-01 -4.72625405e-01 -8.65328610e-01 -1.10191560e+00 -3.23938340e-01 -2.26262406e-01 -7.77895972e-02 1.71482582e-02 1.29003978e+00 2.39613801e-01 1.86075643e-01 -2.51020938e-01 -5.11437356e-01 -9.78954077e-01 -1.10199201e+00 -5.70331633e-01 5.06430745e-01 2.39477262e-01 -6.99845076e-01 -5.22564530e-01 -1.55706584e-01]
[9.928193092346191, 7.9254350662231445]
ec9a9ce9-7e9f-44fd-9923-9e3fae99e126
real-time-multi-view-3d-human-pose-estimation
2106.14729
null
https://arxiv.org/abs/2106.14729v1
https://arxiv.org/pdf/2106.14729v1.pdf
Real-Time Multi-View 3D Human Pose Estimation using Semantic Feedback to Smart Edge Sensors
We present a novel method for estimation of 3D human poses from a multi-camera setup, employing distributed smart edge sensors coupled with a backend through a semantic feedback loop. 2D joint detection for each camera view is performed locally on a dedicated embedded inference processor. Only the semantic skeleton representation is transmitted over the network and raw images remain on the sensor board. 3D poses are recovered from 2D joints on a central backend, based on triangulation and a body model which incorporates prior knowledge of the human skeleton. A feedback channel from backend to individual sensors is implemented on a semantic level. The allocentric 3D pose is backprojected into the sensor views where it is fused with 2D joint detections. The local semantic model on each sensor can thus be improved by incorporating global context information. The whole pipeline is capable of real-time operation. We evaluate our method on three public datasets, where we achieve state-of-the-art results and show the benefits of our feedback architecture, as well as in our own setup for multi-person experiments. Using the feedback signal improves the 2D joint detections and in turn the estimated 3D poses.
['Sven Behnke', 'Simon Bultmann']
2021-06-28
null
null
null
null
['3d-multi-person-pose-estimation']
['computer-vision']
[ 6.96436465e-02 4.16682839e-01 1.93150610e-01 -3.45134377e-01 -7.09584713e-01 -4.04144496e-01 2.86371738e-01 7.45809004e-02 -8.28832448e-01 1.85551226e-01 2.92752028e-01 7.43877828e-01 2.40523845e-01 -6.90443099e-01 -8.74750018e-01 -1.47496700e-01 1.38841523e-02 1.05160582e+00 8.43756378e-01 -1.30077332e-01 -1.06857076e-01 3.57874840e-01 -1.62052596e+00 2.53276199e-01 -1.06151059e-01 1.07462788e+00 1.82452887e-01 1.20963025e+00 6.55415654e-01 1.57565862e-01 -4.83605474e-01 -1.66588932e-01 4.39396590e-01 5.22772074e-02 -2.80423731e-01 4.13677782e-01 6.12043619e-01 -5.36039650e-01 -8.23067278e-02 8.39657903e-01 8.34092379e-01 -1.37852773e-01 8.23524520e-02 -1.10683787e+00 4.96550560e-01 -8.24734047e-02 -3.76574188e-01 -3.58804911e-01 1.28004718e+00 1.59692079e-01 8.18328679e-01 -9.55502331e-01 9.57292557e-01 1.33103299e+00 9.19852257e-01 4.02435392e-01 -8.76109600e-01 -1.08265735e-01 1.34036854e-01 2.40318432e-01 -1.23241079e+00 -4.01243597e-01 7.64845729e-01 -3.09030235e-01 1.10513306e+00 2.76713759e-01 1.13936210e+00 1.09557915e+00 1.87369496e-01 5.30504227e-01 8.49395514e-01 -4.78073001e-01 3.53247464e-01 5.39469607e-02 -1.03056572e-01 1.03672290e+00 2.17242658e-01 6.24026433e-02 -1.15967822e+00 5.69723286e-02 1.10615146e+00 1.98540583e-01 -3.17111984e-03 -7.39191055e-01 -1.32947433e+00 4.38614547e-01 6.83577955e-01 -3.46652895e-01 -6.50515556e-01 5.94468832e-01 2.68108934e-01 1.01820387e-01 3.64715904e-01 -2.86067933e-01 -6.83375120e-01 -1.54302970e-01 -7.26124823e-01 3.54298264e-01 8.46802592e-01 9.41723406e-01 8.33200693e-01 -8.10049057e-01 1.63572505e-01 2.64452010e-01 8.34287643e-01 7.53507137e-01 4.64289263e-02 -1.31278515e+00 5.26827931e-01 7.03309774e-01 2.57615834e-01 -1.02925467e+00 -1.06822670e+00 -2.43104622e-01 -3.58009189e-01 3.35596472e-01 4.11826640e-01 -3.09372306e-01 -4.74093139e-01 1.31464601e+00 9.29372013e-01 -7.75941834e-02 -4.13259447e-01 1.49095464e+00 5.77111840e-01 -5.49897971e-03 -4.49576020e-01 3.69209170e-01 1.93627775e+00 -9.88462687e-01 -4.23149824e-01 -6.79387927e-01 4.35497880e-01 -5.96470237e-01 3.88282061e-01 7.95594037e-01 -1.03844202e+00 -7.74699628e-01 -1.00695014e+00 -2.91705459e-01 -1.92971811e-01 3.89843404e-01 3.11063915e-01 6.16985857e-01 -1.20410550e+00 5.60913324e-01 -1.42237175e+00 -8.17358732e-01 -6.43342361e-02 6.25004292e-01 -7.22043812e-01 -1.44358844e-01 -8.44265282e-01 1.01598048e+00 3.81317466e-01 4.06292349e-01 -5.45981824e-01 -1.05249420e-01 -1.12395012e+00 -4.76231873e-01 7.46587396e-01 -1.61230218e+00 1.20139635e+00 -5.80848873e-01 -1.90756929e+00 9.00580466e-01 4.70174253e-02 -2.92040795e-01 9.99530256e-01 -7.55294681e-01 1.24062397e-01 6.84774280e-01 1.18559144e-01 8.85023594e-01 9.16630208e-01 -9.04357672e-01 -5.44562995e-01 -1.04041326e+00 -7.40052238e-02 5.35559237e-01 -4.10420112e-02 -2.40072936e-01 -1.09128892e+00 -4.03985947e-01 4.16245878e-01 -1.28184628e+00 -4.21821862e-01 5.45775294e-01 -3.42841268e-01 3.39280143e-02 5.15091598e-01 -1.00740969e+00 4.77106988e-01 -1.76705098e+00 6.01854384e-01 5.68502307e-01 6.65248334e-02 -3.53618145e-01 3.49143445e-01 2.66607434e-01 2.32671171e-01 -8.02901804e-01 1.66280195e-01 -9.92017031e-01 8.85276496e-02 3.22352231e-01 5.91708601e-01 8.78247261e-01 -2.37419963e-01 7.31895864e-01 -7.82934427e-01 -4.63139534e-01 7.92863369e-01 6.23238385e-01 -7.96407402e-01 2.86456347e-01 -1.60593614e-01 5.86732328e-01 -3.46928537e-01 5.01817405e-01 3.60194027e-01 -2.06424072e-01 2.30167687e-01 -2.67417133e-01 9.72876325e-03 1.34846598e-01 -1.82766855e+00 2.73713160e+00 -3.88630301e-01 -4.80734669e-02 4.74776119e-01 -6.97902739e-01 6.87551916e-01 4.90900159e-01 3.33671182e-01 -2.42879719e-01 3.14817727e-01 6.88450709e-02 -8.13684940e-01 -3.90108943e-01 4.68789667e-01 2.64408916e-01 -2.90305287e-01 2.88525611e-01 2.17526466e-01 1.45364609e-02 -2.57052690e-01 1.61435366e-01 1.24594915e+00 9.51511860e-01 3.23814929e-01 1.36649851e-02 3.41280103e-01 -2.88235713e-02 2.26563498e-01 5.34627914e-01 3.88306635e-03 7.93047249e-01 2.26118509e-02 -3.63988340e-01 -9.10862505e-01 -1.10203171e+00 4.62790281e-01 1.01835847e+00 1.97545320e-01 -8.88714194e-01 -8.87487352e-01 -5.76771021e-01 4.45226431e-02 -5.75704779e-03 -6.10431790e-01 -3.18610966e-02 -5.47879338e-01 -1.15517162e-01 1.02624953e-01 8.00067723e-01 5.88605642e-01 -5.33421814e-01 -1.55555010e+00 1.62069082e-01 -2.96170950e-01 -1.36928117e+00 -2.11875856e-01 3.84991243e-02 -1.01770139e+00 -1.15076804e+00 -6.84955955e-01 -1.02439463e-01 6.56048179e-01 -1.01661142e-02 9.14846659e-01 -1.91477522e-01 -2.82417953e-01 9.97938573e-01 -2.98360527e-01 -1.77207828e-01 8.22651759e-03 -1.25081569e-01 2.18329206e-01 1.50560914e-02 4.63863835e-02 -4.80833381e-01 -8.56768608e-01 2.88472742e-01 -2.20656842e-01 1.39312789e-01 4.97826338e-01 2.55173564e-01 5.00055730e-01 -4.41920400e-01 -1.12573385e-01 -5.08817673e-01 -2.45988116e-01 -4.25361320e-02 -7.61155069e-01 -1.55824602e-01 -6.33314475e-02 5.30753024e-02 9.51961651e-02 1.37895793e-01 -9.76670384e-01 8.92861664e-01 -2.70013690e-01 -3.67215216e-01 -3.87114316e-01 8.63216743e-02 -3.85503143e-01 1.72406510e-01 6.39603972e-01 -4.05394316e-01 1.65297270e-01 -7.97086775e-01 5.42323232e-01 5.95428765e-01 9.94407475e-01 -4.07055736e-01 3.98455203e-01 9.29754257e-01 2.28309080e-01 -7.57552385e-01 -6.59536898e-01 -8.80312979e-01 -1.13777518e+00 -7.06362486e-01 1.12167549e+00 -1.37439275e+00 -1.01447499e+00 5.87201893e-01 -1.48896420e+00 -6.28726408e-02 -2.92376161e-01 6.92773521e-01 -7.20011055e-01 4.29607093e-01 -4.73104447e-01 -7.87946224e-01 -2.31309935e-01 -1.05783105e+00 1.94866931e+00 -7.48125240e-02 -6.15189970e-01 -9.09643948e-01 4.47778217e-02 6.20441496e-01 -2.75136709e-01 4.59355086e-01 -3.12387973e-01 -1.35591373e-01 -5.77148914e-01 -5.99737704e-01 2.31042415e-01 1.28269896e-01 -4.04984891e-01 -4.98273969e-01 -1.20486081e+00 -1.90507412e-01 -9.08567384e-02 -1.33717507e-01 7.44120955e-01 3.47768217e-01 4.38072801e-01 2.58550525e-01 -6.29598677e-01 3.91943634e-01 1.11932063e+00 -8.94702852e-01 3.08765560e-01 4.22779322e-01 6.85346544e-01 6.56567812e-01 6.27097249e-01 7.14134097e-01 8.57683420e-01 1.17530835e+00 8.41437638e-01 1.31549731e-01 -2.68277764e-01 -1.53860658e-01 7.27559865e-01 6.35137260e-01 -5.03486753e-01 1.22488737e-01 -7.46397138e-01 6.50558472e-02 -2.19435906e+00 -4.89336133e-01 -3.27899367e-01 2.46893096e+00 3.31389219e-01 4.21807766e-01 4.96169120e-01 1.79442734e-01 5.97762406e-01 -1.81586951e-01 -3.09414625e-01 1.50279433e-01 2.90750742e-01 1.35431826e-01 7.97331870e-01 6.02564275e-01 -1.05865538e+00 6.50832295e-01 5.70476675e+00 -7.91864563e-03 -5.24466395e-01 2.90267736e-01 -2.45027989e-01 -4.48046178e-01 3.03601652e-01 -4.87342067e-02 -1.02526224e+00 2.24757865e-01 8.71886253e-01 7.66383052e-01 1.32980332e-01 9.02839780e-01 1.35349631e-01 -6.09330118e-01 -1.37410247e+00 1.24451911e+00 2.78116018e-01 -9.04396892e-01 -6.57189131e-01 1.65647671e-01 2.52617896e-01 2.99643695e-01 -6.48280799e-01 -1.84632853e-01 8.39919075e-02 -2.87830114e-01 1.06156194e+00 9.05893207e-01 6.47681475e-01 -7.42439985e-01 6.74776077e-01 7.52107799e-01 -1.17703199e+00 9.56380442e-02 -1.82127822e-02 -4.44148540e-01 4.60002840e-01 8.37149203e-01 -8.79022002e-01 8.88921320e-01 9.62915659e-01 9.29982245e-01 -5.94390631e-01 6.48173332e-01 -7.29170561e-01 -1.06932178e-01 -1.03055346e+00 2.15134606e-01 -3.57409030e-01 1.30483434e-01 7.45268583e-01 1.05507302e+00 2.71020353e-01 -1.75400689e-01 5.16166568e-01 3.19198817e-01 1.60347372e-01 -3.59886110e-01 -3.10237467e-01 8.39598656e-01 8.12046230e-02 1.44056666e+00 -9.37454998e-01 -4.64232683e-01 -3.44384521e-01 1.59202862e+00 1.33543104e-01 -1.58901408e-01 -7.06654668e-01 -5.95240109e-02 4.32659686e-01 2.05550820e-01 4.70285654e-01 -5.39362013e-01 -1.08412735e-01 -1.39779699e+00 5.59598982e-01 -4.94067222e-01 6.33779407e-01 -9.92794871e-01 -7.64133155e-01 -1.88584123e-02 8.89626332e-03 -9.81510401e-01 -5.77295899e-01 -7.60999382e-01 -1.31787518e-02 5.99299788e-01 -1.00520062e+00 -1.18149972e+00 -7.31033564e-01 7.93617666e-01 2.70643204e-01 3.76145810e-01 1.00989377e+00 1.80818960e-01 -1.60347193e-01 9.27071348e-02 -6.86294019e-01 9.85161737e-02 8.81700039e-01 -1.27268624e+00 4.20997143e-01 8.07272255e-01 2.12124795e-01 2.00826883e-01 6.37138903e-01 -8.77473712e-01 -1.85460567e+00 -6.84093952e-01 7.05789447e-01 -9.26894367e-01 1.81583643e-01 -7.58655787e-01 -1.58824369e-01 9.55781460e-01 -1.31945133e-01 3.08549166e-01 2.98291236e-01 1.06039226e-01 -1.38243571e-01 -7.70679340e-02 -1.11658919e+00 1.72124937e-01 1.40813756e+00 -2.82799929e-01 -7.00131416e-01 3.97617936e-01 5.28179824e-01 -8.66234660e-01 -8.93450141e-01 7.79953524e-02 8.40699911e-01 -1.13833952e+00 1.13412249e+00 4.91020270e-02 -1.01421565e-01 -5.20169079e-01 -1.96924582e-01 -1.07294798e+00 1.48898121e-02 -5.12770593e-01 -2.97757804e-01 6.03207469e-01 -2.71135747e-01 -3.69094312e-01 1.17538738e+00 4.26437825e-01 2.89913937e-02 -1.48563460e-01 -1.26129389e+00 -4.26952720e-01 -1.03173923e+00 -1.00545073e+00 -1.74165778e-02 3.58458459e-02 8.27944372e-03 6.51793897e-01 -3.15676689e-01 6.11193419e-01 1.05181694e+00 -2.49995559e-01 1.23279262e+00 -1.27846611e+00 -7.13754594e-01 2.21578360e-01 -1.02774012e+00 -1.45807481e+00 -2.93783605e-01 -7.07213640e-01 -1.45581970e-02 -1.49559844e+00 -7.59957805e-02 3.38643402e-01 2.80077726e-01 3.45933765e-01 1.36697173e-01 6.35392547e-01 2.57716686e-01 -3.72590199e-02 -1.03471494e+00 1.87741145e-01 7.30914116e-01 4.49514419e-01 1.50748059e-01 6.99165016e-02 -4.68303747e-02 1.30663931e+00 2.50647962e-01 -4.62543845e-01 1.00408234e-01 -6.51236773e-01 6.77202642e-01 2.71087736e-01 1.22524345e+00 -1.44140863e+00 6.51711404e-01 6.03141308e-01 8.87253225e-01 -8.43296885e-01 9.95762289e-01 -1.23347878e+00 3.96369159e-01 7.33103037e-01 1.21134035e-01 8.94174874e-02 -2.34384418e-01 8.68542790e-01 3.42910469e-01 2.26407140e-01 4.21896726e-01 -5.19628108e-01 -7.02733099e-01 -1.69139134e-03 -6.52357787e-02 -3.32813621e-01 9.15056348e-01 -4.32490408e-01 4.50811356e-01 -4.26813006e-01 -1.37162459e+00 2.30758086e-01 5.98494291e-01 4.06773627e-01 5.87234735e-01 -1.28188455e+00 -5.55614889e-01 4.77866918e-01 5.52435629e-02 4.36728477e-01 1.60031110e-01 1.05014217e+00 -5.43394208e-01 2.94992954e-01 -1.55136034e-01 -1.17907798e+00 -1.42180574e+00 1.58378214e-01 3.19185644e-01 -5.51939867e-02 -8.69376898e-01 7.68042743e-01 -3.21207076e-01 -6.10470474e-01 3.67844939e-01 -4.58361238e-01 2.07449242e-01 1.19147010e-01 5.75007915e-01 7.96320438e-01 3.82343858e-01 -6.23534381e-01 -8.04441333e-01 9.70080674e-01 5.25852382e-01 -6.47329330e-01 1.31309760e+00 -4.54094499e-01 -1.55213745e-02 5.13828516e-01 1.15533602e+00 1.35269528e-03 -1.50502944e+00 -2.43461788e-01 -8.56984109e-02 -3.56664032e-01 -9.57437977e-02 -7.25021124e-01 -8.97231281e-01 6.83322251e-01 5.89889765e-01 -2.67460197e-01 9.13452983e-01 2.84832567e-01 6.63982570e-01 4.81037557e-01 8.83922815e-01 -1.53340399e+00 2.86927104e-01 2.24641040e-01 7.94304848e-01 -1.23692381e+00 4.41614002e-01 -5.85518122e-01 -2.37398341e-01 1.20653081e+00 2.38687173e-01 -4.55083281e-01 6.13193154e-01 3.48533005e-01 1.83563214e-02 -3.57313424e-01 -5.92697263e-01 -2.07898185e-01 1.21354125e-01 6.30565703e-01 -2.09363420e-02 7.78523013e-02 2.17784539e-01 4.74820435e-01 -3.12704831e-01 1.45470262e-01 1.61737710e-01 1.06208050e+00 -4.67549801e-01 -1.12000775e+00 -9.50883925e-01 -2.45235994e-01 -2.34710425e-01 5.82249105e-01 -3.07331204e-01 5.93745947e-01 3.40827674e-01 8.59000206e-01 1.21638961e-01 -3.31714660e-01 7.52408266e-01 9.16043967e-02 8.92747045e-01 -7.79155076e-01 -7.05973446e-01 5.10365188e-01 2.98258066e-01 -1.45794010e+00 -5.78741491e-01 -9.44465697e-01 -1.36323667e+00 1.22206159e-01 -1.54835314e-01 -2.55002826e-01 1.09620917e+00 9.01866436e-01 5.58288038e-01 5.29805303e-01 3.15250494e-02 -1.57154179e+00 -5.36513865e-01 -8.90708089e-01 -3.53282988e-01 2.29413092e-01 1.33427933e-01 -8.14881444e-01 7.27198347e-02 3.01602364e-01]
[7.087376117706299, -0.9466521739959717]
1910fc05-ef52-49e6-8a3a-17962e0caac9
an-algorithm-for-automatically-updating-a
2009.03193
null
https://arxiv.org/abs/2009.03193v2
https://arxiv.org/pdf/2009.03193v2.pdf
An Algorithm for Automatically Updating a Forsyth-Edwards Notation String Without an Array Board Representation
We present an algorithm that correctly updates the Forsyth-Edwards Notation (FEN) chessboard character string after any move is made without the need for an intermediary array representation of the board. In particular, this relates to software that have to do with chess, certain chess variants and possibly even similar board games with comparable position representation. Even when performance may be equal or inferior to using arrays, the algorithm still provides an accurate and viable alternative to accomplishing the same thing, or when there may be a need for additional or side processing in conjunction with arrays. Furthermore, the end result (i.e. an updated FEN string) is immediately ready for export to any other internal module or external program, unlike with an intermediary array which needs to be first converted into a FEN string for export purposes. The algorithm is especially useful when there are no existing array-based modules to represent a visual board as it can do without them entirely. We provide examples that demonstrate the correctness of the algorithm given a variety of positions involving castling, en passant and pawn promotion.
['Azlan Iqbal']
2020-09-02
null
null
null
null
['board-games']
['playing-games']
[ 4.20795381e-01 6.57189116e-02 4.07964200e-01 -5.54905720e-02 -5.38830519e-01 -1.09432399e+00 3.63441527e-01 5.01603186e-01 -4.58192229e-01 6.64902627e-01 -3.30187529e-01 -1.07597303e+00 -6.84734210e-02 -1.01376808e+00 -5.63667774e-01 -2.21775770e-01 -1.16213292e-01 4.24054921e-01 8.17620516e-01 -5.67962408e-01 4.80705231e-01 5.25556028e-01 -1.71024632e+00 2.04767928e-01 4.63725239e-01 8.80149662e-01 1.44016996e-01 1.17168581e+00 -3.03354412e-01 9.31239009e-01 -1.03337610e+00 -5.85390389e-01 5.73227584e-01 -4.02600527e-01 -6.23565376e-01 -3.13887924e-01 3.03489029e-01 -4.18499112e-01 1.75361112e-01 9.71508086e-01 1.15265645e-01 -2.26620436e-01 3.86288375e-01 -1.31494057e+00 2.22506836e-01 4.42245811e-01 -6.94488823e-01 -8.17603543e-02 6.35582209e-01 2.20441118e-01 5.95390260e-01 -3.12115312e-01 5.05046666e-01 8.65426123e-01 8.71463180e-01 -1.92571476e-01 -1.17707467e+00 -2.96573341e-01 -2.60452271e-01 -4.66669828e-01 -1.19586515e+00 -1.59363791e-01 2.10314229e-01 -4.21674937e-01 8.96144807e-01 1.08502793e+00 1.10814345e+00 6.59309007e-05 4.27619755e-01 8.79198015e-02 9.52721357e-01 -7.43077099e-01 2.22422108e-01 -7.59083370e-04 2.38180503e-01 6.22968853e-01 7.20547318e-01 -4.01009411e-01 -1.02936223e-01 -3.90010297e-01 1.11431897e+00 -4.29447353e-01 3.94219048e-02 -6.08231366e-01 -1.19749510e+00 2.72592574e-01 -4.21273038e-02 1.12730615e-01 -1.59425586e-01 4.59544361e-01 5.44340014e-01 4.38933641e-01 -1.50547966e-01 6.01260722e-01 -2.01341227e-01 -5.66257000e-01 -1.07393324e+00 6.44096076e-01 1.11764777e+00 9.15440381e-01 7.88280010e-01 1.10817902e-01 2.24812269e-01 3.35451931e-01 9.45667997e-02 -7.17501566e-02 3.13324183e-02 -8.48095596e-01 4.23644841e-01 9.00644958e-01 5.04652262e-01 -1.13030076e+00 -4.31405604e-01 -2.92113423e-01 -3.04472476e-01 1.23790836e+00 8.58279288e-01 -1.09866992e-01 -4.41370785e-01 1.16676724e+00 3.68178844e-01 -4.30713534e-01 -1.02078527e-01 5.03494501e-01 7.22710967e-01 7.04223752e-01 -2.69667208e-01 1.77543715e-01 1.55241656e+00 -6.50808930e-01 -4.05782521e-01 -2.55566299e-01 6.45302474e-01 -1.19466937e+00 1.09136701e+00 6.15252793e-01 -1.59197319e+00 -4.16527867e-01 -1.62758243e+00 -1.74462587e-01 -4.18971539e-01 -2.06221186e-04 7.09179461e-01 1.06780410e+00 -1.20825613e+00 4.08891648e-01 -7.68437922e-01 -1.28413841e-01 -3.20698857e-01 5.70590615e-01 -2.19556794e-01 4.87327605e-01 -6.94300115e-01 1.01894379e+00 5.00175953e-01 2.30662376e-01 -1.09296806e-01 -6.30737066e-01 -7.70870566e-01 1.96313486e-01 2.24523738e-01 -5.05386412e-01 1.42380428e+00 -1.03797638e+00 -1.27786803e+00 8.09614837e-01 4.71197307e-01 -3.50489289e-01 8.39721024e-01 1.84757680e-01 -2.46147051e-01 -2.32982337e-01 7.39933848e-02 3.86438876e-01 3.95393252e-01 -9.63136435e-01 -6.41430140e-01 -1.90783843e-01 8.84749472e-01 2.05289274e-01 -4.33918461e-02 2.94379950e-01 -5.65492690e-01 -5.60120821e-01 6.24157377e-02 -7.15559721e-01 -2.60530382e-01 6.01182356e-02 -2.36729771e-01 2.43105814e-01 3.67415130e-01 -7.03428924e-01 1.75681794e+00 -2.20022488e+00 -2.07014307e-01 6.02161229e-01 1.82260960e-01 9.69091579e-02 1.52392551e-01 9.08434451e-01 -3.11563492e-01 -7.50591010e-02 -8.31523389e-02 5.23584366e-01 2.41807774e-01 -1.30902976e-01 8.86086747e-02 5.59542418e-01 -1.57879800e-01 1.66760102e-01 -5.28165162e-01 -2.54167289e-01 -6.16128594e-02 -1.57478768e-02 -6.32377148e-01 -1.77803040e-01 -1.73003003e-01 -2.77904689e-01 -5.87755479e-02 3.13393652e-01 8.30154300e-01 1.39619363e-02 5.23550212e-01 3.61836776e-02 -8.58335614e-01 3.85359049e-01 -1.88226545e+00 1.42568362e+00 -1.27207190e-01 5.59981883e-01 4.03619081e-01 -4.67502952e-01 1.03932226e+00 3.72882448e-02 3.17318030e-02 -4.84769851e-01 3.73829678e-02 3.69150043e-01 3.21376145e-01 -5.00559956e-02 1.39976609e+00 6.39043003e-02 -4.45050180e-01 5.58140814e-01 -8.45272124e-01 -4.75221425e-01 6.53738379e-01 4.86480355e-01 1.31645226e+00 4.10449922e-01 4.87346053e-01 -3.79922956e-01 5.37043095e-01 5.54616630e-01 2.48772368e-01 5.96923292e-01 5.68401575e-01 4.42517132e-01 1.05852532e+00 -5.05289257e-01 -1.31866848e+00 -9.06328082e-01 -9.38432515e-02 8.57447922e-01 1.94062084e-01 -1.08291352e+00 -7.98293412e-01 1.42289951e-01 -9.68206748e-02 5.97954929e-01 -3.88472021e-01 3.64604384e-01 -5.77287912e-01 -3.19468528e-01 5.98151922e-01 5.36667287e-01 3.23693991e-01 -6.78221822e-01 -1.48399031e+00 5.61334848e-01 5.84138095e-01 -2.72308320e-01 -2.66540885e-01 3.67168814e-01 -5.89822173e-01 -1.10475671e+00 -3.99287403e-01 -7.49697745e-01 7.85127640e-01 1.64750367e-01 9.82727826e-01 4.90312815e-01 -3.45422268e-01 4.37794507e-01 -1.16165243e-01 -4.60341841e-01 -6.08646750e-01 -1.05578229e-01 -6.10796630e-01 -7.40542471e-01 -8.84749517e-02 -3.74019474e-01 -4.40170407e-01 1.81938276e-01 -1.23738587e+00 4.00682420e-01 1.03259690e-01 5.72491884e-01 2.71834850e-01 1.80491462e-01 1.53510258e-01 -9.11385596e-01 8.20879996e-01 -1.47357453e-02 -1.28997946e+00 5.56119122e-02 6.37310445e-02 -2.02508062e-01 6.97218537e-01 4.96778190e-02 -7.29798496e-01 1.19966626e-01 -9.14577171e-02 4.62469190e-01 2.25655302e-01 7.89178312e-01 -1.84877381e-01 -4.89818007e-02 5.38031995e-01 -2.39649251e-01 2.60317445e-01 -3.21414769e-01 1.17284365e-01 6.07072592e-01 7.69619584e-01 -5.80571949e-01 7.65486360e-01 1.58588916e-01 1.20912604e-02 -4.81177539e-01 3.95969898e-01 -9.46015120e-02 -4.34820563e-01 -2.94849157e-01 4.99250799e-01 -5.34623563e-01 -9.65955079e-01 5.27712166e-01 -1.30972767e+00 -3.94648612e-01 -3.11911434e-01 -1.29363284e-01 -4.68373477e-01 4.25604284e-01 -3.83893371e-01 -6.17279351e-01 1.69394180e-01 -1.19690752e+00 5.53192854e-01 2.44338274e-01 -7.48227179e-01 -5.75898290e-01 -7.49632809e-03 -4.48899828e-02 2.96158880e-01 4.82834429e-01 1.22352779e+00 -2.88693279e-01 -6.03560030e-01 -6.70839906e-01 -4.32450324e-03 -2.56827101e-02 -1.69429392e-01 5.56573749e-01 -2.19396934e-01 -2.86451966e-01 -2.92017132e-01 1.10096335e-01 -2.44441093e-04 -1.72633275e-01 8.23712587e-01 -1.08332455e-01 -1.60035834e-01 3.04106355e-01 1.58849657e+00 8.13202739e-01 1.15383029e+00 1.06187487e+00 1.40729517e-01 4.78103489e-01 5.94346583e-01 6.56748712e-01 3.61529350e-01 7.41198003e-01 4.11976844e-01 -2.45350018e-01 5.51417172e-02 -6.68446869e-02 3.51461112e-01 4.78026092e-01 2.28170976e-02 -1.10256612e-01 -1.08033395e+00 1.38241336e-01 -1.59295487e+00 -9.52270091e-01 -7.87940025e-01 2.60640621e+00 6.86783195e-01 4.39211786e-01 3.51552129e-01 5.74208379e-01 5.87018728e-01 8.76155347e-02 -3.10143054e-01 -1.16047359e+00 1.76304474e-01 4.33968663e-01 7.19720721e-01 4.98154163e-01 -7.18321621e-01 4.46261346e-01 6.83421230e+00 6.13134861e-01 -8.33055377e-01 -3.35128307e-01 4.52102393e-01 4.78012487e-02 -5.01443148e-01 3.35512161e-01 -5.65311849e-01 3.36499453e-01 5.64130485e-01 -4.61410373e-01 2.69063622e-01 6.10414803e-01 -9.70391557e-02 -9.76457357e-01 -1.01955390e+00 7.86650360e-01 -1.81767449e-01 -1.41656578e+00 -3.38442951e-01 1.17494330e-01 3.59349787e-01 -9.19297218e-01 -7.51905218e-02 1.18356862e-03 5.79935014e-01 -1.02552748e+00 1.32932341e+00 3.48361194e-01 9.06263113e-01 -9.21070099e-01 3.71065736e-01 1.03456281e-01 -1.19613802e+00 2.04524353e-01 -2.93181717e-01 -6.27509713e-01 -1.24121636e-01 3.41550298e-02 -6.62111998e-01 6.28168404e-01 4.27536488e-01 -2.47061282e-01 -7.08063543e-01 1.53166020e+00 1.06518857e-01 1.74395218e-01 -5.39743781e-01 -1.50213674e-01 3.93954277e-01 -3.00667554e-01 3.60568613e-01 1.01260984e+00 5.98531067e-01 1.02349617e-01 -1.27681345e-01 2.60711223e-01 4.92319316e-01 3.25764388e-01 -4.93861020e-01 2.21153542e-01 4.49067742e-01 1.12232590e+00 -1.25824773e+00 -4.70300585e-01 -7.00316608e-01 7.77411640e-01 -1.08186223e-01 1.22115105e-01 -1.11043763e+00 -9.92020190e-01 4.12691623e-01 6.21729314e-01 3.66615057e-01 -4.62667584e-01 -4.90479082e-01 -5.44877052e-01 9.35121849e-02 -1.42829537e+00 3.21732759e-01 -1.10891867e+00 -5.02639234e-01 5.13759792e-01 -1.52568193e-02 -1.31605422e+00 -4.00550187e-01 -6.20647907e-01 -6.60394669e-01 1.02544522e+00 -6.23152614e-01 -6.34057164e-01 -1.67937741e-01 2.42554054e-01 -1.13598429e-01 2.01116353e-02 8.03941071e-01 2.66906738e-01 -2.73165554e-01 4.82140899e-01 1.66049004e-01 -5.70812300e-02 4.49215323e-01 -1.34762597e+00 3.49355340e-01 9.52859282e-01 -2.92582065e-01 7.96997368e-01 8.78294766e-01 -6.70409501e-01 -1.67197430e+00 -3.60541493e-01 3.32457244e-01 -2.11205781e-01 7.83902645e-01 -4.05934393e-01 -5.16447246e-01 8.85813594e-01 4.97309446e-01 -6.48509145e-01 5.74207246e-01 -1.38022333e-01 -5.76039329e-02 -3.57483178e-01 -8.58510494e-01 9.12265003e-01 6.57323539e-01 -1.12038486e-01 -4.24780011e-01 -4.46880609e-02 6.42411262e-02 -1.04176533e+00 -6.08020782e-01 -7.56980702e-02 9.07695532e-01 -1.26556373e+00 8.35205436e-01 -4.00477707e-01 3.99325639e-01 -1.03551459e+00 -1.23560406e-01 -8.96427929e-01 -3.29121321e-01 -9.26875353e-01 9.11718667e-01 1.08988369e+00 6.07214510e-01 -6.47917867e-01 7.95826197e-01 1.01836145e+00 -3.08787286e-01 -2.98292607e-01 -7.70126820e-01 -7.79183626e-01 -6.96198791e-02 -5.12856722e-01 1.04541194e+00 7.13495553e-01 4.73336339e-01 4.42282967e-02 -1.68865398e-01 3.11261360e-02 8.36083218e-02 1.94933400e-01 1.52975368e+00 -9.41903710e-01 -6.59883142e-01 -5.57537913e-01 -7.88886249e-01 -1.07697380e+00 -6.11615956e-01 -8.48151147e-01 -1.52705118e-01 -1.78155017e+00 -3.23118418e-01 -6.99225247e-01 3.63915771e-01 4.55180883e-01 1.68926433e-01 3.87266695e-01 4.66286659e-01 -1.72754098e-02 -2.59512991e-01 -3.75015885e-01 9.31209207e-01 3.51581834e-02 -2.42430910e-01 -1.39599636e-01 -8.55087221e-01 9.07196879e-01 5.92698097e-01 -3.81110400e-01 -3.12628746e-01 -4.27608341e-01 9.11246240e-01 4.91294175e-01 4.54460382e-01 -1.27212501e+00 5.62353671e-01 1.14484116e-01 4.99863029e-01 -8.04600418e-01 2.71381497e-01 -1.02785957e+00 1.13567686e+00 6.50094509e-01 6.70380592e-02 1.04955602e+00 7.03501999e-01 1.07762376e-02 2.95609813e-02 -7.20610857e-01 4.40618426e-01 -1.70883045e-01 -6.61216855e-01 -5.24438083e-01 -1.03442097e+00 -4.16229188e-01 1.34922254e+00 -9.59633052e-01 -6.40832722e-01 -3.70598495e-01 -6.25821471e-01 2.35926136e-02 1.10517287e+00 -1.36538461e-01 7.16225579e-02 -1.06884778e+00 -2.87919879e-01 3.44826192e-01 -1.54246122e-01 -1.18030399e-01 2.78918017e-02 5.18726528e-01 -1.70704496e+00 1.03622086e-01 -6.85957670e-01 -2.41653994e-01 -1.54290199e+00 3.25611562e-01 1.39266014e-01 -2.92486101e-01 -4.90640223e-01 4.48109061e-01 1.34420972e-02 -2.70092324e-03 -2.04785056e-02 -5.80554605e-01 1.92771837e-01 1.31389081e-01 6.34854794e-01 3.95707846e-01 3.88527751e-01 -2.34361693e-01 -4.26473826e-01 2.27013990e-01 1.21799007e-01 -3.71235669e-01 1.38338017e+00 2.45987877e-01 -6.34569407e-01 2.97235936e-01 3.96282256e-01 7.09719300e-01 -1.04497242e+00 6.79642916e-01 -1.64737090e-01 -7.59544969e-01 -5.21489680e-01 -8.41954291e-01 -6.43187284e-01 4.84715462e-01 1.57848567e-01 7.01523185e-01 9.68237042e-01 -3.74893785e-01 2.64279991e-01 1.69120803e-01 6.67603791e-01 -7.85834670e-01 -4.15074140e-01 2.98616976e-01 9.15158272e-01 -1.49414957e-01 3.88641775e-01 -4.66568202e-01 -2.49731615e-01 1.57428968e+00 5.68053722e-01 -1.78561822e-01 1.44299507e-01 1.15813208e+00 8.31473395e-02 -9.65858176e-02 -5.81656992e-01 1.52959988e-01 -3.54949057e-01 5.55579603e-01 5.81527889e-01 -1.97011605e-02 -8.63630056e-01 5.20232379e-01 -8.01049650e-01 -8.85237232e-02 1.09019101e+00 1.36776912e+00 -5.39969265e-01 -1.55831313e+00 -1.00668883e+00 4.81703401e-01 -3.84068221e-01 -2.64383823e-01 -2.91370839e-01 1.47420263e+00 1.16295680e-01 5.66567361e-01 4.45064068e-01 -1.09891593e-01 4.98815864e-01 -9.10099596e-02 7.30815887e-01 -5.88887095e-01 -1.25283349e+00 1.63226604e-01 6.27191246e-01 -1.80476323e-01 9.13774595e-02 -7.26840913e-01 -1.26234782e+00 -7.90845335e-01 2.49486081e-02 2.16604829e-01 5.01409233e-01 1.49408281e-01 9.36745182e-02 5.04814446e-01 -8.94798040e-02 -5.51048577e-01 -2.14871272e-01 -2.04073116e-01 -7.93561101e-01 -1.56371728e-01 -1.13013856e-01 -3.14781368e-01 1.60019889e-01 3.62873599e-02]
[8.055740356445312, 7.120926856994629]
3221d66c-2495-4dcb-bc24-4c8f939d0b75
learning-residual-flow-as-dynamic-motion-from
1909.06999
null
https://arxiv.org/abs/1909.06999v1
https://arxiv.org/pdf/1909.06999v1.pdf
Learning Residual Flow as Dynamic Motion from Stereo Videos
We present a method for decomposing the 3D scene flow observed from a moving stereo rig into stationary scene elements and dynamic object motion. Our unsupervised learning framework jointly reasons about the camera motion, optical flow, and 3D motion of moving objects. Three cooperating networks predict stereo matching, camera motion, and residual flow, which represents the flow component due to object motion and not from camera motion. Based on rigid projective geometry, the estimated stereo depth is used to guide the camera motion estimation, and the depth and camera motion are used to guide the residual flow estimation. We also explicitly estimate the 3D scene flow of dynamic objects based on the residual flow and scene depth. Experiments on the KITTI dataset demonstrate the effectiveness of our approach and show that our method outperforms other state-of-the-art algorithms on the optical flow and visual odometry tasks.
['In So Kweon', 'Stephen Lin', 'Seokju Lee', 'Sunghoon Im']
2019-09-16
null
null
null
null
['stereo-matching', 'depth-and-camera-motion']
['computer-vision', 'computer-vision']
[-1.70980126e-01 -2.70062834e-01 -2.95661628e-01 -2.29191944e-01 1.28016084e-01 -6.69273257e-01 6.13361061e-01 -7.00433373e-01 -2.96594292e-01 3.57355505e-01 4.08902913e-01 1.76474769e-02 1.26869991e-01 -4.75288093e-01 -5.79528809e-01 -5.24500251e-01 7.14158192e-02 6.05273545e-01 3.75547945e-01 4.39638823e-01 5.63597739e-01 7.78813243e-01 -1.30762148e+00 -1.87445626e-01 3.45473945e-01 5.82551897e-01 4.21664894e-01 1.25238955e+00 -1.51291132e-01 1.66216087e+00 1.52395843e-02 9.46457684e-02 5.95683873e-01 -3.05495441e-01 -1.03881216e+00 6.45012379e-01 1.00160336e+00 -1.13849437e+00 -1.16184878e+00 8.58965039e-01 1.71906948e-01 5.05243480e-01 5.51115215e-01 -1.13789487e+00 -2.22238481e-01 -8.81889388e-02 -4.57661003e-01 4.27339137e-01 6.26800358e-01 4.36409295e-01 8.01046908e-01 -8.34120750e-01 1.24532461e+00 1.44449508e+00 2.20548779e-01 6.08116090e-01 -1.21118283e+00 -2.43344396e-01 1.44959971e-01 2.99067974e-01 -8.70694041e-01 -5.83547056e-01 1.08436048e+00 -1.06047487e+00 8.40557635e-01 -2.70087987e-01 8.96909475e-01 6.28587782e-01 2.39904940e-01 8.48711312e-01 3.48472685e-01 -2.04501733e-01 5.21017909e-02 -1.99656576e-01 -1.62586980e-02 1.00954974e+00 1.96681805e-02 4.44576949e-01 -6.10508978e-01 9.04877260e-02 1.45115340e+00 1.23301566e-01 -5.58191657e-01 -8.93411815e-01 -1.29659581e+00 5.42173982e-01 2.65107691e-01 -3.25799614e-01 -2.07226798e-01 3.68341118e-01 4.43018340e-02 -6.94618970e-02 4.15024519e-01 1.54347435e-01 -2.53164023e-01 -2.78455973e-01 -7.71000743e-01 3.23719531e-01 1.10095739e+00 1.09306824e+00 1.14570069e+00 3.22576225e-01 2.69364476e-01 3.92802477e-01 6.16818488e-01 7.48117447e-01 1.74063623e-01 -2.11967301e+00 6.33960783e-01 4.93057609e-01 4.25351441e-01 -1.16986477e+00 -2.84685135e-01 2.36507788e-01 -5.90864658e-01 3.37388813e-01 8.21106553e-01 -1.37802660e-01 -8.24970603e-01 1.61846375e+00 4.40152079e-01 6.66145265e-01 -1.46048158e-01 1.31485748e+00 7.03590810e-01 7.20768631e-01 -6.15795255e-01 -1.99123368e-01 6.14855170e-01 -1.09687054e+00 -5.55567861e-01 -5.44390023e-01 5.85348129e-01 -8.81694198e-01 2.48067796e-01 1.27333794e-02 -1.44648147e+00 -8.63263190e-01 -8.07579875e-01 -3.28950375e-01 4.17001635e-01 -1.31530523e-01 5.05887091e-01 8.54412168e-02 -1.17464900e+00 5.19536972e-01 -1.16517520e+00 -2.80468047e-01 1.10252656e-01 2.10501343e-01 -3.94151002e-01 -3.20600182e-01 -5.59601068e-01 9.79276955e-01 2.46134847e-01 3.00113171e-01 -1.18333435e+00 -7.06433594e-01 -1.10762155e+00 -1.79473385e-01 1.65225923e-01 -1.28014660e+00 1.21129405e+00 -6.51490211e-01 -1.71609282e+00 9.88104403e-01 -6.54045582e-01 -2.32512504e-01 7.71256924e-01 -3.75379056e-01 4.35424984e-01 5.54889321e-01 -1.25101497e-02 8.02982330e-01 5.19385517e-01 -1.05614185e+00 -8.45868587e-01 -3.05002928e-01 2.10226581e-01 7.12697625e-01 3.10604990e-01 -4.74043399e-01 -7.32713938e-01 9.32257995e-02 4.11637157e-01 -1.23846531e+00 -2.90432364e-01 3.27489167e-01 -3.50681692e-01 2.49390990e-01 9.67458367e-01 -5.30401170e-01 7.24007905e-01 -1.93500459e+00 5.31470835e-01 -2.03556299e-01 3.20646107e-01 -5.51146902e-02 -1.06799742e-03 -1.88759312e-01 8.88991579e-02 -4.83384818e-01 7.19832703e-02 -4.56783146e-01 -4.87924218e-01 3.04595500e-01 -3.47131461e-01 9.08483148e-01 -1.88197121e-02 9.14997280e-01 -1.20614922e+00 -4.02632535e-01 8.46906006e-01 3.87705624e-01 -7.30452538e-01 5.71545780e-01 -3.14699635e-02 8.83588850e-01 -4.23175961e-01 3.19063187e-01 7.51916170e-01 -2.92964280e-01 7.47960508e-02 -2.02058345e-01 -2.41328076e-01 4.92610067e-01 -1.51032495e+00 1.87982607e+00 -1.41871169e-01 1.09358609e+00 9.69551802e-02 -5.78516364e-01 6.05665267e-01 2.80793998e-02 7.78472900e-01 -3.86847794e-01 3.59485224e-02 -8.87887552e-02 -1.43350184e-01 -6.73654258e-01 6.33663535e-01 2.04030186e-01 5.28924108e-01 6.54217124e-01 3.02435815e-01 -5.72667003e-01 2.69707531e-01 4.34008390e-01 9.80799794e-01 7.20793784e-01 4.53117751e-02 -1.61247313e-01 7.39838779e-01 4.75084372e-02 7.24411607e-01 6.12835407e-01 -5.15503883e-01 5.98688900e-01 2.17985973e-01 -8.71797264e-01 -1.38470960e+00 -1.16692054e+00 3.00803393e-01 3.93476278e-01 8.67915511e-01 -1.00430265e-01 -3.43742758e-01 -4.00333017e-01 4.60141040e-02 1.09832942e-01 -3.27772737e-01 8.98772776e-02 -9.67266023e-01 -8.44037905e-02 1.92917120e-02 4.86046463e-01 5.30304492e-01 -8.90499830e-01 -7.19068944e-01 2.88664848e-01 -7.02527642e-01 -1.59245062e+00 -8.89785707e-01 -3.85246009e-01 -1.21341860e+00 -1.27138031e+00 -4.91454959e-01 -7.00689316e-01 6.46707356e-01 9.66689885e-01 1.07524872e+00 -8.04305524e-02 -2.40258828e-01 5.96972704e-01 3.91883641e-01 2.20978022e-01 -3.67809325e-01 -2.81381607e-01 1.91791058e-02 1.48738902e-02 5.74318953e-02 -5.71363866e-01 -7.98972070e-01 3.67315650e-01 -6.11927569e-01 2.12072581e-01 -2.81624347e-01 4.83628243e-01 2.43432507e-01 -3.77319247e-01 -6.79372549e-01 -5.95750153e-01 -4.18450147e-01 -1.83466405e-01 -1.03181541e+00 -3.11619520e-01 -5.92496134e-02 1.63673759e-01 1.44520268e-01 -4.06250626e-01 -1.53117728e+00 5.66430211e-01 3.30185384e-01 -8.70014548e-01 -1.47532210e-01 -1.90059707e-01 8.43419209e-02 -1.23573609e-01 3.98536116e-01 1.54217765e-01 1.13616280e-01 -1.18758507e-01 5.48730016e-01 1.74302444e-01 8.50901186e-01 -2.76960105e-01 9.68246877e-01 1.34563172e+00 2.38254443e-01 -7.59814024e-01 -8.96292448e-01 -1.03063405e+00 -1.07676792e+00 -6.45626545e-01 1.11656654e+00 -1.22961330e+00 -1.03603470e+00 8.56039524e-01 -1.68774867e+00 -3.92975569e-01 -2.63040930e-01 1.00388432e+00 -9.44950879e-01 5.99035561e-01 -9.80990529e-01 -7.82779574e-01 5.41351400e-02 -1.26562667e+00 1.06529760e+00 2.64802873e-01 -1.06014259e-01 -1.47108328e+00 3.13280135e-01 3.69612187e-01 -1.69281319e-01 1.19212925e-01 4.01917309e-01 4.33274597e-01 -1.54369342e+00 1.47650942e-01 -1.93423629e-01 2.53520668e-01 2.10943848e-01 3.50681454e-01 -8.84591877e-01 -1.28501832e-01 2.60813087e-01 1.73247252e-02 7.95105338e-01 8.79908025e-01 4.19332087e-01 -1.02622956e-01 -1.06311314e-01 1.19689035e+00 1.51391077e+00 2.15335339e-01 4.60785568e-01 2.99072474e-01 1.34354568e+00 9.11528468e-01 4.22666878e-01 3.61583710e-01 5.42882383e-01 4.51651335e-01 6.95384800e-01 1.58191323e-01 -2.38844708e-01 -3.55735391e-01 5.40499747e-01 8.80899668e-01 -3.39304745e-01 -1.38689190e-01 -8.58730793e-01 6.22089624e-01 -1.91863132e+00 -1.22057235e+00 -4.99959826e-01 2.06442332e+00 2.29534611e-01 4.28821929e-02 -1.29594997e-01 -2.63567150e-01 7.23347664e-01 4.71458912e-01 -7.53198147e-01 4.01547179e-03 -1.32269263e-01 -5.04071712e-01 6.06652558e-01 1.12796926e+00 -8.55662465e-01 1.09958041e+00 6.52034855e+00 -2.71056533e-01 -1.07003200e+00 -2.25007474e-01 1.20266311e-01 -2.09979266e-01 -2.09356844e-01 3.67776781e-01 -9.48970258e-01 2.18021944e-01 2.86487430e-01 -1.24484390e-01 5.13889730e-01 7.25557387e-01 4.59368944e-01 -2.05442041e-01 -1.35750711e+00 9.84503686e-01 1.25637963e-01 -1.58888125e+00 1.78367384e-02 3.09533238e-01 1.03634584e+00 4.40205038e-01 -2.57072538e-01 -4.19096231e-01 6.89298034e-01 -1.70226276e-01 7.97516465e-01 5.97613394e-01 3.43958914e-01 -2.36748189e-01 4.57218409e-01 5.49710393e-01 -1.18950891e+00 -1.06228860e-02 -3.20681334e-01 -3.33928972e-01 7.32918859e-01 4.94125575e-01 -6.49347246e-01 2.65525222e-01 5.65553367e-01 1.49781334e+00 -7.02929050e-02 9.07477558e-01 -4.20183390e-01 1.56271130e-01 -2.33202770e-01 6.37200177e-01 3.11796635e-01 -5.88648498e-01 1.08526802e+00 9.15158868e-01 3.89047898e-02 2.12062299e-02 2.50692815e-01 9.78435457e-01 3.25954929e-02 -5.40211856e-01 -7.86032557e-01 4.36739266e-01 1.85493916e-01 1.04680157e+00 -4.88850802e-01 -4.87900883e-01 -4.53199089e-01 9.98382926e-01 1.80558532e-01 7.73285806e-01 -3.97175997e-01 2.51284599e-01 1.05995715e+00 1.39460275e-02 1.04973085e-01 -6.29341483e-01 1.59332678e-02 -1.74805236e+00 1.75637063e-02 6.67128041e-02 1.22112796e-01 -1.22116947e+00 -1.00610304e+00 -1.96132995e-02 -2.25587025e-01 -1.41280770e+00 -7.91677833e-01 -5.85482776e-01 -5.07890344e-01 8.81264210e-01 -1.69917822e+00 -5.51967084e-01 -7.15080857e-01 6.90431595e-01 6.35816216e-01 8.61055031e-02 1.08403377e-01 3.43753546e-02 -1.41136184e-01 -5.25049567e-01 -3.47952321e-02 3.56038004e-01 7.03177273e-01 -1.09010351e+00 5.89072168e-01 1.16565263e+00 8.74728039e-02 3.03545624e-01 4.45259720e-01 -4.63063687e-01 -1.55747294e+00 -9.47809279e-01 1.03235257e+00 -9.11867499e-01 6.54922545e-01 -3.21883053e-01 -7.74626255e-01 9.03975308e-01 -2.22430676e-01 3.80221367e-01 -4.91022095e-02 -5.57242692e-01 -2.01216236e-01 -1.01486064e-01 -6.00842357e-01 4.05523032e-01 1.31536233e+00 -8.76314819e-01 -5.75162590e-01 -3.43321338e-02 6.11646414e-01 -9.11315203e-01 -3.93238127e-01 1.02821946e-01 8.31991196e-01 -1.30183542e+00 1.14335370e+00 -5.04366755e-01 7.36588955e-01 -4.33752328e-01 7.75556564e-02 -1.01724911e+00 -4.74111468e-01 -6.90407753e-01 -4.23204809e-01 7.67620027e-01 -1.87957317e-01 -2.56377786e-01 1.24765468e+00 7.87679255e-01 4.64419276e-02 1.73451573e-01 -7.08700657e-01 -4.25811946e-01 -3.92902344e-01 -4.37312752e-01 -6.93429708e-02 1.05840194e+00 -4.07806963e-01 3.13595623e-01 -5.64498365e-01 1.82159781e-01 1.03592241e+00 2.16892228e-01 1.24542046e+00 -1.19138205e+00 -2.90466458e-01 -3.16539943e-01 -7.43719399e-01 -2.00583291e+00 5.80101073e-01 -5.67688644e-01 3.96663427e-01 -1.24692404e+00 2.60756314e-01 2.39043579e-01 3.36601228e-01 -3.32270473e-01 -1.09804653e-01 2.19649449e-02 5.44422030e-01 6.83951497e-01 -4.13411051e-01 2.69573182e-01 1.67851651e+00 -1.21825943e-02 -5.96320868e-01 -2.22392585e-02 1.40116483e-01 1.15425944e+00 2.14757651e-01 -4.06301022e-01 -4.96838927e-01 -9.51894403e-01 -1.10415980e-01 6.22155905e-01 5.22108853e-01 -6.41999185e-01 6.02182984e-01 -3.67055327e-01 4.86629188e-01 -7.06002891e-01 4.09659773e-01 -7.52271175e-01 3.88945602e-02 5.74154615e-01 -2.89365917e-01 -1.08900063e-01 -1.60097569e-01 6.60292268e-01 -8.89810473e-02 7.46682435e-02 9.00212348e-01 -3.32911253e-01 -8.73086452e-01 8.08773696e-01 -4.80139196e-01 1.24631494e-01 6.03240252e-01 -5.32748699e-01 -3.09934616e-01 -6.13582373e-01 -6.54409409e-01 2.53872842e-01 5.53363562e-01 5.87465644e-01 6.86371565e-01 -1.21038806e+00 -6.02091372e-01 5.43579102e-01 -1.05866976e-01 4.35235381e-01 2.70597994e-01 6.39348328e-01 -1.06746531e+00 6.01128101e-01 -2.06127152e-01 -1.20305872e+00 -1.06298506e+00 2.03761175e-01 6.34912729e-01 -2.46354286e-02 -7.15340495e-01 7.00631022e-01 7.47019053e-01 -4.50615495e-01 2.01975733e-01 -3.29360902e-01 7.36524239e-02 -3.53486091e-01 5.01155496e-01 9.86355305e-01 -5.63974798e-01 -9.49329197e-01 -2.34987333e-01 1.04621971e+00 2.03528807e-01 -2.74029553e-01 9.83335614e-01 -8.25864494e-01 -1.31591052e-01 6.37713730e-01 1.42419553e+00 -3.90591651e-01 -2.05147982e+00 -3.90223414e-01 -2.52574295e-01 -8.02936196e-01 9.14498121e-02 9.32033733e-02 -1.24808192e+00 1.05263233e+00 1.51986808e-01 -2.29287580e-01 7.82351792e-01 -1.87932197e-02 6.19393647e-01 3.45451564e-01 4.58715260e-01 -8.08854461e-01 1.26921415e-01 9.12898600e-01 1.97781444e-01 -1.19486010e+00 -8.04577768e-03 -5.92610359e-01 -3.18487406e-01 1.31574893e+00 8.12973440e-01 -5.33907592e-01 5.69046736e-01 1.92728847e-01 2.14716911e-01 1.61234602e-01 -8.37260783e-01 -1.48704171e-01 2.63452381e-01 5.80902100e-01 9.70722809e-02 -5.21692634e-01 5.43757021e-01 -8.56796026e-01 1.07384034e-01 2.16714978e-01 8.93590510e-01 7.58803904e-01 -3.63642663e-01 -6.21755242e-01 -2.09304675e-01 -1.46952301e-01 -2.01848313e-01 2.40463346e-01 -3.53927821e-01 7.27946401e-01 -6.21263571e-02 6.94512010e-01 5.53756535e-01 -7.74326026e-02 3.74807954e-01 -2.69600004e-01 8.42292726e-01 -4.56647754e-01 -8.19361582e-02 1.79466009e-01 -2.04228744e-01 -1.16295040e+00 -1.04884958e+00 -6.11767828e-01 -1.23862946e+00 -5.28340697e-01 -9.33178812e-02 -2.08616287e-01 5.21115363e-01 9.70719874e-01 2.35908050e-02 3.95661183e-02 8.93356562e-01 -1.38725734e+00 -7.97991902e-02 -5.63113928e-01 -6.26086652e-01 7.05456674e-01 9.76229787e-01 -4.82713431e-01 -9.74661052e-01 7.54116952e-01]
[8.61208438873291, -2.027155637741089]
8cf84fe5-87e3-484a-8299-8ac70bd6a9f4
visual-place-recognition-a-tutorial
2303.03281
null
https://arxiv.org/abs/2303.03281v1
https://arxiv.org/pdf/2303.03281v1.pdf
Visual Place Recognition: A Tutorial
Localization is an essential capability for mobile robots. A rapidly growing field of research in this area is Visual Place Recognition (VPR), which is the ability to recognize previously seen places in the world based solely on images. This present work is the first tutorial paper on visual place recognition. It unifies the terminology of VPR and complements prior research in two important directions: 1) It provides a systematic introduction for newcomers to the field, covering topics such as the formulation of the VPR problem, a general-purpose algorithmic pipeline, an evaluation methodology for VPR approaches, and the major challenges for VPR and how they may be addressed. 2) As a contribution for researchers acquainted with the VPR problem, it examines the intricacies of different VPR problem types regarding input, data processing, and output. The tutorial also discusses the subtleties behind the evaluation of VPR algorithms, e.g., the evaluation of a VPR system that has to find all matching database images per query, as opposed to just a single match. Practical code examples in Python illustrate to prospective practitioners and researchers how VPR is implemented and evaluated.
['Tobias Fischer', 'Michael Milford', 'Sourav Garg', 'Peer Neubert', 'Stefan Schubert']
2023-03-06
null
null
null
null
['visual-place-recognition']
['computer-vision']
[ 4.54792157e-02 -2.23991185e-01 -1.23977877e-01 -3.04218531e-01 -5.85912168e-01 -9.91224766e-01 6.27801061e-01 3.49448889e-01 -6.03432715e-01 3.32981348e-01 -1.37227714e-01 -5.78145027e-01 -1.18466653e-01 -5.43401361e-01 -7.64698863e-01 -3.60799700e-01 -2.23430797e-01 4.14094895e-01 4.88502711e-01 -3.65674913e-01 8.77993286e-01 8.91742885e-01 -2.09964466e+00 -7.09162727e-02 3.37097496e-01 6.05536044e-01 5.46070337e-01 7.03478813e-01 -8.06144699e-02 6.07639909e-01 -7.25077748e-01 -5.19463755e-02 3.48282933e-01 5.87925017e-02 -9.67760146e-01 -7.83002563e-03 5.80292761e-01 8.25409368e-02 -5.13344646e-01 9.32682097e-01 3.99264216e-01 4.43685383e-01 5.47329962e-01 -1.81837010e+00 -8.34996104e-01 -1.25538379e-01 -4.99358535e-01 4.26617891e-01 9.45596516e-01 1.48786575e-01 8.16908181e-01 -1.13655436e+00 9.19745862e-01 1.05393732e+00 8.69176209e-01 2.23403111e-01 -1.12807143e+00 6.35960251e-02 1.40941843e-01 5.01813114e-01 -1.93947268e+00 -5.60346425e-01 3.58723789e-01 -5.25437593e-01 1.46384096e+00 4.20230269e-01 6.53593421e-01 6.10794663e-01 1.43656120e-01 9.25062299e-01 7.90378630e-01 -5.36585510e-01 3.51410419e-01 1.32902160e-01 1.88450545e-01 5.00640094e-01 2.07854584e-01 2.23394930e-02 -7.81835079e-01 -1.33363888e-01 9.54218745e-01 -3.83002013e-02 -1.92884549e-01 -9.63649213e-01 -1.48262608e+00 5.11040986e-01 6.42789185e-01 2.37955585e-01 -2.20380038e-01 1.54352605e-01 2.16638684e-01 2.60324210e-01 -2.99963593e-01 5.62954605e-01 -1.03410788e-01 -3.11933160e-01 -8.48506212e-01 3.68903458e-01 7.00175583e-01 1.40309894e+00 1.01340246e+00 -1.83571979e-01 1.31571209e-02 8.79023731e-01 7.25836277e-01 6.14200652e-01 5.12884319e-01 -9.93379354e-01 3.25536758e-01 4.15660918e-01 4.32399720e-01 -1.31385207e+00 -3.19137573e-01 4.01411653e-01 3.33278589e-02 5.47777295e-01 1.20296054e-01 2.24479288e-01 -1.13883102e+00 1.13779545e+00 4.20565791e-02 -2.21954241e-01 2.62982756e-01 9.51428056e-01 1.35380971e+00 7.40430176e-01 -5.41624948e-02 3.47943068e-01 1.30684447e+00 -1.25924885e+00 -5.09207845e-01 -6.90487862e-01 3.50489020e-01 -8.27014863e-01 9.52358246e-01 -1.52341798e-01 -9.52849150e-01 -3.49839509e-01 -1.40348089e+00 -5.72471678e-01 -1.00507140e+00 1.08286366e-01 5.26292264e-01 5.29149532e-01 -1.70601201e+00 3.69580477e-01 -8.07964265e-01 -1.18309617e+00 1.02133334e-01 5.23159504e-01 -6.95453405e-01 -1.16688646e-01 -6.24738753e-01 1.33563375e+00 2.73017973e-01 -5.53153083e-02 -6.37294471e-01 -3.08507264e-01 -1.24229586e+00 -2.30268791e-01 5.70992082e-02 -2.81495035e-01 1.54288113e+00 -5.37496328e-01 -1.07430315e+00 1.33055294e+00 -6.68374836e-01 -3.75209361e-01 5.07382691e-01 1.55522034e-01 -3.63650471e-01 1.05726808e-01 5.77322543e-01 9.29250956e-01 2.48827890e-01 -1.26806021e+00 -9.33901370e-01 -3.12759340e-01 2.06281915e-01 8.75393510e-01 5.16029119e-01 -7.34801963e-02 -8.67541850e-01 -1.91220298e-01 5.05172133e-01 -8.83029401e-01 -2.99418509e-01 3.27791631e-01 -4.51505110e-02 -1.13074675e-01 8.11179042e-01 -3.71361196e-01 6.60981715e-01 -2.63702703e+00 -3.34427655e-01 3.60891372e-01 1.59805343e-01 -1.54273650e-02 -2.11805906e-02 9.17947829e-01 -2.86276964e-03 -6.60114661e-02 -4.53241840e-02 -1.44017383e-01 3.05123210e-01 2.87377387e-01 -5.55820346e-01 8.49579811e-01 -3.19167823e-01 1.00204051e+00 -1.11544752e+00 -4.18819755e-01 6.28869534e-01 4.09862071e-01 1.08083161e-02 -5.92133813e-02 1.92704275e-01 1.03140846e-01 9.31140482e-02 9.27428424e-01 7.05700338e-01 -2.24092290e-01 4.09403890e-02 2.20561743e-01 -8.24513316e-01 2.28912205e-01 -1.22790623e+00 1.63328397e+00 -9.75105166e-02 1.42675161e+00 2.68735196e-02 -5.84777296e-01 1.07763243e+00 1.11833699e-01 3.32134515e-01 -8.78457606e-01 -2.16735110e-01 4.82929170e-01 -5.06005108e-01 -3.28109652e-01 1.20135295e+00 5.33751845e-01 -6.47169501e-02 1.30259991e-01 -2.08064467e-02 6.91115633e-02 3.21712404e-01 -2.52339747e-02 1.13112628e+00 1.73501357e-01 8.46502721e-01 -1.95822299e-01 4.22403276e-01 6.92499876e-01 9.14659500e-02 1.15665936e+00 -7.18325496e-01 7.77029932e-01 4.50490601e-02 -5.35467982e-01 -8.44096720e-01 -1.21862400e+00 -1.70984820e-01 1.18129241e+00 1.05489588e+00 -5.77304602e-01 -1.78330958e-01 -2.09415257e-01 1.79194272e-01 3.72074634e-01 -6.21688843e-01 4.24427360e-01 -4.47322965e-01 -3.75150412e-01 5.09717464e-01 6.89352095e-01 6.29747689e-01 -1.33073997e+00 -1.29878688e+00 -2.70508081e-01 -3.38664949e-01 -8.37905586e-01 -9.97051969e-02 2.16950014e-01 -5.76708257e-01 -1.12890720e+00 -8.42977762e-01 -1.33710825e+00 7.37668037e-01 1.15353811e+00 1.02386165e+00 9.06935260e-02 -2.68444568e-01 1.21664083e+00 -2.89967269e-01 -3.32306892e-01 1.60519630e-01 -1.06066383e-01 -9.37597603e-02 -5.96913755e-01 9.29901540e-01 -3.34532678e-01 -4.53374177e-01 4.40309614e-01 -4.20857459e-01 -3.31116825e-01 4.76654559e-01 4.76012349e-01 9.88421917e-01 -3.87196094e-01 -2.08519340e-01 -2.90819496e-01 6.08107686e-01 -3.70703876e-01 -8.53965759e-01 6.28124058e-01 -3.62243563e-01 -1.45606533e-01 -1.04013190e-01 -2.08363578e-01 -4.74376947e-01 3.01587522e-01 -2.03165159e-01 -2.54925013e-01 -3.70396554e-01 2.74960369e-01 9.59696695e-02 -7.29135156e-01 8.93715560e-01 5.54551959e-01 -8.93664733e-02 -2.18466967e-01 5.08830965e-01 6.99168265e-01 7.81061649e-01 -1.92954272e-01 5.99370420e-01 6.76242530e-01 -1.68122336e-01 -1.28163362e+00 1.71076119e-01 -1.21957266e+00 -7.01030314e-01 -2.64657736e-01 6.93024337e-01 -7.57141888e-01 -8.19910526e-01 2.44444132e-01 -1.17426312e+00 -2.81916112e-01 -3.78597647e-01 2.60394067e-01 -9.13796782e-01 1.98954225e-01 -2.13788018e-01 -7.31784999e-01 -2.16246117e-02 -1.34880412e+00 1.08031595e+00 4.86131042e-01 -1.66715041e-01 -9.45163131e-01 4.13567632e-01 -1.29694924e-01 3.05444658e-01 1.38177067e-01 1.28864884e-01 -6.56794071e-01 -7.78931439e-01 -1.88988194e-01 -2.90779591e-01 -6.01668715e-01 -1.82035506e-01 4.04917588e-03 -9.60902989e-01 -2.01976210e-01 -4.38143462e-01 2.45601665e-02 4.00148302e-01 2.85232425e-01 4.55359697e-01 -7.69972727e-02 -8.93078506e-01 4.88027662e-01 1.71898246e+00 6.11812055e-01 8.20016742e-01 1.17615628e+00 6.13753259e-01 5.71119249e-01 9.28960979e-01 6.74999952e-02 7.66343355e-01 7.92267501e-01 3.38593692e-01 -8.58453363e-02 1.15075046e-02 -3.14840198e-01 1.95244864e-01 2.93350816e-01 -9.31868181e-02 9.89740342e-02 -1.37145960e+00 8.20038617e-01 -2.14094162e+00 -9.53831255e-01 -1.33319467e-01 2.25899577e+00 -4.16347720e-02 -4.43217725e-01 3.39144990e-02 -1.15291379e-01 8.75610173e-01 3.46393794e-01 -3.97273958e-01 -3.63989860e-01 -4.34000120e-02 -2.76818305e-01 9.54626441e-01 5.77361941e-01 -1.46491480e+00 1.05307293e+00 8.04384232e+00 1.37630075e-01 -1.15576613e+00 -1.59732714e-01 -1.15249090e-01 4.10230041e-01 1.10345438e-01 5.09917364e-02 -7.39220083e-01 -2.68613477e-03 5.55694044e-01 -3.88095200e-01 6.25108004e-01 1.44868934e+00 -1.40280932e-01 -3.82124692e-01 -1.23323631e+00 1.72553754e+00 5.15795350e-01 -1.43062222e+00 -2.51392126e-01 4.38656658e-02 3.51485878e-01 6.77324951e-01 2.79303998e-01 2.52180576e-01 1.95216745e-01 -1.03158915e+00 1.10937464e+00 1.88508660e-01 5.21204412e-01 -3.77550513e-01 6.61912978e-01 -3.70889939e-02 -1.40355337e+00 -1.06080122e-01 -5.41734815e-01 -1.55186385e-01 4.10614014e-02 -3.96970153e-01 -1.04261422e+00 4.15471494e-01 1.13707089e+00 8.36332619e-01 -9.60605681e-01 1.84366810e+00 -2.47862637e-01 -4.51819152e-01 -3.30753118e-01 -1.71415687e-01 2.09234133e-01 5.12545404e-04 5.81724405e-01 1.34122276e+00 3.37187082e-01 -8.96239355e-02 4.60515767e-02 5.09606838e-01 2.91731626e-01 1.39614120e-01 -1.25635004e+00 2.44569972e-01 8.94764423e-01 1.07421029e+00 -1.18007672e+00 -8.39859918e-02 -4.10618842e-01 1.13315916e+00 3.60162228e-01 4.63211328e-01 -5.25564671e-01 -8.30853224e-01 7.25342810e-01 5.03720418e-02 5.43823361e-01 -5.68945587e-01 -3.46726209e-01 -8.30550134e-01 9.93644670e-02 -6.78013384e-01 4.06387180e-01 -1.15789711e+00 -8.84956062e-01 4.79221463e-01 1.00157902e-01 -1.53157067e+00 -1.42307833e-01 -8.60343635e-01 -4.67874855e-01 6.73761070e-01 -1.58824134e+00 -7.29562640e-01 -5.70058107e-01 4.71912533e-01 4.27126229e-01 1.10076696e-01 9.32069421e-01 1.56593844e-01 -1.60827443e-01 4.10650641e-01 2.31181517e-01 1.54691145e-01 6.02343559e-01 -1.26261044e+00 8.86883497e-01 1.06393135e+00 4.33577567e-01 1.02092695e+00 7.75377870e-01 -4.04988378e-01 -1.64264226e+00 -7.68689632e-01 9.59282041e-01 -8.78696501e-01 2.76568919e-01 -3.10768962e-01 -5.27997971e-01 1.16977215e+00 3.02986801e-01 -3.40659581e-02 6.64866686e-01 -1.40016750e-01 -2.95600623e-01 1.74294233e-01 -1.35753381e+00 1.02381265e+00 9.60335970e-01 -6.80148125e-01 -9.28456962e-01 2.15714991e-01 3.98752004e-01 -1.03534079e+00 -3.88159543e-01 -3.14420424e-02 5.66837251e-01 -9.12340999e-01 1.23035669e+00 1.19937003e-01 -1.61336899e-01 -1.04937828e+00 -4.99943227e-01 -9.19037640e-01 -4.60198373e-01 -3.87550563e-01 1.55318305e-01 7.58749187e-01 2.76952684e-01 -7.57605851e-01 4.91179764e-01 6.24789238e-01 -1.15928002e-01 -2.31671333e-01 -1.02067757e+00 -8.36851001e-01 -5.75047374e-01 -3.89623344e-01 5.15622437e-01 9.30899024e-01 3.53325009e-01 2.82175671e-02 4.61668670e-02 5.63007951e-01 3.80810022e-01 1.27663821e-01 9.54985142e-01 -9.85578239e-01 2.32766032e-01 -3.36806327e-01 -1.23984182e+00 -1.45755899e+00 -4.40693885e-01 -7.63236105e-01 5.75739264e-01 -2.30985427e+00 -2.22228885e-01 -2.75317937e-01 -3.99936140e-02 6.02330327e-01 4.50828612e-01 4.69802082e-01 4.43313628e-01 6.58487737e-01 -1.11997819e+00 5.19745238e-02 5.65361261e-01 -2.37290457e-01 -4.89821166e-01 -1.48269832e-01 -6.33268476e-01 6.13061070e-01 6.01736486e-01 -2.02418208e-01 -4.94379312e-01 -5.86164415e-01 8.46295506e-02 -3.66225868e-01 5.32020509e-01 -1.23909998e+00 8.80160213e-01 -9.94029790e-02 4.09862429e-01 -1.04870868e+00 5.54542005e-01 -8.62008333e-01 2.31273904e-01 1.27808735e-01 7.01891631e-02 8.92005801e-01 3.47968459e-01 4.55712646e-01 -2.75728762e-01 -2.63670862e-01 5.04115283e-01 -4.78185445e-01 -1.88015997e+00 -2.33438134e-01 -7.17315555e-01 -2.20409423e-01 1.45067894e+00 -7.93289125e-01 -4.99677271e-01 -1.98076829e-01 -5.52626908e-01 3.51369828e-01 1.06533873e+00 6.86358571e-01 8.53046179e-01 -1.33770466e+00 1.99554451e-02 1.63841501e-01 6.36690617e-01 -1.64359465e-01 5.00626825e-02 6.10111773e-01 -1.30121911e+00 6.54512167e-01 -4.25493181e-01 -8.30711603e-01 -1.39540100e+00 7.95036435e-01 3.20097506e-01 4.77245718e-01 -7.98075795e-01 7.18901277e-01 -2.82249060e-02 -6.90829933e-01 6.25593305e-01 -2.48303041e-02 -4.37144458e-01 -1.33961618e-01 7.59100616e-01 4.69211102e-01 1.96298942e-01 -1.01158333e+00 -1.06860185e+00 6.53529406e-01 9.77738500e-02 -5.01353621e-01 1.03009021e+00 -4.51755583e-01 -1.80772349e-01 9.23334837e-01 9.43772376e-01 -1.46292239e-01 -9.94330108e-01 1.50614366e-01 1.11075178e-01 -6.15244448e-01 -4.02290344e-01 -5.32554507e-01 -3.32080990e-01 7.71734834e-01 9.33543682e-01 2.31817216e-01 5.95500886e-01 1.49093494e-01 4.37843017e-02 7.51536846e-01 9.10430431e-01 -1.08955681e+00 -2.08373874e-01 8.05590093e-01 9.58386421e-01 -1.11714768e+00 -2.31928136e-02 -2.74574161e-01 -7.64295697e-01 1.01047027e+00 4.29852754e-01 -9.87932533e-02 5.90741456e-01 -6.33764565e-02 4.88481998e-01 -3.35092455e-01 -1.42432582e-02 -2.41639897e-01 8.52282867e-02 1.23130202e+00 1.40858829e-01 -9.31766480e-02 -8.32944810e-02 -1.01535343e-01 -4.40485120e-01 -2.70109147e-01 4.82075155e-01 1.61290348e+00 -4.90414321e-01 -7.95988917e-01 -6.77423120e-01 -5.80113791e-02 4.67042476e-02 -1.93291518e-03 -5.09778023e-01 9.14082289e-01 -8.65412354e-02 9.45658326e-01 1.91932455e-01 -4.17625695e-01 6.75728798e-01 -1.75723910e-01 4.24629003e-01 -6.08899891e-01 -3.85618299e-01 -5.65146327e-01 -1.59407794e-01 -7.57466674e-01 -4.60711211e-01 -8.33934903e-01 -1.23134863e+00 -4.14296001e-01 1.35886461e-01 6.18996210e-02 9.03009474e-01 6.30418181e-01 5.61770678e-01 6.00714982e-02 4.12860103e-02 -1.27852523e+00 8.33793059e-02 -3.23389471e-01 -5.11416674e-01 -8.56922269e-02 5.16615570e-01 -7.61682093e-01 -1.86409071e-01 -2.40662470e-02]
[7.500015735626221, -1.8724173307418823]
fa7204b7-ae65-4c8b-8100-15e7f619149f
ju_nlp-at-semeval-2016-task-11-identifying
null
null
https://aclanthology.org/S16-1152
https://aclanthology.org/S16-1152.pdf
JU\_NLP at SemEval-2016 Task 11: Identifying Complex Words in a Sentence
null
['B', 'Niloy Mukherjee', 'Dipankar Das', 'Braja Gopal Patra', 'Sivaji yopadhyay']
2016-06-01
null
null
null
semeval-2016-6
['complex-word-identification']
['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]
[-7.303244113922119, 3.6287333965301514]
c696fd50-8a56-44d0-aa4e-e4903704e08d
sok-explainable-machine-learning-for-computer
2208.10605
null
https://arxiv.org/abs/2208.10605v2
https://arxiv.org/pdf/2208.10605v2.pdf
SoK: Explainable Machine Learning for Computer Security Applications
Explainable Artificial Intelligence (XAI) aims to improve the transparency of machine learning (ML) pipelines. We systematize the increasingly growing (but fragmented) microcosm of studies that develop and utilize XAI methods for defensive and offensive cybersecurity tasks. We identify 3 cybersecurity stakeholders, i.e., model users, designers, and adversaries, who utilize XAI for 4 distinct objectives within an ML pipeline, namely 1) XAI-enabled user assistance, 2) XAI-enabled model verification, 3) explanation verification & robustness, and 4) offensive use of explanations. Our analysis of the literature indicates that many of the XAI applications are designed with little understanding of how they might be integrated into analyst workflows -- user studies for explanation evaluation are conducted in only 14% of the cases. The security literature sometimes also fails to disentangle the role of the various stakeholders, e.g., by providing explanations to model users and designers while also exposing them to adversaries. Additionally, the role of model designers is particularly minimized in the security literature. To this end, we present an illustrative tutorial for model designers, demonstrating how XAI can help with model verification. We also discuss scenarios where interpretability by design may be a better alternative. The systematization and the tutorial enable us to challenge several assumptions, and present open problems that can help shape the future of XAI research within cybersecurity.
['Sicco Verwer', 'Robert Baumgartner', 'Simon Dieck', 'Luca Pajola', 'Clinton Cao', 'Daniël Vos', 'Azqa Nadeem']
2022-08-22
null
null
null
null
['computer-security']
['miscellaneous']
[ 2.33200431e-01 6.60317063e-01 -2.36619905e-01 -1.02429733e-01 -1.02811791e-01 -1.33175540e+00 7.34495580e-01 2.11386204e-01 1.93764612e-01 7.47217238e-02 1.16434887e-01 -1.60454464e+00 -4.38301653e-01 -4.06670839e-01 -6.00463986e-01 -1.13004230e-01 1.70603946e-01 1.18172586e-01 -5.60824692e-01 7.29052648e-02 3.97736579e-01 7.03933954e-01 -1.16190243e+00 3.85169148e-01 5.24680316e-01 7.18477309e-01 -8.18952501e-01 5.77818155e-01 1.56922027e-01 9.71958101e-01 -8.96821022e-01 -4.53799248e-01 4.21508580e-01 -2.13154644e-01 -9.63290989e-01 -2.35556155e-01 1.30513296e-01 -4.22292054e-01 2.13943094e-01 9.57340777e-01 -3.72292668e-01 -3.87175620e-01 2.98161924e-01 -2.23192739e+00 -5.58836341e-01 8.90025020e-01 -2.10102633e-01 -1.17211156e-01 1.61497593e-01 8.65898848e-01 7.69891262e-01 -2.07383662e-01 4.62516278e-01 1.18980289e+00 4.26555663e-01 5.69965601e-01 -1.24393916e+00 -9.78882134e-01 3.65336180e-01 4.18198220e-02 -9.16971505e-01 -3.16457152e-01 6.86698139e-01 -7.23162234e-01 1.04980111e+00 9.38452601e-01 4.68171418e-01 1.38122082e+00 2.80791193e-01 3.48722696e-01 1.03069913e+00 -4.91302222e-01 2.62328655e-01 6.40170455e-01 8.77243042e-01 3.51292491e-01 1.04330719e+00 6.61372244e-01 -3.38703275e-01 -6.82036400e-01 2.92566448e-01 5.45455143e-02 -2.01138794e-01 -1.20820686e-01 -1.11218691e+00 8.06784093e-01 1.63208861e-02 2.62936354e-01 -1.68530475e-02 1.79961266e-03 4.70135689e-01 5.27516603e-01 -1.24794759e-01 1.47069860e+00 -9.07687545e-01 -1.26280680e-01 -3.68293971e-01 2.26447001e-01 9.69443381e-01 8.17798495e-01 3.31003159e-01 3.55973959e-01 4.94159192e-01 -4.59690839e-01 7.92647064e-01 6.89388365e-02 -6.74271360e-02 -8.35811436e-01 3.04960728e-01 9.38620627e-01 4.54528660e-01 -9.04328406e-01 -3.73375088e-01 -4.78925169e-01 -2.10433930e-01 1.02861667e+00 5.40617049e-01 -2.75952429e-01 -5.80648780e-01 1.59729338e+00 1.42061502e-01 6.59209117e-03 2.27485910e-01 7.45987117e-01 3.70955437e-01 3.66406798e-01 4.85800713e-01 5.55602685e-02 1.44503748e+00 -5.84114492e-01 -5.36605120e-01 -4.87184316e-01 8.46409917e-01 -4.65666175e-01 1.15617597e+00 7.05040038e-01 -1.06198239e+00 -1.88326493e-01 -1.48238325e+00 2.63287902e-01 -5.89005113e-01 -2.96314150e-01 9.14490283e-01 1.24282694e+00 -4.99878615e-01 4.00029451e-01 -9.07991588e-01 4.39956784e-03 3.12456578e-01 3.76327425e-01 -2.04326585e-01 3.88097435e-01 -1.02645874e+00 1.10202336e+00 2.39744723e-01 3.58018978e-03 -7.89353192e-01 -1.15901744e+00 -9.95614111e-01 5.48491895e-01 4.94344324e-01 -6.99136555e-01 1.33210456e+00 -1.19481254e+00 -8.98319066e-01 3.45082432e-01 5.06664813e-01 -4.76082563e-01 5.04004240e-01 -1.83545142e-01 -4.43447918e-01 -2.61976689e-01 -1.24884985e-01 1.88987106e-01 4.54051733e-01 -1.64288080e+00 -5.37455440e-01 -4.55377430e-01 8.14018607e-01 -4.13197637e-01 -1.76148087e-01 4.26171750e-01 4.03785676e-01 -2.01552689e-01 -2.73388565e-01 -9.51394856e-01 -2.08914895e-02 -2.56010909e-02 -6.67828202e-01 5.01677215e-01 1.15162170e+00 -6.85393214e-01 1.35492563e+00 -2.17811561e+00 -2.67839313e-01 5.46476722e-01 6.02291882e-01 4.80409801e-01 9.76216570e-02 6.23881102e-01 -7.85646319e-01 1.12583911e+00 8.97523239e-02 -2.19369512e-02 4.31771427e-01 -3.15911591e-01 -8.08058739e-01 2.18682274e-01 2.56170332e-01 7.58274317e-01 -7.55162299e-01 1.09549828e-01 3.84904265e-01 2.17675343e-01 -2.93265402e-01 1.07165664e-01 -2.26327538e-01 4.23724174e-01 -2.61854172e-01 8.81717384e-01 6.04638577e-01 -2.42234051e-01 3.42775822e-01 -2.22781133e-02 -3.84398192e-01 3.89893264e-01 -9.78178501e-01 7.09137380e-01 -4.23494816e-01 7.01842844e-01 4.06996429e-01 -2.23568514e-01 2.69057721e-01 4.81450200e-01 -6.02751374e-02 -2.25896146e-02 2.32670143e-01 -8.04351568e-02 5.43048620e-01 -2.60491401e-01 1.90273114e-02 7.29146926e-03 -3.31814028e-02 1.20822001e+00 -5.52183092e-01 1.38274372e-01 -5.21080613e-01 2.50594854e-01 1.17502153e+00 -9.41929296e-02 7.15638041e-01 -8.75532255e-02 3.04267883e-01 4.87607360e-01 5.36023676e-01 8.46604824e-01 -2.76039809e-01 3.20821665e-02 9.62216735e-01 -7.02011168e-01 -8.85992765e-01 -7.63402224e-01 1.51493475e-01 5.33796847e-01 -9.58028808e-03 -8.14647198e-01 -9.03076172e-01 -1.09668911e+00 1.60880178e-01 1.78589368e+00 -6.29555881e-01 -6.81891799e-01 -2.20381647e-01 -1.26534134e-01 9.07150388e-01 5.22133052e-01 -3.98233123e-02 -7.02009022e-01 -1.42962027e+00 -1.79892063e-01 3.57156731e-02 -7.14114428e-01 -8.73436853e-02 9.64126177e-03 -4.44738060e-01 -1.53225112e+00 9.13919687e-01 2.54884779e-01 6.03237450e-01 3.82028133e-01 8.82475853e-01 7.73373067e-01 -2.23038867e-01 8.22720706e-01 -3.79704297e-01 -1.12172484e+00 -1.07903385e+00 -3.44674408e-01 -5.84467947e-02 -5.03092349e-01 4.46736664e-01 -3.13210636e-01 -1.65137440e-01 5.48261344e-01 -1.13007116e+00 2.03253046e-01 2.53480077e-01 5.38841486e-01 -2.26075262e-01 1.71347875e-02 1.06195040e-01 -1.21101582e+00 6.98960721e-01 -5.75990319e-01 -8.72100651e-01 6.78430080e-01 -1.22872865e+00 -6.07081577e-02 6.61491156e-01 -6.01327002e-01 -8.77596915e-01 -4.26501215e-01 4.46384877e-01 -2.94598788e-01 -5.10880113e-01 6.89298272e-01 -6.55317426e-01 -2.62119412e-01 1.00531542e+00 -5.76417804e-01 2.03522995e-01 -1.46851078e-01 4.14626211e-01 3.28516752e-01 1.36673883e-01 -6.03430986e-01 1.22368026e+00 1.93641812e-01 -1.02753967e-01 -1.58076793e-01 -2.63371885e-01 2.76544362e-01 -1.80946529e-01 -2.02534080e-01 5.51281512e-01 -2.98756391e-01 -1.15492308e+00 -4.78214584e-02 -1.28872204e+00 -3.94532174e-01 -2.55569696e-01 3.57060552e-01 -2.28495702e-01 3.36779892e-01 -2.99245298e-01 -1.00266266e+00 -4.15476739e-01 -1.61291790e+00 3.82781982e-01 7.42969140e-02 -9.40711915e-01 -8.70182931e-01 -3.43706042e-01 5.87987363e-01 4.06836629e-01 6.99336469e-01 1.57955253e+00 -1.34016216e+00 -6.54241621e-01 -4.23012733e-01 -4.45621908e-02 9.84675363e-02 2.39441127e-01 5.15181541e-01 -1.17391288e+00 -3.08488887e-02 3.27968538e-01 1.53677282e-03 -3.18112642e-01 -3.02120615e-02 8.77164662e-01 -9.30094421e-01 -3.37445617e-01 5.31763732e-01 9.96714473e-01 6.72946692e-01 5.05846918e-01 7.29581356e-01 6.29036844e-01 1.01456869e+00 8.49294126e-01 3.52620572e-01 -1.20743670e-01 3.05209458e-01 7.72389412e-01 -1.24927223e-01 4.96503770e-01 -9.23245326e-02 2.48175532e-01 -4.49765474e-01 1.58188105e-01 -8.12028870e-02 -1.46403694e+00 -5.16543239e-02 -1.84459925e+00 -9.96428609e-01 -1.92014262e-01 2.06487870e+00 2.14655682e-01 5.01567483e-01 -4.98559773e-02 3.41991246e-01 3.02320778e-01 -1.96520150e-01 -5.92848599e-01 -6.68842316e-01 5.86569726e-01 -1.23600550e-01 3.26967835e-01 6.84272885e-01 -6.86486065e-01 6.57035172e-01 6.08590841e+00 -1.40597448e-01 -9.96240020e-01 -7.06238300e-02 5.71124732e-01 -9.89702344e-02 -7.28634775e-01 7.69072235e-01 -3.79543066e-01 1.25679374e-01 1.16292214e+00 -6.28750205e-01 6.07007802e-01 1.31542170e+00 4.77957159e-01 4.65543151e-01 -1.56733274e+00 3.94607812e-01 -2.46514097e-01 -1.57700908e+00 -5.77440783e-02 4.75986093e-01 -5.97495399e-02 -6.74177408e-01 4.11935121e-01 2.79884152e-02 3.74932647e-01 -1.21510565e+00 1.16393876e+00 2.15463191e-01 4.28705603e-01 -8.25465739e-01 6.21712387e-01 3.28873187e-01 -7.10565507e-01 -6.58224821e-01 2.88288414e-01 -4.83960003e-01 -6.24647997e-02 -1.09467879e-01 -1.06774616e+00 6.09226346e-01 5.20585060e-01 5.10890186e-02 -6.17024004e-01 5.27454495e-01 -4.50240046e-01 8.90109718e-01 -6.00106381e-02 2.97543526e-01 2.75488138e-01 -2.16856971e-01 5.03127217e-01 9.50306356e-01 -3.41397226e-02 3.38155866e-01 -3.03549439e-01 1.34909892e+00 6.92586482e-01 -5.90122998e-01 -9.28129256e-01 -6.20043695e-01 7.49155819e-01 1.33929443e+00 -6.29087925e-01 -1.30819172e-01 -4.08395022e-01 1.92017376e-01 -2.18699247e-01 4.77241248e-01 -9.18188393e-01 -3.04603040e-01 1.33973086e+00 1.89550385e-01 -4.78379339e-01 -2.78821766e-01 -1.13189185e+00 -8.56341839e-01 -1.93023756e-01 -1.90404522e+00 5.22217989e-01 -9.97912169e-01 -7.82574415e-01 3.90246868e-01 4.22131091e-01 -1.06647658e+00 -4.70675260e-01 -7.60739148e-01 -9.35252488e-01 8.55988562e-01 -9.32165146e-01 -1.65118992e+00 -2.34878302e-01 1.20189458e-01 4.35893424e-02 -1.69650957e-01 8.44666481e-01 -3.16835791e-01 -7.05559969e-01 5.17352641e-01 -5.67960382e-01 -2.06714123e-01 4.62416977e-01 -1.11938190e+00 8.92763019e-01 1.25138664e+00 7.85330310e-02 1.64383221e+00 9.61584985e-01 -8.66104424e-01 -1.70072281e+00 -8.08613300e-01 5.71566105e-01 -1.16197646e+00 8.36500466e-01 -3.30069363e-01 -9.62501585e-01 1.34263337e+00 1.86576530e-01 -7.80618131e-01 1.14742196e+00 3.23669434e-01 -7.80566752e-01 1.78501651e-01 -1.36110997e+00 1.06502128e+00 7.35470057e-01 -5.93548238e-01 -5.02557576e-01 2.24722415e-01 9.36048627e-01 -1.39531121e-01 -6.60944581e-01 9.12110731e-02 6.36903524e-01 -9.10889268e-01 9.10148680e-01 -1.22667110e+00 3.78981605e-02 -4.79156435e-01 9.32876840e-02 -8.39409709e-01 -2.76849538e-01 -9.85011697e-01 -5.19867875e-02 1.30397582e+00 6.19625568e-01 -1.08614993e+00 4.07148510e-01 2.04838276e+00 -2.11475685e-01 -3.49457979e-01 -3.92289847e-01 -6.01504326e-01 -8.65749270e-02 -8.75352323e-01 1.17080367e+00 1.28399575e+00 6.60972953e-01 -7.07323023e-04 -3.46483998e-02 8.12837780e-01 4.52632666e-01 -1.20215565e-01 1.00684214e+00 -1.26694751e+00 -3.34741682e-01 -5.21197677e-01 -2.14990363e-01 1.16423322e-02 1.49723142e-01 -6.20414674e-01 -5.07049561e-01 -1.14534402e+00 -5.75294867e-02 -1.08396031e-01 -3.04102898e-02 1.06482923e+00 2.10871045e-02 -7.17012644e-01 6.33794725e-01 2.44698524e-01 2.68226385e-01 -3.63125175e-01 3.45337957e-01 -1.36567548e-01 -3.01171631e-01 -7.27505386e-02 -1.64205790e+00 9.11523104e-01 7.80601203e-01 -5.19281805e-01 -8.96877050e-01 -3.22836339e-01 2.84409642e-01 -9.85047445e-02 7.60143876e-01 -6.60533130e-01 1.82858869e-01 -8.27939212e-01 2.23185167e-01 -1.02592399e-02 -3.80693376e-02 -1.25127411e+00 9.21808779e-01 7.09426045e-01 -3.48466009e-01 3.95624429e-01 6.44833744e-01 4.37098667e-02 2.75522619e-01 -5.04460156e-01 5.18673480e-01 2.91888826e-02 -1.22272067e-01 -7.11772814e-02 -5.99911094e-01 -5.19026458e-01 1.22660851e+00 -2.47691855e-01 -9.04134095e-01 -3.96730036e-01 -4.31112915e-01 1.12367891e-01 9.76813674e-01 2.90405989e-01 3.64800692e-01 -6.20450616e-01 6.43453049e-03 3.40288520e-01 1.92814752e-01 -4.21102434e-01 -6.07436709e-02 4.53561068e-01 -3.94610852e-01 3.98025513e-01 -3.48027527e-01 1.01716556e-01 -1.53962910e+00 9.68799293e-01 2.85959452e-01 -8.95712227e-02 -5.38757920e-01 3.08349788e-01 4.89754349e-01 -2.47707993e-01 2.81321973e-01 -4.31050867e-01 6.43663481e-02 -3.74319851e-01 7.75693953e-01 3.72552097e-01 -3.24728578e-01 -4.59501147e-02 -6.62878454e-01 -1.15880787e-01 -3.47358406e-01 -4.50035781e-01 1.11629045e+00 3.33519012e-01 4.40815510e-03 1.42643914e-01 4.63733375e-01 9.24141239e-03 -9.80548680e-01 5.20003319e-01 2.76420891e-01 -5.85490763e-01 -2.09502622e-01 -1.40527511e+00 -4.78081375e-01 9.54870701e-01 1.40455961e-01 5.73606312e-01 8.25852871e-01 -2.48137698e-01 1.44196779e-01 3.62455308e-01 1.36473328e-01 -6.85337663e-01 -3.21845114e-01 -1.27311960e-01 1.17471397e+00 -7.58151591e-01 2.45853394e-01 -3.85633290e-01 -8.91828477e-01 1.30252433e+00 9.05274808e-01 6.31158531e-01 4.36217755e-01 4.84432191e-01 4.29240286e-01 -4.01418805e-01 -8.93122256e-01 5.37352026e-01 -1.53859541e-01 7.07010567e-01 1.83509693e-01 6.55866414e-02 7.65490085e-02 1.14337099e+00 -2.59674281e-01 -3.19156319e-01 8.82048130e-01 1.09808600e+00 4.06159014e-02 -1.16506433e+00 -9.05245125e-01 9.92919356e-02 -1.53971270e-01 6.84428290e-02 -1.12861896e+00 1.41397548e+00 -3.73781547e-02 1.46297455e+00 -5.98169625e-01 -6.93894148e-01 4.17659193e-01 2.92245716e-01 -1.32208586e-01 -4.48687226e-01 -1.29259706e+00 -2.49151602e-01 4.44475442e-01 -7.28190601e-01 4.27095801e-01 -7.95102775e-01 -1.08016360e+00 -4.81225669e-01 -3.96174848e-01 2.20976904e-01 1.03030503e+00 9.62732613e-01 5.87217808e-01 3.08760494e-01 3.13775808e-01 -2.60978729e-01 -8.21291029e-01 -3.52980733e-01 -1.61042199e-01 1.46724835e-01 1.98531330e-01 -4.35418844e-01 -8.40948939e-01 4.91295569e-02]
[8.716739654541016, 6.153114318847656]
0167121c-d59b-4923-90e5-15fc40c08d6c
hyspa-hybrid-span-generation-for-scalable
2106.15838
null
https://arxiv.org/abs/2106.15838v1
https://arxiv.org/pdf/2106.15838v1.pdf
HySPA: Hybrid Span Generation for Scalable Text-to-Graph Extraction
Text-to-Graph extraction aims to automatically extract information graphs consisting of mentions and types from natural language texts. Existing approaches, such as table filling and pairwise scoring, have shown impressive performance on various information extraction tasks, but they are difficult to scale to datasets with longer input texts because of their second-order space/time complexities with respect to the input length. In this work, we propose a Hybrid Span Generator (HySPA) that invertibly maps the information graph to an alternating sequence of nodes and edge types, and directly generates such sequences via a hybrid span decoder which can decode both the spans and the types recurrently in linear time and space complexities. Extensive experiments on the ACE05 dataset show that our approach also significantly outperforms state-of-the-art on the joint entity and relation extraction task.
['Julia Hockenmaier', 'Heng Ji', 'Chenkai Sun', 'Liliang Ren']
2021-06-30
null
https://aclanthology.org/2021.findings-acl.356
https://aclanthology.org/2021.findings-acl.356.pdf
findings-acl-2021-8
['joint-entity-and-relation-extraction']
['natural-language-processing']
[ 4.23070669e-01 5.00163317e-01 -3.69268984e-01 -1.43802181e-01 -1.14192367e+00 -8.86093676e-01 4.63258207e-01 6.25458241e-01 -1.95548788e-01 9.23875928e-01 1.37269318e-01 -6.45634413e-01 1.01392528e-04 -1.07150710e+00 -6.75858855e-01 -6.78338856e-02 -2.17249721e-01 9.25162971e-01 3.19506079e-01 -1.23771749e-01 9.65852961e-02 1.44096855e-02 -1.13251865e+00 1.53267339e-01 1.02470434e+00 5.88358223e-01 -2.51927584e-01 7.17345774e-01 -6.29585743e-01 8.07308674e-01 -5.23459911e-01 -1.01936269e+00 2.63850927e-01 -5.33542156e-01 -9.52405632e-01 -1.47701964e-01 4.98646110e-01 -1.01648502e-01 -7.22084761e-01 8.51637423e-01 2.76662469e-01 -1.71985850e-01 7.46072948e-01 -1.24298143e+00 -2.12664127e-01 1.35083032e+00 -7.43121386e-01 1.57762617e-01 6.87610507e-01 -3.01245272e-01 1.63938260e+00 -8.21690381e-01 8.93749356e-01 1.03908145e+00 6.59974039e-01 2.62311578e-01 -1.30029702e+00 -6.33454204e-01 8.54401886e-02 2.04795554e-01 -1.47162199e+00 -4.08257216e-01 5.42851627e-01 -2.74636358e-01 1.40321076e+00 4.00171638e-01 4.25100952e-01 7.61823595e-01 -1.82921529e-01 8.25519800e-01 7.96604455e-01 -5.38769722e-01 -1.15206592e-01 -1.76880956e-01 3.04658532e-01 1.00437021e+00 7.79623091e-01 -3.69326025e-01 -7.44917870e-01 -2.59292960e-01 3.44831556e-01 -6.26352906e-01 -1.70015797e-01 6.58854619e-02 -1.22165549e+00 7.34093547e-01 4.25240286e-02 8.48545432e-02 -2.27825969e-01 -2.77750921e-02 5.44645309e-01 2.92495668e-01 4.02426988e-01 5.40468872e-01 -6.89812601e-01 -9.49400365e-02 -9.49894190e-01 3.76102030e-01 1.48240399e+00 1.53382432e+00 6.90069199e-01 -4.56826776e-01 -4.05966431e-01 5.04693389e-01 3.39152738e-02 3.76547307e-01 -3.05551221e-03 -2.12233618e-01 1.31392932e+00 9.80743229e-01 -1.65189311e-01 -9.26160514e-01 -5.83001375e-01 -2.89324135e-01 -6.80240333e-01 -5.22361040e-01 6.53945684e-01 -5.74079514e-01 -7.63193309e-01 1.67484272e+00 5.56844413e-01 -1.78531762e-02 3.63871567e-02 4.25252855e-01 8.75542700e-01 5.40428996e-01 -1.11454919e-01 -1.71344161e-01 1.65897596e+00 -8.23479712e-01 -8.32181275e-01 -5.99880099e-01 9.57675815e-01 -5.12568116e-01 5.76645732e-01 -6.32058010e-02 -1.14092159e+00 1.51428590e-02 -1.06866133e+00 -3.12681586e-01 -3.13356519e-01 1.31818339e-01 7.87710547e-01 5.55832684e-01 -6.71016455e-01 5.73214710e-01 -7.66451478e-01 7.29649439e-02 3.18730950e-01 2.17434853e-01 -4.59040165e-01 1.21584490e-01 -1.33302426e+00 7.79709995e-01 8.03349316e-01 3.35241929e-02 -8.63773823e-02 -8.45623314e-01 -1.25969124e+00 2.08224565e-01 9.20209229e-01 -8.32781136e-01 1.15071142e+00 -6.21869080e-02 -1.10300672e+00 6.36744559e-01 -3.86988908e-01 -7.01616168e-01 3.65641266e-01 -2.20970660e-01 -3.82159203e-01 8.07921365e-02 8.83483514e-02 2.72379816e-01 3.07268292e-01 -6.99187517e-01 -6.29084647e-01 -4.25091833e-01 -6.03355728e-02 2.68090993e-01 -2.85634518e-01 1.70393825e-01 -7.19519198e-01 -5.98632097e-01 1.69471025e-01 -1.01307166e+00 -3.09202790e-01 -4.30001855e-01 -1.17391455e+00 -5.35704195e-01 3.98069084e-01 -1.08624446e+00 1.88953698e+00 -1.54532838e+00 2.05070913e-01 4.95065063e-01 4.68187213e-01 2.46548727e-01 1.50777891e-01 8.30786467e-01 2.10111871e-01 2.98690677e-01 -4.26847935e-01 -2.23339632e-01 2.61506766e-01 1.78120192e-02 -5.01427576e-02 1.87643528e-01 3.10647488e-01 1.14935982e+00 -9.80578780e-01 -9.44472551e-01 -3.89072150e-01 1.85914394e-02 -4.91389632e-01 1.65094286e-01 -4.56955969e-01 -1.71963081e-01 -3.48902643e-01 3.88793677e-01 5.52718759e-01 -4.67960179e-01 8.90411556e-01 -8.18817839e-02 1.42926902e-01 1.12566435e+00 -1.25259864e+00 1.49516273e+00 -3.84621680e-01 6.72148407e-01 -2.62189686e-01 -5.98178267e-01 7.03676283e-01 2.74915457e-01 1.40597835e-01 -2.83688366e-01 -3.26663256e-02 3.00156057e-01 7.92986080e-02 -4.03050810e-01 6.51501596e-01 2.48830110e-01 -5.55470288e-01 4.91518587e-01 1.15718357e-01 1.00658543e-01 9.03872013e-01 7.64491320e-01 1.60380316e+00 -1.41353123e-02 5.52132964e-01 7.14087784e-02 3.60543042e-01 3.96453701e-02 6.46526098e-01 5.97858191e-01 4.96146560e-01 3.62491846e-01 1.14989543e+00 -1.84347525e-01 -1.09989595e+00 -7.75915146e-01 3.42243224e-01 7.98560977e-01 -2.30182767e-01 -1.11440551e+00 -8.19113791e-01 -1.18686652e+00 4.17115949e-02 7.25176990e-01 -5.28232574e-01 8.38750973e-02 -9.30830240e-01 -5.87816000e-01 7.10042775e-01 6.31271243e-01 2.14480072e-01 -8.94347608e-01 -2.71846026e-01 4.69763100e-01 -3.97901684e-01 -1.78009939e+00 -7.73275435e-01 2.24379092e-01 -4.92344648e-01 -1.24565387e+00 -1.47498861e-01 -9.92568791e-01 7.97747493e-01 -2.56605536e-01 1.39357865e+00 1.22547537e-01 -2.34250650e-01 -4.60854948e-01 -3.80291641e-01 -2.51606852e-01 -5.19792020e-01 8.34623754e-01 -4.42938924e-01 -2.16249436e-01 2.91985899e-01 -5.33909976e-01 -3.14654678e-01 9.39098448e-02 -8.27103555e-01 4.16283488e-01 6.84420288e-01 6.76309586e-01 3.72210473e-01 3.69358845e-02 4.70022857e-01 -1.79021955e+00 5.16845524e-01 -3.43488336e-01 -6.98236704e-01 5.58727145e-01 -7.39559650e-01 5.48102796e-01 7.32881188e-01 3.81180383e-02 -7.97723532e-01 2.39718169e-01 -1.91994190e-01 2.19101474e-01 2.25008488e-01 9.54077542e-01 -3.17787528e-01 2.93988615e-01 4.49720830e-01 5.26439473e-02 -4.05380994e-01 -3.86983246e-01 6.25566065e-01 7.12432623e-01 5.62456906e-01 -4.25586551e-01 1.18471432e+00 -4.28210907e-02 2.73322612e-01 -4.47359294e-01 -9.66019213e-01 -5.12673259e-01 -7.91082442e-01 2.87657738e-01 5.72593212e-01 -8.47814083e-01 -5.35560489e-01 3.04534405e-01 -1.10857475e+00 -1.73069835e-01 -1.16817452e-01 3.81067470e-02 -1.00323439e-01 4.89084631e-01 -8.58735621e-01 -7.64440894e-01 -7.56447732e-01 -6.11044586e-01 9.46837902e-01 1.31989911e-01 -4.16692585e-01 -8.66533756e-01 -9.24812779e-02 1.68864682e-01 1.24260327e-02 3.80319566e-01 1.23091090e+00 -9.48655486e-01 -6.89688504e-01 -3.85264546e-01 -3.58300209e-01 -3.44041586e-01 -8.30292031e-02 -1.73924327e-01 -3.29541504e-01 2.96236295e-02 -9.01601374e-01 -2.02957988e-01 9.27157104e-01 -4.03219163e-01 8.58891666e-01 -9.63136256e-01 -6.36977911e-01 4.51295733e-01 1.35491848e+00 -1.06006414e-02 5.65708220e-01 3.53179947e-02 9.03932750e-01 5.67783475e-01 3.90858144e-01 4.35052812e-01 6.94024563e-01 7.85844922e-01 4.65194061e-02 9.63178501e-02 -3.30313981e-01 -8.60360205e-01 1.89320460e-01 1.00343251e+00 2.07562968e-01 -5.46594799e-01 -8.65611076e-01 7.21837521e-01 -1.72460902e+00 -8.15281153e-01 -4.44998026e-01 2.12330270e+00 1.39780366e+00 4.26319987e-01 3.12946886e-01 4.24395561e-01 6.94281340e-01 1.82078436e-01 -3.81781608e-01 -3.51317048e-01 2.95645669e-02 4.50317383e-01 9.83854175e-01 4.97471899e-01 -1.02612007e+00 1.03318036e+00 5.86821938e+00 9.13235009e-01 -3.83121192e-01 -1.64404124e-01 4.27658141e-01 1.74420938e-01 -5.46135068e-01 3.16998780e-01 -1.16916203e+00 3.51745456e-01 1.19198239e+00 -4.73758012e-01 5.10994673e-01 4.82208818e-01 -4.84206706e-01 1.02628849e-01 -1.18471670e+00 6.75664365e-01 -7.19998181e-02 -1.31826341e+00 2.44883224e-02 6.25546128e-02 6.39641702e-01 -1.63951516e-01 -4.50963944e-01 3.28704715e-01 6.61667049e-01 -7.99133837e-01 6.29257798e-01 1.07620932e-01 1.14048147e+00 -7.81607032e-01 7.63794184e-01 3.56923252e-01 -1.61574364e+00 2.50504822e-01 3.65973241e-03 1.03615224e-01 4.40685421e-01 7.67415047e-01 -1.23714900e+00 8.44124317e-01 9.54592451e-02 3.69454771e-01 -6.65986836e-01 9.96320069e-01 -7.55263984e-01 7.47097790e-01 -4.77708012e-01 -4.67338711e-01 -7.75770172e-02 -1.26808733e-01 6.32230818e-01 1.54077911e+00 2.18786955e-01 1.44676417e-01 1.31281823e-01 6.44508004e-01 -6.67879641e-01 1.34553730e-01 -5.28874338e-01 -4.77893233e-01 8.58476222e-01 1.47522247e+00 -8.73360515e-01 -5.43717802e-01 -5.23264349e-01 7.82116711e-01 8.67729187e-01 7.50144273e-02 -6.84723377e-01 -1.14470851e+00 3.68657768e-01 7.44077638e-02 5.50826848e-01 -3.80090356e-01 -4.31621701e-01 -1.15503025e+00 3.38269234e-01 -8.35050941e-01 8.51805151e-01 -2.70419031e-01 -9.83676016e-01 7.16740251e-01 1.98331885e-02 -8.10799837e-01 -5.43647408e-01 -3.08957458e-01 -2.51741678e-01 6.95673585e-01 -1.27352560e+00 -7.91347027e-01 7.14854747e-02 2.52874762e-01 2.70495713e-01 2.38260731e-01 8.91931951e-01 2.21460044e-01 -7.49167323e-01 1.14233327e+00 -2.73478091e-01 7.54385054e-01 3.15438181e-01 -1.66788971e+00 1.18243754e+00 1.10716069e+00 5.63986003e-01 5.45172095e-01 7.30424881e-01 -8.24146688e-01 -1.71115053e+00 -1.17687607e+00 1.64013302e+00 -3.04578990e-01 8.40461016e-01 -7.81240642e-01 -8.45181108e-01 8.87700260e-01 4.09329027e-01 -1.28302425e-01 5.13886690e-01 4.27579075e-01 -6.07121646e-01 5.52042946e-02 -8.76573682e-01 7.05129445e-01 1.45356584e+00 -5.32834589e-01 -4.14706945e-01 4.12336290e-01 8.92811596e-01 -9.73277986e-01 -9.62847769e-01 4.22484696e-01 3.60832214e-01 -3.83878410e-01 5.45034528e-01 -6.00650489e-01 4.66154128e-01 -1.49876431e-01 2.40575448e-01 -1.33466709e+00 -1.63196728e-01 -1.17915070e+00 -6.85439825e-01 1.49369383e+00 1.25537825e+00 -5.88600159e-01 9.25757647e-01 5.04111230e-01 7.58934841e-02 -9.02593791e-01 -7.64894605e-01 -6.66161716e-01 -4.01634902e-01 -2.01752767e-01 8.32490921e-01 6.11347497e-01 4.34109777e-01 1.12171304e+00 -2.44082987e-01 1.67805403e-01 5.36025703e-01 5.90057850e-01 7.61992097e-01 -1.13358569e+00 -4.21147913e-01 -2.86785483e-01 -2.59642750e-01 -9.13880944e-01 1.37173325e-01 -1.26362789e+00 8.02247003e-02 -1.89456916e+00 2.77864486e-01 -5.11227608e-01 1.36086091e-01 7.39462554e-01 -6.01121128e-01 2.52755973e-02 8.47550184e-02 -1.29673943e-01 -7.29356885e-01 1.91391036e-01 9.29618120e-01 -5.12698181e-02 -1.64222583e-01 -5.78669459e-02 -9.89294410e-01 4.21402305e-01 6.94263518e-01 -6.89061940e-01 -1.80126742e-01 -3.12289238e-01 7.78154135e-01 6.10028803e-01 -3.70167911e-01 -7.27259994e-01 3.69720548e-01 1.68350086e-01 -3.10972966e-02 -7.98099577e-01 1.84846800e-02 -4.56255615e-01 1.03701591e-01 2.86177427e-01 -5.49861968e-01 3.17311019e-01 1.18680388e-01 5.97852170e-01 -5.00512570e-02 -2.49029085e-01 1.02570705e-01 7.53962398e-02 -1.67829379e-01 4.74997103e-01 1.26892189e-03 6.04356885e-01 8.43019426e-01 2.15294614e-01 -5.40381730e-01 -2.91607350e-01 -3.51836473e-01 3.34628493e-01 2.04913929e-01 1.69600457e-01 4.31791455e-01 -1.18996286e+00 -9.69075143e-01 9.07663181e-02 4.91359308e-02 2.18960866e-01 -2.11901724e-01 8.00888062e-01 -5.01438975e-01 4.54562306e-01 3.67826760e-01 7.16512948e-02 -1.48702180e+00 4.24696863e-01 -1.43632904e-01 -1.14351392e+00 -7.70481944e-01 8.33031833e-01 -3.41759920e-01 -3.03107202e-01 1.58816472e-01 -3.17347705e-01 -1.78843737e-01 1.20356984e-01 5.26628077e-01 2.56638050e-01 3.57623786e-01 -3.26075852e-01 -4.06661898e-01 2.08187565e-01 -3.76501739e-01 -1.30863473e-01 1.15175569e+00 1.80515632e-01 -2.22147152e-01 -2.30495147e-02 1.16915309e+00 4.23003227e-01 -8.89900565e-01 -5.23422420e-01 6.77466929e-01 -2.77407408e-01 -3.95246387e-01 -7.49296010e-01 -9.99756575e-01 4.06650394e-01 -5.36453307e-01 4.50794131e-01 1.03963625e+00 2.11721539e-01 1.33516467e+00 4.46397543e-01 3.51351589e-01 -8.69502246e-01 -5.14393091e-01 5.34911156e-01 5.47463953e-01 -8.08587313e-01 6.80379346e-02 -1.16698432e+00 -5.53391695e-01 9.57863510e-01 3.81775856e-01 -3.37494500e-02 4.16937351e-01 7.35178590e-01 -5.20745218e-01 -1.14533909e-01 -1.10116279e+00 -4.43481266e-01 4.34915096e-01 3.40845704e-01 4.78882670e-01 1.99130729e-01 -6.34670496e-01 5.59030294e-01 -6.61235094e-01 -2.40768850e-01 4.15646642e-01 8.32364917e-01 -8.89757574e-02 -1.40036416e+00 1.34966627e-01 8.46620202e-01 -7.44487643e-01 -5.14121115e-01 -6.63648486e-01 7.28566110e-01 -2.43693233e-01 1.02770662e+00 -1.19327165e-01 -5.41033387e-01 3.66600782e-01 2.05799833e-01 6.47707820e-01 -8.05738866e-01 -7.12490380e-01 -2.29450688e-01 9.92641866e-01 -2.54186928e-01 9.51955318e-02 -7.27304220e-01 -1.47911799e+00 -3.29337865e-01 -5.39730847e-01 5.24471998e-01 3.95432442e-01 9.25989330e-01 5.03999114e-01 5.20721972e-01 4.35013980e-01 -1.79589823e-01 -4.56171572e-01 -1.12111282e+00 -4.84565377e-01 3.40668559e-01 7.73149878e-02 -2.58805960e-01 -1.14797845e-01 -1.64553806e-01]
[9.429912567138672, 8.657818794250488]
a95a6b43-ccd0-4ad4-a353-76f397d43267
unsupervised-pansharpening-via-low-rank
2305.10925
null
https://arxiv.org/abs/2305.10925v1
https://arxiv.org/pdf/2305.10925v1.pdf
Unsupervised Pansharpening via Low-rank Diffusion Model
Pansharpening is a process of merging a highresolution panchromatic (PAN) image and a low-resolution multispectral (LRMS) image to create a single high-resolution multispectral (HRMS) image. Most of the existing deep learningbased pansharpening methods have poor generalization ability and the traditional model-based pansharpening methods need careful manual exploration for the image structure prior. To alleviate these issues, this paper proposes an unsupervised pansharpening method by combining the diffusion model with the low-rank matrix factorization technique. Specifically, we assume that the HRMS image is decomposed into the product of two low-rank tensors, i.e., the base tensor and the coefficient matrix. The base tensor lies on the image field and has low spectral dimension, we can thus conveniently utilize a pre-trained remote sensing diffusion model to capture its image structures. Additionally, we derive a simple yet quite effective way to preestimate the coefficient matrix from the observed LRMS image, which preserves the spectral information of the HRMS. Extensive experimental results on some benchmark datasets demonstrate that our proposed method performs better than traditional model-based approaches and has better generalization ability than deep learning-based techniques. The code is released in https://github.com/xyrui/PLRDiff.
['Deyu Meng', 'Zongsheng Yue', 'Zeyu Zhu', 'Xiangyong Cao', 'Xiangyu Rui']
2023-05-18
null
null
null
null
['pansharpening']
['computer-vision']
[ 4.07732338e-01 -6.42669022e-01 -1.13803372e-01 5.47497720e-02 -6.66237175e-01 -3.83992732e-01 4.67354864e-01 -4.49923813e-01 -2.81471163e-01 3.84466082e-01 1.24432497e-01 -3.34654003e-01 -5.12739420e-01 -1.09229827e+00 -4.83687103e-01 -1.25049353e+00 1.16477363e-01 1.64793521e-01 -2.11181082e-02 -2.23371565e-01 -1.95093136e-02 5.60567796e-01 -1.04821920e+00 8.39154795e-02 1.31585932e+00 8.42462182e-01 6.79675102e-01 5.90581536e-01 1.91299289e-01 7.43574083e-01 -6.54251575e-02 7.52465725e-02 4.88716841e-01 -5.52413240e-02 -6.54990196e-01 4.79674727e-01 4.80387986e-01 -4.86697197e-01 -5.56909740e-01 1.55891812e+00 1.98141083e-01 3.64126742e-01 4.59128350e-01 -7.21853971e-01 -9.97470558e-01 5.09569764e-01 -1.37310028e+00 2.95485765e-01 -3.51513892e-01 -6.75798729e-02 8.92341316e-01 -8.79020572e-01 4.61599916e-01 9.99647021e-01 6.07853115e-01 -1.10460259e-01 -1.09159076e+00 -6.89936697e-01 3.15452814e-02 2.74099261e-01 -1.41684747e+00 1.01446055e-01 1.09459078e+00 -5.53797483e-01 5.46666265e-01 3.05753201e-01 6.31460905e-01 6.23835087e-01 1.89362109e-01 5.83824515e-01 1.42757368e+00 -1.37186840e-01 -1.91569775e-01 -3.47148091e-01 2.09462717e-01 6.46870255e-01 1.69844091e-01 3.47652972e-01 -1.28260463e-01 -2.28253201e-01 9.68123674e-01 3.94693732e-01 -5.94900608e-01 -3.81775260e-01 -1.24009037e+00 8.77462208e-01 8.68195295e-01 5.11875093e-01 -7.55486310e-01 -1.68940112e-01 7.74517329e-03 1.02484614e-01 6.16013110e-01 6.35563806e-02 -3.12648356e-01 4.46492940e-01 -1.10702336e+00 2.39306223e-02 1.33892924e-01 2.92390645e-01 1.26861560e+00 1.25803322e-01 5.20337164e-01 1.16552758e+00 3.54218721e-01 6.97749197e-01 4.26282167e-01 -9.69947815e-01 4.25752282e-01 4.18862492e-01 7.59285092e-02 -1.36820447e+00 -4.09625828e-01 -5.79128742e-01 -1.41604018e+00 -5.72385788e-02 -1.89515904e-01 -2.12262586e-01 -8.57331932e-01 1.32835150e+00 4.43451762e-01 2.75977850e-01 1.94852352e-02 1.17930067e+00 5.35948575e-01 1.24613857e+00 -3.06542546e-01 -3.08695138e-01 9.94914412e-01 -1.19300997e+00 -5.72049737e-01 -3.30707699e-01 3.94836605e-01 -8.76223445e-01 8.60668182e-01 5.97481608e-01 -5.05724490e-01 -4.15088266e-01 -1.12081909e+00 7.02889040e-02 -3.95141214e-01 5.12895882e-01 8.72308671e-01 4.64896888e-01 -8.33980381e-01 6.30860507e-01 -9.13008928e-01 -1.18795484e-01 9.04193223e-02 5.36667630e-02 -4.18038368e-01 -3.86769325e-01 -1.08194005e+00 5.21724045e-01 4.89789367e-01 5.36001027e-01 -8.46329451e-01 -6.35332882e-01 -5.27673960e-01 -2.22282097e-01 3.45651984e-01 -3.47150624e-01 8.26677561e-01 -1.02553594e+00 -1.19735801e+00 4.88710999e-01 1.57550946e-01 2.20884625e-02 2.57690549e-01 -4.50777382e-01 -6.45226836e-01 4.80094641e-01 1.47048384e-01 2.61712819e-01 1.15087855e+00 -1.55831766e+00 -4.53965217e-01 -4.34725255e-01 -5.92923164e-02 2.40628138e-01 -4.25441206e-01 -9.62610543e-02 -3.27210426e-01 -8.34258616e-01 5.93487620e-01 -9.46332276e-01 -4.79030609e-01 -1.90376207e-01 -4.00822967e-01 3.83339763e-01 1.08622205e+00 -1.13361263e+00 1.02098477e+00 -2.17617249e+00 3.33597124e-01 3.12433720e-01 2.22657681e-01 4.34013784e-01 -4.84636396e-01 4.55591023e-01 -4.79075462e-01 -9.44411755e-02 -6.81173921e-01 1.53761476e-01 -4.63937312e-01 1.75466299e-01 -3.95853490e-01 8.31216872e-01 -1.97694153e-01 5.49000502e-01 -8.71050656e-01 -1.89724445e-01 3.64308953e-01 7.63164580e-01 -1.38681710e-01 -5.40271178e-02 -6.11301437e-02 4.67496514e-01 -6.03688657e-01 6.42631888e-01 1.34927285e+00 -2.49594763e-01 9.75682139e-02 -7.02149153e-01 -4.34430480e-01 -3.31550032e-01 -1.27630222e+00 1.53997004e+00 -4.96639252e-01 3.43323529e-01 3.69765729e-01 -9.49016571e-01 8.30497861e-01 -6.41698837e-02 8.02756846e-01 -6.37192667e-01 -5.53961918e-02 1.57883391e-01 -1.46621898e-01 -2.88649261e-01 6.28630698e-01 -2.65836000e-01 5.86234450e-01 6.54885173e-01 -3.68534803e-01 -9.22023878e-02 1.01097152e-01 1.77100658e-01 4.38587993e-01 7.82995746e-02 -9.86247510e-02 -2.71203816e-01 6.43828988e-01 1.34563819e-01 6.07164323e-01 4.75740939e-01 1.84586402e-02 3.89222056e-01 -1.05251499e-01 -4.57074076e-01 -9.40770388e-01 -1.00235128e+00 -8.84226412e-02 8.45305800e-01 3.26040179e-01 -1.79536536e-01 -6.67172670e-01 -3.51812989e-01 -1.75611749e-01 3.60905081e-01 -4.57620054e-01 -6.18325584e-02 -4.44653541e-01 -1.52577448e+00 6.74648508e-02 2.87739277e-01 1.00235200e+00 -4.38818872e-01 -2.35766433e-02 1.37888297e-01 -4.13512707e-01 -8.78375292e-01 -3.49895865e-01 -3.54957938e-01 -1.02687311e+00 -1.04648507e+00 -7.62530923e-01 -5.43460131e-01 5.34495294e-01 1.10780382e+00 4.49707597e-01 -7.78475702e-02 -6.86237141e-02 2.87229270e-01 -3.58553082e-01 2.35580534e-01 -1.81525648e-01 -1.44063115e-01 -5.50482757e-02 6.01361394e-01 9.04854611e-02 -7.99299002e-01 -6.11617088e-01 2.30735376e-01 -1.43084025e+00 3.19014996e-01 1.04042721e+00 9.00360823e-01 9.35495794e-01 9.77510035e-01 9.11759958e-02 -5.42360425e-01 4.24604654e-01 -3.44104528e-01 -7.85174549e-01 2.31475532e-01 -5.94602287e-01 -2.85432637e-01 4.21953797e-01 -5.96972466e-01 -1.43085051e+00 2.86779523e-01 4.33400199e-02 -5.04509985e-01 1.72370434e-01 1.02199173e+00 -9.27902088e-02 -4.96121854e-01 5.31378746e-01 5.67330956e-01 -1.58008397e-01 -8.68467510e-01 6.27684593e-01 7.61900842e-01 5.17663956e-01 -4.10021424e-01 1.20747197e+00 9.93022919e-01 -1.83549628e-01 -1.43349528e+00 -1.01980233e+00 -6.03231549e-01 -6.88326955e-01 -2.97254920e-01 7.52117574e-01 -1.16535759e+00 -1.08482487e-01 8.29468966e-01 -8.14584672e-01 -5.11806428e-01 1.87148824e-01 6.98243797e-01 -1.84693605e-01 9.46077645e-01 -8.06536257e-01 -5.19066870e-01 -3.83676112e-01 -7.98486888e-01 8.95981848e-01 2.00276256e-01 5.18466115e-01 -8.16091537e-01 2.68238038e-01 7.19422877e-01 3.38756710e-01 6.55418634e-02 9.72147644e-01 1.75928771e-01 -7.65381873e-01 -1.16497137e-01 -6.63300157e-01 6.18947268e-01 2.62656331e-01 1.35241851e-01 -7.22349823e-01 -4.72060502e-01 4.17534649e-01 -9.31605473e-02 1.20565140e+00 5.94306290e-01 1.11392200e+00 -2.39676908e-01 -1.18779242e-01 9.57382321e-01 1.73509550e+00 1.74130930e-03 6.04016364e-01 5.81399024e-01 1.18810093e+00 4.53269005e-01 6.16781533e-01 2.69499868e-01 3.00595313e-01 2.70442903e-01 4.04589236e-01 -3.45189065e-01 -2.19074581e-02 -2.06846029e-01 3.33221078e-01 1.10614419e+00 -4.00727749e-01 1.40130088e-01 -9.11855459e-01 5.46896696e-01 -1.85674345e+00 -9.22596812e-01 -4.28984553e-01 1.95909703e+00 6.07976556e-01 -4.10060316e-01 -3.54249001e-01 4.32475805e-02 7.94808805e-01 8.31547737e-01 -6.16631567e-01 2.55474776e-01 -4.88344520e-01 -1.20184623e-01 8.69915307e-01 7.14692473e-01 -1.29725933e+00 1.02646184e+00 5.36854982e+00 8.28034759e-01 -1.42910004e+00 3.08141410e-01 2.43452430e-01 1.69754595e-01 -3.03009719e-01 -1.92424329e-03 -2.29765132e-01 5.03532663e-02 4.94912237e-01 -8.66374299e-02 6.68895185e-01 5.68449795e-01 4.28287864e-01 -1.58028111e-01 -1.48080677e-01 1.02347445e+00 -5.34043536e-02 -1.19391966e+00 2.57605612e-01 1.88785732e-01 1.04551554e+00 6.78143263e-01 2.72661686e-01 -1.87534720e-01 4.19095069e-01 -5.22137702e-01 3.51381242e-01 8.13608706e-01 5.64420760e-01 -5.88833153e-01 4.12124157e-01 3.52565676e-01 -1.28568339e+00 -3.32434773e-01 -6.60243034e-01 4.51923646e-02 -1.84629336e-02 1.10415208e+00 7.86480028e-03 9.18107390e-01 7.88205564e-01 1.17249358e+00 -4.42414463e-01 8.72440636e-01 -2.96808034e-01 5.51714540e-01 -2.86425084e-01 7.63512075e-01 3.94633025e-01 -1.04082036e+00 6.05147779e-01 9.79646921e-01 4.59113926e-01 2.81225383e-01 3.66635323e-01 8.47600758e-01 1.57643721e-01 1.34759486e-01 -4.30266470e-01 -3.51723552e-01 -5.99352457e-02 1.63526320e+00 -5.25941670e-01 -3.17983508e-01 -5.57895958e-01 8.82959545e-01 -8.99979100e-02 7.59646177e-01 -6.74894452e-01 -7.13315159e-02 5.64012110e-01 -2.92130142e-01 4.34319586e-01 -5.20395041e-01 -1.16179295e-01 -1.63475776e+00 -1.51809201e-01 -1.08983326e+00 3.69376868e-01 -1.14504564e+00 -1.18094552e+00 5.34822524e-01 -2.01781243e-01 -1.16467738e+00 3.11154604e-01 -5.42123258e-01 -4.22126740e-01 1.04308045e+00 -1.87811530e+00 -1.53092885e+00 -5.50679624e-01 6.84820712e-01 2.13444918e-01 7.32744560e-02 4.93666977e-01 3.41658056e-01 -8.84411752e-01 -3.60852867e-01 6.48878098e-01 1.96136702e-02 3.32374007e-01 -8.32478225e-01 7.81138167e-02 1.31191599e+00 3.13310437e-02 3.97812575e-01 5.07493615e-01 -8.18115950e-01 -1.49546647e+00 -1.37500143e+00 4.27809805e-01 2.91289777e-01 1.13258159e+00 2.34048754e-01 -1.11331844e+00 6.59958959e-01 1.19660445e-01 -2.02678829e-01 5.87606013e-01 -2.57755667e-01 -4.55949277e-01 -4.45067912e-01 -7.52716005e-01 4.49051023e-01 6.36373699e-01 -7.60656357e-01 -4.87135082e-01 6.08501792e-01 4.37786490e-01 -3.00325621e-02 -9.11530137e-01 5.20470977e-01 2.97776878e-01 -9.09849107e-01 9.69600737e-01 -1.19873591e-01 5.20650327e-01 -5.38258195e-01 -3.83889467e-01 -1.38243842e+00 -8.67205679e-01 -3.91391397e-01 7.84590319e-02 7.73597658e-01 1.81517646e-01 -6.38208508e-01 5.41343987e-01 1.93062052e-01 -2.37818528e-02 -4.72619593e-01 -4.02883172e-01 -7.28720009e-01 8.85776207e-02 -2.10136503e-01 5.26215971e-01 1.39712167e+00 -6.55244529e-01 1.94481730e-01 -6.38513029e-01 8.86230707e-01 9.67147112e-01 7.44754255e-01 5.21755397e-01 -1.14083993e+00 -3.57533842e-01 -4.17152077e-01 1.52590722e-01 -1.03173018e+00 5.50087392e-02 -7.03328311e-01 -1.75422609e-01 -1.60563314e+00 3.96403044e-01 -3.52231383e-01 -4.40305144e-01 4.45272028e-01 -3.70040722e-03 2.79499799e-01 -2.04631463e-02 5.81139982e-01 2.82381997e-02 7.13216007e-01 1.45971310e+00 -4.80921715e-01 -2.37429872e-01 -1.88670292e-01 -5.79548240e-01 7.85009503e-01 8.73718023e-01 -4.05632913e-01 -3.32037777e-01 -8.03941250e-01 2.56813318e-01 6.27799034e-02 4.79637355e-01 -7.77995169e-01 1.66108236e-01 -4.87912178e-01 1.54146329e-01 -6.99103296e-01 3.40272248e-01 -6.98627591e-01 4.35153574e-01 4.68183637e-01 1.77882329e-01 -3.08488995e-01 1.10970475e-01 5.58240712e-01 -3.73114318e-01 -5.44926599e-02 9.29017484e-01 -2.26967067e-01 -9.69386876e-01 7.47534633e-01 -2.71748334e-01 -6.98140562e-01 5.59307694e-01 8.00622851e-02 -4.96782511e-01 -3.09144497e-01 -6.51232183e-01 8.91803205e-03 4.54950213e-01 3.32387745e-01 6.42791152e-01 -1.22215903e+00 -6.86868250e-01 2.08154827e-01 -1.30233541e-01 -1.20448537e-01 8.90586019e-01 9.98999178e-01 -7.82490790e-01 1.46879539e-01 -3.42762679e-01 -4.57220882e-01 -9.91329014e-01 6.40753388e-01 6.74809992e-01 -1.87626511e-01 -7.28213072e-01 7.00743616e-01 5.62235534e-01 -6.33875251e-01 -4.63558078e-01 -1.51874438e-01 -1.74120471e-01 1.31212622e-01 7.03285158e-01 5.22389591e-01 -9.59431529e-02 -1.22408807e+00 -6.17576130e-02 9.23244417e-01 -3.11791003e-02 -6.88299537e-02 1.68975866e+00 -3.31601083e-01 -7.48584747e-01 8.54918361e-02 1.29686534e+00 4.61131334e-02 -1.30610657e+00 -5.81628621e-01 -2.42419928e-01 -7.00380385e-01 9.64057744e-01 -4.93024051e-01 -1.61449349e+00 9.48888063e-01 6.83957100e-01 1.93398237e-01 1.45364594e+00 -3.62634718e-01 9.01190042e-01 5.68711519e-01 9.15637463e-02 -1.14511657e+00 -8.10349733e-02 2.42145374e-01 9.58345354e-01 -1.19899106e+00 4.07776266e-01 -5.61083853e-01 -4.58867908e-01 1.06228626e+00 2.61373609e-01 -1.28113493e-01 8.30146611e-01 -2.33724624e-01 3.54413956e-01 -3.94668221e-01 -2.10586429e-01 -1.40067816e-01 2.69772619e-01 5.24286151e-01 -2.24789064e-02 3.26896578e-01 -6.31707981e-02 1.38772711e-01 -6.45720512e-02 -7.09067509e-02 5.77663958e-01 5.43805897e-01 -5.96830368e-01 -8.90463889e-01 -6.00822628e-01 3.00442994e-01 -1.83043510e-01 -1.63364232e-01 -2.68378973e-01 4.19461936e-01 6.67023808e-02 8.46775234e-01 -2.16107562e-01 -4.52116668e-01 3.62322293e-03 -4.15935189e-01 1.68919876e-01 -2.88790435e-01 4.86044064e-02 5.08193552e-01 -2.71767318e-01 -5.10527015e-01 -8.48678172e-01 -6.46765888e-01 -8.10757160e-01 -3.21265012e-01 -4.29003596e-01 1.17360979e-01 6.26851618e-01 9.30749774e-01 8.91739950e-02 5.92903048e-02 8.51770520e-01 -1.04714012e+00 -4.04704154e-01 -8.44018698e-01 -1.25327349e+00 8.73008594e-02 5.75361311e-01 -5.25869727e-01 -2.98068970e-01 -3.73009034e-02]
[10.195398330688477, -1.9208978414535522]
074f0690-d17e-46bb-aaaa-a06ef5fba61b
convolutional-oriented-boundaries-from-image
1701.04658
null
http://arxiv.org/abs/1701.04658v2
http://arxiv.org/pdf/1701.04658v2.pdf
Convolutional Oriented Boundaries: From Image Segmentation to High-Level Tasks
We present Convolutional Oriented Boundaries (COB), which produces multiscale oriented contours and region hierarchies starting from generic image classification Convolutional Neural Networks (CNNs). COB is computationally efficient, because it requires a single CNN forward pass for multi-scale contour detection and it uses a novel sparse boundary representation for hierarchical segmentation; it gives a significant leap in performance over the state-of-the-art, and it generalizes very well to unseen categories and datasets. Particularly, we show that learning to estimate not only contour strength but also orientation provides more accurate results. We perform extensive experiments for low-level applications on BSDS, PASCAL Context, PASCAL Segmentation, and NYUD to evaluate boundary detection performance, showing that COB provides state-of-the-art contours and region hierarchies in all datasets. We also evaluate COB on high-level tasks when coupled with multiple pipelines for object proposals, semantic contours, semantic segmentation, and object detection on MS-COCO, SBD, and PASCAL; showing that COB also improves the results for all tasks.
['Luc van Gool', 'Pablo Arbeláez', 'Jordi Pont-Tuset', 'Kevis-Kokitsi Maninis']
2017-01-17
null
null
null
null
['contour-detection']
['computer-vision']
[ 1.58572838e-01 1.33791283e-01 -2.57514138e-02 -3.73979867e-01 -9.82383847e-01 -7.90253103e-01 3.84801477e-01 3.52960110e-01 -4.87943739e-01 1.51142821e-01 -2.21937388e-01 -2.28802964e-01 5.31336963e-01 -8.75946879e-01 -7.49766946e-01 -3.02908808e-01 -3.13853383e-01 6.39019907e-01 1.38600993e+00 -1.93242535e-01 2.21393824e-01 7.86101460e-01 -1.35912800e+00 4.63004380e-01 4.59615797e-01 1.34461498e+00 -8.26751217e-02 9.99399364e-01 -3.25940013e-01 4.98828530e-01 -5.61843991e-01 -2.68162727e-01 2.06467971e-01 -6.95698485e-02 -1.01388323e+00 2.42378131e-01 1.24587286e+00 -8.21593404e-02 3.10274154e-01 9.67540264e-01 3.47674042e-01 -1.90517709e-01 8.24341655e-01 -1.07090318e+00 -4.68212694e-01 3.01549673e-01 -7.90932715e-01 2.45627925e-01 5.06183729e-02 -4.49340604e-02 9.06631172e-01 -1.04590333e+00 8.93591642e-01 1.56435156e+00 1.36259770e+00 5.66605449e-01 -1.39761543e+00 -3.28198969e-01 2.91176319e-01 -3.88758093e-01 -1.03801787e+00 1.06335379e-01 3.55861396e-01 -6.15680814e-01 9.92096424e-01 1.38990255e-02 8.34531486e-01 5.09231150e-01 5.06965742e-02 1.23514211e+00 8.50157440e-01 -3.75801921e-01 3.74361813e-01 -3.85370255e-01 4.08939719e-01 1.00065136e+00 3.10408771e-01 2.33844630e-02 -3.11820298e-01 2.01555118e-01 1.07538044e+00 -5.67700744e-01 -4.06717658e-02 -6.16180539e-01 -1.18004715e+00 7.38515496e-01 6.92629337e-01 3.04232925e-01 -4.74159233e-02 4.14524138e-01 4.17492181e-01 -1.16490781e-01 7.37979293e-01 2.21114367e-01 -5.44190168e-01 2.93556213e-01 -1.27740479e+00 4.86672640e-01 8.75618696e-01 1.09107316e+00 7.65041709e-01 -3.29307863e-03 -2.79118836e-01 8.43864977e-01 9.82039347e-02 4.14999813e-01 6.46063164e-02 -1.26979065e+00 1.63315549e-01 5.89339375e-01 -4.56456877e-02 -8.17756772e-01 -9.82434332e-01 -3.55020136e-01 -5.21355987e-01 7.26982534e-01 7.98042476e-01 -1.45053953e-01 -1.78971195e+00 1.20095217e+00 3.65940511e-01 4.50464115e-02 -1.66978374e-01 9.20717835e-01 1.23616171e+00 5.28011680e-01 2.99019486e-01 4.40713853e-01 1.51688397e+00 -1.18428659e+00 -4.07879472e-01 -6.28305614e-01 5.12849212e-01 -9.11547065e-01 8.48326623e-01 4.33030486e-01 -1.38072228e+00 -8.32683265e-01 -1.03601277e+00 -4.92239892e-01 -5.96843243e-01 9.11454484e-02 1.00958765e+00 6.07440472e-01 -1.55737972e+00 7.72288442e-01 -9.63759422e-01 -5.01599669e-01 8.89097333e-01 6.52075708e-02 -1.30102202e-01 -4.59290333e-02 -6.40305519e-01 6.04213595e-01 7.08866000e-01 7.77661353e-02 -7.65545011e-01 -7.16850817e-01 -1.21073318e+00 8.71714763e-03 2.59901941e-01 -4.27504599e-01 1.31369090e+00 -6.78740263e-01 -1.00221324e+00 1.22013128e+00 1.41058967e-01 -5.51467836e-01 7.81074107e-01 -1.66525021e-01 4.08286713e-02 4.82003778e-01 4.51818973e-01 1.87101769e+00 5.77371001e-01 -1.20407939e+00 -1.06216168e+00 -1.05071001e-01 -2.65166163e-01 -3.26442495e-02 2.11594045e-01 -2.64173113e-02 -9.40041542e-01 -8.45845282e-01 5.38947165e-01 -6.88369811e-01 -3.78372997e-01 5.02376676e-01 -5.34169495e-01 -5.28358817e-01 1.18359363e+00 -6.35985017e-01 5.55360436e-01 -2.06850028e+00 -1.63731173e-01 2.47438792e-02 -3.25833932e-02 1.71183258e-01 -2.49704331e-01 -2.54196554e-01 7.93554112e-02 4.30263162e-01 -8.80798936e-01 -6.25692189e-01 -2.53838241e-01 1.26686797e-01 2.40210041e-01 2.57458448e-01 6.62178874e-01 1.12769961e+00 -7.17402637e-01 -1.10355937e+00 2.51813740e-01 3.94796580e-01 -6.75847113e-01 9.85448286e-02 -4.97630835e-01 1.10988550e-01 -1.58029031e-02 1.13680828e+00 9.75057542e-01 -3.64813656e-01 -1.96930587e-01 -2.63439953e-01 -3.36685628e-01 -3.41634095e-01 -1.18344188e+00 1.70273900e+00 7.90051371e-02 9.35521185e-01 3.82246912e-01 -7.63432443e-01 8.63688529e-01 1.00156873e-01 2.76613176e-01 -5.52990019e-01 -3.09327170e-02 3.45858067e-01 -4.15807277e-01 -7.66821951e-02 4.62053746e-01 3.14993411e-02 -1.09635562e-01 -9.20403823e-02 4.16516662e-01 -9.28304076e-01 5.97590864e-01 1.51132032e-01 6.87833011e-01 4.51961100e-01 1.11306407e-01 -7.14424610e-01 2.35103905e-01 5.37900746e-01 5.30835390e-01 8.46998394e-01 -4.40914094e-01 1.23750722e+00 7.04853594e-01 -6.51104569e-01 -1.00426328e+00 -1.29799557e+00 -4.71997708e-01 1.18943954e+00 4.02308941e-01 1.06053583e-01 -1.13809824e+00 -7.40614474e-01 1.93744063e-01 3.20460647e-01 -7.65643179e-01 5.78004420e-01 -7.04809725e-01 -7.35274673e-01 4.95487213e-01 1.03157985e+00 9.75707054e-01 -1.52050388e+00 -5.04572034e-01 2.19092503e-01 9.62393135e-02 -1.36732900e+00 -5.93029141e-01 4.00640219e-01 -1.06278706e+00 -1.24820876e+00 -1.08872664e+00 -1.40341687e+00 4.24914807e-01 6.55405819e-02 1.44160378e+00 9.85613242e-02 -7.43423760e-01 1.23252548e-01 -1.28837228e-01 -3.42059880e-01 -4.13657933e-01 1.63157955e-01 -8.30159605e-01 -4.56900239e-01 -6.98527843e-02 3.95204164e-02 -6.35674298e-01 3.18442672e-01 -8.65502238e-01 9.68341809e-03 3.77061784e-01 6.81703031e-01 8.06486666e-01 -2.02747747e-01 4.03846681e-01 -1.05927205e+00 2.26521611e-01 1.01236463e-01 -8.50624263e-01 1.51618972e-01 -3.00177187e-01 -2.49957502e-01 -3.82431373e-02 -8.80052671e-02 -1.08025265e+00 3.60108614e-01 -4.72633630e-01 -1.10944480e-01 -6.59576833e-01 -4.63411678e-03 2.22449318e-01 -1.63868383e-01 7.33973861e-01 -2.42210031e-01 -2.01708078e-01 -6.50454700e-01 7.58251548e-01 4.08802181e-01 1.01468921e+00 -5.01032770e-01 4.14514571e-01 8.25277269e-01 -1.78218216e-01 -9.56305325e-01 -1.02517796e+00 -9.17375743e-01 -9.78478312e-01 -1.30515203e-01 1.49506557e+00 -9.08226967e-01 -5.25084853e-01 9.49905515e-01 -1.34092116e+00 -9.00333583e-01 -2.53114730e-01 -1.63602874e-01 -6.02604926e-01 2.89777040e-01 -1.13589823e+00 -5.38239002e-01 -4.55144823e-01 -1.10558546e+00 1.63267934e+00 4.61522758e-01 -1.29982186e-02 -1.11801553e+00 -2.78013587e-01 2.35673249e-01 2.37316743e-01 6.55928195e-01 7.64220178e-01 -3.69926870e-01 -5.18983662e-01 -4.10303753e-03 -8.68066967e-01 4.11365092e-01 -1.97790772e-01 2.33671114e-01 -9.64137435e-01 -3.30941647e-01 -7.64854372e-01 -5.85442722e-01 1.47357619e+00 6.50287569e-01 1.12123406e+00 1.77009597e-01 -5.13084590e-01 7.33623385e-01 1.35886562e+00 -2.70769689e-02 4.47026074e-01 8.23509619e-02 7.17030704e-01 6.28159940e-01 5.32477558e-01 -1.72132283e-01 1.64055943e-01 3.81155491e-01 4.71741170e-01 -9.40898538e-01 -7.58298278e-01 2.03375518e-01 -4.11650017e-02 1.68118134e-01 8.56281966e-02 1.03197014e-02 -1.03647876e+00 9.45528805e-01 -1.63380623e+00 -4.85410511e-01 -5.42348921e-01 1.50829685e+00 9.53991115e-01 7.43192732e-01 3.83399367e-01 -7.28596151e-02 9.30026472e-01 9.96788889e-02 -5.48238695e-01 -4.58735496e-01 -3.27503890e-01 6.67073607e-01 6.60462618e-01 6.86607957e-01 -1.75986218e+00 1.58175075e+00 7.54190636e+00 9.15855289e-01 -1.15673983e+00 1.11336512e-02 1.09478164e+00 6.21306002e-01 6.22935221e-02 -2.33729213e-01 -8.43107760e-01 -1.75791994e-01 1.95827618e-01 5.95665574e-01 -2.01540127e-01 1.18826020e+00 -4.01154131e-01 -3.46044302e-01 -8.92719090e-01 5.33924878e-01 -1.56445041e-01 -1.71275890e+00 -1.20904937e-01 -3.83934587e-01 9.42770064e-01 5.32662630e-01 -3.38769197e-01 2.53395230e-01 4.84037012e-01 -9.36132967e-01 1.18259406e+00 -4.66358922e-02 8.40940773e-01 -4.64442939e-01 7.86399424e-01 8.07742402e-02 -1.57606030e+00 2.04013854e-01 -3.81437927e-01 2.55846471e-01 1.28662080e-01 4.52842236e-01 -7.67363787e-01 3.43490243e-02 1.11078560e+00 7.26837039e-01 -1.03657079e+00 1.20994890e+00 -3.09515506e-01 6.28980756e-01 -5.44215798e-01 1.58664674e-01 6.11544967e-01 1.65077299e-01 7.37696663e-02 1.95829713e+00 -1.90557688e-01 -3.12956125e-02 5.93218029e-01 1.08303642e+00 -4.15217392e-02 -2.75079552e-02 -1.94407389e-01 3.19785804e-01 2.65318930e-01 1.54218745e+00 -1.77375150e+00 -5.71447492e-01 -1.69707805e-01 9.14399564e-01 3.06600511e-01 3.76947314e-01 -5.47059476e-01 -4.06579256e-01 3.26165885e-01 5.91102354e-02 6.49331689e-01 -3.80701661e-01 -5.04517198e-01 -6.77989781e-01 -4.13695306e-01 -3.89064103e-01 6.19175315e-01 -7.88180292e-01 -1.25153911e+00 5.53007662e-01 -6.08334430e-02 -6.01197958e-01 1.45707428e-01 -9.52295780e-01 -6.51932359e-01 5.82395911e-01 -1.67706621e+00 -1.39360321e+00 -5.49890637e-01 2.63303846e-01 7.85525918e-01 5.53544521e-01 6.66836381e-01 -2.33421717e-02 -3.21906626e-01 2.34534755e-01 -4.33753669e-01 7.82496512e-01 3.89268786e-01 -1.82011509e+00 1.24446440e+00 8.14300954e-01 1.67541891e-01 -4.08884995e-02 1.81840584e-01 -7.92521834e-01 -4.16687012e-01 -1.30062354e+00 4.24311101e-01 -2.88993627e-01 2.73882300e-01 -5.59033394e-01 -9.73449171e-01 5.39461672e-01 2.36865327e-01 6.42856121e-01 -4.39256914e-02 -4.77516651e-02 -2.36580104e-01 3.71143073e-01 -1.21534085e+00 5.20147204e-01 1.09180200e+00 -1.21708214e-01 -5.04417241e-01 4.10005569e-01 9.95114684e-01 -8.17452192e-01 -7.48606861e-01 6.77229166e-01 3.57317179e-01 -1.35801184e+00 1.21183586e+00 -3.65153879e-01 1.88776046e-01 -3.07258546e-01 1.77333385e-01 -8.96001637e-01 -1.74199671e-01 -3.26279819e-01 3.35214227e-01 1.09980202e+00 5.43200970e-01 -3.53957295e-01 1.24907875e+00 1.10885896e-01 -3.58785450e-01 -8.75852287e-01 -8.91748071e-01 -6.65618479e-01 3.64705384e-01 -5.44508338e-01 9.51678306e-02 6.97109044e-01 -5.78170478e-01 7.96494484e-02 5.32371283e-01 1.75902933e-01 8.56216967e-01 3.64324868e-01 5.13564706e-01 -1.24019527e+00 1.66448101e-01 -7.40680754e-01 -5.45956671e-01 -1.22265518e+00 -1.22990124e-01 -8.10802877e-01 4.02991831e-01 -1.95942605e+00 -1.74763024e-01 -6.44336700e-01 8.04354325e-02 8.52841854e-01 -3.52899730e-01 9.20121849e-01 2.78818816e-01 3.90859358e-02 -8.11797559e-01 -1.23911932e-01 1.55632603e+00 -2.68274695e-01 -9.39738154e-02 7.16349436e-03 -2.01403931e-01 1.16125858e+00 5.86178839e-01 -3.23218942e-01 1.15541771e-01 -2.73688346e-01 -1.66178927e-01 -4.01470400e-02 5.38345516e-01 -1.39618862e+00 1.37697890e-01 1.35475785e-01 6.60790682e-01 -1.10000205e+00 2.21650109e-01 -3.47758472e-01 -7.04821825e-01 5.00959754e-01 -1.28113821e-01 -2.99439877e-01 7.12802112e-01 4.36438709e-01 -3.00241977e-01 -1.43024266e-01 1.24351752e+00 -4.17930543e-01 -1.24340236e+00 1.57114789e-01 -2.69656271e-01 6.05059445e-01 8.44937384e-01 -5.74241877e-01 -1.55509800e-01 -3.87676293e-03 -1.12687361e+00 4.92226005e-01 4.87747759e-01 4.13396031e-01 5.24220705e-01 -9.70463812e-01 -6.68502688e-01 1.47082567e-01 5.08970581e-03 8.89032960e-01 -7.15843663e-02 3.32714438e-01 -1.40437758e+00 2.90028542e-01 -7.78367966e-02 -1.23631179e+00 -1.03646457e+00 2.67532617e-01 6.99298143e-01 -7.22699761e-02 -7.24393606e-01 1.42441928e+00 3.58235121e-01 -4.85447109e-01 4.15758431e-01 -8.94220829e-01 -4.10652235e-02 2.49865994e-01 1.92546219e-01 1.67236462e-01 2.25102961e-01 -3.23938072e-01 -5.25709391e-01 1.05297065e+00 1.00569561e-01 7.19325384e-03 1.00819647e+00 1.84897438e-01 2.70494241e-02 8.35944936e-02 1.04382885e+00 -3.63983840e-01 -1.82186735e+00 1.51585117e-01 5.44445753e-01 8.87514800e-02 1.84423447e-01 -8.91028225e-01 -1.24756908e+00 1.13045144e+00 6.79124057e-01 2.00205475e-01 8.17710996e-01 2.11285263e-01 8.12510550e-01 2.25092143e-01 3.46825123e-01 -1.14354300e+00 3.60092074e-01 5.31474292e-01 7.27454603e-01 -1.39584291e+00 2.32844567e-03 -1.06186116e+00 -4.06387269e-01 1.32225811e+00 7.20435262e-01 -2.44438663e-01 8.25891972e-01 5.77062249e-01 5.50042808e-01 -3.73265237e-01 -6.67367131e-02 -6.62054121e-01 3.90772730e-01 7.13240623e-01 4.79983598e-01 7.44912773e-02 3.50032412e-02 5.31556681e-02 8.92344937e-02 -2.34664008e-01 3.85967642e-01 9.56881285e-01 -8.21265817e-01 -7.90711999e-01 -4.86914992e-01 1.12291507e-01 -5.06893635e-01 -1.91059753e-01 -5.71228921e-01 1.29360378e+00 3.72725815e-01 8.09383988e-01 4.96406823e-01 3.33717823e-01 3.82855326e-01 9.07819569e-02 3.56126100e-01 -8.64687562e-01 -7.76495516e-01 3.61654043e-01 1.30724281e-01 -8.78619194e-01 -3.87154043e-01 -4.94957030e-01 -1.80368733e+00 1.25323161e-01 -4.58336234e-01 -2.69948561e-02 6.90311730e-01 6.49743259e-01 5.47416694e-02 6.43676221e-01 -1.28172621e-01 -1.37429786e+00 2.15388350e-02 -9.81371999e-01 -4.86534536e-01 6.15083992e-01 2.40943313e-01 -4.45295841e-01 -1.80192083e-01 3.32316756e-01]
[9.473952293395996, 0.1971074342727661]
e96283b6-8475-41ed-a5a4-e75ecd88f553
discrete-simulation-optimization-for-tuning
2201.05978
null
https://arxiv.org/abs/2201.05978v3
https://arxiv.org/pdf/2201.05978v3.pdf
Discrete Simulation Optimization for Tuning Machine Learning Method Hyperparameters
Machine learning (ML) methods are used in most technical areas such as image recognition, product recommendation, financial analysis, medical diagnosis, and predictive maintenance. An important aspect of implementing ML methods involves controlling the learning process for the ML method so as to maximize the performance of the method under consideration. Hyperparameter tuning is the process of selecting a suitable set of ML method parameters that control its learning process. In this work, we demonstrate the use of discrete simulation optimization methods such as ranking and selection (R&S) and random search for identifying a hyperparameter set that maximizes the performance of a ML method. Specifically, we use the KN R&S method and the stochastic ruler random search method and one of its variations for this purpose. We also construct the theoretical basis for applying the KN method, which determines the optimal solution with a statistical guarantee via solution space enumeration. In comparison, the stochastic ruler method asymptotically converges to global optima and incurs smaller computational overheads. We demonstrate the application of these methods to a wide variety of machine learning models, including deep neural network models used for time series prediction and image classification. We benchmark our application of these methods with state-of-the-art hyperparameter optimization libraries such as $hyperopt$ and $mango$. The KN method consistently outperforms $hyperopt$'s random search (RS) and Tree of Parzen Estimators (TPE) methods. The stochastic ruler method outperforms the $hyperopt$ RS method and offers statistically comparable performance with respect to $hyperopt$'s TPE method and the $mango$ algorithm.
['Nomesh Bhojkumar Bolia', 'Aditya Raj Gupta', 'Shobhit Singhal', 'Varun Ramamohan']
2022-01-16
null
null
null
null
['product-recommendation', 'time-series-prediction']
['miscellaneous', 'time-series']
[ 1.06510716e-02 -2.40025759e-01 -5.17668426e-01 -9.04806182e-02 -8.68646562e-01 -3.62108111e-01 2.20551819e-01 6.43622829e-03 -5.13942897e-01 6.86405361e-01 -6.78941071e-01 -4.50823605e-01 -6.67205095e-01 -7.45090783e-01 -6.19117498e-01 -9.94339883e-01 -1.74539968e-01 6.40229762e-01 -1.45946577e-01 1.27671018e-01 4.17715430e-01 5.29123783e-01 -1.44321024e+00 -1.34985015e-01 7.10624158e-01 1.51649976e+00 1.23242065e-01 5.82975864e-01 6.71077967e-02 3.39623421e-01 -6.19763255e-01 -2.61101067e-01 4.58691001e-01 -2.80180275e-01 -5.17067194e-01 -1.57371998e-01 8.02453086e-02 1.31470129e-01 5.87641895e-02 8.39604974e-01 7.08604097e-01 4.86786366e-01 8.06416333e-01 -1.29530942e+00 1.12395577e-01 5.07447541e-01 -8.32821012e-01 2.84894973e-01 -1.60766855e-01 2.42108047e-01 1.04623163e+00 -7.13831306e-01 3.13651651e-01 1.09261644e+00 7.58319736e-01 2.74814039e-01 -1.31446743e+00 -9.11060929e-01 -8.80954042e-02 1.04984954e-01 -1.78197241e+00 -3.17386061e-01 6.52309358e-01 -4.72748190e-01 1.19521165e+00 1.92849904e-01 5.32122850e-01 3.15018594e-01 4.80077207e-01 8.86716843e-01 8.73775423e-01 -5.96529007e-01 6.87338948e-01 2.48781711e-01 1.49761260e-01 9.51569200e-01 1.62836194e-01 3.73065203e-01 -5.04713953e-01 -5.65787315e-01 7.45253503e-01 -2.24278405e-01 -6.32184595e-02 -1.78252086e-01 -8.73028755e-01 1.02902567e+00 2.28364803e-02 -8.84231329e-02 -5.96063077e-01 6.26194715e-01 2.81994104e-01 1.90007612e-01 2.99358368e-01 9.13772702e-01 -8.75560760e-01 -8.89171660e-02 -9.98950243e-01 3.84961069e-01 9.47760880e-01 7.35171020e-01 6.04488492e-01 3.16967160e-01 -5.79015873e-02 1.19191086e+00 2.19589084e-01 3.85004491e-01 6.55704916e-01 -1.24497759e+00 1.44302681e-01 3.78799677e-01 1.22962043e-01 -8.32866549e-01 -4.11250681e-01 -7.05447853e-01 -7.63782918e-01 1.37293220e-01 1.48925141e-01 -1.94864213e-01 -7.18413949e-01 1.59831607e+00 4.12759542e-01 1.61981791e-01 -3.79929692e-02 3.65719676e-01 4.22207236e-01 8.73041213e-01 -8.75607207e-02 -7.27378309e-01 1.05811942e+00 -7.13812649e-01 -4.20423061e-01 -1.04392961e-01 5.64172029e-01 -6.81788206e-01 9.60375726e-01 7.05769718e-01 -1.13232839e+00 -2.86668420e-01 -1.07734931e+00 8.03337276e-01 -1.54191405e-01 2.79994190e-01 6.45650089e-01 6.69417143e-01 -8.34380746e-01 7.52266407e-01 -1.01780808e+00 -1.06083825e-01 3.47498208e-01 8.18851054e-01 1.97257310e-01 3.13053012e-01 -1.00223207e+00 8.40360940e-01 4.86208498e-01 1.46261036e-01 -7.39153147e-01 -8.30435455e-01 -6.52166426e-01 3.33550312e-02 5.18802345e-01 -5.74432969e-01 1.45958579e+00 -8.50796461e-01 -1.82577860e+00 5.74719250e-01 -1.91793069e-01 -7.10758388e-01 3.51549000e-01 -3.06566339e-02 -3.28661278e-02 -1.21194087e-02 -3.34739864e-01 4.46737558e-01 1.05196285e+00 -8.74936879e-01 -5.78905225e-01 -8.40537101e-02 -4.47448164e-01 1.00724965e-01 -2.07864583e-01 -1.34398550e-01 -4.98751223e-01 -5.87160349e-01 7.99074844e-02 -1.06317806e+00 -5.76365292e-01 -3.54440778e-01 -5.38139224e-01 -2.94025689e-01 5.40529966e-01 -2.94171393e-01 1.69929814e+00 -1.86342824e+00 -3.26007083e-02 8.09515119e-01 -5.03853597e-02 2.16148809e-01 7.91876093e-02 4.01093483e-01 -1.66598409e-02 2.28679523e-01 -1.40221059e-01 5.58930524e-02 -1.02677867e-01 5.59794307e-02 -9.08968151e-02 6.71269357e-01 -7.25029185e-02 6.47366464e-01 -3.93817663e-01 -5.30837893e-01 2.08966702e-01 9.13527161e-02 -7.91621327e-01 8.41630250e-02 -4.65751857e-01 -1.92230374e-01 -5.72291195e-01 6.42592490e-01 9.75251123e-02 -6.83476865e-01 3.45560133e-01 -3.51656914e-01 7.48779699e-02 -4.21076156e-02 -1.62377095e+00 7.93438911e-01 -6.45102561e-01 5.17480433e-01 -1.59032598e-01 -1.15634680e+00 9.78354812e-01 2.06229120e-01 7.49911189e-01 -2.89287210e-01 3.08519781e-01 3.72287363e-01 -1.30040839e-01 -3.34636152e-01 2.06745133e-01 4.01321463e-02 8.40892643e-02 4.96520340e-01 -1.05333515e-01 -2.25335732e-01 1.32627606e-01 -1.72961876e-01 9.92079794e-01 -2.51168221e-01 7.72884727e-01 -3.94127280e-01 4.43667680e-01 -9.00831670e-02 5.22067428e-01 1.00734341e+00 -1.58703148e-01 1.31170705e-01 3.71533453e-01 -3.32998395e-01 -8.02469254e-01 -8.06303144e-01 -2.96604782e-01 8.69590938e-01 -8.77807811e-02 -9.52613279e-02 -6.81832016e-01 -3.62713665e-01 1.57230154e-01 8.33003283e-01 -3.53312194e-01 -1.62748635e-01 -6.90660834e-01 -1.15443540e+00 2.34820768e-01 2.22235948e-01 4.52022165e-01 -1.10105669e+00 -8.07401419e-01 3.72534126e-01 9.63882655e-02 -7.56572068e-01 -5.17442226e-01 3.37864697e-01 -1.02833688e+00 -9.34816539e-01 -3.64236981e-01 -3.91569704e-01 5.26164591e-01 -4.10439879e-01 1.03322828e+00 -5.56880422e-02 -5.78031301e-01 3.30369681e-01 -3.91553827e-02 -4.47657257e-01 -4.40339267e-01 2.89398938e-01 1.67976230e-01 -2.56226887e-03 2.00129122e-01 -3.24390292e-01 -7.75716424e-01 5.03868878e-01 -6.44449592e-01 -4.12603974e-01 5.99760056e-01 1.11751366e+00 1.02950227e+00 6.37737811e-01 6.68276310e-01 -7.60448277e-01 8.06403637e-01 -3.93511862e-01 -1.25131989e+00 3.30472767e-01 -1.49350214e+00 4.11619306e-01 5.67035079e-01 -7.73209453e-01 -5.83726168e-01 2.53641456e-01 1.73824102e-01 -8.09063435e-01 4.53105599e-01 7.06260204e-01 8.33540186e-02 -2.10795641e-01 6.84225082e-01 6.34074137e-02 2.18474090e-01 -3.56348038e-01 9.82199982e-03 6.16255879e-01 1.82104886e-01 -5.67063570e-01 4.17469233e-01 6.12780377e-02 2.34795555e-01 -8.37320507e-01 -5.27047217e-01 -4.46374178e-01 8.52802396e-03 -3.05192292e-01 4.83814299e-01 -4.67846006e-01 -1.19587243e+00 4.34659690e-01 -8.64298642e-01 -4.95968223e-01 -2.14295939e-01 5.31118572e-01 -8.94609094e-01 -1.18634023e-01 -2.90580809e-01 -1.16604996e+00 -5.80767155e-01 -1.35537183e+00 8.85293424e-01 3.31139177e-01 -3.71072173e-01 -8.91116679e-01 -1.22161925e-01 1.50269851e-01 2.32893676e-01 3.11151564e-01 1.20316398e+00 -8.03164244e-01 -5.82692742e-01 -4.76709038e-01 2.71181405e-01 3.57050747e-01 -2.44011939e-01 2.68232465e-01 -5.32526195e-01 -3.93704772e-01 -5.98581322e-02 9.64120552e-02 5.79355955e-01 1.17992425e+00 1.46101761e+00 -5.90000927e-01 -5.43393791e-01 7.71524131e-01 1.81433427e+00 8.14715803e-01 4.06469107e-01 5.50572991e-01 4.15285677e-02 4.11328942e-01 6.69444621e-01 7.09460080e-01 -1.40071377e-01 7.16837585e-01 2.16551289e-01 1.74580485e-01 5.18934548e-01 9.63364765e-02 1.21856533e-01 3.73376429e-01 1.16287477e-01 -2.36343101e-01 -1.05063081e+00 2.85005987e-01 -1.96109545e+00 -9.14310634e-01 4.39904362e-01 2.64447999e+00 9.56090212e-01 1.77637637e-01 1.30869880e-01 1.96841031e-01 7.21867800e-01 -1.57910272e-01 -8.85156751e-01 -6.04080200e-01 3.13315421e-01 4.75200653e-01 8.93828034e-01 4.18049634e-01 -9.98873472e-01 7.60647058e-01 6.71766758e+00 1.30216110e+00 -1.07261050e+00 -2.21016645e-01 8.82627308e-01 -4.22516018e-01 7.71633163e-02 -2.96588570e-01 -1.25275445e+00 3.60622078e-01 1.18313384e+00 -2.98131526e-01 8.39705527e-01 1.16694891e+00 5.11207104e-01 -2.71529943e-01 -1.07587850e+00 1.26469505e+00 -3.86182904e-01 -1.64390504e+00 -2.18816966e-01 6.89240694e-02 7.70767093e-01 -1.30815253e-01 2.52937347e-01 1.79757506e-01 4.18484390e-01 -1.19172919e+00 4.76134777e-01 3.30384880e-01 6.42135262e-01 -1.02424204e+00 4.96249646e-01 3.34974587e-01 -9.80639756e-01 -3.89408559e-01 -5.22524230e-02 4.14682150e-01 -2.83506117e-03 9.37796474e-01 -9.77660120e-01 8.63206908e-02 7.45589435e-01 1.34762049e-01 6.81063684e-04 1.04291797e+00 2.57422447e-01 7.62287438e-01 -5.67532003e-01 -5.32046199e-01 7.05253705e-02 -2.01045319e-01 5.95485866e-01 9.79855120e-01 1.81488350e-01 1.24485604e-01 5.92424273e-02 6.70651317e-01 4.72835489e-02 2.17477232e-01 -2.19575956e-01 -2.71363080e-01 9.51102912e-01 8.17305326e-01 -7.22570956e-01 -1.36646450e-01 1.31471053e-01 2.58663118e-01 7.37863109e-02 4.31221545e-01 -7.37565219e-01 -7.06636488e-01 5.53974092e-01 -1.62505440e-03 4.72973764e-01 -6.88199773e-02 -4.55026716e-01 -5.54687679e-01 -2.04994321e-01 -1.12177813e+00 6.70143902e-01 -4.44711208e-01 -1.07972741e+00 5.65673232e-01 4.41635281e-01 -1.18147242e+00 -6.88441157e-01 -6.35003269e-01 -2.30341986e-01 9.06189561e-01 -1.24007678e+00 -2.02631518e-01 1.27577931e-01 3.88700128e-01 6.25675917e-01 -4.79451656e-01 4.44157898e-01 -7.21467361e-02 -9.86102343e-01 8.75739217e-01 4.62199539e-01 -4.05045301e-01 2.38657430e-01 -9.37979996e-01 -2.17612818e-01 4.81418937e-01 -1.96553901e-01 6.47820830e-01 1.03964794e+00 -3.79073590e-01 -1.48468709e+00 -9.82287347e-01 5.19304693e-01 1.51351780e-01 5.18635094e-01 1.90331444e-01 -5.35502017e-01 2.63257086e-01 -2.90138543e-01 -1.61678508e-01 5.69310308e-01 1.65892497e-01 4.59489226e-02 -5.92359722e-01 -1.28914487e+00 7.10167170e-01 3.85662466e-01 -1.18006565e-01 3.74407470e-02 5.80944240e-01 5.19446969e-01 -3.58936578e-01 -1.06549680e+00 6.91353858e-01 7.78325796e-01 -6.93107486e-01 1.23530388e+00 -2.76734084e-01 3.34318638e-01 1.10643037e-01 -2.55691499e-01 -1.16390693e+00 2.58302037e-02 -1.09428060e+00 -4.33187723e-01 8.24937284e-01 7.45626509e-01 -8.92446041e-01 8.44516456e-01 6.57882810e-01 2.66841412e-01 -1.49290895e+00 -9.67895329e-01 -1.00420892e+00 -1.02787033e-01 -4.17975754e-01 5.61058283e-01 5.38583398e-01 -3.14986616e-01 -1.14270404e-01 -6.72763810e-02 1.62817776e-01 7.15100765e-01 1.84785843e-01 5.95861435e-01 -9.46042061e-01 -8.64249289e-01 -8.01107228e-01 -9.38608721e-02 -8.03532958e-01 1.40499860e-01 -5.30970395e-01 2.69998223e-01 -1.00318897e+00 -3.96044292e-02 -7.04094410e-01 -4.54213798e-01 2.06077889e-01 -1.51929006e-01 -2.37166539e-01 -1.15790516e-01 3.83519024e-01 -1.50373951e-01 2.72669166e-01 7.91697145e-01 9.63807628e-02 -7.59142220e-01 4.77418125e-01 -4.09166992e-01 6.69726133e-01 8.57982039e-01 -4.86704856e-01 -4.41262066e-01 1.96700186e-01 2.41083249e-01 4.78686452e-01 1.32159203e-01 -6.15273476e-01 1.91185415e-01 -4.44810331e-01 1.63210958e-01 -4.58537042e-01 3.06364268e-01 -5.58426321e-01 1.88603938e-01 7.28402793e-01 -5.06370783e-01 2.61771470e-01 2.11344406e-01 5.57519078e-01 -2.04255097e-02 -4.86923546e-01 1.34176683e+00 -8.75290111e-02 -4.61901784e-01 4.73026574e-01 -3.53252620e-01 -4.83357871e-04 1.07129002e+00 -2.40923896e-01 -6.83786049e-02 -4.43965286e-01 -4.86694634e-01 3.35325837e-01 -1.03859477e-01 -1.46706671e-01 6.41048491e-01 -9.27771628e-01 -2.71383226e-01 8.39848295e-02 -2.05551758e-01 -1.96015969e-01 -2.16992483e-01 8.56831193e-01 -5.00233352e-01 5.19397318e-01 4.19505745e-01 -6.48630083e-01 -1.21279967e+00 3.01317900e-01 7.71721542e-01 -6.46070480e-01 -2.11805955e-01 9.52252924e-01 -2.10570157e-01 -2.76107877e-01 3.92878681e-01 -1.06187306e-01 8.09447840e-03 -1.65609524e-01 2.00152650e-01 7.96219409e-01 2.20960155e-01 -3.21639888e-02 -3.67800862e-01 5.60122073e-01 3.08650471e-02 -4.74891305e-01 1.26178825e+00 1.56769142e-01 8.98983050e-03 2.87851393e-01 1.33844519e+00 -4.14107054e-01 -1.00545347e+00 -1.93102390e-01 7.45639252e-03 -2.41882414e-01 4.51042533e-01 -7.71437883e-01 -1.09864175e+00 3.84396613e-01 6.75531268e-01 2.15943843e-01 1.44934142e+00 -1.75657570e-01 5.62161863e-01 6.11727774e-01 4.31444824e-01 -1.38690889e+00 -6.44562915e-02 1.44812569e-01 6.20675266e-01 -1.03593397e+00 3.00606161e-01 -1.62540853e-01 -5.55560231e-01 1.03896320e+00 4.73571956e-01 -1.73464864e-01 1.17958939e+00 3.02462250e-01 -1.74923852e-01 -1.83985725e-01 -1.09509671e+00 8.56262669e-02 3.76382232e-01 4.99867313e-02 1.40892461e-01 3.97193432e-02 -5.63547730e-01 4.82712775e-01 -3.18467289e-01 1.50476754e-01 1.38143584e-01 9.06335413e-01 -5.33100545e-01 -9.55165446e-01 -3.02144289e-01 9.37101781e-01 -5.17506063e-01 -1.37074724e-01 2.15516478e-01 8.75597477e-01 -2.95003504e-01 9.19146657e-01 -5.96114472e-02 -3.46714228e-01 1.31059244e-01 2.56883670e-02 4.27168995e-01 -2.39747703e-01 -7.31472194e-01 2.70882159e-01 1.21700205e-01 -6.89762294e-01 -2.23344155e-02 -9.00492489e-01 -1.21821296e+00 -2.77260810e-01 -7.51068473e-01 2.09983051e-01 1.00484729e+00 1.01912594e+00 2.06378683e-01 2.33724684e-01 9.74074960e-01 -5.24480045e-01 -1.16878474e+00 -4.09579992e-01 -6.52047813e-01 -2.05443814e-01 5.25983870e-02 -6.99946642e-01 -5.30766904e-01 -1.03237636e-01]
[6.6205620765686035, 3.9859941005706787]
fe5f4c49-6cbc-4c8b-bd82-89d0a19ce392
an-optimal-algorithm-for-finding-champions-in
2111.13621
null
https://arxiv.org/abs/2111.13621v4
https://arxiv.org/pdf/2111.13621v4.pdf
An Optimal Algorithm for Finding Champions in Tournament Graphs
A tournament graph is a complete directed graph, which can be used to model a round-robin tournament between $n$ players. In this paper, we address the problem of finding a champion of the tournament, also known as Copeland winner, which is a player that wins the highest number of matches. In detail, we aim to investigate algorithms that find the champion by playing a low number of matches. Solving this problem allows us to speed up several Information Retrieval and Recommender System applications, including question answering, conversational search, etc. Indeed, these applications often search for the champion inducing a round-robin tournament among the players by employing a machine learning model to estimate who wins each pairwise comparison. Our contribution, thus, allows finding the champion by performing a low number of model inferences. We prove that any deterministic or randomized algorithm finding a champion with constant success probability requires $\Omega(\ell n)$ comparisons, where $\ell$ is the number of matches lost by the champion. We then present an asymptotically-optimal deterministic algorithm matching this lower bound without knowing $\ell$, and we extend our analysis to three variants of the problem. Lastly, we conduct a comprehensive experimental assessment of the proposed algorithms on a question answering task on public data. Results show that our proposed algorithms speed up the retrieval of the champion up to $13\times$ with respect to the state-of-the-art algorithm that perform the full tournament.
['Rossano Venturini', 'Roberto Trani', 'Franco Maria Nardini', 'Lorenzo Beretta']
2021-11-26
null
null
null
null
['conversational-search']
['natural-language-processing']
[ 2.37949733e-02 5.37983440e-02 3.30135450e-02 -3.24000269e-01 -9.82911944e-01 -8.54283571e-01 -2.76611485e-02 4.32796091e-01 -7.18051195e-01 5.81670165e-01 -4.89764959e-01 -4.16744292e-01 -8.82443488e-01 -1.34595847e+00 -1.04958963e+00 -5.24361432e-01 -2.69808233e-01 1.32246149e+00 3.05489868e-01 -7.17898428e-01 5.02716243e-01 4.22359332e-02 -1.51301801e+00 2.14882404e-01 6.17187023e-01 8.78364980e-01 -1.91768892e-02 9.05092418e-01 -1.41087040e-01 6.32392466e-01 -5.56771040e-01 -1.09353101e+00 6.17053807e-01 -7.43110180e-01 -1.17716289e+00 -3.44173104e-01 3.58599097e-01 2.49333754e-02 -1.44155279e-01 1.07687736e+00 3.19863409e-01 2.91566581e-01 2.11374134e-01 -1.39246809e+00 -1.24759786e-01 1.35358918e+00 -7.25363076e-01 3.91689509e-01 7.66755581e-01 -3.71459454e-01 1.77142489e+00 -3.67840677e-01 7.10276127e-01 1.06987643e+00 3.09488863e-01 1.71573594e-01 -9.84257758e-01 -1.04727149e+00 3.63469422e-02 4.64038938e-01 -1.50388622e+00 9.50526223e-02 7.61458337e-01 3.85175571e-02 5.73168397e-01 8.64549220e-01 5.97809732e-01 1.28865868e-01 -1.55792177e-01 6.54102743e-01 9.23256695e-01 -6.64892435e-01 1.50142953e-01 7.99322873e-02 4.53435093e-01 9.09548402e-01 4.61268485e-01 -1.62183464e-01 -8.23368251e-01 -5.24830461e-01 1.40292391e-01 -2.17693582e-01 -1.59589693e-01 -2.96166927e-01 -6.79162920e-01 9.35297489e-01 3.70165139e-01 3.37208271e-01 -1.73795775e-01 2.45766789e-01 3.96435993e-04 9.64633167e-01 2.13061888e-02 7.09792316e-01 -2.83601016e-01 -1.41355544e-01 -7.46762455e-01 4.36821282e-01 1.37310994e+00 8.95071924e-01 9.65184271e-01 -9.84574795e-01 1.36910319e-01 5.72939515e-01 -7.64668882e-02 4.26173508e-01 -1.16929390e-01 -1.05411875e+00 5.87399185e-01 8.95134151e-01 3.36948633e-01 -1.40042210e+00 -3.95451397e-01 -4.68492270e-01 -5.10455310e-01 -2.44435504e-01 8.16751063e-01 -3.91222797e-02 1.41946241e-01 1.96043110e+00 4.96203423e-01 -1.32735251e-02 -2.87304282e-01 8.47426176e-01 4.33431536e-01 3.12608629e-01 -5.73304713e-01 -2.46662796e-01 1.57562840e+00 -8.94878983e-01 -2.00017899e-01 1.44064112e-03 8.03214550e-01 -7.24970877e-01 9.23954487e-01 6.28897548e-01 -1.35907626e+00 -2.61226416e-01 -7.57046640e-01 1.69799760e-01 -1.05026662e-01 -2.10974887e-01 7.11066365e-01 9.58468258e-01 -8.64064038e-01 5.65112770e-01 -3.07163447e-01 -3.75848830e-01 -1.81546450e-01 6.71360016e-01 -1.21411450e-01 -3.26724440e-01 -1.25172305e+00 4.88106668e-01 -1.67350918e-01 -8.23567063e-02 -6.18999481e-01 -4.73743945e-01 -4.58053827e-01 6.23380654e-02 1.03417671e+00 -8.80473316e-01 1.15925157e+00 -7.93825448e-01 -1.08264101e+00 1.13631356e+00 -2.01075986e-01 -6.65181637e-01 5.40547431e-01 2.10352600e-01 7.18610063e-02 1.56835109e-01 1.37339279e-01 -7.89697692e-02 9.69022736e-02 -8.83278668e-01 -9.17283714e-01 -5.38221478e-01 9.65085268e-01 3.30848306e-01 -1.51360333e-01 1.19419888e-01 -6.23514831e-01 2.15293303e-01 3.45792055e-01 -1.05487216e+00 -4.67330545e-01 -4.35127199e-01 -4.17143792e-01 -7.23500907e-01 -1.74334079e-01 1.65375158e-01 1.32455719e+00 -1.77444935e+00 1.37801141e-01 7.24837959e-01 4.95107353e-01 -2.03654006e-01 -2.13685229e-01 9.47004437e-01 3.54265124e-01 2.82863051e-01 1.59973785e-01 -2.53234565e-01 2.66533673e-01 1.74075410e-01 -7.03555271e-02 5.58217168e-01 -7.95099914e-01 6.66428685e-01 -7.26701260e-01 -3.48727137e-01 -3.48199844e-01 -4.17521477e-01 -8.35889280e-01 3.22744548e-01 -1.73004583e-01 -6.73849583e-02 -4.21971560e-01 2.25223467e-01 9.00647640e-01 -3.53970438e-01 6.73279226e-01 1.52774602e-01 6.61703423e-02 3.75273973e-01 -1.73555052e+00 1.33455861e+00 -3.39449078e-01 5.60984127e-02 5.14229298e-01 -1.13538122e+00 8.71685207e-01 -1.97308615e-01 3.26077074e-01 -5.85668564e-01 3.41319084e-01 4.63289350e-01 1.82500556e-01 -3.79165649e-01 6.98536098e-01 -2.20626101e-01 -6.28801107e-01 9.57835734e-01 -3.77712637e-01 1.01379715e-01 5.61453283e-01 6.50664508e-01 1.38770294e+00 -7.40514398e-01 -3.65709774e-02 -2.11353973e-01 6.38965309e-01 8.22624937e-02 4.96305466e-01 1.56388247e+00 -5.01644835e-02 5.93618602e-02 8.63372028e-01 -3.24811786e-01 -7.18231499e-01 -5.18432915e-01 3.64166349e-01 1.51430738e+00 6.29243970e-01 -6.36748612e-01 -7.07889438e-01 -4.20486361e-01 1.05187856e-02 2.71560222e-01 -7.94759929e-01 -1.13628646e-02 -5.14614761e-01 -4.46972340e-01 5.18047154e-01 -2.88531277e-02 6.79199934e-01 -6.29658759e-01 -3.62435907e-01 5.45551404e-02 -5.64040959e-01 -8.35803688e-01 -6.25583470e-01 5.53147905e-02 -4.58952338e-01 -1.62648118e+00 -1.89050809e-01 -7.95834303e-01 5.97892284e-01 4.44835961e-01 1.31974494e+00 6.39942646e-01 5.90767860e-02 4.93638486e-01 -4.06023830e-01 -2.22074479e-01 -1.35918438e-01 3.15940499e-01 1.98240820e-02 2.78721228e-02 6.77514493e-01 -3.79787683e-01 -8.67308855e-01 8.70950222e-01 -7.73920953e-01 -3.60967994e-01 1.76180467e-01 5.94880581e-01 6.63353622e-01 3.97321492e-01 3.83809507e-01 -1.58656216e+00 7.78172433e-01 -5.16988635e-01 -8.60474825e-01 4.00842756e-01 -4.80925769e-01 7.23201642e-03 8.27278554e-01 -9.59039629e-02 -4.23566967e-01 -1.65682763e-01 -6.38985634e-02 2.52478749e-01 3.97044480e-01 6.65656686e-01 -2.43396666e-02 -3.14914465e-01 6.91879392e-01 -2.32096203e-02 -2.81805068e-01 -3.89254510e-01 3.97546262e-01 5.14271498e-01 2.28448689e-01 -7.29360580e-01 7.56878376e-01 5.01482487e-01 2.35230222e-01 -3.75377148e-01 -8.48263979e-01 -8.79619598e-01 -1.04699150e-01 -1.97228447e-01 3.76521461e-02 -3.91996175e-01 -1.91312432e+00 1.73927754e-01 -7.92299390e-01 3.56265008e-02 -9.73653197e-02 2.82011598e-01 -4.43729192e-01 5.21547019e-01 -3.57829154e-01 -1.09996641e+00 -2.51237422e-01 -7.22333491e-01 3.57981294e-01 2.40597501e-01 -1.42138988e-01 -5.41981936e-01 2.60752082e-01 1.00154269e+00 1.66062340e-01 3.94016830e-03 9.46851611e-01 -1.06430852e+00 -9.31894481e-01 -5.11027098e-01 1.08224444e-01 -3.02340448e-01 -5.12734771e-01 -4.60198402e-01 -1.74119949e-01 -4.18340385e-01 -1.93915412e-01 -2.00642452e-01 7.13575065e-01 9.54217911e-02 8.13328743e-01 -2.14145005e-01 -3.48894835e-01 2.62891471e-01 1.26613450e+00 1.36006355e-01 2.34616801e-01 3.85805994e-01 8.62216279e-02 5.29877126e-01 7.27007687e-01 4.66974854e-01 6.65174007e-01 5.93880475e-01 4.42363739e-01 2.68401206e-01 5.34670353e-01 -3.62264812e-01 -2.13891208e-01 7.47533321e-01 -1.01606630e-01 -4.14990515e-01 -6.07319772e-01 5.79171240e-01 -2.03153396e+00 -7.93649256e-01 -2.73344368e-01 2.64146543e+00 6.54560626e-01 3.36785525e-01 4.98574555e-01 3.95556003e-01 7.45716572e-01 -2.41566360e-01 -5.31566679e-01 -6.10761225e-01 -1.13962032e-01 5.32210886e-01 7.11137533e-01 6.48968637e-01 -5.66693366e-01 7.70204544e-01 5.76341867e+00 8.84173214e-01 -3.05211693e-01 3.05928551e-02 5.36592305e-01 -3.25217336e-01 -3.58962715e-01 3.01494539e-01 -7.33053207e-01 6.00677803e-02 9.79893744e-01 -6.91926479e-01 8.47810924e-01 7.74340153e-01 -3.69648412e-02 -4.45680678e-01 -1.26653707e+00 8.48477840e-01 2.07438573e-01 -1.26684129e+00 -1.36422947e-01 1.30307302e-01 6.85676098e-01 -4.66327280e-01 -6.58397228e-02 3.81781280e-01 6.73052311e-01 -7.41953671e-01 3.63349468e-01 2.85424471e-01 3.46311390e-01 -1.23084116e+00 8.60984802e-01 8.28940034e-01 -1.19069469e+00 -2.50336230e-01 -5.92042029e-01 -3.63116711e-01 3.15330401e-02 3.81855190e-01 -5.93804896e-01 7.78977633e-01 7.50261009e-01 -2.68178821e-01 -9.19020325e-02 1.27571762e+00 3.10981590e-02 6.77094817e-01 -7.30177224e-01 -6.27350271e-01 1.47161454e-01 -3.32527697e-01 4.88210171e-01 6.83685124e-01 2.28952289e-01 6.93328559e-01 2.63878435e-01 2.98690081e-01 -7.25628912e-01 6.00524187e-01 -1.58200413e-01 1.87443838e-01 5.77605069e-01 1.01402736e+00 -7.81931698e-01 -1.15196779e-01 1.00190848e-01 8.29468727e-01 5.33601999e-01 -1.37509540e-01 -7.48525023e-01 -7.37637758e-01 3.72997552e-01 3.40007782e-01 1.89530626e-01 -4.39160056e-02 3.76989581e-02 -8.15412641e-01 3.94726507e-02 -1.10610616e+00 1.02895379e+00 -3.89303297e-01 -1.21595800e+00 6.13668382e-01 -1.69001311e-01 -7.05946267e-01 -2.90758852e-02 -3.01869452e-01 -5.52511692e-01 6.48977399e-01 -1.14829791e+00 -5.56700945e-01 -1.89849138e-01 6.74762726e-01 -3.20466422e-02 1.83399156e-01 6.29268348e-01 3.21019083e-01 -2.77533412e-01 1.04393589e+00 1.02730811e-01 -1.67187043e-02 5.37038863e-01 -1.26497984e+00 -1.36388868e-01 6.21975243e-01 5.84077358e-01 7.36484706e-01 8.85400355e-01 -1.65938437e-01 -1.74158919e+00 -4.07670975e-01 1.26565802e+00 -2.28656799e-01 5.49826860e-01 -2.90105462e-01 -3.28497797e-01 4.81796831e-01 3.62610966e-02 -4.53676373e-01 9.67383564e-01 6.81940258e-01 -2.87506968e-01 -5.87370813e-01 -1.26201689e+00 4.54977244e-01 1.21454859e+00 -4.39988077e-01 -3.39661688e-01 4.89270508e-01 4.01196361e-01 -5.09973049e-01 -5.02120197e-01 7.36563606e-03 5.36715567e-01 -1.11874425e+00 7.68700123e-01 -7.80173957e-01 9.71976295e-02 -2.11057290e-01 -3.94387655e-02 -9.71468091e-01 -2.16987029e-01 -1.04839921e+00 3.41248006e-01 8.50893021e-01 6.49684429e-01 -5.19296467e-01 1.15730631e+00 8.57596576e-01 5.71904838e-01 -8.28309715e-01 -1.17962551e+00 -5.01607180e-01 9.39694196e-02 -4.69436795e-01 7.80881882e-01 6.56885147e-01 3.27758729e-01 3.96761507e-01 -6.32497013e-01 2.50124216e-01 6.82371557e-01 8.60475361e-01 1.10948968e+00 -1.23580587e+00 -8.57397854e-01 -3.41219395e-01 -1.59163281e-01 -1.46418130e+00 -1.74529716e-01 -1.01975691e+00 -1.33730769e-01 -1.41175365e+00 4.30240333e-01 -7.01103687e-01 -2.47670472e-01 7.69729018e-02 3.04765571e-02 3.79711539e-01 1.55472428e-01 3.46141425e-03 -1.33528078e+00 -1.08062871e-01 1.17838252e+00 -2.03105956e-02 -1.06232859e-01 5.59795201e-01 -1.10662377e+00 5.79516351e-01 5.69395602e-01 -8.20431292e-01 -1.82716176e-01 -1.51623473e-01 1.12284410e+00 4.22479689e-01 -8.44031125e-02 -6.75383806e-01 8.90986919e-01 -1.16199441e-01 -5.09375870e-01 -3.57621789e-01 2.17800289e-01 -6.45655096e-01 2.75392383e-01 5.59412301e-01 -7.28988588e-01 4.73131798e-02 -2.86593497e-01 8.48457158e-01 6.52057901e-02 -5.84832609e-01 4.50730413e-01 -2.10443050e-01 6.78875670e-02 3.11559170e-01 -1.42095596e-01 2.40772873e-01 9.86030400e-01 6.57375008e-02 -3.19769531e-01 -8.17214489e-01 -5.78777909e-01 6.16111457e-01 6.45144954e-02 -2.52916180e-02 2.77079314e-01 -8.50754380e-01 -7.67334044e-01 -1.65340126e-01 1.31643504e-01 -4.05368149e-01 4.06842351e-01 8.42111468e-01 -5.26058912e-01 2.14554936e-01 3.79230738e-01 -1.09068178e-01 -1.69359052e+00 4.97245789e-01 4.29115295e-01 -7.34461427e-01 1.05337529e-02 1.28695560e+00 -2.72268116e-01 -3.79954338e-01 3.12707961e-01 -1.30694476e-03 -2.04243343e-02 -1.19788619e-02 4.77624506e-01 4.85285044e-01 5.07664569e-02 -3.97399843e-01 -4.81965512e-01 5.33937931e-01 -2.08261773e-01 -9.82188359e-02 1.16613770e+00 -3.50648165e-01 -3.45388889e-01 8.20716396e-02 9.34257329e-01 4.60297227e-01 -3.22411478e-01 -5.99324107e-01 -2.44176567e-01 -8.08492482e-01 -3.77623439e-01 -6.08330429e-01 -1.04039538e+00 3.82417440e-01 7.92259946e-02 8.28095496e-01 1.06761026e+00 1.74633935e-01 9.13483799e-01 9.79327559e-01 1.01226008e+00 -8.76345575e-01 -1.61724478e-01 2.93404907e-01 4.01125848e-01 -9.32296574e-01 -1.92225084e-01 -4.96500403e-01 -3.48433644e-01 8.17540467e-01 4.02789801e-01 -4.04586613e-01 8.28430951e-01 -6.81868494e-02 -3.22987199e-01 -4.06756163e-01 -9.64807630e-01 -3.48987490e-01 -8.82521942e-02 -1.02039829e-01 6.98715672e-02 1.88984513e-01 -1.06065381e+00 1.13904142e+00 -6.24347687e-01 -6.25720769e-02 6.04398847e-01 7.20301449e-01 -6.89156711e-01 -1.58969057e+00 -1.83030263e-01 4.76240367e-01 -7.60919273e-01 -7.97993839e-02 -7.04349399e-01 4.15748090e-01 1.53115377e-01 1.62120676e+00 -2.27074325e-01 -6.36096537e-01 7.00269938e-01 -2.82606989e-01 4.58964497e-01 -2.60123551e-01 -1.15184355e+00 -4.12490755e-01 2.39823714e-01 -3.73704731e-01 -3.49468291e-01 -2.67588466e-01 -1.05818439e+00 -1.15235102e+00 -6.76898181e-01 1.11570251e+00 3.51026297e-01 8.20681512e-01 2.23540962e-01 -1.66318595e-01 1.04050410e+00 1.45907238e-01 -6.17183864e-01 -6.65689051e-01 -9.82206643e-01 3.74800473e-01 -3.75063241e-01 -3.96338910e-01 -7.11184204e-01 -5.08408129e-01]
[6.729135513305664, 4.97307014465332]
163911d2-0b4c-4956-84d6-af252ddad9d8
priorband-practical-hyperparameter
2306.1237
null
https://arxiv.org/abs/2306.12370v1
https://arxiv.org/pdf/2306.12370v1.pdf
PriorBand: Practical Hyperparameter Optimization in the Age of Deep Learning
Hyperparameters of Deep Learning (DL) pipelines are crucial for their downstream performance. While a large number of methods for Hyperparameter Optimization (HPO) have been developed, their incurred costs are often untenable for modern DL. Consequently, manual experimentation is still the most prevalent approach to optimize hyperparameters, relying on the researcher's intuition, domain knowledge, and cheap preliminary explorations. To resolve this misalignment between HPO algorithms and DL researchers, we propose PriorBand, an HPO algorithm tailored to DL, able to utilize both expert beliefs and cheap proxy tasks. Empirically, we demonstrate PriorBand's efficiency across a range of DL benchmarks and show its gains under informative expert input and robustness against poor expert beliefs
['Frank Hutter', 'Luigi Nardi', 'Marius Lindauer', 'Maciej Janowski', 'Danny Stoll', 'Carl Hvarfner', 'Edward Bergman', 'Neeratyoy Mallik']
2023-06-21
null
null
null
null
['hyperparameter-optimization']
['methodology']
[-5.73178470e-01 1.81588620e-01 -7.09900022e-01 -3.43139589e-01 -1.15954304e+00 -8.70466173e-01 4.89134699e-01 2.39826962e-02 -5.65874100e-01 7.86228836e-01 2.53092617e-01 -4.41383302e-01 -2.44471341e-01 -3.93186897e-01 -6.98094249e-01 -5.67924678e-01 1.42438844e-01 7.67392516e-01 4.49270718e-02 -7.53398240e-02 1.90623641e-01 4.58138496e-01 -1.19888639e+00 -4.04965281e-02 8.10119689e-01 8.47417593e-01 -9.73327011e-02 5.15200555e-01 -6.36698538e-03 4.79886800e-01 -7.12422550e-01 -6.67076588e-01 3.99236709e-01 1.14142306e-01 -6.64488912e-01 -2.73790747e-01 2.77331352e-01 -2.43504554e-01 -2.33516805e-02 1.04319644e+00 9.88004506e-01 3.50835882e-02 4.93576407e-01 -1.19370067e+00 -2.53907233e-01 1.04782546e+00 -5.11682212e-01 2.11396977e-01 -1.32833853e-01 8.24395299e-01 1.29895461e+00 -6.69727147e-01 3.32715213e-01 1.29514182e+00 8.49236012e-01 2.51220345e-01 -1.40728235e+00 -6.93352818e-01 2.15193763e-01 -7.83187598e-02 -1.39973068e+00 -6.73766911e-01 5.34096539e-01 -4.92316723e-01 1.03983581e+00 -1.35779619e-01 7.16768146e-01 1.42489398e+00 -4.74013053e-02 9.05947745e-01 7.75789976e-01 -2.78444290e-01 6.03988051e-01 4.18816924e-01 2.41867527e-01 4.80332345e-01 4.36821193e-01 -3.53915524e-03 -7.37763762e-01 -3.91194880e-01 6.64986491e-01 -5.65236747e-01 -3.83533090e-01 -3.32470298e-01 -9.41720545e-01 8.60894740e-01 -3.34283337e-02 -1.22952700e-01 -3.62526238e-01 4.11344528e-01 3.04962844e-01 1.62714094e-01 3.60823721e-01 1.07270837e+00 -8.65483522e-01 -4.63653654e-01 -7.61779606e-01 5.50497234e-01 1.13092136e+00 9.35098231e-01 6.36197984e-01 -1.43733829e-01 -3.10635030e-01 8.58615398e-01 4.66710091e-01 3.01299900e-01 4.60383713e-01 -1.02714801e+00 4.46503818e-01 5.80783784e-01 2.52168149e-01 -7.99845397e-01 -5.17320096e-01 -9.55017686e-01 -5.01100123e-01 3.57514992e-02 4.80708748e-01 -4.38252360e-01 -6.63354456e-01 1.73275328e+00 3.27048272e-01 6.71583265e-02 -1.07220076e-01 7.58746326e-01 6.41587138e-01 3.37092966e-01 2.03624368e-01 1.98063627e-01 1.07854462e+00 -7.48456419e-01 -3.58122081e-01 -6.64928257e-01 4.26314414e-01 -4.53168750e-01 1.42663181e+00 6.83193922e-01 -1.28818834e+00 -4.08199020e-02 -1.04148114e+00 -3.98414247e-02 -1.08255789e-01 3.25363390e-02 8.67920518e-01 9.05036330e-01 -9.85075533e-01 3.97558331e-01 -8.77774596e-01 6.53990656e-02 6.77852213e-01 5.29884994e-01 1.00681096e-01 2.31899336e-01 -1.17234719e+00 7.82993674e-01 5.35099030e-01 4.00394499e-02 -1.12266660e+00 -1.23585713e+00 -3.22143406e-01 3.60464364e-01 6.42885208e-01 -9.14143264e-01 1.58574212e+00 -4.17067975e-01 -1.79922152e+00 5.29772699e-01 3.27662408e-01 -5.39680481e-01 9.26138520e-01 -5.60123205e-01 1.40006453e-01 -5.57357110e-02 -2.89143920e-01 6.52537942e-01 8.69471967e-01 -9.54798877e-01 -4.76730347e-01 -1.31483972e-01 1.46294385e-01 3.19774270e-01 -5.72990060e-01 -2.07453951e-01 -9.62389350e-01 -4.40772951e-01 -2.58025438e-01 -8.84842098e-01 -3.06204617e-01 -1.84052840e-01 -5.74772835e-01 -2.92128593e-01 2.89299309e-01 -1.94440573e-01 1.45946264e+00 -1.92990756e+00 -1.68247614e-03 3.30849051e-01 3.56142193e-01 2.51350015e-01 -6.20168187e-02 1.20444447e-01 2.79579163e-01 3.63429278e-01 -4.14638259e-02 -3.72876436e-01 4.93508697e-01 -7.67424554e-02 -2.75741637e-01 5.16881406e-01 -2.05538869e-02 9.81549084e-01 -9.21054840e-01 -4.11128730e-01 5.34443744e-02 4.51018035e-01 -9.42441881e-01 2.75095582e-01 -7.13103235e-01 2.01999739e-01 -6.89818203e-01 8.62570047e-01 3.64513934e-01 -7.30085909e-01 3.38007927e-01 -2.88614899e-01 2.65228152e-02 4.05393213e-01 -1.29602838e+00 1.32947135e+00 -5.14948726e-01 5.51106453e-01 1.97108775e-01 -5.02567232e-01 5.64090431e-01 2.31621876e-01 3.32561642e-01 -3.18738550e-01 2.74780333e-01 1.63000062e-01 -9.06519219e-02 -4.70559150e-01 1.48160502e-01 3.24421048e-01 8.50205570e-02 6.41859412e-01 -4.10569645e-02 -1.72039390e-01 1.28006533e-01 -9.57448334e-02 1.25968778e+00 1.80521220e-01 4.73396927e-01 -3.20461780e-01 -6.10407144e-02 1.20211698e-01 5.72570920e-01 1.11282551e+00 -1.45268247e-01 4.76433903e-01 8.33605528e-01 -4.29568142e-01 -9.25019681e-01 -7.83431172e-01 -3.29833895e-01 1.46978056e+00 -3.04935127e-01 -3.55633229e-01 -9.94029224e-01 -5.91141164e-01 2.28721365e-01 7.40990043e-01 -5.92280269e-01 -1.28169566e-01 -3.95469189e-01 -1.32892632e+00 7.57643938e-01 4.87709522e-01 4.89257604e-01 -7.95953035e-01 -8.05256009e-01 2.23555684e-01 1.75733581e-01 -9.54790175e-01 -3.28772366e-01 2.54886121e-01 -7.12997973e-01 -1.16044939e+00 -5.91963947e-01 2.98278388e-02 2.80582607e-01 -9.15530026e-02 1.66485679e+00 -1.70116946e-02 -2.34263495e-01 3.38858247e-01 -1.60057135e-02 -4.24794704e-01 -3.83183092e-01 5.63586116e-01 -9.12561715e-02 -2.83817410e-01 3.21740031e-01 -5.46871364e-01 -9.57865000e-01 3.46328944e-01 -6.36073649e-01 -2.18859822e-01 1.04757595e+00 7.97653794e-01 3.33055884e-01 -3.99137214e-02 4.36131418e-01 -1.22448027e+00 9.41413701e-01 -6.85198009e-01 -9.86488819e-01 2.58794159e-01 -1.21099293e+00 3.30786347e-01 3.45910549e-01 -6.04473352e-01 -1.07806432e+00 -1.86150223e-01 -1.13472529e-01 -4.22355145e-01 -3.65463383e-02 5.39909840e-01 -4.58707750e-01 -6.88968226e-02 1.09857941e+00 -3.44002932e-01 -2.94816464e-01 -6.19651318e-01 3.74247342e-01 4.64554727e-01 4.43205327e-01 -1.03212976e+00 4.81170654e-01 6.16288856e-02 -2.42620096e-01 -5.72620094e-01 -1.21709931e+00 -1.92850128e-01 -2.67746150e-01 3.28767486e-02 6.14838183e-01 -1.07894933e+00 -1.05560231e+00 2.42742702e-01 -8.95860314e-01 -7.95307934e-01 4.31735180e-02 3.14233750e-01 -3.19647491e-01 9.06901881e-02 -1.98020637e-01 -4.23545599e-01 -4.69508827e-01 -1.58998632e+00 9.18047667e-01 3.00318807e-01 -4.08903182e-01 -1.31243348e+00 1.00167535e-01 3.49009335e-01 5.01613438e-01 8.59225690e-02 1.09447801e+00 -8.66352677e-01 -7.64813960e-01 -1.42210156e-01 -1.92661032e-01 1.18152164e-01 -3.96072477e-01 2.00726494e-01 -1.21872449e+00 -2.30498955e-01 -2.79521137e-01 -5.01986682e-01 5.98132312e-01 5.60103953e-01 1.31357479e+00 -4.33766931e-01 -3.07575494e-01 9.84460711e-01 1.23430324e+00 -3.70944589e-01 2.29097247e-01 8.06667030e-01 4.42065090e-01 2.83777535e-01 3.27626944e-01 8.40466857e-01 4.28041637e-01 4.52010900e-01 4.03848767e-01 1.65412247e-01 1.71753317e-01 -1.91282913e-01 2.23458424e-01 2.29132667e-01 1.54213801e-01 -4.84834582e-01 -1.24329090e+00 3.84340137e-01 -1.76501799e+00 -3.63031596e-01 2.39294693e-01 2.13542342e+00 1.42590606e+00 5.47664225e-01 9.09058452e-02 -2.44349629e-01 3.72223973e-01 1.53345644e-01 -1.19346654e+00 4.36729315e-04 -5.02495244e-02 4.39368412e-02 7.23858654e-01 2.71670222e-01 -8.50296199e-01 9.30226624e-01 7.64101934e+00 7.74480402e-01 -8.44866276e-01 -5.08465432e-03 8.37521255e-01 -4.60501730e-01 -4.83965814e-01 -1.19612180e-01 -1.29274797e+00 4.70458090e-01 9.85863328e-01 -1.74526870e-01 5.92234313e-01 1.25344014e+00 1.85401142e-01 -1.55668914e-01 -1.29231250e+00 1.00933242e+00 -4.24012214e-01 -1.50682306e+00 -2.63486266e-01 -3.57624069e-02 7.16251373e-01 2.89219052e-01 4.59569305e-01 5.75602949e-01 1.09387457e+00 -1.32438040e+00 6.64978206e-01 3.20238203e-01 5.26501656e-01 -7.80168056e-01 6.47208512e-01 3.24952990e-01 -4.82142895e-01 -2.02035457e-01 -3.91719639e-01 3.85299861e-01 -1.93400979e-01 9.21667814e-01 -1.23373723e+00 -4.62122485e-02 8.18572581e-01 3.77032727e-01 -7.90747702e-01 1.04817319e+00 -3.43066841e-01 1.00594389e+00 -5.88086426e-01 -4.04337719e-02 2.51125365e-01 6.29277453e-02 3.58688802e-01 1.37612820e+00 2.21414398e-02 -8.14263970e-02 8.25623944e-02 9.89959598e-01 -2.88368493e-01 4.52430956e-02 -1.10153809e-01 -3.15745562e-01 1.01920116e+00 1.19603634e+00 -4.21832412e-01 -9.12563056e-02 -1.53859317e-01 3.22786391e-01 5.67017794e-01 4.14310634e-01 -7.64595032e-01 -1.03835188e-01 9.71164823e-01 -7.99380839e-02 -7.77681172e-02 -2.31198948e-02 -7.66454041e-01 -8.53560507e-01 -2.48873293e-01 -1.45207453e+00 5.90843856e-01 -4.44462627e-01 -1.14985228e+00 2.40135550e-01 2.90527582e-01 -7.77113914e-01 -2.67205149e-01 -6.36934042e-01 -3.92802835e-01 8.98751736e-01 -1.59728873e+00 -7.93844879e-01 -2.73134112e-01 1.91969067e-01 4.69345301e-01 -3.07072908e-01 4.82491434e-01 1.18718959e-01 -9.62926328e-01 1.00136864e+00 1.15003258e-01 -1.66878194e-01 9.03944373e-01 -1.39697349e+00 4.90384579e-01 3.53979975e-01 -2.16977715e-01 9.44607675e-01 1.09415817e+00 -4.58469927e-01 -1.85262167e+00 -7.90935397e-01 1.40544221e-01 -5.37344277e-01 8.65690827e-01 -1.28568202e-01 -8.66341949e-01 7.47569919e-01 1.39280170e-01 -1.95634574e-01 6.81242108e-01 5.57193220e-01 -4.22038496e-01 -6.48378804e-02 -9.89648700e-01 8.90942216e-01 7.65932858e-01 -2.03537583e-01 -3.39942455e-01 4.60854501e-01 4.65105861e-01 -7.01465130e-01 -1.15287721e+00 4.10618335e-01 5.94825447e-01 -1.00948012e+00 1.12606263e+00 -4.95614290e-01 1.29829928e-01 5.49010001e-02 6.26280010e-02 -1.39030898e+00 -1.12363882e-01 -1.15737605e+00 -2.47319698e-01 1.24118483e+00 6.56931579e-01 -6.00553453e-01 8.75433564e-01 1.03107250e+00 -3.29387486e-02 -9.20903206e-01 -4.71780479e-01 -4.62998748e-01 4.63003293e-02 -5.63618779e-01 8.15202653e-01 7.92814612e-01 -4.75143850e-01 3.61206144e-01 -5.20219877e-02 2.26217076e-01 7.53641248e-01 -1.60461083e-01 1.08618295e+00 -1.33357203e+00 -7.81990767e-01 -1.00022292e+00 2.93507218e-01 -9.92812991e-01 2.94353575e-01 -6.38637722e-01 1.01803862e-01 -1.03443527e+00 1.58709988e-01 -6.24671578e-01 -9.20981616e-02 6.59271836e-01 -4.04307872e-01 -1.86829716e-02 -2.91568160e-01 2.25767434e-01 -5.74769318e-01 4.44233060e-01 9.15746570e-01 1.52793095e-01 -4.20835018e-01 -4.36209589e-02 -9.20390725e-01 1.01751292e+00 7.15450764e-01 -3.79795581e-01 -5.24680912e-01 -7.10761130e-01 9.53852832e-01 -3.03588659e-01 1.50822684e-01 -7.21428514e-01 2.61076272e-01 -3.69790584e-01 3.50598872e-01 -3.48261207e-01 9.94645208e-02 -3.26944351e-01 1.54347211e-01 8.04218501e-02 -6.07474387e-01 -1.92313660e-02 2.31810436e-01 5.69689274e-01 2.20572576e-01 -3.72761518e-01 1.05743647e+00 -2.47670799e-01 -6.21870220e-01 3.69154572e-01 -2.48747878e-02 6.24069929e-01 6.66477680e-01 1.68669745e-01 -4.86531168e-01 -2.13898957e-01 -4.61928487e-01 6.27643168e-01 5.17098427e-01 -4.91567627e-02 2.10944846e-01 -7.09339201e-01 -4.97825205e-01 -5.46955690e-02 -6.07062504e-02 2.94685453e-01 -1.33583635e-01 7.54774630e-01 -5.85237384e-01 4.00472850e-01 1.92629904e-01 -4.83551174e-01 -6.86663449e-01 1.53746724e-01 4.93178546e-01 -3.23995322e-01 -6.66370094e-01 1.08230507e+00 -7.42346840e-03 -2.97297984e-01 7.52148330e-01 -3.61076929e-02 1.67840868e-02 1.40707850e-01 4.65015948e-01 5.98061681e-01 1.70662209e-01 2.95171827e-01 -7.82506540e-02 2.06053615e-01 -1.73053995e-01 -2.27723047e-01 1.50902796e+00 -4.10448946e-02 2.11195752e-01 3.34214300e-01 7.07477212e-01 -4.25422639e-02 -1.75505888e+00 -2.42601097e-01 3.05512965e-01 -4.37768012e-01 5.18670559e-01 -9.10930395e-01 -9.04676795e-01 8.12476814e-01 2.55601138e-01 -4.30766046e-02 8.29972565e-01 -1.36916652e-01 5.71437001e-01 8.06922078e-01 2.51549691e-01 -1.23739278e+00 1.06349424e-01 2.84614652e-01 6.94629610e-01 -1.32795930e+00 3.07253331e-01 -7.24799372e-03 -7.66719162e-01 9.57799256e-01 5.91840863e-01 2.08431110e-01 6.86771095e-01 4.51869935e-01 1.37696847e-01 -1.72436774e-01 -1.26891410e+00 5.22368960e-02 1.07438810e-01 3.38221967e-01 3.67802471e-01 -3.83145720e-01 1.99059874e-01 7.54947364e-01 -3.64893228e-01 -6.63113222e-02 1.66300327e-01 5.74973643e-01 -4.97367650e-01 -8.44554126e-01 -3.20503145e-01 2.75855333e-01 -7.10939705e-01 -1.65822029e-01 -1.60867512e-01 9.27504957e-01 -2.50195712e-01 6.19576216e-01 -3.03893775e-01 6.70813536e-03 1.76081106e-01 3.73511249e-03 2.63410181e-01 -7.23022938e-01 -7.84531355e-01 1.99670464e-01 1.84891194e-01 -7.47554839e-01 2.50801653e-01 -5.63049734e-01 -1.00005579e+00 -2.35615626e-01 -2.59118266e-02 9.67127308e-02 7.94463933e-01 9.02828693e-01 5.60968459e-01 2.75093228e-01 4.72072065e-01 -7.10352361e-01 -1.18592620e+00 -8.64585876e-01 -2.72180110e-01 2.58728135e-02 1.93568736e-01 -7.51774430e-01 -5.51969886e-01 -1.36243463e-01]
[8.391520500183105, 3.804572343826294]
c372d8d2-1684-4eda-ba2f-91e9b747fb97
a-quantitative-study-of-nlp-approaches-to
2305.10236
null
https://arxiv.org/abs/2305.10236v1
https://arxiv.org/pdf/2305.10236v1.pdf
A quantitative study of NLP approaches to question difficulty estimation
Recent years witnessed an increase in the amount of research on the task of Question Difficulty Estimation from Text QDET with Natural Language Processing (NLP) techniques, with the goal of targeting the limitations of traditional approaches to question calibration. However, almost the entirety of previous research focused on single silos, without performing quantitative comparisons between different models or across datasets from different educational domains. In this work, we aim at filling this gap, by quantitatively analyzing several approaches proposed in previous research, and comparing their performance on three publicly available real world datasets containing questions of different types from different educational domains. Specifically, we consider reading comprehension Multiple Choice Questions (MCQs), science MCQs, and math questions. We find that Transformer based models are the best performing across different educational domains, with DistilBERT performing almost as well as BERT, and that they outperform other approaches even on smaller datasets. As for the other models, the hybrid ones often outperform the ones based on a single type of features, the ones based on linguistic features perform well on reading comprehension questions, while frequency based features (TF-IDF) and word embeddings (word2vec) perform better in domain knowledge assessment.
['Luca Benedetto']
2023-05-17
null
null
null
null
['reading-comprehension']
['natural-language-processing']
[-2.41323680e-01 1.53578483e-02 -5.91716915e-02 -2.18943551e-01 -8.33310843e-01 -8.97595525e-01 7.02459395e-01 1.10049176e+00 -7.02442169e-01 6.43329084e-01 4.84752089e-01 -4.78080660e-01 -7.94517815e-01 -9.22028065e-01 -3.82563412e-01 -3.27553321e-03 3.60076398e-01 4.78107274e-01 5.26324809e-01 -7.00274587e-01 6.35739744e-01 1.04238294e-01 -1.85433710e+00 -5.91500364e-02 1.43660736e+00 1.00054753e+00 1.38846144e-01 6.28966033e-01 -9.06741202e-01 1.07539427e+00 -6.50699079e-01 -8.59234333e-01 -1.30904883e-01 -3.30757111e-01 -1.31555021e+00 -3.99229795e-01 8.54847193e-01 7.48569816e-02 -2.17164382e-01 9.85695362e-01 4.50175405e-01 3.29327673e-01 6.62075818e-01 -8.60241592e-01 -1.13648105e+00 4.38584417e-01 6.61602393e-02 4.05822784e-01 1.10336471e+00 -2.13298842e-01 1.18235445e+00 -7.30759501e-01 2.91545630e-01 1.25322998e+00 5.27629733e-01 3.57724994e-01 -8.72020841e-01 -1.62527576e-01 6.41056076e-02 7.39464462e-01 -1.07595754e+00 2.50660721e-02 4.98858839e-01 -6.67790949e-01 5.84472537e-01 2.01992273e-01 3.12496781e-01 9.76118505e-01 1.69827919e-02 5.64885020e-01 1.55923772e+00 -8.10233355e-01 2.07423583e-01 4.21225637e-01 9.39347386e-01 3.98472458e-01 2.30313942e-01 -3.82727891e-01 -5.57805300e-01 2.06899043e-04 1.70988590e-01 -2.35356793e-01 -4.77198482e-01 -1.32317752e-01 -1.03990602e+00 1.08128214e+00 -8.43107179e-02 8.45415354e-01 -2.66610440e-02 -5.33441663e-01 1.11676924e-01 8.80168557e-01 5.09756327e-01 8.44366431e-01 -8.96034658e-01 -4.81343895e-01 -8.05354774e-01 6.26389503e-01 1.24370563e+00 7.58206367e-01 5.13437688e-01 -3.51445645e-01 -6.58696711e-01 1.09715569e+00 2.71769792e-01 2.01031938e-01 9.60632861e-01 -6.18400216e-01 5.99862278e-01 9.21309054e-01 -5.80957793e-02 -1.08120418e+00 -4.54844743e-01 -3.19021285e-01 -3.25178027e-01 -1.67921275e-01 1.03042066e+00 -2.46180281e-01 -4.12570477e-01 1.53157270e+00 1.64458767e-01 -1.23384178e-01 1.88913941e-02 6.51985407e-01 1.37937105e+00 5.03443480e-01 1.73092350e-01 1.37438521e-01 1.82626128e+00 -1.04744470e+00 -9.94221807e-01 -1.00384600e-01 5.37788570e-01 -9.05532181e-01 1.42543948e+00 6.26567185e-01 -1.22070324e+00 -6.89092517e-01 -8.56296599e-01 -3.84375989e-01 -1.00401056e+00 -2.40853831e-01 1.68067664e-01 1.03375578e+00 -1.00338888e+00 4.32635576e-01 -4.67052869e-02 -4.65159625e-01 -4.54592630e-02 -1.09019116e-01 -2.84147501e-01 -2.99541652e-01 -1.53141916e+00 1.16229892e+00 9.70178992e-02 -6.44479990e-01 -3.72167587e-01 -1.05940044e+00 -8.82166743e-01 3.50435734e-01 2.16414288e-01 -3.68070513e-01 1.30452251e+00 -7.70969033e-01 -1.79454815e+00 7.56443322e-01 9.80638862e-02 -3.42425376e-01 3.68605375e-01 -5.20434499e-01 -3.40157717e-01 -9.30416398e-03 -7.81469196e-02 2.54690498e-01 3.25673878e-01 -6.86455071e-01 -4.39545512e-01 -2.61283129e-01 4.80298042e-01 1.02628604e-01 -8.58998120e-01 1.15996554e-01 -1.26354219e-02 -5.91121793e-01 -1.86863258e-01 -2.64983237e-01 1.06900632e-01 -2.36272931e-01 1.53288186e-01 -9.90960479e-01 3.18369061e-01 -8.26970935e-01 1.52104771e+00 -1.73492754e+00 1.82274401e-01 -9.20932218e-02 2.70386428e-01 5.17196000e-01 -2.32773766e-01 7.47487962e-01 -4.62807640e-02 2.28164926e-01 3.05694193e-02 -1.54346935e-02 4.20947522e-01 8.08828250e-02 -6.05375469e-02 1.09155066e-01 8.56206715e-02 8.81331444e-01 -1.04169846e+00 -4.91883844e-01 1.47077307e-01 -2.62744613e-02 -4.27211910e-01 5.15051067e-01 -3.22378099e-01 3.64784449e-02 -4.18205231e-01 2.98375815e-01 5.68332136e-01 -5.46199568e-02 -1.20052576e-01 4.15351957e-01 -1.82191551e-01 6.15217149e-01 -1.22389126e+00 1.57956028e+00 -6.52623355e-01 8.06259751e-01 -2.00263560e-01 -1.21692109e+00 1.05394435e+00 3.14691871e-01 4.05593783e-01 -9.56494153e-01 1.40651450e-01 2.15009209e-02 1.27429247e-01 -1.07224464e+00 4.58854765e-01 1.60703868e-01 -2.95997430e-02 3.51357758e-01 5.67864060e-01 -3.40788156e-01 5.84665835e-01 4.16125283e-02 1.11564231e+00 -9.55300778e-02 2.54122913e-01 -5.81013739e-01 9.79091227e-01 -1.86481133e-01 -1.58594564e-01 7.61809647e-01 -2.29574919e-01 5.99252939e-01 6.75602794e-01 1.92534772e-03 -5.92923820e-01 -1.11682677e+00 -4.27320957e-01 1.52970779e+00 -8.66024813e-04 -4.81930852e-01 -9.66151237e-01 -5.84457755e-01 7.90547729e-02 9.40873265e-01 -5.74565947e-01 1.61080465e-01 -3.01808029e-01 -4.25176442e-01 4.64015990e-01 1.78141981e-01 4.72596318e-01 -6.61442876e-01 -2.60176569e-01 2.67785579e-01 -2.44598001e-01 -1.15038526e+00 1.89813618e-02 -9.39209387e-02 -6.25367761e-01 -1.24184859e+00 -8.42195570e-01 -9.26919401e-01 5.91008328e-02 4.36896756e-02 1.77383089e+00 2.09662825e-01 7.17046298e-03 9.63038146e-01 -9.36495602e-01 -6.96508050e-01 -1.81164280e-01 3.98798555e-01 -1.95630968e-01 -1.85269341e-01 7.89608240e-01 -3.96452069e-01 -3.66284192e-01 8.72596055e-02 -1.06515634e+00 -4.95202869e-01 2.91729569e-01 6.40213966e-01 -2.01706635e-03 -3.02292049e-01 8.91970754e-01 -6.91814542e-01 1.40603137e+00 -9.59105074e-01 -4.62940007e-01 4.96407419e-01 -8.07366312e-01 -1.26424469e-02 6.64791346e-01 -4.35735554e-01 -8.95262599e-01 -8.56449604e-01 -6.51421249e-01 6.01728112e-02 -5.11750340e-01 5.98403573e-01 -2.17502508e-02 -2.64069706e-01 8.55426431e-01 2.45645627e-01 -1.84983671e-01 -8.53501081e-01 1.90259993e-01 5.99127233e-01 1.01951517e-01 -7.44041324e-01 7.20910549e-01 -3.42715442e-01 -3.76824290e-01 -9.92431104e-01 -1.13293087e+00 -5.11811912e-01 -6.15778625e-01 -2.47646347e-01 9.24549997e-01 -5.67890644e-01 -7.69464254e-01 3.31741333e-01 -9.97785032e-01 5.05906381e-02 -3.73082191e-01 5.78551948e-01 -1.61246359e-01 4.66744661e-01 -5.48791528e-01 -5.76671660e-01 2.99881212e-02 -1.01195669e+00 6.25032723e-01 6.19116664e-01 -2.70479232e-01 -1.43410468e+00 4.76207167e-01 8.64252150e-01 7.56254792e-01 -5.44389486e-02 1.36687255e+00 -1.22391355e+00 -1.61691755e-01 -1.23811819e-01 2.45773583e-03 5.19300699e-01 -1.74637884e-02 -2.22877085e-01 -8.85922670e-01 4.87502739e-02 2.80795842e-01 -5.05602062e-01 6.81314707e-01 1.19948402e-01 1.30300295e+00 -2.25212872e-01 3.32519799e-01 -1.51206300e-01 1.44207406e+00 -3.37948263e-01 5.60122728e-01 5.21589756e-01 2.14696035e-01 9.80305612e-01 5.50193071e-01 9.47686359e-02 1.02207255e+00 4.21535313e-01 3.97309005e-01 4.65628058e-01 -1.49597302e-01 -1.83349460e-01 4.08132374e-01 1.28443623e+00 1.90658852e-01 -3.56689125e-01 -1.18473899e+00 7.95026541e-01 -1.43991804e+00 -6.62941456e-01 -5.19940615e-01 2.12824035e+00 9.20801044e-01 -1.61384240e-01 9.25256759e-02 3.07591707e-01 1.94598228e-01 1.86887726e-01 9.06529501e-02 -7.62362480e-01 -9.51956660e-02 7.73967862e-01 9.56130028e-02 5.55887461e-01 -8.70782912e-01 5.76529086e-01 6.39037991e+00 8.52023482e-01 -4.88691300e-01 2.19643772e-01 3.85315180e-01 3.89620870e-01 -6.42060220e-01 -2.64144003e-01 -6.49022341e-01 4.55308348e-01 1.27523327e+00 -2.88221240e-01 1.80068195e-01 5.65929055e-01 -8.15321654e-02 -3.85353476e-01 -9.83112991e-01 7.49572515e-01 4.25449878e-01 -9.99012887e-01 -3.07419188e-02 -4.11428124e-01 6.86831653e-01 -2.51636147e-01 -2.19625849e-02 6.26699328e-01 2.55692035e-01 -1.23981214e+00 4.38210458e-01 8.02869081e-01 1.19562328e-01 -5.08276463e-01 9.73617792e-01 6.37181044e-01 -6.92002714e-01 -2.86828995e-01 -4.89374220e-01 -5.23755968e-01 -3.68547291e-01 5.44817448e-01 -3.36214900e-01 7.35183656e-01 8.49722981e-01 3.64496768e-01 -1.11110556e+00 1.14174378e+00 -1.69882193e-01 1.13702023e+00 1.34888580e-02 -5.97116888e-01 2.23800927e-01 -3.23944360e-01 1.55679241e-01 1.17107773e+00 2.09973454e-01 6.24913760e-02 8.96808691e-03 5.93248546e-01 -2.63552275e-02 8.27939212e-01 -2.91455448e-01 3.87401097e-02 3.32861990e-01 1.19672501e+00 -1.93278819e-01 -2.04375282e-01 -9.37441528e-01 3.67541850e-01 4.83840406e-01 6.10741228e-02 -6.66123927e-01 -5.23825407e-01 6.46962225e-01 2.61655837e-01 -2.20567212e-02 -3.44699442e-01 -2.44336620e-01 -1.34272015e+00 4.01399620e-02 -1.15258467e+00 5.53427517e-01 -4.53024507e-01 -1.79969525e+00 2.27840856e-01 6.62053376e-02 -9.43784416e-01 -2.90358011e-02 -8.50204945e-01 -5.93715250e-01 9.31066990e-01 -1.91846037e+00 -4.02129591e-01 -6.29868031e-01 6.03469133e-01 7.77332664e-01 -1.04569659e-01 8.09622526e-01 5.44696391e-01 -4.32670593e-01 6.42115533e-01 3.18621427e-01 1.96018264e-01 9.77602005e-01 -1.66769040e+00 1.32892057e-02 4.65931833e-01 2.70549119e-01 3.36526394e-01 5.82717657e-01 -1.30229890e-01 -1.36765194e+00 -6.01836324e-01 1.40914178e+00 -7.49425292e-01 9.16697562e-01 -1.70181930e-01 -1.32319009e+00 2.24287495e-01 9.65648770e-01 -4.83878553e-01 1.09113181e+00 2.03548282e-01 -3.13591033e-01 6.98850453e-02 -1.11738610e+00 3.54052007e-01 5.80299854e-01 -2.90479511e-01 -1.43936527e+00 6.12664998e-01 6.07205689e-01 -3.24154764e-01 -1.51797748e+00 2.17403755e-01 1.85425341e-01 -1.00964093e+00 9.25489962e-01 -8.46580565e-01 8.40643287e-01 1.37028173e-01 -5.62733002e-02 -1.36559570e+00 -2.02339500e-01 -1.07910767e-01 4.21217903e-02 1.60256827e+00 3.20046782e-01 -6.87565923e-01 5.35027742e-01 4.77782130e-01 1.40897572e-01 -8.11846495e-01 -8.98447692e-01 -6.37613177e-01 9.59066808e-01 -3.05491447e-01 5.81626415e-01 1.20374560e+00 3.60027398e-03 4.36973035e-01 2.07233921e-01 -1.78786501e-01 1.75068066e-01 6.06912300e-02 6.14993453e-01 -1.64838266e+00 -8.89296979e-02 -7.79277325e-01 -4.20071363e-01 -1.12485921e+00 2.71667480e-01 -8.14159274e-01 -2.93123841e-01 -1.65980947e+00 -1.09286055e-01 2.35615503e-02 -1.22658223e-01 1.55182900e-02 -6.08655155e-01 -2.29623541e-01 1.45932838e-01 -5.12604117e-01 -6.26373053e-01 5.67407906e-01 1.25949228e+00 -6.49204329e-02 3.07990253e-01 -6.29405230e-02 -7.03161895e-01 7.99079597e-01 7.19221532e-01 -2.59437591e-01 -4.19409901e-01 -6.82622254e-01 4.59227145e-01 7.78243365e-03 2.00369030e-01 -1.10318196e+00 2.82878906e-01 -2.01386139e-01 2.70165801e-01 -1.36383936e-01 -6.10494278e-02 -6.81401372e-01 -8.24380338e-01 1.94026545e-01 -6.61476254e-01 3.39430064e-01 1.18807033e-01 2.82436728e-01 -6.90176129e-01 -8.96016300e-01 6.56593800e-01 -9.70430747e-02 -6.04769468e-01 -1.39214508e-02 -5.47333062e-01 8.36414576e-01 7.79312670e-01 -8.21538940e-02 -3.39983612e-01 -5.60631096e-01 -4.44114447e-01 5.24615169e-01 -6.78961501e-02 9.88145888e-01 4.41057295e-01 -1.07510662e+00 -1.01415968e+00 -1.22018978e-01 1.92674398e-01 -2.06171408e-01 2.30712086e-01 6.87889755e-01 -5.34782767e-01 7.65967309e-01 -1.97616607e-01 -3.38103026e-01 -1.07693326e+00 3.29163969e-01 3.27896386e-01 -6.45230114e-01 3.99188325e-02 8.46401572e-01 -3.27192873e-01 -8.68271708e-01 2.53426075e-01 -4.91889864e-01 -1.20827937e+00 5.53524673e-01 5.54077148e-01 7.86371112e-01 4.44684952e-01 -2.92463541e-01 4.75502126e-02 7.97494113e-01 1.53061301e-01 1.26500562e-01 1.19334674e+00 -3.94638628e-02 -1.29326954e-01 5.19138455e-01 1.08528280e+00 2.96310574e-01 -3.95792186e-01 -3.73343080e-01 5.10025740e-01 -3.62083822e-01 -1.33103386e-01 -9.29536521e-01 -6.86603129e-01 9.98552382e-01 6.10490620e-01 7.47853816e-01 9.80194569e-01 -8.05405527e-02 5.60676277e-01 3.78607869e-01 2.10364580e-01 -1.01572990e+00 2.15576306e-01 9.59897220e-01 7.07738876e-01 -1.19762051e+00 -2.15621829e-01 -3.27400088e-01 -3.33002150e-01 1.40403962e+00 8.03327620e-01 -1.11178048e-01 8.22261631e-01 -2.59098411e-01 -2.95626335e-02 -1.34217665e-01 -5.82709432e-01 -4.75441754e-01 6.22983158e-01 3.45263034e-01 8.48388791e-01 -1.34121269e-01 -9.06828523e-01 9.08286214e-01 -6.42400980e-01 -1.81930125e-01 6.76549017e-01 7.50574708e-01 -7.45302856e-01 -1.36761761e+00 -4.83898938e-01 6.57815218e-01 -7.66591072e-01 -2.31446907e-01 -5.00695288e-01 8.70560765e-01 9.33057442e-02 1.42948997e+00 2.39274334e-02 -1.04996853e-01 7.06539214e-01 4.54194397e-01 5.55662334e-01 -7.07586467e-01 -1.15011203e+00 -9.17561293e-01 -3.48781538e-03 -2.71726131e-01 -3.33328694e-01 -6.49796247e-01 -5.01304090e-01 -2.83936083e-01 -3.52808565e-01 5.78780174e-01 5.08958817e-01 1.15783405e+00 -6.69092964e-03 5.07251382e-01 5.18023312e-01 1.14794821e-01 -8.96609902e-01 -1.32280016e+00 -4.13721204e-01 5.34665108e-01 2.28420541e-01 -6.41473830e-01 -4.91977066e-01 -2.05830857e-01]
[11.332818984985352, 8.322988510131836]
133393cd-61c3-4942-9951-5235147d687a
a-machine-learning-based-severity-prediction
2203.15151
null
https://arxiv.org/abs/2203.15151v1
https://arxiv.org/pdf/2203.15151v1.pdf
A machine learning-based severity prediction tool for diabetic sensorimotor polyneuropathy using Michigan neuropathy screening instrumentations
Background: Diabetic Sensorimotor polyneuropathy (DSPN) is a major long-term complication in diabetic patients associated with painful neuropathy, foot ulceration and amputation. The Michigan neuropathy screening instrument (MNSI) is one of the most common screening techniques for DSPN, however, it does not provide any direct severity grading system. Method: For designing and modelling the DSPN severity grading systems for MNSI, 19 years of data from Epidemiology of Diabetes Interventions and Complications (EDIC) clinical trials were used. MNSI variables and patient outcomes were investigated using machine learning tools to identify the features having higher association in DSPN identification. A multivariable logistic regression-based nomogram was generated and validated for DSPN severity grading. Results: The top-7 ranked features from MNSI: 10-gm filament, Vibration perception (R), Vibration perception (L), previous diabetic neuropathy, the appearance of deformities, appearance of callus and appearance of fissure were identified as key features for identifying DSPN using the extra tree model. The area under the curve (AUC) of the nomogram for the internal and external datasets were 0.9421 and 0.946, respectively. From the developed nomogram, the probability of having DSPN was predicted and a DSPN severity scoring system for MNSI was developed from the probability score. The model performance was validated on an independent dataset. Patients were stratified into four severity levels: absent, mild, moderate, and severe using a cut-off value of 10.5, 12.7 and 15 for a DSPN probability less than 50%, 75% to 90%, and above 90%, respectively. Conclusions: This study provides a simple, easy-to-use and reliable algorithm for defining the prognosis and management of patients with DSPN.
['Geetika Srivastava', 'Ahmad A. A Bakar', 'Sawal H. M. Ali', 'Iffat Ara', 'Syoji Kobashi', 'Mohammed Alhatou', 'Rayaz Malik', 'Muhammad E. H. Chowdhury', 'Mamun B. I. Reaz', 'Fahmida Haque']
2022-03-28
null
null
null
null
['severity-prediction', 'epidemiology']
['computer-vision', 'medical']
[ 2.06199244e-01 -4.11090195e-01 -6.42826855e-01 -2.34029830e-01 -6.79942191e-01 -4.04554129e-01 -1.35996386e-01 5.83607852e-01 -7.03381777e-01 1.13202238e+00 3.71628821e-01 -3.74195367e-01 -8.97496521e-01 -7.17905819e-01 -1.95397809e-01 -4.55546558e-01 -5.72743773e-01 6.11139536e-01 1.61445603e-01 1.49424300e-01 4.83520001e-01 3.35982293e-01 -1.24580371e+00 2.11212754e-01 1.29046786e+00 1.01413798e+00 5.71796298e-01 7.27060080e-01 2.26860613e-01 4.28576738e-01 -3.98823678e-01 4.87874709e-02 1.66550651e-01 -4.57630306e-01 -5.19223809e-01 -2.84327298e-01 -1.25323934e-02 -4.28949267e-01 1.74450409e-02 7.81247973e-01 9.29210663e-01 -2.55943179e-01 9.29903924e-01 -9.61126685e-01 -3.25602591e-01 4.72144604e-01 -4.76507187e-01 2.26142511e-01 3.37031782e-01 2.95914531e-01 3.03291351e-01 -1.64017081e-01 8.10098529e-01 1.17497241e+00 7.65003026e-01 2.22325325e-01 -1.47697520e+00 -6.30706787e-01 -3.20120007e-01 6.16330326e-01 -1.02116346e+00 2.90596217e-01 -1.65727288e-02 -7.53169417e-01 6.90398514e-01 3.54687989e-01 1.06127155e+00 8.98384869e-01 8.71071637e-01 2.78233513e-02 1.58011460e+00 -2.72989929e-01 2.01834455e-01 -2.49331266e-01 3.01019222e-01 1.28702134e-01 4.57577646e-01 3.34941655e-01 -4.40401174e-02 -2.56857991e-01 9.87250566e-01 -1.70719460e-01 -9.09886360e-02 -3.82158235e-02 -9.20666158e-01 7.34915316e-01 2.76471555e-01 -4.09513265e-01 -7.55700231e-01 -1.87334284e-01 7.55333424e-01 3.72943401e-01 -1.02471091e-01 5.42127565e-02 -5.41043401e-01 -5.94209850e-01 -3.34707618e-01 2.09338069e-01 4.70229119e-01 2.67241031e-01 -1.45020127e-01 -3.64989251e-01 -2.60763526e-01 1.00510550e+00 3.48605573e-01 5.16613960e-01 4.15670067e-01 -9.55086648e-01 3.43470186e-01 8.89090955e-01 -1.68161646e-01 -2.99875408e-01 -9.48124826e-01 -4.54750180e-01 -8.30217659e-01 8.13035846e-01 4.62922275e-01 -5.07446647e-01 -1.30304658e+00 1.46582627e+00 -3.46911371e-01 -5.05741954e-01 2.62201995e-01 1.14615226e+00 3.74853015e-01 2.33496398e-01 3.83130401e-01 -1.98142707e-01 1.47324705e+00 -1.55929863e-01 -2.99831152e-01 9.33190212e-02 5.82807958e-01 -7.16556251e-01 6.30037427e-01 9.37361419e-01 -8.29828560e-01 -3.26471567e-01 -6.28430605e-01 3.48131567e-01 7.55603760e-02 2.75204957e-01 5.95772862e-01 5.00186801e-01 -6.85269237e-01 7.13799059e-01 -8.68298292e-01 -9.69459593e-01 2.42740825e-01 4.73121524e-01 -3.16642731e-01 -3.25036138e-01 -1.17458475e+00 1.50135171e+00 4.02258933e-01 9.62258689e-03 -4.44487035e-01 -5.14012218e-01 -3.33972603e-01 -4.32577521e-01 -2.18249068e-01 -1.11146069e+00 5.90496898e-01 -6.72242165e-01 -1.01339710e+00 4.07407910e-01 -5.91919497e-02 -3.14857572e-01 5.23787320e-01 -2.16696158e-01 -7.16570258e-01 2.07765818e-01 2.94904828e-01 2.27885753e-01 -1.32955119e-01 -7.34657526e-01 -9.44034040e-01 -7.69583404e-01 -6.06087260e-02 4.66025561e-01 4.70379442e-01 3.44584376e-01 2.17259198e-01 -3.85098755e-01 1.75400525e-01 -6.63437307e-01 -6.44250929e-01 2.62864828e-01 -4.42616552e-01 -3.27779055e-01 3.95868927e-01 -9.60218370e-01 9.09418881e-01 -1.87461317e+00 5.47513962e-02 6.68733716e-01 2.30095252e-01 1.91800237e-01 8.80559608e-02 3.80920231e-01 -9.23277289e-02 7.89629444e-02 1.52462631e-01 6.26353800e-01 -1.20380938e-01 2.63338000e-01 6.27125502e-01 2.48711735e-01 1.94483683e-01 3.97569150e-01 -8.79257262e-01 -2.28825167e-01 6.32440627e-01 4.83155131e-01 -4.63228077e-01 -2.71168083e-01 2.98275381e-01 1.54329926e-01 -1.64786234e-01 6.57717824e-01 7.11394072e-01 1.69999197e-01 2.51275629e-01 -1.69036672e-01 -5.08365452e-01 -8.70572925e-02 -1.21788681e+00 1.14172101e+00 1.28951808e-02 5.08901358e-01 -1.23271102e-03 -5.33732772e-01 9.53266203e-01 2.48400822e-01 3.01388621e-01 -7.02150285e-01 1.00041784e-01 5.20643711e-01 6.37994647e-01 -8.95135224e-01 -5.92547715e-01 -5.52512765e-01 1.57623708e-01 -2.15120047e-01 -9.61490646e-02 5.80022991e-01 6.00788951e-01 4.16652393e-03 1.09254158e+00 -1.15240611e-01 5.92422724e-01 -2.07110658e-01 2.54306614e-01 7.36190379e-02 7.14645207e-01 7.00047314e-01 -3.30848098e-01 3.92498046e-01 1.09020591e+00 -4.70174365e-02 -6.59394681e-01 -1.55019951e+00 -8.00278127e-01 2.90443540e-01 1.03180334e-01 -5.85976802e-02 -7.25601241e-02 9.96264890e-02 5.15050709e-01 4.49908584e-01 -4.14305717e-01 -2.84739673e-01 6.22576289e-02 -9.84966636e-01 4.38799113e-01 8.96991193e-01 1.25570670e-01 -6.07253611e-01 -2.78012574e-01 2.85647392e-01 -1.35754764e-01 -2.39556715e-01 1.89369500e-01 3.79354745e-01 -1.13247311e+00 -1.27889884e+00 -9.64911640e-01 -9.98564482e-01 3.45664144e-01 -3.25619400e-01 4.27957535e-01 -4.09913868e-01 -5.04362524e-01 -2.15526059e-01 -2.86822081e-01 -3.15442502e-01 -2.15785578e-01 -5.28939605e-01 2.45339900e-01 -6.91831887e-01 5.66181302e-01 -8.41550291e-01 -1.00958240e+00 2.66706675e-01 -4.39980954e-01 -2.51821071e-01 1.48018146e+00 3.22687447e-01 5.16561985e-01 3.04865763e-02 6.74414575e-01 -3.77670377e-01 6.72968209e-01 -5.14302433e-01 -1.54540852e-01 -1.92824811e-01 -8.94974709e-01 -4.35912400e-01 2.64395773e-01 -5.01975119e-01 -3.88190627e-01 -3.35105509e-01 -6.07538074e-02 1.06658787e-01 -4.81498480e-01 1.04885805e+00 -1.21504245e-02 4.69864672e-03 6.79081023e-01 3.46893668e-02 6.75608158e-01 -6.95458174e-01 -3.25468391e-01 7.87270248e-01 3.51967335e-01 -4.61921483e-01 2.04069003e-01 6.69936836e-02 2.28903741e-01 -7.77045846e-01 -2.60257572e-01 -5.83819628e-01 -2.82588810e-01 -3.17372203e-01 9.82798636e-01 -7.15909660e-01 -9.71980989e-01 6.03241563e-01 -7.11395860e-01 -9.56857726e-02 2.00020209e-01 1.35092974e+00 -3.12655300e-01 5.32411158e-01 -4.98712122e-01 -6.49404526e-01 -4.37846959e-01 -1.15461826e+00 1.87703714e-01 2.72930920e-01 -6.56727731e-01 -6.65042102e-01 4.04692501e-01 2.63866752e-01 5.27449489e-01 8.10616195e-01 1.72026765e+00 -3.02098870e-01 1.31649070e-03 -4.04543579e-01 -5.74053884e-01 6.39010012e-01 2.56607234e-01 2.28381395e-01 1.08198553e-01 1.04407351e-02 -5.15422344e-01 -1.48452416e-01 6.03448570e-01 1.13986373e+00 7.02850580e-01 1.02115080e-01 -2.63480544e-01 3.63143116e-01 2.01982570e+00 9.09753144e-01 1.04842091e+00 7.44733930e-01 2.36215621e-01 4.84933704e-01 3.61964613e-01 3.39328080e-01 2.52019554e-01 5.41969538e-01 5.54188013e-01 -1.86662912e-01 -1.62831798e-01 2.43209586e-01 2.66658038e-01 8.07342827e-02 -9.47941542e-01 1.67524204e-01 -8.74020934e-01 4.37360704e-01 -1.26902139e+00 -8.35887074e-01 -8.91844869e-01 2.21665931e+00 5.70317864e-01 7.70475030e-01 6.02993131e-01 1.69529170e-01 8.32105696e-01 -7.80872464e-01 -4.64599311e-01 -6.99832201e-01 -1.36104599e-01 4.17974651e-01 8.17622066e-01 2.03877375e-01 -6.13405526e-01 7.42632747e-02 6.33354712e+00 8.24417025e-02 -9.83060539e-01 -3.78967583e-01 -1.62747368e-01 -2.16724709e-01 3.77165407e-01 2.06851184e-01 -4.40255344e-01 8.35938275e-01 9.04961765e-01 -2.02700794e-01 8.79858062e-02 3.59158516e-01 9.06996429e-01 -5.35956681e-01 -4.86217827e-01 4.06877816e-01 -4.45370793e-01 -1.05735731e+00 1.97677910e-01 2.44943589e-01 1.43280938e-01 2.18729258e-01 -1.81469899e-02 1.94125563e-01 1.43629715e-01 -8.80166411e-01 3.82911474e-01 9.02392626e-01 1.06944060e+00 -7.13448703e-01 1.61098790e+00 -2.57866383e-01 -6.85355484e-01 -1.90452158e-01 -8.58135670e-02 -3.94123733e-01 6.18157744e-01 5.09989679e-01 -7.59109259e-01 5.56834877e-01 6.31670475e-01 2.98228115e-01 -2.71336138e-01 1.65852582e+00 9.43793207e-02 6.48065329e-01 -2.17824161e-01 -9.00311098e-02 1.97992399e-01 -2.95390934e-01 6.03710175e-01 8.15206468e-01 2.79048473e-01 1.54243475e-02 -1.13069892e-01 5.25945246e-01 7.80474782e-01 1.94520682e-01 1.41677484e-01 1.39996596e-02 3.45791101e-01 6.25925899e-01 -4.78488237e-01 1.53896242e-01 -6.82737172e-01 2.93920487e-01 -4.71466184e-01 3.11947167e-01 -5.25746882e-01 -9.06514704e-01 6.81980133e-01 5.27841747e-01 -2.13020995e-01 2.08802849e-01 -7.73454368e-01 -3.34152699e-01 -4.78504300e-02 -5.63679636e-01 5.92758179e-01 -8.18630636e-01 -1.44974089e+00 2.18816046e-02 -2.28452489e-01 -1.34629953e+00 1.07028179e-01 -9.65620339e-01 -7.25425601e-01 1.48544884e+00 -1.00615108e+00 -7.74539769e-01 -3.30912858e-01 3.63119066e-01 4.05064858e-02 1.94629263e-02 1.09033406e+00 2.05306202e-01 -5.97732484e-01 1.65355176e-01 4.97885257e-01 1.60331070e-01 6.99002922e-01 -1.37968051e+00 -7.07790315e-01 2.88253069e-01 -1.53099275e+00 7.81580269e-01 7.45177984e-01 -1.11363637e+00 -8.20579648e-01 -8.75765622e-01 7.08915472e-01 1.45075843e-01 7.80653536e-01 4.57463562e-01 -3.60178560e-01 4.18844640e-01 -3.83899659e-01 -9.95429933e-01 1.29315960e+00 3.28758657e-01 1.83102161e-01 -5.50064258e-02 -1.30880439e+00 4.51706350e-01 6.85729623e-01 9.80402678e-02 -5.81015766e-01 1.14045411e-01 -1.35282800e-01 -1.37204573e-01 -1.68215728e+00 5.84628701e-01 1.18545771e+00 -7.51214325e-01 9.29267645e-01 -7.54488766e-01 6.98504090e-01 -1.12678379e-01 -1.90359384e-01 -9.48357701e-01 -5.42335153e-01 -1.86341971e-01 3.63432199e-01 7.85752118e-01 4.89165455e-01 -8.43376040e-01 6.35417283e-01 3.67017508e-01 -3.25130224e-01 -9.00506318e-01 -1.03107083e+00 -7.67886937e-01 4.23193015e-02 -4.67085689e-02 -3.29425126e-01 4.96005744e-01 3.94775152e-01 1.66855261e-01 1.90558374e-01 9.39164311e-02 7.72358418e-01 -2.77010381e-01 2.28498772e-01 -1.75388026e+00 -8.45160633e-02 -5.19891620e-01 -1.29076636e+00 -1.06005192e-01 -7.98092067e-01 -1.04198110e+00 -7.18274176e-01 -2.72482944e+00 5.46820760e-01 -3.88572693e-01 -5.81506312e-01 2.74281800e-01 1.35417536e-01 -1.89216673e-01 -4.39349674e-02 1.01948045e-01 3.16315114e-01 -4.89790626e-02 1.07149053e+00 3.12159240e-01 -4.90935445e-01 5.36253393e-01 -8.55692327e-01 5.58411837e-01 1.01240408e+00 -3.16032588e-01 -4.89263713e-01 4.24240753e-02 -2.84254216e-02 3.70235056e-01 7.70089090e-01 -9.77424681e-01 -3.66915971e-01 -6.02554262e-01 8.19563031e-01 -8.74543011e-01 2.79696316e-01 -4.47813302e-01 3.10187191e-01 1.07592452e+00 6.07139319e-02 -2.11493388e-01 1.67104647e-01 3.52129757e-01 1.85780346e-01 -4.50366102e-02 8.28759730e-01 8.53547975e-02 -7.72901356e-01 -2.79339999e-01 -1.00630391e+00 -3.62413436e-01 1.01850843e+00 -8.51128876e-01 -6.13836944e-01 -1.33848637e-01 -1.32015872e+00 3.79754603e-01 3.35070431e-01 1.30837515e-01 6.60131574e-01 -1.15631926e+00 -9.70653057e-01 -2.99374998e-01 1.80954397e-01 -2.22054690e-01 5.91007650e-01 1.57555139e+00 -9.47326183e-01 4.18040395e-01 -1.07385385e+00 -5.00342548e-01 -1.40652418e+00 5.93470410e-02 2.32105449e-01 -5.12762107e-02 -7.49284923e-01 5.35733044e-01 -3.84345502e-01 1.07651934e-01 4.43593472e-01 -5.69286346e-01 -4.16056842e-01 -1.04471579e-01 3.85478318e-01 1.11321664e+00 -1.21102832e-01 -1.94282040e-01 -5.78595459e-01 7.45047808e-01 -9.87213925e-02 -1.99651457e-02 1.25595593e+00 -2.11255953e-01 -3.52976233e-01 2.99021572e-01 9.61860299e-01 -2.69010633e-01 -6.48720920e-01 3.33592027e-01 -3.12348157e-01 -4.40222025e-02 -4.48667929e-02 -1.61904633e+00 -4.07628447e-01 3.16767454e-01 1.45200598e+00 2.98868604e-02 1.23673880e+00 6.01341203e-02 7.14971483e-01 -2.56502807e-01 1.22190826e-01 -8.91635120e-01 -5.45000017e-01 1.68557316e-01 7.14350045e-01 -6.96550071e-01 -1.91801965e-01 -3.94386321e-01 -5.63123882e-01 1.21202612e+00 3.50655019e-01 -2.96978742e-01 5.42824090e-01 -3.64925861e-02 3.19547802e-01 -1.02721706e-01 -5.37246466e-01 -2.40892135e-02 8.21999162e-02 1.17121875e+00 2.11954027e-01 5.93804955e-01 -1.44147265e+00 9.62293446e-01 -1.52769610e-01 5.39781749e-01 6.67240739e-01 8.16268325e-01 -6.67252600e-01 -1.06662321e+00 -2.26522729e-01 1.39331329e+00 -2.70642668e-01 2.43924171e-01 -3.98977488e-01 1.05009758e+00 3.77565503e-01 5.92585385e-01 -7.32958987e-02 -4.99270827e-01 7.95294940e-01 -6.78017884e-02 4.15399373e-01 -3.50003272e-01 -1.24595776e-01 1.51808530e-01 5.45131385e-01 -1.93794921e-01 -3.70821208e-01 -7.63566136e-01 -1.45853055e+00 -4.46529567e-01 -1.32665515e-01 -4.85843308e-02 7.07261086e-01 8.88916969e-01 8.60841200e-02 5.75027943e-01 9.81583595e-02 -3.12380999e-01 -5.13948262e-01 -1.26123691e+00 -1.16187251e+00 -5.52666746e-02 1.56419262e-01 -1.07648957e+00 -3.67021799e-01 -2.89301097e-01]
[14.436646461486816, -1.7782269716262817]
ff26442c-b0fd-4209-86af-9b8452dcca8e
contentctr-frame-level-live-streaming-click
2306.14392
null
https://arxiv.org/abs/2306.14392v1
https://arxiv.org/pdf/2306.14392v1.pdf
ContentCTR: Frame-level Live Streaming Click-Through Rate Prediction with Multimodal Transformer
In recent years, live streaming platforms have gained immense popularity as they allow users to broadcast their videos and interact in real-time with hosts and peers. Due to the dynamic changes of live content, accurate recommendation models are crucial for enhancing user experience. However, most previous works treat the live as a whole item and explore the Click-through-Rate (CTR) prediction framework on item-level, neglecting that the dynamic changes that occur even within the same live room. In this paper, we proposed a ContentCTR model that leverages multimodal transformer for frame-level CTR prediction. First, we present an end-to-end framework that can make full use of multimodal information, including visual frames, audio, and comments, to identify the most attractive live frames. Second, to prevent the model from collapsing into a mediocre solution, a novel pairwise loss function with first-order difference constraints is proposed to utilize the contrastive information existing in the highlight and non-highlight frames. Additionally, we design a temporal text-video alignment module based on Dynamic Time Warping to eliminate noise caused by the ambiguity and non-sequential alignment of visual and textual information. We conduct extensive experiments on both real-world scenarios and public datasets, and our ContentCTR model outperforms traditional recommendation models in capturing real-time content changes. Moreover, we deploy the proposed method on our company platform, and the results of online A/B testing further validate its practical significance.
['Gaofeng Meng', 'Guorui Zhou', 'Fan Yang', 'Xiangyu Wu', 'Shiyao Wang', 'Dong Shen', 'Jiaxin Deng']
2023-06-26
null
null
null
null
['video-alignment', 'click-through-rate-prediction', 'dynamic-time-warping']
['computer-vision', 'miscellaneous', 'time-series']
[ 1.98330879e-02 -8.17341447e-01 -2.46763736e-01 -2.36823887e-01 -5.83750784e-01 -6.58716321e-01 2.92975515e-01 -2.84887310e-02 -2.30608508e-01 1.86202124e-01 5.83981276e-01 5.01165614e-02 -6.08510561e-02 -4.57226157e-01 -6.41343474e-01 -6.30154669e-01 -4.28498209e-01 -3.05214435e-01 4.61754024e-01 -2.39713281e-01 2.44299129e-01 5.59864379e-02 -1.65386569e+00 6.17211878e-01 7.32229590e-01 1.33019638e+00 3.28633636e-01 6.22631729e-01 7.37284571e-02 8.45543802e-01 -3.56513768e-01 -5.09213924e-01 3.94659549e-01 -5.87551184e-02 -6.87305331e-02 1.06432945e-01 4.97606546e-01 -7.38338590e-01 -7.45523036e-01 6.88265026e-01 6.83609962e-01 3.94402027e-01 1.15049474e-01 -1.27128708e+00 -4.20205057e-01 4.07574564e-01 -8.85845661e-01 3.74447972e-01 7.58681357e-01 1.95130557e-01 1.19700575e+00 -1.01399863e+00 5.36631346e-01 1.14667511e+00 4.02951866e-01 -6.19247444e-02 -7.96743870e-01 -7.70641029e-01 5.30241549e-01 7.17957497e-01 -1.29733682e+00 -3.67125064e-01 7.99158633e-01 -2.45949879e-01 6.03721738e-01 4.81166631e-01 8.16013575e-01 1.15604782e+00 9.92217585e-02 9.51037109e-01 4.33296591e-01 -4.97103482e-02 -1.26768097e-01 -2.42785379e-01 -3.71453673e-01 2.67711848e-01 -2.45263979e-01 -1.16949409e-01 -9.70610738e-01 -1.10227294e-01 6.35885477e-01 5.30263007e-01 -4.78852361e-01 -2.27153614e-01 -1.48267531e+00 5.04773021e-01 1.98489636e-01 1.46922633e-01 -3.80325466e-01 3.55857164e-02 4.93995130e-01 3.17081779e-01 5.17233968e-01 -1.43895611e-01 -1.49265632e-01 -4.34480608e-01 -1.09408760e+00 4.19310406e-02 5.12186885e-01 9.41369832e-01 3.23928148e-01 -1.40015423e-01 -4.43587333e-01 9.40476656e-01 3.83341968e-01 5.24711728e-01 4.98046786e-01 -9.91125286e-01 7.41596878e-01 1.49734154e-01 1.21829562e-01 -1.63844907e+00 6.46166280e-02 -4.11352843e-01 -6.53577507e-01 -6.27292275e-01 2.23002627e-01 4.40933742e-02 -2.26770118e-01 1.44385588e+00 3.93740177e-01 8.01150024e-01 -5.37118375e-01 1.21018422e+00 6.13157034e-01 9.09096181e-01 -1.93633035e-01 -6.29115164e-01 1.17643535e+00 -1.02765012e+00 -7.63736308e-01 2.91026950e-01 2.34996319e-01 -9.89579976e-01 1.08992088e+00 6.40984356e-01 -1.23443961e+00 -6.11183226e-01 -9.32963967e-01 3.78078818e-02 2.05474436e-01 -1.07104495e-01 5.26179560e-02 2.56360888e-01 -7.31073022e-01 3.85752410e-01 -6.70126617e-01 -1.37032688e-01 5.08827791e-02 9.23211426e-02 -1.53262436e-01 -3.53357852e-01 -1.09276021e+00 9.39736143e-03 -9.37267765e-02 1.43794805e-01 -7.55697787e-01 -6.78371251e-01 -3.56992900e-01 1.96449250e-01 7.97936916e-01 -2.55796283e-01 1.22833252e+00 -1.20442474e+00 -1.50458264e+00 1.89148650e-01 -3.48404914e-01 -2.31195375e-01 5.24636149e-01 -4.70806360e-01 -6.52109623e-01 5.14592171e-01 -3.32887411e-01 2.53051341e-01 9.61647451e-01 -1.06526768e+00 -9.41524804e-01 -9.25073549e-02 3.13278675e-01 3.38218451e-01 -8.91290188e-01 1.46972403e-01 -1.09397185e+00 -1.04611874e+00 5.93701638e-02 -8.81471157e-01 2.52206415e-01 1.41605750e-01 -1.51997969e-01 2.04262823e-01 9.68074679e-01 -8.55975032e-01 1.70400488e+00 -2.41891026e+00 9.29108262e-02 3.41847867e-01 2.50850886e-01 1.04395747e-01 -4.04192120e-01 6.44758999e-01 1.52572542e-01 -1.84600428e-02 4.14077967e-01 -2.57715374e-01 -1.41775548e-01 -4.71696220e-02 -4.77342248e-01 3.75108659e-01 -3.22247475e-01 4.78704154e-01 -8.98394883e-01 -5.87752938e-01 1.88366741e-01 6.38963938e-01 -9.10106719e-01 3.59707922e-01 -1.16365477e-01 5.46962082e-01 -4.45908368e-01 6.02338850e-01 6.74393594e-01 -2.71471024e-01 1.59382358e-01 -5.64941585e-01 -3.00430596e-01 2.16587558e-01 -1.27235878e+00 1.60644662e+00 -5.32291949e-01 6.15000486e-01 1.32939219e-01 -7.05525935e-01 5.14415860e-01 3.73394549e-01 9.31371629e-01 -1.15105987e+00 -1.07323445e-01 -3.94900925e-02 -2.45940238e-01 -6.98119700e-01 7.44867384e-01 5.60270131e-01 2.81228483e-01 3.15097451e-01 -3.73479396e-01 6.16817296e-01 1.92379743e-01 4.45110351e-01 1.05545866e+00 8.05985630e-02 -2.55263373e-02 4.01449472e-01 5.73387861e-01 -5.59998155e-01 6.08910978e-01 4.56435889e-01 -1.75328478e-01 9.33446467e-01 3.49580288e-01 -2.51634389e-01 -8.44468713e-01 -8.65785837e-01 2.05677062e-01 1.47628295e+00 6.18674695e-01 -7.70921826e-01 -3.33362848e-01 -3.46170962e-01 -1.97424531e-01 4.19545889e-01 -2.68891305e-01 -2.83613009e-03 -6.76202059e-01 -3.79025280e-01 1.05018951e-02 1.14536054e-01 4.33833897e-01 -7.04531372e-01 -4.34818476e-01 3.47151935e-01 -8.77181530e-01 -1.31385040e+00 -1.19698679e+00 -7.61568069e-01 -6.04006588e-01 -9.30960774e-01 -8.18074763e-01 -4.73372221e-01 1.91644251e-01 1.20639992e+00 8.94850075e-01 2.66512632e-01 -1.05672471e-01 7.27279842e-01 -9.19772923e-01 3.08994412e-01 2.19357148e-01 -2.55736530e-01 -1.93171486e-01 5.83703756e-01 -1.53422132e-01 -6.87520087e-01 -1.20322418e+00 7.95252264e-01 -1.04872787e+00 1.61362723e-01 1.91378728e-01 5.92228174e-01 7.10520923e-01 -1.24958921e-02 3.14109921e-01 -3.07465553e-01 4.06549841e-01 -8.65981281e-01 -2.48326451e-01 2.83116579e-01 -3.84339452e-01 -7.16111541e-01 5.77452779e-01 -7.49706089e-01 -1.05079532e+00 -3.97180527e-01 1.59418508e-02 -7.31388450e-01 3.58017802e-01 6.58137143e-01 -2.64149513e-02 1.73238277e-01 3.23116593e-02 1.57651424e-01 -1.92952618e-01 -4.45595950e-01 2.83306003e-01 7.18693554e-01 4.23976779e-01 -4.75785196e-01 7.32382536e-01 6.84049845e-01 -3.10712188e-01 -7.90419757e-01 -6.19171441e-01 -8.30556870e-01 -1.33202329e-01 -6.46995962e-01 4.57405537e-01 -1.23711944e+00 -8.87093246e-01 2.48307109e-01 -9.66908991e-01 2.47736592e-02 2.80850753e-02 5.54158747e-01 -3.08656603e-01 8.61718416e-01 -7.17414975e-01 -6.41854048e-01 -3.81957084e-01 -1.12537348e+00 8.73447239e-01 1.99188769e-01 1.74138650e-01 -5.44272661e-01 3.81592065e-02 4.89104718e-01 4.19879347e-01 -9.07975808e-02 3.82430732e-01 -4.11772966e-01 -7.96556294e-01 -2.69932598e-01 -3.25236320e-01 2.90584601e-02 -4.05207090e-02 4.13247675e-01 -7.07118988e-01 -5.84010363e-01 -1.37477577e-01 -1.60673074e-02 5.89922369e-01 3.43860984e-01 1.51017511e+00 -6.07143402e-01 -7.14072660e-02 6.96061969e-01 1.21315956e+00 2.17571452e-01 7.60772288e-01 2.33998552e-01 7.82235324e-01 5.44938087e-01 1.02254510e+00 9.94551539e-01 6.57820761e-01 1.14434791e+00 7.55488634e-01 1.90172851e-01 3.41858752e-02 -3.67226571e-01 7.04721510e-01 1.20628417e+00 -1.94473311e-01 -6.65419221e-01 -3.02595705e-01 4.68481541e-01 -2.05222964e+00 -1.38396406e+00 2.06399579e-02 2.55718637e+00 5.25314927e-01 -2.39335988e-02 3.90300125e-01 -1.13404028e-01 7.70098150e-01 2.41334736e-01 -4.24533695e-01 8.01969916e-02 -1.51382372e-01 -1.61490649e-01 2.02197492e-01 6.41900301e-02 -8.35180879e-01 3.95838320e-01 5.02740240e+00 1.17599165e+00 -1.34464729e+00 1.70019224e-01 5.87849319e-01 -7.43139088e-01 -4.06442523e-01 -1.61585629e-01 -3.48388255e-01 8.96781087e-01 7.63733029e-01 -6.62107244e-02 6.55939639e-01 4.13412213e-01 9.12572801e-01 -2.67235413e-02 -8.78395319e-01 1.15980601e+00 1.20046817e-01 -1.17592406e+00 -9.75653529e-02 1.12997010e-01 5.98139644e-01 -8.48976523e-02 2.79096335e-01 1.59859002e-01 -2.61642247e-01 -5.68607390e-01 1.00073051e+00 4.83640969e-01 6.46457553e-01 -5.73385477e-01 4.61822301e-01 1.03780039e-01 -1.72661042e+00 -3.09353650e-01 -1.06928617e-01 1.92377627e-01 4.55541432e-01 6.02102578e-01 -1.87743887e-01 6.18468881e-01 9.20842171e-01 1.14516187e+00 -4.58210200e-01 1.37360013e+00 2.23892063e-01 7.49093652e-01 -2.01550946e-01 2.66903490e-01 -9.82804894e-02 -3.12262207e-01 6.67938292e-01 1.24700820e+00 7.85792530e-01 2.18355030e-01 2.07242936e-01 1.91893548e-01 -7.79727250e-02 4.49249625e-01 -1.07943185e-01 5.48437349e-02 6.03773415e-01 1.36471260e+00 -4.98787105e-01 -6.73611686e-02 -9.45686996e-01 1.00088727e+00 -1.03213169e-01 5.05854011e-01 -1.32510817e+00 7.03732949e-03 6.75007224e-01 3.15922379e-01 6.57535136e-01 -2.79174477e-01 3.71136785e-01 -1.54126489e+00 3.54971975e-01 -1.05793989e+00 3.17094088e-01 -7.98264861e-01 -1.27417266e+00 2.29355067e-01 -6.40333295e-02 -1.84216845e+00 -1.59293860e-02 -2.46832073e-02 -7.22572327e-01 2.37172320e-01 -1.45870876e+00 -9.23195422e-01 -5.06326318e-01 8.73789251e-01 7.25433767e-01 7.94026926e-02 1.10803246e-01 9.82896090e-01 -6.65867627e-01 9.64416385e-01 3.76228005e-01 -1.75661296e-01 1.18168974e+00 -5.58170319e-01 -1.34658918e-01 7.90210366e-01 1.68871328e-01 5.20081043e-01 6.81862950e-01 -4.47183698e-01 -1.76150727e+00 -8.86585116e-01 3.85301203e-01 2.11600676e-01 6.76386714e-01 -1.97943285e-01 -8.30980062e-01 3.72113764e-01 8.59195888e-02 1.26275912e-01 7.15956390e-01 -1.44975767e-01 -5.41801751e-01 -4.57401186e-01 -7.95644283e-01 6.89196765e-01 1.09980786e+00 -4.72026438e-01 4.09127697e-02 7.55760297e-02 9.62310374e-01 -2.95593560e-01 -9.10616994e-01 2.89717734e-01 1.06694210e+00 -1.21135485e+00 1.14898098e+00 -9.68521610e-02 5.24265945e-01 -2.98661560e-01 -3.42237353e-01 -8.66938651e-01 -1.16429709e-01 -8.63882124e-01 -3.66099864e-01 1.35236907e+00 -4.85323630e-02 -2.50653833e-01 5.04485309e-01 4.14403170e-01 -5.81817515e-02 -5.97635865e-01 -8.70133102e-01 -4.18467343e-01 -7.88476586e-01 -4.98130113e-01 4.05842513e-01 8.01018536e-01 1.11842006e-01 2.82200892e-02 -9.27739263e-01 1.43856052e-02 3.85245889e-01 2.49823570e-01 8.07389259e-01 -9.35192645e-01 -6.46626830e-01 -1.97759911e-01 -1.66148528e-01 -1.51845121e+00 -3.37373495e-01 -3.28245223e-01 -9.22293141e-02 -1.05639088e+00 4.96618718e-01 -2.40458831e-01 -4.23767239e-01 3.34736258e-02 2.71241106e-02 3.01549256e-01 6.04419172e-01 5.93719065e-01 -1.13304853e+00 6.83581054e-01 1.33018577e+00 -4.73409854e-02 -2.43970007e-01 -1.05664060e-02 -3.89611065e-01 5.59087336e-01 3.13648313e-01 -1.68929115e-01 -4.75542754e-01 -4.86982644e-01 6.32681251e-01 4.19468313e-01 2.27362260e-01 -8.09667170e-01 2.80010223e-01 -2.73868114e-01 3.00642718e-02 -6.94574833e-01 4.91631836e-01 -1.15824950e+00 2.76854903e-01 4.31222431e-02 -3.59517515e-01 2.50759721e-01 -1.87092237e-02 9.51252162e-01 -3.74622554e-01 2.01653495e-01 4.28082883e-01 3.85107070e-01 -4.48049575e-01 6.42599106e-01 -1.80142745e-01 -4.96848822e-02 9.89647090e-01 -3.14836293e-01 -3.38632405e-01 -9.40432250e-01 -4.70262855e-01 3.91371697e-01 4.59545642e-01 6.35337710e-01 7.52238095e-01 -1.35089552e+00 -6.13989592e-01 -5.02711162e-02 1.66238144e-01 -3.54820132e-01 8.16357851e-01 1.13554335e+00 -3.78270537e-01 1.09610178e-01 2.45985966e-02 -6.22519612e-01 -1.46850157e+00 6.69031978e-01 -4.07487564e-02 -2.54938990e-01 -5.56904912e-01 4.82573628e-01 3.65805626e-01 2.79504180e-01 5.82366228e-01 -2.72559077e-02 -3.65540415e-01 2.74388760e-01 8.26664269e-01 4.84898657e-01 -1.13833234e-01 -7.94779539e-01 -8.84421170e-03 5.45029581e-01 -1.72016382e-01 -3.23168933e-02 1.31098664e+00 -9.17771578e-01 1.85854390e-01 3.67108822e-01 1.29917550e+00 4.32195514e-01 -1.35472035e+00 -4.65210021e-01 -5.70119560e-01 -8.96564424e-01 6.86460808e-02 -3.94026637e-01 -1.45996296e+00 7.78777897e-01 7.31725812e-01 2.68144459e-01 1.37952864e+00 -5.22181809e-01 1.29541492e+00 5.70303313e-02 2.72682965e-01 -1.01768935e+00 5.83706796e-01 2.20030218e-01 7.56958425e-01 -1.14259410e+00 5.62090008e-03 -4.36372101e-01 -5.97456694e-01 1.08826256e+00 4.42707062e-01 1.52949706e-01 6.60428643e-01 -1.09001860e-01 -2.69352645e-01 2.51715124e-01 -1.04222703e+00 1.04312703e-01 3.70044678e-01 2.53396600e-01 4.65350538e-01 -1.83094323e-01 -2.84031510e-01 5.04824996e-01 3.36523533e-01 3.01009673e-03 4.93911594e-01 6.10222161e-01 -3.20585370e-01 -9.82808769e-01 -4.22315985e-01 3.36034954e-01 -6.63783848e-01 -1.74557656e-01 2.11291656e-01 3.18446934e-01 -3.68981063e-02 1.32609701e+00 1.43859118e-01 -7.67362237e-01 2.74395704e-01 -5.39131820e-01 2.05811918e-01 -1.07377455e-01 -5.63793302e-01 7.02435017e-01 -1.37910068e-01 -9.06756103e-01 -4.94947374e-01 -7.49626279e-01 -1.02994299e+00 -6.99603617e-01 -2.59263068e-01 -6.27981573e-02 4.87091541e-01 7.38081157e-01 6.05374634e-01 4.52918768e-01 1.23423707e+00 -1.13158381e+00 -3.07294130e-01 -6.22386277e-01 -4.84181762e-01 6.46227002e-01 3.36018980e-01 -4.65944827e-01 -4.05066013e-01 2.12198496e-01]
[10.004186630249023, 0.470393568277359]
c7341715-560b-41da-8edf-62cf57417142
try-to-avoid-attacks-a-federated-data
2211.01592
null
https://arxiv.org/abs/2211.01592v1
https://arxiv.org/pdf/2211.01592v1.pdf
Try to Avoid Attacks: A Federated Data Sanitization Defense for Healthcare IoMT Systems
Healthcare IoMT systems are becoming intelligent, miniaturized, and more integrated into daily life. As for the distributed devices in the IoMT, federated learning has become a topical area with cloud-based training procedures when meeting data security. However, the distribution of IoMT has the risk of protection from data poisoning attacks. Poisoned data can be fabricated by falsifying medical data, which urges a security defense to IoMT systems. Due to the lack of specific labels, the filtering of malicious data is a unique unsupervised scenario. One of the main challenges is finding robust data filtering methods for various poisoning attacks. This paper introduces a Federated Data Sanitization Defense, a novel approach to protect the system from data poisoning attacks. To solve this unsupervised problem, we first use federated learning to project all the data to the subspace domain, allowing unified feature mapping to be established since the data is stored locally. Then we adopt the federated clustering to re-group their features to clarify the poisoned data. The clustering is based on the consistent association of data and its semantics. After we get the clustering of the private data, we do the data sanitization with a simple yet efficient strategy. In the end, each device of distributed ImOT is enabled to filter malicious data according to federated data sanitization. Extensive experiments are conducted to evaluate the efficacy of the proposed defense method against data poisoning attacks. Further, we consider our approach in the different poisoning ratios and achieve a high Accuracy and a low attack success rate.
['Siquan Huang', 'Leyu Shi', 'Ying Gao', 'Chong Chen']
2022-11-03
null
null
null
null
['data-poisoning']
['adversarial']
[-1.57053381e-01 -4.68919426e-01 -1.35305122e-01 7.88998604e-02 -2.19536379e-01 -6.68214381e-01 1.29553929e-01 2.16873765e-01 -3.36407155e-01 2.72376418e-01 1.06854014e-01 -1.63316324e-01 -3.76660675e-01 -9.20043647e-01 -2.99346507e-01 -1.17324328e+00 2.00077564e-01 3.71291161e-01 1.87725034e-02 8.48004073e-02 -1.25726372e-01 5.56156456e-01 -1.33166170e+00 4.01624799e-01 1.10884809e+00 1.07526600e+00 -6.89206049e-02 1.14477307e-01 -1.77285746e-01 6.39480174e-01 -8.52209806e-01 -5.20620346e-01 4.98941481e-01 -2.03915194e-01 -6.15705490e-01 -1.32063314e-01 -3.47545624e-01 -2.87441105e-01 -3.58086139e-01 1.38830137e+00 6.61987007e-01 -3.85598034e-01 4.23265874e-01 -1.61667013e+00 -2.31227726e-01 4.80630457e-01 -2.44305938e-01 1.61975157e-02 1.34222358e-01 2.94612139e-01 3.10106218e-01 -2.06396222e-01 4.43715721e-01 9.90196347e-01 3.37275356e-01 7.61674762e-01 -7.14522660e-01 -9.65630949e-01 -8.26868322e-03 2.53765851e-01 -1.39647794e+00 -2.60146372e-02 7.23090231e-01 -2.57640928e-01 1.86190814e-01 5.39466739e-01 3.63003790e-01 8.76608491e-01 4.79185432e-01 6.64129317e-01 9.08873796e-01 4.35848311e-02 5.14849901e-01 2.89492458e-01 1.86280504e-01 4.32610959e-01 7.88592756e-01 -1.18872590e-01 -2.48864457e-01 -6.60955846e-01 4.62533049e-02 9.88262594e-01 -2.22239748e-01 -3.63371998e-01 -1.06979251e+00 5.59317470e-01 4.07693058e-01 4.20223445e-01 -2.90282816e-01 -3.55089575e-01 6.49903953e-01 1.06801204e-01 8.37779120e-02 8.85016471e-02 -6.80718064e-01 3.30101222e-01 -2.68668383e-01 -1.54026404e-01 8.36381376e-01 7.44922400e-01 4.43125427e-01 -1.09190322e-01 1.46062449e-01 2.32372582e-01 1.69598088e-01 8.56685281e-01 6.16094530e-01 -4.50871259e-01 5.05322218e-01 9.27646399e-01 1.14898533e-01 -1.19688380e+00 -2.42185071e-01 -2.64603466e-01 -1.15690994e+00 -3.94989513e-02 1.57330215e-01 -3.43250573e-01 -5.43153644e-01 1.45700288e+00 7.42586195e-01 2.40489259e-01 5.15753210e-01 7.45340645e-01 4.74292517e-01 4.63287652e-01 2.59449065e-01 -5.82274377e-01 1.30786610e+00 -1.81560263e-01 -1.13483560e+00 7.17792034e-01 6.54049397e-01 -4.56939906e-01 6.42065942e-01 8.31371427e-01 -2.98865497e-01 -2.60520995e-01 -1.11445677e+00 5.27065039e-01 -4.93610591e-01 -3.06984454e-01 5.35297096e-01 1.14557147e+00 -2.17387021e-01 4.03809100e-01 -9.00962353e-01 -4.48569864e-01 7.22006738e-01 8.10995638e-01 -3.17634434e-01 -2.75306553e-01 -1.25143957e+00 5.34304261e-01 3.75389904e-01 -2.28501007e-01 -8.95598173e-01 -5.81849158e-01 -4.77409005e-01 -1.29870653e-01 1.59352466e-01 -6.45339847e-01 8.39785278e-01 -5.37440121e-01 -8.41870964e-01 4.20963824e-01 4.14733738e-01 -5.43880343e-01 3.71297866e-01 3.33913937e-02 -1.01373434e+00 3.09485942e-01 1.28890932e-01 -2.27426499e-01 8.10913861e-01 -1.12190950e+00 -9.74230111e-01 -9.04076219e-01 -3.74082655e-01 6.60180822e-02 -1.19354939e+00 8.24031830e-02 1.28947161e-02 -2.54405171e-01 1.64117292e-01 -5.57269454e-01 -3.57831001e-01 -3.26948762e-01 -4.96206909e-01 -1.61961466e-01 1.33813536e+00 -4.31149572e-01 1.32548141e+00 -2.44303942e+00 -5.85498363e-02 4.10650730e-01 5.16958773e-01 3.71749073e-01 4.14641678e-01 3.52609366e-01 9.84408483e-02 2.66020566e-01 -1.94946155e-01 -2.13014279e-02 -1.79843679e-01 2.36604944e-01 -3.45160335e-01 7.98103392e-01 -4.43790585e-01 2.97036558e-01 -8.03862512e-01 -5.91545045e-01 3.49182785e-01 1.63992554e-01 -2.64005601e-01 2.96420008e-01 -7.90622979e-02 7.36805797e-01 -1.00771940e+00 7.69947290e-01 1.05916321e+00 -1.16474316e-01 4.29292619e-01 -6.62144661e-01 4.85678837e-02 -4.09690231e-01 -1.22972274e+00 1.52734673e+00 -1.44032329e-01 -6.37227297e-01 9.50131491e-02 -7.51049995e-01 9.39968348e-01 4.39998269e-01 1.29198897e+00 -4.53794599e-01 6.55244648e-01 4.27139193e-01 -1.53329208e-01 -8.65469277e-01 -2.43287697e-01 6.83685485e-03 -3.36333513e-01 6.50204837e-01 -1.94843575e-01 5.73372364e-01 -5.06345570e-01 2.59414673e-01 1.32092810e+00 -4.63640541e-01 -9.20959190e-02 -3.28001589e-01 7.41785228e-01 1.46485850e-01 6.87914431e-01 2.29892775e-01 -2.15067759e-01 7.36460164e-02 -1.01834513e-01 -6.55283868e-01 -6.56927645e-01 -1.08123302e+00 -2.91756302e-01 2.84104943e-01 6.73701406e-01 -3.27111065e-01 -8.12246740e-01 -1.12157416e+00 1.88242316e-01 3.89988303e-01 -2.22418234e-01 -7.01667070e-01 -4.18789566e-01 -8.70762825e-01 6.99259579e-01 1.74895655e-02 8.59315336e-01 -9.41824794e-01 -7.27605343e-01 3.79827898e-03 -7.31260329e-02 -6.18575692e-01 -2.36840948e-01 3.11138004e-01 -6.08040929e-01 -1.51141596e+00 -1.73529163e-02 -6.32183313e-01 6.64347827e-01 3.16269189e-01 2.19766542e-01 3.94701600e-01 -3.61474097e-01 3.36991400e-01 -3.07829887e-01 -4.31349814e-01 -5.32516599e-01 -2.13972658e-01 7.98641920e-01 5.39861143e-01 6.69550121e-01 -4.30782974e-01 -6.89723849e-01 3.19245577e-01 -1.20890486e+00 -6.03712678e-01 1.99956074e-01 6.20777428e-01 4.42888826e-01 8.15674305e-01 6.92748129e-01 -9.54695046e-01 4.42401439e-01 -8.81317019e-01 -2.94683248e-01 2.99730778e-01 -6.14884734e-01 -2.24319533e-01 1.16774774e+00 -5.80098867e-01 -8.32498729e-01 2.60475516e-01 4.08377089e-02 -7.48446345e-01 -1.89517438e-01 1.47014618e-01 -1.13858354e+00 -1.38294511e-02 6.41549051e-01 4.27876599e-02 2.69272029e-01 -6.64341390e-01 2.44754091e-01 1.20630395e+00 4.88118857e-01 -2.88353384e-01 1.02091718e+00 7.84942150e-01 -1.14594549e-01 -3.96465391e-01 -3.72782052e-01 -5.54996014e-01 -2.31053531e-01 2.92537492e-02 1.08788919e+00 -7.11711168e-01 -1.23755026e+00 6.07993960e-01 -1.00713718e+00 4.90038931e-01 -1.55331776e-01 4.68019962e-01 -1.83397025e-01 6.60036504e-01 -4.40804929e-01 -7.80543387e-01 -7.96570718e-01 -9.66100812e-01 3.87286127e-01 1.65456295e-01 2.05356792e-01 -6.76639199e-01 -4.22740094e-02 3.09856802e-01 9.67015699e-02 4.96222466e-01 1.04182577e+00 -1.21752954e+00 -4.63286966e-01 -3.19833755e-01 2.58013517e-01 3.01600128e-01 8.50298822e-01 -3.34166348e-01 -8.29452932e-01 -5.16801775e-01 9.75706220e-01 -6.80608898e-02 2.53377050e-01 -2.13791490e-01 1.38107026e+00 -7.01473296e-01 -6.53011560e-01 8.33076596e-01 1.31648076e+00 5.56022763e-01 4.56832826e-01 3.02759886e-01 9.49982107e-01 6.01411462e-01 8.46464694e-01 6.86122954e-01 1.19935468e-01 6.80444613e-02 7.39798903e-01 -5.11130095e-02 3.89765501e-01 -2.05876097e-01 1.56956881e-01 7.69588888e-01 6.02696598e-01 -2.40324393e-01 -7.26706982e-01 2.05338299e-01 -2.01173544e+00 -9.72753763e-01 -1.42419741e-01 2.39923167e+00 5.33480167e-01 -2.21113577e-01 1.17043160e-01 5.29029071e-01 9.70870793e-01 -4.90607530e-01 -9.13076818e-01 -1.04087582e-02 -4.94555011e-02 -7.67406896e-02 6.37611508e-01 -2.36905754e-01 -1.23500597e+00 5.32919466e-01 4.85676622e+00 7.26722836e-01 -1.27535272e+00 3.93410325e-01 5.29174626e-01 6.77110329e-02 -2.43774340e-01 -2.91970581e-01 -5.21085024e-01 8.38891685e-01 7.60124326e-01 -1.24194257e-01 2.69466519e-01 9.32599962e-01 8.40662867e-02 3.08440536e-01 -7.96341777e-01 1.32859290e+00 3.31081152e-02 -1.16976273e+00 2.31844053e-01 2.20682696e-01 5.67653716e-01 -2.37510234e-01 1.22061431e-01 -2.40925759e-01 2.88402319e-01 -7.04994917e-01 4.14121211e-01 4.09606725e-01 7.46766448e-01 -1.20968068e+00 7.15529442e-01 6.11059248e-01 -9.19194520e-01 -8.36460471e-01 -4.30426896e-01 3.22376907e-01 -2.25337759e-01 4.47167426e-01 -4.65493828e-01 9.96695995e-01 1.03341115e+00 5.53498507e-01 -4.18189198e-01 9.96958733e-01 2.69226521e-01 3.73910964e-01 -3.61186743e-01 4.60038669e-02 -4.22285885e-01 -2.37561285e-01 3.96257281e-01 5.13104618e-01 2.29934737e-01 3.47826034e-01 6.04998887e-01 3.49134266e-01 -6.37657493e-02 3.58939797e-01 -8.87112975e-01 1.60894722e-01 7.43671238e-01 1.34948516e+00 -6.11840308e-01 -2.00893298e-01 -1.49913266e-01 9.38529074e-01 -1.59729943e-01 6.77550808e-02 -7.69893408e-01 -3.41904134e-01 7.31182992e-01 1.96947623e-02 -3.77321914e-02 8.94915387e-02 -3.47453862e-01 -1.17066085e+00 -2.01152727e-01 -1.29329157e+00 9.18422282e-01 -6.23860434e-02 -1.64038098e+00 6.57878339e-01 -2.03530863e-01 -1.58641553e+00 3.14162403e-01 -3.32141072e-01 -4.49333787e-01 5.24079978e-01 -8.70916963e-01 -1.23942161e+00 -3.31224173e-01 1.54731369e+00 -1.59281701e-01 -4.71502960e-01 1.02527153e+00 4.74572629e-01 -6.59577668e-01 5.68560839e-01 2.13240594e-01 5.73593490e-02 5.99622965e-01 -6.08816862e-01 -4.03315127e-01 9.38293040e-01 -8.51030499e-02 9.61295187e-01 3.39185059e-01 -1.18838227e+00 -1.89014959e+00 -1.33287525e+00 3.09630901e-01 -5.23979306e-01 4.41318750e-01 -5.40079415e-01 -7.85650849e-01 2.79135227e-01 2.20301803e-02 7.79235736e-02 9.86540139e-01 -2.69402593e-01 -4.11566257e-01 -6.54551029e-01 -1.86292577e+00 3.53476524e-01 7.98025727e-01 -2.90490508e-01 -3.59035194e-01 6.59001946e-01 1.09476423e+00 1.39299512e-01 -9.66051996e-01 4.41816986e-01 5.31703606e-02 -7.01962352e-01 6.72045946e-01 -7.52876759e-01 -3.65195870e-01 -8.44459057e-01 -3.47399175e-01 -6.49966180e-01 -1.90223441e-01 -7.39016473e-01 -1.33132517e-01 1.32960403e+00 -2.05419436e-01 -8.29218626e-01 8.59267712e-01 5.95835268e-01 5.92386946e-02 -3.66942227e-01 -1.11543655e+00 -6.78511441e-01 -1.74380928e-01 -9.21053067e-02 1.11947370e+00 1.19908226e+00 3.22890133e-01 6.56811753e-03 -5.01271129e-01 5.34816384e-01 9.41134393e-01 4.70979847e-02 7.10152388e-01 -1.35713851e+00 -2.75820177e-02 2.35974059e-01 -4.99068230e-01 -1.47307634e-01 -2.04188511e-01 -9.00719762e-01 -3.54921043e-01 -1.11410618e+00 1.20708317e-01 -7.97123194e-01 -7.11656272e-01 6.38666809e-01 7.67317414e-02 -2.14276370e-03 5.75176962e-02 5.21663487e-01 -7.24417448e-01 3.33786547e-01 8.30062628e-01 -3.87120724e-01 -2.86420494e-01 1.27398357e-01 -6.99700117e-01 2.85182744e-01 1.01791966e+00 -5.99486887e-01 -6.44206464e-01 -2.21532077e-01 -2.95105666e-01 6.85274452e-02 1.28779128e-01 -1.20903945e+00 5.96760154e-01 -2.37241939e-01 3.62567872e-01 -5.94401896e-01 -1.76111728e-01 -1.77580416e+00 6.03988349e-01 1.19127667e+00 2.73555756e-01 -9.17096902e-03 -1.19863622e-01 6.60584390e-01 2.30137661e-01 1.27437666e-01 6.10872746e-01 -1.39829153e-02 -2.26401880e-01 7.75748193e-01 -2.89062798e-01 -4.03336376e-01 1.46264660e+00 -1.04546584e-01 -3.63084644e-01 2.32168168e-01 -6.06855989e-01 3.54976386e-01 5.85710585e-01 3.73726398e-01 6.52621746e-01 -1.16011548e+00 -2.74983644e-01 3.41889352e-01 8.15564692e-02 1.00792281e-01 4.29962665e-01 6.41534805e-01 -2.53840297e-01 -1.08192880e-02 -2.39649788e-01 -6.37384117e-01 -1.28055656e+00 1.56587434e+00 3.51541758e-01 6.76622838e-02 -6.19215310e-01 1.90208293e-02 1.50864152e-03 -2.80486256e-01 4.06490177e-01 1.70236573e-01 -2.17913941e-01 -1.51545018e-01 5.94806612e-01 4.33162689e-01 2.62923777e-01 -4.92872864e-01 -7.02331364e-01 2.69176185e-01 -6.22095615e-02 3.17141205e-01 1.10446668e+00 -1.86749950e-01 -4.75648403e-01 -4.96933013e-02 1.13890100e+00 1.62251636e-01 -6.54036701e-01 9.58195329e-02 -7.36429691e-02 -3.49771857e-01 -3.31007957e-01 -7.75061071e-01 -1.29722273e+00 4.33549672e-01 1.01236951e+00 2.28472814e-01 1.39239717e+00 -2.57972419e-01 1.03025246e+00 3.15679163e-01 8.74130070e-01 -6.08980834e-01 1.52561486e-01 -2.37636715e-01 5.96356578e-02 -7.90590942e-01 -9.75928381e-02 -4.11708772e-01 -4.85815763e-01 6.97422683e-01 6.93718910e-01 -7.18905553e-02 9.52855110e-01 2.91193098e-01 2.29681805e-01 -2.66227037e-01 -2.08240166e-01 2.61431515e-01 -4.22215194e-01 9.88768756e-01 -5.22916138e-01 4.10558172e-02 -3.74378890e-01 1.14435339e+00 2.64937039e-02 -2.56298780e-01 1.83299363e-01 7.97803223e-01 -3.05750459e-01 -1.44757199e+00 -6.71461999e-01 4.11037385e-01 -6.50418699e-01 4.01222974e-01 -1.24183089e-01 2.45748177e-01 7.17530847e-01 1.46623528e+00 -3.06427211e-01 -9.12710786e-01 2.42326930e-01 -4.73774485e-02 -2.51330864e-02 -3.79745185e-01 -9.35359120e-01 1.36789083e-01 -5.70443988e-01 -4.54989135e-01 -1.97675526e-01 -4.23096806e-01 -1.69121301e+00 -2.85068274e-01 -2.88795710e-01 6.56044126e-01 7.17004120e-01 6.86376870e-01 4.61185545e-01 2.26052940e-01 1.39989161e+00 4.79606092e-02 -8.73327851e-01 -2.44841337e-01 -6.71077609e-01 8.09233248e-01 1.31687611e-01 -3.82874787e-01 -3.65660101e-01 -2.07590953e-01]
[5.81882905960083, 6.533346176147461]
8ecdc28d-c6b8-4d09-8859-f27622436084
covid-19-epidemiology-as-emergent-behavior-on
2205.0215
null
https://arxiv.org/abs/2205.02150v1
https://arxiv.org/pdf/2205.02150v1.pdf
COVID-19 epidemiology as emergent behavior on a dynamic transmission forest
In this paper we create a compartmental, stochastic process model of SARS-CoV-2 transmission, where the process's mean and variance have distinct dynamics. The model is fit to time series data from Washington from January 2020 to March 2021 using a deterministic, biologically-motivated signal processing approach, and we show that the model's hidden states, like population prevalence, agree with survey and other estimates. Then, in the paper's second half, we demonstrate that the same model can be reframed as a branching process with a dynamic degree distribution. This perspective allows us to generate approximate transmission trees and estimate some higher order statistics, like the clustering of cases as outbreaks, which we find to be consistent with related observations from contact tracing and phylogenetics.
['Mike Famulare', 'Niket Thakkar']
2022-05-04
null
null
null
null
['epidemiology']
['medical']
[ 3.99237454e-01 -6.99861646e-02 -1.50452703e-01 -1.21274605e-01 -7.05527738e-02 -6.24883235e-01 9.29175138e-01 2.06148788e-01 -1.99582547e-01 9.60639358e-01 3.24377894e-01 -5.47073781e-01 -4.14156049e-01 -7.56712794e-01 -2.79955953e-01 -9.70779240e-01 -8.99065852e-01 9.11907375e-01 2.66245633e-01 -1.15469478e-01 -2.84429610e-01 3.70301574e-01 -7.72253513e-01 -2.50752956e-01 6.27467632e-01 2.19012231e-01 -3.89391966e-02 1.03477907e+00 4.89063933e-02 3.01518977e-01 -7.02836156e-01 -3.33430052e-01 1.00677185e-01 -5.10235369e-01 -3.84122193e-01 -1.90309331e-01 -5.15019298e-01 -3.16322416e-01 -3.39323729e-01 7.33576536e-01 2.06353385e-02 -5.00405669e-01 1.14529324e+00 -1.31364143e+00 -3.63881320e-01 4.96323913e-01 -7.71506310e-01 4.40333545e-01 1.51045099e-01 3.87496024e-01 4.80527133e-01 -1.32502556e-01 9.00007963e-01 1.51336026e+00 1.22950244e+00 3.70404124e-01 -1.87105381e+00 -3.27451020e-01 2.09472422e-03 -5.11131525e-01 -1.40339684e+00 1.61771290e-03 3.33454460e-01 -7.17554748e-01 7.74673522e-01 2.39192009e-01 1.26823699e+00 1.34909308e+00 9.45770442e-01 4.96661931e-01 1.07826293e+00 1.32376984e-01 4.68136221e-01 -4.51726466e-01 2.65571237e-01 5.09615958e-01 7.77758002e-01 2.20176384e-01 7.18234405e-02 -1.02228594e+00 8.45922589e-01 4.52909529e-01 -1.88636452e-01 -7.67920688e-02 -1.12121403e+00 9.52660263e-01 -1.87437207e-01 2.67442077e-01 -8.68625343e-01 5.09331644e-01 2.72576436e-02 9.60141793e-02 7.61708021e-01 -2.49245480e-01 -4.91055012e-01 3.29213403e-02 -1.10162222e+00 2.50107169e-01 1.01162744e+00 6.28264964e-01 3.84328663e-01 -1.19612366e-02 -1.35035887e-01 1.64717212e-01 5.79031527e-01 1.15503216e+00 -1.80999756e-01 -7.93943763e-01 -2.54697874e-02 -1.30467340e-02 3.12011331e-01 -7.04637587e-01 -5.90683281e-01 -2.53657490e-01 -1.29064643e+00 -3.55006188e-01 5.41635096e-01 -6.14258289e-01 -9.09369051e-01 2.02693057e+00 2.20799863e-01 4.50243741e-01 -5.62848076e-02 6.49757460e-02 -8.63386020e-02 1.16278481e+00 4.54915851e-01 -8.98462653e-01 1.40046120e+00 -1.65233910e-01 -9.02424335e-01 3.57642978e-01 4.78998125e-02 -2.36418262e-01 1.38519898e-01 2.76235342e-01 -9.46524382e-01 3.63758393e-02 -4.62477744e-01 9.58638728e-01 -4.86319326e-02 -4.99811172e-01 8.74065816e-01 6.20953083e-01 -1.38805771e+00 6.17205441e-01 -1.32112491e+00 -1.04393303e+00 2.16864958e-01 -9.60903987e-03 2.72983760e-01 3.03920448e-01 -1.20820642e+00 4.69208360e-01 -8.57366771e-02 1.95874244e-01 -1.26939094e+00 -6.60181701e-01 -1.95821181e-01 8.50541610e-03 -1.05110973e-01 -1.21023250e+00 8.68465900e-01 -4.74012822e-01 -9.53968823e-01 5.05375206e-01 -5.29118538e-01 -7.50003040e-01 4.12946790e-01 3.69630992e-01 -5.39708316e-01 3.26071143e-01 -1.36836633e-01 1.35920763e-01 7.81355262e-01 -1.27212393e+00 -3.40911031e-01 -5.07635176e-01 -6.61773980e-01 -3.93684864e-01 2.59835184e-01 3.16899776e-01 -1.21438806e-03 -7.34378755e-01 -1.69571385e-01 -1.01018631e+00 -8.22750747e-01 -1.87935218e-01 -3.61960471e-01 -3.34179141e-02 3.19151998e-01 -8.43807876e-01 1.25564170e+00 -1.71025789e+00 1.04897961e-01 4.06747639e-01 5.28837860e-01 -1.51818737e-01 -7.66747370e-02 1.16682875e+00 1.74725652e-01 4.26650673e-01 -9.18167531e-01 -2.35011682e-01 -4.18117851e-01 3.75846654e-01 -3.05328041e-01 8.68076742e-01 3.61784488e-01 6.81351304e-01 -9.91396487e-01 -2.35913053e-01 -1.89230591e-01 4.74312901e-01 -3.48633945e-01 3.13299010e-03 -9.15068612e-02 5.02355218e-01 -4.39173877e-01 3.58003080e-01 7.90069401e-01 -3.63032758e-01 5.24591446e-01 4.84102190e-01 -2.52286166e-01 -3.13465297e-01 -6.03441000e-01 7.60290980e-01 -3.18836272e-02 5.19933343e-01 4.02426600e-01 -7.57648468e-01 6.27414286e-01 3.86684805e-01 7.44830966e-01 3.51013273e-01 5.95730403e-03 -1.95648208e-01 3.84534180e-01 -1.82315618e-01 6.55820817e-02 -3.54158700e-01 -4.47331890e-02 7.66787887e-01 -7.57617876e-02 -1.64508283e-01 3.18233334e-02 2.96762675e-01 1.48229730e+00 -2.58869618e-01 3.41930330e-01 -5.84498584e-01 5.02771214e-02 3.82285595e-01 5.34259737e-01 9.26222205e-01 -3.45467031e-01 -1.07029058e-01 6.89280152e-01 -3.25172395e-01 -1.25826061e+00 -1.78820670e+00 -2.90971041e-01 2.92506784e-01 -3.04287702e-01 -1.57078445e-01 -8.11298728e-01 1.56301200e-01 1.37348026e-01 4.16976869e-01 -9.13644671e-01 -1.78617924e-01 -5.07664621e-01 -1.51373649e+00 7.34269977e-01 -1.93239171e-02 -6.58969628e-03 -7.03866839e-01 -5.81936300e-01 5.19848764e-01 -8.27208683e-02 -6.90191209e-01 -1.90278247e-01 1.21972166e-01 -1.17611575e+00 -1.04949927e+00 -1.04254639e+00 -3.20736319e-01 5.98732471e-01 5.40725254e-02 8.34631085e-01 6.20287023e-02 -4.12408739e-01 5.82429945e-01 1.92452133e-01 -4.53052521e-01 -8.40791285e-01 -2.85977572e-01 4.64179695e-01 -1.12535357e-01 3.14296871e-01 -7.28087425e-01 -8.28932166e-01 -1.61875132e-03 -1.02489746e+00 -2.75256336e-01 2.72352844e-01 4.94652182e-01 3.95397693e-01 1.06462188e-01 4.86589491e-01 -5.25742233e-01 8.12536657e-01 -9.38499272e-01 -6.43803239e-01 3.88365507e-01 -4.46800917e-01 -1.87645495e-01 4.16016996e-01 -5.88795781e-01 -1.00544822e+00 -4.72865589e-02 1.62862971e-01 -4.96205688e-02 -1.86384991e-01 5.08423626e-01 5.03412962e-01 5.51417828e-01 2.68119704e-02 6.11851275e-01 2.93811798e-01 -4.46899325e-01 2.51837373e-01 5.70372760e-01 3.13538611e-01 -2.07596406e-01 9.21461344e-01 1.03409541e+00 2.74033934e-01 -1.32790220e+00 -1.59717456e-01 -3.32966596e-01 -5.04421294e-01 -2.41590723e-01 1.00693178e+00 -9.70999897e-01 -9.71537888e-01 8.00726056e-01 -1.37439191e+00 -5.83572805e-01 -2.67599314e-01 6.80204928e-01 -7.36479342e-01 2.18766376e-01 -1.11450315e+00 -1.61891484e+00 -5.05163334e-02 -5.02866626e-01 1.01657784e+00 -1.28364891e-01 -2.60743946e-01 -1.41331995e+00 9.50188160e-01 -3.08516830e-01 6.18184447e-01 4.83474582e-01 1.13998997e+00 -5.68020761e-01 -3.14392626e-01 1.54502451e-01 6.66695535e-02 -2.44567424e-01 2.91664958e-01 7.38909304e-01 -4.33951706e-01 -3.51424038e-01 3.20732653e-01 6.89937651e-01 7.98318684e-01 1.11815488e+00 3.60033661e-01 -4.51759338e-01 -8.42110455e-01 4.32108998e-01 1.43111062e+00 3.79056990e-01 3.92614067e-01 -3.01751763e-01 2.37848293e-02 1.00448358e+00 1.08311601e-01 6.08114481e-01 3.08691561e-01 2.36715704e-01 -4.81237192e-04 -1.54386103e-01 2.41919354e-01 -5.33654690e-01 3.97472113e-01 1.06954932e+00 -2.72466838e-01 -3.41379493e-01 -1.20931602e+00 7.21902490e-01 -1.67053580e+00 -1.61597776e+00 -6.36489272e-01 2.11742401e+00 8.11984241e-01 -1.59292862e-01 7.35746861e-01 -5.23720443e-01 1.06596518e+00 2.63120849e-02 -4.41097260e-01 -2.27701068e-01 -3.68648529e-01 7.05165043e-02 8.38678420e-01 7.14140356e-01 -6.81289196e-01 7.23547101e-01 9.00705147e+00 5.30491352e-01 -6.18238986e-01 2.04021558e-01 8.06741834e-01 8.18753242e-02 -5.92752695e-01 1.45565405e-01 -4.89024520e-01 6.01853371e-01 1.66869807e+00 -4.97000754e-01 4.08517867e-01 -9.41557884e-02 8.61992180e-01 -2.25808173e-01 -7.08923697e-01 4.03248519e-01 -4.86603558e-01 -1.05493140e+00 -1.13219894e-01 6.07017994e-01 5.94553053e-01 1.91241875e-01 -2.17802465e-01 -8.83585140e-02 1.13775611e+00 -7.90547252e-01 2.93623716e-01 8.83736074e-01 4.59506661e-01 -6.35740101e-01 4.40578252e-01 5.42446375e-01 -1.12694514e+00 1.20368674e-01 -1.67047545e-01 -3.95315997e-02 1.01598513e+00 7.67632604e-01 -7.05201268e-01 2.68840939e-01 4.87170637e-01 5.04249692e-01 -4.12854031e-02 9.46879089e-01 -4.44570631e-02 1.27715981e+00 -7.32544720e-01 -1.52187243e-01 8.37871581e-02 -6.69901133e-01 9.29066598e-01 1.15712726e+00 4.23198849e-01 1.55969381e-01 -1.14793010e-01 9.02682841e-01 6.46709442e-01 -3.37320387e-01 -8.56711924e-01 -2.71880090e-01 4.11351025e-01 7.89444387e-01 -9.34392393e-01 -4.01175171e-01 8.14985577e-03 8.76705825e-01 -4.59589899e-01 8.02810848e-01 -8.97280574e-01 4.85233759e-04 8.82953525e-01 2.31115982e-01 3.45538527e-01 -6.00249648e-01 3.18298668e-01 -1.03979993e+00 -5.85821807e-01 -2.42893606e-01 1.42362431e-01 -5.43195605e-01 -1.58363152e+00 4.60541517e-01 5.31185985e-01 -7.28781998e-01 -5.86464584e-01 -2.18053967e-01 -8.01831543e-01 8.85040104e-01 -9.68659639e-01 -8.16035509e-01 5.61528862e-01 5.30267596e-01 1.60810351e-01 1.85532212e-01 5.79329908e-01 -7.87208974e-02 -5.78372836e-01 -1.37497187e-01 6.23007655e-01 -2.05740467e-01 -8.88407528e-02 -9.16671336e-01 8.90122950e-01 5.83153069e-01 -2.37644643e-01 9.88642693e-01 1.08377814e+00 -1.33142495e+00 -1.33879328e+00 -9.66841877e-01 7.43246794e-01 -3.55123460e-01 1.29399836e+00 -5.17952323e-01 -6.43638968e-01 6.49489522e-01 2.92195708e-01 -5.74755847e-01 7.52633810e-01 -2.73245782e-01 9.06863809e-02 4.20243382e-01 -1.36368585e+00 7.50219584e-01 1.18047142e+00 -2.20790967e-01 -5.81152201e-01 4.50938284e-01 8.18863213e-01 7.43116379e-01 -9.21762109e-01 9.01861787e-02 5.72457731e-01 -3.50016087e-01 8.18628371e-01 -1.04454279e+00 -1.14389345e-01 -2.52492309e-01 -1.01457879e-01 -1.28551650e+00 -4.58227009e-01 -1.04187357e+00 -2.63064206e-02 1.02704430e+00 3.76598060e-01 -1.08806729e+00 3.87624800e-01 -1.25275418e-01 5.86321712e-01 -3.87124062e-01 -1.26499474e+00 -1.20198846e+00 2.58229733e-01 -3.79054435e-02 5.37080228e-01 8.43455374e-01 -4.18014675e-01 2.47859314e-01 -4.94137645e-01 2.93635547e-01 1.11111081e+00 -1.24394245e-01 4.45257485e-01 -1.75193274e+00 -4.06127423e-01 -1.64168239e-01 -1.02915131e-01 -9.25361037e-01 -2.39480555e-01 -2.83409268e-01 9.26111490e-02 -1.40274024e+00 6.04717910e-01 -3.02737206e-01 1.71895087e-01 -1.86703220e-01 1.71027213e-01 -1.97993934e-01 3.47852819e-02 6.16210282e-01 -1.08483426e-01 5.04031003e-01 7.77548790e-01 5.17300218e-02 -2.21716955e-01 3.08430612e-01 -2.11118594e-01 8.38066578e-01 8.62424195e-01 -7.82711744e-01 -2.38943204e-01 1.01565294e-01 2.38360599e-01 5.93456089e-01 4.88454580e-01 -5.19392729e-01 -7.03537911e-02 -5.71965396e-01 1.18159108e-01 -8.36285889e-01 3.58637691e-01 -9.08534348e-01 9.29763973e-01 1.42677200e+00 -9.42397341e-02 5.60779691e-01 8.51702318e-02 1.20392478e+00 3.45420450e-01 9.35688987e-02 2.76720405e-01 -5.82428090e-02 3.10070395e-01 4.75649714e-01 -1.39090323e+00 -2.06640199e-01 1.05747712e+00 1.12597710e-02 -5.05553782e-01 -8.00716221e-01 -9.05449510e-01 3.04081708e-01 6.22837961e-01 -1.27597168e-01 3.16899866e-01 -7.82057703e-01 -1.02830887e+00 -2.82076955e-01 -3.84508729e-01 -7.48601079e-01 3.05265754e-01 1.16690469e+00 -5.61159551e-01 4.20522541e-01 4.88094538e-02 -8.76146734e-01 -9.34702575e-01 9.04388249e-01 2.10316285e-01 -6.04658365e-01 -2.46228248e-01 9.32913721e-02 1.17440104e-01 -2.71084815e-01 -3.06444108e-01 -2.29540557e-01 1.35310069e-01 8.17857757e-02 5.68324983e-01 3.87719423e-01 -9.56098914e-01 -5.86248696e-01 -4.80763763e-01 4.92011338e-01 4.02988821e-01 -5.43348730e-01 1.52876472e+00 -5.31531692e-01 -3.83730948e-01 8.72171462e-01 7.23133981e-01 -2.84876954e-03 -1.25185430e+00 1.39272660e-01 1.95547715e-01 8.65095779e-02 -5.23111224e-01 -4.11436766e-01 -6.52902901e-01 6.99265122e-01 6.45537078e-01 7.65621126e-01 8.33427846e-01 3.01613927e-01 5.75867176e-01 8.91854316e-02 5.62087834e-01 -4.12784249e-01 -7.17584312e-01 1.86716691e-01 5.85948110e-01 -5.53603351e-01 1.58138003e-03 -5.54567873e-01 -1.67383775e-01 6.04527712e-01 -3.78092051e-01 -2.03649148e-01 1.12438202e+00 4.53192294e-01 -4.91215825e-01 -2.98700869e-01 -1.20597458e+00 -1.52326465e-01 -2.99788535e-01 1.15530515e+00 1.45092741e-01 4.52450573e-01 -6.59635663e-01 2.66282558e-01 4.79429513e-02 2.74260174e-02 6.92602992e-01 6.69873655e-01 -4.79800016e-01 -8.70490432e-01 -4.26164389e-01 4.61458653e-01 -4.72216129e-01 -3.87004018e-01 -4.31919575e-01 8.72151315e-01 -2.42215306e-01 8.95431042e-01 5.14498293e-01 5.89159578e-02 -1.53801411e-01 5.78936236e-03 3.43601376e-01 -3.49716872e-01 -6.03499949e-01 3.67333114e-01 -4.43920307e-02 -1.37654722e-01 -5.84378541e-01 -1.01738977e+00 -7.82872915e-01 -7.89258301e-01 1.03338007e-02 1.47935629e-01 5.24304748e-01 5.80031991e-01 1.79405242e-01 2.49845669e-01 7.35454440e-01 -3.94747943e-01 -6.55021071e-01 -9.27717268e-01 -7.94059038e-01 -4.99723628e-02 4.27715063e-01 -1.99186772e-01 -6.51662171e-01 1.09443344e-01]
[6.00039529800415, 4.329865455627441]
205b3029-1e1b-4f25-bd3f-010e9b55b878
write-like-you-synthesizing-your-cursive
null
null
https://onlinelibrary.wiley.com/doi/10.1111/cgf.142621
https://onlinelibrary.wiley.com/doi/10.1111/cgf.142621
Write like you: Synthesizing your cursive online Chinese handwriting via metric-based meta learning
In this paper, we propose a novel Sequence-to-Sequence model based on metric-based meta learning for the arbitrary style transfer of online Chinese handwritings. Unlike most existing methods that treat Chinese handwritings as images and are unable to reflect the human writing process, the proposed model directly handles sequential online Chinese handwritings. Generally, our model consists of three sub-models: a content encoder, a style encoder and a decoder, which are all Recurrent Neural Networks. In order to adaptively obtain the style information, we introduce an attention-based adaptive style block which has been experimentally proven to bring considerable improvement to our model. In addition, to disentangle the latent style information from characters written by any writers effectively, we adopt metric-based meta learning and pre-train the style encoder using a carefully-designed discriminative loss function. Then, our entire model is trained in an end-to-end manner and the decoder adaptively receives the style information from the style encoder and the content information from the content encoder to synthesize the target output. Finally, by feeding the trained model with a content character and several characters written by a given user, our model can write that Chinese character in the user's handwriting style by drawing strokes one by one like humans. That is to say, as long as you write several Chinese character samples, our model can imitate your handwriting style when writing. In addition, after fine-tuning the model with a few samples, it can generate more realistic handwritings that are difficult to be distinguished from the real ones. Both qualitative and quantitative experiments demonstrate the effectiveness and superiority of our method.
['Zhouhui Lian', 'Shusen Tang']
2021-06-04
null
null
null
computer-graphics-forum-2021-6
['style-transfer']
['computer-vision']
[ 2.97615975e-01 -1.64846450e-01 -7.28074387e-02 -4.49249774e-01 -4.16290373e-01 -7.60717273e-01 5.64013958e-01 -6.26708627e-01 -3.57809365e-01 5.99912047e-01 2.15164930e-01 -1.36478484e-01 4.42753315e-01 -7.03577101e-01 -6.63726687e-01 -6.53657854e-01 8.24077427e-01 5.81606686e-01 -8.57726112e-02 -2.28222311e-01 4.38979536e-01 4.31391269e-01 -9.42188799e-01 4.07723904e-01 1.11289418e+00 7.08437383e-01 7.27444530e-01 8.62485945e-01 -2.50453055e-01 1.00281668e+00 -8.76594663e-01 -5.89536607e-01 1.43805251e-01 -9.52250361e-01 -6.48617327e-01 6.43097997e-01 3.09567362e-01 -7.60498703e-01 -6.95635319e-01 1.07776070e+00 5.42906880e-01 -1.14812993e-03 7.35071599e-01 -6.71355069e-01 -1.23098958e+00 6.04387343e-01 -3.82381439e-01 -2.99749345e-01 2.34385177e-01 5.38217366e-01 7.45607495e-01 -8.68094683e-01 6.74895585e-01 1.26424539e+00 1.41914517e-01 9.73456383e-01 -1.03242004e+00 -6.63554728e-01 4.09592599e-01 2.06720997e-02 -9.30967927e-01 -1.67778671e-01 1.02789569e+00 -1.59258187e-01 4.14338589e-01 1.26336562e-02 6.89578831e-01 1.51941514e+00 1.49242952e-01 1.35091662e+00 1.08993661e+00 -2.49304384e-01 1.37201294e-01 1.84487388e-01 -2.87942976e-01 5.51056445e-01 -4.09045517e-01 8.02909881e-02 -2.60818958e-01 2.70098686e-01 1.06361532e+00 1.19343452e-01 -1.92672029e-01 -5.84600270e-02 -1.28339112e+00 5.44164836e-01 3.80413204e-01 1.56908467e-01 -1.86883301e-01 1.23092122e-01 4.51604754e-01 5.54171264e-01 4.12322819e-01 3.91880989e-01 -1.48514099e-02 -3.04952502e-01 -1.05869126e+00 1.84455186e-01 7.72850871e-01 1.40471971e+00 4.78626877e-01 1.28827438e-01 -4.36051995e-01 9.66827393e-01 2.54368149e-02 6.77539229e-01 6.38044298e-01 -6.55389130e-01 7.59622097e-01 4.25959408e-01 3.33887458e-01 -6.27927363e-01 2.73181021e-01 -2.43477568e-01 -1.03761542e+00 4.28174958e-02 3.36700290e-01 -3.04625064e-01 -9.42948580e-01 1.50871778e+00 -2.32350037e-01 -5.83603904e-02 -1.03480920e-01 1.12220478e+00 2.15488642e-01 8.50588322e-01 -2.39404291e-01 -1.08197451e-01 1.05805981e+00 -1.39148796e+00 -8.36347103e-01 -2.50399739e-01 2.38138840e-01 -8.79870057e-01 1.65168774e+00 3.89838189e-01 -1.34822404e+00 -7.71686554e-01 -1.16146374e+00 -2.14673564e-01 5.11202738e-02 5.10943234e-01 8.09015632e-02 2.87452608e-01 -7.08836496e-01 6.20108366e-01 -7.78741062e-01 -2.03919157e-01 4.42600697e-01 -5.08123338e-02 -3.14872409e-03 -1.00836031e-01 -9.50771451e-01 7.76746154e-01 2.41756409e-01 7.44544491e-02 -1.03448629e+00 -3.98075998e-01 -6.28705680e-01 -1.19914720e-02 2.54962444e-01 -6.61414862e-01 1.34749484e+00 -1.44731104e+00 -2.14074636e+00 7.03208745e-01 -3.92735660e-01 -9.99936387e-02 1.15402949e+00 -1.61101416e-01 -3.94479424e-01 8.14580545e-02 -4.88942936e-02 4.58297163e-01 1.16032004e+00 -1.29610109e+00 -6.24204278e-01 -9.17396694e-02 8.48233476e-02 3.93236369e-01 -4.78118062e-01 1.14455735e-02 -8.33274305e-01 -1.10694289e+00 -2.15060011e-01 -8.22495699e-01 8.53910111e-03 1.89267680e-01 -5.30758977e-01 1.43694505e-02 9.01218116e-01 -6.85128331e-01 1.15497959e+00 -2.31231022e+00 4.83728945e-01 -3.93145829e-02 1.97076704e-02 3.21630865e-01 -4.78656113e-01 4.06346172e-01 4.45359766e-01 -3.49346810e-04 -3.19104731e-01 -5.20697296e-01 -7.29686394e-02 2.38355026e-01 -5.25791228e-01 1.06732070e-01 3.50643665e-01 1.04383504e+00 -1.23910725e+00 -2.03916654e-01 -2.65478566e-02 2.16878578e-01 -3.92714590e-01 6.91147447e-01 -4.87442702e-01 5.24843872e-01 -5.35708547e-01 4.78960395e-01 6.08776808e-01 -3.14349651e-01 3.28442425e-01 -3.86866508e-03 2.26005092e-02 1.82729647e-01 -7.52878189e-01 1.70580208e+00 -7.58608103e-01 6.06047094e-01 -1.62278742e-01 -5.67005157e-01 1.06199861e+00 2.42161974e-01 -2.53104001e-01 -7.42278039e-01 2.67240498e-02 2.69149631e-01 -3.80682908e-02 -4.81743872e-01 5.37684679e-01 -1.43233046e-01 3.08838319e-02 9.46142972e-01 6.49022534e-02 -2.66730994e-01 2.02251554e-01 4.57866825e-02 7.93942750e-01 4.04324800e-01 -1.53292909e-01 -8.96602497e-02 6.12413824e-01 -2.88817078e-01 4.41950679e-01 8.02231312e-01 -5.22580743e-02 8.68136227e-01 5.54206312e-01 -3.90024632e-01 -1.61876738e+00 -9.98556614e-01 3.08920205e-01 1.05329370e+00 2.32047081e-01 -1.07091712e-02 -9.05751109e-01 -8.69746804e-01 -1.57230958e-01 7.41381586e-01 -6.39372587e-01 -2.38281623e-01 -8.06363583e-01 -2.12580711e-01 5.56191862e-01 8.02967072e-01 8.54057550e-01 -1.42513764e+00 -3.97419035e-01 4.18733090e-01 -8.89149532e-02 -8.93405259e-01 -1.32999694e+00 -1.97956800e-01 -7.77689159e-01 -7.38105834e-01 -1.37379265e+00 -1.06576478e+00 9.86833692e-01 1.41779900e-01 9.26676810e-01 6.55478835e-02 5.39899766e-02 -2.77959965e-02 -3.67604852e-01 -2.08395705e-01 -7.73667157e-01 2.74233162e-01 -1.57559991e-01 2.42341340e-01 2.18110725e-01 -3.96689504e-01 -6.15478218e-01 3.76004308e-01 -9.11381781e-01 4.37496126e-01 8.11561704e-01 1.13389301e+00 2.68850297e-01 -3.33922088e-01 5.10229051e-01 -1.08525109e+00 8.63463819e-01 -1.28926381e-01 -5.65262377e-01 5.43922961e-01 -3.73518497e-01 2.61759788e-01 1.20735478e+00 -8.26618195e-01 -1.27006567e+00 -4.18880135e-02 -2.89253704e-02 -6.61781073e-01 -2.48097647e-02 2.92193383e-01 -4.08419430e-01 2.92458028e-01 3.73748034e-01 8.84913504e-01 1.78532124e-01 -5.74411929e-01 4.87761617e-01 1.00203788e+00 6.87441468e-01 -7.68924534e-01 8.79102886e-01 2.90626377e-01 -5.79671085e-01 -5.53959489e-01 -8.14632177e-01 1.52948201e-01 -7.43954599e-01 -1.45290762e-01 7.25910842e-01 -7.42645979e-01 -6.18209958e-01 1.00387228e+00 -1.23908114e+00 -7.33102500e-01 -8.59432369e-02 5.51996641e-02 -7.38822222e-01 4.35958594e-01 -1.01231146e+00 -7.82902837e-01 -2.25576967e-01 -1.27347028e+00 9.41751540e-01 2.86986142e-01 -1.19623013e-01 -1.08534348e+00 -1.92604348e-01 4.39756140e-02 5.80147803e-01 -9.58351046e-02 1.01516902e+00 -3.34397912e-01 -5.28642595e-01 -7.55312815e-02 -3.04784983e-01 5.76195717e-01 5.87269366e-01 7.28230551e-02 -8.30201864e-01 -3.85913908e-01 -2.08352748e-02 -4.91166741e-01 7.13803768e-01 -2.01798454e-01 1.33376765e+00 -5.29608727e-01 1.43469155e-01 6.87052608e-01 1.29410505e+00 5.21420181e-01 7.19024479e-01 6.10834770e-02 9.56988275e-01 3.56890291e-01 5.15391469e-01 3.58855516e-01 2.68834025e-01 3.29861462e-01 -1.38049414e-02 8.04525428e-03 -1.70264602e-01 -7.47031391e-01 5.54181874e-01 9.10110176e-01 -1.10417835e-01 -3.06606114e-01 -4.92186517e-01 3.90372545e-01 -1.75573468e+00 -9.54550087e-01 5.29660344e-01 2.08380961e+00 1.34631741e+00 2.73024708e-01 1.72008678e-01 -1.64139330e-01 7.54372716e-01 2.98381716e-01 -9.20547187e-01 -5.69663465e-01 -1.34494454e-01 -4.00261767e-02 2.36430869e-01 4.60559636e-01 -7.76751518e-01 1.29697096e+00 5.78934908e+00 7.62387037e-01 -1.49979687e+00 -2.24831462e-01 8.32567692e-01 -2.06001103e-01 -5.12423873e-01 -1.06358729e-01 -6.42028451e-01 9.18476641e-01 5.11028945e-01 -2.66181558e-01 9.47001159e-01 7.20334113e-01 3.57989639e-01 3.71220708e-01 -1.37948155e+00 7.92491317e-01 1.43686101e-01 -1.28707159e+00 3.84894550e-01 -7.26900473e-02 9.52881575e-01 -4.28428501e-01 1.84208959e-01 2.15242147e-01 2.79654175e-01 -1.05939341e+00 1.14345837e+00 7.38631427e-01 1.10645914e+00 -7.17413604e-01 4.60877210e-01 5.81161141e-01 -8.43825459e-01 -1.01672187e-01 -5.12200713e-01 -3.00428793e-02 1.49403661e-01 3.29782069e-01 -3.12088490e-01 3.55376214e-01 6.27136743e-03 9.09775078e-01 -4.32644010e-01 7.08399773e-01 -5.31676471e-01 5.57130218e-01 1.95335418e-01 -3.84331226e-01 4.91828412e-01 -4.01617765e-01 2.54009306e-01 1.20035160e+00 2.37717971e-01 1.04462234e-02 9.14159045e-02 1.26470566e+00 -3.57063204e-01 -1.98549628e-01 -5.44270039e-01 -3.34609926e-01 5.81632555e-01 1.07595849e+00 -3.31241548e-01 -6.40997589e-01 -4.07199651e-01 1.75061667e+00 4.30552632e-01 6.04691088e-01 -6.65655315e-01 -7.30341315e-01 3.54215711e-01 -1.91419005e-01 2.99306750e-01 -1.18164651e-01 -3.81524533e-01 -1.33289909e+00 1.90123856e-01 -1.16957319e+00 -3.17809016e-01 -9.17753100e-01 -1.50214326e+00 7.54335463e-01 -4.48803306e-01 -1.38805234e+00 -3.60361397e-01 -6.38411522e-01 -9.38665509e-01 1.24225342e+00 -1.17150712e+00 -1.27491951e+00 -2.55988300e-01 3.16869438e-01 9.46764410e-01 -4.23801035e-01 4.98940378e-01 8.79971683e-02 -5.31260967e-01 9.14451003e-01 2.34253168e-01 5.54791987e-01 8.10008466e-01 -1.41849518e+00 7.20799625e-01 7.94654429e-01 -7.18043596e-02 6.45714343e-01 4.44399655e-01 -7.52121091e-01 -1.33257127e+00 -1.14174736e+00 8.01446557e-01 -2.51636654e-01 5.73304534e-01 -6.20328486e-01 -9.90090132e-01 6.53243005e-01 3.90840948e-01 -2.06721261e-01 2.79426485e-01 -3.39240164e-01 -2.97775209e-01 3.88972946e-02 -9.15427208e-01 1.01450884e+00 1.12848294e+00 -8.07645738e-01 -6.90397084e-01 2.30447292e-01 4.31350052e-01 -5.59410512e-01 -5.46633661e-01 -1.20342463e-01 7.29505599e-01 -6.81228638e-01 3.78565878e-01 -7.63088226e-01 1.04637873e+00 -2.32599020e-01 1.40977621e-01 -1.53897393e+00 -4.36613053e-01 -6.63440287e-01 -1.41226530e-01 1.30151474e+00 3.12326193e-01 -2.97555298e-01 7.55276203e-01 4.60162222e-01 -6.21495396e-02 -8.37933600e-01 -3.05293769e-01 -9.08518851e-01 4.36113805e-01 3.98011431e-02 8.27686727e-01 7.02754855e-01 -1.28413677e-01 3.30712229e-01 -7.18540549e-01 -2.68727720e-01 4.96174067e-01 3.87241989e-01 7.38982439e-01 -5.98117411e-01 -6.39204741e-01 -7.07114220e-01 2.01734528e-01 -1.64370596e+00 2.13186547e-01 -7.21318305e-01 1.60510808e-01 -1.26635277e+00 3.78284335e-01 -3.34317476e-01 -1.24914132e-01 1.71811119e-01 -5.43024778e-01 -2.62733344e-02 5.01326323e-01 3.23034525e-01 -3.15957397e-01 7.53762364e-01 1.94903457e+00 -4.68403220e-01 -3.22564803e-02 1.72592968e-01 -6.51539683e-01 4.54193205e-01 6.68088019e-01 -1.26319170e-01 -4.80296433e-01 -7.82444060e-01 -3.37416753e-02 2.81309545e-01 1.94186240e-01 -4.70250249e-01 1.14811406e-01 -3.33712757e-01 5.46082258e-01 -3.79806966e-01 -5.26622077e-03 -5.47052920e-01 -2.96882540e-01 4.86483574e-01 -7.17298925e-01 1.04144357e-01 -2.04346538e-01 4.15659666e-01 -1.83760196e-01 -3.85127783e-01 8.51885438e-01 -3.27567905e-01 -4.48575020e-01 3.58247429e-01 -3.64558548e-01 -5.32116517e-02 6.28544748e-01 -2.87316237e-02 -1.45528451e-01 -5.73808014e-01 -5.82502067e-01 3.08434606e-01 7.78686166e-01 6.08090043e-01 6.48752093e-01 -1.59609389e+00 -7.52432048e-01 3.43909144e-01 6.30660653e-02 1.85740814e-02 1.06632583e-01 3.14946681e-01 -6.77535176e-01 7.66917765e-02 -3.07104707e-01 -2.78701186e-01 -8.02476466e-01 6.82157695e-01 3.31599057e-01 -2.10802436e-01 -6.46705210e-01 6.71634912e-01 2.28351086e-01 -3.14955831e-01 3.23147565e-01 -2.56153136e-01 2.39928648e-01 -2.03743741e-01 8.02801728e-01 2.17104957e-01 -3.68203104e-01 -2.45694116e-01 3.19042504e-02 4.66362685e-01 -5.18146753e-01 -3.13024729e-01 9.88446772e-01 -8.68179649e-02 3.34210731e-02 6.97652102e-01 1.20619869e+00 1.23564117e-02 -1.95071042e+00 -3.78638148e-01 -1.51428863e-01 -5.61813354e-01 -5.22183776e-01 -9.52893794e-01 -1.06542730e+00 1.17663312e+00 2.22156867e-01 -2.20937744e-01 1.17310882e+00 -3.20073247e-01 9.07381535e-01 4.65015560e-01 2.98592567e-01 -1.35875702e+00 4.93035108e-01 6.88708842e-01 1.05047965e+00 -1.13610387e+00 -4.44531262e-01 9.53225717e-02 -1.02082205e+00 1.31392562e+00 7.11587131e-01 -3.81149650e-01 1.85656935e-01 3.04470032e-01 4.10358131e-01 3.75148803e-01 -8.22586894e-01 2.85052717e-01 2.52304137e-01 4.42736953e-01 4.34767842e-01 4.67737280e-02 -7.69825876e-02 6.60309970e-01 -1.88286036e-01 2.72848308e-01 5.81079543e-01 8.07127953e-01 -2.58581132e-01 -1.32563329e+00 3.09328772e-02 3.24682325e-01 -1.37493595e-01 -6.13947883e-02 -5.44151425e-01 3.61813992e-01 -3.18848222e-01 5.81645370e-01 2.67787725e-01 -4.82104450e-01 2.27528006e-01 7.32517317e-02 5.95275521e-01 -6.10974073e-01 -6.11534476e-01 3.10864244e-02 -3.23088974e-01 -2.26071998e-01 2.60033701e-02 -3.65415633e-01 -9.18141246e-01 -3.58474463e-01 -2.80680656e-02 -1.25652596e-01 3.20662975e-01 1.07565308e+00 1.36388376e-01 5.40949762e-01 1.23228109e+00 -8.98774683e-01 -9.90536869e-01 -1.14581406e+00 -5.67174852e-01 5.98541379e-01 3.00606847e-01 -7.28298053e-02 -3.20807248e-02 2.35577852e-01]
[11.616415023803711, -0.2662298381328583]
19338519-6a7e-4be0-ae7e-7e46e308464b
3d-instance-segmentation-of-mvs-buildings
2112.09902
null
https://arxiv.org/abs/2112.09902v2
https://arxiv.org/pdf/2112.09902v2.pdf
3D Instance Segmentation of MVS Buildings
We present a novel 3D instance segmentation framework for Multi-View Stereo (MVS) buildings in urban scenes. Unlike existing works focusing on semantic segmentation of urban scenes, the emphasis of this work lies in detecting and segmenting 3D building instances even if they are attached and embedded in a large and imprecise 3D surface model. Multi-view RGB images are first enhanced to RGBH images by adding a heightmap and are segmented to obtain all roof instances using a fine-tuned 2D instance segmentation neural network. Instance masks from different multi-view images are then clustered into global masks. Our mask clustering accounts for spatial occlusion and overlapping, which can eliminate segmentation ambiguities among multi-view images. Based on these global masks, 3D roof instances are segmented out by mask back-projections and extended to the entire building instances through a Markov random field optimization. A new dataset that contains instance-level annotation for both 3D urban scenes (roofs and buildings) and drone images (roofs) is provided. To the best of our knowledge, it is the first outdoor dataset dedicated to 3D instance segmentation with much more annotations of attached 3D buildings than existing datasets. Quantitative evaluations and ablation studies have shown the effectiveness of all major steps and the advantages of our multi-view framework over the orthophoto-based method.
['Liangliang Nan', 'Ronghua Liang', 'Shufang Lu', 'Yanghui Xu', 'Jiazhou Chen']
2021-12-18
null
null
null
null
['3d-instance-segmentation-1']
['computer-vision']
[ 3.93948078e-01 3.17145824e-01 4.70733374e-01 -4.88465488e-01 -7.55722284e-01 -6.71756744e-01 4.08800334e-01 -1.43663418e-02 -1.61097180e-02 3.53267044e-01 -2.47122109e-01 3.97061296e-02 1.05532847e-01 -1.23585320e+00 -7.28819191e-01 -4.63738292e-01 1.66664049e-01 8.10351789e-01 8.40446055e-01 -3.49063754e-01 2.86817905e-02 6.46803260e-01 -2.07922554e+00 2.58010149e-01 8.81671011e-01 7.35517502e-01 5.36368191e-01 5.81398070e-01 -1.66242067e-02 4.47822688e-03 -2.95568258e-01 -2.72262171e-02 7.36023903e-01 1.75628349e-01 -7.70621300e-01 9.13757205e-01 9.54518616e-01 -4.00108635e-01 1.77881822e-01 7.98353493e-01 4.33718771e-01 4.40026745e-02 3.57856035e-01 -1.09430611e+00 4.82265130e-02 1.98893502e-01 -6.75068378e-01 -8.66922736e-02 4.14971739e-01 1.41642407e-01 9.06048894e-01 -1.01856744e+00 5.45748889e-01 9.66285527e-01 7.93462873e-01 6.09503873e-03 -1.25798476e+00 -2.42908612e-01 5.40056765e-01 -2.79077083e-01 -1.64867735e+00 -1.05190851e-01 9.50217485e-01 -6.97399318e-01 1.08612835e+00 7.24804580e-01 1.20790982e+00 3.42795253e-01 -3.55796427e-01 7.83586860e-01 1.31836975e+00 -2.41988182e-01 1.95827469e-01 1.44905433e-01 1.49043277e-01 7.55682051e-01 1.02014542e-01 -6.10703938e-02 -5.52004017e-02 1.38305187e-01 6.46007657e-01 2.07715437e-01 -2.82606900e-01 -6.19584501e-01 -1.17217469e+00 5.40291667e-01 6.65981710e-01 1.25088423e-01 -4.78882283e-01 -1.73871107e-02 -1.60421550e-01 -2.96212614e-01 6.96075201e-01 9.70565751e-02 -6.30125463e-01 5.09050608e-01 -1.44536996e+00 1.18844181e-01 3.13541889e-01 1.02825856e+00 1.48442113e+00 -1.25681728e-01 3.30197960e-01 7.74391592e-01 3.57695729e-01 7.33539045e-01 -2.87181050e-01 -1.19314957e+00 5.31443954e-01 1.18724382e+00 -1.68747768e-01 -6.91143870e-01 -6.88222408e-01 -3.36642027e-01 -4.02770191e-01 2.52395004e-01 8.88717100e-02 1.80612668e-01 -1.37861431e+00 9.41946447e-01 8.20398629e-01 1.51770571e-02 -2.20517769e-01 8.24652791e-01 1.15647852e+00 4.66972619e-01 -3.88345182e-01 1.61865398e-01 1.28924251e+00 -7.11702228e-01 -8.96506831e-02 -7.36434937e-01 3.82755697e-01 -6.81930006e-01 7.38404453e-01 1.23858996e-01 -1.04755878e+00 -4.36900824e-01 -8.58443737e-01 -8.26653466e-02 -6.31492019e-01 8.95792246e-03 5.82653046e-01 8.93713951e-01 -1.04429209e+00 1.21725477e-01 -6.61957800e-01 -4.19351041e-01 4.94468510e-01 3.20927560e-01 -3.20499510e-01 -1.86108395e-01 -8.57183814e-01 6.11850500e-01 5.51904857e-01 1.34221002e-01 -8.72626066e-01 -7.26130366e-01 -1.05498087e+00 -3.69728506e-01 7.03388393e-01 -8.85863721e-01 6.37397110e-01 -4.77384657e-01 -8.35092962e-01 1.40440261e+00 -4.45706733e-02 -2.30128855e-01 3.23076516e-01 -6.34839088e-02 -4.27066945e-02 -6.17876276e-02 6.01294756e-01 8.63430738e-01 6.21907890e-01 -1.93612194e+00 -1.00457847e+00 -7.77215481e-01 3.09344947e-01 4.71769273e-01 4.79187071e-01 -5.60774982e-01 -8.66663516e-01 -2.74789035e-01 9.29681778e-01 -1.04079413e+00 -5.19778788e-01 -5.81114233e-01 -7.89257228e-01 4.29636598e-01 1.05712700e+00 -8.56768489e-01 1.06163597e+00 -1.96406615e+00 1.86866418e-01 4.44468856e-01 1.39479175e-01 -1.77102074e-01 4.23596650e-01 1.59131899e-01 2.61317249e-02 2.34735027e-01 -8.92231226e-01 -4.15767252e-01 -1.11661784e-01 4.10690486e-01 2.41409335e-02 4.89330083e-01 -2.32737869e-01 7.16289461e-01 -5.67711294e-01 -6.72746718e-01 1.01385450e+00 5.57506025e-01 -5.46884894e-01 -6.54452443e-02 -3.91496658e-01 5.24990857e-01 -3.70716900e-01 1.06473410e+00 1.00232661e+00 4.22160290e-02 -8.51728320e-02 -2.98725784e-01 -3.80900979e-01 1.63628254e-02 -1.56658769e+00 1.70156705e+00 -5.87352514e-01 1.91491812e-01 1.63259417e-01 -5.87205648e-01 8.22236836e-01 9.35217291e-02 7.67773867e-01 -3.47500622e-01 -2.14323759e-01 1.65566936e-01 -8.30811143e-01 -2.56800234e-01 7.63042867e-01 7.43468106e-02 -3.13630611e-01 1.40934095e-01 -5.33180237e-02 -1.10070944e+00 1.89117551e-01 9.83362570e-02 5.19844532e-01 5.04019320e-01 1.61586538e-01 -7.53367320e-02 5.93726277e-01 2.45571077e-01 3.65060270e-01 4.80562210e-01 9.49383378e-02 1.19641685e+00 -1.48788139e-01 -6.55449450e-01 -9.23970401e-01 -1.23467815e+00 -3.26679587e-01 6.54142678e-01 3.15410763e-01 -4.86390918e-01 -9.30746436e-01 -6.45035326e-01 -5.08433506e-02 8.46256137e-01 -4.22904074e-01 6.32212579e-01 -5.48259735e-01 -1.01177478e+00 2.46967569e-01 2.22698182e-01 8.17873240e-01 -4.40093160e-01 -1.03386796e+00 1.33940861e-01 -5.03821194e-01 -1.53624511e+00 -6.91476166e-02 4.70027179e-01 -9.18969154e-01 -1.28310144e+00 -3.42811823e-01 -5.06107569e-01 7.85125375e-01 4.62781966e-01 1.42173338e+00 8.98072347e-02 -3.86406690e-01 8.86323452e-01 -7.63804093e-02 -2.12282538e-01 4.67176586e-02 9.51426849e-02 -4.36727494e-01 -1.94190554e-02 5.14490604e-02 -7.50418961e-01 -5.29825151e-01 5.11994839e-01 -7.11982191e-01 4.65766042e-01 2.90803701e-01 1.37287930e-01 1.20299590e+00 4.03390467e-01 -5.52115977e-01 -9.09789801e-01 -3.34088236e-01 -1.97683722e-01 -9.71160769e-01 3.59445848e-02 -1.66705772e-01 -5.03801823e-01 1.75489187e-02 5.72415471e-01 -9.98067200e-01 7.49868214e-01 -1.56783015e-01 -2.09743723e-01 -9.63826597e-01 3.75933982e-02 -4.97780591e-01 7.27733225e-02 4.39589977e-01 1.25241295e-01 -6.61051989e-01 -3.35968643e-01 5.49160957e-01 4.12973881e-01 3.18884224e-01 -3.42373818e-01 1.06977963e+00 1.02798736e+00 -2.81667951e-02 -1.26747644e+00 -1.05675507e+00 -7.92987049e-01 -1.45640302e+00 -6.65659487e-01 1.32570100e+00 -1.18776882e+00 7.44593441e-02 2.76231200e-01 -9.06502664e-01 -2.67156750e-01 -5.13108194e-01 2.15223894e-01 -6.57444537e-01 1.77314922e-01 -6.56393021e-02 -9.49670434e-01 1.17399283e-01 -1.38610971e+00 1.81015944e+00 -8.36370885e-02 2.00242531e-02 -8.24661911e-01 -8.14735442e-02 1.01242495e+00 -3.18282068e-01 7.07235932e-01 4.19680148e-01 2.33685896e-02 -1.20674014e+00 -2.74596387e-03 1.77888036e-01 2.23054156e-01 8.73279199e-02 2.14365914e-01 -1.29988229e+00 2.31236145e-01 -1.26894906e-01 1.69743136e-01 1.06129718e+00 8.12240362e-01 6.62226558e-01 1.65617600e-01 -5.98890662e-01 8.01073015e-01 1.74083316e+00 -2.46337503e-02 5.35740256e-01 5.34896851e-01 1.15228605e+00 6.58307314e-01 6.62510157e-01 3.81303251e-01 7.49466956e-01 7.35081315e-01 1.03994238e+00 -3.76043618e-01 2.36144885e-02 -1.41850263e-01 -3.03197798e-04 3.68649811e-01 -4.32912976e-01 -1.16049014e-01 -1.30166554e+00 7.40996301e-01 -1.53854501e+00 -8.56049955e-01 -7.97264516e-01 2.23387575e+00 1.00144066e-01 1.57727480e-01 1.46074980e-01 3.84155542e-01 7.88368642e-01 3.80352616e-01 -2.24621683e-01 5.71988337e-02 -3.27600092e-01 1.18635118e-01 8.39335561e-01 8.82646143e-01 -1.63519037e+00 1.02137792e+00 5.25077868e+00 5.89857221e-01 -5.41095316e-01 4.67690751e-02 6.53509974e-01 -8.58041868e-02 -3.13055575e-01 2.85327941e-01 -1.01714361e+00 1.03426844e-01 3.28543603e-01 7.63773263e-01 2.51150519e-01 9.96888578e-01 2.46132210e-01 -5.49184382e-01 -6.55507267e-01 9.68908608e-01 -2.53809919e-03 -1.19182658e+00 -1.50677457e-01 4.07284141e-01 1.24154985e+00 3.96224558e-01 -2.36278117e-01 -8.84305015e-02 4.27491039e-01 -7.15692818e-01 9.50756073e-01 2.86622912e-01 4.56188172e-01 -7.46590912e-01 5.89994490e-01 5.71999550e-01 -1.73660040e+00 1.62381567e-02 -4.72574569e-02 5.66172488e-02 5.34480274e-01 9.62384820e-01 -7.92196989e-01 9.32422101e-01 1.00737119e+00 7.65850365e-01 -9.01885331e-01 7.80775785e-01 -3.22580874e-01 1.99706331e-01 -8.34162831e-01 6.34485185e-01 3.88400108e-01 -6.21749878e-01 6.28633499e-01 8.31326187e-01 2.46555641e-01 4.47074175e-02 5.95466793e-01 9.54747617e-01 3.65273952e-01 -1.16888583e-01 -8.37870300e-01 6.41941428e-01 1.72627792e-01 1.40378094e+00 -1.33201683e+00 -3.66869986e-01 -3.25998038e-01 1.04265916e+00 -1.47512972e-01 2.69346625e-01 -9.53257024e-01 2.98242331e-01 3.34066540e-01 6.20746493e-01 6.98191762e-01 -3.14687610e-01 -5.80086052e-01 -9.69051123e-01 7.35636950e-02 -2.45618075e-01 3.37853551e-01 -8.94514859e-01 -7.65793860e-01 3.43254894e-01 3.37466627e-01 -1.13942218e+00 1.45748079e-01 -1.64402395e-01 -1.10546313e-01 4.98961002e-01 -1.33131480e+00 -1.70022357e+00 -5.83284616e-01 5.01040995e-01 7.47024655e-01 4.64136511e-01 6.32932186e-01 2.43148610e-01 -3.69846761e-01 -5.39538383e-01 -2.29783058e-01 -6.22396693e-02 -7.80481249e-02 -1.51726663e+00 3.27240735e-01 9.64619935e-01 2.79088676e-01 7.75724575e-02 3.88154060e-01 -8.52982819e-01 -7.33559251e-01 -1.58614767e+00 5.81198454e-01 -8.16964924e-01 -3.35077271e-02 -5.06726623e-01 -5.09723067e-01 9.14705515e-01 -1.50327999e-02 -7.95908347e-02 4.22149360e-01 -1.88863918e-01 1.65275663e-01 8.85360315e-02 -1.20594621e+00 2.87775606e-01 1.25655603e+00 -4.76397246e-01 -5.64134300e-01 4.76488471e-01 4.76407886e-01 -7.56281257e-01 -9.87612665e-01 7.40374506e-01 8.05077478e-02 -1.71208405e+00 1.43872666e+00 3.18081319e-01 1.66881531e-01 -8.52984309e-01 -7.02194571e-01 -1.12482607e+00 8.34058002e-02 -1.60002112e-01 3.62929374e-01 1.14190817e+00 3.00772935e-01 -2.40859896e-01 8.67962241e-01 5.30755818e-01 -6.25796258e-01 -3.89120698e-01 -1.09646022e+00 -3.29046071e-01 -3.98181677e-01 -9.65260804e-01 7.73138404e-01 8.38350534e-01 -9.88298655e-01 2.43727624e-01 1.01635389e-01 9.28369403e-01 8.70779991e-01 6.00727856e-01 1.07855356e+00 -1.51092100e+00 3.94708365e-02 -1.49422556e-01 -4.68836337e-01 -9.01906788e-01 -1.13130003e-01 -8.06453943e-01 -7.17789084e-02 -2.10255432e+00 -1.52802199e-01 -5.17663360e-01 4.29560393e-01 2.45462030e-01 -8.75823721e-02 7.17994511e-01 3.18800611e-03 5.67934439e-02 -6.23214841e-01 4.51614201e-01 9.98020649e-01 -2.65094757e-01 -4.17610794e-01 1.73154503e-01 -1.11673482e-01 1.18604410e+00 6.73407078e-01 -3.35901797e-01 -2.61647761e-01 -3.54148597e-01 1.19013608e-01 -1.49324805e-01 7.87400246e-01 -1.28943217e+00 -1.57356292e-01 1.16710320e-01 4.17707235e-01 -1.69734049e+00 9.21628773e-01 -1.28655505e+00 5.72215021e-01 3.36507767e-01 5.24727046e-01 -1.75688311e-01 1.49324685e-01 5.89628398e-01 -4.83900085e-02 7.65858032e-03 7.87379503e-01 -1.10235345e+00 -1.03256869e+00 2.82589883e-01 -4.24034357e-01 5.67556033e-03 1.42650127e+00 -1.05616212e+00 3.29571486e-01 -3.54534909e-02 -1.17629671e+00 2.29929149e-01 9.41819847e-01 6.83431402e-02 7.21309423e-01 -9.68935728e-01 -4.17823404e-01 4.08245146e-01 1.14562258e-01 9.43183064e-01 5.31782448e-01 6.39712512e-01 -7.23413169e-01 4.61620033e-01 1.70395687e-01 -1.34080625e+00 -1.50428355e+00 3.32079619e-01 6.34209991e-01 -1.03344277e-01 -6.95983529e-01 7.29077041e-01 4.82599229e-01 -1.09260833e+00 -1.80655748e-01 -6.93409204e-01 -2.11278111e-01 2.59221971e-01 -2.27377340e-01 4.03531790e-01 3.76473278e-01 -1.19224763e+00 -4.36149597e-01 1.12927091e+00 5.76928198e-01 -1.35783702e-01 1.66204214e+00 -5.48868656e-01 -2.37744480e-01 5.84521532e-01 6.93133235e-01 1.01244442e-01 -1.14958882e+00 1.51130363e-01 -2.69746691e-01 -6.18147910e-01 3.02200437e-01 -4.24786419e-01 -1.10448241e+00 7.39777625e-01 7.19301581e-01 2.96627015e-01 1.02744138e+00 2.33694166e-01 6.01522803e-01 -4.39004228e-02 5.82634091e-01 -1.30953443e+00 -2.37853318e-01 4.08690274e-01 4.83852804e-01 -1.24097967e+00 3.51604670e-01 -9.83284533e-01 -3.75587195e-01 9.85029876e-01 6.79559767e-01 4.77585457e-02 7.19755292e-01 1.95021704e-01 -1.03705272e-01 -6.44962013e-01 3.50724682e-02 -8.22611272e-01 1.61077261e-01 8.64845812e-01 -4.97540832e-02 5.39247915e-02 3.56319189e-01 -1.79768465e-02 -4.61641788e-01 -4.54885811e-01 4.76382434e-01 1.01663756e+00 -6.38079584e-01 -7.43024588e-01 -1.04059780e+00 2.66244918e-01 5.98080158e-02 -1.78990602e-01 -4.39406604e-01 1.00801885e+00 7.96023369e-01 8.51010025e-01 2.29131997e-01 -3.40996385e-01 6.29359365e-01 5.39982058e-02 4.88953143e-01 -8.94739211e-01 -6.15754604e-01 2.79621661e-01 1.30055189e-01 -5.54829240e-01 -9.44181740e-01 -7.24363804e-01 -1.08671629e+00 1.03175774e-01 -5.69493413e-01 -2.86304742e-01 6.68002665e-01 8.78537357e-01 9.85732079e-02 6.13221943e-01 5.49864590e-01 -1.37495315e+00 6.70300722e-01 -3.01611543e-01 -6.32025003e-01 2.40144178e-01 8.69218782e-02 -7.42493033e-01 -3.40576917e-01 2.55043775e-01]
[8.4148530960083, -2.840707540512085]
9cec8d11-d585-4a7c-8b2b-b419536e6280
human-interpretation-and-exploitation-of-self
2112.05364
null
https://arxiv.org/abs/2112.05364v2
https://arxiv.org/pdf/2112.05364v2.pdf
Human Guided Exploitation of Interpretable Attention Patterns in Summarization and Topic Segmentation
The multi-head self-attention mechanism of the transformer model has been thoroughly investigated recently. In one vein of study, researchers are interested in understanding why and how transformers work. In another vein, researchers propose new attention augmentation methods to make transformers more accurate, efficient and interpretable. In this paper, we combine these two lines of research in a human-in-the-loop pipeline to first discover important task-specific attention patterns. Then those patterns are injected, not only to smaller models, but also to the original model. The benefits of our pipeline and discovered patterns are demonstrated in two case studies with extractive summarization and topic segmentation. After discovering interpretable patterns in BERT-based models fine-tuned for the two downstream tasks, experiments indicate that when we inject the patterns into attention heads, the models show considerable improvements in accuracy and efficiency.
['Gabriel Murray', 'Lanjun Wang', 'Linzi Xing', 'Giuseppe Carenini', 'Wen Xiao', 'Raymond Li']
2021-12-10
null
null
null
null
['extractive-summarization']
['natural-language-processing']
[ 3.94183010e-01 7.05647826e-01 -3.12324345e-01 -2.82774568e-01 -7.93141007e-01 -6.13630176e-01 5.19598424e-01 2.20844179e-01 6.62452206e-02 3.80450308e-01 7.49480605e-01 -4.97267514e-01 8.80606696e-02 -4.58532721e-01 -7.62021601e-01 -1.36912167e-01 3.19406599e-01 7.97375679e-01 3.65366906e-01 -6.14812225e-02 5.46518326e-01 1.43859506e-01 -1.42695427e+00 5.98958731e-01 1.22028089e+00 6.80962741e-01 2.07911909e-01 5.23570061e-01 -2.89069563e-01 1.06393731e+00 -5.92267036e-01 -6.65425658e-01 3.17011476e-02 -3.57799500e-01 -1.22079480e+00 2.62613326e-01 4.09126282e-01 -7.93374181e-02 -1.27002094e-02 6.51129901e-01 7.77259842e-02 -1.31455377e-01 4.97461140e-01 -1.15165198e+00 -9.01706755e-01 1.30163980e+00 -7.67688513e-01 5.75814724e-01 1.74605414e-01 3.32660139e-01 1.59688711e+00 -8.12146902e-01 4.19122994e-01 1.32051086e+00 6.89574838e-01 2.23771229e-01 -1.25387287e+00 -4.97752756e-01 4.92327303e-01 3.82818311e-01 -9.06507969e-01 -5.55066407e-01 6.21876895e-01 -3.75643402e-01 1.37082279e+00 5.05469382e-01 7.60736525e-01 8.75943720e-01 2.06529886e-01 1.42377985e+00 7.08923280e-01 -3.76423985e-01 -1.02223463e-01 1.29645288e-01 5.12648225e-01 6.09269440e-01 4.59281743e-01 -4.64420944e-01 -6.67296290e-01 1.43657103e-01 5.86477816e-01 -7.06717446e-02 -1.76459849e-01 1.85565323e-01 -1.10096014e+00 8.33305538e-01 3.75880897e-01 5.49490929e-01 -7.02200353e-01 -3.76409059e-03 2.60373354e-01 -6.75097257e-02 8.58713746e-01 1.19776464e+00 -8.06049764e-01 -3.27408046e-01 -1.07447135e+00 1.45652428e-01 7.12829947e-01 1.01985276e+00 5.62406301e-01 1.24064051e-02 -7.60034502e-01 5.85976124e-01 2.49986708e-01 -2.88722236e-02 7.22188294e-01 -8.43781114e-01 5.40692747e-01 1.06773949e+00 -2.18202367e-01 -7.99471676e-01 -4.08662677e-01 -7.92121887e-01 -5.63595176e-01 -5.40727556e-01 4.04739946e-01 1.43043414e-01 -1.01593280e+00 1.44264901e+00 8.81037787e-02 2.33357511e-02 -2.11771160e-01 6.30196333e-01 7.27249563e-01 6.33143961e-01 2.61717588e-01 3.58429886e-02 1.84117270e+00 -1.50555491e+00 -9.34499860e-01 -7.30346739e-01 5.98591924e-01 -6.64612234e-01 1.59810650e+00 3.61045033e-01 -1.35874188e+00 -4.72368538e-01 -6.99241400e-01 -6.82425141e-01 -1.19185038e-01 8.39164853e-02 6.78099155e-01 2.05500543e-01 -1.14941251e+00 5.05912006e-01 -9.01780963e-01 -5.99658012e-01 8.08378696e-01 2.40990818e-01 2.61129946e-01 2.25577161e-01 -1.02957690e+00 1.05461943e+00 2.37068504e-01 -1.43305942e-01 -8.48854423e-01 -1.02764142e+00 -7.04728901e-01 6.93588197e-01 5.06935775e-01 -9.37933266e-01 1.81117940e+00 -1.05058825e+00 -1.26050198e+00 6.34917259e-01 -5.85420489e-01 -6.40684783e-01 1.89408764e-01 -6.41119599e-01 2.26507723e-01 -1.70968428e-01 4.33409095e-01 6.05568886e-01 7.28950083e-01 -9.60026503e-01 -8.86146009e-01 -4.13431793e-01 1.59548432e-01 2.02074721e-01 -4.78567004e-01 1.56143054e-01 -7.17096925e-01 -6.71289504e-01 -1.29131079e-01 -5.91482043e-01 -1.59029424e-01 -5.66517711e-01 -8.53508711e-01 -6.38377607e-01 8.68466020e-01 -9.95029509e-01 1.55281985e+00 -1.81490290e+00 3.03513497e-01 -1.75870299e-01 5.46786666e-01 1.01708278e-01 -7.47526530e-03 1.82635933e-01 -5.93418032e-02 6.84170723e-01 -2.92037636e-01 -6.47078812e-01 3.17335092e-02 1.36125371e-01 -5.24687946e-01 -1.43387571e-01 5.33069849e-01 1.31000769e+00 -8.84626508e-01 -4.59154427e-01 -1.67542130e-01 8.46604109e-02 -7.16439188e-01 3.19460392e-01 -3.37927073e-01 3.97387981e-01 -5.85305393e-01 5.50778747e-01 2.28302628e-01 -6.03971839e-01 1.04120500e-01 -3.38660389e-01 -1.48839191e-01 8.16490531e-01 -3.39810610e-01 1.24716401e+00 -4.75209713e-01 1.00299251e+00 -1.32209048e-01 -9.91301417e-01 4.62904453e-01 3.08515608e-01 1.03668503e-01 -5.38796067e-01 5.78012429e-02 -9.55619738e-02 2.51323670e-01 -6.97054505e-01 8.73853564e-01 8.84420425e-03 -2.16559097e-02 6.79506719e-01 4.81789596e-02 -4.59447503e-02 4.09228206e-01 3.55767280e-01 1.13852537e+00 4.76056375e-02 2.86438555e-01 -1.68272704e-01 9.51586217e-02 2.78312832e-01 5.38928747e-01 7.87837744e-01 4.71138395e-02 5.83066761e-01 8.00570011e-01 -3.34198117e-01 -1.16248238e+00 -4.67998981e-01 2.71225542e-01 1.42903197e+00 -2.82839835e-01 -7.47507215e-01 -1.01971662e+00 -7.96129465e-01 8.00819974e-03 1.09845948e+00 -8.42951238e-01 -1.87649354e-01 -4.99855399e-01 -5.40291071e-01 7.32201397e-01 9.73474443e-01 4.25164968e-01 -1.32249475e+00 -8.22155714e-01 1.92800611e-01 -6.66641891e-01 -1.10414600e+00 -5.41283190e-01 7.81913772e-02 -1.12512887e+00 -9.90313709e-01 -4.67802346e-01 -3.42348337e-01 5.92011094e-01 1.31782085e-01 1.45801544e+00 1.67666018e-01 1.30108729e-01 1.73481449e-01 -3.32358122e-01 -8.39070439e-01 -2.50744700e-01 6.97958469e-01 -4.04734492e-01 -4.88337688e-02 3.68337631e-01 -3.70426059e-01 -2.91589379e-01 2.24363551e-01 -7.18639612e-01 4.83399063e-01 8.96371603e-01 5.12425423e-01 2.81053245e-01 -1.71324313e-01 4.37901646e-01 -1.02727902e+00 9.39726472e-01 -4.19176310e-01 -1.78049892e-01 2.78282464e-01 -6.88062310e-01 3.04695100e-01 5.17040491e-01 -2.99562067e-01 -1.14758205e+00 -1.85545117e-01 5.25779277e-02 -2.82387674e-01 -6.39631301e-02 7.61462569e-01 -3.18276793e-01 6.10150814e-01 3.80599499e-01 2.18911707e-01 -2.39424154e-01 -5.98482490e-01 4.49217319e-01 5.30060947e-01 3.94666851e-01 -6.65777862e-01 6.44745648e-01 2.09101170e-01 -6.05549634e-01 -5.93332708e-01 -1.32481396e+00 -2.75482982e-01 -5.52577734e-01 1.86876297e-01 8.32583070e-01 -6.62267745e-01 -3.61470580e-01 2.79166520e-01 -1.33791101e+00 -5.28845489e-01 -6.23463988e-01 1.07127596e-02 -3.33022624e-01 1.91593811e-01 -7.19341636e-01 -6.21964455e-01 -7.15629041e-01 -1.24725437e+00 1.31547117e+00 3.40932578e-01 -7.40863025e-01 -9.73818064e-01 -2.04293624e-01 7.54347682e-01 5.55573225e-01 -4.24685717e-01 1.14462066e+00 -1.13202536e+00 -7.22111821e-01 1.05570473e-01 -1.99214846e-01 1.06771268e-01 1.03539303e-01 8.39199498e-02 -1.13626635e+00 1.62212059e-01 -1.12501144e-01 1.34749591e-01 9.13806915e-01 6.33056283e-01 1.53626513e+00 -7.81911254e-01 -5.28499842e-01 4.21133250e-01 6.27187967e-01 -5.27149476e-02 6.82951450e-01 3.52006614e-01 8.94882441e-01 6.71160519e-01 4.22311038e-01 7.89130479e-02 7.61536419e-01 6.10679626e-01 2.52707064e-01 -2.48871073e-01 -4.99735326e-02 -4.61571187e-01 2.42842034e-01 8.94708216e-01 -1.64913327e-01 -4.11184609e-01 -1.03649747e+00 9.91297364e-01 -1.97578251e+00 -9.30955946e-01 -2.68142819e-01 1.69123399e+00 1.00286531e+00 3.16884965e-01 6.37588799e-02 1.14519253e-01 6.15913391e-01 4.87556905e-02 -5.49449503e-01 -6.52362227e-01 1.13201715e-01 2.54091680e-01 3.98264080e-02 3.69811177e-01 -8.71597111e-01 1.22866166e+00 6.45051336e+00 7.86750972e-01 -9.83095765e-01 1.58369526e-01 9.69488263e-01 -7.14375004e-02 -6.44224405e-01 2.42313534e-01 -7.64438450e-01 2.79291630e-01 1.01487243e+00 -4.46973592e-01 2.53486812e-01 7.63236821e-01 4.64107692e-01 -4.48932080e-03 -1.26236296e+00 3.61664027e-01 -5.62703051e-03 -1.42522395e+00 2.29998738e-01 6.27195537e-02 5.11292398e-01 -1.34381101e-01 5.56218736e-02 4.91776496e-01 3.87149274e-01 -9.48580980e-01 9.54871058e-01 3.04675221e-01 3.13751459e-01 -5.28312922e-01 7.06098616e-01 2.77095258e-01 -1.15808260e+00 -1.93087086e-01 -8.75014737e-02 -2.23582327e-01 2.01149702e-01 6.33322001e-01 -1.25035429e+00 3.95353347e-01 8.18236768e-01 8.83904755e-01 -8.44129682e-01 1.02593040e+00 -6.49528980e-01 1.22660804e+00 -9.72088426e-02 8.91183987e-02 2.53115743e-01 1.55466408e-01 6.39100313e-01 1.36831200e+00 6.61502853e-02 1.29777402e-01 -1.65976927e-01 1.30577910e+00 -2.85524070e-01 -8.63501728e-02 -4.24881428e-01 -3.78884286e-01 3.93169194e-01 1.50677443e+00 -8.60799015e-01 -6.41705990e-01 -6.09939247e-02 7.69257307e-01 3.11563104e-01 3.24933529e-01 -8.67907345e-01 -3.04244906e-01 5.25943696e-01 4.74638164e-01 4.52419758e-01 -3.49295177e-02 -7.76338339e-01 -1.10307300e+00 -8.85793716e-02 -1.14997244e+00 2.62673080e-01 -1.00085902e+00 -1.06557679e+00 4.95736718e-01 1.36341691e-01 -4.24914598e-01 -1.32125437e-01 -1.43812329e-01 -1.03175855e+00 7.50253022e-01 -1.28329730e+00 -1.37732697e+00 -2.26100072e-01 1.41165778e-01 1.27924991e+00 1.66099653e-01 4.42684621e-01 9.45238546e-02 -9.06239688e-01 5.57181001e-01 -3.43918741e-01 1.25056759e-01 4.03746367e-01 -1.34611201e+00 9.83173430e-01 1.10191262e+00 2.90709287e-01 9.49100256e-01 6.87601328e-01 -7.03014314e-01 -1.15126109e+00 -1.07381427e+00 1.26420176e+00 -7.42993474e-01 6.14132822e-01 -4.08588409e-01 -1.19150829e+00 1.18623996e+00 6.67562664e-01 -6.43113136e-01 5.10291636e-01 4.33796823e-01 -1.84054792e-01 1.33813173e-01 -8.82624328e-01 5.09151697e-01 1.03682816e+00 -3.52244586e-01 -9.83220935e-01 2.04887033e-01 1.08879852e+00 -4.50250983e-01 -5.39052248e-01 1.61730438e-01 5.77720046e-01 -6.89143002e-01 7.01579988e-01 -9.28245902e-01 8.13125253e-01 -9.09146667e-02 2.84285009e-01 -1.49516010e+00 -6.18816316e-01 -7.05777168e-01 -2.64638156e-01 1.52705991e+00 8.64077032e-01 -4.97888088e-01 5.16273081e-01 6.16374016e-01 -6.90841436e-01 -8.97761405e-01 -5.31374097e-01 -3.05333108e-01 -4.66870144e-02 -5.17192185e-01 7.28501260e-01 9.06886101e-01 8.01850557e-02 9.08941209e-01 -6.98375702e-02 -4.02345024e-02 2.66495407e-01 1.68257073e-01 7.36788332e-01 -1.34548020e+00 -2.12436318e-01 -5.83123922e-01 2.39637658e-01 -1.20501280e+00 1.36540756e-01 -7.99666226e-01 1.19754978e-01 -1.94818068e+00 5.57971418e-01 -2.19922781e-01 -9.77545977e-02 9.62464809e-01 -6.65835917e-01 -1.31887764e-01 3.71088237e-01 3.51171494e-01 -6.47945344e-01 4.76819694e-01 1.06111658e+00 -3.89077842e-01 -2.56192982e-01 -2.82340162e-02 -1.53417027e+00 8.13394248e-01 7.88745821e-01 -4.55050170e-01 -4.01826382e-01 -8.11845720e-01 1.59487844e-01 -4.60890383e-01 2.16457278e-01 -6.73978209e-01 3.40933919e-01 -4.26815748e-02 1.92247212e-01 -6.88176274e-01 3.81727144e-03 -5.00528753e-01 -1.88057214e-01 1.80102721e-01 -5.46733677e-01 3.02570790e-01 5.46917796e-01 2.81140655e-01 -6.12114370e-02 -1.30265981e-01 4.01292503e-01 -1.38042971e-01 -5.37926316e-01 6.85194954e-02 -5.89400887e-01 3.01299423e-01 7.06128299e-01 -2.29702801e-01 -4.04573977e-01 -6.29781663e-01 -5.84995210e-01 3.28582138e-01 1.56154647e-01 5.17344713e-01 4.49364722e-01 -7.94754386e-01 -8.35787773e-01 1.08061343e-01 -4.79106754e-02 2.96267301e-01 -5.96377626e-02 1.08509862e+00 -3.24581005e-02 7.67235518e-01 7.95660317e-02 -6.98711991e-01 -1.20998347e+00 4.10317451e-01 2.62646645e-01 -9.01648939e-01 -4.26945537e-01 8.62315536e-01 4.59764957e-01 -3.06300789e-01 3.75375785e-02 -9.76181269e-01 -2.26399750e-01 2.11981997e-01 2.92224765e-01 3.22322190e-01 2.62389239e-02 -3.79690707e-01 -3.77247661e-01 3.94308388e-01 -3.77257943e-01 -1.23012746e-02 1.53956020e+00 -1.08655661e-01 -1.99142650e-01 2.02589765e-01 6.98977828e-01 7.08014667e-02 -1.06198001e+00 -9.97861400e-02 4.55461979e-01 -1.67528361e-01 1.21347092e-01 -9.78772283e-01 -1.22286558e+00 1.07049477e+00 -8.51998553e-02 5.97467184e-01 1.10652888e+00 3.55168521e-01 8.03672254e-01 9.54145640e-02 -2.19983160e-01 -7.70174682e-01 7.99372494e-02 7.22160101e-01 1.12038207e+00 -9.29454625e-01 1.99418962e-02 -4.08784658e-01 -9.00984764e-01 8.79544616e-01 8.23800385e-01 2.81449407e-01 6.93068877e-02 2.82335311e-01 -3.21560174e-01 -4.78910476e-01 -1.09482706e+00 -1.97935954e-01 5.02450883e-01 4.23049569e-01 7.45325565e-01 -2.71277457e-01 -2.30259709e-02 1.05123949e+00 -3.61012906e-01 6.95912242e-02 5.06455421e-01 7.42115021e-01 -4.89630312e-01 -8.05869818e-01 -2.78954625e-01 6.88503563e-01 -7.16451526e-01 -5.26626706e-01 -7.38504827e-01 7.49986351e-01 -1.76172152e-01 1.02956402e+00 3.43946069e-01 -2.94420034e-01 6.26587510e-01 2.06933394e-01 1.18681774e-01 -9.13387239e-01 -1.08367467e+00 9.58121568e-02 3.15719754e-01 -5.04176497e-01 -1.90615252e-01 -5.82777619e-01 -1.24110425e+00 -7.32433498e-02 -4.89554375e-01 3.49259138e-01 2.53248483e-01 1.10897219e+00 7.95525491e-01 1.06611383e+00 2.29483142e-01 -7.42882729e-01 -3.32849473e-01 -1.48863149e+00 1.51862213e-02 1.86467022e-01 1.36859000e-01 -3.83906245e-01 -3.25309455e-01 3.10582489e-01]
[11.30672836303711, 8.471853256225586]
d27bbb47-cd49-4959-97a9-dd8c68f7ccfd
self-supervised-learning-by-estimating-twin-1
2110.07402
null
https://arxiv.org/abs/2110.07402v4
https://arxiv.org/pdf/2110.07402v4.pdf
Self-Supervised Learning by Estimating Twin Class Distributions
We present TWIST, a simple and theoretically explainable self-supervised representation learning method by classifying large-scale unlabeled datasets in an end-to-end way. We employ a siamese network terminated by a softmax operation to produce twin class distributions of two augmented images. Without supervision, we enforce the class distributions of different augmentations to be consistent. However, simply minimizing the divergence between augmentations will cause collapsed solutions, i.e., outputting the same class probability distribution for all images. In this case, no information about the input image is left. To solve this problem, we propose to maximize the mutual information between the input and the class predictions. Specifically, we minimize the entropy of the distribution for each sample to make the class prediction for each sample assertive and maximize the entropy of the mean distribution to make the predictions of different samples diverse. In this way, TWIST can naturally avoid the collapsed solutions without specific designs such as asymmetric network, stop-gradient operation, or momentum encoder. As a result, TWIST outperforms state-of-the-art methods on a wide range of tasks. Especially, TWIST performs surprisingly well on semi-supervised learning, achieving 61.2% top-1 accuracy with 1% ImageNet labels using a ResNet-50 as backbone, surpassing previous best results by an absolute improvement of 6.2%. Codes and pre-trained models are given on: https://github.com/bytedance/TWIST
['Hang Li', 'Huaping Liu', 'Rufeng Zhang', 'Tao Kong', 'Feng Wang']
2021-10-14
self-supervised-learning-by-estimating-twin
https://openreview.net/forum?id=TLgW66V2CbP
https://openreview.net/pdf?id=TLgW66V2CbP
null
['self-supervised-image-classification', 'semi-supervised-image-classification', 'unsupervised-image-classification']
['computer-vision', 'computer-vision', 'computer-vision']
[ 3.47791225e-01 6.15678072e-01 -5.17779768e-01 -5.66672564e-01 -5.57876766e-01 -5.53626418e-01 4.99385744e-01 -1.53348461e-01 -4.45128262e-01 9.05731499e-01 1.74540043e-01 -2.47849733e-01 3.20173889e-01 -5.10971606e-01 -9.65667188e-01 -9.03773367e-01 2.67018110e-01 6.71606064e-01 -2.49821886e-01 1.38570011e-01 -4.22430262e-02 8.39624554e-02 -1.37164068e+00 2.28136316e-01 7.98711360e-01 1.11757457e+00 1.74353883e-01 3.36557865e-01 -9.79393199e-02 8.31779420e-01 -3.54966253e-01 -4.72607106e-01 3.51059943e-01 -6.14554524e-01 -8.22446287e-01 3.03760558e-01 5.78374803e-01 -2.33801425e-01 -4.77893829e-01 1.11099875e+00 3.01390380e-01 1.12637587e-01 8.38601828e-01 -1.29977584e+00 -8.89348745e-01 8.66180241e-01 -6.56358838e-01 -1.79417536e-01 -1.33649334e-01 3.03769052e-01 1.13168991e+00 -7.91979015e-01 6.31686747e-01 7.79537261e-01 2.96399087e-01 7.56097138e-01 -1.59719968e+00 -8.19465339e-01 2.19897941e-01 5.28645283e-03 -1.24012506e+00 -4.00568783e-01 6.11666739e-01 -2.42038980e-01 6.06831789e-01 1.70150533e-01 4.72251981e-01 1.08924663e+00 -2.48208754e-02 1.00170743e+00 1.07218623e+00 -1.54468819e-01 2.79757291e-01 2.84209043e-01 -1.24063313e-01 7.19613254e-01 9.87043828e-02 -1.61632612e-01 -4.48002517e-01 2.15097144e-02 5.89254558e-01 1.85836688e-01 -2.28685603e-01 -6.15358353e-01 -1.38062525e+00 7.84499586e-01 8.10230792e-01 8.18698294e-03 -3.47016007e-01 1.38363719e-01 1.72499180e-01 2.05242187e-01 5.58796108e-01 4.59507793e-01 -6.96158171e-01 1.91688910e-01 -7.20122993e-01 1.55418172e-01 4.94216144e-01 7.07215905e-01 9.34669375e-01 8.55039880e-02 -9.15185511e-02 7.61885405e-01 3.43575329e-01 4.69460130e-01 6.68205440e-01 -9.50314343e-01 5.32225549e-01 5.56936800e-01 -9.57029760e-02 -6.21647060e-01 -2.78337389e-01 -8.18393648e-01 -1.12106323e+00 1.20079204e-01 6.83445036e-01 -3.16710114e-01 -1.28810143e+00 2.17069983e+00 2.24377289e-01 1.33815736e-01 1.79387033e-01 8.72721255e-01 6.06516838e-01 6.54683053e-01 -2.87153572e-03 -6.50214180e-02 1.14769089e+00 -1.11099577e+00 -5.29706955e-01 -4.97414947e-01 8.37197781e-01 -6.09972894e-01 1.14853537e+00 1.39742494e-01 -9.55870986e-01 -3.08887690e-01 -1.03225851e+00 8.48674253e-02 -1.28044993e-01 3.05243790e-01 5.27861118e-01 2.27510124e-01 -6.40876234e-01 7.74332464e-01 -9.50464368e-01 -4.34829071e-02 7.82265723e-01 4.26284015e-01 -5.39039135e-01 -9.01083276e-02 -8.29859555e-01 6.46608174e-01 4.32923317e-01 -1.28775552e-01 -7.58505762e-01 -9.11540508e-01 -8.19884658e-01 1.68730095e-01 2.92094588e-01 -5.45250654e-01 1.16725469e+00 -1.22636259e+00 -1.40938461e+00 9.25190151e-01 -2.08384380e-01 -6.40020430e-01 5.57378769e-01 -1.49031714e-01 -1.43131698e-02 -1.29492313e-01 2.00249016e-01 1.07133973e+00 7.36052752e-01 -1.08933961e+00 -3.29821795e-01 -2.41011709e-01 -3.06387335e-01 3.33146185e-01 -5.43210030e-01 -5.64764738e-01 -3.53270769e-01 -5.24436533e-01 3.43019158e-01 -1.31213772e+00 -4.60245878e-01 1.91313908e-01 -7.26583898e-01 -2.30619907e-02 5.19257903e-01 -4.11077857e-01 8.36499870e-01 -2.42192793e+00 1.50971472e-01 4.28869613e-02 3.47715646e-01 8.04306269e-02 -3.52204949e-01 -6.29236996e-02 -2.31818795e-01 1.94634378e-01 -5.70912480e-01 -6.01921856e-01 -1.02892078e-01 2.25287572e-01 -4.35436815e-01 5.06339490e-01 3.75069648e-01 9.05529261e-01 -8.61874580e-01 -2.44847178e-01 1.85378775e-01 3.97825927e-01 -6.40115857e-01 2.35578671e-01 -2.59490460e-01 6.13643169e-01 -1.89244241e-01 2.03780740e-01 7.02332199e-01 -6.71674788e-01 2.16296360e-01 -1.22450382e-01 3.42907488e-01 4.00038660e-01 -1.00176013e+00 1.71684647e+00 -4.16067302e-01 6.34195685e-01 -2.65633434e-01 -1.08598971e+00 8.57720315e-01 2.32709914e-01 3.67506981e-01 -5.42297423e-01 -3.04141361e-03 2.85107642e-01 1.64644986e-01 3.89752276e-02 1.05354980e-01 -2.30502173e-01 8.82616714e-02 5.68622231e-01 3.13807368e-01 -1.24572903e-01 1.02974195e-02 4.22597289e-01 7.84751415e-01 6.72169998e-02 1.29507169e-01 -6.56866431e-02 1.26694262e-01 -2.50033200e-01 7.12449431e-01 6.04836583e-01 -3.71571295e-02 1.02244592e+00 7.38895416e-01 -5.25367141e-01 -1.26140571e+00 -1.26776326e+00 -2.56712623e-02 7.43032277e-01 6.94911554e-02 -2.05861434e-01 -7.01901257e-01 -1.09509933e+00 -5.86848427e-03 7.82590806e-01 -7.32675374e-01 -2.67702758e-01 -2.22689867e-01 -7.27461576e-01 2.61228472e-01 4.21817660e-01 5.87442517e-01 -9.46742058e-01 -1.76988631e-01 3.44765224e-02 -2.94785142e-01 -1.14365971e+00 -6.28445745e-01 5.78217685e-01 -8.76493335e-01 -8.61440539e-01 -7.20848799e-01 -7.84366071e-01 1.04084599e+00 3.77051271e-02 9.85668957e-01 -6.94026276e-02 -3.37053314e-02 -3.12644809e-01 -6.13151789e-02 -1.19430594e-01 -1.99509352e-01 3.67643267e-01 6.72475398e-02 9.59940478e-02 1.01535507e-01 -6.08967721e-01 -6.84707820e-01 4.77953225e-01 -8.27312231e-01 4.14118141e-01 6.79475904e-01 1.09486580e+00 7.57262111e-01 -2.89726973e-01 5.75759768e-01 -8.13991547e-01 1.64974794e-01 -7.15686023e-01 -3.78631383e-01 1.41602784e-01 -6.26789749e-01 4.23749477e-01 9.03886318e-01 -6.86722219e-01 -8.14684629e-01 3.72586161e-01 5.26207052e-02 -6.41425908e-01 -1.94909707e-01 4.16702360e-01 -4.33783196e-02 4.75097507e-01 7.87175000e-01 4.44674581e-01 3.59756827e-01 -2.98460096e-01 4.29602116e-01 6.08526409e-01 5.33576548e-01 -3.70742887e-01 7.67523527e-01 4.31684256e-01 -1.16476975e-01 -4.85406995e-01 -1.21899211e+00 -1.99443951e-01 -4.13342744e-01 5.56318276e-02 7.81266689e-01 -9.94342268e-01 -3.71870935e-01 4.58981335e-01 -7.50451028e-01 -6.16769075e-01 -5.82730055e-01 6.75699592e-01 -5.01917541e-01 1.07535541e-01 -4.64525253e-01 -4.66677308e-01 -1.98588580e-01 -1.15468860e+00 8.27958167e-01 3.76649022e-01 -3.01283360e-01 -7.87679791e-01 -2.01166674e-01 5.21337152e-01 3.70415837e-01 -4.62678540e-03 6.75221980e-01 -8.86073709e-01 -5.48341691e-01 -1.66319087e-01 -1.59893841e-01 6.26159191e-01 2.80361056e-01 -1.43074319e-01 -9.28555489e-01 -2.44617432e-01 -2.74123460e-01 -7.40340829e-01 1.06006086e+00 3.57351750e-01 1.40421212e+00 -5.34185767e-01 -1.54616117e-01 7.79051304e-01 1.10916317e+00 -1.30860135e-01 6.65017068e-01 6.96136728e-02 7.54212320e-01 5.30332923e-01 3.19638222e-01 3.73076618e-01 2.75175124e-01 4.24159586e-01 6.01397038e-01 -5.38113900e-02 -1.69565126e-01 -6.14047945e-01 2.13434145e-01 5.17617404e-01 1.81028262e-01 -3.79401833e-01 -7.37552047e-01 4.61324006e-01 -1.80089724e+00 -9.00922239e-01 1.89872965e-01 2.45889759e+00 1.00831318e+00 2.72447497e-01 -1.11043014e-01 -2.08177045e-01 6.69582129e-01 2.92727202e-01 -9.03330803e-01 2.07838099e-02 -2.10066065e-01 6.27119094e-02 6.39987588e-01 5.54350317e-01 -1.15668595e+00 1.00916576e+00 5.34093142e+00 9.39059317e-01 -1.28218484e+00 -8.58597532e-02 1.18354130e+00 -4.19079810e-01 -5.17144322e-01 7.41235316e-02 -5.79257429e-01 6.11973047e-01 7.03718781e-01 -7.35029355e-02 5.11061251e-01 8.99499536e-01 -1.11750968e-01 6.73391819e-02 -9.76713061e-01 8.26939046e-01 -1.70377165e-01 -1.36600101e+00 -1.02546401e-02 1.69055790e-01 9.52896118e-01 4.39247936e-01 3.32973063e-01 2.61990041e-01 3.58831316e-01 -1.07891190e+00 7.22253263e-01 1.98352888e-01 8.92963886e-01 -5.53138673e-01 7.03079402e-01 4.10854340e-01 -8.28609526e-01 1.08965762e-01 -3.14217567e-01 -1.83894888e-01 5.14352992e-02 8.48748982e-01 -8.44945967e-01 1.94539189e-01 4.13609684e-01 7.77094424e-01 -2.96318144e-01 7.94515967e-01 -4.64981586e-01 7.88143098e-01 -6.35046840e-01 4.09511961e-02 2.13482559e-01 -2.15121001e-01 2.95461178e-01 7.68498302e-01 1.66523546e-01 -1.43537655e-01 2.10800245e-01 9.61709976e-01 -6.32915974e-01 7.49215763e-03 -5.42679131e-01 -9.11016613e-02 3.92911643e-01 1.19310820e+00 -7.70898461e-01 -4.91742134e-01 -1.91708162e-01 1.04285443e+00 5.42104244e-01 4.57685649e-01 -7.86117613e-01 -3.03421527e-01 6.69281721e-01 -5.00447340e-02 2.98310131e-01 1.00067914e-01 -6.51614368e-01 -1.34609997e+00 1.50166243e-01 -7.10952938e-01 2.24936783e-01 -6.50817454e-01 -1.31982625e+00 6.57188773e-01 -2.08485782e-01 -1.28568470e+00 -3.10442656e-01 -4.66591448e-01 -6.33204043e-01 8.27071548e-01 -1.30444634e+00 -8.57536197e-01 -1.04115166e-01 2.72766590e-01 3.55571687e-01 -2.37439215e-01 7.39872158e-01 9.60014611e-02 -5.94285667e-01 8.01562309e-01 1.89827695e-01 2.48091832e-01 8.02059472e-01 -1.30699909e+00 3.64833266e-01 5.83489835e-01 3.34164232e-01 4.54302818e-01 6.67899430e-01 -4.98691291e-01 -8.78141463e-01 -1.08015883e+00 9.11826968e-01 -2.20765844e-01 5.95678389e-01 -3.82445216e-01 -1.03232157e+00 8.41030478e-01 7.31799155e-02 4.94291782e-01 6.64808869e-01 2.30466560e-01 -6.60384119e-01 -6.87866881e-02 -1.11090744e+00 7.76524365e-01 9.76875603e-01 -4.33719575e-01 -1.58294410e-01 5.36136806e-01 7.07958460e-01 -4.70608264e-01 -5.87592542e-01 4.05751288e-01 5.86053193e-01 -8.43976796e-01 7.57096469e-01 -6.45571709e-01 9.01349068e-01 -3.10237080e-01 -2.62336165e-01 -1.45647073e+00 -2.09271774e-01 -3.60969990e-01 -2.02806190e-01 9.67545688e-01 9.37250972e-01 -8.66550922e-01 1.18560827e+00 7.03024089e-01 -8.97332877e-02 -1.38026726e+00 -8.75352383e-01 -7.29954362e-01 2.58642912e-01 -2.37175599e-01 4.89316761e-01 1.06235743e+00 -1.45522812e-02 4.61827129e-01 -5.68679988e-01 9.50683001e-03 7.80958951e-01 9.82480347e-02 6.41880214e-01 -8.71123850e-01 -3.49960625e-01 -4.53589469e-01 -3.70825261e-01 -1.17318642e+00 4.13183540e-01 -1.06867599e+00 2.83069815e-03 -1.25303745e+00 4.51019794e-01 -6.08319342e-01 -3.45403641e-01 8.74297976e-01 -3.26749593e-01 4.54507291e-01 3.62185359e-01 3.49383086e-01 -4.87141967e-01 8.47173810e-01 1.28410792e+00 -2.10356042e-01 -1.64901480e-01 3.95953245e-02 -9.03188467e-01 6.60129905e-01 1.17234087e+00 -6.42313659e-01 -5.05793035e-01 -4.10385698e-01 7.78900012e-02 -1.65930405e-01 3.36382508e-01 -8.33533704e-01 5.21761514e-02 -9.99377221e-02 5.94282210e-01 -2.80447841e-01 3.04931551e-01 -6.33886218e-01 -2.44598873e-02 5.30015171e-01 -7.76653111e-01 -3.27188522e-01 4.91127595e-02 3.90166730e-01 -2.06462532e-01 -1.16653673e-01 9.74022627e-01 4.00011661e-03 -1.56874999e-01 5.48107386e-01 -8.90654251e-02 2.38943279e-01 9.95379210e-01 3.46307680e-02 -3.32214475e-01 -5.64263821e-01 -6.88593268e-01 3.72348487e-01 5.97122490e-01 2.31273070e-01 4.10072714e-01 -1.32553971e+00 -8.21653724e-01 3.36367548e-01 1.72962416e-02 3.34518880e-01 3.48877370e-01 6.79692030e-01 -1.88409209e-01 1.15061402e-01 -1.76231220e-01 -5.95068872e-01 -9.96476054e-01 2.15651810e-01 3.62028450e-01 -4.40663308e-01 -3.57141584e-01 9.16325271e-01 5.06886065e-01 -8.98532510e-01 7.64955804e-02 -7.01358914e-02 1.17167838e-01 -1.24962464e-01 4.66736346e-01 -1.31756753e-01 -2.29887739e-01 -3.71231019e-01 -2.84974903e-01 2.30770677e-01 -4.71311808e-01 -1.70180231e-01 1.36040139e+00 9.54965875e-02 1.53567016e-01 4.01765496e-01 1.57507241e+00 -1.93448007e-01 -1.67550206e+00 -4.38273937e-01 -5.00991940e-01 -3.19582373e-01 -4.99028973e-02 -8.97098184e-01 -1.43565953e+00 7.43372679e-01 4.59270418e-01 -4.82733324e-02 9.62955952e-01 3.71340901e-01 5.36528111e-01 3.74064744e-01 -7.49012157e-02 -7.48262048e-01 1.32667556e-01 4.89799052e-01 7.29740798e-01 -1.56283200e+00 2.38623824e-02 -1.69990033e-01 -1.08534300e+00 8.11611414e-01 7.98913240e-01 -1.19655535e-01 5.22896945e-01 2.08305418e-02 8.13980848e-02 1.28144816e-01 -8.85149121e-01 3.84412259e-02 2.18148798e-01 3.60622317e-01 5.39173782e-01 2.63958424e-01 1.63556397e-01 3.04374605e-01 -3.62184197e-01 -8.78536999e-02 3.61575276e-01 6.09379292e-01 -1.14444628e-01 -9.24056113e-01 -6.52867034e-02 8.00197780e-01 -4.48909819e-01 -1.92732006e-01 -2.20166087e-01 4.69189107e-01 -1.54154733e-01 5.35723507e-01 2.83415437e-01 -4.15145278e-01 -8.33934620e-02 1.53442591e-01 2.42495149e-01 -6.02190673e-01 -1.68754622e-01 -2.06996948e-01 -1.13539986e-01 -3.19685161e-01 -1.05136588e-01 -6.34174347e-01 -1.33104169e+00 -2.57076055e-01 -3.46664220e-01 7.60877058e-02 6.05476201e-01 9.73455012e-01 5.14468968e-01 3.51663351e-01 7.62344778e-01 -8.34996760e-01 -8.33624959e-01 -9.92990851e-01 -4.87114400e-01 5.55564582e-01 3.95239472e-01 -3.69729578e-01 -6.96641326e-01 -4.30956408e-02]
[9.477259635925293, 2.6461987495422363]
3850e12a-18c2-4ed4-9a44-e81f297700cc
a-fully-convolutional-deep-auditory-model-for
1612.05082
null
http://arxiv.org/abs/1612.05082v1
http://arxiv.org/pdf/1612.05082v1.pdf
A Fully Convolutional Deep Auditory Model for Musical Chord Recognition
Chord recognition systems depend on robust feature extraction pipelines. While these pipelines are traditionally hand-crafted, recent advances in end-to-end machine learning have begun to inspire researchers to explore data-driven methods for such tasks. In this paper, we present a chord recognition system that uses a fully convolutional deep auditory model for feature extraction. The extracted features are processed by a Conditional Random Field that decodes the final chord sequence. Both processing stages are trained automatically and do not require expert knowledge for optimising parameters. We show that the learned auditory system extracts musically interpretable features, and that the proposed chord recognition system achieves results on par or better than state-of-the-art algorithms.
['Filip Korzeniowski', 'Gerhard Widmer']
2016-12-15
null
null
null
null
['chord-recognition']
['audio']
[ 3.73471826e-01 1.88297257e-02 3.22022825e-01 -3.74162376e-01 -9.90822196e-01 -9.88740683e-01 4.17751193e-01 1.64190054e-01 -7.03547180e-01 4.73896749e-02 2.10780367e-01 -2.19821464e-02 -1.53779849e-01 -6.50035441e-01 -3.99085343e-01 -2.23985031e-01 -4.20705527e-01 3.89235318e-01 1.82176828e-01 -3.54284644e-01 7.16235220e-01 3.74443173e-01 -1.77089715e+00 5.30646145e-01 1.06636278e-01 1.27949119e+00 3.51961404e-02 1.49473071e+00 3.67048904e-02 8.07753682e-01 -6.86346114e-01 -2.39192411e-01 3.48831296e-01 -4.80376989e-01 -1.09077728e+00 -3.13246459e-01 1.91862836e-01 -4.71072905e-02 -1.22353047e-01 6.23572171e-01 6.44246638e-01 2.94965029e-01 5.13016462e-01 -9.82320845e-01 -3.63671452e-01 8.96537304e-01 1.26220763e-01 -1.00840375e-01 4.12243545e-01 1.56725436e-01 1.68299687e+00 -8.87496650e-01 2.66657442e-01 8.44275057e-01 8.63496244e-01 3.12852412e-01 -1.14494145e+00 -5.62296450e-01 -2.87872314e-01 4.42221135e-01 -1.35849154e+00 -7.54965067e-01 9.21149194e-01 -6.26335800e-01 1.30904770e+00 2.77295798e-01 9.10870492e-01 6.58347785e-01 1.39740348e-01 9.14698720e-01 7.56154776e-01 -6.75615907e-01 3.08873296e-01 -6.06809199e-01 -1.18531987e-01 6.70130968e-01 -5.55234432e-01 3.11951339e-01 -1.33189762e+00 6.38170466e-02 7.49880552e-01 -6.53281569e-01 -3.26294340e-02 1.27814729e-02 -1.26597607e+00 6.94010317e-01 2.18173191e-01 1.76438138e-01 -4.74191904e-01 3.03477883e-01 7.90041983e-01 3.89207333e-01 -1.87896103e-01 9.32259202e-01 -6.33878112e-01 -8.21536362e-01 -1.56873894e+00 4.93257403e-01 9.22396421e-01 5.37828386e-01 3.20843130e-01 8.69531035e-02 3.74594703e-02 7.66375184e-01 2.67065465e-01 -2.26724725e-02 4.36232537e-01 -1.06607747e+00 -1.28886148e-01 3.11563253e-01 -3.18849325e-01 -8.51264596e-01 -4.91324335e-01 -5.52736580e-01 -5.66316068e-01 6.05703294e-01 5.82305610e-01 5.46922274e-02 -7.51986921e-01 1.45742130e+00 -1.23756722e-01 1.57025620e-01 9.63464100e-03 9.93031681e-01 6.17775619e-01 2.27220625e-01 -1.62002265e-01 2.67925292e-01 1.31947553e+00 -9.71126497e-01 -4.90062892e-01 -2.06045985e-01 1.01064906e-01 -1.28667438e+00 1.35972416e+00 1.22559631e+00 -1.20265388e+00 -9.53926444e-01 -1.31460619e+00 -4.18600887e-01 8.37047026e-02 1.37631401e-01 7.87488520e-01 5.98410070e-01 -7.99890637e-01 9.70636606e-01 -9.19418752e-01 9.72518623e-02 3.04404438e-01 4.80761409e-01 -2.58292705e-01 6.69624507e-01 -9.66905892e-01 5.18655121e-01 5.05971789e-01 2.20736563e-01 -1.14972615e+00 -5.10211706e-01 -5.31482339e-01 1.74445197e-01 2.31570378e-01 -6.90164208e-01 1.97947025e+00 -7.10086823e-01 -2.13707924e+00 7.47748435e-01 1.11185297e-01 -4.36026365e-01 1.09102629e-01 -7.17094243e-01 -1.24468833e-01 1.72159821e-01 -3.07460546e-01 6.97869897e-01 9.61690545e-01 -5.10237753e-01 -8.08278739e-01 -8.80959630e-02 -1.78458959e-01 -1.03861513e-02 -2.32794002e-01 3.47057104e-01 -3.81552756e-01 -8.98517489e-01 8.35190639e-02 -1.00139391e+00 -2.28383243e-01 7.68218338e-02 -6.59984112e-01 -1.44897193e-01 3.95235091e-01 -7.54768312e-01 1.19614398e+00 -2.12384248e+00 2.53754079e-01 2.00650036e-01 -8.07316452e-02 -9.41298530e-02 -1.21684715e-01 4.66763198e-01 1.74691007e-01 -1.62407741e-01 -4.08086270e-01 -3.91849279e-01 1.84666023e-01 -3.28707881e-02 -5.33585668e-01 1.64976314e-01 3.25990140e-01 6.59147322e-01 -7.71931171e-01 -2.94194818e-01 5.92778511e-02 4.36129093e-01 -9.37577426e-01 5.11694372e-01 -4.67225581e-01 5.27022183e-01 -1.87539682e-02 5.37497282e-01 3.71048227e-02 1.60603538e-01 2.60597616e-02 1.01855709e-04 -3.05265397e-01 9.08317924e-01 -1.33698487e+00 2.33486271e+00 -3.11774731e-01 6.87628508e-01 -9.14708823e-02 -8.31631660e-01 1.15011740e+00 5.29304028e-01 2.04167217e-01 -2.05555350e-01 9.34917033e-02 3.01699519e-01 3.47267777e-01 -2.12778702e-01 8.50632131e-01 -3.95789981e-01 -5.43713808e-01 5.25271893e-01 5.83767354e-01 -3.63983721e-01 -3.59102264e-02 -1.12381965e-01 1.17515826e+00 5.85253417e-01 4.46454644e-01 7.49585107e-02 3.33490431e-01 -2.00901609e-02 6.67677701e-01 7.10606992e-01 -1.22930720e-01 7.50000119e-01 2.58994937e-01 -5.49427092e-01 -1.02007437e+00 -1.10776389e+00 1.88695580e-01 1.54961026e+00 -5.34375012e-01 -1.05296099e+00 -6.59381926e-01 -2.87562221e-01 -3.18075120e-01 3.84092867e-01 -3.98556441e-01 -2.09493056e-01 -4.50981617e-01 6.14731312e-02 1.35389984e+00 7.19002366e-01 1.95556089e-01 -1.67843568e+00 -1.08110094e+00 6.64585471e-01 -5.16415294e-03 -7.68882513e-01 -1.63306430e-01 5.71859300e-01 -5.51608145e-01 -1.01755333e+00 -2.22235084e-01 -8.05903971e-01 -1.89463288e-01 -4.06818688e-01 1.35144246e+00 1.35636069e-02 -5.58216155e-01 7.89241940e-02 -4.98067707e-01 -7.45058596e-01 -3.18091929e-01 4.65608895e-01 1.89416632e-01 -1.81430340e-01 3.24906290e-01 -8.61018956e-01 -5.58358371e-01 -1.42747685e-01 -6.76493824e-01 1.66653603e-01 6.33133829e-01 7.41748989e-01 7.48875856e-01 1.30533939e-02 6.77305996e-01 -4.62557614e-01 7.52122879e-01 1.69168875e-01 -4.00395662e-01 -5.85124902e-02 -4.57286716e-01 2.15912014e-01 7.53173292e-01 -3.53523523e-01 -6.83663309e-01 7.38862455e-01 -5.52691042e-01 -1.56007990e-01 -5.31764328e-01 6.93800926e-01 -1.40088364e-01 1.48318037e-01 7.40085125e-01 2.14247748e-01 -2.65815258e-01 -7.64641047e-01 6.81823790e-01 8.89761269e-01 1.25416434e+00 -6.65605962e-01 6.98365450e-01 1.48491547e-01 -7.21886307e-02 -7.67797828e-01 -8.23095977e-01 -3.22032422e-01 -9.89735484e-01 -3.79186422e-01 8.09483707e-01 -7.40269721e-01 -9.74953055e-01 6.16807103e-01 -9.51925159e-01 -3.03886175e-01 -5.23694277e-01 4.55440223e-01 -1.25595784e+00 1.11053608e-01 -7.72480726e-01 -8.88718486e-01 -6.27106309e-01 -8.10469806e-01 1.07538033e+00 3.01978379e-01 -8.45263660e-01 -5.46292305e-01 4.85533893e-01 2.31895581e-01 1.87258020e-01 3.13080885e-02 6.75306797e-01 -6.73565388e-01 -4.61779743e-01 -1.76328599e-01 2.52930522e-01 3.23683858e-01 6.60905614e-03 2.91183829e-01 -1.55116463e+00 -2.34952997e-02 -4.28094447e-01 -6.87120259e-01 8.44733000e-01 1.21576980e-01 1.14231133e+00 -5.53115457e-02 2.84614682e-01 7.65936196e-01 8.69921505e-01 -1.30601779e-01 4.62174833e-01 5.43298185e-01 3.01289976e-01 3.85998249e-01 5.84433496e-01 6.64751470e-01 2.55244195e-01 5.90028226e-01 4.19086128e-01 2.13906437e-01 -2.32763246e-01 -4.28156316e-01 4.17993695e-01 1.08695400e+00 -3.12564284e-01 2.68676400e-01 -9.97970283e-01 7.43076563e-01 -1.67123771e+00 -1.00242126e+00 1.94589615e-01 1.94392002e+00 1.24376452e+00 4.38581675e-01 4.50037152e-01 9.89625692e-01 1.51055425e-01 -1.13386856e-02 -4.26791281e-01 -5.91878772e-01 6.27149642e-02 1.03714514e+00 -1.65119544e-01 3.92829329e-01 -1.28956890e+00 1.28406084e+00 7.13805819e+00 5.93419790e-01 -1.15500259e+00 -4.45088893e-01 4.58247075e-03 -1.19376503e-01 -3.72612737e-02 3.79243523e-01 -4.59370792e-01 -2.70538807e-01 1.10468757e+00 -1.38044599e-02 5.17443419e-01 8.02405417e-01 6.99961036e-02 2.43428960e-01 -1.20944107e+00 8.76958728e-01 -2.55312711e-01 -1.32937574e+00 -2.20433384e-01 -1.65321589e-01 1.34266233e-02 -1.20844401e-01 6.37307242e-02 4.82383400e-01 2.61578470e-01 -1.25232005e+00 1.14651179e+00 9.08483982e-01 7.35085785e-01 -1.09253156e+00 3.11749011e-01 2.50272065e-01 -1.43741345e+00 -1.18294828e-01 -1.56164810e-01 -7.38601029e-01 2.86076576e-01 5.32077312e-01 -1.16455865e+00 2.03741014e-01 8.00795555e-01 5.11427701e-01 -6.86376393e-01 1.28397071e+00 -6.55148685e-01 1.14412105e+00 -3.27237785e-01 9.04010907e-02 -4.12039645e-02 3.84251684e-01 5.51978767e-01 1.51919806e+00 1.83972165e-01 -1.59785748e-01 1.50256306e-01 6.62289202e-01 1.23868413e-01 2.06404671e-01 -1.24489881e-01 -4.26914304e-01 3.71037990e-01 1.45714200e+00 -8.95040274e-01 -5.13239354e-02 -7.40085244e-02 1.04371953e+00 4.01962012e-01 -2.19039619e-01 -3.66919219e-01 -9.40748215e-01 7.22833574e-01 -1.43944025e-01 6.69414818e-01 -7.01575100e-01 -5.65994382e-01 -8.50792885e-01 -2.73269355e-01 -1.07399392e+00 2.95817345e-01 -6.04510069e-01 -1.12277722e+00 7.27967978e-01 -6.20159388e-01 -1.03908384e+00 -8.23531091e-01 -6.44966304e-01 -7.37836540e-01 7.05646038e-01 -1.09949243e+00 -1.17482936e+00 1.05496086e-01 6.00108743e-01 5.23688257e-01 -5.03406107e-01 1.57258010e+00 -4.72185612e-02 1.03264987e-01 6.78186476e-01 -4.72039372e-01 6.50052071e-01 8.20707977e-01 -1.37745416e+00 9.00576770e-01 6.57118082e-01 1.02981961e+00 5.76041460e-01 7.12150216e-01 -3.41642141e-01 -1.20715463e+00 -5.93187928e-01 8.94453466e-01 -1.49754837e-01 5.48581541e-01 -5.55677712e-01 -7.32408404e-01 3.95560265e-01 2.85289437e-01 -8.68386552e-02 1.44677424e+00 6.42617583e-01 -7.22455442e-01 1.10175759e-01 -5.73357224e-01 3.46528471e-01 9.02879596e-01 -9.46748316e-01 -9.08428311e-01 -2.54755318e-01 3.92094672e-01 -1.41282573e-01 -1.02830625e+00 3.44149321e-01 1.08454311e+00 -9.26006675e-01 8.10102284e-01 -9.05218005e-01 3.89170259e-01 -5.95799387e-01 -3.29879761e-01 -1.20421875e+00 -4.62421507e-01 -8.48895907e-01 -3.63746166e-01 1.08853936e+00 3.87001812e-01 4.27708507e-01 8.42904747e-01 1.23260327e-01 -4.18686926e-01 -4.58413213e-01 -8.97758543e-01 -5.33234715e-01 -1.53388185e-02 -1.03684556e+00 2.93138653e-01 7.92010367e-01 3.71854424e-01 5.93965411e-01 -2.04450876e-01 6.03690231e-03 5.50627410e-01 5.23350596e-01 9.55513060e-01 -1.58844399e+00 -8.87375414e-01 -5.77802658e-01 -8.10064197e-01 -1.02601433e+00 3.55231762e-02 -9.63156879e-01 4.00221825e-01 -1.14753711e+00 -1.33097470e-01 -2.67102510e-01 -5.10901272e-01 8.46387148e-01 -8.09006095e-02 5.51581025e-01 3.40706348e-01 -3.72710312e-03 -5.24029136e-01 5.10108471e-01 8.39670539e-01 4.28717025e-02 -2.04133213e-01 6.33044019e-02 -7.11973190e-01 9.48575377e-01 1.12687337e+00 -4.62458432e-01 -1.08715542e-01 -2.84274161e-01 5.74785113e-01 -6.31178394e-02 3.18124622e-01 -1.53159750e+00 6.30121052e-01 5.84444255e-02 5.76044202e-01 -6.08184218e-01 5.00182152e-01 -1.46464616e-01 -1.71603844e-01 1.86658904e-01 -7.33017802e-01 -5.89346997e-02 2.95769989e-01 2.09166929e-01 -4.93886918e-01 -2.76224613e-01 4.99053210e-01 -1.24272101e-01 -7.28744924e-01 -9.21566561e-02 -7.19656527e-01 5.49307093e-03 3.60516697e-01 -9.79199409e-02 4.25710678e-01 -6.32223785e-01 -1.06208706e+00 -4.62244064e-01 1.06246397e-01 5.42925596e-01 7.42776096e-01 -1.12880158e+00 -8.71435463e-01 3.72732162e-01 6.96006939e-02 2.26122141e-02 4.68198955e-02 2.22141162e-01 -4.84664172e-01 1.50401101e-01 -4.95005757e-01 -4.35136318e-01 -1.29646742e+00 2.48135164e-01 1.70108616e-01 -1.81990281e-01 -6.03887498e-01 1.25051272e+00 -4.53410298e-01 -7.35529661e-01 3.44432950e-01 -2.33851492e-01 -8.64283219e-02 -1.12180607e-02 7.28592575e-01 1.28493607e-02 1.36642978e-01 -5.12048781e-01 -2.72056937e-01 3.85438740e-01 -3.93565930e-02 -6.82046950e-01 1.63385129e+00 4.31784660e-01 1.55465573e-01 8.41101587e-01 5.29204667e-01 3.48324001e-01 -1.23799944e+00 -7.71845877e-02 4.26952958e-01 -2.02204779e-01 2.80196369e-01 -1.00967824e+00 -7.49932885e-01 1.21516645e+00 3.63594979e-01 1.20819055e-01 1.34826982e+00 -2.54189968e-01 9.08142328e-01 6.89580262e-01 2.02737004e-01 -1.18746364e+00 4.93662432e-02 9.38939691e-01 9.25495028e-01 -6.56206131e-01 -1.34639502e-01 4.80487756e-03 -4.55950886e-01 1.48022127e+00 2.16029733e-01 -5.29134631e-01 7.52280533e-01 8.42424035e-01 3.08371723e-01 2.40516867e-02 -9.90877986e-01 -3.63607466e-01 5.10588586e-01 5.68191767e-01 8.92362177e-01 -1.97636634e-02 -3.23133630e-04 1.40873706e+00 -1.23448515e+00 1.95765629e-01 3.79763126e-01 1.02002239e+00 -6.08052313e-01 -1.43389726e+00 -3.02555174e-01 6.57334179e-02 -7.28143871e-01 -2.58979827e-01 -8.10605288e-01 2.94766814e-01 1.83617219e-01 9.60834503e-01 3.47655043e-02 -7.57368863e-01 3.81640792e-01 4.40687031e-01 7.81638443e-01 -5.25588274e-01 -1.08363903e+00 2.58616805e-01 1.63166419e-01 -4.53626901e-01 -2.49150112e-01 -7.47010887e-01 -1.61610532e+00 1.81720287e-01 -2.55212933e-01 2.40557894e-01 7.73528695e-01 7.97826707e-01 3.36663723e-01 7.70873487e-01 3.97595197e-01 -1.18571877e+00 -4.23473805e-01 -1.06459284e+00 -4.90231335e-01 9.32274461e-02 2.48504803e-01 -3.27524871e-01 2.99845133e-02 5.15737414e-01]
[15.831361770629883, 5.314194679260254]
803b72ad-c38b-4b5b-953c-9078bac03782
whether-and-when-does-endoscopy-domain
2303.17636
null
https://arxiv.org/abs/2303.17636v1
https://arxiv.org/pdf/2303.17636v1.pdf
Whether and When does Endoscopy Domain Pretraining Make Sense?
Automated endoscopy video analysis is a challenging task in medical computer vision, with the primary objective of assisting surgeons during procedures. The difficulty arises from the complexity of surgical scenes and the lack of a sufficient amount of annotated data. In recent years, large-scale pretraining has shown great success in natural language processing and computer vision communities. These approaches reduce the need for annotated data, which is always a concern in the medical domain. However, most works on endoscopic video understanding use models pretrained on natural images, creating a domain gap between pretraining and finetuning. In this work, we investigate the need for endoscopy domain-specific pretraining based on downstream objectives. To this end, we first collect Endo700k, the largest publicly available corpus of endoscopic images, extracted from nine public Minimally Invasive Surgery (MIS) datasets. Endo700k comprises more than 700,000 unannotated raw images. Next, we introduce EndoViT, an endoscopy pretrained Vision Transformer (ViT). Through ablations, we demonstrate that domain-specific pretraining is particularly beneficial for more complex downstream tasks, such as Action Triplet Detection, and less effective and even unnecessary for simpler tasks, such as Surgical Phase Recognition. We will release both our code and pretrained models upon acceptance to facilitate further research in this direction.
['Nassir Navab', 'Tobias Czempiel', 'Ege Özsoy', 'Felix Holm', 'Dominik Batić']
2023-03-30
null
null
null
null
['action-triplet-detection', 'video-understanding', 'surgical-phase-recognition']
['computer-vision', 'computer-vision', 'computer-vision']
[ 4.37403917e-01 2.80298442e-01 -3.99613500e-01 -1.95740640e-01 -6.90291464e-01 -8.27258170e-01 2.74339318e-01 2.95454767e-02 -7.03571200e-01 2.55624145e-01 5.53860724e-01 -4.97930944e-01 9.39578488e-02 -2.91886985e-01 -7.50720024e-01 -5.48800766e-01 1.40660509e-01 3.62239145e-02 -2.12021489e-02 -1.04137622e-01 -6.55384287e-02 1.26148477e-01 -1.35663629e+00 7.07071304e-01 7.29870379e-01 8.71248245e-01 3.04784507e-01 6.88566864e-01 3.40032190e-01 7.65265584e-01 -2.57407576e-01 -3.68775904e-01 3.04504812e-01 -4.16715205e-01 -8.83292437e-01 1.51916295e-01 3.07140559e-01 -3.72836202e-01 -4.94287878e-01 1.10812712e+00 4.52458173e-01 -7.03153834e-02 3.84607226e-01 -8.59138608e-01 -4.40004528e-01 6.04326606e-01 -1.83465302e-01 4.62666959e-01 1.01385035e-01 4.38614368e-01 8.75961244e-01 -6.26987934e-01 7.66463280e-01 7.72620738e-01 5.22385180e-01 7.35243857e-01 -8.45136046e-01 -3.47231150e-01 1.07651964e-01 2.30067059e-01 -8.79081011e-01 -3.70943785e-01 6.62474573e-01 -6.70234799e-01 9.43326533e-01 1.81664005e-01 7.69923210e-01 1.25060427e+00 9.36117172e-02 9.11103189e-01 8.12541783e-01 -4.57387626e-01 -5.31934649e-02 7.46042505e-02 1.58503931e-02 1.02209675e+00 2.49536812e-01 5.01631498e-01 -2.95953810e-01 1.40930891e-01 9.64840829e-01 5.90879433e-02 -7.61410177e-01 -6.53013766e-01 -1.57604337e+00 8.77520859e-01 5.65203905e-01 2.76004195e-01 -2.69993216e-01 -2.14020118e-01 5.18437386e-01 4.94620562e-01 -8.38071182e-02 8.04485381e-01 -5.90519786e-01 -3.40292871e-01 -5.88798046e-01 -4.96783018e-01 9.51847970e-01 1.01758552e+00 3.06356102e-01 -2.23063007e-01 1.17814355e-03 7.51857162e-01 1.94623731e-02 -5.01435213e-02 1.01832366e+00 -8.08776259e-01 3.35680723e-01 7.31857359e-01 -2.07744613e-01 -6.72439098e-01 -5.33661246e-01 -3.79410237e-01 -7.72616804e-01 3.40143368e-02 4.07222152e-01 -1.91293880e-01 -1.22744763e+00 1.84078228e+00 2.07394995e-02 1.86894447e-01 2.74120361e-01 1.01517832e+00 1.22440863e+00 1.03534207e-01 3.12100984e-02 -1.64751127e-01 1.57395446e+00 -1.43252170e+00 -4.10679638e-01 -6.04574561e-01 9.58153188e-01 -7.38278449e-01 1.17176235e+00 4.95180786e-01 -7.68154919e-01 -2.38068148e-01 -8.65073442e-01 -2.98889697e-01 -2.90951580e-01 4.06397372e-01 8.09213579e-01 4.73907590e-01 -8.56104612e-01 1.94156229e-01 -1.10313749e+00 -4.43894356e-01 4.33575869e-01 6.04343832e-01 -7.29935348e-01 -2.85437763e-01 -8.92362654e-01 9.00819421e-01 4.36667204e-01 1.35076821e-01 -9.89190638e-01 -7.07871020e-01 -1.21161473e+00 -3.30467708e-02 5.64702749e-01 -1.00037754e+00 1.51811123e+00 -1.05337489e+00 -1.34269202e+00 1.30587113e+00 1.62747219e-01 -5.92807770e-01 4.36335623e-01 -1.54346228e-01 -1.65585309e-01 2.72291988e-01 -2.86788195e-01 7.90237606e-01 7.91812658e-01 -7.77826667e-01 -5.38506567e-01 -1.85604721e-01 5.03737271e-01 2.62099087e-01 -5.11741698e-01 -2.64996141e-01 -8.40189099e-01 -7.53315091e-01 -2.00436860e-01 -1.32765925e+00 -4.90134597e-01 1.44990742e-01 -3.87200564e-01 5.63883111e-02 2.75407344e-01 -6.32758975e-01 1.07155335e+00 -2.49630070e+00 3.34132254e-01 -1.50240943e-01 2.68345237e-01 4.46269721e-01 -2.84612000e-01 5.55238463e-02 -3.18335712e-01 -4.15767804e-02 -1.29391536e-01 -4.67232391e-02 -3.23545963e-01 2.62794048e-01 4.07928489e-02 4.04832721e-01 1.15176141e-02 8.88071358e-01 -1.05784249e+00 -5.60454369e-01 4.41647261e-01 1.86770946e-01 -1.09737766e+00 2.85320282e-01 -1.46333069e-01 6.36413217e-01 -2.36035302e-01 7.79375494e-01 4.64165583e-02 -5.42983770e-01 1.10351600e-01 -7.85382152e-01 8.61849114e-02 2.06385210e-01 -5.80372036e-01 2.32742023e+00 -5.55644214e-01 6.21146798e-01 2.45303139e-01 -1.19855964e+00 1.58177823e-01 4.62722540e-01 8.10740054e-01 -5.84880412e-01 4.01254177e-01 1.32421419e-01 3.74813020e-01 -8.59381139e-01 2.59811640e-01 -2.93876410e-01 -1.34332106e-01 1.16402030e-01 3.33707869e-01 -1.80221677e-01 4.62912560e-01 1.52432108e-02 1.26907539e+00 -1.51418716e-01 6.36627793e-01 -5.73491789e-02 4.38671052e-01 5.74751496e-01 4.65430290e-01 5.85981965e-01 -5.40844977e-01 7.25350976e-01 2.74336249e-01 -4.29009020e-01 -6.91328645e-01 -9.37737167e-01 -1.09444566e-01 8.88071597e-01 2.26785496e-01 -4.67913210e-01 -6.56368494e-01 -8.67496550e-01 -1.65700302e-01 3.38973939e-01 -7.75585949e-01 -5.56691766e-01 -5.76488674e-01 -7.89762497e-01 2.98472345e-01 5.12194276e-01 2.22716421e-01 -1.11340332e+00 -6.14346027e-01 7.77939633e-02 -4.83740211e-01 -1.40210342e+00 -6.99881315e-01 1.52010545e-01 -9.69289899e-01 -1.50574923e+00 -6.75408542e-01 -1.29178262e+00 9.55157697e-01 4.86855626e-01 1.18221474e+00 5.58207324e-03 -7.81796992e-01 7.20419288e-01 -3.66537452e-01 -5.28421402e-01 -5.60944676e-01 1.42964825e-01 -1.80257022e-01 -3.07272136e-01 3.70512128e-01 -2.17957318e-01 -8.69398355e-01 3.65589768e-01 -1.05551231e+00 3.36184025e-01 9.82161164e-01 1.23878157e+00 5.54322600e-01 -2.55022585e-01 -5.98901287e-02 -9.60827827e-01 4.49111104e-01 -2.71646976e-01 -3.76673162e-01 1.75006613e-01 -2.33711585e-01 1.74376089e-03 6.77357614e-01 -5.67135215e-01 -7.59979010e-01 2.57625937e-01 -1.27847746e-01 -7.08378196e-01 -9.92717743e-02 9.24748302e-01 3.45979631e-01 -9.60576162e-02 8.31117332e-01 5.79620786e-02 3.38032335e-01 -2.37808660e-01 2.59618998e-01 4.95121479e-01 7.39916980e-01 -1.58275411e-01 4.92081791e-01 4.86049861e-01 -3.41057241e-01 -8.25882375e-01 -1.21258712e+00 -7.67480910e-01 -3.97621542e-01 1.12052612e-01 8.75128448e-01 -1.15699804e+00 -5.49322188e-01 1.47582799e-01 -7.07485735e-01 -4.57853198e-01 -3.82749051e-01 9.15371716e-01 -7.08315670e-01 3.54925662e-01 -8.16742301e-01 3.92913520e-02 -3.40129197e-01 -1.46758568e+00 8.34330797e-01 1.77012458e-01 -1.87286139e-01 -1.12421000e+00 7.47616291e-02 5.82529187e-01 1.07602596e-01 1.07216708e-01 9.36242461e-01 -5.29614747e-01 -5.31211913e-01 -2.94671059e-01 -1.71020478e-01 4.68836218e-01 4.77336526e-01 -3.75962824e-01 -6.46524310e-01 -5.03283858e-01 1.64958745e-01 -4.81185764e-01 1.01324213e+00 4.57569331e-01 1.34680521e+00 -1.12071887e-01 -4.06682938e-01 9.87106502e-01 1.15451646e+00 -2.93752942e-02 4.34594154e-01 4.58950013e-01 6.60967112e-01 6.11027837e-01 6.96827829e-01 1.41891316e-01 3.08996022e-01 4.38014030e-01 6.51823938e-01 -4.50854808e-01 -2.99002558e-01 -9.23842862e-02 2.36266971e-01 8.93812835e-01 -8.61682072e-02 -1.35386154e-01 -9.24318314e-01 6.52846098e-01 -1.73185861e+00 -7.24255681e-01 4.61708188e-01 2.05996633e+00 1.00576603e+00 -6.27186447e-02 -1.26166955e-01 -3.05047184e-01 5.05432248e-01 -1.82399645e-01 -6.48468852e-01 1.76583361e-02 2.10732371e-01 5.13600819e-02 5.16103745e-01 2.77603745e-01 -1.46448767e+00 8.75960767e-01 5.81526613e+00 3.97403508e-01 -1.37399638e+00 -2.20099911e-02 4.17190909e-01 -2.73914188e-01 2.74644028e-02 -2.63935506e-01 -3.68448734e-01 3.21041822e-01 6.79617167e-01 -2.22346112e-01 3.36984307e-01 9.58255768e-01 2.16191098e-01 9.70458761e-02 -1.49277973e+00 1.35190189e+00 3.02280396e-01 -1.35484791e+00 5.95636740e-02 3.17313038e-02 5.60413361e-01 3.87976497e-01 1.11756705e-01 4.61854011e-01 2.47638479e-01 -1.01371324e+00 2.89970011e-01 1.22699201e-01 9.09829021e-01 -1.63063869e-01 9.02926981e-01 2.25048363e-01 -8.07515562e-01 -2.91598350e-01 -1.33085787e-01 2.66835719e-01 6.19906671e-02 3.81673485e-01 -9.68296409e-01 2.67989516e-01 6.69505179e-01 9.92726624e-01 -4.02567714e-01 1.36667109e+00 -2.06615791e-01 4.56701994e-01 -4.01413381e-01 2.20322251e-01 3.06532443e-01 -4.28561755e-02 4.21078026e-01 1.11347485e+00 2.72931695e-01 2.16179609e-01 3.22881848e-01 2.97702760e-01 -3.05574387e-01 -7.17949793e-02 -6.46777809e-01 -4.55610156e-01 -1.87673867e-01 1.24431765e+00 -6.64118946e-01 -2.80865759e-01 -8.33787560e-01 1.04175687e+00 2.16126308e-01 2.13277981e-01 -8.11402500e-01 -1.57041401e-02 6.48705125e-01 -2.07843050e-01 1.79560900e-01 -1.69798762e-01 -3.49561311e-02 -1.55042911e+00 -4.28878143e-02 -1.20298278e+00 7.69430876e-01 -5.54257333e-01 -1.08757102e+00 6.41786635e-01 -3.38608533e-01 -1.81946194e+00 -4.00193959e-01 -1.12787378e+00 -2.19569385e-01 2.38562927e-01 -1.62467599e+00 -1.07916021e+00 -6.90935552e-01 8.28143656e-01 7.82674611e-01 -7.30335573e-03 9.61682439e-01 4.01902318e-01 -5.36215782e-01 4.90773350e-01 -1.68474853e-01 3.87335449e-01 1.04107225e+00 -1.12726963e+00 -1.32790163e-01 7.79108703e-01 2.57899463e-01 7.48709857e-01 6.70951605e-01 -2.21014172e-01 -1.56023037e+00 -1.06118953e+00 3.71883035e-01 -5.26636899e-01 7.20340073e-01 -6.67938516e-02 -4.65316772e-01 9.11338329e-01 8.62621814e-02 1.17303036e-01 9.48679209e-01 -2.96711028e-02 -2.31282353e-01 1.13529086e-01 -9.27829087e-01 8.30752969e-01 1.19324422e+00 -5.65506637e-01 -8.10584605e-01 6.15620792e-01 6.64699435e-01 -8.23053598e-01 -9.66977656e-01 7.54172564e-01 4.64319080e-01 -8.37119997e-01 1.03296804e+00 -6.97507203e-01 6.04418814e-01 -3.15224193e-02 2.23833233e-01 -1.37121201e+00 -1.43229485e-01 -4.95417953e-01 -1.69852655e-03 4.59046870e-01 5.30992210e-01 -3.36877555e-01 1.00055838e+00 5.66732824e-01 -6.35587037e-01 -6.79550946e-01 -6.28207862e-01 -4.69119877e-01 -4.53637056e-02 -2.58684725e-01 -1.56703621e-01 1.00523090e+00 3.02484244e-01 2.67589092e-01 -1.32273227e-01 1.21322379e-01 1.14827245e-01 3.49549174e-01 7.04472423e-01 -8.47848773e-01 -4.95372534e-01 -4.52886939e-01 -4.58860785e-01 -1.13579905e+00 -1.54636696e-01 -9.78363395e-01 1.95989132e-01 -1.65636182e+00 3.54469091e-01 -1.45836174e-01 -3.97834301e-01 6.84177160e-01 -6.51410818e-02 3.60207111e-01 -3.92571501e-02 2.19259769e-01 -5.50553203e-01 2.92103767e-01 1.58371258e+00 -4.34670329e-01 -3.48037779e-01 8.27482790e-02 -8.36521149e-01 9.65071619e-01 6.29574358e-01 -2.31594220e-01 -6.28355920e-01 -6.90314412e-01 -7.14072511e-02 -2.66348254e-02 2.99300164e-01 -8.58024478e-01 2.93065012e-01 -1.57980416e-02 3.76425274e-02 -2.14355722e-01 2.87013888e-01 -8.91071737e-01 -2.32138917e-01 7.66637206e-01 -2.71448582e-01 -8.93416628e-02 3.87769520e-01 5.70818961e-01 -7.17648923e-01 -1.41031533e-01 7.08011746e-01 -5.35172045e-01 -1.15086269e+00 5.68160355e-01 -2.98417658e-01 2.16137096e-01 1.08053100e+00 -2.55250186e-01 -3.47760230e-01 -1.72869429e-01 -1.12663281e+00 1.81138515e-01 5.31074524e-01 6.91274345e-01 6.53531730e-01 -8.35026681e-01 -4.50104088e-01 2.95181721e-01 5.90652049e-01 2.21753523e-01 4.96429682e-01 1.20064795e+00 -6.62568092e-01 6.00798726e-01 -7.07006752e-02 -7.21234798e-01 -1.30586338e+00 8.53362501e-01 3.78612489e-01 -4.03962493e-01 -8.99973273e-01 9.56960022e-01 8.10876906e-01 -3.56570035e-01 4.04014468e-01 -8.30785990e-01 -3.26209575e-01 -1.86803162e-01 4.37741190e-01 -3.97910297e-01 2.48236969e-01 -2.57946670e-01 -2.74284571e-01 4.40803736e-01 -3.46601009e-01 2.74837255e-01 1.26042771e+00 -3.74647193e-02 1.42280012e-01 1.24676432e-02 1.20315325e+00 -2.31370822e-01 -1.11987972e+00 -1.49138570e-01 -1.72486633e-01 -2.23836675e-01 2.25013420e-01 -8.80399823e-01 -1.14026666e+00 7.73120105e-01 5.59475958e-01 -1.59603596e-01 1.30326486e+00 1.03483468e-01 8.36242914e-01 5.37497163e-01 4.22630399e-01 -8.01260769e-01 2.13572174e-01 4.54741031e-01 8.56195629e-01 -1.60413313e+00 -2.82230020e-01 -7.09536493e-01 -6.67345166e-01 9.77696419e-01 7.39517748e-01 1.30683377e-01 7.45134592e-01 1.80189684e-01 2.57432878e-01 -2.32532397e-01 -6.27890110e-01 -2.87142187e-01 3.94839913e-01 4.52805191e-01 4.46034372e-01 -1.03595853e-01 -1.47983149e-01 7.67200470e-01 -3.69269222e-01 3.46470624e-01 5.86738586e-01 1.10478234e+00 -2.53294166e-02 -9.23723221e-01 5.63214980e-02 6.39552236e-01 -6.04686081e-01 -4.16233242e-01 -1.36503816e-01 7.86669314e-01 -3.06532979e-02 7.67069459e-01 -8.03727955e-02 -1.59969106e-01 3.86682957e-01 -3.89013201e-01 6.79108679e-01 -9.42265391e-01 -4.58537787e-01 3.28041464e-02 3.30589026e-01 -5.96803963e-01 -6.28672421e-01 -3.44923854e-01 -1.08287156e+00 2.66901523e-01 -2.38176182e-01 8.02021623e-02 4.85648394e-01 8.58621180e-01 3.69528413e-01 7.37498581e-01 1.98581070e-01 -6.31415546e-01 -4.56593394e-01 -6.88216507e-01 -1.74207628e-01 7.12602377e-01 5.63111544e-01 -6.22747123e-01 -4.57036674e-01 5.00483453e-01]
[14.250507354736328, -3.159609317779541]
0643d712-7cdd-46e7-9c8b-01850a291397
lagnet-logic-aware-graph-network-for-human
2011.1025
null
https://arxiv.org/abs/2011.10250v3
https://arxiv.org/pdf/2011.10250v3.pdf
Consistency-Aware Graph Network for Human Interaction Understanding
Compared with the progress made on human activity classification, much less success has been achieved on human interaction understanding (HIU). Apart from the latter task is much more challenging, the main cause is that recent approaches learn human interactive relations via shallow graphical models, which is inadequate to model complicated human interactions. In this paper, we propose a consistency-aware graph network, which combines the representative ability of graph network and the consistency-aware reasoning to facilitate the HIU task. Our network consists of three components, a backbone CNN to extract image features, a factor graph network to learn third-order interactive relations among participants, and a consistency-aware reasoning module to enforce labeling and grouping consistencies. Our key observation is that the consistency-aware-reasoning bias for HIU can be embedded into an energy function, minimizing which delivers consistent predictions. An efficient mean-field inference algorithm is proposed, such that all modules of our network could be trained jointly in an end-to-end manner. Experimental results show that our approach achieves leading performance on three benchmarks.
['ShengYong Chen', 'Javen Qinfeng Shi', 'Jianhua Zhang', 'Dongyan Guo', 'Jiajun Meng', 'Zhenhua Wang']
2020-11-20
null
http://openaccess.thecvf.com//content/ICCV2021/html/Wang_Consistency-Aware_Graph_Network_for_Human_Interaction_Understanding_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Wang_Consistency-Aware_Graph_Network_for_Human_Interaction_Understanding_ICCV_2021_paper.pdf
iccv-2021-1
['3d-human-pose-and-shape-estimation']
['computer-vision']
[ 2.69125760e-01 7.01662481e-01 -3.36960196e-01 -5.41701853e-01 -9.52306613e-02 -7.53375366e-02 5.87067366e-01 2.99706552e-02 -8.12702999e-02 4.85446364e-01 3.47467542e-01 -2.91034192e-01 -2.54062027e-01 -7.32496858e-01 -8.57107699e-01 -1.61619455e-01 -1.37420237e-01 5.96775472e-01 1.34897828e-01 1.13448547e-02 -9.55814272e-02 -5.00157401e-02 -1.24291062e+00 3.32206756e-01 1.03192675e+00 9.16200042e-01 -6.13198467e-02 5.83986282e-01 1.97291568e-01 1.43195903e+00 -1.30918309e-01 -5.70103765e-01 3.04601230e-02 -7.87340224e-01 -1.12236774e+00 1.57083496e-01 2.49150515e-01 -3.57695520e-01 -6.72280192e-01 8.01007986e-01 2.30306387e-01 3.55879009e-01 7.12168872e-01 -1.55018759e+00 -8.23661923e-01 7.99096286e-01 -4.82651681e-01 -1.18672788e-01 6.88550830e-01 1.87009543e-01 1.34146333e+00 -4.34428006e-01 6.63392961e-01 1.43029511e+00 5.26445091e-01 4.33881521e-01 -1.27125299e+00 -5.50309718e-01 5.52666187e-01 3.49290609e-01 -1.22337258e+00 -5.73709421e-02 9.86117840e-01 -5.05493939e-01 1.06351590e+00 1.81664452e-01 1.07266045e+00 1.26001799e+00 1.40166342e-01 9.97169495e-01 7.96965420e-01 -3.21742177e-01 1.04343109e-01 -8.53747278e-02 4.07723844e-01 1.16584480e+00 2.44698361e-01 -2.51072258e-01 -7.94436336e-01 -9.90825798e-03 9.98274505e-01 2.25049987e-01 -3.72556657e-01 -6.51209831e-01 -1.12968302e+00 6.08659625e-01 7.82132447e-01 1.31175295e-01 -1.75807714e-01 3.21855217e-01 1.78364441e-01 2.13510960e-01 3.89555961e-01 2.78304249e-01 -3.25183928e-01 1.00448998e-02 -3.56389642e-01 4.70157117e-01 1.09738076e+00 9.43269789e-01 7.77910054e-01 -5.86658537e-01 -4.79854733e-01 5.57992101e-01 5.51402628e-01 -4.45482368e-03 -6.76127672e-02 -1.05008483e+00 5.69495559e-01 1.20637321e+00 -8.35499316e-02 -1.28433061e+00 -5.94325542e-01 -3.70600820e-01 -1.05613077e+00 -1.56332195e-01 3.38580221e-01 1.72591042e-02 -4.70146298e-01 1.96137738e+00 3.02576900e-01 2.40237474e-01 -4.41864192e-01 9.30431783e-01 6.65335000e-01 2.47351095e-01 1.27388269e-01 1.25869244e-01 1.35299909e+00 -1.31437361e+00 -8.75891149e-01 -3.43650401e-01 7.88482785e-01 -1.88355893e-01 1.01484978e+00 2.27569938e-01 -1.14791918e+00 -5.71053088e-01 -1.03448498e+00 -4.57087606e-01 -2.11297482e-01 -6.42896723e-03 1.20720935e+00 2.37330690e-01 -1.01015878e+00 6.07677579e-01 -8.35805953e-01 -4.27229613e-01 6.31008387e-01 3.92775983e-01 -4.08689737e-01 6.16230071e-02 -1.25499773e+00 8.17541897e-01 3.97664100e-01 3.10010791e-01 -5.28081656e-01 -7.18159556e-01 -1.14284396e+00 3.24164450e-01 7.66593575e-01 -1.26830041e+00 1.12532020e+00 -7.86625564e-01 -1.46144569e+00 7.00569808e-01 -1.95200503e-01 -4.33930367e-01 7.38911152e-01 -2.23603770e-01 8.51574987e-02 -5.42281568e-02 -6.84645772e-02 5.03800154e-01 3.63172561e-01 -1.07819963e+00 -3.94228220e-01 -4.06664282e-01 2.58131146e-01 3.60931516e-01 -2.52431691e-01 -2.06657127e-01 -8.13371420e-01 -4.47431505e-01 -1.17026545e-01 -1.04937375e+00 -2.31738403e-01 6.16637878e-02 -6.21317327e-01 -8.00150037e-01 5.18176615e-01 -6.91012681e-01 1.59833050e+00 -1.72341204e+00 5.05639136e-01 4.65556204e-01 7.08583593e-01 -7.31517971e-02 -4.05845828e-02 4.44354504e-01 -5.31556010e-02 9.27808434e-02 -1.54064864e-01 -3.76589984e-01 3.09833378e-01 1.21422350e-01 -4.08300152e-03 3.13403070e-01 1.79577798e-01 1.54779541e+00 -1.09403527e+00 -5.11152864e-01 2.21846268e-01 4.91116434e-01 -8.48202527e-01 5.91230690e-01 -3.92354071e-01 6.03614271e-01 -5.28884768e-01 3.79674286e-01 3.90268326e-01 -8.19423318e-01 7.26011455e-01 -2.52629668e-01 3.48226875e-01 4.30540562e-01 -1.12786365e+00 2.05267143e+00 -2.75223076e-01 4.03775841e-01 -7.11670294e-02 -1.10718203e+00 5.42149186e-01 1.44349545e-01 5.31953871e-01 -5.71420193e-01 2.74626434e-01 -3.34254652e-01 8.36712271e-02 -5.40179968e-01 3.43500376e-02 6.08818114e-01 -7.51465037e-02 5.10115147e-01 1.92980126e-01 2.06738025e-01 1.96347818e-01 5.53781569e-01 1.35520172e+00 3.14854473e-01 3.05793464e-01 -2.04615057e-01 4.31249142e-01 -4.78000313e-01 5.45993924e-01 6.96998715e-01 -2.04068065e-01 2.18183264e-01 7.96303034e-01 -5.14302194e-01 -5.53713977e-01 -9.47798729e-01 4.57344800e-01 1.00296378e+00 3.12533915e-01 -8.68179917e-01 -9.64899004e-01 -9.25375581e-01 -1.44210100e-01 3.04364324e-01 -8.23760569e-01 -3.37763369e-01 -6.78976357e-01 -3.11452538e-01 1.49316728e-01 7.30602920e-01 7.43223131e-01 -9.71501052e-01 -2.16369092e-01 -3.76659222e-02 -4.10748154e-01 -1.14657462e+00 -6.72471941e-01 -2.13413276e-02 -5.10327458e-01 -1.43340743e+00 -1.20174371e-01 -6.99347079e-01 7.77922153e-01 2.95624882e-01 1.33688366e+00 5.17062724e-01 -1.87619954e-01 5.13692200e-01 -1.16828658e-01 -3.77800107e-01 1.06277160e-01 2.16644287e-01 -2.71551222e-01 1.50022924e-01 3.42034549e-01 -8.26184988e-01 -8.03122461e-01 3.45671028e-01 -6.17084265e-01 4.91668195e-01 5.28840125e-01 7.78526902e-01 3.25962305e-01 5.41508719e-02 2.02517346e-01 -1.24159074e+00 6.12850428e-01 -2.73437709e-01 -2.19993845e-01 5.50475061e-01 -6.74843132e-01 1.20898671e-01 3.92975092e-01 -2.54613757e-01 -1.13587582e+00 1.97714016e-01 1.22727707e-01 -2.86590636e-01 -8.49232525e-02 5.06358385e-01 -5.92268884e-01 7.45807821e-03 4.56276298e-01 -1.22053236e-01 -1.18668787e-01 -2.10993782e-01 5.34090996e-01 2.69658476e-01 4.09183085e-01 -6.67531431e-01 7.07921565e-01 3.04278523e-01 4.87624928e-02 -6.55268371e-01 -1.18196034e+00 -3.44291210e-01 -8.01766574e-01 -3.16179514e-01 1.26242363e+00 -9.51147556e-01 -1.38493514e+00 4.06122833e-01 -1.31632006e+00 -6.12724006e-01 2.83734333e-02 2.36274123e-01 -5.55225670e-01 4.15223837e-01 -6.55564189e-01 -8.28252673e-01 -6.21538125e-02 -9.39090371e-01 9.69733834e-01 2.61835396e-01 -5.43518662e-01 -1.19822371e+00 2.08684430e-02 1.03079557e+00 8.37383643e-02 2.09708452e-01 9.32728887e-01 -5.02608478e-01 -9.93878663e-01 -1.06253803e-01 -3.72311354e-01 -3.89744937e-02 -1.02610834e-01 -3.25093299e-01 -8.16181421e-01 -6.39477149e-02 -3.35784405e-01 -5.80999851e-01 7.84689724e-01 2.45744497e-01 1.56510413e+00 -3.21977735e-01 -6.14019573e-01 5.06837249e-01 7.46770620e-01 -8.05021077e-03 6.62807226e-01 -1.49150267e-01 1.26592410e+00 6.09774768e-01 2.64246285e-01 2.01186299e-01 1.08362162e+00 8.28410745e-01 2.93729722e-01 -2.21788481e-01 -1.19938076e-01 -7.40638137e-01 6.64677694e-02 6.30880058e-01 -4.30231273e-01 -3.35703969e-01 -8.10202241e-01 1.94607139e-01 -2.61431026e+00 -1.03435075e+00 -1.42659590e-01 1.81133115e+00 7.16451943e-01 1.79382935e-01 1.78467810e-01 -1.31844552e-02 3.45046133e-01 2.84360319e-01 -4.46182907e-01 4.79904488e-02 3.09219986e-01 1.30388625e-02 -4.94832098e-02 4.77842569e-01 -1.02811658e+00 8.78346145e-01 5.78557682e+00 5.82583606e-01 -4.72844601e-01 -2.07478981e-02 6.40518069e-01 3.05214286e-01 -2.81394929e-01 1.11794889e-01 -4.92592722e-01 2.10859179e-01 5.23657620e-01 1.85071245e-01 6.19916797e-01 6.46149099e-01 2.31683940e-01 -2.85902452e-02 -1.56453907e+00 1.09001410e+00 1.76131651e-02 -1.17469275e+00 1.73479393e-02 2.08991796e-01 6.69307053e-01 -5.37097573e-01 -2.65244573e-01 4.03182954e-01 6.54792190e-01 -1.07834458e+00 5.87025821e-01 8.70639324e-01 2.68445790e-01 -6.70016289e-01 5.14454126e-01 6.04321122e-01 -1.49789858e+00 9.84494165e-02 -2.02383194e-02 -5.24560392e-01 4.05382849e-02 6.28710210e-01 -5.46198666e-01 7.45389402e-01 6.54303908e-01 9.93079662e-01 -5.91002703e-01 6.02601528e-01 -7.12168396e-01 4.77980852e-01 4.30736952e-02 -1.32540762e-01 -8.36504251e-02 -4.49961752e-01 1.18799746e-01 1.06618369e+00 -2.73375005e-01 2.40083024e-01 7.43612051e-01 1.13000786e+00 -3.59341294e-01 -1.01581894e-01 -6.76431715e-01 -2.49248758e-01 2.08250627e-01 1.32569110e+00 -6.28498733e-01 -3.10033292e-01 -6.25706077e-01 1.27689779e+00 1.12119973e+00 4.42979783e-01 -1.00316763e+00 -2.63426244e-01 4.87121731e-01 1.84389800e-01 2.29923055e-02 -2.94709146e-01 -3.75759810e-01 -1.50702536e+00 2.41798937e-01 -7.96694338e-01 5.76300323e-01 -5.94810963e-01 -1.36784625e+00 2.45180294e-01 -5.25057912e-02 -6.59467995e-01 -1.84921682e-01 -5.41859269e-01 -7.83685446e-01 7.49088228e-01 -1.16461170e+00 -1.56181431e+00 -5.89022398e-01 5.81655741e-01 2.31588796e-01 2.33479768e-01 6.50680840e-01 2.50848114e-01 -9.80106413e-01 5.72734654e-01 -7.75628567e-01 4.41395462e-01 4.30181712e-01 -1.40776896e+00 2.78743267e-01 7.20681369e-01 2.61179656e-01 8.98043215e-01 3.92080188e-01 -7.35418200e-01 -1.39050233e+00 -9.88357127e-01 1.11540663e+00 -7.49003947e-01 7.60417819e-01 -7.96470225e-01 -9.54150140e-01 1.09199250e+00 3.58035415e-01 1.09372862e-01 6.64019287e-01 6.19484067e-01 -5.42412221e-01 1.48293778e-01 -4.62696403e-01 8.56723845e-01 1.99143755e+00 -6.95252895e-01 -3.67277831e-01 4.92638201e-01 6.21945500e-01 -5.82745969e-01 -8.30357909e-01 3.38221937e-01 4.92109835e-01 -9.52783644e-01 9.93088186e-01 -8.28467846e-01 5.28072238e-01 -2.58594036e-01 3.23309124e-01 -1.10169637e+00 -6.73321545e-01 -6.83567822e-01 -5.59495330e-01 1.07680464e+00 4.28104818e-01 -3.82573098e-01 7.99166977e-01 1.08373427e+00 -7.68071413e-02 -9.27121699e-01 -4.02009338e-01 -4.96664733e-01 -4.64379728e-01 -3.54167223e-01 4.52188522e-01 1.01478267e+00 4.14731830e-01 1.00148857e+00 -7.17606306e-01 -5.88727258e-02 6.08467937e-01 1.49245843e-01 1.10865283e+00 -1.49362433e+00 -7.58293033e-01 -4.02347535e-01 -2.09198877e-01 -1.33056426e+00 6.39499962e-01 -1.02976418e+00 -1.18288002e-03 -1.92608738e+00 6.39009714e-01 -1.85746569e-02 -1.60405874e-01 6.01409912e-01 -5.10693371e-01 -2.17602313e-01 5.75993098e-02 4.10982221e-02 -1.25516760e+00 6.14022374e-01 1.60187280e+00 -2.76782632e-01 -3.44257772e-01 -7.03323400e-03 -7.74971008e-01 8.92630816e-01 4.47800964e-01 -5.57831563e-02 -8.19934070e-01 -3.39605659e-01 4.57693994e-01 5.00200093e-02 7.28054821e-01 -8.74685049e-01 4.49079186e-01 -1.79868370e-01 4.41288501e-01 -2.89659381e-01 3.33568789e-02 -8.53321373e-01 1.79191515e-01 2.93884158e-01 -6.22468829e-01 -2.43849173e-01 -4.07467872e-01 8.34251761e-01 -1.04801305e-01 4.12556499e-01 3.39868963e-01 2.43743397e-02 -6.14214540e-01 5.59562802e-01 -6.70422688e-02 -1.73844118e-02 8.60614777e-01 1.38935670e-01 -2.71485299e-01 -5.86835563e-01 -6.80303633e-01 5.88176489e-01 1.41346857e-01 4.57992226e-01 2.81571954e-01 -1.39554870e+00 -3.69175851e-01 1.22119682e-02 1.77906245e-01 2.94413865e-01 3.98751110e-01 9.48470056e-01 -1.95589095e-01 4.55754906e-01 2.96169668e-02 -5.02190053e-01 -1.04011381e+00 4.65932131e-01 4.08257812e-01 -8.72316062e-01 -5.88908315e-01 8.06207895e-01 4.01861310e-01 -5.40346444e-01 5.72154939e-01 -4.04518515e-01 -1.02414928e-01 -3.21980417e-01 4.83743668e-01 3.78397465e-01 -1.64875209e-01 -3.41626644e-01 -3.05360734e-01 2.57331461e-01 -1.68051757e-02 1.82043970e-01 1.12195253e+00 -5.06415125e-03 -1.18032619e-01 3.92007321e-01 1.01783490e+00 -4.64827776e-01 -1.32913637e+00 -3.60283047e-01 6.23788238e-02 -2.48136640e-01 -1.54809490e-01 -7.57031798e-01 -9.12093818e-01 7.79146671e-01 2.64925160e-03 3.09293628e-01 9.37862277e-01 1.51050285e-01 7.33875334e-01 5.54285824e-01 4.25553560e-01 -1.00011051e+00 3.60022843e-01 4.77067262e-01 9.02814746e-01 -1.48149359e+00 -7.43554905e-03 -9.28371727e-01 -5.67142785e-01 7.18593121e-01 1.04896605e+00 -5.39185926e-02 7.88526773e-01 -7.08833039e-02 -3.72292817e-01 -5.34077823e-01 -8.49454582e-01 -1.17888317e-01 7.78124273e-01 5.52683353e-01 7.22329497e-01 9.20916125e-02 -3.15571547e-01 9.37619567e-01 -1.39075806e-02 2.34212250e-01 -1.20947763e-01 7.39724517e-01 -1.00728378e-01 -1.04367614e+00 2.45941088e-01 4.16840225e-01 -2.84645781e-02 2.77272075e-01 -7.85410762e-01 7.55416989e-01 1.66092277e-01 1.01763546e+00 -2.82777343e-02 -5.78070343e-01 5.46186507e-01 7.28239194e-02 6.50881350e-01 -5.61758816e-01 -6.03682995e-01 -2.60047078e-01 2.04469338e-01 -1.19138730e+00 -6.57025337e-01 -2.71197796e-01 -1.33712828e+00 -4.13189918e-01 -1.27579868e-01 -1.48351714e-02 9.38649476e-02 1.22983909e+00 4.28541332e-01 8.06813776e-01 2.93940783e-01 -8.30474973e-01 -7.22791329e-02 -8.79644990e-01 -5.56134582e-01 7.58025289e-01 -1.68836743e-01 -7.32974946e-01 -1.54932424e-01 1.20289534e-01]
[8.186402320861816, 0.6635504364967346]
92e30b4e-bfba-4a9c-9ee3-cc1cbe5b32c4
o-type-stars-stellar-parameter-estimation
2210.12791
null
https://arxiv.org/abs/2210.12791v2
https://arxiv.org/pdf/2210.12791v2.pdf
O-type Stars Stellar Parameter Estimation Using Recurrent Neural Networks
In this paper, we present a deep learning system approach to estimating luminosity, effective temperature, and surface gravity of O-type stars using the optical region of the stellar spectra. In previous work, we compare a set of machine learning and deep learning algorithms in order to establish a reliable way to fit a stellar model using two methods: the classification of the stellar spectra models and the estimation of the physical parameters in a regression-type task. Here we present the process to estimate individual physical parameters from an artificial neural network perspective with the capacity to handle stellar spectra with a low signal-to-noise ratio (S/N), in the $<$20 S/N boundaries. The development of three different recurrent neural network systems, the training process using stellar spectra models, the test over nine different observed stellar spectra, and the comparison with estimations in previous works are presented. Additionally, characterization methods for stellar spectra in order to reduce the dimensionality of the input data for the system and optimize the computational resources are discussed.
['Silvana G. Navarro', 'Celia R. Fierro-Santillán', 'Luis J. Corral', 'Miguel Flores R.']
2022-10-23
null
null
null
null
['type']
['speech']
[-9.63154882e-02 -2.51265734e-01 3.72475147e-01 -1.99194565e-01 -1.51418701e-01 -2.04393774e-01 4.38753963e-01 -1.30343392e-01 -3.50303084e-01 5.01335263e-01 -5.44668734e-01 -3.40964258e-01 -4.27866936e-01 -6.52863145e-01 -3.13149065e-01 -1.06099176e+00 3.23289514e-01 6.48141623e-01 -5.78391850e-02 -1.35895804e-01 -1.19746197e-02 9.25820768e-01 -2.01646423e+00 -3.25769037e-01 6.41297996e-01 1.32036138e+00 2.54163861e-01 1.05306756e+00 -1.48011580e-01 7.70812869e-01 -6.57760918e-01 8.87583047e-02 5.57962477e-01 -4.63161498e-01 -5.17856300e-01 3.55728924e-01 6.42354965e-01 3.64115126e-02 -3.89825433e-01 1.01000118e+00 6.94691718e-01 1.04467124e-01 9.67723966e-01 -7.19115794e-01 -3.17996204e-01 5.63645840e-01 -1.19254120e-01 5.19084465e-03 -5.85946441e-01 6.15362167e-01 6.45723045e-01 -6.00684106e-01 1.72517285e-01 8.08832586e-01 8.88081074e-01 3.14141691e-01 -1.18725502e+00 -1.22306399e-01 -6.25524282e-01 3.27778965e-01 -1.32540536e+00 -3.05770487e-01 8.42002749e-01 -8.14643145e-01 1.19907832e+00 -8.66422355e-02 8.08249772e-01 4.15280998e-01 -2.68807203e-01 -1.01540335e-01 1.13675690e+00 -9.01106834e-01 2.44073406e-01 5.06335497e-01 3.75339031e-01 6.44071996e-01 3.79194558e-01 3.25339764e-01 -4.94164079e-01 2.98175123e-02 6.00408018e-01 -6.37638569e-01 -1.05219007e-01 -2.52701700e-01 -8.96951616e-01 8.11630487e-01 1.70321643e-01 3.52436483e-01 -5.42415798e-01 1.70181170e-01 4.35619444e-01 3.51344496e-01 3.89609724e-01 6.10467136e-01 -7.95716226e-01 2.03854650e-01 -7.25658417e-01 2.02454463e-01 9.21500981e-01 2.93425202e-01 8.18090975e-01 7.29606807e-01 2.84747899e-01 9.73999441e-01 6.34171247e-01 9.54531550e-01 8.50845993e-01 -8.53497565e-01 -1.21861413e-01 4.99594271e-01 -4.64827456e-02 -1.66823864e-01 -5.86235404e-01 -6.57509267e-01 -5.76156795e-01 6.58266902e-01 7.43879378e-01 -3.05971295e-01 -6.54874206e-01 1.31089199e+00 3.64082068e-01 -2.36760765e-01 4.74178046e-01 8.38146091e-01 9.98123407e-01 6.11222148e-01 -3.51086497e-01 -3.27237844e-01 1.02609837e+00 -8.03516269e-01 -2.45194301e-01 -2.23310664e-02 4.83639538e-01 -7.91153371e-01 7.46297419e-01 5.84323108e-01 -9.42286789e-01 -9.10669386e-01 -1.13827741e+00 5.26481085e-02 -5.63107073e-01 6.67843461e-01 2.83954889e-01 7.50259757e-01 -9.75818634e-01 9.70484138e-01 -4.15287822e-01 -2.57360369e-01 -4.76055324e-01 2.81012148e-01 7.12066516e-02 1.08125854e+00 -9.97732341e-01 1.09972560e+00 7.64143765e-01 1.70239255e-01 -8.10050249e-01 -4.75604981e-01 -4.37829107e-01 2.56667912e-01 -2.80504208e-02 -6.16165221e-01 1.45618987e+00 -1.18392527e+00 -1.90188885e+00 1.08805990e+00 1.51586428e-01 -7.72054136e-01 1.03186056e-01 2.01215759e-01 -3.72937232e-01 5.90727963e-02 -7.25174308e-01 2.65407562e-01 1.03793621e+00 -7.79742718e-01 -2.34370947e-01 -1.62112221e-01 -5.07618129e-01 -5.14940619e-02 -2.20558062e-01 -5.42629436e-02 1.57230482e-01 1.59133226e-01 2.24807672e-02 -8.26613843e-01 1.20650135e-01 -2.55850583e-01 -7.21060559e-02 -4.93923813e-01 6.12487018e-01 -5.99216044e-01 5.36797881e-01 -1.99618423e+00 2.96397060e-01 2.14064732e-01 4.62610200e-02 4.85248685e-01 2.03637138e-01 2.05181032e-01 -4.70080942e-01 -3.49054873e-01 -1.97418183e-01 -1.21308483e-01 5.16831093e-02 -1.01517595e-01 -4.42727543e-02 5.20657599e-01 -1.75572619e-01 4.18038040e-01 -8.61029625e-02 -1.73909456e-01 5.25893152e-01 4.57341552e-01 1.87959909e-01 4.66854602e-01 -4.01962340e-01 -5.69417775e-02 1.45422397e-02 1.68120191e-01 4.97706711e-01 2.78225821e-02 -1.06916085e-01 -1.55103117e-01 -5.23825765e-01 2.70183891e-01 -1.07424676e+00 1.03056324e+00 -5.02787173e-01 8.57718110e-01 2.91358322e-01 -1.11036348e+00 1.63491940e+00 3.47265005e-01 5.53841650e-01 -4.54434276e-01 5.39993167e-01 4.68986988e-01 3.16793382e-01 -6.78559005e-01 3.84527057e-01 -4.42846894e-01 6.65741503e-01 4.03809398e-01 4.13573146e-01 -5.49773872e-01 2.42484421e-01 -4.93781924e-01 2.59277970e-01 3.58990192e-01 -2.09786873e-02 -4.48533654e-01 9.63100076e-01 -2.06683189e-01 1.92999229e-01 4.69915539e-01 -1.03347018e-01 4.22514826e-01 3.95280778e-01 -5.96945226e-01 -1.73515391e+00 -5.15598595e-01 -3.67418230e-01 7.32918739e-01 -4.15855974e-01 7.85233825e-02 -1.04122269e+00 -2.29936708e-02 1.13655537e-01 5.60511351e-01 -2.48528227e-01 -2.01139793e-01 -1.13857508e-01 -1.10950673e+00 5.13445854e-01 8.67265165e-02 2.48650923e-01 -1.32680333e+00 -9.09111798e-01 -1.31284529e-02 4.15502042e-01 -8.12228322e-01 3.36045474e-01 6.84123099e-01 -8.71473074e-01 -9.79056001e-01 -3.86508852e-01 -4.38930988e-01 5.16890362e-02 -1.46609068e-01 1.10394084e+00 4.75804023e-02 -3.96922797e-01 9.57192779e-02 -1.52247787e-01 -9.33345795e-01 -9.43299055e-01 1.77795485e-01 3.56943935e-01 -5.75428754e-02 6.18544519e-01 -3.34553748e-01 -2.01671496e-01 2.60465108e-02 -6.80648506e-01 -7.00185671e-02 4.61342812e-01 7.32986867e-01 2.54399568e-01 -2.75904220e-02 2.95330793e-01 -1.34441808e-01 2.98661590e-01 -1.24791585e-01 -1.22408724e+00 3.38166803e-01 -1.07807124e+00 6.19009554e-01 8.20476413e-01 -3.89577627e-01 -9.54885244e-01 2.13266194e-01 -4.17329580e-01 -5.61849833e-01 -3.91356081e-01 1.95162803e-01 9.49810967e-02 -1.97273552e-01 1.10280609e+00 1.98365971e-01 4.69497234e-01 -7.47699261e-01 1.81303114e-01 9.50677693e-01 5.93951702e-01 -4.62306738e-01 8.74613822e-01 -2.61573233e-02 3.95830244e-01 -1.06829464e+00 -6.66252851e-01 -6.03373468e-01 -1.00218928e+00 -6.29537165e-01 6.58784568e-01 -7.50806689e-01 -9.24791276e-01 9.02268052e-01 -9.72787023e-01 -3.46502453e-01 -6.82801902e-01 7.94548392e-01 -8.02392304e-01 5.00312030e-01 -5.03008425e-01 -1.18011951e+00 -8.62467408e-01 -1.16326857e+00 7.45555401e-01 6.35224402e-01 3.66654992e-01 -1.08152580e+00 1.65254027e-01 2.65227526e-01 3.33562315e-01 -9.90606472e-02 7.67801583e-01 -4.60043192e-01 -1.93845063e-01 6.86059222e-02 -1.74852923e-01 7.98352063e-01 -3.03040743e-01 4.48328078e-01 -1.36694455e+00 -1.77148253e-01 5.43285251e-01 -5.09035408e-01 1.00898933e+00 5.77952087e-01 9.80738699e-01 3.28392297e-01 4.92402911e-01 7.91207433e-01 1.57290733e+00 1.92939579e-01 4.70566064e-01 4.15877342e-01 4.09852237e-01 6.68743312e-01 2.56676793e-01 5.49125433e-01 -4.42879140e-01 5.36143839e-01 6.16485298e-01 -1.39608935e-01 -2.92090476e-01 2.05758035e-01 5.36303401e-01 7.67213106e-01 -1.59133926e-01 9.59311500e-02 -7.42787361e-01 3.26561332e-01 -1.62312281e+00 -9.32885885e-01 -6.98860943e-01 2.44342351e+00 5.11533499e-01 1.94467127e-01 4.52777088e-01 3.13360900e-01 6.27185106e-01 -8.47981349e-02 -7.01855600e-01 -8.23592484e-01 -2.18088031e-01 2.85536557e-01 8.38641346e-01 5.90035439e-01 -7.30799675e-01 5.58860064e-01 7.27992439e+00 6.11470044e-01 -1.56070852e+00 -6.00743555e-02 3.43764812e-01 -9.76005569e-02 -2.20644698e-02 -1.04627199e-01 -9.31222320e-01 1.73044488e-01 1.56438839e+00 2.23344103e-01 6.40551805e-01 9.41590846e-01 4.33704942e-01 -2.07094029e-01 -5.92900574e-01 1.17043447e+00 1.28048643e-01 -1.01468253e+00 -3.82298678e-01 8.31381828e-02 2.74994940e-01 5.47843218e-01 -2.32305408e-01 3.73936146e-02 -2.07713097e-01 -7.45597899e-01 8.96000326e-01 1.16045284e+00 7.04264700e-01 -6.35352910e-01 8.14590394e-01 3.75624061e-01 -8.25104356e-01 -7.12010190e-02 -8.90544176e-01 -2.93438971e-01 -2.24806905e-01 8.47553372e-01 -1.00242269e+00 6.95377350e-01 7.37358868e-01 2.87897319e-01 -7.69910276e-01 1.02336645e+00 4.08916138e-02 7.48723924e-01 -6.66071057e-01 -4.92948472e-01 1.52943656e-01 -8.09003890e-01 3.56228679e-01 9.55158055e-01 6.04670167e-01 -5.18719733e-01 -3.45319510e-01 1.31309021e+00 2.37082452e-01 2.19046429e-01 -4.14013386e-01 -2.44670108e-01 -8.64990428e-02 1.63064563e+00 -3.88241619e-01 -4.11673337e-01 -2.41221339e-01 1.80188954e-01 2.79651850e-01 1.13948837e-01 -6.58184052e-01 -3.02868605e-01 5.18456161e-01 4.93655056e-02 2.76650310e-01 -3.18517208e-01 -5.10942459e-01 -9.53559756e-01 -2.18520939e-01 -6.24269187e-01 8.20721015e-02 -1.15838397e+00 -1.03558171e+00 5.00192344e-01 -1.62544698e-01 -7.77611494e-01 -4.87265199e-01 -1.45371032e+00 -6.60774469e-01 1.49076021e+00 -1.28273118e+00 -7.52318978e-01 -4.00260717e-01 1.58009827e-02 3.52867693e-01 -8.52380395e-01 6.08395696e-01 -1.60135776e-01 -7.19406545e-01 -2.60511860e-02 7.51360774e-01 -3.04786623e-01 4.57577258e-01 -1.52157724e+00 2.55090743e-01 7.68360734e-01 6.11976348e-02 -1.58958539e-01 1.07426584e+00 -1.75691113e-01 -1.14892578e+00 -6.07367516e-01 6.01012945e-01 -7.03413412e-02 6.80299938e-01 -4.48567085e-02 -9.85244870e-01 1.46208614e-01 4.92294341e-01 -2.35525399e-01 4.13031936e-01 1.41907083e-02 -2.56337017e-01 -3.91985893e-01 -9.25560534e-01 -4.64595892e-02 3.44138235e-01 -6.86302602e-01 -5.69141626e-01 3.18063468e-01 2.48371422e-01 1.60621367e-02 -1.13406074e+00 3.88916075e-01 4.92929757e-01 -1.28764057e+00 8.68426740e-01 -2.72149682e-01 -2.90584937e-02 -5.32621682e-01 3.43433022e-01 -1.26480651e+00 -3.22936088e-01 -4.34828520e-01 -7.15140402e-02 8.98397148e-01 4.68429357e-01 -4.85648960e-01 6.88208699e-01 2.55965859e-01 -1.45194501e-01 -4.06984925e-01 -6.80682242e-01 -6.15123570e-01 2.48600513e-01 -2.64057696e-01 3.85629296e-01 4.49493855e-01 -4.46246862e-01 4.14164990e-01 -1.95496723e-01 2.33474895e-01 6.01236045e-01 1.12438239e-01 6.82463408e-01 -1.66288114e+00 -6.41344786e-01 -7.38119006e-01 -2.51819879e-01 -4.48141396e-01 3.50667715e-01 -7.04383016e-01 1.22708268e-01 -1.09261966e+00 -1.78322624e-02 -8.88769776e-02 -1.75696254e-01 1.30065113e-01 2.86869287e-01 -2.43817866e-02 4.91394699e-02 2.68282443e-01 2.26669624e-01 4.40453738e-01 6.38045430e-01 -1.35214508e-01 -4.45104167e-02 6.41838536e-02 2.87807565e-02 6.22577250e-01 9.54351068e-01 -2.78702974e-01 -1.08079843e-01 1.85249656e-01 3.92494649e-01 -1.63011447e-01 3.66050273e-01 -1.35364723e+00 -6.48098662e-02 5.14808903e-03 4.10022706e-01 -7.27464914e-01 3.00149351e-01 -6.64492249e-01 2.69965261e-01 5.93398392e-01 -1.43506348e-01 -3.22641671e-01 7.07273409e-02 -1.43177614e-01 -1.26339614e-01 -1.29700267e+00 1.35823250e+00 -1.59771800e-01 -5.12652934e-01 -9.03761238e-02 -3.79836708e-01 -6.00959659e-01 7.55625367e-01 -1.42720370e-02 -2.33654469e-01 -2.19746023e-01 -7.50491321e-01 -2.66000986e-01 3.97382826e-01 -9.98726040e-02 1.34021506e-01 -7.73952663e-01 -6.51832104e-01 2.88457483e-01 5.61175644e-02 -1.96042076e-01 5.27470894e-02 7.82071650e-01 -1.14145422e+00 4.40943658e-01 -3.09745371e-01 -5.93995869e-01 -1.47890985e+00 7.30861783e-01 1.56894958e+00 1.34569168e-01 -1.98776484e-01 5.87792456e-01 -2.47514606e-01 -6.54573262e-01 1.31675616e-01 -2.35938713e-01 -2.71308184e-01 -1.27392024e-01 2.26615921e-01 4.32300329e-01 3.54750931e-01 -7.80188739e-01 2.79470295e-01 6.79933429e-01 5.41716874e-01 -3.96423414e-02 1.17262995e+00 -2.04944480e-02 -4.21879172e-01 8.82388294e-01 7.57002890e-01 -2.83816993e-01 -1.24809408e+00 -3.63346159e-01 2.62943413e-02 1.69163778e-01 5.37480712e-01 -8.55578303e-01 -8.82919073e-01 1.09898281e+00 1.08022261e+00 8.45880568e-01 1.14980555e+00 -7.57872164e-02 3.54784906e-01 7.18016982e-01 -3.59857857e-01 -1.56087494e+00 -5.50067902e-01 7.26476312e-01 7.08769202e-01 -1.12282360e+00 4.15937379e-02 -1.12006903e-01 -3.26618701e-01 1.96689689e+00 6.59264863e-01 2.17963427e-01 5.03847361e-01 1.55877218e-01 3.65155041e-01 -2.39469841e-01 -8.06431174e-01 -6.68099463e-01 3.59180272e-01 3.53383958e-01 6.63052738e-01 6.98983520e-02 -2.60500878e-01 -7.29599521e-02 -3.10426652e-01 -3.61074299e-01 6.38609827e-01 3.44196372e-02 -1.06745279e+00 -1.03141618e+00 -9.45142150e-01 2.21970081e-01 2.14340072e-02 1.50392190e-01 -8.50234151e-01 4.86279756e-01 4.29085672e-01 6.71592295e-01 1.55597433e-01 -2.37967253e-01 2.12004647e-01 8.19564760e-01 4.23982203e-01 -1.78703129e-01 -6.72514319e-01 7.60980472e-02 2.54083931e-01 3.15847814e-01 -5.27876794e-01 -7.68730640e-01 -1.02136409e+00 -2.54672498e-01 -5.26668072e-01 3.24258238e-01 1.33521235e+00 1.02648413e+00 -1.00886151e-01 4.52743471e-01 8.55241299e-01 -8.43740106e-01 -8.79743040e-01 -1.60127699e+00 -1.05855167e+00 3.19330662e-01 2.64817804e-01 -5.02495229e-01 -7.51399934e-01 2.99491864e-02]
[7.441359043121338, 3.0807840824127197]
6b0816ae-2232-4c3f-873e-424f68318416
that-s-what-i-said-fully-controllable-talking
2304.03275
null
https://arxiv.org/abs/2304.03275v1
https://arxiv.org/pdf/2304.03275v1.pdf
That's What I Said: Fully-Controllable Talking Face Generation
The goal of this paper is to synthesise talking faces with controllable facial motions. To achieve this goal, we propose two key ideas. The first is to establish a canonical space where every face has the same motion patterns but different identities. The second is to navigate a multimodal motion space that only represents motion-related features while eliminating identity information. To disentangle identity and motion, we introduce an orthogonality constraint between the two different latent spaces. From this, our method can generate natural-looking talking faces with fully controllable facial attributes and accurate lip synchronisation. Extensive experiments demonstrate that our method achieves state-of-the-art results in terms of both visual quality and lip-sync score. To the best of our knowledge, we are the first to develop a talking face generation framework that can accurately manifest full target facial motions including lip, head pose, and eye movements in the generated video without any additional supervision beyond RGB video with audio.
['Joon Son Chung', 'Byeong-Yeol Kim', 'Youshin Lim', 'Jihwan Park', 'Hyeongkeun Lee', 'Jong-Bin Woo', 'Kyeongha Rho', 'Youngjoon Jang']
2023-04-06
null
null
null
null
['talking-face-generation', 'face-generation']
['computer-vision', 'computer-vision']
[ 1.10425659e-01 1.53382733e-01 -7.56678656e-02 -2.53543258e-01 -6.27251744e-01 -6.23087406e-01 6.73004925e-01 -9.55416679e-01 2.16726333e-01 3.54675978e-01 4.91661102e-01 2.18488902e-01 3.42873067e-01 -2.74471045e-01 -4.45463926e-01 -7.24406302e-01 3.60370100e-01 -1.12494767e-01 -2.84445405e-01 -1.22176528e-01 -3.85555811e-02 4.78927523e-01 -1.96490347e+00 2.26812437e-01 5.35388589e-01 1.05730975e+00 -9.49199721e-02 5.76330066e-01 2.53609091e-01 6.59609437e-01 -1.75769582e-01 -5.03556967e-01 2.54046589e-01 -5.17391503e-01 -6.45681262e-01 5.55572569e-01 6.88523710e-01 -5.18027425e-01 -3.91335368e-01 9.78871047e-01 6.55640304e-01 -7.59476274e-02 4.14663017e-01 -1.87599635e+00 -5.22889078e-01 1.48895249e-01 -4.82598662e-01 -5.66701412e-01 8.44596088e-01 4.54170913e-01 8.66327405e-01 -1.12963736e+00 8.04718018e-01 1.57288909e+00 4.27340776e-01 1.00098693e+00 -1.31635571e+00 -9.70652163e-01 4.32121754e-02 2.17799515e-01 -1.59129679e+00 -1.39237332e+00 8.76989007e-01 -3.88567626e-01 3.54256392e-01 3.04597318e-01 6.13213480e-01 1.40067196e+00 -3.26274633e-01 7.00237572e-01 6.68667197e-01 -3.81595761e-01 -1.27743492e-02 -9.88792395e-04 -5.48496604e-01 6.93609297e-01 -1.63906127e-01 -9.69240721e-03 -9.70004559e-01 7.39117563e-02 6.69747889e-01 -2.42203936e-01 -6.79347873e-01 -7.06922054e-01 -1.44226015e+00 6.25867188e-01 -1.74163073e-01 3.23344506e-02 -2.10419655e-01 1.82720348e-01 1.43161044e-02 -2.54356544e-02 1.99359551e-01 1.22030966e-01 -1.25511542e-01 -2.69198060e-01 -8.40520501e-01 -1.37089882e-02 6.26369476e-01 1.24748337e+00 6.84716582e-01 2.33266383e-01 -1.19966574e-01 6.91738725e-01 5.15519440e-01 8.13507557e-01 3.18309456e-01 -1.63634264e+00 3.17369550e-01 1.76925376e-01 2.37769485e-01 -1.04518056e+00 -1.39338940e-01 3.54246438e-01 -7.74550736e-01 2.35283539e-01 1.74319476e-01 -2.75197744e-01 -7.03488171e-01 2.37170625e+00 4.00266588e-01 3.22192639e-01 1.31656170e-01 7.67202795e-01 8.76822054e-01 4.41040903e-01 -2.34946191e-01 -5.57669759e-01 1.21901631e+00 -8.65518272e-01 -1.21436596e+00 6.02290034e-02 1.49872094e-01 -9.69852090e-01 1.01646173e+00 1.20609939e-01 -1.24565327e+00 -6.07673049e-01 -8.41876268e-01 -4.87691537e-02 2.69951671e-01 4.37290937e-01 6.05433643e-01 8.28271210e-01 -1.40335095e+00 2.14438960e-01 -7.69770265e-01 -4.62585449e-01 1.01157553e-01 3.37252229e-01 -8.67653906e-01 5.19445457e-04 -9.52474356e-01 5.49391747e-01 -2.36311615e-01 1.41762838e-01 -7.23139286e-01 -5.06893039e-01 -1.31151414e+00 -1.57950282e-01 2.47994512e-01 -8.76644611e-01 1.33687603e+00 -1.13482714e+00 -2.08331561e+00 9.36849236e-01 -8.41191947e-01 2.58284777e-01 5.68456709e-01 -1.15097299e-01 -4.05834615e-01 2.60493994e-01 1.47490650e-01 1.10920012e+00 1.09181714e+00 -1.47500420e+00 -4.46284354e-01 -2.35041872e-01 -2.81252474e-01 3.31532031e-01 -5.67642391e-01 1.51477799e-01 -9.03563976e-01 -6.97288334e-01 3.02030742e-01 -1.34079528e+00 4.01956379e-01 4.36203599e-01 -4.85985428e-01 2.91955937e-02 1.03661609e+00 -4.21591222e-01 8.68558407e-01 -2.22449708e+00 5.36369264e-01 -3.27587128e-02 1.55313089e-01 -1.70376934e-02 -2.92631805e-01 1.42663896e-01 -2.25625455e-01 5.64389452e-02 1.24563344e-01 -9.75571334e-01 1.07853845e-01 2.46938914e-02 -4.23249424e-01 6.59217060e-01 1.47571832e-01 8.54022264e-01 -6.65002763e-01 -4.88679588e-01 1.65420443e-01 8.99711549e-01 -6.77176416e-01 2.21811071e-01 8.25430006e-02 7.24781752e-01 -3.25979553e-02 8.13752651e-01 7.99474359e-01 8.61311797e-03 3.29599559e-01 -4.52579767e-01 1.37671586e-02 1.68849930e-01 -1.42585135e+00 1.82852066e+00 -3.28382909e-01 8.34606290e-01 4.43019927e-01 -2.49338940e-01 7.27155268e-01 7.27757215e-01 7.03089058e-01 -3.59179795e-01 6.68823570e-02 -1.91259403e-02 -2.80316830e-01 -5.26932418e-01 1.92516699e-01 -1.10948034e-01 1.23280600e-01 3.47573489e-01 2.17529863e-01 -1.81213677e-01 -1.80368260e-01 -1.83838606e-02 5.77567279e-01 1.57042637e-01 -2.80883349e-02 8.99065211e-02 7.11926341e-01 -7.97203183e-01 7.21734464e-01 -5.64840715e-03 -2.82194942e-01 8.54402900e-01 5.73439419e-01 -1.32137816e-02 -8.47187936e-01 -1.19758701e+00 1.43328026e-01 8.45108211e-01 7.62718394e-02 -5.25715351e-01 -8.50632727e-01 -1.82354107e-01 -3.33856642e-01 3.46807957e-01 -6.32706583e-01 -7.43017904e-03 -4.75969255e-01 2.44057383e-02 5.90191245e-01 4.16976839e-01 3.86607558e-01 -8.11836302e-01 -2.53209412e-01 -3.72872710e-01 -7.80354917e-01 -1.43266070e+00 -9.68905628e-01 -7.44734228e-01 -3.13048452e-01 -9.74603355e-01 -8.41701806e-01 -7.12261915e-01 7.89473593e-01 3.43819708e-01 7.11171091e-01 -2.06326529e-01 -1.91296220e-01 6.10999703e-01 -1.37482643e-01 -5.02808169e-02 -2.84897774e-01 -2.97174782e-01 4.97596771e-01 6.19740367e-01 -4.77683842e-02 -6.29314601e-01 -7.15876400e-01 5.00814080e-01 -6.69632971e-01 3.30140918e-01 1.75623447e-01 6.71244919e-01 3.54730695e-01 -2.55771965e-01 1.57658979e-01 -6.63759485e-02 3.06404561e-01 -1.07990168e-01 -4.03831005e-01 2.09804833e-01 -2.37557977e-01 2.25712056e-03 3.07344317e-01 -6.03649259e-01 -1.03674650e+00 4.56323206e-01 -8.62531736e-02 -8.42883348e-01 -2.59223908e-01 -2.67622530e-01 -9.26018357e-01 -4.33066301e-03 2.03057453e-01 2.50240922e-01 2.25226194e-01 -1.64068356e-01 6.88123465e-01 6.57850564e-01 8.31596315e-01 -4.39066052e-01 8.82161975e-01 8.68684351e-01 1.22244447e-01 -1.02859795e+00 -3.41154724e-01 -4.33573067e-01 -7.31086791e-01 -4.49570268e-01 9.63950932e-01 -1.11992455e+00 -1.31836784e+00 7.20024943e-01 -1.18086135e+00 -1.33783907e-01 2.18523312e-02 4.47115511e-01 -8.84297490e-01 5.54777563e-01 -4.79710877e-01 -8.74179363e-01 -1.78022057e-01 -1.47988045e+00 1.43495023e+00 1.31229796e-02 -4.61048871e-01 -5.03248274e-01 -1.94836259e-02 3.67948472e-01 2.00181291e-01 2.26216376e-01 2.59369791e-01 2.37755299e-01 -6.29381657e-01 1.86219916e-01 8.55940804e-02 1.11455768e-01 7.10006714e-01 3.19799334e-01 -1.27752745e+00 -3.57649952e-01 2.25418881e-02 -2.81287670e-01 6.19249284e-01 3.15826178e-01 6.53794348e-01 -5.47603607e-01 -2.01904595e-01 1.01915967e+00 8.09010148e-01 7.96236396e-02 6.45624101e-01 -1.00027412e-01 8.01639676e-01 9.65268433e-01 4.53917265e-01 5.52182257e-01 4.80023563e-01 1.20102692e+00 3.38755995e-01 -8.63668621e-02 -3.68393332e-01 -4.68645543e-01 6.61461949e-01 6.38783157e-01 5.37027791e-02 -1.75446212e-01 -7.33915508e-01 4.64828014e-01 -1.72874892e+00 -1.25768280e+00 2.18620718e-01 1.99571919e+00 7.99142897e-01 -5.14676630e-01 1.67923376e-01 2.30825976e-01 8.34296286e-01 1.64596543e-01 -5.16423821e-01 9.55294967e-02 -1.99472800e-01 -1.08941481e-01 -3.22165005e-02 7.52099633e-01 -1.06285346e+00 1.03465772e+00 6.23554277e+00 4.52999800e-01 -1.42620623e+00 -1.48610234e-01 4.99430388e-01 -4.09730762e-01 -4.01985437e-01 -1.10146068e-01 -7.04697847e-01 3.88493448e-01 5.15968442e-01 -1.14806212e-01 4.97568816e-01 7.17265964e-01 4.82472450e-01 2.87733316e-01 -1.25909257e+00 1.39823711e+00 4.56626028e-01 -1.24815309e+00 1.35530308e-01 1.66576847e-01 6.08076572e-01 -5.58678269e-01 4.51171994e-01 -2.56526142e-01 6.28081188e-02 -1.19855380e+00 9.85954881e-01 6.55007660e-01 1.35858572e+00 -6.87287807e-01 1.41220331e-01 -2.29424075e-03 -1.35444963e+00 1.70772687e-01 1.64071500e-01 2.63802528e-01 3.93125832e-01 -4.76331078e-02 -5.24368882e-01 2.91492522e-01 5.96613646e-01 6.87797904e-01 -1.86910540e-01 5.85686386e-01 -3.39970529e-01 1.54339671e-01 -1.79237604e-01 4.94851559e-01 -3.57091486e-01 -7.67827183e-02 6.77135706e-01 8.07037771e-01 6.14915669e-01 8.20166469e-02 -7.63537958e-02 7.45976210e-01 -1.66303962e-01 -5.38726225e-02 -7.53043592e-01 2.03742102e-01 6.18097723e-01 1.19459748e+00 -2.80401289e-01 5.60129322e-02 -3.32476109e-01 1.28010571e+00 -2.06724271e-01 4.65822101e-01 -8.93148184e-01 -2.99192201e-02 1.42749834e+00 7.47531503e-02 2.33490705e-01 -2.53704160e-01 -2.07796767e-02 -1.43451810e+00 1.76284149e-01 -9.91259277e-01 -6.16387725e-02 -9.44546223e-01 -8.83828580e-01 6.02524877e-01 -1.80612579e-01 -1.31298494e+00 -6.59761608e-01 -4.44004387e-01 -4.16360378e-01 8.83241653e-01 -1.30711651e+00 -1.54809225e+00 -5.09611547e-01 1.15097499e+00 3.99052650e-01 -1.96390480e-01 9.52198267e-01 2.01746702e-01 -5.83931327e-01 1.00281572e+00 -2.33975321e-01 1.85164452e-01 1.08551717e+00 -7.43381143e-01 2.99945831e-01 8.38907599e-01 1.19796015e-01 7.70832181e-01 7.67459154e-01 -2.57651210e-01 -1.62125707e+00 -9.10885453e-01 8.31945300e-01 -5.05569935e-01 3.75198364e-01 -5.63377321e-01 -4.19897974e-01 6.25107527e-01 1.98962525e-01 1.07584618e-01 8.50237489e-01 -1.96652964e-01 -5.86587489e-01 -2.07292601e-01 -9.19122994e-01 8.87860417e-01 1.14787018e+00 -7.67709315e-01 -4.90063056e-02 8.60956907e-02 6.86942875e-01 -4.42094564e-01 -6.59178674e-01 5.55387914e-01 8.32638144e-01 -1.16664195e+00 1.04805017e+00 -2.85280108e-01 2.10034668e-01 -4.49983448e-01 -3.76589626e-01 -1.00749528e+00 2.93273851e-02 -1.37826586e+00 -1.73080876e-01 1.70024073e+00 1.71271145e-01 -4.55059677e-01 7.78490007e-01 7.79609084e-01 2.13179186e-01 -4.12015975e-01 -9.17873502e-01 -5.88304937e-01 -2.02408716e-01 -3.81395042e-01 7.36815870e-01 1.01454055e+00 1.09536059e-01 2.71215022e-01 -8.27623308e-01 2.66141057e-01 5.59809446e-01 1.86665401e-01 1.19008732e+00 -9.59321201e-01 -1.07189745e-03 -3.59777629e-01 -4.84576672e-01 -9.87386346e-01 6.13678217e-01 -3.27920318e-01 1.33495435e-01 -1.14582860e+00 1.45003989e-01 6.84586838e-02 2.80423403e-01 5.29164076e-01 6.59996495e-02 2.83992380e-01 3.93327475e-01 2.56814182e-01 -3.52894098e-01 8.96799922e-01 1.28784406e+00 1.38761118e-01 -2.60405809e-01 -1.22061267e-01 -8.25815260e-01 8.03279519e-01 6.38493955e-01 -3.42822149e-02 -6.49671972e-01 -4.53148812e-01 -8.23091716e-02 3.32513988e-01 3.32905233e-01 -7.88260818e-01 3.58439744e-01 -3.88982475e-01 3.04109156e-01 -2.65537053e-01 8.71367395e-01 -5.54548919e-01 3.49911183e-01 7.89314210e-02 -1.88232005e-01 1.06193520e-01 2.67488897e-01 3.36615562e-01 -3.86176139e-01 5.14305294e-01 8.82875204e-01 2.12374702e-01 -4.91400480e-01 4.77828622e-01 -2.35923707e-01 -1.18049838e-01 1.16096461e+00 -2.62802422e-01 -1.78306594e-01 -1.11840916e+00 -7.28882611e-01 5.19328229e-02 8.49727750e-01 7.04196334e-01 6.61181033e-01 -1.73340678e+00 -6.98723614e-01 4.20488238e-01 1.01873085e-01 -2.83524483e-01 2.65421510e-01 6.86806381e-01 -1.90804213e-01 4.87271339e-01 -2.56499380e-01 -8.24100196e-01 -1.74900985e+00 4.92130488e-01 2.86053598e-01 4.37563658e-01 -4.39807147e-01 8.46945703e-01 4.00051177e-01 -2.39682376e-01 3.77758950e-01 7.02112690e-02 -5.05917333e-02 1.81111559e-01 5.70698619e-01 1.67608455e-01 -1.67287290e-01 -1.38113904e+00 -4.92230654e-01 8.49185467e-01 4.19835448e-01 -6.08501434e-01 9.48489189e-01 -5.43591797e-01 -4.04455140e-03 2.28145137e-01 1.35182035e+00 4.36617136e-01 -1.48216379e+00 1.68718249e-01 -5.43373048e-01 -7.45137393e-01 -2.42361173e-01 -2.59104788e-01 -1.21568465e+00 9.05839443e-01 5.96660614e-01 -2.79474735e-01 1.34263456e+00 -6.90442026e-02 5.98713636e-01 9.99093950e-02 2.42053404e-01 -7.26448715e-01 4.45114046e-01 3.52584839e-01 1.05593359e+00 -1.14869690e+00 -4.29308027e-01 -6.94797397e-01 -8.75180721e-01 1.02739382e+00 5.77528894e-01 5.87393284e-01 4.92490798e-01 2.28004828e-01 2.95014679e-01 1.52457938e-01 -7.84683168e-01 -4.22089338e-01 4.21381235e-01 8.26005161e-01 4.66731638e-01 -2.65379786e-01 3.59608024e-01 2.09553644e-01 -5.57954967e-01 -1.05700999e-01 2.88876265e-01 4.96561050e-01 3.54690105e-02 -1.19859529e+00 -4.67038453e-01 -4.81053263e-01 -3.35058033e-01 5.40302582e-02 -4.88557637e-01 6.35954857e-01 5.60293384e-02 1.36833060e+00 -3.93784940e-02 -4.71094429e-01 3.04461509e-01 2.46940136e-01 5.35368025e-01 -3.97984803e-01 2.82063216e-01 3.25607210e-01 -5.68604283e-02 -1.00159216e+00 -5.07027090e-01 -7.92875051e-01 -9.60398674e-01 -5.52466452e-01 -1.36393398e-01 -1.86121240e-01 5.85487723e-01 5.32797277e-01 5.82611799e-01 1.26663923e-01 8.40824008e-01 -1.20083201e+00 -1.73229232e-01 -8.25951993e-01 -3.64834279e-01 5.77453136e-01 6.86192513e-01 -7.35883296e-01 -4.35795337e-01 5.06030023e-01]
[13.198308944702148, -0.4148283302783966]
b05d01d8-29f0-4ab8-9588-845a1b10a249
forbidden-knowledge-in-machine-learning
1911.08603
null
https://arxiv.org/abs/1911.08603v1
https://arxiv.org/pdf/1911.08603v1.pdf
Forbidden knowledge in machine learning -- Reflections on the limits of research and publication
Certain research strands can yield "forbidden knowledge". This term refers to knowledge that is considered too sensitive, dangerous or taboo to be produced or shared. Discourses about such publication restrictions are already entrenched in scientific fields like IT security, synthetic biology or nuclear physics research. This paper makes the case for transferring this discourse to machine learning research. Some machine learning applications can very easily be misused and unfold harmful consequences, for instance with regard to generative video or text synthesis, personality analysis, behavior manipulation, software vulnerability detection and the like. Up to now, the machine learning research community embraces the idea of open access. However, this is opposed to precautionary efforts to prevent the malicious use of machine learning applications. Information about or from such applications may, if improperly disclosed, cause harm to people, organizations or whole societies. Hence, the goal of this work is to outline norms that can help to decide whether and when the dissemination of such information should be prevented. It proposes review parameters for the machine learning community to establish an ethical framework on how to deal with forbidden knowledge and dual-use applications.
['Thilo Hagendorff']
2019-11-19
null
null
null
null
['vulnerability-detection']
['miscellaneous']
[ 4.42052871e-01 8.25887680e-01 -2.91501671e-01 6.10284284e-02 4.37766872e-03 -7.14432061e-01 8.15965652e-01 2.90483773e-01 -4.71385449e-01 9.29043889e-01 -3.23732905e-02 -8.54638696e-01 -1.11445941e-01 -7.26449192e-01 -5.66747487e-01 -9.00667846e-01 5.31391263e-01 -2.49040991e-01 8.07675440e-03 5.13431989e-02 8.41665328e-01 6.47302449e-01 -1.60192406e+00 5.88999391e-02 6.57438159e-01 3.93225938e-01 -2.86096096e-01 2.35220209e-01 -1.60036668e-01 9.95117307e-01 -8.60485792e-01 -9.28309441e-01 5.55835739e-02 -4.20340329e-01 -9.71929729e-01 -2.34264299e-01 -8.92316774e-02 -1.93647772e-01 5.63459657e-02 1.50135493e+00 1.67181537e-01 -2.59409159e-01 4.85185087e-01 -1.26220834e+00 -8.25694144e-01 7.67425418e-01 -2.56143212e-01 1.75706446e-01 3.59962940e-01 3.87967944e-01 1.62520036e-01 -1.19906351e-01 9.35623288e-01 1.02182686e+00 1.91743121e-01 4.69283789e-01 -9.64147329e-01 -8.69206548e-01 -2.27306888e-01 1.58821940e-01 -1.13202834e+00 -4.17234570e-01 6.39634013e-01 -8.67605448e-01 5.98717868e-01 6.80971920e-01 7.07553923e-01 1.66245401e+00 7.36566722e-01 -5.95910475e-03 1.21446538e+00 -6.60401762e-01 3.87371719e-01 7.25045085e-01 6.71840906e-02 5.45164585e-01 1.06413579e+00 -3.64206992e-02 -4.34490144e-01 -3.79988104e-01 1.70951009e-01 -5.59452176e-01 -2.62423277e-01 5.46249188e-02 -1.13416326e+00 8.82607639e-01 -2.27802575e-01 9.48075235e-01 -2.68485487e-01 -1.34159938e-01 5.84189713e-01 5.27637362e-01 4.87743527e-01 8.53018522e-01 -2.25365534e-01 -5.77941537e-01 -4.30370599e-01 5.90863563e-02 1.02107370e+00 4.31593299e-01 5.19622564e-01 -1.24665923e-01 3.26797754e-01 4.87863392e-01 4.37017053e-01 2.50068367e-01 4.65756387e-01 -8.16191435e-01 -2.41812915e-01 6.00928247e-01 2.31522974e-02 -1.46821880e+00 -9.53851640e-02 3.07478458e-02 -5.46232641e-01 2.62132436e-01 4.42835003e-01 -3.68840009e-01 -4.10134226e-01 1.26835823e+00 3.55034560e-01 -1.04357801e-01 1.86398581e-01 7.41190255e-01 9.57624733e-01 5.46043694e-01 1.90760538e-01 -4.70139146e-01 1.26882052e+00 -1.88882202e-01 -1.11262548e+00 1.44937128e-01 9.83225882e-01 -9.82342958e-01 7.49155104e-01 7.84922898e-01 -7.79340506e-01 7.54701346e-02 -1.09258687e+00 -9.61145572e-03 -9.37149405e-01 -4.55535680e-01 5.06596208e-01 1.56838131e+00 -4.51617420e-01 5.99901438e-01 -5.56060910e-01 -4.62315977e-01 5.68286598e-01 3.35665829e-02 -1.52963653e-01 1.75169900e-01 -1.17746079e+00 1.03782678e+00 7.16309011e-01 1.60066783e-01 -4.98538047e-01 -6.57601535e-01 -4.83259082e-01 -5.01607180e-01 5.93889236e-01 -3.16184193e-01 6.95642769e-01 -1.29810321e+00 -1.33571041e+00 1.38756752e+00 4.10686940e-01 -4.31966126e-01 5.96168876e-01 -2.59742916e-01 -7.32572079e-01 -1.06264897e-01 7.32720494e-02 1.28506586e-01 8.46334517e-01 -1.00455201e+00 -4.58628744e-01 -2.37567320e-01 2.98989773e-01 -4.77830201e-01 -5.80315232e-01 5.09528995e-01 1.75711811e-01 -6.22614086e-01 -3.97985548e-01 -1.07301998e+00 9.53849778e-02 -3.17422271e-01 -7.64943123e-01 -9.91837308e-02 1.11324680e+00 -6.18048131e-01 1.25918877e+00 -2.11958957e+00 -8.88073817e-02 2.94985175e-01 1.67260662e-01 4.68118906e-01 4.43056792e-01 3.22558314e-01 1.45752117e-01 8.94982517e-01 7.00744838e-02 8.50656033e-01 1.48811899e-02 1.26217693e-01 -1.52654737e-01 8.58645260e-01 -2.46272027e-01 4.91278738e-01 -7.74329185e-01 -2.80149877e-01 1.10482022e-01 6.00304723e-01 5.40084764e-02 -1.87789708e-01 -2.84233063e-01 6.99881554e-01 -7.97239542e-01 5.77336431e-01 5.61374605e-01 -5.91328070e-02 3.48866165e-01 1.01736359e-01 -4.83199835e-01 1.88719928e-01 -6.66853249e-01 1.06661117e+00 -1.07450657e-01 9.50645387e-01 -5.11949509e-02 -8.72404873e-01 8.81626189e-01 4.32445735e-01 2.12264240e-01 -4.63553965e-01 4.40084666e-01 2.52452254e-01 2.92901844e-01 -1.01400316e+00 1.77678421e-01 1.03639498e-01 1.10721923e-01 7.20813632e-01 -3.30806285e-01 1.06409136e-02 -6.25577196e-02 -2.44227555e-02 8.52811575e-01 3.17321986e-01 3.56447428e-01 -3.49011123e-01 7.67448545e-01 -1.09262288e-01 3.93708140e-01 4.76132900e-01 -7.17119351e-02 -2.80592501e-01 8.87350321e-01 -5.68496704e-01 -8.95854175e-01 -4.39703286e-01 -2.79867560e-01 1.04278302e+00 -3.95114459e-02 -3.38813990e-01 -1.22464061e+00 -7.83640563e-01 -2.06420004e-01 1.06974661e+00 -4.36731309e-01 -5.08513689e-01 -2.45652366e-02 -5.53714275e-01 8.30576062e-01 -4.07559752e-01 3.77751112e-01 -1.17203593e+00 -1.16927707e+00 5.05318604e-02 4.26777974e-02 -1.05598307e+00 3.35891396e-01 -1.43189490e-01 -4.46070254e-01 -1.33869863e+00 -4.00006056e-01 -2.10333839e-01 5.77380300e-01 -1.11173451e-01 6.63682640e-01 5.65780759e-01 -2.29007617e-01 3.30902249e-01 -6.04830742e-01 -9.84539330e-01 -1.21250308e+00 9.29836854e-02 -1.40234241e-02 3.12477611e-02 6.95886195e-01 -3.98754478e-01 -1.29287377e-01 1.87622845e-01 -1.18681145e+00 -1.75077185e-01 3.95068854e-01 1.20401300e-01 -7.04496503e-02 -4.03753519e-02 8.52541506e-01 -1.39978933e+00 7.43997395e-01 -6.09092176e-01 -4.97693688e-01 4.25096691e-01 -9.38714683e-01 -2.47202963e-01 2.84825534e-01 -4.14813548e-01 -1.21188819e+00 -4.51457947e-01 -2.83059403e-02 -1.00960098e-01 -6.74033463e-01 5.01668811e-01 -5.68597972e-01 -4.81652528e-01 7.96470046e-01 -1.75395124e-02 9.98556837e-02 -2.63302267e-01 1.90044820e-01 1.06429219e+00 7.42492005e-02 -4.46000695e-01 7.42461085e-01 4.62410927e-01 1.11815035e-01 -1.45593262e+00 -5.94007850e-01 -3.76678780e-02 -3.14729571e-01 -6.39649928e-01 9.10617232e-01 -3.14626783e-01 -7.18332589e-01 2.73377955e-01 -1.11611331e+00 1.73856437e-01 7.40309581e-02 4.13032264e-01 -3.77485216e-01 5.53927779e-01 -1.03470765e-01 -8.89012992e-01 -1.89640418e-01 -9.26505446e-01 1.48193225e-01 1.80497095e-01 -5.89043558e-01 -1.00097895e+00 -2.01108009e-01 8.62352490e-01 2.96847224e-01 6.75493181e-01 9.71861303e-01 -9.41357374e-01 -2.97019809e-01 -2.00762898e-01 3.33165795e-01 4.10899758e-01 1.81000367e-01 4.47536439e-01 -1.11745739e+00 5.44267558e-02 3.14284861e-01 -3.28712255e-01 2.59912014e-01 -1.75125763e-01 9.09043372e-01 -9.56829011e-01 -2.56792754e-01 3.38001817e-01 1.09597039e+00 6.04100049e-01 9.82492149e-01 8.43966186e-01 5.86398482e-01 1.17910981e+00 6.29894674e-01 4.82291877e-01 -3.98890734e-01 3.22391987e-01 2.56151378e-01 4.47857112e-01 4.70451146e-01 7.18985051e-02 4.36049998e-01 5.47736585e-01 -5.52532494e-01 -1.14959422e-02 -1.11404240e+00 2.46443585e-01 -1.54721403e+00 -1.10147667e+00 -6.20834053e-01 2.21862173e+00 6.76389158e-01 3.28082770e-01 -1.18486114e-01 8.00983533e-02 7.65190601e-01 6.70689391e-03 -2.60142356e-01 -1.10887158e+00 -4.64397185e-02 -2.26885602e-01 4.67538148e-01 1.00545771e-01 -8.15254152e-01 7.08593667e-01 5.38199472e+00 9.07378614e-01 -1.27681291e+00 3.67757827e-01 6.59677982e-01 1.82981104e-01 -5.97509861e-01 9.65787321e-02 -3.20951283e-01 6.34205163e-01 1.18647432e+00 -6.33361995e-01 2.06907988e-01 6.75244212e-01 4.47203279e-01 -2.96689123e-01 -8.20914745e-01 6.38095617e-01 1.14948284e-02 -1.45035839e+00 -9.23735052e-02 3.91345441e-01 4.67068553e-01 -4.58420336e-01 2.12268103e-02 -1.65965065e-01 2.61418410e-02 -1.04964316e+00 5.61613739e-01 5.78514218e-01 3.69613498e-01 -1.08068955e+00 6.66952610e-01 4.98023719e-01 -2.50880606e-02 1.64879456e-01 -3.69109362e-01 -2.26452544e-01 -8.27772394e-02 6.69233739e-01 -5.41267574e-01 4.84094679e-01 6.75114632e-01 2.97167391e-01 -5.34500062e-01 7.14099944e-01 -5.02332807e-01 8.13054562e-01 1.97469741e-02 -2.95644701e-01 3.38471889e-01 -3.25770885e-01 7.90884912e-01 1.20310080e+00 1.45678863e-01 -5.95260561e-02 -4.98587191e-01 8.61549973e-01 1.76748320e-01 2.14778021e-01 -1.26329982e+00 -7.47534633e-01 3.18523914e-01 1.00119495e+00 -1.08207119e+00 4.17642854e-02 -7.27083027e-01 6.98090971e-01 -3.75639379e-01 2.13549823e-01 -6.95423424e-01 -3.07116181e-01 3.63036245e-01 3.13187182e-01 -1.05657786e-01 3.35243493e-01 -3.89260530e-01 -8.07708025e-01 -2.13164791e-01 -1.08550501e+00 1.59059390e-01 -3.85716528e-01 -9.14106369e-01 2.66858220e-01 8.46716091e-02 -9.61614430e-01 -1.67203486e-01 -5.77054858e-01 -2.68067420e-01 5.26221633e-01 -9.44238901e-01 -1.15389609e+00 3.18907470e-01 1.83148488e-01 -4.30881716e-02 -4.58704323e-01 8.75724733e-01 -1.45694345e-01 -4.98767763e-01 3.02849144e-01 1.90408885e-01 1.28241451e-02 5.30371547e-01 -5.35831094e-01 -2.08350539e-01 6.98111951e-01 6.83566630e-02 6.21967018e-01 9.28194284e-01 -8.55855167e-01 -1.40236402e+00 -7.81805813e-01 9.61873591e-01 -5.12070000e-01 7.72561729e-01 -2.59671628e-01 -1.04113710e+00 6.20175123e-01 4.31769878e-01 -4.27696496e-01 1.19476485e+00 4.23146486e-02 -3.27313811e-01 2.97477514e-01 -1.42792249e+00 7.18948066e-01 5.58406591e-01 -5.10094941e-01 -6.22977316e-01 4.97545302e-01 4.81901139e-01 7.68760666e-02 -1.06185329e+00 -9.19488668e-02 7.16254056e-01 -1.06119466e+00 5.41191995e-01 -8.12416136e-01 4.70631242e-01 1.14008391e-04 2.04496399e-01 -8.02319944e-01 5.12857176e-02 -1.08919907e+00 5.29578887e-02 1.36355126e+00 5.22336066e-01 -9.17575359e-01 6.02695405e-01 9.81107891e-01 9.06865075e-02 -4.38424677e-01 -1.14896977e+00 -7.26794899e-01 4.97614712e-01 -5.73972523e-01 4.57886428e-01 1.64313364e+00 4.89418238e-01 -7.71529181e-03 -2.67845124e-01 2.51252688e-02 4.83604342e-01 -4.08072710e-01 7.58888781e-01 -1.34152055e+00 2.70254225e-01 -5.58764100e-01 -6.78700268e-01 2.42097244e-01 1.95443183e-01 -5.57234228e-01 -5.15693843e-01 -1.01784885e+00 4.46655154e-02 -3.26992990e-03 -5.40721305e-02 2.56202251e-01 3.55205566e-01 8.83181915e-02 3.88302863e-01 2.32063323e-01 -2.54057318e-01 -1.99118629e-01 1.20630634e+00 8.81595816e-03 -3.70372727e-04 -1.63259178e-01 -1.07745576e+00 1.13813794e+00 1.02945495e+00 -5.69519877e-01 -3.68138462e-01 2.88132161e-01 9.39599395e-01 -4.33351368e-01 5.17879665e-01 -8.43514025e-01 2.14089192e-02 -6.54832125e-01 -4.78744470e-02 6.68526590e-02 -2.61674345e-01 -1.06456196e+00 6.50418282e-01 6.79048061e-01 -3.69419128e-01 -5.63211024e-01 1.30785629e-01 3.13986868e-01 2.01428875e-01 -8.75048816e-01 6.39123738e-01 -2.01397538e-01 -3.30594808e-01 -3.71098012e-01 -9.24861968e-01 -2.52301246e-02 1.75161362e+00 -5.78401327e-01 -5.82839668e-01 -1.81811765e-01 -7.14362621e-01 -1.45049646e-01 6.50961161e-01 6.17128193e-01 4.75904137e-01 -8.00155163e-01 -5.69359899e-01 -7.55432174e-02 1.36664063e-01 -5.77429414e-01 2.37464517e-01 7.82953024e-01 -6.06750071e-01 5.48254848e-01 -3.30166370e-01 9.27045718e-02 -1.39551985e+00 9.25441921e-01 4.91431877e-02 2.60337919e-01 -7.82072365e-01 2.11140975e-01 -1.44575194e-01 1.34499222e-01 1.34452403e-01 1.41387358e-01 -4.84500229e-01 2.21206710e-01 5.98579824e-01 8.40125442e-01 2.16616355e-02 -8.77389669e-01 -2.82433599e-01 8.25082213e-02 -3.40355605e-01 9.18898433e-02 1.01349103e+00 -3.13753933e-02 -4.62008595e-01 7.88886428e-01 9.53522921e-01 1.68103963e-01 -5.38569510e-01 4.97393399e-01 1.72473997e-01 -6.32356822e-01 1.13213867e-01 -1.04063404e+00 -7.92102456e-01 6.14165187e-01 3.93995643e-01 1.01265717e+00 7.00756669e-01 -7.86473230e-02 3.15584153e-01 5.75789750e-01 4.03336078e-01 -1.57758248e+00 -3.42267245e-01 1.09288424e-01 1.02201116e+00 -8.99535239e-01 1.30202651e-01 -5.44531941e-01 -5.87785184e-01 1.26184940e+00 3.91693592e-01 4.67839628e-01 7.60351419e-01 3.27987611e-01 2.08359301e-01 -3.15136552e-01 -4.50317591e-01 3.42499852e-01 4.00880575e-02 1.02202547e+00 7.13463306e-01 1.73298866e-01 -1.26308692e+00 4.31594402e-01 -1.24764323e-01 3.71860325e-01 1.02797520e+00 1.08912861e+00 -4.26975757e-01 -1.08605742e+00 -8.80439878e-01 5.39320529e-01 -1.27332675e+00 3.22042078e-01 -9.75071609e-01 8.66896868e-01 6.79178715e-01 1.10881186e+00 -1.96398005e-01 -1.98983520e-01 -6.93434253e-02 2.96473145e-01 3.99660952e-02 -3.39742064e-01 -8.21810424e-01 -1.97499588e-01 5.09641588e-01 -3.47946256e-01 -8.11918318e-01 -6.20378196e-01 -8.97213221e-01 -6.76751077e-01 -2.81838268e-01 2.26401523e-01 8.30482900e-01 1.06229699e+00 2.51757413e-01 2.73469090e-01 7.84354284e-02 -2.03468084e-01 -1.61985353e-01 -6.60654485e-01 -4.90510404e-01 2.18847826e-01 7.94927217e-03 -4.93022949e-01 -4.04438674e-01 2.58097082e-01]
[8.9616060256958, 6.5857834815979]
14468063-0cf1-4d51-a88e-2073f2939fb5
neural-segmental-hypergraphs-for-overlapping
1810.01817
null
http://arxiv.org/abs/1810.01817v1
http://arxiv.org/pdf/1810.01817v1.pdf
Neural Segmental Hypergraphs for Overlapping Mention Recognition
In this work, we propose a novel segmental hypergraph representation to model overlapping entity mentions that are prevalent in many practical datasets. We show that our model built on top of such a new representation is able to capture features and interactions that cannot be captured by previous models while maintaining a low time complexity for inference. We also present a theoretical analysis to formally assess how our representation is better than alternative representations reported in the literature in terms of representational power. Coupled with neural networks for feature learning, our model achieves the state-of-the-art performance in three benchmark datasets annotated with overlapping mentions.
['Bailin Wang', 'Wei Lu']
2018-10-03
neural-segmental-hypergraphs-for-overlapping-1
https://aclanthology.org/D18-1019
https://aclanthology.org/D18-1019.pdf
emnlp-2018-10
['overlapping-mention-recognition', 'nested-named-entity-recognition', 'nested-mention-recognition']
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
[ 1.06194541e-01 7.54021108e-01 -4.88536805e-01 -5.15650988e-01 -6.80131495e-01 -4.68549103e-01 7.18247294e-01 6.12491846e-01 -2.26236433e-01 8.85830641e-01 3.85278106e-01 -1.86273456e-01 -3.06364030e-01 -9.99719501e-01 -9.76405561e-01 -3.09104353e-01 -4.60046262e-01 8.32088053e-01 3.97368759e-01 -3.24298173e-01 3.86778861e-02 2.36727297e-01 -1.43158209e+00 1.26750022e-01 9.09651279e-01 5.74822843e-01 -2.24280715e-01 1.69558555e-01 -2.22254053e-01 6.21417880e-01 -5.82530618e-01 -7.71120787e-01 1.26093328e-01 -5.81855066e-02 -1.18182445e+00 -2.79762834e-01 6.38360560e-01 2.30419129e-01 -8.15081418e-01 8.32537532e-01 2.24233419e-01 2.36258894e-01 8.09487462e-01 -9.78436589e-01 -9.19109941e-01 1.27005959e+00 -3.93351048e-01 2.35970229e-01 4.98330832e-01 -3.56933713e-01 1.64288032e+00 -6.60329938e-01 9.14534092e-01 1.29463243e+00 7.62217402e-01 3.31919700e-01 -1.40981913e+00 -4.09858078e-01 5.12499809e-01 5.07499397e-01 -1.44449449e+00 -2.39198685e-01 4.51312095e-01 -1.40176654e-01 1.56366563e+00 2.81386644e-01 5.12766063e-01 9.07204032e-01 7.39828795e-02 9.22793329e-01 5.15047610e-01 -4.76985544e-01 -9.90940183e-02 -1.86502993e-01 7.01501727e-01 7.76822209e-01 8.42658162e-01 -3.29287723e-02 -3.70748878e-01 -3.68847996e-01 5.20215452e-01 -3.58889222e-01 -1.58906743e-01 -4.49590683e-01 -9.22832310e-01 9.44263160e-01 7.26890564e-01 6.35132849e-01 -3.69959027e-01 4.53444153e-01 3.84565383e-01 2.01630712e-01 4.74742621e-01 7.19178975e-01 -7.08276093e-01 1.76479116e-01 -7.32508600e-01 1.23605184e-01 1.13278985e+00 1.09360111e+00 6.29906297e-01 -1.77384034e-01 -2.70388633e-01 6.28061891e-01 2.76793182e-01 2.49779224e-02 1.73201650e-01 -6.43925130e-01 4.98921305e-01 1.04110074e+00 -1.07985042e-01 -8.99386346e-01 -6.98541105e-01 -4.52074826e-01 -7.68249273e-01 -7.22682655e-01 2.71250784e-01 5.04270196e-02 -7.48092115e-01 1.85020852e+00 2.11155355e-01 6.49149060e-01 1.25600711e-01 3.16791087e-01 1.12751317e+00 5.07128179e-01 1.58750221e-01 -1.83426604e-01 1.48173797e+00 -8.00729990e-01 -9.13661242e-01 -2.25204960e-01 8.20048451e-01 -1.31663591e-01 4.34267968e-01 1.01411734e-02 -1.00428987e+00 -6.44064993e-02 -9.58212554e-01 -2.03920513e-01 -6.16853416e-01 -3.01248524e-02 1.26325572e+00 8.27258766e-01 -9.82904553e-01 7.23020077e-01 -8.32725227e-01 -5.29994130e-01 3.02010864e-01 6.32868528e-01 -4.05734986e-01 9.93669629e-02 -1.72333181e+00 1.16243756e+00 6.94703996e-01 -8.80101919e-02 -3.51007879e-01 -7.10933149e-01 -1.08704996e+00 3.27158570e-01 7.84117758e-01 -8.67011130e-01 9.30999339e-01 -1.16531834e-01 -1.18026412e+00 6.91182673e-01 -2.92841494e-01 -7.65285075e-01 5.18155377e-03 -1.89845785e-01 -5.16334474e-01 -2.18706187e-02 -1.30059287e-01 2.49682486e-01 -7.39699528e-02 -1.18566322e+00 -3.40706259e-01 -4.16111082e-01 5.34643412e-01 -1.15793794e-01 -2.78673649e-01 -3.35583501e-02 -6.57038093e-01 -5.00808299e-01 -7.78589956e-03 -9.63082194e-01 -5.67098856e-01 -6.29042804e-01 -8.66548121e-01 -5.90719640e-01 4.02530223e-01 -3.85541946e-01 1.47762513e+00 -1.45665479e+00 4.47342992e-01 5.86358726e-01 3.11634809e-01 4.70254987e-01 -1.13234133e-01 7.38480091e-01 -8.99118334e-02 3.75534475e-01 -2.92280227e-01 -4.06490684e-01 2.56749868e-01 3.72962892e-01 -4.49574031e-02 4.55078989e-01 2.16181412e-01 1.14498556e+00 -9.88539398e-01 -4.32851076e-01 -1.70912351e-02 5.83945870e-01 -5.07423282e-01 6.45302683e-02 -2.97201991e-01 1.57813787e-01 -5.28939188e-01 6.18609190e-01 5.29362798e-01 -4.67340052e-01 9.78792548e-01 -1.74665302e-01 3.15350473e-01 7.33545363e-01 -1.29934621e+00 1.61724114e+00 -3.14131707e-01 3.11595410e-01 -3.38530511e-01 -1.22223306e+00 9.74843740e-01 3.22586805e-01 3.44106138e-01 -2.41230011e-01 3.69733274e-02 -7.88532421e-02 -8.89747497e-03 -3.90950739e-01 6.53560340e-01 3.41785669e-01 -3.67436230e-01 3.06035221e-01 3.41430604e-01 3.49830627e-01 4.56300825e-01 3.85587037e-01 1.35102046e+00 -1.10322736e-01 7.53961682e-01 -2.32112661e-01 5.12677312e-01 -3.76680017e-01 6.38471782e-01 8.86843026e-01 2.68809557e-01 8.86272714e-02 7.65225291e-01 -4.69595402e-01 -5.93691647e-01 -8.97825778e-01 -2.49482810e-01 1.12771785e+00 7.83901066e-02 -8.74577999e-01 -4.02860105e-01 -8.67435575e-01 1.28630295e-01 7.42085338e-01 -9.94541347e-01 -1.33726209e-01 -6.15823627e-01 -9.62890446e-01 8.86258960e-01 7.63233721e-01 2.84218974e-02 -8.44521284e-01 -3.29638034e-01 2.36485496e-01 -1.21968135e-01 -1.21618497e+00 -1.09077394e-01 3.46003592e-01 -7.33366311e-01 -1.26002097e+00 -1.60823226e-01 -6.92721188e-01 3.98871392e-01 -1.91394150e-01 1.56995928e+00 3.30653638e-01 -2.47484013e-01 1.55865967e-01 -5.49259007e-01 -2.17551410e-01 -1.80418164e-01 6.05362236e-01 -1.07479170e-01 -3.22133183e-01 6.25871956e-01 -6.74770772e-01 -2.84638762e-01 -1.34019196e-01 -8.37129116e-01 -3.57714981e-01 6.84769034e-01 9.09740567e-01 3.22672039e-01 -2.35634223e-01 7.39187896e-01 -1.58345306e+00 4.01507646e-01 -8.77144814e-01 -4.97170001e-01 5.79487860e-01 -7.09708989e-01 4.21126872e-01 3.51072818e-01 -6.72920868e-02 -8.83743823e-01 1.35347828e-01 -1.32351145e-01 1.90726703e-03 -2.06831381e-01 1.07441127e+00 -2.67628282e-01 -1.98663259e-03 3.86667848e-01 -3.70460637e-02 -5.22570968e-01 -7.42810547e-01 7.59431243e-01 2.97687680e-01 4.51476187e-01 -6.78255022e-01 6.17934644e-01 2.03781217e-01 1.39947504e-01 -5.40083766e-01 -8.83640110e-01 -5.94331324e-01 -1.01663697e+00 3.68731618e-01 5.45760632e-01 -7.95419097e-01 -8.02014470e-01 -1.10908963e-01 -1.27367187e+00 2.86613613e-01 -1.32175669e-01 2.95428842e-01 -3.93942714e-01 4.52123493e-01 -7.22419620e-01 -7.28308737e-01 -3.27981681e-01 -8.67035389e-01 8.40188801e-01 1.23877935e-01 -2.14838177e-01 -1.21741760e+00 3.42455328e-01 1.02249086e-01 2.21596405e-01 3.51050407e-01 1.21629083e+00 -1.27417493e+00 -6.02782071e-01 -1.29606873e-01 -1.41468257e-01 -4.16107714e-01 6.82832673e-02 -9.71115846e-03 -8.12679470e-01 -1.62354738e-01 -8.17117929e-01 -2.54760180e-02 1.35190117e+00 3.03770781e-01 1.02712774e+00 -3.97732526e-01 -7.94334471e-01 5.27673960e-01 1.50667393e+00 -4.83755320e-02 6.82326078e-01 3.40458065e-01 5.92687249e-01 5.86746573e-01 4.15945739e-01 3.53487372e-01 8.33483815e-01 9.07364249e-01 5.58150291e-01 1.32316694e-01 1.05916649e-01 -2.97014892e-01 -5.32287918e-02 5.99249423e-01 -3.89701039e-01 -4.61652964e-01 -1.03661966e+00 9.96332824e-01 -2.23410535e+00 -9.80197072e-01 -3.41733098e-01 1.86290622e+00 8.83305073e-01 -1.65204983e-02 2.79777288e-01 -5.46294637e-02 7.60803640e-01 6.35032207e-02 -2.60234743e-01 -4.85517979e-01 -2.04809994e-01 4.79309946e-01 5.84966719e-01 4.02895600e-01 -1.25978541e+00 1.10905898e+00 7.65775490e+00 3.84656847e-01 -2.83158600e-01 -2.96275057e-02 6.40978143e-02 2.20546991e-01 -4.26301479e-01 5.96233122e-02 -9.64498162e-01 9.32843909e-02 1.30161083e+00 -3.55725199e-01 8.34391415e-02 6.04991794e-01 -5.13247967e-01 2.85097063e-01 -1.30619514e+00 3.57015699e-01 4.54520583e-02 -1.64448488e+00 2.47596025e-01 7.22282603e-02 7.71452487e-01 1.06362998e-01 -6.95590675e-02 4.21399862e-01 8.18269432e-01 -1.21407843e+00 5.35938591e-02 4.83815759e-01 3.96137059e-01 -9.86057460e-01 1.18298805e+00 1.41941011e-01 -1.45765245e+00 -1.12268515e-01 -3.66567880e-01 -1.66055392e-02 2.03617945e-01 6.03463173e-01 -9.44000244e-01 1.29271698e+00 3.43673646e-01 8.40861320e-01 -7.57491529e-01 1.12118781e+00 -4.95814055e-01 6.77146673e-01 -1.52521700e-01 -7.93374032e-02 2.59773076e-01 1.31118953e-01 4.56029147e-01 1.65304995e+00 4.26751189e-02 1.56701297e-01 2.10196987e-01 7.30511427e-01 -3.64861816e-01 1.73853487e-01 -8.05871308e-01 -4.12140563e-02 9.39207971e-01 9.39559758e-01 -6.00703478e-01 -4.33363795e-01 -5.54791093e-01 5.02555907e-01 8.46483052e-01 1.19713470e-01 -7.31950045e-01 -3.35889339e-01 5.56769967e-01 -3.23501021e-01 5.29777169e-01 -2.18980312e-02 1.09423231e-02 -1.28209674e+00 -2.54601210e-01 -5.49906850e-01 7.97970414e-01 -1.27295572e-02 -1.68895781e+00 6.42165780e-01 2.75325328e-01 -6.86168075e-01 -5.28058290e-01 -5.54034889e-01 -4.90971923e-01 4.51512814e-01 -1.49396861e+00 -1.43106282e+00 1.69758379e-01 4.18982208e-01 2.32375003e-02 -1.79287903e-02 1.24544191e+00 4.01189148e-01 -6.27798498e-01 9.27335680e-01 5.84396394e-03 3.84841114e-01 4.43010807e-01 -1.59218550e+00 8.81034315e-01 6.18734658e-01 6.05397701e-01 1.02799714e+00 7.85356879e-01 -6.08722270e-01 -1.41307187e+00 -1.05757046e+00 1.29929304e+00 -4.70000952e-01 5.87243140e-01 -2.52678335e-01 -1.09944463e+00 1.44991827e+00 2.80315846e-01 4.05555815e-02 9.94710326e-01 1.01940238e+00 -7.70549715e-01 3.17066580e-01 -1.08240294e+00 2.71590948e-01 1.26347685e+00 -3.33142817e-01 -1.26027989e+00 1.92128435e-01 7.41271436e-01 -4.56366241e-01 -1.18159568e+00 5.47644734e-01 4.88770843e-01 -5.37317991e-01 1.25972152e+00 -1.14382684e+00 2.43958175e-01 5.44344774e-03 -2.84908175e-01 -1.38876152e+00 -5.57019949e-01 -4.94044006e-01 -9.29305434e-01 1.24484527e+00 7.24866271e-01 -6.34500623e-01 6.08098686e-01 5.19825399e-01 2.77984571e-02 -8.39431643e-01 -1.02567351e+00 -1.00608397e+00 3.06683660e-01 -1.15996957e-01 8.35463345e-01 1.09876812e+00 5.26056349e-01 6.81966305e-01 -3.29947442e-01 4.07740295e-01 5.92365384e-01 4.29635167e-01 4.64410990e-01 -1.66477036e+00 -1.90385401e-01 -3.36194217e-01 -8.23099732e-01 -8.82600605e-01 7.79472589e-01 -1.29201162e+00 -3.10915619e-01 -1.86411643e+00 5.62716722e-01 -3.10097426e-01 -6.79952383e-01 7.78392434e-01 -3.56241703e-01 1.89131528e-01 8.10363814e-02 -2.37244144e-01 -7.78908908e-01 5.04163325e-01 5.70309758e-01 -3.00562680e-01 1.43832108e-02 -2.00037524e-01 -9.12689686e-01 7.20671356e-01 5.18323600e-01 -4.81920481e-01 -1.53173417e-01 -4.49689448e-01 3.96966368e-01 -1.10094075e-03 1.11923143e-02 -5.35232425e-01 3.87370437e-01 4.23258282e-02 5.92631586e-02 -6.56572402e-01 4.19819057e-01 -6.59156919e-01 9.07511115e-02 3.55274916e-01 -6.57841861e-01 -3.89896408e-02 2.24945098e-01 7.81455755e-01 -1.91800773e-01 -2.48834029e-01 3.02280426e-01 -5.23926094e-02 -6.29517913e-01 2.08352789e-01 -1.85143143e-01 1.13436408e-01 9.62130427e-01 2.87994951e-01 -5.28443813e-01 -3.35288316e-01 -8.15655828e-01 3.51468980e-01 2.07082286e-01 4.81653988e-01 4.22037661e-01 -1.28479981e+00 -8.25824797e-01 -2.08032191e-01 1.83051005e-01 -4.04039383e-01 1.12373650e-01 6.73912585e-01 -5.36946468e-02 8.69939446e-01 3.64075415e-02 -2.45822564e-01 -1.28713036e+00 7.46043801e-01 1.88321155e-02 -9.32859957e-01 -6.48615003e-01 7.56998658e-01 -1.41118601e-01 -7.68937409e-01 1.83260590e-01 -2.94668466e-01 -5.29097259e-01 1.36440247e-01 3.84316713e-01 3.13699752e-01 1.65883273e-01 -7.20337152e-01 -7.17622042e-01 3.11978310e-01 -3.12877774e-01 3.99575353e-01 1.73935819e+00 -5.09587973e-02 -2.05223113e-01 4.72921968e-01 8.82813513e-01 1.05825312e-01 -5.06529927e-01 -5.27470350e-01 5.72343528e-01 -3.24696511e-01 -2.59172022e-01 -7.93959677e-01 -1.07561505e+00 5.26774228e-01 2.38022450e-02 2.99027294e-01 6.89737678e-01 4.90709692e-01 6.02775097e-01 8.33371401e-01 5.56472659e-01 -8.42517614e-01 -3.71659815e-01 7.55967617e-01 5.98601162e-01 -1.01925218e+00 2.02847615e-01 -9.62278545e-01 -4.55768824e-01 8.90242517e-01 5.68341851e-01 -2.73941219e-01 6.37400091e-01 3.41459066e-01 -5.23310840e-01 -4.77153093e-01 -1.24489832e+00 -6.59671068e-01 4.47738588e-01 7.18275309e-01 7.39593565e-01 2.24242449e-01 -5.47367632e-01 8.74881148e-01 -1.06949992e-01 -3.10620815e-01 5.78442693e-01 6.85826898e-01 -2.13151604e-01 -1.29616702e+00 1.43605560e-01 5.37896693e-01 -6.93102241e-01 -2.52906024e-01 -6.63400173e-01 1.03447914e+00 2.09960081e-02 9.49167788e-01 -2.12529730e-02 -3.65082562e-01 3.85052979e-01 2.42896914e-01 6.62616968e-01 -8.09027851e-01 -7.71039605e-01 -4.05938447e-01 5.41040659e-01 -6.09372854e-01 -6.44155085e-01 -5.98737776e-01 -1.32896698e+00 -2.69705683e-01 -5.82635343e-01 2.76777327e-01 1.85645714e-01 1.03376234e+00 4.83256817e-01 8.01427960e-01 2.53296107e-01 -4.47106004e-01 -3.25409830e-01 -1.08914936e+00 -6.70032322e-01 3.25765073e-01 9.85373557e-02 -9.81278062e-01 2.65609147e-03 -3.62480253e-01]
[9.252662658691406, 8.541504859924316]
0b0aefcf-7e57-4276-884b-e064171de618
automatic-textual-evidence-mining-in-covid-19
2004.12563
null
https://arxiv.org/abs/2004.12563v3
https://arxiv.org/pdf/2004.12563v3.pdf
Automatic Textual Evidence Mining in COVID-19 Literature
We created this EVIDENCEMINER system for automatic textual evidence mining in COVID-19 literature. EVIDENCEMINER is a web-based system that lets users query a natural language statement and automatically retrieves textual evidence from a background corpora for life sciences. It is constructed in a completely automated way without any human effort for training data annotation. EVIDENCEMINER is supported by novel data-driven methods for distantly supervised named entity recognition and open information extraction. The named entities and meta-patterns are pre-computed and indexed offline to support fast online evidence retrieval. The annotation results are also highlighted in the original document for better visualization. EVIDENCEMINER also includes analytic functionalities such as the most frequent entity and relation summarization.
['Aabhas Chauhan', 'Xuan Wang', 'Yingjun Guan', 'Weili Liu', 'Jiawei Han']
2020-04-27
null
null
null
null
['open-information-extraction']
['natural-language-processing']
[ 1.39394164e-01 3.78196687e-01 -1.16003191e+00 -2.06851382e-02 -7.34427035e-01 -7.36221731e-01 7.04957724e-01 1.48981237e+00 -5.85372686e-01 1.20807958e+00 4.62509274e-01 -6.29088640e-01 -4.66292709e-01 -6.91888571e-01 -5.59780717e-01 -1.20153986e-01 -1.49371609e-01 6.28436148e-01 2.78137058e-01 1.45424932e-01 9.12933350e-01 5.12015462e-01 -1.45533979e+00 4.86252248e-01 1.10924041e+00 7.19636798e-01 5.93352318e-02 6.52139962e-01 -6.36521637e-01 1.07606733e+00 -6.30003333e-01 -6.93404317e-01 -2.11271465e-01 8.80990326e-02 -8.66848588e-01 -2.85088152e-01 -4.82373685e-02 1.23413086e-01 -2.41516139e-02 1.07081127e+00 3.25116903e-01 -3.71720612e-01 7.99649835e-01 -1.18749011e+00 -1.71227515e-01 9.14931417e-01 -6.26463890e-01 6.33845806e-01 7.88046241e-01 -7.72525594e-02 8.68078232e-01 -1.42281222e+00 1.52245903e+00 7.56827235e-01 3.12429756e-01 -2.06030339e-01 -6.50600076e-01 -6.49185598e-01 -1.78773209e-01 7.10076153e-01 -1.29999232e+00 -1.11010998e-01 4.32436109e-01 -5.13114631e-01 1.38343489e+00 5.40574193e-01 8.10044825e-01 8.20241928e-01 3.81894320e-01 8.18525970e-01 7.17305660e-01 -7.53386199e-01 3.85913491e-01 1.93862751e-01 7.73904145e-01 9.62976515e-01 9.34945405e-01 -3.79204363e-01 -1.00998116e+00 -8.29692781e-01 -1.39586225e-01 -9.52868760e-02 1.64605796e-01 3.43917161e-01 -1.22761309e+00 5.05743146e-01 -3.07715148e-01 4.38283950e-01 -9.78964627e-01 -3.97136956e-01 1.16101313e+00 1.52439177e-01 5.20781219e-01 5.08969963e-01 -7.86954284e-01 1.37126120e-02 -1.14125419e+00 2.14291453e-01 9.45725560e-01 1.10082841e+00 3.52510899e-01 -4.95092005e-01 -1.96174592e-01 7.42757797e-01 4.40297484e-01 2.07718760e-01 7.59257615e-01 -6.53277636e-01 5.45662165e-01 1.39299405e+00 8.06200728e-02 -1.07114065e+00 -3.49790484e-01 -4.44252836e-03 -3.55418682e-01 -3.89601141e-01 3.97430323e-02 3.03647593e-02 -6.49702907e-01 9.19300675e-01 6.46070063e-01 -1.79332588e-02 2.30876759e-01 2.97814935e-01 1.52819717e+00 4.62934583e-01 3.94708306e-01 -7.59180784e-01 1.94270623e+00 -6.19027734e-01 -1.49925363e+00 2.24659964e-01 7.43913710e-01 -8.24076056e-01 5.73530197e-01 7.89848983e-01 -1.05542481e+00 2.24942639e-01 -1.37318635e+00 -1.38475105e-01 -1.22612250e+00 1.89727083e-01 7.06877530e-01 7.90903121e-02 -3.04965228e-01 2.55738497e-01 -4.21944737e-01 -2.59673297e-01 5.92621744e-01 1.56885386e-01 -7.42531776e-01 1.38555899e-01 -1.07681990e+00 1.10555983e+00 1.18617916e+00 -3.16171765e-01 -6.20910645e-01 -8.14559937e-01 -1.02861309e+00 -9.25674066e-02 1.14318228e+00 -6.60042048e-01 9.87263381e-01 -1.78201497e-02 -7.57304847e-01 1.19736278e+00 -2.75014162e-01 -7.06431150e-01 -2.88112983e-02 -3.24834079e-01 -8.07363093e-01 5.29879451e-01 4.31278944e-01 -1.71708748e-01 1.80060014e-01 -7.60822773e-01 -5.70645154e-01 -6.11037850e-01 -8.06777358e-01 -1.16697446e-01 -1.84539035e-01 6.80923700e-01 -6.19166672e-01 -9.37012374e-01 -1.35963678e-01 -3.84553730e-01 -3.06165665e-01 -2.65541285e-01 -7.95997202e-01 -8.04177642e-01 6.62618220e-01 -7.69949198e-01 1.73518074e+00 -1.51398337e+00 -5.24484873e-01 5.87807357e-01 3.25171441e-01 5.31199053e-02 4.03341502e-01 8.00706327e-01 -1.91705823e-01 4.41459447e-01 -2.30053261e-01 4.64694828e-01 -2.54925251e-01 2.82862276e-01 -3.03486764e-01 4.07098919e-01 9.55516696e-02 8.45989585e-01 -1.10070419e+00 -1.57629347e+00 2.91105751e-02 -2.44541481e-01 -1.84376106e-01 8.78500752e-03 -6.87557697e-01 -1.98681638e-01 -6.47159934e-01 1.08265769e+00 2.65523583e-01 -3.47075492e-01 5.13004541e-01 -1.13248184e-01 -5.46456754e-01 5.43085456e-01 -1.21349156e+00 1.61479652e+00 4.87541184e-02 4.39805120e-01 -1.96842730e-01 -9.29255247e-01 7.50534594e-01 6.42551899e-01 6.41974568e-01 -3.88208628e-01 1.17637374e-01 5.53771079e-01 -4.21007603e-01 -1.12331676e+00 6.79462552e-01 2.77521282e-01 -1.63377762e-01 5.13956308e-01 2.17937455e-01 2.39568442e-01 9.11890745e-01 7.25976408e-01 1.34288073e+00 -1.42421931e-01 1.36078238e+00 -2.09912032e-01 4.95505303e-01 6.74658597e-01 6.01216555e-01 5.45899868e-01 6.00640416e-01 -2.26856232e-01 1.35277033e-01 -4.98129338e-01 -7.52276838e-01 -7.94013143e-01 -3.91882271e-01 8.84682536e-01 -2.05238730e-01 -1.13291061e+00 -1.93304762e-01 -8.24677885e-01 1.00614794e-01 8.36788535e-01 -5.89352250e-01 2.58343369e-01 -3.17984343e-01 -6.08904362e-01 6.42483413e-01 1.54417321e-01 2.48604581e-01 -1.34846520e+00 -8.97510171e-01 4.85427082e-01 -3.91437680e-01 -9.08352971e-01 4.86521646e-02 3.87233943e-01 -6.07231081e-01 -1.63765764e+00 -1.61884382e-01 -4.06635612e-01 4.98814523e-01 -3.87966365e-01 1.14951038e+00 2.98012406e-01 -6.59612417e-01 2.54530367e-02 -3.80922049e-01 -1.29857600e+00 -5.26032150e-01 -2.23772958e-01 -3.32075618e-02 -6.70732260e-01 6.63675725e-01 -2.95757741e-01 -4.19742286e-01 -1.57419667e-01 -1.02005780e+00 -2.58495867e-01 6.76962435e-01 6.25320196e-01 1.04230189e+00 3.76322097e-03 7.79656172e-01 -1.06505990e+00 1.10545301e+00 -9.80439067e-01 -4.90234494e-01 6.63037837e-01 -1.23383844e+00 1.44133061e-01 1.42993718e-01 -1.80456176e-01 -8.54444206e-01 -1.83601379e-01 -1.38109371e-01 4.03707586e-02 -1.99129418e-01 1.45032716e+00 -1.95408110e-02 8.11866045e-01 6.68057919e-01 1.18265085e-01 -3.60054910e-01 -4.36165392e-01 4.52728897e-01 9.41826165e-01 8.71349096e-01 -4.59233642e-01 2.07254350e-01 2.05914095e-01 1.25174344e-01 -7.38395929e-01 -9.12169278e-01 -8.84119928e-01 -4.29231942e-01 -3.09835941e-01 6.54497027e-01 -5.99574149e-01 -9.48378861e-01 -2.97025084e-01 -1.27315319e+00 3.52086514e-01 -4.67911720e-01 5.26121140e-01 -9.11551341e-02 4.50930804e-01 -4.29004699e-01 -6.68687344e-01 -9.93277371e-01 -2.52134502e-01 6.61536992e-01 1.92202434e-01 -6.90475643e-01 -6.81636810e-01 4.68448579e-01 1.36964977e-01 -4.49500918e-01 4.85274166e-01 1.02332318e+00 -1.70447314e+00 -6.56660274e-03 -5.18637061e-01 1.40784174e-01 -4.44046289e-01 1.46910492e-02 4.32727218e-01 -4.10300225e-01 6.54585183e-01 -3.38659078e-01 -2.71754801e-01 6.40484452e-01 -3.44136059e-02 1.34535742e+00 -1.19752669e+00 -9.71516013e-01 -2.56309092e-01 1.19359434e+00 3.27312946e-01 3.80401075e-01 8.19778681e-01 9.84953791e-02 7.15336502e-01 1.17105150e+00 7.07963526e-01 1.78305641e-01 3.56095672e-01 1.50722519e-01 2.46311322e-01 4.31242406e-01 -2.39101961e-01 -2.80335665e-01 7.00030327e-01 -9.25478041e-02 -3.13299298e-01 -1.12556100e+00 8.22815478e-01 -1.88206756e+00 -1.45352614e+00 -7.01137006e-01 1.61232126e+00 1.58439291e+00 4.05834764e-01 7.35821277e-02 4.90165085e-01 5.43269515e-01 -4.32400495e-01 -5.32249033e-01 -5.88094294e-01 -2.94036180e-01 4.70903099e-01 3.89519483e-01 7.42788613e-02 -6.16818130e-01 7.32220054e-01 6.37941217e+00 9.73614633e-01 -4.19971675e-01 8.58694166e-02 2.99496382e-01 -1.26954494e-02 -2.45974526e-01 1.48807392e-01 -7.06901550e-01 4.97096747e-01 1.00814009e+00 -7.72934496e-01 -6.28802776e-01 1.05440032e+00 3.77757490e-01 -6.91446841e-01 -7.27902234e-01 6.36400521e-01 -1.31799623e-01 -2.29635763e+00 2.08828658e-01 3.04398805e-01 3.80199909e-01 1.36314601e-01 -6.68672323e-01 -1.60455525e-01 4.85051185e-01 -5.61780035e-01 7.33097136e-01 6.59353495e-01 6.09693348e-01 -6.77534044e-01 7.80753672e-01 6.07752085e-01 -9.35381770e-01 -7.19307810e-02 -4.07806896e-02 2.68808544e-01 2.27794051e-01 1.18557298e+00 -1.40056145e+00 8.72273505e-01 6.27839684e-01 6.58298254e-01 -6.55131221e-01 1.26457047e+00 -4.53954041e-01 8.91696751e-01 -3.64863485e-01 -3.34184617e-01 -2.00051934e-01 5.70563339e-02 8.05498183e-01 1.90906477e+00 5.07510193e-02 3.41168195e-01 -3.72783616e-02 4.97615546e-01 -3.00059706e-01 6.72321320e-01 -8.10732484e-01 -3.19488883e-01 8.75496924e-01 1.35455370e+00 -9.96161878e-01 -9.92049456e-01 -3.77287604e-02 2.38514647e-01 1.15451314e-01 -4.21264857e-01 -4.05112058e-01 -5.24998248e-01 -2.57306904e-01 1.45038977e-01 1.30190372e-01 6.69636726e-02 -3.06433231e-01 -8.20441902e-01 -2.57624120e-01 -8.80920172e-01 1.32522690e+00 -8.04063559e-01 -1.19798756e+00 1.34115905e-01 3.51756305e-01 -9.63143885e-01 -5.09016454e-01 -4.88857001e-01 -5.93125641e-01 7.38796368e-02 -9.21046615e-01 -8.02806199e-01 2.74269581e-01 6.35563433e-01 7.03144789e-01 -1.17334507e-01 8.97160470e-01 9.89749953e-02 -6.06534541e-01 1.35604739e-01 -1.82341293e-01 2.88709477e-02 6.56169713e-01 -1.32672429e+00 -9.32373703e-02 6.70058548e-01 2.49778271e-01 1.12126052e+00 1.01766837e+00 -1.44551778e+00 -1.37923396e+00 -7.35317886e-01 1.36737621e+00 -3.08865964e-01 8.70699346e-01 8.10680687e-02 -9.32903111e-01 3.60021561e-01 7.11410105e-01 -1.05907328e-01 1.43354797e+00 8.39505717e-02 -3.38913530e-01 4.12612498e-01 -1.25407052e+00 6.34644926e-01 8.42442334e-01 -8.37363303e-02 -1.29089057e+00 5.07657051e-01 4.40833926e-01 -1.31881014e-01 -1.17801738e+00 3.53782803e-01 4.41034496e-01 -1.24794580e-01 7.69106209e-01 -8.89769435e-01 6.37113690e-01 -4.50066239e-01 1.57289430e-01 -4.16503340e-01 3.47681791e-01 -6.22332096e-01 -8.51583004e-01 1.20836174e+00 9.49616432e-01 -9.23675895e-02 3.66292834e-01 3.63150179e-01 -2.39668757e-01 -8.25243592e-01 -8.61020207e-01 -4.57552910e-01 -6.57226682e-01 -6.74091160e-01 4.08509254e-01 1.16139698e+00 1.04827464e+00 5.46314180e-01 2.36036226e-01 -4.50659133e-02 4.70770270e-01 4.38239813e-01 4.71541196e-01 -1.47720861e+00 3.72350425e-01 -3.06357652e-01 -7.05014914e-02 2.97356658e-02 5.15712844e-03 -1.28125715e+00 -1.91401824e-01 -1.89577198e+00 6.63421333e-01 -3.07569765e-02 -2.82216370e-01 7.35225916e-01 2.09059920e-02 -1.55600742e-01 -7.34675467e-01 3.05720389e-01 -9.89909232e-01 8.77904221e-02 6.13069117e-01 -1.67659804e-01 -1.01161592e-01 -4.48180467e-01 -7.16912687e-01 9.99645293e-01 7.42019296e-01 -1.13994813e+00 -7.29213506e-02 5.30813217e-01 9.50172544e-01 3.06337010e-02 4.99136299e-02 -3.64823014e-01 5.96365094e-01 -4.92092252e-01 3.39452028e-01 -1.52098346e+00 -3.06122780e-01 -6.61878169e-01 1.63404569e-01 6.68942809e-01 -4.64086920e-01 4.38981920e-01 3.26834351e-01 6.44989967e-01 -4.94051836e-02 -7.94246912e-01 8.71641114e-02 -2.62202054e-01 -4.73262906e-01 -8.59663039e-02 -9.88590419e-01 -1.37824982e-01 9.25575793e-01 -2.27533072e-01 -7.30988801e-01 1.70304179e-01 -6.45020843e-01 2.42910802e-01 1.81651637e-02 2.48571470e-01 8.83415341e-01 -1.01366138e+00 -6.86260819e-01 -5.37273705e-01 6.28555417e-01 -8.08898956e-02 -2.50878274e-01 9.58458066e-01 -3.40293378e-01 5.32439113e-01 -6.01431297e-04 -5.63714206e-02 -1.66246450e+00 1.00518250e+00 -5.33954382e-01 -7.89742291e-01 -7.86247492e-01 1.64620459e-01 -8.06711137e-01 -9.88081396e-02 1.72082201e-01 -2.34290868e-01 -6.85970962e-01 2.56045341e-01 8.58214259e-01 6.26512110e-01 2.51634628e-01 -1.42514601e-01 -6.88167989e-01 -2.53631860e-01 -3.45895022e-01 -2.99372584e-01 1.76331902e+00 1.10887893e-01 -5.49266458e-01 4.38364536e-01 5.49894750e-01 6.84869170e-01 1.04410678e-01 -5.30038998e-02 1.02536345e+00 8.06033090e-02 -2.54900515e-01 -1.27559817e+00 -2.54704684e-01 -5.24140298e-02 -5.04648946e-02 1.78978592e-01 6.45455241e-01 5.13302922e-01 1.73752099e-01 7.41713107e-01 -6.32212088e-02 -1.44845688e+00 8.47900212e-02 3.60268146e-01 1.16297352e+00 -9.11125422e-01 7.79127300e-01 -5.87548018e-01 -4.25764710e-01 1.07383060e+00 1.93876684e-01 3.15886915e-01 8.36433887e-01 8.02277863e-01 -5.35157144e-01 -1.04138911e+00 -1.21540940e+00 1.92376170e-02 5.15776277e-01 2.29309946e-01 4.48242664e-01 -2.59263039e-01 -1.18112814e+00 1.22941816e+00 -1.17811583e-01 2.68443316e-01 3.47377241e-01 1.34162605e+00 -4.66040194e-01 -9.68367279e-01 -4.07512724e-01 8.25073481e-01 -1.08424330e+00 -2.87772506e-01 -9.66281950e-01 7.69477963e-01 2.64105648e-01 1.04220128e+00 -3.35465729e-01 8.05985406e-02 2.55887032e-01 2.94034630e-01 1.81568220e-01 -6.70647323e-01 -8.20098877e-01 1.08784124e-01 6.70861900e-01 -2.78964907e-01 -5.57443917e-01 -7.61578679e-01 -1.83950758e+00 2.28709862e-01 -6.84394062e-01 6.07246220e-01 1.05117607e+00 1.12226820e+00 5.46133876e-01 2.50954598e-01 -9.24614351e-03 1.75450802e-01 1.87788039e-01 -1.16878891e+00 -1.42049687e-02 -1.09848909e-01 -2.31456757e-01 -6.39734685e-01 1.33213595e-01 4.00657356e-01]
[8.761971473693848, 8.660676956176758]
ca851793-3bdb-47f8-8e1e-ac651f08ffdd
adjusted-asymmetric-accuracy-a-well-behaving
2209.02935
null
https://arxiv.org/abs/2209.02935v1
https://arxiv.org/pdf/2209.02935v1.pdf
Adjusted Asymmetric Accuracy: A Well-Behaving External Cluster Validity Measure
There is no, nor will there ever be, single best clustering algorithm, but we would still like to be able to pinpoint those which are well-performing on certain task types and filter out the systematically disappointing ones. Clustering algorithms are traditionally evaluated using either internal or external validity measures. Internal measures quantify different aspects of the obtained partitions, e.g., the average degree of cluster compactness or point separability. Yet, their validity is questionable because the clusterings they promote can sometimes be meaningless. External measures, on the other hand, compare the algorithms' outputs to the reference, ground truth groupings that are provided by experts. The commonly-used classical partition similarity scores, such as the normalised mutual information, Fowlkes-Mallows, or adjusted Rand index, might not possess all the desirable properties, e.g., they do not identify pathological edge cases correctly. Furthermore, they are not nicely interpretable: it is hard to say what a score of 0.8 really means. Its behaviour might also vary as the number of true clusters changes. This makes comparing clustering algorithms across many benchmark datasets difficult. To remedy this, we propose and analyse a new measure: an asymmetric version of the optimal set-matching accuracy. It is corrected for chance and the imbalancedness of cluster sizes.
['Marek Gagolewski']
2022-09-07
null
null
null
null
['set-matching']
['computer-vision']
[-9.89952385e-02 4.11444418e-02 -1.14232786e-01 -3.65256518e-01 -4.51675087e-01 -8.89934123e-01 4.82453346e-01 7.61329055e-01 -4.10528213e-01 6.51008964e-01 -1.32060319e-01 -3.03997368e-01 -6.42353594e-01 -8.21500242e-01 -1.45967707e-01 -1.10545444e+00 6.46145344e-02 8.48977327e-01 4.89319652e-01 1.29438251e-01 4.70216364e-01 2.63421863e-01 -1.76088870e+00 1.30056098e-01 1.18372834e+00 8.70492160e-01 -5.89151727e-03 2.02665105e-01 2.92508248e-02 3.18196684e-01 -7.14104831e-01 -4.75032985e-01 1.65300872e-02 -5.36595643e-01 -9.16707218e-01 1.82155222e-02 -2.41754442e-01 3.92684102e-01 4.91780967e-01 1.18687034e+00 2.33025700e-01 -1.59865707e-01 1.06574190e+00 -1.37015951e+00 -1.84292182e-01 7.28729784e-01 -5.82309902e-01 1.28262877e-01 4.60184366e-01 5.64914793e-02 1.07560563e+00 -3.62275243e-01 6.52434170e-01 9.37603295e-01 8.35197568e-01 4.22350019e-02 -1.52022314e+00 -3.97011012e-01 -2.28267342e-01 1.36890545e-01 -1.45056856e+00 -1.55891642e-01 3.96210819e-01 -6.93855286e-01 3.71755451e-01 7.38767803e-01 5.66971064e-01 5.32010376e-01 4.52072248e-02 1.38933182e-01 1.46889579e+00 -3.26853156e-01 4.90707695e-01 4.06105995e-01 3.70130618e-03 2.43332729e-01 8.23498428e-01 -1.51989326e-01 -4.28648293e-02 -1.93910629e-01 1.89209491e-01 -2.39500582e-01 -5.42384386e-01 -8.52997959e-01 -1.36289084e+00 7.48988688e-01 3.55049878e-01 8.96151185e-01 -2.01234937e-01 -3.07326645e-01 5.55131316e-01 4.30890203e-01 3.15716088e-01 7.75611162e-01 -3.00806463e-01 -1.63499847e-01 -1.13918900e+00 1.35084078e-01 7.00754106e-01 4.83506382e-01 6.33832693e-01 -5.72746754e-01 3.45290899e-02 5.63923717e-01 1.42749399e-01 -3.67576405e-02 7.51796186e-01 -8.28418255e-01 -1.44914463e-01 9.15293992e-01 4.90015484e-02 -1.38947785e+00 -6.90589309e-01 -3.50416929e-01 -1.09091866e+00 1.85438409e-01 8.69158089e-01 1.66909993e-01 -3.63049001e-01 1.61131239e+00 2.36787096e-01 -2.81795681e-01 -1.00909531e-01 1.09289598e+00 7.11286306e-01 1.26452506e-01 -7.63695315e-02 -4.82619435e-01 1.25521529e+00 -3.72593254e-01 -6.47884548e-01 -4.50376831e-02 8.09352994e-01 -9.79646504e-01 9.74486709e-01 4.27027583e-01 -1.07129061e+00 -4.01439667e-01 -9.27698255e-01 6.24124110e-01 -4.95207250e-01 -1.76141948e-01 3.84051383e-01 8.23513687e-01 -9.37109470e-01 9.61015642e-01 -7.23985255e-01 -3.80709231e-01 4.29127999e-02 2.98496336e-01 -4.95334297e-01 1.65134907e-01 -8.64584267e-01 9.14340019e-01 7.22340465e-01 -1.87779501e-01 -1.15532666e-01 -3.21924567e-01 -3.12101215e-01 2.43678078e-01 3.95350784e-01 -5.06873071e-01 7.15596676e-01 -1.08557451e+00 -8.22413623e-01 1.09374452e+00 1.46567628e-01 -2.56990582e-01 6.55065835e-01 4.79416639e-01 -2.92959362e-01 6.43622968e-03 2.71782517e-01 3.07875276e-01 3.79281729e-01 -1.55884933e+00 -3.59171271e-01 -5.59005618e-01 -2.84078479e-01 9.11785811e-02 -7.78217763e-02 -1.20137610e-01 -1.42278194e-01 -5.50385654e-01 5.83365262e-01 -7.42738068e-01 -2.45045647e-01 -3.66721988e-01 -5.34992516e-01 -1.69506013e-01 3.27282012e-01 -2.57593066e-01 1.51068020e+00 -2.10318184e+00 -5.87321855e-02 7.12209046e-01 3.34166735e-01 2.11149380e-01 3.04079711e-01 2.92578816e-01 -3.81342053e-01 4.26113695e-01 -3.58954132e-01 2.36258015e-01 1.27859905e-01 1.25105992e-01 2.88069576e-01 7.35187232e-01 -1.87338572e-02 4.78526562e-01 -1.08677423e+00 -6.57394290e-01 1.98154554e-01 2.19873443e-01 -3.75464201e-01 -1.73581749e-01 2.10135192e-01 3.94857317e-01 -8.00877213e-02 1.86851099e-01 6.20769203e-01 -4.39027071e-01 3.98425251e-01 -1.53496101e-01 6.59372807e-02 1.78345293e-01 -1.47319889e+00 9.67581570e-01 1.98098108e-01 5.98108888e-01 -1.74213037e-01 -1.24658155e+00 1.07434714e+00 4.09045219e-01 5.04435420e-01 -4.62763160e-01 2.96992332e-01 5.07816911e-01 5.14085591e-01 -2.97211289e-01 3.16131264e-01 -4.92765516e-01 1.98272858e-02 5.18041492e-01 -2.75222450e-01 -6.53214678e-02 3.69000793e-01 1.22720869e-02 1.24969304e+00 -4.42372203e-01 6.08045101e-01 -5.02043068e-01 4.98912513e-01 2.15406828e-02 5.71143329e-01 4.22972798e-01 -1.55431658e-01 8.60130310e-01 8.87180626e-01 -1.37176856e-01 -9.49450612e-01 -1.22698200e+00 -4.29300308e-01 4.51970905e-01 1.72332749e-01 -5.33041894e-01 -9.15294468e-01 -5.94087303e-01 -7.47149289e-02 3.90383720e-01 -4.95552301e-01 -2.24544838e-01 2.16888115e-01 -9.18217659e-01 4.23384279e-01 9.54915211e-02 1.72349080e-01 -7.71763802e-01 -9.74647760e-01 8.02893862e-02 -3.37657124e-01 -7.60528088e-01 4.85602804e-02 2.03001842e-01 -8.96956742e-01 -1.44947290e+00 -5.44380665e-01 -4.94438797e-01 6.90545499e-01 3.09683621e-01 1.47856593e+00 4.94411647e-01 7.01887757e-02 -1.61158144e-01 -6.47233427e-01 -9.48501304e-02 -6.45814419e-01 3.45241576e-02 9.89650488e-02 -5.74482977e-02 6.87557757e-01 -5.19766331e-01 -6.44090593e-01 9.09295619e-01 -9.82642651e-01 -4.55828518e-01 5.44477344e-01 7.13372231e-01 5.02048314e-01 5.54142714e-01 6.10887706e-01 -9.60391819e-01 7.45414138e-01 -3.07752371e-01 -7.81606808e-02 2.95032173e-01 -9.20987964e-01 -3.16614769e-02 6.45695269e-01 -2.47835070e-01 -5.60606420e-01 -2.44493321e-01 3.52009758e-02 -3.72126997e-02 -5.24779737e-01 4.62043703e-01 -8.55562985e-02 2.70479172e-01 9.21223044e-01 -1.83248118e-01 2.29115374e-02 -2.86229432e-01 1.41049862e-01 8.14450383e-01 5.01420796e-01 -3.55501622e-01 6.56061769e-01 5.09905815e-01 -1.39415175e-01 -5.95205486e-01 -6.57311261e-01 -9.50515747e-01 -7.19665349e-01 -1.93299547e-01 6.11775100e-01 -2.84727216e-01 -8.02348435e-01 2.05404654e-01 -8.75044644e-01 -4.96503934e-02 -3.50714058e-01 3.72039348e-01 -6.80063188e-01 7.84061849e-01 -1.49350286e-01 -7.47980833e-01 2.98130214e-02 -1.20572674e+00 7.39713192e-01 -1.16195902e-02 -8.98249149e-01 -9.87539709e-01 -1.31502366e-02 2.74534702e-01 2.64145672e-01 6.34460688e-01 1.02116275e+00 -8.45280111e-01 1.87733665e-01 -3.38337243e-01 -1.96195573e-01 2.07943276e-01 2.48903155e-01 3.40339780e-01 -8.47540677e-01 -3.34956735e-01 2.63095628e-02 2.15875611e-01 7.49045968e-01 4.57715034e-01 1.07923567e+00 -3.17148536e-01 -5.48518658e-01 1.19789183e-01 1.46589339e+00 2.77330071e-01 7.15139389e-01 4.41720366e-01 1.05969489e-01 9.94149387e-01 5.69289148e-01 3.62226129e-01 9.21237916e-02 8.01516056e-01 3.58268321e-01 -9.14307088e-02 3.41958970e-01 1.87344387e-01 5.54300845e-03 8.19219410e-01 -3.49251747e-01 -1.19758666e-01 -1.18443203e+00 4.58205670e-01 -1.76054084e+00 -1.12675357e+00 -6.76423550e-01 2.58444786e+00 6.32075727e-01 6.03419423e-01 5.31425595e-01 1.02447450e+00 9.02651966e-01 -2.50978947e-01 -8.05451125e-02 -4.82980818e-01 -1.10705815e-01 -2.41438732e-01 2.55728424e-01 1.84503838e-01 -9.06389296e-01 3.18789899e-01 5.70917368e+00 9.26258147e-01 -7.18792558e-01 1.86279006e-02 8.89403522e-01 2.03378916e-01 -1.23105861e-01 1.95804313e-01 -2.70735044e-02 6.82393610e-01 9.13385034e-01 -2.11971745e-01 -1.64440670e-03 6.33060753e-01 1.86204404e-01 -4.24834549e-01 -1.04929841e+00 9.42499340e-01 -7.72911459e-02 -8.76956105e-01 -4.40586150e-01 1.94450244e-01 5.27826190e-01 -2.80299544e-01 -1.70861080e-01 -2.56694347e-01 1.59937412e-01 -1.20613599e+00 5.57817101e-01 5.72692215e-01 5.10359824e-01 -8.66931558e-01 1.19102168e+00 4.30045784e-01 -7.63645291e-01 8.17647800e-02 -2.10537404e-01 -1.48886722e-02 -9.07116905e-02 1.00341725e+00 -6.60169721e-01 6.03225052e-01 8.55079651e-01 3.71596485e-01 -8.08768988e-01 1.51497984e+00 -1.31020881e-02 4.30281222e-01 -4.25778538e-01 -6.13473989e-02 1.51786014e-01 -3.72553200e-01 4.71253246e-01 1.06604755e+00 3.44715357e-01 -1.28538162e-01 -2.21873298e-01 7.96537519e-01 4.22282577e-01 3.10533673e-01 -7.48662233e-01 5.80378026e-02 4.36371028e-01 1.34219539e+00 -1.55467319e+00 -2.00327858e-01 -2.19237089e-01 5.80434680e-01 -5.86794317e-02 -1.24913000e-01 -4.70185101e-01 -3.04709047e-01 3.51313442e-01 4.45912272e-01 -5.02015091e-02 1.73986778e-01 -5.71474195e-01 -8.11484933e-01 6.73740506e-02 -9.80570793e-01 5.41383505e-01 -5.55266678e-01 -1.47283530e+00 5.33607244e-01 -1.32322192e-01 -1.44155872e+00 -1.30598634e-01 -4.23332542e-01 -5.22761285e-01 2.44621649e-01 -9.18837190e-01 -2.83681184e-01 -3.14964622e-01 1.64219990e-01 -1.92220569e-01 2.02601850e-01 6.64865911e-01 2.81879038e-01 -3.07170987e-01 5.36918461e-01 3.82515639e-01 3.27121094e-02 8.59789789e-01 -1.53664756e+00 -1.33056015e-01 5.29527664e-01 1.23075373e-01 4.88471597e-01 1.14338958e+00 -2.93808162e-01 -5.01581728e-01 -6.69489503e-01 8.65364313e-01 -6.03617966e-01 5.01494944e-01 -1.54132647e-02 -1.17551255e+00 1.68566316e-01 -2.50877827e-01 -2.84220099e-01 7.34513640e-01 2.27557883e-01 -2.40514904e-01 3.30032744e-02 -1.19511735e+00 3.57174367e-01 8.13720822e-01 1.35177672e-01 -5.14726341e-01 3.16906959e-01 6.86475337e-02 6.13823235e-02 -1.24021494e+00 4.67870712e-01 4.82220322e-01 -1.75117850e+00 9.71679091e-01 -3.99733871e-01 3.00293684e-01 -3.52111608e-01 3.08074635e-02 -1.32337713e+00 -4.36488420e-01 -1.84841782e-01 6.19132876e-01 1.49416590e+00 5.25860012e-01 -6.72816813e-01 7.91258752e-01 3.82291436e-01 2.34586284e-01 -8.83948624e-01 -1.01402581e+00 -1.08292842e+00 1.75006807e-01 -3.06471348e-01 7.64673829e-01 1.43183088e+00 5.34601212e-01 5.26108384e-01 1.95300862e-01 -2.12214082e-01 5.04002512e-01 1.86636403e-01 7.39122331e-01 -1.88134193e+00 -1.39238343e-01 -1.05464935e+00 -8.85941029e-01 -8.56430307e-02 -1.04191042e-01 -1.00162649e+00 -1.15472384e-01 -1.50371063e+00 2.63165891e-01 -7.94297814e-01 -1.64810181e-01 2.04120114e-01 -1.24051079e-01 2.82302767e-01 9.89558920e-02 4.43410724e-01 -7.28064716e-01 -4.24179882e-02 1.02707505e+00 -1.66007970e-02 -1.96992651e-01 7.62629807e-02 -6.99051559e-01 9.76874530e-01 1.03006101e+00 -5.80845594e-01 -2.33302653e-01 3.16878766e-01 6.77955151e-01 -1.60320342e-01 3.97109896e-01 -1.30378568e+00 1.62190691e-01 -5.90735711e-02 3.07734787e-01 -2.68604785e-01 -1.60524949e-01 -8.93067300e-01 5.57046473e-01 6.60782099e-01 -1.56564206e-01 1.51620895e-01 -2.33441308e-01 5.19332647e-01 -3.12136412e-01 -6.41872108e-01 8.56282890e-01 -2.81172246e-01 -2.33336329e-01 -1.75196916e-01 -4.11752045e-01 1.08606793e-01 1.23114645e+00 -7.41820335e-01 -2.50358701e-01 -3.39131027e-01 -6.94596231e-01 1.55863896e-01 1.09638703e+00 1.70933262e-01 1.83663413e-01 -1.17549253e+00 -6.11572504e-01 -1.77485257e-01 3.26929241e-01 1.07268719e-02 -1.01301372e-02 1.29414821e+00 -4.50809181e-01 2.22686917e-01 -1.02138884e-01 -7.98195362e-01 -1.51237917e+00 7.98273742e-01 3.34101230e-01 -3.94708008e-01 -3.96694183e-01 3.46457630e-01 7.53650218e-02 -4.67780054e-01 3.60288061e-02 -2.80646682e-01 -2.74392605e-01 4.59309638e-01 3.31686825e-01 5.48834085e-01 3.72234672e-01 -7.98951447e-01 -5.67749143e-01 5.99189937e-01 3.37091357e-01 2.05831930e-01 1.26447988e+00 -9.57890525e-02 -3.80947053e-01 5.82376838e-01 7.99789667e-01 -2.02738836e-01 -5.30651033e-01 1.73236951e-01 4.32689101e-01 -6.16400659e-01 -2.91316539e-01 -6.18225336e-01 -9.13927317e-01 8.95899415e-01 5.57103634e-01 1.13470936e+00 1.18403208e+00 7.01977536e-02 2.62521625e-01 4.57682312e-02 1.80246294e-01 -1.33729625e+00 -9.22789350e-02 2.45064516e-02 4.54688251e-01 -1.20118022e+00 3.78732458e-02 -4.25607145e-01 -5.83500326e-01 9.71286476e-01 1.90206230e-01 1.71167359e-01 5.29522359e-01 1.84201315e-01 -1.14590734e-01 -4.58367288e-01 -6.29822791e-01 -3.74661416e-01 1.92808494e-01 6.53204501e-01 6.88098729e-01 3.54872078e-01 -8.54691029e-01 4.40906942e-01 -6.18640721e-01 -3.12456161e-01 4.71423537e-01 3.42285782e-01 -6.14836693e-01 -8.30834210e-01 -7.12710142e-01 8.68467867e-01 -6.11283720e-01 5.18739223e-01 -6.99694753e-01 8.53906870e-01 2.58785635e-01 1.15101147e+00 1.73401073e-01 -5.64794421e-01 2.30834112e-01 5.24853244e-02 2.59583503e-01 -3.36578935e-01 -5.74340701e-01 -5.57841659e-02 9.17942449e-03 -2.90888429e-01 -8.34796131e-01 -7.26646543e-01 -1.11518788e+00 -6.44510448e-01 -7.07818449e-01 5.30300558e-01 3.50191355e-01 8.99830937e-01 2.03084033e-02 2.52558351e-01 4.43831891e-01 -5.02414405e-01 -2.31189698e-01 -8.53494763e-01 -6.28632247e-01 8.67545366e-01 -8.97955224e-02 -7.52187788e-01 -6.61607981e-01 -2.27786645e-01]
[7.6595845222473145, 4.515707492828369]
f91aeeb3-a5a7-4c68-9429-97059aff38ea
temporal-dynamics-of-coordinated-online
2301.06774
null
https://arxiv.org/abs/2301.06774v1
https://arxiv.org/pdf/2301.06774v1.pdf
Temporal Dynamics of Coordinated Online Behavior: Stability, Archetypes, and Influence
Large-scale online campaigns, malicious or otherwise, require a significant degree of coordination among participants, which sparked interest in the study of coordinated online behavior. State-of-the-art methods for detecting coordinated behavior perform static analyses, disregarding the temporal dynamics of coordination. Here, we carry out the first dynamic analysis of coordinated behavior. To reach our goal we build a multiplex temporal network and we perform dynamic community detection to identify groups of users that exhibited coordinated behaviors in time. Thanks to our novel approach we find that: (i) coordinated communities feature variable degrees of temporal instability; (ii) dynamic analyses are needed to account for such instability, and results of static analyses can be unreliable and scarcely representative of unstable communities; (iii) some users exhibit distinct archetypal behaviors that have important practical implications; (iv) content and network characteristics contribute to explaining why users leave and join coordinated communities. Our results demonstrate the advantages of dynamic analyses and open up new directions of research on the unfolding of online debates, on the strategies of coordinated communities, and on the patterns of online influence.
['Stefano Cresci', 'Giovanni Da San Martino', 'Preslav Nakov', 'Mauro Conti', 'Maurizio Tesconi', 'Leonardo Nizzoli', 'Serena Tardelli']
2023-01-17
null
null
null
null
['dynamic-community-detection', 'community-detection']
['graphs', 'graphs']
[-1.36463940e-01 -6.18719049e-02 -2.08454385e-01 1.54392153e-01 -3.73154203e-03 -1.15519559e+00 1.04852045e+00 7.14745820e-01 -1.38890058e-01 3.87566417e-01 4.33963716e-01 -6.72591150e-01 -3.93779814e-01 -8.00961018e-01 -3.32972467e-01 -4.56247181e-01 -6.52134418e-01 3.81250143e-01 5.50513089e-01 -4.25706863e-01 3.05500537e-01 3.46320093e-01 -1.19311023e+00 1.33610085e-01 8.25812161e-01 2.50651807e-01 -2.21066326e-01 7.05741704e-01 -7.12005198e-02 8.20551276e-01 -4.41000104e-01 -5.27813971e-01 7.45653063e-02 -5.89637399e-01 -5.71252286e-01 2.30477661e-01 -1.16306737e-01 4.51183952e-02 -3.61044616e-01 9.40907836e-01 7.74031505e-03 -1.70476794e-01 2.96208024e-01 -1.25480282e+00 -2.97423005e-01 9.16095734e-01 -6.16153777e-01 7.16442287e-01 3.63416642e-01 1.72927484e-01 1.32342029e+00 -2.45828480e-01 1.17236221e+00 1.18002295e+00 7.71807373e-01 -2.78326459e-02 -1.60497022e+00 -3.70154202e-01 6.00877643e-01 7.19035119e-02 -1.00549161e+00 -3.90667468e-01 8.15851390e-01 -8.21098208e-01 3.62037629e-01 3.78866315e-01 1.14782703e+00 1.03380358e+00 -6.74255267e-02 3.47340375e-01 1.13882077e+00 -2.63401091e-01 3.80126983e-02 6.03293777e-02 3.84695828e-01 4.52258319e-01 5.26202738e-01 -2.57294178e-01 -3.60946566e-01 -8.00650835e-01 4.27747488e-01 9.48938280e-02 1.66829489e-02 -1.79178879e-01 -1.15735102e+00 1.14183962e+00 2.86115617e-01 9.85916734e-01 -1.87897667e-01 -8.00860971e-02 4.96143460e-01 7.57058322e-01 5.30735612e-01 3.56272638e-01 -3.07240486e-01 -5.87398469e-01 -7.21110463e-01 3.09408486e-01 1.17669642e+00 2.84830868e-01 7.75175631e-01 -5.80915868e-01 3.57981622e-01 4.48965430e-01 3.35644856e-02 3.20065528e-01 2.39352081e-02 -7.39507258e-01 3.67343992e-01 1.04294240e+00 1.45825759e-01 -1.64516687e+00 -4.81114358e-01 -4.14255649e-01 -5.80954373e-01 -3.67330909e-01 9.22460377e-01 -2.43641704e-01 5.83730787e-02 1.61981702e+00 3.20922911e-01 -1.70558259e-01 -4.32937205e-01 4.06764925e-01 1.36521280e-01 2.15454340e-01 -2.25750640e-01 -5.22245824e-01 1.05871165e+00 -3.57363045e-01 -6.05551481e-01 1.58574238e-01 6.43602073e-01 -6.84118629e-01 7.79167712e-01 4.94730026e-02 -8.37247968e-01 7.27438405e-02 -6.06117129e-01 6.97366834e-01 -2.34055147e-01 -5.82650423e-01 9.40038681e-01 8.06579292e-01 -9.85174954e-01 6.34973526e-01 -8.82578731e-01 -6.71371102e-01 1.90689296e-01 -2.85540763e-02 3.75074856e-02 1.42401189e-01 -9.38192666e-01 4.93653595e-01 -3.40181917e-01 1.01408809e-01 -6.04997158e-01 -3.51255119e-01 -2.87946284e-01 -6.19439483e-02 7.31072485e-01 -2.29663681e-02 1.02129102e+00 -1.32934129e+00 -7.42935002e-01 7.22582459e-01 -2.27283567e-01 -2.18807444e-01 9.10994291e-01 4.13139820e-01 -4.83928204e-01 3.22866857e-01 1.80969968e-01 -5.10899246e-01 3.43380481e-01 -1.35375118e+00 -4.23061460e-01 -4.23964143e-01 3.99811156e-02 -2.20919505e-01 -3.77774864e-01 2.07437649e-01 -1.32509083e-01 -3.44342738e-01 2.43185431e-01 -1.09092343e+00 -4.17179793e-01 -4.13386971e-01 -3.66145998e-01 -1.99069276e-01 8.42426956e-01 -5.54841399e-01 1.60639369e+00 -2.12279797e+00 1.13051422e-01 6.35886192e-01 8.45687509e-01 -4.64374647e-02 1.17657326e-01 1.06078994e+00 4.42170143e-01 6.28739178e-01 1.87242806e-01 2.70238798e-02 8.50693602e-03 -9.33365338e-03 -2.58006871e-01 7.65567303e-01 -2.44247541e-01 8.20779681e-01 -9.25471246e-01 -2.38155603e-01 -1.39590427e-01 -1.13414556e-01 -5.44807076e-01 -2.03368127e-01 -2.10736722e-01 6.15833938e-01 -7.06651092e-01 7.17185616e-01 3.79909098e-01 -6.73291743e-01 9.62421000e-01 3.30201536e-01 -6.18788779e-01 2.58032411e-01 -8.71700764e-01 6.07415259e-01 3.22294086e-02 1.08986008e+00 7.12556064e-01 -9.23096240e-01 5.33049643e-01 4.09273505e-02 7.08382726e-01 -4.48264271e-01 2.55432397e-01 2.07004294e-01 6.11410856e-01 -3.97840291e-01 2.95334548e-01 1.59338146e-01 -9.72814634e-02 1.17103481e+00 -4.88456637e-01 3.39885116e-01 4.83503520e-01 6.52640939e-01 1.47058249e+00 -7.93821990e-01 2.06877857e-01 -3.51538897e-01 1.95664793e-01 2.44517967e-01 5.39805710e-01 1.00828016e+00 -6.42354846e-01 -1.03644632e-01 1.36163163e+00 -4.49428737e-01 -1.18429053e+00 -7.25887537e-01 1.01146959e-01 1.19635129e+00 2.35347509e-01 -5.76858878e-01 -4.79191899e-01 -3.80134821e-01 3.12118679e-01 -9.39877480e-02 -7.22644806e-01 2.69337088e-01 -6.90644503e-01 -9.35855389e-01 2.74663270e-01 -1.08597010e-01 1.56688005e-01 -7.66646981e-01 -1.36649564e-01 3.48596126e-01 -3.94207776e-01 -1.11703968e+00 -2.78760135e-01 -2.48268887e-01 -8.65200937e-01 -1.88417637e+00 -2.17560604e-01 -3.66039515e-01 6.01328313e-01 6.29657447e-01 1.10110641e+00 6.59814477e-01 -2.03401357e-01 6.39341772e-01 -3.99767399e-01 -1.37985483e-01 -7.38128781e-01 2.59191424e-01 2.27843761e-01 2.95332968e-01 2.19578013e-01 -1.04366291e+00 -5.47950029e-01 8.16951215e-01 -6.34169102e-01 -3.37150604e-01 2.68098801e-01 2.47156084e-01 -2.65723616e-01 1.89201862e-01 5.88277876e-01 -1.03897691e+00 9.95536566e-01 -8.92430246e-01 -8.31019878e-01 1.20044656e-01 -6.07434988e-01 -4.62023765e-01 4.25905466e-01 -6.11490250e-01 -7.89322376e-01 -6.37980878e-01 5.19228995e-01 2.54645556e-01 4.06162620e-01 7.58070469e-01 4.27522033e-01 -1.09417811e-01 7.29301155e-01 1.34716377e-01 4.09457773e-01 -3.88306081e-01 2.02512115e-01 6.31523788e-01 -2.49221906e-01 -4.77016330e-01 9.29136097e-01 9.67163801e-01 -3.04329932e-01 -1.09647357e+00 -3.61854732e-01 -4.55666721e-01 -6.15877330e-01 -5.74756384e-01 4.73239779e-01 -7.15572596e-01 -1.11319923e+00 4.54402000e-01 -8.02176476e-01 -5.03050566e-01 2.35493377e-01 1.57491550e-01 -2.15720180e-02 6.98448360e-01 -9.39235210e-01 -1.09414840e+00 2.66876400e-01 -6.71564460e-01 2.59038746e-01 1.12277135e-01 -5.34363270e-01 -1.23763001e+00 4.68912095e-01 4.52682316e-01 6.59532964e-01 4.85409081e-01 7.13672042e-01 -7.70467997e-01 -8.66519034e-01 -2.43525192e-01 -6.22238405e-02 -4.16575909e-01 1.38025761e-01 5.43603659e-01 -3.63328785e-01 -3.18336725e-01 -1.22998819e-01 -4.76093078e-03 5.27063072e-01 3.23745281e-01 4.35084879e-01 -5.46904206e-01 -6.23133302e-01 -1.23920357e-02 1.05418634e+00 -3.25419307e-02 1.69988275e-01 3.39239627e-01 3.36071730e-01 9.60353851e-01 7.60013685e-02 7.47990727e-01 3.67186129e-01 5.07331491e-01 3.03856999e-01 5.06044179e-02 4.34343159e-01 -2.12029845e-01 2.99188465e-01 9.66502190e-01 -3.17790061e-01 -8.39449167e-02 -9.80095148e-01 7.22659230e-01 -2.09931278e+00 -1.25079155e+00 -6.88905597e-01 2.02755928e+00 5.94166756e-01 4.16320622e-01 7.18301237e-01 4.90355343e-02 1.03707886e+00 4.37710166e-01 -1.78905472e-01 -1.40307382e-01 -2.17399254e-01 -4.35745716e-01 3.87034297e-01 5.01466453e-01 -6.77157938e-01 6.14397645e-01 7.27498913e+00 2.06727713e-01 -1.01598990e+00 2.32607007e-01 5.63810706e-01 -7.06493563e-04 -5.68676710e-01 4.05126363e-01 -5.42199731e-01 5.07797360e-01 8.44537079e-01 -3.74394625e-01 6.16370857e-01 5.38068295e-01 5.97613990e-01 -2.40302905e-01 -6.67395294e-01 3.10150623e-01 -2.24370241e-01 -1.38409936e+00 -7.15688348e-01 6.57351792e-01 9.77144182e-01 2.14223623e-01 -3.59815359e-01 -3.01854592e-02 7.37996280e-01 -6.02658510e-01 4.68854815e-01 3.00175041e-01 1.85357139e-01 -3.76648217e-01 3.15696627e-01 5.33946157e-01 -1.07941103e+00 -4.21554953e-01 -6.12385534e-02 -5.79617739e-01 2.98305809e-01 1.03278649e+00 -4.83452529e-01 4.46841344e-02 4.03367192e-01 8.41459930e-01 -7.22810090e-01 7.33568013e-01 1.66185424e-02 9.37990487e-01 -3.49610537e-01 -3.48386467e-01 2.87739396e-01 -4.76464361e-01 8.62085819e-01 8.56205702e-01 -1.89168185e-01 -1.63333207e-01 1.97142273e-01 9.09716904e-01 5.44323251e-02 -8.18899646e-02 -6.47091269e-01 -8.49027097e-01 5.63143790e-01 1.20690274e+00 -1.25195062e+00 -1.78307801e-01 -5.34992874e-01 4.50210840e-01 3.81140828e-01 2.70491064e-01 -4.77399111e-01 2.32321411e-01 7.22182751e-01 7.03701258e-01 3.01188141e-01 -7.83076227e-01 -1.68995947e-01 -1.48615658e+00 8.73298123e-02 -8.66598189e-01 4.31658804e-01 1.85721874e-01 -1.44864249e+00 1.68609425e-01 -3.26599121e-01 -7.84077346e-01 -7.63670057e-02 -4.11067307e-02 -1.00809109e+00 2.97421902e-01 -9.40980315e-01 -8.62817109e-01 -3.44954431e-01 5.14017165e-01 4.49211486e-02 -5.31855822e-02 3.37756544e-01 1.61963999e-01 -5.87248623e-01 7.44106770e-02 3.26842546e-01 1.97697252e-01 3.29529524e-01 -8.92466486e-01 2.28078768e-01 7.54307687e-01 9.35336649e-02 7.44793236e-01 7.56797433e-01 -1.04714382e+00 -1.50308204e+00 -5.25506675e-01 1.15722179e+00 -6.49761558e-01 1.57087684e+00 -8.26327801e-01 -6.97836220e-01 6.47966623e-01 -2.94889770e-02 -1.40424699e-01 7.64908373e-01 8.18984807e-01 -3.73285353e-01 1.73216179e-01 -8.24649751e-01 7.97308862e-01 1.20703983e+00 -6.07340395e-01 -1.78301260e-01 6.46335959e-01 4.82810020e-01 3.93143535e-01 -6.30168974e-01 -2.60038823e-01 8.01323652e-01 -1.28836095e+00 4.24240172e-01 -4.45785075e-01 2.90440500e-01 4.71750535e-02 1.18868411e-01 -8.09314549e-01 -5.07479250e-01 -1.01481271e+00 5.72771542e-02 1.20759797e+00 4.41104352e-01 -8.73181641e-01 7.48741746e-01 5.01604617e-01 3.79223496e-01 -4.57839072e-01 -8.47985566e-01 -5.29410243e-01 -5.90027757e-02 -9.04471278e-02 7.43683055e-02 1.38587439e+00 3.28936130e-01 1.10164598e-01 -1.08098835e-01 -2.66316205e-01 5.40314496e-01 2.23053053e-01 9.85570490e-01 -1.61759388e+00 -3.74683768e-01 -7.51485825e-01 -3.76675695e-01 -5.70404828e-01 -1.35137215e-01 -3.62955779e-01 -5.37266135e-01 -1.08438575e+00 4.29274261e-01 -6.27526283e-01 2.50886589e-01 4.62381802e-02 8.19488168e-02 1.08366266e-01 1.99532732e-01 6.96405768e-01 -7.51205623e-01 1.44923210e-01 1.02374601e+00 5.68472315e-03 -5.79653800e-01 1.91286907e-01 -9.88037825e-01 5.00503182e-01 5.33026040e-01 -3.12850982e-01 -8.59892517e-02 2.93207049e-01 8.85246515e-01 1.54404104e-01 2.48129562e-01 -4.51778799e-01 1.62835762e-01 -4.93364215e-01 -1.72281310e-01 -2.14062199e-01 -2.42162824e-01 -7.59717226e-01 2.63687491e-01 8.76712203e-01 -1.82092845e-01 8.67398828e-02 -2.89648950e-01 1.03241765e+00 -1.22973621e-01 1.60150215e-01 5.46787202e-01 -2.94363588e-01 -3.83304022e-02 3.13136168e-02 -9.73353922e-01 6.18509091e-02 1.09971333e+00 -9.01798310e-04 -5.26787043e-01 -7.91454911e-01 -9.34575617e-01 2.65088290e-01 8.09906125e-01 3.29796761e-01 -1.82394043e-01 -9.73264515e-01 -7.32389987e-01 -2.39882037e-01 -1.09636284e-01 -7.91365147e-01 1.18157327e-01 1.49389267e+00 -3.82295370e-01 2.28276029e-01 2.61733621e-01 -5.65986574e-01 -1.45886755e+00 3.97053778e-01 3.34572971e-01 -5.06312668e-01 -6.03231549e-01 1.98531687e-01 -2.19714642e-01 -3.17989320e-01 -9.84932408e-02 8.38839188e-02 -1.46163598e-01 7.51389921e-01 3.41590703e-01 4.76510465e-01 -2.30790228e-01 -5.76682329e-01 -4.34113264e-01 1.03408091e-01 -6.07424527e-02 -5.60248941e-02 1.42026031e+00 -4.79646385e-01 -6.92041576e-01 6.88422978e-01 9.97338176e-01 5.64239383e-01 -8.62347066e-01 -2.55793691e-01 2.80324817e-01 -5.88551402e-01 -5.10077298e-01 -3.02285045e-01 -7.69792914e-01 2.54563183e-01 -1.24855459e-01 1.28867364e+00 6.17296040e-01 3.06783319e-01 3.81826520e-01 2.18452320e-01 2.76573122e-01 -9.29479182e-01 2.25867763e-01 5.36782920e-01 3.86346668e-01 -1.08363330e+00 5.69382347e-02 -6.13435328e-01 -2.75995255e-01 1.03734016e+00 1.43579885e-01 -1.53260395e-01 8.74420285e-01 -6.93549067e-02 -2.07207605e-01 -6.56617582e-01 -9.48780179e-01 -1.74903110e-01 -1.84518874e-01 2.15295225e-01 3.43336850e-01 1.50346950e-01 -9.40125346e-01 4.57597911e-01 -5.45918345e-02 -4.59268272e-01 9.73201573e-01 9.89947557e-01 -5.47471941e-01 -1.13613391e+00 -3.65603834e-01 5.12737751e-01 -5.50897419e-01 1.89653322e-01 -1.21931350e+00 9.74061906e-01 -2.96259880e-01 1.20332742e+00 5.78878745e-02 -3.72914255e-01 -3.21202092e-02 4.87409346e-03 9.89865810e-02 -3.20171207e-01 -8.93479288e-01 8.38383138e-02 4.74661320e-01 -4.25193846e-01 -4.25771862e-01 -1.03499603e+00 -5.27671993e-01 -9.49304998e-01 -4.75129008e-01 4.10170913e-01 5.69485247e-01 9.01416779e-01 4.18275923e-01 5.14493845e-02 1.02498507e+00 -5.22415161e-01 -1.81966856e-01 -1.00717151e+00 -6.93016171e-01 6.28617227e-01 4.01352614e-01 -4.74467486e-01 -1.01048744e+00 -2.02370107e-01]
[7.013131618499756, 5.297226428985596]
16fcece6-2f17-4350-8a82-9e67ec7627fa
improving-spectral-graph-convolution-for
2112.0716
null
https://arxiv.org/abs/2112.07160v2
https://arxiv.org/pdf/2112.07160v2.pdf
A New Perspective on the Effects of Spectrum in Graph Neural Networks
Many improvements on GNNs can be deemed as operations on the spectrum of the underlying graph matrix, which motivates us to directly study the characteristics of the spectrum and their effects on GNN performance. By generalizing most existing GNN architectures, we show that the correlation issue caused by the $unsmooth$ spectrum becomes the obstacle to leveraging more powerful graph filters as well as developing deep architectures, which therefore restricts GNNs' performance. Inspired by this, we propose the correlation-free architecture which naturally removes the correlation issue among different channels, making it possible to utilize more sophisticated filters within each channel. The final correlation-free architecture with more powerful filters consistently boosts the performance of learning graph representations. Code is available at https://github.com/qslim/gnn-spectrum.
['Qiang Zhang', 'Yanming Shen', 'BaoCai Yin', 'Heng Qi', 'Rui Li', 'Mingqi Yang']
2021-12-14
null
null
null
null
['graph-property-prediction', 'graph-regression']
['graphs', 'graphs']
[-1.38942584e-01 4.11965363e-02 -6.66563660e-02 1.99695351e-03 -1.05667993e-01 -6.09546721e-01 6.19312525e-01 -1.56000331e-01 -2.22709313e-01 3.03999871e-01 2.15485871e-01 -7.16584921e-01 -1.86578587e-01 -1.02619386e+00 -7.31426179e-01 -4.07715946e-01 -3.88291061e-01 -2.91630924e-01 2.51163334e-01 -6.62280798e-01 -5.62816449e-02 3.33505690e-01 -1.01814699e+00 -1.34863168e-01 6.66187108e-01 7.37854302e-01 8.05472508e-02 6.90094650e-01 -4.11946699e-02 6.33077383e-01 -6.98933721e-01 -5.06873548e-01 7.84983516e-01 -6.82398319e-01 -4.06648844e-01 -1.85911119e-01 3.99384558e-01 -1.27331033e-01 -1.09496439e+00 1.27673602e+00 3.62525731e-01 1.60344616e-01 2.94819534e-01 -1.12673008e+00 -7.83675849e-01 1.03992069e+00 -5.15233934e-01 5.18619239e-01 7.71675333e-02 2.66645998e-01 1.30491579e+00 -4.31304365e-01 3.36880803e-01 1.23407006e+00 8.55871201e-01 1.74569756e-01 -1.37821281e+00 -8.62158835e-01 2.65488833e-01 6.48834705e-02 -1.50072646e+00 -4.22656536e-01 7.25346208e-01 -9.97205079e-02 8.48550200e-01 2.10452124e-01 8.11919212e-01 1.18639266e+00 2.15855062e-01 4.12704945e-01 7.61202514e-01 -5.27534246e-01 -5.34472615e-02 -2.94734329e-01 -2.00974829e-02 7.99147427e-01 6.74957454e-01 1.29561156e-01 -4.88154560e-01 -6.51075542e-02 9.95968342e-01 -1.41318366e-01 -5.01511931e-01 -5.49692869e-01 -8.12878609e-01 1.02623832e+00 1.00449920e+00 3.33457679e-01 -6.49861842e-02 7.35327721e-01 2.34326169e-01 5.70762932e-01 1.51961520e-01 5.58650196e-01 -1.45403191e-01 7.75070637e-02 -7.16714084e-01 -4.55762632e-02 8.41473877e-01 9.70946133e-01 9.78575289e-01 3.70292306e-01 2.05148175e-01 5.62445223e-01 4.48361397e-01 5.44802129e-01 -5.99608086e-02 -7.06784427e-01 1.97643369e-01 6.22262716e-01 -5.11447310e-01 -1.16570818e+00 -7.90704608e-01 -1.06577504e+00 -8.31169844e-01 -2.05059275e-02 5.26795983e-01 -4.45866108e-01 -9.85325873e-01 1.90793347e+00 -1.32629529e-01 3.05494577e-01 -2.03798428e-01 8.74964297e-01 5.12027144e-01 2.82933027e-01 -1.30244926e-01 3.32981199e-01 1.10588717e+00 -8.18871439e-01 -3.92857075e-01 -4.33197439e-01 6.71168864e-01 -7.75647819e-01 9.43263471e-01 1.46233603e-01 -5.95023334e-01 -3.08362663e-01 -1.38647044e+00 3.09882779e-02 -3.17504108e-01 -1.85477659e-01 1.20334494e+00 1.04858053e+00 -1.43114161e+00 8.38546276e-01 -8.38564992e-01 -4.15863603e-01 3.79371881e-01 5.48379123e-01 1.17739458e-02 -2.01172546e-01 -1.41726577e+00 6.66976213e-01 1.83052227e-01 1.50005728e-01 -4.49613065e-01 -6.27339423e-01 -6.54032767e-01 2.31102496e-01 4.38265055e-01 -6.60109222e-01 8.96172702e-01 -9.74891603e-01 -1.27933645e+00 2.78632432e-01 2.81196356e-01 -5.88026106e-01 1.92704231e-01 -1.81832574e-02 -7.38728642e-01 5.54931574e-02 -2.00546965e-01 3.03834647e-01 8.50120962e-01 -7.53355920e-01 -1.67574674e-01 -8.40801448e-02 3.42734098e-01 1.10534234e-02 -3.78129661e-01 -4.87498432e-01 -6.05957329e-01 -5.81267595e-01 2.67449647e-01 -1.18000674e+00 -4.65267390e-01 -3.71373713e-01 -6.05293512e-01 1.59436062e-01 4.28930372e-01 -2.74072409e-01 1.23570716e+00 -2.33612514e+00 -1.25737682e-01 7.96974719e-01 5.65652907e-01 1.85944602e-01 -4.95440781e-01 7.62298882e-01 -2.22833440e-01 1.49065450e-01 4.05857950e-01 1.23564497e-01 6.61050994e-03 2.37211436e-02 -1.22506805e-01 5.88022292e-01 1.59374237e-01 8.91485214e-01 -8.22925746e-01 3.20017487e-01 8.47164243e-02 5.40041387e-01 -6.41821384e-01 -3.03874701e-01 -1.24984160e-01 2.05139205e-01 -4.18405890e-01 2.73987979e-01 8.06318700e-01 -7.04271376e-01 5.49273491e-01 -2.63478279e-01 8.92093107e-02 5.67130923e-01 -1.19300830e+00 1.55755544e+00 -2.72497654e-01 8.40789616e-01 6.99689165e-02 -8.78534496e-01 8.44804823e-01 -1.10466376e-01 2.79394478e-01 -8.67536485e-01 3.50296974e-01 1.77261710e-01 5.92371643e-01 3.41724843e-01 2.60718882e-01 3.93592529e-02 1.07417457e-01 4.21692908e-01 2.76130021e-01 1.95142478e-01 2.06588924e-01 6.28200710e-01 1.49548531e+00 -3.55321020e-01 4.47423570e-02 -6.92782521e-01 7.33198896e-02 -1.78999692e-01 2.27517098e-01 1.06350517e+00 -1.84239391e-02 1.68891445e-01 6.91768289e-01 -1.91357240e-01 -8.96395028e-01 -1.16640556e+00 3.08590829e-02 1.13982081e+00 1.56959966e-01 -9.57547247e-01 -3.98707151e-01 -5.28274953e-01 2.58256346e-01 2.61453658e-01 -4.80377018e-01 -4.85725969e-01 -2.88263947e-01 -9.62639630e-01 7.34583139e-01 3.53088647e-01 3.62695068e-01 -1.52483255e-01 2.07477789e-02 2.84045339e-02 1.39697373e-01 -9.72680271e-01 -3.94510955e-01 5.14297843e-01 -7.05152690e-01 -1.00741601e+00 -4.44363087e-01 -2.68549323e-01 4.14253503e-01 5.36784351e-01 9.52778101e-01 3.65539312e-01 -1.04898974e-01 3.97517800e-01 -4.33878571e-01 -4.28389668e-01 -2.65161425e-01 3.77808362e-01 9.26302280e-03 -1.86997339e-01 1.59320191e-01 -9.86159086e-01 -7.15665698e-01 -4.12824526e-02 -6.24633312e-01 -2.08832070e-01 7.51847684e-01 6.39158189e-01 -1.30914003e-01 2.04971597e-01 4.79844809e-01 -9.88895774e-01 6.47461236e-01 -3.66052538e-01 -7.41552591e-01 -1.87963113e-01 -7.15415716e-01 3.76659125e-01 8.85013819e-01 -2.76223749e-01 -2.61916101e-01 -1.78755656e-01 -1.73462674e-01 -2.83307910e-01 5.66219747e-01 4.50409949e-01 1.09483890e-01 -5.87342381e-01 9.71421599e-01 -9.16650966e-02 2.82046665e-02 -2.91802019e-01 5.10241032e-01 1.60896838e-01 9.19530019e-02 -1.93741783e-01 1.21633768e+00 4.39276576e-01 1.73249453e-01 -9.89535868e-01 -5.99638343e-01 -2.75912344e-01 -1.50306419e-01 -1.16990358e-01 3.96206975e-01 -1.26189256e+00 -6.91936493e-01 2.53616989e-01 -6.43930614e-01 -4.92667735e-01 3.65005136e-02 6.27858341e-01 -7.14179203e-02 3.75563294e-01 -8.83253455e-01 -3.82066101e-01 -1.13096952e-01 -8.85830581e-01 3.45635325e-01 2.89305598e-01 -1.03240691e-01 -1.01618600e+00 -6.05285987e-02 -8.72838348e-02 7.69409060e-01 -7.95068592e-02 9.70846295e-01 -5.00752926e-01 -9.06524301e-01 -9.90067571e-02 -3.99179548e-01 3.14918578e-01 1.90859206e-03 -1.16426684e-01 -9.23032820e-01 -6.26063704e-01 -2.46624961e-01 -3.25579830e-02 1.15734267e+00 3.42649430e-01 6.92720473e-01 2.52284948e-02 -3.05645347e-01 1.04368079e+00 1.66011333e+00 -1.95002079e-01 6.61288500e-01 3.08901101e-01 9.22175109e-01 -1.25761017e-01 -2.69099653e-01 3.69753152e-01 1.85181007e-01 4.25354183e-01 4.39474761e-01 -2.15810001e-01 -3.16383690e-01 -3.80388886e-01 2.96671778e-01 8.98429990e-01 -1.23326540e-01 -5.80077648e-01 -9.01367009e-01 5.02634645e-02 -1.58231127e+00 -6.90729380e-01 -2.61870980e-01 1.92462957e+00 3.28909755e-01 3.91877234e-01 5.54957576e-02 -3.12100444e-02 6.18034065e-01 6.03858411e-01 -4.85948324e-01 -2.05360323e-01 -1.06104612e-01 5.34586370e-01 1.16207004e+00 4.58191842e-01 -9.35267985e-01 8.85202885e-01 7.12020540e+00 7.91317821e-01 -1.25624871e+00 -1.42566994e-01 2.34588444e-01 -1.18300349e-01 -5.71401536e-01 1.15780905e-01 -6.38581336e-01 3.54065031e-01 1.16115630e+00 -2.57429510e-01 8.37982297e-01 6.39550030e-01 5.31765670e-02 1.26269042e-01 -9.09943342e-01 8.88549805e-01 -1.74669996e-01 -1.42472613e+00 2.10528150e-01 3.54826033e-01 4.86813217e-01 6.21122420e-01 1.87112287e-01 4.37889695e-01 6.69153690e-01 -1.03531051e+00 4.57522631e-01 5.76542243e-02 6.08760297e-01 -7.73443401e-01 3.84383410e-01 -1.16041312e-02 -1.22899830e+00 -1.92706525e-01 -6.10189199e-01 -3.65783215e-01 -2.74486840e-01 8.42583656e-01 -8.61322582e-01 7.40898371e-01 5.21708071e-01 5.82059741e-01 -8.27342570e-01 9.13679302e-01 -1.87370732e-01 6.90122545e-01 -4.76841420e-01 -6.78415820e-02 2.81741828e-01 -3.02118391e-01 3.82143855e-01 1.11372793e+00 3.70711297e-01 -1.55341819e-01 1.06735528e-01 8.59781981e-01 -3.32082570e-01 -8.80318135e-02 -6.17145002e-01 -3.56804371e-01 5.13749838e-01 1.15282476e+00 -9.19775605e-01 8.44389573e-02 -7.68228650e-01 7.91503251e-01 4.00872380e-01 5.84968328e-01 -9.20384645e-01 -3.45703632e-01 6.45031154e-01 3.44630569e-01 4.93163317e-01 -4.47390139e-01 -1.07875913e-01 -1.19455993e+00 -2.90943086e-01 -9.59958613e-01 2.74125189e-01 -3.34725201e-01 -1.21249127e+00 4.51254129e-01 -6.11775041e-01 -1.07500279e+00 1.15535803e-01 -6.78979635e-01 -5.58439851e-01 9.22546208e-01 -1.49441195e+00 -7.83688426e-01 -1.46452948e-01 8.43231380e-01 -3.02578509e-01 -7.81992171e-03 6.34672821e-01 4.81036007e-01 -5.62464535e-01 8.46817374e-01 1.52105331e-01 4.17643696e-01 6.22560620e-01 -1.09497035e+00 7.76827157e-01 1.24577510e+00 4.65483606e-01 1.02681887e+00 6.93911970e-01 -5.41133285e-01 -1.68840992e+00 -8.33778441e-01 2.23069325e-01 -2.18383297e-01 1.04270375e+00 -7.07251310e-01 -6.46765232e-01 8.53945255e-01 3.12505886e-02 3.50503922e-02 6.78285599e-01 7.60665178e-01 -7.54266381e-01 -8.13596919e-02 -4.31077987e-01 8.94605577e-01 1.47433245e+00 -6.77304864e-01 7.36717656e-02 2.36265495e-01 7.42108881e-01 -3.26950550e-01 -6.40448213e-01 2.17820600e-01 2.72021264e-01 -1.07263148e+00 9.21934724e-01 -3.55255544e-01 3.81780677e-02 -2.15098530e-01 -1.23946682e-01 -1.39237666e+00 -7.76933253e-01 -8.85062873e-01 7.20207095e-02 7.57314205e-01 7.17563987e-01 -9.51522291e-01 8.99443209e-01 1.87205732e-01 -1.23512290e-01 -4.77110714e-01 -6.37062192e-01 -9.42057848e-01 -1.23824188e-02 -3.35596532e-01 5.32775462e-01 1.04528856e+00 1.10462634e-02 6.96346939e-01 -3.10516059e-01 3.93408865e-01 4.84546542e-01 5.58703691e-02 7.40612805e-01 -1.11227441e+00 -7.52511382e-01 -5.50240338e-01 -7.14505076e-01 -1.34390223e+00 -2.33180955e-01 -1.14868927e+00 -2.72044778e-01 -1.25215626e+00 -1.10722765e-01 -4.87139583e-01 -4.78383809e-01 3.18053365e-01 -7.10537210e-02 3.25203419e-01 4.66602862e-01 -1.29288994e-02 -5.47655404e-01 2.58308232e-01 1.11482632e+00 1.86360836e-01 -8.65124539e-02 2.77590496e-03 -1.23477435e+00 5.98557353e-01 9.34706151e-01 -2.83038676e-01 -7.00016618e-01 -5.07984757e-01 5.86390853e-01 -3.00155222e-01 1.79585874e-01 -1.28206921e+00 2.31647328e-01 3.02498072e-01 6.09418213e-01 3.85565571e-02 2.21143231e-01 -5.40898442e-01 1.73332900e-01 6.41412258e-01 2.03841776e-02 7.53760338e-03 2.80908167e-01 7.45439947e-01 7.53442198e-02 1.93605348e-01 8.06146979e-01 -1.33658826e-01 -6.29778504e-01 3.43190998e-01 -3.35038722e-01 1.42050847e-01 4.98258740e-01 3.48820239e-02 -7.73944676e-01 -6.75970554e-01 -5.38824677e-01 1.49630800e-01 5.70548177e-01 2.44847685e-01 1.21590212e-01 -1.18941224e+00 -3.86924773e-01 4.75919724e-01 -1.76953748e-01 -5.29479265e-01 4.01801050e-01 9.36764419e-01 -6.25301361e-01 3.76120239e-01 -2.43411109e-01 -3.38730514e-01 -7.45898426e-01 3.73757809e-01 4.28719223e-01 -1.48865163e-01 -7.12512195e-01 1.11853707e+00 1.14026435e-01 -1.53828204e-01 4.02770936e-02 -2.02302411e-01 2.25622371e-01 -2.54560947e-01 2.85816491e-01 1.19777933e-01 2.27245584e-01 -1.30177870e-01 -5.08400619e-01 2.12215707e-01 -1.05383545e-01 2.09867045e-01 1.05522275e+00 -7.28659406e-02 8.41792524e-02 6.88229501e-02 1.16337252e+00 4.21470314e-01 -1.07794857e+00 -2.57923305e-01 -1.21988645e-02 -3.43225062e-01 1.80984527e-01 -4.53236580e-01 -1.20795810e+00 5.96881092e-01 3.28637660e-01 5.80394149e-01 1.06931686e+00 -8.13871995e-02 6.61236465e-01 3.85087281e-01 3.59511346e-01 -8.13235283e-01 9.00556818e-02 8.66514623e-01 3.49819511e-01 -8.08378994e-01 2.40307286e-01 -6.78875089e-01 -1.34768784e-01 1.03636646e+00 3.30840856e-01 -6.18794978e-01 8.61173511e-01 3.90754104e-01 1.09364577e-01 -3.49399209e-01 -6.55239642e-01 -5.40995061e-01 4.73730713e-02 6.85633779e-01 4.51322943e-01 9.99364331e-02 -1.18167438e-01 3.60830039e-01 -5.32489955e-01 -4.49746042e-01 8.19484115e-01 5.93840718e-01 -4.46166843e-01 -1.30087185e+00 -9.57893506e-02 7.41285682e-01 -3.72046441e-01 -4.98953849e-01 -4.23469454e-01 1.03247464e+00 -6.15391284e-02 9.01700675e-01 -5.89878559e-02 -7.94632077e-01 2.79914111e-01 -1.49949506e-01 4.31910008e-01 -7.69139707e-01 -3.86811823e-01 1.22304127e-01 2.72591263e-01 -7.59932280e-01 -6.53791279e-02 -2.59126693e-01 -1.20483708e+00 -9.57176805e-01 -4.44965154e-01 -1.83952823e-02 3.55859071e-01 4.82331991e-01 5.35646737e-01 7.64982581e-01 4.75878268e-01 -5.04230499e-01 -6.18366420e-01 -7.23125815e-01 -9.09098506e-01 1.12118289e-01 3.71329486e-01 -5.38095891e-01 -6.05072498e-01 -6.83373928e-01]
[6.888579368591309, 6.132213115692139]
d1145fb2-3a83-4728-ae03-f24bdf1eb901
transfer-learning-approach-to-bicycle-sharing
2111.0099
null
https://arxiv.org/abs/2111.00990v1
https://arxiv.org/pdf/2111.00990v1.pdf
Transfer Learning Approach to Bicycle-sharing Systems' Station Location Planning using OpenStreetMap Data
Bicycle-sharing systems (BSS) have become a daily reality for many citizens of larger, wealthier cities in developed regions. However, planning the layout of bicycle-sharing stations usually requires expensive data gathering, surveying travel behavior and trip modelling followed by station layout optimization. Many smaller cities and towns, especially in developing areas, may have difficulty financing such projects. Planning a BSS also takes a considerable amount of time. Yet as the pandemic has shown us, municipalities will face the need to adapt rapidly to mobility shifts, which include citizens leaving public transport for bicycles. Laying out a bike sharing system quickly will become critical in addressing the increase in bike demand. This paper addresses the problem of cost and time in BSS layout design and proposes a new solution to streamline and facilitate the process of such planning by using spatial embedding methods. Based only on publicly available data from OpenStreetMap, and station layouts from 34 cities in Europe, a method has been developed to divide cities into micro-regions using the Uber H3 discrete global grid system and to indicate regions where it is worth placing a station based on existing systems in different cities using transfer learning. The result of the work is a mechanism to support planners in their decision making when planning a station layout with a choice of reference cities.
['Piotr Szymański', 'Kamil Raczycki']
2021-11-01
null
null
null
null
['layout-design']
['computer-vision']
[-6.71880484e-01 2.66724080e-01 -3.06875229e-01 -5.86468466e-02 -7.04367578e-01 -6.52719080e-01 4.52591330e-01 -7.18475431e-02 -2.25029320e-01 9.59871113e-01 5.76484382e-01 -1.03601348e+00 -3.32410276e-01 -1.25715089e+00 -2.84439176e-01 -5.60483992e-01 3.30657023e-03 9.04890001e-01 1.21042743e-01 -6.29571140e-01 8.67168307e-02 7.56605983e-01 -1.01746261e+00 -8.73698294e-02 8.09465528e-01 2.98978537e-01 4.00965154e-01 3.02648962e-01 -1.38080359e-01 -7.46338367e-02 -1.73903540e-01 -2.91244864e-01 4.24041450e-01 -8.90532956e-02 -7.58291185e-01 -1.14920698e-01 -5.21847963e-01 -1.15408882e-01 -4.49970841e-01 4.55610454e-01 8.48798037e-01 3.12698603e-01 7.75588274e-01 -1.55591047e+00 -1.49627417e-01 6.56185150e-01 -1.51081860e-01 2.54069895e-01 2.00653702e-01 1.83146253e-01 7.91167796e-01 -7.23790705e-01 2.09323898e-01 9.76037443e-01 8.18441212e-01 -3.42599273e-01 -1.23075140e+00 -6.21763766e-01 -5.11391871e-02 3.17160338e-01 -1.93260145e+00 -4.68412906e-01 4.54132617e-01 -4.36221927e-01 1.14666235e+00 4.45536822e-01 1.05380082e+00 4.37690139e-01 -1.49049703e-02 3.41145098e-01 5.60300887e-01 -4.79029417e-01 3.27604711e-01 3.41789097e-01 -6.03109300e-01 2.48362049e-01 5.01378536e-01 -2.17469856e-01 4.24485862e-01 1.31736500e-02 4.43364292e-01 -1.05432630e-01 -8.28741491e-02 -1.24999456e-01 -1.28109801e+00 8.58577073e-01 8.89049172e-01 8.33303034e-01 -4.49120969e-01 9.12865475e-02 7.94268921e-02 5.22385202e-02 3.81982684e-01 3.00145477e-01 -4.65444624e-01 -3.49940270e-01 -1.10614491e+00 3.13459449e-02 7.64041007e-01 1.13839281e+00 8.53899360e-01 -1.64619699e-01 6.15203530e-02 6.04321659e-01 4.58446622e-01 7.77181208e-01 7.71809369e-02 -7.44647324e-01 8.77153218e-01 6.77915335e-01 5.12815535e-01 -1.31742239e+00 -8.28440189e-01 -2.78113097e-01 -8.93917263e-01 -3.26530397e-01 4.12901103e-01 -4.74989861e-01 -4.07234222e-01 1.19949949e+00 3.32843691e-01 3.09357017e-01 -2.33627751e-01 4.94278312e-01 1.86146542e-01 1.06880426e+00 -1.52980521e-01 2.57803977e-01 1.09965503e+00 -7.47698724e-01 -4.84854609e-01 1.82259232e-01 1.10296607e+00 -6.25425160e-01 7.32952893e-01 -4.02724780e-02 -9.16755617e-01 -3.18846971e-01 -5.45171440e-01 1.38034716e-01 -9.71057475e-01 1.17014565e-01 2.84832776e-01 1.23321557e+00 -1.42935717e+00 1.00104012e-01 -7.47975647e-01 -9.24181759e-01 3.86044323e-01 5.96445024e-01 -1.14205942e-01 -8.40452388e-02 -1.23761404e+00 1.26325488e+00 2.51881689e-01 1.11006916e-01 -2.07751051e-01 -7.10544348e-01 -7.81076252e-01 4.01560426e-01 -8.70912895e-02 -6.92179501e-01 8.52922738e-01 -4.38674897e-01 -1.31488705e+00 9.25448686e-02 7.63745904e-02 -1.02603287e-01 3.36684108e-01 7.44864047e-01 -8.42755854e-01 -4.62209731e-01 4.84184921e-01 7.89524913e-01 2.16652602e-01 -1.08287680e+00 -1.07534504e+00 -2.81759333e-02 8.27629790e-02 1.62037969e-01 -1.64875761e-01 -2.27909461e-01 -3.52281719e-01 -3.18816006e-01 -8.97844881e-02 -1.19391191e+00 -5.24101079e-01 -5.11272013e-01 -3.31365615e-01 -2.87205838e-02 6.78906083e-01 -8.84950519e-01 1.57398975e+00 -1.93034399e+00 -1.53531760e-01 8.78471494e-01 -3.96936119e-01 1.41446114e-01 1.28693189e-02 9.65178192e-01 1.45653002e-02 3.09024096e-01 -2.84789447e-02 -3.66126209e-01 4.58248109e-01 3.30291748e-01 3.86716686e-02 4.68126893e-01 -1.55145600e-01 9.35986280e-01 -9.80305970e-01 -3.57391477e-01 5.57733059e-01 4.27641958e-01 -5.91394484e-01 -3.01828951e-01 3.69109601e-01 5.93416870e-01 -4.10880387e-01 6.11561894e-01 7.57286906e-01 -9.03536379e-02 1.55268520e-01 4.26056013e-02 -7.14643359e-01 3.67131323e-01 -1.27053070e+00 1.25731874e+00 -1.01216817e+00 8.24588835e-01 -1.63033560e-01 -1.27375019e+00 6.34877861e-01 4.68580306e-01 7.00388134e-01 -8.30873013e-01 4.44029272e-02 4.09325778e-01 -1.14206865e-01 -4.88042355e-01 5.27382970e-01 2.20009223e-01 -1.64190263e-01 5.60042799e-01 -6.19260728e-01 -4.18189734e-01 1.69653684e-01 -1.31156087e-01 8.17286134e-01 -2.55192846e-01 3.05582821e-01 -5.25464535e-01 5.05206645e-01 1.34726390e-01 1.47674873e-01 1.26561061e-01 -1.48922399e-01 3.74830991e-01 3.13980728e-01 -5.05118251e-01 -1.20161080e+00 -7.49694586e-01 -1.28970280e-01 6.59263194e-01 -2.41726771e-01 1.00683123e-01 -5.08713543e-01 -1.89721763e-01 3.20523456e-02 9.47768867e-01 -3.98533523e-01 2.08819553e-01 -4.29692864e-01 -7.05260634e-01 4.10381466e-01 1.56412646e-01 5.98861873e-01 -5.72128475e-01 -3.02906036e-01 4.09068435e-01 -6.07749999e-01 -8.24886501e-01 -6.36071026e-01 -5.98744564e-02 -5.77037215e-01 -6.30743802e-01 -1.06698620e+00 -8.76950264e-01 8.51344764e-01 7.13490963e-01 9.13502932e-01 -1.39622420e-01 2.93362379e-01 3.70116830e-01 -2.43851051e-01 -2.68756151e-01 -1.36861801e-01 8.26353252e-01 -1.29623249e-01 -4.14078869e-02 2.21518010e-01 -6.09025002e-01 -8.28448117e-01 9.57115352e-01 -5.47085404e-01 8.30722973e-02 2.51072317e-01 3.08703840e-01 2.16151580e-01 5.78797996e-01 1.00661182e+00 -1.64406657e-01 3.80065739e-01 -1.41200697e+00 -8.08964252e-01 1.81067750e-01 -5.22647619e-01 -3.93890977e-01 4.22864318e-01 1.85777038e-01 -7.45405674e-01 4.92546260e-02 -3.39253813e-01 4.83449519e-01 -1.25998616e-01 7.02410758e-01 -5.08000016e-01 -1.75464582e-02 4.30671781e-01 -4.66273092e-02 -3.17397714e-01 -8.03095326e-02 4.64526474e-01 1.12582219e+00 -2.08554566e-01 -2.24696845e-01 9.44813371e-01 5.73548019e-01 -2.05945626e-01 -1.05193448e+00 1.97983518e-01 -8.32727313e-01 -7.60691464e-01 -3.94350350e-01 6.12481475e-01 -9.51346457e-01 -5.95254779e-01 -1.26341850e-01 -9.09472585e-01 -5.82482755e-01 -1.72072425e-01 4.45043981e-01 -5.55258989e-01 -5.74186519e-02 2.48834789e-01 -6.54296100e-01 3.00251305e-01 -1.24984503e+00 6.46544397e-01 1.11219667e-01 -3.25047255e-01 -1.35903573e+00 3.53629410e-01 4.48258489e-01 9.48860645e-01 -1.07438993e-02 9.32396114e-01 -8.71264115e-02 -8.90021503e-01 -1.34609655e-01 -2.98305511e-01 -1.92945287e-01 5.91709614e-01 -1.36763170e-01 -6.73781157e-01 -3.61937016e-01 -7.56670117e-01 5.21496177e-01 1.08349897e-01 7.44349301e-01 6.65404797e-01 -4.72443849e-01 -7.21556187e-01 3.27438235e-01 1.39187038e+00 4.55005646e-01 1.03708911e+00 6.22699022e-01 3.72696370e-01 6.90340042e-01 4.02086258e-01 5.48372865e-01 1.24068558e+00 7.89312124e-01 4.91883934e-01 -3.88196558e-01 1.21299103e-01 -1.65293694e-01 1.97114691e-01 8.04281294e-01 -2.52784878e-01 -4.96015728e-01 -1.27154088e+00 1.16922045e+00 -1.86751139e+00 -9.75947142e-01 -9.69886333e-02 2.26415300e+00 2.26155624e-01 -1.57722995e-01 6.33870125e-01 2.94763893e-01 5.58217764e-01 -2.42332950e-01 -8.50679055e-02 -4.03783113e-01 1.88166887e-01 -1.18942544e-01 1.11043203e+00 7.24334955e-01 -7.04156101e-01 7.69048929e-01 5.74784565e+00 7.16880798e-01 -1.03112245e+00 1.70473874e-01 8.84589851e-01 3.68654400e-01 -7.58723676e-01 5.20359315e-02 -6.11948133e-01 8.26518655e-01 1.21521795e+00 -2.05941558e-01 7.19051123e-01 4.08693969e-01 1.14998150e+00 -2.65102893e-01 -5.47554791e-01 8.31780136e-01 -6.63316786e-01 -1.58051932e+00 -3.34867030e-01 4.20200199e-01 1.05922699e+00 4.04443979e-01 5.64926676e-02 3.11908334e-01 4.00104254e-01 -1.04347885e+00 5.63715577e-01 4.80627239e-01 6.32893622e-01 -9.60490704e-01 8.44931185e-01 4.81502920e-01 -1.68059635e+00 -2.84166962e-01 -1.86625093e-01 -9.56028625e-02 4.93027031e-01 4.44127500e-01 -1.19316721e+00 6.22875154e-01 5.54457307e-01 5.79516470e-01 -2.98353940e-01 1.43665910e+00 2.15837419e-01 5.74721456e-01 -5.80492437e-01 -4.81981337e-02 6.37030184e-01 -5.81701040e-01 1.01720355e-01 1.20391273e+00 1.25269699e+00 8.74609575e-02 -1.58291027e-01 5.43694198e-01 3.23554337e-01 2.06023544e-01 -1.09189296e+00 1.11427702e-01 6.98700845e-01 1.00818896e+00 -1.07874680e+00 8.09201971e-02 -6.17010117e-01 4.94329572e-01 -2.80732989e-01 5.84880710e-01 -9.19212759e-01 -3.23499769e-01 6.51458323e-01 3.88012081e-01 4.03315306e-01 -6.85380340e-01 -3.24356735e-01 -6.01084173e-01 -5.04045486e-01 -4.12100583e-01 -5.48654981e-02 -6.49836063e-01 -5.24006307e-01 2.80150082e-02 9.95219648e-02 -1.37190723e+00 1.20559325e-02 -2.62604535e-01 -7.55724311e-01 9.72126126e-01 -1.67148471e+00 -1.26905680e+00 -3.78438048e-02 6.43557191e-01 3.13261002e-01 -1.10443160e-01 7.17198849e-01 1.00566101e+00 -5.60132742e-01 2.55548596e-01 5.41914582e-01 -2.93252528e-01 1.93411723e-01 -9.74700809e-01 6.57484353e-01 4.53214198e-01 6.99887648e-02 2.35592484e-01 3.95671785e-01 -4.02821988e-01 -1.00761414e+00 -1.12174344e+00 1.53048837e+00 -1.85699582e-01 6.08093083e-01 -2.29805231e-01 -2.98716605e-01 7.98220634e-01 3.79638493e-01 -5.65281928e-01 8.88393044e-01 -2.20949780e-02 7.14873135e-01 -2.18862310e-01 -1.26096690e+00 8.67033660e-01 6.72807872e-01 -3.22140485e-01 -1.14420392e-01 6.07961833e-01 3.12497646e-01 1.95056528e-01 -5.99660695e-01 -2.65482515e-01 2.57483840e-01 -7.58533835e-01 9.99269068e-01 -7.17000812e-02 -4.08657938e-01 -4.26072061e-01 -3.03408831e-01 -1.85945511e+00 -7.00317621e-01 -7.92521000e-01 7.33326375e-01 1.05313694e+00 7.82105029e-01 -1.01581049e+00 8.22218657e-01 8.34431291e-01 -2.39716321e-01 -3.50398481e-01 -1.29186285e+00 -5.81109166e-01 2.02502817e-01 -6.26833022e-01 1.36120355e+00 9.57397938e-01 7.38872737e-02 -1.98381200e-01 1.27133997e-02 5.84643841e-01 -3.32296565e-02 -3.34792227e-01 9.68974233e-01 -9.43433046e-01 4.49116826e-01 -6.67359650e-01 -3.81301701e-01 -8.85467768e-01 -1.33477569e-01 -9.40490186e-01 -4.04238641e-01 -2.18149066e+00 -5.29795885e-01 -1.35067666e+00 -1.50346413e-01 3.13996464e-01 6.88385010e-01 4.86378074e-02 -2.13112786e-01 -2.24290341e-02 -1.06520310e-01 5.64198136e-01 1.08360982e+00 -3.46257180e-01 -5.18910587e-01 5.11028588e-01 -5.72706699e-01 5.16735971e-01 1.01185703e+00 -3.02192777e-01 -6.02736354e-01 -4.13083762e-01 8.92817795e-01 9.10728797e-02 7.58157969e-02 -1.03929269e+00 4.78844404e-01 -3.01019192e-01 -2.16091245e-01 -9.80476499e-01 2.90135980e-01 -1.63349628e+00 7.86431849e-01 6.97669268e-01 4.01656181e-01 4.07063514e-01 3.18040311e-01 2.73088217e-01 4.61642519e-02 -1.18955016e-01 3.69158775e-01 2.57381678e-01 -4.74735439e-01 2.67252326e-01 -1.01819134e+00 -2.92996168e-01 1.16966724e+00 -6.03025794e-01 3.75499725e-02 -7.28952348e-01 -6.99491262e-01 5.81670403e-01 3.53395581e-01 1.20704792e-01 1.09890342e-01 -1.59707141e+00 -5.18004060e-01 2.32019514e-01 -5.61837703e-02 -2.87074506e-01 2.09205836e-01 1.06999385e+00 -8.71870995e-01 1.05856264e+00 -3.08067232e-01 -1.41929075e-01 -5.63188314e-01 2.43946776e-01 4.44396943e-01 -2.03916624e-01 -3.04382086e-01 4.13919568e-01 -3.72719526e-01 -6.59868479e-01 -8.76109302e-02 -8.86630177e-01 -3.96487117e-01 6.47967279e-01 -1.38474582e-02 9.40186262e-01 2.14903370e-01 -9.60338831e-01 -3.07564497e-01 6.04211152e-01 7.98662424e-01 -2.35933408e-01 1.39717340e+00 -7.24907637e-01 1.26086712e-01 2.07031056e-01 1.02832448e+00 -8.09592977e-02 -1.02581334e+00 1.07705422e-01 1.76531136e-01 -5.63736618e-01 1.02624282e-01 -5.67171335e-01 -1.08862400e+00 5.71538985e-01 6.19484723e-01 6.48654640e-01 9.48901057e-01 -1.26235470e-01 9.58764255e-01 3.22444260e-01 6.97688043e-01 -1.26778090e+00 -4.48841661e-01 4.42312121e-01 6.46842241e-01 -1.36765766e+00 -3.13942492e-01 -3.62390392e-02 -2.57602423e-01 8.47921193e-01 -3.21578741e-01 3.44582379e-01 1.29886723e+00 3.54693457e-02 -1.42727718e-01 6.52961880e-02 -2.72611901e-02 -4.51072574e-01 -8.19463804e-02 8.90487254e-01 -4.84533794e-02 5.78949869e-01 -1.39360905e-01 2.67481714e-01 -2.00451627e-01 1.19483778e-02 5.41387618e-01 6.35261476e-01 -4.54174370e-01 -1.33732617e+00 -6.05861723e-01 2.59925127e-01 1.77149996e-01 -6.27558231e-02 3.32576513e-01 9.65906143e-01 5.08569777e-01 1.08229852e+00 2.00393528e-01 -1.26453787e-01 4.91662234e-01 -1.10145546e-01 -1.18732885e-01 -3.82595867e-01 -2.40807921e-01 -3.47793490e-01 2.95730531e-01 -1.32044256e-01 -3.13048452e-01 -9.03437376e-01 -1.06191432e+00 -7.05195665e-01 -3.10390383e-01 8.65516961e-02 1.05461204e+00 8.81135046e-01 4.63096380e-01 2.73573905e-01 1.00261223e+00 -1.31206727e+00 1.60912156e-01 -6.24807000e-01 -6.19095802e-01 -4.18772668e-01 1.96288452e-01 -7.34503806e-01 -1.02382034e-01 -1.77362084e-01]
[6.1789350509643555, 1.854358434677124]
a547866a-15e2-42b0-8c68-14a8471a3c66
going-deeper-into-semi-supervised-person-re
2107.11566
null
https://arxiv.org/abs/2107.11566v1
https://arxiv.org/pdf/2107.11566v1.pdf
Going Deeper into Semi-supervised Person Re-identification
Person re-identification is the challenging task of identifying a person across different camera views. Training a convolutional neural network (CNN) for this task requires annotating a large dataset, and hence, it involves the time-consuming manual matching of people across cameras. To reduce the need for labeled data, we focus on a semi-supervised approach that requires only a subset of the training data to be labeled. We conduct a comprehensive survey in the area of person re-identification with limited labels. Existing works in this realm are limited in the sense that they utilize features from multiple CNNs and require the number of identities in the unlabeled data to be known. To overcome these limitations, we propose to employ part-based features from a single CNN without requiring the knowledge of the label space (i.e., the number of identities). This makes our approach more suitable for practical scenarios, and it significantly reduces the need for computational resources. We also propose a PartMixUp loss that improves the discriminative ability of learned part-based features for pseudo-labeling in semi-supervised settings. Our method outperforms the state-of-the-art results on three large-scale person re-id datasets and achieves the same level of performance as fully supervised methods with only one-third of labeled identities.
['Mahsa Baktashmotlagh', 'Feras Dayoub', 'Frederic Maire', 'Olga Moskvyak']
2021-07-24
null
null
null
null
['semi-supervised-person-re-identification']
['computer-vision']
[ 8.59858915e-02 -4.03447092e-01 -1.64357364e-01 -6.40749753e-01 -3.83186996e-01 -7.66165495e-01 5.74009359e-01 -9.91195589e-02 -9.15551960e-01 5.83451152e-01 -2.94642057e-03 2.53392577e-01 3.41386557e-01 -6.17908895e-01 -6.65902793e-01 -5.26079059e-01 5.59637368e-01 6.50567889e-01 -7.67217726e-02 1.61864728e-01 -1.62816361e-01 4.21470612e-01 -1.64421892e+00 -1.82789505e-01 7.26506174e-01 9.26611960e-01 -2.16044277e-01 1.04379930e-01 3.73747684e-02 3.56978357e-01 -4.77562785e-01 -1.04790080e+00 7.12693155e-01 -3.60240787e-01 -7.44036019e-01 3.88098061e-01 1.00522208e+00 -6.48025453e-01 -6.64489985e-01 1.27110028e+00 5.25982141e-01 1.73398152e-01 4.28253710e-01 -1.42681313e+00 -6.89003348e-01 2.46397868e-01 -5.64776719e-01 -3.97593528e-02 1.78825244e-01 -1.02090970e-01 7.09408879e-01 -8.57878804e-01 2.91311800e-01 1.03575802e+00 1.00193429e+00 8.70182574e-01 -1.12173307e+00 -1.06676054e+00 2.45298028e-01 2.54091173e-01 -1.91753697e+00 -6.23490155e-01 5.50790310e-01 -4.95523930e-01 6.02138281e-01 -5.82007617e-02 5.15736103e-01 8.73150349e-01 -8.46987903e-01 5.25666356e-01 9.41182256e-01 -4.19465095e-01 2.89173536e-02 3.06645006e-01 2.69448310e-01 6.38509989e-01 5.74570656e-01 -4.87371832e-02 -4.93429810e-01 -1.56199351e-01 8.57801914e-01 5.72229326e-01 -1.18969865e-01 -5.80158949e-01 -1.07292998e+00 5.93607187e-01 3.55374783e-01 1.51197657e-01 -6.83903918e-02 5.42257354e-03 4.03175682e-01 9.86252129e-02 3.69788289e-01 1.38106182e-01 -2.37281770e-01 8.22543427e-02 -9.55927670e-01 1.03666514e-01 6.30640447e-01 1.16316390e+00 1.16735017e+00 -3.84050906e-01 3.96293364e-02 1.01817727e+00 -9.43149382e-04 6.05415046e-01 3.71562988e-01 -7.84289718e-01 5.12444258e-01 9.34902728e-01 4.95874703e-01 -7.53794551e-01 -2.51329452e-01 -3.58625919e-01 -9.83008981e-01 -1.07227609e-01 8.62678826e-01 -2.51869261e-01 -8.12630653e-01 1.96761203e+00 2.68726319e-01 3.09946775e-01 -1.59873396e-01 9.09955442e-01 7.30786443e-01 1.01642467e-01 -3.85192595e-02 2.05848888e-01 1.41811466e+00 -1.21969080e+00 -4.46743399e-01 -4.35362220e-01 4.56328094e-01 -5.06057799e-01 6.05986595e-01 -2.30896935e-01 -7.15756238e-01 -7.82562733e-01 -9.89316285e-01 2.36537140e-02 -6.07769668e-01 7.35123277e-01 5.47351360e-01 9.21563685e-01 -9.90688503e-01 3.66252959e-01 -5.06263971e-01 -7.19998181e-01 5.22957563e-01 7.38733649e-01 -9.04382288e-01 -4.50467527e-01 -9.68719006e-01 6.57156229e-01 3.45836341e-01 3.01527411e-01 -5.04641056e-01 -5.10909736e-01 -9.79685783e-01 1.42862409e-01 2.63697773e-01 -5.31042159e-01 1.06886029e+00 -1.11125457e+00 -1.07695353e+00 1.17551172e+00 -3.95267665e-01 -1.94572940e-01 6.74054682e-01 -2.53122091e-01 -2.87903398e-01 6.69263005e-02 3.24763894e-01 6.81289971e-01 6.11520648e-01 -1.16816783e+00 -7.92417169e-01 -5.47830582e-01 3.83645773e-01 1.65229037e-01 -8.51748705e-01 8.26828554e-02 -8.90690863e-01 -6.25966430e-01 -9.75607932e-02 -1.33881176e+00 -5.89863993e-02 7.38791898e-02 -3.00013214e-01 -3.73086065e-01 5.23494422e-01 -6.32162988e-01 7.03648627e-01 -2.12038398e+00 -1.90373927e-01 6.70085922e-02 3.64249676e-01 5.30374050e-01 -1.08226240e-01 2.06260413e-01 -1.49602205e-01 1.48896908e-03 -3.17382067e-02 -8.82036090e-01 -1.08857229e-01 -7.35139325e-02 6.42627850e-02 7.10726321e-01 -7.87062496e-02 9.22667265e-01 -8.33756447e-01 -4.47761625e-01 3.41803133e-01 3.17529559e-01 -2.39355519e-01 4.44942266e-01 5.74328661e-01 5.26176631e-01 -1.17350958e-01 6.13592923e-01 8.86039734e-01 -3.87750506e-01 1.29523993e-01 -3.61992955e-01 4.14940268e-02 -1.52423605e-01 -1.43031287e+00 1.56618786e+00 -4.60134476e-01 3.91677111e-01 -1.45757675e-01 -1.04664338e+00 7.80364513e-01 3.47836941e-01 5.80369830e-01 -4.28862274e-01 5.85022867e-02 2.86044270e-01 -3.09570163e-01 -9.19404030e-02 1.44200981e-01 2.01956294e-02 -1.65407211e-01 7.62656331e-01 1.40241995e-01 7.87771881e-01 3.83154541e-01 4.39130142e-02 7.13666916e-01 -1.77268777e-02 3.45399946e-01 7.61069506e-02 7.06094742e-01 -1.24223836e-01 8.59159112e-01 9.11829114e-01 -4.02040094e-01 6.44879699e-01 -2.01799914e-01 -8.01577926e-01 -1.24757183e+00 -6.91360772e-01 -1.05311600e-02 1.03115070e+00 4.83496368e-01 -1.44072339e-01 -8.82877409e-01 -9.37842429e-01 1.05465226e-01 1.41085340e-02 -6.94914699e-01 4.02883478e-02 -7.38267839e-01 -7.71517396e-01 5.92687309e-01 8.03722739e-01 9.40802157e-01 -5.71080029e-01 -5.96371479e-02 -8.44884589e-02 -4.60753590e-01 -1.58512378e+00 -9.60489035e-01 -3.07774127e-01 -4.69409585e-01 -1.37684202e+00 -1.20341253e+00 -1.02503705e+00 1.31670034e+00 7.41501689e-01 8.44181836e-01 2.19746441e-01 -1.11534558e-01 5.18148541e-01 -2.72640824e-01 -1.41344965e-01 9.42751616e-02 1.34508058e-01 4.78134692e-01 5.86094558e-01 8.97184253e-01 -3.83173615e-01 -6.36379063e-01 6.56922817e-01 -3.96175206e-01 -6.42468631e-02 4.27296579e-01 9.16209400e-01 4.09295410e-01 1.36536419e-01 4.91824955e-01 -7.81343997e-01 8.43484402e-02 3.24495099e-02 -5.78507125e-01 5.59085071e-01 -4.38928306e-01 -2.48816237e-01 7.71432877e-01 -7.54001081e-01 -1.00201011e+00 5.52578151e-01 3.33979964e-01 -3.79652828e-01 -2.36431465e-01 -1.86849952e-01 -3.34275335e-01 -4.89334613e-01 3.02342981e-01 2.19524458e-01 -1.98686406e-01 -6.48300946e-01 2.16907918e-01 9.37933505e-01 7.33707011e-01 -2.28239879e-01 1.15151989e+00 7.09946573e-01 -1.72074154e-01 -3.29251617e-01 -9.96118486e-01 -8.73851717e-01 -1.17145014e+00 -1.57287553e-01 6.70267105e-01 -1.37822831e+00 -8.47679019e-01 8.77883315e-01 -1.06641388e+00 9.29117575e-02 -1.36220351e-01 5.10624945e-01 -1.02342673e-01 6.61729634e-01 -4.69458789e-01 -6.43452942e-01 -4.28490937e-01 -9.82995093e-01 1.03012478e+00 5.61038375e-01 -6.35180399e-02 -7.34144032e-01 -1.23916157e-01 5.58806539e-01 2.36459821e-01 -1.10996395e-01 2.65236169e-01 -8.45573187e-01 -4.90207970e-01 -8.44482660e-01 -6.08617425e-01 2.93442905e-01 4.07217801e-01 -6.82018220e-01 -1.03685355e+00 -6.26300216e-01 -4.27677691e-01 -5.31165540e-01 9.01435971e-01 -3.55938040e-02 1.12413883e+00 -1.14316806e-01 -5.01068234e-01 7.66299129e-01 1.19743395e+00 -1.87751234e-01 6.79922253e-02 3.39054525e-01 1.13822079e+00 7.21578598e-01 3.48972172e-01 4.82184261e-01 7.33469844e-01 1.02417099e+00 1.52864438e-02 -3.39182734e-01 -2.76784927e-01 -4.22044843e-01 -6.52124882e-02 3.88186038e-01 -2.87263483e-01 -5.44624627e-02 -8.12383473e-01 6.54097974e-01 -2.04601026e+00 -8.74470770e-01 2.50359714e-01 2.52526379e+00 7.38137305e-01 -3.96512061e-01 3.26964855e-01 -3.51340175e-02 1.42011404e+00 -1.80477589e-01 -7.30959594e-01 5.31114459e-01 -6.25486067e-03 -1.79564402e-01 8.20108175e-01 2.60846466e-01 -1.52189875e+00 9.61423397e-01 5.78063297e+00 5.25025427e-01 -7.85428405e-01 1.11260399e-01 5.12931228e-01 -1.18678868e-01 4.01607513e-01 -1.48918360e-01 -1.21380091e+00 6.60416782e-01 5.75687170e-01 -8.54671523e-02 6.32196426e-01 8.95606399e-01 -1.92880169e-01 -1.75416321e-02 -1.39343047e+00 1.67606318e+00 4.91818905e-01 -9.38546956e-01 -7.76958019e-02 9.09713730e-02 8.38250697e-01 -2.14369103e-01 -1.47695974e-01 1.06723770e-01 3.25380176e-01 -8.80349755e-01 6.05030298e-01 3.54870021e-01 1.06869900e+00 -7.68100977e-01 1.19975984e+00 2.62658954e-01 -1.51626229e+00 -2.66326040e-01 -5.44769049e-01 1.98884355e-03 2.10571527e-01 3.60041976e-01 -4.64645803e-01 3.43401998e-01 8.66504490e-01 8.48273635e-01 -8.54163587e-01 1.18730032e+00 -7.06014484e-02 2.27366850e-01 -3.40313345e-01 3.79131019e-01 -2.21773013e-01 -9.42729786e-02 -1.20310165e-01 9.66423213e-01 3.01349431e-01 -8.04677531e-02 5.89593947e-01 5.85188687e-01 -4.77723956e-01 -8.39887857e-02 -2.98476011e-01 1.23044521e-01 7.20664144e-01 1.24343073e+00 -5.27723134e-01 -5.00032425e-01 -7.33573675e-01 1.34075069e+00 5.43426275e-01 3.55911255e-01 -6.61126196e-01 -2.65464276e-01 6.56514049e-01 3.38662229e-02 2.18736738e-01 -9.35938582e-02 -4.64104116e-02 -1.46397305e+00 2.38800198e-01 -5.79244733e-01 4.11904931e-01 -3.36911589e-01 -1.75895500e+00 4.89512205e-01 -2.49255504e-02 -1.33980000e+00 -2.83486456e-01 -4.73969460e-01 -3.03577960e-01 1.09923780e+00 -1.58066189e+00 -1.63439429e+00 -8.85368407e-01 7.77958453e-01 2.24997208e-01 -4.22024012e-01 8.48361552e-01 7.15425491e-01 -9.22013581e-01 1.14960098e+00 1.57989636e-02 9.30917501e-01 1.12267196e+00 -1.04403389e+00 6.38376594e-01 9.99803007e-01 -3.04241162e-02 8.48637879e-01 1.40491471e-01 -5.22116601e-01 -1.05504835e+00 -1.27874887e+00 1.11471331e+00 -4.54748243e-01 1.51789367e-01 -6.52916849e-01 -5.92854738e-01 7.55699337e-01 -3.27872574e-01 4.01837438e-01 1.04569554e+00 1.63932547e-01 -6.02352738e-01 -3.28507721e-01 -1.11813986e+00 3.92526090e-01 1.42107403e+00 -7.54576862e-01 -2.14624479e-01 3.34385544e-01 2.67903090e-01 -2.34006166e-01 -7.38411546e-01 2.83821881e-01 6.61975145e-01 -6.13946140e-01 1.15583563e+00 -3.57612878e-01 -2.41551772e-01 -4.88061905e-01 1.69751346e-01 -9.41339612e-01 -4.56158966e-01 -2.11260051e-01 6.84883669e-02 1.62747884e+00 2.84332894e-02 -8.02570224e-01 1.04519248e+00 1.07930434e+00 4.49345082e-01 -2.18648195e-01 -8.82705748e-01 -8.99867356e-01 -3.10180306e-01 1.73438430e-01 8.22381794e-01 1.01184595e+00 -3.70467305e-01 1.37723386e-01 -7.64500439e-01 2.92980075e-01 8.64726782e-01 2.23745421e-01 1.07344759e+00 -1.46491706e+00 3.77195850e-02 4.25953753e-02 -5.05665779e-01 -1.07462704e+00 5.01052380e-01 -8.05049181e-01 -1.11837588e-01 -1.23010933e+00 8.55777979e-01 -6.97123528e-01 -2.08814263e-01 8.19350123e-01 -4.18974638e-01 7.28524804e-01 3.09460908e-01 7.73273647e-01 -8.29932749e-01 4.51667517e-01 6.83916569e-01 -3.28175813e-01 7.11553767e-02 1.66266814e-01 -7.32498467e-01 7.43484676e-01 7.24830985e-01 -4.98218149e-01 -2.33145103e-01 -5.51853240e-01 -2.16283351e-01 -5.84883630e-01 6.07852042e-01 -1.19818318e+00 5.08245170e-01 1.49480104e-01 7.80323863e-01 -4.36162710e-01 4.71488327e-01 -9.58600938e-01 8.14944804e-02 2.09988743e-01 -2.26323426e-01 1.07186809e-01 -1.30335957e-01 5.68586648e-01 -1.50666356e-01 -4.04796183e-01 8.37596953e-01 -2.93060064e-01 -8.28315973e-01 7.29527473e-01 7.89545253e-02 -3.14965844e-02 1.06372118e+00 -4.16192502e-01 -1.58903122e-01 -3.33902001e-01 -3.65763932e-01 1.70943365e-01 9.93460417e-01 4.56976980e-01 2.14764625e-01 -1.52703226e+00 -4.80121017e-01 1.88861057e-01 4.66322154e-01 -1.15547195e-01 3.27679038e-01 3.15927655e-01 -1.96732894e-01 5.31729877e-01 -3.06296736e-01 -4.53498214e-01 -1.41224861e+00 6.79005802e-01 5.42969823e-01 -1.66647896e-01 -4.41233695e-01 7.26691902e-01 3.56304318e-01 -6.80370331e-01 3.33139658e-01 5.39994836e-01 -3.29356194e-01 -6.31691962e-02 8.72849524e-01 3.67372304e-01 -9.73389670e-02 -1.08479333e+00 -5.01073778e-01 8.45635593e-01 -3.22893888e-01 9.22107399e-02 1.05866373e+00 -3.71166229e-01 -9.61211920e-02 3.21820825e-02 1.20999277e+00 -1.62516966e-01 -1.48457396e+00 -7.29939759e-01 -2.46315956e-01 -6.83175921e-01 -2.84562856e-01 -4.64975029e-01 -1.10172892e+00 7.14019716e-01 8.41024041e-01 -3.47567558e-01 8.86228621e-01 -2.95474641e-02 9.43846643e-01 6.32222652e-01 6.69279337e-01 -1.18972290e+00 3.38516757e-02 2.98467636e-01 3.07955593e-01 -1.66493678e+00 1.43273309e-01 -5.48262298e-01 -4.51909751e-01 9.86728251e-01 9.20798182e-01 1.58146158e-01 5.38321733e-01 -1.77239642e-01 1.54824480e-01 2.62802243e-01 1.82917058e-01 -5.18411398e-01 2.62301475e-01 8.20692599e-01 2.09178820e-01 6.19304329e-02 1.05742641e-01 5.83015203e-01 -7.08460584e-02 2.68717073e-02 1.90143511e-01 7.32758641e-01 -3.57486270e-02 -1.42536557e+00 -3.88023049e-01 2.92247057e-01 -3.08965981e-01 1.63711552e-02 -5.11566043e-01 5.17389715e-01 5.61253369e-01 1.09906733e+00 1.33550882e-01 -3.31897706e-01 2.42129996e-01 -9.07458737e-02 3.79812628e-01 -6.78085208e-01 -3.26407403e-01 -4.22266841e-01 -9.45919976e-02 -2.59280711e-01 -8.65298569e-01 -6.33911014e-01 -7.44560182e-01 -5.83509803e-01 -5.25342941e-01 -6.13496918e-03 3.94944936e-01 1.02066743e+00 3.57311398e-01 -3.27740349e-02 6.90603018e-01 -9.74927604e-01 -5.17496049e-01 -9.20986831e-01 -4.70305026e-01 8.45643401e-01 1.23695567e-01 -9.03794944e-01 -2.56013758e-02 2.40364373e-01]
[14.748485565185547, 1.0100929737091064]
ec4b3b44-c483-45b8-9bb6-2e1f4b2543c2
graph-to-sequence-learning-using-gated-graph
1806.09835
null
http://arxiv.org/abs/1806.09835v1
http://arxiv.org/pdf/1806.09835v1.pdf
Graph-to-Sequence Learning using Gated Graph Neural Networks
Many NLP applications can be framed as a graph-to-sequence learning problem. Previous work proposing neural architectures on this setting obtained promising results compared to grammar-based approaches but still rely on linearisation heuristics and/or standard recurrent networks to achieve the best performance. In this work, we propose a new model that encodes the full structural information contained in the graph. Our architecture couples the recently proposed Gated Graph Neural Networks with an input transformation that allows nodes and edges to have their own hidden representations, while tackling the parameter explosion problem present in previous work. Experimental results show that our model outperforms strong baselines in generation from AMR graphs and syntax-based neural machine translation.
['Trevor Cohn', 'Gholamreza Haffari', 'Daniel Beck']
2018-06-26
graph-to-sequence-learning-using-gated-graph-1
https://aclanthology.org/P18-1026
https://aclanthology.org/P18-1026.pdf
acl-2018-7
['graph-to-sequence']
['natural-language-processing']
[ 6.80024326e-01 8.48274350e-01 -3.91690910e-01 -2.28284940e-01 -9.57746744e-01 -5.03527105e-01 6.65038347e-01 -1.20681889e-01 -1.28031313e-01 7.76489615e-01 3.80163312e-01 -9.22501326e-01 3.93976629e-01 -9.56182063e-01 -9.34809685e-01 -4.08268332e-01 6.01518117e-02 8.69647622e-01 3.01880967e-02 -4.47350144e-01 -3.10676992e-02 2.17309460e-01 -9.03723598e-01 4.98146564e-01 5.62623978e-01 4.67214167e-01 3.40454131e-01 9.02184725e-01 -3.91818881e-01 9.68671918e-01 -5.18306434e-01 -6.16115391e-01 2.51471102e-01 -6.72466755e-01 -8.22253048e-01 -7.63820484e-02 6.73200309e-01 -8.91253948e-02 -3.58869612e-01 8.36067736e-01 5.18113375e-01 1.70379654e-01 4.41450804e-01 -1.00521898e+00 -1.18877637e+00 1.20581186e+00 -1.74214348e-01 6.58715293e-02 4.23393399e-01 1.29342288e-01 1.43779171e+00 -6.13832772e-01 1.02669144e+00 1.47428203e+00 5.32397747e-01 8.05368483e-01 -1.56401789e+00 -2.34564081e-01 3.68745297e-01 1.35072574e-01 -1.06462491e+00 -5.70813775e-01 8.22596908e-01 5.59857711e-02 2.00340676e+00 1.07103683e-01 6.27659798e-01 1.49597943e+00 2.33546630e-01 7.16634989e-01 8.03852260e-01 -8.13042402e-01 -7.44773224e-02 -2.62419671e-01 6.77613020e-02 9.33771193e-01 1.61263570e-01 1.01934463e-01 -3.17807078e-01 1.82978008e-02 8.73816073e-01 -5.49699247e-01 -2.10861623e-01 -2.65817523e-01 -1.06593645e+00 1.20465374e+00 6.12552524e-01 3.05573344e-01 -3.35565358e-01 6.11029148e-01 2.97382265e-01 5.01245260e-01 4.26880687e-01 7.07325459e-01 -3.32755923e-01 1.17780522e-01 -8.15088272e-01 1.68598145e-01 1.00981426e+00 1.14282119e+00 7.39948630e-01 5.84832013e-01 -3.19584221e-01 6.40486121e-01 2.33794674e-01 3.04214120e-01 5.15548348e-01 -7.25683272e-01 8.97246599e-01 6.60850167e-01 -3.54576677e-01 -9.29910541e-01 -4.31208819e-01 -4.78353202e-01 -6.89292014e-01 -3.91910940e-01 2.71315694e-01 -2.30907440e-01 -1.37645710e+00 1.84912145e+00 -9.17047709e-02 1.74147397e-01 2.01461792e-01 6.46244407e-01 8.13490272e-01 9.57940578e-01 -1.70480814e-02 -1.83067322e-01 8.37495387e-01 -1.20500755e+00 -7.06983685e-01 -6.13125265e-01 8.89919698e-01 -4.87760693e-01 1.07282734e+00 1.26201972e-01 -1.41948843e+00 -3.49845707e-01 -8.73674273e-01 -2.82003641e-01 -3.52921277e-01 3.83091345e-02 6.86436057e-01 5.56884527e-01 -1.83511043e+00 7.05715299e-01 -7.74971783e-01 -4.78826225e-01 -6.23296387e-02 6.04044080e-01 -2.32451841e-01 6.97103813e-02 -1.28249907e+00 1.08843589e+00 8.82435203e-01 3.54474604e-01 -6.19533002e-01 -1.79873258e-01 -1.22716260e+00 1.06653452e-01 5.59340119e-01 -1.11488318e+00 1.35202050e+00 -9.47373569e-01 -1.80275118e+00 6.16164744e-01 -2.81881213e-01 -1.07111788e+00 -8.07632040e-03 -2.54276879e-02 -8.72782394e-02 7.75728226e-02 -3.97618890e-01 8.28248739e-01 6.40043736e-01 -9.53813970e-01 -1.56437755e-01 -1.62036568e-01 1.68086097e-01 2.88461119e-01 -1.85861364e-01 4.46349494e-02 -4.01888192e-01 -7.42350221e-01 -8.77238661e-02 -1.15841722e+00 -5.88404238e-01 -9.38020349e-01 -6.83079362e-01 -2.95403183e-01 4.86810386e-01 -8.33226502e-01 1.30841088e+00 -1.40565765e+00 6.97609305e-01 1.77202716e-01 1.70821786e-01 4.08527195e-01 -6.31994069e-01 8.42068672e-01 -6.83140904e-02 3.91345859e-01 -1.01831079e-01 -3.63952756e-01 1.36693969e-01 6.54964805e-01 -4.53688085e-01 -5.42389490e-02 4.94603932e-01 1.65890753e+00 -7.91640043e-01 -4.23860967e-01 5.43464459e-02 5.41158140e-01 -5.37590921e-01 2.87977010e-01 -7.36107528e-01 1.36481881e-01 -2.16617242e-01 4.30057734e-01 1.08541824e-01 -4.42982376e-01 7.57636487e-01 1.58827379e-01 3.90451372e-01 9.03515339e-01 -6.86546981e-01 1.88583791e+00 -5.99314213e-01 5.07049143e-01 -4.62979600e-02 -1.15774465e+00 1.08027852e+00 3.96529496e-01 -2.01243803e-01 -7.05426633e-01 8.40799659e-02 2.09989116e-01 2.47375771e-01 -9.59972739e-02 7.28807688e-01 -6.05634041e-02 3.40393931e-02 3.70599121e-01 4.42422092e-01 -1.99493885e-01 3.87285411e-01 4.22610521e-01 1.20443118e+00 6.02282405e-01 2.43356735e-01 2.59444639e-02 2.44791001e-01 -2.93052793e-02 3.11764657e-01 9.59445596e-01 4.35977221e-01 5.89037359e-01 6.65007770e-01 -4.17625546e-01 -1.21236622e+00 -6.72075331e-01 7.04993010e-01 1.22908509e+00 -5.93444705e-01 -6.81940496e-01 -8.82530928e-01 -7.59713769e-01 -4.86347139e-01 1.05859339e+00 -6.30115449e-01 -2.09716216e-01 -1.27128375e+00 -6.08635008e-01 6.41191006e-01 5.94340146e-01 -1.04119606e-01 -1.46134222e+00 -2.47419417e-01 5.72857380e-01 -1.87671557e-01 -1.27834237e+00 -4.19495791e-01 4.06099856e-01 -1.00947356e+00 -5.92997789e-01 -6.09916329e-01 -1.00541759e+00 5.32740891e-01 -1.85018286e-01 1.57060611e+00 2.16098487e-01 2.13134158e-02 -1.57151505e-01 -4.51496631e-01 -2.37713307e-01 -8.71725023e-01 7.79705644e-01 -5.31472683e-01 -2.70361811e-01 1.14089482e-01 -7.54027128e-01 -8.77217948e-02 -2.18033701e-01 -8.18589926e-01 4.32866305e-01 7.82251537e-01 9.69490945e-01 3.35287809e-01 -7.21064687e-01 5.56070805e-01 -1.26986396e+00 9.40155149e-01 -1.93012148e-01 -5.26485085e-01 3.86558145e-01 -6.92099929e-01 5.99795818e-01 8.52304339e-01 -2.65617251e-01 -8.62329364e-01 1.91982105e-01 -2.27768511e-01 -3.59538853e-01 -1.19210616e-01 8.49588275e-01 -1.39328495e-01 1.74204648e-01 6.75473094e-01 3.03954363e-01 -2.39896297e-01 -3.15689176e-01 8.33304882e-01 2.59162843e-01 4.81305003e-01 -6.66905105e-01 7.31971562e-01 -2.10685387e-01 1.28959775e-01 -4.36378300e-01 -6.88513041e-01 -7.17438981e-02 -6.26162767e-01 2.37849161e-01 6.60829663e-01 -6.68829739e-01 -1.39334217e-01 4.94248495e-02 -1.58521855e+00 -7.47839630e-01 -2.20490992e-01 2.45146677e-01 -7.07333744e-01 4.76480454e-01 -1.06006503e+00 -7.19970167e-01 -5.89995444e-01 -1.08636200e+00 1.16941917e+00 -2.54680932e-01 -3.33096534e-01 -1.21691382e+00 2.64620274e-01 1.30205806e-02 5.66030622e-01 3.33495915e-01 1.19314837e+00 -9.18093145e-01 -7.39193201e-01 2.52480041e-02 -1.56842038e-01 6.84925839e-02 -2.24407643e-01 -2.36623600e-01 -5.87182641e-01 -2.81875283e-01 -4.41989750e-01 -4.15933609e-01 1.06991160e+00 3.85581255e-01 5.87455153e-01 -7.58479893e-01 -2.48951793e-01 6.84340477e-01 1.26944876e+00 3.78010944e-02 7.74867237e-01 1.70461267e-01 9.84029353e-01 5.48551261e-01 -1.94427446e-01 -1.65157050e-01 3.89456093e-01 7.30761170e-01 5.10299921e-01 -2.01088145e-01 -4.14164454e-01 -6.22864008e-01 6.61462247e-01 1.09568214e+00 -1.34786740e-01 -7.94508755e-01 -1.02484274e+00 6.01952493e-01 -2.05307293e+00 -7.62571037e-01 -1.14992991e-01 1.81470978e+00 7.18818426e-01 1.61857828e-01 4.61984891e-03 -3.14184844e-01 7.50698149e-01 3.55563641e-01 -1.14221551e-01 -1.12020862e+00 -2.11690754e-01 6.43044531e-01 5.80712378e-01 7.23205030e-01 -6.98196173e-01 1.58841228e+00 7.08820486e+00 4.78372008e-01 -1.02693975e+00 -1.03997283e-01 3.90256464e-01 2.16591787e-02 -5.66989422e-01 1.98762789e-01 -8.66207242e-01 -1.21688098e-01 1.59055424e+00 -7.52625689e-02 7.44585872e-01 5.89171410e-01 2.19975486e-02 5.37610888e-01 -1.16215134e+00 6.24052525e-01 3.40144515e-01 -1.44630694e+00 6.81470215e-01 1.90028891e-01 5.82820535e-01 2.85828650e-01 -4.35427502e-02 5.06671846e-01 9.81567800e-01 -1.39886403e+00 5.31658888e-01 1.42153695e-01 6.32995844e-01 -6.38227165e-01 5.85652173e-01 2.82594919e-01 -1.07634664e+00 9.36545506e-02 -4.65772092e-01 -3.03955197e-01 3.58487904e-01 1.98694974e-01 -1.44994187e+00 8.60970318e-01 -6.44022748e-02 6.79380774e-01 -6.52887404e-01 5.03570557e-01 -6.91553891e-01 8.70420516e-01 -2.20445946e-01 -1.62530705e-01 7.19833970e-01 -2.98889518e-01 5.97864270e-01 1.51286960e+00 4.06197727e-01 -2.69517452e-01 3.66176158e-01 9.60276186e-01 -4.09015447e-01 2.93377072e-01 -1.13326192e+00 -4.37333316e-01 5.37802726e-02 1.19709492e+00 -6.96756124e-01 -4.60884869e-01 -3.24533880e-01 9.19760346e-01 8.94759476e-01 6.08372033e-01 -6.68026865e-01 -2.70440549e-01 -8.49567503e-02 5.54315187e-02 5.35221219e-01 -4.88650501e-01 8.65987465e-02 -1.35869670e+00 -5.04724868e-02 -1.11178410e+00 5.11558354e-01 -7.86344528e-01 -8.87275934e-01 1.03160930e+00 -3.15559916e-02 -3.93851072e-01 -1.10191393e+00 -7.11614072e-01 -5.16185284e-01 9.01817024e-01 -1.38347435e+00 -1.60608399e+00 4.71859246e-01 3.44028801e-01 5.00492573e-01 -1.27461627e-01 1.17472947e+00 -1.72935516e-01 -5.50751507e-01 6.11406326e-01 -3.49365503e-01 3.33936930e-01 3.51559579e-01 -1.46116459e+00 1.24700642e+00 1.18494320e+00 8.62979054e-01 8.03779364e-01 6.40602529e-01 -8.27238679e-01 -1.64732957e+00 -1.23856103e+00 1.24267471e+00 -3.83986264e-01 7.39010930e-01 -6.98845029e-01 -9.51710880e-01 1.33560836e+00 7.31445968e-01 -2.13268831e-01 2.41915226e-01 3.27642053e-01 -5.00530362e-01 5.45003235e-01 -4.37423587e-01 7.08318293e-01 1.29885352e+00 -5.21081984e-01 -7.61057377e-01 3.21717680e-01 1.08460295e+00 -6.87277317e-01 -4.78145987e-01 3.29966724e-01 1.42785504e-01 -4.71404076e-01 6.77187800e-01 -1.06899512e+00 4.26250011e-01 2.35898383e-02 1.22621553e-02 -1.62176585e+00 -3.98240894e-01 -1.17228639e+00 -4.99779731e-01 9.36239064e-01 1.00002098e+00 -4.93349820e-01 8.46492350e-01 2.21924126e-01 -3.81575972e-01 -6.52628899e-01 -7.32241631e-01 -8.10038686e-01 2.23281175e-01 -3.01492870e-01 4.54376280e-01 7.51266241e-01 -6.19840771e-02 1.07792950e+00 -6.13314688e-01 -1.29231736e-01 3.04065526e-01 1.73922062e-01 6.86778665e-01 -8.93352449e-01 -6.49642587e-01 -4.98611957e-01 -1.68710023e-01 -9.02625799e-01 5.89609146e-01 -1.59115696e+00 8.20464194e-02 -2.10619426e+00 -3.89213748e-02 1.55586705e-01 -1.95457757e-01 8.82349968e-01 -1.61298662e-01 2.10522816e-01 3.35368276e-01 -1.86399713e-01 -5.55134773e-01 2.86164373e-01 9.52912927e-01 -1.06835157e-01 -3.28017563e-01 -5.49475491e-01 -6.43084466e-01 3.48964036e-01 1.00631762e+00 -4.27611798e-01 -3.90123248e-01 -8.40580285e-01 6.91324115e-01 1.90002829e-01 1.85948089e-01 -4.85543281e-01 1.19312838e-01 -7.95547888e-02 6.31915918e-03 -2.17509732e-01 2.43148088e-01 -2.40050003e-01 1.84621170e-01 3.13284278e-01 -6.84139192e-01 4.78479087e-01 1.69717818e-01 4.18900728e-01 -1.14961535e-01 -1.06122278e-01 3.39395821e-01 -5.42580843e-01 -4.35291559e-01 1.61507443e-01 -3.54990155e-01 5.26473746e-02 4.33616251e-01 7.38367140e-02 -3.06760550e-01 -7.36008704e-01 -8.70796859e-01 1.60217341e-02 4.21128035e-01 5.69833636e-01 5.82072854e-01 -1.13387072e+00 -9.63683844e-01 7.37139257e-04 -1.52899727e-01 -1.26499817e-01 -4.63346362e-01 5.85601151e-01 -3.90939683e-01 7.55052567e-01 -1.48122713e-01 -1.98793530e-01 -9.62720096e-01 7.21019030e-01 2.41872311e-01 -1.01277578e+00 -6.77649498e-01 5.56018651e-01 -7.64652491e-02 -6.40405476e-01 5.89074939e-02 -4.28889632e-01 -9.06101465e-02 -1.38498470e-01 1.83598399e-01 3.66917364e-02 2.62949616e-01 -6.39681935e-01 -4.11456302e-02 2.81014681e-01 -2.84146309e-01 -2.73058563e-01 1.31858516e+00 8.13203231e-02 -1.43857569e-01 3.47156644e-01 1.03447270e+00 -1.84005558e-01 -7.53261268e-01 -3.84369344e-01 3.08846027e-01 1.66046172e-01 1.23211900e-02 -7.97689259e-01 -9.68769908e-01 1.06628048e+00 -8.54033753e-02 1.49032339e-01 7.61043310e-01 -6.04462363e-02 1.06115949e+00 7.65472710e-01 1.37681335e-01 -1.01201224e+00 -1.47014111e-01 9.49969530e-01 8.27360153e-01 -8.11534405e-01 -3.62731278e-01 -3.63297671e-01 -4.92523521e-01 1.24260485e+00 4.33587283e-01 -4.05574650e-01 -8.63347426e-02 3.90111387e-01 7.28887767e-02 -5.17419465e-02 -1.21802139e+00 -4.44508284e-01 4.38710868e-01 6.64453804e-01 7.19762504e-01 6.02085888e-02 -3.92179638e-01 2.29888439e-01 -3.73813540e-01 -7.30758384e-02 4.63377953e-01 6.86012506e-01 -2.37638280e-01 -1.65419972e+00 6.91484958e-02 2.91717172e-01 -5.59757948e-01 -7.13747978e-01 -8.33812892e-01 8.62420082e-01 -4.58165556e-01 8.41534257e-01 -4.57752571e-02 -2.88263470e-01 2.02569440e-01 5.36382675e-01 7.41372049e-01 -1.06804287e+00 -9.14108872e-01 2.69908160e-01 6.90829754e-01 -7.53274977e-01 -2.79157192e-01 -2.92741865e-01 -1.21121931e+00 1.65291071e-01 -3.52969468e-01 1.67376831e-01 5.34501493e-01 6.71997964e-01 4.64765072e-01 5.68306744e-01 1.69529006e-01 -7.95861423e-01 -4.85285044e-01 -1.11487162e+00 -1.82757038e-03 2.85739124e-01 1.38892710e-01 2.76983739e-03 -5.31962216e-02 5.76990424e-03]
[10.301071166992188, 8.383410453796387]
ff48c1db-d1c8-40a7-8b2c-0c2a9926752e
faster-lead-optimization-mapper-algorithm-for
2304.04713
null
https://arxiv.org/abs/2304.04713v1
https://arxiv.org/pdf/2304.04713v1.pdf
Faster Lead Optimization Mapper Algorithm for Large-Scale Relative Free Energy Perturbation
In recent years, free energy perturbation (FEP) calculations have garnered increasing attention as tools to support drug discovery. The lead optimization mapper (Lomap) was proposed as an algorithm to calculate the relative free energy between ligands efficiently. However, Lomap requires checking whether each edge in the FEP graph is removable, which necessitates checking the constraints for all edges. Consequently, conventional Lomap requires significant computation time, at least several hours for cases involving hundreds of compounds, and is impractical for cases with more than tens of thousands of edges. In this study, we aimed to reduce the computational cost of Lomap to enable the construction of FEP graphs for hundreds of compounds. We can reduce the overall number of constraint checks required from an amount dependent on the number of edges to one dependent on the number of nodes by using the chunk check process to check the constraints for as many edges as possible simultaneously. Moreover, the output graph is equivalent to that obtained using conventional Lomap, enabling direct replacement of the original Lomap with our method. With our improvement, the execution was tens to hundreds of times faster than that of the original Lomap. https://github.com/ohuelab/FastLomap
['Masahito Ohue', 'Kairi Furui']
2023-04-10
null
null
null
null
['drug-discovery']
['medical']
[ 1.31592274e-01 -1.68310869e-02 5.56222871e-02 1.22885182e-01 -5.70604801e-01 -6.43512309e-01 5.47196977e-02 8.68045270e-01 -3.79145890e-01 1.13020742e+00 -3.64072382e-01 -6.09835386e-01 -1.25833124e-01 -9.69467878e-01 -7.00826347e-01 -6.59231365e-01 -1.37594253e-01 4.04724777e-01 5.15100777e-01 3.15720635e-03 4.35408145e-01 6.68492734e-01 -1.18663406e+00 -4.44440357e-02 1.22826409e+00 6.89280272e-01 2.10702151e-01 1.53297126e-01 9.60522518e-03 1.03188731e-01 -3.19436729e-01 -2.16044396e-01 3.26679200e-02 -5.10637283e-01 -7.71429896e-01 -3.72522116e-01 1.05362624e-01 1.62159160e-01 -9.17034806e-04 1.17829192e+00 7.61130691e-01 1.46128625e-01 3.97725403e-01 -1.04755712e+00 -7.14400187e-02 3.62957239e-01 -4.62800354e-01 -1.34581223e-01 4.86947477e-01 -5.96655160e-02 7.77826786e-01 -1.03261995e+00 6.96235716e-01 7.30712056e-01 5.36421299e-01 2.07844988e-01 -1.44318187e+00 -7.18468189e-01 1.56864605e-03 5.87747037e-01 -1.81341910e+00 -3.46932322e-01 4.83614594e-01 -4.14111078e-01 1.15439105e+00 5.45691967e-01 7.55017221e-01 4.22636181e-01 1.98347300e-01 -6.06269613e-02 8.68238509e-01 -3.81966978e-01 6.29818082e-01 -1.41184434e-01 4.16761674e-02 7.77621925e-01 5.49408913e-01 -1.78756803e-01 -6.21570408e-01 -7.00095356e-01 3.83931309e-01 -1.72399819e-01 -4.05563325e-01 -3.97997379e-01 -8.75903249e-01 8.16027999e-01 4.88493472e-01 2.12406397e-01 -3.86955887e-01 -6.17807731e-02 3.30943257e-01 -1.28701806e-01 9.19267237e-02 7.59928167e-01 -3.19154769e-01 -4.54301713e-03 -8.65050852e-01 1.47742644e-01 8.92454624e-01 6.19871736e-01 8.48061383e-01 -4.30573046e-01 2.12180883e-01 3.84673029e-01 -5.97190559e-02 1.50720298e-01 -3.49453837e-02 -5.17032146e-01 3.22767168e-01 7.99915314e-01 3.47819507e-01 -1.20972872e+00 -6.01947188e-01 -2.06888452e-01 -1.00789917e+00 -3.15679163e-02 6.31309807e-01 -8.71455744e-02 -4.79286373e-01 1.56686330e+00 5.70678949e-01 -2.36057609e-01 -3.37845534e-01 6.66255355e-01 5.89519322e-01 7.90822446e-01 8.48119557e-02 -6.93047762e-01 1.29255605e+00 -7.62350142e-01 -4.81735349e-01 2.69078556e-02 7.80820727e-01 -7.83090711e-01 8.25736582e-01 5.27050436e-01 -9.06318009e-01 2.52308939e-02 -9.60886478e-01 1.91040590e-01 -4.03657049e-01 -4.31204773e-02 6.61802471e-01 4.19825226e-01 -7.69566953e-01 9.55534279e-01 -7.20011353e-01 -5.17588519e-02 3.19198787e-01 7.14164257e-01 -5.94039977e-01 -1.12335443e-01 -1.23324263e+00 6.91344023e-01 6.70735538e-01 2.74115771e-01 -3.35477799e-01 -8.19526494e-01 -6.96670592e-01 3.25024933e-01 7.57171392e-01 -3.63779992e-01 5.68838179e-01 -3.27718973e-01 -1.33136570e+00 8.96804631e-02 -4.71461028e-01 -2.17266730e-03 3.37590218e-01 3.18211883e-01 -1.63707659e-01 1.27573147e-01 -5.88050187e-02 3.16045403e-01 1.41028851e-01 -6.50072336e-01 2.74029132e-02 -3.24419737e-01 -1.42660826e-01 4.61040698e-02 -1.76422298e-01 7.42805153e-02 -5.82124114e-01 -2.76073098e-01 2.26108536e-01 -1.04668331e+00 -5.38315356e-01 -8.91095251e-02 -6.12543285e-01 -2.12277472e-01 2.93789625e-01 -5.70293903e-01 1.64072132e+00 -1.91836786e+00 2.77782470e-01 7.40104795e-01 3.30824345e-01 2.31507316e-01 1.45862987e-02 9.31711316e-01 -2.24020630e-01 2.55528808e-01 -4.21545893e-01 4.94758576e-01 -2.42785230e-01 -3.04687381e-01 2.46408775e-01 5.64875066e-01 -8.17602575e-02 6.56213999e-01 -8.04786861e-01 -3.10900271e-01 3.85967046e-02 5.67320347e-01 -7.44268954e-01 -2.99429744e-01 -2.65461445e-01 5.09034455e-01 -5.60626507e-01 4.67367887e-01 1.01955295e+00 -6.47955775e-01 7.87418425e-01 -2.78525889e-01 -4.26524073e-01 4.41527128e-01 -1.29995251e+00 1.30700195e+00 -1.02450654e-01 4.67522368e-02 -2.82796830e-01 -6.44670069e-01 7.94769943e-01 2.56623685e-01 7.03827322e-01 -6.61292076e-01 -4.29349579e-02 5.59704661e-01 1.32806793e-01 1.24225147e-01 2.59636156e-02 -9.17557701e-02 -7.99754634e-02 3.23511422e-01 -4.95699853e-01 3.39969933e-01 5.70538163e-01 2.31724247e-01 1.09198654e+00 -7.21268728e-02 5.32491386e-01 -3.11703175e-01 7.62289584e-01 1.46384344e-01 8.02596152e-01 1.19984768e-01 1.85398936e-01 1.47516513e-03 7.79840112e-01 -5.83120704e-01 -8.18097234e-01 -5.46109498e-01 -3.19132566e-01 4.14024204e-01 2.74020016e-01 -1.27882171e+00 -8.47169876e-01 -3.62268925e-01 -6.62228316e-02 2.95157582e-01 -3.17771524e-01 3.93386409e-02 -5.21547973e-01 -1.04731369e+00 1.53107613e-01 1.59038320e-01 4.26470011e-01 -9.15785372e-01 -4.79570657e-01 5.89375556e-01 -1.05100544e-02 -7.50070274e-01 -4.73604172e-01 2.65311062e-01 -5.79686224e-01 -1.43969667e+00 -3.87958854e-01 -4.17370677e-01 1.02500498e+00 -5.43979555e-02 5.88019609e-01 2.28116348e-01 -4.47694987e-01 -6.98294163e-01 -1.25230670e-01 -1.74339190e-01 -1.26193389e-01 -1.91196576e-02 5.69216460e-02 -4.02363688e-01 2.19222501e-01 -7.85707474e-01 -8.17856967e-01 4.97956395e-01 -6.94186389e-01 2.00586379e-01 2.80439526e-01 6.87547982e-01 1.00285113e+00 3.27128142e-01 5.90189099e-01 -8.37656915e-01 4.38585013e-01 -1.85341030e-01 -1.12358761e+00 2.92486429e-01 -7.71770239e-01 2.83839643e-01 8.99708927e-01 -2.69389153e-01 -4.68947411e-01 3.68780136e-01 -2.95014590e-01 8.86235312e-02 4.29430485e-01 8.89106393e-01 -2.92573750e-01 -4.59282547e-01 4.67514306e-01 1.32253706e-01 -1.63515985e-01 -4.88502830e-01 4.26750556e-02 4.21535343e-01 -5.87763898e-02 -4.06328768e-01 4.57965642e-01 1.60645787e-02 6.22271419e-01 -6.63017750e-01 -2.40069732e-01 -1.38823271e-01 -4.11580026e-01 6.63736686e-02 6.21948421e-01 -4.56550688e-01 -1.29887927e+00 1.82912320e-01 -1.02239144e+00 -2.20060065e-01 2.17148930e-01 4.38646525e-01 -7.03917816e-02 6.96246862e-01 -4.26305324e-01 -2.86275566e-01 -5.57683945e-01 -1.47809339e+00 3.87445480e-01 -5.08469641e-02 -2.89981276e-01 -6.98625982e-01 6.02241904e-02 1.07534133e-01 3.48599583e-01 6.42986417e-01 1.21846938e+00 -2.67017752e-01 -6.91078722e-01 -3.35345834e-01 -1.35336578e-01 -3.47752243e-01 2.79771954e-01 -9.29632112e-02 -3.10341030e-01 -4.62509632e-01 -3.47440809e-01 6.19727792e-03 5.06532431e-01 2.82350957e-01 1.37276685e+00 -4.48712766e-01 -6.53827190e-01 3.92918050e-01 1.48025501e+00 3.57713103e-01 7.59452641e-01 2.57908970e-01 6.19492829e-01 3.40709925e-01 6.69062495e-01 5.53138793e-01 -4.78441492e-02 1.01263964e+00 2.73516327e-01 -9.68653932e-02 3.09156686e-01 -1.26194537e-01 1.48876280e-01 5.65176249e-01 -2.21112877e-01 -2.91299254e-01 -9.80194807e-01 4.00878377e-02 -1.80802035e+00 -7.49715507e-01 -3.44131112e-01 2.63635135e+00 1.06604552e+00 -1.47486106e-01 1.35001257e-01 1.99716970e-01 7.23997593e-01 -2.01173514e-01 -7.55124509e-01 -4.64268267e-01 3.08586866e-01 3.06341529e-01 5.65015078e-01 7.75972068e-01 -5.29154241e-01 7.89953470e-01 5.94499445e+00 9.33535397e-01 -1.16034019e+00 -2.58337379e-01 4.21039850e-01 -1.17786355e-01 -1.68525606e-01 3.90442401e-01 -7.24157870e-01 6.63120329e-01 9.87285972e-01 -4.77376759e-01 4.50723708e-01 5.55744767e-01 3.46176416e-01 -4.51812983e-01 -8.94677281e-01 9.67881620e-01 -3.55762810e-01 -1.46702409e+00 1.33981202e-02 4.14142787e-01 4.70786989e-01 -2.94003546e-01 -4.51888204e-01 -3.80053520e-01 -1.57980204e-01 -1.07214594e+00 1.63295388e-01 9.11476910e-02 9.06273246e-01 -1.12464380e+00 5.85548937e-01 3.11338514e-01 -1.41032851e+00 3.73556405e-01 -4.46347505e-01 5.77004105e-02 1.93850577e-01 9.22224939e-01 -7.84466624e-01 8.03159893e-01 5.07873297e-01 3.67922336e-01 -3.81246716e-01 1.17021024e+00 -1.35142118e-01 9.60593298e-02 -6.01833045e-01 -7.56126121e-02 3.62770911e-03 -5.60638070e-01 3.45043242e-01 7.18586087e-01 3.74346048e-01 1.48225710e-01 2.37634838e-01 6.95817471e-01 -1.56995580e-01 4.87704426e-01 -1.28497124e-01 -2.01235056e-01 7.31989384e-01 1.06635571e+00 -7.81808078e-01 -2.54371077e-01 -1.64001405e-01 7.33650625e-01 4.56641883e-01 9.95961800e-02 -8.97533119e-01 -6.31954372e-01 3.48104239e-01 5.29364407e-01 1.90714672e-01 -1.43895701e-01 1.26698881e-01 -7.99166262e-01 1.53028056e-01 -8.80242944e-01 3.50462019e-01 -3.78286242e-01 -7.94763327e-01 7.23760307e-01 -1.90173015e-01 -9.47841942e-01 1.43690720e-01 -5.41646123e-01 -4.08866018e-01 1.01894259e+00 -1.25652206e+00 -2.80916780e-01 -2.45222896e-01 4.48022962e-01 -8.21282193e-02 4.52933311e-01 1.12659216e+00 4.59580988e-01 -9.73728895e-01 5.71676433e-01 1.32148966e-01 -5.04519761e-01 6.76077306e-01 -7.89262831e-01 1.01765364e-01 5.22961199e-01 -3.88835162e-01 7.45909333e-01 6.74754918e-01 -7.77695239e-01 -1.44683576e+00 -8.33797038e-01 1.15180802e+00 1.73251376e-01 4.71621096e-01 -3.94922942e-01 -1.04098785e+00 9.63896587e-02 -1.58177137e-01 -6.08603954e-02 9.32425380e-01 -1.48908868e-02 -9.97952744e-02 8.13693777e-02 -1.04135549e+00 6.61383748e-01 9.17528152e-01 -1.69368610e-01 3.48519355e-01 6.81074023e-01 3.07036698e-01 -5.55763602e-01 -1.21389198e+00 3.87911171e-01 3.14330637e-01 -7.23725140e-01 7.97592461e-01 -1.53340101e-01 -5.78006953e-02 -8.48812521e-01 1.69713289e-01 -9.26277220e-01 -5.51904202e-01 -8.01725209e-01 -6.20033592e-02 6.66518331e-01 7.99640715e-01 -8.66515875e-01 6.38956070e-01 8.26061964e-01 1.55133465e-02 -1.16420245e+00 -1.05500412e+00 -7.35722780e-01 -1.96202815e-01 7.33860359e-02 7.98885047e-01 8.61049116e-01 5.63178480e-01 4.58566815e-01 -1.81205004e-01 3.92586514e-02 3.09489369e-01 2.33236298e-01 5.23072183e-01 -1.28965545e+00 -5.44332325e-01 -1.65875450e-01 -2.99896449e-01 -5.95069349e-01 -2.47378498e-01 -1.10667264e+00 -3.84356856e-01 -1.59958220e+00 4.02642727e-01 -6.62421584e-01 -1.71128318e-01 8.45729291e-01 -1.29164413e-01 2.98285186e-01 -1.50592513e-02 1.08643822e-01 -4.03559238e-01 3.78752798e-01 1.23452508e+00 5.73424622e-02 -4.79614139e-01 -2.57586241e-01 -7.35038400e-01 5.54363251e-01 9.86123741e-01 -6.53349161e-01 -1.69204533e-01 9.05605480e-02 8.84331644e-01 4.34051529e-02 6.86253682e-02 -7.74902225e-01 1.81761324e-01 -1.99355617e-01 3.15126449e-01 -6.79536104e-01 1.46712512e-01 -7.13149548e-01 9.46015120e-01 7.35307992e-01 2.16212615e-01 -1.16152167e-02 4.79484111e-01 3.78553778e-01 -1.37410507e-01 -1.79046839e-01 5.58047950e-01 4.13315110e-02 -1.90667778e-01 4.22034830e-01 -2.23220780e-01 -4.86405969e-01 1.16904700e+00 -2.21472964e-01 -3.44113111e-01 1.55549005e-01 -8.21861863e-01 2.64347166e-01 7.34367788e-01 -3.74316186e-01 5.30771077e-01 -1.16693604e+00 -1.60552427e-01 -9.28153917e-02 4.33031023e-02 2.08508208e-01 2.70725876e-01 1.14227414e+00 -9.32511747e-01 6.44670010e-01 -7.48431534e-02 -1.79986328e-01 -1.66621661e+00 8.43369782e-01 2.43742019e-01 -5.03297627e-01 -6.99682593e-01 3.64730775e-01 -3.39709334e-02 1.20343924e-01 -1.62844330e-01 -8.70210119e-03 8.17551464e-02 -1.23508632e-01 5.66913128e-01 5.43031752e-01 2.86105186e-01 -2.89700538e-01 -8.52621377e-01 6.64484560e-01 -2.63369799e-01 4.72293615e-01 1.41649127e+00 3.34999770e-01 -4.84074235e-01 -3.43498021e-01 9.79844928e-01 3.36778075e-01 -9.99233246e-01 2.16418713e-01 -1.51297942e-01 -3.31168979e-01 5.09532876e-02 -6.81913018e-01 -6.72385931e-01 4.94832933e-01 2.80829132e-01 -2.28584230e-01 1.20511067e+00 -7.42230639e-02 6.68784916e-01 5.14124751e-01 6.65765584e-01 -8.81038725e-01 -1.43941864e-01 1.58661559e-01 7.71368623e-01 -8.71594191e-01 5.32855749e-01 -9.75180924e-01 -2.23224327e-01 1.08743799e+00 2.49982730e-01 1.96100935e-01 4.79329824e-01 -9.97891743e-03 -7.46067941e-01 -3.73564154e-01 -4.61456090e-01 2.42540717e-01 2.94113964e-01 4.93549518e-02 3.82964700e-01 3.41477990e-02 -9.79124546e-01 4.39023823e-01 3.18067968e-02 5.74661605e-02 3.56799424e-01 1.02198195e+00 -5.25206566e-01 -1.85266113e+00 -2.19201252e-01 2.56782591e-01 -2.56904602e-01 -4.15881932e-01 -6.32711351e-01 6.58510804e-01 3.11816018e-02 9.29886281e-01 -3.20931882e-01 -1.34539321e-01 3.80583018e-01 6.00234382e-02 6.11917794e-01 -4.04989630e-01 -3.27830940e-01 1.41411498e-01 1.50001779e-01 -7.10959673e-01 -1.52195513e-01 -1.50095940e-01 -1.60728014e+00 -6.97534382e-01 -6.87295377e-01 6.79713011e-01 4.89020705e-01 6.04956150e-01 8.41128588e-01 3.33050609e-01 6.23079717e-01 -5.32019258e-01 -2.69716270e-02 -6.08610451e-01 -4.94143814e-01 -2.74702087e-02 -2.97840744e-01 -7.61239707e-01 -1.32066131e-01 -3.69913936e-01]
[4.951566696166992, 5.599321365356445]
302c9849-8058-4e9a-a247-f216314b0e9f
exploring-visual-prompts-for-whole-slide
2303.13122
null
https://arxiv.org/abs/2303.13122v1
https://arxiv.org/pdf/2303.13122v1.pdf
Exploring Visual Prompts for Whole Slide Image Classification with Multiple Instance Learning
Multiple instance learning (MIL) has emerged as a popular method for classifying histopathology whole slide images (WSIs). However, existing approaches typically rely on pre-trained models from large natural image datasets, such as ImageNet, to generate instance features, which can be sub-optimal due to the significant differences between natural images and histopathology images that lead to a domain shift. In this paper, we present a novel, simple yet effective method for learning domain-specific knowledge transformation from pre-trained models to histopathology images. Our approach entails using a prompt component to assist the pre-trained model in discerning differences between the pre-trained dataset and the target histopathology dataset, resulting in improved performance of MIL models. We validate our method on two publicly available datasets, Camelyon16 and TCGA-NSCLC. Extensive experimental results demonstrate the significant performance improvement of our method for different MIL models and backbones. Upon publication of this paper, we will release the source code for our method.
['Hao Chen', 'Kwang-Ting Cheng', 'Lisheng Wang', 'Zhengjie ZHU', 'Zhongchen Zhao', 'Yi Lin']
2023-03-23
null
null
null
null
['whole-slide-images', 'multiple-instance-learning']
['computer-vision', 'methodology']
[ 6.94197893e-01 1.68931052e-01 -3.43243927e-01 -4.92616594e-01 -1.36080468e+00 -4.66121405e-01 4.01963830e-01 4.17251408e-01 -5.07991493e-01 7.14083135e-01 -2.73634563e-03 -2.23898917e-01 -1.40197277e-01 -6.01514339e-01 -6.50944412e-01 -8.88651192e-01 1.90524086e-01 3.63272429e-01 2.99582839e-01 1.58392098e-02 2.02862322e-01 3.81102115e-01 -1.14233184e+00 8.29276979e-01 7.54497170e-01 8.40570152e-01 1.48421153e-01 8.83425653e-01 -3.76435250e-01 9.25015211e-01 -4.41580951e-01 -3.01061153e-01 2.20909752e-02 -3.73266697e-01 -1.24830377e+00 1.89359516e-01 3.17831546e-01 -2.71800552e-02 -9.72666368e-02 1.03272521e+00 3.32483977e-01 -3.53172451e-01 8.11828136e-01 -1.06938434e+00 -1.69361815e-01 5.76765180e-01 -5.90725303e-01 1.87709153e-01 -9.74873826e-02 1.59704953e-01 8.15342367e-01 -6.10951722e-01 9.01416361e-01 9.07108963e-01 9.43459094e-01 5.84847629e-01 -1.21137369e+00 -6.94001853e-01 -1.28165260e-01 3.19170356e-01 -1.39468348e+00 -4.21180815e-01 6.49299264e-01 -4.37296063e-01 5.78334868e-01 3.82101923e-01 3.18060905e-01 7.93278635e-01 2.17986718e-01 8.22695196e-01 1.55698001e+00 -5.82427859e-01 2.89818309e-02 4.10642296e-01 1.92934245e-01 6.34935021e-01 8.48868787e-02 -8.44008997e-02 -2.24624291e-01 -2.68951952e-01 5.97200811e-01 1.29525363e-01 -3.48390937e-01 -1.66463256e-01 -1.18592191e+00 4.89844054e-01 5.72063327e-01 4.19336230e-01 -2.77587384e-01 -1.85449690e-01 4.01731700e-01 2.92149216e-01 4.10110950e-01 4.63380307e-01 -4.03420895e-01 2.40324020e-01 -8.10453176e-01 -7.32266158e-02 6.52896225e-01 5.18784046e-01 8.07724357e-01 -7.76282012e-01 -2.84379721e-01 8.04165840e-01 9.50678438e-02 2.08538715e-02 7.51644075e-01 -3.80791187e-01 5.40850610e-02 9.18104053e-01 -2.13972300e-01 -7.22590804e-01 -3.50210875e-01 -4.51614618e-01 -7.52306879e-01 -1.78272501e-01 4.71079677e-01 1.39470801e-01 -1.08144736e+00 1.33323717e+00 3.80480289e-01 6.94102407e-01 3.88444990e-01 6.30632281e-01 1.09953690e+00 2.79211104e-01 2.12396026e-01 -7.24645704e-02 1.36817527e+00 -8.72980297e-01 -4.11585659e-01 -1.16459198e-01 8.73701751e-01 -6.37846410e-01 9.06029284e-01 1.21429019e-01 -6.76570714e-01 -2.88826644e-01 -7.99054325e-01 1.35374531e-01 -4.84918088e-01 1.58749580e-01 3.46996814e-01 2.67348975e-01 -8.98991048e-01 4.18206662e-01 -8.84992301e-01 -6.55282199e-01 6.31764472e-01 5.32494724e-01 -5.49069881e-01 -3.36356342e-01 -8.47618878e-01 6.52905345e-01 4.85321075e-01 -5.00892438e-02 -8.34064364e-01 -1.17463982e+00 -5.44461310e-01 -8.23644400e-02 2.68731922e-01 -2.98588961e-01 1.33960128e+00 -1.12575388e+00 -1.22973812e+00 1.20032513e+00 -5.93396053e-02 -3.64914566e-01 5.06592572e-01 4.01149482e-01 -2.63633400e-01 4.04515266e-01 -7.48780519e-02 7.26364136e-01 5.49150705e-01 -1.35238636e+00 -7.49761164e-01 -2.46765554e-01 4.38755676e-02 -4.66388203e-02 -5.70687890e-01 -8.97162259e-02 -4.51838881e-01 -3.92608136e-01 -1.38262108e-01 -1.02645576e+00 -5.34487009e-01 -1.19901024e-01 -6.53728902e-01 -1.50392935e-01 6.61246538e-01 -5.99870086e-01 9.27707255e-01 -2.20954227e+00 -2.59597927e-01 3.69077981e-01 3.12990636e-01 3.78334880e-01 -3.30390245e-01 1.65093079e-01 -1.16265789e-01 2.60382026e-01 -2.00027883e-01 -1.08228102e-01 -4.31095630e-01 2.86447793e-01 7.34914541e-02 4.46360797e-01 4.57299203e-01 8.58699679e-01 -1.04650676e+00 -8.92100453e-01 -4.24051704e-03 3.66203070e-01 -3.20083410e-01 3.43479335e-01 -7.79118761e-02 7.13127434e-01 -4.11098599e-01 6.23539805e-01 6.67510808e-01 -6.02033913e-01 4.68360871e-01 -4.77426589e-01 2.67621368e-01 1.86805744e-02 -7.87591159e-01 1.42621410e+00 -4.28163022e-01 4.00132447e-01 -1.68925360e-01 -9.93050098e-01 5.83587468e-01 3.95884573e-01 5.31522810e-01 -4.86201555e-01 1.31600469e-01 2.06967801e-01 2.96717174e-02 -6.45298243e-01 -2.00830381e-02 -2.98378468e-01 1.27491176e-01 1.85547143e-01 5.61252013e-02 -1.79585926e-02 2.77845562e-01 -1.41548440e-02 1.37715411e+00 -3.17120433e-01 5.41669428e-01 -1.79255277e-01 8.66770983e-01 3.79104048e-01 6.89710855e-01 6.43877923e-01 -3.36242527e-01 5.89295387e-01 4.77589220e-01 -5.78713298e-01 -7.01900601e-01 -8.73798251e-01 -3.78013313e-01 7.34721541e-01 6.10838160e-02 -2.59364963e-01 -6.84000194e-01 -1.19492662e+00 -5.41680641e-02 1.78685546e-01 -8.04390967e-01 -4.57169004e-02 -4.34166640e-01 -9.78069425e-01 6.37370169e-01 4.58285719e-01 4.29985791e-01 -7.62394905e-01 -2.57823676e-01 1.50285393e-01 -2.07568541e-01 -1.21402717e+00 -2.63276756e-01 1.23155192e-01 -1.09068859e+00 -1.28070247e+00 -5.64579844e-01 -9.70693946e-01 1.36420691e+00 2.47187451e-01 1.03291559e+00 4.77577299e-01 -7.79688537e-01 1.69680566e-01 -2.39471629e-01 -6.11756623e-01 -8.36298227e-01 1.80124700e-01 -3.81391287e-01 1.90390483e-01 4.34978843e-01 -8.00996348e-02 -6.18093193e-01 2.07281291e-01 -1.19743359e+00 3.06689024e-01 1.09799218e+00 1.06499660e+00 1.02691174e+00 1.58287641e-02 4.44629252e-01 -1.53575492e+00 3.34549427e-01 -4.99219805e-01 -2.16652796e-01 6.00782514e-01 -5.64161181e-01 9.47153121e-02 6.07170820e-01 -3.17188829e-01 -1.04907286e+00 3.34729135e-01 -9.30440500e-02 -2.15609640e-01 -3.38849783e-01 8.90891016e-01 2.29651704e-01 -3.86093527e-01 6.90393329e-01 2.13243857e-01 2.51459092e-01 -3.38609338e-01 -1.46802083e-01 8.62920761e-01 5.14333308e-01 -3.50281179e-01 6.77031875e-01 6.31299496e-01 2.81881187e-02 -5.93202412e-01 -1.05073500e+00 -7.21417844e-01 -6.65568650e-01 -4.10973839e-02 5.85209310e-01 -8.42724264e-01 -2.97588915e-01 5.41099310e-01 -7.62709975e-01 -3.91463727e-01 2.60355305e-02 2.41384715e-01 -2.30690122e-01 1.23053968e-01 -7.70726085e-01 -1.34571403e-01 -3.63072008e-01 -1.23739743e+00 1.10225749e+00 4.50825483e-01 -8.74036700e-02 -1.32742894e+00 3.70210946e-01 4.98319983e-01 4.24477160e-01 3.67752075e-01 1.22650564e+00 -8.92390370e-01 -3.93657535e-01 -3.95454675e-01 -4.36669081e-01 2.01765075e-01 5.64127743e-01 1.87101752e-01 -1.10831833e+00 -4.20046479e-01 -3.13900679e-01 -4.22195464e-01 7.69363225e-01 1.63747996e-01 1.42921031e+00 -2.01213643e-01 -8.20258260e-01 6.55175745e-01 1.79673457e+00 -3.42387296e-02 5.77028692e-01 3.59052181e-01 5.07988155e-01 5.61211944e-01 7.11539865e-01 9.52655599e-02 6.00867093e-01 3.51728529e-01 2.27025285e-01 -6.69230461e-01 -2.72599131e-01 -2.67695892e-03 -3.43703516e-02 7.31450260e-01 3.59223261e-02 4.84560579e-02 -1.17858720e+00 7.01714396e-01 -1.63744652e+00 -6.27245188e-01 1.10833004e-01 1.94603264e+00 1.26113331e+00 -1.85313374e-02 -2.34380841e-01 1.39255961e-02 6.18673921e-01 -3.11575383e-01 -5.00217438e-01 -2.63204258e-02 1.94164559e-01 2.75527179e-01 4.87860560e-01 2.88723469e-01 -1.14939821e+00 8.07621837e-01 6.47830582e+00 8.09470832e-01 -1.57560134e+00 9.96329933e-02 9.76548493e-01 6.09816201e-02 -3.24333236e-02 -1.75404206e-01 -6.01801753e-01 2.38312557e-01 9.83444691e-01 -2.71232963e-01 -2.17694297e-01 6.27005577e-01 7.20223337e-02 8.54618922e-02 -1.24102616e+00 8.16996574e-01 -2.77714096e-02 -1.66771817e+00 1.02056779e-01 8.47106874e-02 8.12664866e-01 -2.29135025e-02 -5.13047688e-02 2.49174789e-01 3.05313587e-01 -1.04590404e+00 -4.35607359e-02 3.80403370e-01 8.69310081e-01 -6.28719270e-01 1.20516384e+00 7.84341320e-02 -9.54039156e-01 1.13705575e-01 -3.07152182e-01 3.26045185e-01 -5.85057795e-01 6.32898033e-01 -1.68122756e+00 6.99068606e-01 4.97771591e-01 7.52233565e-01 -9.24637198e-01 1.13869667e+00 -2.33153086e-02 8.39568615e-01 2.38644611e-02 9.66198668e-02 1.11900829e-01 3.86730939e-01 -2.66495207e-03 1.43227875e+00 7.46674910e-02 5.51139712e-02 3.91396344e-01 3.55244488e-01 -8.88335109e-02 2.15218589e-01 -3.53951424e-01 -1.56130075e-01 3.00804138e-01 1.69177890e+00 -9.47237134e-01 -3.91679317e-01 -4.98229086e-01 5.61967611e-01 3.95804435e-01 2.70772517e-01 -6.90477490e-01 -3.92163515e-01 4.87338662e-01 8.08891505e-02 1.01103693e-01 3.19983155e-01 -1.36431217e-01 -9.90154326e-01 -3.68756145e-01 -9.83255267e-01 7.41525054e-01 -3.50019813e-01 -1.48422706e+00 7.03409076e-01 -1.57358080e-01 -1.35754180e+00 3.08706448e-03 -7.15863109e-01 -6.80559278e-01 6.07332706e-01 -2.04711986e+00 -1.38767731e+00 -5.87902188e-01 5.89893222e-01 3.65499318e-01 3.35185714e-02 8.86832833e-01 1.59786761e-01 -6.23100042e-01 8.67873371e-01 1.75900698e-01 3.52748573e-01 1.05810273e+00 -1.28067362e+00 -1.87568426e-01 5.55242121e-01 -1.09345801e-01 4.52543020e-01 6.20848596e-01 -3.72857898e-01 -1.38994658e+00 -1.39433801e+00 7.68713892e-01 -3.09821188e-01 6.56655729e-01 -4.04735394e-02 -9.98848498e-01 7.50738263e-01 1.91586375e-01 3.56674850e-01 1.35829985e+00 -6.74790740e-02 -2.15080991e-01 -3.16126227e-01 -1.45451760e+00 5.18136382e-01 6.20686710e-01 -3.78982246e-01 -2.40124151e-01 3.94906163e-01 3.18451315e-01 -6.09939754e-01 -1.22640145e+00 5.68373382e-01 3.82240117e-01 -5.58735609e-01 8.33144784e-01 -7.84601688e-01 4.96136904e-01 -3.46699417e-01 1.59426734e-01 -1.43619442e+00 -2.87197202e-01 -1.10191673e-01 2.97050506e-01 1.19169259e+00 5.44646502e-01 -6.44737422e-01 9.08144236e-01 3.91848713e-01 -5.11851534e-02 -9.89813387e-01 -6.38733506e-01 -5.20460427e-01 1.45262536e-02 -1.30857125e-01 6.68157935e-01 1.03773260e+00 1.40845388e-01 -9.79206935e-02 2.06766367e-01 3.69165212e-01 6.72558427e-01 1.70538619e-01 7.07140267e-01 -1.17260981e+00 -2.78954834e-01 -2.05681264e-01 -6.47386193e-01 -3.50626618e-01 1.83256939e-01 -1.25533867e+00 2.64973640e-01 -1.47779000e+00 7.70725310e-01 -7.16653049e-01 -9.46062505e-01 8.18276942e-01 -5.52753806e-01 5.24351776e-01 -1.53542355e-01 2.70487189e-01 -7.86019921e-01 -1.22855462e-01 1.30795813e+00 -3.98453772e-01 -5.19386195e-02 -1.33329242e-01 -9.61184502e-01 6.26596153e-01 7.66725659e-01 -6.39706910e-01 -4.02031839e-01 -2.27558985e-01 -1.91089943e-01 -1.55166477e-01 4.05343413e-01 -9.92868721e-01 3.08647066e-01 -3.47014368e-01 4.78677452e-01 -3.52292597e-01 -1.76610142e-01 -6.27893865e-01 1.82395786e-01 7.49393761e-01 -5.54561913e-01 -2.76196808e-01 2.70010561e-01 4.80170935e-01 -5.40181994e-01 -2.15959921e-01 9.43735361e-01 -2.76888847e-01 -8.52867365e-01 5.13164163e-01 -1.33465528e-01 -1.61026731e-01 1.12172365e+00 -1.80675656e-01 -6.59796417e-01 4.68229130e-02 -6.07644320e-01 3.71597975e-01 5.10689080e-01 1.44390240e-01 5.73881388e-01 -1.11241508e+00 -8.83885086e-01 9.56009403e-02 5.01620889e-01 1.03523254e-01 4.68290150e-01 1.19491911e+00 -6.14370644e-01 4.03801084e-01 -1.90364122e-01 -8.51163387e-01 -1.58625722e+00 3.44468921e-01 4.40754205e-01 -9.09220219e-01 -3.08141679e-01 8.63790631e-01 3.95300418e-01 -6.48152769e-01 6.91811228e-03 -4.38941479e-01 -1.76536053e-01 -2.48151392e-01 6.81339145e-01 -6.42600432e-02 3.10626149e-01 -3.55076879e-01 -5.39142907e-01 3.81086975e-01 -7.00953603e-01 4.04861838e-01 1.37235725e+00 1.67891517e-01 -1.88413367e-01 2.14502722e-01 1.34788299e+00 -1.77755013e-01 -1.16416204e+00 -4.23236430e-01 2.05138624e-01 -3.10994625e-01 6.47798181e-02 -8.71837199e-01 -1.11658752e+00 6.36808395e-01 7.81461418e-01 -5.72571643e-02 1.38090563e+00 1.57278866e-01 7.25703061e-01 2.36295417e-01 2.90528268e-01 -8.09360564e-01 2.11824879e-01 1.79940641e-01 4.80697542e-01 -1.48514688e+00 -1.17595695e-01 -6.30721092e-01 -4.57407176e-01 1.20801342e+00 7.70659387e-01 3.61977220e-02 5.66093445e-01 3.56618822e-01 5.08705318e-01 -5.25035523e-02 -8.55132699e-01 -1.28850460e-01 1.67290658e-01 5.91598690e-01 5.35370469e-01 8.94550905e-02 -8.01594779e-02 6.44422233e-01 1.18722975e-01 4.33331579e-01 5.65989614e-01 1.17762160e+00 -2.32051373e-01 -1.44910049e+00 -2.59801567e-01 7.23679662e-01 -6.60506427e-01 5.88806085e-02 -5.09617865e-01 8.84543598e-01 4.83839624e-02 6.50331616e-01 1.27997249e-02 -3.49800110e-01 1.54810920e-01 2.80651310e-03 5.74236095e-01 -7.80929744e-01 -6.61116064e-01 -1.25983477e-01 -4.67185006e-02 -2.43088081e-01 -5.64834774e-01 -5.92494249e-01 -1.44372654e+00 5.47535680e-02 -1.55421257e-01 5.98356165e-02 4.88408566e-01 1.06256628e+00 4.60616410e-01 5.76858282e-01 7.17359126e-01 -4.73129570e-01 -3.85004133e-01 -7.67807782e-01 -2.49989286e-01 6.66906476e-01 4.19542283e-01 -4.33541119e-01 -2.31085613e-01 3.79290491e-01]
[15.078362464904785, -2.763939380645752]
11500d73-5ea1-47db-b69d-00a23888c217
ckd-transbts-clinical-knowledge-driven-hybrid
2207.0737
null
https://arxiv.org/abs/2207.07370v1
https://arxiv.org/pdf/2207.07370v1.pdf
CKD-TransBTS: Clinical Knowledge-Driven Hybrid Transformer with Modality-Correlated Cross-Attention for Brain Tumor Segmentation
Brain tumor segmentation (BTS) in magnetic resonance image (MRI) is crucial for brain tumor diagnosis, cancer management and research purposes. With the great success of the ten-year BraTS challenges as well as the advances of CNN and Transformer algorithms, a lot of outstanding BTS models have been proposed to tackle the difficulties of BTS in different technical aspects. However, existing studies hardly consider how to fuse the multi-modality images in a reasonable manner. In this paper, we leverage the clinical knowledge of how radiologists diagnose brain tumors from multiple MRI modalities and propose a clinical knowledge-driven brain tumor segmentation model, called CKD-TransBTS. Instead of directly concatenating all the modalities, we re-organize the input modalities by separating them into two groups according to the imaging principle of MRI. A dual-branch hybrid encoder with the proposed modality-correlated cross-attention block (MCCA) is designed to extract the multi-modality image features. The proposed model inherits the strengths from both Transformer and CNN with the local feature representation ability for precise lesion boundaries and long-range feature extraction for 3D volumetric images. To bridge the gap between Transformer and CNN features, we propose a Trans&CNN Feature Calibration block (TCFC) in the decoder. We compare the proposed model with five CNN-based models and six transformer-based models on the BraTS 2021 challenge dataset. Extensive experiments demonstrate that the proposed model achieves state-of-the-art brain tumor segmentation performance compared with all the competitors.
['Chu Han', 'Zaiyi Liu', 'Guoqiang Han', 'Changhong Liang', 'Biao Huang', 'Zeyan Xu', 'Xipeng Pan', 'Bingjiang Qiu', 'Zhenwei Shi', 'Bingchao Zhao', 'Huan Lin', 'Hao Chen', 'Cheng Lu', 'Jiatai Lin', 'Jianwei Lin']
2022-07-15
null
null
null
null
['brain-tumor-segmentation', 'clinical-knowledge']
['medical', 'miscellaneous']
[ 2.35391811e-01 4.28370647e-02 -1.84315190e-01 -5.04075110e-01 -1.04378939e+00 3.31485197e-02 4.38166261e-01 -2.19409347e-01 -4.69145626e-01 5.59567094e-01 2.09428340e-01 -4.91217464e-01 -1.96359634e-01 -5.91322422e-01 -4.93633419e-01 -8.32382619e-01 2.72820085e-01 4.61235255e-01 4.51068878e-01 -1.62510037e-01 -2.22378924e-01 2.52951175e-01 -9.10413146e-01 3.50907564e-01 1.04692864e+00 1.52296126e+00 5.30182779e-01 3.05815309e-01 -3.78633142e-01 9.59103644e-01 -2.23729581e-01 -2.93226480e-01 1.64496973e-01 -3.22048932e-01 -1.04127443e+00 6.97836280e-02 1.56757876e-01 -3.06147575e-01 -6.33306205e-01 1.12823057e+00 6.84749722e-01 -4.05102819e-01 5.03114760e-01 -1.09035122e+00 -6.17358983e-01 6.87292993e-01 -7.99283981e-01 5.29475212e-01 -2.55885839e-01 2.20912784e-01 5.06110966e-01 -6.46924675e-01 3.73468280e-01 6.53460026e-01 7.50624299e-01 6.29500926e-01 -7.39737332e-01 -7.20484197e-01 6.84891418e-02 4.67348248e-01 -1.27631819e+00 -1.86465770e-01 5.57033062e-01 -4.33265477e-01 7.71499336e-01 1.98476240e-01 1.02221775e+00 1.01409411e+00 4.03253257e-01 1.12389195e+00 1.25751734e+00 -1.72345471e-02 -2.01751143e-01 -2.02023223e-01 1.57967612e-01 9.28514779e-01 -2.24790368e-02 5.70352131e-04 -3.18547994e-01 2.95028925e-01 8.18069637e-01 2.69531906e-01 -5.66845059e-01 -3.55378121e-01 -1.65130901e+00 6.95303738e-01 1.00433004e+00 6.75367355e-01 -3.66673082e-01 2.64224529e-01 6.24983847e-01 -2.88728923e-02 4.37501341e-01 -1.30417854e-01 -3.69407296e-01 1.38400301e-01 -1.08332038e+00 -1.90492272e-01 2.21759900e-01 9.97082174e-01 1.66586310e-01 -1.43889114e-02 -4.79596645e-01 8.91534925e-01 1.09792694e-01 4.52549368e-01 1.10367060e+00 -5.43336309e-02 4.68450397e-01 5.97521067e-01 -5.24251163e-01 -3.84854913e-01 -7.49057651e-01 -8.24880779e-01 -1.24781895e+00 -3.89330477e-01 2.63180077e-01 9.99631956e-02 -1.60539842e+00 1.48593616e+00 2.98594832e-01 4.24127936e-01 -8.73259604e-02 9.67433035e-01 1.27355862e+00 8.25916082e-02 1.20644346e-01 -4.22809310e-02 1.60610151e+00 -1.29668522e+00 -7.24714100e-01 -1.01026036e-01 7.43197918e-01 -5.23878813e-01 6.27802372e-01 1.48763638e-02 -1.00016057e+00 -3.14878076e-01 -8.90547693e-01 -2.13506579e-01 -3.50236475e-01 1.00656308e-01 7.70020008e-01 6.18735194e-01 -1.14015508e+00 3.77729572e-02 -1.30750811e+00 -1.71058968e-01 9.31008279e-01 6.21322453e-01 -3.38669300e-01 -1.07360654e-01 -1.19439852e+00 1.14267910e+00 3.13092858e-01 4.06091660e-01 -1.08524859e+00 -1.00147498e+00 -7.18884528e-01 -1.40176669e-01 4.02031183e-01 -9.89479840e-01 1.43956542e+00 -8.55461419e-01 -1.36259377e+00 7.79168010e-01 -1.07731722e-01 -5.36350071e-01 6.53752387e-01 1.43294588e-01 -4.35145587e-01 2.52564579e-01 2.29881987e-01 9.35463846e-01 6.37536287e-01 -8.46798539e-01 -8.52161288e-01 -5.15651107e-01 -2.52574533e-01 2.75305182e-01 -2.90629864e-01 -2.48296887e-01 -7.46976256e-01 -6.39456570e-01 1.79743394e-01 -8.18034470e-01 -3.96667928e-01 -1.86205357e-01 -6.38827085e-01 9.91054848e-02 8.98488820e-01 -7.30964780e-01 8.27987492e-01 -1.85177791e+00 3.93167853e-01 2.14193895e-01 6.54776096e-01 1.05923414e-01 5.79559691e-02 -4.04559553e-01 -3.20829481e-01 -4.58524339e-02 -3.95737529e-01 -2.51584768e-01 -2.46319965e-01 1.55578658e-01 1.21271208e-01 4.85092670e-01 9.10828728e-03 1.47853100e+00 -7.48107731e-01 -6.71262324e-01 3.34966272e-01 4.98208284e-01 -3.30329835e-01 5.69813363e-02 2.06872359e-01 9.05460954e-01 -6.40791416e-01 9.48340595e-01 7.14210808e-01 -4.53544706e-01 -1.92911953e-01 -6.75994992e-01 1.14775732e-01 -4.95610833e-02 -3.70110363e-01 2.03413677e+00 -4.90598470e-01 2.92246640e-01 -1.21698240e-02 -1.33575237e+00 4.39102173e-01 4.76646334e-01 9.75248516e-01 -1.04660594e+00 5.09166896e-01 5.37046850e-01 2.76292294e-01 -6.64051950e-01 -6.12589680e-02 -4.40384924e-01 9.62139443e-02 -4.86156940e-02 3.97273242e-01 -1.87524423e-01 -9.21558067e-02 4.55652922e-02 1.09450245e+00 -2.28131652e-01 1.78739905e-01 -2.24385217e-01 6.92762673e-01 -3.44890654e-02 5.21622002e-01 4.19583589e-01 -5.67830503e-01 5.98991752e-01 2.92823434e-01 -3.07306498e-01 -6.31921530e-01 -1.00914073e+00 -3.55733037e-01 4.78912652e-01 1.32471040e-01 -8.11378360e-02 -7.86776006e-01 -9.33812797e-01 -1.11291960e-01 2.08349526e-01 -9.49303091e-01 -2.25888297e-01 -5.47876418e-01 -9.68966544e-01 6.20245039e-01 8.35749865e-01 8.83951962e-01 -7.59407043e-01 -6.84138358e-01 9.55961645e-02 -5.83762944e-01 -1.30578136e+00 -6.61566377e-01 5.25359929e-01 -9.68991458e-01 -1.09913373e+00 -1.21973681e+00 -8.95588934e-01 6.32152438e-01 3.91390651e-01 7.23557413e-01 6.75498173e-02 -3.96596789e-01 3.49006563e-01 -4.27019626e-01 -3.49754691e-01 -7.07431808e-02 4.50597107e-01 -3.73606950e-01 7.17472732e-02 1.86575517e-01 -4.62303877e-01 -6.96613312e-01 2.04580918e-01 -9.52890277e-01 6.40318811e-01 1.03103840e+00 1.23179293e+00 6.73049808e-01 -5.35671152e-02 5.12400150e-01 -7.24645495e-01 3.08368534e-01 -6.68275714e-01 -2.09418967e-01 4.58998531e-01 -3.69685382e-01 -9.33044851e-02 4.35040861e-01 -1.84849635e-01 -8.01821828e-01 1.04354180e-01 -3.33684146e-01 -6.24798954e-01 -3.22756097e-02 7.63945401e-01 2.03558840e-02 -4.33451086e-01 1.09845638e-01 6.08640492e-01 6.10797144e-02 -2.08015412e-01 2.03223601e-01 5.83303273e-01 6.79094732e-01 -3.94781202e-01 5.59912384e-01 4.67791617e-01 -2.61382461e-02 -4.18139964e-01 -9.34911430e-01 -5.59860826e-01 -6.93288326e-01 -3.51900250e-01 1.15512621e+00 -8.98075104e-01 -6.29450917e-01 7.45463967e-01 -8.39234829e-01 -1.34346768e-01 -1.58670023e-01 4.94633734e-01 -5.77459157e-01 2.14199096e-01 -6.68979824e-01 -1.91710651e-01 -5.86508095e-01 -2.08950448e+00 1.20190740e+00 3.14382315e-01 5.08688092e-01 -9.20293808e-01 -4.24821764e-01 6.76845372e-01 8.57067227e-01 2.65701830e-01 9.00017083e-01 -6.24342859e-01 -5.50630927e-01 -7.74917472e-03 -5.39252698e-01 2.25026280e-01 3.10086578e-01 -7.46346951e-01 -9.06855822e-01 -2.88931102e-01 1.27530292e-01 -4.52942759e-01 1.14553249e+00 7.07440615e-01 1.46429026e+00 3.05795461e-01 -5.63081324e-01 1.01534605e+00 1.39429379e+00 3.25824946e-01 4.66660202e-01 4.07511562e-01 1.10268223e+00 1.53068647e-01 7.22330213e-02 3.69175635e-02 6.61452830e-01 5.83306372e-01 6.73377216e-01 -3.75777483e-01 -3.97108078e-01 1.27200678e-01 -7.94980954e-03 1.32188642e+00 5.61918393e-02 1.11836158e-01 -1.20118749e+00 6.44113660e-01 -1.61152148e+00 -5.69286525e-01 -5.45612648e-02 1.70062542e+00 8.04311574e-01 5.69315962e-02 -1.53955311e-01 -3.45145948e-02 6.59436703e-01 -3.08787189e-02 -7.90675879e-01 1.31926224e-01 -1.13488615e-01 3.14334929e-01 8.62661242e-01 8.19598138e-02 -1.28534353e+00 7.32287288e-01 5.76948595e+00 1.22093225e+00 -1.55432630e+00 7.02815473e-01 8.35515678e-01 -5.07887378e-02 -1.17987707e-01 -3.71813834e-01 -4.82488126e-01 4.58278060e-01 7.96250582e-01 -3.41430306e-04 1.96209788e-01 4.76035178e-01 -1.05269730e-01 -2.20080856e-02 -9.71891105e-01 1.12015975e+00 7.38815516e-02 -1.40021348e+00 2.36530211e-02 1.03256345e-01 5.54950714e-01 5.17734647e-01 2.98092574e-01 3.59840006e-01 -4.73986752e-02 -1.20932400e+00 9.71842110e-01 5.86752415e-01 1.04364431e+00 -5.86464345e-01 9.63455200e-01 1.58319831e-01 -1.36791575e+00 -3.66263650e-02 5.98155893e-02 7.10953951e-01 1.43045321e-01 5.81211686e-01 -8.96211743e-01 1.13008463e+00 7.62847781e-01 9.96753514e-01 -7.78114915e-01 1.24999547e+00 1.61661506e-01 6.36035025e-01 -2.95018286e-01 4.13099796e-01 5.56784689e-01 2.28098538e-02 2.22040534e-01 1.10115790e+00 4.06834632e-01 8.19549784e-02 1.75722852e-01 8.35864544e-01 -5.00145778e-02 2.23168842e-02 -1.42790347e-01 1.36375323e-01 6.04261719e-02 1.44658732e+00 -9.60290790e-01 -4.34255838e-01 -7.01953888e-01 8.19874287e-01 1.96954176e-01 9.58087221e-02 -1.09766293e+00 -1.37142763e-02 1.64467081e-01 -1.07215665e-01 4.32637811e-01 8.50814357e-02 -4.68477041e-01 -1.33399940e+00 -1.21390909e-01 -8.25646460e-01 3.56475025e-01 -7.56763697e-01 -1.29055023e+00 9.79115248e-01 -2.45487262e-02 -1.26366234e+00 2.48957485e-01 -5.44685483e-01 -4.80411440e-01 7.60088265e-01 -2.00810719e+00 -1.79827726e+00 -5.41374385e-01 9.39415574e-01 4.95876670e-01 -7.22641349e-02 5.51468909e-01 5.85765541e-01 -7.46859789e-01 5.61929941e-01 -2.00447924e-02 2.70501673e-01 5.96657693e-01 -1.20310974e+00 2.25691553e-02 6.22047901e-01 -4.49699461e-01 2.33478144e-01 8.17042962e-02 -5.21966457e-01 -1.25335574e+00 -1.31001949e+00 3.43511939e-01 -6.77155852e-02 7.25869656e-01 -8.63542221e-03 -7.87203431e-01 7.30184555e-01 3.55598956e-01 6.18955433e-01 6.10921383e-01 -4.08544570e-01 -2.27740094e-01 -2.00352296e-01 -1.13680625e+00 3.41933012e-01 1.01006675e+00 -4.49128300e-01 -5.50666511e-01 4.04818773e-01 7.04334736e-01 -8.76504302e-01 -1.07173073e+00 7.91110694e-01 3.52104276e-01 -7.35900879e-01 1.02074456e+00 -3.49003077e-01 5.04503965e-01 -1.39700964e-01 -9.19349790e-02 -1.37811601e+00 -3.26246172e-01 6.72439933e-02 2.28855312e-01 7.69551516e-01 1.90868884e-01 -6.22602165e-01 6.28644168e-01 3.14045697e-01 -7.59980083e-01 -1.26011312e+00 -1.31095350e+00 -5.29595435e-01 3.41961384e-01 -4.34052110e-01 6.98206663e-01 9.82177556e-01 -7.66669288e-02 9.31927040e-02 -1.80810288e-01 9.45153460e-02 5.09727478e-01 4.75751422e-02 1.69755384e-01 -8.61308634e-01 -6.15304522e-02 -9.05374229e-01 -5.18663704e-01 -1.02120936e+00 -1.25048682e-02 -1.41830862e+00 -7.05857649e-02 -1.67158699e+00 6.86691880e-01 -4.67874825e-01 -8.06554973e-01 7.65267849e-01 -8.74282420e-02 3.91143650e-01 9.53271985e-02 1.42182708e-01 -6.17140174e-01 7.48044372e-01 1.87221086e+00 -5.90827942e-01 3.15126181e-01 -3.71545881e-01 -6.53762937e-01 6.01378679e-01 5.14753342e-01 -2.66352057e-01 -3.50789666e-01 -7.05753028e-01 -4.29986358e-01 4.49035794e-01 5.43224216e-01 -1.17348528e+00 5.88299811e-01 -3.39028873e-02 4.84904855e-01 -7.61419475e-01 2.68385440e-01 -9.58813846e-01 -8.07173476e-02 5.92243731e-01 -1.61010370e-01 2.41827616e-03 2.17745125e-01 3.92833471e-01 -5.16981065e-01 1.84678555e-01 9.69864666e-01 -1.78293452e-01 -6.11952960e-01 8.79158795e-01 -2.59545892e-01 -5.22083901e-02 1.13503122e+00 -2.52100021e-01 -3.11129183e-01 7.56231770e-02 -8.04418087e-01 4.53605503e-01 -1.46474868e-01 4.66455162e-01 7.86221683e-01 -1.38295269e+00 -5.91196060e-01 1.82474315e-01 2.17577487e-01 2.16194823e-01 6.51116848e-01 1.87035608e+00 -3.84681821e-01 7.44085491e-01 -4.14125741e-01 -8.49885821e-01 -8.05562854e-01 3.18974286e-01 8.57930660e-01 -5.52080333e-01 -8.13387394e-01 9.07651007e-01 4.63997394e-01 -3.63352984e-01 1.41598657e-01 -8.39535177e-01 -2.65227914e-01 -2.56976664e-01 3.47021043e-01 -2.36201391e-01 4.81813729e-01 -9.15063679e-01 -5.28065026e-01 5.75003684e-01 -4.45115536e-01 1.24219783e-01 1.40097106e+00 -6.72588497e-02 -9.08490494e-02 1.33041635e-01 1.34020400e+00 -5.71782291e-01 -8.45836818e-01 -4.98227060e-01 -1.66256607e-01 -2.13132575e-01 6.48903370e-01 -1.14909267e+00 -2.01224256e+00 1.00210500e+00 8.62929881e-01 -1.57438189e-01 1.42517269e+00 -3.72948684e-02 1.43263793e+00 -9.21927020e-02 3.89718205e-01 -7.11367846e-01 -5.32762073e-02 4.52227652e-01 7.72620678e-01 -1.39458942e+00 -4.37647462e-01 -4.97558028e-01 -7.01723695e-01 1.17345154e+00 7.18072474e-01 1.76124722e-01 8.90616953e-01 3.71099263e-01 8.45189020e-03 -4.54060942e-01 -5.35433233e-01 -3.30088049e-01 3.64024729e-01 4.89247829e-01 3.57901067e-01 2.10787550e-01 -2.04719022e-01 1.01688802e+00 -6.74855709e-02 3.74917835e-01 2.20316127e-01 8.48752856e-01 -2.22686335e-01 -8.84888887e-01 -1.88244283e-01 8.71874154e-01 -4.97098535e-01 -2.16604993e-01 9.71101522e-02 7.39199996e-01 4.60755289e-01 5.36854088e-01 -2.16675401e-01 -5.94335139e-01 2.88247466e-01 -4.67877500e-02 5.86821973e-01 -3.19404840e-01 -7.24511325e-01 2.82840103e-01 -3.16889465e-01 -5.13735533e-01 -6.21112347e-01 -5.98045528e-01 -1.34435952e+00 -3.87446992e-02 -5.11511385e-01 -1.97265819e-02 6.15239382e-01 1.27159905e+00 1.59992889e-01 1.22728229e+00 4.01185602e-01 -7.86415875e-01 -4.00497556e-01 -9.20413196e-01 -4.75797087e-01 1.48758158e-01 3.74538064e-01 -9.64808822e-01 1.21547848e-01 -1.86498418e-01]
[14.571869850158691, -2.3734941482543945]
0f2c5361-6768-4c85-8bc9-275dfee8de54
ranpac-random-projections-and-pre-trained
2307.02251
null
https://arxiv.org/abs/2307.02251v1
https://arxiv.org/pdf/2307.02251v1.pdf
RanPAC: Random Projections and Pre-trained Models for Continual Learning
Continual learning (CL) aims to incrementally learn different tasks (such as classification) in a non-stationary data stream without forgetting old ones. Most CL works focus on tackling catastrophic forgetting under a learning-from-scratch paradigm. However, with the increasing prominence of foundation models, pre-trained models equipped with informative representations have become available for various downstream requirements. Several CL methods based on pre-trained models have been explored, either utilizing pre-extracted features directly (which makes bridging distribution gaps challenging) or incorporating adaptors (which may be subject to forgetting). In this paper, we propose a concise and effective approach for CL with pre-trained models. Given that forgetting occurs during parameter updating, we contemplate an alternative approach that exploits training-free random projectors and class-prototype accumulation, which thus bypasses the issue. Specifically, we inject a frozen Random Projection layer with nonlinear activation between the pre-trained model's feature representations and output head, which captures interactions between features with expanded dimensionality, providing enhanced linear separability for class-prototype-based CL. We also demonstrate the importance of decorrelating the class-prototypes to reduce the distribution disparity when using pre-trained representations. These techniques prove to be effective and circumvent the problem of forgetting for both class- and domain-incremental continual learning. Compared to previous methods applied to pre-trained ViT-B/16 models, we reduce final error rates by between 10\% and 62\% on seven class-incremental benchmark datasets, despite not using any rehearsal memory. We conclude that the full potential of pre-trained models for simple, effective, and fast continual learning has not hitherto been fully tapped.
['Anton Van Den Hengel', 'Ehsan Abbasnejad', 'Amin Parveneh', 'Dong Gong', 'Mark D. McDonnell']
2023-07-05
null
null
null
null
['continual-learning']
['methodology']
[ 2.45597139e-01 1.03455380e-01 3.71716321e-02 -7.15849325e-02 -5.05842566e-01 -3.49939555e-01 6.78271890e-01 2.18024582e-01 -6.12948239e-01 9.74954247e-01 -7.56063759e-02 -2.69422412e-01 -1.45259380e-01 -6.15130007e-01 -8.91962767e-01 -8.27983499e-01 -2.93190181e-02 4.17996496e-01 5.17968178e-01 -2.58211672e-01 1.76368773e-01 4.55363661e-01 -2.14751601e+00 3.99580866e-01 8.29695106e-01 7.73531854e-01 4.42215472e-01 5.53569734e-01 -2.45908126e-01 6.36392772e-01 -6.06621027e-01 -2.00177968e-01 5.66026010e-02 -4.41071868e-01 -5.27295411e-01 8.83062929e-03 2.37907603e-01 -9.63937789e-02 -3.04673761e-01 5.11174917e-01 5.70495903e-01 1.25040591e-01 6.66995406e-01 -9.76575077e-01 -7.22629726e-01 5.98853648e-01 -3.98375899e-01 3.62644464e-01 1.51806340e-01 3.44023824e-01 6.17101133e-01 -1.44781613e+00 4.45619375e-01 8.36812973e-01 9.89078164e-01 8.04237187e-01 -1.40379369e+00 -7.14375079e-01 2.08287299e-01 3.89659882e-01 -1.41233146e+00 -6.02702856e-01 7.16006815e-01 -2.78478414e-01 1.23548269e+00 3.60757075e-02 9.50821280e-01 1.29338574e+00 3.92003924e-01 8.88139665e-01 1.01544118e+00 -7.49896109e-01 4.01907414e-01 4.76953596e-01 3.05478454e-01 5.27700484e-01 4.24455553e-01 1.80787131e-01 -9.67113912e-01 6.15137182e-02 3.36492032e-01 4.16834444e-01 -3.67619008e-01 -6.12268865e-01 -8.37967455e-01 6.23747766e-01 1.08870380e-01 4.14407283e-01 -3.76872182e-01 -1.73744597e-02 3.32605273e-01 5.32960951e-01 4.85425681e-01 4.67516452e-01 -5.98929524e-01 -1.22876309e-01 -1.28669858e+00 1.37347530e-03 5.61311007e-01 8.40394557e-01 1.00356090e+00 3.23336512e-01 -2.10485682e-01 6.27567291e-01 -1.24168567e-01 3.39303523e-01 9.60171282e-01 -4.44561779e-01 2.03384176e-01 6.45865858e-01 -1.36471257e-01 -5.98649979e-01 -3.65560710e-01 -7.65819550e-01 -8.25218022e-01 3.20741773e-01 2.18704894e-01 9.17551816e-02 -1.21676970e+00 1.88236117e+00 1.75636247e-01 3.90201926e-01 2.19974533e-01 4.09557730e-01 3.23355794e-01 5.28701127e-01 1.23124272e-01 -4.39442664e-01 8.28411102e-01 -8.62174988e-01 -5.57503998e-01 -2.19619051e-01 5.32539308e-01 -4.83975589e-01 1.50430536e+00 6.73795760e-01 -1.05670321e+00 -7.04013765e-01 -1.27044451e+00 -5.62544959e-03 -5.38032651e-01 -1.05899505e-01 5.07223368e-01 6.69805288e-01 -1.19165814e+00 9.30185497e-01 -1.10059083e+00 -2.63974071e-01 6.31005824e-01 5.10110080e-01 -3.34404290e-01 -2.00893268e-01 -1.08302009e+00 9.96382654e-01 4.42336172e-01 -5.98984510e-02 -1.09731281e+00 -9.81445670e-01 -6.08220041e-01 2.32316718e-01 2.06325442e-01 -8.95989835e-01 9.27861810e-01 -9.89569068e-01 -1.69309163e+00 3.87210757e-01 -2.09267631e-01 -7.39344656e-01 4.60418344e-01 -6.00953937e-01 -2.38067582e-01 -9.27198827e-02 -2.63087451e-01 6.50925994e-01 1.35877645e+00 -1.08968115e+00 -4.04427022e-01 -1.88339412e-01 -1.89957336e-01 2.37348557e-01 -1.00886643e+00 -5.88663697e-01 -1.58138692e-01 -6.10665083e-01 6.62262738e-02 -8.18571806e-01 -2.11996902e-02 -3.69938239e-02 1.48092359e-01 -5.96414171e-02 1.07193482e+00 -3.39575648e-01 1.31847227e+00 -2.23063779e+00 2.77165055e-01 -3.09263945e-01 9.04518068e-02 6.56716943e-01 -1.95704937e-01 4.22476858e-01 -1.61097199e-01 -1.71311513e-01 -5.45781314e-01 -9.46415484e-01 -4.19495672e-01 4.60473925e-01 -8.37095618e-01 2.73631096e-01 4.56667215e-01 9.05678153e-01 -8.52453947e-01 -8.50218832e-02 9.31218565e-02 7.65879631e-01 -7.43508279e-01 1.41770259e-01 -2.19110593e-01 3.00012678e-01 3.07183623e-01 4.13432807e-01 7.15562463e-01 -1.41443461e-01 -6.54362002e-03 1.92948163e-01 -8.06483999e-02 2.73150086e-01 -1.04228163e+00 2.02921891e+00 -6.20412529e-01 5.00940740e-01 -3.67740899e-01 -9.86622453e-01 8.58061254e-01 1.46712467e-01 1.98646951e-02 -6.88798666e-01 -3.49765509e-01 2.92493552e-01 -3.08593094e-01 -1.52558669e-01 6.70254588e-01 -5.11237085e-01 4.44672555e-02 5.05548239e-01 5.89254320e-01 -9.43409931e-03 -8.93316884e-03 1.11689299e-01 1.27179468e+00 4.12341714e-01 1.38999611e-01 -1.86357111e-01 3.30151767e-01 -1.50604472e-01 5.68402529e-01 8.51031899e-01 1.05200909e-01 9.06472266e-01 1.76010244e-02 -5.54999948e-01 -8.30418408e-01 -1.22406530e+00 -5.24611138e-02 1.17562449e+00 -2.28045862e-02 -4.69837219e-01 -2.60747731e-01 -7.44078696e-01 6.66337013e-02 9.95640993e-01 -6.61728621e-01 -7.83530474e-01 -7.11169839e-01 -8.64089727e-01 3.71821672e-01 3.91613215e-01 2.91711301e-01 -1.00608861e+00 -8.53618681e-01 4.43539530e-01 1.28993139e-01 -5.30892432e-01 -1.51583463e-01 8.53208601e-01 -1.29804814e+00 -7.54658759e-01 -8.06481957e-01 -4.44799453e-01 7.53689826e-01 4.61169749e-01 9.60693896e-01 4.75385785e-02 -4.16274756e-01 5.94377279e-01 -2.61398345e-01 -2.98744202e-01 -2.36841500e-01 5.40446222e-01 4.78391528e-01 4.25051711e-02 2.45254710e-01 -9.94492769e-01 -5.26232898e-01 1.87824089e-02 -1.08821285e+00 6.21017329e-02 7.88507283e-01 1.31543446e+00 4.20334458e-01 -2.94322539e-02 9.12239075e-01 -9.15751278e-01 3.86645615e-01 -6.22872412e-01 -1.24555953e-01 1.73171088e-01 -1.05377150e+00 3.53787541e-01 9.09312665e-01 -9.85021949e-01 -1.19384933e+00 2.32199263e-02 -7.06131980e-02 -9.04217660e-01 1.22738145e-01 2.76552588e-01 -1.52899111e-02 1.73859969e-01 9.45430160e-01 7.25151718e-01 2.05588415e-02 -5.79327166e-01 3.94186884e-01 3.36567789e-01 4.34434533e-01 -5.18792510e-01 7.35355437e-01 4.68026131e-01 -2.45400041e-01 -6.60924554e-01 -8.84348273e-01 -3.08262229e-01 -7.19956338e-01 7.59093761e-02 1.74225196e-01 -9.97014940e-01 -1.93621859e-01 6.22535765e-01 -1.11974382e+00 -5.37754476e-01 -1.09722018e+00 3.59526932e-01 -5.11552870e-01 2.77103782e-01 -5.07139444e-01 -7.29429245e-01 -3.39278489e-01 -7.07427919e-01 7.05060899e-01 1.56631470e-01 -1.70246363e-01 -8.87030184e-01 2.27890432e-01 -2.33389989e-01 9.37418520e-01 -2.45912358e-01 8.73202920e-01 -5.39813161e-01 -5.73297083e-01 -7.63731375e-02 1.84253782e-01 5.42114258e-01 1.43942088e-01 -4.48174953e-01 -1.21560907e+00 -7.30093360e-01 2.01467514e-01 -4.24451381e-01 1.42019010e+00 -1.45878121e-01 9.80886459e-01 -3.34677517e-01 -4.21432674e-01 6.28843069e-01 1.34862542e+00 2.48120707e-02 6.87579930e-01 2.17304453e-01 3.71725917e-01 2.87332624e-01 3.42707992e-01 5.93165338e-01 2.74740040e-01 3.81520182e-01 2.23243579e-01 1.25215828e-01 -6.47604883e-01 -4.80511487e-01 5.83476007e-01 1.04822779e+00 7.17317462e-02 -8.19491595e-02 -7.41730750e-01 7.20714748e-01 -1.65895474e+00 -8.72880518e-01 3.40031743e-01 2.40713072e+00 1.14791477e+00 4.19665217e-01 -2.83450246e-01 4.64476466e-01 4.71126437e-01 -1.47909030e-01 -8.26838911e-01 -2.12278187e-01 -2.62896091e-01 5.75663745e-01 1.02504559e-01 3.67917895e-01 -7.30942667e-01 1.01550579e+00 5.99500322e+00 8.11324477e-01 -1.34655499e+00 4.03190225e-01 4.48271304e-01 -3.37740600e-01 -3.53353292e-01 1.62062630e-01 -1.29284739e+00 4.91523921e-01 1.26872563e+00 -5.97337335e-02 3.66202444e-01 8.85155618e-01 -4.03746128e-01 -2.13442519e-01 -1.16224790e+00 7.70210624e-01 2.39701271e-01 -1.36488152e+00 2.74393260e-01 -1.98062181e-01 6.81201220e-01 -1.20438196e-01 4.39367563e-01 9.47682321e-01 1.96453314e-02 -7.18919039e-01 8.55522394e-01 7.35256195e-01 1.04665530e+00 -6.36065602e-01 5.02123117e-01 5.78926861e-01 -8.38357747e-01 -4.11697298e-01 -5.93713403e-01 -2.50223458e-01 -8.19805660e-04 9.03822005e-01 -1.02989280e+00 4.76407826e-01 7.45696425e-01 7.09460616e-01 -8.17300200e-01 1.05333817e+00 -2.12596744e-01 9.04378355e-01 -4.30603802e-01 2.44155392e-01 -3.52318257e-01 2.87587315e-01 4.91604984e-01 1.20346081e+00 7.20591605e-01 -2.93630004e-01 -3.46020877e-01 7.22161710e-01 1.02442741e-01 -1.96947873e-01 -7.00325429e-01 2.58033156e-01 3.77002805e-01 9.04037654e-01 -6.34877443e-01 -3.78044724e-01 -1.72121301e-01 1.28513229e+00 6.98204100e-01 2.84614861e-01 -6.67188644e-01 -4.04812545e-01 3.69631499e-01 5.53245068e-01 6.56837046e-01 -3.73273462e-01 -3.39173734e-01 -1.25832629e+00 -4.33302894e-02 -5.92554152e-01 3.24188143e-01 -5.32426238e-01 -1.08674359e+00 7.17417240e-01 -6.17490374e-02 -1.19612134e+00 -3.02783936e-01 -2.12105229e-01 -5.15792906e-01 5.47508299e-01 -1.83728170e+00 -1.13343000e+00 -1.96639046e-01 7.60202348e-01 7.18983114e-01 -2.67811477e-01 1.02053058e+00 3.64089102e-01 -5.01202822e-01 8.42301130e-01 5.69922552e-02 -7.82271802e-01 8.37846041e-01 -9.86254454e-01 5.13165444e-02 8.46874058e-01 1.26911163e-01 9.61661577e-01 7.00046301e-01 -5.69325984e-01 -1.49726105e+00 -1.12412381e+00 8.89539301e-01 -5.39938688e-01 3.56115311e-01 -7.27960169e-01 -1.38451433e+00 6.71469331e-01 3.98021638e-02 2.00115532e-01 6.44589007e-01 1.55871287e-01 -5.24815738e-01 -3.95283192e-01 -9.30294812e-01 4.65530276e-01 1.24870861e+00 -3.74474049e-01 -8.95314157e-01 -8.01748931e-02 8.78839672e-01 -1.72851712e-01 -2.13735402e-01 3.19074810e-01 3.82256210e-01 -1.01051247e+00 9.22978818e-01 -4.18509662e-01 6.24299496e-02 -2.59748042e-01 7.25011677e-02 -1.28452539e+00 -2.07661241e-01 -7.32099175e-01 -5.93902409e-01 1.23146105e+00 3.07317555e-01 -7.34534979e-01 7.58026123e-01 1.99773401e-01 -4.32741195e-01 -7.25244343e-01 -1.16648448e+00 -9.06416953e-01 2.12870121e-01 -5.04267275e-01 4.03962255e-01 7.33472764e-01 -4.66070212e-02 5.43827295e-01 -5.24018824e-01 -1.18917249e-01 2.67055064e-01 3.44416648e-02 6.48013353e-01 -1.28733099e+00 -5.25498807e-01 -1.32361874e-01 -1.39220059e-01 -1.06918740e+00 4.99527231e-02 -1.04696000e+00 -3.23239453e-02 -1.05797029e+00 8.29655454e-02 -7.65147150e-01 -6.98275208e-01 9.29201543e-01 -1.42462447e-01 2.34075010e-01 1.64591253e-01 4.73674595e-01 -6.90136373e-01 1.06375778e+00 8.94001722e-01 -1.20942704e-01 -5.46461463e-01 2.32085586e-02 -6.80812657e-01 3.38037431e-01 6.61780596e-01 -7.85613835e-01 -6.43605173e-01 -2.80815005e-01 3.30664039e-01 -3.06078196e-01 2.40030870e-01 -1.49745834e+00 5.95763922e-01 2.61036873e-01 4.83487159e-01 -5.03501475e-01 5.20332992e-01 -6.27273560e-01 2.54583240e-01 6.40495539e-01 -1.52193010e-01 -4.82672453e-03 5.52961707e-01 8.87948811e-01 -8.48301128e-02 -2.28308201e-01 7.17368066e-01 -1.66900024e-01 -5.89938819e-01 3.45028155e-02 -4.87620831e-01 -2.05778345e-01 1.01688397e+00 -2.96704590e-01 -2.50052154e-01 -5.61691672e-02 -9.66639280e-01 -1.76630214e-01 4.24081117e-01 3.83170784e-01 9.62896168e-01 -1.05707514e+00 -3.95707220e-01 7.12087691e-01 -3.00370231e-02 6.05522357e-02 4.20883179e-01 8.25784743e-01 4.34670299e-02 3.25269699e-01 -3.33759785e-01 -6.13355160e-01 -8.01441550e-01 8.50600243e-01 3.27522531e-02 -4.46104258e-01 -7.60020494e-01 8.80539834e-01 8.96249041e-02 -3.74118239e-01 2.45579153e-01 -1.61636293e-01 -3.53492461e-02 4.00475711e-01 5.36919236e-01 2.95224577e-01 4.73443121e-01 -6.15756474e-02 -2.50682443e-01 3.89887452e-01 -5.22145450e-01 -7.48995366e-03 1.60925639e+00 -2.28290409e-01 2.52989739e-01 8.30214739e-01 9.05553639e-01 -1.31014571e-01 -1.61049986e+00 -3.13487500e-01 -5.47987670e-02 -1.43779114e-01 -1.65004015e-01 -7.11955488e-01 -7.69093454e-01 1.33920491e+00 7.88340569e-01 -2.21539751e-01 1.12329662e+00 -3.23338240e-01 7.22873509e-01 4.30375546e-01 7.04673588e-01 -9.96863365e-01 6.48678899e-01 6.29005373e-01 8.61051798e-01 -8.55930686e-01 2.13029943e-02 1.34951904e-01 -3.89647990e-01 1.15838492e+00 5.63158572e-01 1.77764408e-02 6.86646402e-01 2.50266433e-01 -3.05388629e-01 2.12588087e-01 -1.20380211e+00 -1.08543046e-01 -1.29316896e-01 6.03252828e-01 4.41620089e-02 -3.69166255e-01 -1.44070953e-01 8.04172695e-01 3.91619578e-02 2.91667044e-01 5.70985377e-01 1.22468364e+00 -5.20687103e-01 -1.16886270e+00 -1.08051769e-01 2.84839243e-01 -1.79699615e-01 -2.78756171e-01 3.17821205e-01 6.31009638e-01 2.74508387e-01 5.19268453e-01 3.97295952e-02 -5.57094514e-01 2.16594607e-01 6.32539034e-01 4.33427632e-01 -8.20864499e-01 -5.07850230e-01 -2.67639846e-01 -4.25693065e-01 -1.93497434e-01 -4.90363948e-02 -7.79251635e-01 -1.02885234e+00 -2.39177719e-02 -4.66236383e-01 4.46537696e-03 4.59656984e-01 7.01029718e-01 8.80197585e-01 4.84483629e-01 4.56642896e-01 -1.05288208e+00 -6.95317805e-01 -9.37021792e-01 -3.94389242e-01 1.11909300e-01 4.78229612e-01 -9.11881506e-01 -6.96889460e-01 7.66213462e-02]
[9.78907299041748, 3.4468982219696045]
4718c9c1-a0f0-462d-ad42-48582b8144b6
faster-riemannian-newton-type-optimization-by
2302.11076
null
https://arxiv.org/abs/2302.11076v1
https://arxiv.org/pdf/2302.11076v1.pdf
Faster Riemannian Newton-type Optimization by Subsampling and Cubic Regularization
This work is on constrained large-scale non-convex optimization where the constraint set implies a manifold structure. Solving such problems is important in a multitude of fundamental machine learning tasks. Recent advances on Riemannian optimization have enabled the convenient recovery of solutions by adapting unconstrained optimization algorithms over manifolds. However, it remains challenging to scale up and meanwhile maintain stable convergence rates and handle saddle points. We propose a new second-order Riemannian optimization algorithm, aiming at improving convergence rate and reducing computational cost. It enhances the Riemannian trust-region algorithm that explores curvature information to escape saddle points through a mixture of subsampling and cubic regularization techniques. We conduct rigorous analysis to study the convergence behavior of the proposed algorithm. We also perform extensive experiments to evaluate it based on two general machine learning tasks using multiple datasets. The proposed algorithm exhibits improved computational speed and convergence behavior compared to a large set of state-of-the-art Riemannian optimization algorithms.
['Tingting Mu', 'Yian Deng']
2023-02-22
null
null
null
null
['type']
['speech']
[-2.71209568e-01 -1.36296809e-01 1.55943200e-01 -2.96830922e-01 -7.93829739e-01 -4.68764931e-01 2.38595635e-01 -2.35493913e-01 -5.03546655e-01 5.89044988e-01 -6.61314800e-02 -1.41829014e-01 -4.79010403e-01 -2.89894283e-01 -4.12659347e-01 -9.19862151e-01 -2.87351817e-01 6.87867254e-02 1.46307945e-02 -2.20790282e-01 5.02646387e-01 5.40135443e-01 -9.87677395e-01 -4.66080904e-01 1.23438764e+00 6.75439656e-01 9.35079306e-02 4.69390959e-01 3.48882288e-01 2.45660111e-01 6.26130030e-02 -5.63130677e-01 4.42979097e-01 -2.65893400e-01 -7.46080816e-01 5.29355407e-01 2.39518285e-01 2.54358929e-02 -1.69225499e-01 1.24499440e+00 4.22143728e-01 5.99955022e-01 6.62964523e-01 -9.52643573e-01 -8.25531244e-01 -9.62883383e-02 -7.36169577e-01 2.49178126e-01 -7.37818629e-02 -2.47220322e-01 1.00912309e+00 -1.45866990e+00 4.67525661e-01 1.08252406e+00 6.68870330e-01 3.30487847e-01 -1.01414466e+00 4.33579944e-02 9.62441489e-02 -2.01844587e-03 -1.63145518e+00 -2.75662869e-01 8.83840799e-01 -3.04512620e-01 5.20317674e-01 3.36024284e-01 2.31301591e-01 3.63109976e-01 2.02539250e-01 6.20817065e-01 8.26675892e-01 -1.11762382e-01 1.91380590e-01 2.22288132e-01 6.42002895e-02 1.06281316e+00 1.97408989e-01 -2.64168173e-01 -1.60491511e-01 -2.05363497e-01 8.79968643e-01 2.60649920e-02 -4.19644684e-01 -6.02239847e-01 -1.28824496e+00 1.03464127e+00 6.19921148e-01 2.08407894e-01 -2.60627657e-01 -2.74487704e-01 -7.52041936e-02 1.63145512e-01 6.21497452e-01 3.11519444e-01 -2.68645704e-01 -8.14083517e-02 -5.74040771e-01 5.57183288e-02 7.47916818e-01 1.01560974e+00 8.16767991e-01 1.49223074e-01 1.72178462e-01 8.86594176e-01 5.05265415e-01 3.62833261e-01 3.68561208e-01 -9.72018719e-01 7.12349951e-01 6.78265750e-01 2.19121546e-01 -1.43958759e+00 -6.93256736e-01 -5.07495224e-01 -1.07051170e+00 1.09843463e-01 6.23915315e-01 -2.79390305e-01 -1.46548841e-02 1.43109512e+00 8.05620432e-01 4.81759273e-02 -6.49828315e-02 1.42182124e+00 1.63500294e-01 4.90466326e-01 -5.08393288e-01 -3.08050692e-01 9.41510081e-01 -1.07164550e+00 -6.71481490e-01 1.72913834e-01 1.00260627e+00 -9.12593484e-01 1.34243023e+00 2.41610065e-01 -1.07555676e+00 -2.52473354e-01 -1.16751277e+00 -1.77715555e-01 -8.68281946e-02 5.57589829e-01 7.16760278e-01 6.68014288e-01 -9.55326498e-01 1.00209939e+00 -9.50268030e-01 -2.05860332e-01 3.98990475e-02 4.24571335e-01 -4.15477991e-01 2.17042059e-01 -5.75412393e-01 7.04875350e-01 -1.64229363e-01 5.90567172e-01 -3.41943741e-01 -5.61725795e-01 -9.51140344e-01 -3.19343865e-01 3.61062169e-01 -4.47107345e-01 8.17275167e-01 -6.41123056e-01 -1.89694011e+00 6.68761492e-01 -3.15045744e-01 -3.55475917e-02 7.19747722e-01 -5.03201187e-01 -1.49812639e-01 1.58188850e-01 -3.22452225e-02 -4.85339127e-02 1.00597608e+00 -6.86448991e-01 -1.70569897e-01 -6.24003112e-01 -1.59985144e-02 5.55050611e-01 -5.99411130e-01 1.19655328e-02 -5.48074782e-01 -6.53081119e-01 3.96583974e-01 -1.27601779e+00 -4.70507652e-01 1.07355595e-01 -2.89566755e-01 -8.43942836e-02 7.14735687e-01 -4.40838367e-01 1.20786786e+00 -1.95965266e+00 8.28175902e-01 8.05256143e-02 2.89008379e-01 3.64905223e-02 -1.08284922e-02 2.44929031e-01 1.21401645e-01 1.11326054e-01 -6.00152552e-01 -6.83347106e-01 -9.78923813e-02 -2.06799716e-01 -1.79750882e-02 1.26338422e+00 1.94708392e-01 6.87272787e-01 -7.96002567e-01 -3.74638528e-01 2.14959830e-01 5.59226394e-01 -8.15945745e-01 2.00247407e-01 4.38889205e-01 6.66600823e-01 -7.11232245e-01 4.06502426e-01 7.50040948e-01 -3.93131644e-01 -3.57938707e-02 -1.98904887e-01 -3.18303347e-01 -7.85416737e-02 -1.49586391e+00 1.92513573e+00 -3.92228514e-01 3.12191248e-01 6.17775142e-01 -1.16190076e+00 8.54793549e-01 2.12289337e-02 6.08527303e-01 6.41048476e-02 3.37859541e-01 3.18486601e-01 -1.81639239e-01 -4.99782145e-01 6.27584815e-01 -2.15605795e-02 1.98653683e-01 4.25498635e-01 -1.81898609e-01 -1.86000541e-01 1.65013790e-01 1.07639283e-01 6.68538451e-01 8.11162665e-02 7.13580400e-02 -8.96836162e-01 1.06536281e+00 -3.00319970e-01 5.30867457e-01 1.18110798e-01 -2.53047496e-01 7.20211625e-01 1.44238651e-01 -2.47363880e-01 -8.12540233e-01 -9.75643218e-01 -4.74734396e-01 7.72308826e-01 3.62908810e-01 -3.71495694e-01 -9.65482295e-01 -6.65001392e-01 -1.72335193e-01 2.54680425e-01 -4.94731694e-01 -3.02443415e-01 -5.99835336e-01 -1.31285024e+00 2.74759326e-02 9.89433154e-02 5.30538082e-01 -3.76317143e-01 -1.04312018e-01 -1.60263795e-02 -6.48169294e-02 -1.14341807e+00 -1.18650484e+00 -5.53906441e-01 -1.32289457e+00 -1.29828548e+00 -7.47129858e-01 -8.90007257e-01 1.13690734e+00 6.11668050e-01 6.64725006e-01 1.74092203e-01 -2.25576729e-01 5.22984684e-01 -1.46946743e-01 2.74838477e-01 -1.20101348e-02 2.57508844e-01 5.06780028e-01 6.28055573e-01 2.34143622e-02 -3.88652265e-01 -6.79727793e-01 9.16421652e-01 -8.41596007e-01 -3.67331177e-01 4.33377802e-01 6.87858760e-01 7.02832758e-01 -6.32744059e-02 4.38703567e-01 -5.70905805e-01 8.92751276e-01 -5.47565818e-01 -9.42900300e-01 2.08950043e-01 -6.88185930e-01 4.39158231e-01 7.67546952e-01 -4.06739116e-01 -9.93185103e-01 2.59587448e-02 2.05874249e-01 -4.04299110e-01 5.50370097e-01 5.11641145e-01 7.44662061e-02 -6.19900882e-01 4.49142486e-01 4.74274866e-02 2.00396478e-01 -7.13580012e-01 4.78906870e-01 5.64393103e-01 2.92341620e-01 -5.83135605e-01 1.08089817e+00 7.78490722e-01 3.12304527e-01 -1.09229875e+00 -8.70062947e-01 -5.85357845e-01 -1.15578556e+00 -2.14130446e-01 7.92127013e-01 -5.78572094e-01 -7.37001240e-01 3.90026271e-01 -8.83110881e-01 -2.69282945e-02 1.31209150e-01 8.47640336e-01 -6.22117519e-01 7.87087619e-01 -6.28282905e-01 -7.65258789e-01 -2.80136973e-01 -1.19987476e+00 9.93896842e-01 2.13726878e-01 2.71026015e-01 -1.57705092e+00 2.66501337e-01 2.45758697e-01 3.28035146e-01 9.48918611e-02 3.45093936e-01 -1.43392354e-01 -5.89730322e-01 -2.65430480e-01 -7.07026795e-02 2.31605649e-01 4.20373410e-01 5.50895650e-03 -3.32684755e-01 -5.53166747e-01 6.40588760e-01 -3.96577688e-03 4.27227437e-01 1.96842402e-01 9.61432397e-01 -2.87384272e-01 5.70201464e-02 1.00623250e+00 1.26684451e+00 -4.12215114e-01 3.26761067e-01 1.97040409e-01 8.52279484e-01 6.55773580e-01 7.93976665e-01 4.08089936e-01 4.97122973e-01 6.02721155e-01 3.55872601e-01 2.28338428e-02 4.02360350e-01 1.21207954e-02 3.42305154e-01 1.47914720e+00 -3.37085783e-01 5.40268958e-01 -5.31180024e-01 4.60374504e-01 -2.13411236e+00 -5.17915905e-01 -3.79492223e-01 2.65596414e+00 7.08355129e-01 -2.46014014e-01 1.05545871e-01 5.10068052e-02 8.45377147e-01 -5.02282083e-02 -6.02454722e-01 -1.66281804e-01 -7.25362599e-02 -2.56677449e-01 4.14182484e-01 8.74016404e-01 -1.27060032e+00 9.50178325e-01 6.39500237e+00 6.12482190e-01 -1.00562680e+00 1.21421814e-01 3.38783979e-01 -2.36193277e-02 -7.45715126e-02 -1.47409454e-01 -7.66321123e-01 1.98978707e-02 5.93848348e-01 -2.43915409e-01 8.05876434e-01 9.01156604e-01 4.44042534e-01 1.57129034e-01 -8.40425551e-01 1.15936434e+00 1.29445806e-01 -1.15512049e+00 -2.95596480e-01 1.65779442e-01 9.74144340e-01 3.52242254e-02 1.96168646e-01 -2.21917443e-02 -1.87187478e-01 -8.10083330e-01 3.57253522e-01 5.35443008e-01 4.14666891e-01 -7.21541047e-01 4.85183924e-01 2.71570653e-01 -1.36594093e+00 3.94793116e-02 -5.61849654e-01 -8.16466883e-02 3.18807393e-01 5.55526912e-01 -2.56066620e-01 5.62291622e-01 5.62767446e-01 9.96151388e-01 -6.58511519e-01 1.06228030e+00 -1.29243836e-01 2.89674938e-01 -4.41909224e-01 -1.97981745e-01 4.33505923e-01 -1.46061504e+00 9.34526682e-01 8.21995378e-01 2.39299580e-01 7.04226047e-02 -6.54622987e-02 8.97264898e-01 2.08556019e-02 6.88988507e-01 -5.92534363e-01 1.06355533e-01 3.99727933e-02 1.71368015e+00 -7.07691014e-01 2.29161888e-01 -6.62961304e-01 1.11724687e+00 7.12994754e-01 4.81137246e-01 -8.72580469e-01 -4.90129501e-01 7.90205598e-01 -4.94430363e-02 2.41488338e-01 -1.06482971e+00 -2.48186544e-01 -1.66900337e+00 4.04248267e-01 -3.54518801e-01 3.54641706e-01 -2.14901596e-01 -1.22196889e+00 4.84073460e-01 -2.15437427e-01 -1.44837856e+00 3.20165157e-02 -7.51954257e-01 -7.33988881e-01 5.56849480e-01 -1.35053730e+00 -7.56966114e-01 -1.22858815e-01 8.80215049e-01 4.65344638e-01 -1.09075606e-01 4.12589699e-01 4.61659402e-01 -9.28280711e-01 5.43533325e-01 5.24194777e-01 -9.91744250e-02 4.34725553e-01 -1.37806487e+00 -2.29906682e-02 1.03611803e+00 3.33195850e-02 9.20643330e-01 4.41857725e-01 -3.00904661e-01 -1.93074298e+00 -1.13321888e+00 4.60357159e-01 -3.53578061e-01 1.01722038e+00 -2.51361609e-01 -9.37082231e-01 5.48904300e-01 -1.96545437e-01 7.20743164e-02 6.09590769e-01 6.66554049e-02 6.37600943e-02 -1.94272473e-02 -1.15468049e+00 8.28893304e-01 1.04654622e+00 -5.10596454e-01 -2.44562909e-01 5.18769503e-01 3.99749845e-01 -2.74724275e-01 -1.16739178e+00 3.34657222e-01 7.15133846e-02 -7.11373210e-01 9.19483542e-01 -7.14709461e-01 -4.09344994e-02 -4.82395351e-01 -2.05387861e-01 -1.27793002e+00 -1.49380907e-01 -1.40228820e+00 -3.35668951e-01 1.01034296e+00 3.90277445e-01 -8.67301464e-01 6.39305413e-01 9.25787866e-01 -2.19567418e-01 -9.40240443e-01 -1.01770532e+00 -8.91620994e-01 3.39358896e-01 -1.94435075e-01 2.12458633e-02 8.14629078e-01 2.70719916e-01 3.87374967e-01 -4.95763451e-01 3.42062235e-01 1.01633954e+00 2.32620820e-01 7.30112672e-01 -9.37964737e-01 -2.09021438e-02 -3.50451976e-01 -3.41690242e-01 -1.28589857e+00 1.55455366e-01 -1.03733242e+00 -2.47358859e-01 -1.03660703e+00 -7.35943615e-02 -3.24851006e-01 3.36674042e-03 -2.04548031e-01 -4.42983449e-01 1.39848068e-01 -9.08623915e-03 4.50089753e-01 -6.51223898e-01 1.25130010e+00 1.49590349e+00 1.76400349e-01 -4.70412433e-01 2.96575665e-01 -5.29385924e-01 8.79559815e-01 8.56721640e-01 -2.78309733e-01 -4.34622347e-01 -5.03122509e-01 1.84351429e-01 -2.52503902e-01 2.21119523e-02 -7.88012147e-01 1.49155810e-01 -1.12360373e-01 -2.08233371e-01 -1.51879802e-01 3.32393408e-01 -6.07354939e-01 -2.65955091e-01 2.49892145e-01 -4.65504713e-02 2.74420619e-01 -1.19900472e-01 6.93435729e-01 -7.31925741e-02 -3.55500013e-01 1.12273586e+00 2.21761897e-01 -2.54567802e-01 6.29132450e-01 -1.67566180e-01 2.18327150e-01 1.04509389e+00 1.75632268e-01 1.67114869e-01 -3.95130932e-01 -7.46603608e-01 2.42959708e-01 5.16073465e-01 4.94518191e-01 7.69131541e-01 -1.53862751e+00 -5.58406293e-01 5.29520139e-02 -2.05842912e-01 -7.45989233e-02 -4.61152680e-02 1.44999564e+00 -6.83401048e-01 3.02634209e-01 2.77641743e-01 -7.05052912e-01 -1.07028961e+00 6.00982070e-01 5.62531769e-01 -1.33087635e-01 -5.89335263e-01 5.63513756e-01 6.28629848e-02 -7.76914597e-01 6.59960657e-02 -2.81519443e-01 -1.51409313e-01 -3.08362335e-01 4.76547420e-01 8.45454812e-01 -1.88653450e-02 -8.76087248e-01 -3.19819570e-01 1.13690698e+00 2.38077492e-01 -1.14699543e-01 1.20656002e+00 -6.12278402e-01 -3.01436394e-01 4.28689510e-01 1.72645557e+00 1.60821646e-01 -1.38034809e+00 -2.74171948e-01 1.23912551e-01 -5.68646669e-01 1.88791424e-01 1.81505665e-01 -1.24671388e+00 8.03977370e-01 4.33965117e-01 6.75423294e-02 7.50052631e-01 -2.80847847e-01 6.19971991e-01 6.95208788e-01 3.96285176e-01 -1.27300215e+00 2.65464395e-01 4.20585960e-01 1.11424327e+00 -1.57437766e+00 2.06947818e-01 -6.54241979e-01 -6.22251272e-01 1.19337082e+00 4.80104983e-01 -5.13032973e-01 1.12001538e+00 -3.31850827e-01 -4.02309932e-02 -1.88526198e-01 -1.73307419e-01 -7.53307641e-02 7.27525949e-01 1.07275590e-01 4.04145777e-01 -1.31235242e-01 -7.14066148e-01 4.61322486e-01 -8.15685019e-02 -4.71770674e-01 5.14829636e-01 6.54272139e-01 -1.99255109e-01 -8.93461585e-01 -3.12121391e-01 5.04288003e-02 -5.47734141e-01 -2.28086244e-02 -2.30929360e-01 6.67634189e-01 -8.19664419e-01 1.19811726e+00 -2.86915749e-01 -2.42646679e-01 7.41853118e-02 -3.51115525e-01 4.50151056e-01 -3.22839528e-01 -2.61226743e-01 7.87454993e-02 -2.05859676e-01 -7.07218826e-01 -5.01346111e-01 -9.57506955e-01 -1.37425351e+00 -1.98814228e-01 -4.58268791e-01 4.33886826e-01 8.39801311e-01 9.79782939e-01 6.10576570e-01 -4.24698479e-02 1.18032837e+00 -1.00822222e+00 -1.02885664e+00 -8.73703659e-01 -8.18261504e-01 5.70943594e-01 2.10878700e-01 -8.02571833e-01 -7.88357794e-01 -9.45749059e-02]
[7.53929328918457, 4.153052806854248]
f77596c8-f015-4fdd-9560-9fd856d28e6d
multi-fidelity-black-box-optimization-with
null
null
https://icml.cc/Conferences/2018/Schedule?showEvent=2264
http://proceedings.mlr.press/v80/sen18a/sen18a.pdf
Multi-Fidelity Black-Box Optimization with Hierarchical Partitions
Motivated by settings such as hyper-parameter tuning and physical simulations, we consider the problem of black-box optimization of a function. Multi-fidelity techniques have become popular for applications where exact function evaluations are expensive, but coarse (biased) approximations are available at much lower cost. A canonical example is that of hyper-parameter selection in a learning algorithm. The learning algorithm can be trained for fewer iterations – this results in a lower cost, but its validation error is only coarsely indicative of the same if the algorithm had been trained till completion. We incorporate the multi-fidelity setup into the powerful framework of black-box optimization through hierarchical partitioning. We develop tree-search based multi-fidelity algorithms with theoretical guarantees on simple regret. We finally demonstrate the performance gains of our algorithms on both real and synthetic datasets.
['Kirthevasan Kandasamy', 'Sanjay Shakkottai', 'Rajat Sen']
2018-07-01
null
null
null
icml-2018-7
['physical-simulations']
['miscellaneous']
[ 6.44573793e-02 1.18606180e-01 -4.49688673e-01 -2.75038630e-01 -1.41923273e+00 -6.26758397e-01 2.96851099e-01 2.84171522e-01 -4.99811172e-01 1.25027037e+00 -1.54421311e-02 -1.57602951e-01 -4.82108176e-01 -5.91171563e-01 -9.56871986e-01 -8.53361905e-01 -2.25649834e-01 6.27213418e-01 -2.69676447e-01 1.63597554e-01 1.76266402e-01 2.93736458e-01 -1.20696497e+00 -8.05998519e-02 8.28855336e-01 1.00468075e+00 -2.86384791e-01 9.64038432e-01 5.69815040e-01 3.00663918e-01 -2.50486255e-01 -3.97801131e-01 4.30876195e-01 -4.08225775e-01 -9.23151314e-01 1.38519153e-01 2.08836094e-01 -1.75223768e-01 -3.12901914e-01 8.43656600e-01 5.00854611e-01 4.66598332e-01 6.30474091e-01 -9.15963590e-01 1.27736776e-05 6.51860237e-01 -5.01037121e-01 -2.14988947e-01 5.66906184e-02 4.86100733e-01 1.33698094e+00 -7.79556394e-01 3.09046626e-01 1.22144651e+00 7.47312367e-01 1.74519196e-01 -1.72013807e+00 -3.67470354e-01 1.36909574e-01 -2.93309301e-01 -1.30491102e+00 -4.49875712e-01 3.11404467e-01 -3.34138513e-01 7.58976460e-01 2.94501215e-01 6.63126469e-01 6.90446615e-01 2.18765587e-01 6.15421474e-01 1.02162230e+00 -3.05674672e-01 5.31510174e-01 1.74963728e-01 -2.49781702e-02 7.78481960e-01 4.09543544e-01 4.56337810e-01 -5.78946531e-01 -6.28988564e-01 6.93201005e-01 -1.77989662e-01 -4.74524856e-01 -6.22842491e-01 -1.11813557e+00 1.03444016e+00 4.06532884e-01 -2.03398466e-01 -2.36357197e-01 7.08600283e-01 3.53773445e-01 3.42182010e-01 3.55255187e-01 8.70749235e-01 -5.99589050e-01 -3.24608415e-01 -1.14967752e+00 7.01376021e-01 9.69066203e-01 7.10708261e-01 7.82538652e-01 -1.69986904e-01 -2.86395639e-01 6.34961069e-01 2.48007372e-01 3.56013417e-01 6.24492168e-02 -1.36713696e+00 2.16635704e-01 9.17778239e-02 6.28819227e-01 -3.09411675e-01 -5.66082120e-01 -7.59712279e-01 -6.79105580e-01 5.34979701e-01 6.94108248e-01 -4.82417762e-01 -7.91778922e-01 1.73069417e+00 5.99897563e-01 5.18970676e-02 -1.48463309e-01 1.09729874e+00 -8.03668946e-02 7.62302160e-01 -2.31562883e-01 -4.37510759e-01 8.37920606e-01 -1.05315256e+00 -3.68191302e-01 -1.73198894e-01 8.79403889e-01 -5.42645395e-01 1.15892529e+00 6.84868157e-01 -1.50098026e+00 -7.42444694e-02 -1.11436915e+00 4.22250703e-02 7.22494423e-02 -1.79914162e-01 9.13907111e-01 7.26462364e-01 -7.66375482e-01 1.28825593e+00 -7.92847455e-01 6.88455105e-02 5.19292235e-01 4.39956576e-01 -8.24221224e-03 6.25213683e-02 -8.37135911e-01 7.71495402e-01 3.28865528e-01 -1.49427876e-01 -1.25088394e+00 -1.24134219e+00 -3.87208015e-01 2.19153568e-01 6.43699825e-01 -1.01848292e+00 1.51166987e+00 -7.72165000e-01 -1.64750135e+00 4.94500428e-01 9.06734914e-02 -3.91911119e-01 8.97670925e-01 -2.42990792e-01 2.00128943e-01 -1.38652444e-01 -2.71400392e-01 2.11778447e-01 9.47625279e-01 -1.05091405e+00 -4.16618764e-01 -2.24669471e-01 2.25697756e-01 3.09028357e-01 -3.82782854e-02 -2.64663696e-01 -2.19206870e-01 -5.45199513e-01 -1.05293833e-01 -1.13376617e+00 -4.68476504e-01 1.91194907e-01 -3.75770628e-01 2.87118495e-01 4.16202992e-01 -2.93046862e-01 1.37390292e+00 -1.71933484e+00 2.83327341e-01 4.35528934e-01 2.78880835e-01 -1.07505374e-01 2.17940390e-01 3.83775443e-01 1.09673694e-01 4.01410073e-01 -4.89731401e-01 -1.45399362e-01 9.26827267e-02 -1.90762449e-02 -7.49889538e-02 7.47565627e-01 -6.43588603e-02 8.68192494e-01 -8.19710732e-01 -2.80192286e-01 -4.97311540e-02 2.78767437e-01 -1.03984606e+00 5.01157790e-02 -3.25373381e-01 3.87131453e-01 -5.67761779e-01 4.41907555e-01 3.37150306e-01 -6.66182339e-01 1.67506158e-01 -1.87601224e-01 7.92231262e-02 2.57657349e-01 -1.50676942e+00 1.71994340e+00 -7.90017545e-01 3.08710963e-01 3.47375959e-01 -7.95074940e-01 1.61821917e-01 1.68995261e-01 5.41346014e-01 -2.04994723e-01 9.79004800e-02 1.57454759e-01 3.00833932e-03 -1.00246713e-01 3.82267356e-01 -2.96066314e-01 -1.83940172e-01 7.89395750e-01 -2.34839823e-02 -3.00102741e-01 1.29193410e-01 1.38798371e-01 1.28197348e+00 3.59642208e-01 3.38387847e-01 -5.73877156e-01 1.27188146e-01 9.62883681e-02 4.55092818e-01 1.03980160e+00 -6.89570531e-02 6.23819351e-01 3.99742872e-01 -2.24110708e-01 -1.50546587e+00 -9.53212917e-01 -5.15950978e-01 1.04246223e+00 4.25600894e-02 -5.29767275e-01 -8.41405213e-01 -5.86249352e-01 4.64015812e-01 4.86432225e-01 -5.92635572e-01 -1.96296439e-01 -4.63784993e-01 -1.08051550e+00 2.73322493e-01 2.25259379e-01 1.03169262e-01 -2.97628433e-01 -5.10606289e-01 3.57641399e-01 3.88675570e-01 -7.50227511e-01 -4.88015801e-01 3.71233821e-01 -1.01406860e+00 -9.14799809e-01 -6.16722703e-01 -1.76155925e-01 5.56382418e-01 -2.22996771e-01 1.26413774e+00 3.04200441e-01 -3.49352986e-01 2.52086848e-01 1.04015879e-01 3.70521694e-02 -2.55826890e-01 1.68341517e-01 2.58908123e-01 -8.30997601e-02 -5.54464340e-01 -5.51260471e-01 -8.25664282e-01 3.56500477e-01 -5.15997291e-01 -4.06881534e-02 4.97855157e-01 1.28586447e+00 8.02496791e-01 3.77809592e-02 4.61161137e-01 -1.14454758e+00 5.27673900e-01 -3.89498621e-01 -1.02136660e+00 3.10257316e-01 -9.23670650e-01 5.51852763e-01 8.10785949e-01 -5.39283693e-01 -9.84394073e-01 -2.70727668e-02 1.20201647e-01 -4.80718642e-01 4.31756794e-01 3.03878754e-01 -6.69955760e-02 -6.05130315e-01 7.84975886e-01 -2.93432832e-01 -2.32162997e-01 -5.04972160e-01 4.40077245e-01 2.65142471e-01 3.45408380e-01 -1.15841508e+00 8.81157994e-01 2.33075008e-01 3.34814012e-01 -4.97878283e-01 -1.10949731e+00 -1.00113593e-01 -1.62227243e-01 -1.72629327e-01 3.87677699e-01 -9.28866327e-01 -1.07320738e+00 -4.95623909e-02 -6.40715897e-01 -6.32850766e-01 -6.78108633e-01 4.97102559e-01 -9.63220716e-01 5.13085686e-02 -3.11126292e-01 -8.71788085e-01 -1.76702589e-01 -1.16894603e+00 1.05475926e+00 2.56729841e-01 -1.54862136e-01 -1.13034999e+00 1.85406148e-01 1.87170699e-01 3.54497701e-01 2.65636325e-01 1.00425446e+00 -2.54844904e-01 -7.73204386e-01 -1.22912310e-01 3.84112149e-02 1.12905294e-01 -2.65688270e-01 7.70998886e-03 -1.10833931e+00 -6.68161750e-01 -9.62865427e-02 -7.21443713e-01 8.47406089e-01 6.14490509e-01 1.51314604e+00 -5.29733598e-01 -4.65557396e-01 1.05844796e+00 1.65969777e+00 -4.19129252e-01 1.27512336e-01 1.44716322e-01 4.88327205e-01 1.43344417e-01 8.44302297e-01 8.05639327e-01 -7.29513615e-02 6.72043800e-01 1.51872933e-01 5.86531125e-02 1.04225382e-01 -2.85566121e-01 1.55089676e-01 3.79299462e-01 1.38779715e-01 -2.82520354e-01 -8.73863995e-01 2.97848433e-01 -2.03122640e+00 -7.09207177e-01 1.86852053e-01 2.55184102e+00 1.05668890e+00 3.69009405e-01 2.86002189e-01 1.18210003e-01 5.26860476e-01 4.83350493e-02 -1.01722622e+00 -4.15635049e-01 -3.17802243e-02 2.96181530e-01 8.61574888e-01 8.31756055e-01 -9.03680742e-01 4.62851435e-01 7.89780760e+00 1.12806129e+00 -7.49478579e-01 2.33604565e-01 1.06092393e+00 -8.28943193e-01 -5.26620150e-01 4.31552440e-01 -7.71258652e-01 4.36595649e-01 1.10394931e+00 -5.02718389e-01 9.23670650e-01 8.08064103e-01 2.60951787e-01 -4.12791252e-01 -1.41176343e+00 9.58128393e-01 -6.10469222e-01 -1.61438251e+00 -4.85619396e-01 1.55026272e-01 1.14297843e+00 4.20261174e-02 8.45762938e-02 2.32337624e-01 5.12332261e-01 -1.28762794e+00 5.42645574e-01 4.66660023e-01 9.77258861e-01 -8.59545946e-01 3.15605938e-01 3.14206451e-01 -8.58293593e-01 -2.10182145e-01 -4.30330902e-01 3.78804002e-03 2.02055737e-01 9.45357084e-01 -4.93481219e-01 2.77573794e-01 4.70955700e-01 1.90771177e-01 -9.69787687e-02 1.32911658e+00 1.60073116e-01 7.66146243e-01 -7.47700691e-01 -9.25015658e-02 1.27941102e-01 -3.76763105e-01 6.33409381e-01 9.93960500e-01 1.51421770e-01 1.28176823e-01 2.09239513e-01 9.55409765e-01 -3.79965931e-01 -9.35246572e-02 -3.04071903e-01 -1.63554773e-01 6.18977368e-01 1.35988629e+00 -3.23515683e-01 -2.65684634e-01 -2.11258307e-01 6.05457902e-01 4.68364507e-01 4.19599861e-01 -9.96222198e-01 -1.66065738e-01 8.60905588e-01 2.12809950e-01 2.44567007e-01 -2.38940999e-01 -6.01998806e-01 -9.81889009e-01 -4.31997143e-02 -8.15409839e-01 3.71645212e-01 -4.68860924e-01 -1.07331598e+00 -2.09315032e-01 -1.23992711e-01 -8.42415094e-01 -2.84912825e-01 -4.43945944e-01 -3.89316052e-01 7.91391730e-01 -1.36888433e+00 -5.48105776e-01 7.28159696e-02 1.23846054e-01 2.59920657e-01 1.31816015e-01 6.53598964e-01 2.89318413e-01 -6.94792628e-01 8.60476434e-01 7.17703581e-01 -5.00895381e-01 5.42836487e-01 -1.37167728e+00 3.52246240e-02 5.31326771e-01 -8.47421214e-02 4.16033655e-01 8.73663902e-01 -5.23613513e-01 -1.81494772e+00 -9.19418991e-01 2.57618010e-01 -3.30685526e-01 7.22231448e-01 -3.29352230e-01 -7.72184432e-01 4.63877678e-01 -2.11365402e-01 1.88757986e-01 4.32485193e-01 2.84273356e-01 -3.85092571e-02 -1.17487997e-01 -1.31258857e+00 3.81037951e-01 1.03595626e+00 -3.64747316e-01 1.97080106e-01 7.74439812e-01 6.08123779e-01 -7.05886245e-01 -1.38538361e+00 3.71806979e-01 6.12147093e-01 -7.73972750e-01 9.58439946e-01 -7.82958150e-01 1.78184971e-01 -1.18575953e-01 -1.21542081e-01 -1.38391101e+00 -3.31619352e-01 -1.33588266e+00 -5.54857075e-01 8.40713859e-01 7.10231841e-01 -5.18574178e-01 8.57950151e-01 9.93526638e-01 1.72144458e-01 -1.37867367e+00 -7.91457772e-01 -8.52487564e-01 2.77371585e-01 -2.79767662e-01 4.70332086e-01 5.48500955e-01 -9.02858302e-02 2.68454820e-01 -4.72991675e-01 1.48349211e-01 8.58405352e-01 3.00409526e-01 6.53094232e-01 -9.33695316e-01 -8.84152293e-01 -5.45184910e-01 -1.97084658e-02 -1.11346066e+00 3.87412049e-02 -8.42051804e-01 -3.97960432e-02 -9.24442232e-01 3.95571232e-01 -8.06786299e-01 -7.16937482e-02 2.23195463e-01 -3.44288468e-01 5.11277746e-03 -2.67229453e-02 1.61787629e-01 -6.04141533e-01 7.14046121e-01 1.23862290e+00 1.21290907e-01 -8.05041865e-02 6.59487545e-02 -5.76078892e-01 5.88546395e-01 6.12031043e-01 -6.73747122e-01 -2.96062768e-01 -2.27212176e-01 5.75514555e-01 3.65701973e-01 2.24783584e-01 -9.57260489e-01 1.16647638e-01 -3.55907977e-01 5.05299568e-01 7.68541321e-02 3.55730504e-01 -5.84706664e-01 2.09531680e-01 4.51337039e-01 -6.60481632e-01 -8.52795616e-02 1.77028328e-01 6.66661561e-01 2.61386842e-01 -4.47351426e-01 1.27966595e+00 -8.41188580e-02 1.13467038e-01 5.39215386e-01 -4.11627032e-02 5.97188175e-01 9.91663456e-01 -7.76480464e-03 -1.68568283e-01 -4.11563724e-01 -7.01341808e-01 4.87462789e-01 7.50386715e-01 -2.66248256e-01 2.48114467e-01 -1.20784509e+00 -5.69349527e-01 -1.34160370e-01 -1.68486491e-01 1.04604147e-01 -6.54616281e-02 8.10631692e-01 -3.90391707e-01 1.90317348e-01 3.00779849e-01 -6.34318829e-01 -8.85094404e-01 4.12444592e-01 6.60306275e-01 -3.74365181e-01 -5.58274806e-01 9.40810204e-01 1.97932512e-01 -1.93583518e-01 2.72807360e-01 -8.66220444e-02 6.19399011e-01 -3.41140151e-01 2.70900100e-01 5.66600561e-01 4.88397032e-02 2.83614025e-02 -1.38611764e-01 6.56045318e-01 -8.23966265e-02 -4.29394871e-01 1.26459825e+00 -4.90628881e-03 2.24647433e-01 4.15702313e-01 1.18124878e+00 9.65442136e-03 -1.63259602e+00 -1.90514311e-01 -2.16706112e-01 -7.21553206e-01 4.70887065e-01 -9.04504240e-01 -9.07051384e-01 6.68494761e-01 5.07696688e-01 1.43365994e-01 9.52546656e-01 -2.40540341e-01 6.77317142e-01 6.41263604e-01 4.35999572e-01 -1.40471482e+00 -1.36219487e-01 1.08029783e-01 6.67255044e-01 -1.25926352e+00 5.81625044e-01 -2.63846833e-02 -3.69537681e-01 1.00086272e+00 3.61423165e-01 -2.60451943e-01 7.09486723e-01 3.95739406e-01 -5.45734227e-01 2.54382417e-02 -1.20901752e+00 7.79087171e-02 1.43715203e-01 4.45712060e-02 2.34567508e-01 -1.09478710e-02 -1.36979178e-01 4.11550224e-01 -1.88607275e-01 -2.05385834e-01 4.85064268e-01 8.70656610e-01 -5.35020530e-01 -1.07039857e+00 -2.46296510e-01 7.76380122e-01 -5.07006407e-01 8.77695233e-02 -1.38211688e-02 6.59092605e-01 -4.05492812e-01 6.89710021e-01 -2.90828347e-01 -1.46196932e-01 3.38825323e-02 -8.05846155e-02 8.46141219e-01 -3.81156772e-01 -6.92902207e-01 8.92204419e-02 3.52386802e-01 -9.87006724e-01 -8.63279589e-03 -7.54209638e-01 -1.31149185e+00 -5.94536960e-01 -5.27382493e-01 3.37241322e-01 7.18662560e-01 9.01107669e-01 2.88516074e-01 5.07919014e-01 9.87481594e-01 -9.34046268e-01 -1.21239078e+00 -4.10585701e-01 -4.36714321e-01 1.33469135e-01 6.27593815e-01 -7.09332705e-01 -5.55236638e-01 -4.36418027e-01]
[6.705516338348389, 4.052088260650635]
6a406546-05d3-4641-a484-117c89f64954
knn-box-a-unified-framework-for-nearest
2302.13574
null
https://arxiv.org/abs/2302.13574v1
https://arxiv.org/pdf/2302.13574v1.pdf
kNN-BOX: A Unified Framework for Nearest Neighbor Generation
Augmenting the base neural model with a token-level symbolic datastore is a novel generation paradigm and has achieved promising results in machine translation (MT). In this paper, we introduce a unified framework kNN-BOX, which enables quick development and interactive analysis for this novel paradigm. kNN-BOX decomposes the datastore-augmentation approach into three modules: datastore, retriever and combiner, thus putting diverse kNN generation methods into a unified way. Currently, kNN-BOX has provided implementation of seven popular kNN-MT variants, covering research from performance enhancement to efficiency optimization. It is easy for users to reproduce these existing works or customize their own models. Besides, users can interact with their kNN generation systems with kNN-BOX to better understand the underlying inference process in a visualized way. In the experiment section, we apply kNN-BOX for machine translation and three other seq2seq generation tasks, namely, text simplification, paraphrase generation and question generation. Experiment results show that augmenting the base neural model with kNN-BOX leads to a large performance improvement in all these tasks. The code and document of kNN-BOX is available at https://github.com/NJUNLP/knn-box.
['Jiajun Chen', 'Sizhe Liu', 'Siheng Zhao', 'ShuJian Huang', 'Yunzhe Lv', 'Qianfeng Zhao', 'Wenhao Zhu']
2023-02-27
null
null
null
null
['paraphrase-generation', 'question-generation', 'paraphrase-generation']
['computer-code', 'natural-language-processing', 'natural-language-processing']
[ 2.95810878e-01 7.33291581e-02 -3.28310490e-01 -3.78811270e-01 -9.73689914e-01 -5.91682136e-01 4.79374945e-01 -9.99857262e-02 -3.49397480e-01 9.40083265e-01 3.10105920e-01 -7.16166139e-01 1.60682470e-01 -8.08916628e-01 -8.12612534e-01 -2.67068177e-01 6.04651392e-01 6.73089921e-01 -2.73528814e-01 -6.56429410e-01 9.56983566e-02 5.01448251e-02 -1.14329588e+00 6.87177598e-01 1.26598656e+00 4.53018934e-01 4.12875652e-01 6.77874684e-01 -2.86026180e-01 3.74914706e-01 -8.16678703e-01 -8.65347862e-01 3.49879861e-01 -6.33996606e-01 -7.24322021e-01 -8.12056005e-01 3.17707568e-01 -3.16256911e-01 -9.54295695e-02 1.02743554e+00 8.84072483e-01 2.44338527e-01 3.94872576e-01 -1.20377052e+00 -1.04992890e+00 1.21316051e+00 -1.71797931e-01 9.09039006e-02 3.04305106e-01 2.68692791e-01 9.18722332e-01 -1.18216515e+00 7.51652658e-01 1.39312220e+00 6.17180586e-01 6.40264392e-01 -1.12013948e+00 -8.87440503e-01 -2.29521826e-01 1.53760239e-01 -1.17105842e+00 -5.35341680e-01 4.61293250e-01 -5.13730645e-02 1.17159808e+00 5.09884715e-01 4.72602904e-01 1.24656582e+00 1.10782437e-01 1.12911022e+00 7.33447015e-01 -6.33755505e-01 -1.24325201e-01 1.22946247e-01 7.22636729e-02 5.25272727e-01 1.29079118e-01 -8.78565697e-05 -5.61817050e-01 5.71843907e-02 6.79134786e-01 -3.40703040e-01 -1.46202207e-01 3.12802464e-01 -1.59318435e+00 7.58627892e-01 3.41032237e-01 3.69780809e-01 -1.11983098e-01 2.90612519e-01 6.67149425e-01 5.35396338e-01 6.41328275e-01 7.87079990e-01 -5.54059803e-01 -5.36666989e-01 -9.04602170e-01 5.67975581e-01 6.86256826e-01 1.42963541e+00 4.98029023e-01 2.19722956e-01 -8.42533290e-01 1.12309039e+00 -2.59514749e-01 8.55365872e-01 9.23172951e-01 -7.83172309e-01 9.64600623e-01 4.50740904e-01 -1.70143411e-01 -7.26215005e-01 -2.71912724e-01 -4.05100435e-01 -1.20022488e+00 -4.85617906e-01 2.60809690e-01 -6.31739438e-01 -8.11558723e-01 1.80815172e+00 1.67591706e-01 -2.45369792e-01 1.25738204e-01 7.54404783e-01 1.11138320e+00 1.04130089e+00 -3.00739259e-01 -1.33535648e-02 1.33311510e+00 -1.52945220e+00 -9.62442756e-01 6.26593549e-03 1.04344857e+00 -9.74924266e-01 1.37011337e+00 2.52214253e-01 -1.22923470e+00 -7.43962467e-01 -7.06069410e-01 -6.93934560e-01 -6.68547809e-01 5.26286602e-01 6.30710006e-01 2.48121545e-01 -1.18093944e+00 8.43166053e-01 -6.08879328e-01 -4.15567487e-01 1.01747684e-01 3.03913534e-01 -2.29786649e-01 5.57105467e-02 -1.73755717e+00 1.12381172e+00 7.71829009e-01 3.71947497e-01 -4.13610637e-01 -9.01972234e-01 -9.50016022e-01 -4.93293405e-02 3.46896172e-01 -1.30496347e+00 1.79068494e+00 -7.88392663e-01 -1.73360968e+00 4.17805403e-01 -4.07388002e-01 -3.87661546e-01 4.26351815e-01 -5.08373499e-01 -2.64709890e-01 -4.49183673e-01 -8.71285051e-03 7.27272511e-01 5.33209145e-01 -6.23705983e-01 -2.31899887e-01 3.14745940e-02 -1.00644074e-01 4.10981834e-01 -2.25759313e-01 2.03836352e-01 -5.54923832e-01 -1.20626068e+00 -4.74478692e-01 -8.81517231e-01 -1.59862489e-01 -5.01103103e-01 -8.08281898e-01 -3.68366748e-01 4.13018763e-01 -1.02943516e+00 1.50019085e+00 -1.65448201e+00 4.29293752e-01 -1.39566675e-01 -3.19896899e-02 6.83143020e-01 -5.46901226e-01 8.15961540e-01 -8.93588737e-02 2.33464062e-01 -3.09968770e-01 -4.36845481e-01 1.63832873e-01 1.39102921e-01 -3.43333691e-01 -2.70950705e-01 2.73443699e-01 1.57270491e+00 -1.06116378e+00 -3.12860966e-01 1.10615224e-01 1.92901328e-01 -5.93940735e-01 1.94549546e-01 -4.34998810e-01 2.69546688e-01 -2.30347887e-01 6.52698159e-01 5.87839425e-01 1.94974348e-01 4.46832813e-02 -7.84793273e-02 -1.61232486e-01 6.16839528e-01 -8.48940551e-01 2.01395559e+00 -6.49880648e-01 7.20297098e-01 -3.59991759e-01 -6.84342861e-01 8.90683949e-01 2.42090881e-01 -2.14765221e-01 -7.79837489e-01 1.69639766e-01 4.87871587e-01 1.56024294e-02 -4.18777823e-01 1.09863770e+00 6.15687249e-03 -1.73102096e-01 6.63673222e-01 2.73176700e-01 -3.16667587e-01 6.17580652e-01 2.99034476e-01 6.91851497e-01 3.19236279e-01 3.56118351e-01 -1.26021355e-01 3.16011518e-01 7.77928298e-03 4.61484224e-01 8.95764530e-01 3.24885845e-01 4.96181428e-01 3.72971147e-01 -8.96329582e-02 -1.37846923e+00 -8.13081443e-01 1.39400974e-01 1.30077219e+00 -3.37985486e-01 -7.30380893e-01 -1.02532125e+00 -4.85213310e-01 -8.64685327e-02 1.05751610e+00 -5.02914965e-01 -2.33238831e-01 -8.79775167e-01 -7.31082082e-01 1.15446734e+00 3.84048283e-01 4.71479893e-01 -1.34004331e+00 -5.78341112e-02 1.41729593e-01 -6.95509255e-01 -7.97665656e-01 -7.57063568e-01 -7.43817464e-02 -9.51425016e-01 -3.27062994e-01 -8.54359210e-01 -7.64657319e-01 4.58183169e-01 1.13201044e-01 1.35434306e+00 1.76796876e-02 3.98005731e-02 -3.26722592e-01 -7.31363356e-01 -7.15633869e-01 -7.71125078e-01 8.01981270e-01 -1.32370200e-02 -4.41850662e-01 1.05174512e-01 -5.92701793e-01 -2.36099958e-01 -3.06408554e-02 -9.69573855e-01 7.35383689e-01 7.81581938e-01 7.83042431e-01 4.84036535e-01 -6.06316090e-01 6.46375239e-01 -1.06111526e+00 1.09906518e+00 -4.49863374e-01 -3.87285143e-01 3.58053386e-01 -5.29708326e-01 2.01475099e-01 9.90006983e-01 -2.64916629e-01 -9.23054755e-01 -5.17776549e-01 -3.38120997e-01 -1.85835361e-01 9.44718346e-02 8.02406549e-01 -3.16844136e-01 5.68810403e-01 4.98571277e-01 4.43787068e-01 8.93124714e-02 -7.32583702e-01 9.41366732e-01 8.41823459e-01 5.41688383e-01 -6.18894219e-01 9.53482747e-01 -3.13297749e-01 -3.57327610e-01 -4.28839087e-01 -7.65189409e-01 6.04581833e-02 -4.35689628e-01 7.45429397e-02 4.80374128e-01 -6.85773194e-01 -3.55766654e-01 6.63595736e-01 -1.57913876e+00 -5.64967334e-01 -3.54785979e-01 3.19647729e-01 -4.18776482e-01 2.40532026e-01 -7.57356942e-01 -2.46670172e-01 -9.26799893e-01 -1.16105771e+00 1.08361387e+00 2.05399975e-01 -3.95899177e-01 -9.43720937e-01 9.24794078e-02 4.36216325e-01 7.19218493e-01 -9.19198245e-02 1.23104298e+00 -1.03530860e+00 -3.94737184e-01 -1.20697236e-02 -1.59989208e-01 5.60908437e-01 7.17100278e-02 -3.33802477e-02 -6.79775476e-01 -7.95387328e-02 -2.13595331e-01 -1.42494440e-01 7.19052911e-01 2.05771193e-01 1.25751030e+00 -5.96772075e-01 -6.58923313e-02 9.31599796e-01 1.06899977e+00 1.39169842e-01 6.85296357e-01 3.44695479e-01 8.06645036e-01 4.77610737e-01 5.57543099e-01 5.24692200e-02 3.74335140e-01 9.04734850e-01 7.61943310e-02 -3.16288292e-01 -2.62664109e-01 -4.28829491e-01 6.60442829e-01 1.51500571e+00 -1.30190477e-01 -5.60127854e-01 -8.61929893e-01 2.39490032e-01 -2.03643155e+00 -8.46884787e-01 -2.70628512e-01 1.91007805e+00 1.25170076e+00 -2.11443692e-01 -6.52301013e-02 -2.53625125e-01 7.99187481e-01 -8.84431899e-02 -5.02205551e-01 -8.84907365e-01 -2.36193225e-01 4.34166729e-01 2.37557650e-01 7.46269584e-01 -7.12608814e-01 1.42861152e+00 5.56985188e+00 1.27698708e+00 -9.52604830e-01 2.51563728e-01 4.93060201e-01 -3.29547316e-01 -4.75406140e-01 -4.92365286e-02 -9.29934204e-01 5.74976802e-01 1.04797840e+00 -7.10693061e-01 7.26431131e-01 5.28411746e-01 5.50037622e-01 1.79952428e-01 -1.22230554e+00 8.52133036e-01 5.80154061e-02 -1.46547103e+00 7.15933859e-01 -2.33491138e-01 6.84004903e-01 6.14522025e-02 -2.54027601e-02 7.62612939e-01 4.16564107e-01 -1.08062279e+00 6.32456779e-01 5.79589188e-01 9.16064262e-01 -7.53747880e-01 8.57689977e-01 5.68259656e-01 -6.40081227e-01 1.71683401e-01 -4.40722883e-01 -8.30961019e-02 4.27373022e-01 8.07653368e-01 -9.65201080e-01 1.17408979e+00 2.50769466e-01 5.96778572e-01 -6.30560458e-01 7.91721106e-01 -3.58180195e-01 5.75657129e-01 -9.94875059e-02 -2.14296773e-01 1.37944743e-01 -4.50244993e-01 6.98104382e-01 1.59308600e+00 5.27291059e-01 -1.51204303e-01 -2.70174563e-01 9.86904502e-01 -3.98231715e-01 4.09837544e-01 -5.34105480e-01 -1.20390877e-01 5.79661071e-01 1.30677056e+00 -1.95327789e-01 -6.39872670e-01 -3.54858413e-02 1.16350830e+00 3.70040894e-01 3.66317511e-01 -1.00424671e+00 -8.64549696e-01 5.55961668e-01 -1.78572148e-01 6.40314221e-02 -2.11503983e-01 -3.30893546e-01 -1.34610522e+00 1.93475470e-01 -1.49561012e+00 1.73591487e-02 -1.08210659e+00 -9.42159474e-01 8.51904690e-01 2.03492567e-01 -1.06756389e+00 -5.67079604e-01 -4.65510458e-01 -3.73832345e-01 1.14176893e+00 -1.27792144e+00 -1.20893288e+00 -6.94426075e-02 3.60998362e-01 8.02266777e-01 -2.93125719e-01 8.95500422e-01 4.35367644e-01 -8.90424550e-01 1.07867396e+00 4.90501821e-01 4.02893573e-01 8.86299551e-01 -1.12055099e+00 1.05157173e+00 9.72405434e-01 8.13880488e-02 9.73072529e-01 5.07489502e-01 -6.60632491e-01 -1.50425422e+00 -1.40269125e+00 1.38724589e+00 -5.24223685e-01 6.35244071e-01 -7.45441437e-01 -7.28856981e-01 7.28108346e-01 5.65590620e-01 -6.34507239e-01 5.86767495e-01 4.97224405e-02 -7.03674629e-02 -1.52415648e-01 -6.00940824e-01 1.08701444e+00 1.11060047e+00 -3.86671990e-01 -6.17643476e-01 5.36369503e-01 1.34785342e+00 -7.87466764e-01 -7.27270246e-01 4.24978286e-01 3.62938106e-01 -5.97125649e-01 5.15136361e-01 -9.54904318e-01 8.54108512e-01 -1.62589997e-01 -5.16237170e-02 -1.72070742e+00 -1.44660324e-01 -9.75530684e-01 -1.66554198e-01 1.40711224e+00 8.65452945e-01 -7.10500896e-01 3.32451552e-01 1.70991182e-01 -5.48070371e-01 -1.01352000e+00 -6.55585229e-01 -7.72336543e-01 3.71127784e-01 -6.02790713e-01 1.05059886e+00 9.05643165e-01 -1.11020967e-01 6.77600026e-01 -4.71586347e-01 -3.96006137e-01 9.95762199e-02 5.39757684e-02 1.08731186e+00 -5.89783967e-01 -4.07672942e-01 -6.72652245e-01 1.78542495e-01 -1.24414003e+00 5.08532152e-02 -1.55955338e+00 -7.60986730e-02 -1.75045049e+00 2.32648611e-01 -2.29810715e-01 4.37579639e-02 6.26320243e-01 -5.04146159e-01 2.26528242e-01 5.13750911e-01 2.47343644e-01 -2.10261017e-01 7.07105339e-01 1.49906051e+00 -2.08663363e-02 -2.61277974e-01 -1.22952551e-01 -9.78888869e-01 3.47529590e-01 1.05694008e+00 -4.59720731e-01 -3.47185612e-01 -1.05119634e+00 4.84174669e-01 -8.79547745e-02 1.45255819e-01 -5.86880207e-01 6.57807067e-02 2.27715690e-02 6.89437836e-02 -6.59825742e-01 1.67519599e-01 -1.95802629e-01 1.67203546e-01 3.04602176e-01 -6.28074050e-01 6.05787754e-01 4.17421281e-01 -1.39678344e-01 -2.84505635e-01 -3.78389776e-01 3.20707768e-01 -1.92196473e-01 -2.16845855e-01 1.70896351e-01 -2.77749717e-01 1.95787787e-01 4.06964391e-01 5.09868674e-02 -6.26736820e-01 -5.77826679e-01 -3.36362213e-01 3.25713128e-01 1.42823070e-01 7.54556596e-01 3.89569908e-01 -1.44886386e+00 -1.05475676e+00 1.48889154e-01 -9.68463793e-02 1.53742298e-01 1.17377669e-01 9.11116540e-01 -6.71905220e-01 8.00864816e-01 8.37643910e-03 -1.07637875e-01 -1.00512028e+00 3.73618603e-01 3.32604080e-01 -7.06041753e-01 -2.52714902e-01 9.46723342e-01 -1.10937871e-01 -1.08242822e+00 -5.78149743e-02 -6.79424047e-01 6.71615973e-02 -2.01380178e-02 4.54066128e-01 4.94105250e-01 2.94160485e-01 -2.57414043e-01 -6.70755878e-02 1.07893467e-01 -2.81674832e-01 -1.87338457e-01 1.09021914e+00 5.08818626e-02 -5.34487307e-01 3.79391819e-01 9.92367804e-01 -7.86625370e-02 -3.56637239e-01 -3.00583810e-01 -1.57649279e-01 -1.01929307e-01 -3.73330981e-01 -1.08079314e+00 -8.17659199e-01 9.18834627e-01 5.89212775e-02 -1.52555510e-01 1.12514567e+00 -3.44861388e-01 1.29776466e+00 6.55159473e-01 6.68388754e-02 -1.05353832e+00 -2.44481891e-01 1.05259478e+00 1.20688510e+00 -8.33090603e-01 -1.45323232e-01 -2.29994193e-01 -5.95302761e-01 9.50668573e-01 5.97340226e-01 2.31841624e-01 1.86040565e-01 1.54144347e-01 9.56247151e-02 1.92597985e-01 -9.87444878e-01 6.20632395e-02 3.00034493e-01 3.94816011e-01 7.10291624e-01 2.61585176e-01 -5.61126769e-01 9.40712929e-01 -9.61537123e-01 1.00595623e-01 4.75142002e-01 5.00693321e-01 -5.27460091e-02 -1.59298491e+00 -2.38425031e-01 5.08461356e-01 -3.05866510e-01 -9.25935328e-01 -5.35866559e-01 7.96822011e-01 1.56949654e-01 6.87561274e-01 -1.17482103e-01 -5.90788186e-01 6.18572891e-01 3.31640720e-01 3.18837762e-01 -8.12998772e-01 -1.04968071e+00 -2.49277055e-01 3.90299737e-01 -4.69098806e-01 6.06464259e-02 -5.27872145e-01 -8.84955525e-01 -5.97449601e-01 -3.17740560e-01 2.30947733e-01 6.90853000e-01 7.51118124e-01 7.66657233e-01 4.82683659e-01 3.20466012e-01 -6.73005283e-01 -6.65811002e-01 -1.40398514e+00 -2.25795582e-02 3.30544226e-02 5.61922416e-02 -2.84131914e-02 6.92479312e-02 6.29092604e-02]
[11.63825798034668, 9.803398132324219]
819d3b38-5f9c-4e65-8c2b-be965541cbe1
semi-supervised-outlier-detection-using
null
null
https://openreview.net/forum?id=BkS3fnl0W
https://openreview.net/pdf?id=BkS3fnl0W
Semi-supervised Outlier Detection using Generative And Adversary Framework
In a conventional binary/multi-class classification task, the decision boundary is supported by data from two or more classes. However, in one-class classification task, only data from one class are available. To build an robust outlier detector using only data from a positive class, we propose a corrupted GAN(CorGAN), a deep convolutional Generative Adversary Network requiring no convergence during training. In the adversarial process of training CorGAN, the Generator is supposed to generate outlier samples for negative class, and the Discriminator as an one-class classifier is trained to distinguish data from training datasets (i.e. positive class) and generated data from the Generator (i.e. negative class). To improve the performance of the Discriminator (one-class classifier), we also propose a lot of techniques to improve the performance of the model. The proposed model outperforms the traditional method PCA + PSVM and the solution based on Autoencoder.
['Matthias Schubert', 'Jindong Gu', 'Volker Tresp']
2018-01-01
null
null
null
iclr-2018-1
['one-class-classifier']
['methodology']
[ 3.40644449e-01 2.22770169e-01 2.15491042e-01 -8.71232226e-02 -5.23033559e-01 -6.16547644e-01 6.01422131e-01 -6.61146417e-02 -2.28062853e-01 8.04417789e-01 -4.89142567e-01 -1.71624005e-01 2.75592446e-01 -1.28083098e+00 -8.02703381e-01 -1.06675911e+00 1.79032251e-01 4.44763780e-01 1.37062117e-01 2.08709463e-01 2.84681674e-02 4.93321180e-01 -1.37038958e+00 5.35917401e-01 8.49594414e-01 1.24085081e+00 -3.52959782e-01 7.84546375e-01 -1.95540190e-02 6.62760317e-01 -1.19930446e+00 -4.40714389e-01 6.87524498e-01 -1.00003314e+00 -1.59572750e-01 -7.66956657e-02 1.05170004e-01 -2.54939705e-01 -1.07644029e-01 1.45132315e+00 5.69657266e-01 -9.50263813e-02 8.65609825e-01 -1.82996428e+00 -6.27133191e-01 4.21215564e-01 -2.87961990e-01 -6.78127408e-02 8.18778053e-02 3.09462726e-01 2.80015260e-01 -6.08599842e-01 3.73910159e-01 8.71357441e-01 4.99176353e-01 7.43787229e-01 -8.19150627e-01 -8.88132632e-01 -1.95580825e-01 -7.47071356e-02 -1.40081632e+00 -1.46351486e-01 1.11388707e+00 -4.35551643e-01 2.80081838e-01 2.96048969e-01 6.91323817e-01 1.61382258e+00 4.95637447e-01 5.60971558e-01 1.25235641e+00 -1.07298873e-01 6.70252562e-01 3.10311913e-01 -2.12008163e-01 2.12473586e-01 5.34485638e-01 4.18085724e-01 -4.55934107e-02 -4.70824629e-01 4.80632871e-01 3.21675152e-01 -1.67099610e-01 -1.95229277e-01 -6.11670196e-01 8.45801532e-01 3.17144066e-01 3.67010266e-01 -4.79187101e-01 4.62370999e-02 3.98393869e-01 5.50656796e-01 1.31649137e-01 1.48327276e-01 -2.31568441e-01 -1.85123757e-02 -9.64615524e-01 1.24028502e-02 8.92497718e-01 7.55277157e-01 6.95531845e-01 6.02742195e-01 -9.13744792e-02 2.90321440e-01 2.35158041e-01 6.05008900e-01 1.12276578e+00 -2.25981399e-01 3.78841758e-01 8.02007735e-01 -1.79456756e-01 -9.38930035e-01 -1.75344031e-02 -7.61643112e-01 -1.11744404e+00 7.49283493e-01 2.98611403e-01 -2.90048867e-01 -1.33935606e+00 1.50593174e+00 4.30920601e-01 4.16118979e-01 6.10215008e-01 6.73935115e-01 6.02516949e-01 7.74157822e-01 -3.12315792e-01 -1.85506985e-01 7.80207217e-01 -8.11979949e-01 -5.90532184e-01 -1.69063732e-01 6.25061393e-01 -6.30850971e-01 7.30707824e-01 8.44303310e-01 -4.31504637e-01 -6.08952820e-01 -1.39048767e+00 4.94709015e-01 -6.87379658e-01 3.69668722e-01 3.11017364e-01 9.42188203e-01 -3.03105891e-01 5.75675249e-01 -7.12353408e-01 5.65827973e-02 5.30925930e-01 3.93721104e-01 -6.26835823e-01 -2.35260185e-02 -1.00644076e+00 4.54823881e-01 7.46098220e-01 1.63381204e-01 -1.33268309e+00 -9.51908901e-02 -6.35602176e-01 -3.03954501e-02 -3.66655248e-03 -4.06194240e-01 5.92042148e-01 -1.70469904e+00 -1.43122339e+00 5.66755056e-01 4.76056337e-01 -5.83469689e-01 1.00314415e+00 6.74449652e-02 -7.09493458e-01 -8.12044144e-02 6.99161261e-04 1.50944486e-01 1.28040230e+00 -1.39651835e+00 -5.26274920e-01 -2.62938559e-01 -4.28296804e-01 -2.33867094e-01 -1.50445893e-01 -2.70399183e-01 2.72166342e-01 -7.63519287e-01 2.56252855e-01 -8.90248001e-01 1.13662638e-01 -3.44381332e-01 -7.62413800e-01 -6.45813271e-02 1.40308964e+00 -6.93889618e-01 8.88493240e-01 -2.49684906e+00 -2.95705557e-01 6.16616488e-01 -2.05172300e-02 4.95727390e-01 4.93293293e-02 2.06725895e-01 -5.44796884e-01 1.97241351e-01 -5.35770357e-01 -9.67146158e-02 -2.98016906e-01 5.43811262e-01 -5.28380632e-01 5.94378054e-01 3.82344991e-01 4.50568527e-01 -7.65988648e-01 2.11381659e-04 1.80385243e-02 2.10361466e-01 -1.73617721e-01 5.37406266e-01 8.39640498e-02 5.97963750e-01 -3.03938270e-01 6.79308712e-01 9.64753449e-01 2.49082088e-01 -2.62590021e-01 8.39141235e-02 4.62774843e-01 -2.57361025e-01 -1.60732257e+00 1.02028358e+00 -2.53515095e-01 2.78497845e-01 -4.02793646e-01 -9.94353473e-01 1.27442634e+00 4.43784595e-01 -3.79229337e-03 -2.03294113e-01 3.64820838e-01 4.21853691e-01 5.09233892e-01 -3.87277573e-01 -3.22933830e-02 -1.42158223e-02 -5.40610999e-02 2.64284074e-01 2.55670041e-01 1.00415386e-01 -1.83317617e-01 -1.91376641e-01 1.22015786e+00 -5.82119413e-02 2.86230505e-01 1.46461084e-01 5.14057934e-01 -2.63273656e-01 9.50794220e-01 8.60269248e-01 -8.60172510e-02 8.08800697e-01 6.81402206e-01 -5.31593859e-01 -1.17429936e+00 -9.14950311e-01 6.67591617e-02 1.21953316e-01 -8.98792669e-02 1.62341699e-01 -6.68508708e-01 -1.11496174e+00 1.54994680e-02 6.92330301e-01 -7.15217888e-01 -6.57581866e-01 -3.06336433e-01 -9.65451539e-01 8.97380531e-01 3.46978039e-01 7.80395508e-01 -1.06748068e+00 -2.78235286e-01 6.51372448e-02 2.80468673e-01 -7.16874659e-01 1.20469064e-01 5.11000693e-01 -7.05214977e-01 -1.46827888e+00 -4.14488703e-01 -6.87630594e-01 8.88454258e-01 -5.52622020e-01 4.97204065e-01 4.34703715e-02 6.78108707e-02 -1.73980996e-01 -5.55285931e-01 -6.01223469e-01 -9.47843552e-01 -1.60958692e-01 2.67820120e-01 6.23521328e-01 3.17709088e-01 -6.80901229e-01 -2.59427875e-01 1.87190458e-01 -1.14754701e+00 -3.00139517e-01 7.38056481e-01 1.02483582e+00 4.54299301e-01 5.53548992e-01 6.60552979e-01 -1.12128580e+00 4.21408981e-01 -8.28363121e-01 -5.54272592e-01 -2.92125600e-03 -5.22327006e-01 -8.99470225e-02 1.36706674e+00 -9.05313909e-01 -6.31566167e-01 6.08870499e-02 -2.82570928e-01 -7.22689867e-01 -5.23270667e-01 2.19157457e-01 -6.60632908e-01 -1.25871211e-01 9.10782039e-01 4.68063027e-01 -2.04833388e-01 -3.12443823e-01 -7.61057213e-02 8.92771482e-01 5.81098139e-01 -2.46982440e-01 1.08248937e+00 1.92069530e-01 1.30032644e-01 -5.27387738e-01 -2.39650249e-01 4.22906578e-02 -4.57868487e-01 -2.94657294e-02 7.00504124e-01 -8.44221532e-01 -1.95421532e-01 9.19614732e-01 -1.16015518e+00 -1.01271726e-01 -4.79488581e-01 6.37807548e-01 -2.28021190e-01 1.43114150e-01 -3.15197319e-01 -1.03466749e+00 -3.65017802e-01 -1.07868350e+00 6.80105031e-01 3.95128101e-01 2.25069165e-01 -6.66890383e-01 1.11086844e-02 -1.34735093e-01 7.69506022e-02 9.35632467e-01 6.97180271e-01 -1.62321746e+00 -3.94324064e-01 -1.01994550e+00 4.77836013e-01 1.04146433e+00 -3.90025266e-02 2.46485904e-01 -1.21574342e+00 -4.76325989e-01 6.01079404e-01 -1.89095065e-01 6.04232490e-01 -2.42979988e-01 1.21792734e+00 -5.86107492e-01 -1.44423589e-01 7.86435306e-01 1.49508989e+00 6.81089640e-01 8.77241135e-01 1.72740176e-01 6.67913437e-01 -6.14476092e-02 3.27607036e-01 2.38809586e-01 -3.51299435e-01 1.24218859e-01 7.50413716e-01 4.15184163e-02 6.52809888e-02 -3.65909427e-01 4.88984615e-01 6.75547302e-01 1.45184144e-01 -6.30287349e-01 -6.14839017e-01 2.58706897e-01 -1.61050987e+00 -1.03817236e+00 -1.51227489e-01 2.44563031e+00 5.70889652e-01 3.77158135e-01 -1.69636488e-01 7.07699955e-01 9.98037636e-01 -1.49746075e-01 -6.11591756e-01 -3.05211425e-01 -4.50008392e-01 4.02713388e-01 3.78650308e-01 1.47274226e-01 -1.17468941e+00 5.25075018e-01 5.05494356e+00 9.97744381e-01 -1.38936853e+00 -3.71069717e-03 7.10448921e-01 2.50007153e-01 6.77932873e-02 1.49263099e-01 -4.86613393e-01 1.00052631e+00 7.74287224e-01 2.55807996e-01 3.24963659e-01 1.28983724e+00 -2.06935525e-01 -1.07627772e-01 -1.28677142e+00 9.50761855e-01 3.85071367e-01 -6.40400708e-01 6.38792589e-02 1.47232518e-01 8.12539577e-01 -4.21172947e-01 -1.00642368e-02 4.17125195e-01 2.16705129e-01 -9.51732457e-01 5.75611591e-01 6.20389998e-01 6.05773151e-01 -8.99878442e-01 1.30968356e+00 8.24506581e-01 -5.91131985e-01 -2.89512187e-01 -4.35007006e-01 -6.39126971e-02 -4.53312308e-01 8.85465145e-01 -6.58531845e-01 8.54165733e-01 5.16699851e-01 2.36317873e-01 -7.48986006e-01 1.20269883e+00 -4.49294120e-01 8.05802047e-01 -4.83635098e-01 3.08528662e-01 1.05722345e-01 -2.86316365e-01 5.05959988e-01 7.39099503e-01 6.98728144e-01 -1.15862168e-01 2.33463585e-01 5.20809352e-01 -1.70758143e-01 -6.25816807e-02 -9.31287110e-01 5.25101414e-03 4.50926870e-01 1.11835802e+00 -6.91526353e-01 -4.78275418e-01 -1.65295348e-01 1.24611008e+00 6.23014905e-02 2.67353892e-01 -8.56656551e-01 -7.74047077e-01 1.60362169e-01 -1.62182391e-01 1.42892197e-01 1.61856949e-01 -3.34646732e-01 -1.13302410e+00 3.06232363e-01 -9.61627185e-01 4.16822672e-01 -7.00487435e-01 -1.45501745e+00 6.64593220e-01 -5.72501123e-01 -1.71245706e+00 -4.24470425e-01 -6.53766453e-01 -1.04610670e+00 8.25121105e-01 -9.44698989e-01 -1.31349480e+00 -3.88108939e-01 6.48111284e-01 -1.42791748e-01 -6.13633394e-01 8.80003035e-01 2.90312171e-01 -5.84457695e-01 9.22220945e-01 2.76489168e-01 6.61102831e-01 6.62101984e-01 -1.09949172e+00 -1.10533185e-01 1.37764704e+00 7.75343105e-02 2.21813917e-01 5.25251091e-01 -9.19946909e-01 -1.02526748e+00 -1.37739086e+00 3.23481470e-01 -1.77945912e-01 2.96952575e-01 -3.34397525e-01 -8.34774971e-01 7.57695854e-01 -5.38333394e-02 3.45465183e-01 9.34560478e-01 -6.54182434e-01 -1.71119094e-01 -1.96690321e-01 -1.75973225e+00 2.37958357e-01 3.12643200e-01 -3.96214485e-01 -5.93714356e-01 2.46389821e-01 4.00285482e-01 -5.89702010e-01 -7.19340980e-01 3.80353689e-01 2.85056442e-01 -9.18203831e-01 5.35474062e-01 -4.75353926e-01 3.35140079e-01 -7.62538373e-01 -1.32878572e-01 -1.39039457e+00 1.04754686e-01 -3.73322815e-01 -3.05527210e-01 1.25510800e+00 1.36181936e-01 -9.36819315e-01 7.65084445e-01 1.98085353e-01 -3.00147593e-01 -5.61479270e-01 -1.15685868e+00 -9.63846684e-01 1.22274108e-01 -1.83886066e-01 8.27306747e-01 1.11691749e+00 -7.13146865e-01 -6.81567565e-02 -5.91299951e-01 5.85320294e-01 3.68900657e-01 -3.99824679e-02 1.04758894e+00 -1.26289964e+00 -4.67327207e-01 4.94645424e-02 -9.79114830e-01 -3.71433407e-01 -4.81029786e-02 -9.32800055e-01 -2.70684123e-01 -9.35720146e-01 -3.56722862e-01 -4.02066022e-01 -3.50707531e-01 5.57580292e-01 -2.31264774e-02 3.36304516e-01 3.89797315e-02 1.06519364e-01 6.48296773e-02 4.35311317e-01 8.53260040e-01 -8.38152841e-02 -1.87076405e-01 4.46576715e-01 -3.88022035e-01 6.78990662e-01 9.84940827e-01 -9.77628469e-01 -2.71410972e-01 1.08765811e-02 8.39658678e-02 3.44734974e-02 5.77374637e-01 -1.64758778e+00 1.99120477e-01 -3.37430127e-02 1.15493894e+00 -5.40687799e-01 1.20808087e-01 -1.21696806e+00 5.40939152e-01 7.42867053e-01 4.78242598e-02 8.63082930e-02 -2.46700525e-01 7.03651011e-01 -3.20348710e-01 -5.92909515e-01 8.58143747e-01 -7.82114193e-02 -1.29213557e-01 3.86549234e-01 -2.20503882e-01 -1.01480454e-01 1.41472924e+00 -3.57413918e-01 -4.72519875e-01 -2.97392253e-02 -9.04887140e-01 -6.65647909e-02 6.16612434e-01 1.65508613e-01 6.53515816e-01 -1.65395927e+00 -4.57613468e-01 7.23017991e-01 8.38133916e-02 3.41042638e-01 -1.47928223e-02 2.75236398e-01 -5.30079961e-01 -3.75226289e-01 -3.92560959e-01 -4.02578026e-01 -8.27853143e-01 7.36985207e-01 4.83454317e-01 -1.96502522e-01 -1.85229003e-01 5.03338039e-01 -2.37493679e-01 -5.23222268e-01 5.06954826e-02 -5.38906790e-02 -6.28666207e-02 -7.83488080e-02 3.00458461e-01 3.32833976e-01 2.12112486e-01 -4.57661748e-01 -9.86758471e-02 6.90487549e-02 3.33156675e-01 8.78000259e-03 1.13168919e+00 7.02463627e-01 -3.70868802e-01 6.01754010e-01 1.19901502e+00 3.00399989e-01 -9.57461059e-01 2.18917966e-01 -3.89558882e-01 -4.45065618e-01 -4.36056942e-01 -7.95893192e-01 -1.10396183e+00 8.80352676e-01 1.05863822e+00 3.82934004e-01 1.22268486e+00 -6.64896250e-01 6.22342825e-01 2.32423857e-01 1.41742587e-01 -1.04011965e+00 4.22537364e-02 2.47548267e-01 5.88871777e-01 -1.19265091e+00 -1.96046054e-01 -6.79629073e-02 -5.43414176e-01 1.41671658e+00 8.76892567e-01 -5.71654320e-01 7.06842601e-01 3.19838256e-01 8.35316554e-02 1.58754572e-01 -3.59974176e-01 2.36101151e-01 4.00838396e-03 8.00334036e-01 -1.14644893e-01 -2.13829726e-02 -2.02108428e-01 8.80642116e-01 -4.83973265e-01 -1.50199756e-01 7.08639622e-01 7.60505021e-01 -2.85643220e-01 -1.12910450e+00 -5.95528424e-01 7.30605423e-01 -4.51619834e-01 1.63317844e-01 -3.25401336e-01 7.01514661e-01 7.60366619e-01 8.80221307e-01 2.86285549e-01 -6.76620483e-01 1.01784945e-01 4.89599377e-01 -7.87622258e-02 -5.06632805e-01 -6.42873585e-01 -2.77066201e-01 -3.19803208e-01 -3.26364100e-01 -6.93210214e-03 -3.14822525e-01 -9.47905123e-01 -1.71951607e-01 -4.70898479e-01 3.08750719e-01 5.90547979e-01 7.88729846e-01 1.15137674e-01 5.70837080e-01 1.00764811e+00 -3.66458029e-01 -6.96507692e-01 -1.04333031e+00 -6.29625082e-01 4.41281408e-01 3.32284838e-01 -3.53537172e-01 -8.58244836e-01 -1.14064582e-01]
[7.6405487060546875, 2.3042726516723633]
d1195e52-29a3-4bee-9061-24787013b744
highres-net-multi-frame-super-resolution-by
null
null
https://openreview.net/forum?id=HJxJ2h4tPr
https://openreview.net/pdf?id=HJxJ2h4tPr
HighRes-net: Multi-Frame Super-Resolution by Recursive Fusion
Generative deep learning has sparked a new wave of Super-Resolution (SR) algorithms that enhance single images with impressive aesthetic results, albeit with imaginary details. Multi-frame Super-Resolution (MFSR) offers a more grounded approach to the ill-posed problem, by conditioning on multiple low-resolution views. This is important for satellite monitoring of human impact on the planet -- from deforestation, to human rights violations -- that depend on reliable imagery. To this end, we present HighRes-net, the first deep learning approach to MFSR that learns its sub-tasks in an end-to-end fashion: (i) co-registration, (ii) fusion, (iii) up-sampling, and (iv) registration-at-the-loss. Co-registration of low-res views is learned implicitly through a reference-frame channel, with no explicit registration mechanism. We learn a global fusion operator that is applied recursively on an arbitrary number of low-res pairs. We introduce a registered loss, by learning to align the SR output to a ground-truth through ShiftNet. We show that by learning deep representations of multiple views, we can super-resolve low-resolution signals and enhance Earth observation data at scale. Our approach recently topped the European Space Agency's MFSR competition on real-world satellite imagery.
['Samira E. Kahou', 'Vincent Michalski', 'Julien Cornebise', 'Israel Goytom', 'Yoshua Bengio', 'Michel Deudon', 'Kris Sankaran', 'Zhichao Lin', 'Md Rifat Arefin', 'Alfredo Kalaitzis']
2020-01-01
null
null
null
iclr-2020-1
['de-aliasing', 'multi-frame-super-resolution']
['computer-vision', 'computer-vision']
[ 6.27746522e-01 2.12477937e-01 2.25812435e-01 -4.85586703e-01 -1.52701259e+00 -3.92202467e-01 8.93515766e-01 -5.59921682e-01 -3.18767816e-01 8.93728137e-01 6.33415222e-01 2.01257512e-01 -2.07747310e-01 -1.07744145e+00 -8.72682214e-01 -8.31698358e-01 -3.05677980e-01 2.50242412e-01 -1.72640532e-01 -7.99583256e-01 -2.21506178e-01 6.92665040e-01 -1.46548343e+00 4.77079749e-01 6.06284499e-01 6.87072039e-01 1.97036237e-01 8.25797677e-01 5.63754082e-01 8.18794250e-01 -1.86423481e-01 -1.52312189e-01 5.22033274e-01 -3.73906344e-01 -9.02788758e-01 8.77721421e-03 9.26403046e-01 -4.34135169e-01 -3.79650891e-01 1.30209577e+00 5.35485327e-01 1.56581685e-01 2.88107902e-01 -7.13682234e-01 -8.12754154e-01 5.48898816e-01 -1.12260199e+00 5.22458017e-01 2.58144408e-01 9.86712128e-02 1.10670459e+00 -1.05581033e+00 8.96528125e-01 1.54353273e+00 1.16110063e+00 2.22684696e-01 -1.84585738e+00 -4.13818806e-01 -9.84548107e-02 -1.44536436e-01 -1.43271434e+00 -4.74421501e-01 5.25786579e-01 -4.42061812e-01 8.15641880e-01 3.29864234e-01 3.44393909e-01 1.02807593e+00 2.25354537e-01 1.28176019e-01 1.51683438e+00 -7.60118216e-02 -1.53395310e-01 -5.79497099e-01 -2.72060394e-01 3.10962915e-01 -4.58335318e-02 7.36964881e-01 -5.82068563e-01 -3.16218771e-02 1.08517694e+00 4.32962142e-02 -5.49293220e-01 -7.94031471e-02 -1.28379500e+00 9.24190342e-01 1.04457593e+00 3.56687129e-01 -5.22362947e-01 3.17983508e-01 -6.87165260e-02 4.75718886e-01 1.10708129e+00 3.89851451e-01 -4.27173257e-01 6.05212033e-01 -1.35060179e+00 3.07403803e-01 1.34480059e-01 2.95788705e-01 9.64428246e-01 3.27172369e-01 1.73007265e-01 6.16206110e-01 1.42848596e-01 7.68318236e-01 6.05949946e-02 -1.05430818e+00 3.03361118e-01 -5.99618405e-02 2.70750999e-01 -1.00221169e+00 -6.73639059e-01 -7.87568927e-01 -1.41543758e+00 6.59912348e-01 -1.00769170e-01 -2.10583389e-01 -9.09239888e-01 2.11122608e+00 3.76449466e-01 5.20035207e-01 3.45340788e-01 1.27095973e+00 7.23745584e-01 7.31547713e-01 -2.00152546e-01 -1.35271445e-01 1.27716601e+00 -4.88813132e-01 -5.31673670e-01 -3.79479080e-01 1.38987839e-01 -5.22661865e-01 6.79382086e-01 1.78044736e-01 -1.13973546e+00 -7.59883046e-01 -1.00152993e+00 -2.90923715e-01 -3.09077650e-01 -2.07092121e-01 2.37129569e-01 1.50649667e-01 -1.66995049e+00 9.14610147e-01 -8.40614676e-01 -7.75264278e-02 5.55175245e-01 1.36273995e-01 -7.46893883e-01 -8.39772969e-02 -1.57916462e+00 1.17791188e+00 6.06087372e-02 2.57196039e-01 -1.03729308e+00 -1.02032316e+00 -1.03032231e+00 -4.56450358e-02 1.37738651e-02 -1.00259411e+00 7.12226093e-01 -1.21664786e+00 -1.02221680e+00 1.16882634e+00 1.85165942e-01 -7.37018406e-01 5.53058863e-01 -5.21171629e-01 -5.93457460e-01 8.32700431e-02 3.97381693e-01 7.44311392e-01 1.14660728e+00 -1.51319230e+00 -6.84306741e-01 -7.08900809e-01 4.51622084e-02 4.41101402e-01 5.58139622e-01 -4.10553664e-02 4.13888276e-01 -8.24490070e-01 3.08703065e-01 -7.77302623e-01 -4.78112400e-01 3.29425260e-02 -3.34321149e-02 4.72149372e-01 7.59586394e-01 -1.08025122e+00 4.61181819e-01 -2.18937111e+00 5.64934194e-01 -1.24714427e-01 4.77916151e-01 -9.48125273e-02 -4.17558938e-01 2.65383422e-02 -5.75748622e-01 1.18843079e-01 -6.63176954e-01 -2.25315824e-01 -3.44568700e-01 8.65160599e-02 -7.61515081e-01 9.70437229e-01 4.42476809e-01 8.74241471e-01 -1.09721529e+00 3.21270563e-02 2.89467722e-01 1.18821239e+00 -3.29814345e-01 5.89505136e-02 1.82322502e-01 8.73040557e-01 3.46989967e-02 3.33511472e-01 1.05792153e+00 -1.91390216e-01 -1.40082151e-01 -6.72493756e-01 -3.99551600e-01 3.10729984e-02 -1.32896984e+00 1.98158288e+00 -6.25495195e-01 6.11769140e-01 5.83693147e-01 -7.36552417e-01 7.60047615e-01 1.08499117e-01 5.92893422e-01 -8.25673819e-01 -4.56678659e-01 1.20781958e-01 -4.96548861e-01 -1.25244021e-01 8.05949569e-01 -7.15190411e-01 -8.01721439e-02 2.70604551e-01 1.67641461e-01 -3.34861070e-01 -3.81630540e-01 1.28847346e-01 7.18375146e-01 5.44837534e-01 2.66694784e-01 -3.68673772e-01 4.55273479e-01 -1.16479419e-01 3.82719934e-01 5.84027827e-01 2.75991559e-01 1.05868900e+00 -7.29943514e-02 -7.80639768e-01 -1.30719399e+00 -1.22972155e+00 -2.40751788e-01 1.09782124e+00 2.57598255e-02 2.21020952e-01 -3.83709639e-01 -2.23040774e-01 -1.78026199e-01 6.12779737e-01 -8.25583458e-01 -4.35987525e-02 -7.54584789e-01 -1.07959235e+00 4.99432087e-01 3.06357473e-01 6.89461052e-01 -7.15591073e-01 -8.66057575e-01 1.47577375e-01 -5.51240921e-01 -1.14216042e+00 -2.28846535e-01 2.43332550e-01 -8.20093215e-01 -7.39180326e-01 -8.54595065e-01 -2.32611775e-01 2.95952588e-01 4.22973365e-01 1.43825567e+00 -3.44410002e-01 -3.10052544e-01 1.94782898e-01 -1.12142570e-01 2.28475839e-01 -2.56539255e-01 -1.09527335e-01 1.94947887e-02 1.36636049e-01 -3.23534936e-01 -1.01943648e+00 -6.20073676e-01 -1.58115160e-02 -1.08542430e+00 1.48094475e-01 5.87216437e-01 1.00548339e+00 9.41304386e-01 -1.48129210e-01 2.51428038e-01 -5.97698987e-01 1.10518157e-01 -3.92880201e-01 -6.19980276e-01 3.40410322e-02 -2.30489507e-01 2.61160463e-01 1.45897627e-01 1.23383343e-01 -1.28475451e+00 1.63743988e-01 -2.24122867e-01 -5.39692640e-01 -3.51899192e-02 6.25487030e-01 4.89969775e-02 -1.82284713e-01 1.12661290e+00 2.19523773e-01 -1.95586681e-01 -4.83890265e-01 9.25240576e-01 3.14119607e-01 1.21096528e+00 -1.64182290e-01 1.18943906e+00 1.10443223e+00 2.67200202e-01 -8.78394902e-01 -1.29097462e+00 -3.07128191e-01 -8.11657727e-01 -2.16771048e-02 9.97272611e-01 -1.73833668e+00 4.29842025e-02 2.28475481e-01 -9.87568021e-01 -2.79409349e-01 -7.28436530e-01 3.23098093e-01 -6.40276849e-01 2.03165069e-01 -3.21431160e-01 -5.88949382e-01 -5.82257450e-01 -8.68211448e-01 1.66118848e+00 1.03204250e-01 1.81398034e-01 -7.55300879e-01 4.91462946e-01 2.11535454e-01 7.14753568e-01 7.85872161e-01 7.24487379e-02 -6.81032566e-03 -6.36863887e-01 3.19335610e-01 -5.34676969e-01 2.95479178e-01 -1.82337351e-02 -4.89414513e-01 -1.23714685e+00 -6.41953349e-01 2.63901711e-01 -4.66255546e-01 1.30125713e+00 5.73372364e-01 7.46529400e-01 -3.79486024e-01 6.84214244e-03 1.36292994e+00 1.77145886e+00 -5.95134616e-01 1.08334601e+00 3.80126089e-01 7.75877118e-01 4.45561826e-01 4.24176365e-01 3.12427700e-01 2.98294902e-01 7.03451455e-01 8.46552193e-01 -6.76344931e-01 -4.56904501e-01 4.07675505e-02 3.68534982e-01 -1.52277164e-02 -5.96561849e-01 1.45343080e-01 -6.31037712e-01 5.41622996e-01 -1.54937661e+00 -1.58881974e+00 -1.94275513e-01 2.24391317e+00 7.58991122e-01 -4.25381482e-01 -2.20136911e-01 -2.12850958e-01 7.17566550e-01 9.14359629e-01 -4.90552366e-01 2.57066458e-01 -7.57000089e-01 4.99962389e-01 7.51672924e-01 9.74819362e-01 -1.49845040e+00 9.32859302e-01 5.80604744e+00 6.03873551e-01 -1.26515353e+00 4.64384645e-01 6.70469820e-01 5.09291962e-02 -4.67234880e-01 -9.34442803e-02 -6.24962509e-01 -8.46918672e-02 8.17214012e-01 9.57272351e-02 5.24011612e-01 5.12594402e-01 1.39949292e-01 1.65942430e-01 -6.77102506e-01 1.00614130e+00 1.05503567e-01 -1.66636610e+00 2.05370281e-02 4.83083911e-02 9.54498827e-01 8.23200583e-01 1.91728398e-01 2.80593187e-02 6.92104578e-01 -1.23956764e+00 6.97833180e-01 7.36806154e-01 1.33599961e+00 -8.34101558e-01 5.51626444e-01 -3.17006819e-02 -1.40689683e+00 6.06769770e-02 -3.70986611e-01 2.46633306e-01 4.47780013e-01 7.65037537e-01 -7.37203434e-02 1.05567455e+00 8.47550035e-01 9.87008929e-01 -3.40026528e-01 4.51022685e-01 -1.59318343e-01 -3.58461551e-02 -3.04265589e-01 1.32525849e+00 1.72799557e-01 -2.21083090e-01 9.18375909e-01 1.20149076e+00 5.55942297e-01 4.11648184e-01 5.01910113e-02 1.06834733e+00 -1.05345748e-01 -5.90207934e-01 -6.73950911e-01 5.16733229e-01 -3.50842588e-02 1.45641470e+00 -3.17249715e-01 -1.12346105e-01 -1.88131139e-01 1.08400166e+00 1.30293563e-01 3.50880057e-01 -7.44665444e-01 7.90146813e-02 8.30731571e-01 3.30156147e-01 2.94449747e-01 -4.12364379e-02 -1.38923749e-01 -1.43846262e+00 -3.20544481e-01 -9.68300343e-01 4.98624921e-01 -1.22003639e+00 -1.32788813e+00 9.30800676e-01 -2.27575190e-02 -1.38850272e+00 -2.15796396e-01 5.89278080e-02 -2.32246295e-01 1.24322534e+00 -2.04660320e+00 -1.66856694e+00 -5.20716786e-01 6.01916254e-01 4.04992789e-01 6.00572638e-02 6.24678075e-01 1.69282287e-01 1.14746042e-01 -1.07472852e-01 3.22300494e-02 -5.39257452e-02 5.53781986e-01 -1.12388623e+00 6.62418306e-01 1.20867634e+00 2.21315295e-01 4.69050407e-02 9.25478220e-01 -5.04775286e-01 -1.04646564e+00 -1.44098842e+00 7.56345570e-01 -3.51536363e-01 6.33329570e-01 1.10098228e-01 -1.04919517e+00 8.73955309e-01 2.07609639e-01 5.66538334e-01 1.58988595e-01 -7.82727599e-02 -5.05690694e-01 -2.03512073e-01 -1.30451798e+00 1.71650618e-01 9.90743279e-01 -7.95683622e-01 -5.79769194e-01 2.85044044e-01 7.70716608e-01 -6.33046448e-01 -9.98536348e-01 6.22684419e-01 4.08814758e-01 -1.17942011e+00 1.56061304e+00 -3.90193313e-01 7.35463560e-01 -6.04704142e-01 -6.06121361e-01 -1.51479888e+00 -7.38110662e-01 -5.62752664e-01 2.27822155e-01 6.72410965e-01 1.67276949e-01 -5.07438779e-01 2.00150311e-01 1.07730635e-01 -2.69384652e-01 -2.22533002e-01 -1.12628663e+00 -5.08338928e-01 1.26048923e-01 -3.05112451e-02 5.59030294e-01 1.30199254e+00 -8.24715614e-01 4.88949209e-01 -8.05055916e-01 8.48309398e-01 1.18370473e+00 2.54283071e-01 5.83543956e-01 -1.40216911e+00 -3.61104727e-01 -3.04826319e-01 -2.23181173e-01 -6.76497161e-01 -5.56184277e-02 -8.37681711e-01 5.38937673e-02 -1.46849120e+00 2.06686556e-01 2.03504935e-02 -1.56547248e-01 3.02253574e-01 -1.44020289e-01 5.99354923e-01 1.32920459e-01 3.82608831e-01 -2.71669626e-01 5.25198400e-01 1.10345399e+00 -6.06007315e-02 -2.24090870e-02 -3.62539440e-01 -5.70650995e-01 7.37781346e-01 3.97240698e-01 -5.82582474e-01 -2.34599970e-02 -6.30461156e-01 5.70780814e-01 4.61560369e-01 9.19621348e-01 -9.07333314e-01 -1.17719933e-01 -8.50318968e-02 6.71685338e-01 -6.00734949e-01 4.52411771e-01 -6.43389046e-01 7.70363748e-01 4.00578439e-01 -4.76211280e-01 -7.01017445e-03 2.44526528e-02 5.94527602e-01 -2.59838730e-01 4.10641760e-01 1.36260545e+00 -3.46318364e-01 -7.20877230e-01 4.58350420e-01 1.78813353e-01 2.43544474e-01 5.00419974e-01 1.47676483e-01 -4.60285157e-01 -4.02149945e-01 -1.07596636e+00 7.69891739e-02 5.74116468e-01 2.37562031e-01 5.89922071e-01 -1.23309934e+00 -1.54409540e+00 2.91231364e-01 -1.65057063e-01 3.37688237e-01 6.07892871e-01 7.81853020e-01 -4.05020773e-01 -1.72437340e-01 -3.43502611e-01 -6.99243128e-01 -1.02481353e+00 2.87888676e-01 8.15416455e-01 -7.26282060e-01 -1.00825906e+00 8.91889215e-01 4.52499479e-01 -5.88307679e-01 -3.96780670e-01 -1.17187738e-01 -2.84703434e-01 2.38335788e-01 8.65122676e-01 2.48741284e-01 -1.69552248e-02 -1.20521522e+00 -2.64679939e-01 7.84869850e-01 3.05019647e-01 -6.13240957e-01 1.98115218e+00 -4.48626906e-01 -1.73752770e-01 2.29280308e-01 1.02491188e+00 -1.96024418e-01 -1.66189957e+00 -5.01488507e-01 -3.25299501e-01 -5.73099613e-01 6.94302857e-01 -8.11363459e-01 -1.46004426e+00 7.17889130e-01 9.35441136e-01 -4.50892560e-02 1.22567821e+00 -9.50665548e-02 5.73359668e-01 1.56985387e-01 4.12280113e-01 -7.46149898e-01 -9.14148465e-02 4.29129750e-01 1.53479314e+00 -1.37966192e+00 3.61357421e-01 1.24474019e-01 -5.32871008e-01 8.69690239e-01 1.26263022e-01 -4.42925066e-01 6.19200766e-01 3.49971652e-01 -1.14458047e-01 -4.19839442e-01 -5.30420005e-01 -4.85747188e-01 3.55756462e-01 8.15483630e-01 2.61053979e-01 8.86360854e-02 2.41430447e-01 1.76096335e-01 -8.37160125e-02 8.49409029e-02 5.57455301e-01 5.63488662e-01 -5.17929733e-01 -4.85736847e-01 -7.91089535e-01 1.64153278e-01 -3.46299767e-01 -4.10833418e-01 -5.23862150e-03 6.39183640e-01 1.97060704e-01 5.23943484e-01 1.75319925e-01 -1.95611864e-01 3.51083368e-01 -4.03049648e-01 4.77200150e-01 -3.62088770e-01 -3.63988638e-01 2.46285066e-01 -1.77030191e-02 -9.94030297e-01 -9.60102975e-01 -8.42208743e-01 -7.95588613e-01 -2.63860613e-01 -7.68159935e-03 -1.63890362e-01 3.81081045e-01 7.76865482e-01 3.30027789e-01 5.05001426e-01 7.98050880e-01 -1.70934916e+00 -6.49830103e-01 -8.50994229e-01 -6.28618062e-01 2.99393982e-01 1.01661265e+00 -4.23799962e-01 -6.30106330e-01 2.07307175e-01]
[10.378314018249512, -1.9152246713638306]
d22f69d7-e328-4a44-9169-956dbe84bcd7
debatekg-automatic-policy-debate-case
2307.0409
null
https://arxiv.org/abs/2307.04090v1
https://arxiv.org/pdf/2307.04090v1.pdf
DebateKG: Automatic Policy Debate Case Creation with Semantic Knowledge Graphs
Recent work within the Argument Mining community has shown the applicability of Natural Language Processing systems for solving problems found within competitive debate. One of the most important tasks within competitive debate is for debaters to create high quality debate cases. We show that effective debate cases can be constructed using constrained shortest path traversals on Argumentative Semantic Knowledge Graphs. We study this potential in the context of a type of American Competitive Debate, called Policy Debate, which already has a large scale dataset targeting it called DebateSum. We significantly improve upon DebateSum by introducing 53180 new examples, as well as further useful metadata for every example, to the dataset. We leverage the txtai semantic search and knowledge graph toolchain to produce and contribute 9 semantic knowledge graphs built on this dataset. We create a unique method for evaluating which knowledge graphs are better in the context of producing policy debate cases. A demo which automatically generates debate cases, along with all other code and the Knowledge Graphs, are open-sourced and made available to the public here: https://github.com/Hellisotherpeople/DebateKG
['Allen Roush']
2023-07-09
null
null
null
null
['knowledge-graphs', 'argument-mining']
['knowledge-base', 'natural-language-processing']
[ 1.62960421e-02 1.08131409e+00 -6.45222485e-01 -3.46603334e-01 -1.30334115e+00 -1.18958056e+00 1.06792796e+00 6.02910995e-01 -2.88289607e-01 1.14020908e+00 9.16703224e-01 -1.22931027e+00 -6.39372170e-01 -9.49597657e-01 -8.74189079e-01 1.84881892e-02 2.81428486e-01 1.11761725e+00 3.81222457e-01 -3.83451790e-01 6.18411124e-01 -1.11771598e-01 -1.36384583e+00 7.04470754e-01 1.17836642e+00 4.84519750e-01 -4.15299088e-01 4.29313987e-01 -6.55781448e-01 8.11879635e-01 -6.47861481e-01 -1.06705177e+00 2.55386710e-01 -4.79996622e-01 -1.55104351e+00 -4.11973387e-01 5.21513104e-01 3.79545480e-01 8.04815143e-02 9.08973336e-01 3.32415849e-01 -7.52623454e-02 3.05887699e-01 -1.41952789e+00 -3.73308390e-01 1.60352731e+00 -1.27713010e-01 3.12297553e-01 4.79521930e-01 -2.88471609e-01 1.57096267e+00 -3.67275983e-01 1.54329276e+00 1.54095340e+00 3.69434506e-01 2.13475227e-01 -8.05527806e-01 -7.06575394e-01 -2.17504464e-02 5.98056972e-01 -4.02120739e-01 -4.00145262e-01 7.52534449e-01 -1.76483050e-01 6.96435928e-01 5.01425326e-01 9.08457160e-01 1.03531432e+00 -2.02602461e-01 7.05876648e-01 1.52202713e+00 -5.45960248e-01 3.35744053e-01 -1.86511010e-01 6.21439040e-01 4.60254818e-01 8.54968429e-01 -2.79538363e-01 -5.31090915e-01 -8.45346808e-01 2.17473298e-01 -8.12874675e-01 -8.72929301e-03 -4.26676065e-01 -8.88440788e-01 1.27504432e+00 1.97160140e-01 3.87459129e-01 1.67799846e-03 5.66326566e-02 4.03890461e-01 5.23043752e-01 5.19894660e-01 8.06122601e-01 -5.55186570e-01 -5.67061365e-01 -5.32931745e-01 9.02702749e-01 1.44029331e+00 4.72665250e-01 5.13755679e-01 -7.82648385e-01 1.87846586e-01 5.77052534e-01 1.86416253e-01 3.43037099e-01 2.03013327e-02 -1.53304267e+00 9.36550021e-01 1.03569853e+00 3.30526561e-01 -9.11924481e-01 -2.81319082e-01 -4.42207009e-01 3.10947508e-01 1.08484710e-04 7.84731448e-01 -2.71216691e-01 -4.74860668e-01 1.38429475e+00 8.81213129e-01 -2.10525349e-01 3.51608008e-01 6.26231253e-01 1.02136934e+00 2.64818937e-01 1.16875567e-01 3.57314646e-02 1.61867797e+00 -8.43754113e-01 -7.80630052e-01 -2.13807691e-02 1.06076992e+00 -9.46214139e-01 9.35890615e-01 1.47198468e-01 -1.11702037e+00 3.22609395e-01 -7.49765396e-01 -2.26751462e-01 -5.07203579e-01 -6.17248535e-01 1.02918601e+00 4.41903889e-01 -6.48263931e-01 3.37197840e-01 -1.02301337e-01 -2.55812496e-01 8.51121426e-01 -2.65522331e-01 -1.88103363e-01 -8.04896578e-02 -1.58211637e+00 1.03420091e+00 3.24811310e-01 -5.05920768e-01 -2.59307384e-01 -7.03020513e-01 -8.55595589e-01 -3.74079533e-02 1.26117837e+00 -7.49953210e-01 1.35206020e+00 -7.75030017e-01 -9.37659800e-01 1.04660296e+00 9.94353965e-02 -5.18019259e-01 7.60654032e-01 -3.90970498e-01 -4.17574078e-01 1.47288769e-01 7.22694695e-01 2.43356317e-01 1.86185930e-02 -1.26399779e+00 -8.06077659e-01 -2.75131494e-01 6.51396453e-01 1.78633586e-01 2.36024618e-01 4.17120785e-01 -5.09308688e-02 -4.28344935e-01 -1.52353123e-01 -9.80010211e-01 -1.60807729e-01 -3.12550157e-01 -4.71443981e-01 -8.01237404e-01 7.96050370e-01 -8.17440689e-01 1.18814397e+00 -1.31432831e+00 -5.75480983e-02 5.90076387e-01 7.58852884e-02 1.14480376e-01 1.70910284e-01 5.30579686e-01 7.38055184e-02 7.93242097e-01 -3.09613705e-01 6.11319542e-01 1.51695415e-01 3.78013760e-01 -4.78331953e-01 1.94813117e-01 -2.68362910e-01 1.22188067e+00 -1.07012093e+00 -7.58310676e-01 -2.46363536e-01 -9.07162651e-02 -6.93084121e-01 -6.03219688e-01 -8.23511660e-01 1.95333406e-01 -1.06444347e+00 4.56428349e-01 2.10293308e-01 -4.48787421e-01 8.23270917e-01 -4.69022989e-02 -2.83768207e-01 8.70541990e-01 -1.06485093e+00 1.90057886e+00 -7.73656964e-02 6.37732685e-01 2.33362749e-01 -9.88694310e-01 7.22498238e-01 1.71720624e-01 2.55126208e-01 -8.24970126e-01 4.15213257e-01 5.09996533e-01 4.14370298e-02 -4.56203073e-01 4.07350481e-01 2.35100955e-01 -4.57755148e-01 1.07746053e+00 -5.52236676e-01 -3.82438511e-01 5.74227154e-01 7.16190219e-01 1.01693571e+00 3.68116707e-01 2.93488503e-01 -5.73829353e-01 1.97346374e-01 1.01504099e+00 5.97988427e-01 6.73608124e-01 7.34205022e-02 -6.58747479e-02 8.82663429e-01 -5.96290469e-01 -1.21754420e+00 -6.59104049e-01 -3.49022858e-02 6.92161083e-01 1.61679104e-01 -8.18301380e-01 -5.75263083e-01 -9.48664367e-01 2.19494924e-01 1.03961313e+00 -5.47189415e-01 5.86016953e-01 -9.27041054e-01 -3.65927428e-01 6.01606786e-01 2.19798490e-01 4.98877257e-01 -8.90278280e-01 -8.24129522e-01 2.73638338e-01 -8.23722005e-01 -1.11231291e+00 1.75525233e-01 -2.53709346e-01 -4.44094032e-01 -2.07401323e+00 3.78911234e-02 -3.94031852e-01 2.70996064e-01 -1.17822267e-01 1.27140999e+00 4.25784826e-01 -3.92983586e-01 3.69102865e-01 -4.53351349e-01 -7.68581033e-01 -6.67204857e-01 8.73015597e-02 -6.48994386e-01 -5.88002324e-01 2.89978653e-01 -2.56461680e-01 -4.45285469e-01 9.64916199e-02 -7.95158803e-01 2.24666864e-01 -8.59726518e-02 4.34626997e-01 3.35539609e-01 -3.61705244e-01 5.55869460e-01 -1.38500202e+00 1.16963136e+00 -6.46842897e-01 -5.54199159e-01 6.34646773e-01 -8.22770774e-01 2.42177948e-01 1.14976034e-01 3.98967117e-01 -1.18327045e+00 -8.00810814e-01 2.37716325e-02 5.59589386e-01 1.71399802e-01 6.59448564e-01 1.09818742e-01 3.02431524e-01 9.96148586e-01 -7.56231487e-01 6.89074472e-02 -4.60418463e-01 1.09657192e+00 4.57273275e-01 2.43675873e-01 -1.22996712e+00 8.42619538e-01 7.93372154e-01 -1.88454941e-01 -1.43458486e-01 -1.19132876e+00 -2.72272527e-01 6.63086493e-03 -2.19945461e-01 6.61429346e-01 -6.89688802e-01 -6.57130778e-01 -2.32970163e-01 -8.31845641e-01 -4.71268654e-01 -4.99895722e-01 2.79620230e-01 -5.49730241e-01 3.35177690e-01 -2.00690597e-01 -4.28417385e-01 -4.50080067e-01 -7.14636683e-01 4.83106613e-01 3.41793001e-01 -7.09756851e-01 -1.11745036e+00 1.16620265e-01 1.18529212e+00 3.44429702e-01 7.05029547e-01 1.10403597e+00 -1.02702510e+00 -1.16547056e-01 3.03303808e-01 -1.13759302e-01 -2.69515187e-01 -2.44970351e-01 2.40169033e-01 -3.85811001e-01 1.13507278e-01 -3.48053664e-01 -4.81329471e-01 6.52975380e-01 2.14860782e-01 6.27947807e-01 -6.53152585e-01 -5.39678574e-01 -1.00492105e-01 1.02265811e+00 3.66448727e-03 5.28295934e-01 1.19842696e+00 2.30266064e-01 8.07141006e-01 7.85242379e-01 1.01761118e-01 9.38694119e-01 6.10791147e-01 2.26102415e-02 1.02378249e-01 -2.15858743e-01 -2.81728446e-01 -1.83436081e-01 4.88367707e-01 -2.16076821e-01 -1.31517336e-01 -1.34170318e+00 7.94121206e-01 -2.29744220e+00 -1.16711426e+00 -4.99001682e-01 1.47253633e+00 1.00505030e+00 3.20849955e-01 -4.42157313e-02 1.57584086e-01 7.83708155e-01 1.36207968e-01 -1.97070524e-01 -7.71617055e-01 -4.80384976e-01 7.15218723e-01 3.52946848e-01 8.61734331e-01 -6.46622241e-01 1.05585849e+00 5.36588717e+00 8.23573768e-01 -4.26372916e-01 2.91592866e-01 7.12074637e-01 -1.12817355e-01 -1.22409928e+00 1.07386971e+00 -4.33092415e-01 1.66554183e-01 8.44925761e-01 -8.96941066e-01 1.82513267e-01 7.13527024e-01 2.03607515e-01 -1.51065856e-01 -4.51633185e-01 4.36148316e-01 -2.35798344e-01 -2.28984189e+00 1.80743799e-01 2.76959371e-02 8.92080903e-01 -1.10378228e-01 -4.59700376e-01 1.58357993e-01 1.24101865e+00 -8.91225159e-01 8.31156194e-01 1.59800291e-01 4.37999815e-01 -6.36063039e-01 5.18594325e-01 3.16072777e-02 -7.66212583e-01 -2.77973831e-01 3.83695178e-02 -2.13510662e-01 4.46975559e-01 6.43269598e-01 -7.04922378e-01 1.15929401e+00 6.13698542e-01 5.55919051e-01 -3.81469250e-01 8.23768556e-01 -8.53891551e-01 8.84102523e-01 -4.34811741e-01 4.20701243e-02 5.26919842e-01 -1.40077993e-01 7.50976562e-01 7.89408326e-01 9.00744125e-02 4.10328627e-01 1.57763988e-01 5.66643119e-01 -4.28540677e-01 2.68708020e-01 -2.46747464e-01 -1.26074985e-01 9.72968102e-01 1.14448822e+00 -8.14970016e-01 -6.81945264e-01 -7.45992213e-02 -4.57592346e-02 3.05019170e-01 2.45382056e-01 -7.22912252e-01 -1.54146940e-01 3.98253053e-01 2.53868908e-01 1.07868806e-01 1.10470332e-01 -3.34449708e-01 -1.00653875e+00 1.09630287e-01 -1.21674597e+00 1.08957922e+00 -6.81454539e-01 -9.11081612e-01 1.37951866e-01 3.91999006e-01 -3.69247705e-01 -2.66869158e-01 -2.68464744e-01 -7.47768700e-01 3.52471352e-01 -1.57849777e+00 -1.13297963e+00 -1.50797442e-01 4.08617973e-01 3.66264015e-01 2.74714589e-01 6.03240013e-01 -1.71275791e-02 -4.25302312e-02 3.01749315e-02 -2.30244026e-01 3.57729271e-02 6.89071953e-01 -1.03265560e+00 6.31894231e-01 5.03767788e-01 3.93457025e-01 5.47273278e-01 8.33492160e-01 -9.86051917e-01 -1.15494263e+00 -4.86055315e-01 1.24879944e+00 -8.26433361e-01 1.16035950e+00 -7.49564841e-02 -7.71205604e-01 5.82500875e-01 5.08729875e-01 -4.98795450e-01 6.74241602e-01 3.12080234e-01 -5.49568474e-01 4.90042895e-01 -1.03114796e+00 5.40575802e-01 1.46379697e+00 -3.63837183e-01 -1.40669501e+00 4.24198002e-01 5.46552241e-01 -5.60214102e-01 -8.06840897e-01 3.16676676e-01 5.08901596e-01 -7.07813263e-01 6.67912602e-01 -1.01418436e+00 6.28526807e-01 -2.93013573e-01 1.39780059e-01 -1.17094338e+00 2.10967779e-01 -9.68306959e-01 5.56337796e-02 9.90890861e-01 9.51799214e-01 -1.02220571e+00 5.71353853e-01 7.24139512e-01 -1.05468929e-02 -7.10057676e-01 -1.10550249e+00 -4.50114280e-01 5.05530357e-01 -4.36853915e-01 7.81865299e-01 1.52649140e+00 2.83641040e-01 3.31129313e-01 3.22626382e-01 -3.42418879e-01 7.79412270e-01 8.85836065e-01 8.66553426e-01 -1.54933476e+00 -1.14390831e-02 -4.00210768e-01 1.82721272e-01 -3.82220060e-01 1.58753276e-01 -1.24738681e+00 -6.15177095e-01 -2.48465896e+00 2.56961048e-01 -8.23018372e-01 1.99238464e-01 6.64857864e-01 -2.01085463e-01 -1.77034065e-01 1.67142391e-01 4.54100341e-01 -6.87948048e-01 -9.62475985e-02 1.30716991e+00 1.92365013e-02 3.02121043e-02 -6.04537845e-01 -1.36664855e+00 7.73911357e-01 9.17030394e-01 -6.50157392e-01 -3.24254751e-01 -4.20852184e-01 8.94055068e-01 -2.17557028e-01 3.57371271e-01 -4.45810944e-01 2.96804488e-01 -2.96857744e-01 -1.67278275e-01 -3.38278443e-01 -2.95026809e-01 -1.50518388e-01 4.65347797e-01 4.70309228e-01 -6.39214396e-01 2.35572889e-01 1.08495422e-01 4.30928499e-01 -1.75250605e-01 -8.16794857e-02 9.72700864e-02 -3.63608688e-01 -6.86212361e-01 -2.67997831e-01 3.09690107e-02 9.48731542e-01 1.10055017e+00 -9.55891088e-02 -1.46215439e+00 -4.26047385e-01 -7.87403405e-01 5.43207347e-01 5.85343242e-01 6.57933712e-01 -6.22585416e-03 -9.58800972e-01 -1.22522378e+00 -7.33032405e-01 -4.73312242e-03 -4.84941155e-03 -1.19591512e-01 6.57393336e-01 -5.28551280e-01 4.01343495e-01 -9.39337835e-02 1.27687141e-01 -1.34159863e+00 9.63499099e-02 -1.48460314e-01 -5.17901480e-01 -8.60983193e-01 4.32702452e-01 -3.46204817e-01 -4.41540807e-01 -3.54165822e-01 -1.47028789e-01 -3.92648935e-01 3.05990130e-01 2.83492982e-01 3.37382764e-01 -3.46890241e-01 -4.09913480e-01 -5.19974411e-01 3.44868153e-01 1.06709056e-01 -3.25639397e-01 1.63688660e+00 6.82927296e-02 -2.22828314e-01 -4.25440911e-03 5.87631762e-01 6.72242165e-01 -5.97844541e-01 -1.48442417e-01 5.95798194e-01 -3.82837743e-01 -1.98130533e-01 -1.22450125e+00 -7.74016798e-01 6.19634753e-03 -2.46606663e-01 4.81232911e-01 1.30010292e-01 6.10825956e-01 6.55471921e-01 2.97510356e-01 3.15263659e-01 -1.51455665e+00 -2.63453513e-01 4.30253118e-01 9.77041900e-01 -8.53835523e-01 3.36416543e-01 -7.87933767e-01 -6.44233704e-01 9.63295996e-01 -2.47088037e-02 2.86079682e-02 4.08488005e-01 1.80278376e-01 3.74589175e-01 -1.07840312e+00 -8.24476719e-01 -6.89541548e-02 -1.96407959e-01 2.65328944e-01 2.90223420e-01 2.73563802e-01 -1.27462125e+00 7.00276077e-01 -6.90389514e-01 -1.35421500e-01 6.39864147e-01 1.18252778e+00 -3.72014582e-01 -1.54381740e+00 -4.20232683e-01 5.18660843e-01 -8.33858132e-01 -2.35260323e-01 -1.07345772e+00 1.07389820e+00 -9.71415415e-02 1.23116517e+00 -6.97920918e-02 3.07085007e-01 3.31101894e-01 7.77184814e-02 2.65586257e-01 -3.97610933e-01 -9.51343834e-01 -5.79415739e-01 1.58491504e+00 -5.93128264e-01 -9.00045812e-01 -5.71306109e-01 -1.60711193e+00 -6.01864517e-01 -3.80073994e-01 9.94426429e-01 9.78808701e-01 1.06878221e+00 6.75358891e-01 1.78075626e-01 -8.89985412e-02 2.30023071e-01 -2.35098764e-01 -4.07354623e-01 1.63379773e-01 9.62947428e-01 -4.97673452e-01 -8.87469828e-01 -4.26653534e-01 -1.55861095e-01]
[9.493098258972168, 9.537003517150879]
1d4a79e5-9536-478c-a50b-ab06a455b20f
visual-subtitle-feature-enhanced-video
2208.11307
null
https://arxiv.org/abs/2208.11307v2
https://arxiv.org/pdf/2208.11307v2.pdf
Visual Subtitle Feature Enhanced Video Outline Generation
With the tremendously increasing number of videos, there is a great demand for techniques that help people quickly navigate to the video segments they are interested in. However, current works on video understanding mainly focus on video content summarization, while little effort has been made to explore the structure of a video. Inspired by textual outline generation, we introduce a novel video understanding task, namely video outline generation (VOG). This task is defined to contain two sub-tasks: (1) first segmenting the video according to the content structure and then (2) generating a heading for each segment. To learn and evaluate VOG, we annotate a 10k+ dataset, called DuVOG. Specifically, we use OCR tools to recognize subtitles of videos. Then annotators are asked to divide subtitles into chapters and title each chapter. In videos, highlighted text tends to be the headline since it is more likely to attract attention. Therefore we propose a Visual Subtitle feature Enhanced video outline generation model (VSENet) which takes as input the textual subtitles together with their visual font sizes and positions. We consider the VOG task as a sequence tagging problem that extracts spans where the headings are located and then rewrites them to form the final outlines. Furthermore, based on the similarity between video outlines and textual outlines, we use a large number of articles with chapter headings to pretrain our model. Experiments on DuVOG show that our model largely outperforms other baseline methods, achieving 77.1 of F1-score for the video segmentation level and 85.0 of ROUGE-L_F0.5 for the headline generation level.
['Ba Yuan', 'Zhiwei Hu', 'Guohong Fu', 'Sujian Li', 'Wenjie Li', 'Min Cao', 'Yuanhang Li', 'Tangkun Zhang', 'Jingwen Wang', 'Derui Wang', 'Wenrui Xie', 'Ziqiang Cao', 'Qi Lv']
2022-08-24
null
null
null
null
['headline-generation']
['natural-language-processing']
[ 3.54001254e-01 4.00860906e-02 -3.50092709e-01 -1.58051774e-01 -6.80124164e-01 -8.48863363e-01 4.68439639e-01 1.47685427e-02 -1.44035399e-01 4.75638300e-01 4.37785000e-01 -2.50749826e-01 4.10741985e-01 -4.18023884e-01 -8.87549520e-01 -3.80865633e-01 1.82525471e-01 1.38869667e-02 2.42952541e-01 1.74266651e-01 4.01603729e-01 1.45611959e-02 -1.40735912e+00 5.06377161e-01 1.15006435e+00 1.02726638e+00 4.16704118e-01 8.00474107e-01 -4.49359477e-01 8.76025975e-01 -8.75324488e-01 -4.31644917e-01 -3.42225879e-02 -7.92172611e-01 -9.81473744e-01 6.79109335e-01 6.73198819e-01 -4.14802074e-01 -5.74832499e-01 1.04189253e+00 4.36895527e-03 2.09796354e-01 7.49568522e-01 -1.35333085e+00 -4.51873332e-01 8.34998548e-01 -8.09858501e-01 2.12983996e-01 7.50122309e-01 4.26381007e-02 1.11569369e+00 -7.95829594e-01 9.34796572e-01 1.03324080e+00 3.77156049e-01 5.44405639e-01 -7.65974641e-01 -3.80995482e-01 4.97856796e-01 2.77563035e-01 -1.31388533e+00 -1.42959192e-01 8.08983922e-01 -6.86026871e-01 7.00303495e-01 3.19062740e-01 7.57458985e-01 8.87912273e-01 4.48332867e-03 1.38676548e+00 4.80318636e-01 -3.76196861e-01 6.21725954e-02 3.83348428e-02 4.97034676e-02 6.06848955e-01 -7.53242671e-02 -7.33027935e-01 -4.42590177e-01 2.82746941e-01 5.59647501e-01 -6.26031533e-02 -5.67327678e-01 -1.52796701e-01 -1.19711685e+00 5.95722735e-01 1.53034598e-01 1.34375229e-01 -3.77335101e-01 6.65630493e-03 5.56285143e-01 -7.37624317e-02 2.43032411e-01 3.02534074e-01 -1.62576348e-01 -2.31384009e-01 -1.29074740e+00 5.18289030e-01 7.51067460e-01 1.20869374e+00 6.01053536e-01 -2.68439263e-01 -4.34142560e-01 8.55860293e-01 1.01956241e-01 3.15293700e-01 4.66556758e-01 -1.01915479e+00 8.80486608e-01 7.95683801e-01 -1.69884730e-02 -1.10992217e+00 3.30109894e-02 -6.58954978e-02 -6.47010267e-01 -3.54355842e-01 3.13777983e-01 -3.03367823e-01 -9.14812803e-01 1.34317911e+00 1.37135565e-01 -7.19716481e-04 -1.51874423e-02 8.24220717e-01 1.11755931e+00 1.18328261e+00 2.75359564e-02 -2.90657729e-01 1.60445118e+00 -1.23927641e+00 -7.56259799e-01 -2.22378284e-01 6.34832859e-01 -8.25384617e-01 9.75751340e-01 4.09604341e-01 -1.03211653e+00 -7.73427188e-01 -9.14746046e-01 -1.61007747e-01 -1.05591223e-01 4.57462698e-01 9.28364843e-02 3.21088344e-01 -8.24666917e-01 3.42760623e-01 -5.09490967e-01 -4.00882304e-01 5.03725290e-01 -1.10492446e-01 -2.83589691e-01 -1.42499551e-01 -9.78824675e-01 8.93492848e-02 7.63520002e-01 -6.41575232e-02 -6.76593840e-01 -5.46680152e-01 -1.01914477e+00 5.61943613e-02 8.11724544e-01 -5.27362347e-01 1.24039853e+00 -1.21373487e+00 -1.09434962e+00 7.78914750e-01 -3.22052896e-01 -3.65957946e-01 4.73726064e-01 -4.81073022e-01 -2.52432287e-01 6.52064145e-01 2.56473929e-01 1.05258954e+00 1.02447796e+00 -1.12786710e+00 -1.08300471e+00 -9.28961933e-02 1.38104275e-01 4.90898699e-01 -3.54734570e-01 3.83714847e-02 -1.24353516e+00 -1.06518924e+00 -7.98274651e-02 -8.82207692e-01 2.43150592e-02 -3.86123329e-01 -8.52602601e-01 -4.62826550e-01 1.13748002e+00 -1.13697934e+00 1.83367288e+00 -2.04219317e+00 3.86188418e-01 1.77971199e-02 3.41988742e-01 2.76966393e-01 -9.94100124e-02 3.64390850e-01 -8.25487170e-03 4.31780905e-01 -1.13344826e-01 3.02021615e-02 -1.68402828e-02 -2.81931728e-01 -2.72222400e-01 -1.49927691e-01 1.03513777e-01 9.23606694e-01 -7.99913943e-01 -6.01851642e-01 2.23910855e-03 1.17375784e-01 -6.97224259e-01 4.66103643e-01 -4.64723825e-01 1.83156207e-01 -4.70084548e-01 5.41577756e-01 3.71124029e-01 -2.95845181e-01 5.56910001e-02 -3.82101387e-01 -1.92997441e-01 3.12127359e-02 -1.10147870e+00 1.50673497e+00 -1.44477412e-01 9.58856225e-01 -2.86447614e-01 -8.69944274e-01 6.75298631e-01 2.32421607e-01 4.36165541e-01 -3.28136712e-01 -2.24962439e-02 -2.59930058e-03 -3.05392891e-01 -9.92080808e-01 7.90413797e-01 3.93522561e-01 -4.83310550e-01 2.76238948e-01 -8.70137960e-02 -8.45964700e-02 8.22170854e-01 7.93825388e-01 8.92771780e-01 4.64693904e-01 3.99573505e-01 2.15023100e-01 6.99437916e-01 1.22143142e-01 4.46483254e-01 6.27287805e-01 6.02049986e-03 7.75379598e-01 1.04405570e+00 -3.47054005e-01 -1.05643499e+00 -6.86893463e-01 5.44249475e-01 9.85214651e-01 2.07991749e-01 -8.74645591e-01 -1.38139629e+00 -9.23838258e-01 -3.68110865e-01 5.20693719e-01 -4.71598685e-01 1.80521086e-01 -7.54354179e-01 -2.76155412e-01 3.50581795e-01 6.15366697e-01 6.98593438e-01 -1.22129381e+00 -5.83764851e-01 1.01185046e-01 -7.89713681e-01 -1.36684871e+00 -1.12620223e+00 -5.17364919e-01 -5.01430392e-01 -1.11554313e+00 -1.05175126e+00 -9.45061743e-01 6.98796451e-01 2.97704160e-01 9.81899023e-01 1.56130061e-01 -2.30982065e-01 4.76874918e-01 -7.47877896e-01 -1.55739799e-01 -3.53541225e-01 2.46739089e-01 -3.33845019e-01 1.62381053e-01 2.80691028e-01 9.36584622e-02 -6.17606282e-01 2.56594777e-01 -1.09500551e+00 4.75384742e-01 4.45751071e-01 4.39200848e-01 5.39760172e-01 4.91645187e-02 2.72861928e-01 -1.02302694e+00 5.23723125e-01 -2.60785580e-01 -3.59737545e-01 4.09084469e-01 4.78966720e-02 -1.29251733e-01 7.40480423e-01 -3.50290418e-01 -1.02357948e+00 2.55951047e-01 -4.57934365e-02 -4.12450224e-01 -3.59107196e-01 5.20312965e-01 -3.92025948e-01 4.20006990e-01 1.40905350e-01 4.27307934e-01 -3.77842158e-01 -3.09591055e-01 4.07932490e-01 8.60512674e-01 7.64867544e-01 -3.11268896e-01 7.13258088e-01 4.80532944e-02 -5.18114388e-01 -1.29608572e+00 -8.61657619e-01 -5.85441649e-01 -6.70696378e-01 -6.62369251e-01 1.12868679e+00 -7.87790716e-01 -4.31173563e-01 4.70867217e-01 -1.22894371e+00 -2.67327726e-01 5.13146259e-02 2.70120293e-01 -5.59338808e-01 8.07829440e-01 -4.51774538e-01 -4.88514036e-01 -3.10978413e-01 -1.31063926e+00 9.95040476e-01 5.18979311e-01 -4.99946475e-01 -6.97147906e-01 -3.32312047e-01 5.68447053e-01 -1.81729525e-01 2.22419560e-01 8.92896712e-01 -5.89139283e-01 -6.30724132e-01 -1.63907766e-01 -3.90303433e-01 2.70005107e-01 -2.35246792e-02 2.98048496e-01 -5.92447162e-01 -3.98412533e-02 -4.66046363e-01 -1.45661667e-01 9.17141259e-01 4.70984191e-01 1.44123054e+00 -6.24384522e-01 -4.26664472e-01 6.21133566e-01 1.07824242e+00 5.36819339e-01 7.69303501e-01 4.11264271e-01 1.00418508e+00 7.07766771e-01 7.75488913e-01 3.70260596e-01 5.77635705e-01 7.37327933e-01 2.77203113e-01 4.23141494e-02 -1.12132117e-01 -6.27512217e-01 6.16962671e-01 8.84505391e-01 3.65069695e-02 -6.97145522e-01 -7.49112368e-01 5.34233212e-01 -1.70506072e+00 -1.09771490e+00 -1.35701463e-01 1.87180865e+00 6.10695958e-01 1.69616923e-01 4.35814291e-01 4.75946479e-02 9.53557611e-01 3.50530326e-01 -3.84754926e-01 -2.01102600e-01 1.23911701e-01 -4.33066756e-01 2.28586525e-01 1.78065434e-01 -1.40178955e+00 1.12755132e+00 5.23160124e+00 1.01216638e+00 -9.69108164e-01 -4.22479898e-01 9.64182496e-01 1.28118277e-01 -1.59141511e-01 -1.69114023e-02 -1.05780625e+00 7.96010852e-01 6.02626324e-01 -1.42984748e-01 2.62860715e-01 6.91228211e-01 5.00128806e-01 -2.59891748e-01 -8.78039956e-01 1.16255939e+00 4.54328120e-01 -1.27340877e+00 4.44722116e-01 -3.13337684e-01 9.34377730e-01 -6.12619221e-01 -1.05988577e-01 3.78467053e-01 -2.22867012e-01 -8.37597609e-01 9.49024022e-01 3.28245670e-01 8.73124480e-01 -6.38415456e-01 6.26546443e-01 2.01153308e-01 -1.55064034e+00 3.02440114e-02 -1.64355636e-01 2.79392332e-01 3.30934465e-01 2.39860281e-01 -7.15901136e-01 5.30737817e-01 6.79588974e-01 9.09429193e-01 -6.40430510e-01 1.38477969e+00 -5.02725303e-01 6.02218568e-01 2.12568194e-02 -5.38838916e-02 4.01813447e-01 -2.44121924e-01 4.04220223e-01 1.48123622e+00 5.15542030e-01 2.07541034e-01 4.87283349e-01 4.28471088e-01 -5.22260010e-01 3.65353435e-01 -3.73141378e-01 -3.47396702e-01 4.48593378e-01 1.29734075e+00 -1.15226460e+00 -6.74459398e-01 -5.42554736e-01 1.27502275e+00 -7.95058459e-02 4.11581904e-01 -1.01698172e+00 -9.28011298e-01 2.93123484e-01 7.23717436e-02 5.93102038e-01 5.10944687e-02 1.22711428e-01 -1.23914528e+00 5.05350493e-02 -1.09608746e+00 4.61046964e-01 -1.08159459e+00 -6.29281580e-01 8.42184663e-01 2.24213526e-02 -1.25195420e+00 -3.16934705e-01 -2.99365610e-01 -6.75912917e-01 3.34505975e-01 -1.11769855e+00 -8.45006526e-01 -5.97649336e-01 1.79253235e-01 1.39977753e+00 1.78391337e-01 -3.29553597e-02 1.83841422e-01 -8.40643287e-01 4.27016616e-01 -6.72282130e-02 5.48460841e-01 6.37345910e-01 -1.13943398e+00 5.18908381e-01 1.05656505e+00 2.30273873e-01 4.00344700e-01 6.66993558e-01 -9.39045131e-01 -1.00098300e+00 -1.19225979e+00 9.08765435e-01 -2.09964782e-01 6.01793826e-01 -2.10617423e-01 -8.79314005e-01 6.53076410e-01 3.37610096e-01 -4.77451891e-01 4.94864553e-01 -5.71469307e-01 -1.23793021e-01 1.15596712e-01 -5.20218313e-01 6.64125562e-01 9.39237237e-01 -2.38929316e-01 -6.79986775e-01 3.33376795e-01 7.64442086e-01 -5.07455587e-01 -5.39201558e-01 1.24057151e-01 4.37984526e-01 -7.26513386e-01 6.74190223e-01 -3.47240955e-01 8.77287269e-01 -3.90908420e-01 2.68832088e-01 -1.14254379e+00 7.06916302e-02 -7.96408892e-01 -7.91265741e-02 1.54175198e+00 3.17170829e-01 2.23799944e-01 8.09854984e-01 3.97120327e-01 -2.08626822e-01 -6.25000536e-01 -3.31236720e-01 -4.66811776e-01 -4.78487253e-01 -3.20351869e-01 2.70155221e-01 5.98352253e-01 1.55159086e-01 3.99034619e-01 -5.40442348e-01 -1.09304063e-01 3.88643682e-01 1.98558390e-01 9.12021875e-01 -9.00603235e-01 -1.60185844e-02 -5.38919806e-01 -1.45534754e-01 -1.60375547e+00 -1.05894789e-01 -6.23462677e-01 2.53803611e-01 -1.75581825e+00 5.08273780e-01 1.36840433e-01 2.07701877e-01 3.67311686e-01 -5.07281959e-01 3.65726382e-01 4.27898943e-01 2.77567744e-01 -1.21453500e+00 3.14540058e-01 1.06586051e+00 -2.08526656e-01 -4.20624226e-01 -5.13656251e-02 -6.53503418e-01 9.36165094e-01 5.67444086e-01 -9.93691161e-02 -4.10593152e-01 -2.08606571e-01 3.64830717e-02 2.23225236e-01 7.93790519e-02 -1.10637629e+00 6.10142164e-02 -1.94683194e-01 3.94855976e-01 -9.90281284e-01 -8.44194666e-02 -4.22561705e-01 -1.01289954e-02 2.94011801e-01 -5.51402450e-01 2.37993211e-01 2.93463171e-02 5.00205219e-01 -3.33097667e-01 -5.69101095e-01 3.25902104e-01 -2.39985004e-01 -1.16905892e+00 3.67428720e-01 -7.90198565e-01 3.38660330e-01 1.15116882e+00 -4.28950697e-01 -1.82624012e-01 -4.93860424e-01 -6.47964597e-01 3.72715235e-01 6.52569354e-01 5.83128333e-01 7.49768674e-01 -1.14152741e+00 -5.58574975e-01 -3.05904206e-02 1.31044745e-01 1.03936438e-02 4.52169597e-01 6.29308701e-01 -8.84800911e-01 5.65341830e-01 3.09674609e-02 -4.61908877e-01 -1.56984282e+00 5.50426960e-01 -1.15169488e-01 -1.33622661e-01 -7.06393957e-01 8.33194673e-01 5.85121334e-01 2.97057152e-01 4.42917645e-01 -4.04480487e-01 -6.99172139e-01 3.78087640e-01 9.54202056e-01 3.20946425e-01 -3.31921875e-01 -8.65588248e-01 -6.53113872e-02 7.22454965e-01 -3.20909679e-01 1.64034083e-01 1.06836224e+00 -3.53397846e-01 8.25668499e-02 1.17967755e-01 1.26817536e+00 -2.31787302e-02 -1.49897408e+00 7.23898783e-02 2.23165274e-01 -3.16299111e-01 -2.83398986e-01 -5.28391600e-01 -1.03269756e+00 8.59796882e-01 5.41754030e-02 3.35619181e-01 1.19676065e+00 6.61363751e-02 1.11529410e+00 1.25537261e-01 -7.83627108e-02 -1.11271584e+00 4.39662188e-01 4.87778664e-01 7.04337358e-01 -9.17578399e-01 1.63075607e-02 -6.50278866e-01 -1.01791477e+00 1.28821230e+00 7.55448580e-01 1.75268799e-01 -1.02207825e-01 -6.27548620e-02 -4.41955067e-02 4.00308929e-02 -4.99159008e-01 -7.75575116e-02 6.18707657e-01 3.93141836e-01 4.36598510e-01 -1.73808888e-01 -3.47003847e-01 5.79954803e-01 -1.89707801e-01 1.52403368e-02 8.25136542e-01 6.38674319e-01 -7.22964227e-01 -8.08310211e-01 -4.39769924e-01 5.08937955e-01 -7.47694910e-01 2.13349424e-02 -4.36755896e-01 6.66636586e-01 -2.25648563e-02 8.44215751e-01 1.24523297e-01 -4.02370691e-01 1.78956181e-01 4.70699631e-02 1.87883243e-01 -7.01169550e-01 -3.44039768e-01 4.47046638e-01 -1.58030819e-02 -4.49972123e-01 -4.27465975e-01 -7.65612125e-01 -1.28494799e+00 2.25654691e-02 -1.08293099e-02 4.48062122e-01 4.27550286e-01 9.84545231e-01 1.77078769e-01 7.29563773e-01 3.85773689e-01 -7.91477621e-01 1.44764200e-01 -7.36620247e-01 -1.82120413e-01 6.54712260e-01 5.03657162e-02 -2.72816211e-01 -1.78010732e-01 6.49277747e-01]
[10.473539352416992, 0.5773870348930359]
def221f4-464c-4f55-b4fa-c2011ba7b1af
embrace-opportunities-and-face-challenges
2305.18616
null
https://arxiv.org/abs/2305.18616v1
https://arxiv.org/pdf/2305.18616v1.pdf
Embrace Opportunities and Face Challenges: Using ChatGPT in Undergraduate Students' Collaborative Interdisciplinary Learning
ChatGPT, launched in November 2022, has gained widespread attention from students and educators globally, with an online report by Hu (2023) stating it as the fastest-growing consumer application in history. While discussions on the use of ChatGPT in higher education are abundant, empirical studies on its impact on collaborative interdisciplinary learning are rare. To investigate its potential, we conducted a quasi-experimental study with 130 undergraduate students (STEM and non-STEM) learning digital literacy with or without ChatGPT over two weeks. Weekly surveys were conducted on collaborative interdisciplinary problem-solving, physical and cognitive engagement, and individual reflections on ChatGPT use. Analysis of survey responses showed significant main effects of topics on collaborative interdisciplinary problem-solving and physical and cognitive engagement, a marginal interaction effect between disciplinary backgrounds and ChatGPT conditions for cognitive engagement, and a significant interaction effect for physical engagement. Sentiment analysis of student reflections suggested no significant difference between STEM and non-STEM students' opinions towards ChatGPT. Qualitative analysis of reflections generated eight positive themes, including efficiency, addressing knowledge gaps, and generating human-like responses, and eight negative themes, including generic responses, lack of innovation, and counterproductive to self-discipline and thinking. Our findings suggest that ChatGPT use needs to be optimized by considering the topics being taught and the disciplinary backgrounds of students rather than applying it uniformly. These findings have implications for both pedagogical research and practices.
['Tan Lay Poh', 'Low Kin Yew', 'Preman Rajalingam', 'Annabel Chen Shen-Hsing', 'Peter Seow', 'Tianlong Zhong', 'Chenyu Hou', 'Xiuyi Fan', 'Gaoxia Zhu']
2023-05-23
null
null
null
null
['sentiment-analysis']
['natural-language-processing']
[-2.96580583e-01 3.82498980e-01 -5.76518118e-01 1.25409871e-01 -5.95097303e-01 -8.64736736e-01 5.15309513e-01 6.34149790e-01 -2.28449464e-01 3.36202174e-01 7.09408462e-01 -9.55101788e-01 -3.73699903e-01 -7.92634964e-01 -5.55437803e-01 -3.93841147e-01 7.66349673e-01 -3.03185344e-01 1.94490850e-01 -2.95574576e-01 1.03296030e+00 2.13611439e-01 -1.45897007e+00 1.29228264e-01 1.32118750e+00 2.09438875e-01 2.81080931e-01 2.96248108e-01 -6.07095659e-01 1.28835201e+00 -9.17462885e-01 -4.72672552e-01 -2.81610459e-01 -9.15611446e-01 -7.40808606e-01 1.49233878e-01 3.48186105e-01 -1.45956427e-01 1.48723125e-01 8.44672740e-01 6.20548368e-01 4.14851397e-01 -1.49207097e-02 -7.98675716e-01 -1.07316017e+00 7.07742572e-01 -3.61836851e-01 2.28948295e-01 7.41115153e-01 2.19693050e-01 6.49618924e-01 -5.07875621e-01 7.69838214e-01 8.90438378e-01 7.27254689e-01 2.40519255e-01 -9.25628364e-01 -9.15996194e-01 1.96246654e-01 2.72108912e-01 -9.57008600e-01 -3.38951387e-02 3.45532417e-01 -7.86674500e-01 5.31063974e-01 2.23865077e-01 1.72713983e+00 8.83855164e-01 6.17833793e-01 1.30115733e-01 1.47059274e+00 -4.19718087e-01 3.62608880e-01 8.81329775e-01 1.10390924e-01 2.65501589e-01 3.14505726e-01 -6.00378692e-01 -6.33919418e-01 8.25519115e-02 7.98589587e-01 -5.13310507e-02 -3.44238073e-01 3.60878170e-01 -1.07416010e+00 8.20443571e-01 1.77420303e-02 6.89390540e-01 -2.31192768e-01 -2.25115195e-01 3.35551947e-01 5.81900716e-01 4.50915128e-01 5.45908272e-01 -1.77401215e-01 -1.08526587e+00 -3.21838945e-01 5.43825254e-02 1.13623869e+00 8.04919541e-01 1.79784313e-01 -2.13428125e-01 9.73509178e-02 8.11169446e-01 3.34614187e-01 1.12928592e-01 4.27438378e-01 -9.21544194e-01 -1.28890544e-01 8.67296517e-01 -3.92116755e-01 -9.57528353e-01 1.92856297e-01 -5.05121648e-01 8.08690414e-02 1.33475736e-01 5.02975464e-01 -4.75417584e-01 -2.81545192e-01 1.25981665e+00 1.73436925e-01 -9.67801213e-02 -1.58992186e-01 8.19499612e-01 1.25761592e+00 5.49451053e-01 4.41106409e-01 -1.84095591e-01 1.39879489e+00 -8.48129451e-01 -1.03890896e+00 -7.46138766e-03 1.10346866e+00 -1.30111814e+00 1.22473395e+00 5.76536894e-01 -1.38222730e+00 -3.57221454e-01 -6.74368560e-01 -1.48356080e-01 -3.62704694e-01 -2.63756931e-01 3.11320871e-01 1.36594117e+00 -1.04756224e+00 3.21949601e-01 -3.63126397e-01 -6.24747038e-01 4.63476658e-01 -6.35315999e-02 -1.32855013e-01 -2.88042307e-01 -6.16503716e-01 1.00933480e+00 -7.96666518e-02 -3.25524122e-01 -2.90166646e-01 -1.39309788e+00 -2.72008449e-01 1.02443047e-01 4.53027546e-01 -5.00286579e-01 1.34483480e+00 -9.52845454e-01 -2.04197073e+00 5.77385426e-01 2.35404298e-01 3.78783405e-01 2.45980322e-01 -3.59239846e-01 -1.91676363e-01 1.07239723e-01 1.21785566e-01 2.00713247e-01 -2.18071088e-01 -6.68506980e-01 -4.81899559e-01 -6.53927922e-02 1.84271857e-01 5.56679606e-01 -8.71433794e-01 1.34750575e-01 8.14745277e-02 -2.25594267e-01 4.30909395e-01 -7.11903334e-01 -4.45927493e-02 -1.73045360e-02 3.04469496e-01 -5.45687318e-01 8.87625396e-01 -6.40957355e-01 1.12607872e+00 -2.07016301e+00 -4.04440761e-01 1.85207818e-02 3.57288957e-01 -1.21849373e-01 1.22701019e-01 1.03951085e+00 2.16709182e-01 6.29158556e-01 6.43667877e-01 3.18323433e-01 6.38973713e-02 8.22578557e-03 2.40787938e-01 3.66592884e-01 -2.54085720e-01 5.31428337e-01 -1.36122262e+00 -2.10600197e-01 4.07256722e-01 6.38672411e-01 -5.89904308e-01 -3.95275503e-02 1.34707615e-01 4.99821037e-01 -3.77476931e-01 4.63066787e-01 5.71863115e-01 -3.47325146e-01 4.07249063e-01 7.02676177e-01 -1.14428794e+00 9.25614297e-01 -7.75871158e-01 1.36836278e+00 -8.60359013e-01 1.12207055e+00 6.38358817e-02 -6.22400403e-01 9.86680031e-01 6.18763447e-01 3.23288113e-01 -9.34463322e-01 2.92707562e-01 2.99337357e-01 3.57862890e-01 -6.49450958e-01 3.69470268e-01 -2.20019013e-01 3.88336182e-01 7.65761614e-01 -5.42845577e-02 -5.43923199e-01 7.58373216e-02 4.14102048e-01 1.03131020e+00 6.62091672e-02 -1.85728371e-02 -8.17033947e-01 -5.41413538e-02 1.14535578e-02 1.50580227e-01 6.03153110e-01 -8.83536041e-02 1.03118926e-01 6.96831226e-01 8.36659148e-02 -4.10507172e-01 -5.60701132e-01 -6.16843477e-02 1.15645003e+00 1.18778102e-01 -6.07148409e-01 -5.10269642e-01 -2.29490072e-01 -4.35431421e-01 9.81682837e-01 -1.21793196e-01 9.57212523e-02 -1.01426437e-01 -9.71650332e-02 -2.08225310e-01 5.34084290e-02 7.42204905e-01 -8.65653038e-01 -7.55653977e-01 3.32347631e-01 -1.81164682e-01 -1.03931737e+00 -3.03884357e-01 -1.54467076e-01 -9.54288483e-01 -8.34740341e-01 -6.15942717e-01 -1.05061090e+00 7.53932595e-01 6.82910144e-01 8.04550052e-01 4.07241523e-01 7.39031136e-02 1.01637506e+00 -4.29612815e-01 -6.23861849e-01 -3.06400687e-01 -3.63322943e-02 -5.51008821e-01 -8.65107954e-01 5.30186236e-01 -6.40416026e-01 -7.37730563e-01 2.33186230e-01 -7.00572312e-01 2.64051050e-01 5.00077605e-01 2.87056327e-01 -1.36458471e-01 -1.01051154e-02 7.65729010e-01 -8.80821526e-01 6.80379987e-01 -7.46856391e-01 -1.34660006e-01 -9.41166133e-02 -5.02239764e-01 -9.53614295e-01 2.77125120e-01 -6.27527297e-01 -1.25028658e+00 -1.07835937e+00 -2.90100396e-01 2.11693257e-01 -1.38048068e-01 1.02879643e+00 1.21655248e-01 -5.72794974e-01 7.46892273e-01 -1.09703995e-01 1.68225199e-01 5.68574034e-02 -2.88541645e-01 5.34355223e-01 -1.29935473e-01 -8.59301448e-01 3.44768107e-01 -1.25838295e-01 -4.17593390e-01 -1.40315402e+00 -8.06366920e-01 -4.14428771e-01 -6.33958131e-02 -1.07646549e+00 8.17836940e-01 -1.15869999e+00 -1.31077242e+00 1.03923857e-01 -5.64828634e-01 -7.65052319e-01 -2.77213395e-01 1.24563885e+00 1.01014078e-01 9.39465091e-02 -6.72259510e-01 -7.13900268e-01 2.60892928e-01 -1.09658539e+00 7.26663917e-02 9.96287823e-01 -4.12148386e-01 -1.48055768e+00 -7.46429956e-04 1.19677508e+00 5.86239338e-01 2.11588830e-01 9.59668577e-01 -5.14224529e-01 -6.64288819e-01 2.09041461e-02 1.26821414e-01 2.55584657e-01 1.69217438e-01 1.26278773e-01 -6.12110436e-01 6.94370270e-03 2.58411080e-01 -2.94483811e-01 -1.98581636e-01 2.49134123e-01 6.04315162e-01 -5.90699613e-01 -2.02291291e-02 -3.19419920e-01 1.48976231e+00 3.91452760e-01 5.02922893e-01 7.98618138e-01 4.16854680e-01 8.54805887e-01 4.27721709e-01 3.63144845e-01 4.32705790e-01 -7.23033817e-03 5.64336404e-02 5.29176474e-01 1.91284139e-02 -2.09328368e-01 6.31051898e-01 1.68909562e+00 -8.23946521e-02 -3.03280000e-02 -1.08376384e+00 9.52568054e-01 -1.23395109e+00 -8.62672865e-01 -8.40149164e-01 2.07096529e+00 5.84028661e-01 3.71314496e-01 1.00934513e-01 -1.94432870e-01 3.34102511e-01 -2.24886253e-01 8.59127715e-02 -8.63780320e-01 2.11545020e-01 5.45985878e-01 1.58601373e-01 2.11700261e-01 3.57465982e-01 4.85992223e-01 5.77085543e+00 4.06365454e-01 -1.30723083e+00 2.60955185e-01 6.30160630e-01 2.01010611e-02 -8.63244295e-01 4.20791119e-01 -5.61967134e-01 2.76941836e-01 1.05115962e+00 -5.61533272e-01 -2.20096916e-01 5.63587785e-01 4.48618442e-01 -4.76835191e-01 -5.85267365e-01 3.35346967e-01 -1.56761199e-01 -1.32843137e+00 -6.10372126e-01 3.30533385e-01 1.26387632e+00 -4.48820829e-01 2.56544687e-02 4.54548478e-01 3.62866789e-01 -6.80444360e-01 6.71451151e-01 -2.97203101e-02 6.66185329e-03 -6.55511796e-01 5.85721612e-01 2.57465541e-01 -6.66597545e-01 -2.32863352e-02 -2.52549276e-02 -9.86970365e-01 -2.65802234e-01 4.38571543e-01 -7.33621478e-01 3.29316676e-01 7.11812794e-01 5.61296701e-01 -1.75058544e-01 9.87468541e-01 -3.33492309e-01 1.08701479e+00 4.81785610e-02 -5.33734620e-01 3.62339824e-01 -6.88188970e-01 2.54685521e-01 8.36317599e-01 7.10171163e-01 6.17889225e-01 -7.42796361e-02 7.55299151e-01 2.35419199e-01 5.02350688e-01 -4.06357050e-01 -5.34966707e-01 9.68261838e-01 1.24004233e+00 -1.19251752e+00 -2.09885776e-01 -9.16000307e-01 1.37420967e-01 -3.52345824e-01 3.04259509e-01 -4.65896189e-01 -3.21122557e-01 6.56712711e-01 7.64883041e-01 -8.10913183e-03 -2.30140045e-01 -8.64971220e-01 -5.03342032e-01 -2.68144399e-01 -9.39458072e-01 -1.89713582e-01 -3.52792859e-01 -9.44505990e-01 -3.88466001e-01 -1.93612978e-01 -8.32849443e-01 5.90227008e-01 -2.78532326e-01 -1.14897072e+00 9.74930286e-01 -9.69795048e-01 -4.67778981e-01 -5.93907595e-01 -2.55400781e-02 4.00728256e-01 4.59944397e-01 4.70984817e-01 1.33371383e-01 -4.99417901e-01 3.12625974e-01 -9.57716629e-03 -5.38277090e-01 7.79505610e-01 -9.16633427e-01 -2.76535511e-01 3.50267619e-01 -7.20339835e-01 8.50830197e-01 6.94870234e-01 -6.87243223e-01 -1.60815156e+00 -3.61880183e-01 1.25606453e+00 -1.57861605e-01 7.45649993e-01 7.78193548e-02 -9.01422620e-01 6.36485338e-01 8.95817041e-01 -9.05606985e-01 1.40409541e+00 3.23580265e-01 1.79704979e-01 4.49507177e-01 -1.02976167e+00 1.09588349e+00 1.02388144e+00 -4.72341359e-01 -2.82776356e-01 5.67930341e-01 4.82258230e-01 -6.79421842e-01 -1.66517019e+00 -3.16187829e-01 5.41790187e-01 -1.06515777e+00 5.66898763e-01 2.71950245e-01 9.12447870e-01 3.16232055e-01 4.22645688e-01 -1.09546864e+00 -1.21343039e-01 -8.68904948e-01 6.67116940e-01 1.26108527e+00 2.19832789e-02 -8.66915643e-01 9.54952359e-01 1.07073760e+00 -9.46359277e-01 -7.75364339e-01 -4.99905765e-01 -4.17237878e-01 4.96946573e-01 -9.53865722e-02 3.06319669e-02 1.59798265e+00 6.99357867e-01 9.81846601e-02 2.85299987e-01 -2.57361591e-01 3.34446542e-02 -2.40318835e-01 7.65966356e-01 -1.24511552e+00 -9.63215828e-02 -8.07086647e-01 -1.74904376e-01 -7.95971870e-01 -3.76201481e-01 -6.28629029e-01 -3.38064253e-01 -2.02678180e+00 2.29856502e-02 -6.21035881e-02 5.28545320e-01 2.85961151e-01 9.11589786e-02 -3.01168650e-01 2.36321136e-01 -1.38932109e-01 -1.40304491e-01 1.67102873e-01 2.02058578e+00 5.19299567e-01 -7.52414346e-01 -2.31488869e-01 -1.32367873e+00 3.64892274e-01 8.27179074e-01 -2.21496731e-01 -6.99488282e-01 -1.06132187e-01 6.40324295e-01 2.41592541e-01 1.48599923e-01 -9.83860850e-01 4.17899758e-01 -6.42599344e-01 1.94659039e-01 -8.71799514e-02 -5.30369207e-02 -6.91932797e-01 2.02691868e-01 4.91540164e-01 -2.53458679e-01 7.95193613e-02 5.42285442e-01 -1.88537478e-01 -1.46371290e-01 -5.05964398e-01 4.85509932e-01 -1.19277880e-01 2.02690251e-02 -5.99394739e-01 -1.16488969e+00 1.47185937e-01 1.18201077e+00 -7.39413083e-01 -5.97770095e-01 -7.50703573e-01 -5.93473911e-01 1.59796059e-01 3.97257239e-01 3.61162215e-01 4.14679170e-01 -9.89400983e-01 -5.59067607e-01 -1.52816564e-01 -3.47467303e-01 1.17291464e-02 7.59780705e-01 1.25688207e+00 -7.26828337e-01 3.66969943e-01 -4.57036436e-01 -3.54696453e-01 -1.26756060e+00 -2.68113464e-01 -1.34808064e-01 1.54518411e-01 -7.10931063e-01 8.91391575e-01 8.76834914e-02 -4.89106476e-01 2.34087363e-01 -2.09987625e-01 -1.59646556e-01 3.38928312e-01 4.98234630e-01 8.46421361e-01 5.84530123e-02 -8.17834660e-02 2.39413664e-01 3.38786364e-01 -1.88970268e-01 -1.86532199e-01 1.22193539e+00 -8.03091004e-02 -2.05093231e-02 7.48827934e-01 9.29560244e-01 3.13432962e-01 -5.42459488e-01 1.44489452e-01 -7.03132004e-02 -8.57160032e-01 -3.38391066e-02 -8.87552857e-01 -6.09825194e-01 6.49029672e-01 2.42775768e-01 3.75136614e-01 8.05315733e-01 -2.31057823e-01 4.38270688e-01 -5.84727153e-02 -1.15037911e-01 -1.25864089e+00 6.37117088e-01 4.36436117e-01 7.56192386e-01 -7.49611199e-01 -6.11686185e-02 -6.80260122e-01 -4.74647492e-01 1.21655214e+00 9.57440615e-01 2.02352241e-01 8.72901320e-01 -2.35887945e-01 1.08760849e-01 -3.72623980e-01 -8.09496582e-01 5.43118477e-01 -1.10226184e-01 3.29406023e-01 1.31960869e+00 -8.53630975e-02 -1.01993060e+00 1.70420900e-01 -5.33896625e-01 2.66232461e-01 1.31380403e+00 1.43584228e+00 -6.14453912e-01 -1.16354227e+00 -5.95290542e-01 4.77725893e-01 -6.32271707e-01 1.28750399e-01 -5.19810855e-01 1.12611341e+00 9.20087025e-02 9.90356147e-01 1.89529553e-01 -1.63162753e-01 1.57331884e-01 1.63682356e-01 4.89725500e-01 -8.77060533e-01 -1.45072162e+00 2.31051561e-03 6.86845779e-02 2.16758147e-01 -2.70156831e-01 -9.04993892e-01 -1.07410181e+00 -1.13240635e+00 -5.83041191e-01 5.68070412e-01 9.86208618e-01 9.17584658e-01 3.51965904e-01 7.78824449e-01 1.72050551e-01 -2.87587225e-01 -4.83952537e-02 -1.03650022e+00 -3.49822044e-01 -3.29126567e-01 -1.25090078e-01 -1.77803844e-01 -3.76226723e-01 -2.74271309e-01]
[10.205437660217285, 7.318281173706055]
b6b6f355-2d30-4f99-b222-5b173d060de6
causal-discovery-with-unobserved-variables-a
2305.05281
null
https://arxiv.org/abs/2305.05281v2
https://arxiv.org/pdf/2305.05281v2.pdf
Causal Discovery with Unobserved Variables: A Proxy Variable Approach
Discovering causal relations from observational data is important. The existence of unobserved variables, such as latent confounders or mediators, can mislead the causal identification. To address this issue, proximal causal discovery methods proposed to adjust for the bias with the proxy of the unobserved variable. However, these methods presumed the data is discrete, which limits their real-world application. In this paper, we propose a proximal causal discovery method that can well handle the continuous variables. Our observation is that discretizing continuous variables can can lead to serious errors and comprise the power of the proxy. Therefore, to use proxy variables in the continuous case, the critical point is to control the discretization error. To this end, we identify mild regularity conditions on the conditional distributions, enabling us to control the discretization error to an infinitesimal level, as long as the proxy is discretized with sufficiently fine, finite bins. Based on this, we design a proxy-based hypothesis test for identifying causal relationships when unobserved variables are present. Our test is consistent, meaning it has ideal power when large samples are available. We demonstrate the effectiveness of our method using synthetic and real-world data.
['Yizhou Wang', 'Yu Qiao', 'Xinwei Sun', 'Mingzhou Liu']
2023-05-09
null
null
null
null
['causal-discovery', 'causal-identification']
['knowledge-base', 'reasoning']
[ 1.60295591e-01 1.93031222e-01 -8.76391768e-01 -3.90561998e-01 -4.77931589e-01 -3.62512559e-01 3.62592340e-01 1.93013430e-01 -2.60271728e-02 1.29495347e+00 4.30334836e-01 -5.40068567e-01 -4.47988153e-01 -1.18528938e+00 -8.40520382e-01 -6.72294080e-01 -2.79096395e-01 2.38734439e-01 -9.73453224e-02 3.54813904e-01 1.72798559e-01 9.33814049e-02 -1.20858550e+00 -1.38298944e-01 1.41624713e+00 4.52158064e-01 -1.70380145e-01 5.87263480e-02 1.57838389e-01 5.89888275e-01 -4.16680306e-01 -9.95173380e-02 -2.69852523e-02 -5.56293726e-01 -5.51593125e-01 -1.61473364e-01 2.79378816e-02 -5.59337556e-01 6.32255105e-03 1.13613117e+00 3.39297146e-01 -1.53126061e-01 8.23601663e-01 -1.59006250e+00 -7.30941892e-01 1.04874134e+00 -6.20194674e-01 -4.18118462e-02 1.81516513e-01 -1.06904931e-01 9.40806031e-01 -7.96524465e-01 3.33795518e-01 1.52333653e+00 7.12363839e-01 1.09712459e-01 -1.34361732e+00 -1.11745703e+00 3.98016602e-01 -1.61839515e-01 -1.36846197e+00 -4.45630997e-01 6.08197808e-01 -6.13874197e-01 -4.23232606e-03 4.01199669e-01 4.85492170e-01 1.14914155e+00 3.06034952e-01 2.34625936e-01 1.12667477e+00 -4.30808157e-01 6.61378443e-01 -1.74938217e-01 7.74924159e-02 3.66112679e-01 9.13348973e-01 3.60123485e-01 -2.17140064e-01 -7.74048805e-01 1.05973959e+00 3.69787961e-01 -2.49658123e-01 -3.45023349e-02 -1.19705009e+00 1.13937402e+00 2.23258451e-01 1.98133774e-02 -5.33313870e-01 2.40858272e-01 1.68858364e-01 1.47577584e-01 5.10238945e-01 2.67018646e-01 -3.79415244e-01 2.09973752e-01 -7.12430596e-01 2.60386825e-01 4.81298685e-01 8.00082386e-01 3.46535265e-01 -2.71592021e-01 -5.49098730e-01 4.16535258e-01 2.50069588e-01 6.16351068e-01 3.95814590e-02 -8.77248824e-01 3.66571605e-01 5.52807093e-01 5.35096884e-01 -9.91367519e-01 -1.52946442e-01 -1.58233702e-01 -1.21910620e+00 -8.80450234e-02 6.38779879e-01 -3.89930755e-01 -8.58422637e-01 2.08117223e+00 6.45779669e-01 3.11763197e-01 -1.99247196e-01 1.01657736e+00 2.48710766e-01 4.49484229e-01 3.42279166e-01 -9.26658511e-01 1.28261018e+00 -1.06671579e-01 -1.11225808e+00 2.88083851e-01 3.87826622e-01 -5.05952001e-01 1.10819376e+00 1.79460004e-01 -9.43398416e-01 -2.28159413e-01 -6.15256786e-01 2.67393202e-01 1.08612113e-01 -1.71842739e-01 8.66067290e-01 4.52423722e-01 -2.64524281e-01 3.81877780e-01 -8.02281320e-01 -6.01552166e-02 2.01414809e-01 2.58558035e-01 6.49034008e-02 6.46706531e-03 -1.68945634e+00 2.27914035e-01 1.38955951e-01 8.97854865e-02 -9.47502017e-01 -9.79751825e-01 -4.25174445e-01 2.94899225e-01 6.13329947e-01 -8.89879584e-01 9.87908721e-01 -6.78396404e-01 -9.86757159e-01 3.20614755e-01 -3.33867311e-01 -5.68683036e-02 6.58096254e-01 8.82188231e-02 -3.92853796e-01 -2.70847768e-01 6.21197343e-01 -7.26849884e-02 6.39106750e-01 -1.15720272e+00 -6.07113779e-01 -4.43543822e-01 5.86271286e-02 -2.44955629e-01 -1.81762293e-01 -1.20525472e-01 -5.33584654e-02 -7.39903927e-01 3.01843375e-01 -6.63272321e-01 -5.27427912e-01 -2.28849038e-01 -5.34423828e-01 -3.70044440e-01 3.38540792e-01 -2.93604642e-01 1.38212192e+00 -2.22263098e+00 -4.06721413e-01 2.97429442e-01 4.71787125e-01 -4.21348095e-01 2.93912053e-01 4.25644785e-01 -1.32294014e-01 5.56224167e-01 -2.71175474e-01 2.45833799e-01 -1.13480717e-01 2.48967901e-01 -4.92474139e-01 7.56773293e-01 5.29489778e-02 5.22208631e-01 -9.15832400e-01 -5.97167552e-01 -1.28054023e-01 8.78859833e-02 -7.64300942e-01 3.39663059e-01 -8.48113969e-02 6.35479093e-01 -8.92220438e-01 5.54810047e-01 8.33369613e-01 -3.97233188e-01 3.29910189e-01 1.24602184e-01 -4.77206200e-01 3.16203535e-01 -1.38604891e+00 8.81051719e-01 -1.54881969e-01 -1.32084396e-02 -1.02816708e-02 -1.17744350e+00 8.90804112e-01 6.02224529e-01 5.83790779e-01 -3.45324069e-01 4.42315591e-03 2.09689230e-01 -5.71076050e-02 -6.07530296e-01 -1.35052428e-01 -5.52447498e-01 -2.03607455e-01 2.89754709e-03 -6.84494197e-01 5.22202671e-01 -1.30748793e-01 -3.53225656e-02 1.07974315e+00 -6.02891743e-01 6.88399076e-01 -4.93926644e-01 2.38217339e-02 4.40630876e-02 1.22192383e+00 9.59860802e-01 -9.15818033e-04 2.43806645e-01 1.02212262e+00 -1.67617500e-01 -8.52281272e-01 -1.22323263e+00 -5.33366978e-01 5.94429672e-01 2.34336004e-01 2.96603758e-02 -3.66789848e-01 -5.69726884e-01 3.03871900e-01 5.61594605e-01 -8.68701756e-01 -2.53513426e-01 -4.32489634e-01 -1.06371009e+00 3.09293777e-01 5.57282805e-01 2.77133286e-01 -5.27091265e-01 -3.35131407e-01 3.26567173e-01 -2.37808123e-01 -5.32068372e-01 -3.49889398e-01 6.70731887e-02 -9.63868141e-01 -1.22410476e+00 -4.90831137e-01 -4.45692271e-01 8.75092566e-01 1.23949535e-01 8.14672649e-01 -6.89940974e-02 2.15732440e-01 -2.96259403e-01 -2.39174485e-01 -3.50786507e-01 -2.67059535e-01 -2.78013915e-01 1.66940570e-01 -1.18460342e-01 2.80042142e-01 -6.42627656e-01 -7.13599265e-01 3.30665320e-01 -8.01429331e-01 -5.82167804e-02 4.28099781e-01 1.08099294e+00 5.27571976e-01 3.16713959e-01 9.89085793e-01 -1.08513355e+00 5.32293797e-01 -9.21326637e-01 -9.53650355e-01 2.34580472e-01 -8.60603154e-01 8.20292905e-03 7.41771221e-01 -8.20192635e-01 -1.04529333e+00 -3.47324103e-01 2.94893295e-01 -2.36239299e-01 -1.11568503e-01 9.38637972e-01 -6.85287237e-01 5.34978569e-01 5.35529137e-01 -5.17765820e-01 -3.57334256e-01 -4.41481739e-01 2.15809047e-01 6.51671588e-01 2.95512676e-01 -8.11818242e-01 4.79799449e-01 6.20939851e-01 2.94535488e-01 -2.70756334e-01 -6.99601293e-01 -2.09092543e-01 -3.01452160e-01 2.52266318e-01 5.90107083e-01 -7.35251844e-01 -1.07821357e+00 -2.62077898e-01 -1.11098444e+00 -1.99639097e-01 -2.38980591e-01 1.03287280e+00 -3.79234433e-01 -9.90487412e-02 -3.63382041e-01 -1.14384508e+00 1.66094407e-01 -9.77450907e-01 6.59809172e-01 -3.00630555e-02 -4.82931852e-01 -9.88017917e-01 2.58779824e-01 -2.01577172e-01 -1.71264067e-01 5.88375509e-01 1.13411260e+00 -3.04963291e-01 -4.72565532e-01 9.68964919e-02 -2.87438840e-01 -4.52249616e-01 4.24273133e-01 1.67676061e-01 -6.37300789e-01 -1.38595730e-01 1.53510738e-02 1.25476494e-01 7.02557743e-01 9.59158897e-01 1.43857002e+00 -9.12244976e-01 -6.89197183e-01 4.21307743e-01 1.23334432e+00 3.39073181e-01 4.65974361e-01 6.73026370e-04 5.54164886e-01 6.81245744e-01 7.54294693e-01 8.49036157e-01 3.44227582e-01 4.49719489e-01 3.23076189e-01 -4.74727631e-01 4.30622011e-01 -6.68581605e-01 -5.67106642e-02 3.84656876e-01 -6.31554201e-02 -3.13916445e-01 -8.94708216e-01 7.06992745e-01 -1.96366751e+00 -9.92998838e-01 -6.73593342e-01 2.59919977e+00 1.30735409e+00 -3.70464660e-02 8.88906047e-02 2.08495438e-01 1.04386747e+00 -3.77882183e-01 -4.78951693e-01 -2.11730838e-01 1.19008332e-01 -2.69527137e-01 8.27202916e-01 5.24671316e-01 -8.16134393e-01 3.71373147e-01 6.90793133e+00 5.19122660e-01 -9.73555446e-01 1.47585899e-01 6.26383185e-01 1.80691496e-01 -6.91282630e-01 4.55052882e-01 -7.10195065e-01 8.18191767e-01 8.09834421e-01 -4.60856199e-01 -5.49134659e-03 5.33853114e-01 1.07174611e+00 -2.60198116e-02 -1.11057758e+00 3.94397408e-01 -8.73675942e-01 -9.66054201e-01 -6.31505102e-02 4.46837842e-01 8.73844802e-01 -7.58764029e-01 -6.63679913e-02 -2.73059751e-03 7.09027946e-01 -1.19499016e+00 5.14667928e-01 5.92943370e-01 1.01650190e+00 -7.49350429e-01 6.92719162e-01 4.47187543e-01 -8.01674962e-01 -1.31489396e-01 -6.14979148e-01 -5.84910750e-01 2.57144481e-01 1.50127923e+00 -6.88382447e-01 3.39925021e-01 4.38653916e-01 4.09179121e-01 1.18125342e-01 9.79433417e-01 -3.29126835e-01 1.14532495e+00 -2.12230921e-01 2.46997997e-01 -1.48230702e-01 -1.93116084e-01 2.49735937e-01 7.97016382e-01 5.29360235e-01 6.23549759e-01 4.13634449e-01 1.23160970e+00 -1.13028459e-01 5.03140800e-02 -6.73747420e-01 2.09677429e-03 1.06103098e+00 5.25217295e-01 -5.31771898e-01 -4.07369584e-01 -4.47347939e-01 3.09708178e-01 1.35514960e-01 7.03749418e-01 -1.01409471e+00 5.09689562e-03 5.58464766e-01 2.99825460e-01 -2.23194629e-01 3.69778946e-02 -8.01534891e-01 -1.04891229e+00 1.79276168e-02 -6.77688420e-01 6.74336433e-01 -1.64901897e-01 -1.52305126e+00 -3.02528113e-01 3.29418093e-01 -1.04902434e+00 -4.84064259e-02 1.04430482e-01 -6.49627447e-01 1.04843211e+00 -1.12495577e+00 -8.46091747e-01 -1.81192964e-01 4.57082927e-01 -4.55792025e-02 4.66230363e-01 6.11072361e-01 3.75592470e-01 -7.09349632e-01 5.52679121e-01 1.86009452e-01 6.98653385e-02 8.41896474e-01 -1.04510713e+00 -4.35890667e-02 6.71094954e-01 -5.05919158e-01 1.09606600e+00 9.20859396e-01 -1.20885670e+00 -9.98677015e-01 -1.08688366e+00 1.00177729e+00 -1.12795882e-01 7.36085296e-01 -3.27713042e-01 -1.10058379e+00 7.61933565e-01 -3.41396391e-01 9.76079926e-02 6.20040774e-01 6.59007847e-01 -6.32570237e-02 -1.00058787e-01 -1.14393961e+00 7.10070431e-01 1.04415298e+00 -1.06947415e-01 -5.58205068e-01 1.82209104e-01 9.23931599e-01 3.73189934e-02 -1.15898895e+00 8.29412401e-01 5.95056713e-01 -6.62681639e-01 7.83525944e-01 -7.26150095e-01 5.96499383e-01 -2.63960749e-01 9.02733803e-02 -1.23154593e+00 -4.94410872e-01 -3.93810272e-01 7.47538507e-02 1.55555272e+00 3.65563035e-01 -8.42502594e-01 4.90436643e-01 8.56335580e-01 2.92930782e-01 -3.96265924e-01 -9.76361573e-01 -7.98162520e-01 2.64164001e-01 -1.92650497e-01 9.86742675e-01 1.44399989e+00 1.94890648e-01 2.27457806e-01 -5.07820129e-01 4.91167307e-01 8.58659327e-01 3.86199623e-01 6.61840141e-01 -1.49128425e+00 -1.89140767e-01 -1.03032105e-01 2.50675865e-02 -7.60549426e-01 6.25810474e-02 -2.92179227e-01 -5.61294779e-02 -1.27480888e+00 5.32955945e-01 -1.00427616e+00 -1.06632285e-01 4.47174817e-01 -7.06023455e-01 -4.81124163e-01 -4.84063268e-01 3.96371394e-01 4.83606979e-02 7.46563196e-01 1.44023609e+00 -5.67599162e-02 -4.87249881e-01 1.73899531e-01 -8.41757715e-01 7.05812395e-01 8.69567871e-01 -8.66416693e-01 -4.55212682e-01 1.82558857e-02 2.51277268e-01 5.27252913e-01 6.38341606e-01 -1.52266994e-01 -2.96637341e-02 -9.92311060e-01 3.10885549e-01 -5.23540378e-01 -2.91048288e-01 -8.11628759e-01 5.12597322e-01 5.25521040e-01 -6.46206141e-01 -2.13498235e-01 -2.71894187e-01 6.05230629e-01 -1.20467305e-01 2.68425234e-02 5.33067465e-01 1.18949831e-01 -2.95416196e-03 2.38662586e-01 -2.81882524e-01 8.90706405e-02 8.54802430e-01 1.39005408e-01 -3.22146058e-01 -2.95623869e-01 -5.67492962e-01 4.61561948e-01 4.00752813e-01 -5.33572724e-03 3.66025686e-01 -1.65392017e+00 -7.81602979e-01 7.94485509e-02 1.14359930e-01 -8.12378526e-02 2.67891511e-02 1.14441133e+00 2.25278065e-01 5.14300644e-01 1.81259826e-01 -3.83861721e-01 -8.77363384e-01 1.07746041e+00 1.64262913e-02 8.00535306e-02 -5.02513289e-01 2.37462536e-01 7.56280303e-01 -4.80935127e-02 9.22124013e-02 -5.36989093e-01 -5.89581691e-02 6.91433111e-03 5.10870874e-01 6.67820573e-01 -4.55595791e-01 6.32200018e-02 -3.58082145e-01 2.70624518e-01 4.84430134e-01 -1.47795588e-01 9.38122392e-01 -4.32539821e-01 -3.12103629e-01 7.53042519e-01 1.08777857e+00 4.75294560e-01 -1.23786533e+00 -5.64859062e-02 4.12043242e-04 -6.93500578e-01 -4.10511270e-02 -3.68600607e-01 -6.22035086e-01 5.38127780e-01 3.65437120e-01 6.06566429e-01 1.01739228e+00 -9.30370856e-03 2.40300328e-01 -3.12358886e-01 3.28018844e-01 -7.40016997e-01 -5.36555767e-01 -9.23274923e-03 6.97428882e-01 -1.31623006e+00 8.91910419e-02 -7.93837309e-01 -1.18455969e-01 6.97509527e-01 4.50602084e-01 3.71331386e-02 5.11701465e-01 2.67112792e-01 -2.54254162e-01 -2.06222832e-01 -7.92717755e-01 8.40730965e-02 1.04487203e-01 3.10064614e-01 7.29129195e-01 6.77759647e-01 -1.05995977e+00 8.17346513e-01 -3.77691090e-02 3.01414788e-01 5.05370080e-01 5.62598526e-01 -3.17666292e-01 -8.16870153e-01 -1.00962913e+00 4.81195152e-01 -6.79863632e-01 -9.77567062e-02 -9.00423080e-02 8.00068021e-01 2.38137737e-01 1.26912928e+00 1.48172513e-01 1.74755484e-01 3.46384078e-01 -3.40775788e-01 -1.82582494e-02 -4.02718186e-01 1.66428417e-01 3.15426409e-01 3.97857977e-03 -4.06134635e-01 -5.05631208e-01 -6.75541520e-01 -1.26951230e+00 -7.20536351e-01 -5.86287916e-01 5.51498234e-01 -1.83968525e-02 9.57204998e-01 3.44485827e-02 5.81192255e-01 9.52990890e-01 3.34812850e-02 -7.66647756e-01 -8.36351037e-01 -7.41513550e-01 2.68440336e-01 4.95877355e-01 -1.03428304e+00 -5.53191900e-01 1.09234668e-01]
[7.92462682723999, 5.285148620605469]
33f8e546-7b45-4128-b121-67036aa5d675
190503556
1905.03556
null
https://arxiv.org/abs/1905.03556v1
https://arxiv.org/pdf/1905.03556v1.pdf
Cycle-IR: Deep Cyclic Image Retargeting
Supervised deep learning techniques have achieved great success in various fields due to getting rid of the limitation of handcrafted representations. However, most previous image retargeting algorithms still employ fixed design principles such as using gradient map or handcrafted features to compute saliency map, which inevitably restricts its generality. Deep learning techniques may help to address this issue, but the challenging problem is that we need to build a large-scale image retargeting dataset for the training of deep retargeting models. However, building such a dataset requires huge human efforts. In this paper, we propose a novel deep cyclic image retargeting approach, called Cycle-IR, to firstly implement image retargeting with a single deep model, without relying on any explicit user annotations. Our idea is built on the reverse mapping from the retargeted images to the given input images. If the retargeted image has serious distortion or excessive loss of important visual information, the reverse mapping is unlikely to restore the input image well. We constrain this forward-reverse consistency by introducing a cyclic perception coherence loss. In addition, we propose a simple yet effective image retargeting network (IRNet) to implement the image retargeting process. Our IRNet contains a spatial and channel attention layer, which is able to discriminate visually important regions of input images effectively, especially in cluttered images. Given arbitrary sizes of input images and desired aspect ratios, our Cycle-IR can produce visually pleasing target images directly. Extensive experiments on the standard RetargetMe dataset show the superiority of our Cycle-IR. In addition, our Cycle-IR outperforms the Multiop method and obtains the best result in the user study. Code is available at https://github.com/mintanwei/Cycle-IR.
['Bo Yan', 'Xuejing Niu', 'Weimin Tan', 'Chumin Lin']
2019-05-09
null
null
null
null
['image-retargeting']
['computer-vision']
[ 2.73942560e-01 1.11721545e-01 -2.10056216e-01 -8.77146274e-02 -3.82551819e-01 -4.56860006e-01 4.46300179e-01 -3.07671428e-01 -3.36787492e-01 5.24985492e-01 2.19908446e-01 -3.80774438e-01 1.50304392e-01 -7.25917220e-01 -8.92688811e-01 -6.27032459e-01 5.59928119e-01 -2.28579566e-01 5.24594724e-01 -4.29590106e-01 4.70869660e-01 3.92678142e-01 -1.36966538e+00 6.18247576e-02 9.65681434e-01 7.49290049e-01 6.90615177e-01 5.80798566e-01 6.79122955e-02 7.43566573e-01 -5.09095430e-01 -8.84314328e-02 3.08585078e-01 -5.45933425e-01 -8.34669650e-01 2.99139880e-02 2.69349784e-01 -4.27247405e-01 -3.71313781e-01 1.22100198e+00 7.09051251e-01 1.93429351e-01 2.71313518e-01 -1.15582144e+00 -1.18192375e+00 5.89375794e-01 -9.19515848e-01 4.24946159e-01 2.33937576e-01 3.20788175e-01 7.65558541e-01 -1.04915130e+00 3.05494308e-01 1.10557103e+00 3.44265670e-01 4.88081515e-01 -1.36672544e+00 -7.77559817e-01 3.01948965e-01 2.97529250e-01 -1.54032898e+00 -3.26452851e-01 9.87747371e-01 -3.14267784e-01 6.62069678e-01 3.69368792e-01 5.75809658e-01 9.08321798e-01 2.40501538e-01 7.58376539e-01 9.42189693e-01 -4.53617871e-01 5.69764040e-02 7.30872676e-02 -2.67200410e-01 5.11693418e-01 1.90015107e-01 2.71711737e-01 -3.22288483e-01 2.90311992e-01 1.17126656e+00 1.86900452e-01 -6.66626573e-01 -5.22934139e-01 -1.37527466e+00 7.17386186e-01 1.11811388e+00 3.45552921e-01 -1.67563498e-01 8.27320442e-02 1.22983761e-01 1.79536983e-01 2.99992114e-01 8.01016748e-01 -3.52285922e-01 3.43787432e-01 -6.93948030e-01 -2.05650385e-02 1.30848378e-01 8.06791842e-01 1.03809440e+00 7.30325282e-02 -4.67290998e-01 8.44747424e-01 1.28875956e-01 3.31954211e-01 8.63975346e-01 -7.21758723e-01 2.19842374e-01 6.17510200e-01 2.05754548e-01 -1.40886283e+00 -4.98631895e-01 -4.74076837e-01 -8.89701307e-01 3.65998745e-01 2.55494062e-02 -1.01340272e-01 -1.17351544e+00 1.79273379e+00 2.46546194e-01 4.73191701e-02 -1.24150597e-01 1.37897050e+00 8.89630318e-01 6.07733727e-01 2.04230007e-03 -1.19978683e-02 1.11910880e+00 -1.27326918e+00 -4.78102803e-01 -3.53346944e-01 4.55328673e-01 -7.97242999e-01 1.58105600e+00 -9.49924998e-03 -9.36862707e-01 -6.46747589e-01 -1.13612962e+00 -1.95001602e-01 -2.19553128e-01 2.05298066e-01 5.60234427e-01 4.12347317e-01 -1.22793031e+00 2.67878860e-01 -4.61643636e-01 -4.36445773e-01 3.69032383e-01 3.91402692e-01 -2.61427939e-01 -8.66738260e-02 -1.11571717e+00 8.15564394e-01 6.19325936e-01 1.34641692e-01 -7.93807328e-01 -6.77791655e-01 -9.13593233e-01 1.82073534e-01 5.17222166e-01 -6.72743797e-01 1.17625642e+00 -1.54272139e+00 -1.67754328e+00 6.86277926e-01 7.42363259e-02 -8.37227404e-02 4.02072400e-01 -1.89265117e-01 -2.31330767e-01 3.88423800e-02 1.06897980e-01 1.05292845e+00 1.10870349e+00 -1.35555601e+00 -5.51384330e-01 2.70572845e-02 4.05440599e-01 4.55717921e-01 -3.18310976e-01 -1.17233850e-01 -6.93421304e-01 -1.04098856e+00 -1.47247594e-02 -9.38707590e-01 -3.90796036e-01 1.45798773e-02 -5.60455084e-01 1.87250346e-01 7.68968701e-01 -2.97733784e-01 1.06456470e+00 -2.25427866e+00 1.59032911e-01 -6.28240034e-02 3.79612654e-01 4.96311605e-01 -4.65093732e-01 9.00281519e-02 -3.72482836e-01 1.37358069e-01 -1.70074597e-01 1.54114380e-01 -3.65717053e-01 -9.94094536e-02 -4.26909119e-01 3.83216977e-01 7.90453330e-02 1.31457710e+00 -9.71152306e-01 -4.03985679e-01 3.38766396e-01 4.26256895e-01 -5.84080398e-01 2.80649662e-01 -9.79477987e-02 5.93788981e-01 -3.11150551e-01 3.40004236e-01 7.45018840e-01 -3.49781573e-01 -6.08411692e-02 -5.62868476e-01 -1.80742726e-01 9.05649550e-03 -8.03880930e-01 1.71679878e+00 -4.02590215e-01 6.94555640e-01 -3.79536092e-01 -8.37668777e-01 7.83752799e-01 3.56037822e-03 2.08203882e-01 -1.14411747e+00 2.54768848e-01 3.18412818e-02 -3.80942449e-02 -3.70816886e-01 6.17075801e-01 -1.99897271e-02 9.05140787e-02 5.47382176e-01 -2.51412868e-01 8.22617114e-02 -1.79855004e-01 2.09801525e-01 7.50928581e-01 1.07787013e-01 3.73244971e-01 -2.27010936e-01 3.96493316e-01 -1.49526909e-01 5.34357846e-01 5.58291495e-01 -1.51870340e-01 1.07247090e+00 1.64373457e-01 -5.77560663e-01 -8.84144425e-01 -8.61196458e-01 8.32949132e-02 1.20328593e+00 7.81875968e-01 -6.97988123e-02 -5.94321132e-01 -6.91874743e-01 -3.93032551e-01 5.54312110e-01 -6.57433331e-01 -5.48123777e-01 -4.82388854e-01 -7.87807882e-01 1.84081092e-01 5.35761118e-01 7.92634070e-01 -1.25765884e+00 -6.74737811e-01 2.24263426e-02 -2.62925118e-01 -7.68111348e-01 -1.07529783e+00 -1.17982946e-01 -6.32673562e-01 -9.10824418e-01 -1.03257442e+00 -9.38124239e-01 1.00762630e+00 9.47379410e-01 7.32781112e-01 2.39657149e-01 -2.21495017e-01 1.82171557e-02 -4.81863439e-01 -9.57075432e-02 -3.94282863e-02 1.94231540e-01 -8.05597156e-02 5.82220219e-02 -7.43864523e-03 -5.25429726e-01 -1.00652516e+00 6.73172534e-01 -1.21154368e+00 5.13187945e-01 8.58181596e-01 8.68339598e-01 4.78907049e-01 3.94839719e-02 6.75253272e-01 -5.79440296e-01 5.54437101e-01 -1.85831830e-01 -5.40884256e-01 2.30914325e-01 -5.76835155e-01 1.79670200e-01 5.16118050e-01 -7.16713965e-01 -9.46929395e-01 6.70819804e-02 5.84794208e-02 -5.66951513e-01 1.24651738e-01 4.14551973e-01 -3.59194279e-01 -3.90272617e-01 8.39691758e-01 2.83022642e-01 -2.38048747e-01 -3.69545192e-01 6.73277974e-01 4.83701795e-01 5.97736955e-01 -2.80242205e-01 8.77836466e-01 3.06468934e-01 -3.71390879e-01 -4.05049384e-01 -8.36429536e-01 -8.76277015e-02 -5.12352407e-01 -8.77098218e-02 6.84461057e-01 -8.75994861e-01 -5.14193118e-01 4.14411008e-01 -1.04746842e+00 -5.22156775e-01 -2.74957687e-01 3.59085172e-01 -3.46825272e-01 3.84451121e-01 -2.45314404e-01 -1.32397056e-01 -3.15546751e-01 -1.25823045e+00 8.51789176e-01 5.69677293e-01 4.88454811e-02 -7.25718856e-01 1.76755525e-02 3.19987312e-02 5.89734972e-01 -3.52196470e-02 7.78761029e-01 -2.34429225e-01 -7.30513334e-01 1.29937427e-02 -7.37728119e-01 1.60904571e-01 5.67980886e-01 -2.44384050e-01 -9.38470364e-01 -4.61175948e-01 -2.50774086e-01 -1.64046481e-01 9.43621516e-01 4.47861344e-01 1.33846092e+00 -5.37562370e-01 -3.46727252e-01 7.92182744e-01 1.32024550e+00 2.26120993e-01 9.85304594e-01 5.09229183e-01 1.06300116e+00 2.90683061e-01 5.52531540e-01 1.63661122e-01 5.10426104e-01 8.46766174e-01 5.77792645e-01 -6.49131238e-01 -3.91805172e-01 -3.86982083e-01 2.63849854e-01 5.00213683e-01 4.29773964e-02 -3.17713708e-01 -6.62252665e-01 6.66062653e-01 -1.86536324e+00 -7.34338880e-01 2.59828180e-01 2.38126230e+00 1.00698733e+00 -5.32971062e-02 -3.15624080e-03 -6.67145848e-02 9.86940503e-01 1.40266731e-01 -7.45718241e-01 -2.68300891e-01 -5.43312617e-02 -1.38036311e-01 5.06505132e-01 5.04433393e-01 -1.12708831e+00 1.20056665e+00 5.82690239e+00 8.14671993e-01 -1.64580011e+00 5.86217344e-02 6.96417511e-01 -1.04260676e-01 -4.54294831e-01 3.87404561e-02 -4.93240774e-01 4.65488702e-01 3.61753851e-01 -3.10215592e-01 5.23727059e-01 6.97963297e-01 2.02696145e-01 -5.85056208e-02 -8.01588416e-01 1.18076074e+00 1.84398457e-01 -1.26008964e+00 2.58734077e-01 -1.57507539e-01 8.63804281e-01 -1.58069015e-01 3.15618098e-01 1.66951030e-01 1.97058141e-01 -1.00701010e+00 7.57542014e-01 4.13940161e-01 9.83073592e-01 -7.33042598e-01 5.67563951e-01 -6.59181923e-03 -1.03955674e+00 -1.39958542e-02 -5.12315392e-01 -1.83425620e-02 -7.33065084e-02 6.29996300e-01 -5.89875340e-01 4.17398244e-01 7.71874845e-01 7.25520313e-01 -8.39704037e-01 1.23778570e+00 -4.21627134e-01 2.29791343e-01 -4.61486503e-02 2.82595903e-01 6.43938035e-03 8.19419995e-02 5.45216262e-01 8.84935081e-01 2.75705278e-01 1.53753823e-02 1.10656381e-01 8.71822119e-01 -2.28359327e-01 1.66268006e-01 -5.94567180e-01 1.22104160e-01 2.90459305e-01 1.35680020e+00 -7.66931295e-01 -1.10442586e-01 -3.60939324e-01 1.34997499e+00 3.41033906e-01 7.18098640e-01 -9.03883338e-01 -6.45850599e-01 4.81350452e-01 1.22110575e-01 4.80927706e-01 7.37803876e-02 -3.03817958e-01 -1.30359209e+00 -4.20991741e-02 -8.65309477e-01 1.33786768e-01 -1.06476843e+00 -9.66054380e-01 8.71168375e-01 -1.92892566e-01 -1.37166369e+00 1.16168223e-01 -2.51359731e-01 -6.26826406e-01 7.75628507e-01 -1.57463884e+00 -1.22683549e+00 -6.02558196e-01 6.42604470e-01 5.08265793e-01 1.69604402e-02 3.84443909e-01 2.59278566e-01 -6.84819043e-01 8.39252830e-01 -2.70989716e-01 -5.40259741e-02 9.27961230e-01 -1.01276350e+00 4.18058187e-01 1.03589177e+00 5.13034388e-02 4.86599475e-01 7.11239636e-01 -4.71799970e-01 -1.08157337e+00 -1.37351704e+00 4.43263173e-01 -3.81375998e-01 3.12790811e-01 -1.61612093e-01 -9.75319743e-01 7.19660759e-01 3.88551503e-01 1.02898329e-01 2.80177087e-01 -2.50993729e-01 -3.85432690e-01 -1.43485978e-01 -9.36231673e-01 1.02083290e+00 9.47758555e-01 -2.83685595e-01 -4.73379314e-01 1.14357077e-01 1.16160512e+00 -4.21248198e-01 -3.89167994e-01 3.93247187e-01 4.38572109e-01 -8.34494948e-01 1.01970685e+00 -2.62704372e-01 3.74761552e-01 -7.28928983e-01 4.88420911e-02 -1.41110837e+00 -6.83227479e-01 -6.74771786e-01 2.01863244e-01 1.12918460e+00 3.39609027e-01 -6.95287943e-01 4.43654925e-01 5.50371587e-01 -1.16544552e-01 -6.19254887e-01 -5.58373570e-01 -6.56139612e-01 -9.29607004e-02 -5.90845309e-02 7.68827558e-01 9.85337138e-01 -1.08373135e-01 5.39126277e-01 -5.81656158e-01 1.97486967e-01 3.72862339e-01 3.18563670e-01 7.75504708e-01 -7.80990005e-01 -1.68362409e-01 -4.78659570e-01 -4.62749183e-01 -1.14364791e+00 -8.78379419e-02 -7.65139461e-01 1.39168426e-01 -1.49223542e+00 4.19312358e-01 -4.68074381e-01 -4.45648104e-01 9.47193682e-01 -5.36739528e-01 6.41903937e-01 2.29583636e-01 3.83374691e-01 -5.19256055e-01 9.24857914e-01 1.61958182e+00 -2.60650754e-01 -5.40986955e-01 -2.83278972e-01 -1.28794086e+00 5.40101826e-01 9.87874687e-01 -3.02842945e-01 -6.42412305e-01 -6.11082196e-01 2.81423599e-01 -3.37326974e-01 5.84144354e-01 -8.62717927e-01 1.67950988e-01 -3.48800540e-01 4.24194038e-01 -2.51890987e-01 -5.59779536e-03 -5.92801332e-01 -8.27433821e-03 2.68757999e-01 -3.24570030e-01 9.19403508e-02 4.19171989e-01 4.45684642e-01 -9.08980444e-02 2.48294249e-02 1.08283913e+00 5.38554452e-02 -1.01165950e+00 3.17719489e-01 -1.69661656e-01 -9.66731682e-02 1.08889043e+00 -3.03066313e-01 -4.42367375e-01 -4.08088207e-01 -4.11753088e-01 6.24434054e-02 7.75786936e-01 6.94715679e-01 7.98376441e-01 -1.46922910e+00 -5.26160717e-01 2.43637338e-01 2.51171559e-01 -1.04947388e-01 3.89169723e-01 7.60935545e-01 -4.25261050e-01 3.51370186e-01 -3.94260794e-01 -5.06950617e-01 -9.02393937e-01 1.06740797e+00 4.56952244e-01 8.99814069e-02 -8.41654301e-01 7.70333946e-01 1.01765943e+00 -2.91894644e-01 -6.21979907e-02 -2.79183149e-01 -2.98830926e-01 -2.95120627e-01 7.31777191e-01 -4.74644303e-02 -5.68334423e-02 -5.69644332e-01 -2.36283690e-01 7.94709682e-01 -3.92538190e-01 5.54974191e-02 9.43869114e-01 -4.99313533e-01 6.51680455e-02 6.78773178e-03 1.10135376e+00 -1.77195460e-01 -1.45983243e+00 -3.83963943e-01 -4.82003182e-01 -7.21780419e-01 2.74592757e-01 -7.96909571e-01 -1.49674630e+00 7.22230017e-01 8.52000594e-01 -1.36568725e-01 1.56548238e+00 -8.47238153e-02 7.18054116e-01 1.62003815e-01 3.08332980e-01 -6.34073973e-01 4.50731248e-01 2.49597505e-01 1.20635188e+00 -1.33053195e+00 -4.73575443e-02 -1.91660941e-01 -8.46257210e-01 7.02311516e-01 9.28534031e-01 -3.28488485e-03 4.08998013e-01 -1.80913687e-01 2.06982478e-01 -1.21569961e-01 -4.35580105e-01 -2.66737700e-01 4.22587425e-01 6.20065153e-01 1.40310988e-01 -7.50166923e-02 -2.24787861e-01 4.19861257e-01 -9.33291540e-02 -1.83187053e-02 5.92983186e-01 6.70886338e-01 -5.29832602e-01 -9.02997434e-01 -3.61393660e-01 1.69852182e-01 -2.12769493e-01 -2.30526075e-01 -2.90997118e-01 6.37849450e-01 -4.40770909e-02 8.35938275e-01 -7.51286447e-02 -6.00859523e-01 2.76457429e-01 -7.01157689e-01 3.51687372e-01 -6.69523358e-01 -2.56795377e-01 4.06825505e-02 -5.17108858e-01 -6.95162714e-01 -2.93905705e-01 -3.82733718e-02 -1.14578545e+00 -1.98385835e-01 -3.92252773e-01 -1.09536171e-01 3.92390102e-01 7.44177580e-01 6.21695876e-01 6.25985861e-01 7.42494047e-01 -1.13614452e+00 3.67665850e-02 -8.42382789e-01 -3.64733368e-01 2.46823967e-01 4.68738437e-01 -7.57544398e-01 -1.57727346e-01 1.50204703e-01]
[11.20089340209961, -0.9535713195800781]
2923a88f-6216-4581-ae8e-a4e5eeaf39e3
multi-epoch-learning-for-deep-click-through
2305.19531
null
https://arxiv.org/abs/2305.19531v1
https://arxiv.org/pdf/2305.19531v1.pdf
Multi-Epoch Learning for Deep Click-Through Rate Prediction Models
The one-epoch overfitting phenomenon has been widely observed in industrial Click-Through Rate (CTR) applications, where the model performance experiences a significant degradation at the beginning of the second epoch. Recent advances try to understand the underlying factors behind this phenomenon through extensive experiments. However, it is still unknown whether a multi-epoch training paradigm could achieve better results, as the best performance is usually achieved by one-epoch training. In this paper, we hypothesize that the emergence of this phenomenon may be attributed to the susceptibility of the embedding layer to overfitting, which can stem from the high-dimensional sparsity of data. To maintain feature sparsity while simultaneously avoiding overfitting of embeddings, we propose a novel Multi-Epoch learning with Data Augmentation (MEDA), which can be directly applied to most deep CTR models. MEDA achieves data augmentation by reinitializing the embedding layer in each epoch, thereby avoiding embedding overfitting and simultaneously improving convergence. To our best knowledge, MEDA is the first multi-epoch training paradigm designed for deep CTR prediction models. We conduct extensive experiments on several public datasets, and the effectiveness of our proposed MEDA is fully verified. Notably, the results show that MEDA can significantly outperform the conventional one-epoch training. Besides, MEDA has exhibited significant benefits in a real-world scene on Kuaishou.
['Han Li', 'Dongying Kong', 'Jian Liang', 'Zhongxiang Fan', 'Zhaocheng Liu']
2023-05-31
null
null
null
null
['click-through-rate-prediction']
['miscellaneous']
[-3.41715477e-02 -4.36331391e-01 -2.80759126e-01 -1.80450846e-02 -4.01175082e-01 9.89170372e-03 3.86405855e-01 1.07432283e-01 -2.95531422e-01 3.13131154e-01 2.05850210e-02 -6.21422112e-01 -2.54245102e-01 -5.84616840e-01 -7.30187416e-01 -7.68529117e-01 2.21911003e-03 -1.72017328e-02 1.75792038e-01 -3.35503787e-01 1.48022830e-01 1.40180349e-01 -1.32256222e+00 1.57676011e-01 9.76690650e-01 1.15341747e+00 2.87020087e-01 2.65448004e-01 -8.25253502e-02 7.28340447e-01 -4.47668999e-01 -4.72890645e-01 4.18030709e-01 -8.96988995e-03 -3.86099815e-01 2.98544299e-03 3.99212688e-02 -4.87901539e-01 -6.75722599e-01 7.42161274e-01 2.97855675e-01 8.63988250e-02 2.97156692e-01 -1.17504072e+00 -6.72590673e-01 5.45203924e-01 -7.69875109e-01 2.28177562e-01 -5.39033674e-02 1.98956415e-01 1.43568718e+00 -1.12297630e+00 1.89350516e-01 9.06329691e-01 6.98095262e-01 1.72046572e-01 -1.13946915e+00 -8.25176418e-01 3.64183873e-01 1.95448384e-01 -1.42574549e+00 -5.35512492e-02 9.27253366e-01 -3.79129052e-01 5.78420222e-01 1.14064766e-02 5.61983347e-01 1.41814482e+00 3.48427624e-01 1.15312684e+00 9.31611061e-01 -2.16673419e-01 5.51533811e-02 2.50630409e-01 1.98497579e-01 4.15924251e-01 2.44915485e-01 1.53495714e-01 -3.27590376e-01 -1.19940706e-01 8.63377571e-01 3.43480408e-01 5.06838001e-02 -1.93172172e-01 -1.05540276e+00 9.75779653e-01 7.12200582e-01 3.96119118e-01 -4.48891133e-01 -1.63845960e-02 4.95906115e-01 3.74181479e-01 4.59867626e-01 4.21346515e-01 -5.86024582e-01 -2.13739142e-01 -5.81773758e-01 4.04844619e-02 3.52696180e-01 8.92811716e-01 5.30088902e-01 1.85377419e-01 -1.38322920e-01 1.00602436e+00 2.19315320e-01 1.37043566e-01 8.25416267e-01 -3.65076751e-01 5.91361046e-01 7.36113369e-01 -1.07931169e-02 -1.03856635e+00 -2.08816186e-01 -1.13964283e+00 -9.04997826e-01 -2.86230236e-01 3.57675642e-01 -1.12982288e-01 -5.99499404e-01 1.54512489e+00 2.03165412e-01 3.99311632e-01 -7.11039603e-02 7.94980049e-01 3.79209697e-01 6.33521318e-01 1.74375549e-01 -5.53910285e-02 1.18292499e+00 -1.13030148e+00 -8.38079751e-01 -2.46355921e-01 1.01253033e+00 -8.17455173e-01 1.42760146e+00 5.47250152e-01 -4.29325879e-01 -8.38915467e-01 -1.15448236e+00 1.40143409e-01 -3.46483976e-01 5.81433296e-01 9.79190171e-01 4.16913986e-01 -2.79331744e-01 5.89696944e-01 -7.06615806e-01 -1.33861527e-01 3.65871340e-01 2.18865260e-01 -1.27403885e-01 -2.48059452e-01 -1.28957486e+00 4.36033338e-01 3.11033726e-01 4.45890933e-01 -4.60269898e-01 -7.24483550e-01 -3.24508518e-01 1.18934549e-01 5.91892362e-01 -1.95535347e-01 1.23217475e+00 -7.40506887e-01 -1.35901988e+00 3.44546214e-02 1.38780177e-02 -5.57556987e-01 3.30277920e-01 -6.17294669e-01 -7.02221334e-01 -3.87177020e-01 -2.38238752e-01 1.30312279e-01 9.45869863e-01 -1.17795122e+00 -4.55384135e-01 -2.51674414e-01 7.67922327e-02 -1.38087541e-01 -1.11220253e+00 -4.27712858e-01 -4.87071574e-01 -9.18277919e-01 -5.43452799e-02 -1.03558826e+00 -3.22746664e-01 -2.15024322e-01 -3.74547958e-01 -3.12715918e-01 9.02997971e-01 -4.32061106e-01 1.86569846e+00 -2.39660096e+00 -1.97999999e-01 3.48502137e-02 3.09412241e-01 5.75797617e-01 -3.67970198e-01 6.09609962e-01 -1.41431287e-01 1.05801292e-01 2.43651494e-01 -4.04909432e-01 -2.76450673e-03 2.48976499e-01 -6.31778240e-01 2.65246540e-01 1.44626677e-01 8.61161888e-01 -6.33617401e-01 -3.02074283e-01 6.69619963e-02 4.03248370e-01 -7.06237137e-01 2.31075600e-01 -2.48483896e-01 3.10446113e-01 -7.01657653e-01 6.11217499e-01 7.40957975e-01 -7.20238447e-01 1.65023431e-01 -3.56116414e-01 -2.66117305e-01 1.91919759e-01 -9.70825672e-01 1.41225457e+00 -5.39173067e-01 5.68517864e-01 -4.18380469e-01 -1.03860140e+00 1.00529599e+00 1.76899016e-01 7.96995759e-01 -1.09203136e+00 2.96703011e-01 3.45273435e-01 2.49062121e-01 -4.17056888e-01 7.09844410e-01 8.58285278e-02 1.33910999e-01 3.24938387e-01 -4.38826889e-01 7.31304884e-01 -5.33180078e-03 1.93635941e-01 1.06384552e+00 -3.28924119e-01 -1.66288257e-01 1.09635919e-01 3.83431554e-01 -1.40627772e-01 4.69927281e-01 7.26399481e-01 -4.64369245e-02 3.01313072e-01 3.46288323e-01 -5.16088128e-01 -1.07593453e+00 -8.54258358e-01 -2.25371659e-01 1.02015877e+00 2.06944898e-01 -6.44298971e-01 -1.99823335e-01 -8.39493573e-01 1.06541611e-01 5.79989612e-01 -6.26368761e-01 -4.76785064e-01 -5.83564341e-01 -9.16409373e-01 2.85944045e-01 6.85082853e-01 5.88923573e-01 -7.11484075e-01 -5.93614653e-02 3.77721995e-01 -1.34919479e-01 -1.38455844e+00 -3.59966606e-01 1.69376105e-01 -1.15643013e+00 -8.09544086e-01 -5.60924292e-01 -5.72523117e-01 6.23129249e-01 5.56263626e-01 6.28601313e-01 2.95773357e-01 -8.73923153e-02 -1.12424649e-01 -5.77512622e-01 -3.36201817e-01 -1.66644957e-02 5.59237897e-01 1.22677244e-01 2.52373129e-01 3.61151576e-01 -3.86010140e-01 -8.39562893e-01 6.05719447e-01 -9.64806318e-01 -1.55113548e-01 1.11207783e+00 9.57831144e-01 3.78686905e-01 3.82892132e-01 6.26489043e-01 -8.37712049e-01 7.57572412e-01 -7.96411037e-01 -4.57779437e-01 2.72933356e-02 -9.94033456e-01 -1.45009700e-02 1.02452374e+00 -8.03495049e-01 -8.81628990e-01 -2.23087162e-01 -2.61671126e-01 -6.91627920e-01 3.28708827e-01 8.97858024e-01 2.75906384e-01 1.03640221e-01 3.85515898e-01 2.71855235e-01 -5.84993660e-02 -8.26327205e-01 2.48856008e-01 7.08177447e-01 2.84850784e-02 -5.10583162e-01 1.21476150e+00 3.23000759e-01 -2.64145404e-01 -7.23497331e-01 -1.19568574e+00 -6.18799388e-01 -3.60341907e-01 -1.49620309e-01 4.19917256e-01 -9.47392166e-01 -6.50826514e-01 3.67900312e-01 -8.28277528e-01 -1.73262909e-01 -1.66470319e-01 7.02735484e-01 -4.86075617e-02 3.85532320e-01 -5.20354986e-01 -6.45326376e-01 -1.68469474e-01 -1.04706764e+00 6.46206319e-01 1.94758072e-01 6.87988847e-02 -1.02839267e+00 6.64425120e-02 3.60582590e-01 5.87325811e-01 -2.79742956e-01 1.00948083e+00 -8.84572506e-01 -6.17597818e-01 -5.09099424e-01 -2.83133209e-01 5.84149599e-01 2.10212901e-01 -6.62719011e-02 -8.26552093e-01 -6.62240744e-01 -1.58678189e-01 -4.03293908e-01 6.02738202e-01 1.71101950e-02 1.55207443e+00 -3.17988485e-01 -3.25257421e-01 3.04487109e-01 1.40034628e+00 2.03733504e-01 6.15152895e-01 5.92674971e-01 6.14360452e-01 5.52236959e-02 1.04908514e+00 6.52248263e-01 1.91390261e-01 8.32839727e-01 5.57328582e-01 -3.92753929e-01 1.93463698e-01 -5.53192139e-01 2.52194822e-01 1.21455348e+00 4.26302366e-02 -2.91264474e-01 -5.64790666e-01 4.35986727e-01 -1.82783616e+00 -6.10974371e-01 -2.16900751e-01 2.20273256e+00 5.55813551e-01 5.23300409e-01 1.63095057e-01 4.26902801e-01 2.95901835e-01 2.43879363e-01 -6.17005646e-01 -2.23692805e-01 3.13208550e-02 1.18827380e-01 4.59098428e-01 -4.52879593e-02 -9.77532268e-01 7.32208014e-01 5.89219856e+00 1.02836597e+00 -1.35786700e+00 1.51733711e-01 4.58491564e-01 6.16929643e-02 -1.39794469e-01 1.91531349e-02 -8.90747726e-01 4.61619854e-01 8.72923255e-01 -1.36720732e-01 2.34157443e-01 1.14007020e+00 3.26393634e-01 4.81300622e-01 -9.32894826e-01 8.50700021e-01 -2.18366340e-01 -1.17279708e+00 6.54090270e-02 4.88295943e-01 7.80700207e-01 1.16006341e-02 5.19723654e-01 9.52991188e-01 1.00299483e-02 -7.70821154e-01 3.49742472e-01 2.99280465e-01 4.46966380e-01 -6.19866967e-01 7.94055820e-01 3.45574111e-01 -1.32636034e+00 -5.88109672e-01 -3.88665646e-01 7.66439214e-02 -1.14570901e-01 8.51721168e-01 -8.01777303e-01 7.37239540e-01 5.47814131e-01 8.98048759e-01 -7.34925926e-01 1.21710694e+00 9.12318528e-02 9.17996526e-01 -1.98551714e-01 -1.12251624e-01 3.76433581e-01 -8.71395692e-02 3.79005432e-01 8.34265113e-01 3.82356107e-01 -1.47355348e-01 1.05790474e-01 7.24259377e-01 -1.17714494e-01 2.87290007e-01 -5.08982360e-01 -4.01552230e-01 4.83001828e-01 1.12731957e+00 -1.84934378e-01 -1.21403672e-01 -7.32191741e-01 8.23045492e-01 2.39371091e-01 4.39466119e-01 -1.25467980e+00 -3.97548854e-01 6.55735314e-01 3.56578887e-01 5.70395470e-01 -4.28323984e-01 -2.80003637e-01 -1.17127550e+00 2.52534062e-01 -8.24019432e-01 2.85981357e-01 -3.53181928e-01 -1.40251482e+00 5.40299356e-01 -3.65689158e-01 -1.72978497e+00 2.02684253e-01 -5.90866625e-01 -5.72511852e-01 4.28171277e-01 -1.68662131e+00 -8.40722322e-01 -3.09075415e-01 5.59804499e-01 6.43959761e-01 -1.83730394e-01 4.59973931e-01 7.57469893e-01 -1.11893904e+00 1.04084361e+00 4.12857234e-01 2.43451633e-02 6.02041185e-01 -8.93799365e-01 9.10904258e-02 6.44488513e-01 8.23328346e-02 1.01015925e+00 6.14084244e-01 -4.37033415e-01 -1.57081366e+00 -1.18377793e+00 6.89414799e-01 -1.43483311e-01 1.04342973e+00 -4.36088294e-01 -1.14863217e+00 6.52261853e-01 -5.62439598e-02 7.11432938e-03 6.69849873e-01 4.58206892e-01 -4.34108764e-01 -4.20873344e-01 -5.58034599e-01 6.63220584e-01 8.21621478e-01 -4.80569899e-01 -1.72797471e-01 4.38290209e-01 7.99424827e-01 -2.00481266e-01 -1.20158446e+00 5.61478257e-01 6.02527559e-01 -6.63322031e-01 9.50477660e-01 -5.97727537e-01 6.56059682e-01 -1.20618366e-01 -2.23839283e-01 -9.66635048e-01 -4.54792351e-01 -5.48509121e-01 -4.29031610e-01 1.08553755e+00 4.54329908e-01 -7.58740008e-01 8.40276480e-01 2.48547167e-01 -4.73663621e-02 -1.17984998e+00 -7.70054102e-01 -1.10463512e+00 1.48028880e-01 -4.89292502e-01 5.46965420e-01 9.49787736e-01 -2.99156308e-01 3.75080377e-01 -6.77086055e-01 1.89151898e-01 4.87083554e-01 6.57408983e-02 1.06693637e+00 -1.20568490e+00 -3.84136796e-01 -2.49071866e-01 -3.62144969e-02 -1.63409579e+00 -1.95328996e-01 -6.52236283e-01 -2.60041773e-01 -1.19001770e+00 5.69962412e-02 -8.58781934e-01 -7.29131997e-01 3.99683744e-01 -2.42679000e-01 -5.48006631e-02 1.73771694e-01 4.96296853e-01 -5.28130054e-01 1.05917108e+00 1.42931378e+00 1.58864073e-02 -2.45407864e-01 1.20933630e-01 -8.12265635e-01 4.52321529e-01 9.00375068e-01 -4.18506354e-01 -5.94174087e-01 -5.13278663e-01 2.90853560e-01 -2.52303571e-01 1.69254467e-01 -1.00239027e+00 1.76657647e-01 -5.04924767e-02 9.90459546e-02 -7.04124749e-01 4.01446879e-01 -1.09996140e+00 -6.80929273e-02 4.35714722e-01 -2.57849693e-01 1.06762126e-01 3.01578999e-01 9.12467778e-01 -2.50776321e-01 8.49535018e-02 5.54840922e-01 4.24383253e-01 -5.09107113e-01 4.38761294e-01 -3.24048460e-01 -2.42179245e-01 9.50505435e-01 -1.36442423e-01 -2.89948016e-01 -3.38999443e-02 -3.73520166e-01 2.51730174e-01 6.84809834e-02 8.08554888e-01 6.08930886e-01 -1.52438128e+00 -3.95032585e-01 3.26934993e-01 2.85287201e-01 -3.24365675e-01 4.44599956e-01 1.11098099e+00 -2.61749551e-02 5.41458666e-01 3.20895374e-01 -6.26243055e-01 -9.85557020e-01 6.52500510e-01 -6.98854923e-02 -6.52508140e-01 -7.15618014e-01 4.87158656e-01 1.71272784e-01 -2.95231104e-01 3.83409858e-01 -2.96252549e-01 -7.46552721e-02 -1.37276068e-01 2.38487467e-01 2.62502193e-01 1.93770409e-01 -9.92457345e-02 -8.25972259e-02 4.54667747e-01 -7.42498577e-01 4.41062480e-01 1.29848742e+00 -5.00633791e-02 4.42180723e-01 4.65492696e-01 1.44153190e+00 -4.66071181e-02 -1.06457853e+00 -5.35737872e-01 -2.40150034e-01 -6.99720621e-01 1.09160967e-01 -5.52633941e-01 -1.43652725e+00 7.92274892e-01 5.38901985e-01 2.38273144e-01 1.08948302e+00 -3.47198635e-01 1.38891900e+00 5.08251727e-01 1.89915806e-01 -1.31100190e+00 7.88400769e-01 4.07160968e-01 7.52750397e-01 -1.39023566e+00 1.20791290e-02 -1.83864743e-01 -6.70847952e-01 9.67281997e-01 7.96526432e-01 -6.44674450e-02 9.07459676e-01 -1.02824770e-01 -2.03908145e-01 -1.85397878e-01 -9.66232836e-01 6.76803887e-02 7.58771822e-02 -5.47512174e-02 3.59677017e-01 -5.36850467e-02 -4.31339532e-01 7.84376502e-01 -1.48288161e-02 5.80312647e-02 2.16990352e-01 1.01299024e+00 -2.13823304e-01 -1.26706040e+00 1.59386303e-02 5.26836693e-01 -4.09026712e-01 -1.12989791e-01 -1.08383715e-01 1.02605474e+00 -1.18443593e-02 9.82037783e-01 -1.90987155e-01 -1.05692506e+00 5.47984123e-01 -2.53091276e-01 -4.73797321e-02 -2.86378652e-01 -4.56642300e-01 1.94253623e-01 -1.31837502e-01 -4.98758048e-01 2.78648678e-02 -5.53827047e-01 -1.03168809e+00 -4.17059720e-01 -7.31651306e-01 1.55470267e-01 6.01691484e-01 7.91023314e-01 4.80566919e-01 8.07673037e-01 1.19085038e+00 -1.88228086e-01 -9.91241217e-01 -1.00386381e+00 -5.12678802e-01 3.75947148e-01 2.65153021e-01 -8.41912508e-01 -6.94136083e-01 -3.69540423e-01]
[10.128704071044922, 5.484592914581299]
4c090674-4cc9-4bd0-ba08-64bef6d21d73
out-of-vocabulary-challenge-report
2209.06717
null
https://arxiv.org/abs/2209.06717v1
https://arxiv.org/pdf/2209.06717v1.pdf
Out-of-Vocabulary Challenge Report
This paper presents final results of the Out-Of-Vocabulary 2022 (OOV) challenge. The OOV contest introduces an important aspect that is not commonly studied by Optical Character Recognition (OCR) models, namely, the recognition of unseen scene text instances at training time. The competition compiles a collection of public scene text datasets comprising of 326,385 images with 4,864,405 scene text instances, thus covering a wide range of data distributions. A new and independent validation and test set is formed with scene text instances that are out of vocabulary at training time. The competition was structured in two tasks, end-to-end and cropped scene text recognition respectively. A thorough analysis of results from baselines and different participants is presented. Interestingly, current state-of-the-art models show a significant performance gap under the newly studied setting. We conclude that the OOV dataset proposed in this challenge will be an essential area to be explored in order to develop scene text models that achieve more robust and generalized predictions.
['Dimosthenis Karatzas', 'Ron Litman', 'Shai Mazor', 'Aviad Aberdam', 'Oren Nuriel', 'Ali Furkan Biten', 'Andrés Mafla', 'Sergi Garcia-Bordils']
2022-09-14
null
null
null
null
['scene-text-recognition']
['computer-vision']
[ 7.19536901e-01 -2.18812287e-01 -2.16248911e-02 -5.52489996e-01 -6.64213121e-01 -6.85211241e-01 1.02494621e+00 2.54604042e-01 -4.94390398e-01 2.86982685e-01 2.11147502e-01 -1.08764760e-01 4.48437303e-01 -3.91087919e-01 -8.76954019e-01 -4.52418834e-01 4.15593117e-01 8.23598862e-01 5.99624336e-01 9.74361226e-03 6.41559958e-01 3.10249686e-01 -1.60889804e+00 7.85536051e-01 6.91943705e-01 9.53572690e-01 4.43595529e-01 1.04252684e+00 -3.14569980e-01 1.12012649e+00 -8.50717425e-01 -5.93568265e-01 2.16062859e-01 -1.19751796e-01 -7.63252437e-01 5.33188641e-01 1.51187325e+00 -2.73542076e-01 -5.93708217e-01 8.20444643e-01 5.31623065e-01 2.32833833e-01 1.00589931e+00 -9.97384667e-01 -9.27589536e-01 4.08099651e-01 -3.28211606e-01 3.19484949e-01 2.54953593e-01 3.69499922e-02 8.78805757e-01 -1.11789215e+00 8.99244189e-01 9.93937433e-01 3.39609921e-01 5.28832257e-01 -1.06333232e+00 -2.22886711e-01 3.69209111e-01 3.57364386e-01 -1.35967517e+00 -4.50586468e-01 5.68102777e-01 -6.87217414e-01 1.25362945e+00 5.68682969e-01 3.40176105e-01 1.51002717e+00 8.56024325e-02 1.28662026e+00 1.13204372e+00 -5.74155807e-01 1.20684855e-01 4.10353094e-01 4.01373267e-01 4.20841157e-01 3.83688986e-01 -2.85552174e-01 -6.91755533e-01 2.87691385e-01 1.72659695e-01 -3.78846586e-01 -4.48816597e-01 -6.05045080e-01 -1.31888759e+00 6.74945712e-01 -3.01373675e-02 1.14948325e-01 1.30931914e-01 2.65013445e-02 5.82358181e-01 1.40165806e-01 4.25111026e-01 2.63543040e-01 -2.03259423e-01 -1.75452501e-01 -1.09745729e+00 3.35539401e-01 7.83148646e-01 1.28561103e+00 4.28689390e-01 3.71539086e-01 -6.24859154e-01 1.15277076e+00 3.62785697e-01 8.57882082e-01 5.81681132e-01 -2.94039577e-01 8.66713226e-01 5.42896152e-01 -8.79043490e-02 -7.13198125e-01 -1.35938957e-01 -1.92775088e-03 -5.59436381e-01 -6.38259500e-02 3.42052221e-01 2.65476704e-01 -1.55920613e+00 8.72106135e-01 4.31980891e-03 2.26972938e-01 2.10802302e-01 8.21689844e-01 1.31940830e+00 8.59056950e-01 -1.79367978e-02 4.15521562e-01 1.24754989e+00 -1.54485130e+00 -6.52966559e-01 -6.45595074e-01 6.76913202e-01 -1.21089041e+00 1.05779815e+00 5.43469191e-01 -6.86271608e-01 -4.87574786e-01 -1.15881050e+00 -3.53901148e-01 -7.67036855e-01 5.35209060e-01 3.28582108e-01 7.57431388e-01 -1.08060968e+00 1.76472202e-01 -3.40847582e-01 -7.65464187e-01 4.30836976e-01 6.83288500e-02 -3.61811697e-01 -4.19895470e-01 -5.98245323e-01 8.85241151e-01 5.39400518e-01 -3.05704847e-02 -1.21021402e+00 -3.78068149e-01 -9.34813559e-01 -3.72242779e-01 5.04725218e-01 -4.75528091e-02 1.24202955e+00 -9.06411827e-01 -1.26047051e+00 1.32584524e+00 -1.59693718e-01 -6.55881643e-01 1.05819118e+00 -4.53067929e-01 -5.08695602e-01 1.73419729e-01 2.03281343e-01 6.91196024e-01 1.13332820e+00 -1.27271461e+00 -4.51139629e-01 -2.56124586e-01 -4.94326562e-01 4.84157115e-01 -1.41160354e-01 4.00779769e-02 -1.04802358e+00 -8.07229877e-01 -1.98026761e-01 -9.46758449e-01 2.38136724e-01 -2.47704566e-01 -9.15799737e-01 -3.03923845e-01 1.05147994e+00 -5.01626313e-01 8.09028327e-01 -1.97809863e+00 -7.74357840e-02 -3.16151619e-01 -3.28305624e-02 3.18715900e-01 -2.86602795e-01 6.71663105e-01 -8.12855735e-03 9.73122865e-02 -1.56320184e-01 -6.78012371e-01 6.81893378e-02 -6.60622120e-02 -8.46514225e-01 5.28276622e-01 2.08662465e-01 9.78789687e-01 -3.86202961e-01 -4.80156511e-01 7.98617125e-01 1.46286413e-01 -8.67797136e-02 2.06299275e-01 -5.23096502e-01 8.82875919e-02 -4.34159309e-01 8.12935472e-01 6.87765956e-01 -1.71060964e-01 -3.42404664e-01 3.95998284e-02 -1.10507801e-01 -2.43256554e-01 -1.04744279e+00 1.65378034e+00 9.26423073e-02 1.56643140e+00 -5.89381456e-01 -9.54647779e-01 9.98890936e-01 3.15644592e-02 -8.49010609e-03 -7.58175790e-01 3.21978748e-01 -3.07902936e-02 -3.77891928e-01 -5.97022712e-01 1.02707791e+00 4.06625092e-01 -1.22963458e-01 -1.53132558e-01 1.55842453e-01 -5.16132593e-01 5.62192202e-01 2.73050666e-01 8.06602299e-01 2.94183969e-01 3.51485610e-02 -3.90723139e-01 6.11118138e-01 4.97838229e-01 -1.36724189e-01 1.25443435e+00 -4.43530917e-01 1.18654180e+00 3.25972110e-01 -5.54550707e-01 -1.21154201e+00 -9.73801732e-01 -4.66467559e-01 1.06351125e+00 3.26322913e-01 -3.27460289e-01 -6.48935199e-01 -5.12374699e-01 -5.38680442e-02 8.84894073e-01 -1.00383210e+00 2.24650770e-01 -4.07647401e-01 -5.49909234e-01 8.00624669e-01 4.92511034e-01 7.13036537e-01 -1.18457758e+00 -3.95291507e-01 1.84599701e-02 -4.80589345e-02 -1.77548313e+00 -4.17487532e-01 1.87916473e-01 -6.83321536e-01 -1.02041268e+00 -1.07562709e+00 -1.06307590e+00 3.46965998e-01 4.91250336e-01 1.03280246e+00 -1.50027469e-01 -6.55861378e-01 7.07984149e-01 -7.90517092e-01 -8.22533607e-01 -4.88053530e-01 -2.37538088e-02 -2.44920671e-01 1.71267077e-01 7.15948462e-01 4.13965195e-01 -1.66552916e-01 3.97045761e-01 -9.08233881e-01 3.10925573e-01 4.29183096e-01 6.97144866e-01 5.53877354e-01 -4.02191132e-01 -1.07222155e-01 -8.31009328e-01 2.02491254e-01 -1.25836534e-03 -8.64997685e-01 5.98015964e-01 -2.16976017e-01 -2.42169887e-01 5.85542381e-01 -5.13320327e-01 -1.12019527e+00 1.50795653e-01 2.23433465e-01 -4.31693733e-01 -6.57076895e-01 1.00968853e-01 7.37598389e-02 -5.22830412e-02 7.80919790e-01 8.79916668e-01 -5.32001197e-01 -5.48636973e-01 2.86942601e-01 1.04724276e+00 7.11717486e-01 -4.53319252e-01 6.43678904e-01 5.13490677e-01 -4.03935909e-01 -1.70864511e+00 -6.68717802e-01 -1.00888658e+00 -7.92044461e-01 -2.32067928e-01 1.20929217e+00 -1.14798343e+00 -1.58548877e-01 1.11469030e+00 -1.09393334e+00 -8.35704267e-01 -9.98282805e-02 3.16282153e-01 -5.23790777e-01 7.63525844e-01 -2.50206947e-01 -7.12900162e-01 -2.27923051e-01 -1.19141567e+00 1.69101357e+00 3.66980210e-02 2.00987488e-01 -8.65065873e-01 1.64936185e-01 8.11920822e-01 9.98829380e-02 1.07801281e-01 6.09865785e-01 -1.02012110e+00 -7.77975678e-01 -5.56149662e-01 -3.27683359e-01 3.54619533e-01 -2.46929750e-01 1.31309271e-01 -1.27191973e+00 -3.11831355e-01 -5.86858869e-01 -6.85626864e-01 1.30283117e+00 2.28098541e-01 1.09192252e+00 2.48334870e-01 -2.87926763e-01 5.82525671e-01 1.66139269e+00 4.56784070e-02 9.77557123e-01 5.61657488e-01 9.73321915e-01 5.56341171e-01 6.20256603e-01 3.24328840e-01 2.23693520e-01 8.44940901e-01 4.59347546e-01 1.75509706e-01 -4.42975432e-01 -2.94404417e-01 3.34472656e-01 4.17793483e-01 1.99971452e-01 -1.05151725e+00 -1.22518384e+00 6.60888970e-01 -1.65237749e+00 -7.08913803e-01 -6.63177907e-01 2.07723570e+00 3.82560551e-01 8.49155784e-02 -1.12631790e-01 1.18015803e-01 8.47370803e-01 4.23425674e-01 -4.46286291e-01 -3.79351437e-01 -6.27386928e-01 -9.01250355e-03 7.96098173e-01 2.96309382e-01 -1.55770612e+00 1.59454405e+00 6.22068453e+00 8.90210629e-01 -1.22155941e+00 -2.78282523e-01 4.92750555e-01 1.50448084e-01 2.80911684e-01 -1.28583148e-01 -1.15605021e+00 2.38932282e-01 6.72958255e-01 1.15895197e-01 2.19668597e-02 1.09040809e+00 -1.09791614e-01 -2.09625214e-01 -1.06676686e+00 1.11231637e+00 8.48828554e-01 -1.35414314e+00 5.10780811e-01 -1.74472645e-01 1.04367805e+00 7.13588893e-01 3.25976200e-02 3.83814692e-01 3.04234792e-02 -1.14525545e+00 1.08810568e+00 2.51313388e-01 9.99884844e-01 -4.12843330e-03 6.28767192e-01 2.38908276e-01 -9.69792247e-01 1.06449313e-01 -7.96268284e-01 3.18444937e-01 -3.80265638e-02 9.66163054e-02 -9.41641510e-01 4.49205726e-01 8.47895324e-01 9.49465990e-01 -1.20419848e+00 1.27749002e+00 -9.29220468e-02 6.13015115e-01 -1.89010262e-01 -5.89288771e-01 4.17392194e-01 4.82218154e-02 5.39190531e-01 1.63991642e+00 -1.04235545e-01 -2.00270057e-01 1.41829282e-01 5.58604598e-01 -1.53513998e-01 5.25503635e-01 -8.53120923e-01 -2.02692404e-01 2.77779885e-02 1.01461291e+00 -9.13162231e-01 -6.43108070e-01 -4.91044760e-01 1.43266833e+00 5.96273504e-02 5.12495100e-01 -7.49720037e-01 -3.49038571e-01 6.86294883e-02 -1.25876933e-01 5.54026723e-01 -1.34649098e-01 -1.86672956e-01 -1.58773577e+00 1.11344092e-01 -8.35114956e-01 1.98103085e-01 -1.03833532e+00 -1.15824986e+00 6.97096407e-01 5.66531904e-03 -1.11683190e+00 1.30412221e-01 -1.23511529e+00 -4.63569760e-01 5.65287888e-01 -1.61188734e+00 -1.50261748e+00 -8.79130423e-01 6.12618923e-01 1.40529168e+00 -4.81464773e-01 6.51047528e-01 -4.30366024e-02 -5.07797003e-01 6.63125157e-01 5.85545242e-01 4.67948586e-01 9.10460353e-01 -1.30200458e+00 7.67678797e-01 8.84421706e-01 4.66365784e-01 -1.29588410e-01 6.56665802e-01 -8.22252333e-01 -1.37090862e+00 -1.34312201e+00 8.03192139e-01 -1.11482191e+00 4.84598517e-01 -9.20577705e-01 -8.76211524e-01 8.13591242e-01 2.26804018e-01 5.16538322e-02 3.35729808e-01 -5.09574294e-01 -4.41946298e-01 3.98932368e-01 -5.98837733e-01 7.13508427e-01 7.35134125e-01 -3.34329545e-01 -7.84896374e-01 5.45281708e-01 5.18100262e-01 -6.71131730e-01 -3.79304886e-01 2.95133710e-01 3.84633541e-01 -9.02626872e-01 7.94321001e-01 -6.15621388e-01 6.30396485e-01 -2.73560118e-02 -5.72242975e-01 -6.41542852e-01 8.00274462e-02 -3.06830794e-01 1.21888779e-01 1.11454153e+00 3.64625722e-01 -2.66222954e-01 9.34912920e-01 2.53218114e-01 -3.45302552e-01 -3.18165034e-01 -8.14461946e-01 -9.05364633e-01 2.77650148e-01 -8.82721186e-01 -2.08939463e-02 8.59550595e-01 -5.31910062e-01 3.97115558e-01 -6.26247108e-01 -1.46411166e-01 6.59044802e-01 -9.57372785e-02 1.26241314e+00 -1.06408465e+00 3.02827125e-03 -4.57352638e-01 -9.16219115e-01 -1.38256025e+00 -7.54233599e-02 -8.49739492e-01 1.96671158e-01 -1.62805700e+00 2.91064769e-01 2.23530410e-03 2.60967910e-01 1.34897903e-01 -9.59852040e-02 5.68786740e-01 4.84568596e-01 3.29632044e-01 -9.01736140e-01 5.83288372e-01 1.07984078e+00 -5.92244506e-01 8.22831094e-02 -2.21114546e-01 -1.05011694e-01 6.57905161e-01 5.38895607e-01 -1.78598672e-01 -2.73756534e-01 -6.67376399e-01 -2.66723156e-01 -3.72866303e-01 4.80570674e-01 -1.01525831e+00 1.27257824e-01 -1.24834456e-01 5.53029597e-01 -1.17301798e+00 6.13210440e-01 -5.93755603e-01 -4.24127370e-01 1.95351556e-01 -4.55290586e-01 -2.20467359e-01 4.03908312e-01 7.84748971e-01 -4.13550958e-02 -4.89625007e-01 7.31172562e-01 1.51453957e-01 -1.32934582e+00 1.35190949e-01 -5.76827109e-01 3.36226910e-01 1.25691593e+00 -6.14576459e-01 -5.94886720e-01 -2.41822422e-01 -5.16520560e-01 4.86652851e-02 5.97005844e-01 1.05868232e+00 7.69964278e-01 -8.67647707e-01 -1.06182134e+00 -6.15773257e-03 9.35950041e-01 -2.80010477e-02 3.75044137e-01 4.68010277e-01 -1.08509529e+00 9.10117567e-01 4.58375327e-02 -1.04179180e+00 -1.54001820e+00 5.62308431e-01 2.60277092e-01 -6.49009421e-02 -8.56067002e-01 8.78175855e-01 3.50864798e-01 -3.58140171e-01 5.00955522e-01 -1.34433135e-01 -3.47649485e-01 -1.81821883e-01 6.20578289e-01 2.55443931e-01 1.74492538e-01 -9.78234231e-01 -2.32016489e-01 8.63657713e-01 -3.01058203e-01 6.50320016e-03 9.80049610e-01 -9.27981827e-03 2.83343881e-01 5.88222802e-01 1.21640873e+00 -1.35690019e-01 -1.11707628e+00 -3.35713297e-01 2.53612618e-03 -5.85130274e-01 -6.08976297e-02 -9.53582466e-01 -5.73129058e-01 1.14920783e+00 8.30188215e-01 -1.03798777e-01 6.37816787e-01 8.35529342e-02 3.30289721e-01 8.33607018e-01 2.55317390e-02 -1.32159221e+00 3.11698675e-01 7.71204054e-01 8.96529198e-01 -1.59907889e+00 -2.92908140e-02 -4.71895158e-01 -1.03038406e+00 1.36959159e+00 7.96027958e-01 -2.37281695e-01 1.95073158e-01 -4.81675044e-02 2.08721906e-01 -2.97140002e-01 -6.35728717e-01 -1.80127323e-01 5.54021060e-01 7.43008137e-01 3.23345393e-01 -7.03407973e-02 9.20501649e-02 1.60717368e-01 -1.26698032e-01 -4.06137615e-01 1.03816950e+00 9.70075369e-01 -4.59080607e-01 -8.50627720e-01 -3.83654952e-01 5.70779443e-01 -4.96604711e-01 -3.79373372e-01 -9.42502856e-01 1.05049551e+00 -3.58406901e-01 9.92564678e-01 8.19699466e-03 -1.16660573e-01 3.82522881e-01 1.11345537e-01 4.32234406e-01 -7.35790253e-01 -1.93843946e-01 -9.82845724e-02 1.47846416e-01 -1.94368735e-01 -1.57584101e-01 -6.40294313e-01 -8.05948555e-01 -9.34147388e-02 -5.98976612e-01 -2.98960388e-01 9.83425617e-01 7.98212707e-01 -1.18677029e-02 3.91128242e-01 2.13025302e-01 -7.64416456e-01 -4.18952078e-01 -1.12476254e+00 -5.03645182e-01 6.42873466e-01 2.65268654e-01 -4.34724599e-01 -1.52038068e-01 5.09902656e-01]
[11.954167366027832, 2.239657163619995]
e441c20c-43c5-48e0-81cf-b45ecdde7713
musicnn-pre-trained-convolutional-neural
1909.06654
null
https://arxiv.org/abs/1909.06654v1
https://arxiv.org/pdf/1909.06654v1.pdf
musicnn: Pre-trained convolutional neural networks for music audio tagging
Pronounced as "musician", the musicnn library contains a set of pre-trained musically motivated convolutional neural networks for music audio tagging: https://github.com/jordipons/musicnn. This repository also includes some pre-trained vgg-like baselines. These models can be used as out-of-the-box music audio taggers, as music feature extractors, or as pre-trained models for transfer learning. We also provide the code to train the aforementioned models: https://github.com/jordipons/musicnn-training. This framework also allows implementing novel models. For example, a musically motivated convolutional neural network with an attention-based output layer (instead of the temporal pooling layer) can achieve state-of-the-art results for music audio tagging: 90.77 ROC-AUC / 38.61 PR-AUC on the MagnaTagATune dataset --- and 88.81 ROC-AUC / 31.51 PR-AUC on the Million Song Dataset.
['Xavier Serra', 'Jordi Pons']
2019-09-14
null
null
null
null
['audio-tagging']
['audio']
[-7.73507282e-02 -7.71435276e-02 -3.43848705e-01 -2.38953382e-02 -1.38784277e+00 -8.33745599e-01 1.75355181e-01 -2.79610485e-01 -3.99330705e-01 2.85562724e-01 4.59786534e-01 2.27435619e-01 -7.44789317e-02 -5.36284328e-01 -6.79232717e-01 -5.26525974e-01 -4.14038420e-01 2.86583215e-01 -2.52949214e-03 -5.90859689e-02 1.76224664e-01 -1.70666084e-01 -1.44704533e+00 6.76172078e-01 1.71211764e-01 1.02934599e+00 7.16288686e-02 1.02647555e+00 1.17551714e-01 5.48827648e-01 -5.51229894e-01 -2.52103746e-01 1.33481249e-01 -4.39832926e-01 -8.66406083e-01 -5.11651218e-01 5.51181674e-01 3.14350352e-02 -4.42269206e-01 1.08839500e+00 8.81608009e-01 1.38211727e-01 2.35791430e-01 -1.07279646e+00 -8.43423188e-01 1.50167096e+00 -4.42858100e-01 2.08638743e-01 1.36312947e-01 2.13933766e-01 1.61675692e+00 -4.98573124e-01 4.29989636e-01 8.34272325e-01 9.35716510e-01 7.47570574e-01 -9.61877048e-01 -1.17683160e+00 -1.78775758e-01 1.94785580e-01 -1.40277743e+00 -4.95952129e-01 9.07935560e-01 -5.41232586e-01 8.28726828e-01 2.39447683e-01 7.48000741e-01 1.11805367e+00 -2.04582021e-01 9.74932492e-01 5.59673369e-01 -2.30090499e-01 -1.56325951e-01 -5.23256183e-01 -2.04628631e-01 4.95363325e-01 -3.87299299e-01 1.14169970e-01 -9.89779472e-01 -1.96474120e-01 1.15728331e+00 -3.94230723e-01 -1.72031581e-01 2.86169142e-01 -1.46947157e+00 4.78180617e-01 6.25247598e-01 6.90834165e-01 -1.16663478e-01 8.60423446e-01 7.95939505e-01 2.58597881e-01 5.09082019e-01 1.03588331e+00 -5.94183087e-01 -6.79481149e-01 -1.39138949e+00 4.36321110e-01 4.33241278e-01 7.37652302e-01 2.63093382e-01 3.59635383e-01 -4.74092245e-01 1.16437638e+00 1.18098468e-01 1.01188488e-01 4.95488882e-01 -1.23378444e+00 2.50030965e-01 8.48886520e-02 -1.70737773e-01 -2.82902360e-01 -4.56887811e-01 -1.00583243e+00 -7.46346354e-01 -7.53080621e-02 3.90710324e-01 -6.73461482e-02 -6.53812051e-01 2.00108576e+00 -2.04663187e-01 6.83939576e-01 -3.75084192e-01 1.25893593e+00 1.15673184e+00 4.59745020e-01 2.55921394e-01 3.68525267e-01 1.54267442e+00 -1.25976145e+00 -4.47519690e-01 1.33130297e-01 5.34980595e-01 -1.22125864e+00 1.55976844e+00 5.71559787e-01 -1.20292759e+00 -9.63442028e-01 -9.55723286e-01 -1.09576263e-01 -1.50166318e-01 7.25815117e-01 7.03511536e-01 2.37056315e-01 -1.04556668e+00 1.15899599e+00 -8.21236789e-01 -2.08435819e-01 5.65649748e-01 3.52157682e-01 6.64175861e-03 7.48537779e-01 -1.09236026e+00 7.85421953e-02 4.50369596e-01 -1.64220572e-01 -1.34635782e+00 -1.02029860e+00 -3.65885466e-01 2.29812428e-01 6.36028349e-02 -7.03846991e-01 1.72816122e+00 -1.18178809e+00 -1.55993283e+00 1.22681797e+00 3.15694571e-01 -5.50540805e-01 3.64048004e-01 -4.99923080e-01 -5.41957974e-01 -3.88394259e-02 2.00000033e-01 1.05788505e+00 4.86202508e-01 -5.97725451e-01 -4.73433435e-01 -5.67503422e-02 8.54781270e-02 6.82294443e-02 -2.12687537e-01 2.47611359e-01 -6.82007253e-01 -1.26242900e+00 -2.68679172e-01 -1.21080363e+00 7.72482250e-03 -3.71369272e-01 -9.39869404e-01 -2.47243702e-01 3.33674341e-01 -6.99871421e-01 1.39521623e+00 -2.54559779e+00 -1.31684214e-01 -2.21188039e-01 -1.89064935e-01 1.98143557e-01 -5.88821769e-01 2.39708185e-01 -3.23248833e-01 2.45283678e-01 -1.13938272e-01 -3.04415256e-01 3.48675609e-01 -4.47013795e-01 -4.34155732e-01 1.01935133e-01 2.65589897e-02 1.09172034e+00 -9.07020152e-01 -5.68721956e-03 6.52278811e-02 5.55648267e-01 -6.15006626e-01 1.01641215e-01 -3.71493191e-01 8.35722625e-01 -3.01607311e-01 6.72861695e-01 1.62285343e-01 -1.37671560e-01 -4.20871004e-02 -1.01719908e-01 -8.28177109e-02 9.29916263e-01 -8.82557094e-01 2.68144441e+00 -4.53393489e-01 8.23843539e-01 -1.23775937e-01 -4.25314337e-01 8.65389168e-01 5.84709764e-01 6.36489749e-01 -3.18508148e-01 2.74575859e-01 3.00866485e-01 1.80261940e-01 -1.27586558e-01 5.65504014e-01 -1.23130582e-01 -4.46754456e-01 5.29909909e-01 5.64351678e-01 1.20682664e-01 3.30764279e-02 -1.61474928e-01 1.16011453e+00 5.58549941e-01 -1.78025559e-01 -2.37759575e-01 1.31718725e-01 -1.42600387e-01 5.85037589e-01 6.21602714e-01 1.49693685e-02 1.03564763e+00 2.57614732e-01 -4.32760060e-01 -1.06132472e+00 -9.79882240e-01 2.76430300e-03 1.75745487e+00 -4.44067180e-01 -9.07238305e-01 -8.07336986e-01 -1.76607698e-01 -3.19799036e-02 3.82870793e-01 -5.98664761e-01 -1.87496841e-01 -3.25848728e-01 -3.90442461e-01 1.47877419e+00 7.41969526e-01 4.84083235e-01 -1.67412746e+00 -3.02244842e-01 4.17974740e-01 -3.41416925e-01 -6.63556039e-01 -6.64737523e-01 3.88901383e-01 -6.59071267e-01 -1.05700421e+00 -8.26110542e-01 -8.20792615e-01 -3.03109199e-01 -2.62180090e-01 1.39070284e+00 8.29260238e-03 -1.30050555e-01 1.42871276e-01 -4.91364926e-01 -7.23337829e-01 -8.18915293e-02 7.99769580e-01 -4.96396050e-02 -1.39635861e-01 3.63592923e-01 -1.03743649e+00 -8.15953493e-01 1.74695224e-01 -5.83190143e-01 -6.99380413e-02 1.15309618e-01 5.63396394e-01 1.01656735e+00 -5.74956357e-01 7.72875845e-01 -5.42673886e-01 5.45532525e-01 -1.42998382e-01 -4.13085550e-01 -2.06306219e-01 -1.30499914e-01 -3.13834637e-01 4.22425419e-01 -8.37874889e-01 -2.40714192e-01 8.76248777e-02 -4.78899598e-01 -6.82293713e-01 -3.26261818e-01 3.25729519e-01 -1.69394314e-02 4.45846617e-01 7.10418999e-01 -1.64939955e-01 -6.32364035e-01 -1.14442766e+00 4.98519212e-01 9.00039494e-01 1.04294944e+00 -6.86865628e-01 5.81449986e-01 1.57879367e-01 -3.29796106e-01 -4.81801093e-01 -1.27636909e+00 -6.96601391e-01 -4.85931545e-01 -1.64807469e-01 1.06837809e+00 -1.16857004e+00 -8.45342338e-01 3.95847827e-01 -9.29570436e-01 -8.52438748e-01 -7.21902490e-01 4.59367573e-01 -9.93162334e-01 -1.63868681e-01 -1.02597773e+00 -6.82709038e-01 -8.31164837e-01 -6.93262279e-01 1.17252445e+00 2.79903114e-01 -7.71109760e-01 -5.77558696e-01 5.15964091e-01 3.97462875e-01 2.11109251e-01 1.52868256e-01 8.17789674e-01 -7.90347159e-01 -3.14258635e-01 4.68028933e-02 1.23074576e-02 2.77211159e-01 -1.52592555e-01 3.82515527e-02 -1.76708949e+00 -1.08390898e-01 -8.81374180e-01 -3.93555313e-01 1.21708965e+00 8.09844375e-01 1.55788517e+00 -6.25181186e-04 3.99269797e-02 1.04266918e+00 8.78157854e-01 3.93565707e-02 5.33856332e-01 5.31295896e-01 6.65088356e-01 -1.46569386e-02 4.79671478e-01 4.35704917e-01 7.77610242e-02 1.09251308e+00 4.27091092e-01 1.85854658e-02 -5.97077131e-01 -6.08685553e-01 3.12748134e-01 9.41605449e-01 -4.65645611e-01 3.05561498e-02 -7.73077309e-01 8.49779308e-01 -2.00640535e+00 -1.05402553e+00 -8.57304484e-02 2.05644107e+00 8.73695910e-01 -1.58836901e-01 6.49225295e-01 4.07622278e-01 8.07881415e-01 2.94448555e-01 -5.26070595e-01 -3.13598484e-01 -1.33063011e-02 6.40081942e-01 -4.22532484e-02 6.43764855e-03 -1.65275848e+00 1.37164712e+00 5.24229050e+00 1.14844871e+00 -1.20091355e+00 6.34828806e-01 3.10206950e-01 -5.76111376e-01 1.69367805e-01 -7.65991881e-02 -4.21445787e-01 4.33424652e-01 1.14266324e+00 2.03740135e-01 7.52888978e-01 7.88316429e-01 1.43466339e-01 4.10588235e-01 -8.50347042e-01 1.26293528e+00 -3.76427591e-01 -1.34478354e+00 -1.46147162e-01 4.28692028e-02 6.12972379e-01 5.03597736e-01 3.11714321e-01 4.99955833e-01 6.50025532e-02 -9.47889984e-01 1.09618032e+00 5.08278489e-01 1.24192166e+00 -9.07087862e-01 6.31652951e-01 -2.67469615e-01 -1.48031926e+00 1.02269366e-01 -4.05252635e-01 -1.17392734e-01 1.41676381e-01 4.08736885e-01 -5.64162374e-01 4.77794707e-01 1.16516674e+00 8.93382847e-01 -3.73753846e-01 1.55226636e+00 -4.21759784e-01 1.17431045e+00 -1.56528130e-01 2.66781449e-01 1.73377559e-01 1.84791908e-01 7.56769717e-01 1.45358753e+00 5.39794624e-01 -4.71575707e-01 -1.19380623e-01 9.19987023e-01 -5.11801004e-01 3.29356253e-01 -2.58237690e-01 -4.69824404e-01 2.34094977e-01 1.53909719e+00 -7.13781655e-01 -1.19102202e-01 1.51937038e-01 8.17699075e-01 1.07146159e-01 2.29489461e-01 -1.05439913e+00 -5.88031173e-01 9.35013771e-01 2.37255618e-01 3.01090658e-01 -5.57344407e-02 -2.03845143e-01 -1.06193507e+00 -4.24248368e-01 -7.81119585e-01 6.04713976e-01 -1.22717786e+00 -1.18656981e+00 7.45063663e-01 -6.33496284e-01 -1.50780916e+00 -7.38376975e-02 -3.79404098e-01 -9.91883576e-01 5.59699535e-01 -1.08665109e+00 -1.34220386e+00 -1.24171808e-01 5.45457363e-01 4.69670862e-01 -5.55553615e-01 1.16429925e+00 6.78486049e-01 -1.99331492e-01 7.73092389e-01 -2.69925654e-01 7.76034832e-01 9.40080822e-01 -1.37672079e+00 5.52721202e-01 3.20088267e-01 1.04806340e+00 2.36590371e-01 3.42065901e-01 -3.16910893e-01 -6.79235160e-01 -1.42609942e+00 6.81339204e-01 -3.07540625e-01 8.24228883e-01 -4.35207307e-01 -7.63659239e-01 7.47894466e-01 2.19528317e-01 -3.34140062e-01 1.17292631e+00 7.61532545e-01 -6.27212882e-01 -1.15127675e-01 -4.45857942e-01 5.40371001e-01 1.30280232e+00 -9.35384691e-01 -3.53479803e-01 2.75536776e-01 7.59550631e-01 -3.94839257e-01 -1.07139492e+00 3.82900238e-01 9.87344205e-01 -6.93601310e-01 8.46359551e-01 -7.08740950e-01 4.49459225e-01 -4.52626556e-01 -9.18416455e-02 -1.08703434e+00 -6.03884399e-01 -9.71122444e-01 -3.61595526e-02 1.47652042e+00 6.69349372e-01 1.17349558e-01 6.71408474e-01 -4.12598431e-01 -5.43507278e-01 -4.85116541e-01 -1.08582473e+00 -9.13435102e-01 -5.22993924e-03 -1.00365603e+00 7.10297108e-01 1.15571988e+00 1.18526416e-02 2.64062643e-01 -3.80865186e-01 -2.01198190e-01 1.24310762e-01 2.47322619e-01 9.03903902e-01 -1.22190595e+00 -7.93535113e-01 -8.14079285e-01 -4.86761004e-01 -6.33403480e-01 2.01216608e-01 -1.51174021e+00 -1.37452886e-01 -1.28361571e+00 2.54640758e-01 -3.21759880e-01 -1.09675634e+00 1.16596901e+00 2.75380105e-01 1.09599185e+00 5.14794707e-01 2.64041543e-01 -8.06960762e-01 4.02224243e-01 1.09420300e+00 -2.81317323e-01 -4.53544050e-01 1.23718038e-01 -6.69373870e-01 7.70118296e-01 1.30095911e+00 -8.39349568e-01 2.60230117e-02 -3.89316350e-01 4.13056195e-01 -1.82426065e-01 6.52175128e-01 -1.41438460e+00 -1.46421641e-01 4.42541838e-01 2.36712754e-01 -3.95374775e-01 5.06320059e-01 -3.10046095e-02 3.74877900e-01 1.22882672e-01 -6.50157034e-01 -1.93836525e-01 4.66736555e-01 5.52967712e-02 -4.26730633e-01 -2.16746405e-01 5.92304170e-01 -8.84303898e-02 -3.69874954e-01 2.79268891e-01 -1.72941014e-02 2.37105995e-01 1.58649743e-01 1.84027538e-01 -2.78346896e-01 -5.77883840e-01 -1.27470398e+00 -3.09683502e-01 1.41991153e-01 7.52688766e-01 -5.38904034e-02 -1.72098565e+00 -7.07833469e-01 -1.59536183e-01 2.24027321e-01 -3.65523845e-01 4.30384666e-01 8.98916364e-01 -2.01859698e-01 3.07686538e-01 -3.85127187e-01 -5.21633804e-01 -1.23236918e+00 1.11523077e-01 3.36246699e-01 -2.52773046e-01 -4.85136449e-01 1.15340817e+00 -7.29860663e-02 -6.56559825e-01 3.38885784e-01 -4.14389104e-01 1.70202643e-01 8.14791918e-02 3.67314577e-01 1.26898050e-01 -4.99460995e-02 -4.74442661e-01 -3.23891699e-01 7.30610371e-01 6.07695341e-01 -3.25737625e-01 1.63229954e+00 4.78648216e-01 4.98859398e-02 8.97908688e-01 8.45351338e-01 1.90099373e-01 -9.82275724e-01 8.66860971e-02 4.14501727e-02 6.25173971e-02 1.24673225e-01 -1.17715955e+00 -1.27535439e+00 1.26863849e+00 7.71749198e-01 1.32087797e-01 1.18323481e+00 3.06780159e-01 7.61949003e-01 1.47696603e-02 2.52024293e-01 -9.65968311e-01 -6.67176023e-02 4.97288406e-01 1.06231618e+00 -5.47983408e-01 -4.09184068e-01 5.03536835e-02 -6.10734820e-01 9.97955203e-01 3.12699348e-01 -5.56591392e-01 7.36163497e-01 2.11845368e-01 2.75987893e-01 -6.07679635e-02 -5.87084532e-01 -5.63381732e-01 8.10323358e-01 4.59360182e-01 1.11073196e+00 4.90447640e-01 -2.37467680e-02 1.52858162e+00 -9.99535441e-01 8.27322155e-02 8.20451155e-02 3.19930673e-01 -9.01715457e-02 -1.21794629e+00 2.75496878e-02 2.80664831e-01 -9.46414351e-01 -3.77136022e-01 -6.21716917e-01 5.92402160e-01 3.96989346e-01 8.67702842e-01 1.60807014e-01 -9.37317073e-01 2.58548796e-01 5.10690093e-01 5.07219911e-01 -7.30237603e-01 -1.33755374e+00 8.04025412e-01 1.90039992e-01 -5.65777600e-01 -4.13973302e-01 -3.47563356e-01 -1.31039178e+00 -4.02918719e-02 -1.88718691e-01 3.43650758e-01 5.47151506e-01 4.37084466e-01 5.96880913e-01 9.24516439e-01 4.82958704e-02 -1.10509396e+00 7.60572404e-02 -1.50937295e+00 -9.33424652e-01 3.27581644e-01 -5.37791215e-02 -4.01633799e-01 7.23114014e-02 7.65705481e-02]
[15.801839828491211, 5.251217365264893]
a20e611b-2c59-49b2-b2d7-269aa935eaa9
designing-deep-convolutional-neural-networks-1
null
null
https://ieeexplore.ieee.org/document/9870218
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9870218
Designing Deep Convolutional Neural Networks using a Genetic Algorithm for Image-based Malware Classification
In recent years, deep Convolutional Neural Networks (CNNs) have shown great potential in malware classification. CNNs, which are originally designed for image processing, identify malware binaries visualised as images. Despite offering promising performance, these human-designed networks are very large requiring more resources to train and deploy them. Evolutionary algorithms have been successfully used in designing deep neural networks automatically for different application domains. In this work, we use a Genetic Algorithm (GA) to optimise the CNN topology and hyperparameters for image-based malware classification. Computational experiments with two different malware datasets, Malimg and Microsoft Malware, show that the GA-evolved networks are very competitive to the networks designed by experts in classifying malware, yet they are also considerably smaller in size comparison.
['Vijay Varadharajan', 'Raymond Chiong', 'Nasimul Noman', 'Cornelius Paardekooper']
2022-07-18
null
null
null
ieee-congress-on-evolutionary-computation-cec
['malware-classification']
['miscellaneous']
[ 7.53813311e-02 -4.16921675e-01 6.49881512e-02 -1.23666406e-01 6.48243248e-01 -5.54251075e-01 6.42494917e-01 -3.57011974e-01 -5.42149782e-01 5.83327234e-01 -6.75075471e-01 -7.75028288e-01 1.66174859e-01 -8.67205918e-01 -6.17183626e-01 -7.42625356e-01 -4.42308217e-01 5.92843592e-01 3.49796593e-01 -3.73632997e-01 5.22526205e-01 8.23781788e-01 -1.48173606e+00 2.65349656e-01 5.86120009e-01 9.80481505e-01 1.24041945e-01 1.02198350e+00 -2.14416757e-01 6.02735519e-01 -1.10539114e+00 -4.83321041e-01 3.42358828e-01 -3.78836930e-01 -5.91822684e-01 -4.55057696e-02 3.80119234e-02 9.78826359e-03 -4.97839540e-01 1.21880877e+00 2.90419251e-01 -2.23429993e-01 6.46953046e-01 -1.37201464e+00 -8.56137335e-01 4.77942675e-01 -1.06946923e-01 6.08410954e-01 -2.22126186e-01 4.58068639e-01 3.07699710e-01 -3.02714676e-01 7.05731273e-01 1.24416471e+00 9.15836871e-01 1.03042221e+00 -9.85780180e-01 -7.73243606e-01 -5.14801383e-01 6.05508268e-01 -1.18899143e+00 -2.13197410e-01 7.38172948e-01 -4.98270005e-01 1.50692582e+00 2.50860691e-01 1.03200579e+00 1.49710870e+00 4.08776075e-01 3.59092444e-01 7.13880837e-01 -2.49730274e-01 2.48172462e-01 4.64526236e-01 -5.87126613e-02 8.26274157e-01 5.44069588e-01 5.88499427e-01 2.55068153e-01 -3.13855000e-02 6.40465200e-01 2.07364261e-01 -1.61902323e-01 -2.63308048e-01 -7.05355465e-01 1.49691892e+00 9.32370543e-01 7.12665617e-01 -4.19533968e-01 1.38256595e-01 9.46845412e-01 5.30138075e-01 1.18207388e-01 1.04951322e+00 -3.38050663e-01 -3.02548446e-02 -6.03755832e-01 4.48672436e-02 7.95438051e-01 4.55121785e-01 3.72894228e-01 7.35953391e-01 7.55223483e-02 7.01163232e-01 -3.59404013e-02 1.31964967e-01 9.51872587e-01 -3.44118625e-01 7.39353523e-03 9.28358376e-01 -6.43203318e-01 -1.39380395e+00 -5.33716202e-01 -2.50291795e-01 -1.19062495e+00 3.59344155e-01 1.44761488e-01 -9.43659618e-02 -1.25646210e+00 1.19756114e+00 -1.34663224e-01 4.66268882e-02 1.92179412e-01 4.27751750e-01 7.72637308e-01 8.58801007e-01 1.30320340e-01 1.38538510e-01 1.16281331e+00 -8.06636930e-01 -1.69304326e-01 -2.49472946e-01 2.67631948e-01 -2.84595430e-01 5.45823336e-01 4.42230731e-01 -6.23586357e-01 -6.21402025e-01 -1.08105528e+00 7.67183304e-01 -1.03742242e+00 -2.18491688e-01 5.54455817e-01 1.20576298e+00 -1.36147356e+00 6.66895211e-01 -6.33637488e-01 -5.02768755e-01 9.56792653e-01 7.80683100e-01 -5.73454536e-02 4.16417509e-01 -1.02764606e+00 8.13572168e-01 1.27362061e+00 -3.37998965e-04 -1.35659647e+00 -8.92548040e-02 -5.05549610e-01 2.34444678e-01 -1.74057763e-02 -2.51154453e-01 8.60661864e-01 -1.58117139e+00 -1.41443586e+00 9.45332706e-01 6.34470046e-01 -1.01636302e+00 2.11193338e-01 7.53482878e-01 -5.06872416e-01 1.09952010e-01 -7.46145248e-01 9.83628690e-01 1.48353755e+00 -1.31970930e+00 -4.12562162e-01 -1.84015155e-01 -7.27200359e-02 -5.57744324e-01 -8.51647735e-01 4.18108910e-01 -6.75283074e-02 -6.90307379e-01 -4.77917701e-01 -9.24594522e-01 -1.82601675e-01 -4.86506373e-01 -4.38455641e-01 -2.07680151e-01 1.70209515e+00 -6.85498834e-01 8.94968629e-01 -1.79431868e+00 2.47483537e-01 4.41649884e-01 4.11492079e-01 1.28661406e+00 -1.44879684e-01 -1.71765238e-01 -2.19360098e-01 4.32985097e-01 -2.14793757e-01 1.47856995e-01 -2.22556323e-01 4.39017743e-01 -5.97626669e-03 3.32600027e-01 4.55233276e-01 1.20873523e+00 -5.90003371e-01 -4.66771275e-01 2.74191856e-01 6.58766806e-01 -4.68593955e-01 3.58288705e-01 -3.70449156e-01 3.60340595e-01 -1.69128075e-01 9.51373398e-01 4.75995660e-01 -1.78222448e-01 1.92276120e-01 1.14692777e-01 2.40624189e-01 -4.30344373e-01 -8.45994279e-02 6.41852617e-01 -2.56959558e-01 1.34747672e+00 1.10688195e-01 -1.53250599e+00 1.05701303e+00 1.98373258e-01 1.32561058e-01 -7.31289685e-01 9.31389749e-01 1.15518570e-01 6.92486107e-01 -6.72791719e-01 2.64461696e-01 2.17664793e-01 2.46940285e-01 3.24113190e-01 1.92414939e-01 -1.73648164e-01 3.65139097e-01 -3.52907270e-01 1.11396432e+00 -5.77544093e-01 3.66637945e-01 -1.18655339e-01 7.78589845e-01 3.11497629e-01 2.11791471e-01 6.50095880e-01 -5.09398520e-01 1.26892939e-01 6.24032378e-01 -1.11589420e+00 -1.59122729e+00 -5.33399105e-01 -3.55341285e-02 9.17274594e-01 -2.57561922e-01 -5.45411184e-02 -1.24101102e+00 -9.82211828e-01 -2.28441730e-01 3.76679391e-01 -8.28210354e-01 -3.83665711e-01 -9.86680686e-01 -9.55990076e-01 9.62153852e-01 4.15913373e-01 5.73381543e-01 -1.81191540e+00 -1.16802299e+00 1.69615552e-01 6.02847338e-01 -8.33147943e-01 1.85299903e-01 5.34042597e-01 -7.80799150e-01 -1.25774181e+00 -6.34579480e-01 -9.48250115e-01 8.73302102e-01 -1.13819689e-01 1.09601784e+00 7.44870365e-01 -7.51618445e-01 -6.91240951e-02 -5.69001138e-01 -7.33618379e-01 -1.07544327e+00 2.65011996e-01 -1.23191275e-01 -3.04051846e-01 3.79796803e-01 -4.76540208e-01 -1.45235091e-01 2.40421936e-01 -1.09763515e+00 -3.82261097e-01 6.93423212e-01 1.21378255e+00 -5.87433763e-02 6.33249640e-01 2.04641417e-01 -7.61561930e-01 1.13110971e+00 -5.74703276e-01 -7.59421587e-01 1.88182294e-01 -5.44646800e-01 -1.69754475e-01 1.20108736e+00 -8.76236558e-01 -4.53684628e-01 -3.51875424e-02 -1.28062919e-01 -9.01848316e-01 -4.07908112e-01 8.35110396e-02 1.73946485e-01 -8.19499433e-01 9.67633903e-01 2.00381964e-01 3.09179366e-01 -3.92181352e-02 -2.26541042e-01 7.65259087e-01 3.09454590e-01 -4.36055921e-02 9.42392826e-01 6.85052723e-02 -1.23807358e-05 -7.92080820e-01 2.22907573e-01 7.53702000e-02 -4.97611105e-01 -2.07067236e-01 9.86210227e-01 -1.03934452e-01 -9.40530837e-01 8.31152856e-01 -1.26623142e+00 -2.39174441e-01 -2.53241770e-02 -2.77088940e-01 -1.51328847e-01 8.28908291e-03 -5.22964060e-01 -6.18050039e-01 -5.98748863e-01 -1.58213985e+00 4.33359504e-01 6.00899637e-01 -6.09742142e-02 -1.25705647e+00 -3.58348973e-02 -7.10903704e-02 9.90377426e-01 3.84325773e-01 1.24460614e+00 -1.10523605e+00 -1.74302533e-01 -3.65011424e-01 -5.61810851e-01 6.25831008e-01 -5.80687337e-02 4.24451649e-01 -8.82637024e-01 -5.00224531e-01 -1.04559369e-01 -3.04990858e-01 8.20417106e-01 3.85206282e-01 1.77194297e+00 -5.49210906e-01 -5.97890615e-01 8.57345581e-01 1.43574321e+00 1.06607568e+00 8.41915369e-01 7.70353615e-01 8.22129607e-01 4.52577233e-01 -2.00483054e-01 1.20206557e-01 -4.94320750e-01 5.01784563e-01 1.03715813e+00 2.86816270e-03 1.27046764e-01 4.34568048e-01 3.11305791e-01 4.61547047e-01 -2.54540712e-01 -6.19663477e-01 -1.29660451e+00 1.99226633e-01 -1.34168327e+00 -9.91084933e-01 2.91531742e-01 1.53174782e+00 3.73031110e-01 2.28915229e-01 3.34423751e-01 1.82392865e-01 8.25960457e-01 -7.40109012e-02 -5.54721594e-01 -1.05743277e+00 -2.63269335e-01 7.11037397e-01 6.81746721e-01 -1.88638762e-01 -1.42814112e+00 1.01153147e+00 6.89685392e+00 9.78637576e-01 -1.61978006e+00 1.39782935e-01 8.08352590e-01 1.45729616e-01 3.15216899e-01 -5.35835147e-01 -3.79535735e-01 6.58489645e-01 1.29496205e+00 1.77029148e-01 7.91269362e-01 1.16975605e+00 -2.92705566e-01 3.38967770e-01 -4.65736896e-01 9.21816528e-01 1.45795181e-01 -1.71902966e+00 1.10191345e-01 3.37519586e-01 8.64637196e-01 5.74352778e-02 3.49328190e-01 3.58150691e-01 3.24539095e-01 -1.55717599e+00 4.56251502e-01 3.72456089e-02 6.12105131e-01 -1.13263500e+00 1.10723221e+00 8.97601098e-02 -9.55949068e-01 -8.80675673e-01 -6.51883662e-01 3.06932837e-01 -3.80849600e-01 -8.53908509e-02 -1.21506929e+00 8.10851306e-02 1.00694013e+00 2.63855070e-01 -1.02182817e+00 1.08676422e+00 1.02090120e-01 4.35223758e-01 1.30751804e-01 -6.03997767e-01 6.72971487e-01 1.02568775e-01 5.46503127e-01 1.33655858e+00 2.47435823e-01 -5.04845679e-01 -1.31150797e-01 1.00234509e+00 -5.78462183e-02 -2.36773178e-01 -7.75600255e-01 -8.49301517e-01 1.86151758e-01 1.30280030e+00 -1.25023019e+00 -4.66324061e-01 2.70757616e-01 8.57147038e-01 1.67441845e-01 -2.69099455e-02 -1.07643485e+00 -6.24616444e-01 9.06231165e-01 -2.27549762e-01 5.87858200e-01 -9.66666266e-03 -1.53004125e-01 -4.88991678e-01 -6.82623923e-01 -1.43502355e+00 1.27703980e-01 -3.90855342e-01 -1.22311306e+00 1.39046347e+00 -1.23035222e-01 -9.50577259e-01 -6.44130290e-01 -1.58075511e+00 -1.01568401e+00 2.64145195e-01 -8.00565839e-01 -9.00495350e-01 -3.99359703e-01 5.27799964e-01 5.27727604e-01 -1.02803731e+00 6.77055478e-01 2.28406057e-01 -1.07003284e+00 4.94154602e-01 1.27737716e-01 3.28097850e-01 -1.40120104e-01 -8.42950821e-01 6.27877831e-01 7.40156889e-01 -3.58754210e-02 4.55667853e-01 6.66129947e-01 -5.90103567e-01 -1.33939898e+00 -1.17577457e+00 1.46417290e-01 -5.16404659e-02 4.46715653e-01 -4.16709960e-01 -9.59766328e-01 2.06339285e-01 5.26100874e-01 -2.89813101e-01 2.06840411e-01 -5.17285585e-01 -4.06999707e-01 2.71550655e-01 -1.68198109e+00 3.47134441e-01 8.32383692e-01 -1.48249120e-01 -1.33903086e-01 4.12814736e-01 5.83940506e-01 -2.88275331e-01 -2.07052931e-01 4.08997864e-01 2.95332581e-01 -1.18018234e+00 1.16502237e+00 -9.49166238e-01 2.99622208e-01 1.14927687e-01 2.58636773e-01 -1.44845474e+00 -3.24581623e-01 -3.92482102e-01 -2.56557256e-01 7.10658133e-01 1.59426078e-01 -8.26207161e-01 1.04202712e+00 -5.75944968e-02 -1.75498128e-02 -7.67992973e-01 -7.91951835e-01 -9.90598798e-01 -3.42329182e-02 -6.56073093e-02 7.56916821e-01 8.75985980e-01 -5.46437442e-01 -1.07911639e-01 -2.01547444e-01 -4.02212411e-01 3.18574309e-01 -4.45647955e-01 4.23861802e-01 -1.59144270e+00 -1.45936936e-01 -1.21829307e+00 -1.00579202e+00 1.38589934e-01 6.17853463e-01 -6.31856263e-01 -2.49456130e-02 -7.68276632e-01 1.51271060e-01 -4.52557862e-01 -2.25869510e-02 5.69365978e-01 3.06121975e-01 6.06325924e-01 1.02259919e-01 1.16438605e-01 7.42508247e-02 3.26068364e-02 8.65873575e-01 -5.24679840e-01 -3.88364233e-02 -8.52163136e-02 -3.05454314e-01 7.03120887e-01 1.28574491e+00 -5.49546957e-01 -2.19777003e-01 -2.07863748e-01 -5.39243445e-02 -8.11644435e-01 4.02528226e-01 -1.28305197e+00 1.57130752e-02 3.64448428e-02 8.79581392e-01 -3.40717226e-01 3.13884318e-01 -9.02065516e-01 5.23295641e-01 1.22031403e+00 1.33865744e-01 4.80203003e-01 4.96545076e-01 2.04058141e-01 -1.53775975e-01 -7.46042430e-01 1.14717925e+00 -4.82378513e-01 -9.71416354e-01 3.93378913e-01 -7.66484737e-01 -3.64672512e-01 1.48427320e+00 -6.77927315e-01 -3.92167628e-01 -8.05453509e-02 -2.25116804e-01 -3.47200662e-01 6.01394832e-01 5.04895329e-01 8.09614122e-01 -1.00606692e+00 -3.85280550e-01 3.76978546e-01 -2.82805711e-01 -5.13287604e-01 -8.11131075e-02 3.74243557e-01 -1.24427748e+00 8.93598080e-01 -1.07015169e+00 -6.89297378e-01 -1.65793586e+00 8.94160688e-01 6.42893493e-01 -3.29661518e-01 -1.70015216e-01 1.18033314e+00 -2.45922491e-01 -4.98072535e-01 3.54603946e-01 2.00574547e-01 -5.24311125e-01 -3.97600010e-02 4.65405047e-01 2.98153013e-01 7.60265887e-02 -8.05143178e-01 -3.30178648e-01 1.25383481e-01 -4.76955436e-02 5.58992088e-01 1.55626345e+00 6.47360444e-01 -4.57962096e-01 -4.66445625e-01 1.46138537e+00 -8.20958197e-01 -7.27454245e-01 4.01941508e-01 1.71914563e-01 -4.19270605e-01 -3.90527025e-02 -5.33099532e-01 -1.79391420e+00 9.69162703e-01 9.16108549e-01 8.62127304e-01 1.37039018e+00 -2.85046965e-01 6.53990328e-01 6.61673188e-01 1.08730219e-01 -7.24537313e-01 3.81525010e-01 8.42403650e-01 6.96328759e-01 -1.09008765e+00 -4.00169730e-01 7.93450475e-02 -3.14938247e-01 1.66510308e+00 1.01995695e+00 -3.00229549e-01 4.08538193e-01 4.30935264e-01 -9.51353312e-02 -5.92023849e-01 -4.61518824e-01 2.33042892e-02 1.53353542e-01 1.20154381e+00 9.19035822e-02 1.48351401e-01 1.63454562e-01 -2.14333475e-01 -2.56469667e-01 -4.59976196e-01 2.16805443e-01 9.71292675e-01 -5.40926278e-01 -1.05799413e+00 -7.11170197e-01 5.53187907e-01 -3.29929203e-01 -3.72598697e-05 -1.08728981e+00 8.20105672e-01 4.69049275e-01 6.09697461e-01 2.93801099e-01 -9.87208962e-01 -6.21727705e-02 -7.07415640e-02 6.22475982e-01 -1.59707546e-01 -1.31972647e+00 -6.67320669e-01 -8.98588002e-02 -3.42313021e-01 -7.21128434e-02 -1.05757825e-01 -8.29422593e-01 -7.13482857e-01 -2.97111571e-01 6.14732280e-02 1.06017494e+00 6.60030365e-01 2.31232062e-01 7.18098521e-01 7.10532427e-01 -1.41478586e+00 -4.04853582e-01 -9.42846894e-01 -7.41175096e-03 1.81895062e-01 1.70820337e-02 -6.10809982e-01 -2.69999266e-01 -1.83284178e-01]
[14.390206336975098, 9.66279411315918]
a9d97bab-1f7c-47c9-92bb-44953b685f9c
multimodal-transformer-distillation-for-audio
2210.15563
null
https://arxiv.org/abs/2210.15563v1
https://arxiv.org/pdf/2210.15563v1.pdf
Multimodal Transformer Distillation for Audio-Visual Synchronization
Audio-visual synchronization aims to determine whether the mouth movements and speech in the video are synchronized. VocaLiST reaches state-of-the-art performance by incorporating multimodal Transformers to model audio-visual interact information. However, it requires high computing resources, making it impractical for real-world applications. This paper proposed an MTDVocaLiST model, which is trained by our proposed multimodal Transformer distillation (MTD) loss. MTD loss enables MTDVocaLiST model to deeply mimic the cross-attention distribution and value-relation in the Transformer of VocaLiST. Our proposed method is effective in two aspects: From the distillation method perspective, MTD loss outperforms other strong distillation baselines. From the distilled model's performance perspective: 1) MTDVocaLiST outperforms similar-size SOTA models, SyncNet, and PM models by 15.69% and 3.39%; 2) MTDVocaLiST reduces the model size of VocaLiST by 83.52%, yet still maintaining similar performance.
['Jyh-Shing Roger Jang', 'Hung-Yi Lee', 'Chung-Che Wang', 'Haibin Wu', 'Xuanjun Chen']
2022-10-27
null
null
null
null
['audio-visual-synchronization', 'audio-visual-synchronization']
['audio', 'computer-vision']
[-1.56105042e-01 -5.94385900e-02 -3.13077956e-01 3.53887826e-02 -9.30877686e-01 -5.73082864e-01 6.49260283e-01 -2.29548469e-01 -3.63040775e-01 1.64324284e-01 6.42581999e-01 -1.40625790e-01 4.39684361e-01 -2.13516966e-01 -7.36355782e-01 -6.72001481e-01 3.30696166e-01 3.14948022e-01 1.53376728e-01 -9.16433260e-02 -7.28462338e-02 -6.38196841e-02 -1.43214500e+00 4.53770608e-01 6.10440075e-01 1.04455364e+00 3.93276364e-01 6.56826437e-01 3.17414454e-03 9.62506950e-01 -4.91117835e-01 -5.26000261e-01 -1.68904930e-01 -3.96610856e-01 -6.55364394e-01 -4.06173259e-01 8.32680047e-01 -3.15300167e-01 -7.26554751e-01 9.01622117e-01 8.79049242e-01 1.13513485e-01 5.44725180e-01 -1.43710685e+00 -7.03636110e-01 8.67671072e-01 -6.93330884e-01 2.65872419e-01 3.31658959e-01 4.00691181e-01 1.52250624e+00 -1.06412804e+00 2.52119392e-01 1.69558144e+00 5.69738090e-01 5.83979487e-01 -1.19168091e+00 -1.03564525e+00 4.73969251e-01 4.26319450e-01 -1.27236640e+00 -7.03007281e-01 7.44527459e-01 -2.52136201e-01 1.12218654e+00 3.02752316e-01 7.42892504e-01 1.40838122e+00 1.22052073e-01 1.10683572e+00 8.45598161e-01 1.07491314e-02 -2.23510534e-01 7.70691559e-02 -1.75482243e-01 4.82963622e-01 -3.63672763e-01 1.50089011e-01 -9.77272153e-01 4.56707850e-02 7.14528203e-01 -2.09757835e-01 -3.50061119e-01 3.96311507e-02 -1.25842595e+00 7.51248002e-01 2.62475759e-01 1.80705041e-01 3.81441712e-02 4.26115513e-01 6.78860307e-01 3.16970229e-01 4.68379825e-01 2.62383044e-01 -4.24948245e-01 -4.46659088e-01 -1.01164174e+00 1.08841797e-02 4.48738724e-01 9.22004580e-01 3.43897879e-01 4.22174394e-01 -3.06699812e-01 1.01003814e+00 6.90147519e-01 8.27410996e-01 7.15927720e-01 -1.12227654e+00 6.30755544e-01 2.57852107e-01 -4.19400871e-01 -8.22848380e-01 -1.31070718e-01 -3.05700719e-01 -6.28971517e-01 -2.42196232e-01 9.17845294e-02 5.52831367e-02 -7.58598924e-01 1.92706442e+00 1.34889618e-01 4.96802837e-01 -4.83983010e-02 7.65884817e-01 1.45572174e+00 9.39470589e-01 6.89982250e-02 -1.17750615e-01 1.36718750e+00 -1.17312264e+00 -9.23470080e-01 -3.46966743e-01 2.80106902e-01 -1.05847168e+00 1.64597428e+00 3.00941408e-01 -1.30240965e+00 -7.29625940e-01 -8.80177855e-01 -2.53391236e-01 2.03689009e-01 2.30846867e-01 1.69123277e-01 7.97467902e-02 -1.22289658e+00 3.46127391e-01 -9.29201365e-01 -2.14433342e-01 1.39177695e-01 3.13464910e-01 -1.63574517e-01 2.47384369e-01 -1.19336653e+00 7.07460105e-01 -2.06161421e-02 -2.66851522e-02 -1.46324396e+00 -9.57365632e-01 -1.16175020e+00 2.71639764e-01 3.04715425e-01 -7.11943388e-01 1.70763338e+00 -7.01206505e-01 -1.75073409e+00 7.35854983e-01 -5.26040792e-01 -4.00874734e-01 4.40668166e-01 -4.75498259e-01 -4.17558461e-01 3.62460732e-01 -3.79459970e-02 8.93929124e-01 1.08594382e+00 -1.16149020e+00 -4.67659682e-01 9.65141924e-04 6.89725904e-03 4.31217313e-01 -4.63361144e-01 -5.82025386e-03 -9.84030485e-01 -8.22863638e-01 -1.74630582e-01 -1.00506198e+00 2.35803306e-01 1.00973643e-01 -6.25330627e-01 -3.63687366e-01 9.54422951e-01 -5.72367668e-01 1.42946064e+00 -2.47748089e+00 2.37652451e-01 -4.47774440e-01 4.48842555e-01 2.91732937e-01 -4.82311606e-01 4.86645848e-01 -1.90340057e-01 9.91700962e-02 9.88785699e-02 -1.01246667e+00 1.51711136e-01 1.37274772e-01 -5.43452144e-01 3.26956004e-01 2.61130910e-02 8.09772134e-01 -6.82301581e-01 -6.60325706e-01 4.13889825e-01 7.99371600e-01 -9.58959877e-01 3.71018231e-01 -1.15259036e-01 3.35389763e-01 1.28858075e-01 6.53690577e-01 4.42939878e-01 -1.23524860e-01 1.25362650e-01 -6.38356090e-01 3.21430378e-02 6.54757738e-01 -6.35767519e-01 1.74354959e+00 -7.68304288e-01 1.20955408e+00 8.28804970e-02 -5.57144821e-01 4.57205623e-01 6.23653710e-01 3.76765668e-01 -9.01599288e-01 9.65621918e-02 -6.41673356e-02 2.37544049e-02 -4.69909877e-01 5.57152808e-01 -1.20970219e-01 -6.51305215e-03 2.36876816e-01 1.11226246e-01 -1.50483027e-01 -8.68142843e-02 4.08568352e-01 6.36380494e-01 -9.28711239e-03 -1.59403771e-01 -4.16269228e-02 3.75377804e-01 -6.68121338e-01 5.20861089e-01 3.35569531e-01 -2.13006943e-01 6.33420527e-01 6.07496321e-01 1.89539179e-01 -7.77969897e-01 -1.29792488e+00 9.07398537e-02 1.25462866e+00 2.96076745e-01 -7.07160056e-01 -6.55742764e-01 -5.21823525e-01 4.59572673e-02 7.71514952e-01 -5.63364625e-01 -2.86132038e-01 -5.55135071e-01 -3.12880099e-01 7.74701953e-01 5.58897614e-01 4.99422669e-01 -8.43911409e-01 -9.10094306e-02 2.08083913e-02 -7.61627078e-01 -1.35402799e+00 -1.21115410e+00 -2.00404912e-01 -6.91803694e-01 -8.28628421e-01 -7.41514921e-01 -7.01995552e-01 1.86163202e-01 4.43804085e-01 1.14192605e+00 -2.37835616e-01 7.73540810e-02 4.26714063e-01 -1.66342363e-01 -2.26458043e-01 -2.60443747e-01 1.38292171e-03 1.42180592e-01 -4.32345234e-02 1.87587082e-01 -7.35668063e-01 -5.54520547e-01 6.24222696e-01 -5.87503731e-01 -5.83709255e-02 3.91039789e-01 7.47378945e-01 5.22575855e-01 -2.59059876e-01 5.16440988e-01 -3.95923465e-01 5.42813241e-01 -3.40059638e-01 -4.00826156e-01 -6.90498725e-02 -5.54761648e-01 2.89986134e-02 3.40955585e-01 -9.20357287e-01 -8.83766294e-01 -2.80270100e-01 -2.91936189e-01 -1.06539702e+00 3.51409286e-01 4.43908393e-01 -3.46498072e-01 3.81359905e-01 1.64698228e-01 3.23059529e-01 2.32936606e-01 -5.99432528e-01 4.98943269e-01 6.36583209e-01 7.27520406e-01 -3.47706288e-01 5.79753816e-01 4.02347356e-01 -5.72380304e-01 -9.90215719e-01 -7.61783600e-01 -4.83599454e-01 -1.08242378e-01 -1.00107200e-01 9.12951231e-01 -1.28377092e+00 -1.13130295e+00 4.93733257e-01 -1.35114110e+00 -4.97545034e-01 -1.63012594e-01 6.54398978e-01 -4.07674134e-01 5.57213902e-01 -7.59225726e-01 -6.16208971e-01 -3.81985158e-01 -1.38418138e+00 1.17081571e+00 -2.94201858e-02 -3.00245643e-01 -1.06321502e+00 1.72688380e-01 5.09186029e-01 3.77350211e-01 -1.38328612e-01 6.29714847e-01 -3.55227768e-01 -3.05810630e-01 1.66190311e-01 -2.12467670e-01 4.09627616e-01 1.14745930e-01 3.13577950e-01 -1.37535119e+00 -3.97403359e-01 -9.35880244e-02 -2.30599418e-01 9.32686746e-01 5.83471775e-01 1.00309741e+00 -5.09100676e-01 -1.99072942e-01 5.88778496e-01 8.59997094e-01 3.18234593e-01 5.40259957e-01 -8.62625539e-02 8.70869040e-01 3.10384899e-01 3.91208529e-01 4.03962135e-01 7.79512823e-01 1.03708386e+00 8.39788556e-01 -3.19535464e-01 -5.91765046e-01 -6.58916414e-01 1.07814467e+00 1.43177116e+00 3.45026821e-01 -4.24011499e-01 -5.92166662e-01 6.99752688e-01 -1.88534951e+00 -1.01504612e+00 1.11345455e-01 2.01833248e+00 9.76561189e-01 8.32049996e-02 3.87962013e-01 -5.90869971e-03 6.60997510e-01 4.90293980e-01 -5.61807334e-01 -3.23139489e-01 -3.88274454e-02 -1.50762081e-01 -2.14422983e-03 6.02637410e-01 -1.00960243e+00 1.00314391e+00 5.88870239e+00 9.19708312e-01 -1.27312279e+00 2.84061283e-01 1.84550583e-01 -5.19472480e-01 -2.69898057e-01 -2.12207451e-01 -1.02143097e+00 6.26917899e-01 1.04637575e+00 -2.91388065e-01 4.07563746e-01 6.24132395e-01 5.18413007e-01 1.31900489e-01 -1.38823676e+00 1.36425829e+00 2.71141529e-01 -1.11936867e+00 2.32266575e-01 1.57732740e-01 3.57505381e-01 3.46104652e-01 6.23428464e-01 4.84453946e-01 3.19884598e-01 -9.80480790e-01 1.04279208e+00 9.48776454e-02 9.28234398e-01 -7.92462945e-01 6.14272892e-01 4.82146181e-02 -1.72021842e+00 8.62692595e-02 -1.28828064e-01 3.06448579e-01 3.90235692e-01 1.93011299e-01 -7.52052426e-01 2.32659236e-01 8.57778132e-01 1.01689422e+00 -3.69694918e-01 9.27940071e-01 -3.20086509e-01 1.02466846e+00 -2.96856284e-01 2.30084330e-01 4.13443118e-01 1.32854849e-01 6.53457344e-01 1.19167125e+00 2.62326390e-01 -4.10933524e-01 3.58255282e-02 6.33664966e-01 -3.07649612e-01 -1.56221101e-02 -4.78734195e-01 -2.58253455e-01 8.60287070e-01 9.97864485e-01 3.89827415e-02 -2.53487617e-01 -4.38175797e-01 8.72780561e-01 2.58203652e-02 3.74801278e-01 -1.24637032e+00 -2.62370765e-01 1.00441253e+00 1.23041645e-01 5.10502160e-01 -1.47557974e-01 7.37708881e-02 -1.14439857e+00 -1.83906406e-01 -1.18053579e+00 3.53364527e-01 -9.78324771e-01 -1.17759502e+00 8.39741588e-01 9.93616655e-02 -1.42858338e+00 -2.48960257e-01 -2.63363898e-01 -7.44927168e-01 7.00704455e-01 -1.61245024e+00 -1.24027252e+00 -2.45300919e-01 7.43168592e-01 7.92338908e-01 -2.57513732e-01 6.29923105e-01 6.05329752e-01 -7.57882774e-01 1.12706900e+00 -9.54088643e-02 4.52377871e-02 1.04024172e+00 -1.06343102e+00 4.44208741e-01 6.03774548e-01 2.76496470e-01 7.22575426e-01 7.23313093e-01 -1.65034220e-01 -1.29439306e+00 -1.06983840e+00 9.25861061e-01 -4.82134521e-01 8.91192973e-01 -4.04420257e-01 -8.01786005e-01 6.99068964e-01 7.94606745e-01 -1.46750957e-01 7.15890646e-01 1.67045817e-02 -6.81519866e-01 -2.90895194e-01 -7.40405321e-01 8.42752516e-01 6.93026960e-01 -1.08575416e+00 -5.30761719e-01 1.14752613e-01 1.22254395e+00 -4.01800126e-01 -8.66907656e-01 3.00582387e-02 6.74266934e-01 -6.70340538e-01 1.03397477e+00 -2.50213653e-01 5.00651419e-01 -2.84308642e-01 -2.81016707e-01 -1.32121527e+00 -5.63792177e-02 -1.08712220e+00 -3.56505841e-01 1.48224699e+00 4.50946808e-01 -6.53077483e-01 3.72825682e-01 7.82891959e-02 -3.05283308e-01 -5.42741239e-01 -1.31732500e+00 -8.89436722e-01 6.54067919e-02 -6.23926461e-01 4.48565930e-01 8.56841087e-01 1.53027430e-01 8.97953689e-01 -7.44247019e-01 4.54254895e-02 3.79433483e-01 -1.08644955e-01 8.00331116e-01 -9.75975454e-01 -3.84499520e-01 -5.93145788e-01 -5.07442579e-02 -1.53682435e+00 2.81887352e-01 -8.33468556e-01 -1.63696393e-01 -1.38520992e+00 1.92348719e-01 1.10655375e-01 -2.90866882e-01 5.96290648e-01 -3.42687815e-02 3.16241056e-01 5.81156313e-01 2.53362298e-01 -4.84849334e-01 1.07611680e+00 1.40622258e+00 -3.91500473e-01 -1.43743783e-01 -7.80675039e-02 -6.66651487e-01 7.63859808e-01 5.64713180e-01 -3.94183099e-01 -7.41605341e-01 -5.89782655e-01 -2.21530478e-02 1.74677879e-01 3.13653886e-01 -6.97208822e-01 2.49272898e-01 1.95817977e-01 -2.92627066e-01 -7.76416898e-01 8.69063139e-01 -6.73700035e-01 -7.09783658e-02 3.51817727e-01 -4.60792840e-01 4.93587852e-01 6.30941808e-01 5.75772822e-01 -4.46394026e-01 2.17358738e-01 8.46211851e-01 2.09938452e-01 -3.43357235e-01 3.24181885e-01 -4.94026065e-01 3.79495114e-01 6.49058461e-01 -4.80455607e-02 -4.74236339e-01 -8.67033005e-01 -7.09104776e-01 3.21077615e-01 2.00021967e-01 5.57089388e-01 7.09524751e-01 -1.47604167e+00 -6.96673453e-01 -1.05103068e-02 6.16143877e-03 -1.78271741e-01 3.54880273e-01 1.10882115e+00 -1.17918603e-01 4.11583245e-01 2.27131441e-01 -8.53252411e-01 -1.72759104e+00 4.00365323e-01 3.30830842e-01 -1.78041056e-01 -6.15029871e-01 1.04084194e+00 7.05572784e-01 -3.74567538e-01 6.59895062e-01 -6.26402438e-01 -1.59161389e-01 3.02021146e-01 3.89463365e-01 3.24937791e-01 -2.52338111e-01 -6.92000151e-01 -4.58937347e-01 7.03536570e-01 -3.63930255e-01 -4.51605320e-01 1.02382326e+00 -2.56019711e-01 1.81351036e-01 4.70307350e-01 1.40012348e+00 1.32232517e-01 -1.34603381e+00 -4.95039731e-01 -5.30407846e-01 -1.35315984e-01 3.29064906e-01 -7.01015353e-01 -1.35541773e+00 1.41440284e+00 4.83893037e-01 -1.21458285e-01 1.01613295e+00 1.64273560e-01 1.04596686e+00 8.01805779e-02 -2.34855324e-01 -9.14752662e-01 5.60299873e-01 5.80282569e-01 1.10114884e+00 -1.09021842e+00 -2.12618351e-01 -2.17327029e-01 -9.84201550e-01 7.65832484e-01 6.86909378e-01 3.60808223e-01 6.42758429e-01 2.65146583e-01 1.83884516e-01 -1.18296795e-01 -1.26302516e+00 -1.35632887e-01 5.06372333e-01 4.08140033e-01 5.43525279e-01 -2.36589253e-01 2.16941565e-01 5.45169890e-01 -3.53285819e-01 -4.02593940e-01 2.58428305e-01 4.07031387e-01 -5.63269034e-02 -7.10232794e-01 -5.23862876e-02 -2.70175450e-02 -4.52694833e-01 -5.65631270e-01 -3.56113940e-01 8.77853274e-01 -3.30612123e-01 1.25500226e+00 2.37080723e-01 -6.86086893e-01 4.58402216e-01 -3.56987208e-01 1.83698028e-01 -4.19758350e-01 -6.61669850e-01 7.73303628e-01 1.20652474e-01 -5.48478842e-01 -2.90836304e-01 -6.76474988e-01 -1.13957095e+00 -4.54402000e-01 -3.58270794e-01 1.49764404e-01 4.12333488e-01 7.27421701e-01 4.68153268e-01 8.24275076e-01 7.35459566e-01 -9.30594146e-01 -5.77346444e-01 -1.00983977e+00 -3.39698136e-01 4.54601310e-02 5.67323625e-01 -7.39433050e-01 -6.06801927e-01 1.74612217e-02]
[14.581631660461426, 5.155244827270508]
d18fa40a-f1ad-4b85-97f0-38c139a4f7c8
a-hierarchical-variational-neural-uncertainty-1
2110.03446
null
https://arxiv.org/abs/2110.03446v1
https://arxiv.org/pdf/2110.03446v1.pdf
A Hierarchical Variational Neural Uncertainty Model for Stochastic Video Prediction
Predicting the future frames of a video is a challenging task, in part due to the underlying stochastic real-world phenomena. Prior approaches to solve this task typically estimate a latent prior characterizing this stochasticity, however do not account for the predictive uncertainty of the (deep learning) model. Such approaches often derive the training signal from the mean-squared error (MSE) between the generated frame and the ground truth, which can lead to sub-optimal training, especially when the predictive uncertainty is high. Towards this end, we introduce Neural Uncertainty Quantifier (NUQ) - a stochastic quantification of the model's predictive uncertainty, and use it to weigh the MSE loss. We propose a hierarchical, variational framework to derive NUQ in a principled manner using a deep, Bayesian graphical model. Our experiments on four benchmark stochastic video prediction datasets show that our proposed framework trains more effectively compared to the state-of-the-art models (especially when the training sets are small), while demonstrating better video generation quality and diversity against several evaluation metrics.
['Anoop Cherian', 'Narendra Ahuja', 'Moitreya Chatterjee']
2021-10-06
a-hierarchical-variational-neural-uncertainty
http://openaccess.thecvf.com//content/ICCV2021/html/Chatterjee_A_Hierarchical_Variational_Neural_Uncertainty_Model_for_Stochastic_Video_Prediction_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Chatterjee_A_Hierarchical_Variational_Neural_Uncertainty_Model_for_Stochastic_Video_Prediction_ICCV_2021_paper.pdf
iccv-2021-1
['video-prediction']
['computer-vision']
[ 2.88252890e-01 1.21234238e-01 -3.52808982e-02 -4.21723366e-01 -9.95840251e-01 -2.98313737e-01 7.29452372e-01 -1.29309535e-01 -9.92587209e-02 9.99559224e-01 2.27484137e-01 -1.47924637e-02 -1.37722567e-01 -6.30790710e-01 -1.07595170e+00 -9.11847234e-01 -2.73006931e-02 4.13001865e-01 3.25049698e-01 2.45356098e-01 9.39739197e-02 -7.59094954e-02 -1.48150253e+00 1.07347168e-01 1.05000257e+00 1.38450384e+00 3.39133888e-01 5.10985076e-01 2.49504894e-01 1.05726314e+00 -4.36822414e-01 -8.03914249e-01 4.81183007e-02 -5.58866620e-01 -3.42969179e-01 5.13996929e-02 3.58028382e-01 -3.44732434e-01 -3.08089405e-01 1.27771902e+00 1.68525144e-01 1.29572883e-01 1.03015780e+00 -1.14392102e+00 -2.05199420e-01 6.85829520e-01 -3.11036140e-01 1.82837218e-01 1.37325600e-01 2.37004131e-01 1.14868009e+00 -7.39743173e-01 4.42889214e-01 1.23375511e+00 7.58562922e-01 4.25343841e-01 -1.42754948e+00 -6.03439331e-01 1.30939826e-01 2.47839406e-01 -1.37290537e+00 -5.24003625e-01 7.18982518e-01 -7.96613514e-01 3.24417293e-01 -4.70720828e-02 4.30795461e-01 1.41904497e+00 4.25698280e-01 8.44925463e-01 7.80865788e-01 -4.30642255e-02 4.80740219e-01 1.43140122e-01 -2.70834655e-01 5.84327281e-01 1.57860234e-01 1.21646084e-01 -7.44897306e-01 -2.52717063e-02 5.87074876e-01 -2.30828464e-01 -4.69764352e-01 -4.52405453e-01 -9.46286619e-01 6.78726614e-01 7.31845647e-02 -5.35922088e-02 -5.63536227e-01 5.04609466e-01 2.00768366e-01 -8.79279375e-02 6.01742864e-01 1.15600161e-01 -1.82523414e-01 -5.05020142e-01 -1.37157893e+00 4.62964803e-01 7.25013673e-01 7.38499463e-01 5.49109995e-01 1.54131711e-01 -6.17550194e-01 5.63665032e-01 6.15260601e-01 3.33597928e-01 9.53386575e-02 -1.25773025e+00 5.68211854e-01 -1.11885117e-02 3.39222282e-01 -1.04887843e+00 1.10009409e-01 -6.70160830e-01 -9.25161064e-01 2.17951730e-01 5.14306545e-01 -2.76629388e-01 -5.09774208e-01 2.00317502e+00 4.95439172e-02 7.10862398e-01 -1.53940007e-01 9.11993563e-01 3.67836803e-01 8.16482544e-01 1.28512517e-01 -3.97689521e-01 8.13301861e-01 -6.92072272e-01 -5.62787890e-01 -1.17789574e-01 1.47404030e-01 -4.27163780e-01 7.13699222e-01 5.82582116e-01 -1.14891064e+00 -5.94734192e-01 -1.07965052e+00 3.98254603e-01 4.19221014e-01 8.43219012e-02 2.31457174e-01 7.14645922e-01 -8.99770379e-01 1.07552576e+00 -1.09100795e+00 1.36321589e-01 5.31396508e-01 4.02890630e-02 7.45511875e-02 9.36326459e-02 -1.19330049e+00 6.36298597e-01 4.08392549e-01 1.54753178e-01 -1.31316411e+00 -9.04824495e-01 -6.29271805e-01 2.89589703e-01 5.27620733e-01 -8.58552337e-01 1.02167594e+00 -8.76510859e-01 -1.71076584e+00 3.23457837e-01 -8.82366970e-02 -9.05583382e-01 1.03011823e+00 -4.31445181e-01 -8.93646851e-03 6.42500371e-02 -1.47652552e-01 5.64856768e-01 1.25918972e+00 -1.37836730e+00 -5.72603226e-01 1.36387488e-02 9.18225944e-02 -1.77450199e-02 -8.38561133e-02 -3.59571040e-01 -4.87959027e-01 -7.80278683e-01 -1.58535272e-01 -9.83304858e-01 -1.79176018e-01 8.30108523e-02 -3.51698369e-01 -7.48885572e-02 5.00943005e-01 -6.84732318e-01 1.30085766e+00 -2.02428389e+00 4.14443165e-01 -3.27423750e-03 2.57362694e-01 1.32406056e-01 3.48292291e-01 4.51872684e-02 3.25000823e-01 1.14199467e-01 -4.25842494e-01 -6.70001030e-01 2.14700252e-01 1.72257200e-01 -5.24663687e-01 3.83912414e-01 3.23261112e-01 6.10971332e-01 -1.04415298e+00 -5.80075204e-01 2.19800606e-01 7.28346348e-01 -6.92386806e-01 3.22843075e-01 -6.24528229e-01 5.60392201e-01 -5.16564190e-01 1.34449139e-01 4.96496290e-01 -3.64028871e-01 1.11256279e-01 -2.37383306e-01 2.62076676e-01 7.02323616e-02 -1.25950634e+00 1.60256839e+00 -4.76635367e-01 8.95128012e-01 -2.23317370e-01 -8.37758720e-01 7.30660856e-01 2.88713932e-01 3.39680851e-01 1.28485551e-02 1.13178201e-01 8.73740390e-02 -2.55566448e-01 -3.33376706e-01 4.46568221e-01 -1.74625546e-01 2.85384178e-01 1.62049428e-01 2.14347228e-01 7.43852034e-02 2.85970747e-01 2.00260386e-01 9.13572848e-01 6.34908497e-01 1.29906252e-01 -2.33712554e-01 5.20186961e-01 -5.67812681e-01 9.16558981e-01 9.42886591e-01 -1.32843956e-01 9.36842859e-01 9.12062824e-01 -1.11385457e-01 -9.85593379e-01 -1.25579083e+00 -8.52870941e-02 5.37318110e-01 -2.88848635e-02 -4.75389034e-01 -9.82127607e-01 -5.72882652e-01 -1.53385192e-01 9.79640067e-01 -4.82126713e-01 -1.21736571e-01 -1.41902417e-01 -7.11003482e-01 2.70753086e-01 3.37930530e-01 2.41182074e-01 -5.88059187e-01 -7.13713229e-01 3.95632654e-01 -3.87886524e-01 -1.41261399e+00 -2.39454538e-01 -2.25514889e-01 -7.82098651e-01 -7.13829994e-01 -8.08027148e-01 8.31273943e-02 2.14988202e-01 -2.91526496e-01 1.35956216e+00 -2.10992798e-01 1.99704826e-01 2.51992643e-01 -2.79251665e-01 -1.71160415e-01 -6.43125236e-01 -2.98934489e-01 1.02672011e-01 3.89373988e-01 -3.04821748e-02 -6.33207917e-01 -6.78896904e-01 2.00524896e-01 -8.24064493e-01 2.76524782e-01 4.53051239e-01 8.54403198e-01 6.31917834e-01 2.82154202e-01 2.94781566e-01 -6.79740667e-01 3.34929109e-01 -6.51385188e-01 -8.83017421e-01 2.35015854e-01 -5.40656149e-01 3.99630666e-01 5.97246587e-01 -2.02063754e-01 -1.28105581e+00 -1.60148203e-01 3.98902446e-02 -7.77475536e-01 -3.51785659e-03 5.57326317e-01 -9.90758613e-02 2.12806925e-01 4.56213683e-01 2.75641233e-01 -2.42332131e-01 -3.00832927e-01 8.33903477e-02 3.09729218e-01 5.17596126e-01 -7.84402728e-01 4.91357416e-01 4.94363099e-01 4.41915989e-01 -5.63154638e-01 -1.08447874e+00 -1.59246288e-02 -4.14568573e-01 -6.28331184e-01 7.94236004e-01 -1.01482534e+00 -6.03669286e-01 3.86857510e-01 -1.21889389e+00 -2.29319453e-01 -1.98820382e-01 7.11802602e-01 -8.53688121e-01 4.20313150e-01 -3.39268118e-01 -1.27930856e+00 8.44726712e-03 -1.42222118e+00 1.16526139e+00 7.30526075e-02 -1.47940934e-01 -9.85134184e-01 -3.70916314e-02 2.37676874e-01 2.82836854e-01 5.38560510e-01 8.14399481e-01 -2.32692108e-01 -8.60906959e-01 -3.62151004e-02 -2.00374022e-01 6.58267796e-01 -3.70436609e-01 3.09318244e-01 -1.03849840e+00 3.79372612e-02 7.29963481e-02 -1.65458068e-01 1.06342733e+00 7.47788787e-01 1.21119940e+00 -2.92702764e-01 -7.01355413e-02 4.47691530e-01 1.50156891e+00 -1.32747427e-01 6.99533522e-01 -8.88211057e-02 5.49738288e-01 4.86700982e-01 6.11986279e-01 9.62850809e-01 3.39356303e-01 7.86083043e-01 6.98326170e-01 7.16155410e-01 6.20291336e-03 -3.94039631e-01 7.25414872e-01 6.66456878e-01 -1.28546402e-01 -6.29106879e-01 -8.06233168e-01 3.01171809e-01 -2.12793326e+00 -1.20709169e+00 -4.87589762e-02 2.38613105e+00 7.68147945e-01 4.89219040e-01 -3.72469090e-02 1.07666917e-01 6.01944745e-01 3.43019694e-01 -5.53451657e-01 1.39162630e-01 -5.20322807e-02 -2.30067521e-01 3.60620975e-01 4.92440224e-01 -1.10421717e+00 6.48831844e-01 5.80419636e+00 1.14822304e+00 -9.40164626e-01 1.19287744e-01 1.03679168e+00 -1.91383108e-01 -3.33867431e-01 -7.38387853e-02 -7.97239900e-01 9.02425468e-01 1.24205554e+00 -1.49953917e-01 2.86362559e-01 6.89351976e-01 3.71599764e-01 -1.82639658e-01 -1.50745177e+00 1.08126581e+00 -4.18517254e-02 -1.48317468e+00 1.45748645e-01 9.65768024e-02 8.65737319e-01 1.53724486e-02 3.23725879e-01 2.77784139e-01 2.77870029e-01 -9.65414882e-01 1.21152401e+00 1.17605615e+00 4.65865076e-01 -7.11646914e-01 7.48634040e-01 4.92018789e-01 -8.81712437e-01 3.61376926e-02 -4.12587255e-01 1.26609862e-01 3.82394403e-01 1.10814726e+00 -4.47914779e-01 3.81226867e-01 7.49441147e-01 8.22487891e-01 -3.51931989e-01 1.05929828e+00 -2.02597752e-01 8.82655382e-01 -3.76272738e-01 -6.90896157e-03 1.85844168e-01 -5.25541127e-01 9.24303591e-01 1.15207160e+00 6.26631141e-01 -2.15705931e-01 -1.28941059e-01 1.30795240e+00 -1.17720850e-01 -2.87382036e-01 -1.79097995e-01 -4.66287509e-02 1.85379103e-01 8.78626049e-01 -4.81969625e-01 -2.46666059e-01 -1.53950244e-01 9.10836816e-01 1.63117200e-01 4.89991456e-01 -1.16758871e+00 1.64391831e-01 7.05236733e-01 -5.00629432e-02 4.28758085e-01 -1.46938831e-01 -1.77484900e-01 -1.21331918e+00 3.85709703e-01 -7.62984514e-01 5.78216836e-02 -8.73262286e-01 -1.23381066e+00 5.72996676e-01 9.90065634e-02 -1.51259339e+00 -6.81406796e-01 -4.88078564e-01 -4.35471535e-01 7.33645260e-01 -1.32223535e+00 -7.88602054e-01 -1.52950183e-01 1.17572136e-01 5.97527504e-01 -2.45849177e-01 4.10525888e-01 1.78361908e-01 -5.18929541e-01 5.03953636e-01 4.19040173e-01 -2.15667486e-01 3.90443772e-01 -1.21485865e+00 1.17330059e-01 1.05263722e+00 2.82197356e-01 2.32189104e-01 1.36205637e+00 -3.96410108e-01 -1.15064895e+00 -1.17040563e+00 4.96100515e-01 -6.19979322e-01 8.17590535e-01 -2.05312863e-01 -8.35330427e-01 2.86337107e-01 -1.04389369e-01 1.90794751e-01 5.01234591e-01 -9.13368613e-02 -2.08846286e-01 -8.85720402e-02 -1.07262444e+00 6.13458335e-01 9.30017531e-01 -4.29340899e-01 -1.50135592e-01 1.63886145e-01 7.33683288e-01 -3.04415017e-01 -9.04001296e-01 3.70237738e-01 7.59701550e-01 -1.48132813e+00 7.51306832e-01 -3.45874846e-01 1.01120508e+00 -2.80618072e-01 -3.79619986e-01 -1.30926836e+00 -2.20446542e-01 -6.94134057e-01 -6.72001421e-01 1.34588671e+00 3.88776004e-01 -1.24629855e-01 9.33577895e-01 6.90298915e-01 9.08880681e-02 -9.19921279e-01 -1.14222836e+00 -8.00558448e-01 -1.50929376e-01 -9.35529888e-01 4.85167801e-01 2.70478666e-01 -4.68650818e-01 -2.15720125e-02 -7.27001965e-01 2.79142827e-01 1.05898929e+00 -2.53614902e-01 5.43448389e-01 -1.22856057e+00 -6.88873947e-01 -5.81318378e-01 -8.13224554e-01 -1.19072664e+00 3.49814713e-01 -4.11492229e-01 3.24065357e-01 -1.21783245e+00 2.86116332e-01 -2.02770844e-01 -3.69834930e-01 -3.13604444e-01 -3.27421755e-01 -1.79353673e-02 3.46720099e-01 2.62756407e-01 -8.31003964e-01 9.71476674e-01 9.22528028e-01 -8.40972513e-02 1.55137956e-01 1.18369028e-01 -3.62687111e-01 7.82417059e-01 4.38190490e-01 -6.15045726e-01 -5.16191304e-01 -3.44579786e-01 4.61492777e-01 2.90099680e-01 5.78861475e-01 -1.40986836e+00 1.21127374e-01 -8.81386995e-02 2.15673938e-01 -4.20649797e-01 4.63649720e-01 -8.64511788e-01 3.91797781e-01 1.90829515e-01 -5.30605912e-01 -5.06095231e-01 -1.42914221e-01 1.10213387e+00 -2.75892973e-01 -3.51530522e-01 8.33151042e-01 1.80729181e-01 -4.77358609e-01 4.67734933e-01 -2.05364540e-01 1.74517497e-01 8.90036166e-01 2.58967225e-02 1.06516704e-01 -8.02067637e-01 -6.85313344e-01 1.14007220e-01 2.55887210e-01 2.87155211e-01 5.47848642e-01 -1.28140378e+00 -8.76855910e-01 -1.79076731e-01 -7.00529069e-02 -1.15914211e-01 3.28109026e-01 8.21819425e-01 -3.59710187e-01 3.59707206e-01 2.06545800e-01 -9.05373991e-01 -8.42084527e-01 2.63441294e-01 4.35852766e-01 -5.11563540e-01 -3.58171880e-01 9.57908571e-01 4.35873330e-01 2.47092262e-01 4.05192971e-01 -3.49739432e-01 -9.61881354e-02 -3.07241306e-02 4.20297980e-01 5.12838244e-01 -2.38633782e-01 -7.42483497e-01 -1.22112289e-01 3.39199722e-01 2.32950926e-01 -2.47827664e-01 1.11169374e+00 -3.27096820e-01 3.07268679e-01 6.72568262e-01 1.12995601e+00 -2.85859913e-01 -2.07593179e+00 -3.78352523e-01 -8.53147656e-02 -5.20656288e-01 2.65796125e-01 -3.73705804e-01 -1.08446920e+00 1.00006342e+00 6.93762958e-01 1.23425022e-01 8.62163484e-01 -2.04640523e-01 6.96711957e-01 1.92388728e-01 4.72091526e-01 -1.02642965e+00 -1.58740923e-01 2.84575850e-01 9.01285350e-01 -1.41393495e+00 1.03789583e-01 -2.94623911e-01 -7.84555435e-01 1.07846022e+00 2.35206574e-01 -1.96841314e-01 8.64044249e-01 2.35526264e-02 -3.88742328e-01 2.42271930e-01 -9.85431910e-01 1.89133957e-01 4.45469886e-01 4.28060174e-01 3.67444932e-01 -2.83051264e-02 1.17378764e-01 8.07090819e-01 -7.48606473e-02 2.77148664e-01 3.81937712e-01 3.80335987e-01 -2.72033006e-01 -6.52484000e-01 -1.32117659e-01 4.66908813e-01 -7.62078106e-01 -9.62632969e-02 2.71391779e-01 2.32311070e-01 8.52694586e-02 9.58956301e-01 5.01203276e-02 -4.19646382e-01 -1.03756569e-01 -1.32210195e-01 4.31003541e-01 -2.17622489e-01 -5.95642589e-02 6.77633807e-02 -4.18028161e-02 -6.42706454e-01 -5.78521967e-01 -9.49268579e-01 -7.67493904e-01 -2.70762354e-01 -2.12879911e-01 1.06413417e-01 5.29166162e-01 1.06620657e+00 2.54307806e-01 5.88194251e-01 5.26006520e-01 -1.08083892e+00 -8.73108447e-01 -7.51792252e-01 -5.42301059e-01 4.39648688e-01 3.10119033e-01 -8.91613126e-01 -4.90542680e-01 3.76282036e-01]
[7.137290954589844, 3.7200658321380615]
a4cf25ff-1e89-4ba6-a751-ec58ae4955d7
prototypical-classifier-for-robust-class
2110.11553
null
https://arxiv.org/abs/2110.11553v1
https://arxiv.org/pdf/2110.11553v1.pdf
Prototypical Classifier for Robust Class-Imbalanced Learning
Deep neural networks have been shown to be very powerful methods for many supervised learning tasks. However, they can also easily overfit to training set biases, i.e., label noise and class imbalance. While both learning with noisy labels and class-imbalanced learning have received tremendous attention, existing works mainly focus on one of these two training set biases. To fill the gap, we propose \textit{Prototypical Classifier}, which does not require fitting additional parameters given the embedding network. Unlike conventional classifiers that are biased towards head classes, Prototypical Classifier produces balanced and comparable predictions for all classes even though the training set is class-imbalanced. By leveraging this appealing property, we can easily detect noisy labels by thresholding the confidence scores predicted by Prototypical Classifier, where the threshold is dynamically adjusted through the iteration. A sample reweghting strategy is then applied to mitigate the influence of noisy labels. We test our method on CIFAR-10-LT, CIFAR-100-LT and Webvision datasets, observing that Prototypical Classifier obtains substaintial improvements compared with state of the arts.
['Min-Ling Zhang', 'Yu-Feng Li', 'Jiang-Xin Shi', 'Tong Wei']
2021-10-22
null
null
null
null
['learning-with-noisy-labels', 'learning-with-noisy-labels']
['computer-vision', 'natural-language-processing']
[ 1.09074682e-01 1.72010750e-01 -4.95796859e-01 -7.58058786e-01 -6.36888802e-01 -3.23857874e-01 3.35256517e-01 3.10348839e-01 -4.95108187e-01 6.71777248e-01 -4.47747745e-02 -9.14231986e-02 -1.48874596e-01 -6.85336173e-01 -4.86004889e-01 -8.26403975e-01 2.82320380e-01 4.23121095e-01 8.79301038e-03 7.76528716e-02 1.17279567e-01 1.07340254e-01 -1.60301459e+00 2.73077130e-01 7.49692678e-01 1.30556071e+00 -3.68731767e-01 2.76054084e-01 -1.55422434e-01 8.49705994e-01 -7.32373297e-01 -4.67066973e-01 1.76268771e-01 -2.73984462e-01 -5.80931306e-01 6.72167763e-02 7.65667498e-01 -9.65314135e-02 -3.06089103e-01 1.15600121e+00 8.16461682e-01 1.33733195e-03 7.50321031e-01 -1.41730726e+00 -3.98923010e-01 7.77378559e-01 -6.74972892e-01 3.70450437e-01 -3.28884065e-01 2.83819269e-02 1.22954679e+00 -1.03997576e+00 2.47818604e-01 1.26081836e+00 8.40579093e-01 5.10103226e-01 -1.34483075e+00 -1.14007711e+00 3.71358156e-01 2.48054698e-01 -1.14369476e+00 -4.19443578e-01 9.30355549e-01 -4.43032473e-01 4.20402735e-01 2.03894570e-01 3.67013782e-01 1.23779857e+00 -3.07957325e-02 7.70069480e-01 1.12180424e+00 -4.62596029e-01 4.11614478e-01 4.41018462e-01 7.71475255e-01 3.20088148e-01 4.30775851e-01 1.47682995e-01 -5.26761174e-01 -2.25627944e-01 8.79105479e-02 -7.82133043e-02 -2.43508577e-01 -3.17236602e-01 -1.00581121e+00 8.50039124e-01 5.91640532e-01 1.24399491e-01 -8.48626569e-02 1.59019276e-01 7.30693877e-01 4.34195429e-01 6.98687613e-01 3.71244371e-01 -4.22462493e-01 1.36407971e-01 -8.12816501e-01 2.09705904e-01 6.49237633e-01 6.20166004e-01 7.04946876e-01 7.10867643e-02 -4.24540430e-01 1.29657531e+00 2.52653956e-01 2.61961550e-01 7.28288651e-01 -6.51579261e-01 4.08672184e-01 6.89932585e-01 -7.66752809e-02 -1.12461722e+00 -6.39686286e-01 -1.05013311e+00 -1.18623793e+00 1.88685954e-01 4.78736669e-01 -1.10949852e-01 -9.48983490e-01 1.71470094e+00 4.29950684e-01 2.77061574e-02 -9.18287262e-02 8.97574365e-01 8.76381695e-01 3.85579020e-01 1.06848203e-01 8.45857412e-02 1.13494229e+00 -9.73951578e-01 -7.92998135e-01 -3.22155297e-01 9.03055787e-01 -4.12622988e-01 1.15949845e+00 5.23893476e-01 -4.94659156e-01 -5.92553675e-01 -1.24143660e+00 1.37729093e-01 -2.25389451e-01 2.61331826e-01 3.32945436e-01 8.32242489e-01 -6.77734911e-01 7.29617715e-01 -4.97416079e-01 1.68225646e-01 8.69162798e-01 2.68064946e-01 -2.52130181e-01 -2.07137331e-01 -1.17681623e+00 6.24588132e-01 5.22220314e-01 7.72578716e-02 -7.34720826e-01 -6.10506296e-01 -6.37942314e-01 1.32721901e-01 3.05211693e-01 -1.41865596e-01 1.31392562e+00 -1.07223606e+00 -1.21477151e+00 9.09758389e-01 2.07613349e-01 -5.02904773e-01 7.67311037e-01 -8.50852057e-02 -3.56636286e-01 -3.69076580e-01 -8.35411400e-02 6.81983590e-01 8.37245107e-01 -1.25766623e+00 -6.32537186e-01 -6.28526866e-01 -3.32453437e-02 1.13550596e-01 -8.75938833e-01 -2.25261867e-01 -1.99827161e-02 -6.22380018e-01 4.90015149e-01 -7.59264469e-01 -1.18010856e-01 2.48672348e-02 -7.65253603e-01 -5.14596939e-01 8.87774765e-01 -1.06430635e-01 1.28667390e+00 -2.28366160e+00 -4.36348259e-01 2.05467895e-01 4.94004220e-01 2.79091328e-01 -4.61972132e-02 2.31898930e-02 -3.63843173e-01 2.58308232e-01 7.66793415e-02 -3.86416107e-01 2.42933780e-01 2.40035981e-01 -2.24786207e-01 6.98345482e-01 1.97635904e-01 4.70875531e-01 -8.65424275e-01 -2.79118776e-01 1.71252917e-02 2.30327293e-01 -5.14944255e-01 6.99910056e-03 -1.11569483e-02 2.16674861e-02 -1.68994814e-01 5.52801192e-01 7.45157301e-01 -3.21363419e-01 1.31030485e-01 -3.62834752e-01 3.41241896e-01 3.78381699e-01 -1.38554180e+00 7.88140178e-01 -4.83408153e-01 5.36617279e-01 -2.78865010e-01 -1.34433973e+00 1.21842539e+00 2.21384302e-01 1.25687614e-01 -6.91701174e-01 3.51984143e-01 3.23502213e-01 2.71914244e-01 -5.17803013e-01 1.82828844e-01 -2.45099247e-01 4.34652865e-02 3.22369695e-01 7.95095880e-03 2.06142679e-01 6.84034303e-02 -9.55734774e-02 8.91647935e-01 -3.09208274e-01 1.47164628e-01 -2.18441725e-01 1.22854821e-01 -1.42344445e-01 8.28675985e-01 9.71997797e-01 -5.16310215e-01 5.13623893e-01 6.43369198e-01 -6.90373361e-01 -8.89867008e-01 -6.34249151e-01 -5.30243874e-01 1.34727383e+00 1.18190654e-01 -7.80932754e-02 -5.73338151e-01 -1.09290683e+00 6.23078123e-02 5.81357598e-01 -7.28079557e-01 -3.81138504e-01 -2.79298931e-01 -1.36145127e+00 5.91669738e-01 6.09818697e-01 4.74518836e-01 -9.04453993e-01 -4.92980212e-01 1.60635591e-01 -6.39367104e-03 -9.34285879e-01 -7.37497807e-02 6.32649779e-01 -8.10478270e-01 -1.09144545e+00 -3.66805822e-01 -7.70641923e-01 6.75276041e-01 1.06323339e-01 1.41920328e+00 2.51678973e-01 -6.33037165e-02 -2.61887163e-01 -3.02891105e-01 -6.56492114e-01 -3.23051244e-01 2.20429406e-01 2.37764373e-01 1.48261204e-01 6.12359166e-01 -4.59815085e-01 -5.88154078e-01 4.96362358e-01 -9.05808985e-01 -1.09990671e-01 4.05978709e-01 1.30666482e+00 5.33196330e-01 3.14590156e-01 1.10423183e+00 -1.20560479e+00 4.32486534e-01 -5.35642505e-01 -5.28397977e-01 -1.24930993e-01 -8.26854467e-01 -7.21331388e-02 7.46372938e-01 -7.09044635e-01 -6.34714544e-01 -5.97387590e-02 -3.95764619e-01 -2.39430860e-01 -2.42061526e-01 4.33384418e-01 -2.60722458e-01 1.86266467e-01 9.54115629e-01 -2.54209757e-01 -2.96665817e-01 -4.64295894e-01 -4.79521155e-02 1.11735928e+00 2.73491085e-01 -3.73799294e-01 5.21440327e-01 4.19398665e-01 -3.14029485e-01 -4.35209781e-01 -1.44369256e+00 -4.32185799e-01 -3.16931039e-01 -1.79741606e-01 2.25173697e-01 -8.62363100e-01 -4.96360540e-01 5.52360475e-01 -8.09488237e-01 -2.53511727e-01 -3.15320134e-01 3.98127973e-01 -1.42768353e-01 -2.54174750e-02 -5.15363753e-01 -7.60418773e-01 -2.97960937e-01 -1.11820257e+00 8.45816374e-01 1.77642852e-01 -3.38910699e-01 -7.86635458e-01 -5.74585125e-02 3.36860120e-01 2.64807820e-01 2.31076509e-01 1.10126567e+00 -1.22574282e+00 -9.97076556e-02 -6.01982713e-01 -4.01514351e-01 7.11623251e-01 5.32929227e-03 -1.04041055e-01 -1.47560942e+00 -4.34723467e-01 -8.85736868e-02 -8.11711013e-01 9.56384659e-01 1.69397175e-01 1.62484300e+00 -3.83080751e-01 -1.97537586e-01 4.01734859e-01 1.43454528e+00 4.74372283e-02 3.78648520e-01 4.15661335e-01 6.38920367e-01 4.68699038e-01 4.66749310e-01 3.39822680e-01 2.98004150e-01 7.51677692e-01 6.73183978e-01 -1.61383390e-01 -3.33011337e-02 -5.78898787e-02 2.54825652e-02 6.74848676e-01 5.21892607e-01 -3.95169377e-01 -1.01839638e+00 5.56883216e-01 -1.57515001e+00 -5.12750685e-01 -9.73817781e-02 2.14187360e+00 9.53447938e-01 5.65946281e-01 -4.14297245e-02 8.47294033e-01 6.84958518e-01 1.75336301e-01 -8.21124256e-01 -1.58233404e-01 -1.15644485e-01 -1.09903105e-01 4.17932510e-01 2.01619342e-01 -1.24812520e+00 5.70487857e-01 5.65195560e+00 9.96865511e-01 -1.22478473e+00 2.92923510e-01 1.31443810e+00 -1.83443889e-01 -1.63855657e-01 -4.19637501e-01 -1.00328088e+00 5.86399436e-01 7.51301885e-01 3.14293683e-01 -1.78665787e-01 9.90491152e-01 6.19740300e-02 5.21257073e-02 -1.23566902e+00 1.05489826e+00 5.52142411e-02 -1.07021308e+00 -3.87710541e-01 -6.45744726e-02 8.03276658e-01 1.01657882e-01 2.65350074e-01 7.52479613e-01 4.32277769e-01 -1.06406915e+00 7.47562408e-01 -3.07262372e-02 5.75474501e-01 -8.43904257e-01 1.21144485e+00 5.58499694e-01 -3.42045873e-01 -4.37458009e-01 -5.27190924e-01 -1.09698482e-01 -3.65983576e-01 1.40442157e+00 -6.39995635e-01 8.81287158e-02 9.37580109e-01 5.49863577e-01 -6.60669863e-01 1.18154359e+00 -8.59892890e-02 1.09162641e+00 -3.47367793e-01 1.72976833e-02 1.45099625e-01 1.86329424e-01 2.41954252e-01 1.02370548e+00 1.62022058e-02 -3.98345709e-01 3.02963197e-01 5.08062720e-01 -5.24153352e-01 8.51100013e-02 -4.61502016e-01 2.57721007e-01 6.04439557e-01 1.19242620e+00 -6.39531016e-01 -5.37764847e-01 -1.03513062e-01 5.31456351e-01 5.10079265e-01 2.67748326e-01 -7.77729928e-01 -3.31937551e-01 3.76459658e-01 4.20668162e-02 1.12106204e-01 3.76420677e-01 -7.14748740e-01 -1.03961682e+00 1.09140798e-01 -1.04785454e+00 3.67719382e-01 -2.59983212e-01 -1.50599194e+00 7.38698661e-01 -3.68356287e-01 -1.23087442e+00 6.33987486e-02 -6.11196876e-01 -3.68303180e-01 4.77756888e-01 -1.63877606e+00 -6.97129011e-01 -4.95882988e-01 7.33196065e-02 4.15186524e-01 -9.09952261e-03 6.38485312e-01 5.97594202e-01 -9.99543726e-01 9.88626361e-01 3.74658495e-01 3.35846275e-01 8.95938098e-01 -1.25192726e+00 1.19884290e-01 4.71278757e-01 1.69230893e-01 1.30123049e-02 6.82081282e-01 -2.22228810e-01 -6.80190384e-01 -1.33429515e+00 9.20021117e-01 -2.55939722e-01 4.66840714e-01 -5.74469805e-01 -9.95573163e-01 4.12794352e-01 -9.80333388e-02 5.76756239e-01 6.81003749e-01 3.61944288e-01 -7.12971389e-01 -4.95053947e-01 -1.14590311e+00 4.44006830e-01 9.23934996e-01 -1.94822282e-01 -2.54019260e-01 6.53834105e-01 4.93719101e-01 -5.39126217e-01 -6.74133837e-01 7.31388032e-01 4.96647000e-01 -1.16631138e+00 7.11192846e-01 -6.59148693e-01 4.77884114e-01 -7.17998520e-02 -2.76426166e-01 -1.57689226e+00 -2.68349528e-01 6.44665118e-03 -6.19757622e-02 1.31180120e+00 3.77026290e-01 -5.55553615e-01 9.21700180e-01 1.99601516e-01 -2.70660296e-02 -1.07547748e+00 -9.50662434e-01 -7.58157969e-01 7.49012157e-02 -4.41312492e-01 5.52932203e-01 1.11656761e+00 -3.53621274e-01 4.61638063e-01 -4.07821774e-01 1.25903979e-01 5.73062420e-01 -7.60936737e-02 7.02249348e-01 -1.53137076e+00 -1.88075840e-01 -3.67281973e-01 -4.31650162e-01 -7.11718678e-01 2.42911741e-01 -8.33925605e-01 3.17354947e-01 -1.01302385e+00 7.32535645e-02 -1.02865291e+00 -6.86084092e-01 7.05184996e-01 -3.18003565e-01 7.79331028e-01 -3.04960348e-02 1.45962745e-01 -7.22357094e-01 6.35116518e-01 1.00340652e+00 -3.92237246e-01 6.94116950e-02 1.72654927e-01 -8.95168364e-01 8.60217750e-01 9.24452364e-01 -8.44220877e-01 -3.53483230e-01 -2.63569951e-01 4.31201428e-01 -5.52430391e-01 3.55441988e-01 -1.16033816e+00 -9.04342309e-02 7.70737082e-02 4.72272784e-01 -4.85542268e-01 1.20875135e-01 -8.12703431e-01 -2.72654414e-01 3.24755073e-01 -8.44139278e-01 -2.11995587e-01 7.47373551e-02 6.87137127e-01 -3.05693656e-01 -4.66525078e-01 1.12974930e+00 1.77538574e-01 -3.13861996e-01 1.77947372e-01 -2.36819685e-01 4.06980067e-01 7.26483285e-01 -1.02701798e-01 -4.10546839e-01 -2.55253047e-01 -5.74955046e-01 2.80413032e-01 7.66279325e-02 4.22513485e-01 3.51405501e-01 -1.42033637e+00 -7.19749510e-01 2.28157550e-01 4.02904272e-01 1.92842305e-01 1.71320856e-01 8.03198457e-01 -7.66478404e-02 6.74744397e-02 1.91757351e-01 -9.24835086e-01 -1.12008858e+00 2.70450085e-01 5.18096507e-01 -4.09754068e-01 -3.69035810e-01 1.02995956e+00 1.46331847e-01 -8.71982276e-01 7.04171419e-01 -3.39346915e-01 -2.91957945e-01 3.19596350e-01 6.29977405e-01 3.41951668e-01 6.01190865e-01 -3.66209507e-01 -2.48490900e-01 2.62026072e-01 -4.09254849e-01 5.19505322e-01 1.22140276e+00 -8.12812671e-02 1.67101130e-01 6.97317243e-01 1.19842637e+00 -3.27219397e-01 -1.13959658e+00 -5.15332162e-01 1.81736439e-01 -3.70944440e-01 2.37350017e-01 -9.12144184e-01 -1.23239660e+00 1.09208131e+00 1.06191754e+00 3.97498578e-01 1.07023239e+00 -3.26291233e-01 6.05809331e-01 4.35801417e-01 1.37467042e-01 -1.21796596e+00 2.87364632e-01 5.40574431e-01 5.77861667e-01 -1.60445535e+00 -2.02427819e-01 -2.27884263e-01 -4.51067597e-01 8.28694701e-01 7.78804183e-01 -1.37555093e-01 6.52807355e-01 2.41800860e-01 4.27962035e-01 -9.05613601e-02 -8.39861393e-01 1.11046985e-01 7.03519806e-02 3.67328167e-01 4.10587460e-01 -2.30971687e-02 -7.12227002e-02 5.99925041e-01 -1.67681247e-01 3.75421606e-02 3.45655918e-01 5.83048463e-01 -4.94122863e-01 -8.72948766e-01 -4.65004981e-01 8.13649416e-01 -6.44498587e-01 -4.87455428e-02 -9.89906937e-02 6.23177946e-01 4.02202785e-01 9.08202171e-01 2.23674163e-01 -4.73238021e-01 4.16516602e-01 1.38185978e-01 -2.15948634e-02 -5.17418623e-01 -5.50982594e-01 -1.51938692e-01 7.85250291e-02 -2.99564987e-01 -2.73252428e-01 -2.22038001e-01 -8.11930895e-01 -4.27828878e-02 -7.44794965e-01 5.75772002e-02 4.42730814e-01 9.09063339e-01 1.82790428e-01 7.48633623e-01 7.11536229e-01 -7.50358164e-01 -1.12679362e+00 -1.24190128e+00 -5.59907556e-01 5.68316221e-01 5.07479787e-01 -8.16865742e-01 -6.93032265e-01 -2.34820127e-01]
[9.241744995117188, 3.804807186126709]
3f1d62b3-04ac-4f0c-93ea-0fa3735bda6b
missing-modality-meets-meta-sampling-m3s-an
2210.03428
null
https://arxiv.org/abs/2210.03428v1
https://arxiv.org/pdf/2210.03428v1.pdf
Missing Modality meets Meta Sampling (M3S): An Efficient Universal Approach for Multimodal Sentiment Analysis with Missing Modality
Multimodal sentiment analysis (MSA) is an important way of observing mental activities with the help of data captured from multiple modalities. However, due to the recording or transmission error, some modalities may include incomplete data. Most existing works that address missing modalities usually assume a particular modality is completely missing and seldom consider a mixture of missing across multiple modalities. In this paper, we propose a simple yet effective meta-sampling approach for multimodal sentiment analysis with missing modalities, namely Missing Modality-based Meta Sampling (M3S). To be specific, M3S formulates a missing modality sampling strategy into the modal agnostic meta-learning (MAML) framework. M3S can be treated as an efficient add-on training component on existing models and significantly improve their performances on multimodal data with a mixture of missing modalities. We conduct experiments on IEMOCAP, SIMS and CMU-MOSI datasets, and superior performance is achieved compared with recent state-of-the-art methods.
['Gaoang Wang', 'Guanhong Wang', 'Junhao Zhu', 'Minghua Yang', 'Haozhe Chi']
2022-10-07
null
null
null
null
['multimodal-sentiment-analysis', 'multimodal-sentiment-analysis']
['computer-vision', 'natural-language-processing']
[ 3.88536543e-01 -2.15690613e-01 -5.75395346e-01 -5.69312572e-01 -1.14557493e+00 -2.53504157e-01 7.14678228e-01 -5.91546781e-02 -3.51321727e-01 7.05235898e-01 4.59237605e-01 1.97683543e-01 2.81503409e-01 -4.26205277e-01 -6.75498128e-01 -5.09096324e-01 6.97540283e-01 1.38498515e-01 -2.16492549e-01 -3.95120770e-01 6.63952529e-02 -3.27061921e-01 -1.91159344e+00 1.05813313e+00 6.81334376e-01 8.97330880e-01 -5.44226617e-02 5.08019447e-01 -4.55064744e-01 1.07025313e+00 -3.71453583e-01 -7.26267993e-01 -2.74789482e-01 -5.61209738e-01 -9.36941087e-01 3.75995129e-01 2.36651391e-01 -2.69973695e-01 -8.34597424e-02 9.22126353e-01 6.25828266e-01 4.28908169e-02 7.07147658e-01 -1.65099776e+00 -2.77997464e-01 6.62070930e-01 -7.30697572e-01 -2.87695915e-01 7.41631806e-01 -3.48216176e-01 6.10906780e-01 -1.08860505e+00 4.91196573e-01 1.23179173e+00 4.24723417e-01 9.50849533e-01 -8.82371783e-01 -5.78840315e-01 1.77296668e-01 4.34399277e-01 -1.06973207e+00 -7.91723788e-01 9.82680559e-01 -1.61063686e-01 5.63279152e-01 3.18498343e-01 2.10236669e-01 1.56394613e+00 -2.56045401e-01 1.34010410e+00 1.41342521e+00 -5.54064095e-01 2.87766486e-01 4.19158697e-01 6.36075288e-02 5.07460356e-01 -2.50158936e-01 -5.47079504e-01 -1.25409937e+00 -3.44023079e-01 1.43504426e-01 3.58565658e-01 -1.49959361e-03 -9.84845981e-02 -1.49166393e+00 4.05113667e-01 -1.36862963e-01 5.62606491e-02 -2.02171102e-01 -2.40926266e-01 6.09279573e-01 3.77282560e-01 3.87192309e-01 -2.75055647e-01 -5.36975861e-01 -5.35293698e-01 -9.22129393e-01 -1.45850042e-02 5.35968781e-01 8.94772768e-01 7.56436288e-01 -7.41538182e-02 5.55704720e-02 1.24164879e+00 4.16051060e-01 7.56379187e-01 6.58697367e-01 -1.17424345e+00 8.65266562e-01 9.06177759e-01 1.69974133e-01 -4.91462320e-01 -5.73454440e-01 7.72890374e-02 -9.05663729e-01 -1.89438865e-01 2.30507880e-01 -3.34846526e-01 -6.99122310e-01 1.83857095e+00 5.79608560e-01 1.86547518e-01 3.43440056e-01 6.54376745e-01 1.35245562e+00 5.34692824e-01 4.88818660e-02 -3.83765012e-01 1.37973416e+00 -1.05391848e+00 -1.07824576e+00 -1.39132529e-01 7.21499979e-01 -7.09117353e-01 1.10074675e+00 6.23182237e-01 -1.13634133e+00 -4.15131122e-01 -8.47444296e-01 -2.42997836e-02 -6.05409622e-01 2.31401026e-01 6.41719401e-01 8.69541705e-01 -7.53987193e-01 1.16396748e-01 -8.50323021e-01 -3.74847382e-01 3.26148570e-01 2.03656569e-01 -7.22442567e-01 -3.34479749e-01 -9.97560263e-01 7.08404064e-01 3.02835554e-01 8.27750936e-02 -7.43635476e-01 -4.45819765e-01 -9.79104817e-01 -3.92283410e-01 4.20592874e-01 -6.95932031e-01 1.24094439e+00 -1.09581113e+00 -1.56124890e+00 7.57556796e-01 -8.71000767e-01 1.01460211e-01 2.59780526e-01 4.69576195e-02 -9.08727527e-01 3.21998566e-01 -1.97166517e-01 6.07819974e-01 1.06503272e+00 -1.46887481e+00 -7.10117042e-01 -4.67668384e-01 2.27307796e-01 5.44787228e-01 -7.44249284e-01 -2.61985492e-02 -3.88562799e-01 -3.04665029e-01 1.03590138e-01 -9.31703091e-01 2.84411639e-01 -4.41187799e-01 -4.76468951e-01 -4.57432084e-02 7.58841455e-01 -5.14398873e-01 1.15374291e+00 -2.01666188e+00 4.31833774e-01 -1.90037102e-01 -5.87275513e-02 -4.34722267e-02 -2.85091370e-01 6.57799363e-01 8.04047212e-02 -1.08462729e-01 -2.57054627e-01 -1.23978817e+00 1.62920415e-01 2.05805346e-01 -9.86505598e-02 2.47566104e-01 -1.83250517e-01 7.42566228e-01 -6.87604487e-01 -6.74058318e-01 3.61309797e-01 5.40825546e-01 -2.20650986e-01 -1.27123399e-02 4.27370034e-02 5.88388979e-01 -7.02763721e-02 1.32757747e+00 7.08085477e-01 -2.74011314e-01 2.67998368e-01 -5.54235160e-01 1.82272702e-01 -1.89044863e-01 -1.20522630e+00 2.24535012e+00 -5.46082020e-01 2.65167505e-01 9.56348032e-02 -8.80950451e-01 3.19818586e-01 7.18332469e-01 6.01825118e-01 -7.44782031e-01 4.04058248e-01 1.59238636e-01 -4.58423108e-01 -7.55657196e-01 8.33937466e-01 -2.26938635e-01 -2.67985702e-01 5.33664167e-01 4.79603678e-01 4.45007086e-01 2.07052931e-01 1.74751014e-01 7.57808387e-01 2.61904329e-01 1.05536804e-01 5.39223611e-01 6.71489298e-01 -1.21370994e-01 3.61832023e-01 5.99147022e-01 -2.71246433e-01 7.05993295e-01 3.22544724e-01 1.08446077e-01 -6.57845497e-01 -8.07450712e-01 -2.86146458e-02 1.46462035e+00 2.31258720e-01 -5.25766373e-01 -6.25062585e-01 -7.50927567e-01 -5.15880346e-01 4.38302666e-01 -6.68385088e-01 -5.59555702e-02 6.00062795e-02 -1.22364414e+00 4.99453127e-01 4.89163458e-01 5.77083886e-01 -9.27355289e-01 -1.59943804e-01 -1.63363859e-01 -9.75185394e-01 -1.38018155e+00 -1.36111332e-02 -1.45864666e-01 -9.71253216e-01 -9.65130806e-01 -8.31859171e-01 -5.40333867e-01 6.93913579e-01 4.76594359e-01 8.10000837e-01 -3.78504604e-01 2.47532636e-01 7.99193978e-01 -7.76957095e-01 -5.51959872e-01 -2.51425058e-01 2.34078884e-01 5.44983745e-02 5.70903480e-01 5.93430161e-01 -4.33870018e-01 -4.63921934e-01 1.78392664e-01 -1.07448447e+00 2.67025590e-01 5.45645773e-01 7.78505087e-01 5.02093911e-01 1.56811178e-02 7.71162629e-01 -9.47259724e-01 3.42751652e-01 -8.01118910e-01 2.54707336e-01 3.73157024e-01 -1.34581760e-01 -3.26554865e-01 3.14052701e-01 -5.34679294e-01 -1.28114343e+00 8.39181468e-02 -7.05085099e-02 -3.88705581e-01 -4.35579389e-01 7.10915864e-01 -4.93154705e-01 3.37918364e-02 1.63871557e-01 4.33099270e-01 3.48669402e-02 -5.78014076e-01 2.50142187e-01 1.11315560e+00 4.31321412e-01 -5.48658311e-01 3.72307807e-01 9.37933326e-01 -4.03777361e-02 -8.61036777e-01 -9.64270413e-01 -5.18821895e-01 -4.74928558e-01 -4.38746244e-01 6.03266597e-01 -1.22930741e+00 -8.07701766e-01 7.93996811e-01 -7.39541054e-01 3.20985098e-03 4.63648476e-02 5.81964850e-01 -4.58082646e-01 4.25411433e-01 -6.42832577e-01 -1.05282760e+00 -1.39860898e-01 -1.01128542e+00 1.31031072e+00 2.06570655e-01 -1.49617031e-01 -1.03556001e+00 4.40599546e-02 1.13074636e+00 1.91017464e-01 6.77808747e-02 6.10374153e-01 -4.53655452e-01 3.20797302e-02 -3.72906089e-01 1.22808881e-01 3.53975177e-01 1.81635320e-01 -1.99315250e-01 -1.52036607e+00 -2.49945238e-01 -5.05723245e-02 -7.73249805e-01 8.37963760e-01 2.08083913e-01 9.37329829e-01 -5.98078631e-02 -2.51770411e-02 2.24926591e-01 1.27681422e+00 -1.18202955e-01 5.90043724e-01 2.85787731e-01 7.96642363e-01 6.92390025e-01 7.05111861e-01 6.48230135e-01 1.04784358e+00 2.88881153e-01 6.47885263e-01 1.14876507e-02 1.72272652e-01 -9.16768834e-02 6.55590117e-01 1.32277906e+00 -5.99796847e-02 -2.60126740e-01 -4.20971155e-01 7.38909066e-01 -2.12432122e+00 -1.17113817e+00 -5.01501337e-02 2.11961269e+00 7.61061788e-01 -1.98864877e-01 3.87872964e-01 4.64125812e-01 6.07740939e-01 2.21739158e-01 -4.57516849e-01 -2.47140095e-01 -4.29153293e-01 -3.26651275e-01 -2.76740193e-02 1.44228399e-01 -1.18149495e+00 3.83336663e-01 5.97114849e+00 8.71043444e-01 -8.98841441e-01 6.63566887e-01 2.91499913e-01 -5.42883635e-01 -4.28417802e-01 -1.85699537e-01 -7.61544645e-01 6.90053344e-01 1.00496519e+00 4.88413513e-01 4.15419966e-01 4.50060219e-01 -1.13242738e-01 -4.18693095e-01 -1.08578348e+00 1.45721877e+00 5.28608859e-01 -9.57698584e-01 4.78832722e-02 -3.84742618e-02 8.99318993e-01 -8.22620187e-03 2.07386956e-01 4.14528936e-01 -3.50445300e-01 -7.82232702e-01 7.32388437e-01 7.57807374e-01 6.54263198e-01 -9.61241782e-01 9.56795394e-01 5.69797993e-01 -9.94490862e-01 -1.03744313e-01 -1.21947765e-01 2.56119445e-02 2.86235332e-01 4.00582731e-01 -8.16480368e-02 9.58123803e-01 6.44587994e-01 7.78562784e-01 -5.93059599e-01 5.77917278e-01 1.91842407e-01 5.27570009e-01 -2.17129230e-01 1.11106165e-01 -2.37901703e-01 2.00116150e-02 2.88688034e-01 9.86053765e-01 3.32469434e-01 -2.51902014e-01 -7.16229454e-02 1.83872610e-01 -2.92515129e-01 1.20559752e-01 -4.56759125e-01 -7.97919109e-02 4.73654121e-01 1.40885353e+00 -2.23949611e-01 -4.70863789e-01 -1.00965846e+00 1.01662290e+00 4.75733578e-02 3.27486038e-01 -7.32381284e-01 -6.45745397e-02 2.51641095e-01 -4.29973006e-01 4.58505899e-02 2.15261847e-01 -2.32659459e-01 -1.75052202e+00 1.34727269e-01 -9.90673304e-01 6.09885693e-01 -1.05509961e+00 -1.40650976e+00 3.88376057e-01 9.52160880e-02 -1.67583096e+00 -2.57236511e-01 -3.43904883e-01 -9.47068930e-02 6.10185504e-01 -1.66749227e+00 -1.74256587e+00 -3.50179255e-01 9.49049354e-01 5.39824605e-01 -3.31836224e-01 9.53616738e-01 6.76798940e-01 -6.58596933e-01 7.08073914e-01 7.36196786e-02 -8.08705091e-02 9.87184703e-01 -8.88599515e-01 -5.95950603e-01 5.50610304e-01 -3.38992067e-02 4.48703825e-01 5.79134524e-01 -4.00209814e-01 -1.81384420e+00 -8.65781546e-01 8.30516815e-01 -8.09559941e-01 4.82298106e-01 -1.98996291e-01 -5.98578453e-01 6.88546062e-01 5.92079997e-01 -3.05292994e-01 1.37812507e+00 2.81455010e-01 -3.88272852e-01 -5.76392226e-02 -1.19587922e+00 5.71597397e-01 6.22475207e-01 -8.26648831e-01 -5.78655541e-01 4.27534990e-02 3.84289950e-01 -4.14317131e-01 -1.01640093e+00 5.45121074e-01 7.33147621e-01 -9.26751435e-01 7.36973584e-01 -5.21129370e-01 6.87299252e-01 -3.33285302e-01 -6.92790747e-01 -1.13288593e+00 4.39284503e-01 -2.48269379e-01 -6.88892543e-01 1.58555353e+00 4.57542390e-01 -3.14831793e-01 7.41743684e-01 6.72825933e-01 9.69515741e-02 -3.90075028e-01 -9.45936024e-01 -3.16810429e-01 -3.28099936e-01 -9.02238727e-01 7.20755577e-01 1.26622450e+00 5.74198067e-01 3.29385906e-01 -9.25162137e-01 7.52335861e-02 7.90596068e-01 7.33342320e-02 8.83122146e-01 -8.69201601e-01 -1.91513866e-01 -3.89580168e-02 -9.94372368e-02 -7.43640304e-01 2.29684576e-01 -6.29345775e-01 -3.58009070e-01 -1.41882372e+00 6.11115515e-01 2.18026906e-01 -5.58892190e-01 6.07006907e-01 -9.03903246e-02 6.43267930e-01 1.66042581e-01 2.24859923e-01 -1.08842134e+00 7.88165808e-01 1.12816238e+00 -2.27219582e-01 -1.31708696e-01 -9.74204682e-04 -8.19229901e-01 1.00382149e+00 6.51962817e-01 -3.43728155e-01 -5.99201143e-01 -3.53225321e-01 7.13856459e-01 3.28750968e-01 2.92685360e-01 -8.32390726e-01 2.97877043e-01 -2.28940442e-01 2.82110691e-01 -1.09132683e+00 1.10810518e+00 -9.96170044e-01 1.12764500e-01 -2.21036896e-01 -3.22105020e-01 -1.83297202e-01 7.14109540e-02 5.08049190e-01 -5.47867298e-01 -2.33799994e-01 2.37608910e-01 -1.60457730e-01 -8.06950271e-01 5.33261113e-02 -3.95504743e-01 -1.58861667e-01 6.02014899e-01 -2.09922716e-01 -6.45792305e-01 -4.35119182e-01 -9.02220666e-01 2.30070934e-01 4.36673045e-01 5.63649833e-01 6.49950624e-01 -1.63065910e+00 -3.83907765e-01 -6.79222792e-02 5.95936239e-01 -3.51284266e-01 9.43061233e-01 1.31711602e+00 1.75129458e-01 3.69032286e-02 -1.36985794e-01 -5.44699669e-01 -1.42493999e+00 3.11387867e-01 1.81673840e-02 1.65917978e-01 5.79947904e-02 4.05989885e-01 -2.18113765e-01 -9.60647285e-01 1.74977198e-01 1.03947312e-01 -4.93808985e-01 5.58970928e-01 7.94595838e-01 6.37569845e-01 6.86374679e-02 -1.03347766e+00 -3.66312295e-01 4.54178154e-01 1.67671114e-01 -4.92582858e-01 1.12663352e+00 -7.45100141e-01 -1.76066101e-01 1.29384565e+00 1.15785933e+00 -5.28852530e-02 -7.51397908e-01 -5.72367251e-01 -4.99121368e-01 -2.86498129e-01 -1.24797978e-01 -7.83294916e-01 -9.71176505e-01 9.69760656e-01 7.06721008e-01 5.62994294e-02 1.34983706e+00 -5.10679185e-03 7.76949227e-01 3.88761997e-01 4.42652881e-01 -1.53230095e+00 -1.72483008e-02 2.59718746e-01 3.92902195e-01 -1.79762506e+00 -7.29112653e-03 -1.84453219e-01 -1.10506189e+00 8.18374872e-01 6.33625567e-01 5.29045463e-01 6.93345964e-01 5.26782572e-02 1.93237886e-01 1.70136578e-02 -9.26185668e-01 -2.33457640e-01 2.51373917e-01 4.98635292e-01 3.24587286e-01 3.30960974e-02 -9.68258530e-02 8.45564723e-01 2.95585454e-01 2.63209313e-01 6.16259456e-01 1.36648619e+00 -1.82912663e-01 -1.23322415e+00 -5.51833510e-01 3.55297148e-01 -5.44557273e-01 2.93872897e-02 -2.37673193e-01 5.19086719e-01 1.96149617e-01 1.34032667e+00 -2.05536023e-01 -6.23797774e-01 2.47730196e-01 4.56704229e-01 6.36414289e-01 -3.03582579e-01 -3.97943586e-01 1.81487232e-01 1.70806810e-01 -5.21559894e-01 -1.43482149e+00 -7.93153703e-01 -9.10187364e-01 -3.74620765e-01 -1.78213477e-01 -1.81202397e-01 8.69410157e-01 1.43486094e+00 3.87747228e-01 4.81305152e-01 5.94239533e-01 -9.82106447e-01 -3.25165503e-02 -1.06777418e+00 -7.00799584e-01 6.04105115e-01 4.08349067e-01 -5.69514573e-01 -2.42491692e-01 3.32504362e-01]
[13.158853530883789, 5.102975368499756]
039a378c-844f-4076-8988-288b27f3a5ff
split-learning-in-6g-edge-networks
2306.12194
null
https://arxiv.org/abs/2306.12194v2
https://arxiv.org/pdf/2306.12194v2.pdf
Split Learning in 6G Edge Networks
With the proliferation of distributed edge computing resources, the 6G mobile network will evolve into a network for connected intelligence. Along this line, the proposal to incorporate federated learning into the mobile edge has gained considerable interest in recent years. However, the deployment of federated learning faces substantial challenges as massive resource-limited IoT devices can hardly support on-device model training. This leads to the emergence of split learning (SL) which enables servers to handle the major training workload while still enhancing data privacy. In this article, we offer a brief overview of key advancements in SL and articulate its seamless integration with wireless edge networks. We begin by illustrating the tailored 6G architecture to support edge SL. Then, we examine the critical design issues for edge SL, including innovative resource-efficient learning frameworks and resource management strategies under a single edge server. Additionally, we expand the scope to multi-edge scenarios, exploring multi-edge collaboration and mobility management from a networking perspective. Finally, we discuss open problems for edge SL, including convergence analysis, asynchronous SL and U-shaped SL.
['Kaibin Huang', 'Xianhao Chen', 'Guanqiao Qu', 'Zheng Lin']
2023-06-21
null
null
null
null
['management', 'edge-computing']
['miscellaneous', 'time-series']
[-4.65268672e-01 1.11041650e-01 -5.16887307e-01 -1.12172708e-01 -2.11242318e-01 -6.96862996e-01 -9.08239111e-02 -5.55558860e-01 1.42408162e-01 1.02804029e+00 7.38601387e-02 -9.17861044e-01 -4.71740156e-01 -7.18094230e-01 -2.43717939e-01 -5.92289686e-01 -3.46014917e-01 2.65291572e-01 -2.78671175e-01 2.10004747e-01 -3.81664097e-01 5.29682577e-01 -1.33794439e+00 -6.27615303e-02 8.59284699e-01 1.54556882e+00 -1.87055673e-02 5.32647789e-01 -1.99680880e-01 8.25563610e-01 -3.31856042e-01 -7.29834437e-01 6.85735345e-01 -1.25027016e-01 -7.78467417e-01 1.58322379e-01 1.37624711e-01 -2.32817441e-01 -5.44216394e-01 7.92893529e-01 1.02053130e+00 8.58413801e-02 -1.75637916e-01 -1.72907424e+00 -8.16564709e-02 5.40636122e-01 1.45897076e-01 1.66329294e-01 -3.16065326e-02 -3.27950448e-01 8.83267462e-01 -6.32915556e-01 8.30893934e-01 2.95985758e-01 8.20703685e-01 5.74468136e-01 -3.54856491e-01 -5.77902138e-01 4.12259042e-01 4.69989419e-01 -1.09494150e+00 -9.68268037e-01 6.71452701e-01 1.93678856e-01 7.56943524e-01 5.86551547e-01 7.83531785e-01 6.53602004e-01 -1.56505957e-01 7.03713059e-01 5.98528862e-01 -3.67588073e-01 4.61507112e-01 2.20111519e-01 -3.68598133e-01 6.00239933e-01 6.35426044e-01 -7.27200955e-02 -6.71326160e-01 -2.75189608e-01 5.18468201e-01 7.56801665e-02 -2.87613153e-01 -7.28120327e-01 -5.86161852e-01 2.42102623e-01 1.47301316e-01 2.58019984e-01 -5.80670774e-01 1.57835320e-01 4.58559722e-01 5.42245746e-01 7.84227550e-01 -2.42457569e-01 -8.26564968e-01 -3.73951793e-01 -1.12614894e+00 -3.85187924e-01 1.20118940e+00 1.33676219e+00 6.86952889e-01 5.96210547e-02 7.16854036e-02 5.55627197e-02 3.41322064e-01 1.20122261e-01 -4.62317392e-02 -1.29776347e+00 3.33185375e-01 3.01054120e-01 4.46552113e-02 -4.94842112e-01 -4.13490832e-01 -1.40743995e+00 -9.66054797e-01 -1.17967404e-01 1.28283754e-01 -9.19094801e-01 -3.71084064e-02 1.67825496e+00 5.69253385e-01 8.61924052e-01 1.06378146e-01 6.28533840e-01 7.56073713e-01 2.45517190e-03 -2.54380345e-01 -6.49913430e-01 9.49466109e-01 -1.40302944e+00 -8.33839536e-01 -1.33026227e-01 7.83825457e-01 -4.57291752e-01 3.37382644e-01 1.90542400e-01 -1.29500878e+00 4.78895585e-04 -6.93846166e-01 3.10603082e-01 -3.14151525e-01 9.23897102e-02 1.05662513e+00 1.20537734e+00 -1.53666461e+00 3.51883143e-01 -7.46540129e-01 -6.28780007e-01 7.73108304e-01 7.11450279e-01 -1.95922881e-01 -2.16349900e-01 -9.65379715e-01 1.96496844e-01 -4.80671898e-02 -2.18427747e-01 -4.56531882e-01 -7.82465160e-01 -2.98004985e-01 4.79555964e-01 4.24719661e-01 -1.54459071e+00 1.16147673e+00 -7.26527452e-01 -1.69948471e+00 5.95509171e-01 -2.10471511e-01 -3.79976392e-01 6.21078551e-01 7.20679536e-02 -8.99958611e-01 1.45417362e-01 -1.15074456e-01 -1.60099968e-01 5.53357780e-01 -1.05268586e+00 -9.97000575e-01 -3.96242768e-01 1.75556794e-01 2.13880360e-01 -7.74266660e-01 4.65597250e-02 -4.84530270e-01 -3.13257605e-01 -6.14197478e-02 -9.37902093e-01 -4.88094687e-01 -5.22467960e-03 7.17970580e-02 7.21516907e-02 1.22469687e+00 6.09340728e-04 1.62430441e+00 -2.34835958e+00 -3.66408676e-01 2.65429050e-01 8.21122408e-01 2.19210371e-01 3.10242057e-01 4.53617036e-01 3.15288007e-01 1.34909600e-01 5.24929881e-01 -8.88538957e-01 7.08898483e-03 2.18116745e-01 3.35281268e-02 1.61539882e-01 -8.99017632e-01 1.02543867e+00 -1.00697803e+00 -3.30971360e-01 -1.48623422e-01 2.54603237e-01 -9.21893477e-01 1.71125069e-01 1.10059395e-01 6.72087908e-01 -5.95343947e-01 9.62366462e-01 7.23865271e-01 -7.96204627e-01 5.28994203e-01 -9.32425782e-02 -1.17303878e-01 5.82604744e-02 -1.05394351e+00 1.57068968e+00 -7.30418801e-01 3.60861003e-01 8.97260129e-01 -9.46652472e-01 6.51165068e-01 9.45844829e-01 1.05932689e+00 -5.31111248e-02 2.53917634e-01 6.77711844e-01 -5.02494335e-01 -3.65424752e-01 3.47703516e-01 -6.16546310e-02 1.79657757e-01 6.24003470e-01 3.08487937e-02 7.74425745e-01 -3.65108043e-01 2.53864855e-01 1.32388377e+00 -3.48804533e-01 1.78553134e-01 -6.67031258e-02 4.38656688e-01 -6.39354527e-01 9.26689208e-01 6.80994809e-01 -8.54134142e-01 2.67023817e-02 2.35907119e-02 -4.98173177e-01 -1.57183811e-01 -9.16996837e-01 3.10627401e-01 1.05924344e+00 2.22443894e-01 -1.01084828e+00 -6.32098138e-01 -1.11061442e+00 -2.18696788e-01 2.08838999e-01 9.51820239e-02 -1.05487928e-01 2.31571943e-02 -6.44127905e-01 4.28815633e-01 2.34168649e-01 1.05371737e+00 -3.98630053e-01 -3.43711406e-01 1.75425604e-01 -3.44303638e-01 -1.31180382e+00 -5.10768831e-01 2.30535835e-01 -1.03652513e+00 -8.57518911e-01 -3.54472458e-01 -9.35824335e-01 4.17991519e-01 6.85491025e-01 1.37461817e+00 5.26730809e-03 1.19825087e-01 1.06826949e+00 -2.83703566e-01 -1.75090164e-01 2.40098834e-01 6.63063407e-01 4.63754892e-01 4.36091691e-01 3.37177247e-01 -1.25713265e+00 -9.80168700e-01 4.73961681e-01 -3.06162894e-01 -2.65028894e-01 3.48517627e-01 4.74859387e-01 4.06900615e-01 3.45259756e-01 1.13395691e+00 -1.01183176e+00 2.91282922e-01 -9.14590716e-01 -1.14840478e-01 4.46152240e-01 -1.06843495e+00 -5.96292675e-01 7.93777108e-01 5.98229282e-02 -1.18936396e+00 4.72360803e-03 1.20211951e-01 -5.22101641e-01 1.71074092e-01 3.67148042e-01 -7.30073333e-01 -6.01688147e-01 3.44926298e-01 -1.61769688e-01 -1.95602298e-01 -3.84639770e-01 5.36033273e-01 1.17008555e+00 2.42451087e-01 -5.81042945e-01 5.88289797e-01 7.27495492e-01 1.48655221e-01 -6.04130983e-01 -6.70202315e-01 -4.99195784e-01 5.16375620e-03 -4.71032321e-01 2.53984720e-01 -1.14605772e+00 -9.03212965e-01 7.96270147e-02 -7.83080220e-01 -3.54045570e-01 -6.66355371e-01 6.46347821e-01 -7.16956913e-01 2.15346411e-01 -6.21524155e-01 -7.79906929e-01 -6.23357415e-01 -7.87509620e-01 4.76597369e-01 3.95029664e-01 3.31791848e-01 -1.31829047e+00 -2.93793559e-01 6.43605053e-01 1.12608910e+00 -1.58892289e-01 3.99562418e-01 -5.50632417e-01 -1.02593589e+00 -1.55792981e-01 1.87321547e-02 -6.45986274e-02 2.65760750e-01 -5.51058948e-01 -1.02657330e+00 -7.67432511e-01 1.82282433e-01 2.16386050e-01 2.49591038e-01 5.28640211e-01 1.21256030e+00 -2.79666752e-01 -6.69009089e-01 1.39070642e+00 1.33779073e+00 -1.08347647e-02 4.39734966e-01 1.20359480e-01 3.74742419e-01 1.47814989e-01 3.71419877e-01 9.32947814e-01 5.29019773e-01 2.21448526e-01 7.01788783e-01 -1.14324912e-01 -1.39750496e-01 -2.24598944e-01 1.49312541e-01 1.14406025e+00 -3.95797342e-01 -5.70694029e-01 -3.18326652e-01 2.21814379e-01 -2.09758329e+00 -8.90381634e-01 9.34266001e-02 2.13338423e+00 -2.12238990e-02 -2.26058941e-02 3.10512781e-01 7.31132030e-02 7.91361570e-01 7.66462013e-02 -8.85995865e-01 6.08889498e-02 -1.18531771e-01 3.65781933e-02 8.12497497e-01 2.15585798e-01 -8.12892199e-01 9.76983726e-01 6.72367096e+00 6.63120270e-01 -1.15220833e+00 6.06845021e-01 8.22342157e-01 -1.45618945e-01 -4.30613250e-01 1.67115912e-01 -5.13319314e-01 3.40061069e-01 9.96219695e-01 -6.22219384e-01 7.53883421e-01 1.21376467e+00 3.38120043e-01 5.03149688e-01 -6.08779132e-01 1.46361041e+00 -4.29660171e-01 -1.64722407e+00 -3.35074931e-01 4.09280986e-01 1.02580404e+00 5.05728543e-01 4.41161543e-03 1.81983173e-01 8.70036557e-02 -4.71887767e-01 1.89146891e-01 4.87511665e-01 1.17156744e+00 -8.40182126e-01 5.40844560e-01 2.98974097e-01 -1.64219630e+00 -4.43998069e-01 -8.52486417e-02 -3.74894917e-01 4.10625011e-01 8.41877460e-01 -1.81338713e-01 1.07423604e+00 8.50289345e-01 8.62523377e-01 -1.46993205e-01 1.34402287e+00 2.72044688e-01 5.37536323e-01 -2.52262473e-01 3.75609398e-01 -3.25921327e-01 -3.86154324e-01 5.95383346e-01 6.01314306e-01 8.04292381e-01 6.85067698e-02 2.29571477e-01 1.84502840e-01 -5.39346278e-01 1.77634731e-01 -5.86684167e-01 8.34431797e-02 9.29630399e-01 1.61652935e+00 -6.09503329e-01 -2.69513607e-01 -9.00514722e-01 1.08609021e+00 1.70827702e-01 5.79533160e-01 -5.28585613e-01 -3.28838229e-01 1.24171972e+00 2.22886965e-01 1.48557395e-01 -3.39487106e-01 -3.35444570e-01 -1.39515758e+00 -1.14215307e-01 -7.08142281e-01 6.20130420e-01 -3.64730805e-01 -1.18788552e+00 7.42173731e-01 -6.97100937e-01 -1.34048283e+00 1.10203192e-01 -3.25182915e-01 -7.93820202e-01 3.31942171e-01 -1.51480401e+00 -1.00709093e+00 -4.50849742e-01 9.06230450e-01 1.58612076e-02 -5.44027388e-01 9.40577149e-01 1.14048946e+00 -6.13585711e-01 1.02950943e+00 3.90827924e-01 -3.92157972e-01 6.89998388e-01 -8.27549517e-01 1.53497919e-01 8.99152339e-01 6.67002425e-02 4.23148096e-01 1.10929087e-01 -4.23883528e-01 -1.59834743e+00 -1.20153952e+00 1.07606518e+00 -5.67424782e-02 3.26150715e-01 -2.78702378e-01 -9.64955613e-02 9.48513746e-01 -6.26296327e-02 8.79078507e-01 1.30360186e+00 1.81557670e-01 2.32496232e-01 -5.13025641e-01 -1.45193040e+00 5.99660575e-01 1.85140586e+00 -6.45017445e-01 7.67308831e-01 4.46371406e-01 7.08171606e-01 -2.37825230e-01 -9.13388014e-01 2.22474635e-01 3.31277817e-01 -1.13212061e+00 6.32744670e-01 -8.22260678e-01 -7.82874346e-01 -2.53202885e-01 -2.73214281e-01 -1.05739951e+00 -5.73829353e-01 -1.73635006e+00 -8.66685688e-01 1.22067046e+00 1.36812208e-02 -1.02358186e+00 1.73477089e+00 7.35891581e-01 -3.14294785e-01 -7.96905756e-01 -1.20340300e+00 -8.84454846e-01 -5.08221090e-01 -5.82911491e-01 9.42170620e-01 9.29065824e-01 2.99286693e-01 2.31174529e-01 -4.39402908e-01 1.89410374e-01 5.54978848e-01 8.74457695e-03 6.42946005e-01 -1.39295816e+00 -5.46515882e-01 -9.32586938e-02 -4.13605601e-01 -1.40653372e+00 3.81259173e-02 -1.20408189e+00 -8.10702980e-01 -1.46131563e+00 -2.95031786e-01 -1.01892054e+00 -4.72366154e-01 2.11261168e-01 2.04640120e-01 2.02295616e-01 3.05918396e-01 1.60580188e-01 -1.19502449e+00 4.33892101e-01 1.01701796e+00 4.19594377e-01 -1.58485308e-01 9.63930905e-01 -1.10926092e+00 4.93331134e-01 1.20067513e+00 -5.68432882e-02 -9.26254153e-01 -3.77684146e-01 5.37661493e-01 2.06971794e-01 -1.54256210e-01 -1.12039828e+00 6.96616054e-01 9.87230912e-02 -1.13856405e-01 -4.43328284e-02 -2.04252973e-02 -1.52188158e+00 4.98912632e-01 1.76093593e-01 4.98994172e-01 1.13754645e-02 -3.17506790e-01 6.60424948e-01 9.54839811e-02 2.74826020e-01 2.19083011e-01 -1.00353705e-02 -5.33775985e-01 1.09148252e+00 -4.43028241e-01 1.28871635e-01 1.26376987e+00 -3.56312722e-01 -1.69220746e-01 -8.78444433e-01 -1.11868680e+00 4.68136460e-01 4.53152418e-01 -8.42872486e-02 1.04097188e-01 -1.26873374e+00 -1.91820294e-01 2.98535317e-01 -2.59989649e-01 -2.17733398e-01 4.03133631e-01 9.89249170e-01 -2.12205544e-01 2.65805155e-01 9.15965214e-02 -7.77721778e-02 -1.16253150e+00 6.57026291e-01 7.50869989e-01 -2.93976337e-01 -4.08082247e-01 6.79837823e-01 -3.46901268e-01 -4.15240526e-01 5.51891625e-01 2.84562141e-01 3.25295001e-01 -2.63606399e-01 2.86045730e-01 9.39897835e-01 4.52385277e-01 -1.13316476e-01 -4.28079426e-01 1.62515700e-01 5.25428832e-01 2.78498888e-01 1.11346781e+00 -7.94991434e-01 -8.11778456e-02 -6.95922673e-02 9.91943061e-01 3.94893229e-01 -1.17800474e+00 -2.69017607e-01 -7.80650750e-02 -3.83839399e-01 4.27442700e-01 -6.10026717e-01 -1.69516850e+00 3.04710776e-01 7.21606016e-01 2.00565025e-01 1.55655932e+00 9.67988558e-03 1.12421620e+00 2.20051706e-01 1.08917248e+00 -1.04678988e+00 -3.29717666e-01 3.91772419e-01 -3.46047968e-01 -9.23300624e-01 -2.96136826e-01 -7.60143459e-01 -2.07902491e-02 1.07209671e+00 4.27796274e-01 4.77517188e-01 1.46559501e+00 2.65375823e-01 7.79585317e-02 -2.63117999e-01 -7.86375821e-01 -4.08970922e-01 -3.57441574e-01 6.57885849e-01 3.10589194e-01 1.75308689e-01 -3.04130316e-01 1.06907439e+00 -4.06564660e-02 4.55355883e-01 2.18817890e-01 1.02320206e+00 -1.98732123e-01 -1.69343352e+00 9.62965488e-02 5.75543880e-01 -6.54456794e-01 4.05639559e-02 -7.10028931e-02 1.10293865e-01 4.94431645e-01 1.18067527e+00 3.25915851e-02 -5.87557971e-01 -1.27176672e-01 1.02310866e-01 1.63381994e-01 -4.02742684e-01 -5.84622562e-01 -3.22908223e-01 2.69021273e-01 -5.25767922e-01 -2.95412064e-01 -4.26478237e-01 -9.46294308e-01 -7.06130624e-01 -4.02310938e-01 6.07722104e-01 7.31236100e-01 6.81454539e-01 1.32542992e+00 4.68315005e-01 1.12090075e+00 -4.74846005e-01 -2.22856358e-01 -1.52263552e-01 -8.74506652e-01 -3.57124656e-01 1.49127066e-01 -1.98527217e-01 -5.65869510e-01 -3.93183589e-01]
[5.949120044708252, 5.652664661407471]
170aa451-3fab-4b6d-9374-ab732f143f07
3dias-3d-shape-reconstruction-with-implicit
2108.08653
null
https://arxiv.org/abs/2108.08653v1
https://arxiv.org/pdf/2108.08653v1.pdf
3DIAS: 3D Shape Reconstruction with Implicit Algebraic Surfaces
3D Shape representation has substantial effects on 3D shape reconstruction. Primitive-based representations approximate a 3D shape mainly by a set of simple implicit primitives, but the low geometrical complexity of the primitives limits the shape resolution. Moreover, setting a sufficient number of primitives for an arbitrary shape is challenging. To overcome these issues, we propose a constrained implicit algebraic surface as the primitive with few learnable coefficients and higher geometrical complexities and a deep neural network to produce these primitives. Our experiments demonstrate the superiorities of our method in terms of representation power compared to the state-of-the-art methods in single RGB image 3D shape reconstruction. Furthermore, we show that our method can semantically learn segments of 3D shapes in an unsupervised manner. The code is publicly available from https://myavartanoo.github.io/3dias/ .
['Kyoung Mu Lee', 'Reyhaneh Neshatavar', 'JaeYoung Chung', 'Mohsen Yavartanoo']
2021-08-19
null
http://openaccess.thecvf.com//content/ICCV2021/html/Yavartanoo_3DIAS_3D_Shape_Reconstruction_With_Implicit_Algebraic_Surfaces_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Yavartanoo_3DIAS_3D_Shape_Reconstruction_With_Implicit_Algebraic_Surfaces_ICCV_2021_paper.pdf
iccv-2021-1
['3d-shape-representation']
['computer-vision']
[-2.94760764e-02 1.26948550e-01 4.08754796e-02 -2.01664791e-01 -6.68522894e-01 -6.27722383e-01 5.95526934e-01 -1.22576065e-01 -8.85760691e-03 2.27266222e-01 2.57166084e-02 -3.22993964e-01 1.13019250e-01 -1.14519477e+00 -7.62027979e-01 -6.28802836e-01 1.26402780e-01 8.57467890e-01 2.69838810e-01 -8.29225704e-02 1.20593712e-01 1.17927599e+00 -1.32777965e+00 1.09752648e-01 6.85445845e-01 1.21339941e+00 1.69757888e-01 3.29936802e-01 -4.07352090e-01 3.58398147e-02 -3.68636437e-02 -1.68965161e-01 6.24045670e-01 -7.35529587e-02 -6.09866619e-01 1.37775674e-01 3.07068318e-01 -4.93734390e-01 -3.97245407e-01 8.17948699e-01 5.34930170e-01 3.73112895e-02 9.50598359e-01 -9.55777764e-01 -6.46488547e-01 -2.84790415e-02 -3.30094993e-01 -6.17003262e-01 3.55976969e-01 1.02439925e-01 9.07903314e-01 -1.28958523e+00 4.83376205e-01 1.34827590e+00 7.97167599e-01 5.19227087e-01 -1.44293928e+00 -5.99689126e-01 5.21308556e-02 -3.65219295e-01 -1.69295311e+00 -4.38718110e-01 1.08581960e+00 -3.03565860e-01 9.76779222e-01 2.86361486e-01 8.19943190e-01 7.16135144e-01 -2.50174344e-01 7.37106323e-01 1.00426853e+00 -2.72792071e-01 2.63608098e-01 -3.98455262e-01 -2.75428057e-01 9.16453958e-01 2.39832059e-01 -5.29306047e-02 -6.85594976e-02 -4.51407433e-01 1.72482634e+00 2.61080772e-01 -2.85999421e-02 -7.54688799e-01 -8.69092524e-01 7.67447412e-01 7.26881981e-01 1.89858049e-01 -3.07367504e-01 5.61989546e-01 -1.08207017e-01 -1.10993147e-01 4.07118052e-01 1.69747368e-01 -4.35095876e-01 1.66007336e-02 -5.93799949e-01 2.28593811e-01 7.87458479e-01 1.16904318e+00 1.03826475e+00 2.36525014e-01 4.47940618e-01 7.21061289e-01 4.25222993e-01 6.84902608e-01 -4.17574383e-02 -1.19436383e+00 -3.07610095e-03 8.89572501e-01 -4.76294197e-02 -8.80751967e-01 -3.04308355e-01 -2.34656110e-02 -8.84635210e-01 4.83257324e-01 3.18798691e-01 3.49723727e-01 -1.08684838e+00 1.24502683e+00 5.31541824e-01 1.80100143e-01 -2.84710735e-01 8.45504165e-01 1.09673846e+00 5.28642416e-01 -1.86668351e-01 3.19790870e-01 1.09101903e+00 -6.10016882e-01 -1.70855105e-01 1.47437304e-01 3.69581580e-01 -6.28014863e-01 1.08405077e+00 1.45645142e-01 -1.23807919e+00 -2.85779029e-01 -6.75655842e-01 -4.98720616e-01 -9.88845378e-02 1.71149373e-01 9.56541777e-01 4.49519157e-01 -9.95517015e-01 5.70214808e-01 -1.04742289e+00 -1.07996993e-01 8.33115578e-01 5.15742302e-01 -3.16990018e-01 2.13430990e-02 -4.94584441e-01 5.74342191e-01 2.41698965e-01 -2.15823874e-01 -7.41274893e-01 -6.93785250e-01 -9.21000898e-01 -4.41680066e-02 2.89425433e-01 -1.06417036e+00 1.15352249e+00 -6.21601462e-01 -1.77921176e+00 1.05349052e+00 -1.14355609e-01 7.27316784e-03 4.38665211e-01 -1.41630545e-01 3.51916075e-01 2.79092699e-01 -2.81034201e-01 6.47743285e-01 8.54972482e-01 -1.59196055e+00 1.92969143e-02 -2.40474790e-01 2.89215088e-01 2.51425236e-01 2.42291540e-01 -2.43177027e-01 -6.15280807e-01 -7.21091986e-01 7.11046159e-01 -8.04201722e-01 -4.73599643e-01 7.15163112e-01 -2.30162814e-01 -2.83241957e-01 6.29103601e-01 -2.97139376e-01 6.21263444e-01 -1.99727166e+00 1.73011988e-01 2.66776651e-01 1.87168509e-01 1.23244546e-01 -1.70342505e-01 3.32382739e-01 1.93301767e-01 3.15685004e-01 -7.60398328e-01 -5.55664122e-01 9.49482024e-02 5.54028094e-01 -3.71213734e-01 3.91156316e-01 3.10396552e-01 1.06274712e+00 -8.09585035e-01 -2.16184482e-01 2.75385410e-01 9.66587245e-01 -6.50361776e-01 2.05768391e-01 -3.97669226e-01 6.67045295e-01 -9.06313479e-01 9.31278408e-01 8.59296560e-01 -3.06163430e-01 -8.01678002e-02 -1.91004768e-01 7.45511847e-03 6.92570388e-01 -1.18923748e+00 2.00561428e+00 -3.33222508e-01 -2.10015535e-01 1.80428490e-01 -9.19104815e-01 1.15739059e+00 2.27382481e-01 7.31508970e-01 -3.05915534e-01 2.00857684e-01 4.46171075e-01 -3.70538116e-01 1.22263052e-01 1.53829530e-01 -4.02387351e-01 5.84768802e-02 6.82147264e-01 -7.82617554e-02 -1.00511813e+00 -5.14468670e-01 -1.24334663e-01 7.69231439e-01 6.98846996e-01 4.22476321e-01 -1.63848281e-01 2.29915485e-01 -1.26838937e-01 5.34093320e-01 4.87505883e-01 3.26114208e-01 9.34064150e-01 2.38339797e-01 -6.32194519e-01 -1.21046042e+00 -1.29975855e+00 -3.68667990e-01 5.27301013e-01 1.79416403e-01 -4.73505467e-01 -4.92750674e-01 -3.42669874e-01 2.97081202e-01 5.82868516e-01 -2.89647788e-01 2.35023677e-01 -8.37070644e-01 -4.14824307e-01 5.38908124e-01 5.82794785e-01 3.07948351e-01 -8.77999544e-01 -6.65859163e-01 -2.92180888e-02 1.91434428e-01 -1.13567126e+00 -2.67320931e-01 -9.05527994e-02 -1.42472959e+00 -9.55100000e-01 -5.98497570e-01 -7.07785070e-01 1.13593674e+00 3.89635682e-01 1.09634662e+00 4.05550003e-01 -2.41477147e-01 6.43903255e-01 -2.23613441e-01 -4.33403522e-01 -2.95731038e-01 -1.72211528e-01 -5.40565960e-02 -2.11119696e-01 1.24811092e-02 -1.13384771e+00 -6.60352290e-01 8.65769014e-02 -9.31872129e-01 3.92280847e-01 5.29030740e-01 4.92046803e-01 1.30017972e+00 -1.71405256e-01 -1.76083278e-02 -5.94926298e-01 1.53778210e-01 -2.14438692e-01 -6.37155116e-01 -1.65541217e-01 -1.12112194e-01 1.11031748e-01 4.89926934e-01 -3.90081316e-01 -8.33347321e-01 5.18750072e-01 -3.93742204e-01 -6.46204889e-01 -4.49915111e-01 1.86222374e-01 -1.46754980e-01 -3.58612806e-01 4.14506644e-01 4.42054272e-01 5.74785620e-02 -1.00149906e+00 5.81806123e-01 1.91142023e-01 3.52319688e-01 -1.10674965e+00 1.09529793e+00 9.26751375e-01 3.56314301e-01 -9.92272198e-01 -6.22671187e-01 -2.76351541e-01 -1.15903652e+00 2.48301700e-01 4.59843904e-01 -9.09007251e-01 -5.67057073e-01 4.51424122e-01 -1.31447041e+00 -5.78126252e-01 -5.10723650e-01 1.25067085e-01 -9.90067244e-01 5.45595586e-01 -6.15067124e-01 -8.39744925e-01 -4.96933669e-01 -9.91532743e-01 1.35248232e+00 -7.54594132e-02 -9.49785262e-02 -8.01785409e-01 -1.19142078e-01 8.85325223e-02 1.92382231e-01 6.30240142e-01 9.95622575e-01 -1.64509282e-01 -1.01641893e+00 -8.60725865e-02 -2.36998618e-01 9.57834274e-02 2.82688409e-01 -5.73910326e-02 -8.94424081e-01 -1.27572909e-01 6.81835264e-02 -2.63284504e-01 6.60819232e-01 2.84225255e-01 1.48547626e+00 -3.63031775e-01 -1.98384032e-01 1.13564479e+00 1.50950122e+00 -1.15890853e-01 5.79755783e-01 1.04118399e-01 9.25964296e-01 1.98447660e-01 1.22482628e-01 6.01408899e-01 4.80402350e-01 5.78923941e-01 5.58280885e-01 -1.25447422e-01 -2.73113906e-01 -4.65360045e-01 1.48082327e-03 1.00047803e+00 -6.54092729e-01 3.64535719e-01 -9.82993126e-01 3.24299783e-01 -1.68170476e+00 -5.05962431e-01 -1.65489778e-01 2.26573396e+00 9.00993884e-01 -8.30014646e-02 1.05844237e-01 8.85948241e-02 1.84641019e-01 -1.30321048e-02 -5.72619498e-01 -2.95494705e-01 -1.75479308e-01 5.85737526e-01 3.33419651e-01 5.29344559e-01 -8.85225713e-01 1.11648238e+00 6.20781898e+00 8.66468072e-01 -1.09250259e+00 -5.30896410e-02 3.53189647e-01 9.23684314e-02 -7.96472549e-01 1.54017255e-01 -6.63750947e-01 4.55663837e-02 3.64747316e-01 -6.44582734e-02 4.64761734e-01 7.85984218e-01 -3.24415229e-02 1.27996966e-01 -9.82236326e-01 1.14789534e+00 -1.14516065e-01 -1.35455143e+00 4.53251749e-01 1.13467172e-01 6.30722284e-01 1.40569910e-01 -6.20571598e-02 -1.23996086e-01 2.39530608e-01 -1.22952425e+00 8.69502664e-01 6.54368997e-01 1.07856095e+00 -5.84260285e-01 3.77892256e-02 5.38618505e-01 -1.40218580e+00 3.99804890e-01 -7.58546233e-01 -1.73040494e-01 1.48925647e-01 5.95805109e-01 -6.22388601e-01 5.08305132e-01 5.12986422e-01 5.98103285e-01 -2.07380876e-01 1.02980316e+00 -5.53305328e-01 3.99693996e-01 -8.62216711e-01 2.32392177e-01 1.08440183e-01 -6.46672785e-01 6.65823400e-01 1.06219196e+00 4.94336307e-01 8.46182466e-01 2.32882217e-01 1.38532925e+00 -1.84501782e-01 9.44413394e-02 -8.82418036e-01 3.20133977e-02 4.89046454e-01 1.11656606e+00 -8.49649966e-01 -1.81044698e-01 -3.55458379e-01 8.93565238e-01 4.24149215e-01 3.58283401e-01 -5.06951988e-01 1.31026655e-02 7.02941000e-01 3.44541669e-01 4.30943489e-01 -8.85175884e-01 -8.08930159e-01 -1.13819420e+00 4.11097556e-02 -5.13001978e-01 1.44546121e-01 -7.79648840e-01 -1.27985752e+00 3.96992773e-01 9.55048501e-02 -1.24286735e+00 1.67125031e-01 -8.65731299e-01 -5.88357389e-01 7.91815042e-01 -1.53839767e+00 -1.47955430e+00 -2.30342478e-01 5.76903343e-01 2.25629762e-01 2.96659350e-01 1.23482084e+00 -1.24911070e-01 -8.21778737e-03 2.63821721e-01 -1.17618561e-01 2.44305521e-01 1.05558090e-01 -1.13259721e+00 7.17573643e-01 4.15310293e-01 2.35219702e-01 6.15521610e-01 1.57480821e-01 -6.03069246e-01 -1.86880672e+00 -7.86308467e-01 5.28469145e-01 -6.12324178e-01 3.54533404e-01 -5.02874136e-01 -1.09081721e+00 7.67393529e-01 -3.96364897e-01 2.47763768e-01 6.95288420e-01 -2.34961659e-01 -5.48836768e-01 3.42853367e-01 -1.22901261e+00 7.28796184e-01 1.52251291e+00 -5.49613714e-01 -6.24396861e-01 1.97982609e-01 6.93667054e-01 -7.47621119e-01 -1.12091005e+00 5.70400178e-01 5.80201209e-01 -8.35624874e-01 1.48426449e+00 -3.14967066e-01 1.99680954e-01 -4.42919135e-01 -3.57786775e-01 -9.86489713e-01 -3.84553134e-01 -6.21389985e-01 -4.52865720e-01 7.24208653e-01 3.12745236e-02 -6.99946284e-01 8.09634745e-01 7.83461392e-01 -2.17494488e-01 -1.08169174e+00 -1.05151129e+00 -8.57910693e-01 4.04596150e-01 -5.64565241e-01 8.92485023e-01 7.99741387e-01 -5.00900805e-01 1.16551258e-02 4.07483838e-02 1.55626267e-01 7.72352457e-01 7.09778428e-01 9.29581225e-01 -1.45089328e+00 -1.75830364e-01 -6.18144274e-01 -4.03566629e-01 -1.54973638e+00 9.32822675e-02 -1.19431782e+00 -2.90848285e-01 -1.64173806e+00 -2.99927704e-02 -1.13537562e+00 1.24644963e-02 7.83498406e-01 2.01895416e-01 3.88162434e-01 2.67424613e-01 3.76768976e-01 -2.20474496e-01 8.95760894e-01 1.70352101e+00 -1.49365338e-02 -2.87995517e-01 -2.58199088e-02 -5.59073687e-01 9.87839282e-01 8.75917852e-01 -3.85034204e-01 -1.10449940e-01 -6.87280416e-01 -1.86453089e-02 -1.73010990e-01 4.48687643e-01 -5.56878150e-01 -3.57073396e-02 -2.38691345e-01 4.18076307e-01 -7.87233353e-01 8.41946840e-01 -8.90941203e-01 3.81216556e-01 3.77074242e-01 2.11483464e-01 -2.46474743e-01 3.51628661e-01 2.71411330e-01 2.11590394e-01 -2.52589136e-01 8.02356720e-01 -4.79528338e-01 -3.43885928e-01 8.34976912e-01 -1.07363248e-02 -1.32878721e-01 6.13775969e-01 -5.24534702e-01 1.72205493e-01 -3.19143564e-01 -5.24179876e-01 -2.22717866e-01 1.02506340e+00 4.74990495e-02 9.82218325e-01 -1.71906352e+00 -5.68810761e-01 3.32680136e-01 -1.70248523e-01 8.44177425e-01 -1.67482197e-01 5.80639660e-01 -8.16717744e-01 1.62735000e-01 -2.24864140e-01 -7.40348816e-01 -8.33535731e-01 2.70209879e-01 4.00696307e-01 1.29291788e-01 -1.07604647e+00 7.28217542e-01 5.01181781e-01 -9.41273451e-01 1.49496738e-02 -6.34498358e-01 3.10880691e-01 -4.12248731e-01 1.64489135e-01 2.88109958e-01 -1.70037448e-01 -8.84062707e-01 -2.47505009e-01 1.19907761e+00 6.04025066e-01 1.70542136e-01 1.68623495e+00 2.53158569e-01 -2.86257386e-01 3.93661797e-01 1.05479932e+00 2.51657534e-02 -1.47303593e+00 -3.58157516e-01 -3.15559566e-01 -6.40561819e-01 -5.65598346e-02 -3.64290446e-01 -9.46319938e-01 9.68624294e-01 3.41800600e-02 -1.00614123e-01 1.05400550e+00 2.79360682e-01 8.20582807e-01 5.59587419e-01 7.86610067e-01 -5.28842568e-01 -1.40320025e-02 7.23086476e-01 1.36961925e+00 -1.03959751e+00 2.67809957e-01 -9.32609737e-01 -2.37056389e-01 1.34303284e+00 1.33480325e-01 -5.81893027e-01 1.01195896e+00 3.64676297e-01 -1.18253402e-01 -3.11665148e-01 -2.91511267e-01 -3.18598270e-01 5.20925105e-01 7.31619596e-01 2.57024199e-01 2.89994031e-01 6.36280281e-04 5.58897078e-01 -3.08429211e-01 -1.67330101e-01 2.32691541e-01 8.24750900e-01 -3.15319449e-01 -1.27134740e+00 -4.06485736e-01 1.75772324e-01 -1.62685290e-01 -5.53801768e-02 -4.47509587e-01 7.12191939e-01 -1.35337546e-01 2.23483264e-01 9.77279618e-02 -4.02426422e-02 3.98991764e-01 7.90749714e-02 9.04333174e-01 -7.80226111e-01 -1.77871034e-01 3.37155879e-01 -2.41445526e-01 -6.77496731e-01 -4.59768862e-01 -6.35009050e-01 -1.76991069e+00 -2.38986075e-01 -4.27981429e-02 -4.14174259e-01 5.76176882e-01 6.13129854e-01 5.85235775e-01 -8.64200592e-02 5.94694912e-01 -1.49790609e+00 -4.21443582e-01 -5.07613242e-01 -4.40658003e-01 4.78894800e-01 9.87589136e-02 -8.69891942e-01 -2.16568083e-01 -2.71522589e-02]
[8.643349647521973, -3.6448042392730713]
6caa5239-725b-48a7-99a8-7ea93e6c36a8
physiomtl-personalizing-physiological
2203.12595
null
https://arxiv.org/abs/2203.12595v1
https://arxiv.org/pdf/2203.12595v1.pdf
PhysioMTL: Personalizing Physiological Patterns using Optimal Transport Multi-Task Regression
Heart rate variability (HRV) is a practical and noninvasive measure of autonomic nervous system activity, which plays an essential role in cardiovascular health. However, using HRV to assess physiology status is challenging. Even in clinical settings, HRV is sensitive to acute stressors such as physical activity, mental stress, hydration, alcohol, and sleep. Wearable devices provide convenient HRV measurements, but the irregularity of measurements and uncaptured stressors can bias conventional analytical methods. To better interpret HRV measurements for downstream healthcare applications, we learn a personalized diurnal rhythm as an accurate physiological indicator for each individual. We develop Physiological Multitask-Learning (PhysioMTL) by harnessing Optimal Transport theory within a Multitask-learning (MTL) framework. The proposed method learns an individual-specific predictive model from heterogeneous observations, and enables estimation of an optimal transport map that yields a push forward operation onto the demographic features for each task. Our model outperforms competing MTL methodologies on unobserved predictive tasks for synthetic and two real-world datasets. Specifically, our method provides remarkable prediction results on unseen held-out subjects given only $20\%$ of the subjects in real-world observational studies. Furthermore, our model enables a counterfactual engine that generates the effect of acute stressors and chronic conditions on HRV rhythms.
['Shirley You Ren', 'XuanLong Nguyen', 'Bo Li', 'Ding Zhao', 'Agni Kumar', 'Gregory Darnell', 'Jiacheng Zhu']
2022-03-19
null
null
null
null
['heart-rate-variability']
['medical']
[ 2.24271595e-01 -3.50859284e-01 -3.43113035e-01 -4.64143455e-01 -2.92262644e-01 -2.77387142e-01 1.33584350e-01 2.87847072e-01 -1.88405216e-01 1.21859121e+00 2.74204940e-01 -1.66667029e-01 -4.52356130e-01 -6.46896660e-01 -4.42483276e-01 -8.27790082e-01 -4.59973335e-01 4.34181422e-01 -5.59684753e-01 7.28297383e-02 -1.88935548e-01 1.35028824e-01 -1.27506626e+00 -1.95110872e-01 9.76349056e-01 9.46935177e-01 -1.09146304e-01 7.19390631e-01 2.97262877e-01 2.74229974e-01 -5.97931147e-01 -2.44128015e-02 1.09941937e-01 -7.37600505e-01 -2.22109839e-01 -2.75763422e-01 1.95238784e-01 1.11837961e-01 8.45553540e-03 4.43381041e-01 8.60267580e-01 1.86541364e-01 6.55792177e-01 -1.05307567e+00 -4.92772937e-01 2.80614406e-01 -2.81479120e-01 4.86072838e-01 6.56728223e-02 4.33300853e-01 7.49121666e-01 -1.96321189e-01 -9.43990946e-02 7.92412698e-01 1.03955078e+00 3.61337215e-01 -1.66027904e+00 -5.21999359e-01 -1.29771903e-01 -4.57371995e-02 -1.25680995e+00 -5.37108481e-01 6.32058382e-01 -6.64474010e-01 4.31140006e-01 4.57287073e-01 8.63984346e-01 1.61845875e+00 8.90098989e-01 -1.26034155e-01 1.39416194e+00 2.46451452e-01 3.53857398e-01 7.30567947e-02 3.56434286e-02 4.87888962e-01 3.42606246e-01 3.24013799e-01 -6.65615439e-01 -5.64344108e-01 6.75300360e-01 5.12293696e-01 -3.07050049e-01 -1.08702099e-02 -1.47556603e+00 3.54629099e-01 7.37210736e-02 -8.58265907e-02 -6.48647189e-01 1.53072318e-02 6.57244384e-01 3.10806751e-01 7.24664688e-01 3.42055798e-01 -8.67650092e-01 -1.04239630e-02 -8.33233595e-01 1.49705961e-01 7.96831071e-01 3.59273911e-01 5.29936314e-01 5.61429970e-02 -6.08316779e-01 8.25447977e-01 5.60108460e-02 8.37884843e-01 5.86127341e-01 -1.15370178e+00 1.18343249e-01 2.63437539e-01 2.30784118e-01 -9.62115824e-01 -1.00664806e+00 -9.37002361e-01 -1.57214475e+00 -4.99512911e-01 4.25642282e-01 -2.89638698e-01 -3.76236349e-01 1.92623997e+00 3.53486955e-01 5.10524511e-01 -2.05881462e-01 9.62148070e-01 4.54925895e-01 2.29626015e-01 4.45144355e-01 -9.49014127e-01 1.57761967e+00 -3.86853814e-01 -7.28833437e-01 -1.70102775e-01 3.26414645e-01 -2.48950765e-01 1.17384744e+00 3.47297490e-01 -6.17469311e-01 -7.31290460e-01 -6.66262507e-01 3.14005613e-01 -1.75714895e-01 -4.70449924e-02 5.30128002e-01 9.07546341e-01 -5.80497086e-01 8.14738452e-01 -7.88173497e-01 -4.82220232e-01 3.91936451e-01 2.01086160e-02 1.24453999e-01 3.25092338e-02 -1.32725716e+00 6.48010373e-01 -1.73660833e-02 4.08307463e-01 -8.37271035e-01 -1.44841838e+00 -6.28891766e-01 -2.63052676e-02 1.37140393e-01 -1.44640958e+00 6.37955427e-01 -3.16024661e-01 -1.48209846e+00 5.72461307e-01 -3.42034608e-01 -4.10791546e-01 5.23459613e-01 -1.87598526e-01 -4.93433326e-01 -8.99739340e-02 -7.98508897e-02 -6.23985380e-02 1.02989197e+00 -7.69790947e-01 2.37644866e-01 -6.23008251e-01 -7.58900583e-01 5.52723221e-02 -1.03315093e-01 -4.19238448e-01 6.19281471e-01 -4.44790423e-01 -4.31348085e-02 -8.89791429e-01 -2.86538929e-01 -1.53587997e-01 -3.89734060e-01 2.60296106e-01 2.94209927e-01 -5.57175100e-01 1.25001311e+00 -1.89981878e+00 5.70170581e-02 1.02274574e-01 5.66211045e-01 -1.02149308e-01 2.51923680e-01 2.96615660e-01 1.58863589e-01 2.60341734e-01 -2.68381327e-01 -2.55524397e-01 -1.07187554e-01 2.69694686e-01 -2.42772743e-01 7.24578559e-01 -2.44904488e-01 9.46560144e-01 -9.51961517e-01 -3.97301555e-01 2.24532261e-01 4.74299997e-01 -1.56034604e-01 1.56287268e-01 1.21134263e-03 1.03475606e+00 -2.74505854e-01 6.42822862e-01 2.04905212e-01 -3.79852295e-01 3.58388275e-01 -8.56945738e-02 9.94338766e-02 -6.99667856e-02 -6.04584932e-01 1.49347568e+00 -6.17116272e-01 2.21206382e-01 -3.72246355e-01 -1.21495938e+00 1.19981229e+00 4.82081592e-01 8.07432532e-01 -7.79837132e-01 -8.78254771e-02 5.11934571e-02 1.34849995e-01 -7.97080934e-01 -2.52023697e-01 -7.33210206e-01 -2.45411158e-01 4.65858638e-01 -1.89841211e-01 3.51972759e-01 -3.66609395e-01 -4.85539883e-01 1.12999988e+00 1.44313782e-01 7.30656147e-01 -5.57952762e-01 2.51343101e-01 -4.90890831e-01 1.05844128e+00 8.74279916e-01 -6.65312052e-01 3.95231634e-01 5.00114799e-01 -9.58177865e-01 -8.61413360e-01 -1.23176968e+00 -4.85889882e-01 9.51261163e-01 -3.42039764e-01 5.72202206e-02 -1.52009428e-01 -1.74880534e-01 3.86828005e-01 3.77862185e-01 -7.40823328e-01 -4.15673345e-01 -3.57243806e-01 -1.69062734e+00 7.44264603e-01 1.98902056e-01 1.85274705e-01 -8.27273607e-01 -7.46872306e-01 3.88745040e-01 -5.71877182e-01 -1.02108526e+00 -4.15747911e-01 7.55672008e-02 -1.12512732e+00 -9.85530615e-01 -5.41793823e-01 7.35570937e-02 1.57628879e-01 -5.61168455e-02 1.30228984e+00 -1.78354159e-01 -7.69653440e-01 1.99122176e-01 1.52820110e-01 -5.66729128e-01 -1.65182561e-01 2.27271706e-01 6.21900976e-01 3.16862762e-01 2.28984162e-01 -1.14480913e+00 -1.22511351e+00 4.31783915e-01 -2.53821880e-01 -2.93621480e-01 4.02795523e-01 7.56253839e-01 7.95159221e-01 -3.31535369e-01 1.27644730e+00 -9.10469830e-01 4.14481342e-01 -8.85404468e-01 -1.81751356e-01 7.46970251e-02 -1.00169098e+00 -1.18732378e-01 8.02200437e-01 -4.28225785e-01 -8.11534703e-01 -1.93852022e-01 5.09007573e-01 -4.23214585e-01 -1.77551359e-01 3.23026210e-01 1.27039641e-01 5.29884875e-01 9.60419774e-01 4.25684720e-01 1.87711522e-01 -3.06132466e-01 2.50458531e-02 3.74679506e-01 3.61813873e-01 -8.69356394e-01 4.31640923e-01 3.29781294e-01 6.35168314e-01 -1.13825369e+00 -1.20805204e+00 -3.05661201e-01 -5.88802814e-01 -3.24926764e-01 9.58386302e-01 -1.12775946e+00 -1.40995860e+00 1.98633835e-01 -5.85276783e-01 -5.80485225e-01 -2.02527925e-01 6.60215378e-01 -7.27661788e-01 -3.35373394e-02 -4.88451362e-01 -1.05371094e+00 -5.44492364e-01 -5.76283991e-01 8.71814430e-01 1.01011328e-01 -5.54426610e-01 -1.42986465e+00 5.54632425e-01 5.57742655e-01 6.17886543e-01 8.54047835e-01 1.06762886e+00 -4.78673160e-01 -1.86336771e-01 7.22181872e-02 2.38525122e-02 2.05558419e-01 4.06085700e-01 -2.61468768e-01 -1.24008358e+00 -1.60375342e-01 1.87916070e-01 -2.54063785e-01 6.42063022e-01 8.79212976e-01 1.29880524e+00 -4.05306309e-01 -2.88051844e-01 7.45987594e-01 1.28549778e+00 -1.01377666e-01 4.52581495e-01 -2.48503804e-01 7.98024118e-01 6.05653882e-01 1.63533874e-02 7.10378528e-01 6.92824781e-01 4.23308522e-01 1.19613573e-01 -1.19306199e-01 2.95338869e-01 -2.72028539e-02 1.73524588e-01 9.18678582e-01 -5.23028076e-01 1.67942360e-01 -8.41419339e-01 2.44424582e-01 -1.71980762e+00 -1.22051978e+00 -3.49232078e-01 2.70232582e+00 9.03028309e-01 -1.56757787e-01 4.60454375e-01 -2.77511716e-01 4.69314396e-01 1.93931967e-01 -1.02331769e+00 -3.38391751e-01 -1.38407797e-01 2.07784280e-01 4.51380223e-01 1.46390602e-01 -8.47056687e-01 -6.59387112e-02 6.93732166e+00 -1.76189259e-01 -1.08030403e+00 3.19990695e-01 1.02467322e+00 -5.55414200e-01 -8.11162591e-03 -3.96458119e-01 -4.09900129e-01 6.29504204e-01 1.58317828e+00 -4.94172037e-01 5.79713345e-01 3.84546638e-01 1.03816855e+00 8.27012062e-02 -1.10866308e+00 9.06177163e-01 -4.51761894e-02 -8.20801616e-01 -5.69647193e-01 3.31617832e-01 4.46267217e-01 1.30934402e-01 -6.45913091e-03 5.27333915e-01 -2.04836980e-01 -1.01276159e+00 1.08011467e-02 1.22962499e+00 8.48191798e-01 -4.94318753e-01 6.87682390e-01 4.27441657e-01 -1.01174343e+00 -2.77958512e-01 -3.45612913e-01 -3.16074252e-01 8.98379982e-02 1.40551054e+00 -6.68220341e-01 4.74800438e-01 4.48054463e-01 6.82433665e-01 -2.38783613e-01 7.50143468e-01 2.60281295e-01 7.32227147e-01 -1.83069423e-01 3.40529501e-01 -5.94199121e-01 -4.64956224e-01 5.25915980e-01 9.72204506e-01 4.68181789e-01 3.74439359e-02 3.78189087e-01 9.22448516e-01 -5.31909764e-02 1.29604056e-01 -7.71489739e-01 1.50618702e-01 3.71491551e-01 1.37215364e+00 -4.57415044e-01 -8.99232328e-02 -1.62148952e-01 7.91421831e-01 6.74168346e-03 4.76786286e-01 -8.87126386e-01 2.46059112e-02 1.13060784e+00 2.60878146e-01 -2.73325294e-01 -1.31783694e-01 -6.22081459e-01 -1.23053372e+00 1.19012944e-01 -5.97471833e-01 4.40672815e-01 -3.12079877e-01 -1.44079101e+00 1.66399166e-01 -7.01446384e-02 -1.11884212e+00 -1.67404003e-02 -5.26182055e-02 -7.39834964e-01 1.18388939e+00 -1.53578603e+00 -6.07726753e-01 -5.13821840e-01 1.88134432e-01 3.44954967e-01 1.47053421e-01 1.06270158e+00 3.57281744e-01 -1.06451619e+00 3.04483682e-01 2.23117303e-02 -2.72678524e-01 1.02202106e+00 -1.29581749e+00 5.39597012e-02 3.64109218e-01 -4.04200166e-01 6.00626707e-01 7.47458875e-01 -6.33390784e-01 -1.32506514e+00 -1.50932205e+00 7.99848676e-01 -7.30255485e-01 4.99245673e-01 -3.61773729e-01 -9.14808333e-01 3.52292597e-01 -4.19904888e-01 4.62181598e-01 1.38740933e+00 5.47604203e-01 -2.28028506e-01 -6.24440014e-01 -1.07361257e+00 1.65076777e-01 1.10862780e+00 -4.61604208e-01 -2.85330534e-01 4.25150961e-01 6.89290881e-01 -2.54631966e-01 -1.61963665e+00 2.24516883e-01 9.58800673e-01 -9.09535348e-01 1.05903387e+00 -7.40401626e-01 2.76744157e-01 -7.24011138e-02 6.77231252e-02 -1.56753576e+00 -2.73406237e-01 -8.86445165e-01 -3.78677666e-01 8.08529615e-01 2.95323402e-01 -1.02451956e+00 2.56895661e-01 6.62169695e-01 -7.13138506e-02 -8.02254617e-01 -8.39163959e-01 -8.54699492e-01 -1.91119656e-01 -2.59118795e-01 6.02046013e-01 1.12102950e+00 3.09131807e-03 4.28684145e-01 -6.75003648e-01 1.13289401e-01 1.17826080e+00 1.43550530e-01 4.78806436e-01 -1.76022565e+00 -3.91929209e-01 -1.19179383e-01 -8.01040158e-02 -1.96956679e-01 1.14074081e-01 -7.63875306e-01 -1.70175821e-01 -1.12502587e+00 3.34101796e-01 -4.70594108e-01 -7.49807596e-01 2.37780675e-01 -4.54612017e-01 1.45768330e-01 -1.84810087e-01 2.59434193e-01 -2.57086843e-01 5.94585180e-01 1.25067365e+00 2.41925508e-01 -6.50590122e-01 3.25969547e-01 -7.09791958e-01 3.48921001e-01 1.17914677e+00 -6.28490984e-01 -6.37406886e-01 1.90319031e-01 3.89536321e-01 5.62695742e-01 6.73940182e-01 -8.42446148e-01 -2.66379893e-01 -4.53945488e-01 7.18048573e-01 6.96538612e-02 1.74544364e-01 -4.36505556e-01 5.15153944e-01 7.05674112e-01 -3.20504665e-01 6.83601946e-02 -5.52687161e-02 9.61211920e-01 3.62840146e-01 6.62864268e-01 5.63769996e-01 -2.85946548e-01 2.94897050e-01 5.16082823e-01 -2.27208838e-01 3.74491900e-01 7.33493567e-01 3.71312164e-02 -5.44680595e-01 -1.38843790e-01 -9.70492959e-01 2.89694786e-01 1.12242326e-02 2.50508636e-01 3.19607407e-01 -1.16866481e+00 -9.24447119e-01 2.03732237e-01 1.30128384e-01 -2.36424997e-01 6.98699534e-01 1.49136198e+00 1.39567843e-02 2.93730468e-01 -1.28780857e-01 -7.19375968e-01 -6.54614747e-01 5.41231334e-01 5.64680815e-01 -2.21209917e-02 -7.14680493e-01 -5.39456308e-02 4.95123006e-02 -4.00884986e-01 -1.30369172e-01 -4.73454148e-01 -1.22437123e-02 3.43251079e-01 5.38818538e-01 8.14168990e-01 1.09460643e-02 -1.02686450e-01 -2.29300380e-01 2.80080885e-01 5.25935888e-01 4.14173514e-01 1.13098192e+00 -5.32527447e-01 -9.94482338e-02 1.33072793e+00 1.00307477e+00 -4.68080997e-01 -1.08302438e+00 -2.30706111e-01 -2.14211076e-01 -1.24317318e-01 -1.88444704e-01 -7.55253434e-01 -6.93204105e-01 7.99823701e-01 7.16794133e-01 4.11251336e-01 1.09500444e+00 -3.57883662e-01 7.60206521e-01 3.17877918e-01 2.19719797e-01 -8.25017631e-01 -1.38355955e-01 -6.60682023e-02 5.54273367e-01 -1.24897742e+00 1.72909215e-01 4.14963532e-03 -6.21231735e-01 8.46527278e-01 2.76322931e-01 3.19225103e-01 8.83470237e-01 -1.66097477e-01 1.68727905e-01 -8.59267861e-02 -1.05099165e+00 1.46538466e-01 8.96191150e-02 7.00539887e-01 4.54780191e-01 4.87996519e-01 -2.08035484e-01 6.10056579e-01 -1.89802259e-01 1.72894165e-01 4.02425677e-01 2.58698344e-01 -2.26635098e-01 -6.22347534e-01 -2.66394585e-01 9.87809002e-01 -5.23629308e-01 2.22064946e-02 2.55255848e-01 3.68728936e-01 6.53325096e-02 5.90356708e-01 8.02931264e-02 -1.16144665e-01 4.51478183e-01 7.13415861e-01 2.67754585e-01 -5.73336303e-01 -4.98327762e-01 -1.20857887e-01 -1.16168588e-01 -6.23019814e-01 -4.72735405e-01 -9.02733445e-01 -7.23380625e-01 -4.12675917e-01 4.50055331e-01 -1.64481610e-01 5.26647866e-01 1.16910934e+00 6.73600018e-01 8.80547345e-01 9.72483814e-01 -7.12327600e-01 -6.40828907e-01 -9.15141702e-01 -7.45358706e-01 2.90222615e-01 6.90179467e-01 -6.74162328e-01 -5.00784874e-01 3.05872798e-01]
[13.708244323730469, 3.1468279361724854]
1a65d517-1bb0-4ddd-8848-55cdc2edf1d5
sequential-end-to-end-network-for-efficient
2103.10148
null
https://arxiv.org/abs/2103.10148v1
https://arxiv.org/pdf/2103.10148v1.pdf
Sequential End-to-end Network for Efficient Person Search
Person search aims at jointly solving Person Detection and Person Re-identification (re-ID). Existing works have designed end-to-end networks based on Faster R-CNN. However, due to the parallel structure of Faster R-CNN, the extracted features come from the low-quality proposals generated by the Region Proposal Network, rather than the detected high-quality bounding boxes. Person search is a fine-grained task and such inferior features will significantly reduce re-ID performance. To address this issue, we propose a Sequential End-to-end Network (SeqNet) to extract superior features. In SeqNet, detection and re-ID are considered as a progressive process and tackled with two sub-networks sequentially. In addition, we design a robust Context Bipartite Graph Matching (CBGM) algorithm to effectively employ context information as an important complementary cue for person matching. Extensive experiments on two widely used person search benchmarks, CUHK-SYSU and PRW, have shown that our method achieves state-of-the-art results. Also, our model runs at 11.5 fps on a single GPU and can be integrated into the existing end-to-end framework easily.
['Duoqian Miao', 'Zhengjia Li']
2021-03-18
null
null
null
null
['person-search']
['computer-vision']
[-1.79423109e-01 -2.47957140e-01 9.92040038e-02 -3.44708890e-01 -4.28714722e-01 -1.41381323e-01 3.29930335e-01 -1.09693468e-01 -1.01866496e+00 4.67701197e-01 3.78954053e-01 1.15927957e-01 -1.18202250e-02 -8.70606065e-01 -3.81468922e-01 -1.86105773e-01 6.44027144e-02 7.08941221e-01 4.06027079e-01 -1.89987466e-01 -2.06745192e-01 4.94281530e-01 -1.71737480e+00 3.22301276e-02 7.67887592e-01 7.79493332e-01 2.07150411e-02 6.56628728e-01 2.54653066e-01 3.69099528e-01 -5.31323075e-01 -8.81832540e-01 5.40556848e-01 -1.05920069e-01 -7.17320502e-01 -1.78147152e-01 9.48225498e-01 -5.55623710e-01 -9.77887094e-01 1.15868640e+00 1.10490394e+00 5.24600685e-01 1.10838734e-01 -1.16557026e+00 -6.90747082e-01 2.29292512e-01 -8.53750646e-01 3.05813104e-01 6.82739377e-01 4.87941384e-01 9.26677585e-01 -9.22871470e-01 4.75223720e-01 1.56848502e+00 9.30992544e-01 7.98736334e-01 -1.01292264e+00 -8.61545444e-01 3.95897895e-01 2.57363290e-01 -1.71250665e+00 -2.88824499e-01 4.68867838e-01 -6.28763437e-02 9.36501265e-01 3.35485697e-01 1.00105512e+00 9.95618999e-01 -3.04364473e-01 8.80214155e-01 6.50265396e-01 -1.02187686e-01 -2.84818500e-01 -1.29358619e-01 3.46261024e-01 9.53766882e-01 4.80888873e-01 4.57382947e-01 -4.75857168e-01 -2.85819411e-01 8.77394319e-01 3.12177926e-01 -3.02343160e-01 -9.48936492e-02 -1.03466749e+00 6.01512671e-01 9.71705437e-01 -1.71358380e-02 -2.15172514e-01 2.85445541e-01 5.39221466e-01 1.77851282e-02 1.99200079e-01 -2.12263957e-01 -1.06140092e-01 1.41431659e-01 -9.15065110e-01 6.90778732e-01 6.89594865e-01 1.16792083e+00 5.75142264e-01 -5.26105642e-01 -8.08912218e-01 1.01577616e+00 3.70311439e-01 6.82052314e-01 1.98518232e-01 -5.21258235e-01 5.89380980e-01 8.58883262e-01 9.55126882e-02 -1.16275489e+00 -5.63259721e-01 -7.58744359e-01 -1.13536942e+00 -3.00091743e-01 6.07788920e-01 2.50737630e-02 -1.03646767e+00 1.69242239e+00 4.29220498e-01 3.60919476e-01 -3.65523219e-01 1.61739218e+00 1.31596887e+00 2.17630669e-01 2.72762984e-01 5.26471853e-01 1.89230418e+00 -1.50471568e+00 -2.20625252e-01 -4.16772604e-01 3.36172074e-01 -4.64823633e-01 8.07962894e-01 -1.48663059e-01 -1.07990324e+00 -9.85170364e-01 -7.29239047e-01 -4.27740127e-01 -3.48821938e-01 4.26163256e-01 4.54265922e-01 6.93798602e-01 -1.12574434e+00 4.05445218e-01 -2.49741837e-01 -5.26972413e-01 6.08743727e-01 5.55481136e-01 -4.61451352e-01 -2.68172473e-01 -1.29180741e+00 4.95145380e-01 2.78431833e-01 5.56591570e-01 -3.97561163e-01 -4.20926422e-01 -7.02449262e-01 2.64656395e-01 4.15243566e-01 -1.33501768e+00 9.69198406e-01 -6.09373152e-01 -1.15818417e+00 1.23676622e+00 -3.24956119e-01 -2.94717312e-01 8.62292647e-01 -3.02118272e-01 -3.40565145e-01 1.19192354e-01 3.03985894e-01 7.92854190e-01 5.53072989e-01 -7.60700703e-01 -9.00191963e-01 -7.09283829e-01 8.31484571e-02 1.87513515e-01 -2.74241805e-01 4.74535793e-01 -1.33825958e+00 -6.59519434e-01 -1.43000185e-01 -1.08645546e+00 -4.39277798e-01 3.04052532e-01 -4.08367932e-01 -6.18821919e-01 2.92994857e-01 -7.94033051e-01 1.12769663e+00 -1.80572653e+00 8.57538171e-03 3.31629872e-01 5.96735597e-01 4.58430767e-01 -2.38061413e-01 -6.90292418e-02 4.14716937e-02 -3.15140337e-01 1.19371958e-01 -6.92713439e-01 1.99505374e-01 -2.70593554e-01 2.56054461e-01 6.47907138e-01 -1.54762030e-01 1.24244380e+00 -9.48842108e-01 -6.68891490e-01 8.45932513e-02 4.51009035e-01 -4.14511830e-01 1.67911902e-01 3.48155737e-01 2.64894366e-01 -4.03424680e-01 8.72129321e-01 9.06311512e-01 -4.22610968e-01 -1.01181082e-01 -1.40798643e-01 9.34487954e-02 -1.96145654e-01 -1.28578532e+00 1.80685544e+00 2.40938459e-02 3.30377996e-01 1.02813374e-02 -5.77271760e-01 8.38159740e-01 -1.78200841e-01 1.60216123e-01 -9.79065299e-01 1.78759679e-01 9.36679468e-02 -2.33205482e-01 -1.14182577e-01 6.92745686e-01 6.24381125e-01 -5.65983467e-02 2.99355775e-01 -1.44203469e-01 8.94779682e-01 1.43088982e-01 1.88678011e-01 1.04234350e+00 3.34917270e-02 3.25179324e-02 -2.52882272e-01 8.69798601e-01 -1.18555762e-01 8.21836531e-01 1.14438212e+00 -7.68936217e-01 6.17022276e-01 -3.04307584e-02 -9.39653873e-01 -1.03722346e+00 -8.37752879e-01 1.79175138e-01 1.28176689e+00 6.38770461e-01 -5.45254827e-01 -7.71980524e-01 -7.89001763e-01 2.65346587e-01 -2.26763844e-01 -4.64258879e-01 3.41615863e-02 -9.58202541e-01 -7.28241742e-01 9.32880938e-01 8.08223844e-01 9.49905694e-01 -1.19150782e+00 -4.61738169e-01 2.24822298e-01 -2.39379868e-01 -1.24253571e+00 -1.13609958e+00 -5.38185477e-01 -4.34395552e-01 -1.30594921e+00 -1.32816637e+00 -9.96146679e-01 8.03768218e-01 6.34604454e-01 1.27981961e+00 6.06256485e-01 -7.40280211e-01 2.77435303e-01 -1.32403657e-01 -3.02555854e-03 4.94304478e-01 2.26364329e-01 5.77490665e-02 -6.23989031e-02 8.58649790e-01 -8.56829807e-02 -1.20762670e+00 5.91737092e-01 -2.86989957e-01 2.73313224e-02 4.10801739e-01 8.43089402e-01 5.07699668e-01 -1.93796590e-01 1.68042243e-01 -3.90688926e-01 6.20963216e-01 -1.60196349e-02 -6.47036910e-01 5.18627644e-01 -4.24135387e-01 -2.02699780e-01 3.53501320e-01 -4.36428010e-01 -8.97567809e-01 1.80850744e-01 -2.32288942e-01 -3.41256678e-01 -1.27634734e-01 -1.32736072e-01 -8.83615389e-02 -3.52765858e-01 5.10569334e-01 2.52369106e-01 -1.32883966e-01 -4.20434207e-01 2.74740189e-01 6.15720272e-01 7.90282786e-01 -5.25089204e-01 9.63924825e-01 5.58515251e-01 -1.30846605e-01 -4.12249893e-01 -6.95497036e-01 -9.88740683e-01 -4.86698896e-01 -3.42679232e-01 8.18234324e-01 -1.31094217e+00 -1.44490480e+00 6.93540215e-01 -1.25725174e+00 -6.96193427e-02 6.97609633e-02 1.99115992e-01 2.91905063e-03 5.63427031e-01 -7.56215513e-01 -7.59969115e-01 -1.11597514e+00 -1.02663112e+00 1.38276994e+00 7.48309076e-01 3.65282292e-03 -5.75143814e-01 -2.62719430e-02 6.27488196e-01 3.75975430e-01 -4.25029024e-02 -9.97767318e-03 -4.38860446e-01 -7.83775210e-01 -4.18377310e-01 -1.02017891e+00 -3.34719926e-01 -3.54273647e-01 -6.28221929e-01 -8.96249473e-01 -7.94901431e-01 -7.08299398e-01 -1.04025930e-01 1.26834583e+00 1.19284421e-01 1.18096626e+00 -5.93632832e-02 -7.26873040e-01 8.93315852e-01 1.24644399e+00 -3.90562832e-01 5.87412953e-01 4.36517149e-01 9.56686854e-01 6.03343248e-01 4.69198167e-01 3.39433730e-01 7.58370399e-01 1.04222298e+00 -2.22694478e-03 -3.58950615e-01 -4.81627434e-01 -4.55006331e-01 -4.90289107e-02 1.45907745e-01 -4.00064379e-01 -1.32894486e-01 -8.35910797e-01 4.42743719e-01 -2.44549012e+00 -1.07006371e+00 -2.86066324e-01 2.07151031e+00 4.14322346e-01 -6.56710044e-02 7.11798966e-01 -3.92381161e-01 1.21629632e+00 7.00136274e-02 -6.00292742e-01 4.78199124e-01 -1.27775908e-01 1.95465237e-02 7.24846840e-01 1.94768161e-01 -1.24434209e+00 1.05465829e+00 5.60397148e+00 9.20045853e-01 -4.61341202e-01 1.87039927e-01 4.64648455e-01 -1.78185076e-01 1.58267364e-01 -1.88931823e-01 -1.31366730e+00 6.04845226e-01 2.89574772e-01 9.04603526e-02 4.51552808e-01 9.50461924e-01 -3.36486287e-02 8.61675572e-03 -1.05131996e+00 1.65862286e+00 1.33178294e-01 -1.09356391e+00 -1.37802631e-01 8.48972425e-02 4.05235946e-01 -1.18977666e-01 -1.64630234e-01 5.04225194e-01 1.34865865e-01 -9.07825947e-01 6.10635698e-01 6.49765432e-01 7.86645889e-01 -9.51499343e-01 9.78211224e-01 2.14084178e-01 -1.89250481e+00 -2.33549818e-01 -7.92380452e-01 8.46739188e-02 2.53858835e-01 4.48750854e-01 -2.77661443e-01 6.23059928e-01 1.26190960e+00 5.17189920e-01 -8.88698339e-01 1.33824623e+00 4.95503396e-02 -6.60813823e-02 -4.30966407e-01 -1.81987390e-01 6.45333976e-02 -3.83661203e-02 4.75567639e-01 1.50925004e+00 -1.40570905e-02 1.52635545e-01 4.66578245e-01 9.73107159e-01 -3.25715512e-01 -8.22974592e-02 -9.66707692e-02 5.68620741e-01 3.08219433e-01 1.28439999e+00 -6.93165064e-01 -4.27643836e-01 -5.68168283e-01 1.31717002e+00 7.13878453e-01 2.95970142e-01 -9.74084496e-01 -4.78624135e-01 7.66010106e-01 2.18387693e-01 2.21871018e-01 6.76765144e-02 3.88388187e-02 -1.38288200e+00 3.15884858e-01 -7.85657763e-01 7.38024890e-01 -4.33559835e-01 -1.75003850e+00 5.89563668e-01 -3.11255008e-01 -8.82884026e-01 2.73359686e-01 -4.07917112e-01 -5.07636786e-01 1.20049691e+00 -1.65400791e+00 -1.43161094e+00 -1.07373297e+00 9.03501809e-01 3.88767660e-01 -3.81030083e-01 4.71706271e-01 7.87086546e-01 -8.71025264e-01 1.29110479e+00 -4.70722646e-01 7.88311064e-01 8.50834906e-01 -9.99325335e-01 1.01638198e+00 1.04659545e+00 -1.85090885e-01 7.82424152e-01 1.35157421e-01 -9.24458265e-01 -1.17752445e+00 -1.24189377e+00 9.52758431e-01 -3.50529462e-01 2.19854847e-01 -4.87236410e-01 -6.39261246e-01 3.40225279e-01 -1.48715764e-01 3.97315234e-01 3.91469955e-01 3.45462561e-01 -4.58312988e-01 -2.35402927e-01 -1.18128371e+00 7.31919825e-01 1.94418466e+00 -5.81223786e-01 -3.06308478e-01 3.54414403e-01 6.76360428e-01 -5.90430975e-01 -4.91532505e-01 2.38480717e-01 7.08904266e-01 -1.00009322e+00 1.41474652e+00 -5.09591639e-01 -2.22214997e-01 -4.82769728e-01 3.02208424e-01 -7.42120743e-01 -8.26982796e-01 -6.48703039e-01 -1.86757356e-01 1.02517104e+00 -1.71837527e-02 -7.22929120e-01 1.18483341e+00 9.00677443e-01 3.06580156e-01 -6.34909272e-01 -8.81267369e-01 -8.53154778e-01 -5.35567462e-01 -6.06557056e-02 9.59634006e-01 4.22620654e-01 -3.36927056e-01 1.56759247e-01 -5.92668772e-01 2.49391168e-01 1.21250081e+00 1.42922655e-01 1.03769183e+00 -1.36065495e+00 -3.28712106e-01 -5.24213254e-01 -5.50981462e-01 -1.50864947e+00 9.82086733e-02 -9.02887225e-01 1.78759061e-02 -1.38381839e+00 8.62238824e-01 -6.17041290e-01 -2.54730463e-01 4.07099307e-01 -6.80335760e-01 3.65991563e-01 4.86245632e-01 3.94815803e-01 -9.83107328e-01 5.15182674e-01 1.14348841e+00 -3.31063062e-01 -2.25052699e-01 1.23982012e-01 -5.57894528e-01 5.25036633e-01 5.42105734e-01 -3.37946355e-01 5.22010066e-02 -4.51369345e-01 1.38073489e-01 -8.49791542e-02 9.82284844e-01 -1.11834455e+00 9.08218384e-01 3.18458974e-01 8.70737493e-01 -8.54556978e-01 3.40247959e-01 -6.28205597e-01 -4.72120047e-02 5.83177865e-01 -1.34334564e-01 2.94620186e-01 -5.49481213e-02 7.39463329e-01 -1.21225312e-03 2.30774879e-02 6.95546508e-01 -2.51262635e-01 -8.47046554e-01 9.21805978e-01 3.03561598e-01 5.09531908e-02 8.99515569e-01 -4.88402188e-01 -3.55097920e-01 -1.35727227e-01 -5.68422616e-01 6.73611999e-01 3.86315197e-01 5.59496999e-01 7.94972658e-01 -1.43920159e+00 -8.72251034e-01 1.08923882e-01 1.82685450e-01 -3.56175401e-03 4.90505964e-01 5.09537935e-01 -4.84690070e-01 4.79270577e-01 1.90523136e-02 -5.70089996e-01 -1.54831254e+00 5.91154337e-01 5.55537283e-01 -4.82453585e-01 -9.02420163e-01 1.09155869e+00 8.31501931e-02 -5.12227178e-01 6.53008401e-01 2.80657023e-01 -1.68187857e-01 -2.08499685e-01 8.01783442e-01 5.91313541e-01 -3.27972591e-01 -7.34271646e-01 -6.41089201e-01 7.66409874e-01 -2.39590347e-01 2.19318241e-01 1.02754056e+00 -9.86485928e-02 -6.55812025e-02 -7.23127127e-01 1.06157696e+00 -3.58589113e-01 -1.08202195e+00 -5.35243630e-01 -1.82769358e-01 -6.99282587e-01 -2.18073636e-01 -5.46299994e-01 -1.17441082e+00 3.83628875e-01 9.63475168e-01 -3.57001424e-01 9.52651620e-01 1.15366783e-02 1.37739730e+00 4.50726360e-01 5.95822453e-01 -1.39592946e+00 3.31680663e-02 3.81833464e-01 6.87358797e-01 -1.38762951e+00 1.21948034e-01 -5.78617156e-01 -2.40342438e-01 8.92450631e-01 1.07851577e+00 -1.97272047e-01 4.44118589e-01 -1.95870370e-01 -3.02858680e-01 -2.31148973e-01 -8.55970010e-02 -7.58714080e-01 5.99090457e-01 6.22876167e-01 1.51526146e-02 9.66286957e-02 -1.21227853e-01 6.25398636e-01 -6.30110055e-02 1.64142758e-01 -2.60262132e-01 6.56321585e-01 -3.09877753e-01 -9.18868363e-01 -5.08211076e-01 3.40053856e-01 -1.75995663e-01 -5.71751706e-02 -3.13575774e-01 5.35808623e-01 2.88198441e-01 9.37870204e-01 -1.49111208e-02 -3.55822980e-01 6.64145768e-01 -4.21673208e-01 3.87208581e-01 -2.57742792e-01 -1.03634310e+00 -3.25349927e-01 -2.57457048e-02 -9.23149526e-01 -2.22795635e-01 -3.47731709e-01 -9.40060318e-01 -7.71421552e-01 -2.72134453e-01 -1.15900494e-01 8.67171362e-02 7.14501500e-01 5.73147237e-01 4.13407832e-01 1.86333179e-01 -7.84342885e-01 -4.28293556e-01 -8.38623166e-01 -1.92543387e-01 6.46659434e-01 1.40353948e-01 -5.73265314e-01 9.27845165e-02 -4.59020466e-01]
[14.812895774841309, 0.8178457021713257]
a035b659-2aec-479a-8b2c-fec6f94f74bd
hiding-in-plain-sight-differential-privacy
2307.00268
null
https://arxiv.org/abs/2307.00268v1
https://arxiv.org/pdf/2307.00268v1.pdf
Hiding in Plain Sight: Differential Privacy Noise Exploitation for Evasion-resilient Localized Poisoning Attacks in Multiagent Reinforcement Learning
Lately, differential privacy (DP) has been introduced in cooperative multiagent reinforcement learning (CMARL) to safeguard the agents' privacy against adversarial inference during knowledge sharing. Nevertheless, we argue that the noise introduced by DP mechanisms may inadvertently give rise to a novel poisoning threat, specifically in the context of private knowledge sharing during CMARL, which remains unexplored in the literature. To address this shortcoming, we present an adaptive, privacy-exploiting, and evasion-resilient localized poisoning attack (PeLPA) that capitalizes on the inherent DP-noise to circumvent anomaly detection systems and hinder the optimal convergence of the CMARL model. We rigorously evaluate our proposed PeLPA attack in diverse environments, encompassing both non-adversarial and multiple-adversarial contexts. Our findings reveal that, in a medium-scale environment, the PeLPA attack with attacker ratios of 20% and 40% can lead to an increase in average steps to goal by 50.69% and 64.41%, respectively. Furthermore, under similar conditions, PeLPA can result in a 1.4x and 1.6x computational time increase in optimal reward attainment and a 1.18x and 1.38x slower convergence for attacker ratios of 20% and 40%, respectively.
['Hung La', 'Md Tamjid Hossain']
2023-07-01
null
null
null
null
['anomaly-detection']
['methodology']
[-5.39688766e-02 -5.05936856e-04 1.15786292e-01 2.89893985e-01 -7.67288625e-01 -1.12830448e+00 4.10440922e-01 6.16121829e-01 -8.34380507e-01 1.02597880e+00 -3.84693146e-01 -3.67574841e-01 -9.81669649e-02 -8.76797616e-01 -7.63823092e-01 -9.28852916e-01 -3.29332858e-01 -2.49510616e-01 4.87740338e-02 -2.03857556e-01 1.00486040e-01 4.82176930e-01 -8.28032434e-01 -3.55657339e-01 9.92711842e-01 8.54144990e-01 -5.10638833e-01 6.08810425e-01 6.03079259e-01 9.17372525e-01 -1.16049051e+00 -6.97757721e-01 7.58344471e-01 -1.62440985e-01 -5.09814739e-01 -3.62643987e-01 -8.58917162e-02 -8.08555841e-01 -5.44885635e-01 1.33760607e+00 5.03021121e-01 -1.34804677e-02 3.24706554e-01 -1.72972703e+00 -4.31206673e-01 8.97174478e-01 -1.00775099e+00 2.02334166e-01 3.01793307e-01 5.37197351e-01 6.41228080e-01 7.13603124e-02 2.10062981e-01 1.06742036e+00 5.78272641e-01 8.03346395e-01 -1.25378036e+00 -1.15710604e+00 1.32280231e-01 1.12198507e-02 -1.36010993e+00 -1.91184059e-01 6.16971791e-01 4.92744707e-02 4.95302081e-01 3.10527235e-01 4.71725821e-01 1.38518012e+00 4.57217664e-01 5.52334011e-01 1.41344428e+00 -7.11319819e-02 5.79331160e-01 1.05629385e-01 -2.02940390e-01 4.45191354e-01 8.32425117e-01 3.28353703e-01 -4.48722214e-01 -7.40961254e-01 8.33965719e-01 -6.16837256e-02 -1.29729569e-01 -9.51227099e-02 -7.33852684e-01 8.56499434e-01 2.36961022e-01 -3.00687432e-01 -5.74624062e-01 5.00283778e-01 4.50740516e-01 5.76483727e-01 3.12922820e-02 5.91523767e-01 -1.73477575e-01 -1.09994315e-01 -7.28345215e-02 5.42053521e-01 9.18463290e-01 7.44607389e-01 4.49952394e-01 3.21004391e-01 -5.30882254e-02 -4.13333140e-02 1.94164976e-01 7.01642036e-01 2.43035946e-02 -1.20847416e+00 5.61534464e-01 3.04584950e-01 4.37452525e-01 -1.06086111e+00 -1.51603535e-01 -5.89993000e-01 -6.14189088e-01 4.18725103e-01 6.89478219e-01 -8.21801662e-01 -2.21008390e-01 2.16325331e+00 6.23068392e-01 2.36861899e-01 6.67341530e-01 7.88347423e-01 6.22281395e-02 3.08528066e-01 3.37140977e-01 -4.39920932e-01 1.23832464e+00 -3.77621859e-01 -5.87432921e-01 5.61478361e-02 4.39645678e-01 -2.12594166e-01 8.77227008e-01 3.30051988e-01 -9.28602159e-01 3.66207957e-01 -1.10788620e+00 7.62316704e-01 -8.54473934e-02 -7.00760484e-01 6.04134738e-01 1.05468464e+00 -5.27415395e-01 2.06735671e-01 -8.02670240e-01 -1.35350466e-01 7.12321937e-01 3.82183313e-01 -1.53778329e-01 1.95803463e-01 -1.18649685e+00 5.75466156e-01 1.05439015e-01 -3.10419083e-01 -1.43576908e+00 -1.00060999e+00 -5.24107575e-01 1.37812085e-03 1.04654205e+00 -5.11552751e-01 9.86711800e-01 -6.43182456e-01 -1.59376216e+00 2.77649432e-01 5.66081643e-01 -1.23970890e+00 9.67933834e-01 -3.10460806e-01 -1.73488915e-01 4.42571461e-01 -7.89013281e-02 9.55621228e-02 6.76201999e-01 -1.38963962e+00 -3.98017526e-01 -4.92605090e-01 6.69013143e-01 3.62733841e-01 -4.72444355e-01 2.16483083e-02 5.06480813e-01 -7.97250628e-01 -7.36639798e-01 -9.31222141e-01 -5.78682542e-01 -7.45016709e-02 -3.26325566e-01 1.74898431e-01 8.21607709e-01 -2.63948172e-01 9.00716126e-01 -2.16862106e+00 -3.31162125e-01 3.58950138e-01 3.95921081e-01 4.21846896e-01 -2.48809326e-02 5.62737107e-01 6.26218915e-01 7.39003494e-02 -2.82808691e-01 -2.41429154e-02 1.41761690e-01 7.81072676e-02 -4.75788891e-01 8.24694097e-01 -7.13229179e-02 7.17197895e-01 -1.05997956e+00 -1.81896705e-02 -6.36478439e-02 1.55286998e-01 -6.57829404e-01 2.48861492e-01 -1.54369593e-01 5.49808025e-01 -7.74956465e-01 7.54944861e-01 8.14361215e-01 7.79291764e-02 4.30587709e-01 3.01202893e-01 2.42178980e-02 -3.96545172e-01 -1.04442573e+00 1.15831697e+00 -7.31813088e-02 1.72358200e-01 3.90450686e-01 -5.70454895e-01 6.31160617e-01 3.01824152e-01 4.51140434e-01 -7.20860362e-01 4.92198318e-01 1.12324737e-01 -6.27347901e-02 -6.27858937e-02 3.75314713e-01 -1.50343180e-01 -4.42560315e-01 7.90023744e-01 -4.70791787e-01 5.01260161e-01 -2.99386919e-01 2.79767096e-01 1.53528404e+00 -3.87413651e-01 3.81464511e-01 -2.85960883e-01 4.77997780e-01 -9.76358205e-02 8.18940282e-01 1.27917361e+00 -8.96008432e-01 -3.33381593e-01 6.40874624e-01 -2.89506942e-01 -6.89858377e-01 -1.07470691e+00 3.25295240e-01 8.87036622e-01 5.72678924e-01 -1.86221093e-01 -8.82802784e-01 -9.49171543e-01 4.39368039e-01 7.03018546e-01 -3.77679110e-01 -5.24914920e-01 -3.83096188e-01 -9.06995952e-01 1.35388315e+00 2.26069361e-01 1.05949295e+00 -7.09660053e-01 -9.56626296e-01 9.09632742e-02 1.95129514e-01 -1.12240136e+00 -4.62664992e-01 -5.72039448e-02 -3.13030243e-01 -1.26480615e+00 -2.55632043e-01 6.52188610e-04 6.89472198e-01 2.81764805e-01 4.06069815e-01 -5.16732745e-02 -6.36337996e-02 8.28216314e-01 -3.50686282e-01 -4.77469116e-01 -5.90872705e-01 -2.05413491e-01 3.93032879e-01 8.47774222e-02 3.06118011e-01 -7.27833748e-01 -6.97175443e-01 1.68544278e-01 -1.05747449e+00 -7.48418152e-01 2.89672911e-01 6.01752877e-01 1.70992047e-01 1.34317189e-01 1.12790501e+00 -7.44208515e-01 1.00962853e+00 -7.82728076e-01 -1.02327228e+00 2.66204774e-01 -6.34294868e-01 -5.61168268e-02 1.02384138e+00 -7.71801353e-01 -8.27823043e-01 -9.68729109e-02 1.77300081e-01 -7.14385271e-01 -1.63432956e-01 7.23284632e-02 -1.08388029e-01 -6.01392627e-01 6.95357263e-01 2.56130368e-01 2.38365233e-01 1.55115435e-02 2.59823263e-01 4.26210582e-01 3.44238698e-01 -7.87600338e-01 1.04410970e+00 6.02459669e-01 2.29926586e-01 -5.35487413e-01 -3.72500509e-01 2.41057038e-01 2.86450148e-01 -1.38989881e-01 5.77207863e-01 -1.04274559e+00 -1.48182058e+00 8.35630178e-01 -7.46544242e-01 -4.13581789e-01 -3.33812773e-01 3.50894302e-01 -3.96630496e-01 7.58487821e-01 -6.69616342e-01 -1.22819591e+00 -6.20534122e-01 -1.02795613e+00 4.87245768e-02 4.90693420e-01 1.55432209e-01 -7.53875077e-01 -1.16657607e-01 3.72386038e-01 5.54991603e-01 6.10290527e-01 5.71686924e-01 -9.80720043e-01 -5.93026102e-01 -1.49306148e-01 2.13919923e-01 1.17898375e-01 -5.91528527e-02 -4.44782436e-01 -6.83696985e-01 -8.40280831e-01 2.25801468e-01 -5.67992806e-01 1.56578779e-01 -1.48947433e-01 9.58131969e-01 -9.86839533e-01 6.53940439e-02 4.22315747e-01 1.40023863e+00 5.08119226e-01 2.80253530e-01 5.79184771e-01 3.68208736e-01 2.62184560e-01 7.09947586e-01 1.08249605e+00 4.94298697e-01 2.84320652e-01 9.42361176e-01 3.31495106e-01 5.39090693e-01 -4.24792558e-01 6.24085844e-01 -3.05273682e-02 2.03044489e-01 -2.48171687e-01 -4.48018312e-01 3.20490330e-01 -1.59552193e+00 -8.90721023e-01 1.81895465e-01 2.50571156e+00 9.81752098e-01 3.72673124e-02 4.79510218e-01 -5.34892082e-02 7.51874328e-01 9.34784487e-02 -1.05852211e+00 -3.87708247e-01 -5.19144386e-02 -8.03280771e-02 1.10355616e+00 3.86564672e-01 -9.00339305e-01 8.71814907e-01 5.60820246e+00 7.50376165e-01 -8.74102652e-01 2.25517854e-01 6.04151964e-01 -1.65495574e-01 -1.34469301e-01 2.75874119e-02 -5.04407048e-01 4.90958929e-01 9.54333663e-01 -5.75526297e-01 7.58818746e-01 7.96481729e-01 5.24943471e-02 1.49616664e-02 -6.81751192e-01 7.19928384e-01 -2.24200219e-01 -9.22401071e-01 -8.70296210e-02 3.81788760e-01 6.15164697e-01 -3.04568201e-01 2.88184822e-01 2.78929383e-01 8.87459040e-01 -8.28115284e-01 5.67774773e-01 2.69959327e-02 4.62496310e-01 -1.51540506e+00 6.37181997e-01 3.69078785e-01 -8.99447381e-01 -4.85399306e-01 -1.58314064e-01 -1.89705759e-01 -1.55940637e-01 3.68753336e-02 -5.20674646e-01 6.38925850e-01 5.73575974e-01 -4.53878334e-03 -1.09494366e-01 5.94927788e-01 -3.27993631e-01 5.81592917e-01 -4.76270199e-01 -4.08388488e-02 4.19636399e-01 -6.30120561e-02 9.79318321e-01 5.81921577e-01 -1.07864872e-01 2.27524504e-01 1.91113263e-01 8.79992306e-01 -3.33172917e-01 -2.27915987e-01 -5.92584372e-01 4.03406583e-02 1.10573304e+00 1.07572687e+00 -2.14044973e-01 -1.64397736e-03 4.86262105e-02 9.54794884e-01 2.75654614e-01 4.95554477e-01 -1.20222604e+00 -4.98585910e-01 9.79010403e-01 -2.84827173e-01 1.36673585e-01 -1.66501701e-01 -1.27638191e-01 -1.00079489e+00 -1.97154023e-02 -1.12791789e+00 5.32120883e-01 1.13427497e-01 -1.57416081e+00 2.50495523e-01 -1.97327375e-01 -8.76147687e-01 2.07977220e-02 3.88740115e-02 -7.57726967e-01 4.86763924e-01 -1.49608421e+00 -7.46351182e-01 -6.08937582e-03 9.31433678e-01 -4.55382876e-02 -3.06534499e-01 7.05630720e-01 1.55141219e-01 -8.97439003e-01 1.30639398e+00 2.19395086e-01 1.65582985e-01 3.55238259e-01 -1.04707551e+00 1.42008532e-02 1.10166395e+00 -3.71069431e-01 6.53823137e-01 8.19575071e-01 -5.91415763e-01 -2.03222990e+00 -1.30858314e+00 -1.24146782e-01 -3.12170595e-01 8.40062559e-01 -2.57403880e-01 -7.03953326e-01 5.16027629e-01 1.15512691e-01 1.43478751e-01 6.29095018e-01 -5.21290183e-01 -6.47623956e-01 -2.94306874e-01 -1.99575579e+00 1.05758297e+00 7.10399210e-01 -4.15014684e-01 -1.06819838e-01 -3.68625834e-03 8.82008076e-01 -3.17858040e-01 -1.09297848e+00 1.89434327e-02 3.70096684e-01 -6.60046935e-01 8.55589867e-01 -5.67770958e-01 -1.01443984e-01 -2.64110178e-01 -1.99765697e-01 -1.10561490e+00 1.78619385e-01 -1.37772489e+00 -3.58768106e-01 1.22015131e+00 5.35637699e-02 -1.20044017e+00 6.51018858e-01 5.97159266e-01 4.97040153e-01 -5.21938264e-01 -1.29317629e+00 -1.00132668e+00 4.26371306e-01 -1.92124128e-01 7.01720834e-01 9.80459630e-01 1.09309688e-01 -2.41325974e-01 -5.63965857e-01 7.39911795e-01 1.23520958e+00 -2.62514651e-01 8.73725176e-01 -5.49704552e-01 -4.13474083e-01 -1.93960428e-01 -2.14697719e-01 -5.36052823e-01 2.20845029e-01 -4.81421977e-01 -2.59211510e-01 -6.20784640e-01 1.15223050e-01 -4.81849670e-01 -4.78362888e-01 6.89484119e-01 -2.44719416e-01 5.92027232e-02 3.64422321e-01 2.54574176e-02 -6.84158325e-01 7.42587149e-01 7.88067877e-01 -2.77196453e-03 -1.97638944e-01 -3.76552083e-02 -9.02737141e-01 4.16769713e-01 1.13453996e+00 -7.33532071e-01 -5.45423090e-01 -9.16959066e-03 2.75747627e-02 2.69050360e-01 5.31005621e-01 -8.16782475e-01 2.18174875e-01 -5.00132263e-01 -9.75404084e-02 1.10726552e-02 6.63992167e-02 -9.13615346e-01 1.47918060e-01 1.00446045e+00 -3.69598687e-01 5.60134873e-02 3.01698893e-01 1.07156575e+00 3.67739469e-01 -1.23386038e-02 7.82172799e-01 -2.00289413e-01 -1.22405633e-01 2.03393534e-01 -6.83210135e-01 1.63119555e-01 1.69478834e+00 -2.80890837e-02 -7.37376392e-01 -4.42796588e-01 -1.39709041e-01 6.15868032e-01 5.61149716e-01 7.61769935e-02 5.30225635e-01 -9.74684298e-01 -6.30444348e-01 -9.60935280e-03 -2.02772081e-01 -2.94584155e-01 3.46028507e-01 7.79238522e-01 -1.79685161e-01 -3.97887528e-02 -3.60526025e-01 -5.06051909e-03 -1.24738359e+00 7.04979956e-01 5.58554769e-01 -2.97374785e-01 -5.85220218e-01 4.55492705e-01 6.53076917e-03 -3.67530808e-02 4.63481396e-01 1.71662092e-01 1.59039408e-01 -4.23679411e-01 5.58604598e-01 7.78892398e-01 -4.16668057e-01 -2.48479411e-01 -4.42166537e-01 -2.09546220e-02 -3.49635065e-01 -1.00432411e-01 9.58101809e-01 -1.50643840e-01 1.87875524e-01 -1.78085506e-01 6.95450425e-01 3.16522121e-01 -1.66493964e+00 -1.73475757e-01 -2.19053909e-01 -6.51563823e-01 -1.20427668e-01 -9.00715470e-01 -1.08495772e+00 2.53644705e-01 5.26071846e-01 5.10422170e-01 1.05133653e+00 -4.76422340e-01 7.69236505e-01 4.71568823e-01 8.79163325e-01 -8.49505723e-01 9.77772698e-02 2.56503314e-01 4.47936088e-01 -9.29641128e-01 -7.38777146e-02 -1.65701807e-01 -1.00437236e+00 5.62282205e-01 7.96875417e-01 -2.54137367e-01 6.91025481e-02 3.93853873e-01 -1.03498600e-01 -9.97475982e-02 -7.01371789e-01 2.19194412e-01 -7.69707739e-01 7.82961845e-01 -6.31707728e-01 1.91870227e-01 -2.66234130e-01 9.47800815e-01 2.27715939e-01 -3.69670093e-01 1.00600982e+00 1.43631494e+00 -2.52036065e-01 -9.29816544e-01 -4.98510301e-01 7.71278739e-02 -9.20193434e-01 2.38480315e-01 -4.47517127e-01 6.29924178e-01 -2.27784514e-01 1.23353827e+00 -2.36770481e-01 -3.39571625e-01 2.59936184e-01 -3.92431289e-01 2.52934724e-01 1.02763660e-01 -1.09728456e+00 -1.58258319e-01 -1.32487565e-01 -5.99763632e-01 -2.05036834e-01 -3.26094151e-01 -1.43972695e+00 -8.17079961e-01 -1.77051276e-01 3.36405843e-01 2.82190382e-01 6.96954548e-01 5.75230598e-01 2.18525976e-01 1.03700113e+00 -9.56882387e-02 -1.33926558e+00 -4.67566013e-01 -7.81039894e-01 1.43837020e-01 3.59617352e-01 -6.12372160e-01 -6.82239771e-01 -6.92942560e-01]
[5.676334381103516, 7.338398456573486]
ec236a05-b63a-4057-9f51-d8e12b130e63
cost-sensitive-bert-for-generalisable-1
2003.11563
null
https://arxiv.org/abs/2003.11563v1
https://arxiv.org/pdf/2003.11563v1.pdf
Cost-Sensitive BERT for Generalisable Sentence Classification with Imbalanced Data
The automatic identification of propaganda has gained significance in recent years due to technological and social changes in the way news is generated and consumed. That this task can be addressed effectively using BERT, a powerful new architecture which can be fine-tuned for text classification tasks, is not surprising. However, propaganda detection, like other tasks that deal with news documents and other forms of decontextualized social communication (e.g. sentiment analysis), inherently deals with data whose categories are simultaneously imbalanced and dissimilar. We show that BERT, while capable of handling imbalanced classes with no additional data augmentation, does not generalise well when the training and test data are sufficiently dissimilar (as is often the case with news sources, whose topics evolve over time). We show how to address this problem by providing a statistical measure of similarity between datasets and a method of incorporating cost-weighting into BERT when the training and test sets are dissimilar. We test these methods on the Propaganda Techniques Corpus (PTC) and achieve the second-highest score on sentence-level propaganda classification.
['Harish Tayyar Madabushi', 'Michael Castelle', 'Elena Kochkina']
2020-03-16
null
null
null
null
['propaganda-detection']
['natural-language-processing']
[ 2.13367254e-01 -1.11408994e-01 -2.54064560e-01 -5.11197627e-01 -5.61600268e-01 -6.16634369e-01 1.19523787e+00 7.82024920e-01 -5.88473797e-01 7.04122245e-01 5.78850865e-01 -4.64638531e-01 -9.34355184e-02 -7.49777079e-01 -1.79046646e-01 -5.85077107e-01 -6.13796078e-02 6.14439905e-01 1.72266588e-02 -7.34489560e-01 6.32562399e-01 1.91301420e-01 -1.42306840e+00 5.98528385e-01 8.43521416e-01 6.55193031e-01 -4.44883496e-01 6.89263880e-01 -3.14047277e-01 1.12664855e+00 -1.34299779e+00 -6.15238070e-01 -9.76730436e-02 -6.07437670e-01 -1.05497944e+00 -1.13252766e-01 4.76767540e-01 5.40268868e-02 1.47248492e-01 9.30457771e-01 5.33004761e-01 -2.45815560e-01 9.44635153e-01 -8.82703066e-01 -2.99016684e-01 7.56429791e-01 -6.71680868e-01 6.72915995e-01 4.05252248e-01 -5.02703011e-01 8.85504007e-01 -4.63775277e-01 7.00838149e-01 1.34935474e+00 7.96471179e-01 3.98902118e-01 -1.27352822e+00 -4.50765073e-01 3.83949094e-02 2.77051210e-01 -6.23256743e-01 -3.29819232e-01 9.68683779e-01 -7.00521111e-01 7.93155670e-01 5.10246933e-01 7.06814528e-01 1.30940473e+00 3.33994269e-01 7.96518862e-01 1.07204795e+00 -6.03053212e-01 2.46378526e-01 3.50305825e-01 3.48390788e-01 2.68410653e-01 2.21019432e-01 -4.89465535e-01 -5.25522709e-01 -4.87950921e-01 -1.62823364e-01 -4.77450013e-01 -6.80692568e-02 1.54407159e-03 -1.17742169e+00 1.34339142e+00 1.03845298e-01 8.08748245e-01 -1.97897345e-01 -4.29519713e-02 9.46006358e-01 8.08179617e-01 1.13666928e+00 9.48339999e-01 -4.09520477e-01 -3.39248508e-01 -1.01900423e+00 4.63821381e-01 1.02738750e+00 2.31364340e-01 1.28710777e-01 9.70674604e-02 -7.22016618e-02 9.00036573e-01 -1.55940264e-01 3.70898783e-01 8.42563272e-01 -2.61284262e-01 6.28083169e-01 5.82060277e-01 -9.61998105e-02 -1.39323485e+00 -7.58601725e-01 -6.46978080e-01 -9.18119788e-01 1.63478106e-01 5.12531161e-01 -2.08544314e-01 -6.86178327e-01 1.72559690e+00 4.31375653e-01 -6.15292966e-01 8.61696806e-03 5.32467008e-01 7.88105786e-01 6.43912613e-01 -7.30053382e-03 -6.44563913e-01 1.31998384e+00 -5.00584722e-01 -9.09999430e-01 -3.32176775e-01 9.72829759e-01 -9.67871189e-01 8.65335584e-01 4.72795486e-01 -8.18482339e-01 -2.78193690e-02 -1.08829272e+00 2.01196536e-01 -6.03493571e-01 -4.88250017e-01 6.38772905e-01 8.72055113e-01 -4.75383699e-01 5.23843110e-01 -4.56946582e-01 -3.09096724e-01 5.71283996e-01 7.00090826e-03 -2.76984364e-01 3.90159249e-01 -1.38388169e+00 1.29060650e+00 2.85335928e-01 -2.57396340e-01 -4.09067124e-01 -4.82501388e-01 -5.39687812e-01 -1.50390729e-01 2.10088044e-01 -1.93392754e-01 1.12315226e+00 -1.37197363e+00 -1.09101975e+00 1.03590310e+00 3.31697226e-01 -4.68358755e-01 5.89844406e-01 -1.57183073e-02 -6.11504316e-01 -1.51653707e-01 1.37460023e-01 -7.58717433e-02 1.13627708e+00 -8.00079703e-01 -4.46206778e-01 -4.40716952e-01 -1.43708631e-01 1.60124809e-01 -7.14344323e-01 5.93741179e-01 6.10995829e-01 -8.51343811e-01 -4.35190238e-02 -5.04938245e-01 -3.96709703e-02 -5.84520400e-01 -4.48618442e-01 -2.62952030e-01 8.71113181e-01 -6.26079023e-01 1.14761198e+00 -1.88015914e+00 2.77818620e-01 2.66163975e-01 3.76221061e-01 4.54905123e-01 2.28629913e-02 6.26369059e-01 -1.00101173e-01 3.09481233e-01 4.04065549e-02 -3.00357435e-02 -2.50160228e-03 8.87563080e-03 -3.84035319e-01 7.24424660e-01 1.57021686e-01 3.36059511e-01 -8.99729609e-01 -3.51634711e-01 -2.28285894e-01 8.33606645e-02 -5.35591543e-01 -2.28916615e-01 -2.33960077e-01 1.71649843e-01 -4.46638614e-01 1.75779730e-01 3.83374095e-01 -1.41588628e-01 2.18149111e-01 1.88228711e-01 -1.92705467e-01 7.25066245e-01 -8.82917106e-01 1.15766585e+00 -3.21481347e-01 1.11899579e+00 1.61720719e-02 -1.29730606e+00 8.61865461e-01 2.69004464e-01 3.59524697e-01 -8.47454071e-01 5.60933709e-01 1.76342040e-01 2.36559480e-01 -5.93279302e-01 5.49720466e-01 -4.45273757e-01 -4.69945341e-01 8.93143117e-01 -1.88283220e-01 -3.07041556e-01 5.19370854e-01 4.30014253e-01 1.35652137e+00 -8.43134582e-01 4.74532694e-01 -4.73599225e-01 3.52027357e-01 4.56672728e-01 3.46737802e-01 7.98376858e-01 -1.26018301e-01 2.68052489e-01 8.60087872e-01 -7.93457150e-01 -1.08889592e+00 -7.50626862e-01 -3.70479167e-01 1.26376247e+00 -1.33470699e-01 -5.91914177e-01 -5.29997706e-01 -9.56494212e-01 -1.08051099e-01 6.10221565e-01 -7.36364722e-01 -5.37535176e-02 -6.11131608e-01 -1.37032318e+00 4.69532937e-01 -1.67110339e-02 1.00014776e-01 -9.70398784e-01 -8.43017936e-01 4.71718252e-01 -2.42622912e-01 -6.65185571e-01 1.75120845e-01 4.93544966e-01 -5.54344237e-01 -1.06599915e+00 -3.08476269e-01 -6.83495641e-01 4.02195036e-01 1.91427916e-02 1.32463253e+00 1.16623156e-01 -4.10886824e-01 1.00810342e-01 -7.68257856e-01 -7.84899652e-01 -1.04356444e+00 2.97660887e-01 3.55713442e-02 -1.70928568e-01 2.67192394e-01 -5.54272592e-01 -1.24322504e-01 1.83382463e-02 -1.00370729e+00 2.31643066e-01 2.31120482e-01 1.06141794e+00 -3.03023368e-01 2.88878024e-01 1.08651674e+00 -1.17036343e+00 1.11839211e+00 -5.43695867e-01 -1.29285842e-01 -9.55819711e-02 -4.23207134e-01 -3.20483409e-02 6.74826801e-01 -4.24549639e-01 -8.49766672e-01 -6.38588309e-01 -3.88423860e-01 5.94027340e-01 -1.34649247e-01 8.44303131e-01 3.99582982e-01 2.63416916e-01 1.27224588e+00 -2.35381007e-01 1.85441554e-01 -4.49033588e-01 2.76580960e-01 1.10409582e+00 1.86554730e-01 -7.68208504e-03 8.48622620e-01 5.46262622e-01 -3.11371267e-01 -7.86802173e-01 -1.25760877e+00 -5.66610217e-01 -3.95309508e-01 -8.84428546e-02 4.85764444e-01 -6.17709696e-01 -2.06681892e-01 5.45720041e-01 -1.23929667e+00 -5.28164059e-02 -3.15537423e-01 3.35748613e-01 -3.58635396e-01 1.90186962e-01 -6.13891959e-01 -8.18781793e-01 -3.71379733e-01 -4.81064469e-01 5.83623111e-01 -2.18621388e-01 -6.52200818e-01 -1.00439811e+00 3.60126555e-01 4.57592815e-01 6.37690246e-01 4.78639424e-01 1.36008120e+00 -1.20867157e+00 5.06688833e-01 -5.10734975e-01 7.48445541e-02 4.37867850e-01 2.01850787e-01 -1.26170546e-01 -7.97265589e-01 -2.17301190e-01 5.55338204e-01 -6.06862187e-01 8.93478274e-01 1.34490073e-01 6.78776979e-01 -8.23904037e-01 -3.32593592e-03 -1.66589200e-01 9.19967234e-01 3.89420725e-02 5.55296361e-01 6.89312816e-01 3.72688502e-01 9.88456309e-01 4.80636477e-01 4.73124623e-01 1.55837879e-01 7.39719808e-01 4.65909183e-01 -2.48319760e-01 6.31972849e-02 3.14324558e-01 3.11324805e-01 9.59955871e-01 1.01661362e-01 -3.26227605e-01 -9.74354982e-01 4.64485019e-01 -1.69726586e+00 -1.60951686e+00 -4.29630280e-01 1.87728739e+00 1.16685915e+00 3.00216407e-01 2.36902058e-01 8.33534658e-01 5.54752767e-01 4.91896749e-01 -5.10302074e-02 -1.02011168e+00 -3.43835205e-01 9.53647792e-02 -1.05144875e-02 4.09504473e-01 -1.42791212e+00 4.72608298e-01 6.32870722e+00 9.76734698e-01 -1.13752115e+00 2.26724401e-01 6.77594244e-01 -2.95776874e-01 -3.16162348e-01 -4.18754041e-01 -3.63477588e-01 6.02708757e-01 8.54186714e-01 -1.70238107e-01 3.33330669e-02 6.17873132e-01 2.13359937e-01 -5.62282801e-02 -9.15156424e-01 6.52803481e-01 4.57982510e-01 -1.46707666e+00 -1.96094424e-01 -1.11930497e-01 7.93638587e-01 1.47474602e-01 1.30773494e-02 2.77910113e-01 1.15615301e-01 -9.46346760e-01 8.69487345e-01 -2.38547146e-01 2.23838687e-01 -6.89552724e-01 9.73681867e-01 6.87581122e-01 -2.42949173e-01 -2.40304887e-01 -1.63397968e-01 -5.08816659e-01 1.60339698e-01 1.09827721e+00 -8.86930883e-01 3.06247741e-01 5.35110056e-01 5.86241245e-01 -6.90019846e-01 7.65507162e-01 -5.19946590e-02 7.72079945e-01 -2.24969253e-01 -4.41961288e-01 3.11321616e-01 2.71470487e-01 7.88207591e-01 1.37838268e+00 -1.78157333e-02 -4.26661044e-01 -9.88964960e-02 2.28453711e-01 1.44133359e-01 5.88206410e-01 -5.30976415e-01 -1.07224055e-01 3.03927064e-03 1.18669140e+00 -6.83148146e-01 -4.31911975e-01 2.16076206e-02 4.11509693e-01 3.43690604e-01 -1.64013505e-01 -5.89517474e-01 -3.14029932e-01 3.07660282e-01 2.70647883e-01 -5.07169664e-02 -4.12117224e-03 -1.75506771e-01 -1.20115888e+00 -1.09963061e-03 -1.16821241e+00 4.70262289e-01 -2.48428598e-01 -1.54009938e+00 6.55379117e-01 -1.02433220e-01 -1.04791582e+00 -5.21289051e-01 -4.15381312e-01 -6.54005289e-01 3.07601303e-01 -1.24199224e+00 -7.77226925e-01 -8.45996216e-02 2.76152164e-01 5.28634250e-01 -1.52540237e-01 8.46318781e-01 3.46768141e-01 -1.63990662e-01 2.19146639e-01 3.57157648e-01 1.81021199e-01 7.57534862e-01 -1.21689272e+00 1.47584662e-01 5.66710353e-01 1.70051754e-01 1.42960384e-01 1.29874623e+00 -4.04139549e-01 -9.13996220e-01 -8.56786728e-01 1.30758834e+00 -5.15411496e-01 1.05119526e+00 -7.61300802e-01 -7.83433855e-01 1.42937914e-01 2.56416321e-01 -5.46885073e-01 9.03428733e-01 5.09797513e-01 -7.38758624e-01 1.07268825e-01 -1.14446187e+00 4.74772036e-01 8.30288053e-01 -2.11534753e-01 -9.46778774e-01 9.51241970e-01 3.54748756e-01 -2.45875627e-01 -4.45402533e-01 2.24161685e-01 4.03612942e-01 -1.02738333e+00 5.99493563e-01 -1.09100056e+00 6.94152057e-01 7.18516409e-02 1.03726894e-01 -1.61987388e+00 -1.47571385e-01 -4.28709447e-01 1.53616771e-01 1.13232851e+00 4.94381785e-01 -5.99850178e-01 6.58253193e-01 -1.17037810e-01 -2.13520918e-02 -5.25875807e-01 -1.21640885e+00 -7.66974211e-01 2.53450692e-01 -3.70056719e-01 2.59784281e-01 1.45235121e+00 5.16194403e-01 8.07165146e-01 -5.80020785e-01 -5.34978449e-01 2.92216808e-01 2.73171604e-01 6.43625379e-01 -1.62558949e+00 -1.00044213e-01 -7.33998060e-01 -8.15739393e-01 -4.06159163e-01 1.69297487e-01 -8.83920670e-01 -4.97135632e-02 -1.43011045e+00 2.38099605e-01 -4.75082427e-01 1.24240391e-01 3.24861884e-01 7.07724914e-02 3.85464698e-01 9.89089254e-03 1.65006608e-01 -4.28121626e-01 4.21896905e-01 9.51826453e-01 -3.56045991e-01 -1.61222145e-01 6.00056015e-02 -8.48553240e-01 8.76377761e-01 1.04066646e+00 -7.92694390e-01 -1.44497633e-01 -2.75976688e-01 8.47222567e-01 -3.54658127e-01 1.66645303e-01 -8.58049750e-01 3.98554094e-02 -1.35979474e-01 1.69496924e-01 -5.15683651e-01 2.61122733e-01 -4.75300550e-01 -7.89813027e-02 6.83725893e-01 -6.41663730e-01 1.64900571e-01 -4.03610766e-02 5.14259934e-01 -2.39357978e-01 -4.52017456e-01 7.81944752e-01 7.91215003e-02 8.04123748e-03 -3.02431673e-01 -9.38184261e-01 2.79995710e-01 8.60386014e-01 8.30192715e-02 -1.12438095e+00 -4.80164021e-01 -5.18112063e-01 -8.74457881e-02 2.20283926e-01 5.06475031e-01 9.53935757e-02 -1.16816270e+00 -1.14724267e+00 5.83785065e-02 3.74191962e-02 -3.17988038e-01 1.09445602e-01 1.12643480e+00 -5.17051280e-01 7.80788809e-02 -1.28034234e-01 -3.60207170e-01 -1.56600320e+00 3.00752133e-01 3.41922678e-02 -5.43749094e-01 -4.68170732e-01 6.14888668e-01 -1.63727269e-01 -1.63439825e-01 5.64920716e-02 2.87922136e-02 -5.98038793e-01 7.82955050e-01 7.92782128e-01 2.22643867e-01 4.59728986e-01 -7.27707684e-01 -2.77633578e-01 5.35148047e-02 -4.47307229e-01 4.88632508e-02 1.69445872e+00 1.65509343e-01 -5.00635207e-01 5.95874488e-01 1.09387767e+00 2.69644856e-01 -1.47795975e-01 -1.05399974e-01 1.72125682e-01 -3.95573080e-01 9.95556638e-03 -9.21951413e-01 -6.81245565e-01 7.80506432e-01 2.30977520e-01 1.21701956e+00 7.34266222e-01 9.89846140e-02 3.62777919e-01 3.95190835e-01 5.56094572e-02 -1.46291018e+00 3.66151690e-01 8.22111964e-01 1.03513956e+00 -1.08160996e+00 3.14946622e-01 -2.11359411e-01 -4.07157004e-01 9.57020819e-01 6.14578091e-02 -5.26478980e-04 3.55799288e-01 2.97747433e-01 1.69474468e-01 -5.36154449e-01 -9.45169806e-01 1.63123354e-01 1.22505732e-01 4.34100688e-01 4.72892225e-01 2.58731171e-02 -8.90610278e-01 2.11278901e-01 -5.07374942e-01 -6.79507494e-01 8.48360121e-01 1.11476147e+00 -7.69271970e-01 -9.20035779e-01 -5.30477762e-01 1.00375473e+00 -8.24369013e-01 -4.17653546e-02 -8.21315765e-01 7.26280928e-01 -2.73839477e-02 1.15241575e+00 1.77213490e-01 -3.53751153e-01 1.55716076e-01 5.65084703e-02 3.02621573e-01 -7.01442957e-01 -1.18855429e+00 -6.95153549e-02 7.85880327e-01 -1.18245989e-01 -7.16727793e-01 -5.87305129e-01 -6.84059560e-01 -6.23249769e-01 -5.72088659e-01 6.10221207e-01 8.99685025e-01 1.09617150e+00 -4.31235740e-03 4.20921952e-01 7.55531847e-01 -7.66962469e-01 -9.08922136e-01 -1.13283277e+00 -5.03543496e-01 6.76816165e-01 4.82280314e-01 -5.16317487e-01 -8.32143843e-01 -1.13174528e-01]
[8.532018661499023, 10.441838264465332]
7be64161-ac44-4c11-84fe-eea43721c4fc
discovering-implicit-discourse-relations
null
null
https://aclanthology.org/E14-1068
https://aclanthology.org/E14-1068.pdf
Discovering Implicit Discourse Relations Through Brown Cluster Pair Representation and Coreference Patterns
null
['Attapol Rutherford', 'Nianwen Xue']
2014-04-01
null
null
null
eacl-2014-4
['implicit-discourse-relation-classification']
['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]
[-7.229660511016846, 3.6934990882873535]
bd656032-433e-486f-ba04-e6db30b413f2
massive-language-models-can-be-accurately
2301.00774
null
https://arxiv.org/abs/2301.00774v3
https://arxiv.org/pdf/2301.00774v3.pdf
SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot
We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy. This is achieved via a new pruning method called SparseGPT, specifically designed to work efficiently and accurately on massive GPT-family models. We can execute SparseGPT on the largest available open-source models, OPT-175B and BLOOM-176B, in under 4.5 hours, and can reach 60% unstructured sparsity with negligible increase in perplexity: remarkably, more than 100 billion weights from these models can be ignored at inference time. SparseGPT generalizes to semi-structured (2:4 and 4:8) patterns, and is compatible with weight quantization approaches. The code is available at: https://github.com/IST-DASLab/sparsegpt.
['Dan Alistarh', 'Elias Frantar']
2023-01-02
null
null
null
null
['common-sense-reasoning']
['reasoning']
[ 1.12837411e-01 6.28502548e-01 -4.27685738e-01 -4.20475096e-01 -9.16230679e-01 -2.90602148e-01 4.35585618e-01 -1.00806490e-01 -1.11901395e-01 7.16627657e-01 2.42135197e-01 -4.39742327e-01 -4.38408628e-02 -7.03133523e-01 -5.71711540e-01 -5.35975933e-01 -5.93434758e-02 1.30282295e+00 1.33674189e-01 -1.17164575e-01 -2.10353836e-01 1.39487684e-01 -1.41272271e+00 3.08327973e-01 8.11237216e-01 8.38220239e-01 4.28630441e-01 6.26068532e-01 -1.04477920e-01 6.15993381e-01 -4.29627031e-01 -7.24038720e-01 2.57507056e-01 -5.75362444e-02 -7.87743747e-01 8.47048163e-02 5.65375924e-01 -3.39881450e-01 -4.83848780e-01 9.35307503e-01 3.73456210e-01 -1.91345096e-01 3.62752736e-01 -1.05066967e+00 -6.00713193e-01 1.40338409e+00 -4.78977263e-01 2.00157166e-01 -1.13336392e-01 1.26393940e-02 1.55067182e+00 -1.00253272e+00 4.84207660e-01 1.54849064e+00 6.62949085e-01 6.13508582e-01 -1.44254136e+00 -9.22573328e-01 3.52576822e-01 2.96210088e-02 -1.43709898e+00 -6.28404915e-01 2.67627299e-01 -1.43973157e-02 1.40293825e+00 3.01490903e-01 9.37433183e-01 1.31007349e+00 -9.52940807e-02 9.89536285e-01 8.06768239e-01 -3.13322306e-01 1.04396902e-01 -2.17208147e-01 2.22926468e-01 9.85926092e-01 5.01436353e-01 -2.74423540e-01 -6.07827902e-01 -5.26300132e-01 5.89646935e-01 -5.31988703e-02 3.17502059e-02 2.28747185e-02 -1.06105661e+00 9.76734221e-01 6.48030546e-03 3.45438272e-01 -8.01091194e-02 5.33709824e-01 1.12341963e-01 2.43368283e-01 7.25350320e-01 1.89895913e-01 -7.44817078e-01 -5.87560594e-01 -1.34248376e+00 2.53891408e-01 8.78241658e-01 1.31122684e+00 9.62881505e-01 5.86665630e-01 -1.94821298e-01 1.06879592e+00 2.35465333e-01 7.74334133e-01 6.42695129e-01 -1.05416822e+00 4.78108853e-01 1.93062857e-01 -2.37003684e-01 -4.64564830e-01 -3.39668021e-02 -6.76645398e-01 -9.07165051e-01 -5.65733314e-01 1.30185008e-01 -2.39505664e-01 -1.32324445e+00 1.78786159e+00 2.00627178e-01 1.83371425e-01 -3.39630812e-01 2.76079178e-01 5.99432707e-01 7.60462999e-01 4.47374582e-02 -1.46029696e-01 1.31462705e+00 -9.47239220e-01 -3.64876240e-01 -7.69999146e-01 6.50047839e-01 -8.59468937e-01 1.08136332e+00 6.28006637e-01 -1.34443212e+00 -1.90064341e-01 -7.59271324e-01 -1.47843868e-01 1.14524998e-01 1.79518387e-01 1.02435315e+00 6.22304440e-01 -1.15623164e+00 6.58499539e-01 -1.14932108e+00 -3.49724501e-01 7.07029521e-01 4.32188511e-01 -9.72712487e-02 -2.05928266e-01 -1.00177288e+00 6.80421829e-01 5.20832300e-01 -2.94311911e-01 -1.13514972e+00 -1.07599497e+00 -6.76674187e-01 3.47027719e-01 2.95771986e-01 -7.99897671e-01 1.46932268e+00 -6.73121929e-01 -1.36378467e+00 8.24536979e-01 -5.83884120e-01 -8.94676030e-01 1.58217907e-01 -3.16695303e-01 -2.58056790e-01 -1.44464239e-01 -6.75401017e-02 8.52669537e-01 1.10246658e+00 -7.86045790e-01 -6.53730452e-01 -1.83658227e-01 -2.10380152e-01 -5.35159521e-02 -6.10881925e-01 -5.93505017e-02 -6.78270578e-01 -8.78807485e-01 2.27976680e-01 -8.68291378e-01 -2.49187723e-01 -4.63586092e-01 -6.07544839e-01 -3.51755559e-01 4.66500521e-01 -5.66534936e-01 1.40891683e+00 -1.81549323e+00 2.53940105e-01 6.29863366e-02 5.73206961e-01 3.06246698e-01 -2.96273232e-01 4.65788871e-01 4.69507799e-02 7.84397945e-02 -2.76363641e-01 -8.73873830e-01 3.75967234e-01 7.15526462e-01 -4.93470669e-01 9.26772803e-02 4.43470702e-02 9.18863714e-01 -8.15105081e-01 -4.86632288e-01 -4.42976780e-05 2.63934165e-01 -9.36271071e-01 -6.69293478e-02 -5.04457295e-01 -1.31450847e-01 -1.99952140e-01 9.03172314e-01 5.91485918e-01 -5.65454900e-01 4.82336819e-01 -3.41167562e-02 3.53807837e-01 7.00609446e-01 -7.70865321e-01 1.58781958e+00 -2.87618279e-01 5.21759748e-01 -3.53033021e-02 -9.01365161e-01 8.01880181e-01 1.48751482e-01 4.83405799e-01 -3.56411487e-01 1.30549356e-01 2.12066337e-01 2.05319505e-02 2.35470384e-01 5.25675714e-01 -1.64855391e-01 -1.76337156e-02 4.41791832e-01 8.48693192e-01 -1.74146086e-01 7.62246668e-01 4.52227533e-01 1.12379324e+00 -3.48610848e-01 -5.56073058e-03 -3.31228197e-01 -3.58550280e-01 -1.42945379e-01 8.97765934e-01 7.99625754e-01 2.22631469e-01 4.94899929e-01 4.70176101e-01 -4.64032352e-01 -1.29541123e+00 -1.09769082e+00 -1.14031933e-01 1.30881941e+00 -5.60285866e-01 -1.19500744e+00 -7.84928739e-01 -3.54286790e-01 1.10139512e-01 8.51179540e-01 -3.20600510e-01 -5.19123860e-02 -5.42888761e-01 -1.15130115e+00 6.61045134e-01 3.85389566e-01 1.30427659e-01 -6.62198424e-01 1.41769843e-02 4.29705054e-01 -1.60042644e-01 -1.17753613e+00 -3.90768170e-01 4.96096194e-01 -1.20616508e+00 -6.70729637e-01 -5.50627291e-01 -5.22687435e-01 4.62074637e-01 6.66690394e-02 1.61539102e+00 3.98356095e-02 -2.66258895e-01 2.30790190e-02 -2.62598634e-01 -3.62004340e-01 -3.03523600e-01 6.72052562e-01 3.22142355e-02 -3.67986292e-01 4.91142184e-01 -1.14757514e+00 -4.05503660e-01 -1.86033698e-03 -5.09595871e-01 2.76291549e-01 6.36063874e-01 7.04213738e-01 7.72227287e-01 -1.21477768e-01 1.87170714e-01 -1.02335215e+00 4.25923705e-01 -5.89138389e-01 -5.67180753e-01 1.48586705e-01 -8.14837813e-01 1.74712121e-01 3.43187630e-01 -5.07654309e-01 -7.65411854e-01 -2.38843381e-01 -5.07146001e-01 -7.49755383e-01 1.95651397e-01 2.89849430e-01 1.07537888e-01 1.98647350e-01 4.30872560e-01 3.57243478e-01 -3.93339723e-01 -8.03677917e-01 5.77328742e-01 3.65481913e-01 1.40007019e-01 -8.63541067e-01 8.17460001e-01 2.68534631e-01 -2.30372846e-01 -7.82424450e-01 -1.13570178e+00 -1.56347126e-01 -1.57057554e-01 4.57211971e-01 2.06670925e-01 -1.24653566e+00 -2.09450677e-01 3.71082753e-01 -7.69684017e-01 -7.72149622e-01 -6.05374813e-01 1.50515169e-01 -4.20372367e-01 3.49014968e-01 -9.44568515e-01 -5.70201993e-01 -9.02939975e-01 -7.21310794e-01 1.18862927e+00 -1.28453821e-01 -3.90754074e-01 -8.72736573e-01 -9.75720398e-03 2.96907812e-01 4.56110358e-01 -4.09647971e-01 1.08641374e+00 -6.91610396e-01 -4.48315829e-01 1.61036104e-02 -5.78432344e-02 4.24672186e-01 1.98050458e-02 1.37281772e-02 -8.10078681e-01 -5.28570354e-01 -1.91684231e-01 -4.15274113e-01 1.11080122e+00 3.37021261e-01 1.31776905e+00 -5.52667379e-01 -4.39862907e-01 9.45277393e-01 1.07039404e+00 -2.03241080e-01 5.30820668e-01 -2.01867357e-01 7.35751152e-01 -1.42629609e-01 2.22485051e-01 7.02282846e-01 4.24745709e-01 3.79001826e-01 2.63671249e-01 1.40302315e-01 -3.62029642e-01 -6.57497227e-01 3.13270807e-01 1.42995393e+00 -9.77061689e-02 -3.45578283e-01 -8.69268298e-01 7.38191307e-01 -1.65949559e+00 -9.25057232e-01 4.26109098e-02 1.96734083e+00 1.22533011e+00 3.80093902e-01 -7.16432333e-02 1.66894943e-01 4.97305602e-01 3.58386457e-01 -4.07830596e-01 -3.86614710e-01 -2.66317368e-01 8.32914233e-01 6.06768727e-01 5.95954478e-01 -8.20189714e-01 1.37336183e+00 7.26872396e+00 1.54505312e+00 -9.07802820e-01 4.95042026e-01 9.25291002e-01 -5.93193650e-01 -6.51648402e-01 2.59000540e-01 -1.37213504e+00 5.72736442e-01 1.26738906e+00 -2.86473185e-01 6.26771331e-01 1.05894291e+00 -1.99982285e-01 2.66292125e-01 -8.57163131e-01 9.42484140e-01 -1.09087020e-01 -1.32737577e+00 3.17945898e-01 2.14167148e-01 8.17553163e-01 6.45466328e-01 1.79564089e-01 6.02270663e-01 8.64241838e-01 -8.14006805e-01 7.58500159e-01 1.53463170e-01 8.08770835e-01 -6.71353102e-01 1.29105955e-01 2.69780725e-01 -1.15108633e+00 -4.14426699e-02 -9.35404062e-01 4.67990227e-02 2.88411975e-01 1.18034089e+00 -8.66515636e-01 1.34752616e-01 7.13976920e-01 8.20295036e-01 -4.41392809e-01 7.75460362e-01 -4.52176809e-01 1.30724895e+00 -8.78512383e-01 1.56932414e-01 2.93995112e-01 -3.78640071e-02 5.53910971e-01 1.12858641e+00 7.80379772e-01 -1.92235023e-01 -9.11809318e-03 7.49854624e-01 -4.25746471e-01 -1.99609116e-01 -1.74360722e-01 -3.20272386e-01 8.23328972e-01 1.05039167e+00 -3.80189002e-01 -6.73668921e-01 -9.60186794e-02 8.04274619e-01 6.86705291e-01 1.63562536e-01 -1.00305986e+00 -1.11415230e-01 8.62305820e-01 2.87681699e-01 8.43110800e-01 -1.77345708e-01 1.48996249e-01 -1.52314782e+00 -1.73206463e-01 -1.07780278e+00 3.74797553e-01 -7.25041687e-01 -1.28036070e+00 7.20901430e-01 2.68445592e-02 -5.26200116e-01 -3.03921252e-01 -4.47006375e-01 -3.98498386e-01 6.71953917e-01 -1.18345714e+00 -1.12698841e+00 7.18404353e-02 6.30710483e-01 6.52434766e-01 -2.20038176e-01 9.30793107e-01 4.73932296e-01 -6.40692651e-01 8.50228369e-01 1.61635071e-01 -2.37905368e-01 4.34323967e-01 -1.33976853e+00 9.60274637e-01 7.04515576e-01 5.53811431e-01 6.83376431e-01 6.94569707e-01 -5.28578103e-01 -1.39016104e+00 -1.30194056e+00 1.20782185e+00 -3.90809327e-01 9.41771209e-01 -6.60680413e-01 -6.54505014e-01 1.13534749e+00 1.99079961e-01 5.19419946e-02 6.29988194e-01 5.73293686e-01 -6.34820044e-01 -1.35082930e-01 -9.23935831e-01 5.52340388e-01 1.44442403e+00 -2.83691645e-01 -5.78092575e-01 6.26683116e-01 1.01267886e+00 -4.92493063e-01 -1.09922647e+00 3.02385926e-01 3.62711310e-01 -7.53600776e-01 9.93813634e-01 -5.64680815e-01 5.23282774e-02 2.08570212e-01 -3.34423393e-01 -1.17850125e+00 -6.31778300e-01 -1.18474066e+00 -7.92130709e-01 1.05994809e+00 6.24974787e-01 -7.81273067e-01 9.93982673e-01 1.20879069e-01 -2.18072802e-01 -9.59887385e-01 -1.10145235e+00 -7.68934667e-01 2.21334666e-01 -6.64661288e-01 8.46044362e-01 5.61130285e-01 -2.27402806e-01 3.79898548e-01 -6.68626726e-01 -1.36843115e-01 8.42569828e-01 1.87496483e-01 5.58687031e-01 -1.24479914e+00 -7.27855444e-01 -3.53335768e-01 -1.68827564e-01 -1.43773866e+00 8.47062618e-02 -1.03940046e+00 -3.97421986e-01 -1.37103522e+00 3.41106504e-01 -6.62373483e-01 -2.53477871e-01 9.39926088e-01 -8.62438828e-02 4.03671026e-01 1.44051433e-01 2.60912776e-01 -5.27240753e-01 7.57446945e-01 9.36121404e-01 -2.48402417e-01 3.10439497e-01 -8.32904652e-02 -8.80507886e-01 6.91698611e-01 1.08451855e+00 -7.30541110e-01 -5.24492979e-01 -8.25061679e-01 3.12853158e-01 -3.07091594e-01 -6.08944185e-02 -1.00363839e+00 -7.38835558e-02 -3.08548436e-02 1.57505527e-01 -5.68259418e-01 7.09067523e-01 -3.44662577e-01 2.49433249e-01 2.81870246e-01 -8.35439749e-03 1.66456159e-02 2.66967237e-01 3.15635353e-01 -1.13076475e-02 -2.52172619e-01 6.67206109e-01 -2.28562728e-01 -2.32209340e-01 7.17928767e-01 -3.48384351e-01 3.84526908e-01 4.43161517e-01 1.32418260e-01 -3.82742852e-01 -3.22252333e-01 -8.17278147e-01 6.44898042e-02 1.43078625e-01 1.22326016e-01 4.23522234e-01 -1.32383025e+00 -7.19868481e-01 2.74751544e-01 -9.59532931e-02 -2.56923623e-02 2.00573474e-01 6.53806150e-01 -2.74873525e-01 5.88644147e-01 2.40554139e-01 -5.43503940e-01 -1.15091658e+00 2.32703373e-01 2.20685825e-01 -5.80217063e-01 -7.15758085e-01 1.19649315e+00 3.10694966e-02 -2.50216424e-01 1.52259573e-01 -6.73682988e-01 5.75436532e-01 -1.06788218e-01 4.29751843e-01 1.91570997e-01 2.34377176e-01 -2.78775752e-01 3.86951678e-03 3.75435084e-01 -3.78656507e-01 -1.37508968e-02 1.48614275e+00 2.54383445e-01 -2.43397802e-01 1.52200505e-01 8.81073952e-01 -4.13556062e-02 -1.16199946e+00 -2.86960751e-01 -1.62727356e-01 -3.01400095e-01 2.09941134e-01 -5.28914034e-01 -1.29262805e+00 8.80289018e-01 1.40362144e-01 1.98706254e-01 9.49067414e-01 1.74666822e-01 1.22161663e+00 7.05291927e-01 6.58519745e-01 -8.34246993e-01 -1.18672624e-01 7.85043836e-01 5.64196467e-01 -6.96209669e-01 -1.17977999e-01 -3.18500906e-01 -3.33922327e-01 5.72655916e-01 2.54099250e-01 -3.03601921e-01 6.69481218e-01 5.99710703e-01 -5.22958815e-01 -1.61776140e-01 -1.50661945e+00 -1.98994339e-01 2.86104046e-02 5.28517544e-01 4.12359059e-01 3.44181508e-01 1.12646841e-01 6.77303731e-01 -6.91421986e-01 -2.09707335e-01 4.60646674e-02 6.97240472e-01 -8.23262930e-01 -1.44468665e+00 4.30159792e-02 8.02482724e-01 -4.91638660e-01 -7.53410518e-01 -1.02808937e-01 5.93408287e-01 8.29741135e-02 7.29779005e-01 2.44220987e-01 -3.19002897e-01 -9.71144065e-03 2.62584269e-01 8.00315201e-01 -8.24990869e-01 -3.75617474e-01 2.99053460e-01 2.87998766e-01 -6.21583462e-01 -1.25503791e-02 -5.77268779e-01 -8.67019951e-01 -7.88595080e-01 -1.45566434e-01 2.14995146e-01 2.35457122e-01 6.63993418e-01 5.99579632e-01 3.13315541e-01 1.33515805e-01 -6.97861612e-01 -5.89781344e-01 -1.37615967e+00 -5.61438084e-01 -7.70130679e-02 -1.30723149e-01 -3.73205483e-01 -4.53311652e-01 -1.26589823e-03]
[8.711661338806152, 3.586719036102295]
21c7c943-3b12-415e-9769-a4f551bb421b
a-majorization-minimization-gauss-newton
2304.1394
null
https://arxiv.org/abs/2304.13940v1
https://arxiv.org/pdf/2304.13940v1.pdf
A Majorization-Minimization Gauss-Newton Method for 1-Bit Matrix Completion
In 1-bit matrix completion, the aim is to estimate an underlying low-rank matrix from a partial set of binary observations. We propose a novel method for 1-bit matrix completion called MMGN. Our method is based on the majorization-minimization (MM) principle, which yields a sequence of standard low-rank matrix completion problems in our setting. We solve each of these sub-problems by a factorization approach that explicitly enforces the assumed low-rank structure and then apply a Gauss-Newton method. Our numerical studies and application to a real-data example illustrate that MMGN outputs comparable if not more accurate estimates, is often significantly faster, and is less sensitive to the spikiness of the underlying matrix than existing methods.
['Boaz Nadler', 'Eric C. Chi', 'Xu Han', 'Xiaoqian Liu']
2023-04-27
null
null
null
null
['low-rank-matrix-completion', 'matrix-completion']
['methodology', 'methodology']
[ 6.11135721e-01 -1.27811968e-01 -2.61336356e-01 -6.40584007e-02 -1.07125068e+00 -4.28204209e-01 4.63543564e-01 -1.47594333e-01 -3.84653538e-01 7.76159942e-01 4.09168333e-01 -4.34900373e-01 -2.61124969e-01 -2.23386362e-01 -6.77370310e-01 -7.02956617e-01 -4.68564034e-01 3.34631860e-01 -5.18070936e-01 -2.30791494e-01 2.25375384e-01 1.60686597e-01 -6.29813969e-01 2.36250013e-01 4.91985410e-01 8.71686876e-01 3.46224532e-02 1.13818002e+00 4.69717741e-01 7.22352684e-01 -2.22022161e-01 -5.93390584e-01 6.75821364e-01 -1.23306252e-01 -5.60223877e-01 4.99358118e-01 4.17662382e-01 -3.55614841e-01 -7.02453017e-01 1.32215989e+00 2.67183661e-01 3.39081138e-03 5.85105360e-01 -8.12947631e-01 -2.94320017e-01 6.05576038e-01 -8.71899188e-01 1.34327516e-01 5.31071424e-01 -1.91188589e-01 1.10922897e+00 -1.24904203e+00 3.80475670e-01 1.38316441e+00 1.01798177e+00 -4.13602637e-03 -1.74096453e+00 -3.97919625e-01 -3.83190244e-01 -1.42456651e-01 -1.72943294e+00 -7.74218500e-01 4.32941258e-01 -2.85737991e-01 3.01354498e-01 3.71019870e-01 1.56581908e-01 7.58556247e-01 2.32124612e-01 7.32104421e-01 1.23438370e+00 -3.13493699e-01 2.32998401e-01 -2.61927933e-01 1.40646040e-01 5.10910034e-01 6.72716260e-01 3.27188969e-02 -6.28329694e-01 -9.58504796e-01 6.44288838e-01 4.37042229e-02 -4.92420435e-01 -3.16117197e-01 -1.45818496e+00 9.55128789e-01 -7.11070225e-02 2.48692418e-03 -5.58596671e-01 6.19736016e-01 1.85247973e-01 4.29439425e-01 1.81222394e-01 -5.95857501e-02 -1.15956374e-01 -8.82918760e-02 -1.22677767e+00 8.24167132e-02 1.20467830e+00 9.14393485e-01 9.84899938e-01 1.13291748e-01 -1.11758485e-01 5.89578927e-01 5.42935550e-01 8.31475794e-01 6.15386181e-02 -9.26922917e-01 7.79266536e-01 -2.62926936e-01 1.02640025e-01 -1.22598982e+00 -2.01478258e-01 -5.45389891e-01 -1.27242136e+00 -1.22200422e-01 3.32530349e-01 -2.49798954e-01 -6.65160120e-01 1.56506693e+00 6.67872354e-02 4.06648397e-01 2.64099121e-01 8.87615144e-01 1.46590382e-01 5.88654101e-01 -5.67128837e-01 -4.06245708e-01 1.08331978e+00 -3.58554363e-01 -9.02245760e-01 -3.07061732e-01 4.01749462e-01 -1.06483746e+00 6.10134065e-01 7.08764911e-01 -1.03508246e+00 -2.29375958e-01 -1.30630147e+00 4.86408733e-02 2.56694704e-01 4.91654158e-01 1.03100526e+00 1.12902677e+00 -1.19076025e+00 6.03462398e-01 -5.61926901e-01 8.84018764e-02 8.23759958e-02 7.08974898e-01 -6.17140055e-01 -5.38263440e-01 -8.53981972e-01 1.98619395e-01 2.14666009e-01 1.82944074e-01 -7.92406082e-01 -4.70691144e-01 -8.36586893e-01 -1.75821438e-01 2.37016156e-01 -5.64132512e-01 1.05435717e+00 -6.18780196e-01 -1.38036537e+00 5.14384449e-01 -5.07373393e-01 -5.65430701e-01 2.23511353e-01 -3.50868672e-01 -4.14098918e-01 1.38807639e-01 -6.91733807e-02 -4.73278500e-02 1.50223804e+00 -1.11110139e+00 -2.38048419e-01 -2.36004874e-01 -2.45826855e-01 -1.39436916e-01 -1.59281164e-01 -2.62926966e-01 -6.69310272e-01 -9.40987408e-01 5.86469114e-01 -1.13970804e+00 -7.94055521e-01 -2.95992166e-01 -6.35942996e-01 5.34101963e-01 3.05179358e-01 -7.20902801e-01 1.42190230e+00 -2.38420415e+00 2.61084586e-01 8.93572628e-01 5.71612775e-01 -1.15585215e-01 -2.42392495e-01 8.88480186e-01 -3.01140279e-01 -2.04371050e-01 -5.35921216e-01 -6.73363805e-01 -2.24550694e-01 2.00303853e-01 -5.34853220e-01 1.06854331e+00 -1.32701546e-01 4.04855162e-01 -1.04792190e+00 5.43294065e-02 -1.93189755e-01 4.87651438e-01 -7.60487199e-01 -1.31419450e-01 2.23284781e-01 2.38425598e-01 -1.50309399e-01 5.94904661e-01 1.12464607e+00 -4.69402760e-01 6.22006595e-01 -6.70246184e-01 1.58963129e-01 8.00530761e-02 -1.96402454e+00 1.73958516e+00 -2.72312492e-01 8.85995150e-01 6.49021149e-01 -9.33043301e-01 6.38278961e-01 3.31967115e-01 5.62247276e-01 -7.87726417e-02 1.37466386e-01 3.95075083e-01 -3.35515216e-02 1.34477634e-02 7.55629241e-01 -4.53230068e-02 -3.53728719e-02 6.69235051e-01 -4.09887172e-02 -2.65152194e-02 4.23125148e-01 8.82639229e-01 1.41672528e+00 -5.93560517e-01 4.75394994e-01 -4.82536882e-01 3.98759604e-01 -3.58400583e-01 4.93560940e-01 1.02038276e+00 2.84775138e-01 6.87448263e-01 4.36501265e-01 -1.35317609e-01 -1.02290022e+00 -1.10946226e+00 -2.54877478e-01 5.28367698e-01 -1.55238286e-01 -1.03466499e+00 -3.80357653e-01 -1.76360890e-01 1.72797292e-01 1.06332496e-01 -3.76140863e-01 2.74567574e-01 -2.88022965e-01 -1.35586095e+00 5.43459356e-01 1.16926348e-02 1.86931655e-01 1.61617696e-01 2.18200326e-01 3.34127367e-01 -1.76931396e-01 -1.38242042e+00 -6.90740228e-01 2.75999933e-01 -1.01968706e+00 -1.04771161e+00 -6.97614789e-01 -3.90423149e-01 1.10285151e+00 5.05917311e-01 1.02056193e+00 -9.41193178e-02 -1.33080482e-01 5.99649191e-01 -1.35762110e-01 1.85319915e-01 -3.23978335e-01 -5.03917336e-01 5.49489856e-01 6.73351884e-01 1.08437546e-01 -7.79186189e-01 -3.72083485e-01 5.14702052e-02 -1.08088708e+00 -2.22736642e-01 1.03215945e+00 1.04048455e+00 7.13222027e-01 2.22635359e-01 6.98013529e-02 -1.21263599e+00 8.90856624e-01 -5.48458040e-01 -7.47517526e-01 -1.29632547e-01 -6.36854827e-01 3.56835723e-01 4.99160200e-01 -5.45934200e-01 -5.11468053e-01 5.77602327e-01 1.93971142e-01 -5.71052909e-01 6.03024065e-01 7.35984862e-01 1.28053725e-02 -5.45694053e-01 8.28564823e-01 2.75099188e-01 -8.63784552e-02 -5.06076992e-01 6.72535658e-01 5.42015851e-01 7.98691094e-01 -7.29713082e-01 1.35093224e+00 8.28788519e-01 4.53763425e-01 -1.06269097e+00 -6.20226800e-01 -6.67385757e-01 -5.11699557e-01 2.28006065e-01 6.58882335e-02 -1.36054230e+00 -8.07394862e-01 8.27972591e-02 -9.42449152e-01 -2.48417497e-01 -4.22320850e-02 7.58578002e-01 -5.74879408e-01 1.00470376e+00 -7.36281633e-01 -8.61041188e-01 -2.28481367e-01 -8.53721797e-01 7.99493134e-01 -3.01275432e-01 -9.48436931e-02 -8.87655377e-01 4.25553173e-01 -8.73594917e-03 2.56244540e-01 -1.78963140e-01 5.17235041e-01 -7.94014186e-02 -5.74280500e-01 -4.85734969e-01 -4.47185755e-01 3.57742757e-01 8.07633623e-02 -3.97252947e-01 -6.95607543e-01 -8.29530776e-01 2.58837372e-01 -1.57183096e-01 8.98369908e-01 2.68222064e-01 8.42598736e-01 -6.84251487e-01 -1.31262287e-01 9.48803902e-01 1.66744375e+00 -6.31911695e-01 6.81721330e-01 -2.68651605e-01 7.34496415e-01 1.32020757e-01 5.03855109e-01 1.02830839e+00 1.19695097e-01 6.06314361e-01 1.93020076e-01 1.79451978e-04 4.93788570e-02 -6.84625283e-02 5.88308096e-01 1.09311712e+00 1.61946550e-01 6.46978244e-03 -6.79843068e-01 4.18294370e-01 -1.96760809e+00 -8.11140180e-01 -5.38863003e-01 2.45942831e+00 1.00467587e+00 -1.11715250e-01 4.99921963e-02 3.19063336e-01 6.04730368e-01 1.98364988e-01 -2.29845390e-01 -1.09496251e-01 -3.66691411e-01 2.71709174e-01 1.26210713e+00 6.08624697e-01 -1.06033492e+00 5.05562007e-01 8.23501778e+00 9.94119763e-01 -6.27543509e-01 4.11075763e-02 4.78671134e-01 6.52678907e-02 -3.07203591e-01 3.67225736e-01 -7.54506469e-01 3.08181822e-01 8.90823483e-01 -9.47977901e-02 9.00953233e-01 5.63512087e-01 7.20625222e-02 -1.40913129e-01 -1.27364278e+00 1.63082850e+00 1.57666758e-01 -1.33524036e+00 -6.07740618e-02 4.10337657e-01 1.04762387e+00 1.29710883e-01 2.49376982e-01 -1.93362236e-01 5.44907212e-01 -1.04088664e+00 2.53541738e-01 4.62779045e-01 1.11275768e+00 -5.64917266e-01 3.84889275e-01 8.80353004e-02 -1.25697899e+00 -1.47196695e-01 -6.52546942e-01 -2.49052912e-01 3.21695209e-01 1.48457527e+00 -7.81362116e-01 4.35812116e-01 2.27144107e-01 7.13003635e-01 -1.97080806e-01 1.13727701e+00 -9.60348099e-02 8.45382392e-01 -7.31266975e-01 3.89168978e-01 5.53229749e-02 -4.42318797e-01 8.43186557e-01 1.48050928e+00 3.28858703e-01 2.78709143e-01 3.81010056e-01 4.20709282e-01 -1.72930852e-01 4.59035635e-02 -4.83708233e-01 -2.45621055e-01 1.94513410e-01 1.26909149e+00 -4.09532011e-01 -3.43763351e-01 -5.79954088e-01 1.10152650e+00 1.16765633e-01 6.34300292e-01 -1.88945130e-01 -3.25113505e-01 6.57740414e-01 -7.05479905e-02 4.28559721e-01 -7.20838070e-01 -4.13136154e-01 -1.37505305e+00 -3.39582786e-02 -1.38425684e+00 3.88090909e-01 -2.34627053e-01 -1.21613026e+00 1.93634942e-01 -3.76810074e-01 -1.33120263e+00 -3.96974593e-01 -6.02016389e-01 -3.68448168e-01 8.31395984e-01 -1.18956351e+00 -4.60799098e-01 6.41295388e-02 8.21843088e-01 4.58788909e-02 -1.66418493e-01 6.25984967e-01 5.58981478e-01 -3.44859451e-01 5.74492335e-01 7.18852282e-01 9.15318206e-02 6.18338168e-01 -1.16983521e+00 3.85689259e-01 1.25643063e+00 3.38460326e-01 9.38852787e-01 8.97955239e-01 -7.61421442e-01 -2.23803759e+00 -8.12228441e-01 6.51885390e-01 -1.77198648e-01 8.72758508e-01 -3.81564617e-01 -4.09364820e-01 7.43269563e-01 -1.57998681e-01 1.51352674e-01 9.39034700e-01 2.52460510e-01 -5.74135780e-01 8.71670526e-03 -8.61170232e-01 5.43372154e-01 6.49893463e-01 -8.86215091e-01 -2.13378504e-01 6.72492027e-01 1.15440525e-01 -3.73659074e-01 -8.43600094e-01 3.51692140e-01 3.55130702e-01 -6.07570171e-01 1.23597717e+00 -2.92374104e-01 1.46936953e-01 -6.40000939e-01 -8.81012261e-01 -9.28135633e-01 -5.74412227e-01 -1.44089437e+00 -5.46735644e-01 6.67015970e-01 2.32005432e-01 -2.49661341e-01 9.64483261e-01 2.76888490e-01 2.64561653e-01 -4.19428408e-01 -8.30674648e-01 -7.50263512e-01 -7.07182884e-01 -7.42752492e-01 1.01149581e-01 8.04070055e-01 6.94605112e-02 4.70403314e-01 -1.12448812e+00 5.11581302e-01 1.29402471e+00 -1.15800664e-01 1.15570867e+00 -9.07705545e-01 -1.00789320e+00 5.29980510e-02 -4.79793519e-01 -1.61985624e+00 -1.79359823e-01 -9.19382095e-01 -4.54394929e-02 -9.93526459e-01 3.49522233e-01 -5.17801046e-01 -1.11060843e-01 1.97966069e-01 -9.03068557e-02 7.67574370e-01 1.26255631e-01 3.08883250e-01 -5.91196477e-01 3.35771382e-01 7.61597991e-01 -1.61314666e-01 -3.82931530e-02 2.54701257e-01 -7.68212199e-01 4.65285480e-01 3.28085721e-01 -5.46234727e-01 -2.77035385e-01 -2.56073833e-01 7.75749028e-01 2.64335930e-01 6.72376007e-02 -1.06355810e+00 3.82746637e-01 1.61113471e-01 3.69007826e-01 -6.35353327e-01 6.19118690e-01 -6.66090369e-01 2.92591482e-01 5.57096660e-01 3.77294607e-02 4.68117446e-02 -3.90092172e-02 9.65593576e-01 -1.47599623e-01 -3.25996935e-01 5.67496002e-01 2.48953387e-01 -4.29443449e-01 4.55698520e-01 -5.68385363e-01 -1.06127532e-02 3.29453140e-01 -4.25288007e-02 5.04653230e-02 -8.69434714e-01 -7.40949273e-01 -1.13375686e-01 3.67052019e-01 -2.99246669e-01 6.72761738e-01 -1.53837502e+00 -1.06030297e+00 3.46303493e-01 -2.33997032e-02 -4.58569884e-01 3.82228824e-03 9.99297678e-01 -2.94105768e-01 4.10936177e-01 4.02877122e-01 -5.96301913e-01 -1.08024228e+00 4.81810510e-01 -1.59903616e-01 -5.22599518e-01 -4.57596689e-01 7.74625421e-01 1.90030351e-01 -7.08909240e-03 1.32285342e-01 1.59541905e-01 3.12563986e-01 -2.98395693e-01 8.93007815e-01 4.37031806e-01 -2.34504677e-02 -7.54482865e-01 -1.69026092e-01 5.32467484e-01 -9.53261033e-02 -4.96243566e-01 9.64513183e-01 -3.44205707e-01 -3.86605501e-01 2.36053184e-01 1.41197324e+00 6.96534455e-01 -1.06946027e+00 -6.99364603e-01 7.10338503e-02 -8.25184584e-01 1.71061575e-01 -9.96762440e-02 -9.53867555e-01 4.83697563e-01 4.22493130e-01 -3.99163775e-02 1.09867239e+00 -4.69382465e-01 6.25736177e-01 8.18099380e-01 4.23375607e-01 -8.42759311e-01 6.46238998e-02 7.49959290e-01 5.97815573e-01 -1.03230655e+00 6.05763555e-01 -6.45897985e-01 -1.40396968e-01 1.17478764e+00 -3.02313566e-01 -3.60557497e-01 7.99197316e-01 4.01456177e-01 -1.76328108e-01 -4.31843214e-02 -6.56837702e-01 -4.71046939e-02 4.62788463e-01 6.06862426e-01 2.61432916e-01 1.37674779e-01 -4.02455151e-01 4.10930723e-01 -2.35493600e-01 -2.34761178e-01 7.27108359e-01 6.54920101e-01 -2.62155473e-01 -1.41605639e+00 -9.47178364e-01 6.70896292e-01 -7.39386320e-01 -5.69039345e-01 -6.08722456e-02 9.83599499e-02 -4.54274267e-01 1.08193398e+00 -3.40783030e-01 -5.90616822e-01 -2.42235586e-01 -5.08937120e-01 4.54100430e-01 -6.46894753e-01 -2.15536654e-02 4.25831556e-01 1.54685870e-01 -8.04901659e-01 -2.02850372e-01 -7.84638405e-01 -8.66282225e-01 -5.74973285e-01 -2.46839225e-01 3.00948441e-01 8.20051968e-01 7.57205665e-01 2.81520307e-01 -4.30047065e-02 9.25836444e-01 -9.22723114e-01 -8.66669536e-01 -6.01666212e-01 -9.27009761e-01 2.65903234e-01 6.13971472e-01 -1.48948491e-01 -5.26609302e-01 -1.85005330e-02]
[6.968489646911621, 4.662806510925293]
ed64d0ba-7e42-408e-9522-c6981d56d4e7
sysnoise-exploring-and-benchmarking-training
2307.0028
null
https://arxiv.org/abs/2307.00280v1
https://arxiv.org/pdf/2307.00280v1.pdf
SysNoise: Exploring and Benchmarking Training-Deployment System Inconsistency
Extensive studies have shown that deep learning models are vulnerable to adversarial and natural noises, yet little is known about model robustness on noises caused by different system implementations. In this paper, we for the first time introduce SysNoise, a frequently occurred but often overlooked noise in the deep learning training-deployment cycle. In particular, SysNoise happens when the source training system switches to a disparate target system in deployments, where various tiny system mismatch adds up to a non-negligible difference. We first identify and classify SysNoise into three categories based on the inference stage; we then build a holistic benchmark to quantitatively measure the impact of SysNoise on 20+ models, comprehending image classification, object detection, instance segmentation and natural language processing tasks. Our extensive experiments revealed that SysNoise could bring certain impacts on model robustness across different tasks and common mitigations like data augmentation and adversarial training show limited effects on it. Together, our findings open a new research topic and we hope this work will raise research attention to deep learning deployment systems accounting for model performance. We have open-sourced the benchmark and framework at https://modeltc.github.io/systemnoise_web.
['Xianglong Liu', 'Fengwei Yu', 'Tianzi Xiao', 'Yunchen Zhang', 'Yongqiang Yao', 'Jian Hu', 'Yanfei Wang', 'Aishan Liu', 'Ruihao Gong', 'Yuhang Li', 'Yan Wang']
2023-07-01
null
null
null
null
['instance-segmentation', 'benchmarking', 'benchmarking']
['computer-vision', 'miscellaneous', 'robots']
[-1.92298159e-01 -3.11744392e-01 2.89306760e-01 -2.75138468e-01 -6.82676315e-01 -1.04415846e+00 6.29501998e-01 -2.55373538e-01 -3.11531842e-01 4.24721330e-01 -2.05031380e-01 -7.95016348e-01 1.82075977e-01 -6.08769178e-01 -1.10897458e+00 -5.85255563e-01 2.42442358e-02 -2.61412747e-02 1.75781026e-01 -2.83289522e-01 -1.65515333e-01 6.41028702e-01 -1.26552331e+00 8.25326368e-02 5.76121628e-01 9.37955320e-01 -1.58663884e-01 8.48746419e-01 2.56798893e-01 1.03329349e+00 -1.44117522e+00 -6.97155893e-01 7.25577056e-01 -7.04087391e-02 -6.87508881e-01 -5.06072938e-01 6.20918036e-01 -5.77519357e-01 -7.56089389e-01 1.16627145e+00 1.12140143e+00 -3.19532663e-01 2.28119656e-01 -1.55516028e+00 -4.74876285e-01 9.11941707e-01 -3.20639104e-01 4.49450672e-01 -1.58703864e-01 9.50604379e-01 3.06336343e-01 -4.28055942e-01 2.26843834e-01 1.25409448e+00 1.06480861e+00 7.26959169e-01 -1.20748770e+00 -1.21008790e+00 1.20247761e-02 -1.77481174e-01 -1.27047133e+00 -7.24835098e-01 5.05043626e-01 -5.92845798e-01 8.94588470e-01 3.44447076e-01 -3.37289982e-02 1.79401565e+00 6.03471518e-01 3.54513139e-01 1.06729114e+00 3.43910069e-03 4.48855460e-01 1.88892752e-01 4.28160548e-01 1.99627906e-01 3.34784091e-01 3.39986593e-01 -1.30624756e-01 -1.65536791e-01 3.73096138e-01 -1.47584915e-01 -1.25958681e-01 5.65902352e-01 -7.23139882e-01 2.25983202e-01 4.20808703e-01 2.55348265e-01 -2.43927926e-01 7.03295887e-01 7.59397447e-01 6.46416605e-01 2.59426624e-01 5.60261369e-01 -8.66643131e-01 -3.89433742e-01 -6.08586907e-01 8.81484523e-02 9.08519566e-01 1.11910379e+00 6.44427180e-01 4.59022582e-01 -1.82175323e-01 5.34586608e-01 -1.28699958e-01 6.82394147e-01 3.46003830e-01 -8.82609427e-01 3.59981805e-01 9.12019238e-02 -1.58339024e-01 -8.60197186e-01 -3.70435834e-01 -7.80756831e-01 -8.86053026e-01 1.34609282e-01 2.52395719e-01 -7.96318471e-01 -7.65278518e-01 1.78556025e+00 -9.02668666e-03 5.58302343e-01 6.35989010e-02 6.01403117e-01 1.01602638e+00 1.77252710e-01 1.97663501e-01 2.68823236e-01 1.19356799e+00 -7.82353520e-01 -5.95891654e-01 -3.17242175e-01 5.73474228e-01 -9.53896224e-01 1.25151074e+00 4.82293755e-01 -7.81275690e-01 -7.57694125e-01 -1.03586113e+00 3.69340122e-01 -4.47654843e-01 -6.34892732e-02 3.12748015e-01 1.13148117e+00 -1.12341940e+00 6.46717906e-01 -1.04404449e+00 -2.56523728e-01 5.61754167e-01 2.87875950e-01 -1.54629678e-01 2.01905876e-01 -1.32991385e+00 8.58450711e-01 -8.89428332e-02 6.04463108e-02 -1.55118108e+00 -1.09130836e+00 -5.77118993e-01 -1.57238707e-01 4.00768161e-01 -6.42717600e-01 1.61634076e+00 -7.31791794e-01 -1.23738241e+00 5.03981769e-01 2.84863800e-01 -8.72289240e-01 5.60631871e-01 -5.91634214e-01 -7.66833127e-01 -4.50566977e-01 -1.59059912e-01 2.59836376e-01 8.02190542e-01 -1.42734122e+00 -5.32174744e-02 -7.49927834e-02 4.08779293e-01 -4.33438361e-01 -5.79448700e-01 3.57963502e-01 -2.79434949e-01 -7.03690112e-01 -5.65158784e-01 -9.22000825e-01 -2.08371684e-01 -4.75985646e-01 -7.81503320e-01 2.63329566e-01 9.01430666e-01 -5.31948388e-01 1.38287795e+00 -2.27162123e+00 -5.96955478e-01 -1.77510038e-01 3.96234363e-01 7.29333341e-01 -3.25191230e-01 5.34252286e-01 -2.29598105e-01 6.63004458e-01 -3.60447653e-02 -4.74245042e-01 1.32442191e-01 1.43381342e-01 -7.37536013e-01 3.25130671e-01 3.12398851e-01 8.41072321e-01 -6.21265948e-01 2.41782125e-02 1.67638168e-01 4.69574571e-01 -3.99546444e-01 2.88717389e-01 1.99985802e-02 3.24770987e-01 -1.97596103e-01 7.43091881e-01 9.77130115e-01 1.41859055e-01 -3.54843140e-01 -3.09146106e-01 6.32940158e-02 3.42629313e-01 -8.78966570e-01 1.09696174e+00 -5.68001032e-01 8.16465318e-01 1.09742597e-01 -5.88540375e-01 7.08946168e-01 1.08774565e-01 -8.98021385e-02 -5.94902575e-01 4.82166499e-01 8.54006112e-02 2.83717841e-01 -5.30341029e-01 4.73738968e-01 8.63325298e-02 -3.08848739e-01 2.45757997e-01 3.03724427e-02 -1.66322425e-01 -2.59303749e-01 2.33055249e-01 1.72382355e+00 -4.91508931e-01 -1.50807038e-01 -1.84869424e-01 -7.97595829e-02 -2.48675570e-01 4.85297322e-01 1.30241609e+00 -4.00630265e-01 5.90608478e-01 6.07958794e-01 -2.76349008e-01 -6.81858122e-01 -1.05951679e+00 -1.82300553e-01 9.62522388e-01 -5.01708835e-02 -4.05176848e-01 -1.21247482e+00 -7.89641440e-01 6.39173295e-03 8.05807352e-01 -6.35683596e-01 -5.98868906e-01 -4.16578501e-01 -8.65696073e-01 1.50689614e+00 5.72486758e-01 7.07473695e-01 -1.03243279e+00 -3.24978501e-01 -5.51346913e-02 3.07856202e-01 -1.36727631e+00 -3.16201717e-01 4.09053862e-01 -5.52330375e-01 -1.00580561e+00 -1.55700818e-01 -5.05074084e-01 3.16659421e-01 2.84525484e-01 1.35177028e+00 2.65134662e-01 -3.04099351e-01 2.93289632e-01 -2.81244159e-01 -7.48459518e-01 -8.30371618e-01 2.93555826e-01 5.62229037e-01 -4.21387285e-01 1.17536373e-01 -6.22935355e-01 -5.17982781e-01 4.22910750e-01 -1.11134136e+00 -4.75692362e-01 3.42631727e-01 4.67554152e-01 1.46413013e-01 2.89934009e-01 5.18018425e-01 -9.64820087e-01 8.99272859e-01 -7.43541837e-01 -6.81452930e-01 8.59696195e-02 -3.55591416e-01 -2.33002141e-01 9.50767815e-01 -5.37376046e-01 -7.77579129e-01 -3.16587120e-01 -4.66991305e-01 -7.54151285e-01 -4.88571554e-01 2.43593305e-01 -3.93654525e-01 -8.23175237e-02 1.11318767e+00 -7.46044815e-02 -3.38033557e-01 -5.22559702e-01 1.78917825e-01 7.31280029e-01 6.25165164e-01 -6.14672840e-01 1.13766325e+00 2.60368288e-01 -4.58473295e-01 -7.76787877e-01 -5.45793116e-01 3.72883603e-02 -2.03105897e-01 -9.96391997e-02 5.84545910e-01 -1.14628577e+00 -6.95706308e-01 1.14954233e+00 -1.19980311e+00 -9.04093325e-01 -1.50475755e-01 -2.73396429e-02 8.38321820e-02 1.83167443e-01 -7.48995066e-01 -7.58289099e-01 -5.14157057e-01 -1.38590944e+00 9.16377127e-01 2.46079549e-01 -8.02204385e-03 -7.62665808e-01 -1.48843927e-03 2.41137594e-01 8.41565430e-01 3.16314965e-01 4.35096234e-01 -8.67512107e-01 -3.09926540e-01 -2.27085009e-01 -9.62634664e-03 9.08816695e-01 -2.16030367e-02 4.15821731e-01 -1.43387723e+00 -5.90539634e-01 3.29543650e-01 -3.87624055e-01 5.86369574e-01 9.92378816e-02 1.48188508e+00 -3.24692160e-01 -6.38332441e-02 8.02923203e-01 1.34725332e+00 2.28629395e-01 7.11295962e-01 4.20770705e-01 7.05459058e-01 1.54822737e-01 3.44017386e-01 3.50206912e-01 -3.84357497e-02 4.41617936e-01 6.93729162e-01 -3.20199698e-01 -1.67022616e-01 4.06922251e-02 7.57106423e-01 5.58336318e-01 4.82776314e-01 -4.10988331e-01 -1.35143101e+00 2.88657516e-01 -1.25182557e+00 -7.22051620e-01 -1.81004986e-01 1.91051817e+00 7.37664998e-01 4.72314000e-01 -1.56727180e-01 9.75706056e-02 6.74233079e-01 7.14433938e-02 -8.07336390e-01 -1.91716731e-01 -1.34397715e-01 3.32970053e-01 7.66272962e-01 2.66085953e-01 -1.11203742e+00 1.22032285e+00 6.73987532e+00 9.04081702e-01 -1.32844090e+00 3.59562874e-01 9.79695261e-01 -2.49607220e-01 -6.26621917e-02 -2.07815841e-01 -7.60492980e-01 7.84784853e-01 1.33841360e+00 -1.63073063e-01 2.27199972e-01 1.03755164e+00 4.55245793e-01 1.05981156e-01 -1.04467630e+00 7.34972775e-01 -3.14726502e-01 -1.10125375e+00 -1.33312479e-01 -1.45474577e-03 6.46754503e-01 6.66279554e-01 3.63283783e-01 6.66101694e-01 7.41361856e-01 -1.12380552e+00 8.09387922e-01 3.39319527e-01 8.11722875e-01 -7.88851678e-01 1.05188775e+00 3.77017528e-01 -8.36710632e-01 -9.42948684e-02 -3.21158350e-01 -2.79167980e-01 -6.93560615e-02 6.54154718e-01 -9.20152962e-01 2.37998992e-01 9.70733404e-01 3.23090434e-01 -1.00743151e+00 8.64044905e-01 -1.03727184e-01 1.20531619e+00 -1.89044088e-01 3.30652922e-01 -1.05511369e-02 4.26778197e-01 4.94131446e-01 1.27496767e+00 8.66395608e-02 -1.44083247e-01 -1.24413475e-01 9.90783334e-01 -3.95087361e-01 -6.33106053e-01 -6.31976068e-01 9.34624695e-04 9.60838079e-01 1.23103666e+00 -5.13785660e-01 -2.04468966e-01 -1.29369795e-01 8.46987545e-01 2.10277848e-02 4.76153314e-01 -1.44789052e+00 -3.14544052e-01 1.29619527e+00 4.19548899e-02 -2.52114773e-01 -6.83025941e-02 -5.64956367e-01 -9.45580482e-01 1.29142433e-01 -1.21616030e+00 -8.56200084e-02 -5.25573850e-01 -1.40275753e+00 9.57665324e-01 -1.12072088e-01 -1.04846144e+00 1.57266751e-01 -4.41964328e-01 -1.00002229e+00 7.89293289e-01 -1.17527020e+00 -7.15571821e-01 -4.74818677e-01 3.57873917e-01 3.66772681e-01 -4.69665080e-01 7.36072540e-01 5.66042662e-01 -1.29102302e+00 1.18498766e+00 2.68015265e-01 4.64068502e-01 8.95154417e-01 -1.14172673e+00 1.42304766e+00 1.27732491e+00 -2.60969490e-01 9.22105789e-01 8.56785834e-01 -5.43812394e-01 -1.29536235e+00 -1.51422071e+00 2.19080910e-01 -9.37118888e-01 9.02135849e-01 -6.91431403e-01 -9.12453830e-01 6.30683601e-01 3.35944623e-01 1.67199135e-01 6.73314154e-01 -1.90190852e-01 -4.97467965e-01 -2.23357826e-01 -1.15565383e+00 6.52076304e-01 1.03082490e+00 -6.73906088e-01 -7.50570446e-02 1.62802190e-01 1.40720141e+00 -5.91416478e-01 -8.66362154e-01 3.41924042e-01 1.52300373e-01 -1.12199605e+00 8.97515595e-01 -7.11846173e-01 3.83570790e-01 -2.34629661e-01 -2.19477907e-01 -1.26931357e+00 -9.85899102e-03 -8.60224187e-01 -7.22407997e-02 1.69388425e+00 4.95052904e-01 -8.83445203e-01 4.83196199e-01 8.30782652e-01 -4.05180454e-01 -5.70925355e-01 -7.35756040e-01 -1.08948541e+00 4.03394967e-01 -7.72061229e-01 9.64316189e-01 1.07243705e+00 -7.76861072e-01 1.95885655e-02 -3.62481177e-01 4.84355778e-01 3.86260659e-01 -7.01973975e-01 1.24527657e+00 -6.32218957e-01 -5.00344038e-01 -4.70475823e-01 -2.65604705e-01 -6.60179377e-01 3.12496312e-02 -4.80191976e-01 8.65340326e-03 -9.21970487e-01 -2.06824820e-02 -6.30428731e-01 -3.66614848e-01 6.41450226e-01 -1.94969520e-01 2.63242304e-01 4.28629488e-01 1.59061268e-01 -6.30611718e-01 3.31237972e-01 7.58682787e-01 -1.55996785e-01 5.81436567e-02 1.61802471e-01 -1.04604268e+00 6.09844744e-01 1.17274106e+00 -7.52715826e-01 -2.79431909e-01 -9.59302723e-01 2.78184693e-02 -4.12396848e-01 4.98544574e-01 -1.50059688e+00 9.72623825e-02 1.70788690e-01 8.11982155e-03 -7.28694648e-02 -3.61079238e-02 -9.02789354e-01 4.22293156e-01 5.82073867e-01 -8.88839960e-02 3.06824416e-01 8.44614148e-01 2.58752167e-01 1.21824935e-01 -4.77794409e-02 7.33477533e-01 2.55998731e-01 -5.59484184e-01 2.84513563e-01 -5.00678897e-01 1.82296708e-01 9.20416594e-01 1.46833330e-01 -7.91300237e-01 -2.33442307e-01 -4.47978795e-01 1.97855055e-01 4.99727935e-01 7.56548166e-01 2.49969080e-01 -1.08707225e+00 -6.44029260e-01 1.45432860e-01 -9.37866867e-02 -6.91034412e-03 3.94602090e-01 5.82310081e-01 -6.01905942e-01 -7.97909275e-02 -4.44630999e-03 -5.67624331e-01 -1.34196770e+00 4.17602450e-01 8.55101764e-01 1.42988302e-02 -8.80955830e-02 1.13106775e+00 1.69843689e-01 -8.05505097e-01 6.45895779e-01 -4.06950593e-01 3.75968426e-01 -2.02019557e-01 4.67152447e-01 4.62394446e-01 4.71287549e-01 -5.40496707e-01 -4.74129826e-01 6.74657375e-02 -1.85829729e-01 2.89636314e-01 1.02011335e+00 2.23708063e-01 -3.37191895e-02 3.42376381e-01 1.16024446e+00 -2.31903076e-01 -1.25678146e+00 -2.02492520e-01 -3.39485943e-01 -2.52489835e-01 -4.72087646e-03 -1.12476480e+00 -1.32630229e+00 9.24098969e-01 8.42123389e-01 5.67153990e-01 1.23816681e+00 -8.63855705e-02 8.72856140e-01 4.43975896e-01 4.78042632e-01 -7.16608584e-01 8.30775574e-02 7.78753519e-01 7.83926368e-01 -1.27427113e+00 -4.33685511e-01 -1.49788097e-01 -5.36930740e-01 5.33772230e-01 1.07799506e+00 2.00148579e-02 7.78953195e-01 1.06204963e+00 4.71872777e-01 -4.78860773e-02 -8.35515440e-01 1.08671345e-01 -3.04548591e-01 6.17476702e-01 4.60741669e-01 1.29448548e-01 3.56061757e-01 1.08816457e+00 -6.09244943e-01 -2.52423614e-01 5.78748047e-01 8.54408324e-01 -9.80463475e-02 -9.84328449e-01 -5.74915469e-01 4.84765947e-01 -6.76201999e-01 -3.26133907e-01 -6.49461389e-01 6.94843352e-01 3.41992766e-01 1.16537249e+00 -2.50601806e-02 -1.09784138e+00 7.06043303e-01 -2.26413071e-01 -4.41421978e-02 -4.90313262e-01 -1.33507109e+00 -3.77651095e-01 3.09002046e-02 -7.00097263e-01 2.85601079e-01 -4.63120848e-01 -9.43127930e-01 -8.11635315e-01 -3.64834100e-01 -1.20085970e-01 6.96151495e-01 7.14688122e-01 6.76117837e-01 1.04300451e+00 6.12572730e-01 -8.44400048e-01 -8.94643664e-01 -1.13573444e+00 -2.05518693e-01 2.99761832e-01 4.76406932e-01 -3.01213622e-01 -7.04578459e-01 -1.15779042e-02]
[5.678770542144775, 7.7696661949157715]
28ef6a6a-0e79-4e6b-9884-ac02820b5ffc
supervised-prototypical-contrastive-learning
2210.08713
null
https://arxiv.org/abs/2210.08713v2
https://arxiv.org/pdf/2210.08713v2.pdf
Supervised Prototypical Contrastive Learning for Emotion Recognition in Conversation
Capturing emotions within a conversation plays an essential role in modern dialogue systems. However, the weak correlation between emotions and semantics brings many challenges to emotion recognition in conversation (ERC). Even semantically similar utterances, the emotion may vary drastically depending on contexts or speakers. In this paper, we propose a Supervised Prototypical Contrastive Learning (SPCL) loss for the ERC task. Leveraging the Prototypical Network, the SPCL targets at solving the imbalanced classification problem through contrastive learning and does not require a large batch size. Meanwhile, we design a difficulty measure function based on the distance between classes and introduce curriculum learning to alleviate the impact of extreme samples. We achieve state-of-the-art results on three widely used benchmarks. Further, we conduct analytical experiments to demonstrate the effectiveness of our proposed SPCL and curriculum learning strategy. We release the code at https://github.com/caskcsg/SPCL.
['Songlin Hu', 'Hui Xue', 'Longtao Huang', 'Xiaohui Song']
2022-10-17
null
null
null
null
['imbalanced-classification', 'emotion-recognition-in-conversation']
['miscellaneous', 'natural-language-processing']
[-1.34943843e-01 -1.76756233e-01 -2.02268571e-01 -7.97451556e-01 -6.90906167e-01 -4.13721293e-01 3.88718992e-01 2.48832285e-01 -2.94165820e-01 5.15888929e-01 2.25346223e-01 -2.59837229e-02 2.17988230e-02 -3.74196917e-01 -3.45252484e-01 -6.53517842e-01 1.32953778e-01 2.02551022e-01 -2.43846461e-01 -4.79046166e-01 1.82896897e-01 2.59155426e-02 -1.35030329e+00 5.31813860e-01 1.13097179e+00 1.15306449e+00 -2.11082965e-01 4.82486665e-01 -2.95515358e-01 1.00642371e+00 -8.04557621e-01 -4.92150456e-01 -8.95698741e-03 -5.36382377e-01 -7.40076840e-01 1.26731068e-01 5.89790381e-02 8.28621760e-02 -8.03881884e-02 9.94172156e-01 6.03915095e-01 2.57586449e-01 4.82281923e-01 -1.69189584e+00 -2.62669712e-01 6.25914454e-01 -6.62576973e-01 1.71055424e-03 3.62473696e-01 -2.95064580e-02 1.12465036e+00 -8.13795924e-01 1.50262982e-01 1.46187341e+00 4.14933592e-01 7.70592332e-01 -8.49941790e-01 -1.00953507e+00 6.13174736e-01 3.91161591e-01 -9.04519260e-01 -4.09433126e-01 1.28035748e+00 -1.30631953e-01 5.79400420e-01 3.98021340e-01 5.26244640e-01 1.19855094e+00 -2.36493513e-01 1.18807817e+00 1.30430889e+00 -3.94760638e-01 3.11173797e-01 3.82619888e-01 3.58487457e-01 4.60920513e-01 -4.02224422e-01 -4.82925028e-01 -7.55763233e-01 -1.61774442e-01 3.86700034e-02 3.54888290e-02 -3.98599565e-01 -2.33607352e-01 -6.55615449e-01 8.54226768e-01 1.08931944e-01 4.09670100e-02 -3.15713789e-03 -7.34808445e-02 8.08753371e-01 6.19129658e-01 7.19200850e-01 2.27308184e-01 -5.19681633e-01 -5.11715174e-01 -2.41707489e-01 1.55398801e-01 8.70242715e-01 6.60732746e-01 4.20735925e-01 -2.36194000e-01 -2.67151743e-02 1.32153273e+00 9.07918364e-02 1.89231142e-01 4.65702146e-01 -8.57731283e-01 5.41983724e-01 6.13885641e-01 -2.39505127e-01 -1.09799004e+00 -4.62185323e-01 -3.38898093e-01 -9.25590634e-01 -2.13087738e-01 2.20788032e-01 -3.50304395e-01 -1.68651640e-01 1.95082700e+00 5.05987883e-01 2.08605543e-01 2.39531577e-01 9.60672379e-01 8.92314434e-01 7.36960232e-01 1.06164269e-01 -4.25882906e-01 1.08321381e+00 -1.20316160e+00 -8.68799508e-01 -7.63464794e-02 7.74759471e-01 -6.43757343e-01 1.56119573e+00 5.39505064e-01 -7.80809760e-01 -2.48696193e-01 -9.07086313e-01 -6.98022991e-02 -7.12129846e-02 -7.76255690e-03 7.43776441e-01 5.81480622e-01 -4.52117026e-01 2.16661990e-01 -4.80025083e-01 1.24260284e-01 2.33721390e-01 1.53374359e-01 9.29832738e-03 -1.85856130e-02 -1.30510056e+00 5.14219344e-01 1.22866482e-01 2.23511994e-01 -4.16608363e-01 -6.78744495e-01 -7.51099408e-01 1.41163751e-01 4.76550370e-01 2.81444304e-02 1.38780534e+00 -1.28391242e+00 -1.98838294e+00 7.50278115e-01 -3.26661170e-02 -1.71739787e-01 5.94876170e-01 -2.76466280e-01 -3.38980496e-01 -4.55333032e-02 -3.63892943e-01 4.31535602e-01 4.95152444e-01 -1.07683337e+00 -5.29400229e-01 -2.56004214e-01 3.01963955e-01 4.90266770e-01 -6.35407209e-01 1.38459116e-01 -3.41338575e-01 -5.85275173e-01 -5.18860715e-03 -8.75149131e-01 2.58769542e-02 -1.56791478e-01 -2.12232113e-01 -6.86526775e-01 7.90389717e-01 -3.13921690e-01 1.19539356e+00 -2.29647756e+00 1.41263098e-01 7.37352967e-02 1.70888126e-01 8.91630724e-02 -2.30603099e-01 2.17739224e-01 -6.11114912e-02 -7.99047649e-02 -2.16759339e-01 -3.37398082e-01 3.42103541e-01 -4.26160209e-02 -2.88622141e-01 3.17135692e-01 1.76726416e-01 4.85742241e-01 -8.30568671e-01 -4.19262469e-01 -5.59134930e-02 2.42387623e-01 -8.06029499e-01 4.47923273e-01 -2.68108934e-01 3.01813006e-01 -3.68622780e-01 4.46083188e-01 5.66614211e-01 -1.86649814e-01 4.75601822e-01 -9.42665711e-02 1.42269775e-01 5.59756696e-01 -1.02932096e+00 1.59500134e+00 -7.54037559e-01 5.41301489e-01 1.59060568e-01 -1.44423127e+00 1.14316201e+00 2.65562236e-01 3.83382350e-01 -8.19263577e-01 1.93343133e-01 5.51306494e-02 -2.23608455e-03 -4.15844560e-01 4.36837375e-01 -2.24109158e-01 -5.39042294e-01 4.59253550e-01 -8.46918821e-02 4.32980619e-03 9.21915751e-03 1.64318323e-01 6.66808248e-01 -1.83332160e-01 1.92830130e-01 -2.45468616e-01 6.11261547e-01 -2.05887675e-01 1.16408002e+00 3.25510114e-01 -6.20089471e-01 2.12042704e-01 1.03742123e+00 -2.83534080e-01 -4.76218820e-01 -5.36322951e-01 -5.12154102e-02 1.38740218e+00 2.92719692e-01 -3.69562894e-01 -5.87431312e-01 -7.99703300e-01 -1.97147742e-01 4.92466569e-01 -3.68137509e-01 -3.27388227e-01 -4.40570801e-01 -8.98148119e-01 4.57805455e-01 2.64495254e-01 5.00073075e-01 -1.02595305e+00 -3.03781182e-01 4.62733619e-02 -5.40437222e-01 -1.11405075e+00 -4.77384448e-01 1.55204371e-01 -3.33351731e-01 -9.94578660e-01 -3.42651516e-01 -8.65039527e-01 3.56455833e-01 2.74312735e-01 1.24414599e+00 1.28631487e-01 -7.49150217e-02 2.23879769e-01 -4.67108846e-01 -5.19711137e-01 -2.24739939e-01 1.87048092e-01 -9.32244491e-03 1.21363916e-01 5.96783757e-01 -4.90467250e-01 -5.97071469e-01 2.43469238e-01 -5.81138492e-01 2.25051180e-01 5.37834428e-02 9.42112923e-01 3.12007397e-01 1.36882439e-01 1.01635015e+00 -1.03745139e+00 8.62628639e-01 -5.72266877e-01 -4.08015430e-01 2.09586903e-01 -3.42853189e-01 -1.98478386e-01 7.48300433e-01 -6.38976753e-01 -1.14750957e+00 -3.32242399e-01 -1.43270269e-01 -2.80954033e-01 -6.47613686e-03 4.50367868e-01 -4.53803658e-01 2.99684763e-01 1.70995608e-01 -7.01132864e-02 -4.73249964e-02 -2.28326023e-01 8.74904916e-02 9.66394365e-01 1.69775859e-01 -9.29823518e-01 1.79683357e-01 1.76767007e-01 -5.51806509e-01 -7.30023682e-01 -1.16049480e+00 -4.86075133e-01 1.06391055e-03 -4.07806545e-01 4.46482718e-01 -1.00336730e+00 -1.10793507e+00 5.28811753e-01 -9.48286951e-01 -6.07478082e-01 4.60677668e-02 4.05269057e-01 -4.19436574e-01 1.54937416e-01 -8.58156621e-01 -9.46907759e-01 -4.02205795e-01 -1.09198439e+00 6.53802156e-01 4.76298183e-01 -1.83939144e-01 -8.83974850e-01 1.15984738e-01 6.17757142e-01 2.55666137e-01 1.69666469e-01 8.88469040e-01 -6.93488657e-01 1.88810267e-02 1.13219820e-01 -3.76692936e-02 5.00296831e-01 3.35240327e-02 -8.83010551e-02 -1.05505061e+00 -1.79687232e-01 1.89866230e-01 -9.02804017e-01 6.77316546e-01 -8.85769278e-02 1.56563747e+00 -2.96290785e-01 2.26398915e-01 3.96637857e-01 1.09585309e+00 2.22933844e-01 4.50229734e-01 1.95188135e-01 5.89286745e-01 1.03071833e+00 8.63141119e-01 6.12779677e-01 6.88715041e-01 6.44390404e-01 1.19618222e-01 -1.82495341e-02 3.98034722e-01 -4.11905535e-02 4.49104339e-01 1.36632347e+00 3.80156219e-01 -2.06660122e-01 -8.25702190e-01 3.70775461e-01 -1.90182161e+00 -7.62007236e-01 1.45572141e-01 1.88284755e+00 1.35384679e+00 2.32990086e-01 6.71849027e-02 1.60082161e-01 7.55352318e-01 2.94320375e-01 -6.37449861e-01 -7.10952520e-01 -1.15252875e-01 3.73447867e-04 -1.10500276e-01 5.44863760e-01 -1.03718245e+00 8.56257200e-01 4.70919323e+00 1.02073634e+00 -1.29073715e+00 8.91979486e-02 1.18507242e+00 -4.18858021e-01 -3.59516084e-01 -3.49419266e-01 -6.22267663e-01 6.46902800e-01 7.86165833e-01 -2.02555984e-01 3.97321939e-01 9.18623865e-01 2.51335561e-01 1.43853292e-01 -8.85778964e-01 1.23064792e+00 1.55647144e-01 -9.30944979e-01 -3.96914124e-01 -4.19367790e-01 5.60050607e-01 -2.88489938e-01 5.87585531e-02 8.08463395e-01 1.31230831e-01 -8.41521025e-01 4.73010540e-01 1.52146384e-01 3.90927076e-01 -1.11443138e+00 7.43755937e-01 2.70085931e-01 -8.28412652e-01 -2.63209436e-02 -2.82772779e-01 -1.69195846e-01 -2.98425287e-01 6.34304047e-01 -3.15165550e-01 2.27744818e-01 7.09215701e-01 6.08572721e-01 -1.47176638e-01 5.00981271e-01 -1.26890346e-01 6.94347680e-01 -1.77923724e-01 -3.26316684e-01 1.77194774e-01 -2.30830818e-01 2.05361933e-01 1.36025345e+00 -1.87198132e-01 2.62245178e-01 3.87381792e-01 5.00887215e-01 -4.63557631e-01 4.97135490e-01 -1.12887919e-01 1.27131775e-01 7.73483992e-01 1.38409019e+00 -3.62429410e-01 -2.17070118e-01 -3.90861660e-01 7.96872020e-01 6.60715878e-01 1.56929716e-01 -9.47996318e-01 -4.20873672e-01 9.26815808e-01 -4.54493761e-01 -1.05517112e-01 2.05992609e-01 -1.81591645e-01 -1.25602412e+00 3.16455930e-01 -1.42691588e+00 3.34328175e-01 -2.57314831e-01 -1.55680120e+00 4.09920782e-01 -2.59861708e-01 -1.12898672e+00 1.67405792e-02 -4.46304947e-01 -7.68646419e-01 4.08324540e-01 -1.63594413e+00 -5.80473840e-01 -5.44242680e-01 5.54393649e-01 7.40927219e-01 -3.44926082e-02 7.12595761e-01 3.44489455e-01 -1.00886071e+00 9.32607949e-01 6.96391193e-03 1.86273217e-01 1.05296087e+00 -1.26269841e+00 -2.03620210e-01 2.35349819e-01 -3.72649878e-01 2.12114066e-01 5.92984259e-01 -2.30940264e-02 -1.22999084e+00 -9.37635005e-01 7.02175140e-01 2.14695498e-01 6.24686658e-01 -6.47617877e-01 -1.01624680e+00 3.40201050e-01 2.33844548e-01 -8.12377110e-02 1.03896463e+00 4.95299816e-01 -6.82690501e-01 -3.90459806e-01 -1.16488802e+00 6.11158609e-01 7.18914032e-01 -5.25556028e-01 -3.37629259e-01 4.10361260e-01 8.30049276e-01 -5.58099806e-01 -7.44223952e-01 3.52290541e-01 5.87906957e-01 -9.44974065e-01 5.67839861e-01 -6.22403204e-01 7.57587850e-01 1.39682025e-01 -1.79184452e-01 -1.40792191e+00 1.82742819e-01 -5.94101429e-01 -4.28622402e-02 1.43706739e+00 3.16376895e-01 -6.41663849e-01 7.62174010e-01 6.88666582e-01 -5.50775006e-02 -1.09274244e+00 -6.45533025e-01 -5.12078643e-01 2.65408158e-01 -3.73741120e-01 5.49186289e-01 1.34698725e+00 4.43051219e-01 6.27764821e-01 -5.05270302e-01 -6.12366982e-02 5.48470557e-01 3.50155085e-01 7.93855011e-01 -1.09844232e+00 -4.79852051e-01 -5.07545829e-01 -2.08395831e-02 -9.30708766e-01 6.79608226e-01 -5.89558780e-01 1.51507348e-01 -8.44040155e-01 3.89479637e-01 -6.83163702e-01 -4.99573916e-01 3.80062789e-01 -4.76933300e-01 3.36364545e-02 1.99491635e-01 -2.27415618e-02 -1.08838928e+00 1.00867891e+00 9.65364754e-01 -1.85193256e-01 -2.77200729e-01 -1.87330768e-01 -8.21751356e-01 7.63183773e-01 1.40194917e+00 -4.13182080e-01 -5.74402094e-01 -1.62920132e-01 1.78539649e-01 1.32961497e-01 3.19189839e-02 -5.74388146e-01 9.17970017e-02 -3.81128788e-01 -9.08888206e-02 -4.01130080e-01 3.49910080e-01 -4.19809282e-01 -5.06545246e-01 2.50334501e-01 -8.07061911e-01 -4.95107509e-02 1.13715939e-01 3.89131039e-01 -4.48818892e-01 -6.18964657e-02 9.13472176e-01 1.58304527e-01 -4.09896344e-01 4.82248440e-02 -1.44611180e-01 5.95971465e-01 7.65660763e-01 3.62099707e-01 -4.70542908e-01 -5.52679658e-01 -3.22051764e-01 6.83667123e-01 1.34944871e-01 5.56726515e-01 4.87360805e-01 -1.27422452e+00 -7.16204226e-01 -5.48434444e-02 2.62812585e-01 5.13019376e-02 4.70380813e-01 8.00119817e-01 -2.71321148e-01 1.83513269e-01 -3.74568217e-02 -4.45252806e-01 -1.53329515e+00 2.63625145e-01 4.74103630e-01 -3.60917270e-01 -4.28053945e-01 8.10092568e-01 3.52696896e-01 -9.65068817e-01 7.00990677e-01 -6.35093870e-03 -1.62300736e-01 1.02455452e-01 6.00358903e-01 1.79058343e-01 3.79770715e-03 -1.79268211e-01 -2.82673776e-01 1.84005480e-02 -3.51836205e-01 6.79660141e-02 1.28755391e+00 -1.12443335e-01 -1.21905029e-01 6.86846793e-01 1.32261944e+00 -2.47633364e-02 -1.21179235e+00 -3.37968260e-01 -2.85974853e-02 -2.09689409e-01 4.95138988e-02 -6.65594935e-01 -1.16480803e+00 8.06512475e-01 5.96720040e-01 9.37036946e-02 1.24721944e+00 -1.82501644e-01 8.54809582e-01 4.42108005e-01 5.65574840e-02 -1.39201939e+00 4.28399593e-01 6.16757751e-01 7.58447826e-01 -1.44967818e+00 -2.61947125e-01 -5.19527197e-01 -9.65806305e-01 9.06528950e-01 8.73204648e-01 -5.41170351e-02 5.44267416e-01 2.61001676e-01 4.54610944e-01 -1.02621026e-01 -1.17695558e+00 1.99700505e-01 -1.10903636e-01 1.23327933e-01 7.65995204e-01 3.25300753e-01 -5.89008927e-01 8.41869354e-01 -3.10773522e-01 -4.06761289e-01 4.70406175e-01 7.62125194e-01 -1.43215120e-01 -1.10173619e+00 -4.49770642e-03 1.03642039e-01 -5.22920609e-01 -1.55844660e-02 -4.60679114e-01 6.11473918e-01 -1.42439649e-01 1.11091101e+00 1.65217519e-01 -3.97500932e-01 2.87141383e-01 6.16904981e-02 2.85457462e-01 -3.56522858e-01 -7.49416590e-01 3.07820048e-02 2.07912549e-01 -5.44245899e-01 -5.21458030e-01 -4.82738465e-01 -1.33101749e+00 -4.57182348e-01 -2.15287358e-01 5.56189179e-01 5.00390947e-01 8.28474402e-01 2.83938378e-01 5.41945219e-01 1.16605937e+00 -2.48136699e-01 -6.01181209e-01 -1.01850951e+00 -5.34533203e-01 6.34400368e-01 1.11203462e-01 -5.74210763e-01 -5.72442055e-01 -3.56851459e-01]
[13.037490844726562, 6.081747055053711]
4543cbcd-6a34-41b6-962f-1a1636a43d6b
capturing-emerging-complexity-in-lenia
2305.09378
null
https://arxiv.org/abs/2305.09378v2
https://arxiv.org/pdf/2305.09378v2.pdf
Capturing Emerging Complexity in Lenia
This research project investigates Lenia, an artificial life platform that simulates ecosystems of digital creatures. Lenia's ecosystem consists of simple, artificial organisms that can move, consume, grow, and reproduce. The platform is important as a tool for studying artificial life and evolution, as it provides a scalable and flexible environment for creating a diverse range of organisms with varying abilities and behaviors. Measuring complexity in Lenia is a key aspect of the study, which identifies the metrics for measuring long-term complex emerging behavior of rules, with the aim of evolving better Lenia behaviors which are yet not discovered. The Genetic Algorithm uses neighborhoods or kernels as genotype while keeping the rest of the parameters of Lenia as fixed, for example growth function, to produce different behaviors respective to the population and then measures fitness value to decide the complexity of the resulting behavior. First, we use Variation over Time as a fitness function where higher variance between the frames are rewarded. Second, we use Auto-encoder based fitness where variation of the list of reconstruction loss for the frames is rewarded. Third, we perform combined fitness where higher variation of the pixel density of reconstructed frames is rewarded. All three experiments are tweaked with pixel alive threshold and frames used. Finally, after performing nine experiments of each fitness for 500 generations, we pick configurations from all experiments such that there is a scope of further evolution, and run it for 2500 generations. Results show that the kernel's center of mass increases with a specific set of pixels and together with borders the kernel try to achieve a Gaussian distribution.
['Stefano Nichele', 'Aarati Shrestha', 'Sanyam Jain']
2023-05-16
null
null
null
null
['artificial-life']
['miscellaneous']
[ 3.32172140e-02 -3.64219218e-01 4.36517328e-01 3.51915330e-01 5.77438712e-01 -6.52706504e-01 7.05064654e-01 4.03967425e-02 -5.83434045e-01 8.32005978e-01 -1.96959808e-01 -1.49764195e-02 -3.08034271e-01 -1.07629395e+00 -6.38000309e-01 -1.08486187e+00 -5.50221980e-01 1.45176634e-01 5.56653500e-01 -3.26566398e-01 2.78620243e-01 5.34378409e-01 -2.10702467e+00 -7.54363015e-02 8.55908096e-01 5.35451949e-01 3.13345969e-01 1.12692976e+00 1.73428968e-01 4.19611067e-01 -9.23644245e-01 -2.15577558e-01 5.25059581e-01 -8.60859990e-01 -2.60978401e-01 1.51096955e-01 -1.33960739e-01 1.87123224e-01 1.30285397e-01 9.29420114e-01 5.23172796e-01 1.11197382e-01 5.98188400e-01 -1.15335310e+00 -5.32832384e-01 5.20490408e-01 -1.93411395e-01 3.71604353e-01 1.47680104e-01 7.59519517e-01 3.67192924e-01 -1.75749749e-01 7.28944421e-01 1.25889683e+00 5.14053226e-01 5.06893218e-01 -1.17018759e+00 -3.38806033e-01 -2.41962835e-01 -6.60033971e-02 -1.20877743e+00 -2.98585624e-01 2.49207035e-01 -2.61834383e-01 8.64039421e-01 5.76202571e-01 1.34031415e+00 7.42307067e-01 7.18979418e-01 3.18056941e-02 9.12937880e-01 -5.42197406e-01 5.85234523e-01 1.70065090e-01 -4.94162530e-01 6.86123312e-01 6.24281704e-01 2.48378813e-01 -1.38380364e-01 -1.11038089e-01 8.25819492e-01 -2.91109413e-01 -2.52221882e-01 -1.19574159e-01 -1.29231715e+00 3.48006755e-01 1.62851378e-01 6.80460632e-01 -5.46589136e-01 4.91183966e-01 -7.95634836e-02 5.13741434e-01 -1.10565238e-01 9.12980139e-01 -4.98377711e-01 -4.45067585e-01 -5.45325398e-01 3.01066458e-01 8.19754481e-01 4.01441455e-01 6.79319441e-01 1.09283797e-01 -3.06193143e-01 7.68552244e-01 6.53367639e-02 4.95992690e-01 9.05213594e-01 -1.13672030e+00 -3.07500303e-01 9.65068638e-01 -1.20951265e-01 -1.00496328e+00 -2.67517447e-01 -5.60301423e-01 -6.43496394e-01 6.91377282e-01 4.29161191e-01 -4.26612109e-01 -9.22996283e-01 1.62794828e+00 2.53086239e-01 3.92371237e-01 2.22595260e-01 5.77632189e-01 1.69133872e-01 7.48612344e-01 -7.14725405e-02 -4.95967209e-01 1.10381365e+00 -6.73719049e-01 -4.30685669e-01 1.43932179e-01 2.69469529e-01 -4.79256094e-01 1.11572945e+00 3.14590842e-01 -1.22523808e+00 -4.22785491e-01 -1.16871893e+00 8.11866164e-01 -7.24961221e-01 -1.15033828e-01 4.04742301e-01 8.87972772e-01 -1.20444965e+00 8.42299759e-01 -7.70182312e-01 -6.70750022e-01 1.37966901e-01 1.88886806e-01 1.24505594e-01 3.75570863e-01 -9.24983025e-01 7.98540413e-01 2.53004223e-01 -3.86389315e-01 -8.70013893e-01 -5.35070300e-01 -3.76293123e-01 1.46585748e-01 -1.09070092e-01 -9.67587590e-01 4.63059634e-01 -1.31740630e+00 -1.57030213e+00 6.31580949e-01 2.65148818e-01 -6.09398127e-01 7.04432249e-01 4.45928961e-01 -4.55052555e-01 -3.59753162e-01 -1.70255482e-01 7.97538459e-01 7.74632454e-01 -1.29313040e+00 -5.76118588e-01 -1.78529203e-01 -8.68225470e-02 3.50016505e-01 -4.96732295e-01 -1.70157671e-01 -4.37716633e-01 -5.53394318e-01 -3.16548020e-01 -1.13389122e+00 -2.73047000e-01 -3.91670577e-02 1.16555884e-01 2.34302849e-01 7.93565631e-01 -1.70512319e-01 1.35501552e+00 -2.13710618e+00 2.50274926e-01 3.07769984e-01 -2.22474769e-01 1.31439999e-01 -2.89116681e-01 2.28180736e-01 3.68907839e-01 4.64996427e-01 -4.54537809e-01 2.10949078e-01 -3.08768719e-01 2.69001096e-01 4.73933339e-01 1.81588724e-01 1.53614014e-01 4.90342200e-01 -8.04192066e-01 -2.71744549e-01 3.38084772e-02 5.72311997e-01 -5.25176406e-01 -9.20948535e-02 -3.76221925e-01 2.91063279e-01 -2.00349897e-01 3.98233414e-01 4.30832386e-01 3.83482664e-03 -4.70294207e-02 2.57435918e-01 -5.26913524e-01 -5.49531996e-01 -1.27643561e+00 9.49217975e-01 -2.90751994e-01 5.56947470e-01 -2.25968763e-01 -4.37953830e-01 1.01673508e+00 -7.79916793e-02 4.42071736e-01 -4.25575376e-01 4.21183109e-01 1.41052812e-01 4.79088187e-01 -5.02797842e-01 2.15213522e-01 3.03010255e-01 2.93751687e-01 4.08854693e-01 -2.02105284e-01 -3.96705180e-01 7.67957687e-01 -2.31696308e-01 1.46981835e+00 2.67756395e-02 7.26732984e-02 -4.09070492e-01 7.86267042e-01 -5.88908903e-02 4.30061609e-01 6.92168057e-01 -3.25658172e-01 4.15180027e-01 4.24568981e-01 -4.68788117e-01 -1.40251160e+00 -9.05490816e-01 -9.18272957e-02 8.45114470e-01 3.70573312e-01 3.60728445e-04 -9.87744212e-01 -1.03156060e-01 -5.85810952e-02 6.93905115e-01 -7.96306312e-01 -2.77532339e-01 -5.25435865e-01 -1.05186665e+00 5.15814900e-01 -2.91029692e-01 8.13441515e-01 -1.61423898e+00 -1.32585025e+00 9.73340720e-02 4.29214716e-01 -4.80942845e-01 -9.07105431e-02 1.25577182e-01 -7.92793393e-01 -9.03544188e-01 -7.43988633e-01 -5.83849430e-01 6.64323866e-01 -1.27506167e-01 1.03003299e+00 6.49512768e-01 -7.28420854e-01 2.04178005e-01 -3.05756867e-01 -3.94335568e-01 -8.00452113e-01 -1.60013944e-01 -2.22190972e-02 -2.92588294e-01 -2.70941406e-01 -7.42606342e-01 -5.96573174e-01 3.60347748e-01 -9.63057339e-01 -2.83977747e-01 4.94436890e-01 6.28171802e-01 5.78110278e-01 6.54098392e-01 3.94403934e-01 -3.40151966e-01 1.08197391e+00 -4.62499559e-01 -7.07829654e-01 5.10547400e-01 -5.35688758e-01 1.42130166e-01 7.15170801e-01 -7.98820376e-01 -6.80863261e-01 -1.73233956e-01 4.18930680e-01 -1.23238228e-01 -3.21066193e-02 5.64439818e-02 -1.85916666e-02 -9.29692835e-02 8.78734767e-01 4.15139616e-01 2.84430802e-01 7.58688822e-02 1.17668152e-01 2.85416216e-01 3.17906171e-01 -2.55693674e-01 5.54094493e-01 3.82301658e-01 5.93096167e-02 -8.38794649e-01 3.14203233e-01 1.61644295e-01 -1.59260064e-01 -7.31043279e-01 8.32931399e-01 -1.45864233e-01 -6.62662268e-01 6.78483725e-01 -5.83814919e-01 -4.38459605e-01 -6.12286389e-01 2.82667756e-01 -5.29106498e-01 -7.74177685e-02 -2.15616137e-01 -1.00101316e+00 -3.42940003e-01 -9.91178989e-01 3.97429138e-01 8.88136029e-01 -9.40950811e-02 -9.40791905e-01 3.83662790e-01 -2.97279656e-01 6.80436254e-01 4.46827233e-01 7.50804365e-01 -1.62809044e-01 -4.83758241e-01 6.75922483e-02 2.86003172e-01 2.33089238e-01 3.37666452e-01 7.09541023e-01 -3.49017799e-01 -2.84924507e-01 -1.73402742e-01 3.24170113e-01 6.95064306e-01 4.56377149e-01 7.64207900e-01 -2.92296886e-01 -3.14518869e-01 5.45560181e-01 1.41138220e+00 7.60114908e-01 9.44007814e-01 9.47589993e-01 -8.62259492e-02 3.69027793e-01 2.10759193e-01 6.18428826e-01 -3.47443894e-02 3.30915481e-01 5.85492313e-01 1.15330301e-01 -2.21860364e-01 3.69042575e-01 5.52087665e-01 4.18441325e-01 -4.29413825e-01 -5.82848966e-01 -7.74399221e-01 5.03368080e-01 -1.45700300e+00 -1.17327666e+00 2.07831100e-01 2.46276188e+00 5.25259733e-01 2.06359088e-01 3.46020848e-01 2.56865621e-01 9.02784944e-01 -1.73272863e-01 -7.58084774e-01 -5.93957007e-01 -4.92165893e-01 -2.15661284e-02 6.45315230e-01 2.66652077e-01 -6.65882587e-01 7.48925269e-01 6.82631969e+00 3.54866236e-01 -1.27850103e+00 -3.02668780e-01 7.31239855e-01 -1.99232191e-01 -3.18206459e-01 -5.53133450e-02 -4.87834424e-01 9.07366633e-01 9.39033449e-01 -3.66904974e-01 9.65428412e-01 4.13967371e-01 4.37407225e-01 -4.04766381e-01 -4.25802380e-01 5.72078943e-01 -4.08262052e-02 -1.18057096e+00 8.84199813e-02 1.72467023e-01 1.01620460e+00 -1.22656889e-01 1.07297458e-01 2.22308375e-02 5.15859246e-01 -9.08667088e-01 7.82051504e-01 8.82502854e-01 2.05779672e-01 -8.55930984e-01 6.31237686e-01 3.99529338e-01 -8.75620425e-01 -5.38671076e-01 -4.01967376e-01 -5.70778996e-02 -6.11137748e-02 4.16617751e-01 -5.55954337e-01 2.44813785e-02 9.42658663e-01 2.32684955e-01 -8.52510512e-01 1.52882326e+00 2.13760927e-01 4.56920892e-01 -6.08508825e-01 -5.93795359e-01 6.68302774e-02 -4.46291298e-01 9.04314220e-01 9.95571971e-01 8.57829869e-01 -2.04548687e-01 -3.12978268e-01 1.01808941e+00 1.62826404e-01 9.42823738e-02 -4.77446258e-01 -3.83163318e-02 7.60154605e-01 1.00354552e+00 -1.11785698e+00 -1.56114519e-01 2.17614621e-01 9.31954503e-01 -2.87856609e-01 1.79545492e-01 -8.34373832e-01 -2.77260870e-01 5.60598731e-01 2.23819330e-01 3.26291949e-01 -1.24827564e-01 -2.71564841e-01 -6.10671103e-01 -3.24524879e-01 -9.19745564e-01 1.61438450e-01 -7.46217370e-01 -8.41486275e-01 6.61075175e-01 -2.22320586e-01 -9.41103816e-01 -1.21940538e-01 -2.22870499e-01 -8.69545996e-01 5.42141259e-01 -9.37265038e-01 -5.98032057e-01 -5.70782900e-01 3.44457477e-01 4.56574589e-01 -5.06494999e-01 5.82659543e-01 -9.59830657e-02 -6.75599992e-01 3.22152793e-01 2.80545771e-01 -4.70758647e-01 2.16364637e-01 -9.34986293e-01 3.05137306e-01 8.22631657e-01 -1.89279586e-01 4.27827567e-01 9.27566469e-01 -7.84720659e-01 -1.15799737e+00 -9.25355554e-01 3.02571625e-01 -9.19853151e-02 3.40969533e-01 6.02532178e-02 -6.49582028e-01 2.16331203e-02 2.90432215e-01 -1.66179106e-01 2.13859603e-01 -6.39287531e-01 3.99456888e-01 -1.07135467e-01 -1.56678855e+00 9.83132005e-01 9.71687078e-01 4.42764610e-01 1.19946979e-01 -8.32245946e-02 6.36568129e-01 1.28285155e-01 -7.58117855e-01 2.66693980e-01 7.79938281e-01 -1.30891931e+00 7.99669385e-01 -6.08671457e-02 2.33116657e-01 -6.61006510e-01 1.19072787e-01 -1.35857296e+00 -5.19153774e-01 -7.11037576e-01 1.67124808e-01 1.32487798e+00 6.94721222e-01 -7.43189573e-01 5.78297615e-01 4.95432466e-02 2.11993560e-01 -7.01578557e-01 -7.07360327e-01 -1.02250314e+00 -1.27953058e-02 2.31768176e-01 8.93665135e-01 6.46998465e-01 -3.86389315e-01 -1.61300302e-01 -1.27379186e-02 -1.31532088e-01 4.59465355e-01 -3.88488442e-01 6.81124628e-01 -1.10428905e+00 -2.80763596e-01 -8.61965418e-01 -6.18595302e-01 -1.47555262e-01 -4.95346993e-01 -3.66617948e-01 -4.93376218e-02 -1.35659933e+00 7.22446442e-02 -5.13554633e-01 -1.27658933e-01 1.12866633e-01 -7.62437936e-03 1.62301585e-01 2.93817431e-01 1.33382797e-01 -1.54677927e-01 3.43866706e-01 1.28157806e+00 4.88942713e-02 -6.20693445e-01 -1.12539336e-01 -2.59463638e-01 5.80055475e-01 8.56541216e-01 -3.32567006e-01 -4.90223706e-01 -1.39378428e-01 2.62867361e-01 -2.33498514e-01 1.86015531e-01 -1.71284330e+00 2.56216936e-02 -3.53455693e-01 5.32978058e-01 -1.22670949e-01 1.16766103e-01 -8.77673507e-01 7.19132960e-01 1.00951159e+00 -8.43880400e-02 3.55604202e-01 1.38755366e-01 3.35149914e-01 3.57504278e-01 -4.50870007e-01 1.09762752e+00 -3.55685413e-01 -4.24452901e-01 -1.06125347e-01 -7.95784950e-01 -1.67397797e-01 1.54504120e+00 -8.11987400e-01 -1.34647921e-01 -2.45362014e-01 -4.93982077e-01 1.66931823e-01 1.16357911e+00 4.19739544e-01 2.21642509e-01 -9.79044676e-01 -7.86553085e-01 2.92251378e-01 -2.60194808e-01 -5.13136208e-01 -1.54734850e-01 2.84942925e-01 -9.54630077e-01 -2.70522416e-01 -6.19663835e-01 -5.45134902e-01 -1.33118534e+00 4.65286642e-01 8.10730100e-01 -9.66670290e-02 -4.11235601e-01 7.60074615e-01 -3.43356520e-01 -1.24822050e-01 3.88318188e-02 -2.56413341e-01 -4.99657631e-01 -6.47029653e-02 4.50164407e-01 6.18088007e-01 -2.27623999e-01 -4.43504125e-01 -3.13505590e-01 6.98790252e-01 5.74589014e-01 -2.90775210e-01 1.44580555e+00 -2.06953287e-01 -2.97533035e-01 4.15404826e-01 6.30378187e-01 2.88826935e-02 -1.33936286e+00 6.86040580e-01 -2.23308846e-01 -4.25632745e-01 -1.85821027e-01 -9.83469248e-01 -1.01630437e+00 5.80398552e-02 1.10564947e+00 5.98017633e-01 1.35551405e+00 -2.45069012e-01 3.09901416e-01 1.54706851e-01 2.13592529e-01 -9.26691353e-01 1.95368245e-01 5.32979548e-01 7.86476433e-01 -6.00949824e-01 1.33073023e-02 8.70663300e-02 -4.12872612e-01 9.67053115e-01 6.18245900e-01 -2.15605184e-01 4.39635843e-01 5.88938177e-01 -1.49459675e-01 -3.26742493e-02 -1.07707191e+00 -1.21281236e-01 -2.00685292e-01 7.81000614e-01 1.29734814e-01 -2.56902482e-02 -7.81757355e-01 1.10710822e-01 -5.23924708e-01 1.22537985e-02 6.74083233e-01 9.21471894e-01 -8.11820269e-01 -9.67067778e-01 -5.57215691e-01 2.83273518e-01 -2.66030073e-01 7.02386498e-02 -4.77638453e-01 5.88484347e-01 7.13298500e-01 7.52105057e-01 3.29903692e-01 -2.31099427e-01 2.12382942e-01 -3.11521500e-01 4.15193111e-01 1.49838459e-02 -9.94981170e-01 -2.25451276e-01 -1.21606492e-01 -8.87821764e-02 -1.87830374e-01 -9.18693602e-01 -1.29580474e+00 -4.76252854e-01 -3.06072325e-01 1.47772610e-01 8.36218178e-01 4.86380428e-01 3.94795209e-01 6.95436776e-01 6.62556410e-01 -6.45828187e-01 2.19085347e-02 -7.48206377e-01 -2.87229985e-01 3.91144395e-01 -6.10288884e-03 -6.00149274e-01 -6.29502356e-01 3.38769197e-01]
[5.63350248336792, 4.060884475708008]
52c14191-bd8a-4af1-bc4d-62756e02ef3d
pseudo-labels-refinement-with-intra-camera
2304.12634
null
https://arxiv.org/abs/2304.12634v1
https://arxiv.org/pdf/2304.12634v1.pdf
Pseudo Labels Refinement with Intra-camera Similarity for Unsupervised Person Re-identification
Unsupervised person re-identification (Re-ID) aims to retrieve person images across cameras without any identity labels. Most clustering-based methods roughly divide image features into clusters and neglect the feature distribution noise caused by domain shifts among different cameras, leading to inevitable performance degradation. To address this challenge, we propose a novel label refinement framework with clustering intra-camera similarity. Intra-camera feature distribution pays more attention to the appearance of pedestrians and labels are more reliable. We conduct intra-camera training to get local clusters in each camera, respectively, and refine inter-camera clusters with local results. We hence train the Re-ID model with refined reliable pseudo labels in a self-paced way. Extensive experiments demonstrate that the proposed method surpasses state-of-the-art performance.
['Jinjun Wang', 'Sanping Zhou. Qianxin Huang', 'Kangyi Wu', 'Pengna Li']
2023-04-25
null
null
null
null
['person-re-identification', 'unsupervised-person-re-identification']
['computer-vision', 'computer-vision']
[-1.02348343e-01 -5.07050753e-01 6.36970848e-02 -4.91849601e-01 -5.63161731e-01 -6.36974335e-01 5.68220437e-01 7.32265264e-02 -7.39683270e-01 4.74879175e-01 2.14713901e-01 5.18777430e-01 1.06444217e-01 -3.46266657e-01 -3.68368208e-01 -6.91155910e-01 4.55955744e-01 5.06541133e-01 2.07293987e-01 4.88070011e-01 1.00362040e-01 1.71860337e-01 -1.64319265e+00 1.32552430e-01 7.95537651e-01 4.80098903e-01 -5.33473503e-04 3.99995089e-01 2.36987114e-01 6.09276891e-01 -3.80713522e-01 -7.43791878e-01 3.23662817e-01 -3.80564511e-01 -7.71711409e-01 8.60512733e-01 5.65691292e-01 -3.46791208e-01 -3.13755214e-01 1.49950016e+00 3.90699744e-01 3.69046628e-01 6.76887751e-01 -1.17282009e+00 -7.75258064e-01 1.25716716e-01 -9.80844855e-01 1.01667061e-01 3.94444108e-01 1.08022228e-01 5.06374359e-01 -9.86645699e-01 4.70901549e-01 1.31883514e+00 7.42444813e-01 9.52801108e-01 -1.49897361e+00 -8.12658310e-01 5.25748253e-01 5.15755534e-01 -2.02103686e+00 -5.91398418e-01 6.82217419e-01 -4.06281948e-01 2.65975505e-01 2.28807449e-01 3.83866310e-01 9.96872544e-01 -6.85616612e-01 6.33835435e-01 1.18490958e+00 -3.77006114e-01 2.37732735e-02 4.82492507e-01 2.69373596e-01 4.46526229e-01 5.03596127e-01 -1.08362153e-01 -4.28050697e-01 -3.30722257e-02 5.54283023e-01 4.33693826e-01 -2.08647102e-01 -4.20919091e-01 -1.34823000e+00 3.87023658e-01 3.05225998e-01 2.34992027e-01 -8.23323876e-02 -1.14091203e-01 3.48442376e-01 -2.00773473e-03 1.93274200e-01 -8.12490508e-02 -2.51297466e-02 2.03471228e-01 -8.48380446e-01 5.76258712e-02 2.25651652e-01 1.26700282e+00 9.42845047e-01 -5.39658546e-01 -1.81075171e-01 1.10725379e+00 1.81677908e-01 4.98406261e-01 4.65136141e-01 -1.05854774e+00 3.05428922e-01 5.78725696e-01 4.21795070e-01 -1.15016186e+00 -2.69496590e-01 -2.78359443e-01 -1.18836725e+00 -1.80307180e-01 5.49182773e-01 -3.61911394e-02 -6.76819682e-01 1.57431161e+00 4.60340828e-01 5.68979144e-01 -1.28222257e-01 1.10586476e+00 6.66311860e-01 2.51028717e-01 4.09083456e-01 -2.82303154e-01 1.42763102e+00 -1.23395288e+00 -4.90291655e-01 -1.27441794e-01 2.49148697e-01 -6.12194121e-01 5.40790975e-01 3.15482944e-01 -7.95164526e-01 -1.15654302e+00 -6.13501966e-01 2.16638535e-01 -9.72617865e-02 3.72809589e-01 1.23662695e-01 8.55868518e-01 -1.13392425e+00 2.15343028e-01 -3.89918387e-01 -7.64245391e-01 3.66127640e-01 4.29059744e-01 -7.04320788e-01 -3.89490902e-01 -6.93771720e-01 5.32865226e-01 4.98636335e-01 5.35463989e-02 -8.18826318e-01 -4.08029079e-01 -6.31305695e-01 -1.63739741e-01 3.39144647e-01 -7.75274575e-01 8.98585320e-01 -1.13186169e+00 -1.22100616e+00 1.18182635e+00 -5.95869184e-01 -3.93451005e-02 5.06977022e-01 -1.02689564e-01 -6.71750069e-01 1.71710700e-01 4.58589375e-01 8.26568961e-01 1.05398488e+00 -1.69814622e+00 -1.10988963e+00 -4.41228777e-01 -3.99108231e-01 4.21080202e-01 -5.38694024e-01 2.20893696e-01 -1.02478707e+00 -5.44857383e-01 4.45365533e-02 -1.08729005e+00 -2.26159975e-01 -4.41337258e-01 -3.74666750e-01 -4.18207079e-01 4.45352167e-01 -5.90702891e-01 1.07790256e+00 -2.24230623e+00 -3.90349254e-02 2.97918260e-01 4.28768814e-01 1.73035234e-01 -6.30593523e-02 -9.67456177e-02 -1.73982263e-01 -1.82426438e-01 1.00971777e-02 -7.58832097e-01 -1.32347509e-01 -1.02431312e-01 2.51587480e-01 7.47610748e-01 -1.43611923e-01 5.45714378e-01 -9.18470204e-01 -9.40287411e-01 4.24942672e-01 3.43134344e-01 -3.57239217e-01 3.33433956e-01 4.91382688e-01 7.36398101e-01 -2.34528780e-01 6.51045442e-01 1.07066488e+00 -3.20425123e-01 2.24149257e-01 -3.85224074e-01 -8.67012069e-02 -6.05502605e-01 -1.44969630e+00 1.56378722e+00 5.74889407e-02 1.52442634e-01 -5.57743199e-02 -9.30117309e-01 8.28112960e-01 1.05317451e-01 5.01262248e-01 -4.82801259e-01 -9.49456263e-03 -1.96933120e-01 -5.12446404e-01 -2.92500407e-01 4.18328941e-01 1.11518845e-01 -6.23800457e-02 4.11265463e-01 -5.59044890e-02 7.44909585e-01 2.11715549e-01 1.58080354e-01 6.17330551e-01 -6.74641728e-02 6.51036501e-02 -1.75204560e-01 9.68590915e-01 -1.51975960e-01 7.45209992e-01 8.56319129e-01 -7.36714721e-01 9.28417206e-01 -3.73318076e-01 -4.92155850e-01 -1.13472116e+00 -1.05160427e+00 3.14669013e-02 1.16825902e+00 8.13487411e-01 -3.04936796e-01 -1.03280294e+00 -8.43882203e-01 -7.79179931e-02 2.04845577e-01 -5.45445204e-01 4.44096550e-02 -3.63630354e-01 -9.22396064e-01 4.18201536e-01 3.92923504e-01 9.31446373e-01 -6.57364547e-01 2.77881138e-02 9.54109281e-02 -5.53949356e-01 -1.27873302e+00 -9.86437619e-01 -4.18290287e-01 -5.10655046e-01 -1.16738057e+00 -1.13991284e+00 -1.26790643e+00 1.13585269e+00 9.08615291e-01 9.49124157e-01 8.56297165e-02 -7.53537491e-02 6.53891385e-01 -4.13832039e-01 3.64104927e-01 -6.67762011e-02 -3.80225107e-02 4.77038056e-01 6.90743089e-01 9.32654023e-01 -3.39260221e-01 -9.90960479e-01 7.01950312e-01 -4.30663496e-01 -3.81497815e-02 3.76373589e-01 7.25209594e-01 6.40837491e-01 4.75642115e-01 3.77473503e-01 -7.65076697e-01 3.30098689e-01 -2.80313045e-01 -2.28955179e-01 6.10475540e-01 -6.37034893e-01 -2.48482898e-01 4.52832609e-01 -5.17730415e-01 -1.37064207e+00 4.69202429e-01 3.87636572e-01 -3.65560651e-01 -7.91376770e-01 -3.54320407e-01 -4.66165155e-01 -1.32270128e-01 3.54057699e-01 4.29220915e-01 -3.11241865e-01 -5.31792760e-01 4.72725868e-01 1.01237893e+00 1.03207445e+00 -5.79387784e-01 9.69403088e-01 7.60848820e-01 -5.01278877e-01 -4.45921838e-01 -6.83058441e-01 -1.18871748e+00 -1.06326985e+00 -5.20197213e-01 9.52103913e-01 -1.42459846e+00 -8.36155713e-01 6.49148464e-01 -1.06020415e+00 1.79952502e-01 -2.08189636e-02 4.60888535e-01 -1.03993729e-01 8.23684692e-01 -5.48139572e-01 -7.08301902e-01 -2.16336012e-01 -9.19794440e-01 9.98005092e-01 7.02189386e-01 -1.32636830e-01 -6.52321637e-01 3.34984288e-02 4.63227510e-01 -1.26352310e-01 -1.14366271e-01 4.67171744e-02 -3.98670852e-01 -3.90548646e-01 -1.40207574e-01 -7.92591631e-01 2.75028527e-01 3.61930281e-01 -4.39584315e-01 -1.07517409e+00 -4.56939876e-01 -2.37873748e-01 3.83918127e-03 8.69746566e-01 1.28208846e-01 9.90956306e-01 -1.34161502e-01 -6.82260215e-01 5.96235871e-01 1.33674312e+00 -1.04036972e-01 3.71659487e-01 4.41473782e-01 1.13957179e+00 7.79736876e-01 5.62826276e-01 5.47077358e-01 8.75973701e-01 7.24851966e-01 -1.94540471e-01 -1.88369885e-01 -2.27955267e-01 -2.76114076e-01 2.50409842e-01 7.51623571e-01 -2.29036957e-01 7.97223896e-02 -6.28286719e-01 8.03808212e-01 -2.11409736e+00 -1.05830061e+00 -2.21006662e-01 2.35075307e+00 6.95128918e-01 -2.63007075e-01 5.92851043e-01 -9.03952494e-02 1.57988596e+00 -2.50415236e-01 -5.12595356e-01 3.43610853e-01 -1.22841999e-01 -4.84065711e-01 7.40334153e-01 2.75295019e-01 -1.44842267e+00 1.01272607e+00 5.87634277e+00 8.85884941e-01 -3.54282707e-01 3.45053822e-01 9.17011082e-01 1.51305825e-01 2.65724927e-01 -1.20121323e-01 -1.09306455e+00 8.69989395e-01 4.66669172e-01 -1.08527258e-01 6.07614934e-01 8.62751007e-01 4.07142006e-02 -1.07678488e-01 -1.16426754e+00 1.78452742e+00 3.99491012e-01 -6.92077160e-01 -6.00861385e-03 -8.76787603e-02 1.06655526e+00 -4.74380106e-01 -4.63517196e-02 -2.85899993e-02 5.86376429e-01 -7.19640851e-01 7.36818969e-01 7.06808567e-01 8.58430088e-01 -9.32036042e-01 7.47475028e-01 1.88344091e-01 -1.58832049e+00 -1.94727212e-01 -6.33559704e-01 1.02629445e-01 9.05444771e-02 3.35301667e-01 -4.02592570e-01 4.84253228e-01 1.14688289e+00 1.09901202e+00 -1.15903854e+00 1.11942554e+00 1.24794364e-01 2.95885831e-01 -1.34591013e-01 3.37299764e-01 -1.85956791e-01 -2.80924737e-01 1.71898872e-01 1.41817629e+00 1.54609650e-01 1.40218392e-01 4.99863207e-01 4.63890731e-01 5.23921940e-03 -1.14710905e-01 -2.24247932e-01 7.37358987e-01 7.82823622e-01 1.23912501e+00 -9.08282101e-01 -5.75819135e-01 -6.25269175e-01 1.72263074e+00 3.28410059e-01 6.12359047e-01 -7.19941497e-01 -2.70706713e-02 6.31445467e-01 8.38018209e-02 2.27741331e-01 -2.93660909e-02 -2.90560480e-02 -1.41583753e+00 -1.30363673e-01 -7.12457299e-01 7.91437864e-01 -5.76551437e-01 -1.76138616e+00 3.47406298e-01 4.66660187e-02 -1.28004038e+00 5.98676689e-02 -1.82741031e-01 -2.18683675e-01 5.65128207e-01 -1.46309626e+00 -1.39369237e+00 -7.75268257e-01 1.12440372e+00 5.09600341e-01 -2.80077308e-01 6.34083152e-01 6.54956460e-01 -8.73420715e-01 9.51264977e-01 2.88884014e-01 4.78516012e-01 1.24652553e+00 -1.10165775e+00 2.17773050e-01 1.08521760e+00 -7.75024071e-02 6.15918458e-01 3.38084221e-01 -5.53624034e-01 -7.92774558e-01 -1.47923923e+00 8.23082924e-01 -6.44395828e-01 1.51809767e-01 -3.66043329e-01 -6.41494989e-01 5.31933784e-01 5.71477041e-02 8.80620629e-02 6.84820831e-01 1.14736520e-01 -4.32384491e-01 -4.54899132e-01 -1.09276080e+00 4.64401305e-01 1.33129299e+00 -5.40304005e-01 -2.83170104e-01 2.36438870e-01 3.77462238e-01 9.12437737e-02 -6.82965398e-01 9.42031294e-02 4.18714464e-01 -9.39198375e-01 1.20327711e+00 -9.02049765e-02 -3.38722110e-01 -8.83378863e-01 7.09081441e-02 -9.58095431e-01 -8.33172321e-01 -4.30196106e-01 3.44556481e-01 1.89809191e+00 -2.38541439e-01 -4.37650681e-01 8.13896358e-01 1.02959538e+00 4.71400648e-01 4.02126729e-01 -7.25778580e-01 -6.83625042e-01 -3.31178963e-01 -7.64553528e-03 7.15473115e-01 1.10339952e+00 -6.55021369e-02 1.79582462e-01 -6.39864147e-01 5.29125094e-01 1.33577633e+00 -1.38786156e-02 9.31261539e-01 -1.30759108e+00 -1.93663146e-02 -2.50985920e-01 -4.33156788e-01 -1.00329578e+00 2.55961955e-01 -6.05642259e-01 1.91358700e-01 -1.02900338e+00 1.09338570e+00 -6.14386201e-01 -5.06000340e-01 1.94193289e-01 -6.33465946e-01 7.83027768e-01 2.41274521e-01 7.63290584e-01 -1.55905187e+00 3.30127358e-01 7.08585560e-01 -4.04168218e-01 -1.00991443e-01 -2.48123407e-01 -8.08289647e-01 6.79967582e-01 6.63454473e-01 -5.74196041e-01 -9.67747420e-02 -4.26624835e-01 -3.55789453e-01 -5.24814904e-01 5.74666739e-01 -1.42811048e+00 7.09814191e-01 2.01418810e-02 8.37266922e-01 -3.63347232e-01 -5.91073260e-02 -1.05845606e+00 3.30609351e-01 3.31622995e-02 -2.81204402e-01 6.02614880e-02 -2.42107898e-01 8.90539169e-01 -1.93327188e-01 -1.03758737e-01 9.47051823e-01 -3.57983053e-01 -9.51887310e-01 3.57818037e-01 -3.56405795e-01 -1.20411240e-01 1.19511819e+00 -4.36015129e-01 -1.79918595e-02 -2.50118375e-01 -7.97892272e-01 3.97215903e-01 8.54848504e-01 3.16177011e-01 5.15451908e-01 -1.52196693e+00 -8.70482564e-01 1.73531771e-01 3.83628219e-01 -5.59529886e-02 5.96914232e-01 5.42639554e-01 -1.50537074e-01 2.48578787e-01 -9.45122987e-02 -9.06153858e-01 -1.58020020e+00 9.52963710e-01 2.20575571e-01 -8.79486650e-02 -4.70098823e-01 7.79398322e-01 5.28893590e-01 -3.79859269e-01 3.06069076e-01 4.39479172e-01 -4.49031025e-01 1.68606942e-03 8.05794358e-01 5.36904871e-01 -3.91221195e-01 -1.20408177e+00 -6.20680034e-01 1.01119828e+00 -3.26175034e-01 4.72361818e-02 8.23550045e-01 -9.34913278e-01 9.84701235e-03 3.91197689e-02 1.08111238e+00 -2.11983472e-01 -1.47291088e+00 -4.66445923e-01 -4.60379794e-02 -6.63940310e-01 -2.31502205e-01 -4.36734289e-01 -9.88724887e-01 3.90121758e-01 1.10750091e+00 -1.72577709e-01 1.19143271e+00 1.70490239e-02 8.73094559e-01 1.36229515e-01 5.43607354e-01 -1.46736681e+00 2.34009936e-01 -2.75275595e-02 4.55517396e-02 -1.52109241e+00 9.73166078e-02 -4.77891266e-01 -8.66775155e-01 6.27061963e-01 7.43810773e-01 -1.10348739e-01 4.82873529e-01 -1.92534208e-01 6.98678195e-02 2.69036293e-01 -5.36067858e-02 -5.96051395e-01 1.51691079e-01 1.10915542e+00 -1.59320682e-01 1.38397887e-01 1.21664524e-01 7.08927691e-01 3.36279273e-01 -8.02114513e-03 7.92896897e-02 4.15539652e-01 -3.26721251e-01 -1.19656682e+00 -8.77102256e-01 1.53993191e-02 -1.51038885e-01 8.62321854e-02 -3.39263618e-01 4.48016733e-01 5.94337463e-01 1.41662169e+00 1.52870312e-01 -5.64897716e-01 1.38587564e-01 -1.22725904e-01 3.73185366e-01 -3.52628052e-01 -4.75081384e-01 1.87693670e-01 -1.86230987e-01 -3.64232242e-01 -9.47933197e-01 -9.79071438e-01 -8.00886929e-01 -6.23479247e-01 -3.71780217e-01 2.69525468e-01 -1.86417135e-03 6.83049560e-01 4.92908955e-01 8.31288844e-02 8.79011810e-01 -8.40702176e-01 6.32434338e-02 -7.96052933e-01 -6.29878521e-01 1.06233215e+00 1.89002395e-01 -5.37250936e-01 -1.49660736e-01 8.20384979e-01]
[14.792131423950195, 1.0468157529830933]
1decab8c-ce02-43f7-adad-163101d72317
from-node-to-graph-joint-reasoning-on-visual
null
null
https://ieeexplore.ieee.org/document/9706663
https://openaccess.thecvf.com/content/WACV2022/papers/Nie_From_Node_To_Graph_Joint_Reasoning_on_Visual-Semantic_Relational_Graph_WACV_2022_paper.pdf
From Node to Graph: Joint Reasoning on Visual-Semantic Relational Graph for Zero-Shot Detection
Zero-Shot Detection (ZSD), which aims at localizing andrecognizing unseen objects in a complicated scene, usuallyleverages the visual and semantic information of individ-ual objects alone. However, scene understanding of hu-man exceeds recognizing individual objects separately: thecontextual information among multiple objects such as vi-sual relational information (e.g. visually similar objects)and semantic relational information (e.g. co-occurrences)is helpful for understanding of visual scene. In this pa-per, we verify that contextual information plays a more im-portant role in ZSD than in traditional object detection.To make full use of such information, we propose a newend-to-end ZSD methodGRaphAligningNetwork (GRAN)based on graph modeling and reasoning which simultane-ously considers visual and semantic information of multipleobjects instead of individual objects. Specifically, we for-mulate a Visual Relational Graph (VRG) and a SemanticRelational Graph (SRG), where the nodes are the objectsin the image and the semantic representations of classes re-spectively and the edges are the relevance between nodesin each graph. To characterize mutual effect between twomodalities, the two graphs are further merged into a hetero-geneous Visual-Semantic Relational Graph (VSRG), wheremodal translators are designed for the two subgraphs to en-able modal information to transform into a common spacefor communication, and message passing among nodes isenforced to refine their representations. Comprehensive ex-periments on MSCOCO dataset demonstrate the advantageof our method over state-of-the-arts, and qualitative anal-ysis suggests the validity of using contextual information.
['Xilin Chen', 'Ruiping Wang', 'Hui Nie']
2022-02-15
null
null
null
winter-conference-on-applications-of-computer-5
['zero-shot-object-detection']
['computer-vision']
[ 6.58382080e-04 6.65444136e-02 -2.70771474e-01 -2.75271475e-01 -1.98022798e-01 -7.01165617e-01 8.92972350e-01 3.90871823e-01 -7.23466724e-02 4.08143997e-01 2.82132894e-01 -1.80026278e-01 -2.90090829e-01 -1.14464808e+00 -6.25352085e-01 -5.03514409e-01 -6.94682896e-02 2.22116902e-01 3.63863170e-01 -2.21291348e-01 -1.36528119e-01 5.71345747e-01 -1.79972529e+00 4.39022243e-01 7.86009371e-01 7.28541970e-01 4.27006453e-01 3.52755934e-01 -3.58190775e-01 9.12477732e-01 -4.22389388e-01 -2.45967627e-01 -6.61597326e-02 -6.01926565e-01 -7.91060627e-01 5.42105913e-01 2.95838803e-01 9.12243575e-02 -5.35606325e-01 1.30399466e+00 -7.29038790e-02 2.17759088e-01 6.70063436e-01 -1.57681513e+00 -8.06958973e-01 6.64339125e-01 -5.56219816e-01 2.62459099e-01 4.73612875e-01 -1.42196000e-01 1.24376035e+00 -9.54912186e-01 9.14745331e-01 1.77731776e+00 -6.14474379e-02 2.43493900e-01 -1.38014829e+00 -5.94724834e-01 5.86697876e-01 5.28530061e-01 -1.45304906e+00 -1.78928182e-01 8.37757826e-01 -4.78875756e-01 7.86025584e-01 4.83887494e-01 7.79958010e-01 7.08704591e-01 1.34285027e-02 8.90413523e-01 9.59599674e-01 -3.72248173e-01 -1.84611343e-02 2.05861270e-01 3.88900638e-01 7.88231254e-01 3.92488003e-01 3.15566026e-02 -6.49019480e-01 2.75286883e-01 6.98219299e-01 2.63595879e-01 -1.24955893e-01 -3.88154536e-01 -1.26075852e+00 3.77850264e-01 9.31004584e-01 3.76141429e-01 -2.28938356e-01 -1.76134780e-02 1.64816037e-01 3.02211970e-01 8.49093944e-02 1.83861718e-01 1.05844781e-01 3.81710708e-01 -3.70357960e-01 1.49160053e-03 4.59947675e-01 1.09909391e+00 1.07468355e+00 -8.14536512e-02 -9.71308053e-02 6.71185672e-01 4.51192856e-01 5.22770464e-01 1.87715203e-01 -6.93099380e-01 5.14513254e-01 1.18381870e+00 -4.65671360e-01 -1.42080593e+00 -3.73665214e-01 -1.40479431e-01 -8.61921310e-01 1.15488529e-01 1.65352494e-01 2.05664024e-01 -1.14892209e+00 1.80025280e+00 4.64796245e-01 2.51776516e-01 1.64548412e-01 1.08378112e+00 1.48877954e+00 5.52738965e-01 3.14149201e-01 -4.12823223e-02 1.71097291e+00 -6.03209257e-01 -7.16487467e-01 -3.31877500e-01 6.43584490e-01 -4.78311658e-01 9.33677316e-01 1.97444800e-02 -7.79324889e-01 -5.41120708e-01 -9.79498804e-01 -2.10342869e-01 -7.19116926e-01 -3.27527791e-01 5.83505809e-01 3.87868546e-02 -8.34564209e-01 2.05568507e-01 -5.69587946e-01 -6.66128695e-01 4.43596095e-01 9.76830423e-02 -7.09068537e-01 -1.25224292e-01 -1.12053597e+00 6.42778158e-01 8.12386751e-01 1.19518131e-01 -1.17102051e+00 -3.61609578e-01 -1.10142136e+00 7.78887495e-02 8.68757784e-01 -5.58222115e-01 6.42048895e-01 -1.04461813e+00 -7.90101826e-01 1.08689249e+00 -1.16008885e-01 9.29182544e-02 1.68672670e-02 2.75652766e-01 -6.76833212e-01 3.06813151e-01 1.44839764e-01 6.01274312e-01 5.27018487e-01 -1.72687805e+00 -6.06026292e-01 -5.31465471e-01 4.95358974e-01 7.15778172e-01 -2.84158885e-01 6.60466030e-02 -8.67034733e-01 -5.96496999e-01 6.84181929e-01 -6.12109900e-01 1.21945135e-01 8.41093808e-03 -1.09208822e+00 -1.29313961e-01 1.10991001e+00 -4.12014931e-01 1.19812500e+00 -2.32654214e+00 5.43063939e-01 5.10215819e-01 6.62466884e-01 -1.16452478e-01 -3.36860746e-01 4.66593176e-01 -4.06492323e-01 -2.28084512e-02 -3.23030911e-02 -2.68306322e-02 -5.46916574e-02 5.35854518e-01 4.05270047e-02 3.00498754e-01 -1.61666647e-02 1.02356529e+00 -1.17830098e+00 -7.66872406e-01 4.17123497e-01 2.25161895e-01 -1.82874694e-01 2.12205902e-01 -2.15595111e-01 1.65476993e-01 -5.23927510e-01 8.44516754e-01 3.01616311e-01 -5.07906258e-01 3.53751332e-01 -3.91649246e-01 1.90165713e-01 -4.93907779e-02 -1.35400879e+00 1.38361084e+00 -2.23387286e-01 4.44438756e-01 4.76178303e-02 -9.14953053e-01 8.00204515e-01 -2.45774705e-02 2.15401798e-01 -7.12955475e-01 2.04588607e-01 -1.93193734e-01 2.60682479e-02 -6.56423509e-01 2.71203220e-01 -1.02039792e-01 -3.79554741e-02 2.09448919e-01 1.27077594e-01 -3.22576500e-02 3.38721395e-01 7.90813982e-01 7.34829545e-01 -1.08288355e-01 3.65578294e-01 -1.88565254e-01 5.33212602e-01 1.38679650e-02 3.08129847e-01 7.11412489e-01 9.94100049e-02 3.08686316e-01 7.34473467e-01 3.30518708e-02 -6.23442888e-01 -1.37709904e+00 1.26259565e-01 1.05830288e+00 8.87042582e-01 -7.63425291e-01 -4.81718272e-01 -7.70062804e-01 4.90229242e-02 7.10545123e-01 -7.01673687e-01 -2.09374726e-01 -2.56477911e-02 -3.84161860e-01 1.62426561e-01 5.50199926e-01 5.66016078e-01 -9.13064539e-01 -2.77623743e-01 -2.81293660e-01 -1.17121577e-01 -1.13576722e+00 -2.36990064e-01 -6.68486021e-03 -5.79704344e-01 -1.28157663e+00 -3.96286584e-02 -8.01300406e-01 9.74470675e-01 6.50205851e-01 9.95776236e-01 2.54293948e-01 -3.62182617e-01 5.11343658e-01 -3.01319718e-01 -1.21237166e-01 -2.90396780e-01 -3.52541506e-01 -1.62154734e-01 1.82859376e-01 4.49955821e-01 -5.54719687e-01 -3.33217412e-01 4.19330597e-01 -9.01203215e-01 4.76457089e-01 6.17874980e-01 6.15814328e-01 7.65605032e-01 2.16283336e-01 1.11208074e-01 -9.90915298e-01 3.62143904e-01 -5.95887721e-01 -3.79523009e-01 5.68573058e-01 -3.22237104e-01 -8.14955309e-02 2.67633438e-01 -3.05666327e-01 -1.06395686e+00 -1.56045377e-01 3.81039739e-01 -6.34682059e-01 -2.30620489e-01 6.65009439e-01 -7.03155041e-01 2.33192861e-01 6.45544410e-01 1.59262046e-01 -1.49814352e-01 -2.43379116e-01 7.75663018e-01 3.62819105e-01 5.70782661e-01 -4.95441616e-01 9.18751478e-01 6.58818185e-01 2.92466611e-01 -8.37366879e-01 -7.81675518e-01 -6.10990584e-01 -6.38279974e-01 -4.72624958e-01 1.06381953e+00 -8.26990664e-01 -7.88641691e-01 2.80301958e-01 -9.81434226e-01 1.06189445e-01 -2.72862941e-01 4.22071010e-01 -1.40293226e-01 3.17433804e-01 -6.35107696e-01 -6.45162642e-01 3.29309642e-01 -9.17344332e-01 1.03429449e+00 3.02983522e-01 -5.15734665e-02 -1.07925594e+00 -3.83581072e-01 1.82086557e-01 -3.06135267e-01 2.13818118e-01 1.20710874e+00 -7.25517988e-01 -9.12320077e-01 -8.59194100e-02 -7.91365027e-01 6.20470010e-02 1.75719395e-01 -1.34005040e-01 -8.77308071e-01 -6.31809607e-02 -3.31656247e-01 2.94052362e-02 7.04037488e-01 5.82857765e-02 1.10853004e+00 -1.80981979e-01 -6.68073237e-01 3.92832965e-01 1.51591229e+00 1.91800117e-01 6.22796655e-01 6.33874014e-02 1.18665016e+00 7.78259456e-01 4.73079562e-01 1.64007485e-01 5.50911665e-01 5.94433546e-01 5.91778517e-01 -3.29464614e-01 -4.11870360e-01 -5.71955740e-01 1.36160269e-01 6.03594363e-01 -3.22205454e-01 -3.54183763e-01 -8.53310645e-01 3.79586816e-01 -1.98056793e+00 -9.65735376e-01 -3.48070592e-01 2.08471155e+00 4.88605887e-01 -3.07930652e-02 3.52583304e-02 -2.10262716e-01 9.32077229e-01 2.88779199e-01 -4.71452981e-01 1.44125566e-01 -3.68663013e-01 -4.43257511e-01 2.21348226e-01 5.05126178e-01 -8.04649532e-01 1.09150374e+00 5.33261871e+00 8.48187506e-01 -7.18198180e-01 3.16308662e-02 3.88869733e-01 2.09820881e-01 -5.88489592e-01 1.89329132e-01 -5.71754098e-01 1.19042031e-01 2.88713634e-01 -2.62629569e-01 6.21506751e-01 6.99082136e-01 3.36606540e-02 -3.74953866e-01 -1.16141367e+00 1.06025589e+00 3.86997581e-01 -1.09659302e+00 3.83056551e-01 1.98672730e-02 3.87429625e-01 -2.59353280e-01 -1.02262065e-01 9.17783901e-02 6.99425936e-01 -9.55278337e-01 7.02011168e-01 7.35693693e-01 6.44154966e-01 -6.30532682e-01 5.02651989e-01 2.15059996e-01 -1.55021250e+00 -2.97047123e-02 -2.25117639e-01 8.57734680e-02 1.23950332e-01 3.80292684e-01 -7.14125574e-01 9.37961221e-01 6.78893685e-01 1.09116817e+00 -6.65248871e-01 7.33855486e-01 -6.71707213e-01 2.39857718e-01 -6.41204789e-02 6.04767688e-02 3.71945649e-02 -3.38779539e-01 7.88262427e-01 9.92984056e-01 2.13082924e-01 3.95985514e-01 4.56176668e-01 1.08107150e+00 -4.58204746e-02 1.10248543e-01 -7.10349262e-01 -2.51318544e-01 4.92042452e-01 1.32525873e+00 -1.11823916e+00 -5.72080851e-01 -5.58324277e-01 8.88221383e-01 4.14942741e-01 8.23484063e-01 -7.26148486e-01 -4.38171476e-01 5.91811776e-01 1.17317244e-01 2.19940580e-02 -2.09199578e-01 -2.26862341e-01 -1.24949169e+00 -1.82344541e-01 -5.49056530e-01 8.34525049e-01 -1.08138192e+00 -1.25906265e+00 3.80021363e-01 4.12692159e-01 -1.18557537e+00 1.11913703e-01 -3.50704551e-01 -3.87154579e-01 7.45420694e-01 -1.28319180e+00 -1.42617893e+00 -5.78020096e-01 8.87263358e-01 2.86450356e-01 -1.26326466e-02 5.63507795e-01 -8.27956721e-02 -5.49303651e-01 1.70322806e-01 -3.16005796e-01 2.16012135e-01 2.13878855e-01 -1.08148468e+00 -3.03231589e-02 1.08817506e+00 3.97409171e-01 7.83630908e-01 7.11783350e-01 -1.10592067e+00 -1.51793909e+00 -1.06529760e+00 7.95917571e-01 -3.53461772e-01 8.42546463e-01 -6.88505590e-01 -1.07433593e+00 7.67543435e-01 -6.77779466e-02 4.54384349e-02 3.52363318e-01 3.68435502e-01 -5.49584806e-01 -4.87987660e-02 -7.49736547e-01 8.85833025e-01 1.42236149e+00 -8.97607863e-01 -7.54566729e-01 3.40287536e-01 9.44761634e-01 -3.50389808e-01 -4.94225234e-01 1.81022406e-01 1.48657188e-01 -8.85789633e-01 1.12053406e+00 -7.03754783e-01 3.82688135e-01 -6.41649663e-01 -3.16032588e-01 -1.14226830e+00 -3.32774162e-01 -5.30456267e-02 -8.29795823e-02 1.22314405e+00 2.46189162e-01 -4.72665340e-01 5.61145067e-01 4.83457744e-01 -2.87732095e-01 -3.15884680e-01 -6.60953581e-01 -9.15768862e-01 -4.81863528e-01 -5.21948218e-01 7.07595229e-01 1.21167839e+00 2.81462461e-01 6.20303690e-01 -8.06613863e-02 5.08009315e-01 5.64045072e-01 1.50907978e-01 6.16663754e-01 -1.26941979e+00 -2.30807602e-01 -3.17139953e-01 -9.22961295e-01 -7.54513919e-01 2.36475214e-01 -1.27070236e+00 -1.35531470e-01 -1.86216283e+00 4.93575543e-01 -1.75378531e-01 -3.07932109e-01 8.31236064e-01 -2.33025834e-01 9.25440565e-02 4.07412887e-01 1.47496372e-01 -8.81907880e-01 5.14720380e-01 1.63316226e+00 -3.33941102e-01 -3.25053811e-01 -5.42336762e-01 -6.25588417e-01 6.00416541e-01 2.78291643e-01 -1.79324627e-01 -8.42430055e-01 -1.54662982e-01 2.70933509e-01 1.30643576e-01 8.34840178e-01 -5.94971120e-01 3.28804284e-01 -4.26859498e-01 2.91929275e-01 -5.86722195e-01 4.31444675e-01 -1.05213487e+00 3.83352190e-01 2.46325970e-01 -1.36394501e-01 -9.82526615e-02 9.06260535e-02 8.27429354e-01 -5.09540796e-01 1.01038717e-01 4.06874210e-01 -2.52926469e-01 -1.26818383e+00 3.25116515e-01 -1.75881341e-01 2.56269495e-03 1.28994608e+00 -5.14780104e-01 -5.94151676e-01 -4.01743442e-01 -9.79369760e-01 3.89604956e-01 4.59630519e-01 5.99454820e-01 8.18585455e-01 -1.48441887e+00 -4.65966731e-01 2.76653647e-01 5.06025732e-01 -8.81739557e-02 6.49558485e-01 6.99535429e-01 -1.49729908e-01 1.20383620e-01 -3.29620875e-02 -7.74813414e-01 -1.69391286e+00 8.74611616e-01 3.04404378e-01 3.71822089e-01 -6.23930216e-01 9.03112948e-01 9.93149340e-01 -3.32054347e-01 5.57376742e-02 -3.25606585e-01 -5.38816452e-01 2.34203905e-01 4.00238127e-01 2.99552172e-01 -2.79547125e-01 -1.04794896e+00 -3.84720355e-01 4.01646525e-01 3.65604647e-02 6.59705922e-02 1.09957349e+00 -3.49381059e-01 -6.41377091e-01 6.58714056e-01 1.02761662e+00 -2.78810054e-01 -8.98287714e-01 -4.16651160e-01 -1.61643803e-01 -6.80575967e-01 -5.11395074e-02 -6.25483155e-01 -1.09247863e+00 7.76856840e-01 3.40976954e-01 2.69888282e-01 1.05489564e+00 7.00729311e-01 5.28327487e-02 1.40635625e-01 3.14365357e-01 -9.29727793e-01 3.09495807e-01 3.04526299e-01 8.37670922e-01 -1.10299051e+00 4.54572774e-02 -1.11301923e+00 -8.40161383e-01 1.04331732e+00 7.72597730e-01 1.65062413e-01 6.80437863e-01 -1.16488367e-01 -1.45203486e-01 -7.70781279e-01 -5.11716247e-01 -6.43018901e-01 8.55577528e-01 6.51368856e-01 2.76000742e-02 3.28813970e-01 2.25298628e-02 4.64983284e-01 -7.84474704e-03 -4.66666967e-01 1.79967001e-01 8.56274843e-01 -3.70177269e-01 -7.48445868e-01 -4.11266237e-01 3.71844798e-01 1.63639501e-01 -1.09141782e-01 -4.76617634e-01 9.91838634e-01 3.27274352e-01 1.12152660e+00 2.12885514e-01 -5.94527185e-01 3.80238235e-01 -4.56704408e-01 3.15784186e-01 -8.67505491e-01 -1.93042502e-01 1.55765712e-01 1.84915930e-01 -6.19208694e-01 -5.58629215e-01 -5.05303681e-01 -1.58454084e+00 -2.55861729e-01 -3.08580965e-01 -2.30710641e-01 4.37626183e-01 1.14361155e+00 1.97684586e-01 7.68394470e-01 2.81100541e-01 -4.34798717e-01 4.27587867e-01 -3.21585864e-01 -9.57624674e-01 8.25364172e-01 1.34514973e-01 -8.12779844e-01 -3.74543011e-01 4.59027775e-02]
[10.304455757141113, 1.6601642370224]
78d1ea68-bcad-4082-9b78-a3bc528844d9
hatemm-a-multi-modal-dataset-for-hate-video
2305.03915
null
https://arxiv.org/abs/2305.03915v1
https://arxiv.org/pdf/2305.03915v1.pdf
HateMM: A Multi-Modal Dataset for Hate Video Classification
Hate speech has become one of the most significant issues in modern society, having implications in both the online and the offline world. Due to this, hate speech research has recently gained a lot of traction. However, most of the work has primarily focused on text media with relatively little work on images and even lesser on videos. Thus, early stage automated video moderation techniques are needed to handle the videos that are being uploaded to keep the platform safe and healthy. With a view to detect and remove hateful content from the video sharing platforms, our work focuses on hate video detection using multi-modalities. To this end, we curate ~43 hours of videos from BitChute and manually annotate them as hate or non-hate, along with the frame spans which could explain the labelling decision. To collect the relevant videos we harnessed search keywords from hate lexicons. We observe various cues in images and audio of hateful videos. Further, we build deep learning multi-modal models to classify the hate videos and observe that using all the modalities of the videos improves the overall hate speech detection performance (accuracy=0.798, macro F1-score=0.790) by ~5.7% compared to the best uni-modal model in terms of macro F1 score. In summary, our work takes the first step toward understanding and modeling hateful videos on video hosting platforms such as BitChute.
['Animesh Mukherjee', 'Manish Gupta', 'Binny Mathew', 'Punyajoy Saha', 'Rohit Raj', 'Mithun Das']
2023-05-06
null
null
null
null
['video-classification', 'hate-speech-detection']
['computer-vision', 'natural-language-processing']
[-1.14409029e-01 -1.83048293e-01 -1.62103549e-01 2.85166889e-01 -5.93624890e-01 -8.30279768e-01 5.00821173e-01 8.26200917e-02 -1.83425143e-01 3.65759879e-01 2.86352873e-01 1.93346977e-01 2.11688131e-01 -1.77501470e-01 -5.67564309e-01 -7.43302941e-01 3.92037071e-02 -3.32632989e-01 1.74261346e-01 -1.03310540e-01 5.22949278e-01 2.44937301e-01 -1.65045738e+00 5.04231334e-01 4.93723214e-01 6.87481225e-01 -7.24560721e-03 8.66742790e-01 2.92213291e-01 1.27049470e+00 -1.04902148e+00 -8.53626251e-01 -6.21462986e-02 -3.10877711e-01 -5.79223752e-01 1.53522581e-01 7.25959539e-01 -8.28270555e-01 -7.34791219e-01 1.28596270e+00 6.80426657e-01 -1.67509049e-01 4.59029198e-01 -1.42658305e+00 -8.86957407e-01 3.36260229e-01 -6.50828004e-01 6.12381637e-01 5.37129998e-01 1.58064574e-01 4.94731456e-01 -7.84497261e-01 7.52406657e-01 1.10176325e+00 6.26182556e-01 5.90810120e-01 -6.50612235e-01 -9.39167619e-01 -3.74003053e-01 6.16454065e-01 -1.45139456e+00 -6.80571556e-01 8.26416016e-01 -1.10615432e+00 6.94884777e-01 1.03499576e-01 6.94595158e-01 1.63612568e+00 7.10299537e-02 9.36173439e-01 8.58698547e-01 -3.31704557e-01 -2.39713907e-01 3.48865002e-01 -1.49087995e-01 6.39771819e-01 1.49249867e-01 -5.88466227e-01 -7.45656788e-01 -5.98594546e-02 3.69949758e-01 1.54867738e-01 -2.87452966e-01 1.63547456e-01 -8.33437800e-01 9.23481762e-01 -6.19383156e-02 6.07734919e-01 -2.04488933e-01 3.31745483e-02 8.40089321e-01 1.27209783e-01 5.78088224e-01 5.16831398e-01 1.19363882e-01 -5.36638319e-01 -1.15823877e+00 2.83219576e-01 4.61575985e-01 6.76347911e-01 4.42346424e-01 -1.06454790e-01 -1.10092290e-01 9.11195278e-01 8.16232041e-02 7.20252216e-01 2.01139614e-01 -8.76513720e-01 4.54308957e-01 2.48185933e-01 1.13580171e-02 -1.64092851e+00 -8.75081941e-02 2.63172928e-02 -5.12721479e-01 -2.33353704e-01 3.43990833e-01 -3.32316816e-01 -1.00602663e+00 1.34775758e+00 1.46686375e-01 1.04734771e-01 -5.43458879e-01 8.48001957e-01 5.71357727e-01 8.56977582e-01 2.22009748e-01 -2.67300665e-01 1.53551090e+00 -7.97989130e-01 -1.25408614e+00 -1.93428621e-02 7.78843224e-01 -1.16428256e+00 6.64349139e-01 5.92019796e-01 -7.22963274e-01 -1.56017661e-01 -9.44836676e-01 -1.73434749e-01 -7.20813751e-01 -1.11832097e-01 1.60788044e-01 8.58737290e-01 -7.44375527e-01 1.76864877e-01 -5.02350748e-01 -5.04843891e-01 6.14089131e-01 -1.29806057e-01 -5.89896381e-01 -4.36020315e-01 -1.31017756e+00 1.01464212e+00 2.10139409e-01 -4.85724024e-02 -1.20919287e+00 -6.99462056e-01 -7.75381386e-01 -2.54025221e-01 5.58525801e-01 1.08597375e-01 9.76086795e-01 -1.06805253e+00 -8.70613158e-01 9.40672934e-01 1.22276116e-02 -3.79084826e-01 1.77383378e-01 -6.53946698e-01 -6.47343040e-01 6.48037553e-01 1.26418576e-01 5.58410585e-01 1.42973268e+00 -1.13746333e+00 -4.99423116e-01 -2.58261681e-01 1.72325850e-01 -1.59692273e-01 -1.01051450e+00 7.59705245e-01 -2.58560717e-01 -8.50418329e-01 -8.15372527e-01 -1.03107202e+00 8.54382098e-01 -2.51716703e-01 -5.17234325e-01 -5.70382848e-02 1.67226672e+00 -1.48442960e+00 1.98227870e+00 -2.27982116e+00 2.29442045e-01 -1.20130748e-01 6.01415157e-01 7.67172933e-01 1.77603334e-01 7.59508014e-01 6.58298731e-02 4.19741899e-01 1.60621434e-01 -3.31414789e-01 4.27549286e-03 -2.08460107e-01 -2.78615594e-01 8.39525104e-01 1.01021841e-01 4.03215080e-01 -9.36691046e-01 -4.76602554e-01 3.09312195e-01 8.95758867e-01 -5.98377824e-01 3.84723932e-01 1.94381848e-01 1.22661605e-01 -7.59445280e-02 9.92013514e-01 4.91251379e-01 1.88452005e-01 -2.29431689e-01 -1.26460329e-01 -2.47941479e-01 -1.45337477e-01 -4.85380471e-01 1.30035496e+00 -1.84554026e-01 1.28013921e+00 2.92652220e-01 -4.22240764e-01 3.63974810e-01 6.43384874e-01 5.06153643e-01 -4.92935240e-01 3.60461324e-01 1.16054825e-01 -1.50250450e-01 -1.19093144e+00 6.37363970e-01 1.52963847e-01 5.31538995e-03 5.76801486e-02 2.54764497e-01 2.89899379e-01 3.13493639e-01 3.89298499e-01 1.29074192e+00 -1.04273975e-01 1.34562999e-01 1.78772569e-01 4.94049847e-01 -1.47004887e-01 7.52731860e-02 4.23748583e-01 -8.50994110e-01 6.34186029e-01 7.31327295e-01 -3.56552035e-01 -1.18910825e+00 -5.04576087e-01 1.44906063e-02 1.33369029e+00 -1.80561453e-01 -8.91563833e-01 -1.22873831e+00 -6.96717143e-01 -1.46833837e-01 5.38170874e-01 -7.23983705e-01 -1.80668086e-01 -5.40002525e-01 -5.41906893e-01 8.99233401e-01 -3.35609466e-02 4.01781768e-01 -1.06287944e+00 -4.62752730e-01 -2.19318345e-01 -3.06957215e-01 -1.29180694e+00 -6.01795316e-01 -2.89926112e-01 2.35640034e-01 -1.20756865e+00 -9.62330580e-01 -5.96727192e-01 3.81435961e-01 6.03337407e-01 3.82928818e-01 3.65672916e-01 -3.11069518e-01 4.19454604e-01 -7.16638088e-01 -3.09221625e-01 -4.56464857e-01 -3.08019370e-02 1.64619926e-02 2.36814991e-01 5.60178936e-01 -2.74182349e-01 -4.86557573e-01 6.02304563e-02 -1.20301390e+00 -1.47091627e-01 1.98857576e-01 4.41046119e-01 -1.14699282e-01 2.12224618e-01 1.02924094e-01 -6.32952094e-01 3.86698276e-01 -1.21909046e+00 -5.90498038e-02 -1.74736276e-01 4.67952453e-02 -8.46314788e-01 7.88396299e-01 -6.17194176e-01 -7.09017456e-01 -2.39350006e-01 7.02751353e-02 -9.77931917e-01 -3.48716497e-01 1.88326046e-01 5.99212386e-02 1.76403131e-02 4.05909300e-01 -1.38076231e-01 -2.10975498e-01 -4.53762680e-01 9.54776108e-02 1.11965907e+00 3.36823702e-01 -9.38345790e-02 8.56220007e-01 4.64987218e-01 -3.20144087e-01 -1.38733757e+00 -9.28148210e-01 -7.89600670e-01 -6.16994023e-01 -6.88883126e-01 1.25308883e+00 -9.06923831e-01 -5.05727828e-01 1.05655527e+00 -1.28250945e+00 8.33340883e-02 7.76807129e-01 1.70821562e-01 -9.51397121e-02 7.06707656e-01 -8.80101621e-01 -1.23077106e+00 -2.06165925e-01 -1.14818919e+00 1.08558857e+00 -6.54224977e-02 -2.87820905e-01 -8.24194610e-01 -1.83790792e-02 8.40989232e-01 2.98584342e-01 4.63063687e-01 7.97771394e-01 -7.62612820e-01 -2.69764066e-01 -3.88233036e-01 -3.60060900e-01 4.50616211e-01 -4.99247480e-03 3.47898096e-01 -1.28023040e+00 -4.02751386e-01 -1.08959466e-01 -5.36163509e-01 5.59360802e-01 1.82436153e-01 9.20325816e-01 -5.66317737e-01 -1.97685570e-01 2.22611442e-01 1.29507279e+00 3.19342315e-01 8.25559318e-01 5.29872477e-01 1.14778876e+00 7.84638643e-01 4.69396651e-01 7.41962492e-01 2.41230756e-01 6.60610318e-01 6.39708757e-01 4.06498194e-01 -2.89091408e-01 -4.17564064e-01 8.19160342e-01 9.63451207e-01 1.58339906e-02 -5.36819041e-01 -1.04671109e+00 8.62694323e-01 -1.67419958e+00 -1.55675209e+00 -2.04097275e-02 1.94179785e+00 4.63365316e-01 -1.21748574e-01 5.67031682e-01 1.59317493e-01 1.25087440e+00 3.92983675e-01 -6.73000962e-02 -4.75623757e-01 -7.37265497e-02 -3.02410364e-01 4.09723878e-01 1.37789443e-01 -1.52722311e+00 8.50931585e-01 5.62412548e+00 1.06064725e+00 -1.14374185e+00 3.97868305e-01 5.01983821e-01 -4.59170610e-01 2.02939913e-01 -3.59502554e-01 -6.38266444e-01 1.07757425e+00 1.30477846e+00 1.97926179e-01 6.89939678e-01 7.57720828e-01 4.28589791e-01 -5.53643852e-02 -7.35178709e-01 1.29400289e+00 7.86222637e-01 -1.16330397e+00 -1.31263554e-01 2.77832806e-01 6.51086926e-01 -1.93857312e-01 3.05824339e-01 3.45332652e-01 -5.17382979e-01 -1.16619551e+00 9.61649239e-01 2.34425694e-01 6.16074324e-01 -8.93541276e-01 8.71062100e-01 2.25477815e-01 -7.28508532e-01 -2.35109955e-01 -9.82795954e-02 2.48248335e-02 1.72427356e-01 4.39676374e-01 -6.78195000e-01 -6.60253763e-02 9.25096452e-01 8.53516817e-01 -6.30804360e-01 9.76068437e-01 -1.07777819e-01 6.91074371e-01 5.42796254e-02 1.93429232e-01 1.98481083e-01 1.93375602e-01 6.04038596e-01 1.60843754e+00 5.00499964e-01 -2.46893615e-01 -1.56931475e-01 3.84466588e-01 -3.21728230e-01 6.40265867e-02 -9.27570820e-01 -5.97061515e-01 6.50676847e-01 1.27258682e+00 -4.10600066e-01 -5.60455211e-02 -6.40070021e-01 1.13969433e+00 2.27142185e-01 -6.91094771e-02 -1.29298878e+00 -6.27773881e-01 8.24332237e-01 3.05103958e-01 2.86951870e-01 -1.74867198e-01 3.67293179e-01 -1.13687241e+00 -3.19550261e-02 -1.12111485e+00 1.75216541e-01 -8.55119169e-01 -1.15638983e+00 3.07971954e-01 -5.19239418e-02 -8.77942562e-01 -9.82588083e-02 -5.09728909e-01 -1.69131264e-01 3.89602363e-01 -1.24033272e+00 -1.35009181e+00 -1.66528866e-01 3.66343826e-01 6.89637780e-01 -2.54586875e-03 4.99194980e-01 8.53322148e-01 -1.00382614e+00 4.37676311e-01 -9.18023214e-02 5.80185235e-01 1.04798222e+00 -8.64243627e-01 -7.09612295e-02 1.14075220e+00 -2.21738085e-01 6.24392271e-01 8.71070027e-01 -8.06900084e-01 -1.55898952e+00 -8.88693750e-01 6.72317147e-01 -8.24825466e-01 1.32656121e+00 -3.27936381e-01 -9.19337094e-01 4.80286956e-01 7.11221457e-01 -2.58634120e-01 8.30829263e-01 -2.19884276e-01 -6.04434311e-01 4.69486684e-01 -1.08614492e+00 3.58662188e-01 8.23152542e-01 -1.04225838e+00 -4.16117251e-01 4.25857544e-01 5.52147150e-01 -2.05157995e-01 -9.77061629e-01 -1.32779464e-01 7.29274571e-01 -1.00147808e+00 6.31350398e-01 -5.49686253e-01 7.78903484e-01 -2.35199183e-01 -1.20771743e-01 -1.04951191e+00 -2.97196209e-01 -9.36408758e-01 -6.89735889e-01 1.57452214e+00 -1.19563892e-01 1.13399573e-01 4.49013442e-01 4.16080356e-01 -1.71052292e-01 -3.90272260e-01 -8.22862506e-01 -3.31932336e-01 -1.42048016e-01 -2.90879011e-01 -7.91850016e-02 1.39959860e+00 5.45080751e-02 1.67041823e-01 -1.25218070e+00 1.93969682e-01 4.33708519e-01 -5.86910605e-01 6.26359761e-01 -8.20179343e-01 2.06438392e-01 -3.19758326e-01 -6.35870874e-01 -4.56159115e-01 2.87897497e-01 -4.62116927e-01 -2.50154704e-01 -1.03690350e+00 6.32535815e-01 2.73250759e-01 -2.94195246e-02 5.86884499e-01 4.52361442e-02 7.74026334e-01 5.52105248e-01 3.64282370e-01 -8.93208981e-01 -1.31510105e-03 1.04275298e+00 -1.72667906e-01 2.29202375e-01 -5.92151225e-01 -4.78852630e-01 7.51247346e-01 6.16121709e-01 -4.88924503e-01 -1.31049469e-01 -4.10487980e-01 3.41040820e-01 -3.37964475e-01 4.09285814e-01 -9.25134003e-01 -7.63230473e-02 -3.02682612e-02 1.95114240e-01 -3.99803340e-01 5.95960617e-01 -7.37656891e-01 -4.64543328e-02 2.94691890e-01 -5.11020198e-02 5.73608279e-02 1.55132607e-01 5.10978520e-01 -2.08207548e-01 -3.70626539e-01 7.40191579e-01 1.03453055e-01 -7.63877511e-01 1.99159551e-02 -8.52525055e-01 4.28748280e-02 1.16124427e+00 -1.60751924e-01 -9.94866431e-01 -5.67627668e-01 -4.19663638e-01 -1.30918249e-01 6.36318564e-01 6.54578805e-01 4.84051585e-01 -1.00342596e+00 -5.02348065e-01 -7.02520981e-02 2.24429742e-01 -8.14763486e-01 4.32985336e-01 1.01799226e+00 -6.59743488e-01 5.87035120e-01 -4.22640234e-01 -3.50810029e-02 -1.47059953e+00 8.62367570e-01 -5.81696481e-02 3.39053124e-01 -3.72216403e-01 4.83850837e-01 2.03176439e-02 4.71258223e-01 2.71650910e-01 3.28041852e-01 -4.22119528e-01 5.73766470e-01 9.99391198e-01 7.50017643e-01 -1.03350885e-01 -1.44171512e+00 -4.68612611e-01 2.51667410e-01 -1.51684120e-01 3.83191615e-01 1.19119298e+00 -1.76050186e-01 -2.68320501e-01 3.18562388e-01 1.60884154e+00 5.05001128e-01 -8.45435679e-01 4.45385695e-01 -1.25633836e-01 -8.86860073e-01 1.59142345e-01 -5.99380016e-01 -9.63164985e-01 8.51550937e-01 3.67212385e-01 7.49256611e-01 8.57288063e-01 -7.46128932e-02 1.26464236e+00 7.99624324e-02 2.10481316e-01 -1.13189840e+00 2.87990749e-01 4.40474361e-01 7.01625705e-01 -1.27292848e+00 -3.22171673e-02 -3.97190541e-01 -6.10747874e-01 1.07123780e+00 5.61112285e-01 2.82473937e-02 3.65828872e-01 3.92624550e-02 -5.46145998e-03 -2.89071351e-01 -4.25797015e-01 -1.02323085e-01 2.58004218e-01 6.74796581e-01 5.57378590e-01 -1.33798033e-01 -5.91987781e-02 3.72184724e-01 3.31476331e-02 -2.97454685e-01 8.40265334e-01 8.21691692e-01 -6.30492747e-01 -6.50522351e-01 -6.62921011e-01 3.78295839e-01 -1.11833000e+00 -3.00740525e-02 -7.24307358e-01 8.45690846e-01 4.37306494e-01 1.12294114e+00 -1.69574037e-01 -7.13230669e-01 -1.08642213e-01 1.00190558e-01 3.19370359e-01 -4.93661225e-01 -6.46509051e-01 1.07956402e-01 2.80233294e-01 -3.97420913e-01 -5.79381585e-01 -6.01354539e-01 -5.71080208e-01 -8.34993422e-01 -2.75746971e-01 -7.60683492e-02 7.19751298e-01 9.15906549e-01 4.51626092e-01 1.60687044e-01 4.29454714e-01 -1.04005384e+00 1.89996913e-01 -9.31960702e-01 -3.96354079e-01 6.51439190e-01 6.80396736e-01 -8.51298213e-01 -6.86267614e-01 2.23991141e-01]
[8.659400939941406, 10.576521873474121]
5b7f945c-c4f5-48de-a728-70273cf324ee
learning-and-planning-in-complex-action
2104.06303
null
https://arxiv.org/abs/2104.06303v1
https://arxiv.org/pdf/2104.06303v1.pdf
Learning and Planning in Complex Action Spaces
Many important real-world problems have action spaces that are high-dimensional, continuous or both, making full enumeration of all possible actions infeasible. Instead, only small subsets of actions can be sampled for the purpose of policy evaluation and improvement. In this paper, we propose a general framework to reason in a principled way about policy evaluation and improvement over such sampled action subsets. This sample-based policy iteration framework can in principle be applied to any reinforcement learning algorithm based upon policy iteration. Concretely, we propose Sampled MuZero, an extension of the MuZero algorithm that is able to learn in domains with arbitrarily complex action spaces by planning over sampled actions. We demonstrate this approach on the classical board game of Go and on two continuous control benchmark domains: DeepMind Control Suite and Real-World RL Suite.
['David Silver', 'Simon Schmitt', 'Mohammadamin Barekatain', 'Ioannis Antonoglou', 'Julian Schrittwieser', 'Thomas Hubert']
2021-04-13
null
null
null
null
['game-of-go']
['playing-games']
[ 4.21740472e-01 4.94999707e-01 -4.98398483e-01 1.16983518e-01 -7.72246659e-01 -6.31587029e-01 7.43664145e-01 8.14036354e-02 -7.26617754e-01 1.43042576e+00 3.38428319e-02 -6.22232139e-01 -5.76294124e-01 -8.89715493e-01 -6.64181471e-01 -7.40992785e-01 -3.41469377e-01 7.17615604e-01 3.92197698e-01 -2.30431676e-01 3.25199306e-01 3.20405960e-01 -1.40570140e+00 5.01653329e-02 7.36329079e-01 6.79684520e-01 1.81422651e-01 6.33587360e-01 2.18997911e-01 8.22776914e-01 -4.31511253e-01 3.07646215e-01 5.37330925e-01 -5.47091722e-01 -9.71135795e-01 3.64844471e-01 -1.76220015e-01 -5.62314391e-01 2.50856970e-02 9.35814440e-01 5.03024042e-01 6.30745232e-01 3.91313761e-01 -1.14184725e+00 2.16713443e-01 8.57758105e-01 -2.54735053e-01 -9.74977091e-02 3.80702734e-01 7.79053688e-01 1.05875075e+00 4.33020517e-02 6.33564770e-01 1.49719441e+00 1.07778281e-01 7.50874758e-01 -1.55535126e+00 -2.43475512e-01 5.24986863e-01 -3.04951929e-02 -6.30414784e-01 -7.54255131e-02 2.91713566e-01 -2.12221295e-01 1.06485057e+00 2.25364596e-01 9.82149184e-01 9.86049712e-01 9.69358608e-02 1.01150572e+00 1.40076888e+00 -2.98459321e-01 1.00564766e+00 -4.25280243e-01 -4.62223083e-01 5.38762748e-01 2.04768851e-01 7.24070430e-01 1.23785213e-02 -4.01088119e-01 8.58726084e-01 -4.38881740e-02 -1.46712750e-01 -8.30630243e-01 -1.37920594e+00 9.09371376e-01 -8.48689601e-02 2.93241926e-02 -6.54802680e-01 5.19042432e-01 4.55939919e-01 5.90690076e-01 5.78302480e-02 1.11225665e+00 -7.84273803e-01 -5.11015654e-01 -4.20814425e-01 8.65138888e-01 9.17320549e-01 4.47068840e-01 7.05473423e-01 2.33725384e-01 -5.66732585e-01 3.88123930e-01 5.42084053e-02 3.94657493e-01 3.70542526e-01 -1.64586246e+00 4.38120842e-01 3.57909918e-01 7.70258963e-01 -3.33777219e-01 -4.36218828e-01 -2.57817209e-01 -4.27641988e-01 1.02452052e+00 5.36482990e-01 -5.59867322e-01 -7.10202277e-01 1.69826543e+00 4.52647954e-01 2.68001437e-01 3.03384751e-01 6.54858410e-01 -2.14442924e-01 4.75717992e-01 -1.51220724e-01 -4.71492499e-01 8.23425651e-01 -8.21676731e-01 -5.33398092e-01 -2.54029334e-01 6.96045458e-01 -2.56993677e-02 1.21840537e+00 8.45007777e-01 -1.05781996e+00 -1.77121297e-01 -9.19371068e-01 6.92436337e-01 -1.34136766e-01 -2.13934660e-01 8.18059444e-01 3.69227558e-01 -9.47847128e-01 1.13185668e+00 -1.07940614e+00 -8.36503431e-02 4.62541819e-01 6.19728088e-01 -1.60084456e-01 7.46706054e-02 -9.50026989e-01 8.56977105e-01 9.56241071e-01 -1.74913719e-01 -1.40250552e+00 -3.39978069e-01 -6.01080000e-01 5.58478348e-02 1.37499022e+00 -4.38188672e-01 1.88688385e+00 -6.93338394e-01 -2.12702680e+00 5.01176342e-02 5.18880188e-01 -8.57778728e-01 7.22355902e-01 -1.85521603e-01 2.69857887e-02 -1.44197404e-01 -1.66825384e-01 4.98199403e-01 8.73272121e-01 -8.13173652e-01 -8.46557498e-01 -1.18056022e-01 6.10428691e-01 3.90845835e-01 1.64167374e-01 -4.74859118e-01 2.91332185e-01 -2.46351629e-01 -4.53077406e-01 -1.10371423e+00 -9.95985270e-01 -2.85882473e-01 -2.68980205e-01 -3.11475635e-01 5.02687037e-01 8.27075765e-02 1.06968284e+00 -1.74612796e+00 4.12470251e-01 2.55559027e-01 -1.43330261e-01 4.43787932e-01 -3.81696433e-01 5.91473341e-01 2.15167269e-01 -7.28431204e-03 -4.15661395e-01 1.37669057e-01 4.45852280e-01 7.64629900e-01 -3.76432210e-01 3.62901688e-01 -4.29694504e-02 7.62326837e-01 -1.38551104e+00 -7.17150122e-02 4.23783839e-01 -2.88049817e-01 -9.42828417e-01 3.90311815e-02 -1.02598548e+00 5.65852046e-01 -1.01182342e+00 2.42593259e-01 1.98987976e-01 2.02481657e-01 4.72997546e-01 6.27822757e-01 -2.53408313e-01 1.98050112e-01 -1.55129218e+00 1.56554925e+00 -4.24618959e-01 2.48784959e-01 7.59317540e-03 -1.16513562e+00 6.50845528e-01 2.46187314e-01 6.96521163e-01 -5.79441786e-01 6.78202882e-02 1.19626768e-01 1.38150215e-01 -1.98755935e-01 3.18107516e-01 -2.22492009e-01 -1.25021383e-01 7.55419135e-01 -9.79797244e-02 -4.41263974e-01 5.83744943e-01 -2.10854039e-01 1.53554392e+00 3.03833067e-01 8.29645455e-01 -1.64749116e-01 4.62568045e-01 6.68643937e-02 5.71514845e-01 1.23211038e+00 -1.99029699e-01 -1.14537152e-02 1.17426956e+00 -4.56997186e-01 -8.45352054e-01 -8.97754431e-01 2.06052914e-01 9.90653872e-01 -1.42750606e-01 -4.28302795e-01 -5.45668423e-01 -9.79603410e-01 1.13005526e-01 1.02138853e+00 -6.21499300e-01 -1.09771237e-01 -5.95838189e-01 -4.19935644e-01 1.24149144e-01 2.77661651e-01 2.49472141e-01 -1.67970097e+00 -1.28723562e+00 5.86564600e-01 3.86947364e-01 -7.09718406e-01 -8.57680142e-02 3.69620621e-01 -9.44345117e-01 -1.31717157e+00 -2.84169704e-01 -1.59974381e-01 3.34220946e-01 -2.95727104e-01 1.03039789e+00 -2.02885032e-01 -6.50552437e-02 7.27977812e-01 -1.83381736e-01 -4.91538256e-01 -6.04570270e-01 5.06300069e-02 4.21986096e-02 -2.52559811e-01 -1.69388384e-01 -4.87160563e-01 -4.91597265e-01 1.71175927e-01 -9.44538713e-01 1.71093736e-02 7.07144856e-01 1.04226446e+00 6.79832935e-01 2.55228698e-01 6.07840061e-01 -8.14750075e-01 1.04908121e+00 -2.40176693e-01 -1.22308457e+00 2.58427680e-01 -4.08122152e-01 6.27539039e-01 6.48609579e-01 -5.92287958e-01 -7.78993309e-01 2.62074023e-01 3.19717475e-03 -5.18500432e-02 -2.22882137e-01 3.99906844e-01 -8.27465281e-02 2.13973388e-01 5.37663400e-01 1.90360889e-01 2.31314555e-01 -3.78409535e-01 4.63413626e-01 1.70598194e-01 1.21685147e-01 -1.04819000e+00 4.84915137e-01 2.56755680e-01 2.44261548e-01 -5.57870805e-01 -4.69617873e-01 -1.80342998e-02 -3.14945847e-01 -8.42261761e-02 7.19405949e-01 -4.03899491e-01 -1.13970721e+00 7.74548873e-02 -6.84523106e-01 -1.21710551e+00 -9.75408256e-01 5.49030781e-01 -1.38272631e+00 3.50209884e-02 -9.53160152e-02 -9.65968966e-01 1.96927547e-01 -1.33412886e+00 8.94138157e-01 1.82693169e-01 9.60047729e-03 -9.12053764e-01 4.47881162e-01 -4.36026275e-01 2.60625184e-01 5.17323256e-01 7.67261744e-01 -7.11205125e-01 -5.62245309e-01 1.38485655e-01 4.82835919e-01 3.24976504e-01 1.88640237e-01 -2.50443250e-01 -3.44419569e-01 -5.96231401e-01 -1.76575661e-01 -6.71479940e-01 5.82842112e-01 4.19723034e-01 1.31626046e+00 -4.68653023e-01 -2.10673183e-01 1.21130094e-01 1.36265886e+00 4.88994747e-01 5.54969251e-01 5.75724244e-01 1.80764776e-02 2.00725392e-01 1.14528966e+00 8.63679290e-01 -2.01533049e-01 6.51342154e-01 7.34935880e-01 2.59863615e-01 5.41584551e-01 -3.33106607e-01 5.32807529e-01 -3.16102564e-01 -2.08527878e-01 -2.40292653e-01 -7.15171039e-01 3.82136166e-01 -2.21452761e+00 -1.22013307e+00 5.75900495e-01 2.30883002e+00 9.73120928e-01 3.90880972e-01 4.44005013e-01 5.97202871e-03 4.38115597e-01 1.22929834e-01 -9.11087990e-01 -5.54155946e-01 3.68194669e-01 4.32904750e-01 5.84201455e-01 5.78270555e-01 -1.01031816e+00 8.91548216e-01 6.44304466e+00 8.58594060e-01 -8.24165583e-01 -2.53825605e-01 4.59492028e-01 -2.93547869e-01 -1.65053457e-01 2.01040074e-01 -6.30116105e-01 1.37092143e-01 8.61508906e-01 -2.62026966e-01 9.90939736e-01 1.06949437e+00 5.14249384e-01 -2.97242939e-01 -1.17527461e+00 4.66485292e-01 -9.41186607e-01 -1.31450438e+00 -3.78304899e-01 2.94428647e-01 1.01000035e+00 -6.91737160e-02 -1.49925966e-02 8.00226331e-01 1.04834342e+00 -9.50581372e-01 3.46736938e-01 2.41115972e-01 4.86198932e-01 -9.62545455e-01 3.85007203e-01 6.81871712e-01 -7.30132222e-01 -5.84617019e-01 -3.69438291e-01 -3.80859882e-01 1.67672917e-01 1.06916852e-01 -1.12314260e+00 2.28031397e-01 1.07435450e-01 6.14510655e-01 -1.18224792e-01 1.09630060e+00 -4.66318756e-01 6.38303518e-01 -2.80935168e-01 -3.39628488e-01 8.19611728e-01 -3.54756862e-01 6.80936337e-01 5.98501861e-01 3.98328722e-01 2.04807445e-01 6.39679551e-01 5.75699687e-01 4.67563242e-01 -2.56216794e-01 -8.10904860e-01 -1.47377372e-01 2.60116518e-01 8.84958267e-01 -7.08283663e-01 -2.95121610e-01 -1.97355121e-01 4.87176478e-01 3.05281639e-01 4.75010365e-01 -8.52378607e-01 -1.30939722e-01 9.63636696e-01 -1.19384862e-01 6.09822631e-01 -3.30071032e-01 2.19250143e-01 -8.79233122e-01 -2.90752023e-01 -1.30911493e+00 4.03640002e-01 -2.38228291e-01 -8.72448146e-01 1.44797429e-01 2.87915200e-01 -1.22320914e+00 -8.30740333e-01 -7.16802478e-01 -5.49373806e-01 4.72454458e-01 -1.19152594e+00 -1.83584601e-01 3.34517777e-01 6.06353462e-01 6.10920310e-01 -1.76156119e-01 7.84750640e-01 -4.25249726e-01 -4.49600637e-01 -2.20703129e-02 4.12249982e-01 -3.16370577e-01 2.57343769e-01 -1.69884205e+00 1.81803331e-01 5.10422528e-01 -1.71286017e-01 2.11052835e-01 8.05091918e-01 -5.10279775e-01 -1.58321071e+00 -9.74737465e-01 -1.27586380e-01 -9.91452709e-02 7.99259663e-01 -1.27848357e-01 -5.61212778e-01 6.38948917e-01 2.72378325e-01 7.59213045e-02 1.42043279e-02 8.61552730e-02 3.67520481e-01 -8.75547305e-02 -1.12690127e+00 8.69971693e-01 1.08437157e+00 8.85463059e-02 -5.13910472e-01 3.93634856e-01 7.55032122e-01 -5.41277766e-01 -7.16446936e-01 3.97738963e-01 3.37864071e-01 -8.83734703e-01 8.47994089e-01 -1.02433538e+00 1.10129721e-01 -2.37182051e-01 -9.09924060e-02 -1.76387048e+00 -1.43336937e-01 -1.18009889e+00 -3.39793265e-01 6.65603995e-01 2.91548669e-01 -8.46198916e-01 6.04852319e-01 3.92715365e-01 5.65415882e-02 -1.03991044e+00 -1.11712539e+00 -1.04554045e+00 1.89873010e-01 -4.57858235e-01 8.59686375e-01 3.68179649e-01 1.19836189e-01 1.24571897e-01 -4.07722980e-01 -2.34035358e-01 5.09562194e-01 7.21796751e-02 8.21982265e-01 -9.68091130e-01 -9.78941739e-01 -6.32590473e-01 -6.40992001e-02 -1.01664948e+00 2.01862961e-01 -3.89222890e-01 2.92582750e-01 -1.44835377e+00 -2.82015741e-01 -5.49697757e-01 -3.67695242e-01 6.77550912e-01 1.24675237e-01 -6.55545413e-01 3.24849844e-01 -3.93950760e-01 -9.95994866e-01 6.65617406e-01 1.62679636e+00 -1.08520813e-01 -5.14347196e-01 4.78821844e-01 -4.45824921e-01 6.46056890e-01 1.06431007e+00 -3.66719097e-01 -8.96594644e-01 -3.53710800e-02 3.42786968e-01 5.71742594e-01 2.04510689e-01 -1.00359225e+00 -2.85406888e-01 -1.10081983e+00 -1.56543970e-01 -3.31432849e-01 3.27257328e-02 -6.37754798e-01 -1.26096666e-01 9.37163293e-01 -7.42650151e-01 -6.67335913e-02 1.43272489e-01 8.26909006e-01 6.41240925e-02 -3.04907888e-01 6.61731958e-01 -4.84386146e-01 -8.87591243e-01 4.12797153e-01 -6.51490331e-01 3.41396511e-01 1.27094913e+00 1.61379173e-01 -1.08818233e-01 -3.58661592e-01 -8.27937186e-01 4.12592053e-01 2.72929609e-01 1.71610132e-01 5.05369127e-01 -1.11395454e+00 -6.29826486e-01 -2.79164454e-03 -2.27353871e-01 1.93354711e-02 -1.33349478e-01 6.05350912e-01 -7.84800872e-02 4.85086262e-01 -3.49320501e-01 -2.67342746e-01 -8.96265984e-01 7.41035223e-01 4.79325145e-01 -8.87563705e-01 -6.56292140e-01 1.22152455e-01 -1.77013632e-02 -6.14112794e-01 2.50493973e-01 -7.38239706e-01 -2.05326334e-01 -2.92128205e-01 6.25602186e-01 4.58386004e-01 -3.59950989e-01 2.51274467e-01 -6.55370578e-02 -1.95363443e-02 2.67780066e-01 -5.81789374e-01 1.50830865e+00 1.98180825e-01 2.24257156e-01 2.95000851e-01 6.50492966e-01 -3.69642526e-01 -1.83726215e+00 -6.24128878e-02 1.72079831e-01 -5.26588678e-01 -8.27571675e-02 -9.01284218e-01 -7.00261950e-01 5.81966400e-01 4.37333465e-01 4.08142269e-01 1.14850140e+00 -3.27934682e-01 2.65461564e-01 9.40404773e-01 7.91620910e-01 -1.28076828e+00 1.81257099e-01 6.10055685e-01 8.51502299e-01 -9.84246492e-01 9.23427939e-03 3.77967209e-01 -8.45503151e-01 1.07406676e+00 6.95561230e-01 -2.56234407e-01 1.97672561e-01 2.34251544e-01 -5.65093279e-01 7.12303594e-02 -1.27763915e+00 -5.70508003e-01 -5.91204725e-02 6.95043027e-01 -6.52228966e-02 2.61644393e-01 -5.19342721e-01 1.27649963e-01 1.37394086e-01 2.18783975e-01 6.61428988e-01 1.30587983e+00 -8.67162764e-01 -1.60085988e+00 -3.00967157e-01 5.39299548e-01 -1.52757227e-01 3.26614499e-01 -1.77758217e-01 9.78370547e-01 -2.16512103e-02 7.24172235e-01 -8.37523341e-02 -1.19514819e-02 2.90252954e-01 -1.52576223e-01 7.27495730e-01 -7.37022340e-01 -2.67845005e-01 6.80881506e-03 3.31159770e-01 -1.06018174e+00 -2.17754409e-01 -9.23152626e-01 -1.31899869e+00 -2.64040157e-02 1.23662516e-01 2.55874962e-01 3.35254252e-01 1.14253402e+00 1.66726559e-01 6.25526845e-01 6.47342086e-01 -8.40860963e-01 -1.34425545e+00 -6.87500954e-01 -6.04338467e-01 1.62299611e-02 5.57125151e-01 -8.48417521e-01 -5.59174344e-02 -6.48439109e-01]
[4.17844820022583, 1.9549046754837036]
478d88b0-3989-423c-8481-a3cafef254d6
an-approach-based-on-combination-of-features
2004.11699
null
https://arxiv.org/abs/2004.11699v1
https://arxiv.org/pdf/2004.11699v1.pdf
An approach based on Combination of Features for automatic news retrieval
Nowadays, according to the increasingly increasing information, the importance of its presentation is also increasing. The internet has become one of the main sources of information for users and their favorite topics. It also provides access to more information. Understanding this information is very important for providing the best set of information resources for users. Content providers now need a precise and efficient way to retrieve news with the least human help. Data mining has led to the emergence of new methods for detecting related and unrelated documents. Although the conceptual relationship between documents may be negligible, it is important to provide useful information and relevant content to users. In this paper, a new approach based on the Combination of Features (CoF) for information retrieval operations is introduced. Along with introducing this new approach, we proposed a dataset by identifying the most commonly used keywords in documents and using the most appropriate documents to help them with the abundance of vocabulary. Then, using the proposed approach, techniques of text categorization, evaluation criteria and ranking algorithms, the data were analyzed and examined. The evaluation results show that using the combination of features approach improves the quality and effects on efficient ranking.
['Sasan Harifi', 'Mohammad Moradi', 'Elham Ghanbari', 'Mehrdad Maeen']
2020-04-16
null
null
null
null
['text-categorization']
['natural-language-processing']
[-1.90254763e-01 -4.41615015e-01 -3.51503491e-01 -1.08725682e-01 -3.91352624e-01 -6.13112748e-01 6.38200581e-01 1.11772323e+00 -5.19145310e-01 6.32445455e-01 3.48846018e-01 1.51772320e-01 -9.02830482e-01 -1.00198507e+00 1.93540212e-02 -4.58542943e-01 1.70693528e-02 6.33446455e-01 4.18743193e-01 -4.83827412e-01 1.09476304e+00 6.64771438e-01 -2.27795362e+00 2.81874120e-01 9.82754886e-01 1.15366387e+00 7.46210158e-01 1.54148445e-01 -8.66611123e-01 5.94066679e-01 -6.28106177e-01 -9.96980667e-02 1.07562937e-01 -3.65762770e-01 -7.88100600e-01 -1.39094472e-01 -2.29999378e-01 -3.09541523e-01 2.78238356e-01 8.50282371e-01 2.63434261e-01 4.82889831e-01 8.02800655e-01 -8.45341146e-01 -3.49826992e-01 5.03465712e-01 -5.55708289e-01 3.77993733e-01 5.48403144e-01 -9.06298280e-01 9.86194015e-01 -5.84997773e-01 5.69660544e-01 1.08455300e+00 2.46178601e-02 -1.12463944e-01 -5.33997416e-01 -3.81303698e-01 2.34369170e-02 6.48067355e-01 -1.26362920e+00 -1.46649675e-02 7.23835945e-01 -4.08445716e-01 6.32365108e-01 3.34994674e-01 6.81994617e-01 3.12716395e-01 1.38211086e-01 2.84730434e-01 9.58169162e-01 -9.54140663e-01 6.17660843e-02 8.91731977e-01 9.01991963e-01 3.41621995e-01 6.73539698e-01 -4.05910999e-01 -2.90358186e-01 -2.95255870e-01 -8.03677365e-02 4.31097150e-01 -2.75664926e-01 -1.64070167e-03 -7.26382792e-01 8.17086279e-01 1.77166060e-01 1.15481794e+00 -5.75976253e-01 -5.48183382e-01 4.84103322e-01 7.93911070e-02 2.99309373e-01 6.85789287e-01 -2.30419457e-01 -1.62111431e-01 -7.53463626e-01 2.58352607e-01 8.41405809e-01 5.03148735e-01 6.68833494e-01 -6.36697412e-01 3.93666364e-02 8.80129516e-01 4.42601293e-01 3.15460145e-01 9.51705754e-01 -4.34571534e-01 1.65435582e-01 1.34552777e+00 1.04929440e-01 -1.74081743e+00 -4.31287646e-01 -5.71398437e-01 -5.22096634e-01 -1.92688286e-01 8.23860839e-02 3.70907336e-01 -4.43360955e-01 1.17749500e+00 2.22686425e-01 -6.01109624e-01 2.04628962e-03 6.28995597e-01 8.30912232e-01 9.93894577e-01 -7.98165128e-02 -4.59178090e-01 1.62561309e+00 -5.10574520e-01 -9.87980723e-01 4.94546086e-01 3.73376757e-01 -1.32229555e+00 8.31585348e-01 5.06282389e-01 -5.96337318e-01 -4.78807271e-01 -8.21781039e-01 9.24536362e-02 -8.92253101e-01 2.10531026e-01 4.75071996e-01 2.86829561e-01 -8.67926836e-01 5.88635325e-01 -2.17795044e-01 -8.67067754e-01 -2.27735922e-01 2.46527731e-01 -1.72462091e-01 -1.93167225e-01 -1.27033305e+00 1.06154692e+00 7.13878930e-01 -3.50624561e-01 -1.19074583e-01 -2.02876776e-01 -1.81448162e-01 1.98927313e-01 4.22044069e-01 -2.47199759e-01 7.84388006e-01 -9.34933007e-01 -8.01567078e-01 4.01402622e-01 -5.93477748e-02 -3.64495486e-01 1.08558252e-01 -2.31765136e-01 -6.15178227e-01 4.99399364e-01 2.87074327e-01 -2.70775650e-02 5.02308309e-01 -1.08807087e+00 -1.16687346e+00 -5.03174007e-01 -3.35190818e-02 3.86416465e-01 -8.80784035e-01 4.58839357e-01 -5.83722413e-01 -2.80573845e-01 2.46106580e-01 -4.32287306e-01 2.33913258e-01 -4.33242947e-01 -9.64809582e-02 -5.61395288e-01 9.36577559e-01 -8.39353800e-01 1.57494795e+00 -1.92646861e+00 7.18212351e-02 7.71480262e-01 2.50524700e-01 1.85317189e-01 5.69658399e-01 1.01767325e+00 4.33113933e-01 2.35548303e-01 2.18481451e-01 4.17630553e-01 -3.07393700e-01 -3.21719088e-02 -1.78367093e-01 -9.77168232e-02 -6.36917710e-01 3.53838503e-02 -7.09197462e-01 -7.35693872e-01 1.90105990e-01 4.44444299e-01 -1.12487182e-01 2.20253348e-01 7.58295432e-02 8.17515701e-02 -1.01856530e+00 5.75610280e-01 1.90366864e-01 -1.82762995e-01 2.48990487e-02 -3.20997685e-01 -4.57063913e-01 4.38409075e-02 -1.21552360e+00 9.24576879e-01 -5.68084002e-01 5.55535913e-01 -3.39611709e-01 -1.14763463e+00 1.07373285e+00 3.85012776e-01 6.21953905e-01 -8.41082692e-01 5.35162985e-01 5.53097248e-01 -4.74307500e-02 -7.63616383e-01 5.33441961e-01 4.20320481e-01 1.60942167e-01 5.05897582e-01 -2.89582580e-01 3.38281691e-01 8.44898701e-01 5.21396637e-01 5.34081936e-01 -2.64547169e-01 7.05528736e-01 -3.21217656e-01 8.56738985e-01 2.42227972e-01 -4.89652064e-03 4.58031863e-01 3.03155452e-01 -5.27359210e-02 1.70689404e-01 -3.59113842e-01 -7.99867451e-01 -1.97237000e-01 -2.23670557e-01 7.75177896e-01 1.95404589e-01 -4.19432104e-01 -6.71405494e-01 -2.04496473e-01 2.40521096e-02 6.04557276e-01 -4.34381336e-01 6.75127134e-02 -8.41553807e-02 -4.75738794e-01 -2.85751168e-02 -2.79138237e-01 7.05041409e-01 -8.99251878e-01 -6.50756299e-01 2.24617377e-01 -3.34410906e-01 -4.26339537e-01 -4.85183634e-02 -5.09214066e-02 -1.01400149e+00 -1.10622609e+00 -8.61500502e-01 -7.29929268e-01 7.17219710e-01 8.12436879e-01 7.65417695e-01 5.52143514e-01 -2.21758887e-01 3.28840554e-01 -1.14313364e+00 -5.07712841e-01 -4.12282825e-01 4.22512978e-01 -3.69187258e-02 -4.91960980e-02 4.96948987e-01 -3.03992301e-01 -4.74803001e-01 1.76128462e-01 -1.00288844e+00 -2.13661596e-01 5.03361821e-01 5.53517878e-01 3.64045173e-01 7.95531034e-01 6.70636475e-01 -8.84775460e-01 9.97365236e-01 -6.19276106e-01 -5.64736843e-01 4.76607472e-01 -1.20361948e+00 2.87097186e-01 6.93009734e-01 -8.98424909e-03 -1.13988769e+00 -6.54904068e-01 9.04660746e-02 3.40473145e-01 -1.18223898e-01 8.45133901e-01 -7.72633702e-02 -2.31050793e-02 4.96658802e-01 1.18383139e-01 -1.52525499e-01 -8.37265134e-01 -6.70857355e-02 1.11613786e+00 -2.71852911e-01 -3.22083265e-01 3.73978347e-01 1.74213007e-01 2.95616034e-02 -1.10937405e+00 -7.26591587e-01 -1.11879182e+00 -4.67299342e-01 -4.50744301e-01 6.69566929e-01 -1.81038648e-01 -6.27803028e-01 1.10853262e-01 -9.86860216e-01 1.00690269e+00 1.64731517e-01 6.39514863e-01 2.11360261e-01 6.46805704e-01 1.72998924e-02 -9.60119069e-01 -4.43852574e-01 -9.11173999e-01 3.28357458e-01 3.62860620e-01 -2.41101697e-01 -7.04304934e-01 -7.06535531e-04 4.40556228e-01 7.09709823e-01 -6.28870279e-02 1.31479394e+00 -1.20607495e+00 -3.16643447e-01 -6.25141084e-01 -1.22470997e-01 1.65027767e-01 4.72565979e-01 2.27912888e-02 -4.39458281e-01 1.30970359e-01 -2.33188621e-03 6.51924908e-02 7.74613082e-01 2.19288170e-01 1.03699410e+00 -5.58273911e-01 -6.25273287e-01 -4.97499257e-02 1.53839707e+00 9.32290316e-01 4.15374428e-01 8.64902079e-01 3.49695891e-01 1.02291775e+00 9.37720656e-01 6.73244596e-01 1.04197741e-01 6.97717309e-01 2.43102655e-01 2.15855390e-01 3.14404458e-01 -4.84336987e-02 -2.44633898e-01 1.00199604e+00 -2.32861593e-01 -3.37077588e-01 -7.25106001e-01 3.81066382e-01 -1.56595957e+00 -1.21745121e+00 -6.34900630e-02 2.39217615e+00 4.07545716e-01 7.91808292e-02 1.27001092e-01 7.23141372e-01 7.25313187e-01 -2.64267027e-01 8.70785862e-02 -3.82210791e-01 1.30688831e-01 9.28813592e-03 3.08735877e-01 3.50481361e-01 -7.51100659e-01 3.72183532e-01 4.79858398e+00 9.23998296e-01 -9.88617539e-01 -1.83641523e-01 5.16914487e-01 3.23992312e-01 -3.73380959e-01 5.96807897e-02 -1.04466677e+00 7.13806033e-01 7.22576201e-01 -7.87793040e-01 2.78226823e-01 9.21426237e-01 5.10234952e-01 -6.88653588e-01 -4.21301603e-01 7.27067232e-01 2.71126866e-01 -1.12525427e+00 4.43085641e-01 -2.61184052e-02 4.40935731e-01 -6.55783534e-01 -2.21480414e-01 -1.55706510e-01 -3.02109838e-01 -3.88194203e-01 4.25780386e-01 6.46839440e-01 -1.43317701e-02 -1.14425039e+00 1.19199634e+00 5.25575161e-01 -1.08808589e+00 -1.88927129e-01 -3.62539500e-01 -6.67650178e-02 -1.74730122e-01 5.77393293e-01 -6.57200634e-01 7.98278332e-01 7.84829497e-01 5.40074468e-01 -5.22943079e-01 1.37452865e+00 1.86174482e-01 3.34246010e-01 -2.66553551e-01 -8.43345702e-01 8.39259848e-02 -4.09038872e-01 3.32811922e-01 9.75129247e-01 6.04970157e-01 1.30058482e-01 4.67240997e-02 1.04747504e-01 2.52560914e-01 1.16738248e+00 -6.47134244e-01 -2.14706063e-01 7.00660825e-01 1.34053862e+00 -1.32384205e+00 -4.21448767e-01 -2.59723991e-01 5.59647918e-01 -2.03014258e-03 -3.97894792e-02 -2.71567225e-01 -8.67181540e-01 -9.19548422e-02 6.06931269e-01 -2.04211041e-01 -1.00421384e-01 2.53651410e-01 -6.16529644e-01 -4.28790972e-02 -7.95593321e-01 5.98533571e-01 -2.83593088e-01 -1.00011885e+00 7.94900954e-01 4.18539613e-01 -1.34065974e+00 -2.17942595e-01 -4.44752485e-01 -1.04131006e-01 7.86889255e-01 -1.45132434e+00 -6.43656611e-01 -3.64313543e-01 4.04937595e-01 5.58985174e-01 -3.61057580e-01 7.86311150e-01 5.35301507e-01 -1.68951556e-01 -4.03755568e-02 6.93327487e-01 -3.36442709e-01 6.91865385e-01 -8.54958892e-01 -6.60663724e-01 6.60283983e-01 6.56404048e-02 9.18521643e-01 6.41129911e-01 -6.68835938e-01 -1.11261177e+00 -2.59140879e-01 1.26010144e+00 1.59503311e-01 4.11214441e-01 3.24171364e-01 -8.39222252e-01 4.70438367e-03 3.15686762e-01 -9.56137121e-01 8.70836854e-01 5.83791025e-02 1.41431078e-01 -5.13140440e-01 -1.15274262e+00 4.45258737e-01 2.10976020e-01 -1.14575950e-02 -8.38287354e-01 3.06602627e-01 5.03357112e-01 2.96799958e-01 -4.03810024e-01 -7.43278340e-02 6.16702080e-01 -9.94351149e-01 6.80864573e-01 -2.31960177e-01 1.62475750e-01 -4.54949945e-01 -1.25688732e-01 -1.13236785e+00 -2.98094451e-01 -3.38756293e-02 1.55820563e-01 1.46010029e+00 3.98487747e-01 -5.93326688e-01 3.75171244e-01 3.92997056e-01 2.51886666e-01 -5.83483875e-01 -4.62148547e-01 -3.72702688e-01 -6.21794820e-01 1.09895607e-02 4.14175242e-01 8.03310275e-01 4.30320464e-02 1.38899654e-01 -1.88971028e-01 -2.40485132e-01 4.82081622e-01 2.69692093e-01 4.27098423e-01 -1.83751321e+00 -1.52298644e-01 -5.93127072e-01 -4.88614231e-01 -5.21253645e-01 -4.45848107e-01 -9.00843084e-01 -4.68755931e-01 -2.02231121e+00 4.11608577e-01 -3.23091835e-01 -3.87544662e-01 3.64864059e-02 -6.65806606e-02 -8.28482881e-02 9.31165218e-02 7.49217689e-01 -4.26193297e-01 -3.64253484e-02 9.71671224e-01 1.28868192e-01 -2.43017286e-01 2.59729445e-01 -9.01892900e-01 6.17716968e-01 7.92212784e-01 -6.49585545e-01 -6.13347828e-01 -4.06471919e-03 6.28649294e-01 -3.13131958e-01 -2.95273513e-01 -9.61551845e-01 3.11354786e-01 -1.93528220e-01 4.16951239e-01 -7.75334001e-01 2.62413453e-02 -1.29100120e+00 1.80014819e-01 4.27496135e-01 -4.59527671e-01 3.16865236e-01 -3.65490019e-02 3.42951536e-01 -5.62055945e-01 -8.22686195e-01 4.45007175e-01 -3.31062376e-01 -8.42465937e-01 -1.67683691e-01 -5.77962697e-01 -2.78393388e-01 1.09069967e+00 -1.66216865e-01 -1.38640791e-01 -4.94928151e-01 -3.47139478e-01 1.29542753e-01 3.12361151e-01 5.18000245e-01 4.76251453e-01 -9.14989829e-01 -3.42384189e-01 -2.49738738e-01 2.45950773e-01 -5.88514090e-01 2.53473371e-01 4.37002420e-01 -7.12173641e-01 6.23654127e-01 -4.17295098e-01 -1.09649003e-02 -1.60425329e+00 6.04124784e-01 -3.31517190e-01 -3.91462684e-01 -2.69208461e-01 4.65883583e-01 -2.95983344e-01 2.62802184e-01 4.40380126e-01 -9.08456668e-02 -1.38119185e+00 7.22662210e-01 8.36441398e-01 7.22529173e-01 2.72829294e-01 -7.14765549e-01 -1.84343323e-01 8.83264124e-01 -4.28124994e-01 -2.09089160e-01 1.09089386e+00 -2.49987274e-01 -5.34555912e-01 4.97339219e-01 9.97710466e-01 3.55575740e-01 3.69813666e-02 -9.85075831e-02 4.48891193e-01 -5.80694139e-01 3.27169865e-01 -8.66568685e-01 -7.30693042e-01 7.07807302e-01 5.81034839e-01 8.23153079e-01 1.33897817e+00 -2.00911865e-01 4.79034871e-01 5.97710013e-01 5.51546812e-01 -1.26875412e+00 -9.49829221e-02 2.53423989e-01 6.07914090e-01 -1.13281858e+00 3.59393716e-01 -3.29861671e-01 -3.97685826e-01 1.35216713e+00 1.32656112e-01 2.40189537e-01 9.63897288e-01 -6.20968826e-02 2.21210122e-02 -2.32964680e-01 -3.76117706e-01 -3.45973670e-01 5.35207808e-01 2.49071509e-01 8.02068114e-01 -3.94925535e-01 -1.45503175e+00 5.11430502e-01 5.16399518e-02 -1.89825311e-01 2.66099095e-01 9.93710935e-01 -1.18703330e+00 -1.54894125e+00 -6.21309519e-01 9.07034278e-01 -8.71578395e-01 9.45354346e-03 -4.46641803e-01 8.21942627e-01 5.03294170e-02 1.22434413e+00 -1.98441133e-01 -2.62824684e-01 2.43366033e-01 -4.29583453e-02 -3.36489640e-03 -4.49172437e-01 -4.53186095e-01 1.13865286e-02 -9.70624387e-03 4.29677224e-04 -6.14047706e-01 -3.42450649e-01 -1.17836738e+00 -2.45508611e-01 -6.77036405e-01 1.00966930e+00 1.12740481e+00 9.43482935e-01 3.80131811e-01 4.09669787e-01 9.14812267e-01 -4.20272827e-01 -2.03279868e-01 -8.89374971e-01 -5.57777882e-01 4.55113113e-01 -2.87593931e-01 -9.13675725e-01 -3.72630030e-01 -2.19594628e-01]
[10.493752479553223, 7.4236650466918945]
98d73b06-480c-4358-b362-ed2738644578
a-simple-framework-for-contrastive-learning
2002.05709
null
https://arxiv.org/abs/2002.05709v3
https://arxiv.org/pdf/2002.05709v3.pdf
A Simple Framework for Contrastive Learning of Visual Representations
This paper presents SimCLR: a simple framework for contrastive learning of visual representations. We simplify recently proposed contrastive self-supervised learning algorithms without requiring specialized architectures or a memory bank. In order to understand what enables the contrastive prediction tasks to learn useful representations, we systematically study the major components of our framework. We show that (1) composition of data augmentations plays a critical role in defining effective predictive tasks, (2) introducing a learnable nonlinear transformation between the representation and the contrastive loss substantially improves the quality of the learned representations, and (3) contrastive learning benefits from larger batch sizes and more training steps compared to supervised learning. By combining these findings, we are able to considerably outperform previous methods for self-supervised and semi-supervised learning on ImageNet. A linear classifier trained on self-supervised representations learned by SimCLR achieves 76.5% top-1 accuracy, which is a 7% relative improvement over previous state-of-the-art, matching the performance of a supervised ResNet-50. When fine-tuned on only 1% of the labels, we achieve 85.8% top-5 accuracy, outperforming AlexNet with 100X fewer labels.
['Simon Kornblith', 'Mohammad Norouzi', 'Geoffrey Hinton', 'Ting Chen']
2020-02-13
null
https://proceedings.icml.cc/static/paper_files/icml/2020/6165-Paper.pdf
https://proceedings.icml.cc/static/paper_files/icml/2020/6165-Paper.pdf
icml-2020-1
['self-supervised-image-classification', 'self-supervised-person-re-identification']
['computer-vision', 'computer-vision']
[ 3.84483397e-01 2.34936565e-01 -5.57084441e-01 -4.10957903e-01 -5.69587231e-01 -4.17355835e-01 7.85861731e-01 1.71490699e-01 -5.66242456e-01 7.37413704e-01 2.20646068e-01 -1.73064992e-01 1.17406771e-01 -5.75235248e-01 -9.38100398e-01 -4.95120734e-01 -1.59925133e-01 4.20404106e-01 2.34736204e-01 -2.64075458e-01 1.42413706e-01 5.37328780e-01 -1.78797948e+00 4.29331601e-01 7.55000412e-01 1.37688172e+00 2.38467693e-01 6.42754495e-01 -4.81720157e-02 1.34337616e+00 -3.77618968e-01 -6.74735010e-02 4.06666875e-01 -3.25264812e-01 -9.88348663e-01 1.59235429e-02 9.95890021e-01 -2.77636230e-01 -4.72918272e-01 7.91545749e-01 1.06769145e-01 1.51682392e-01 9.28614020e-01 -1.07685959e+00 -7.35227108e-01 6.08561754e-01 -4.40273225e-01 4.59637046e-01 -1.57267630e-01 2.61065662e-01 1.28458810e+00 -8.84237885e-01 4.87947375e-01 1.08328784e+00 6.07090771e-01 6.73319638e-01 -1.49465072e+00 -8.76571655e-01 2.49446422e-01 3.12195141e-02 -1.16069639e+00 -7.11356819e-01 6.15335286e-01 -4.52959985e-01 1.18688858e+00 7.78871402e-02 6.11781359e-01 8.33444893e-01 -3.31615098e-02 7.24720597e-01 1.41347814e+00 -6.04707658e-01 2.22147349e-03 3.68715137e-01 3.67960274e-01 8.22769701e-01 6.55082539e-02 3.26128900e-01 -4.61089581e-01 1.24501720e-01 7.16596901e-01 1.86117366e-01 1.26689542e-02 -3.07704628e-01 -9.50371623e-01 8.15790594e-01 1.01646972e+00 3.21296215e-01 -1.26687273e-01 3.79083872e-01 5.09564102e-01 6.04160726e-01 8.51670980e-01 6.93619311e-01 -4.60382640e-01 1.90233752e-01 -1.06967378e+00 -2.02905580e-01 5.02156258e-01 8.27279747e-01 1.00666928e+00 5.85146010e-01 -1.04486018e-01 9.52312708e-01 4.05052826e-02 4.65716422e-01 7.00603247e-01 -8.29669416e-01 2.16352105e-01 7.17749536e-01 -3.21314543e-01 -4.89669651e-01 -4.45311904e-01 -8.59858930e-01 -8.94096792e-01 4.43978727e-01 3.13377917e-01 1.60262376e-01 -1.11171997e+00 1.68061602e+00 -3.41156572e-01 1.12156481e-01 -1.77224609e-03 4.08238828e-01 8.63804758e-01 6.50740743e-01 4.36750263e-01 -3.22384506e-01 1.02886462e+00 -1.21679294e+00 -2.92032242e-01 -4.74656403e-01 6.35923803e-01 -5.32684982e-01 1.15579486e+00 1.82506189e-01 -1.23201001e+00 -9.45386767e-01 -1.28510714e+00 -2.51917467e-02 -3.83660644e-01 1.81120425e-01 6.73680782e-01 3.83395851e-01 -1.37482738e+00 8.77079964e-01 -5.97339213e-01 -1.22134447e-01 9.36025500e-01 4.96020764e-01 -3.94340366e-01 -4.32444699e-02 -7.56110430e-01 1.05295312e+00 4.63438570e-01 -5.95858216e-01 -1.05598521e+00 -9.65763152e-01 -8.83750737e-01 1.85833931e-01 2.93243863e-02 -4.18575048e-01 1.18970096e+00 -1.53159618e+00 -1.24897981e+00 1.31975782e+00 -2.04168916e-01 -9.31535363e-01 4.03652459e-01 -2.61944324e-01 -2.58088559e-01 2.87608176e-01 -2.81014591e-02 1.07550371e+00 1.10383105e+00 -1.27975953e+00 -5.99371612e-01 -8.73128250e-02 -5.20297699e-02 2.91856676e-01 -5.96282840e-01 -1.19797915e-01 -1.10630706e-01 -7.26117432e-01 -1.36984721e-01 -8.31870437e-01 -2.47435495e-01 8.53838846e-02 -7.39700869e-02 -2.10836604e-01 6.16044462e-01 -3.71786386e-01 8.56046081e-01 -2.18379354e+00 -1.57143757e-01 4.44634072e-02 5.74192941e-01 5.22730887e-01 -3.97353590e-01 -1.02755509e-01 -4.27065104e-01 6.19070157e-02 -1.53401077e-01 -5.16782403e-01 -2.98936009e-01 2.00751089e-02 -5.96575081e-01 4.20584619e-01 4.04534608e-01 1.06296408e+00 -8.86398256e-01 -3.86526108e-01 3.65755409e-01 3.17709178e-01 -4.55030859e-01 2.98238635e-01 -1.44695565e-01 3.05349648e-01 4.03834060e-02 3.55558753e-01 4.58686501e-01 -6.13156319e-01 9.39715207e-02 -2.33354807e-01 3.76734883e-02 6.13422513e-01 -6.36416912e-01 1.30708671e+00 -6.93035901e-01 8.85575414e-01 -3.69400203e-01 -1.32362556e+00 1.13024771e+00 -5.74649218e-03 3.45140219e-01 -1.03186715e+00 4.29806020e-03 1.26157075e-01 -1.50417000e-01 1.13960579e-01 3.82569432e-01 -2.06502587e-01 4.24695581e-01 5.04259586e-01 6.19912267e-01 8.56083930e-02 8.95404145e-02 2.03072906e-01 1.05865455e+00 3.24320942e-02 5.45275688e-01 -4.60718960e-01 3.48210663e-01 3.10208852e-04 2.29233444e-01 8.65744293e-01 -1.11773595e-01 5.03146470e-01 2.69783825e-01 -6.59933805e-01 -1.05778491e+00 -9.88062501e-01 -2.35749289e-01 1.63287222e+00 -2.06076458e-01 -4.01813984e-01 -4.08527672e-01 -8.42971146e-01 -4.57701795e-02 6.11373067e-01 -7.92604804e-01 -3.38974893e-01 -6.00605905e-01 -6.77523255e-01 3.67172480e-01 8.34756792e-01 6.81644440e-01 -1.09892917e+00 -3.01899880e-01 -6.70567155e-02 2.84782737e-01 -1.16001379e+00 -1.58490583e-01 6.35427773e-01 -1.22663593e+00 -9.47776735e-01 -4.88986880e-01 -1.05084240e+00 8.56938601e-01 4.98553842e-01 1.53555799e+00 3.00585598e-01 -1.93785995e-01 3.63104582e-01 -1.76388547e-01 -2.38537803e-01 -6.11548603e-01 3.50503981e-01 -1.78591553e-02 -2.70987451e-01 1.23458654e-01 -7.61653066e-01 -4.85155225e-01 1.33193493e-01 -6.40268624e-01 1.42593876e-01 7.27554083e-01 9.52481985e-01 6.71674371e-01 -1.52468130e-01 6.86030805e-01 -1.29860699e+00 2.28234366e-01 -3.74047220e-01 -5.77245831e-01 1.07205078e-01 -9.97600317e-01 4.11764652e-01 9.15526986e-01 -6.35481954e-01 -8.86751950e-01 2.38275111e-01 -1.16034687e-01 -4.54803348e-01 -1.29909471e-01 1.20034777e-01 3.73605311e-01 -2.80079484e-01 1.08577561e+00 3.53948474e-01 4.49053235e-02 -2.65120029e-01 5.46452045e-01 4.30117965e-01 4.79286969e-01 -3.34010839e-01 9.30898845e-01 4.43831772e-01 -4.92897816e-02 -7.32450366e-01 -1.17302120e+00 -5.36930919e-01 -7.51502812e-01 -5.26044518e-02 5.26907384e-01 -1.24681211e+00 -4.67387855e-01 2.59545326e-01 -6.89533532e-01 -7.82676756e-01 -7.79820681e-01 2.34933913e-01 -6.42485559e-01 -2.40193550e-02 -7.85374880e-01 -6.89491034e-01 -4.99197006e-01 -8.17462385e-01 5.55804372e-01 1.03401102e-01 -1.84459150e-01 -1.15624619e+00 6.50098398e-02 3.23030412e-01 6.80894315e-01 -9.25768241e-02 8.14274490e-01 -9.68832314e-01 -3.96545261e-01 -6.69635460e-02 -4.50326949e-01 8.32667530e-01 1.11143030e-01 -3.68373394e-01 -1.37308598e+00 -4.54330027e-01 -2.60003120e-01 -9.99892533e-01 1.42099214e+00 3.54800671e-01 1.22251713e+00 -3.98894012e-01 -1.97492823e-01 7.49791145e-01 1.36065030e+00 -1.56995624e-01 7.33354568e-01 3.83439124e-01 7.36346126e-01 2.92456627e-01 2.70857453e-01 1.76246494e-01 2.46160045e-01 4.37228441e-01 3.88051510e-01 -4.24930364e-01 -5.77228904e-01 -4.85943675e-01 3.57242823e-01 7.70084858e-01 -1.65702462e-01 3.75969648e-01 -7.30351567e-01 3.67248297e-01 -1.54524004e+00 -8.68197739e-01 2.22460568e-01 2.21374869e+00 1.07576025e+00 4.97132927e-01 2.49198034e-01 9.60898399e-02 4.37265724e-01 4.62616026e-01 -7.60620415e-01 -1.86395913e-01 -2.51786172e-01 6.54279172e-01 7.65996456e-01 5.05110383e-01 -1.32236421e+00 1.25450528e+00 7.68723059e+00 8.62144530e-01 -1.20401788e+00 9.46545228e-02 9.33628678e-01 1.02559691e-02 -4.98479642e-02 -8.00451338e-02 -8.71794999e-01 1.88795105e-01 1.09767067e+00 9.69971493e-02 4.87422585e-01 1.17475426e+00 -3.18433583e-01 2.05635637e-01 -1.29893982e+00 1.06081414e+00 3.69826108e-01 -1.50582480e+00 2.38790140e-01 -6.48836121e-02 1.04525793e+00 3.16024333e-01 3.68398458e-01 6.47499323e-01 5.04461408e-01 -1.35935783e+00 5.60923934e-01 3.82129014e-01 1.05152392e+00 -5.88523269e-01 3.92589718e-01 2.37021491e-01 -1.07315826e+00 -1.99640796e-01 -6.95333719e-01 -2.51597881e-01 -4.95407760e-01 4.79696721e-01 -7.88180172e-01 -4.77489457e-02 6.35844469e-01 1.09414470e+00 -9.24339831e-01 8.58963907e-01 -3.63412350e-01 9.83246326e-01 -1.05366655e-01 8.63360390e-02 1.27402619e-01 2.72694737e-01 1.69628114e-01 1.45265520e+00 -2.66106427e-01 -1.33742332e-01 1.56054333e-01 8.28890324e-01 -6.01112664e-01 1.28824189e-01 -5.67100763e-01 5.17505594e-02 4.21928704e-01 1.20914316e+00 -5.37007511e-01 -6.63549781e-01 -3.75228584e-01 8.83923173e-01 8.78189445e-01 2.43538260e-01 -4.01527256e-01 -9.49582905e-02 4.28081483e-01 2.71381795e-01 3.40352952e-01 -7.29805604e-02 -3.57309759e-01 -1.13888800e+00 -2.76625723e-01 -9.52866435e-01 3.36953551e-01 -6.83442593e-01 -1.30925512e+00 8.13303709e-01 -1.28382474e-01 -1.31244814e+00 -3.11035603e-01 -8.30788851e-01 -5.45708835e-01 5.92482209e-01 -1.79675376e+00 -1.16980219e+00 -3.86596441e-01 5.57767451e-01 5.40779233e-01 -6.21938050e-01 1.06515741e+00 -7.14109242e-02 -3.34335655e-01 8.43090951e-01 1.79377407e-01 1.40748501e-01 7.32595861e-01 -1.27500820e+00 4.22228515e-01 6.36235178e-01 4.26703393e-01 3.80702317e-01 5.01988769e-01 -2.40787089e-01 -9.38023210e-01 -1.16857862e+00 6.68993235e-01 -3.15007865e-01 6.81909919e-01 -3.51702005e-01 -9.54655111e-01 9.33709025e-01 2.23859757e-01 4.74686056e-01 6.06724024e-01 2.59535253e-01 -9.81252491e-01 -4.26884234e-01 -1.03007436e+00 4.33618784e-01 1.10993087e+00 -6.74139142e-01 -6.32001638e-01 4.37649846e-01 7.07347572e-01 -1.27165765e-01 -6.67534292e-01 3.60802412e-01 3.26923162e-01 -8.75019550e-01 1.06117594e+00 -8.46463501e-01 6.04278922e-01 1.64954200e-01 -1.52015105e-01 -1.33831584e+00 -5.99105239e-01 -1.90083817e-01 -2.58956850e-01 9.59422350e-01 4.30873275e-01 -6.38132930e-01 7.60627687e-01 2.83607811e-01 -6.51637539e-02 -7.28545547e-01 -5.34049749e-01 -8.80576611e-01 3.70747417e-01 -3.49305570e-01 4.92456146e-02 9.68408704e-01 -7.32825324e-02 6.37588561e-01 -4.75670964e-01 -3.80011916e-01 7.20370173e-01 -3.08590177e-02 6.03454828e-01 -1.39962757e+00 -2.29876727e-01 -4.97040510e-01 -4.99736428e-01 -1.35325873e+00 5.62453389e-01 -1.20593739e+00 -1.90028235e-01 -1.28212798e+00 5.28839469e-01 -6.76982284e-01 -6.84755087e-01 8.40959966e-01 -8.90536457e-02 5.89589715e-01 3.30505818e-01 5.39974809e-01 -8.38632762e-01 4.87990916e-01 9.33567584e-01 -2.91795015e-01 -1.91796884e-01 -1.62057251e-01 -8.63940775e-01 7.79071569e-01 9.20517147e-01 -4.10975486e-01 -4.87257808e-01 -1.57927439e-01 -2.76708845e-02 -4.53052729e-01 2.83774227e-01 -1.20563841e+00 1.81594938e-01 9.44785122e-03 6.81358516e-01 -2.11424977e-01 3.35895836e-01 -5.29819548e-01 -5.28014004e-01 6.09610319e-01 -9.16878581e-01 -2.23781198e-01 4.23593700e-01 3.82752389e-01 -1.72039494e-01 -2.93706506e-01 1.24230564e+00 -2.39171743e-01 -8.27917159e-01 1.70853615e-01 -3.16898882e-01 3.36770386e-01 6.89353466e-01 -2.62296312e-02 -6.00852013e-01 -3.61052901e-01 -6.21728837e-01 -8.40921327e-02 3.62814903e-01 2.64235228e-01 6.83741331e-01 -1.33496094e+00 -6.62191451e-01 2.23957226e-01 2.06296697e-01 -3.38576376e-01 1.87036227e-02 4.51253921e-01 -3.72404009e-01 3.13182920e-01 -5.60230970e-01 -5.24257600e-01 -1.15356052e+00 5.19537628e-01 3.42096716e-01 -4.93798405e-01 -6.23674273e-01 8.20752740e-01 3.30080062e-01 -2.80874103e-01 3.94916385e-01 -1.40447661e-01 -3.59770536e-01 -4.40361258e-03 7.89080918e-01 2.50672311e-01 -6.54075146e-02 -6.63834155e-01 -2.01552227e-01 4.43954438e-01 -5.18541574e-01 1.76175490e-01 1.53445959e+00 2.06333607e-01 1.83840662e-01 5.16542971e-01 1.45788944e+00 -2.78542519e-01 -1.53059161e+00 -5.69749057e-01 -1.96893960e-01 -9.84620377e-02 2.66388416e-01 -7.28411257e-01 -1.18460071e+00 8.38124394e-01 7.88661838e-01 -4.57930528e-02 1.10621107e+00 1.07798390e-01 3.88189912e-01 6.16855025e-01 1.79482594e-01 -1.01535833e+00 5.26024759e-01 6.48335636e-01 8.50552857e-01 -1.53242660e+00 2.73129582e-01 -2.18127966e-01 -7.33956039e-01 1.03960013e+00 6.08727694e-01 -6.56800091e-01 7.21092105e-01 1.46401361e-01 -5.88360950e-02 -3.09813768e-02 -9.52337861e-01 -3.71348172e-01 6.40026093e-01 5.94780147e-01 5.33615232e-01 -6.75055310e-02 2.26730734e-01 3.16193312e-01 -2.44241804e-01 -2.01227218e-01 1.42207876e-01 8.15544665e-01 -6.86872661e-01 -7.34010279e-01 9.58703980e-02 7.78872371e-01 -3.09957206e-01 -4.62034911e-01 -2.28307456e-01 7.46601880e-01 -1.05764009e-01 6.95469558e-01 3.25627983e-01 -4.54030186e-01 2.60941625e-01 1.87609151e-01 5.04543662e-01 -7.01529324e-01 -6.57579422e-01 -3.01793277e-01 -1.13590419e-01 -4.56821561e-01 -5.37304819e-01 -2.89039642e-01 -1.12420762e+00 -2.55347162e-01 -1.94645345e-01 -2.70415395e-01 3.48240048e-01 9.17779267e-01 1.89168692e-01 5.41473150e-01 8.00270677e-01 -9.72835183e-01 -7.03429818e-01 -1.03727806e+00 -5.06001651e-01 5.19462168e-01 4.32356924e-01 -6.64218009e-01 -6.27019167e-01 2.16811046e-01]
[9.383039474487305, 2.6947383880615234]
f3fd3baa-5cb8-46cb-911c-eeb4e9cb9dbb
grafit-learning-fine-grained-image
2011.12982
null
https://arxiv.org/abs/2011.12982v1
https://arxiv.org/pdf/2011.12982v1.pdf
Grafit: Learning fine-grained image representations with coarse labels
This paper tackles the problem of learning a finer representation than the one provided by training labels. This enables fine-grained category retrieval of images in a collection annotated with coarse labels only. Our network is learned with a nearest-neighbor classifier objective, and an instance loss inspired by self-supervised learning. By jointly leveraging the coarse labels and the underlying fine-grained latent space, it significantly improves the accuracy of category-level retrieval methods. Our strategy outperforms all competing methods for retrieving or classifying images at a finer granularity than that available at train time. It also improves the accuracy for transfer learning tasks to fine-grained datasets, thereby establishing the new state of the art on five public benchmarks, like iNaturalist-2018.
['Hervé Jégou', 'Matthieu Cord', 'Matthijs Douze', 'Alexandre Sablayrolles', 'Hugo Touvron']
2020-11-25
null
http://openaccess.thecvf.com//content/ICCV2021/html/Touvron_Grafit_Learning_Fine-Grained_Image_Representations_With_Coarse_Labels_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Touvron_Grafit_Learning_Fine-Grained_Image_Representations_With_Coarse_Labels_ICCV_2021_paper.pdf
iccv-2021-1
['learning-with-coarse-labels']
['computer-vision']
[ 1.61255121e-01 -2.96903193e-01 -8.25279713e-01 -4.63316768e-01 -1.45844007e+00 -9.69617128e-01 9.68292952e-01 3.35068166e-01 -6.69902682e-01 6.54482007e-01 3.03338706e-01 2.24050865e-01 -5.04544079e-01 -7.28700757e-01 -8.69247317e-01 -6.16127372e-01 -4.70579602e-02 6.86508775e-01 -4.51911129e-02 4.47810799e-01 2.98843563e-01 3.21242720e-01 -1.99427366e+00 6.20260596e-01 6.13605082e-01 1.69828463e+00 3.16400407e-03 1.71917588e-01 -1.97318390e-01 7.03050613e-01 -2.49005198e-01 -2.47482598e-01 4.18979466e-01 3.68392706e-01 -1.13022280e+00 -1.29040971e-01 1.41547906e+00 -4.55053836e-01 -2.18472257e-01 8.77690017e-01 1.94601953e-01 1.61167786e-01 1.35008836e+00 -1.07924914e+00 -1.11579537e+00 3.42602700e-01 -3.76838237e-01 1.18932903e-01 8.02565739e-02 -1.51322126e-01 1.33290958e+00 -9.42469299e-01 5.77050209e-01 1.48566496e+00 6.63129210e-01 5.62026143e-01 -1.45466483e+00 -8.85771096e-01 3.63083392e-01 1.80597574e-01 -1.89846134e+00 -2.06695125e-01 3.06753457e-01 -6.17053628e-01 9.54718411e-01 2.29354165e-02 1.02906898e-01 1.15884280e+00 -5.56255095e-02 7.21327722e-01 1.34893131e+00 -2.22180471e-01 9.73009020e-02 5.20014130e-02 3.29392999e-01 6.90705180e-01 -5.03755547e-03 6.24673590e-02 -4.85710293e-01 -3.68415833e-01 5.58385909e-01 4.27067429e-01 -5.28419111e-03 -6.46815717e-01 -1.45987320e+00 8.92588615e-01 9.75862622e-01 3.42617720e-01 -2.15715453e-01 4.66678530e-01 5.54613531e-01 5.12408197e-01 1.06620932e+00 6.89984381e-01 -7.40242779e-01 3.34473401e-01 -1.37036657e+00 1.78638510e-02 6.79334044e-01 9.02491868e-01 8.89410377e-01 -5.89016736e-01 -7.71885931e-01 9.40617621e-01 1.84194043e-01 6.08760834e-01 5.70594609e-01 -1.08073533e+00 2.73196787e-01 5.36553681e-01 1.40496090e-01 -5.99635124e-01 1.64807625e-02 -6.94091678e-01 -8.98379624e-01 -4.26214151e-02 2.96055108e-01 6.26545966e-01 -1.25403798e+00 1.73766720e+00 8.51559192e-02 1.90724149e-01 -3.12447250e-01 8.04751992e-01 8.97822678e-01 6.68641627e-01 4.29427236e-01 1.61084220e-01 1.30524325e+00 -1.36614811e+00 -3.25946778e-01 -1.56375591e-03 4.11995351e-01 -5.27006447e-01 1.23709476e+00 3.46182555e-01 -5.88443696e-01 -7.55326450e-01 -7.27273643e-01 -2.61692971e-01 -8.78714859e-01 3.12074989e-01 7.47844577e-01 9.25025940e-02 -1.46291888e+00 7.32634783e-01 -3.96007031e-01 -2.33481705e-01 7.91680872e-01 1.55223772e-01 -6.43566608e-01 -3.71110767e-01 -1.16093278e+00 7.66513407e-01 4.64647204e-01 -2.46626943e-01 -1.36102271e+00 -1.05443919e+00 -6.95499301e-01 2.49181226e-01 1.54468656e-01 -7.76717961e-01 1.04856074e+00 -6.41394317e-01 -9.54204321e-01 1.53070533e+00 9.93900746e-02 -5.79453349e-01 4.99295175e-01 -2.95954525e-01 -6.02214132e-03 2.30925098e-01 7.81143785e-01 1.30768239e+00 1.16850233e+00 -1.07464635e+00 -7.00442493e-01 -3.34771991e-01 1.60750300e-01 1.29178345e-01 -6.21631026e-01 -4.26243573e-01 -5.39709032e-01 -7.18674839e-01 -1.44907370e-01 -9.41633403e-01 1.23777047e-01 3.79016608e-01 -3.20813775e-01 -1.11201787e+00 4.25680697e-01 -2.27698773e-01 9.00430262e-01 -2.13707495e+00 2.04418138e-01 -3.11680287e-02 4.06553417e-01 5.08331992e-02 -5.75812519e-01 3.25141609e-01 2.60120369e-02 2.46006384e-01 1.78456947e-01 -5.78132153e-01 5.22161543e-01 1.54222446e-02 -8.27888012e-01 5.06333768e-01 1.18544348e-01 1.23159695e+00 -9.75484073e-01 -6.86180294e-01 6.38463274e-02 4.21571225e-01 -3.43662024e-01 3.74843568e-01 -2.47191399e-01 1.27309263e-01 -6.46795213e-01 7.58521795e-01 4.84304756e-01 -7.79547989e-01 -5.10079563e-01 -4.27474171e-01 1.75076842e-01 9.04771388e-02 -6.68466210e-01 2.15647817e+00 -7.25076854e-01 4.40209508e-01 -3.78653675e-01 -8.26890528e-01 6.36323571e-01 5.26961051e-02 2.38050595e-01 -9.61922467e-01 -3.92309368e-01 2.67814279e-01 -9.09575582e-01 -1.37026697e-01 2.94328868e-01 6.84999600e-02 -4.92421478e-01 4.41133201e-01 5.07573724e-01 -7.47848675e-02 1.59055397e-01 4.18756247e-01 1.01579523e+00 1.74621210e-01 1.62374545e-02 -4.46833432e-01 2.17591614e-01 -1.47975951e-01 -3.32361050e-02 1.40453947e+00 -7.04910830e-02 5.02959847e-01 2.67430358e-02 -6.66875482e-01 -1.01864684e+00 -1.24784768e+00 -6.07256949e-01 1.88875151e+00 2.17271462e-01 -4.42558348e-01 -4.92223889e-01 -1.03172135e+00 5.98152518e-01 3.64321284e-02 -1.22243738e+00 -4.29472864e-01 8.41457471e-02 -2.37931922e-01 6.12448752e-01 4.74017322e-01 6.09483838e-01 -1.07846415e+00 4.07237038e-02 -1.06552064e-01 -4.47368383e-01 -1.03702521e+00 -5.85847199e-01 3.96729112e-01 -7.45965123e-01 -8.41878295e-01 -9.91495609e-01 -7.79056907e-01 4.95864570e-01 3.04403961e-01 1.76213586e+00 1.74198434e-01 -4.92819011e-01 5.66045582e-01 -3.79655600e-01 5.67356832e-02 2.03221425e-01 5.68437397e-01 6.92454427e-02 -9.27045650e-04 4.90164042e-01 -2.15632409e-01 -8.60046685e-01 1.50048792e-01 -9.08542454e-01 -4.10832494e-01 7.17224419e-01 1.05311954e+00 9.68888938e-01 1.04697153e-01 5.10958314e-01 -7.52184749e-01 4.28908378e-01 -5.79487741e-01 -5.84112823e-01 3.74953300e-01 -8.75218272e-01 3.41787010e-01 4.60079312e-01 -5.00272930e-01 -7.03673422e-01 -1.88453272e-01 3.18460822e-01 -7.67767310e-01 -5.58291376e-01 9.87736285e-02 2.80785710e-01 -2.32938558e-01 5.95235348e-01 6.94673806e-02 -4.62655067e-01 -9.17377710e-01 7.24920511e-01 7.23852575e-01 4.59046811e-01 -7.41688788e-01 8.08990955e-01 4.84659016e-01 -1.68089122e-01 -4.16120543e-04 -1.74229741e+00 -9.99659359e-01 -7.10563123e-01 2.47382283e-01 9.10583913e-01 -1.34978569e+00 -6.17109716e-01 3.51762533e-01 -7.66439080e-01 -4.35120344e-01 -4.81412172e-01 1.56436548e-01 -6.37118399e-01 -5.71119040e-02 -8.29010189e-01 -1.98027313e-01 -4.46100503e-01 -7.36621201e-01 2.07065606e+00 -7.79582262e-02 7.90885240e-02 -9.34640110e-01 1.36505052e-01 3.13401461e-01 6.07302606e-01 -4.74217758e-02 7.53320694e-01 -5.33388674e-01 -8.78853500e-01 -2.21667737e-01 -7.59633243e-01 2.98177242e-01 4.76353429e-02 -3.84673417e-01 -1.20693469e+00 -7.47913718e-01 -5.94891250e-01 -1.16044426e+00 1.83980250e+00 2.34780982e-01 1.59567940e+00 -5.34180343e-01 -6.91580772e-01 6.35861933e-01 1.63965285e+00 -7.86051154e-01 1.53926075e-01 5.26360571e-01 6.90691769e-01 5.53866863e-01 8.84270072e-01 1.01030551e-01 4.45158392e-01 8.52926612e-01 6.30535424e-01 -7.91237056e-02 -5.19429564e-01 -3.84820461e-01 -2.76124626e-01 2.83015281e-01 3.07946712e-01 -9.50561017e-02 -8.17215860e-01 8.92529011e-01 -1.77602279e+00 -9.07875419e-01 4.95424241e-01 2.02045059e+00 1.14399099e+00 -1.79466873e-01 -1.41787082e-01 -2.66413271e-01 6.02221429e-01 3.55700225e-01 -6.85653806e-01 9.81015526e-03 -7.40363402e-03 2.73355812e-01 6.54606640e-01 3.08079034e-01 -1.83258140e+00 1.14865017e+00 6.23821402e+00 1.47077131e+00 -8.31352055e-01 2.59385735e-01 6.33127213e-01 -1.57697365e-01 -1.55079111e-01 -4.38657939e-01 -1.01665998e+00 5.75425327e-01 7.93839335e-01 3.06749403e-01 5.86667538e-01 8.80612075e-01 -5.37420690e-01 2.26904944e-01 -1.57555246e+00 1.10672200e+00 2.06866771e-01 -1.44745779e+00 4.33886737e-01 6.16603121e-02 1.03423405e+00 6.16680145e-01 4.92940396e-01 7.14211881e-01 4.96020734e-01 -1.20347142e+00 7.20960200e-01 7.86006868e-01 1.42712736e+00 -3.79194796e-01 6.81877077e-01 3.76753300e-01 -1.15291202e+00 -3.07229400e-01 -6.13852739e-01 1.92792103e-01 -4.23124760e-01 5.81695914e-01 -6.23708963e-01 2.15618819e-01 1.23056149e+00 1.04572928e+00 -1.02609050e+00 1.15618086e+00 -9.90436599e-02 5.46138227e-01 -2.21237332e-01 3.26651633e-01 6.00146413e-01 4.25670236e-01 4.97187898e-02 1.48074377e+00 -1.12117037e-01 -3.32754344e-01 5.57175875e-01 6.72337890e-01 -7.35830426e-01 -1.66020647e-01 -6.94123924e-01 1.08144619e-01 6.40480518e-01 1.55357957e+00 -4.67319310e-01 -4.66760397e-01 5.68349026e-02 1.08633757e+00 8.63456786e-01 4.70313162e-01 -4.40637499e-01 -2.84647971e-01 5.45274675e-01 -9.35042948e-02 4.18600261e-01 2.54812479e-01 2.90204018e-01 -1.28803110e+00 -2.68159453e-02 -6.56780243e-01 7.34981298e-01 -7.30286002e-01 -2.10445833e+00 7.22469628e-01 8.13444331e-02 -1.33354819e+00 -4.47179317e-01 -5.71292579e-01 1.48948327e-01 8.51845920e-01 -2.14827776e+00 -1.58485115e+00 -4.62567270e-01 6.86826587e-01 4.87188876e-01 -1.54808983e-01 1.24348199e+00 4.24152613e-01 1.57605052e-01 8.19260776e-01 4.80511487e-01 9.16164890e-02 1.26733589e+00 -1.60046160e+00 9.41412821e-02 6.40006289e-02 4.04397666e-01 5.35937786e-01 2.15528324e-01 -2.22664759e-01 -8.05638015e-01 -1.51034832e+00 8.31069231e-01 -8.76119673e-01 7.01286852e-01 -5.47843575e-01 -8.84365380e-01 5.37869155e-01 3.57294604e-02 7.78083444e-01 5.62826097e-01 2.80719280e-01 -1.36916351e+00 -4.19147819e-01 -1.19136548e+00 -1.31890535e-01 1.04259467e+00 -1.27063918e+00 -7.00022638e-01 7.26013601e-01 1.01687241e+00 -1.58922106e-01 -1.11919117e+00 5.12884557e-01 6.60387456e-01 -2.97860771e-01 1.45147705e+00 -9.14963484e-01 5.38362086e-01 -1.90349787e-01 -3.90407145e-01 -1.25785816e+00 -7.40931690e-01 1.31879851e-01 -1.68120384e-01 1.18550932e+00 1.67869613e-01 -5.15580118e-01 6.77789450e-01 1.66235238e-01 2.33976170e-01 -7.51238048e-01 -8.56623590e-01 -9.20434415e-01 4.76097316e-01 1.38519168e-01 5.02224863e-01 9.86151516e-01 -6.81133270e-01 2.65801698e-01 -2.26511672e-01 -9.70128328e-02 1.24271142e+00 6.94441617e-01 3.64830226e-01 -1.57801294e+00 -1.43030033e-01 -4.52640057e-01 -4.71516043e-01 -1.14683545e+00 8.16080868e-01 -1.22572958e+00 8.30886662e-02 -1.55180907e+00 8.58649433e-01 -8.46419871e-01 -1.01733768e+00 8.75634730e-01 -5.06806932e-02 1.28792202e+00 2.08160996e-01 8.02260637e-01 -1.66665804e+00 3.84617180e-01 1.02166891e+00 -6.87043130e-01 4.80108082e-01 -2.50715703e-01 -6.17256522e-01 5.28765857e-01 3.94038379e-01 -7.69196630e-01 -1.66375846e-01 -7.05462515e-01 9.73951742e-02 -3.86967659e-01 7.41072476e-01 -7.65599012e-01 2.87283450e-01 1.98000208e-01 5.98130763e-01 -6.33184850e-01 3.40188086e-01 -8.17440510e-01 -2.27366567e-01 1.51557624e-02 -1.04049540e+00 -5.85419714e-01 6.29741699e-02 8.56860220e-01 -4.30993706e-01 1.08400304e-02 5.30370951e-01 -1.61642700e-01 -9.29492950e-01 7.54495561e-01 2.46832207e-01 2.79221773e-01 6.41110063e-01 1.92431167e-01 -5.59637427e-01 -4.95241620e-02 -8.64869475e-01 2.79975951e-01 5.72041214e-01 7.00424671e-01 1.86453491e-01 -1.72898042e+00 -6.80228949e-01 -6.10554628e-02 8.49136233e-01 -1.73479021e-01 2.31757209e-01 3.11809301e-01 5.15808500e-02 1.03261554e+00 -2.44133502e-01 -9.31063235e-01 -9.05423224e-01 8.00292850e-01 3.40716243e-01 -6.69225812e-01 -2.69949704e-01 1.05724847e+00 5.87784708e-01 -6.81827784e-01 6.30348325e-01 2.05508508e-02 -3.90743285e-01 2.61980057e-01 5.69805562e-01 3.66647914e-02 2.57947475e-01 -6.13440037e-01 -3.95660162e-01 8.55324864e-01 -3.92000169e-01 2.87791282e-01 1.30425274e+00 -3.62766951e-01 -2.97781497e-01 5.98026991e-01 1.83081579e+00 -3.68095607e-01 -1.47167468e+00 -8.28384578e-01 2.42618937e-02 -5.35317421e-01 2.95975655e-01 -1.29896128e+00 -8.11784625e-01 8.14259708e-01 9.16560709e-01 3.48230153e-01 1.00282431e+00 6.23703420e-01 4.43030506e-01 7.74015248e-01 7.93597758e-01 -8.59702289e-01 3.55393261e-01 4.97770965e-01 8.79451096e-01 -1.65791547e+00 -1.71164513e-01 -1.52996806e-02 -3.29045355e-02 6.49388313e-01 4.01007354e-01 -5.35843492e-01 6.53465807e-01 -2.12109342e-01 -1.24003470e-01 -1.66278347e-01 -9.05520201e-01 -3.26525092e-01 1.01264966e+00 4.63393152e-01 3.06725562e-01 1.70154110e-01 1.03416003e-01 3.26923132e-01 2.01293409e-01 3.17914523e-02 -3.74800473e-01 4.65452373e-01 -2.78085440e-01 -9.53606725e-01 -6.93268999e-02 9.69487071e-01 -6.79981351e-01 -4.48740184e-01 -3.33291292e-01 4.14294779e-01 3.00196379e-01 7.44619310e-01 3.88696849e-01 1.70885734e-02 6.62174588e-03 -1.53035685e-01 5.38563430e-01 -8.50553215e-01 -3.64830583e-01 -2.66755730e-01 -1.94572195e-01 -1.05173802e+00 -5.11012256e-01 -3.12637538e-01 -5.30156016e-01 -1.13134749e-01 -2.72192836e-01 4.27164555e-01 4.66602951e-01 8.21539342e-01 4.99178767e-01 2.71440923e-01 7.49959588e-01 -1.14014089e+00 -8.07112038e-01 -9.42280173e-01 -7.04905748e-01 7.01039016e-01 7.22455680e-01 -7.82308638e-01 -5.96479475e-01 -3.20322737e-02]
[9.719943046569824, 2.103008985519409]
33527806-cb2a-4805-aacf-4cc970e61684
untrimmednets-for-weakly-supervised-action
1703.03329
null
http://arxiv.org/abs/1703.03329v2
http://arxiv.org/pdf/1703.03329v2.pdf
UntrimmedNets for Weakly Supervised Action Recognition and Detection
Current action recognition methods heavily rely on trimmed videos for model training. However, it is expensive and time-consuming to acquire a large-scale trimmed video dataset. This paper presents a new weakly supervised architecture, called UntrimmedNet, which is able to directly learn action recognition models from untrimmed videos without the requirement of temporal annotations of action instances. Our UntrimmedNet couples two important components, the classification module and the selection module, to learn the action models and reason about the temporal duration of action instances, respectively. These two components are implemented with feed-forward networks, and UntrimmedNet is therefore an end-to-end trainable architecture. We exploit the learned models for action recognition (WSR) and detection (WSD) on the untrimmed video datasets of THUMOS14 and ActivityNet. Although our UntrimmedNet only employs weak supervision, our method achieves performance superior or comparable to that of those strongly supervised approaches on these two datasets.
['Luc van Gool', 'Limin Wang', 'Yuanjun Xiong', 'Dahua Lin']
2017-03-09
untrimmednets-for-weakly-supervised-action-1
http://openaccess.thecvf.com/content_cvpr_2017/html/Wang_UntrimmedNets_for_Weakly_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Wang_UntrimmedNets_for_Weakly_CVPR_2017_paper.pdf
cvpr-2017-7
['weakly-supervised-action-localization', 'weakly-supervised-action-recognition']
['computer-vision', 'computer-vision']
[ 5.89073122e-01 2.36952603e-02 -6.26570880e-01 -5.13591707e-01 -6.10088527e-01 -3.40289742e-01 6.98424399e-01 -4.95207846e-01 -6.13666654e-01 5.99206984e-01 2.68031478e-01 -6.59226552e-02 3.01619917e-02 -3.51940811e-01 -7.30396450e-01 -7.21624076e-01 -2.34321713e-01 3.24758887e-01 5.40259004e-01 3.42471510e-01 9.30420030e-03 1.72503665e-01 -1.57364500e+00 6.18889570e-01 4.78247017e-01 1.21711016e+00 -1.06816471e-01 8.52563262e-01 4.51155663e-01 1.47723961e+00 -2.24774420e-01 -6.78315461e-02 2.59264380e-01 -6.18698061e-01 -7.79697835e-01 4.75678533e-01 5.82463443e-01 -6.56572104e-01 -6.83728456e-01 5.51799119e-01 1.22387856e-01 1.97943777e-01 4.58003134e-01 -1.42366612e+00 -3.81483108e-01 7.30835021e-01 -2.84438968e-01 2.59465277e-01 2.93725133e-01 3.80458713e-01 7.65568197e-01 -8.17524195e-01 6.49372816e-01 1.00201118e+00 5.31551838e-01 7.10552275e-01 -1.09529805e+00 -5.09081900e-01 3.85282010e-01 5.98054171e-01 -1.02761936e+00 -8.66565943e-01 5.53407371e-01 -5.28561711e-01 1.11419821e+00 1.26589909e-01 5.79422355e-01 1.55224097e+00 -1.28895789e-01 1.17541540e+00 7.33892262e-01 -2.32788876e-01 3.56410861e-01 -2.35910475e-01 5.06777596e-03 6.42709255e-01 -2.28999525e-01 3.00022990e-01 -6.35383844e-01 1.02778070e-01 8.58419359e-01 3.25445473e-01 -2.10585445e-01 -4.59543765e-01 -1.46477818e+00 4.38125759e-01 2.37512350e-01 3.02036107e-01 -3.88103604e-01 3.54367137e-01 6.53901219e-01 3.50172848e-01 4.32619005e-01 2.10802838e-01 -7.09239066e-01 -6.30117893e-01 -8.71653318e-01 -2.63211280e-01 5.85065246e-01 9.25206125e-01 4.00852531e-01 2.01809332e-01 -3.35780680e-01 4.92763996e-01 2.08796654e-02 3.18506837e-01 6.69450939e-01 -1.11015439e+00 4.86056805e-01 7.19842196e-01 1.38466567e-01 -6.82629704e-01 -2.91509867e-01 -1.11371927e-01 -6.90429389e-01 1.07388861e-01 4.36778307e-01 -2.03318968e-01 -1.21358681e+00 1.60835278e+00 1.14255488e-01 6.86671615e-01 5.52371480e-02 1.03619349e+00 5.80468178e-01 4.75555122e-01 2.44512483e-01 -1.83455572e-01 8.26205969e-01 -1.47619843e+00 -6.05044425e-01 -3.80733252e-01 8.50397825e-01 -2.75761634e-01 8.91973197e-01 4.08673286e-01 -9.95889962e-01 -5.15327811e-01 -9.86397326e-01 3.58366151e-03 -1.67458698e-01 6.17332757e-01 6.75500572e-01 -5.09766676e-02 -7.20137060e-01 7.67640948e-01 -1.37681317e+00 -3.82890314e-01 6.03092432e-01 3.61904949e-01 -6.92364812e-01 -1.12907507e-01 -1.00717545e+00 7.72186637e-01 4.26925808e-01 1.95129842e-01 -1.45268059e+00 -3.65564317e-01 -1.02862191e+00 -1.68680288e-02 8.40259433e-01 -3.01988840e-01 1.45637357e+00 -1.53937101e+00 -1.50818956e+00 7.97624946e-01 -2.69648223e-03 -8.70554626e-01 7.55974293e-01 -3.67749691e-01 -4.15143907e-01 4.54903305e-01 -3.01352283e-03 4.97636288e-01 1.25142312e+00 -6.02870524e-01 -6.42227471e-01 -2.24964932e-01 1.66083083e-01 2.30970494e-02 -2.76769966e-01 1.11158431e-01 -5.92133343e-01 -8.35776806e-01 -1.18484803e-01 -8.62430692e-01 -1.92003608e-01 1.89297587e-01 -1.40955165e-01 -2.80753911e-01 1.06325090e+00 -5.61689615e-01 1.00327146e+00 -2.26247287e+00 3.76894206e-01 -3.20721567e-01 3.99468355e-02 5.00955045e-01 -4.84280109e-01 3.30654949e-01 -2.73261130e-01 -2.79669344e-01 -1.84913367e-01 -3.78827572e-01 -6.05747625e-02 5.15612543e-01 -1.88465610e-01 5.27254701e-01 3.82454306e-01 9.64812636e-01 -1.21725953e+00 -4.78954136e-01 5.07331610e-01 2.92571783e-01 -4.38360542e-01 4.87290114e-01 -4.57002759e-01 5.58510959e-01 -4.24508661e-01 7.80848086e-01 4.14767452e-02 -2.46027157e-01 3.74416232e-01 -2.23066598e-01 2.64781434e-02 1.60535231e-01 -8.87201726e-01 1.78773725e+00 -3.33070874e-01 4.70545322e-01 -1.03994861e-01 -1.28882110e+00 4.83880460e-01 4.72517550e-01 8.10224950e-01 -5.95634103e-01 1.61929369e-01 -3.42547074e-02 -1.98029190e-01 -9.39667821e-01 -3.11877560e-02 -3.68620791e-02 2.30950173e-02 4.88007426e-01 4.37900364e-01 5.84229708e-01 4.89788949e-01 2.19142437e-01 1.48701513e+00 7.88936675e-01 1.90762863e-01 1.69441417e-01 5.27588367e-01 -6.15448132e-02 8.47637653e-01 4.89508212e-01 -3.91540974e-01 5.76792955e-01 5.32567859e-01 -6.51050687e-01 -6.50061667e-01 -8.38923037e-01 3.21983099e-01 1.53587866e+00 -1.00960918e-01 -5.10909498e-01 -5.98571837e-01 -1.21041203e+00 -1.07030123e-01 4.44089532e-01 -8.88541579e-01 -3.70906800e-01 -7.56704807e-01 -3.46529812e-01 5.22665083e-01 1.17280316e+00 5.09579957e-01 -1.35397613e+00 -7.11331964e-01 1.32583886e-01 -1.60023466e-01 -1.36356735e+00 -7.01068461e-01 3.46945763e-01 -9.09423172e-01 -1.51468468e+00 -4.59202379e-01 -6.31950617e-01 7.29592800e-01 1.52257293e-01 9.51495528e-01 8.30623880e-02 -1.02268994e-01 2.96173483e-01 -6.91779852e-01 -3.53818089e-02 -1.12252809e-01 -1.30245328e-01 1.63875073e-01 5.17061412e-01 4.69632208e-01 -6.64769828e-01 -4.90356892e-01 6.65252507e-01 -9.21670079e-01 1.45598516e-01 8.61708701e-01 7.89000690e-01 5.41977167e-01 -1.31038338e-01 4.98820007e-01 -7.27404177e-01 -2.42357090e-01 -3.72230679e-01 -4.85101014e-01 2.25385830e-01 -3.41393888e-01 7.43556097e-02 7.08771408e-01 -7.39910543e-01 -1.01959157e+00 6.06298327e-01 1.06651550e-02 -9.08632636e-01 -3.09440434e-01 3.89508426e-01 -2.38753542e-01 1.63912386e-01 5.18207192e-01 3.47223043e-01 -5.61177135e-02 -7.01689839e-01 1.91807836e-01 5.87824464e-01 7.57915258e-01 -1.53617010e-01 5.92277825e-01 7.12018371e-01 -2.79434055e-01 -5.97299635e-01 -1.08164585e+00 -4.63437974e-01 -1.11321425e+00 -2.61642635e-01 1.03732944e+00 -9.47197497e-01 -4.75433230e-01 7.00546563e-01 -7.19275594e-01 -7.82677531e-01 -4.34150279e-01 7.07554638e-01 -8.53717327e-01 3.63843173e-01 -6.41199112e-01 -6.09009445e-01 -6.44665211e-02 -6.92816734e-01 1.09055507e+00 -2.55111247e-01 -1.58948898e-01 -8.43176663e-01 -8.59022699e-03 6.82591081e-01 1.74606696e-01 3.29106838e-01 4.96881306e-01 -7.97906280e-01 -5.38746953e-01 -5.74604750e-01 -4.38698903e-02 6.16663575e-01 2.19911098e-01 -1.08931540e-02 -8.91165853e-01 -3.09982389e-01 -1.44206718e-01 -8.61912489e-01 1.19200695e+00 2.60118663e-01 1.38519752e+00 -4.64070976e-01 -2.46896237e-01 5.23146927e-01 9.01071548e-01 2.23146930e-01 7.80772448e-01 1.46037892e-01 7.22336173e-01 2.81868160e-01 8.93501461e-01 3.85884047e-01 1.24429993e-01 8.75638306e-01 6.07050300e-01 -3.72650698e-02 -1.21909961e-01 -3.84784818e-01 9.02837753e-01 5.65868318e-01 -4.47560966e-01 -7.84190223e-02 -5.79707086e-01 5.51539302e-01 -2.38643193e+00 -1.24268329e+00 1.91193968e-01 2.09370279e+00 7.74752915e-01 2.13147044e-01 3.69068891e-01 1.18637405e-01 5.30257046e-01 3.95091861e-01 -8.58629405e-01 6.78329691e-02 1.89351231e-01 -4.20890525e-02 2.21775487e-01 2.71074533e-01 -1.56790221e+00 9.24835205e-01 6.35942841e+00 5.58768332e-01 -1.03924680e+00 1.61578685e-01 2.51882851e-01 -5.58825850e-01 5.87256491e-01 -1.87950507e-02 -4.17700350e-01 5.11287510e-01 1.08257806e+00 3.12407434e-01 2.69599348e-01 1.03993940e+00 4.88402039e-01 -8.64064991e-02 -1.62455285e+00 9.84228015e-01 1.99979767e-01 -1.12587309e+00 -2.89494079e-02 -2.36872986e-01 5.53131163e-01 1.78946014e-02 -2.86799967e-01 6.00721061e-01 1.73763335e-01 -1.01380920e+00 6.76272273e-01 5.41271925e-01 7.57870495e-01 -3.97362739e-01 7.69960165e-01 4.49398309e-01 -1.07987010e+00 -3.85311455e-01 -1.04124598e-01 -3.40835273e-01 3.07514101e-01 2.56512016e-01 -4.83133823e-01 2.97537595e-01 5.72670043e-01 1.51535034e+00 -5.02094984e-01 6.79024577e-01 -5.49664378e-01 7.85006762e-01 4.41363491e-02 4.54585314e-01 3.54212493e-01 2.57741157e-02 3.02453816e-01 9.78037775e-01 -5.84599040e-02 7.53169209e-02 4.78436351e-01 2.76818603e-01 -8.88953283e-02 -4.10636753e-01 -5.08111894e-01 -4.32429373e-01 -5.06904051e-02 1.19785345e+00 -5.05257547e-01 -6.59382463e-01 -6.10762835e-01 1.04671896e+00 2.64750302e-01 2.90926158e-01 -1.05345857e+00 -4.76006046e-02 5.92098475e-01 1.50802463e-01 5.63968599e-01 -1.23530388e-01 1.86334699e-01 -1.69948852e+00 1.37626141e-01 -1.08062816e+00 7.76177049e-01 -7.03662217e-01 -1.08038032e+00 4.82636034e-01 -4.55863103e-02 -1.60840929e+00 -5.17826140e-01 -6.00125849e-01 -5.91008604e-01 1.23273097e-01 -1.23343468e+00 -1.19534767e+00 -2.91165799e-01 8.06728840e-01 8.93888414e-01 2.23364364e-02 6.58854723e-01 3.74649346e-01 -8.83497953e-01 2.92284697e-01 -2.54252195e-01 5.83951116e-01 6.69182003e-01 -1.13220012e+00 1.05110757e-01 9.10711527e-01 3.21075439e-01 1.44004092e-01 3.79152507e-01 -5.15537024e-01 -1.32717204e+00 -1.53962648e+00 7.24872231e-01 -7.80061126e-01 7.72680521e-01 -4.16899413e-01 -7.34934270e-01 1.22507608e+00 4.14011255e-02 3.60169679e-01 6.34804308e-01 -1.23763636e-01 -3.88947427e-01 -1.43889800e-01 -7.82782376e-01 4.24328536e-01 1.45996487e+00 -4.50993687e-01 -8.70240986e-01 4.85575169e-01 5.87642491e-01 -1.95436195e-01 -8.32591772e-01 4.69223738e-01 5.72695792e-01 -8.64748597e-01 7.64456332e-01 -1.21187556e+00 4.66514826e-01 -3.01028967e-01 5.13618290e-02 -9.84633327e-01 -2.04861745e-01 -5.87010384e-01 -7.82606423e-01 8.52689683e-01 4.02511358e-01 -3.31698209e-01 8.58709037e-01 6.04474187e-01 -1.61610857e-01 -7.25744843e-01 -8.31014097e-01 -1.07941091e+00 -5.01776636e-01 -5.18111825e-01 2.54564524e-01 9.66658890e-01 1.59754425e-01 3.65419745e-01 -8.41118157e-01 -1.50083035e-01 5.18574357e-01 2.98232436e-01 6.26970947e-01 -7.73877442e-01 -4.05187368e-01 -2.03932337e-02 -6.62747443e-01 -1.16424465e+00 2.28191078e-01 -5.61580837e-01 1.80919781e-01 -1.20012534e+00 2.29766086e-01 1.39581814e-01 -5.66460609e-01 1.10542619e+00 1.20785780e-01 4.55504984e-01 4.44888370e-03 2.49656692e-01 -1.24141550e+00 7.40582466e-01 1.06923008e+00 -1.94731116e-01 -2.38337472e-01 8.41885433e-02 -1.98403195e-01 1.17201018e+00 6.39843643e-01 -4.41274732e-01 -5.91961503e-01 -5.39033294e-01 -2.26695403e-01 9.35744420e-02 6.44459009e-01 -1.12248373e+00 8.35280120e-02 -4.71041381e-01 4.18186009e-01 -4.42850471e-01 3.58565003e-01 -8.91494215e-01 -5.65328561e-02 4.42418247e-01 -5.91058195e-01 -3.44451517e-01 -1.07052051e-01 7.82074153e-01 -3.32256287e-01 8.08177143e-02 8.00386846e-01 -2.12720886e-01 -1.10013735e+00 5.60660720e-01 -4.92987067e-01 1.13656907e-03 1.36663353e+00 -1.88805237e-01 -1.38827428e-01 -3.44933838e-01 -1.16101360e+00 2.61776090e-01 2.70046830e-01 6.79928064e-01 6.92016304e-01 -1.35041523e+00 -4.50506866e-01 2.51586199e-01 1.66265279e-01 -1.20912656e-01 1.85676053e-01 1.16661263e+00 -1.12652943e-01 4.86729681e-01 -1.79073527e-01 -5.08812904e-01 -1.33914423e+00 8.40175807e-01 2.77217954e-01 -3.26790601e-01 -7.15957344e-01 5.34502804e-01 -1.00490209e-02 -2.35861197e-01 5.84032416e-01 -3.70386332e-01 -1.43251166e-01 3.91196646e-02 7.14493334e-01 3.79805267e-01 -1.66282952e-01 -5.45442462e-01 -5.77888787e-01 3.89345467e-01 -7.08631948e-02 1.62502572e-01 1.59819889e+00 1.75132483e-01 1.07490443e-01 4.40679461e-01 1.06605577e+00 -6.43412888e-01 -1.92230785e+00 -2.72141814e-01 1.86360314e-01 -5.28920114e-01 -3.27455662e-02 -8.64316225e-01 -1.20391381e+00 8.38478029e-01 3.14607382e-01 -1.79075941e-01 1.28108656e+00 -2.82971244e-02 8.52048695e-01 5.75829506e-01 4.27507490e-01 -1.24337363e+00 5.03542602e-01 4.27064121e-01 9.68430459e-01 -1.28476071e+00 -1.56656712e-01 -1.87172160e-01 -7.55036891e-01 8.96706820e-01 1.10874021e+00 -1.19875275e-01 4.90547270e-01 1.48647577e-01 1.32901430e-01 -1.18256047e-01 -1.23734903e+00 -2.85725653e-01 2.86249906e-01 5.21608889e-01 1.82402544e-02 -2.18792900e-01 -8.37261304e-02 6.17429435e-01 5.89449286e-01 5.26997626e-01 3.11974794e-01 1.32921076e+00 -2.67311275e-01 -8.84275377e-01 2.83777919e-02 4.38289165e-01 -3.07911962e-01 1.92756593e-01 -6.07533872e-01 6.59978747e-01 1.52232960e-01 8.81681442e-01 7.00593069e-02 -6.60448909e-01 3.77944767e-01 1.63952380e-01 3.50764126e-01 -7.46651828e-01 -1.96952254e-01 -8.40926617e-02 2.80420601e-01 -1.18362069e+00 -8.88487279e-01 -5.32121003e-01 -1.11433399e+00 7.43652508e-02 -1.35621279e-01 -2.80259661e-02 6.10031523e-02 1.12598157e+00 5.40757716e-01 3.76908332e-01 5.61153769e-01 -1.04279399e+00 -7.41636992e-01 -1.22068977e+00 -6.33663237e-01 6.32524073e-01 4.08806413e-01 -7.96228826e-01 -3.47079426e-01 6.31215572e-01]
[8.472122192382812, 0.6460627317428589]
153a6211-ecd1-4c00-9496-b1792a9bc3bb
latent-graph-attention-for-enhanced-spatial
2307.04149
null
https://arxiv.org/abs/2307.04149v1
https://arxiv.org/pdf/2307.04149v1.pdf
Latent Graph Attention for Enhanced Spatial Context
Global contexts in images are quite valuable in image-to-image translation problems. Conventional attention-based and graph-based models capture the global context to a large extent, however, these are computationally expensive. Moreover, the existing approaches are limited to only learning the pairwise semantic relation between any two points on the image. In this paper, we present Latent Graph Attention (LGA) a computationally inexpensive (linear to the number of nodes) and stable, modular framework for incorporating the global context in the existing architectures, especially empowering small-scale architectures to give performance closer to large size architectures, thus making the light-weight architectures more useful for edge devices with lower compute power and lower energy needs. LGA propagates information spatially using a network of locally connected graphs, thereby facilitating to construct a semantically coherent relation between any two spatially distant points that also takes into account the influence of the intermediate pixels. Moreover, the depth of the graph network can be used to adapt the extent of contextual spread to the target dataset, thereby being able to explicitly control the added computational cost. To enhance the learning mechanism of LGA, we also introduce a novel contrastive loss term that helps our LGA module to couple well with the original architecture at the expense of minimal additional computational load. We show that incorporating LGA improves the performance on three challenging applications, namely transparent object segmentation, image restoration for dehazing and optical flow estimation.
['Dilip K. Prasad', 'Deepak K. Gupta', 'Himanshu Buckchash', 'Yash Bhambhu', 'Ayush Singh']
2023-07-09
null
null
null
null
['optical-flow-estimation', 'image-to-image-translation', 'image-restoration', 'graph-attention', 'image-to-image-translation']
['computer-vision', 'computer-vision', 'computer-vision', 'graphs', 'miscellaneous']
[ 1.96571693e-01 1.46815181e-01 -2.16827869e-01 -1.84065700e-01 -1.16769254e-01 -2.86286652e-01 3.88544679e-01 3.17309648e-01 -3.42718482e-01 4.27958786e-01 6.67339265e-02 -1.03555374e-01 -2.52185632e-02 -1.04396510e+00 -7.43726075e-01 -6.94584370e-01 7.76806921e-02 8.84637162e-02 4.98113602e-01 -3.85059603e-02 1.33862749e-01 7.18014061e-01 -1.19959140e+00 1.40639190e-02 9.57042873e-01 9.47220504e-01 5.31160712e-01 3.16128582e-01 -1.14903890e-01 7.63920486e-01 -1.79871023e-01 -3.96833569e-01 2.61360914e-01 -5.06288886e-01 -7.22752035e-01 1.83098480e-01 6.18951619e-01 -2.34455273e-01 -3.98151696e-01 1.09007967e+00 4.08439219e-01 2.05706939e-01 3.69181901e-01 -1.10032272e+00 -7.15143144e-01 3.60742986e-01 -6.39737248e-01 3.04500461e-01 -1.01168461e-01 2.75761664e-01 1.16791332e+00 -5.20761788e-01 6.82805300e-01 1.11035442e+00 4.50256318e-01 3.72703671e-01 -1.39346826e+00 -2.90302068e-01 6.37870669e-01 4.51304376e-01 -1.24008930e+00 -3.37654114e-01 1.20193648e+00 -2.83359945e-01 8.26233447e-01 1.50269583e-01 8.90896142e-01 7.70955563e-01 7.14082494e-02 5.00189722e-01 8.16204429e-01 -4.43915427e-01 2.55321920e-01 -4.19638045e-02 -6.34679794e-02 8.96806359e-01 2.07215548e-01 -2.56195813e-01 -5.25949478e-01 1.52783513e-01 1.14410126e+00 2.66127139e-02 -4.86736357e-01 -6.21175528e-01 -1.03034878e+00 7.51805782e-01 1.18647838e+00 4.24310982e-01 -3.68253618e-01 5.06248236e-01 1.64887622e-01 3.50526683e-02 3.81734222e-01 4.06618804e-01 -2.74442524e-01 2.86518067e-01 -6.65104449e-01 -2.37498760e-01 4.10853952e-01 6.96690857e-01 1.04762614e+00 3.47313434e-02 -2.16230050e-01 4.48472053e-01 2.80244768e-01 1.46247193e-01 2.81923801e-01 -1.06499803e+00 3.77120942e-01 8.29530716e-01 -3.64242904e-02 -1.50544202e+00 -3.69668156e-01 -6.33741975e-01 -9.51830149e-01 3.03073227e-01 4.61894184e-01 1.65288091e-01 -8.22619557e-01 1.95273840e+00 3.80249918e-01 4.63435531e-01 -3.14470291e-01 1.08246374e+00 4.16218013e-01 4.99565661e-01 1.75234098e-02 -6.41044006e-02 1.39311767e+00 -1.16084123e+00 -6.01906896e-01 -5.36291897e-01 5.24897873e-01 -6.19312644e-01 1.25692272e+00 -1.07666090e-01 -1.11457455e+00 -4.66007978e-01 -9.67408955e-01 -5.00801265e-01 -2.98253506e-01 2.46625021e-02 7.96388745e-01 4.11066473e-01 -1.29195511e+00 6.20502651e-01 -8.58452916e-01 -1.47613227e-01 7.21953213e-01 4.34256583e-01 -1.33571729e-01 -1.79468140e-01 -9.03562725e-01 6.38750851e-01 5.29117435e-02 1.59230471e-01 -4.20427412e-01 -6.91626191e-01 -8.82562578e-01 4.07434613e-01 5.01530528e-01 -1.09055161e+00 5.20559072e-01 -1.25064600e+00 -1.56518948e+00 6.18198752e-01 -2.74007738e-01 -4.66339767e-01 4.05635893e-01 2.15846580e-02 2.04031199e-01 4.29634631e-01 -2.83337850e-02 9.99454975e-01 8.75938714e-01 -1.05630898e+00 -2.86685437e-01 -4.32664424e-01 3.50748599e-01 3.31946760e-01 -4.25991476e-01 -3.03147584e-01 -7.77040720e-01 -8.02725971e-01 1.03615396e-01 -9.20404911e-01 -4.15667892e-01 5.11454105e-01 -1.22029237e-01 -6.31476799e-03 9.18108642e-01 -4.95177269e-01 9.65638161e-01 -2.15461493e+00 4.50245380e-01 9.59563479e-02 4.28417772e-01 3.15556258e-01 -3.44052434e-01 7.00559318e-02 1.58102080e-01 4.04300215e-03 -3.30426186e-01 -4.24977988e-01 -3.31162751e-01 3.74073714e-01 4.72849347e-02 3.59925121e-01 3.21455419e-01 1.30075407e+00 -9.36567366e-01 -3.49773437e-01 3.75311166e-01 8.02535951e-01 -8.65608454e-01 -3.64967883e-02 -2.44885221e-01 6.11826479e-01 -4.48728025e-01 1.08873270e-01 5.23740351e-01 -5.43210745e-01 1.89493507e-01 -3.83465916e-01 1.36509508e-01 2.53309280e-01 -9.84657407e-01 1.98550677e+00 -8.06662440e-01 8.08429480e-01 2.44173393e-01 -9.81701255e-01 6.93099201e-01 2.95652878e-02 5.12895286e-01 -7.06074953e-01 2.01115653e-01 3.76884714e-02 2.69126724e-02 -2.45048895e-01 2.37996385e-01 1.86146200e-01 3.09841901e-01 2.36217827e-01 -1.75395533e-01 4.43888232e-02 3.79013689e-03 3.26676905e-01 7.86400259e-01 6.97967634e-02 9.31993648e-02 -2.75080889e-01 5.55467188e-01 -4.53931510e-01 5.29360890e-01 4.11659718e-01 -6.73696548e-02 5.67676365e-01 5.60335279e-01 -4.37020510e-01 -7.39476919e-01 -7.15831280e-01 1.07606992e-01 8.01422596e-01 5.54682553e-01 -3.63609672e-01 -8.69361520e-01 -6.24656916e-01 -3.05636078e-01 3.31736267e-01 -5.41187406e-01 -1.97356269e-01 -6.43312991e-01 -4.35776919e-01 -9.50033143e-02 5.47661364e-01 6.70737386e-01 -9.44922447e-01 -8.37228894e-01 1.86358690e-01 -2.96974391e-01 -1.41020441e+00 -8.76014411e-01 -1.30199298e-01 -1.08684146e+00 -9.92119133e-01 -5.94040155e-01 -7.71238267e-01 1.01502800e+00 3.71559829e-01 9.33474898e-01 3.10632288e-01 -4.12798822e-01 2.16690809e-01 -2.74278790e-01 7.25060329e-02 -3.69534679e-02 1.24083959e-01 -4.82987642e-01 3.55877399e-01 -5.02056554e-02 -8.57867777e-01 -9.25116718e-01 3.09498310e-01 -9.68318462e-01 3.50399584e-01 4.58702415e-01 8.35108876e-01 5.76410949e-01 -9.19397399e-02 3.59147519e-01 -8.05612147e-01 2.36824468e-01 -5.82549125e-02 -6.64509177e-01 6.72709346e-02 -5.34753859e-01 2.07819074e-01 5.92052341e-01 -3.50150377e-01 -8.16004872e-01 1.44794798e-02 6.33277595e-02 -5.04069567e-01 2.26610258e-01 3.50985736e-01 -3.38728040e-01 -4.44171995e-01 2.89106399e-01 -3.42273433e-03 3.16206813e-02 -2.86036670e-01 7.09774077e-01 9.97905359e-02 3.45928043e-01 -3.13915163e-01 4.41562891e-01 6.69036686e-01 4.36047912e-01 -6.79855883e-01 -5.91233611e-01 -2.00757444e-01 -7.45428026e-01 -4.02135365e-02 7.99403965e-01 -7.19719052e-01 -7.61322796e-01 3.85344028e-01 -1.09897900e+00 -4.84294027e-01 -4.98178124e-01 2.92467862e-01 -4.43457991e-01 5.12525260e-01 -5.66884756e-01 -3.58893275e-01 -2.83893466e-01 -1.41670990e+00 9.89728928e-01 1.97227329e-01 5.76904975e-03 -1.38108695e+00 -4.99341339e-01 3.25830072e-01 6.12264872e-01 2.24416092e-01 1.10248065e+00 1.20945759e-01 -1.11921012e+00 8.88870284e-02 -6.15212917e-01 2.25298181e-01 1.72763720e-01 -2.63072848e-01 -9.41680789e-01 -3.00257891e-01 3.22421901e-02 1.79945603e-01 1.03117681e+00 5.85028231e-01 1.22804880e+00 -4.01916921e-01 -2.84255385e-01 8.25129211e-01 1.43263173e+00 -1.68236390e-01 7.53523529e-01 8.57410058e-02 1.22329831e+00 7.02058733e-01 1.26853481e-01 6.34470582e-02 4.36274320e-01 1.02025914e+00 6.28941119e-01 -4.64991301e-01 -7.53962696e-01 -1.56815931e-01 1.98079690e-01 7.21522033e-01 -1.30429119e-01 -3.83728862e-01 -6.01161957e-01 5.53830445e-01 -1.97035754e+00 -6.45204008e-01 -3.91118750e-02 2.20290709e+00 6.12315297e-01 1.82449613e-02 -2.68550724e-01 -7.82734081e-02 6.10909522e-01 3.90745610e-01 -5.53557575e-01 -2.57930338e-01 -8.66734702e-03 1.07397743e-01 4.90640998e-01 7.89713502e-01 -9.21258748e-01 1.03620648e+00 5.24799442e+00 8.76244962e-01 -1.25922823e+00 2.07900524e-01 9.12925720e-01 -1.98667213e-01 -4.75178689e-01 1.76572785e-01 -3.55596364e-01 4.56723064e-01 5.04327774e-01 1.38407782e-01 5.27121305e-01 4.38326865e-01 2.85404205e-01 -8.82350728e-02 -1.03964937e+00 1.07014918e+00 -3.46485786e-02 -1.51896453e+00 2.14348972e-01 1.43928602e-01 8.19838405e-01 3.76292542e-02 1.11704722e-01 -3.79620880e-01 -1.56580344e-01 -7.71008074e-01 6.24803305e-01 3.62443835e-01 6.19521976e-01 -7.19523430e-01 6.01805985e-01 1.09737389e-01 -1.22172010e+00 -9.43072513e-02 -2.86040694e-01 1.91329326e-02 2.46549845e-01 7.76748955e-01 -3.06544691e-01 5.48450410e-01 5.85373461e-01 7.81140447e-01 -5.25283754e-01 8.28133047e-01 -4.03260499e-01 2.58634120e-01 -3.30817193e-01 2.00639695e-01 4.19044763e-01 -4.54384506e-01 7.96681225e-01 8.00416887e-01 1.04667298e-01 -3.70704196e-02 2.11770207e-01 1.16179276e+00 -4.08652991e-01 1.67629436e-01 -5.19171774e-01 3.02401870e-01 2.19498754e-01 1.22345340e+00 -1.03615654e+00 -1.13707714e-01 -3.88812095e-01 1.25417960e+00 5.69682777e-01 4.78895038e-01 -7.50168443e-01 -3.00457180e-01 5.80846131e-01 4.86773103e-01 4.28486943e-01 -4.00336355e-01 -4.31947678e-01 -1.04780197e+00 2.17584059e-01 -4.12213743e-01 1.62321582e-01 -7.54195392e-01 -9.85215187e-01 5.62299311e-01 -3.22454631e-01 -1.01233530e+00 1.77621737e-01 -4.32640880e-01 -6.37544632e-01 8.86215329e-01 -2.00940180e+00 -1.30384886e+00 -5.15715182e-01 5.93867660e-01 4.09853369e-01 2.42246330e-01 4.84703481e-01 4.21705604e-01 -5.57310283e-01 5.69012165e-01 -2.72326529e-01 1.05906829e-01 5.54969132e-01 -9.50520694e-01 3.81128728e-01 1.04491866e+00 3.28964919e-01 6.92287505e-01 3.00373882e-01 -4.79436785e-01 -1.21338451e+00 -1.18542051e+00 6.81777298e-01 -1.82658419e-01 5.29409587e-01 -2.96631604e-01 -9.39777195e-01 4.66345519e-01 2.80965716e-01 4.57871020e-01 2.84662873e-01 -1.08318016e-01 -3.72281164e-01 -3.14629376e-01 -9.83835042e-01 8.07461083e-01 1.14134181e+00 -5.36394119e-01 2.84677185e-02 1.74829543e-01 8.69238853e-01 -2.47394666e-01 -5.68482280e-01 1.57484829e-01 1.98271200e-01 -9.95884955e-01 1.02668154e+00 -1.60558730e-01 3.62394124e-01 -4.08232957e-01 2.31109887e-01 -1.07264376e+00 -3.41116667e-01 -7.89177239e-01 -1.09373376e-01 1.13747871e+00 2.29621589e-01 -9.44100380e-01 7.97411740e-01 7.40964651e-01 -7.47109801e-02 -9.22865331e-01 -1.04186642e+00 -4.38591361e-01 -2.13673279e-01 -1.58673972e-01 4.79784906e-01 9.77551937e-01 -3.51722509e-01 3.39439988e-01 -2.73035765e-01 2.15440512e-01 5.37350714e-01 2.70374686e-01 4.36648995e-01 -1.04673934e+00 -4.33659971e-01 -6.82319999e-01 -7.29192197e-01 -1.36412990e+00 -8.91221538e-02 -1.06760883e+00 -2.82461315e-01 -1.50348341e+00 1.17979504e-01 -6.17282033e-01 -2.45281056e-01 5.06849945e-01 -4.13576394e-01 4.57042038e-01 3.96276295e-01 1.69874132e-01 -3.61550301e-01 6.89421237e-01 1.60251880e+00 -1.56467810e-01 -2.99774319e-01 -9.52452794e-02 -7.39297569e-01 7.68180907e-01 5.97923577e-01 -4.37271357e-01 -6.68094158e-01 -8.31288815e-01 1.77840009e-01 -2.37433091e-01 5.85277855e-01 -7.30170250e-01 3.14048469e-01 -2.37861108e-02 1.06188871e-01 9.29752514e-02 3.42353553e-01 -1.01768494e+00 -5.46441749e-02 3.38308394e-01 -1.74165532e-01 -1.71225205e-01 1.70866594e-01 8.05085540e-01 -2.89186776e-01 -1.35989040e-01 9.83909905e-01 -8.14376250e-02 -5.51921129e-01 5.32629550e-01 -5.89251034e-02 -1.06486037e-01 1.08803272e+00 -3.58875275e-01 -1.49226755e-01 -3.82715940e-01 -6.80545509e-01 8.36788192e-02 6.55645192e-01 3.34599942e-01 5.54855049e-01 -1.16892493e+00 -3.00562263e-01 4.30176973e-01 -1.46072730e-01 2.65280306e-01 4.05607402e-01 9.97868061e-01 -7.62336195e-01 2.63854474e-01 -2.26125374e-01 -6.66445851e-01 -1.14412093e+00 6.30569220e-01 4.13159758e-01 -3.08886826e-01 -8.95025432e-01 8.97872388e-01 5.60869038e-01 6.39505014e-02 1.50778279e-01 -5.41167736e-01 -1.33617669e-01 5.00465336e-04 2.91197211e-01 2.63737649e-01 -4.82904427e-02 -6.57788575e-01 -2.76263237e-01 9.88579094e-01 6.17614649e-02 1.59587190e-01 1.17260611e+00 -4.96307790e-01 -3.00721675e-01 9.82712135e-02 1.17866266e+00 1.99755449e-02 -1.54731011e+00 -3.51323158e-01 -2.45914936e-01 -6.88907743e-01 5.68008363e-01 -4.83609527e-01 -1.64033520e+00 1.13887978e+00 5.32346189e-01 5.23840114e-02 1.33689177e+00 1.42402798e-02 8.57840657e-01 -1.08846545e-01 3.96377414e-01 -7.51654387e-01 2.50776529e-01 1.49808750e-01 6.79872930e-01 -1.01161754e+00 9.46735293e-02 -7.64903188e-01 -5.40342450e-01 9.71997321e-01 4.51615453e-01 -7.20975325e-02 5.66259682e-01 1.58824459e-01 -1.05032817e-01 -1.86625466e-01 -4.45318162e-01 -2.06175432e-01 4.71377999e-01 5.27478158e-01 2.23315790e-01 -1.28548145e-01 -1.67016640e-01 1.27324760e-01 3.33764732e-01 -1.70194566e-01 4.05950636e-01 4.99890834e-01 -2.53007531e-01 -1.24418664e+00 6.97388723e-02 1.79484874e-01 -3.69598299e-01 -2.29959249e-01 -1.11714453e-01 4.88232613e-01 2.60728806e-01 7.63197541e-01 2.85010695e-01 1.02655031e-01 1.70735002e-01 -2.79264897e-01 5.67389488e-01 -4.71631169e-01 -3.73160452e-01 1.94438949e-01 -3.22207659e-01 -8.78538489e-01 -6.49279773e-01 -3.08996260e-01 -1.15113854e+00 -1.91646010e-01 -3.33346665e-01 -2.42333591e-01 5.42212009e-01 7.26634920e-01 7.71136642e-01 6.63275301e-01 5.50556183e-01 -1.03360415e+00 7.66150579e-02 -5.15678108e-01 -2.97673672e-01 4.58582222e-01 2.99891472e-01 -5.65413356e-01 -1.52865618e-01 3.32018808e-02]
[9.836669921875, -0.13856558501720428]
fc59bfcc-e0ea-4c72-9636-4bde819b1f9c
total-text-a-comprehensive-dataset-for-scene
1710.104
null
http://arxiv.org/abs/1710.10400v1
http://arxiv.org/pdf/1710.10400v1.pdf
Total-Text: A Comprehensive Dataset for Scene Text Detection and Recognition
Text in curve orientation, despite being one of the common text orientations in real world environment, has close to zero existence in well received scene text datasets such as ICDAR2013 and MSRA-TD500. The main motivation of Total-Text is to fill this gap and facilitate a new research direction for the scene text community. On top of the conventional horizontal and multi-oriented texts, it features curved-oriented text. Total-Text is highly diversified in orientations, more than half of its images have a combination of more than two orientations. Recently, a new breed of solutions that casted text detection as a segmentation problem has demonstrated their effectiveness against multi-oriented text. In order to evaluate its robustness against curved text, we fine-tuned DeconvNet and benchmark it on Total-Text. Total-Text with its annotation is available at https://github.com/cs-chan/Total-Text-Dataset
['Chee Seng Chan', 'Chee Kheng Chng']
2017-10-28
null
null
null
null
['curved-text-detection']
['computer-vision']
[ 1.24470308e-01 -2.74962574e-01 3.41549478e-02 -3.55521679e-01 -4.10391510e-01 -8.26050937e-01 8.75079572e-01 -8.91937762e-02 -3.08414370e-01 1.03771247e-01 1.99261695e-01 -2.95269519e-01 1.71937793e-01 -8.27691138e-01 -5.88808179e-01 -5.95371485e-01 5.62223256e-01 8.14367354e-01 6.43995464e-01 -2.62484968e-01 3.86623353e-01 1.83198333e-01 -1.31078362e+00 3.46288741e-01 7.07360864e-01 5.60614049e-01 2.55318731e-01 6.45835102e-01 -3.44901770e-01 3.98672730e-01 -3.34826589e-01 -7.16681838e-01 4.65795338e-01 -1.83700636e-01 -6.75507367e-01 3.40643436e-01 7.18422294e-01 -9.98105854e-02 -4.57797617e-01 1.09238410e+00 6.06199622e-01 -1.53770283e-01 9.39205766e-01 -1.22985756e+00 -2.75321603e-01 7.34920800e-01 -1.20937765e+00 2.84507602e-01 2.09304094e-01 -1.95825715e-02 9.87916470e-01 -9.88934040e-01 8.92828405e-01 1.31769884e+00 9.59326327e-01 9.84360650e-02 -7.14868367e-01 -5.02348363e-01 1.00285813e-01 -4.37870435e-02 -1.30297220e+00 5.79910837e-02 6.08553886e-01 -5.75329900e-01 7.80720651e-01 3.60075742e-01 5.60325205e-01 1.38371265e+00 3.78516763e-01 1.25301552e+00 1.14979851e+00 -3.97865683e-01 -1.23351522e-01 -3.85968238e-02 3.53876054e-01 5.06748557e-01 3.10645580e-01 -3.06010336e-01 -3.43162686e-01 1.46982327e-01 3.29583704e-01 6.45310581e-02 -2.36004889e-01 -3.37701261e-01 -1.51923501e+00 6.69129789e-01 6.15323186e-02 4.60553646e-01 1.67397559e-01 -2.36683496e-04 7.05680907e-01 -7.04946667e-02 5.16588151e-01 -2.55405568e-02 -4.46013600e-01 -2.39792466e-01 -1.16664910e+00 4.27440375e-01 6.04103625e-01 1.24414992e+00 4.10582930e-01 1.09811395e-01 -1.59055442e-01 1.13541603e+00 2.01810449e-01 9.44704652e-01 7.15311527e-01 -1.39969647e-01 9.64652359e-01 8.67905080e-01 -5.04633367e-01 -1.10583961e+00 -7.54206479e-01 -5.51586270e-01 -1.12005091e+00 -3.29491287e-01 3.97917986e-01 -1.40275568e-01 -1.16719770e+00 7.74967194e-01 4.81393397e-01 -9.80321094e-02 -3.52880239e-01 7.86606729e-01 8.49069178e-01 4.54392761e-01 -3.31332684e-01 3.21090102e-01 1.44575059e+00 -1.11085105e+00 -6.63914800e-01 -1.21021912e-01 9.14302826e-01 -1.38101232e+00 1.05830491e+00 7.78869927e-01 -6.62783265e-01 -1.96083769e-01 -9.64844286e-01 -4.21129353e-02 -7.29704618e-01 4.61041987e-01 2.79027492e-01 7.97535181e-01 -8.03260207e-01 1.18433699e-01 -7.54598260e-01 -8.34036469e-01 4.06651974e-01 1.31212603e-02 -2.30815448e-02 -6.51611984e-02 -6.37458086e-01 5.15144587e-01 3.83054018e-01 -2.88302489e-02 -4.61989313e-01 -2.16432080e-01 -4.48335171e-01 -3.43906194e-01 7.74156332e-01 -6.01343930e-01 1.05662262e+00 -6.71038508e-01 -1.11404276e+00 1.13182008e+00 -1.77072529e-02 -3.45227838e-01 1.32450461e+00 -4.13103223e-01 -4.63686675e-01 1.10067636e-01 3.27471673e-01 5.39023817e-01 1.06771958e+00 -1.05111599e+00 -6.79147124e-01 -6.67695343e-01 -5.71373940e-01 3.53305489e-01 -3.03574890e-01 -8.43288749e-03 -9.03014481e-01 -1.21629930e+00 5.42815506e-01 -1.01845622e+00 1.55243680e-01 -1.97406262e-01 -1.12603092e+00 -2.55759895e-01 1.30021346e+00 -3.10989052e-01 1.12437809e+00 -1.86428940e+00 -6.36496544e-02 4.27914970e-02 2.74729341e-01 1.27295986e-01 2.60530025e-01 7.03387856e-01 -1.94087978e-02 3.39294910e-01 -2.62014687e-01 -3.50100040e-01 -6.42042374e-03 -2.36624889e-02 -4.52618837e-01 9.19980645e-01 -3.13997895e-01 8.00751388e-01 -3.34787458e-01 -8.39498460e-01 7.54613936e-01 2.72418886e-01 -2.97694415e-01 -4.25315827e-01 -4.03949440e-01 9.24999863e-02 -7.02536523e-01 8.13459396e-01 1.12598395e+00 -5.13730422e-02 -2.14360461e-01 -1.87539190e-01 -3.16592157e-01 -4.14670616e-01 -1.35704482e+00 1.36035216e+00 2.58336037e-01 9.98519480e-01 -1.33796036e-01 -1.07313836e+00 7.62668133e-01 5.63703291e-02 6.63445532e-01 -7.62053013e-01 5.68794549e-01 1.66612417e-01 -3.47612709e-01 -6.20530665e-01 1.00232637e+00 2.97922015e-01 7.09385425e-02 2.99845971e-02 -2.22657770e-01 -4.29097444e-01 4.35395569e-01 4.58846092e-01 8.31516922e-01 2.26482451e-01 -5.50585501e-02 -3.69544774e-01 4.12509501e-01 2.82708049e-01 1.85754821e-01 8.05529237e-01 -1.15991220e-01 1.19876301e+00 3.26141387e-01 -3.74479711e-01 -1.11628795e+00 -7.92439640e-01 -7.21751750e-01 1.06263232e+00 3.26867968e-01 -5.44094920e-01 -8.24007690e-01 -5.73456287e-01 -1.44816041e-01 6.08298242e-01 -7.23868191e-01 5.80281138e-01 -5.50239503e-01 -1.15023375e+00 1.12174070e+00 3.09440762e-01 1.01679647e+00 -7.60394692e-01 -5.94288528e-01 -8.47306475e-02 -4.07548130e-01 -1.45860028e+00 -5.07611871e-01 2.29744464e-01 -9.05261695e-01 -1.22216690e+00 -1.13019180e+00 -6.07523024e-01 5.47282219e-01 5.35404563e-01 1.16603994e+00 -1.01467066e-01 -6.47051990e-01 5.01260340e-01 -7.51634955e-01 -6.66195035e-01 -1.31595597e-01 2.74240136e-01 -5.70801616e-01 9.38395038e-02 2.19732955e-01 1.13923185e-01 -6.10319257e-01 7.25212991e-01 -1.08480215e+00 3.34717184e-01 3.57737124e-01 6.20238483e-01 4.66296732e-01 8.44316855e-02 5.63772842e-02 -9.73812878e-01 3.78756553e-01 -3.16809237e-01 -5.96031129e-01 1.56189710e-01 -4.50727999e-01 -3.76416981e-01 3.82708013e-01 -1.89911246e-01 -1.02807653e+00 7.80595914e-02 -2.37393111e-01 -1.62883848e-01 -2.90522575e-01 3.75288874e-01 2.71366268e-01 3.76548082e-01 6.30584776e-01 3.62065256e-01 -7.80716181e-01 -4.98683184e-01 2.95668900e-01 9.09344494e-01 2.69409418e-01 -5.72859526e-01 8.64556611e-01 9.31132853e-01 2.97330245e-02 -1.60946476e+00 -6.48153663e-01 -9.73650098e-01 -7.50822067e-01 -2.35596284e-01 1.21898973e+00 -7.05292642e-01 -2.52389550e-01 1.05451274e+00 -9.56493855e-01 -4.08280134e-01 2.28992715e-01 1.99401781e-01 -2.72405207e-01 8.13710213e-01 -3.52058381e-01 -6.68517411e-01 -2.94954181e-01 -1.20732999e+00 1.66341352e+00 -1.08816260e-02 2.81884789e-01 -1.03385293e+00 -8.65624398e-02 7.01734781e-01 1.08922027e-01 2.81630129e-01 5.91517031e-01 -7.04739213e-01 -3.27455103e-01 -1.75883785e-01 -3.10310334e-01 -6.50897697e-02 -2.62824208e-01 4.14271086e-01 -8.72510493e-01 -3.02966386e-01 -1.65761754e-01 -1.53235286e-01 1.04098570e+00 4.54083145e-01 1.28433645e+00 2.22610041e-01 -5.37144303e-01 7.67301857e-01 1.46644628e+00 -7.89462551e-02 4.89528567e-01 7.14863718e-01 9.40410674e-01 2.80380666e-01 6.85422778e-01 5.05094826e-01 3.88563186e-01 7.73849726e-01 5.32752514e-01 -1.84507355e-01 -1.21546581e-01 1.80544909e-02 1.06847994e-02 8.13068986e-01 2.96868887e-02 -7.74999499e-01 -1.36834943e+00 4.81177837e-01 -1.65961254e+00 -9.13693547e-01 -1.12392306e+00 1.86317265e+00 3.78170878e-01 2.55073756e-01 2.34839648e-01 3.47119302e-01 8.52606833e-01 2.05390736e-01 -4.39838558e-01 7.64517870e-04 -6.36407673e-01 -2.21395642e-01 9.82462764e-01 -1.39741048e-01 -1.43960011e+00 1.27552104e+00 5.34516764e+00 1.25792909e+00 -1.32488465e+00 -3.17379609e-02 6.55336201e-01 2.61357039e-01 2.19845235e-01 -2.89314598e-01 -9.75084424e-01 3.66697788e-01 2.35103101e-01 -8.75605643e-02 -5.81405759e-02 8.87902558e-01 2.84033656e-01 -3.54717553e-01 -6.36884749e-01 1.09616518e+00 2.74681538e-01 -1.08186960e+00 1.92773357e-01 1.15935780e-01 7.61622250e-01 6.70703590e-01 9.87010747e-02 2.05996633e-01 9.46511775e-02 -9.50404763e-01 1.03593516e+00 -1.00956224e-01 8.68117094e-01 -4.03289318e-01 6.44842982e-01 4.69472587e-01 -1.33690393e+00 1.11372463e-01 -5.08278191e-01 4.85656857e-01 -8.77596140e-02 5.83038568e-01 -8.40251327e-01 9.70927536e-01 1.05827153e+00 9.90846992e-01 -1.06423163e+00 1.09494209e+00 2.32792005e-01 1.00620878e+00 -6.14639461e-01 -3.32468771e-03 4.90112066e-01 -4.32721883e-01 7.64485955e-01 1.69929934e+00 2.91511744e-01 -3.49858820e-01 3.89992148e-01 4.25386280e-01 -7.07969740e-02 5.08291423e-01 -6.30608261e-01 5.77389598e-02 1.09999873e-01 1.29721594e+00 -1.54869068e+00 -4.05699283e-01 -3.91411871e-01 9.70670521e-01 -4.01378453e-01 3.15040946e-01 -1.09889758e+00 -3.51705015e-01 -3.24866921e-01 2.24099010e-01 3.30298096e-01 -2.51634836e-01 -5.75005233e-01 -1.39272606e+00 8.48703459e-02 -9.56464410e-01 3.90011519e-01 -9.67071056e-01 -1.01313984e+00 6.62187517e-01 5.45325093e-02 -1.33718896e+00 3.97400528e-01 -8.38992596e-01 -4.73398507e-01 3.45404536e-01 -1.15915990e+00 -1.61672735e+00 -7.03596354e-01 8.80449533e-01 1.37834060e+00 -1.39896378e-01 3.23976040e-01 3.60842973e-01 -8.49837363e-01 5.03783047e-01 5.82124531e-01 3.82664233e-01 8.71118486e-01 -1.45541346e+00 5.17226219e-01 7.72158802e-01 2.51542151e-01 1.77018657e-01 9.56522524e-01 -7.88935244e-01 -1.53440535e+00 -1.11437094e+00 1.39636427e-01 -5.42847216e-01 6.80884659e-01 -4.55792397e-01 -6.77305996e-01 8.62784505e-01 4.53284830e-01 -2.34181434e-01 -8.56617745e-03 -2.81862676e-01 -1.95653111e-01 3.31823379e-01 -9.48631585e-01 8.09168100e-01 9.48910952e-01 8.98131803e-02 -6.56726837e-01 5.57148278e-01 2.81241715e-01 -7.00163066e-01 -3.64983469e-01 3.75174105e-01 2.45910481e-01 -1.32350779e+00 7.62080014e-01 2.76114255e-01 4.61808532e-01 -2.38466829e-01 -2.13247687e-01 -8.89884770e-01 4.35278922e-01 -4.52722281e-01 4.90713298e-01 1.28179955e+00 8.34244192e-02 -8.02883625e-01 9.45592642e-01 -2.91738808e-01 -4.82356012e-01 -7.12536693e-01 -9.05681491e-01 -6.73708916e-01 4.96097535e-01 -5.95403135e-01 1.74500316e-01 1.16169262e+00 -4.35585856e-01 2.43589833e-01 -2.02963501e-01 -6.22769408e-02 6.90485835e-01 -9.98881757e-02 1.09016025e+00 -1.25878990e+00 3.79444435e-02 -5.57358325e-01 -4.01611000e-01 -1.31327319e+00 -4.05790150e-01 -1.09788775e+00 -3.04729491e-01 -1.65429497e+00 1.24225080e-01 -3.68813157e-01 4.89554524e-01 3.08995962e-01 1.84385046e-01 4.31416065e-01 2.06326663e-01 3.25533241e-01 -6.35727704e-01 4.43719357e-01 1.31697476e+00 -2.33607814e-01 1.45815667e-02 2.36419290e-02 -1.04244076e-01 1.16164494e+00 9.31092441e-01 -5.19280136e-01 -3.39201599e-01 -4.36199397e-01 5.12510478e-01 -1.83201462e-01 1.44691318e-01 -1.14174283e+00 2.56370097e-01 -4.84638624e-02 3.46382260e-01 -1.73916423e+00 1.52742252e-01 -9.30727959e-01 -1.50125727e-01 1.64537326e-01 -1.03956414e-02 6.26429319e-02 1.26045674e-01 5.56158245e-01 -1.36100471e-01 -4.20347750e-01 8.26286972e-01 -4.88309413e-02 -4.99004066e-01 8.56851041e-02 -4.49171454e-01 6.11009240e-01 1.11768448e+00 -7.07163095e-01 -5.15055895e-01 -2.13100594e-02 -2.91153342e-01 4.40017223e-01 5.30283988e-01 6.51988626e-01 4.34344858e-01 -5.49966693e-01 -7.20096111e-01 -1.43245921e-01 2.62886733e-01 3.40010345e-01 2.90066689e-01 1.11520350e+00 -1.07451928e+00 6.22769356e-01 1.11430891e-01 -1.19505608e+00 -1.60550833e+00 2.00203180e-01 5.40357351e-01 -1.66517526e-01 -1.13670218e+00 4.21002775e-01 4.13486272e-01 -5.87299049e-01 3.22784573e-01 -4.55214143e-01 -9.69136357e-02 8.81740600e-02 -1.17247570e-02 6.18196309e-01 2.61775553e-01 -9.14311647e-01 -2.66311675e-01 1.18672597e+00 -1.22875594e-01 -1.79671705e-01 1.25538230e+00 -2.49550566e-01 1.25470787e-01 3.85419518e-01 8.99183750e-01 1.86355874e-01 -9.23205614e-01 2.72521023e-02 1.29571229e-01 -4.93750244e-01 9.62191448e-02 -9.35819209e-01 -1.12720370e+00 1.03188705e+00 6.77492142e-01 3.45585972e-01 8.32338810e-01 -1.07272901e-01 6.14593446e-01 5.40650427e-01 2.67358184e-01 -1.39319777e+00 2.87374377e-01 6.39590442e-01 8.67287040e-01 -1.32408857e+00 2.05857962e-01 -6.26168132e-01 -7.74579108e-01 1.31756866e+00 4.65434819e-01 -2.06049830e-02 5.23963809e-01 4.40006107e-01 1.44385040e-01 -5.13989687e-01 -2.95075506e-01 -1.79373175e-01 1.73189089e-01 3.69542539e-01 6.10236168e-01 3.71721992e-03 -3.19934070e-01 8.55520591e-02 -5.33074796e-01 -2.18743369e-01 9.30460989e-01 8.35664272e-01 -3.27021152e-01 -7.87431836e-01 -7.23123670e-01 6.65720940e-01 -8.62057030e-01 -6.67615533e-02 -4.88130659e-01 1.11553204e+00 3.04207224e-02 1.03595304e+00 -1.46316364e-01 -5.51621504e-02 4.66831774e-01 -5.84257245e-02 7.06523806e-02 -3.35811406e-01 -7.97035813e-01 5.54351807e-01 -8.10368732e-02 -2.77442247e-01 -1.86318800e-01 -8.63136649e-01 -1.36942971e+00 -4.12361979e-01 -6.83843255e-01 -2.96244889e-01 9.29579318e-01 7.62207806e-01 -5.09132668e-02 5.65775394e-01 3.43733728e-01 -8.30303907e-01 -3.67928714e-01 -1.12054241e+00 -5.73045969e-01 5.67720771e-01 -6.64590672e-02 -5.87833166e-01 -5.64986229e-01 2.74272591e-01]
[12.061807632446289, 2.278257369995117]
992dbc01-1620-4524-a1c2-4d0381ec9f65
lyricsim-a-novel-dataset-and-benchmark-for
2306.01325
null
https://arxiv.org/abs/2306.01325v1
https://arxiv.org/pdf/2306.01325v1.pdf
LyricSIM: A novel Dataset and Benchmark for Similarity Detection in Spanish Song LyricS
In this paper, we present a new dataset and benchmark tailored to the task of semantic similarity in song lyrics. Our dataset, originally consisting of 2775 pairs of Spanish songs, was annotated in a collective annotation experiment by 63 native annotators. After collecting and refining the data to ensure a high degree of consensus and data integrity, we obtained 676 high-quality annotated pairs that were used to evaluate the performance of various state-of-the-art monolingual and multilingual language models. Consequently, we established baseline results that we hope will be useful to the community in all future academic and industrial applications conducted in this context.
['Elena González-Blanco', 'Salvador Ros', 'Víctor Fresno', 'Pedro Hernández', 'Adrián Ghajari', 'Alejandro Benito-Santos']
2023-06-02
null
null
null
null
['semantic-textual-similarity', 'semantic-similarity']
['natural-language-processing', 'natural-language-processing']
[-5.84401563e-02 -3.91437829e-01 -5.91761582e-02 -5.15369594e-01 -1.13398957e+00 -1.11713541e+00 4.44006354e-01 2.79937565e-01 -6.18500948e-01 7.66909659e-01 4.36230004e-01 3.45549196e-01 1.39010116e-01 -2.91251093e-01 -3.50230068e-01 -1.76821336e-01 4.17209193e-02 7.20251918e-01 2.23668665e-01 -3.78038853e-01 2.35215530e-01 4.78259176e-02 -1.52986538e+00 5.08182466e-01 5.99947512e-01 7.30359614e-01 8.11558217e-02 4.25862581e-01 2.47692987e-01 7.47619152e-01 -7.36381710e-01 -8.80497575e-01 1.81271687e-01 -4.86912429e-01 -1.05671203e+00 -5.46366751e-01 7.59492218e-01 3.21903229e-01 4.03985158e-02 1.10730898e+00 8.16593468e-01 2.97633082e-01 3.99654627e-01 -1.07870877e+00 -6.68434918e-01 9.65198219e-01 8.41268972e-02 1.27417997e-01 8.56029093e-01 -1.18901268e-01 1.66645670e+00 -9.46826160e-01 1.03512990e+00 9.21017468e-01 8.39183152e-01 5.24928153e-01 -9.49316978e-01 -8.90178859e-01 -4.01797295e-01 3.91619682e-01 -1.78279674e+00 -6.55736625e-01 6.83567464e-01 -4.06416953e-01 9.28776801e-01 5.21208167e-01 5.61760604e-01 1.16120625e+00 -2.80184507e-01 5.70072711e-01 1.04464865e+00 -4.93415594e-01 -1.81209862e-01 2.01617554e-01 -3.52443904e-02 3.84324968e-01 -3.18984628e-01 -1.38607845e-01 -9.40746129e-01 -2.85263956e-01 2.72588581e-01 -7.94482410e-01 -3.64245415e-01 -3.60149503e-01 -1.66677737e+00 7.69735634e-01 2.99652308e-01 6.85656011e-01 1.38047725e-01 -3.18692893e-01 7.81926155e-01 5.12200356e-01 4.73149896e-01 1.14099777e+00 -4.64271665e-01 -5.26110590e-01 -8.71628046e-01 3.45137089e-01 1.08189058e+00 1.09579420e+00 3.78973097e-01 -1.65642217e-01 -8.30380395e-02 1.31907344e+00 -5.40587120e-02 4.70984817e-01 5.46779573e-01 -1.25801134e+00 4.61426526e-01 2.91158825e-01 -1.21294037e-02 -9.12489176e-01 -2.60199696e-01 -4.79401976e-01 -4.27792370e-01 -6.33753359e-01 4.01461840e-01 1.80830091e-01 -4.77098674e-02 1.89826989e+00 -2.31611282e-02 2.58898228e-01 1.03715576e-01 9.37362492e-01 1.11052954e+00 3.04035306e-01 -2.60476470e-02 -2.10868955e-01 1.33164322e+00 -1.24487770e+00 -6.08755052e-01 1.83684424e-01 5.11906505e-01 -1.58031607e+00 1.41755271e+00 2.90375233e-01 -1.00456643e+00 -9.11810279e-01 -9.07584667e-01 -1.67607710e-01 -3.06121618e-01 2.00724423e-01 2.11385399e-01 4.33849186e-01 -9.10685599e-01 5.19580543e-01 -1.52910769e-01 -8.24682713e-01 -5.71882762e-02 6.38090372e-02 -3.89191568e-01 1.49299458e-01 -1.51070738e+00 8.38010967e-01 4.40491647e-01 -2.17253745e-01 -9.46570754e-01 -8.53927851e-01 -4.70396340e-01 -4.28160220e-01 3.02189231e-01 -5.09000421e-01 1.55557740e+00 -7.63691723e-01 -1.13757300e+00 1.41818357e+00 -2.12904345e-02 -3.16365778e-01 2.41274998e-01 -3.05249214e-01 -8.06960583e-01 -1.37756780e-01 5.77694356e-01 7.64722764e-01 2.05051973e-01 -9.61913586e-01 -6.27631187e-01 -2.27610782e-01 -5.58778048e-02 5.58519721e-01 -5.74394703e-01 5.73604405e-01 -7.13851571e-01 -1.06460011e+00 -3.16524833e-01 -1.27604580e+00 1.93710327e-01 -5.11296034e-01 -2.70784318e-01 -2.82727867e-01 4.10595983e-01 -8.20268154e-01 1.36351156e+00 -2.33108306e+00 4.79072630e-01 -1.01073824e-01 -2.31315270e-02 -2.99504725e-03 -4.02383238e-01 6.75994515e-01 1.57618046e-01 -4.21135426e-02 -1.66863129e-01 -3.83454651e-01 5.18324561e-02 -5.92896603e-02 -4.58961129e-01 3.04075032e-01 -3.72374892e-01 6.87148809e-01 -1.19216943e+00 -5.92996359e-01 -1.27081394e-01 2.05991864e-01 -4.27827150e-01 6.13955796e-01 -1.08560234e-01 7.46329427e-01 -1.10445082e-01 6.28229558e-01 -1.69398963e-01 9.71449241e-02 2.87561148e-01 -4.91441637e-01 -4.89199758e-02 6.69340372e-01 -8.07582557e-01 2.44599891e+00 -6.04762912e-01 8.15421581e-01 -1.06337473e-01 -5.09457231e-01 1.15196145e+00 4.38221335e-01 5.97577870e-01 -6.37872398e-01 -8.81315544e-02 5.98982215e-01 -4.99836132e-02 -2.08926648e-01 1.06018281e+00 8.04703310e-02 -6.60163939e-01 4.55171853e-01 4.54992622e-01 -6.62393332e-01 6.25956237e-01 1.13025121e-01 7.09175825e-01 2.16208044e-02 4.51159388e-01 -5.41383266e-01 7.77074039e-01 2.78812051e-01 4.60066527e-01 4.56904262e-01 -2.49378562e-01 6.10218346e-01 -1.44763030e-02 -5.22085309e-01 -1.29381955e+00 -1.16077828e+00 -2.70615101e-01 1.44677246e+00 -1.15870982e-01 -9.94425833e-01 -7.22490489e-01 -6.80014491e-01 -3.32381308e-01 6.14129663e-01 -2.12228164e-01 1.52879491e-01 -4.61922616e-01 -2.64730841e-01 1.23388255e+00 3.03212315e-01 3.16782862e-01 -1.19155455e+00 -7.83136114e-02 9.60128307e-02 -6.59044206e-01 -1.25967824e+00 -9.53813612e-01 -4.05650400e-02 -2.81449080e-01 -1.32475519e+00 -5.47833383e-01 -1.24223697e+00 -6.92685023e-02 -1.91573761e-02 1.73187673e+00 -2.18832850e-01 -1.39611110e-01 4.53330398e-01 -5.79525888e-01 -2.14972034e-01 -5.92822134e-01 5.82898557e-01 3.00541073e-01 -3.96936357e-01 5.79082668e-01 -4.10969228e-01 -1.06721290e-03 6.18906319e-01 -5.60599208e-01 -2.94286638e-01 -2.38199625e-02 6.35015070e-01 7.11155236e-01 -2.77615309e-01 6.03701532e-01 -7.46518314e-01 9.46642399e-01 -2.17285648e-01 -4.69543576e-01 3.25006634e-01 -3.28120232e-01 -3.29099000e-01 8.59531820e-01 -3.25498968e-01 -6.78117692e-01 1.13083785e-02 -1.90137193e-01 -2.33502537e-01 -1.89408839e-01 5.51716387e-01 -5.19014113e-02 -1.70644131e-02 6.90441906e-01 -4.19876911e-02 -5.49908519e-01 -8.61711383e-01 6.10737801e-01 1.03056622e+00 1.09813488e+00 -7.68867373e-01 4.87097412e-01 -1.44248351e-01 -3.57509375e-01 -7.99155831e-01 -1.11368084e+00 -7.37436593e-01 -8.15291047e-01 -2.75465220e-01 9.08606052e-01 -1.18474972e+00 -3.84882957e-01 3.65775019e-01 -1.10143924e+00 -1.12036839e-01 -3.81658643e-01 3.41370612e-01 -6.91378057e-01 2.49734849e-01 -5.78260601e-01 -1.72410905e-01 -4.04149771e-01 -9.46330607e-01 1.07241786e+00 -1.01903975e-01 -9.69543159e-01 -1.12788630e+00 8.03029180e-01 6.56247914e-01 2.33779401e-01 -1.41269282e-01 6.27584696e-01 -1.12349403e+00 2.51434427e-02 -4.29724418e-02 1.74691290e-01 3.33407611e-01 1.60369933e-01 -1.43114299e-01 -1.00983393e+00 -3.71844262e-01 -2.02463061e-01 -9.60372329e-01 3.83680016e-01 -2.90682524e-01 9.52737451e-01 8.76419898e-03 1.70911342e-01 4.57372606e-01 8.78149748e-01 1.66402590e-02 6.59074038e-02 4.31556702e-01 8.87137413e-01 6.67938828e-01 9.25832450e-01 1.44195154e-01 4.66088146e-01 1.35920787e+00 -1.02275573e-01 2.24189237e-01 -3.28293085e-01 -4.59530681e-01 2.90324658e-01 1.74543953e+00 -3.02706845e-02 -1.70094907e-01 -1.05517137e+00 8.13651443e-01 -1.71978176e+00 -8.31740022e-01 -8.60930011e-02 2.10618138e+00 1.32291496e+00 -2.87201822e-01 2.74986446e-01 -4.24141027e-02 6.49215817e-01 3.18998069e-01 -9.05293897e-02 -4.48167026e-01 -4.32629377e-01 1.68691695e-01 4.28488590e-02 3.83636296e-01 -1.15003502e+00 1.25785780e+00 7.19846916e+00 1.20961177e+00 -7.53949344e-01 1.93682626e-01 1.94636062e-01 -1.20572299e-01 -2.78046936e-01 -9.53596085e-02 -7.49685287e-01 2.63543636e-01 1.10077178e+00 -3.76704067e-01 8.35183680e-01 7.44997025e-01 -8.21471959e-02 4.15162832e-01 -1.18508768e+00 9.42754209e-01 3.95013303e-01 -1.07749403e+00 -1.50470853e-01 -4.71542835e-01 1.02324450e+00 4.77434486e-01 -2.50157684e-01 3.35745722e-01 2.74528623e-01 -8.32479179e-01 7.99738884e-01 4.70764995e-01 9.00486171e-01 -6.84315979e-01 7.40992725e-01 4.79208957e-03 -1.38491988e+00 3.19683015e-01 -2.06029475e-01 -1.36228083e-02 2.42206842e-01 3.14318836e-01 -8.17681611e-01 7.78129280e-01 8.19787204e-01 1.27320623e+00 -8.94195259e-01 8.88317168e-01 -3.71360600e-01 5.92966914e-01 -8.93821865e-02 -1.17808782e-01 1.69316214e-02 -8.88648406e-02 7.52253592e-01 1.24163330e+00 8.40782002e-02 -4.88882929e-01 4.26264822e-01 6.20887280e-01 -3.34392548e-01 7.15646148e-01 -5.61302662e-01 -3.00047129e-01 7.29645073e-01 1.26029444e+00 -4.23656046e-01 -1.98092520e-01 -3.49802703e-01 9.86454189e-01 5.11199474e-01 -7.76196923e-03 -7.63549805e-01 -4.16356415e-01 6.63496017e-01 -2.07680270e-01 -1.61956340e-01 -1.97574332e-01 -6.30274117e-02 -1.18084240e+00 -1.17684670e-01 -1.22681379e+00 6.57677710e-01 -9.21651840e-01 -1.78065002e+00 1.08718848e+00 -4.40603532e-02 -1.30819798e+00 -5.41539550e-01 -2.52811670e-01 -1.56725943e-01 6.32116258e-01 -9.67292309e-01 -1.27016878e+00 -2.34189987e-01 5.14470160e-01 6.28320277e-01 -8.32293391e-01 1.27812815e+00 7.11229622e-01 -1.14633039e-01 6.29360020e-01 -1.68653801e-02 2.08378136e-01 1.49992549e+00 -1.15579283e+00 3.86765003e-01 4.69783187e-01 1.09387064e+00 6.61759496e-01 6.74073219e-01 -4.81320351e-01 -8.51443589e-01 -1.07011199e+00 1.38198996e+00 -8.19456041e-01 1.16561508e+00 -2.76435673e-01 -6.82579517e-01 5.44857264e-01 4.71569240e-01 -3.10814023e-01 1.05142331e+00 3.31957310e-01 -5.16141117e-01 -1.01310134e-01 -5.99206328e-01 3.77962142e-01 1.12589121e+00 -1.21802557e+00 -8.98756385e-01 2.82638878e-01 8.19809556e-01 -4.67744023e-01 -1.40228486e+00 6.11757338e-01 7.53819287e-01 -6.50740862e-01 8.57266724e-01 -6.88278437e-01 2.02440664e-01 -3.29892367e-01 -5.91141105e-01 -1.23687530e+00 -8.99337158e-02 -5.34987569e-01 4.96552229e-01 1.57983696e+00 5.44230998e-01 7.05829933e-02 3.35225493e-01 -1.98364928e-01 -4.10726577e-01 -1.99346885e-01 -9.53404725e-01 -9.27205861e-01 5.96335270e-02 -6.17075980e-01 5.77295303e-01 1.29908681e+00 6.19791038e-02 8.17063868e-01 -4.96519953e-01 -3.17509234e-01 2.97127783e-01 2.85972476e-01 8.85738552e-01 -9.60552037e-01 -2.19291031e-01 -4.86306310e-01 -3.75073642e-01 -6.37546122e-01 6.35967195e-01 -1.25679207e+00 5.09904623e-02 -8.30362320e-01 3.71079683e-01 -4.44993168e-01 -3.59861642e-01 4.57164466e-01 -1.01766638e-01 1.17818105e+00 2.05746904e-01 4.30777282e-01 -1.21612501e+00 5.17431080e-01 8.07297230e-01 -1.66668117e-01 -1.21673308e-01 -9.47862342e-02 -3.95315081e-01 6.25698984e-01 8.05727005e-01 -5.77037156e-01 -2.33194649e-01 -5.20162046e-01 2.21472338e-01 -4.46663141e-01 7.27850497e-02 -1.06558597e+00 1.52208716e-01 6.98630959e-02 -1.96762443e-01 -3.94031346e-01 4.16513652e-01 -6.53652489e-01 3.44960660e-01 1.39677256e-01 -6.57413065e-01 2.74906337e-01 8.14389586e-02 8.37996006e-02 -8.36185455e-01 -3.05098087e-01 4.38337207e-01 2.44748909e-02 -8.34413409e-01 -1.10560432e-01 -9.01293606e-02 6.18907988e-01 8.31521988e-01 4.58799332e-01 -1.66065380e-01 -5.38117349e-01 -6.31249964e-01 2.00103633e-02 8.38415861e-01 1.09137392e+00 4.98261191e-02 -1.71617091e+00 -9.49218392e-01 -5.61514869e-02 7.64060736e-01 -6.91654265e-01 -2.02404663e-01 6.17158532e-01 -3.94651562e-01 7.39036202e-01 -2.69395739e-01 -4.19372529e-01 -1.57323086e+00 2.95165360e-01 1.07224144e-01 -2.05042571e-01 1.67880580e-02 6.57078743e-01 -4.08124983e-01 -8.88688982e-01 3.09037268e-01 1.73980996e-01 -3.74778152e-01 1.24644615e-01 2.89453536e-01 4.35512722e-01 1.23069838e-01 -1.22370672e+00 -4.62031037e-01 6.34005010e-01 4.44922835e-01 -3.32750678e-01 1.04010761e+00 -6.14297353e-02 -4.49611157e-01 1.07632923e+00 1.17322242e+00 5.72606742e-01 -3.77555728e-01 -5.16315587e-02 1.54987007e-01 -4.27192390e-01 -2.93237835e-01 -9.21899021e-01 -7.93254554e-01 5.52177250e-01 3.96113932e-01 2.46114582e-01 9.85843480e-01 1.87871680e-01 7.35251427e-01 5.03956020e-01 4.33781385e-01 -1.21328080e+00 -8.92761126e-02 6.92440212e-01 8.83928776e-01 -1.11516023e+00 -8.95345509e-02 -2.97105640e-01 -7.94292629e-01 9.65313435e-01 5.79036951e-01 6.23511225e-02 3.44555944e-01 4.76591177e-02 2.88410336e-01 -3.09925266e-02 -6.87261164e-01 -1.24978729e-01 9.74229813e-01 6.23993635e-01 1.10256803e+00 1.00378945e-01 -7.53615975e-01 8.98684323e-01 -7.74085402e-01 -2.89100081e-01 6.59242645e-02 4.44794029e-01 -1.06725499e-01 -1.58663523e+00 -6.55734614e-02 -1.04927786e-01 -5.85853934e-01 -3.35629404e-01 -1.07272720e+00 7.33679175e-01 2.14571521e-01 1.06162012e+00 -1.06043592e-01 -5.07479846e-01 5.06716251e-01 3.11849639e-02 4.84724492e-01 -7.48414338e-01 -9.65848684e-01 6.02048300e-02 7.26849020e-01 -5.00959396e-01 -7.03859508e-01 -7.13134408e-01 -8.27122331e-01 -6.21827804e-02 -1.07940383e-01 6.05620384e-01 5.14246583e-01 7.57165015e-01 8.58781934e-02 3.99830639e-01 5.13313055e-01 -4.23801124e-01 -3.24495703e-01 -1.14913583e+00 -4.94709492e-01 7.62580156e-01 -3.49967718e-01 -3.88202965e-01 -1.35339081e-01 2.25334197e-01]
[10.91490364074707, 9.703363418579102]
ac998abd-08b1-45f4-8c63-53c71ac90edb
conditional-training-with-bounding-map-for
2103.12277
null
https://arxiv.org/abs/2103.12277v1
https://arxiv.org/pdf/2103.12277v1.pdf
Conditional Training with Bounding Map for Universal Lesion Detection
Universal Lesion Detection (ULD) in computed tomography plays an essential role in computer-aided diagnosis. Promising ULD results have been reported by coarse-to-fine two-stage detection approaches, but such two-stage ULD methods still suffer from issues like imbalance of positive v.s. negative anchors during object proposal and insufficient supervision problem during localization regression and classification of the region of interest (RoI) proposals. While leveraging pseudo segmentation masks such as bounding map (BM) can reduce the above issues to some degree, it is still an open problem to effectively handle the diverse lesion shapes and sizes in ULD. In this paper, we propose a BM-based conditional training for two-stage ULD, which can (i) reduce positive vs. negative anchor imbalance via BM-based conditioning (BMC) mechanism for anchor sampling instead of traditional IoU-based rule; and (ii) adaptively compute size-adaptive BM (ABM) from lesion bounding box, which is used for improving lesion localization accuracy via ABMsupervised segmentation. Experiments with four state-of-the-art methods show that the proposed approach can bring an almost free detection accuracy improvement without requiring expensive lesion mask annotations.
['S. Kevin Zhou', 'Hu Han', 'Long Chen', 'Han Li']
2021-03-23
null
null
null
null
['medical-object-detection']
['computer-vision']
[ 4.20984000e-01 2.59786695e-01 -5.99663079e-01 -1.01212390e-01 -1.39115560e+00 -9.70431720e-04 3.81397665e-01 4.25826967e-01 -3.90471011e-01 6.33455634e-01 1.73629578e-02 -6.11863375e-01 -4.07118537e-02 -5.97876430e-01 -4.62027609e-01 -9.95962381e-01 1.80940017e-01 5.96693575e-01 9.88152385e-01 2.03895301e-01 3.23810935e-01 6.87246263e-01 -9.78136539e-01 3.26799542e-01 1.00341630e+00 1.06486821e+00 6.17301106e-01 5.04349709e-01 -8.68867338e-02 6.26445770e-01 -2.74425268e-01 4.43531573e-02 3.74171942e-01 -3.72524053e-01 -6.78386509e-01 1.19100645e-01 2.28834137e-01 -2.81609923e-01 5.83241433e-02 9.30857658e-01 6.96585774e-01 -2.68171608e-01 1.24203837e+00 -7.73600757e-01 -3.73382010e-02 4.04794008e-01 -1.32457101e+00 4.42405850e-01 -8.56991112e-02 1.47906467e-01 5.44952154e-01 -9.15120661e-01 5.36539197e-01 8.89569163e-01 9.41801846e-01 5.06514847e-01 -1.18412054e+00 -6.52168930e-01 -2.19480209e-02 -3.75220478e-02 -1.58204758e+00 -4.90307286e-02 4.90628034e-01 -6.57562315e-01 6.32020056e-01 5.57033956e-01 5.61512053e-01 4.72981542e-01 5.23216784e-01 9.03537035e-01 1.04200685e+00 -6.57068789e-01 1.26950279e-01 2.27835447e-01 6.28456771e-02 9.10108149e-01 4.88758236e-01 -2.63403058e-01 1.21009894e-01 -3.19257706e-01 1.27865112e+00 1.59740336e-02 -4.22360688e-01 -5.55515468e-01 -1.06673634e+00 8.81663561e-01 7.76425362e-01 3.44690204e-01 -5.36830306e-01 -1.83913596e-02 4.38830733e-01 -3.79393816e-01 4.65244830e-01 9.50268656e-02 -3.34611312e-02 5.77439427e-01 -1.08197117e+00 9.57750082e-02 2.08807975e-01 7.02364028e-01 3.64090711e-01 -4.01682496e-01 -5.71767032e-01 9.09842432e-01 2.62975514e-01 2.57497102e-01 8.19585025e-01 -3.83732915e-01 3.63018155e-01 8.50757182e-01 -1.64165854e-01 -5.06451428e-01 -6.06630683e-01 -6.20120406e-01 -1.06430316e+00 3.95335972e-01 4.62712288e-01 5.25571850e-05 -1.52979243e+00 1.29448926e+00 7.52012908e-01 1.02241203e-01 -5.72467148e-01 9.25837696e-01 6.43422484e-01 1.96826667e-01 3.58418465e-01 -5.17187655e-01 1.46367502e+00 -8.89864564e-01 -4.79031265e-01 -2.43123770e-01 1.10141253e+00 -7.19201148e-01 9.63724554e-01 1.05506644e-01 -1.04363883e+00 -2.35047728e-01 -9.84686375e-01 1.28250152e-01 6.79704770e-02 4.01046306e-01 7.38970637e-01 8.52279544e-01 -9.18278158e-01 2.96550635e-02 -1.09297872e+00 -3.86016220e-01 9.07465219e-01 5.15750289e-01 -2.42489517e-01 -7.23656565e-02 -6.94046199e-01 1.03506708e+00 4.03786659e-01 -5.18572256e-02 -9.06752884e-01 -1.00482178e+00 -6.63210750e-01 -5.65628111e-01 5.28641343e-01 -6.22010767e-01 1.11525381e+00 -5.41947603e-01 -1.07685030e+00 1.06758690e+00 -2.12118030e-01 -5.49020827e-01 7.87512779e-01 2.20505074e-01 2.49347687e-02 3.48990440e-01 4.93818730e-01 8.10093045e-01 7.26699412e-01 -1.23104024e+00 -8.82631958e-01 -6.23811543e-01 -3.25373858e-01 3.05745929e-01 -1.64866075e-01 -1.17029533e-01 -5.78469813e-01 -7.23470151e-01 7.50037730e-01 -7.00263917e-01 -8.14614296e-01 4.44634378e-01 -5.54802001e-01 -1.90946743e-01 7.33832717e-01 -7.04359055e-01 1.64893866e+00 -1.72167468e+00 -2.67544240e-01 3.58649045e-01 2.14012489e-01 4.96330746e-02 5.03623784e-01 -4.18112218e-01 -8.94134790e-02 7.82444924e-02 -5.49251854e-01 -2.51312792e-01 -5.33905804e-01 4.88747284e-02 1.90482065e-01 8.02024245e-01 2.00315118e-01 6.64312303e-01 -5.60106277e-01 -1.23441303e+00 3.88789266e-01 1.51485890e-01 -8.53718162e-01 7.23508000e-02 3.61111239e-02 5.68709254e-01 -7.01643229e-01 1.01572871e+00 7.32364595e-01 -2.97355056e-01 -5.64982966e-02 -5.04960120e-01 1.07423766e-02 -1.11575566e-01 -1.35352206e+00 1.43554556e+00 -3.77440453e-01 1.19926944e-01 1.56245023e-01 -1.05935776e+00 5.45088053e-01 4.32017297e-01 6.43399775e-01 -4.02646989e-01 3.01597118e-01 5.81772506e-01 1.29241839e-01 -4.98945266e-01 5.88196144e-03 -4.70195323e-01 6.01604469e-02 2.79827327e-01 -3.91844451e-01 -2.86926866e-01 2.73758918e-03 9.90614593e-02 1.06565166e+00 -1.53410792e-01 8.68374765e-01 -3.53595138e-01 8.47940028e-01 2.64220566e-01 6.32102609e-01 8.19353878e-01 -6.14639223e-01 8.25483024e-01 4.00415212e-01 -7.59009570e-02 -7.86689222e-01 -8.19219470e-01 -8.53915691e-01 5.41392982e-01 1.83004156e-01 2.13999540e-01 -8.60631824e-01 -8.93163979e-01 -1.27747774e-01 3.93161356e-01 -8.95763516e-01 7.95772448e-02 -8.75114858e-01 -1.24603677e+00 4.65707779e-01 6.82002485e-01 2.78188348e-01 -7.49751627e-01 -7.87099659e-01 3.25101197e-01 -2.30663847e-02 -8.74495029e-01 -4.16039675e-01 3.96555245e-01 -1.16951334e+00 -1.13975096e+00 -1.22305179e+00 -9.00329471e-01 1.15156460e+00 2.36407384e-01 7.60927796e-01 2.13994443e-01 -7.83278108e-01 -6.08871840e-02 -3.05071592e-01 -4.33828443e-01 -4.10275787e-01 5.76074831e-02 -3.20706725e-01 -1.50663301e-01 4.74159680e-02 -1.68259263e-01 -9.09450471e-01 5.86373448e-01 -8.48004401e-01 3.05395037e-01 1.02659416e+00 9.64372158e-01 1.00542903e+00 -1.00154318e-01 6.48810804e-01 -1.07070136e+00 2.71070123e-01 -4.94491130e-01 -3.29130828e-01 2.36151114e-01 -4.98275548e-01 -3.73247236e-01 -2.49594543e-02 -5.40162265e-01 -1.00903881e+00 2.09474713e-01 -1.69218481e-01 -2.34426126e-01 -1.19586013e-01 1.85479179e-01 2.55622357e-01 -4.20374215e-01 8.58321488e-01 1.11503983e-02 6.58280682e-03 -1.70169622e-01 -1.96858831e-02 7.48316169e-01 4.16967839e-01 -4.24088180e-01 4.81904477e-01 7.36019433e-01 2.11835966e-01 -5.52524626e-01 -7.21689224e-01 -8.49527657e-01 -6.95087433e-01 -2.47872517e-01 9.74368393e-01 -7.81943083e-01 -1.53833166e-01 2.25725845e-01 -8.26918483e-01 -1.68930173e-01 -2.71031886e-01 5.74023306e-01 -3.76158893e-01 2.75104672e-01 -6.28836393e-01 -7.57940531e-01 -5.21664500e-01 -1.59000218e+00 1.28903198e+00 2.68127322e-01 -1.84935033e-01 -7.58933067e-01 -2.74508595e-01 4.04845715e-01 3.47992748e-01 4.85815376e-01 9.96799886e-01 -4.72707301e-01 -5.64699709e-01 -6.56759202e-01 -4.92608517e-01 8.13803971e-02 8.09320435e-02 -2.48539433e-01 -6.74055934e-01 -3.20502967e-01 -9.75734834e-03 -1.48920283e-01 8.40496123e-01 1.04833984e+00 1.28200305e+00 3.35043669e-01 -1.04934108e+00 5.21511674e-01 1.42182219e+00 3.47821638e-02 4.20050740e-01 3.80620599e-01 5.33219337e-01 1.75648138e-01 9.45117295e-01 4.01829809e-01 -5.56594096e-02 6.56526029e-01 5.80909550e-01 -3.39556813e-01 -5.80357254e-01 -8.30525756e-02 -1.62493050e-01 2.69739717e-01 2.39377283e-02 5.56515940e-02 -1.09155273e+00 7.52790809e-01 -1.56703818e+00 -3.29222560e-01 -4.66274053e-01 2.03051829e+00 1.01388371e+00 4.91113722e-01 3.47886160e-02 2.37824738e-01 9.40477073e-01 -1.83046028e-01 -5.62184751e-01 2.17372566e-01 2.84509540e-01 2.21767679e-01 6.47352874e-01 4.25351769e-01 -1.27380264e+00 7.09435046e-01 5.59035063e+00 1.53414333e+00 -1.05081403e+00 4.85882759e-01 9.16625023e-01 1.65150538e-01 6.44081458e-02 4.58331127e-03 -1.14022279e+00 2.08149582e-01 1.00720920e-01 3.12422842e-01 -6.21781588e-01 9.11694825e-01 1.54234007e-01 -8.53887200e-01 -8.81824374e-01 7.55284607e-01 -2.00863220e-02 -1.30908310e+00 1.36773661e-01 9.39199701e-02 7.28556335e-01 -3.00960064e-01 -2.03361481e-01 2.67431825e-01 -2.84092814e-01 -8.22648942e-01 5.61862648e-01 2.80461341e-01 1.14584935e+00 -4.16634083e-01 8.80870104e-01 5.35557210e-01 -1.19732165e+00 1.10353477e-01 -3.91358197e-01 4.38914269e-01 2.96602428e-01 6.84890449e-01 -1.40840280e+00 3.87802333e-01 5.25581419e-01 1.89872339e-01 -6.35697007e-01 1.42567241e+00 8.13105926e-02 5.92566311e-01 -4.62029010e-01 2.38623973e-02 2.29526520e-01 1.65699959e-01 6.30186379e-01 1.22519720e+00 2.20683351e-01 4.83754188e-01 2.81226873e-01 6.11174166e-01 2.50863791e-01 3.53827655e-01 -2.41951998e-02 5.67135334e-01 3.70713145e-01 1.24264228e+00 -1.30301797e+00 -4.27289814e-01 -1.17816366e-01 6.55544579e-01 2.91555990e-02 -5.88708557e-02 -9.35191751e-01 -3.68223619e-03 -7.76594505e-02 7.72505283e-01 1.11234359e-01 1.11082979e-01 -6.80338025e-01 -5.39523184e-01 -1.60037920e-01 -4.34385598e-01 7.30967760e-01 -4.31979686e-01 -1.00509477e+00 3.81720483e-01 -8.22388157e-02 -1.39141452e+00 1.79321855e-01 -5.87645710e-01 -6.39245868e-01 7.95797527e-01 -1.36916399e+00 -1.28854883e+00 -3.32152635e-01 4.99830663e-01 7.58135200e-01 2.19472110e-01 5.79832733e-01 4.54786450e-01 -6.21719420e-01 8.33642483e-01 -2.66196996e-01 7.45039955e-02 8.36550713e-01 -1.16451812e+00 -3.83816957e-01 6.67567670e-01 -5.92173040e-01 3.00124824e-01 4.36840028e-01 -8.61398876e-01 -8.76432955e-01 -1.15921021e+00 1.92675531e-01 -3.45599502e-01 4.88564014e-01 1.74830094e-01 -8.97807240e-01 5.23645401e-01 -3.42644423e-01 4.91850942e-01 5.08381963e-01 -5.45431376e-01 2.95003116e-01 3.21488529e-02 -1.59035552e+00 6.50939047e-01 7.03170478e-01 8.98819938e-02 -3.47811371e-01 5.12722611e-01 4.63128418e-01 -7.78109193e-01 -7.78198302e-01 1.05122817e+00 3.10500294e-01 -9.70038593e-01 1.08869898e+00 -2.75209278e-01 1.41567066e-01 -3.19677472e-01 -1.32497642e-02 -6.85271859e-01 -2.92033017e-01 -1.50960572e-02 5.33488765e-02 8.47437501e-01 4.63061661e-01 -5.93007982e-01 9.30500388e-01 3.80323529e-01 -5.43701828e-01 -1.41919589e+00 -1.20309818e+00 -3.48947853e-01 1.22666717e-01 -4.39578861e-01 1.92487556e-02 6.71777427e-01 -2.94861585e-01 -1.42765939e-01 1.89848080e-01 4.13537383e-01 7.37314284e-01 -1.70753390e-01 6.15009546e-01 -9.49489772e-01 -1.75384328e-01 -6.82653069e-01 -4.02391285e-01 -1.05025661e+00 -5.33064365e-01 -9.05404389e-01 8.64013955e-02 -1.66644752e+00 5.75133741e-01 -9.15126562e-01 -2.73929983e-01 3.64922822e-01 -4.60404575e-01 6.65848017e-01 -4.57034796e-01 3.55806142e-01 -2.83768445e-01 1.03398427e-01 1.64275169e+00 1.22915685e-01 -5.00452667e-02 2.30300114e-01 -4.78024840e-01 9.87026155e-01 4.90821511e-01 -6.46252394e-01 -9.80729088e-02 7.95809552e-02 -3.57581586e-01 4.73592997e-01 4.71279532e-01 -1.12313759e+00 3.20292503e-01 -1.74490377e-01 6.79310799e-01 -1.00204837e+00 2.07348898e-01 -6.09340012e-01 -2.86713988e-01 8.65426064e-01 -4.23993394e-02 -5.10720968e-01 1.80889174e-01 6.38103127e-01 -9.72508341e-02 -4.38096732e-01 1.35671353e+00 -2.79404879e-01 -4.53634739e-01 4.19218659e-01 -2.64626831e-01 -1.06487021e-01 1.50974798e+00 -6.38076246e-01 1.01385511e-01 2.37563968e-01 -8.90837252e-01 2.58552283e-01 1.17377438e-01 -6.50181845e-02 3.97823423e-01 -1.23243010e+00 -8.40750694e-01 2.32413530e-01 8.33456665e-02 4.84189391e-01 3.55320722e-01 1.63494658e+00 -7.40555227e-01 4.66177881e-01 2.44617268e-01 -1.08198059e+00 -1.26639891e+00 1.14840463e-01 6.30815864e-01 -7.36258507e-01 -7.32302368e-01 1.31306159e+00 5.81511319e-01 -3.06191295e-01 2.81088948e-01 -3.74798626e-01 -1.38161346e-01 3.06660328e-02 3.11390400e-01 1.60295635e-01 3.27949107e-01 -5.20252705e-01 -3.77932429e-01 6.58652365e-01 -5.38739145e-01 1.21363692e-01 8.70778799e-01 -4.28502038e-02 9.48076025e-02 2.57624686e-02 8.11562657e-01 -1.51491731e-01 -9.47316945e-01 -2.14339897e-01 -8.70808214e-03 -3.77035171e-01 5.04337370e-01 -7.67884731e-01 -7.73667336e-01 8.48656595e-01 1.08059084e+00 -6.28722906e-02 1.04109180e+00 2.45953023e-01 9.33731377e-01 -3.18199694e-01 4.63403732e-01 -8.45491707e-01 9.58016068e-02 -8.75449851e-02 8.99460793e-01 -1.43709385e+00 3.17029655e-01 -7.58976221e-01 -5.13966560e-01 9.88458395e-01 7.90162921e-01 -1.14945523e-01 7.49062121e-01 3.54234397e-01 -1.06944896e-01 -1.02326080e-01 -1.85707256e-01 -2.50478715e-01 4.32611942e-01 3.08839053e-01 4.11417305e-01 2.90298343e-01 -6.37944698e-01 6.99279904e-01 3.60276610e-01 -1.78845748e-01 1.47102416e-01 1.07070982e+00 -9.90346491e-01 -8.60698164e-01 -9.05153811e-01 9.12582397e-01 -4.97046411e-01 6.55742437e-02 1.46441013e-01 1.15097344e+00 3.96704495e-01 3.84135425e-01 -2.17295557e-01 2.28504270e-01 2.60722607e-01 -2.06048355e-01 5.43631673e-01 -8.59717250e-01 -3.75827909e-01 5.34458816e-01 -2.52977461e-01 -2.72334397e-01 -9.09446403e-02 -8.07564139e-01 -1.56748748e+00 4.23559248e-01 -1.00261104e+00 -1.04792058e-01 4.50154006e-01 9.16416764e-01 -2.41769910e-01 6.59733117e-01 3.49041820e-01 -9.35978472e-01 -6.94169760e-01 -1.11612105e+00 -4.69859600e-01 5.96440099e-02 2.13493884e-01 -9.72325802e-01 -3.49414319e-01 -1.38271317e-01]
[15.068865776062012, -2.3907337188720703]